2022-08-01 03:00:58 <user2> just finished my custom alpine iso with xfce :)
2022-08-01 03:01:34 <user2> it's working perfectly now
2022-08-01 03:02:08 <user2> alpine is very good!
2022-08-01 03:53:32 <sam_> sorry psykose not had a chance to read back over the log yet
2022-08-01 03:53:40 <psykose> nothin to be sorry for
2022-08-01 03:53:53 <psykose> i just nuked the rest of libexecinfo
2022-08-01 04:26:34 <sam_> oh nice
2022-08-01 04:26:44 <sam_> i bet 100 quid all of it was completely unnecessary
2022-08-01 04:27:27 <psykose> every single one was a random stacktrace for no reason
2022-08-01 04:27:29 <psykose> half in sigsegv
2022-08-01 04:27:31 <psykose> : )
2022-08-01 04:27:53 <psykose> but i bet the same, how bout we get a tenner each?
2022-08-01 05:18:55 <sam_> wfm thx
2022-08-01 06:14:08 <fcolista> psykose, i'm going to review and merge today your MR..i was away, sorry for the delay
2022-08-01 06:14:17 <psykose> mm? take your time :)
2022-08-01 06:14:38 <psykose> i don't think i even have one
2022-08-01 06:16:02 <fcolista> just to let you know that I've not forgot :D
2022-08-01 06:16:07 <fcolista> you have
2022-08-01 06:16:59 <psykose> are you suure
2022-08-01 06:18:03 <fcolista> https://gitlab.alpinelinux.org/alpine/aports/-/merge_requests/36967
2022-08-01 06:18:10 <fcolista> https://gitlab.alpinelinux.org/alpine/aports/-/merge_requests/36970
2022-08-01 06:18:13 <psykose> that's not me
2022-08-01 06:18:20 <fcolista> https://gitlab.alpinelinux.org/alpine/aports/-/merge_requests/37098
2022-08-01 06:18:23 <panekj> psykose: hi patrycja
2022-08-01 06:18:45 <panekj> fcolista: that's ptrc
2022-08-01 06:19:01 <fcolista> Arehttps://gitlab.alpinelinux.org/alpine/aports/-/merge_requests/36948
2022-08-01 06:19:03 <psykose> afternoon kaspar
2022-08-01 06:19:22 <panekj> it's mornin'
2022-08-01 06:19:24 <fcolista> Patrycja Rosa is not psykose ?
2022-08-01 06:19:26 <fcolista> :D
2022-08-01 06:19:26 <fcolista> ok
2022-08-01 06:19:40 <fcolista> psykose, nevermind :D
2022-08-01 06:20:04 <psykose> you got a laugh out of me, it's something :D
2022-08-01 06:31:39 <sam_> btw re sed
2022-08-01 06:31:41 <sam_> https://git.savannah.gnu.org/cgit/gnulib.git/commit/?id=6ee6c7189f4bafbb603ef043fdf44b84c3f7532a
2022-08-01 06:31:43 <sam_> i.e. it got fixed a year ago but it's gnulib hell
2022-08-01 06:31:45 <sam_> so you have to beg sed to make a release 
2022-08-01 06:31:47 <sam_> why do people hate releases
2022-08-01 06:32:25 <psykose> : )
2022-08-01 06:33:28 <ptrc> i love when the bugfix patch also contains 10 other random things, so you have to look carefully for the actual fix
2022-08-01 06:34:45 <psykose> binutils is worse, out of 9 hunks you keep 1
2022-08-01 06:35:54 <ptrc> fcolista: oh, thanks :) (it's me btw)
2022-08-01 06:36:13 <fcolista> ptrc, yeah now I got it :D
2022-08-01 09:40:12 <wener> Ariadne Hi, is it possible to add local file to apko without a @local repo ?
2022-08-01 10:20:04 <ikke> ncopa: !26263
2022-08-01 13:53:03 <Ariadne> wener: no
2022-08-01 14:40:58 <wener> ok, melange is the only way
2022-08-01 15:16:03 <psykose> ikke: uhh
2022-08-01 15:16:10 <psykose> busybox is missing from the 3.16 mirror
2022-08-01 15:16:20 <ikke> hmmm
2022-08-01 15:16:24 <psykose> ah no
2022-08-01 15:16:27 <psykose> used a wrong url
2022-08-01 15:16:28 <psykose> hmm
2022-08-01 15:16:37 <psykose> ok no idea then
2022-08-01 15:16:42 <psykose> but for some reason the 3.16 busybox is broken
2022-08-01 15:16:44 <ikke> I think they are missing alpine-base
2022-08-01 15:16:49 <ikke> or not?
2022-08-01 15:16:53 <psykose> no, broken with any list
2022-08-01 15:16:57 <psykose> works on edge and 3.15
2022-08-01 15:17:39 <psykose> sure, let's try the really dumb fix first since the package does exist
2022-08-01 15:18:03 <ikke> https://tpaste.us/4omJ 
2022-08-01 15:18:07 <psykose> oh
2022-08-01 15:18:07 <psykose> no
2022-08-01 15:18:08 <psykose> it is missing
2022-08-01 15:18:09 <psykose> https://dl-cdn.alpinelinux.org/alpine/v3.16/main/x86/
2022-08-01 15:18:10 <psykose> not here
2022-08-01 15:18:22 <psykose> only busybox-initscripts
2022-08-01 15:18:24 <ikke> I see
2022-08-01 15:18:36 <psykose> exists for x86_64
2022-08-01 15:19:04 <psykose> did you do any cleanups recently
2022-08-01 15:19:27 <ikke> no
2022-08-01 15:19:38 <panekj> https://build.alpinelinux.org/buildlogs/build-3-16-x86/main/busybox/busybox-1.35.0-r16.log
2022-08-01 15:19:59 <psykose> huh
2022-08-01 15:20:14 <psykose> it just.. failed to build and vanished? wtf
2022-08-01 15:20:57 <psykose> terminated looks like it was stopped 'intentionally' at the time, but why would it not get picked up on restart
2022-08-01 15:21:14 <psykose> i wonder how many random packages are missing now
2022-08-01 15:21:29 <ikke> psykose: it may have to do with those hangs
2022-08-01 15:21:39 <psykose> which ones specifically
2022-08-01 15:21:44 <ikke> bb
2022-08-01 15:21:45 <psykose> the sigpipe ones?
2022-08-01 15:21:47 <ikke> yes
2022-08-01 15:21:56 <psykose> yes but that's isolated
2022-08-01 15:22:03 <psykose> it shouldn't matter if it 'hangs' or does anything
2022-08-01 15:22:03 <ikke> isolated how?
2022-08-01 15:22:09 <psykose> in the sense of the apk is not there
2022-08-01 15:22:14 <psykose> and buildrepo should.. build it
2022-08-01 15:22:21 <psykose> it was skipping it this whole time but the apk isn't there
2022-08-01 15:22:26 <psykose> that's the much worse bug
2022-08-01 15:22:38 <ikke> Sadly I cannot investigate anymore since you already bumped it
2022-08-01 15:22:42 <panekj> wonderful software
2022-08-01 15:22:59 <psykose> aye, perhaps should have waited a bit in retrospect :/
2022-08-01 15:23:06 <psykose> i did not think buildrepo was actually skipping it
2022-08-01 15:23:18 <psykose> some really weird edge case
2022-08-01 15:23:28 <psykose> i don't think the build would matter as it wouldn't check any of that
2022-08-01 15:24:02 <psykose> are the folders just symlinked between local and 'repo' or is there some other hierarchy? if it's more than that it's possible it's only in one place
2022-08-01 15:24:06 <ikke> Maybe it _is_ an edge case, but now we never know
2022-08-01 15:24:33 <ikke> psykose: it just places packages in ~/packages like it would if you do it locally
2022-08-01 15:24:41 <psykose> right, and then how do they reach the mirror
2022-08-01 15:24:42 <ikke> nothing fancy going on there
2022-08-01 15:24:48 <ikke> rsync
2022-08-01 15:24:53 <ikke> by aports-build
2022-08-01 15:24:56 <psykose> hm
2022-08-01 15:25:16 <psykose> no way rsync would skip it if it was there, but also if it wasn't buildrepo wouldn't skip a build..
2022-08-01 15:25:36 <ikke> psykose: there _was_ a bug where rsync would skip files
2022-08-01 15:25:44 <psykose> persistently on the same file?
2022-08-01 15:25:48 <ikke> yes
2022-08-01 15:25:51 <psykose> might be that
2022-08-01 15:25:59 <psykose> or a really fucked up buildrepo edgecase
2022-08-01 15:26:00 <ikke> It should be fixed
2022-08-01 15:26:00 <psykose> oh well
2022-08-01 15:26:07 <ikke> but I'm checking it
2022-08-01 15:26:15 <psykose> speaking of it might be time to update all the builders :)
2022-08-01 15:26:38 <ikke> It would skip files where a tmpfile for that file was present on the destination (because the sync was interrupted or similar)
2022-08-01 15:27:55 <psykose> are tmpfiles in some consistent non-final-mirror directory? like it tries to rsync :/mirror and the tmp is in :/tmp/asfasfas.apk
2022-08-01 15:28:25 <ikke> They are placed in the destination directory
2022-08-01 15:28:31 <psykose> with the same name or renamed
2022-08-01 15:28:34 <ikke> renamed
2022-08-01 15:28:36 <psykose> ah
2022-08-01 15:28:43 <ikke> .~xxxx
2022-08-01 15:28:45 <ikke> something like that
2022-08-01 15:28:55 <ikke> A directory
2022-08-01 15:29:07 <ikke> (due to --delay-updates)
2022-08-01 15:29:17 <psykose> can you check for that in the mirror dir real quick
2022-08-01 15:29:21 <psykose> (yes it's late but)
2022-08-01 15:30:01 <ikke> psykose: I was already doing that
2022-08-01 15:30:05 <psykose> aha
2022-08-01 15:31:10 <ikke> https://github.com/WayneD/rsync/issues/192
2022-08-01 15:32:36 <ikke> https://tpaste.us/Mkx7 
2022-08-01 15:33:44 <psykose> if they stick around do they consistently block a package?
2022-08-01 15:33:52 <psykose> that's certainly something for edge
2022-08-01 15:34:20 <ikke> no, it should not
2022-08-01 15:34:22 <ikke> that was the bug
2022-08-01 15:34:35 <psykose> welp
2022-08-01 15:34:45 <psykose> guess i learnt a lesson then
2022-08-01 15:34:51 <psykose> simply don't do anything for a bit :p
2022-08-01 15:35:02 <ikke> yeah
2022-08-01 15:35:04 <psykose> now we'll never know
2022-08-01 15:38:05 <psykose> that was broken for like 2 weeks
2022-08-01 15:38:10 <psykose> and it took that long for someone to notice
2022-08-01 15:38:15 <psykose> really speaks to the usage of x86 :p
2022-08-01 15:38:26 <panekj> :P
2022-08-01 15:38:53 <psykose> granted, the only way to even see that is to try a clean chroot bootstrap ..? if you already have busybox, you won't see these kinds of errors
2022-08-01 15:48:38 <ncopa> packages in community that does not build or have failing tests with openssl3: thrift, fractal, git-branchless, gnome-shortwave, vdr
2022-08-01 15:49:09 <ncopa> i intend to merge openssl3 this week
2022-08-01 15:49:22 <panekj> 🎉
2022-08-01 15:49:39 <ikke> psykose: or everyone is using bash :P
2022-08-01 15:49:46 <ncopa> i originally planned to do it today but this day didnt go as expected
2022-08-01 15:50:03 <panekj> + coreutils
2022-08-01 15:50:09 <ikke> panekj: obviously
2022-08-01 15:50:17 <panekj> not really :P
2022-08-01 16:00:59 <skarnet> <ncopa> i originally planned to do it today but this day didnt go as expected
2022-08-01 16:01:04 <skarnet> story of my life
2022-08-01 16:02:27 <psykose> now that execinfo is gone what's the next thing i should nuke from aports
2022-08-01 16:12:30 <skarnet> give me a week and I’ll make you a list
2022-08-01 16:17:48 <psykose> nooo ur just gonna put my name on it 10 times
2022-08-01 16:19:59 <Ermine> psykose's death note
2022-08-01 16:23:24 <skarnet> main/psykose: demote to unmaintained
2022-08-01 16:23:33 <psykose> : ' (
2022-08-01 16:23:49 <panekj> revert main/psykose: demote to unmaintained
2022-08-01 16:24:26 <skarnet> community/panekj: why was this moved from testing again
2022-08-01 19:32:03 <JuniorJPDJ> hello, is there a chance that py3-cryptography and rust enablement on ibm and riscv will be backported to 3.16?
2022-08-01 20:11:33 <ptrc> JuniorJPDJ: i'm pretty sure upgrading py3-cryptography would be breaking for some people
2022-08-01 20:13:38 <JuniorJPDJ> Ok, thanks
2022-08-01 20:13:44 <ptrc> ah, maybe not, they seem to guarantee api stability ( https://cryptography.io/en/latest/api-stability/ )
2022-08-01 20:14:40 <panekj> it's already late enough
2022-08-01 20:18:31 <JuniorJPDJ> there are some incompatibilities between 3.4.8 and 37.0
2022-08-01 20:18:39 <JuniorJPDJ> https://cryptography.io/en/latest/changelog/
2022-08-01 20:19:37 <JuniorJPDJ> those doesnt look like a big things, but anyway those exist
2022-08-01 20:20:15 <panekj> just because releast notes doesn't note many things, doesn't mean new release don't introduce new bugs
2022-08-01 20:20:39 <panekj> typing hard yh
2022-08-01 21:38:48 <user2> setup-user should have a -s parameter for setting the sell, no?
2022-08-01 23:45:55 <Guest6833> it leaked my ip!
2022-08-02 06:46:41 <panekj> happens to the best of us
2022-08-02 07:30:42 <psykose> ncopa: i pushed the go rebuilds too
2022-08-02 07:53:45 <Danct12> ptrc: !37122 lgtm
2022-08-02 08:24:51 <ncopa> psykose: thanks
2022-08-02 08:25:08 <psykose> and i'll backport both once it's cleared
2022-08-02 08:25:55 <ncopa> i think i will push openssl3 today or tomorrow
2022-08-02 08:26:24 <psykose> you can do it now
2022-08-02 08:26:47 <psykose> i'm awake and will be here to fix whatever random stuff needs fixing if needed
2022-08-02 08:27:57 <ncopa> there are a few things that needs fix. im working on community/fractal as we speak
2022-08-02 08:28:17 <ncopa> ok. i'll rebase again and push and we take it from there
2022-08-02 08:28:59 <ncopa> btw... what is the proper way to update a dependency in Cargo.lock?
2022-08-02 08:29:17 <ncopa> i assume manually pasting in the checksum there is not the right way
2022-08-02 08:29:20 <psykose> no
2022-08-02 08:29:36 <psykose> cargo update --precise VER nameofthing
2022-08-02 08:29:40 <psykose> then diff .toml and .lock
2022-08-02 08:29:47 <psykose> ..usually
2022-08-02 08:29:49 <psykose> more or less that
2022-08-02 08:30:15 <psykose> tell me what specifically and i'll go double check the commands
2022-08-02 08:31:06 <ncopa> i updated openssl-sys in fractal, and some other.
2022-08-02 08:31:08 <psykose> or well, actually without the --precise
2022-08-02 08:31:11 <psykose> sure, one moment
2022-08-02 08:31:25 <ncopa> i think gnome -shortwave is next
2022-08-02 08:32:28 <ncopa> git-branchless and gnome-shortwave
2022-08-02 08:32:30 <psykose> what version is the fixed one
2022-08-02 08:32:38 <ncopa> that i dont know yet
2022-08-02 08:32:41 <psykose> aha
2022-08-02 08:32:54 <ncopa> a version that supports openssl3
2022-08-02 08:33:06 <psykose> if latest 0.9 works
2022-08-02 08:33:07 <psykose> then
2022-08-02 08:33:19 <psykose> https://img.ayaya.dev/hKLQE7eO80Oi
2022-08-02 08:33:20 <psykose> here
2022-08-02 08:33:31 <psykose> `cargo update -p openssl-sys`
2022-08-02 08:33:34 <ncopa> community/vdr also needs fix. i think there was some download error or something. dont remember
2022-08-02 08:33:48 <psykose> yeah that one is broken half the time :/ and doesn't support 3 anyway
2022-08-02 08:35:43 <psykose> yeah >=v0.9.64
2022-08-02 08:35:49 <psykose> so that patch should fix that
2022-08-02 08:37:20 <psykose> shortwave has an actual upgrade
2022-08-02 08:38:30 <ncopa> im pushing openssl3 now
2022-08-02 08:38:38 <psykose> ok
2022-08-02 08:38:42 <ncopa> pushed...
2022-08-02 08:47:06 <ncopa> i working on git-branchless
2022-08-02 08:47:16 <psykose> think there was a circle somewhere
2022-08-02 08:48:13 <ncopa> a circle?
2022-08-02 08:49:05 <psykose> oh, no
2022-08-02 08:49:15 <psykose> is >3 valid for 3.0.1 or similar?
2022-08-02 08:49:19 <psykose> or does it want '4'
2022-08-02 08:50:49 <ncopa> 3.0.1>3 
2022-08-02 08:51:08 <ncopa> 3 is equivalent to 3.0.0
2022-08-02 08:54:35 <psykose> aha
2022-08-02 08:55:58 <psykose> just found it weird ca-certificates built first, usually that's caused by circular deps
2022-08-02 08:57:45 <ncopa> hm.. yeah
2022-08-02 08:58:03 <psykose> can't find anything obvious
2022-08-02 08:58:25 <ikke> If only we had tools to detect circular deps
2022-08-02 08:59:27 <psykose> we do now, there is only a mildly related thing
2022-08-02 08:59:51 <psykose> will fix
2022-08-02 09:01:02 <ncopa> i might have messed something up while rebasing
2022-08-02 09:18:27 <psykose> you bumped all the kernels again
2022-08-02 09:18:45 <psykose> litespeed missed a pkgrel
2022-08-02 09:20:58 <ncopa> bah. it was due to the rebase
2022-08-02 09:21:53 <psykose> i will fix
2022-08-02 09:21:53 <psykose> dw
2022-08-02 09:21:57 <psykose> just going through the list
2022-08-02 09:22:08 <ncopa> thank you
2022-08-02 09:22:21 <psykose> (and no need to rush since will take long time anyways) :)
2022-08-02 09:23:58 <ncopa> i wonder if we can remove testing/php7*
2022-08-02 09:31:57 <psykose> one thing uses it
2022-08-02 09:32:03 <ikke> kopano apparently still depends on it 
2022-08-02 09:32:09 <psykose> ye
2022-08-02 09:32:20 <ncopa> i wonder if makedepends_host is used when calculating buildtime deps
2022-08-02 09:32:42 <ncopa> i guess its not, and that is the reason to why it tries to build ca-certificates first
2022-08-02 09:41:52 <ikke> ncopa: re kopano: #13471
2022-08-02 09:45:22 <omni> seems to be merged https://gitlab.alpinelinux.org/alpine/aports/issues/13471
2022-08-02 09:45:30 <omni> but after the latest release https://github.com/Kopano-dev/kopano-core/blob/master/RELNOTES.txt
2022-08-02 09:45:42 <omni> oh, w8
2022-08-02 09:46:06 <omni> the first link was supposed to be https://github.com/Kopano-dev/kopano-core/commit/466e22976c2fb3129cf99ed67328637e66e09b04
2022-08-02 09:52:03 <psykose> did someone take armv7 down or did it go poof`
2022-08-02 09:58:32 <psykose> ncopa: good work on all of this, btw
2022-08-02 09:58:34 <psykose> took a long time
2022-08-02 10:19:27 <ikke> psykose: seems to be still there?
2022-08-02 10:19:42 <psykose> it's not building anything
2022-08-02 10:20:01 <ikke> mqtt-exec crashed
2022-08-02 10:20:03 <psykose> (or the status page is broken i guess)
2022-08-02 10:20:05 <psykose> haha
2022-08-02 10:21:00 <ikke> Aug  2 09:31:25 build-edge-armv7 daemon.info supervise-daemon[434]: stopping /usr/bin/mqtt-exec, pid 435
2022-08-02 10:21:11 <ikke> Someone stopped it
2022-08-02 10:21:30 <ikke> ncopa: did you stop it?
2022-08-02 10:33:58 <ncopa> i stopped it to build openssl3
2022-08-02 10:51:54 <ikke> ncopa: I started it again, I hope that was not an issue
2022-08-02 10:52:04 <ncopa> no problem. thanks
2022-08-02 11:50:18 <psykose> ncopa: fb1ece8758eab07a2dc027f800e48945bb24709e
2022-08-02 11:51:10 <psykose> (you accidentally downgraded it to .18 instead of up to .58)
2022-08-02 12:24:44 <psykose> fixed
2022-08-02 12:35:18 <ncopa> wow... thats a..
2022-08-02 12:35:21 <ncopa> thanks
2022-08-02 12:36:21 <sindibad363> Hello everyone. I have a small issue, and a small problem. I am trying to run pihole on alpine linux which is arm64, which is on proxmox which is on raspberry pi. I tried to follow a script from gitlab but it didn't work when it go time to use or compile pihole-FTL. I was wondering if someone can nudge me in right direction. I am trying to compile it myself but I am getting some errors and am no sure what am i doing wrong.
2022-08-02 12:39:39 <sindibad363> If Anyone is wondering what script it is this https://gitlab.com/yvelon/pi-hole
2022-08-02 12:42:55 <panekj> pihole is not supported on alpine
2022-08-02 12:43:12 <sindibad363> meaning it wont compile ?
2022-08-02 12:43:46 <panekj> nope
2022-08-02 12:44:19 <panekj> you should wait for script maintainer response instead of cross-posting to other places
2022-08-02 12:44:40 <psykose> i don't see any errors
2022-08-02 12:44:56 <panekj> \/opt/pihole/gravity.sh: line 78: /usr/bin/pihole-FTL: Operation not permitted
2022-08-02 12:45:48 <sindibad363> well i know that but after that didn't worked i tried to compile it
2022-08-02 12:46:29 <sindibad363> but now it is showing me this https://pastebin.com/z3BVp2iT
2022-08-02 12:47:27 <psykose> add readline-dev
2022-08-02 12:47:53 <sindibad363> already added
2022-08-02 13:02:47 <psykose> v5.16.1-dirty
2022-08-02 13:02:47 <psykose> sumire < ./build/pihole-FTL --version
2022-08-02 13:02:58 <psykose> this is one of the funniest cmake uses i've seen
2022-08-02 13:03:04 <psykose> i guess it builds if you know how to patch it
2022-08-02 15:56:04 <Ariadne> skarnet: the TSC has decided to approve your proposal, with some modifications.  I outlined them in the bug.  the deadline for implementing the proposal is one month.
2022-08-02 16:07:10 <Ariadne> the modifications in question should, in concurrence with other TSC proposals, make it possible to have a busybox-less alpine system in 3.17
2022-08-02 16:07:49 <Ariadne> (though such a configuration will largely remain experimental)
2022-08-02 16:18:46 <mercenary> \o/
2022-08-02 16:20:32 <skarnet> Ariadne, and the rest of the TSC: awesome, thank you. I’ll begin to work on the spec tomorrow.
2022-08-02 16:20:33 <skarnet> \o/
2022-08-02 16:21:12 <Ariadne> skarnet: as we are doing other work to decouple busybox, please coordinate with us :)
2022-08-02 16:21:37 <skarnet> Sure. Are you going to touch busybox-initscripts, or can I hack at it?
2022-08-02 16:21:47 <Ariadne> busybox-initscripts is being sent to a farm upstate
2022-08-02 16:22:04 <skarnet> yes but I mean
2022-08-02 16:22:05 <psykose> it's putting on a mask and being renamed to busybox-openrc (which sounds like the same thing)
2022-08-02 16:22:13 <psykose> but the goals are now different
2022-08-02 16:22:18 <psykose> so send it to a farm upstate
2022-08-02 16:22:26 <psykose> :)
2022-08-02 16:22:33 <Ariadne> psykose: i mean as a distinct package
2022-08-02 16:22:40 <psykose> of course
2022-08-02 16:22:50 <Ariadne> skarnet: yes, feel free to do the changes outlined by the TSC, i will review them
2022-08-02 16:23:03 <panekj> BB alternative?
2022-08-02 16:23:05 <psykose> excited to see how well the shell thing goes too
2022-08-02 16:23:08 <psykose> i assume it will be huge
2022-08-02 16:23:21 <psykose> but i prefer quality to random size
2022-08-02 16:23:21 <skarnet> what’s the goal of busybox-openrc if not to host the rest of busybox-initscripts after mdev scripts have been extracted?
2022-08-02 16:23:39 <skarnet> sounds to me like busybox-openrc should be transient as well
2022-08-02 16:23:50 <Ariadne> skarnet: well, there is `busybox-openrc`, `busybox-mdev-openrc`, and `busybox-extras-openrc` (for crap like telnetd)
2022-08-02 16:23:58 <psykose> that is the goal, it's the init.d equivalent for everything in busybox
2022-08-02 16:24:06 <psykose> just.. if there's other providers
2022-08-02 16:24:07 <Ariadne> skarnet: we want them to be alongside the busybox aport
2022-08-02 16:24:10 <psykose> they should provide the virtual
2022-08-02 16:24:24 <psykose> instead of as you saw whatever mdev is :)
2022-08-02 16:24:33 <skarnet> yeah but my point was that the -openrc scripts should be service by service, not implementation-tied
2022-08-02 16:24:44 <Ariadne> skarnet: that comes later :)
2022-08-02 16:24:51 <skarnet> yeah so it’s transient
2022-08-02 16:25:09 <skarnet> might as well have kept the busybox-initscripts name :P
2022-08-02 16:25:18 <Ariadne> it does not fit current alpine policy
2022-08-02 16:25:23 <skarnet> true
2022-08-02 16:25:47 <skarnet> what’s the policy regarding fitting scripts in bb-openrc or bb-extras-openrc?
2022-08-02 16:26:02 <Ariadne> is the service in busybox or busybox-extras
2022-08-02 16:26:34 <skarnet> makes sense, but...
2022-08-02 16:26:38 <Ariadne> psykose: i think that freebsd shell will not be much larger than ash
2022-08-02 16:26:43 <skarnet> then what’s the policy on starting the services by default
2022-08-02 16:26:55 <Ariadne> alpine does not "start services by default"
2022-08-02 16:26:56 <skarnet> if someone installs bb-extras, should *all* the crap be started at boot?
2022-08-02 16:27:06 <Ariadne> no
2022-08-02 16:27:20 <skarnet> well as I’ve been subtly hinting at for the past 6 months
2022-08-02 16:27:20 <psykose> Ariadne: i really can't guess tbh
2022-08-02 16:27:31 <skarnet> busybox is in a special place wrt Alpine policies
2022-08-02 16:27:39 <skarnet> and yes, bb services *are* started by default
2022-08-02 16:27:47 <Ariadne> some bb services are
2022-08-02 16:27:51 <Ariadne> because you need them to boot
2022-08-02 16:28:01 <skarnet> so alpine doesn’t start service by default, except when it does
2022-08-02 16:28:01 <Ariadne> but we will not be adding any more
2022-08-02 16:28:33 <psykose> exceptions for the core of the system are more like implementation details
2022-08-02 16:28:35 <Ariadne> look, you got what you wanted, just maintain the status quo for now
2022-08-02 16:28:35 <skarnet> okay, so I suppose there should be a setup-bb-extras script or something
2022-08-02 16:28:43 <skarnet> yes
2022-08-02 16:28:51 <psykose> also, what bb-extra automatic startup are you referring to?
2022-08-02 16:28:55 <psykose> i haven't heard of this
2022-08-02 16:29:02 <skarnet> I am asking questions about how to implement the spec
2022-08-02 16:29:07 <psykose> aha
2022-08-02 16:29:17 <psykose> well, try to make it a bit better than now, at least
2022-08-02 16:29:20 <psykose> that gets you my approval :)
2022-08-02 16:29:23 <psykose> (from my heart)
2022-08-02 16:29:36 <psykose> and assume it's not followed already
2022-08-02 16:29:38 <Ariadne> skarnet: just split the initscripts but keep everything else the same, no `setup-bb-extras`
2022-08-02 16:29:50 <skarnet> okay
2022-08-02 16:30:38 <skarnet> can you confirm that today, if someone installs bb-extras, they’re on their own without a setup-* or even prewritten openrc scripts to help them?
2022-08-02 16:30:45 <Ariadne> yes, i can
2022-08-02 16:30:48 <skarnet> okay
2022-08-02 16:30:55 <Ariadne> well
2022-08-02 16:31:02 <skarnet> so there will be no regression, and small improvements. All good.
2022-08-02 16:31:03 <Ariadne> i cannot confirm "prewritten openrc scripts to help them"
2022-08-02 16:31:08 <psykose> 'or even prewritten' well.. you need something to start them at all
2022-08-02 16:31:11 <psykose> is that not there already
2022-08-02 16:31:14 <Ariadne> because there are initscripts for telnetd at least
2022-08-02 16:31:19 <skarnet> I will check
2022-08-02 16:31:26 <Ariadne> but there is no setup-*
2022-08-02 16:34:57 <Ariadne> skarnet: the TSC is also open to deprecating and removing mdev entirely in the future.  but this would have to be 3.18 or 3.19 at the earliest, and it would need a solid migration plan.
2022-08-02 16:35:20 <Ariadne> again, no love for busybox, etc :)
2022-08-02 16:35:50 <psykose> .post-upgrade if version<=x; then migrate
2022-08-02 16:35:54 <psykose> surely that never fails, right
2022-08-02 16:36:12 <skarnet> I’ll be there for the deprecation and migration plan
2022-08-02 16:38:45 <Ariadne> psykose: the keywords are "integration testing"
2022-08-02 16:39:42 <skarnet> who can I ask for help with integration testing when the time comes?
2022-08-02 16:40:05 <skarnet> plan-wise, infrastructure-wise, operation-wise
2022-08-02 16:40:36 <Ariadne> there will be a TSC work item to coordinate this type of testing initiative
2022-08-02 16:40:41 <skarnet> cool
2022-08-02 18:08:51 <ikke> go 1.19 released
2022-08-02 18:08:58 <ikke> https://go.dev/doc/go1.19
2022-08-02 18:09:34 <panekj> The compiler now uses a jump table to implement large integer and string switch statements. Performance improvements for the switch statement vary but can be on the order of 20% faster. (GOARCH=amd64 and GOARCH=arm64 only)
2022-08-02 18:10:02 <panekj> pretty much the only important thing from that changelog
2022-08-02 18:11:21 <ikke> go1.19 was kept smaller due to 1.18 generics
2022-08-02 18:11:29 <ikke> (less time for 1.19)
2022-08-02 18:14:19 <panekj> fixing generics (:
2022-08-02 18:18:24 <psykose> Ariadne: ash was a 68k (just blocks so 4k accuracy) diff (from removing it), chimera sh is 138k (default cflags)
2022-08-02 18:18:27 <psykose> seems perfectly ok to me
2022-08-02 18:19:00 <ikke> How is shell feature compattibiity?>
2022-08-02 18:19:09 <psykose> iirc it also has line editing and stuff, so more things
2022-08-02 18:19:12 <Ariadne> freebsd sh and ash are basically equivalent
2022-08-02 18:19:57 <psykose> the build system is meson (but this is easy to cross build i guess), and yes, as q66 said it has to be factored out a bit
2022-08-02 18:20:04 <psykose> i had to package two things to build it even though they're unused, haha
2022-08-02 18:20:11 <psykose> but yeah it's alright
2022-08-02 18:20:18 <skarnet> freebsd sh builds with meson?
2022-08-02 18:20:22 <psykose> the port
2022-08-02 18:20:27 <psykose> freebsd sh doesn't build on linux at all afaik
2022-08-02 18:20:39 <psykose> https://github.com/chimera-linux/chimerautils
2022-08-02 18:20:41 <skarnet> napalm all computers then the world
2022-08-02 18:21:02 <psykose> it was at least 10x as good as autotools, i guess we should aim for 20x
2022-08-02 18:21:37 <skarnet> I mean
2022-08-02 18:21:37 <psykose> and if meson keeps being as good as it is other implementations would become more viable too
2022-08-02 18:21:54 <psykose> ninja was so good samurai showed up and it's the smallest thing imaginable for the same usage
2022-08-02 18:22:03 <psykose> and there are like 3 other ports of meson by now
2022-08-02 18:22:13 <psykose> muon, some other c++ thing, something else i forgot
2022-08-02 18:22:14 <psykose> etc
2022-08-02 18:22:22 <skarnet> are people really suggesting replacing busybox ash, because it’s part of a generic do-it-all tool, with sh from chimerautils, which is ANOTHER generic do-it-all tool?
2022-08-02 18:22:27 <skarnet> AM I THE ONLY ONE AROUND HERE
2022-08-02 18:22:33 <skarnet> <cocks gun>
2022-08-02 18:22:45 <psykose> yes, due to ash, well
2022-08-02 18:22:58 <skarnet> due to ash WHAT
2022-08-02 18:23:02 <psykose> there was a lot of man hours spent on broken builders due to an ash parsing bug that segfaults
2022-08-02 18:23:08 <psykose> and then a lot more on at least 20 busybox cves
2022-08-02 18:23:09 <skarnet> do you really think we’re lacking shells on Linux?
2022-08-02 18:23:09 <psykose> and so it goes
2022-08-02 18:23:14 <psykose> no, not really
2022-08-02 18:23:47 <psykose> well, in all honesty, there isn't much choice
2022-08-02 18:24:05 <psykose> there's a bunch of ports from shells outside linux, ash, zsh, and bash
2022-08-02 18:24:08 <skarnet> so why do people have to spelunk and find the most unbootstrappable, un-standalone project possible to replace our shit
2022-08-02 18:24:10 <psykose> and then a bunch of new stuff that is 5MB
2022-08-02 18:24:24 <psykose> what makes it unbootstrappable
2022-08-02 18:24:30 <skarnet> has Alpine subscribed to the competition for most edgy distro?
2022-08-02 18:24:38 <psykose> this took less effort than *stares at busybox build system* all of that
2022-08-02 18:24:46 <skarnet> that’s a low bar
2022-08-02 18:24:50 <psykose> it is, and it was beaten
2022-08-02 18:24:52 <psykose> so..
2022-08-02 18:24:58 <psykose> it's.. better, if anything
2022-08-02 18:25:03 <skarnet> I suspect even freaking bash would be easier to build
2022-08-02 18:25:06 <panekj> get rid of whole bb
2022-08-02 18:25:09 <ikke> skarnet: do you have a good replacement suggestion?
2022-08-02 18:25:16 <psykose> *checks*
2022-08-02 18:25:19 <skarnet> I do! zsh. Which is bash, but simpler.
2022-08-02 18:25:26 <psykose> bash seems to be about even, shrug
2022-08-02 18:25:37 <psykose> the only reason this took any effort at all is because it was not ported to be standalone
2022-08-02 18:25:43 <skarnet> even *dash* would do the job, and I hate dash
2022-08-02 18:25:51 <psykose> and it would be made standalone in 1 business day the moment anyone cares
2022-08-02 18:25:53 <psykose> even i can implement that
2022-08-02 18:25:55 <skarnet> again, we. are. not. lacking. choice.
2022-08-02 18:26:04 <psykose> i am not sure what you mean
2022-08-02 18:26:21 <ikke> many aports would fail to build on dash
2022-08-02 18:26:30 <ikke> I mean, the APKBUILDs
2022-08-02 18:26:38 <panekj> dash would be a regression?
2022-08-02 18:26:38 <psykose> do you mean choice for /bin/sh? yes, there's a few options
2022-08-02 18:26:52 <skarnet> yes, I do mean choice for /bin/sh, wasn’t it the question?
2022-08-02 18:26:58 <psykose> zsh is terrible, and has the worst build/test system i've ever seen, so that's out
2022-08-02 18:27:06 <psykose> dash fails because we actually use non-posix features
2022-08-02 18:27:12 <skarnet> zsh is way better than bash
2022-08-02 18:27:13 <psykose> then there's a bunch more similar ports of shit like the openbsd shell
2022-08-02 18:27:23 <psykose> (no better than this)
2022-08-02 18:27:25 <psykose> and.. that's it
2022-08-02 18:27:29 <psykose> okay that's all the choices
2022-08-02 18:28:20 <skarnet> pdksh isn’t a thing anymore?
2022-08-02 18:28:55 <skarnet> istg if I have to write a shell
2022-08-02 18:29:11 <psykose> so, we can evaluate this one. it's 2x the size of the shell itself, +- the same featureset, and *maybe* more usable with a better codebase (which is the real gain)
2022-08-02 18:29:16 <psykose> pkdsh, as in, https://github.com/Orc/pdksh ?
2022-08-02 18:29:25 <psykose> no, this is some random shit
2022-08-02 18:29:29 <psykose> where is the actual upstream, hmm
2022-08-02 18:29:32 <panekj> https://github.com/multishell/pdksh ?
2022-08-02 18:29:54 <psykose> nope
2022-08-02 18:30:30 <skarnet> looks legit at first sight, why nope?
2022-08-02 18:30:42 <psykose> it's a git repo with 30 commits
2022-08-02 18:31:08 <skarnet> that’s a good sign
2022-08-02 18:31:09 <panekj> probably because pdksh didn't used git or any vcs at all
2022-08-02 18:31:12 <psykose> (i assume it's a shell with an actual upstream ..?)
2022-08-02 18:31:17 <skarnet> means they didn’t deviate from the old source too much
2022-08-02 18:31:25 <psykose> no vcs is fine, but there has to be somewhere it comes from
2022-08-02 18:31:28 <panekj> I assume there is no upstream
2022-08-02 18:31:46 <skarnet> it’s a public domain thing, I suspect you can find old tarballs all over the net
2022-08-02 18:31:46 <psykose> so, literally unmaintained. damn, really getting to choose from the 'worse than ash' pile here
2022-08-02 18:31:49 <ptrc> apparently there's a few ftp mirrors
2022-08-02 18:32:06 <panekj> psykose: it's public domain, what do you expect
2022-08-02 18:32:09 <skarnet> well look at the history of your beloved ash
2022-08-02 18:32:19 <panekj> any reasonable alternative would have some kind of licence
2022-08-02 18:32:21 <ptrc> or at least that's what freebsd uses in their packaging system, because the website they link ( http://www.cs.mun.ca/~michael/pdksh/ ) is dead
2022-08-02 18:33:04 <skarnet> using unmaintained shit hasn’t stopped Alpine before
2022-08-02 18:33:07 <skarnet> but I get you
2022-08-02 18:33:31 <psykose> so we should add more unmaintained shit, or
2022-08-02 18:33:39 <psykose> i thought you preferred to fix and improve things
2022-08-02 18:33:39 <skarnet> however I really can’t believe that we can’t find anything better than freaking chimerautils
2022-08-02 18:33:41 <panekj> well, it would be quite a decision going from broken supported thing to broken unsupported thing
2022-08-02 18:33:56 <psykose> chimerautils for 5 seconds until it gets put in its own box
2022-08-02 18:34:01 <skarnet> aiui it was about giving users a *choice*
2022-08-02 18:34:14 <skarnet> personally if those are the alternatives we get I’m staying with ash
2022-08-02 18:34:19 <panekj> you can still package whatever shell you want (:
2022-08-02 18:34:28 <psykose> there are lots of shells packaged and you can use any of them
2022-08-02 18:34:33 <psykose> seems like choice to me..?
2022-08-02 18:34:34 <skarnet> or switching to zsh but apparently it’s not a contender
2022-08-02 18:34:49 <skarnet> okay so there are 2 questions
2022-08-02 18:34:51 <Newbyte> all right how about pwsh
2022-08-02 18:34:58 <panekj> Newbyte: yes
2022-08-02 18:34:59 <psykose> it's even packed
2022-08-02 18:35:02 <ikke> zsh-5.9-r0 installed size:
2022-08-02 18:35:03 <psykose> you can use pwsh if you want
2022-08-02 18:35:06 <ikke> 6620 KiB
2022-08-02 18:35:07 <skarnet> 1. what shell do we want as default /bin/sh
2022-08-02 18:35:16 <skarnet> 2. what shell do we want to interpret APKBUILDs
2022-08-02 18:35:22 <skarnet> are these the same question?
2022-08-02 18:35:25 <psykose> yes
2022-08-02 18:35:27 <panekj> yes
2022-08-02 18:35:32 <ikke> yes
2022-08-02 18:35:37 <skarnet> maybe that’s a problem
2022-08-02 18:35:40 <psykose> (i mean, yes, you have a point, those don't have to be the same)
2022-08-02 18:35:59 <panekj> the problem is we use extensions 
2022-08-02 18:36:02 <skarnet> APKBUILDs legit use advanced constructs
2022-08-02 18:36:03 <psykose> but it's also a lot easier if /bin/sh is not some random (checks notes) port of pdksh, while apkbuilds are bash or something
2022-08-02 18:36:07 <skarnet> because development is complex
2022-08-02 18:36:24 <psykose> sure, and ash supports them
2022-08-02 18:36:29 <ikke> the shell that interprets APKBUILDs needs to be early in the bootstrap
2022-08-02 18:36:31 <psykose> so does freebsd sh (i guess)
2022-08-02 18:36:35 <psykose> so does zsh
2022-08-02 18:36:36 <skarnet> so I would suggest pinning a shell spec for APKBUILDs first
2022-08-02 18:36:41 <skarnet> and stick to it
2022-08-02 18:36:45 <psykose> we do
2022-08-02 18:36:47 <skarnet> have abuild use /bin/ash or something
2022-08-02 18:36:48 <psykose> the spec is /bin/sh
2022-08-02 18:36:49 <panekj> that's what we trying to do
2022-08-02 18:36:55 <skarnet> the spec obviously is not /bin/sh
2022-08-02 18:36:58 <psykose> it is
2022-08-02 18:37:01 <psykose> and /bin/sh is ash
2022-08-02 18:37:01 <skarnet> because /bin/sh means posix
2022-08-02 18:37:02 <psykose> so it's ash
2022-08-02 18:37:06 <psykose> :)
2022-08-02 18:37:12 <skarnet> well if you’re using ash features then the spec is ash
2022-08-02 18:37:20 <skarnet> and abuild should use #!/bin/ash
2022-08-02 18:37:35 <skarnet> and we can decide how to implement ash other than with bb ash
2022-08-02 18:37:52 <skarnet> that’s a different question from what /bin/sh needs to be
2022-08-02 18:38:21 <ikke> In a minimal chroot, i'd like to at least of a shell that's usable
2022-08-02 18:38:24 <psykose> yes, but also not
2022-08-02 18:38:35 <psykose> the features 'used in apkbuilds' are also things people kinda want to even use
2022-08-02 18:38:38 <ikke> I don't consider dash usable as an interactive shell
2022-08-02 18:38:54 <skarnet> again those are two different things
2022-08-02 18:38:55 <panekj> not only you 
2022-08-02 18:39:02 <skarnet> you want /bin/sh to have nice interactive features
2022-08-02 18:39:05 <skarnet> that’s legit
2022-08-02 18:39:12 <psykose> and that's what we're going for
2022-08-02 18:39:25 <skarnet> APKBUILD sh does not need nice interactive features, it just needs local and a few bash-like extensions
2022-08-02 18:39:57 <psykose> are you saying there is a shell that can be /bin/sh that has some 'nice interactive features' but does not have any bash extensions that would be using in an APKBUILD? because it does not really exist
2022-08-02 18:40:13 <psykose> things that fall into the former also implement most of the latter
2022-08-02 18:40:16 <skarnet> I really believe that before you start tampering with shells, it’s important to scan your main shell apps, like abuild, and pin the sets of features
2022-08-02 18:40:46 <skarnet> what I do know and am saying is that changing shells without pinning specs WILL break things
2022-08-02 18:41:03 <psykose> i agree
2022-08-02 18:41:12 <skarnet> and if you want things not to break, you should start by making abuild depend on ash
2022-08-02 18:41:17 <skarnet> can be revisited later
2022-08-02 18:41:49 <skarnet> then we can have a conversation on what /bin/sh should be
2022-08-02 18:41:55 <Newbyte> why do we need to replace ash again?
2022-08-02 18:42:03 <panekj> broken crap
2022-08-02 18:42:04 <skarnet> because busybox is bad
2022-08-02 18:42:05 <ikke> Newbyte: buggy hard to read code
2022-08-02 18:42:09 <psykose> go read the tsc issue
2022-08-02 18:42:19 <psykose> before someone asks for the 18th time
2022-08-02 18:42:24 <ikke> https://gitlab.alpinelinux.org/alpine/tsc/-/issues/39
2022-08-02 18:42:40 <psykose> and it is merely an exploration
2022-08-02 18:42:47 <psykose> equally likely nothing gets changed
2022-08-02 18:42:52 <psykose> anyways
2022-08-02 18:42:59 <skarnet> your bet is 50%?
2022-08-02 18:43:09 <skarnet> mine is at a solid 80% chance nothing gets changed
2022-08-02 18:43:16 <psykose> skarnet: that doesn't really make sense because ash isn't standalone, and the moment /bin/sh is replaced it would be gone
2022-08-02 18:43:18 <skarnet> will reevaluate in the following months
2022-08-02 18:43:58 <skarnet> psykose: now you understand why I want to separate functionality from implementation in busybox
2022-08-02 18:44:00 <psykose> so "depend on ash" is just a meaningless gesture in the meantime
2022-08-02 18:44:15 <psykose> yes, things need to be evaluated if they still work, this would be part of the checking-other-shells process
2022-08-02 18:44:24 <skarnet> “depend on ash” is a meaningful gesture, that, yes, would pin busybox when you want to run abuild
2022-08-02 18:44:33 <skarnet> it’s the price to pay for not breaking shit
2022-08-02 18:44:39 <ikke> the goal is to get rid of ash
2022-08-02 18:44:45 <psykose> what do you mean by 'pin'?
2022-08-02 18:45:14 <ikke> ash not handling APKBUILDs very well (what psykose referred to earlier), is one of the reasons why we are even looking at something else
2022-08-02 18:45:20 <skarnet> pin: enforce #!/bin/ash shebangs, add an explicit dependency from abuild to busybox, or better, to a virtual package named ash, implemented by busybox
2022-08-02 18:45:37 <psykose> that doesn't do anything useful in the meantime, and is just extra work to remove them after
2022-08-02 18:45:43 <psykose> that's just an explicit way of tracking 'what needs changing'
2022-08-02 18:45:49 <psykose> slash checking
2022-08-02 18:45:56 <skarnet> ...
2022-08-02 18:46:01 <skarnet> it makes shit work
2022-08-02 18:46:10 <skarnet> it prevents regressions
2022-08-02 18:46:18 <psykose> huh? no it wouldn't
2022-08-02 18:46:22 <psykose> i feel like you're misunderstanding
2022-08-02 18:46:33 <skarnet> what am I not understanding
2022-08-02 18:46:34 <psykose> currently, there is a /bin/sh, and things rely on it being ash
2022-08-02 18:46:39 <psykose> at some point
2022-08-02 18:46:42 <skarnet> yes and that’s bad
2022-08-02 18:46:43 <psykose> (assuming this goes through)
2022-08-02 18:46:51 <psykose> there will be one singular `apk upgrade`
2022-08-02 18:47:01 <psykose> and busybox will not have ash anymore, and some other shell will be /bin/sh
2022-08-02 18:47:02 <psykose> one step
2022-08-02 18:47:03 <psykose> that's it
2022-08-02 18:47:09 <skarnet> and shit will break
2022-08-02 18:47:18 <psykose> yes, and if you put #!/bin/ash on things, it would also break
2022-08-02 18:47:21 <psykose> because the thing would just not exist
2022-08-02 18:47:24 <psykose> what is the difference
2022-08-02 18:47:33 <skarnet> follow your thought to its logical conclusion
2022-08-02 18:47:41 <ikke> psykose: I suppose he wants to break it 2 steps
2022-08-02 18:47:42 <skarnet> “the thing would just not exist” is a bad idea
2022-08-02 18:47:50 <skarnet> ikke gets it
2022-08-02 18:47:55 <ikke> replace /bin/sh separate from removing ash
2022-08-02 18:48:03 <skarnet> the way to perform a smooth transition is to have a period where you have both
2022-08-02 18:48:28 <psykose> sure
2022-08-02 18:48:39 <Ariadne> the end result is that busybox no longer exists in alpine at all, but this will take several versions
2022-08-02 18:48:46 <skarnet> 1. make sure everything that depends on ash is marked as depending on ash
2022-08-02 18:48:51 <skarnet> 2. introduce new /bin/sh
2022-08-02 18:49:10 <skarnet> 3. over time, incrementally, switch stuff to /bin/sh, see what breaks, fix it
2022-08-02 18:49:16 <skarnet> 4. when everything’s fixed, remove ash
2022-08-02 18:49:34 <psykose> ah, okay
2022-08-02 18:49:43 <psykose> i did think of that before, but discarded it for some reason
2022-08-02 18:49:57 <psykose> sure, that works
2022-08-02 18:50:28 <Ariadne> i'm not convinced that anything actually depends on ash
2022-08-02 18:50:43 <psykose> anything that is not strictly posix sh does, because ash might provide the feature
2022-08-02 18:50:48 <psykose> ah, you mean explicitly already
2022-08-02 18:50:48 <Ariadne> maybe we should couple mdev with this :))))
2022-08-02 18:50:50 <psykose> yeah, nothing does
2022-08-02 18:51:13 <ikke> not explicitly, but implicitly lots of things depends on posix sh + extensions
2022-08-02 18:51:22 <psykose> i guess step one is provides="ash" on main/busybox
2022-08-02 18:51:28 <kunkku> scripts could contain "ashisms"?
2022-08-02 18:51:39 <psykose> then once we get another sh we disable the busybox sh link
2022-08-02 18:51:41 <Ariadne> okay
2022-08-02 18:51:43 <Ariadne> guys
2022-08-02 18:51:44 <skarnet> if scripts contain ashisms then they should explicitly depend on ash
2022-08-02 18:51:44 <ikke> kunkku: just use shellcheck on sh mode
2022-08-02 18:51:52 <Ariadne> both are almquist shells
2022-08-02 18:52:03 <Ariadne> what "ashism" could busybox ash have that freebsd ash does not
2022-08-02 18:52:16 <psykose> the issue is it's hard to say for sure
2022-08-02 18:52:22 <psykose> it would have to be a feature-by-feature comparison
2022-08-02 18:52:25 <Ariadne> both are ash
2022-08-02 18:52:40 <Ariadne> i suggested freebsd ash because it is the closest to what we already have
2022-08-02 18:52:41 <panekj> is there any truly truly for sure 100% POSIX shell out there anyway?
2022-08-02 18:52:48 <psykose> okay, i will go build chimera ash and replace it, give me 10 minutes to push it
2022-08-02 18:52:49 <ikke> panekj: dash?
2022-08-02 18:52:51 <psykose> hope nothing breaks
2022-08-02 18:52:53 <skarnet> what feature could <mutated strain> have that <separately mutated strain> does not? :think:
2022-08-02 18:53:02 <skarnet> what could possibly go wrong
2022-08-02 18:53:05 <Ariadne> skarnet: denys syncs from freebsd ash
2022-08-02 18:53:14 <skarnet> does he now?
2022-08-02 18:53:16 <Newbyte> having a 100% POSIX shell for APKBUILDs would suck
2022-08-02 18:53:18 <Ariadne> yes
2022-08-02 18:53:22 <ikke> Newbyte: We agree
2022-08-02 18:53:28 <Newbyte> very good
2022-08-02 18:53:32 <panekj> ikke: afaik, dash has some kind of extension
2022-08-02 18:53:33 <ptrc> Ariadne: wasn't bb ash forked from debian's dash?
2022-08-02 18:53:38 <Ariadne> no
2022-08-02 18:53:47 <Ariadne> bb ash has nothing to do with dash
2022-08-02 18:53:49 <ptrc> or rather, "derived", as the bb source says
2022-08-02 18:54:10 <psykose> Ariadne: i know that they are more or less the same, but it has to be investigated with an actual list of features / tested, to make sure, so random thing somewhere doesn't explode
2022-08-02 18:54:17 <skarnet> exactly
2022-08-02 18:54:23 <ptrc> https://git.busybox.net/busybox/tree/shell/ash.c#n29
2022-08-02 18:54:30 <skarnet> changing shells definitely isn’t something you can yolo
2022-08-02 18:54:45 <psykose> given how close they are, perhaps a literal /bin/ash isn't needed, and a side-by-side would be fine, but even then, it can't just be handwaved away
2022-08-02 18:55:23 <Ariadne> ptrc: they are all derived from each other, who cares
2022-08-02 18:56:09 <Ariadne> anyway you are all doing it wrong
2022-08-02 18:56:41 <panekj> how big is zsh?
2022-08-02 18:56:44 <ptrc> i just thought that if dash has added some feature in the meantime, migrating to freebsd ash might break something
2022-08-02 18:56:44 <psykose> huge
2022-08-02 18:56:45 <ikke> 6MB
2022-08-02 18:56:48 <ptrc> panekj: >6MB
2022-08-02 18:56:51 <psykose> that's the full one
2022-08-02 18:56:54 <psykose> i did test the no features one
2022-08-02 18:56:55 <Ariadne> jesus fucking christ
2022-08-02 18:56:57 <psykose> (still huge)
2022-08-02 18:57:31 <psykose> Ariadne: what is the correct way of doing 'thing' then
2022-08-02 18:58:11 <Ariadne> the way we did this with pkg-config was introducing pkgconf with an optional package that replaced /usr/bin/pkg-config
2022-08-02 18:58:15 <Ariadne> do that with chimera sh
2022-08-02 18:58:22 <Ariadne> then rebuild the entire system
2022-08-02 18:58:24 <panekj> so some random public domain source code with no upstream or 6MB shell? :P
2022-08-02 18:58:48 <psykose> sure, i'll do it as soon as q66 fixes the build
2022-08-02 18:58:49 <ikke> The suggestion was to use the ported freebsd sh
2022-08-02 18:58:57 <skarnet> panekj: or some mutated thing from a mutated thing from FreeBSD
2022-08-02 18:58:58 <psykose> though i did just build it anyway
2022-08-02 18:58:59 <Ariadne> okay
2022-08-02 18:59:01 <Ariadne> i'm done with this
2022-08-02 18:59:02 <ikke> at least, evaluate
2022-08-02 18:59:09 <Ariadne> forget it
2022-08-02 18:59:13 <psykose> Ariadne: or would you like me to throw it in already
2022-08-02 18:59:15 <Ariadne> we will just keep busybox forever
2022-08-02 18:59:17 <skarnet> I really don’t want to write a shell
2022-08-02 18:59:21 <Ariadne> no
2022-08-02 18:59:23 <Ariadne> don't bother
2022-08-02 18:59:27 <Ariadne> we are keeping busybox forever
2022-08-02 18:59:30 <Ariadne> we are keeping mdev forever
2022-08-02 18:59:36 <psykose> my god i just want to do things
2022-08-02 18:59:36 <Ariadne> because god forbid we cannot take risks
2022-08-02 18:59:36 <ikke> Ariadne: please..
2022-08-02 18:59:44 <Ariadne> even though the TSC has already decided we can take risks
2022-08-02 18:59:54 <psykose> do you want me to do anything or not?
2022-08-02 19:00:14 <skarnet> if you somehow interpreted what I said as “we should not take risks” then go take two days of sleep
2022-08-02 19:00:41 <Ariadne> no i am talking about whatever this zsh conversation is
2022-08-02 19:00:48 <psykose> it's just random fluff
2022-08-02 19:00:53 <panekj> true
2022-08-02 19:00:54 <Ariadne> well, it is annoying
2022-08-02 19:01:45 <psykose> ikke: think you have some diskspace to free
2022-08-02 19:01:49 <Ariadne> so.  if somebody wants to seriously take this, then
2022-08-02 19:01:56 <Ariadne> 1. package chimera's freebsd sh port
2022-08-02 19:01:58 <ikke> The only thing we have committed to so far is that we are going to _evaluate_ another shell
2022-08-02 19:02:06 <Ariadne> 2. add "chimera-binsh" package which adds `/bin/sh`
2022-08-02 19:02:18 <Ariadne> 3. rebuild world with "chimera-binsh" in world, and see what breaks
2022-08-02 19:03:12 <Ariadne> this is how we did it with pkg-config --> pkgconf until we proved that pkgconf was ready
2022-08-02 19:03:29 <psykose> what was the way to allow something to replace /bin/sh again (if it even needed anything to not conflict)
2022-08-02 19:03:37 <psykose> aside from that, sure, give me 10 minutes
2022-08-02 19:03:40 <Ariadne> you just do it
2022-08-02 19:03:54 <psykose> ok
2022-08-02 19:04:09 <Ariadne> the way busybox works is there's a hook that installs all the missing symlinks
2022-08-02 19:04:15 <Ariadne> so if you have one already there
2022-08-02 19:04:21 <Ariadne> it won't overwrite it
2022-08-02 19:04:28 <Ariadne> though that brings up a point
2022-08-02 19:04:38 <Ariadne> alpine triggers need /bin/sh to run
2022-08-02 19:04:51 <Ariadne> so if you revert the change
2022-08-02 19:04:58 <Ariadne> you'll need to manually do /bin/busybox --install -s
2022-08-02 19:05:23 <ikke> psykose: 105G available now :=
2022-08-02 19:05:26 <ikke> :+
2022-08-02 19:05:29 <psykose> ^^
2022-08-02 19:05:38 <Ariadne> there is one feature missing btw
2022-08-02 19:05:44 <Ariadne> command_not_found_handler()
2022-08-02 19:05:46 <ikke> rm -fr .cargo .cache/yarn go .cache/go-build
2022-08-02 19:05:56 <Ariadne> but the patch that added it to busybox ash is likely easy to port
2022-08-02 19:06:35 <ikke> I say psykose packages it, and then we can do some initial tests
2022-08-02 19:06:45 <Ariadne> forbidden
2022-08-02 19:07:05 <Ariadne> skarnet must write a shell
2022-08-02 19:07:10 <Ariadne> s6sh
2022-08-02 19:07:11 <psykose> what is the licence for this shit
2022-08-02 19:07:12 <Ariadne> lets goooo
2022-08-02 19:07:35 <Ariadne> BSD-3-Clause
2022-08-02 19:08:14 <Ariadne> btw panekj this does raise the possibility of allowing zsh-binsh too
2022-08-02 19:08:29 <ikke> Ariadne: ftr, it was skarnet who brought up zsh
2022-08-02 19:08:38 <Ariadne> omg why
2022-08-02 19:08:38 <skarnet> yes and I regret all my life choices
2022-08-02 19:09:06 <Ariadne> ikke: can we convenene emergency TSC meeting to cancel the previous TSC decision giving skarnet what he wanted :D
2022-08-02 19:09:12 <skarnet> why? because when people start talking about ash extensions, in general they bring up bash very quickly
2022-08-02 19:09:25 <skarnet> and zsh is my stfu answer to any mention of bash
2022-08-02 19:09:38 <Ariadne> don't worry, only kidding
2022-08-02 19:09:39 <skarnet> because it does everything bash does, but better and smaller
2022-08-02 19:09:46 <Ariadne> as eudev maintainer i want this shit too
2022-08-02 19:09:48 <Newbyte> smaller?
2022-08-02 19:09:48 <skarnet> this time, for once nobody brought bash
2022-08-02 19:09:54 <Ariadne> smaller...?
2022-08-02 19:10:00 <Ariadne> bash ain't no 6MB
2022-08-02 19:10:02 <skarnet> so my mention of zsh was counterproductive
2022-08-02 19:10:04 <skarnet> yes, smaller
2022-08-02 19:10:07 <Newbyte> bash is smaller in terms of binary size in Alpine
2022-08-02 19:10:11 <Newbyte> way smaller
2022-08-02 19:10:17 <Ariadne> yeah its only 1MB!
2022-08-02 19:10:18 <skarnet> my zsh ain’t 6 MB
2022-08-02 19:10:26 <Newbyte> Alpine's is
2022-08-02 19:10:28 <skarnet> and its RAM footprint is smaller
2022-08-02 19:10:36 <ikke> You removed all usefull features? :P
2022-08-02 19:10:40 <skarnet> well maybe that’s an Alpine problem
2022-08-02 19:10:47 <skarnet> no, I removed all useless features :P
2022-08-02 19:11:10 <Newbyte> how big is your zsh?
2022-08-02 19:11:15 <skarnet> 1.5MB
2022-08-02 19:11:20 <Newbyte> that's still bigger than bash
2022-08-02 19:11:26 <Ariadne> .5MB bigger
2022-08-02 19:11:37 <skarnet> there’s something there
2022-08-02 19:11:42 <skarnet> because my bash is 2 MB
2022-08-02 19:12:04 <skarnet> so you guys did something to compress bash to the maxx and inflate zsh :P
2022-08-02 19:12:20 <Ariadne> maybe inflation is just the zsh maintainer's thing
2022-08-02 19:12:23 <Ariadne> don't judge
2022-08-02 19:12:50 <Ariadne> 
ACTION looks up who actually maintains zsh in alpine

2022-08-02 19:12:58 <ikke> jirutka perhaps?
2022-08-02 19:13:02 <Ariadne> haha probably should have done that first
2022-08-02 19:13:07 <Ariadne> no, its ncopa apparently
2022-08-02 19:13:10 <ikke> ah
2022-08-02 19:13:23 <skarnet> --enable-everything :P
2022-08-02 19:13:24 <Ariadne> anyway!
2022-08-02 19:14:00 <Ariadne> sometimes i miss my old job :(
2022-08-02 19:14:10 <Newbyte> what was your old job?
2022-08-02 19:14:13 <Ariadne> i used to be able to just, like, rebuild all of alpine on a supercomputer
2022-08-02 19:14:18 <Ariadne> in like 2 hours
2022-08-02 19:15:20 <ikke> Ariadne: effectively it's jirukta 2 years 
2022-08-02 19:15:38 <Ariadne> Newbyte: i worked for the US department of energy national lab system :)
2022-08-02 19:15:49 <psykose> anyway it's packaged
2022-08-02 19:15:51 <psykose> now for some rebuilds
2022-08-02 19:15:53 <Ariadne> is it
2022-08-02 19:15:55 <Ariadne> IS IT
2022-08-02 19:15:56 <psykose> yes
2022-08-02 19:16:06 <psykose> https://git.alpinelinux.org/aports/commit/?id=7c8d0d96f01d
2022-08-02 19:16:10 <Ariadne> >libxo
2022-08-02 19:16:11 <Ariadne> gross
2022-08-02 19:16:19 <psykose> it's not for sh
2022-08-02 19:16:19 <psykose> haha
2022-08-02 19:16:21 <psykose> nothing is
2022-08-02 19:16:35 <psykose> sh is only libedit
2022-08-02 19:17:01 <Newbyte> what's wrong with libxo?
2022-08-02 19:17:16 <Ariadne> i mean, nothing, if you think it is normal for /bin/ls to output XML
2022-08-02 19:17:41 <psykose> it can be built text-only
2022-08-02 19:17:43 <ikke> psykose: so you cannot install this side-by-side?
2022-08-02 19:17:46 <psykose> at which point i wonder why it exists
2022-08-02 19:17:51 <skarnet> I was hesitating between “executive officer” and “GenZ kiss” but of course it has to be XML
2022-08-02 19:17:59 <psykose> ah, right, it conflicts
2022-08-02 19:18:09 <Ariadne> yeah it needs
2022-08-02 19:18:13 <psykose> ole replaces
2022-08-02 19:18:51 <Ariadne> amove + binsh() { provides="/bin/sh"; ln -sf /bin/sh /bin/cash }
2022-08-02 19:19:12 <Ariadne> ...look, if we went with freebsd ash it would be fash, and i don't think that was a good idea, so i decided chimera ash was better
2022-08-02 19:19:37 <Ariadne> the binsh virtual does *not* need replaces
2022-08-02 19:19:42 <Ariadne> apk has no awareness of the busybox symlinks
2022-08-02 19:19:42 <panekj> skarnet: "xo" has existed before gen z could even tal
2022-08-02 19:19:46 <panekj> s/tal/talk
2022-08-02 19:19:46 <alpine-meetbot> panekj meant to say: skarnet: "xo" has existed before gen z could even talk
2022-08-02 19:20:13 <Newbyte> let's go with fash to weed out all corporate use from Alpine
2022-08-02 19:20:20 <Ariadne> psykose: anyway if you need a powerful machine to rebuild world on, my old x86_64 server is still racked up in my colo
2022-08-02 19:20:30 <Ariadne> i can like
2022-08-02 19:20:35 <Ariadne> reinstall it with a fresh alpine if you want
2022-08-02 19:20:41 <skarnet> panekj: yeah but before GenZ the ones who used it were GenX Facebook dwellers and those are dead to me
2022-08-02 19:21:22 <Ariadne> Newbyte: same energy as licensing my new projects AGPL just to own the corporates
2022-08-02 19:21:39 <panekj> > let's go with fash to weed out all corporate use from Alpine
2022-08-02 19:21:40 <psykose> Ariadne: i have the nld5-dev1 container which should work
2022-08-02 19:21:40 <ikke> AGPL All The Things
2022-08-02 19:21:42 <panekj> time to drop s390x
2022-08-02 19:21:57 <psykose> also the provides= doesn't work still
2022-08-02 19:21:57 <psykose> hm
2022-08-02 19:22:11 <Ariadne> oh, right
2022-08-02 19:22:17 <psykose> ah
2022-08-02 19:22:20 <ikke> provides="cmd:/bin/sh"?
2022-08-02 19:22:23 <Ariadne> busybox needs
2022-08-02 19:22:27 <Ariadne> busybox-binsh package
2022-08-02 19:22:29 <ikke> although, that should be detected
2022-08-02 19:22:30 <Ariadne> for that to work
2022-08-02 19:22:37 <ptrc> ikke: '/bin/sh' is a special virtual package iirc
2022-08-02 19:22:52 <ikke> rgiht
2022-08-02 19:22:53 <ikke> right
2022-08-02 19:23:24 <ikke> https://gitlab.alpinelinux.org/alpine/abuild/-/blob/master/abuild.in#L1111
2022-08-02 19:24:18 <ikke> https://gitlab.alpinelinux.org/alpine/aports/-/blob/master/main/busybox/APKBUILD#L20
2022-08-02 19:24:21 <Ariadne> that came from last time we tried to make /bin/sh selectable :P
2022-08-02 19:24:29 <psykose> the real error is `trying to overwrite bin/sh` so i forget where to put the overrides
2022-08-02 19:24:48 <Ariadne> wait does busybox package actually have a symlink
2022-08-02 19:25:03 <psykose> looks like it, at least /bin/sh is a sym
2022-08-02 19:25:14 <ikke> https://pkgs.alpinelinux.org/contents?file=sh&path=&name=busybox&branch=edge
2022-08-02 19:25:39 <Ariadne> huh
2022-08-02 19:25:46 <Ariadne> yeah
2022-08-02 19:25:57 <Ariadne> busybox package does register bin/sh symlink explicitly
2022-08-02 19:26:00 <Ariadne> ok
2022-08-02 19:26:05 <skarnet> there is a *lot* to untangle with busybox, that’s exactly why I wrote my proposal
2022-08-02 19:26:12 <skarnet> bb is hardcoded everywhere
2022-08-02 19:26:19 <ikke> https://gitlab.alpinelinux.org/alpine/aports/-/blob/master/main/busybox/APKBUILD#L204
2022-08-02 19:26:27 <Ariadne> i guess what we can do is
2022-08-02 19:26:35 <Ariadne> add busybox-binsh package
2022-08-02 19:26:40 <Ariadne> thats what adelie did
2022-08-02 19:27:03 <psykose> so move that symlink and the provides stuff to a subpackage i guess
2022-08-02 19:27:04 <Ariadne> move the /bin/sh provide to it
2022-08-02 19:27:06 <Ariadne> yep
2022-08-02 19:27:06 <psykose> sure
2022-08-02 19:27:24 <ikke> Ariadne: the comment says it's done to prevent the issue you mentioned earler
2022-08-02 19:27:40 <Ariadne> yeah
2022-08-02 19:27:44 <Ariadne> anyway we can move it
2022-08-02 19:27:49 <Ariadne> apk will do the right thing
2022-08-02 19:28:11 <Ariadne> set provide_priority=1 on it
2022-08-02 19:28:27 <Ariadne> then for freebsd sh, put that at like 10
2022-08-02 19:28:51 <Ariadne> skarnet: tbh a busybox-symlinks package would be nice for `apko`
2022-08-02 19:28:51 <ikke> highest wins
2022-08-02 19:28:56 <Ariadne> oh, right
2022-08-02 19:28:58 <panekj> make it empty
2022-08-02 19:29:01 <panekj> like linux-firmware
2022-08-02 19:29:40 <skarnet> ikke: really? that’s not what the alpine wiki says iirc
2022-08-02 19:29:53 <psykose> !37197
2022-08-02 19:29:56 <ikke> skarnet: bb sets provider_priority to 100
2022-08-02 19:30:07 <ikke> skarnet: afaik, it's compared like versions
2022-08-02 19:30:20 <skarnet> hm
2022-08-02 19:30:26 <panekj> > A numeric value which is used by apk-tools to break ties when choosing a virtual package to satisfy a dependency. Higher values have higher priority.
2022-08-02 19:30:50 <Ariadne> yeah its higher sorry
2022-08-02 19:30:53 <panekj> alpine wiki
2022-08-02 19:31:18 <Ariadne> psykose: did you test apk upgrade -Ua :)
2022-08-02 19:31:21 <skarnet> ANYWAY, I am currently witnessing you people yoloing a busybox shell change *live*, so next time I am submitting a MR that modifies low-level shit I don’t want to hear ANYTHING
2022-08-02 19:31:25 <psykose> doing that first
2022-08-02 19:31:33 <ikke> skarnet: we don't _change_ the shell
2022-08-02 19:32:01 <Ariadne> skarnet: we are yoloing the very early stage of an approved SCP
2022-08-02 19:32:33 <Ariadne> doing whatever needs to be done to collect data for the next stage was approved
2022-08-02 19:32:52 <psykose> works, says installs binsh
2022-08-02 19:33:06 <psykose> still cant install the other though
2022-08-02 19:33:13 <psykose> conflicts: busybox-1.35.0-r22[cmd:sh=0_git20220731-r0]
2022-08-02 19:33:17 <psykose> because of the cmd trace
2022-08-02 19:33:18 <psykose> weh
2022-08-02 19:34:13 <Ariadne> hmm, whats up with this
2022-08-02 19:34:23 <Ariadne> https://gitlab.alpinelinux.org/alpine/aports/-/jobs/790129
2022-08-02 19:34:31 <Ariadne> wait what
2022-08-02 19:34:32 <psykose> it missed a rel
2022-08-02 19:34:35 <psykose> but was renamed for openssl3
2022-08-02 19:34:38 <Ariadne> why would cmd:sh
2022-08-02 19:34:42 <Ariadne> be in busybox
2022-08-02 19:34:49 <Ariadne> if it is moved to busybox-binsh
2022-08-02 19:34:59 <psykose> hmm
2022-08-02 19:35:15 <ikke> symlink tracing?
2022-08-02 19:35:27 <Ariadne> yeah sure but /bin/sh is now in the virtual
2022-08-02 19:35:32 <Ariadne> er, in the other package
2022-08-02 19:35:35 <psykose> oh
2022-08-02 19:35:50 <psykose> no, i forgot to move it
2022-08-02 19:35:55 <psykose> sorry i'm just stupid
2022-08-02 19:37:03 <psykose> skarnet: yo we aint' yoloing anything homeslice
2022-08-02 19:37:21 <Ariadne> yolooooo
2022-08-02 19:37:32 <psykose> the french shall never forgive
2022-08-02 19:37:39 <Ariadne> yeah this was just an unanticipated thing
2022-08-02 19:37:46 <Ariadne> we are just collecting data at the moment
2022-08-02 19:38:01 <skarnet> so that’s how you call it huh
2022-08-02 19:38:08 <Ariadne> i mean
2022-08-02 19:38:15 <Ariadne> yes, because that's what we are doing
2022-08-02 19:38:25 <Ariadne> introducing a new /bin/sh provider that is not selected by default
2022-08-02 19:38:32 <Ariadne> and then explicitly selecting it
2022-08-02 19:38:35 <Ariadne> and rebuilding world
2022-08-02 19:38:41 <Ariadne> to make sure everything is good
2022-08-02 19:38:47 <Ariadne> or not good
2022-08-02 19:38:59 <Ariadne> either way, rebuild world -> present data to TSC
2022-08-02 19:39:02 <skarnet> all right, all right :)
2022-08-02 19:39:06 <psykose> this same step would need to be taken even if we did exactly what you wanted to
2022-08-02 19:39:26 <Ariadne> this step lets you have zsh as /bin/sh too
2022-08-02 19:39:27 <Ariadne> :D
2022-08-02 19:39:44 <psykose> the intermediate upgrade fails
2022-08-02 19:39:46 <psykose> ERROR: busybox-1.35.0-r22.post-upgrade: script exited with error 127
2022-08-02 19:39:47 <psykose> haha
2022-08-02 19:39:54 <psykose> because busybox loses the sym
2022-08-02 19:39:59 <psykose> and it's added after
2022-08-02 19:40:08 <psykose> sigh
2022-08-02 19:40:14 <psykose> that one is a bit hard to solve
2022-08-02 19:40:18 <Ariadne> hmm
2022-08-02 19:40:24 <psykose> (it works with fix, but)
2022-08-02 19:40:26 <psykose> it is a roadblock
2022-08-02 19:40:27 <Ariadne> maybe have symlink in both and replaces=busybox
2022-08-02 19:40:32 <Ariadne> and then turn off cmd tracing?
2022-08-02 19:40:33 <skarnet> no
2022-08-02 19:40:42 <skarnet> use #!/bin/busybox sh
2022-08-02 19:40:42 <psykose> cmd: tracing is part of tracedeps
2022-08-02 19:40:49 <psykose> i guess that's not relevant for busybox as it doesn't use most things
2022-08-02 19:40:58 <skarnet> in post-upgrade and post-uninstall or something
2022-08-02 19:41:00 <psykose> o
2022-08-02 19:41:04 <psykose> ur a genius skarnet
2022-08-02 19:41:10 <psykose> actual 400iq gigachad
2022-08-02 19:41:12 <psykose> lets go bestie
2022-08-02 19:41:14 <Ariadne> hmm
2022-08-02 19:41:17 <Ariadne> it should work
2022-08-02 19:41:26 <Ariadne> technically apk does not care what is in the shebang
2022-08-02 19:41:31 <psykose> and the moment we change actual sh, we change it again
2022-08-02 19:41:40 <Ariadne> well no
2022-08-02 19:41:43 <Ariadne> busybox's script
2022-08-02 19:41:50 <skarnet> if it’s in a busybox post-upgrade script you’re guaranteed to have busybox
2022-08-02 19:41:50 <Ariadne> should just always have #!/bin/busybox sh
2022-08-02 19:41:56 <Ariadne> ^
2022-08-02 19:41:56 <psykose> what if busybox has no ash
2022-08-02 19:41:59 <psykose> ah
2022-08-02 19:42:09 <Ariadne> we will always provide ash in the busybox package
2022-08-02 19:42:15 <psykose> ah, okay
2022-08-02 19:42:15 <psykose> sure
2022-08-02 19:42:19 <Ariadne> because mkinitfs
2022-08-02 19:42:20 <skarnet> if busybox has no ash it’s because we haven’t BURNED IT YET
2022-08-02 19:42:39 <Ariadne> though nmeum is working on replacing alpine's initramfs
2022-08-02 19:42:51 <Ariadne> with one that is just a static executable
2022-08-02 19:43:01 <skarnet> blah
2022-08-02 19:43:16 <Ariadne> written in skarnet's favorite language
2022-08-02 19:43:25 <skarnet> execline?
2022-08-02 19:43:26 <skarnet> :P
2022-08-02 19:43:33 <Ariadne> the one with the gopher
2022-08-02 19:43:44 <ikke>  yaay
2022-08-02 19:43:48 <psykose> upgrade is clean
2022-08-02 19:43:57 <Ariadne> that change has merit regardless
2022-08-02 19:44:08 <psykose> override to chimera is clean
2022-08-02 19:44:14 <Ariadne> hell yeah
2022-08-02 19:44:15 <Ariadne> ship it
2022-08-02 19:44:38 <psykose> moving back (via apk del chimera-binsh; apk add busybox-binsh), doesn't even need a reinstall
2022-08-02 19:44:57 <Ariadne> yeah
2022-08-02 19:45:15 <Ariadne> haha you can remove that comment about the symlink now :D
2022-08-02 19:45:23 <Ariadne> because it was for that specific script
2022-08-02 19:45:29 <Ariadne> hooray one technical debt paid off
2022-08-02 19:45:44 <psykose> whiiich comment
2022-08-02 19:45:52 <ikke> :)
2022-08-02 19:45:53 <Ariadne> the one about /bin/sh being needed for postinst
2022-08-02 19:45:58 <Ariadne> we odn't need it anymore
2022-08-02 19:46:08 <psykose> o, yeah
2022-08-02 19:46:12 <psykose> it's deleted in the diff
2022-08-02 19:46:13 <psykose> (i think)
2022-08-02 19:46:42 <psykose> if it's L203
2022-08-02 19:47:08 <ikke> The one I linked to before
2022-08-02 19:47:34 <ikke> yes, 203
2022-08-02 19:47:41 <psykose> ok
2022-08-02 19:48:16 <Ariadne> sent it
2022-08-02 19:48:24 <psykose> sikk
2022-08-02 19:48:35 <psykose> now to beg q66 for a build fix
2022-08-02 19:48:40 <psykose> and for me to go spelunking
2022-08-02 19:48:43 <psykose> i guess i just run buildrepo
2022-08-02 19:51:43 <psykose> (also the builders are fucked for a long time in case anyone wants to go and actually use it)
2022-08-02 19:51:51 <psykose> you'd have to build them yourself
2022-08-02 19:52:17 <psykose> haha, x86_64 is 13/289 on community let alone testing
2022-08-02 19:52:40 <psykose> at least a few people will have really broken worlds because of main upgrading first
2022-08-02 19:52:44 <psykose> if only it was possible to push at once
2022-08-02 19:54:04 <psykose> ikke: small request
2022-08-02 19:54:12 <psykose> can you kill all of edge and restart it, so openssl1.1 rebuilds
2022-08-02 19:54:27 <psykose> since it's broken otherwise, but it will take a really long time to loop around
2022-08-02 19:54:30 <ikke> ok
2022-08-02 19:55:43 <psykose> we also need to fix armv7/s390x mariadb
2022-08-02 19:55:43 <psykose> weh
2022-08-02 19:55:54 <ikke> yup
2022-08-02 19:56:01 <ikke> a test failure, right?
2022-08-02 19:56:11 <psykose> what's weird is s390x segfaults and armv7 sigbus's
2022-08-02 19:56:13 <psykose> but armhf passed
2022-08-02 19:56:15 <psykose> yea
2022-08-02 19:56:18 <psykose> same test on both
2022-08-02 19:56:23 <psykose> one of the aes subtests
2022-08-02 20:00:17 <Ariadne> 
ACTION mutters things about OpenSSL

2022-08-02 20:00:42 <Ariadne> skarnet: if you want to complain about yolo changes, openssl 3 is a better place to look :D
2022-08-02 20:01:59 <ikke> psykose: I everything should have been restarted now
2022-08-02 20:02:11 <psykose> awesome, thanks
2022-08-02 20:02:23 <psykose> haha, yeah, it was much more yolo
2022-08-02 20:02:31 <psykose> but i think it'll be mostly ok
2022-08-02 20:03:07 <skarnet> I’m not complaining about yolo changes, I’m writing them down for the next time someone’s whining that a proposed change of mine is too risky. :P
2022-08-02 20:11:23 <psykose> Ariadne: maybe we should yolo libintl next? /s
2022-08-02 20:11:55 <Ariadne> does that one have an SCP in flight yet
2022-08-02 20:12:48 <psykose> no, aside from me spending a bunch of time on it and concluding it was ok
2022-08-02 20:12:54 <psykose> also i found an issue with chimera binsh already
2022-08-02 20:12:58 <psykose> -l is not an option..?
2022-08-02 20:13:26 <skarnet> what’s sh -l supposed to do?
2022-08-02 20:13:40 <ikke> login shell
2022-08-02 20:13:52 <skarnet> :stare:
2022-08-02 20:14:05 <Ariadne> that's not part of POSIX or BSD
2022-08-02 20:14:08 <Ariadne> that's a GNUism
2022-08-02 20:14:17 <psykose> for this it's uh
2022-08-02 20:14:19 <psykose> sh -
2022-08-02 20:14:22 <psykose> that's the equivalent to -l
2022-08-02 20:14:26 <Ariadne> none the less, it should be trivial to implement
2022-08-02 20:14:26 <ikke> nod
2022-08-02 20:14:28 <psykose> well, it's a breaking change for whoever used -l
2022-08-02 20:14:34 <psykose> (until that, yeah)
2022-08-02 20:14:36 <psykose> like me
2022-08-02 20:14:37 <skarnet> try “exec -a -/bin/sh -- /bin/sh”
2022-08-02 20:14:48 <psykose> i had -l in the ssh command and couldn't ssh back to the server
2022-08-02 20:14:48 <psykose> haha
2022-08-02 20:14:58 <skarnet> no really
2022-08-02 20:15:14 <skarnet> shells think they’re a login shell when the first character of their invocation name is a -
2022-08-02 20:15:20 <Ariadne> adding -l seems trivial
2022-08-02 20:15:31 <Ariadne> i suspect chimera will want that
2022-08-02 20:15:58 <skarnet> it’s not even official yet and the additional mutations on the mutated thing derived from a mutated thing begin
2022-08-02 20:16:08 <Ariadne> look s6sh
2022-08-02 20:16:10 <Ariadne> does not exist
2022-08-02 20:16:22 <psykose> i sourced /etc/profile
2022-08-02 20:16:23 <psykose> and got a
2022-08-02 20:16:27 <skarnet> it will be called smash
2022-08-02 20:16:27 <psykose> `${HOSTNAME%%.*}:$PWD$`
2022-08-02 20:16:41 <skarnet> literally “suck my ass shell”
2022-08-02 20:16:49 <psykose> also this does not read .profile for some reason ..?
2022-08-02 20:17:10 <Ariadne> weird
2022-08-02 20:17:13 <skarnet> or “sadomasochistic almquist shell”
2022-08-02 20:17:22 <Ariadne> ${HOSTNAME%%.*}:$PWD$ works on dash
2022-08-02 20:17:25 <Ariadne> 
ACTION shrugs

2022-08-02 20:17:30 <psykose> it works locally, just not on the server
2022-08-02 20:17:35 <psykose> what a strange bug
2022-08-02 20:17:58 <Ariadne> the main thing we want to know about is "does abuild explode"
2022-08-02 20:18:07 <skarnet> full distro rebuild go
2022-08-02 20:18:14 <psykose> yeah i know, just also have to work out these little things
2022-08-02 20:18:19 <Ariadne> although admittedly
2022-08-02 20:18:23 <Ariadne> this is not looking good :D
2022-08-02 20:18:36 <psykose> i mean it's only broken on ssh to this box for some reason
2022-08-02 20:18:39 <psykose> and i am utterly confused
2022-08-02 20:18:55 <Ariadne> mksh it is
2022-08-02 20:18:58 <skarnet> how do you say “I told you so” without looking like an insufferable prick
2022-08-02 20:19:05 <psykose> ok peaches
2022-08-02 20:19:09 <psykose> what would you rather i did instead
2022-08-02 20:19:10 <psykose> literally nothing?
2022-08-02 20:19:15 <psykose> all you did was sit here and complain
2022-08-02 20:19:16 <psykose> now silence
2022-08-02 20:19:41 <skarnet> no you misunderstand me, I applaud and encourage your patience and efforts
2022-08-02 20:19:42 <psykose> i prefer to do-by-testing and not do-by-talking-for-8-years
2022-08-02 20:19:57 <Ariadne> anyway, we should also test with mksh
2022-08-02 20:20:24 <Ariadne> or, i dunno
2022-08-02 20:20:24 <psykose> ikke: you were talking about some lxd ssh login .profile somethingsomethingsomething bug before with these containers
2022-08-02 20:20:27 <psykose> do you remember some specifics
2022-08-02 20:20:49 <ikke> psykose: only thing I'm aware of is with rv64
2022-08-02 20:20:52 <psykose> hmm
2022-08-02 20:21:07 <ikke> where qemu-user prevents sh from seeing it as a login shel
2022-08-02 20:21:09 <ikke> shell
2022-08-02 20:21:35 <psykose> wait it works
2022-08-02 20:21:45 <psykose> but it breaks if i have `RemoteCommand exec /bin/sh` ???
2022-08-02 20:21:47 <psykose> what
2022-08-02 20:21:57 <ikke> that's not a login shell?
2022-08-02 20:22:03 <psykose> any arguments
2022-08-02 20:22:06 <skarnet> docker exec isn’t a login shell
2022-08-02 20:22:08 <Guest6913> why does it always leak my IP when I join?
2022-08-02 20:22:11 <psykose> -i would make it one
2022-08-02 20:22:19 <ikke> Guest6913: that's irc
2022-08-02 20:22:39 <Guest6913> i dont use irc very often
2022-08-02 20:22:55 <ikke> Guest6913: it's from a time where everything was much more innocent and showing people your IP as not a big deal
2022-08-02 20:23:01 <psykose> but i'm really confused
2022-08-02 20:23:03 <skarnet> docker exec -it just gives you a terminal, doesn’t guarantee it’s a login shell
2022-08-02 20:23:08 <Ariadne> we have recorded a CRIME on your IP address, please pay the fine via moneygram
2022-08-02 20:23:16 <psykose> why does `RemoteCommand exec /bin/sh -i` fail
2022-08-02 20:23:18 <ikke> skarnet: this is not docker exec..
2022-08-02 20:23:30 <Ariadne> haha imagine alpine actually using docker
2022-08-02 20:23:31 <psykose> it doesn't with ash
2022-08-02 20:23:52 <Ariadne> this is lxc fam
2022-08-02 20:24:21 <psykose> if i run `exec /bin/sh -i` myself after connecting (by ssh docs this is the same) it also works
2022-08-02 20:24:25 <psykose> utter confusion
2022-08-02 20:25:31 <skarnet> sigh, I know it’s lxc, but that doesn’t change a remote command execution in the container... okay -i may have a different meaning with lxc remote but the documentation is... um...
2022-08-02 20:25:49 <Ariadne> :D
2022-08-02 20:27:05 <psykose> why does the container matter at all? this is an ssh command it executes after connecting 'with the users shell'
2022-08-02 20:29:32 <skarnet> okay, please hear me out
2022-08-02 20:29:42 <Guest6913> does anyone know how to change the default shell for the 'adduser' command? i know i can pass the -s flag, but some scripts like setup-user don't allow that by default. i found a wiki 'Change the default shell' but it doesnt tell me how to do this
2022-08-02 20:29:53 <skarnet> why are you specifying a RemoteCommand in your ssh config?
2022-08-02 20:30:52 <psykose> a workaround for a bug so i could pass -l to busybox before, but who cares?
2022-08-02 20:31:28 <skarnet> I care because the way ssh and friends pass arguments on the command line is super brittle
2022-08-02 20:31:48 <psykose> aha
2022-08-02 20:31:51 <Ariadne> what if we switched to powershell
2022-08-02 20:32:08 <Habbie> does powershell even compile against musl?
2022-08-02 20:32:20 <Ariadne> yeah
2022-08-02 20:32:22 <skarnet> and if something isn’t working as you expect, start by removing all the legacy workarounds you have
2022-08-02 20:32:27 <Habbie> Ariadne, ok, cool :)
2022-08-02 20:32:33 <psykose> that is the only one
2022-08-02 20:32:40 <Ariadne> i look forward to APKBUILD.ps1
2022-08-02 20:32:43 <psykose> anyway, i figured out why it's broken
2022-08-02 20:32:49 <psykose> well
2022-08-02 20:32:50 <psykose> not why
2022-08-02 20:33:17 <psykose> a sub-shell of chimera-sh breaks
2022-08-02 20:33:44 <psykose> if you run `sh` or `sh -i` or `exec sh` etc, then . /etc/profile, it's fucked
2022-08-02 20:33:53 <psykose> if it was the initial login shell (via login or ssh), it works
2022-08-02 20:33:58 <psykose> if you enter anything past that, it's broken
2022-08-02 20:34:07 <skarnet> 
ACTION remains silent, more silent than he’s ever been, so silent you can’t even tell he’s here

2022-08-02 20:34:21 <psykose> do u know why
2022-08-02 20:34:33 <Ariadne> because we should all be using execline
2022-08-02 20:34:35 <skarnet> no but I can experiment
2022-08-02 20:34:53 <skarnet> I have suspicions
2022-08-02 20:34:54 <psykose> merely being in anything and execing into it then trying to source is broken
2022-08-02 20:34:57 <psykose> so.. weh
2022-08-02 20:34:58 <panekj> the shell is practically written
2022-08-02 20:35:01 <psykose> seems like an actual bug
2022-08-02 20:35:08 <panekj> it just has super weird frontend and new one will take years
2022-08-02 20:35:23 <psykose> aside from that, back to the rebuilds
2022-08-02 20:36:38 <ikke> Guest6913: apparently either from SHELL, or it uses the same shall as the current user
2022-08-02 20:37:52 <skarnet> psykose: so is it a subshell, or a new explicit shell process (or the same process but after one exec) that fails? subshells are when you do “a | b” for instance, a and b run in subshells
2022-08-02 20:37:53 <psykose> ./APKBUILD: ${_i/...}: Bad substitution
2022-08-02 20:37:57 <psykose> buildrepo fails instantly
2022-08-02 20:38:05 <psykose> experiment going real well instantly
2022-08-02 20:38:21 <psykose> skarnet: not a subshell like that, i mean just typing anything with `sh`
2022-08-02 20:38:50 <psykose> i.e. i am not in chimera sh currently. if i type `sh` or `exec sh` or `exec sh -i` then . /etc/profile it will be broken
2022-08-02 20:39:22 <psykose> the only way for it to not be broken is to log in as it directly, either via ssh on an account with /bin/sh as login, or su -, etc
2022-08-02 20:39:32 <psykose> anything else breaks
2022-08-02 20:39:35 <psykose> (so far)
2022-08-02 20:39:39 <Ariadne> ok i think we have spent enough time on freebsd sh
2022-08-02 20:39:41 <Ariadne> lets try
2022-08-02 20:39:43 <Ariadne> mksh
2022-08-02 20:39:47 <ikke> breaks meaning?
2022-08-02 20:39:53 <Ariadne> that was going to be my next suggestion
2022-08-02 20:40:04 <psykose> well.. if i source profile i get ${HOSTNAME%%.*}:$PWD$
2022-08-02 20:40:18 <ikke> ok, your prompt is broken
2022-08-02 20:40:36 <Ariadne> its because its ='' i guess
2022-08-02 20:40:42 <Ariadne> try mksh :D
2022-08-02 20:40:49 <psykose> okie
2022-08-02 20:41:08 <Ariadne> worst case we make abuild depend on bash
2022-08-02 20:41:10 <Ariadne> :D
2022-08-02 20:41:17 <ikke> CVS repository :D
2022-08-02 20:41:30 <skarnet> yeah I think it’s just a case of prompt interpretation
2022-08-02 20:41:39 <skarnet> worst case it’s the prompt that’s broken, that’s not a big deal
2022-08-02 20:41:45 <Ariadne> no
2022-08-02 20:41:50 <Ariadne> freebsd sh fails for another reason
2022-08-02 20:41:54 <Ariadne> it cant parse APKBUILDs
2022-08-02 20:42:00 <skarnet> yeah that’s something else
2022-08-02 20:42:01 <Ariadne> because they have bashisms
2022-08-02 20:42:20 <Ariadne> i mean we can delete the bashisms
2022-08-02 20:42:21 <Ariadne> but
2022-08-02 20:42:27 <Ariadne> i feel like people might complain
2022-08-02 20:42:33 <skarnet> I thought the whole *point* of using this thing is because it was ash-compatible and had a good chance of understanding APKBUILDs
2022-08-02 20:42:33 <ikke> means we need to exec a lot more in global scope 
2022-08-02 20:42:43 <psykose> >>> mksh: Fetching https://www.mirbsd.org/MirOS/dist/mir/mksh/mksh-R59c.tgz
2022-08-02 20:42:44 <ikke> yeah, that's what I thought as well
2022-08-02 20:42:47 <psykose> curl: (35) error:0A000152:SSL routines::unsafe legacy renegotiation disabled
2022-08-02 20:42:48 <psykose> x D
2022-08-02 20:42:56 <ikke> psykose: ouch
2022-08-02 20:43:01 <Ariadne> skarnet: i mean, it is ash
2022-08-02 20:43:02 <ikke> well, they are still using cvs as well
2022-08-02 20:43:05 <psykose> it's because of openssl3, they bumped some stuff
2022-08-02 20:43:08 <psykose> i bet it works from 1.1
2022-08-02 20:43:19 <Ariadne> ok
2022-08-02 20:43:19 <Ariadne> ok
2022-08-02 20:43:21 <Ariadne> new plan
2022-08-02 20:43:23 <psykose> yeah, it does
2022-08-02 20:43:26 <Ariadne> we just extricate ash from busybox
2022-08-02 20:43:31 <Ariadne> and clean it up
2022-08-02 20:43:34 <Ariadne> and by clean it up
2022-08-02 20:43:37 <Ariadne> i mean rewrite the parser
2022-08-02 20:43:39 <psykose> it's tied into the rest of bb isn't it
2022-08-02 20:43:43 <psykose> yes!!! in rust!!!!!!!!!!!!!!
2022-08-02 20:43:45 <Ariadne> not really
2022-08-02 20:43:47 <ikke> psykose: https://github.com/MirBSD/mksh/archive/refs/tags/mksh-R59c.tar.gz
2022-08-02 20:43:52 <psykose> thanks bestie
2022-08-02 20:44:20 <Ariadne> we can call it
2022-08-02 20:44:21 <Ariadne> smash
2022-08-02 20:44:26 <Ariadne> for smash-stacking shell
2022-08-02 20:44:28 <Ariadne> hell yeah
2022-08-02 20:44:42 <Ariadne> er, stack-smashing shell
2022-08-02 20:44:44 <psykose> ] conftest.c:10:5: error: 'bounded' attribute directive ignored [-Werror=attributes]
2022-08-02 20:44:48 <Ariadne> ok
2022-08-02 20:44:48 <Ariadne> fuck it
2022-08-02 20:44:49 <psykose> my god why does every shell suck shit
2022-08-02 20:44:50 <Ariadne> FUCK IT
2022-08-02 20:44:55 <Habbie> i preferred smash-stacking
2022-08-02 20:44:57 <psykose> (this is probably gcc12)
2022-08-02 20:44:58 <Ariadne> i will just
2022-08-02 20:45:02 <Ariadne> extricate busybox ash
2022-08-02 20:45:05 <psykose> no, i have 11
2022-08-02 20:45:06 <psykose> sigh
2022-08-02 20:45:11 <Ariadne> and start fixing it
2022-08-02 20:45:18 <psykose> Ariadne: it is a very dark path
2022-08-02 20:45:22 <psykose> very incredibly dark
2022-08-02 20:45:38 <Ariadne> i can use the freebsd one as reference
2022-08-02 20:45:48 <Ariadne> i'm sure denys hasn't deviated *that* much
2022-08-02 20:45:50 <ikke> Ariadne: while you're at it, can you pretty please add alt + . :P
2022-08-02 20:46:01 <psykose> what's that
2022-08-02 20:46:05 <skarnet> it’s going to end in SMASH, I can feel it
2022-08-02 20:46:13 <ikke> psykose: recall last argument
2022-08-02 20:46:25 <psykose> give example
2022-08-02 20:46:31 <ikke> echo foo
2022-08-02 20:46:37 <ikke> echo <alt+.>
2022-08-02 20:46:40 <psykose> oo
2022-08-02 20:46:41 <ikke> -> echo foo
2022-08-02 20:46:56 <psykose> asd <alt+.>
2022-08-02 20:46:59 <psykose> -> asd foo
2022-08-02 20:47:02 <ikke> yes
2022-08-02 20:47:11 <psykose> wow fish supports this
2022-08-02 20:47:13 <psykose> didn't even know
2022-08-02 20:47:14 <psykose> thanks
2022-08-02 20:47:21 <ikke> :D
2022-08-02 20:47:26 <Ariadne> how dark could it be
2022-08-02 20:47:32 <psykose> look
2022-08-02 20:47:33 <Ariadne> i rewrote pkg-config from scratch
2022-08-02 20:47:34 <psykose> i'm just saying
2022-08-02 20:47:43 <psykose> you may or may not end up with a heroin addiction in 5 years
2022-08-02 20:47:45 <Ariadne> i'm sure i can fork ash and fix it
2022-08-02 20:47:58 <Ariadne> bold assuming i don't have one now
2022-08-02 20:48:05 <psykose> scary statement bestie
2022-08-02 20:48:07 <psykose> but yeah idk
2022-08-02 20:48:08 <ikke> psykose: can you add your findings to the TSC ticket?
2022-08-02 20:48:09 <psykose> go for it
2022-08-02 20:48:10 <psykose> all of this sucks
2022-08-02 20:48:12 <skarnet> you already have a plushie addiction
2022-08-02 20:48:14 <psykose> ikke: sure thing
2022-08-02 20:48:37 <panekj> maybe alpine should use my shell
2022-08-02 20:48:40 <Ariadne> task bbash and fbsd ash and bolt them together np
2022-08-02 20:48:53 <Ariadne> since when did systemd have a shell
2022-08-02 20:49:07 <ikke> systemd-shell
2022-08-02 20:49:16 <skarnet> may I suggest delaying the frankensteinization of shells and keeping busybox ash until it’s the last bb applet standing
2022-08-02 20:50:18 <Ariadne> skarnet: we have enhancements we want to make to ash which are impossible because the code is shit
2022-08-02 20:50:54 <panekj> ariadne: it's written in rust and it's called j(ack)sh(it) :P
2022-08-02 20:52:14 <skarnet> oh, horsesh is an excellent shell name
2022-08-02 20:52:25 <skarnet> or bullsh
2022-08-02 20:52:32 <panekj> (;
2022-08-02 20:52:33 <ikke> nice domain names
2022-08-02 20:52:59 <ikke> I bet they're already taken
2022-08-02 20:53:37 <skarnet> Ariadne: what kind of enhancements and what do you need them for?
2022-08-02 20:53:55 <Ermine> apparently ash is beyond repair?
2022-08-02 20:54:23 <panekj> it was just said
2022-08-02 20:54:52 <Ariadne> Ermine: not beyond repair, but needing to be forked to be repaired properly
2022-08-02 20:54:57 <Ariadne> skarnet: libedit support, for one
2022-08-02 20:55:09 <Ariadne> (libedit/readline is what gets you that alt + . stuff)
2022-08-02 20:55:29 <skarnet> yeah you’re not adding that to a bb codebase
2022-08-02 20:55:43 <Ariadne> we can if we extricate it :P
2022-08-02 20:56:22 <panekj> fish would be great if it didn't completely break any kind of posix compat
2022-08-02 20:56:44 <Ariadne> i think most are happy with bb ash, just not with the code quality
2022-08-02 20:57:01 <ikke> I think that sums it up
2022-08-02 20:57:23 <psykose> ikke: done
2022-08-02 20:57:28 <ikke> thx
2022-08-02 20:58:57 <skarnet> so I’m looking at https://man.openbsd.org/editline.3
2022-08-02 20:59:20 <skarnet> and I see: EditLine *el_init(const char *prog, FILE *fin, FILE *fout, FILE *ferr);
2022-08-02 20:59:31 <skarnet> ah yes, I love it when an external library forces stdio on me
2022-08-02 20:59:44 <psykose> stdio good
2022-08-02 20:59:47 <skarnet> and then
2022-08-02 20:59:49 <skarnet> The editline library is designed to work with blocking I/O only.
2022-08-02 20:59:59 <psykose> : )
2022-08-02 21:00:02 <Ariadne> i mean
2022-08-02 21:00:09 <skarnet> welp if that’s the stuff you want then I’m DEFINITELY not writing a shell
2022-08-02 21:00:12 <Ariadne> the current shell
2022-08-02 21:00:16 <Ariadne> is blocking i/o only
2022-08-02 21:00:18 <Ariadne> sooo
2022-08-02 21:00:26 <skarnet> oh, for sure
2022-08-02 21:00:49 <skarnet> but wanting to add libedit to it doesn’t sound like an order of magnitude of an improvement, code quality-wise
2022-08-02 21:01:07 <Ariadne> its a quality of life improvement
2022-08-02 21:01:13 <Ariadne> though we would probably just use GNU readline
2022-08-02 21:01:17 <Ariadne> since busybox ash is GPL anyway
2022-08-02 21:01:33 <skarnet> readline is actively worse XD
2022-08-02 21:02:04 <Ariadne> 
ACTION shrugs

2022-08-02 21:02:33 <skarnet> well it’s your time and sanity
2022-08-02 21:02:54 <Ariadne> i just want a shell i can actually debug
2022-08-02 21:03:02 <skarnet> I’ma stop looking at shell stuff and go play violent videogames, fantasizing I’m napalming the whole software world
2022-08-02 21:03:41 <ikke> jirutka suggest yash as a shell
2022-08-02 21:04:37 <skarnet> yash queen
2022-08-02 21:04:39 <Ariadne> actually
2022-08-02 21:04:40 <Ariadne> yash
2022-08-02 21:04:44 <Ermine> dumb question: what about linenoise?
2022-08-02 21:04:46 <Ariadne> is basically already this
2022-08-02 21:04:48 <Ariadne> no
2022-08-02 21:04:50 <Ariadne> fuck linenoise
2022-08-02 21:04:57 <Ermine> well
2022-08-02 21:06:13 <orbea> 
ACTION reads yash and approves

2022-08-02 21:08:12 <Ariadne> psykose: try with yash
2022-08-02 21:08:15 <psykose> yea
2022-08-02 21:08:17 <psykose> doing that now
2022-08-02 21:08:22 <Ariadne> yash should work
2022-08-02 21:08:46 <skarnet> yash is looking good indeed
2022-08-02 21:08:59 <skarnet> now there’s what’s printed on the box and what’s delivered inside
2022-08-02 21:10:23 <ikke> don't read anything about parameter substitution like ${a/b/c}
2022-08-02 21:10:52 <ikke> psykose: speaking about html docs :P
2022-08-02 21:12:27 <Ariadne> yes, the bash parameter substitutions work
2022-08-02 21:12:40 <ikke> ok
2022-08-02 21:12:44 <Ariadne> no echo alt+.
2022-08-02 21:12:46 <Ariadne> though
2022-08-02 21:12:47 <Ariadne> :(
2022-08-02 21:14:01 <Ariadne> yash is 339KB on arm64
2022-08-02 21:14:18 <Ariadne> kinda fat, but i think it provides enough quality of life improvements that its justifiable
2022-08-02 21:14:52 <ikke> what dependencies?
2022-08-02 21:14:57 <psykose> ncurses
2022-08-02 21:15:19 <Ariadne> ncurses is a typical dependency on an alpine system anyway
2022-08-02 21:15:25 <psykose> ikke: what about html docs
2022-08-02 21:15:33 <ikke> psykose: docs/*.html
2022-08-02 21:15:35 <psykose> Ariadne: yes, but i really wish it wasn't :/
2022-08-02 21:15:44 <Ariadne> me too, but that's more about ncurses existing
2022-08-02 21:15:46 <psykose> i fucking hate ncurses and the whole TERM ecosystem
2022-08-02 21:15:51 <ikke> sorry, doc/*.html
2022-08-02 21:15:52 <psykose> broken pile of garbage
2022-08-02 21:16:41 <panekj> I hate whole unix shell situation
2022-08-02 21:16:42 <psykose> okay first impression
2022-08-02 21:16:45 <psykose> yash tab completion
2022-08-02 21:16:46 <psykose> is really good
2022-08-02 21:16:51 <psykose> this is better than bash/zsh
2022-08-02 21:16:57 <psykose> (stock)
2022-08-02 21:17:07 <Ariadne> can it run abuild
2022-08-02 21:17:10 <psykose> yeah next test
2022-08-02 21:17:26 <Ariadne> it supports more bash extensions than bbash
2022-08-02 21:17:51 <ikke> mkdir {a,b}
2022-08-02 21:17:54 <psykose> bbash
2022-08-02 21:18:00 <psykose> i missed the {a,x} stuff
2022-08-02 21:18:05 <Ariadne> hell yeah
2022-08-02 21:18:06 <Ariadne> that works
2022-08-02 21:18:08 <Ariadne> :D
2022-08-02 21:18:33 <ikke> oh, even autosuggestion
2022-08-02 21:18:37 <Ermine> yay
2022-08-02 21:19:41 <orbea> fwiw its possible to disable ncurses in yash
2022-08-02 21:19:56 <Ariadne> yeah but what features does that cost you
2022-08-02 21:19:58 <ikke> process substitution
2022-08-02 21:19:58 <orbea> but it offers some quality of life improvements
2022-08-02 21:20:13 <Ariadne> (i guess we could have yash-minimal and yash)
2022-08-02 21:20:15 <psykose> ikke: how to use buildrepo without the lua : bad argument #1 to 'pairs' thing
2022-08-02 21:20:34 <Ariadne> `buildrepo main`?
2022-08-02 21:20:38 <ikke> make sure you execute it in the correct dir (or pass it with -d)
2022-08-02 21:20:40 <ikke> yes
2022-08-02 21:20:48 <orbea> iirc yash without ncurses is fine for /bin/sh, but kind of limited with login shells
2022-08-02 21:21:23 <psykose> i'm in aports root and -a "$PWD" and it doesn't work
2022-08-02 21:21:24 <psykose> hm
2022-08-02 21:21:37 <ikke> psykose: you need to pass the repo
2022-08-02 21:21:38 <Ariadne> hmm
2022-08-02 21:21:42 <Ariadne> no
2022-08-02 21:21:43 <Ariadne> i verify
2022-08-02 21:21:45 <psykose> yes, that's the first arg
2022-08-02 21:21:46 <Ariadne> with yash as /bin/sh
2022-08-02 21:21:50 <Ariadne> buildrepo main fails
2022-08-02 21:21:53 <psykose> it fails with ash too
2022-08-02 21:21:56 <psykose> not shell related
2022-08-02 21:22:00 <psykose> i've ran into this before
2022-08-02 21:22:02 <psykose> just forgot why it fails
2022-08-02 21:22:18 <psykose> (buildrepo is not very good either)
2022-08-02 21:22:30 <ikke> let me rewrite it in go :P
2022-08-02 21:22:43 <psykose> o
2022-08-02 21:22:44 <psykose> nvm
2022-08-02 21:22:46 <psykose> ash fixes it
2022-08-02 21:22:48 <psykose> sigh
2022-08-02 21:22:53 <psykose> forgot i had it swapped
2022-08-02 21:22:56 <psykose> well, another shell failure to debug
2022-08-02 21:23:00 <psykose> can't even source aports
2022-08-02 21:23:09 <psykose> was looking much better so far though
2022-08-02 21:23:16 <q66_> psykose: i don't understand what the subshell breakage is about
2022-08-02 21:23:19 <q66_> what exactly breaks
2022-08-02 21:23:36 <psykose> i dunno, trying to source . /etc/profile doesn't work
2022-08-02 21:23:39 <Ariadne> let me figure out the buildrepo
2022-08-02 21:23:41 <psykose> if you want me to test specific things i can
2022-08-02 21:23:44 <q66_> psykose: well what does it say
2022-08-02 21:23:51 <Ariadne> abuild runs the main/gcc APKBUILD fine with yash
2022-08-02 21:23:51 <psykose> nothing, it sources, and breaks PS1
2022-08-02 21:23:58 <Ariadne> which is pretty esoteric
2022-08-02 21:24:37 <q66_> psykose: breaks how
2022-08-02 21:24:43 <q66_> do you set PS1 in /etc/profile
2022-08-02 21:25:12 <psykose> it's set there by default yes
2022-08-02 21:25:15 <psykose> q66_: https://img.ayaya.dev/W3xSZgnJbbYX.png
2022-08-02 21:25:42 <q66_> that looks like it sources indeed
2022-08-02 21:25:44 <orbea> fwiw this is my PS1 in yash: PS1='ct.$USER@g.$HOSTNAME:ci.${PWD/$HOME/\~}\$ '
2022-08-02 21:25:49 <orbea> err
2022-08-02 21:25:52 <q66_> psykose: what is the value of PS1 there?
2022-08-02 21:25:52 <orbea> paste fail
2022-08-02 21:26:00 <orbea> PS1='\fct.$USER@\fg.$HOSTNAME:\fci.${PWD/$HOME/\~}\$ '
2022-08-02 21:26:10 <psykose> https://img.ayaya.dev/6PqP9nezylgn is /etc/profile
2022-08-02 21:26:30 <psykose> (ignore the comments i forgot to remove them)
2022-08-02 21:26:41 <psykose> the rest is stock alpine /etc/profile after the PATH
2022-08-02 21:26:51 <psykose> tbh that else is never hit normally
2022-08-02 21:27:02 <orbea> you could also consider mksh is you haven't already, that is also a very good shell for /bin/sh
2022-08-02 21:48:58 <ikke> psykose: I just spent effort to remove all exec from global scope
2022-08-02 21:49:31 <psykose> ah
2022-08-02 21:49:35 <Ariadne> psykose: it hates running libtool
2022-08-02 21:49:35 <psykose> yes it's all
2022-08-02 21:49:46 <psykose> instead of first match with /./-
2022-08-02 21:49:46 <psykose> ugh
2022-08-02 21:49:57 <psykose> yeah that one is really needed for all the stuff we already have
2022-08-02 21:50:04 <psykose> Ariadne: like, any libtool script?
2022-08-02 21:50:11 <Ariadne> yeah they all die
2022-08-02 21:50:12 <psykose> what about slibtool? :p
2022-08-02 21:50:12 <Ariadne> :D
2022-08-02 21:50:16 <psykose> we can port to that too
2022-08-02 21:50:16 <Ariadne> well
2022-08-02 21:50:17 <Ariadne> not that one
2022-08-02 21:50:19 <psykose> ok
2022-08-02 21:50:23 <psykose> shall i port to slibtool?
2022-08-02 21:50:25 <psykose> any blockers?
2022-08-02 21:50:29 <psykose> i'll test all of aports
2022-08-02 21:50:34 <Ariadne> i think we should SCP that
2022-08-02 21:50:38 <Ariadne> and just fork bbash
2022-08-02 21:50:41 <psykose> or i can just push it and yolo
2022-08-02 21:50:50 <psykose> :D
2022-08-02 21:50:57 <Ariadne> using slibtool would improve build times
2022-08-02 21:51:00 <psykose> (i'm not sure why we haven't already)
2022-08-02 21:51:05 <psykose> was there any singular issue?
2022-08-02 21:51:10 <psykose> because i know it works well
2022-08-02 21:51:13 <psykose> for packaging and speed
2022-08-02 21:51:20 <Ariadne> have not tried in a long while
2022-08-02 21:51:32 <psykose> ok, i will go try a real buildrepo on main/ with slib
2022-08-02 21:52:30 <Ariadne> i think if we can extricate bbash we can refactor it to not suck
2022-08-02 21:52:42 <Ariadne> and then add shit from yash we want
2022-08-02 21:52:44 <Ariadne> :)
2022-08-02 21:53:07 <psykose> or we add these like two missing yash features
2022-08-02 21:53:12 <psykose> and once aports builds it's good to go
2022-08-02 21:53:16 <psykose> seems easier than new stuff
2022-08-02 21:54:39 <psykose> wrt libtool- isn't it the case autoconf projects have their libtool vendored-in anyway? so it's an improvement only for generated libtool via autoreconf or whatnot
2022-08-02 21:56:22 <Ariadne> well
2022-08-02 21:56:28 <Ariadne> idk how to debug the libtool dying issue
2022-08-02 21:56:38 <Ariadne> i will fuck with it later
2022-08-02 21:57:30 <psykose> was it eval:1: syntax error: `(' must be followed by `)' in a function definition
2022-08-02 21:57:44 <Ariadne> yes
2022-08-02 21:57:47 <psykose> haha
2022-08-02 21:57:54 <psykose> sure, guess we can figure it out later
2022-08-02 21:58:20 <Ariadne> i suspect something is parsing wrong
2022-08-02 21:58:33 <Ariadne> because otherwise
2022-08-02 21:58:35 <Ariadne> yash would be great
2022-08-02 21:58:42 <Ariadne> perfect fit for alpine
2022-08-02 21:58:49 <ikke> yeah, seems nice
2022-08-02 21:59:04 <ikke> jirutka tries to contact the author(s)
2022-08-02 21:59:22 <psykose> is pkgs.a.o down again
2022-08-02 21:59:50 <ikke> psykose: linode has issues again
2022-08-02 21:59:54 <psykose> :/
2022-08-02 22:00:13 <ikke> https://status.linode.com/incidents/8qnvzrfz7hg2
2022-08-02 22:00:18 <psykose> speaking of pkgs.a.o
2022-08-02 22:00:23 <psykose> weren't we meaning to improve it a bit
2022-08-02 22:00:32 <psykose> maybe i should write a replacement
2022-08-02 22:00:56 <ikke> psykose: pmos also has a replacement
2022-08-02 22:01:01 <ikke> We have been looking at that
2022-08-02 22:01:05 <psykose> i do remember all that
2022-08-02 22:01:16 <psykose> and i specifically want to actually improve it to something that is not $scriptinglanguage
2022-08-02 22:01:52 <ikke> What would you suggest instead?
2022-08-02 22:05:03 <panekj> what pmos has
2022-08-02 22:05:18 <panekj> oh wait
2022-08-02 22:05:25 <panekj> it's not shell topic anymore ;\
2022-08-02 22:05:33 <ikke> nope
2022-08-02 22:05:40 <ptrc> panekj: https://gitlab.com/postmarketOS/apkbrowser
2022-08-02 22:05:49 <panekj> I'm aware
2022-08-02 22:06:34 <panekj> we talked about it on friday in #alpine-infra (:
2022-08-02 22:06:40 <psykose> i would say even go is better and you know me
2022-08-02 22:06:56 <psykose> i just don't like scripting languages
2022-08-02 22:07:15 <panekj> I dont like python
2022-08-02 22:07:18 <psykose> no compiler is a huge pain without writing a huge testsuite and random stuff breaks all the time unintentionally
2022-08-02 22:07:28 <psykose> that and performance
2022-08-02 22:07:29 <psykose> meh
2022-08-02 22:08:49 <psykose> for an actual web service that would be seldom touched and just renders some pages it's not a crazy idea to write something else, but i would have to spend some time
2022-08-02 22:08:56 <psykose> and go over the full featureset
2022-08-02 22:09:44 <panekj> > some time go I started rewriting aports-turbo (in rust, yes, kill me)
2022-08-02 22:10:01 <psykose> if you get it mostly done i could finish the rest
2022-08-02 22:10:14 <psykose> or i could do the same for fun
2022-08-02 22:10:17 <psykose> lots of things to do though
2022-08-02 22:10:17 <panekj> it renders the page
2022-08-02 22:10:21 <psykose> wow
2022-08-02 22:10:45 <panekj> getting it to function likewith aports-turbo
2022-08-02 22:10:55 <panekj> s/likewith/like
2022-08-02 22:10:55 <alpine-meetbot> panekj meant to say: getting it to function like aports-turbo
2022-08-02 22:11:06 <panekj> is a bit harder
2022-08-02 22:11:19 <psykose> anyway if it's wanted i could work on it, but i would need access to the $everything to deploy it
2022-08-02 22:11:32 <ptrc> rewriting aports-turbo in go could make use of alpine/go and skip some time (though well, alpine/rust would be cool too :p )
2022-08-02 22:11:54 <panekj> > I would do it in Go and it would take me much less time, but given that Go won't enter `main`, and rust will, I decided to give rust a try (:
2022-08-02 22:12:08 <psykose> i don't think rust will enter main anytime soon
2022-08-02 22:12:14 <psykose> 2 year support is impossible
2022-08-02 22:12:16 <psykose> even 1 is pushing it
2022-08-02 22:12:31 <panekj> Go will never enter
2022-08-02 22:12:38 <panekj> so there is that
2022-08-02 22:12:39 <psykose> based what i'm seeing in https://github.com/rust-lang/libs-team/issues/72
2022-08-02 22:12:47 <psykose> it might not even be possible to have rust in community
2022-08-02 22:12:48 <psykose> so
2022-08-02 22:12:59 <psykose> (slight /s but like lol)
2022-08-02 22:13:51 <panekj> that issue says about at least N-1
2022-08-02 22:13:59 <psykose> N-1 is 6 weeks
2022-08-02 22:14:58 <psykose> imagine not updating your rust compiler for 3 months (12 weeks) and suddenly a random bump of stuff doesn't build anymore because libc broke, lmao
2022-08-02 22:30:40 <sam_> owned
2022-08-02 22:34:27 <psykose> Ariadne: looks like the way to use this is to add MAKEFLAGS="LIBTOOL=slibtool"?
2022-08-02 22:35:21 <orbea> psykose: you want rlibtool in almost all cases
2022-08-02 22:35:51 <psykose> not mentioned in the readme for some reason
2022-08-02 22:35:53 <orbea> it will then respect the share/static configure arguments
2022-08-02 22:36:02 <orbea> *Shared
2022-08-02 22:36:22 <psykose> sure
2022-08-02 22:36:45 <psykose> looks like it doesn't provide the /usr/bin/libtoolize stuff anymore, does that break anything if it's gone
2022-08-02 22:36:45 <orbea> but it has the caveat that the build needs the actual generated libtool file and to not use /usr/bin/libtool directly
2022-08-02 22:37:00 <orbea> yea, you need that for autoreconf
2022-08-02 22:37:08 <psykose> so.. it's not a full replacement?
2022-08-02 22:37:13 <orbea> slibtoolize was planned, but midipix is pretty busy lately
2022-08-02 22:37:16 <psykose> aw
2022-08-02 22:37:22 <psykose> well, guess i'll wait for that then
2022-08-02 22:37:30 <psykose> don't want to just add another shitty libtool
2022-08-02 22:38:02 <orbea> there was a lot of effort that has gone into gentoo to make it work, it does in almost all cases, but a lot of build systems had to be improved for that...
2022-08-02 22:38:15 <psykose> yeah, fixing random builds is fine by me :)
2022-08-02 22:38:34 <psykose> i just want to fully migrate over to something better and fix a few stragglers, not have two things live side by side for a while
2022-08-02 22:39:11 <psykose> also would be nice if it didn't need a makeflag but that's a bit too ideal
2022-08-02 22:41:33 <orbea> you can see the current remaining issue at gentoo here https://bugs.gentoo.org/showdependencytree.cgi?id=765709&hide_resolved=1
2022-08-02 22:41:43 <psykose> thanks!
2022-08-02 22:41:50 <orbea> but yea, this doesn't solve ltdl or libtoolize yet...
2022-08-02 22:47:36 <orbea> actually now that i thought about it there is https://dev.midipix.org/cross/sltdl
2022-08-02 22:47:54 <orbea> but its scant on documentation and idk how well it works for the few ltdl cases
2022-08-02 22:48:49 <orbea> you can see them by searching the issues page for "undefined reference to lt_PROGRAM_LTX_preloaded_symbols"
2022-08-02 22:50:03 <psykose> haha, there's even a bautomake
2022-08-02 22:50:11 <psykose> i guess it won't end until all of autotools is replaced
2022-08-02 22:56:50 <orbea> libtool is probably the worst part of autotools currently
2022-08-02 22:57:16 <orbea> in that is seemingly unmaintained and unmaintainable
2022-08-02 22:57:46 <orbea> does anyone understand how it even works anymore? :P
2022-08-02 22:57:48 <omni> 323 is a nice nuber
2022-08-02 22:57:51 <omni> number*
2022-08-02 22:58:32 <psykose> hey it got a version bump recently
2022-08-02 22:58:47 <psykose> took them 7 years but they did it
2022-08-02 22:59:56 <orbea> yea, it got a little work recently https://git.savannah.gnu.org/cgit/libtool.git/log/
2022-08-02 23:00:50 <sam_> the new maintainer is nice as well
2022-08-02 23:03:51 <orbea> any idea how to tell them that -no-undefined is broken in libtool? :)
2022-08-02 23:07:17 <sam_> email them personally or bug-libtool
2022-08-02 23:07:58 <orbea> will have to try to remmeber to bug them about that when I Have a chance
2022-08-02 23:12:58 <sam_> cc me if/when u do
2022-08-02 23:13:27 <psykose> cute cc: sam_ 
2022-08-02 23:31:13 <sam_> <3
2022-08-03 05:59:34 <ncopa> morning
2022-08-03 05:59:42 <ncopa> i dont think we decided that busybox must go
2022-08-03 05:59:59 <ncopa> or maybe there was a meeting where I didnt attend?
2022-08-03 06:00:47 <psykose> lol
2022-08-03 06:18:09 <nmeum> Ariadne: btw: any updates on the gcc 12 front? 🥺
2022-08-03 06:18:42 <ncopa> We never decided that busybox must go. what we discussed is here: https://gitlab.alpinelinux.org/alpine/tsc/-/blob/master/minutes/2022-08-02.md
2022-08-03 06:19:24 <ncopa> we discussed: 1) if we should replace busybox sh with something else? conclusion was that it needs more investigation.
2022-08-03 06:19:37 <ncopa> and that we need to postpone that decision
2022-08-03 06:19:58 <ncopa> 2) if we drop s390x. decision was no. not at this time
2022-08-03 06:20:42 <psykose> :plea
2022-08-03 06:20:43 <ncopa> 3) if we should rework busybox-initscripts and device manager handling. decision was yes
2022-08-03 06:22:03 <ncopa> we did talk about replacing busybox when discussing the other issues, but concluded that needs to be handled in separate agenda 
2022-08-03 06:22:15 <ncopa> as separate agenda item
2022-08-03 06:27:27 <nmeum> right
2022-08-03 06:28:51 <psykose> why do the builders feel so slow for these rebuilds
2022-08-03 06:29:07 <psykose> i've been awake for forever and it's barely a quarter of community/
2022-08-03 06:29:09 <psykose> grr
2022-08-03 07:41:13 <DylanVanAssche> Alpine is switching to openssl3, could it cause build issues with for example gst-plugins-bad? https://gitlab.alpinelinux.org/dylanvanassche/aports/-/jobs/790841#L463
2022-08-03 07:42:48 <ptrc> you probably just need to wait for the rebuild to finish
2022-08-03 07:43:03 <panekj> Caused by:
2022-08-03 07:43:03 <panekj>   [77] Problem with the SSL CA cert (path? access rights?) (error setting certificate file: /etc/ssl/certs/ca-certificates.crt)
2022-08-03 07:43:03 <panekj> error: failed to download from `https://crates.io/api/v1/crates/backtrace/0.3.66/download`
2022-08-03 07:43:11 <panekj> (:
2022-08-03 07:43:13 <ddevault> after an upgrade today I started getting openssl issues as well
2022-08-03 07:43:20 <ddevault> SSL error connecting imap.migadu.com (141.95.144.66:993): unable to get local issuer certificate
2022-08-03 07:43:20 <ptrc> because currently only 2 things outside testing/ depend on openssl1.1-compat-dev
2022-08-03 07:44:20 <ddevault> it would be nice if when edge is going through big risky upgrades that it was somehow communicated to users
2022-08-03 07:50:46 <DylanVanAssche> ptrc: Thanks will try later then 🙂 
2022-08-03 07:51:41 <ptrc> ddevault: there was quite a bit of discussion about openssl3 migration over the past 2 weeks and an announcement from ncopa yesterday morning that it's going live soon
2022-08-03 07:52:01 <ddevault> oh? where was that announcement posted?
2022-08-03 07:52:27 <ptrc> here, i guess it should have been in #alpine-linux instead
2022-08-03 07:52:54 <ddevault> or on ~alpine/announce
2022-08-03 07:52:58 <ddevault> or at least devel
2022-08-03 07:53:57 <panekj> pfff, who uses email these days
2022-08-03 07:54:36 <ptrc> 
ACTION rolls eyes

2022-08-03 07:59:37 <ncopa> I pushed openssl3 yesterday. I can write an announcement to ~alpine/devel
2022-08-03 08:05:51 <psykose> DylanVanAssche: done for you
2022-08-03 08:07:13 <psykose> (i added the codecs)
2022-08-03 08:07:18 <psykose> how does one even use these
2022-08-03 08:07:18 <ncopa> re busybox ash, are there any currently unresolved issues in ash?
2022-08-03 08:07:34 <psykose> i think the ash patches just got applied upstream
2022-08-03 08:07:36 <psykose> probably not
2022-08-03 08:08:51 <DylanVanAssche> psykose: Ah thanks! Well PulseAudio has support for those and probes them at runtime to enable them. Better than shipping pulseaudio-modules-bt as an out-of-tree module again
2022-08-03 08:08:58 <psykose> ah
2022-08-03 08:09:11 <psykose> so.. pulseaudio bluetooth can use ldac via gst-bad
2022-08-03 08:09:39 <DylanVanAssche> Yes since 15.0 https://www.freedesktop.org/wiki/Software/PulseAudio/Notes/15.0/#supportforldacandaptxbluetoothcodecsplussbcxqsbcwithhigher-qualityparameters
2022-08-03 08:10:11 <psykose> oh wow
2022-08-03 08:10:12 <DylanVanAssche> But that these GStreamer elements were missing when running PulseAudio in debug mode
2022-08-03 08:10:23 <DylanVanAssche> *I saw
2022-08-03 08:10:27 <psykose> guess the modules-bt really is completely obsolete then
2022-08-03 08:10:36 <psykose> since gst-bad gives the same (if it works well)
2022-08-03 08:10:42 <psykose> well, hope it works when it builds in a day
2022-08-03 08:12:01 <DylanVanAssche> Indeed! Well I looked for it because of this pmOS issue: https://gitlab.com/postmarketOS/pmaports/-/issues/1628 Curious as well if it works properly
2022-08-03 08:13:15 <psykose> if you have an ldac/aptx device you can rebuild locally and check
2022-08-03 08:13:19 <psykose> shouldn't take super long
2022-08-03 08:14:08 <DylanVanAssche> I tried to build it with pmbootstrap but got stuck there as well with the same error 😛 I think I have a headset with aptX support, will test 
2022-08-03 08:16:26 <psykose> ah, right
2022-08-03 08:58:57 <ncopa> there will be openssl issues while upgrading now, til community repo is done
2022-08-03 09:00:04 <psykose> yes
2022-08-03 09:01:11 <ncopa> seems like its difficult to uninstall chimera-binsh: $ doas apk del chimera-binsh | tpaste
2022-08-03 09:01:11 <ncopa> https://tpaste.us/bnl4
2022-08-03 09:04:12 <psykose> yes, the provides= mechanism with apk doesn't work instantly for that
2022-08-03 09:04:15 <psykose> you have to del then add
2022-08-03 09:04:29 <psykose> it goes for everything that is virtual provides in multiple places
2022-08-03 09:16:46 <ncopa> hum, apk del --force chimera-binsh does not seem to work either
2022-08-03 09:18:57 <psykose> yeah i have no idea how virtuals are supposed to function
2022-08-03 09:19:01 <psykose> they're all like this
2022-08-03 09:19:35 <psykose> or with the other ways (provides= of a real pkgname, provides of the default cmd:) they're just broken entirely instead, so i dunno
2022-08-03 09:20:17 <ncopa> ah... i didnt get that i have to apk add busybox-binsh
2022-08-03 09:20:29 <psykose> yeah, though it's still two steps
2022-08-03 09:20:51 <ikke> You need something else to provide what you try to uninstall 
2022-08-03 09:20:51 <psykose> one would assume the moment you delete chimera-binsh, it would automatically pull busybox-binsh because it just has a higher priority
2022-08-03 09:21:02 <psykose> but it's not how it works
2022-08-03 09:21:05 <ikke> Yeah 
2022-08-03 09:21:45 <psykose> i would inquire more but the response would either be 'fixed in apk3' or 'working as intended', so i'm not sure if there's anything to really do
2022-08-03 09:22:05 <ncopa> be436fc192ab4bf6f22d67926243cedff3e95fa6
2022-08-03 09:22:39 <ncopa> its high risk change
2022-08-03 09:22:44 <psykose> it's safe on upgrade
2022-08-03 09:24:17 <ncopa> hm... would have been good to have waited til the dust after openssl3 upgrade had settled, but ok...
2022-08-03 09:24:44 <psykose> i don't see how it makes a difference
2022-08-03 09:25:42 <psykose> i don't disagree with being cautious, and i tested the upgrade paths
2022-08-03 09:25:47 <psykose> what scenarios are you thinking of
2022-08-03 09:26:26 <ncopa> well, interesting things may happen if /bin/sh doesnt exist
2022-08-03 09:26:50 <ncopa> also not sure how may downstream scripts rely on busybox providing /bin/sh
2022-08-03 09:27:11 <psykose> well
2022-08-03 09:27:20 <psykose> 1) i tested it doesn't disappear /bin/sh
2022-08-03 09:27:26 <psykose> 2) that has nothing to do with openssl
2022-08-03 09:27:33 <psykose> 3) busybox is still the only real provider of /bin/sh
2022-08-03 09:27:50 <psykose> the only way to get the other is to explicitly add it from testing
2022-08-03 09:28:45 <psykose> which goes to 4) that is to do with the SCP itself, not the specific change binsh there, which could stay forever regardless
2022-08-03 09:30:34 <psykose> uunless you mean `apk add busybox` to provide sh by itself without anything pulling /bin/sh downstream-wise, which this would no longer provide since it wouldn't pull in binsh
2022-08-03 09:31:36 <psykose> that's the only thing i can think of, it could be fixed with package() { depends="/bin/sh" } in busybox
2022-08-03 09:32:22 <psykose> so, yeah, i suppose that can be added
2022-08-03 09:32:25 <ncopa> might still be ok, due to packages having depend=/bin/sh will still pull in busybox-binsh
2022-08-03 09:32:32 <ncopa> it might not break anything
2022-08-03 09:32:45 <ncopa> its just a risky change, IMHO
2022-08-03 09:32:46 <psykose> that is the only breaking case (which does not apply to aports)
2022-08-03 09:32:48 <psykose> well
2022-08-03 09:32:57 <ncopa> and things may break in unexpected ways
2022-08-03 09:33:07 <psykose> unexpected implies nothing can be done
2022-08-03 09:33:11 <psykose> otherwise it would be expected
2022-08-03 09:33:32 <psykose> by that logic everyone should have their commit rights revoked and nobody should ever do anything, because something unexpected could happen
2022-08-03 09:33:35 <ncopa> and when that happens, it might not be evident if breakage is due to bin sh change or openssl change
2022-08-03 09:33:44 <psykose> ok, i will never push anything
2022-08-03 09:33:45 <psykose> you got me
2022-08-03 09:34:08 <ncopa> what i mean is touching /bin/sh has consequences
2022-08-03 09:34:16 <ncopa> if it breaks
2022-08-03 09:34:36 <ncopa> if /bin/foobar breaks, it does not have same consequences
2022-08-03 09:34:59 <ncopa> and yeah, things in edge may break occationally
2022-08-03 09:35:27 <psykose> the issue with this conversation is that you are not proposing anything
2022-08-03 09:36:20 <ncopa> well, im just expressing that i think it could have been wise to wait a few days with a risky change like that. nothing to do about it now
2022-08-03 09:37:02 <ncopa> alpine edge is currently broken anyway now, and we should try fix that
2022-08-03 09:37:14 <psykose> all there is to do is wait
2022-08-03 09:37:51 <ncopa> yeah
2022-08-03 09:47:37 <hcs> Is there any reason why Alpine 3.16 uses BusyBox 1.35.0 (unstable)?
2022-08-03 09:51:19 <psykose> 1.35 + 26 patches
2022-08-03 09:51:25 <ncopa> we updated busybox in jaunary 6, and thought it would be enough time to shake out issues for release in May
2022-08-03 09:51:43 <ncopa> apparently upstream has not yet pushed 1.35.1 (stable)?
2022-08-03 09:52:02 <psykose> mostly because it's very 'tightly integrated' and when you actively develop around something and test it 'unstable' versions are just the patches you would want to backport half the time anyway
2022-08-03 09:52:28 <psykose> afaik busybox does not have a very 'just made a stable release, now lets break everything on git' policy
2022-08-03 09:52:31 <hcs> I'm having difficulties with `grep -r` even with a directory only containing a single file. It just hangs forever. I'm quite certain that this bug was introduced in 1-35
2022-08-03 09:52:50 <psykose> can you post a reproduction
2022-08-03 09:52:53 <ncopa> also definition 'stable' of upstream busybox is a bit different of what we define as stable
2022-08-03 09:53:00 <psykose> yup
2022-08-03 09:54:07 <ncopa> busybox definition of 'stable' seems to be 'not so many bugs', while our definition is 'no breaking changes in the behavior'
2022-08-03 09:54:33 <hcs> psykose yes I can, I can reproduce with latest alpine docker image and on metal in three steps
2022-08-03 09:54:39 <ptrc> hcs: are you specifying a path to the directory? gnu grep allows for running just `grep -r keyword`, whereas with busybox you have to add the path too
2022-08-03 09:54:59 <ptrc> (and this behaviour has been around for longer than just 1.35)
2022-08-03 09:55:01 <psykose> ptrc: is that a change in behaviour though, or are like the past 5 busybox versions the same
2022-08-03 09:55:03 <hcs> ptrc: has that been a thing since always?
2022-08-03 09:55:08 <psykose> very long
2022-08-03 09:55:21 <psykose> what made you 'quite certain' it was 1-35? did you test 1-34?
2022-08-03 09:55:27 <ncopa> hcs: can you please create an issue, with the details on how to reproduce?
2022-08-03 09:55:38 <psykose> i.e. in 3.15 or 3.14
2022-08-03 09:55:43 <psykose> those still have the older versions
2022-08-03 09:56:41 <hcs> User error: busybox grep requires path as argument to grep. two of my developers highlighted the issue with a recent build of our alpine-based firmware, so I assumed (my bad) that this was a behaviour change.
2022-08-03 09:57:01 <psykose> happens, no worries 
2022-08-03 09:58:22 <hcs> I'll just ship gnu grep to make them happy. Thanks for the swift resolution of the problem :D
2022-08-03 10:47:04 <psykose> ikke: for some reason zabbix fails on s390x only
2022-08-03 10:47:46 <ikke> undefined reference to `SSL_get_peer_certificate'
2022-08-03 10:48:13 <psykose> yeah, not sure why
2022-08-03 10:55:21 <psykose> maybe some dependency was miscompiled, or s390x has special semantics for something relating to this via cgo
2022-08-03 10:55:22 <psykose> weh
2022-08-03 10:57:27 <ikke> fun
2022-08-03 11:20:30 <fcolista> hehe
2022-08-03 11:20:31 <fcolista> ERROR: openssl-dev-3.0.5-r1: trying to overwrite usr/include/openssl/ui.h owned by openssl1.1-compat-dev-1.1.1q-r1.
2022-08-03 11:20:53 <fcolista> so...should I wait that the repo are syncd I suppose..
2022-08-03 11:22:13 <psykose> on what specifically
2022-08-03 11:22:31 <fcolista> (193/244) Installing openssl-dev (3.0.5-r1)
2022-08-03 11:22:31 <fcolista> ERROR: openssl-dev-3.0.5-r1: trying to overwrite usr/include/openssl/aes.h owned by openssl1.1-compat-dev-1.1.1q-r1.
2022-08-03 11:22:39 <psykose> i mean in what build ^^
2022-08-03 11:22:54 <fcolista> ah sorry. 
2022-08-03 11:22:56 <fcolista> gvm-libs
2022-08-03 11:22:58 <psykose> ah
2022-08-03 11:23:01 <ncopa> i think it is because of community packages not yet being uploaded
2022-08-03 11:23:28 <psykose> yeah, libssh-dev depends openssl1.1-dev
2022-08-03 11:23:39 <psykose> so it pulls both until community rebuilt
2022-08-03 11:23:51 <ncopa> fcolista: you can work around it by rebuilding libssh locally
2022-08-03 11:23:52 <fcolista> yeah..i was wondering that I had to wait that all repos are in sync
2022-08-03 11:24:02 <fcolista> ncopa, thx. I'll do that
2022-08-03 11:24:19 <ncopa> (and all other dependencies that links to openssl)
2022-08-03 11:24:25 <fcolista> yeah..
2022-08-03 12:13:16 <skarnet> stupid kyua-related question
2022-08-03 12:13:31 <skarnet> when a test fails with: failed: atf-check failed; see the output of the test for details
2022-08-03 12:13:42 <skarnet> *how* do you see the output of the test?
2022-08-03 12:13:50 <skarnet> output is stored in a binary format database
2022-08-03 12:13:58 <skarnet> what’s the command to extract it?
2022-08-03 12:14:22 <skarnet> (also, who designs a fucking test framework that stores output in a binary database?)
2022-08-03 12:14:52 <psykose> kyua report
2022-08-03 12:14:57 <psykose> kyua report --results-file=
2022-08-03 12:15:10 <skarnet> thanks
2022-08-03 12:15:27 <psykose> it's a binary database but it's also sqlite so it could be a lot worse
2022-08-03 12:15:36 <skarnet> that’s never an argument
2022-08-03 12:15:40 <skarnet> it can *always* be worse
2022-08-03 12:15:51 <psykose> yes, but i consider that to be kinda even to json
2022-08-03 12:15:58 <psykose> ..for small sizes
2022-08-03 12:16:04 <psykose> for large outputs, i'd rather have sqlite
2022-08-03 12:16:20 <psykose> (between not-just-text options)
2022-08-03 12:16:22 <skarnet> kyua-report: not found
2022-08-03 12:16:35 <psykose> lol
2022-08-03 12:17:04 <skarnet> agh, my bad
2022-08-03 12:17:19 <skarnet> but uh
2022-08-03 12:17:38 <skarnet> “kyua report” tells me exactly the same thing
2022-08-03 12:17:52 <skarnet> a report of the failed tests
2022-08-03 12:18:08 <skarnet> and tells me to see the output of the test for details
2022-08-03 12:18:22 <skarnet> I have the report on my terminal, I’m looking for the test details
2022-08-03 12:19:43 <psykose> hmm
2022-08-03 12:22:02 <psykose> no idea, i always just saw the test std* output when they failed
2022-08-03 12:22:28 <skarnet> yeah I’m trying to extract information from it but I wondered whether there was more
2022-08-03 12:22:34 <skarnet> the message seemed to imply there was more
2022-08-03 12:24:37 <Ariadne> kyua debug
2022-08-03 12:24:47 <Ariadne> will run the test and dump the output
2022-08-03 12:35:00 <psykose> aha
2022-08-03 12:35:02 <psykose> convenient
2022-08-03 12:36:04 <skarnet> didn’t work for me (missing arguments) but it’s fine, apparently there isn’t more information than what gets dumped to stdout at the end
2022-08-03 12:36:24 <skarnet> so I’m making do with the diffs
2022-08-03 12:39:09 <psykose> i couldn't even figure out how to run one test
2022-08-03 12:40:01 <skarnet> alpine-conf is blessed with a “make check” instruction that hides the kyua details :)
2022-08-03 12:43:19 <psykose> yeah but i mean like
2022-08-03 12:43:21 <psykose> literally one test
2022-08-03 12:43:26 <psykose> not all of them
2022-08-03 12:46:07 <Ariadne> kyua debug <test name>
2022-08-03 12:46:11 <Ariadne> geeze
2022-08-03 12:46:38 <Ariadne> where testname is foo:bar stuff kyua prints
2022-08-03 12:46:50 <skarnet> hindsight is 20/20
2022-08-03 12:50:02 <psykose> i tested literally everything except also adding path/ in front
2022-08-03 12:50:13 <psykose> grr
2022-08-03 14:06:12 <psykose> a16bitsysop: fyi the rspamd tests segfault on aarch64 again
2022-08-03 14:06:15 <psykose> you should take a look
2022-08-03 14:42:29 <panekj> is ssl3 rebuild complete?
2022-08-03 14:44:06 <ncopa> for main and community for some of the architectures, yes
2022-08-03 14:44:11 <ncopa> for x86_64, not sure
2022-08-03 14:44:17 <ncopa> for s390x, no
2022-08-03 14:44:31 <ncopa> also, there are a bunch of packafges in testing that needs rebuild and fixes
2022-08-03 14:44:31 <panekj> I guess I'll just wait for x64
2022-08-03 14:46:58 <ncopa> i was not able to reproduce zabbix error on s390x in my dev env
2022-08-03 15:00:50 <Ariadne> i look forward to all the openssl 3 bugs
2022-08-03 15:01:25 <panekj> mhm, why rust segfaults
2022-08-03 15:07:16 <panekj> rust/cargo seems to be broken on x64 since the rebuild
2022-08-03 15:07:26 <ncopa> seems like zabbix builds after a rm ~/.cache/go-build
2022-08-03 15:08:06 <ncopa> panekj: have you got all the updates?
2022-08-03 15:08:10 <ikke> ncopa: oh, interesting
2022-08-03 15:08:35 <panekj> fetch https://dl-cdn.alpinelinux.org/alpine/edge/main/x86_64/APKINDEX.tar.gz
2022-08-03 15:08:35 <panekj> fetch https://dl-cdn.alpinelinux.org/alpine/edge/community/x86_64/APKINDEX.tar.gz
2022-08-03 15:08:35 <panekj> root@6347a1a736cc /lapce [SIGSEGV]# apk upgrade -Uia
2022-08-03 15:08:35 <panekj> OK: 1305 MiB in 225 packages
2022-08-03 15:08:35 <panekj> After this operation, 0 B of additional disk space will be used.
2022-08-03 15:08:59 <ncopa> if rust (from community) is built with openssl1.1 and curl (main) with openssl3, you'll have segfault yes
2022-08-03 15:09:47 <ncopa> what version of rust do you have installed?
2022-08-03 15:09:49 <panekj> ok, rust must be in queue still (60/71) Upgrading rust (1.62.1-r0 -> 1.62.1-r1) 
2022-08-03 15:09:57 <Ariadne> yes, it is -r2
2022-08-03 15:10:12 <ncopa> yeah. need tp be a bit patient... sorry
2022-08-03 15:11:22 <panekj> slight panic because added daily builds and switched from glibc to musl build in one project, and it failed on the first try :P
2022-08-03 15:12:15 <ncopa> you could use aarch64, which should be good to go now ;)
2022-08-03 15:12:29 <ncopa> not sure why x86_64 is so much behind
2022-08-03 15:12:42 <ncopa> https://build.alpinelinux.org/
2022-08-03 15:12:46 <ncopa> 57% done
2022-08-03 15:12:49 <Ariadne> some packages broke tests
2022-08-03 15:12:56 <Ariadne> on x86_64
2022-08-03 15:13:00 <Ariadne> but they worked on aarch64
2022-08-03 15:13:01 <ncopa> hum, ok
2022-08-03 15:13:27 <ncopa> i think they passed on my dev box
2022-08-03 15:13:38 <panekj> yes, my aarch64 will be fine but I need x64 build (aarch64 too but it's borked) and it's running on GitHub (:
2022-08-03 15:13:49 <ncopa> i might have skipped one or two
2022-08-03 15:14:11 <ncopa> panekj: builders are working on it, as fast as it can
2022-08-03 15:14:20 <ncopa> again, sorry for the inconvenience
2022-08-03 15:15:13 <panekj> no need to apologise
2022-08-03 15:15:36 <panekj> that upgrade was long overdue
2022-08-03 15:22:59 <Ariadne> is it really
2022-08-03 15:23:09 <ikke> ncopa: seems we need to do the same on x86_64 for zabixx
2022-08-03 15:23:14 <Ariadne> what if we upgraded the entire world to never using openssl ever again
2022-08-03 15:23:56 <panekj> Ariadne: do it, i dare you
2022-08-03 15:24:15 <Ariadne> i want to, unfortunately people like the self-abuse kit that is openssl
2022-08-03 15:24:15 <panekj> s6ssl when?
2022-08-03 15:24:56 <panekj> mhm, it would be nice if topics in build mqtt were grouped per alpine version
2022-08-03 15:25:32 <ikke> what if someone wants it per arch?
2022-08-03 15:25:35 <ikke> :)
2022-08-03 15:26:05 <ikke> so would be per arch + version
2022-08-03 15:27:46 <panekj> build -> {arch,version} -> build-{edge,etc.}-{armhf,armv7,etc.}
2022-08-03 15:30:01 <skarnet> panekj: it’s already there, has been for a while :P
2022-08-03 15:31:15 <panekj> I guess it would make it s8l
2022-08-03 15:31:27 <skarnet> lol
2022-08-03 15:31:35 <ncopa> ikke: ok im on it
2022-08-03 15:32:08 <ikke> ncopa: alright
2022-08-03 15:32:46 <skarnet> seriously, contributing bearssl ports to software in order to slowly remove dependencies to openssl would be awesome
2022-08-03 15:35:59 <panekj> bearssl bindings for rust, last updated 5 years ago
2022-08-03 15:38:00 <Ariadne> rust want to use rustls which is problematic for other reasons
2022-08-03 15:42:10 <ncopa> seems like everytime the x86_64 builder is fixed there is a new pile of packages to be built
2022-08-03 15:43:19 <ncopa> but we are making progress. slowly
2022-08-03 15:43:49 <ncopa> testing repo is broken though. will try fix as much of those tomorrow
2022-08-03 15:45:21 <ikke> As long as there is progress, we are getting somewhere :)
2022-08-03 17:20:45 <skarnet> the eternal question when you start fiddling with stuff
2022-08-03 17:20:55 <skarnet> how did that test even ever pass in the first place?
2022-08-03 17:21:15 <skarnet> that shit was never supposed to work
2022-08-03 18:00:20 <ncopa> what test?
2022-08-03 18:00:30 <ncopa> but yeah. been there
2022-08-03 18:01:16 <skarnet> setup-xorg-base
2022-08-03 18:01:26 <skarnet> doesn’t work when there’s a local repo in /etc/apk/repositories
2022-08-03 18:01:43 <skarnet> and it looks like the logic is completely broken, both inside and outside sed expressions
2022-08-03 18:01:50 <skarnet> still figuring out what it’s supposed to be doing
2022-08-03 18:02:04 <ncopa> the test or the script?
2022-08-03 18:02:38 <skarnet> the script
2022-08-03 18:02:50 <skarnet> the test fails because the script fails on sed invocation
2022-08-03 18:03:15 <ncopa> which of the sed invocations?
2022-08-03 18:03:43 <skarnet> the big one
2022-08-03 18:03:52 <skarnet> having dinner atm, can’t exactly focus, but I’ll fix it
2022-08-03 18:03:58 <ncopa> ok
2022-08-03 18:04:19 <ncopa> "i solved the problem I had with a regexp. Now I have two problems"
2022-08-03 18:05:31 <skarnet> exactly :D
2022-08-03 18:06:40 <ncopa> seems like it was me who introduced it 83c8d7fdac85ce2e4ffd919333c6976cee6f7c5d
2022-08-03 18:07:36 <skarnet> please don’t touch anything, I don’t wanna deal with merges
2022-08-03 18:07:54 <skarnet> I think I know what’s happening
2022-08-03 18:09:51 <ncopa> 👍
2022-08-03 18:29:07 <skarnet> O happy day (o happy day), o happy day, when the test passed (when the test passed), oh when it passed (oh when they passed), when the test passed (oh when it passed), it washed my pain away
2022-08-03 18:30:00 <skarnet> yeah the choirs got it wrong, they didn’t have much time to rehearse okay?
2022-08-03 18:36:43 <skarnet> gdi I can’t do it this way AAAAAAA
2022-08-03 18:37:08 <skarnet> o woeful day (o woeful day)
2022-08-03 19:14:50 <skarnet> There we go, https://gitlab.alpinelinux.org/alpine/alpine-conf/-/merge_requests/100
2022-08-03 19:15:08 <skarnet> when in doubt, don’t be smart
2022-08-03 19:15:30 <skarnet> when merged, this should fix !14046
2022-08-03 19:15:38 <skarnet> uh no
2022-08-03 19:15:41 <skarnet> #14046
2022-08-03 19:39:40 <Guest7035> does anyone know where the 'uname' command takes its values from? are they hardcoded or something?
2022-08-03 19:40:06 <panekj> busybox
2022-08-03 19:40:23 <ikke> Guest7035: it's a syscall
2022-08-03 19:40:30 <skarnet> https://man7.org/linux/man-pages/man2/sysinfo.2.html
2022-08-03 19:40:51 <ikke> https://www.man7.org/linux/man-pages/man2/uname.2.html
2022-08-03 19:40:53 <skarnet> uh, wrong one
2022-08-03 19:40:54 <Guest7035> so its hardcoded in the kernel? and to change it i would have to recompile the kernel?
2022-08-03 19:41:00 <skarnet> yeah, what ikke said
2022-08-03 19:41:00 <ikke> yes
2022-08-03 19:41:06 <Guest7035> ah
2022-08-03 19:41:08 <Guest7035> thanks
2022-08-03 19:41:17 <ikke> it would not make sense to be able to change the uname at runtime
2022-08-03 19:41:30 <Guest7035> yeah, i guess not
2022-08-03 20:09:30 <Ariadne> ikke: it might for cases where you are crosscompiling
2022-08-03 20:09:46 <Ariadne> julia's binarybuilder creates a fake environment where uname looks like the target
2022-08-03 20:10:39 <ikke> right, uname -m
2022-08-03 20:22:36 <Ariadne> man this openssl 3 rebuild sure is something
2022-08-03 20:22:55 <Ariadne> half my entire team is blocked in their work because php and python are segfaulting
2022-08-03 20:23:22 <Ariadne> time to upgrade them all to arm64 :))))
2022-08-03 20:23:28 <ikke> :D
2022-08-03 20:23:42 <panekj> > build-edge-x86_64 | failed
2022-08-03 20:23:45 <panekj> :c
2022-08-03 20:24:37 <Ariadne> ikke: can you put the builder in "fuck it" mode
2022-08-03 20:24:49 <Ariadne> so we can at least get an updated community repo
2022-08-03 20:24:50 <Ariadne> :D
2022-08-03 20:24:58 <ikke> heh
2022-08-03 20:25:04 <ikke> sure
2022-08-03 20:25:14 <panekj> wow, everything has updated rust except for x64 (and rv64 but who cares)
2022-08-03 20:25:35 <Ariadne> yeah there are people who are also being hit by rust segfaulting
2022-08-03 20:25:38 <Ariadne> it is very cool
2022-08-03 20:26:57 <panekj> I was wondering if it's possible to donate compute (container or something) for builders
2022-08-03 20:27:19 <Ariadne> its not
2022-08-03 20:27:21 <Ariadne> sadly
2022-08-03 20:29:13 <ikke> btw, nut, tut, hut, last 3 builds on rv64 :D
2022-08-03 20:31:37 <panekj> psykose: did jirutka ever responded to you about luajit?
2022-08-03 20:32:18 <panekj> (outside of gitlab ofc)
2022-08-03 20:38:31 <Guest7035> is it taking significantly longer to build alpine from mkimage? it used to build in 10 minutes or so in my machine, now its taking over 40 minutes
2022-08-03 20:46:36 <Hello71> setarch
2022-08-03 20:47:53 <Ariadne> Hello71: setarch does not work for `aarch64` on `x86_64` and such
2022-08-03 20:48:25 <Hello71> yes but you can set your kernel version to 2.6
2022-08-03 20:49:08 <Ariadne> :P
2022-08-03 20:50:16 <Guest7035> ls
2022-08-03 20:50:20 <Guest7035> oops
2022-08-03 20:55:58 <humm> sh: oops: not found
2022-08-03 20:56:24 <Guest7035> lol
2022-08-03 20:56:41 <Guest7035> i added a fixed workdir to the mkimage and now its build very fast
2022-08-03 23:25:26 <Guest7035> how can i change the default shell for 'adduser'? without passing a -s flag
2022-08-03 23:25:29 <Guest7035> if thats even possible
2022-08-03 23:26:35 <panekj> why
2022-08-03 23:27:05 <panekj> you can change shell after creating the user
2022-08-03 23:27:19 <panekj> just sed etc/passwd
2022-08-03 23:27:46 <panekj> but I don't see why would you ever want to change shell when doing adduser and not do it via the flag
2022-08-03 23:53:09 <Guest7035> the setup-user script, which is run in setup-alpine, does not allow you to pass a default shell (or at least i couldnt find any way to do it)
2022-08-03 23:53:24 <Guest7035> so the users created during the setup are all gonna have /bin/ash as their default shell
2022-08-03 23:53:38 <Guest7035> but i want them to use /bin/zsh by default, since i am going to customize it
2022-08-04 00:24:15 <Guest7035> MESA-LOADER: failed to open zink ... failed to open driver: bochs-drm ... failed to open libEGL.so.1 ...
2022-08-04 00:24:21 <Guest7035> happens when i startx
2022-08-04 00:25:01 <Guest7035> any idea what this could be? i have 'mesa' installed, xf86-video-*, xf86-input-*
2022-08-04 00:31:27 <Guest7035> also, lightdm doesnt show anything, just a black window with a blinking cursor in the top left
2022-08-04 00:36:20 <Guest7035> strange, when i install xfce, Xorg starts working, along with lightdm
2022-08-04 00:36:35 <Guest7035> maybe it has some dependency that makes everything work that i havent installed
2022-08-04 00:42:04 <Guest7035> 'mesa-egl' seems to have done the job
2022-08-04 01:01:07 <dalias> uhg why does clipper-dev have no .a file when the only conceivable way you could want to use this is to static link it??
2022-08-04 01:26:25 <panekj> Guest6553: just saying, but general support is in #alpine-linux
2022-08-04 01:26:47 <panekj> welp, they quit
2022-08-04 05:45:44 <psykose> panekj: never spoken to jirutka anywhere else
2022-08-04 05:47:06 <DylanVanAssche> psykose: aptX works with gst-plugins-bad now: https://gitlab.com/postmarketOS/pmaports/-/issues/1628#note_1050316819 🙂 
2022-08-04 05:47:38 <psykose> that's good
2022-08-04 06:04:26 <ncopa> morning
2022-08-04 06:04:33 <psykose> dalias: added
2022-08-04 06:04:35 <psykose> morning
2022-08-04 06:07:15 <ncopa> how are things going with the openssl3 mess i created?
2022-08-04 06:07:23 <ncopa> where is it burning the most?
2022-08-04 06:08:18 <psykose> i think nodejs-current is broken (tried a fix, will see) and anything that used npm
2022-08-04 06:08:24 <psykose> webpack tries to use md4 hash
2022-08-04 06:08:31 <psykose> and everything used old webpack, so
2022-08-04 06:38:39 <psykose> everything fixed
2022-08-04 06:38:52 <psykose> only electron might need to be fixed and i'll do that one
2022-08-04 06:55:07 <psykose> also seems to clear
2022-08-04 06:55:42 <psykose> ikke: did you add the noskip thing to s390x
2022-08-04 06:55:51 <psykose> (since i see zabbix wasn't changed but passed)
2022-08-04 06:59:07 <ikke> psykose: ncopa found out cleaning go build cache fixed it. 
2022-08-04 06:59:14 <psykose> ..
2022-08-04 06:59:17 <psykose> :/
2022-08-04 06:59:33 <psykose> one of the hard problems in computer science eh
2022-08-04 06:59:35 <ikke> And it was not only s390x
2022-08-04 06:59:38 <ikke> Yup
2022-08-04 06:59:51 <ikke> x86_64 failed as well 
2022-08-04 06:59:55 <psykose> i guess the go cache did not expect for things to be stale based on external library linking
2022-08-04 07:00:01 <psykose> which.. makes sense
2022-08-04 07:00:11 <psykose> the google engineers definitely don't give a fuck about anything outside of go itself
2022-08-04 07:00:27 <psykose> i guess i'll add GOMODCACHE or whatever it was to more things as i go through stuff
2022-08-04 07:00:32 <ikke> 
ACTION wants to adjust to the go modcache path for all aports

2022-08-04 07:00:37 <psykose> i mean, same
2022-08-04 07:00:37 <ikke> Heh
2022-08-04 07:00:48 <psykose> i wish we didn't have to type it into every apkbuild
2022-08-04 07:00:55 <psykose> that
2022-08-04 07:01:01 <psykose> 's why i generally don't do it
2022-08-04 07:01:14 <psykose> (that, and it's also very annoying when rebuilding locally)
2022-08-04 07:01:22 <psykose> i want to use the cache for local testing so it doesn't take 5 minutes on a clean build
2022-08-04 07:01:24 <ptrc> would be cool to have it as an abuild option
2022-08-04 07:01:26 <psykose> yes
2022-08-04 07:01:33 <psykose> for rust especially
2022-08-04 07:01:41 <ptrc> that, or in buildrepo which the builders use
2022-08-04 07:01:45 <psykose> running it with cargo set to srcdir is an extra minute of git cloning it every time
2022-08-04 07:08:20 <psykose> ikke: what was the correct variable for it
2022-08-04 07:08:23 <psykose> for go
2022-08-04 07:11:07 <panekj> GOPATH?
2022-08-04 07:19:56 <ikke> GOMODCAHCE 
2022-08-04 07:20:24 <ikke> GOPATH should not be used anymore 
2022-08-04 07:21:09 <ikke> There is also GOCACHE 
2022-08-04 07:21:23 <ikke> probably more relevant to this 
2022-08-04 07:22:34 <psykose> modcache is for downloaded
2022-08-04 07:22:39 <psykose> cache is for built objects
2022-08-04 07:22:43 <psykose> so.. just cache
2022-08-04 07:22:52 <psykose> modcache is safe cause that doesn't get used in any way, it's just some code
2022-08-04 07:23:12 <psykose> (and unless versions/sum/checksums/whatever break, that would not cause the above)
2022-08-04 07:23:17 <psykose> i trust that one more
2022-08-04 07:24:27 <ikke> Yes, but for my sake, please set both :-p
2022-08-04 07:26:19 <psykose> trust me, the former should be fine..
2022-08-04 07:26:22 <psykose> there's also gotmpdir
2022-08-04 07:29:08 <psykose> and they of course don't support being set to the same dir
2022-08-04 07:31:28 <psykose> really wish i could set these in abuild
2022-08-04 07:34:07 <ikke> psykose: I mean, to alleviate disk usage on the builders which I have to wipe every other day 
2022-08-04 07:34:16 <psykose> ah, sure
2022-08-04 07:34:25 <psykose> will fix all of aports, dw
2022-08-04 07:34:30 <psykose> gimme a bit
2022-08-04 07:34:47 <psykose> but i mean.. this really is going to be the same shit set in every single apkbuild
2022-08-04 10:25:05 <donoban> Hi, probably it's already known, I'm getting message like: ERROR: openssl1.1-compat-doc-1.1.1q-r1: trying to overwrite usr/share/man/man3/OSSL_STORE_INFO.3ssl.gz owned by openssl-doc-3.0.5-r1.
2022-08-04 10:25:17 <donoban> should I remove openssl1.1-compat?
2022-08-04 10:26:10 <ikke> do you have it explicitly installed (ie, is it in /etc/apk/world)?
2022-08-04 10:26:41 <donoban> yes
2022-08-04 10:26:53 <ikke> https://pkgs.alpinelinux.org/contents?file=&path=&name=OSSL_STORE_INFO.3ssl.gz&branch=edge&arch=x86_64
2022-08-04 10:26:59 <ikke> no package seems to include that file it appears
2022-08-04 10:27:14 <donoban> uhM
2022-08-04 10:27:31 <donoban> I have many more
2022-08-04 10:28:05 <donoban> http://ix.io/46Ds
2022-08-04 10:31:23 <psykose> ikke: https://pkgs.alpinelinux.org/contents?file=OSSL_STORE_INFO.3ssl.gz&path=&name=&branch=edge&arch=x86_64
2022-08-04 10:31:39 <ikke> oof
2022-08-04 10:33:57 <psykose> 937e2f9d3fd6e1d01e8c32dec39e81fa5554b6b2 fixed
2022-08-04 10:35:24 <donoban> ok
2022-08-04 10:59:36 <ismael> hi, reporting broken package in testing: fdm depends on so:lib{crypto,ssl}.so.3 
2022-08-04 11:00:02 <ismael> fdm-2.1-r1
2022-08-04 11:00:09 <psykose> installs fine
2022-08-04 11:00:12 <psykose> let me guess
2022-08-04 11:00:21 <psykose> you're on 3.16, but you installed packages from testing, which now need openssl3
2022-08-04 11:50:08 <fcolista> what's the quickest way with alpine to run 32bit app on 64bit?
2022-08-04 11:50:13 <fcolista> "can't load IA 32 .so on a AMD 64 platform"
2022-08-04 11:50:49 <fcolista> it's an old java app
2022-08-04 11:51:11 <fcolista> i could use a kvm with a 32bit os, but i don't want to virtualize it
2022-08-04 11:51:44 <Ermine> qemu-user?
2022-08-04 11:51:45 <ismael> psykose: ah, thanks, some leftover line in apk config :)
2022-08-04 11:54:42 <psykose> :)
2022-08-04 12:10:43 <Hello71> fcolista: chroot
2022-08-04 12:13:41 <panekj> Ermine: you don't need qemu for x86 on x64
2022-08-04 12:14:12 <ncopa> fcolista: chroot or docker or lxc
2022-08-04 12:14:21 <fcolista> it's a gui app..
2022-08-04 12:14:33 <fcolista> maybe lxc with ssh -x it's the quickest 
2022-08-04 12:14:49 <panekj> chroot via bwrap probably will be the easiest
2022-08-04 12:15:02 <ncopa> you can run gui apps in docker as well, need to share some socket
2022-08-04 12:15:38 <ncopa> https://l10nn.medium.com/running-x11-applications-with-docker-75133178d090
2022-08-04 12:16:24 <fcolista> ncopa, panekj, Hello71, thanks
2022-08-04 12:16:31 <fcolista> never used bwrap btw
2022-08-04 12:36:28 <psykose> ikke: for some reason `electron` dissappeared from x86_64
2022-08-04 12:36:40 <ikke> checking
2022-08-04 12:36:46 <psykose> not in pkgs.a.o either
2022-08-04 12:37:27 <panekj> https://build.alpinelinux.org/buildlogs/build-edge-x86_64/testing/electron/electron-19.0.10-r2.log
2022-08-04 12:38:22 <panekj> NPM script 'webpack' failed with code '1':
2022-08-04 12:38:24 <psykose> not relevant
2022-08-04 12:39:20 <ikke> It's indeed not in the repo, but buildrepo -n testing does show it is going to build it
2022-08-04 12:39:28 <psykose> so something deleted it from the mirror
2022-08-04 12:39:32 <psykose> well
2022-08-04 12:39:37 <ikke> https://tpaste.us/enlO 
2022-08-04 12:39:37 <psykose> it would be being built regardless
2022-08-04 12:39:39 <psykose> there was a rel bump
2022-08-04 12:39:54 <ikke> maybe because I used -k?
2022-08-04 12:40:00 <psykose> where on what
2022-08-04 12:40:07 <ikke> on x86_64 for buildrepo
2022-08-04 12:40:15 <ikke> (--keep-going)
2022-08-04 12:40:17 <ptrc> -k = keep going for buildrepo
2022-08-04 12:40:31 <psykose> that would mean `buildrepo -k -p ..` destroys packages
2022-08-04 12:40:41 <psykose> and you should drop one of the two
2022-08-04 12:40:48 <psykose> but it also has to get fixed
2022-08-04 12:40:49 <ikke> right
2022-08-04 12:40:51 <psykose> because that makes no sense
2022-08-04 12:41:02 <ikke> We normally use that only for new builders
2022-08-04 12:41:08 <ikke> where it's less of an issue
2022-08-04 12:42:01 <psykose> can you restart x86_64
2022-08-04 12:42:20 <ikke> Is it stuck?
2022-08-04 12:42:24 <ikke> ah
2022-08-04 12:42:26 <psykose> no, but it will take a while
2022-08-04 12:42:30 <psykose> may as well go to the new one
2022-08-04 12:43:01 <psykose> was it it still in -k, haha
2022-08-04 12:43:08 <ikke> no
2022-08-04 12:43:22 <ikke> Just untill it finished community yerstday
2022-08-04 12:43:32 <ikke> yesterday, if I can type
2022-08-04 12:43:47 <psykose> if it finished community then instantly got removed that would mean electron was removed for other reasons
2022-08-04 12:44:27 <psykose> (also you just killed that but instead of failing it went to the next package)
2022-08-04 12:44:30 <psykose> which look strange
2022-08-04 12:44:33 <ikke> hmm,
2022-08-04 12:44:37 <ikke> maybe it is still
2022-08-04 12:44:40 <psykose> i assume it's still running in -k
2022-08-04 12:44:59 <psykose> but yeah -k and -p should conflict
2022-08-04 12:44:59 <ikke> I did remove it and restart mqtt-exec
2022-08-04 12:45:21 <psykose> and then a final -p without -k would delete everything anyway
2022-08-04 12:45:32 <ikke> yes
2022-08-04 12:45:38 <ikke> I now restarted everything again
2022-08-04 12:46:49 <psykose> snipe something to double check
2022-08-04 12:46:57 <psykose> pkill gcc :p
2022-08-04 12:48:34 <ikke> Or just wait for hte next build failure :P
2022-08-04 12:51:00 <psykose> there wont be one
2022-08-04 12:51:21 <psykose> unless dotnet fails
2022-08-04 12:51:24 <psykose> so i guess maybe :)
2022-08-04 12:51:29 <psykose> also the arm dotnets are stuck
2022-08-04 12:51:36 <psykose> they ran out of space and are probably hanged now
2022-08-04 12:51:41 <psykose> since you saw the tiny space before
2022-08-04 12:51:48 <psykose> idk why the build itself doesn't fail when out of space
2022-08-04 12:51:58 <psykose> like the other time
2022-08-04 12:58:22 <ikke> yeah, gzip is using 100% cpu
2022-08-04 12:58:49 <ikke> Seems to still do something
2022-08-04 13:05:43 <psykose> ah
2022-08-04 13:05:46 <psykose> it calls tar -z
2022-08-04 13:05:51 <psykose> ok, i will 'improve' that a little
2022-08-04 13:06:08 <ikke> using pigz or something like that/
2022-08-04 13:06:10 <ikke> ?
2022-08-04 13:06:25 <Ariadne> the improvements will continue until openssl 3 ruins everyone's day again
2022-08-04 13:07:46 <psykose> ikke: yeah
2022-08-04 13:50:33 <andypost[m]> Is there index of ruins, I file #13020
2022-08-04 13:51:09 <andypost[m]> The #14020
2022-08-04 13:51:47 <panekj> just look at whatever is not built with openssl3
2022-08-04 13:51:49 <panekj> (:
2022-08-04 13:52:11 <ptrc> funny bug? weird behaviour in (abuild|apk-tools) discovered by psykose: if a package has the same script twice in $install, the resulting tar has the script, then a hardlink to itself(?), which results in an empty file being created
2022-08-04 13:52:29 <ikke> lol
2022-08-04 13:52:40 <psykose> and an exec format error on the resulting install
2022-08-04 13:53:45 <psykose> add tpm2-tss-fapi, bump the pkgrel, upgrade (to trigger the upgrade script)
2022-08-04 13:53:45 <psykose> haha
2022-08-04 13:54:17 <ptrc> also seen with:
2022-08-04 13:54:20 <ptrc> $ curl https://dl-cdn.alpinelinux.org/alpine/edge/community/x86_64/tpm2-tss-fapi-3.1.0-r1.apk | tar tvzf -
2022-08-04 13:54:55 <ptrc> ooor just `apk fetch --stdout tpm2-tss-fapi | tar tvzf -`
2022-08-04 13:55:06 <ptrc> i always forget apk has a fetch function
2022-08-04 14:12:52 <ncopa> psykose: do we need openssl3 with dotnet6? I'm not sure it will work
2022-08-04 14:13:07 <ncopa> 7c724040b82d57bb16e6b7b395117c23109a38ed
2022-08-04 14:13:30 <ncopa> does not look like its ready upstream: https://github.com/dotnet/runtime/issues/46526
2022-08-04 14:14:17 <psykose> guess it wasn't
2022-08-04 14:15:24 <ncopa> apk info -R dotnet6 shows that it should be ok to keept in on openssl1.1-compat for now. it does not seem to link against other lib that sues openssl3
2022-08-04 14:15:36 <ncopa> s/sues/uses/
2022-08-04 14:15:36 <alpine-meetbot> ncopa meant to say: apk info -R dotnet6 shows that it should be ok to keept in on openssl1.1-compat for now. it does not seem to link against other lib that uses openssl3
2022-08-04 14:15:49 <panekj> > 6.0 will still ship with OSSL3 support, it's just not as clean as I'd like.
2022-08-04 14:16:43 <ncopa> i thikn we can keep dotnet6 on openssl 1.1 for now
2022-08-04 14:16:49 <psykose> no, it's supported
2022-08-04 14:16:57 <psykose> even the linked issue of the guy has a fixed commit on v6.0.0
2022-08-04 14:17:00 <ncopa> openssl3?
2022-08-04 14:17:05 <psykose> these people are talking from beta pre 6.0 in 2021
2022-08-04 14:18:21 <psykose> the original person with the issue was on 3.1
2022-08-04 14:18:58 <psykose> ikke: you should restart the arm ones since they would fail anyway
2022-08-04 14:18:59 <ncopa> this comment claims the target was 6.0. https://github.com/dotnet/runtime/issues/46526#issuecomment-852235778 but did thye actually reach that target? the issue is still open and milestone set to 8.0
2022-08-04 14:19:35 <psykose> in the same way every fixed issue in aports gets assigned to the next release, i bet :p
2022-08-04 14:19:51 <psykose> that or it's just open so they can improve it
2022-08-04 14:20:30 <panekj> nah. 6 should support ssl3 just fine
2022-08-04 14:20:41 <panekj> they wanted to even go as far as to backport that to 5 and 3.1
2022-08-04 14:20:48 <psykose> fedora has it on 3
2022-08-04 14:20:57 <ncopa> seems like didnt make it to 6: https://github.com/dotnet/runtime/issues/46526#issuecomment-882932636
2022-08-04 14:21:10 <psykose> no, that's to get off the deprecated apis
2022-08-04 14:21:14 <psykose> the ones that are.. there
2022-08-04 14:21:14 <psykose> and work
2022-08-04 14:21:18 <ncopa> ok
2022-08-04 14:21:26 <ncopa> maybe we should ask maintainer
2022-08-04 14:21:33 <panekj> ncopa: you linked to comment that says 6.0 will ship with OSSL3 support
2022-08-04 14:22:15 <panekj> and as psykose said, 7.0 is target for moving off deprecated api
2022-08-04 14:22:22 <ncopa> oh.. l didnt read the last sentence :)
2022-08-04 14:22:23 <panekj> 7.0/8.0
2022-08-04 14:22:41 <ncopa> ok. so it should work. but it doesnt build currently
2022-08-04 14:22:54 <panekj> is ayakael still not on irc...
2022-08-04 14:23:21 <ncopa> it should be fixable atleast. good
2022-08-04 14:24:43 <panekj> mold 1.4 is to be released tomorrow
2022-08-04 14:25:49 <psykose> what about it
2022-08-04 14:26:17 <panekj> just giving you a heads up
2022-08-04 14:26:19 <panekj> (:
2022-08-04 14:59:04 <ptrc> testing a posix-strict shell, day 2: ifupdown-ng doesn't work, because `[` is technically not defined as a shell builtin and they wipe the whole environment, so no PATH
2022-08-04 14:59:06 <ptrc> this is so fun
2022-08-04 14:59:26 <ikke> ouch
2022-08-04 14:59:43 <ptrc> and by ifupdown-ng i mean the executors bundled with it, which are shell scripts
2022-08-04 14:59:58 <psykose> i guess there's a reason for `test` but i forget why everyone hates it
2022-08-04 15:01:33 <psykose> ah no, it's also not builtin
2022-08-04 15:01:48 <psykose> guess u cannot have any conditionals in posix sh while being builtin only
2022-08-04 15:01:48 <ptrc> mhm 
2022-08-04 15:02:03 <ptrc> builtin only *without any path whatsoever*
2022-08-04 15:04:16 <orbea> ptrc: [ seems to be fully posix? https://pubs.opengroup.org/onlinepubs/9699919799.2018edition/utilities/test.html
2022-08-04 15:04:26 <ptrc> i know that it is
2022-08-04 15:04:35 <orbea> the rational explains that test -o and test -a are potentially problematic
2022-08-04 15:04:44 <ptrc> but it's just in `Utilities`
2022-08-04 15:05:05 <ptrc> not in `Shell Command Language` -> `Special Built-in Utilities`
2022-08-04 15:05:05 <ikke> But why does it make sense to wipe the env?
2022-08-04 15:05:19 <orbea> its not specificied that 'test' is a builtin either
2022-08-04 15:05:22 <ptrc> it doesn't, and i'm not sure why ifupdown-ng does it; trying to find it in the source code now
2022-08-04 15:05:36 <psykose> indeed it's not specified
2022-08-04 15:05:38 <psykose> ur point
2022-08-04 15:06:05 <ikke> ptrc: oh, now I get it, ifupdown-ng wipes the env
2022-08-04 15:07:26 <ptrc> https://github.com/ifupdown-ng/ifupdown-ng/blob/ifupdown-ng-0.12.1/libifupdown/lifecycle.c#L298
2022-08-04 15:17:35 <ptrc> $ emptyenv ash -c 'echo $PATH'
2022-08-04 15:17:38 <ptrc>  /sbin:/usr/sbin:/bin:/usr/bin
2022-08-04 15:17:51 <ptrc> well, that makes sense then
2022-08-04 15:17:59 <ptrc> (as in, how it ever worked before)
2022-08-04 15:20:52 <orbea> ptrc: from posix: "If PATH is unset or is set to null, the path search is implementation-defined."
2022-08-04 15:21:01 <orbea> https://pubs.opengroup.org/onlinepubs/9699919799.2018edition/basedefs/V1_chap08.html
2022-08-04 15:21:14 <ptrc> yup, and in this case the path search is "don't search anywhere"
2022-08-04 15:21:25 <orbea> that in yash? :P
2022-08-04 15:21:39 <ptrc> in every shell i know
2022-08-04 15:22:01 <orbea> most shells have some default path if its unset
2022-08-04 15:22:51 <ptrc> yes, that's unrelated to the snippet, trying to find anything about default PATH in the spec rn
2022-08-04 15:24:15 <psykose> when's posix 2022 shell scripting being definede
2022-08-04 15:26:28 <ptrc> ask these guys: https://en.wikipedia.org/wiki/The_Open_Group#Members
2022-08-04 15:26:28 <alpine-meetbot> [WIKIPEDIA] The Open Group#Members | "The Open Group is a global consortium that seeks to "enable the achievement of business objectives" by developing "open, vendor-neutral technology standards and certifications." It has over 840 member organizations and provides a number of services, including strategy, management, innovation and research..."
2022-08-04 15:26:37 <ptrc> oh thank you alpine-meetbot 
2022-08-04 15:27:19 <ptrc> well, maybe it's just not a good "business objective" to have a decent base shell specification /s
2022-08-04 15:35:21 <ptrc> slowly losing my sanity reading the specification: "assuming /bin/sh is bad, you should do `getconf PATH` and search there"
2022-08-04 15:38:20 <psykose> hey it works
2022-08-04 15:39:50 <orbea> ptrc: if you havent seen it already you might find the second example interesting here :P https://pubs.opengroup.org/onlinepubs/9699919799.2018edition/utilities/command.html
2022-08-04 15:40:07 <orbea> the "Start off a "secure shell script" in which the script avoids being spoofed by its parent:"
2022-08-04 15:40:39 <skarnet> ptrc: the shell specification is batshit insane. That’s one of the cases where the lack of good implementations is directly tied to the suckiness of the spec.
2022-08-04 15:41:11 <orbea> the original bourne shell is pretty hard to use by todays standards
2022-08-04 15:42:07 <skarnet> we should really defined a set of “posix + a few extensions” (like local, some non-posix substitutions, etc.) and declare that this standard will be called “ash”
2022-08-04 15:42:13 <skarnet> s/defined/define/
2022-08-04 15:42:13 <alpine-meetbot> skarnet meant to say: we should really define a set of “posix + a few extensions” (like local, some non-posix substitutions, etc.) and declare that this standard will be called “ash”
2022-08-04 15:42:38 <skarnet> and then the implementation of this standard could be busybox ash or something else, maybe yash if it does fit the bill
2022-08-04 15:42:39 <psykose> does it have to be called ash
2022-08-04 15:42:49 <panekj> alpine shell
2022-08-04 15:43:08 <ikke> alsh doesn't sound nice
2022-08-04 15:43:12 <skarnet> call it mash if you want
2022-08-04 15:43:41 <ptrc> modern alpine shell? :p
2022-08-04 15:43:41 <panekj> ikke: why not?
2022-08-04 15:43:44 <skarnet> minimal alpine shell
2022-08-04 15:43:49 <ptrc> works too
2022-08-04 15:44:07 <skarnet> but the idea is that it’s a _standard_, not an implementation
2022-08-04 15:44:10 <psykose> but yeah i agree
2022-08-04 15:47:12 <psykose> is there a posix changelog somewhere
2022-08-04 15:47:16 <psykose> for what actually changed between versions
2022-08-04 15:48:15 <ptrc> if you can find older versions, you can just download a tarball and diff them
2022-08-04 15:48:40 <ptrc> but finding that stuff is nearly impossible
2022-08-04 15:50:55 <psykose> i am not diffing a 4k page document
2022-08-04 15:51:26 <psykose> that's why i asked :p maybe somebody did it and described the actual changes
2022-08-04 15:53:16 <ikke> ftr, doing some maintenance / deployment on pkgs.a.o, so it might be unavailable for a moment
2022-08-04 15:53:45 <orbea> ptrc: some of the pages has a CHANGE HISTORY section, but I dont see anywhere its all collected into one place
2022-08-04 15:54:06 <orbea> example: https://pubs.opengroup.org/onlinepubs/9699919799/functions/_Exit.html#tag_16_01_11
2022-08-04 15:54:30 <ptrc> yeah and the changes are like
2022-08-04 15:54:33 <ptrc> "Austin Group Interpretation 1003.1-2001 #031 is applied"
2022-08-04 15:54:48 <ptrc> (thanks algitbot, always so helpful)
2022-08-04 15:55:22 <psykose> my floppy :(
2022-08-04 16:08:32 <ikke> pkgs.a.o is back
2022-08-04 16:08:44 <ikke> let me know if you notice issues
2022-08-04 20:08:47 <Thermi> Hi, https://gitlab.alpinelinux.org/alpine/apk-tools/-/issues/10852 is laying around for a month now with now activity. It's a bug in apk-tools that needs fixing so packages that are actually not conflicting can be installed.
2022-08-04 20:12:27 <psykose> it's definitely a real issue, but out of curiosity what were you using ucspi-tcp6 for
2022-08-04 20:13:05 <Thermi> needed to satisfy dependency
2022-08-04 20:14:50 <psykose> nothing depends on it though
2022-08-04 20:15:05 <psykose> well, <=3.14 tinysshd did
2022-08-04 20:15:19 <Thermi> I need to take a look, don't remember it from the top of my head
2022-08-04 20:15:44 <Thermi> I think it was tinysshd though
2022-08-04 20:16:08 <psykose> sadly community before 3.16 is technically unsupported
2022-08-04 20:16:18 <psykose> but if you are actually on 3.14 for some reason i can fix it by changing the dependency
2022-08-04 20:21:26 <Thermi> How would tcpserver be provided?
2022-08-04 20:22:04 <Thermi> nvm
2022-08-04 20:22:07 <Thermi> found the commit that changes it
2022-08-04 20:53:24 <skarnet> can port it to s6-networking, if it hasn’t been done yet
2022-08-04 20:53:43 <skarnet> there’s an ipv6 superserver that actually works :P
2022-08-04 20:53:49 <skarnet> (and is maintained)
2022-08-04 21:05:09 <psykose> you already did
2022-08-04 21:05:22 <psykose> 909942755f24aee41aaee0bff81d0df62c0ead00
2022-08-04 21:06:28 <psykose> it was backported into 3.15 at the time
2022-08-04 21:06:40 <psykose> (i think(
2022-08-04 21:12:14 <skarnet> ah, so that was tinyssh
2022-08-04 21:12:28 <skarnet> I knew I had ported *something* away from ucspi-tcp6, forgot what it was
2022-08-04 22:57:15 <umk3> hello, what is the requirement to recommend a pkg upgrade from edge to latest-stable?
2022-08-04 22:58:09 <umk3> for example, the latest waybar 0.9.13 package has fixed a problem with xwayland that makes compositors crash 
2022-08-04 22:58:19 <umk3> and latest-stable version is 0.9.12
2022-08-04 22:59:42 <sam_> psykose: so you yeeted libexecinfo, how much stuff needed patching?
2022-08-04 22:59:44 <sam_> psykose: btw, another kill idea: IM6, maybe sdl1?
2022-08-04 22:59:56 <psykose> sdl1 is dead except for one single thing
2022-08-04 23:00:01 <psykose> netsurf-fb does not work with it
2022-08-04 23:00:19 <psykose> sdl12-compat shims to sdl2 fine, but sdl2 does not support directfb
2022-08-04 23:00:21 <sam_> does it not work with the weird sdl2 shi-
2022-08-04 23:00:22 <psykose> and without directfb it fails
2022-08-04 23:00:39 <psykose> the equivalent new one (drm? videodriver) uhh fails
2022-08-04 23:00:46 <psykose> it actually just hangs the entire tty if you try
2022-08-04 23:00:47 <psykose> or something
2022-08-04 23:00:54 <psykose> as for patching, hmmm
2022-08-04 23:01:30 <psykose> uhh just grep the last 2 weeks for `execinfo` in aports
2022-08-04 23:01:33 <psykose> all the commits that show up
2022-08-04 23:01:40 <psykose> not that much
2022-08-04 23:01:44 <psykose> some of them are just removing the declaration
2022-08-04 23:02:09 <psykose> most were already patched, but kept it because of an #include (but the backtrace() was patched)
2022-08-04 23:02:16 <psykose> it was like ~30 things, maybe
2022-08-04 23:02:31 <sam_> ohhhh
2022-08-04 23:02:31 <psykose> now, im6..
2022-08-04 23:02:37 <psykose> the only thing that uses it is uh
2022-08-04 23:02:40 <sam_> i looked at some and i was wondering why so few needed patching but now i get it
2022-08-04 23:03:25 <psykose> nothing
2022-08-04 23:03:30 <psykose> ruby-rmagick swapped to 7
2022-08-04 23:03:37 <psykose> inkscape swapped to graphicsmagick
2022-08-04 23:03:40 <psykose> can be gone
2022-08-04 23:04:05 <psykose> (we have less stuff than gentoo though, what kind of things use 6 there?)
2022-08-04 23:05:32 <psykose> umk3: you can open a merge request if you want
2022-08-04 23:08:04 <umk3> i've never done that
2022-08-04 23:19:57 <psykose> i can try backport it i suppose
2022-08-04 23:20:10 <psykose> sam_: a gitlab employee just emailed me directly because of something i removed, lmao
2022-08-04 23:21:40 <sam_> LOL
2022-08-04 23:21:42 <sam_> psykose: lemme look, iirc it was literally 2 things i'd never heard of, and we managed to get both onto GM
2022-08-04 23:21:46 <sam_> app-misc/graphlcd-base and media-radio/xastir
2022-08-04 23:22:37 <psykose> aha
2022-08-04 23:23:03 <psykose> our xastir uses neither (?) don't have the other
2022-08-04 23:47:54 <Ariadne> psykose: tell the gitlab employee that our new policy is to delete packages that gitlab wants to save money
2022-08-04 23:48:11 <psykose> lol do you want to know what they were doing
2022-08-04 23:48:25 <Ariadne> no, probably not, i have been debugging woodpecker for the past 3 hours
2022-08-04 23:48:29 <psykose> it's a 'security code scanning' container
2022-08-04 23:48:35 <psykose> and they apk add from testing mono for nugety
2022-08-04 23:48:37 <psykose> nuget*
2022-08-04 23:48:38 <psykose> hahaha
2022-08-04 23:48:57 <Ariadne> but we have real .net now
2022-08-04 23:49:16 <psykose> we do, though idk what any of this does
2022-08-04 23:49:29 <Ariadne> i looked at gitlab CI blahblah stuff
2022-08-04 23:49:32 <Ariadne> for work
2022-08-04 23:49:34 <Ariadne> and it was like
2022-08-04 23:49:37 <Ariadne> staring into an abyss
2022-08-04 23:49:41 <Ariadne> of shit code
2022-08-04 23:49:42 <psykose> https://gitlab.com/gitlab-org/security-products/analyzers/security-code-scan/-/blob/master/Dockerfile
2022-08-04 23:50:37 <Ariadne> well
2022-08-04 23:50:40 <Ariadne> that seems very not great
2022-08-04 23:56:25 <psykose> Ariadne: is there any ci that doesn't feel like that, haha
2022-08-04 23:56:39 <Ariadne> tekton is pretty reasonable
2022-08-04 23:56:49 <Ariadne> the only downside is that it has kubernetes as a dependency
2022-08-04 23:57:23 <Ariadne> sourcehut's CI is also reasonable
2022-08-04 23:58:58 <psykose> woo i got a git short-sha of only numbers
2022-08-04 23:59:01 <psykose> lucky me
2022-08-04 23:59:20 <psykose> ah yeah, i remember sourcehut looking pretty good
2022-08-04 23:59:26 <psykose> wanted to look at selfhosting..
2022-08-04 23:59:46 <Ariadne> my main gripes with CI systems are that they love running workloads as root, fail to retain state between pipeline stages, and make running workloads as nonroot a pain in the ass
2022-08-05 00:01:22 <Ariadne> state persistence and nonroot are likely solveable in woodpecker, which would allow me to say it is acceptable
2022-08-05 00:03:04 <skarnet> don’t modern CI systems run their workloads in a container?
2022-08-05 00:03:30 <Ariadne> they do
2022-08-05 00:03:45 <Ariadne> but end users won't build the software as root
2022-08-05 00:03:56 <Ariadne> so it is good to validate that software is built as nonroot :)
2022-08-05 00:04:18 <skarnet> yeah
2022-08-05 00:04:25 <Ariadne> i don't think that running in a disposable container is an excuse to YOLO running as root
2022-08-05 00:05:13 <Ariadne> (and again, tekton does not ever run a workload as root, unless you explicitly tell it to do so)
2022-08-05 00:05:20 <skarnet> testing ‘make install’ or equivalent needs root though
2022-08-05 00:05:41 <Ariadne> yes/no
2022-08-05 00:05:48 <Ariadne> can test with `make install DESTDIR=...`
2022-08-05 00:06:06 <skarnet> on smart packages, yes
2022-08-05 00:06:07 <Ariadne> but even still, you could just run that step with root privileges
2022-08-05 00:06:38 <skarnet> that means your wrapper needs to understand the semantics of the build system
2022-08-05 00:06:53 <Ariadne> s/wrapper/config file
2022-08-05 00:07:01 <skarnet> true
2022-08-05 00:07:03 <Ariadne> and they all use config files :p
2022-08-05 00:07:44 <Ariadne> the problem with tekton is that you have to run it on kubernetes
2022-08-05 00:07:50 <Ariadne> otherwise it is an excellent CI system
2022-08-05 00:11:07 <Ariadne> but for example, in tekton, you can have a "log in to OCI registry" task, followed by building your stuff, followed by "upload your built OCI image to the registry" task, and it will actually preserve the state through the entire build, because the homedir is an ephemeral docker volume
2022-08-05 00:11:37 <Ariadne> woodpecker/drone promise the ability to do this, but it falls flat because the only preserved state is the source code workspace
2022-08-05 00:11:57 <skarnet> people can’t modelize
2022-08-05 00:12:21 <skarnet> and builds are a complex thing with lots of moving targets
2022-08-05 00:12:25 <skarnet> so they’re hard to modelize
2022-08-05 00:12:52 <Ariadne> yeah, but i mean, i feel like woodpecker could improve significantly by preserving the homedir across build stages :P
2022-08-05 00:13:04 <Ariadne> it would make their setup far more composable
2022-08-05 00:13:25 <Ariadne> this is one of the huge things github actions has going for it, like tekton, they are very good at preserving state
2022-08-05 00:13:41 <Ariadne> so you can stitch building blocks together
2022-08-05 00:13:46 <Ariadne> and, it actually works
2022-08-05 00:14:10 <skarnet> well yeah it should be a staple of any forge
2022-08-05 00:15:06 <psykose> maybe that posix sh guy should write one in posix sh
2022-08-05 00:15:25 <Ariadne> which posix sh guy
2022-08-05 00:15:31 <psykose> yes
2022-08-05 00:15:58 <Ariadne> woodpecker is even worse, it will let you choose an OCI image that is nonroot
2022-08-05 00:16:00 <Ariadne> and it will run it
2022-08-05 00:16:07 <Ariadne> and it will even run it nonroot
2022-08-05 00:16:18 <Ariadne> but then the HOME env variable is hardcoded to /root
2022-08-05 00:16:25 <Ariadne> so, you are screwed :)
2022-08-05 00:17:14 <minimal> Ariadne: are you intending to package woodpecker for Alpine? I started work on it many months ago but it got sidelined
2022-08-05 00:17:28 <Ariadne> minimal: no, i deploy it on kubernetes
2022-08-05 09:50:05 <HRio> anyone else that have VM crashes (just hangs during was seems like io) with the kernel in edge right now? downgraded to the kernel from 3.15 and no problems (running edge in the VM)?
2022-08-05 09:57:18 <psykose> interesting
2022-08-05 10:24:03 <ikke> HRio: No, haven't noticed any freezes
2022-08-05 10:42:08 <HRio> host is running xen with an AMD CPU, tried to upgrade the XEN version and host kernel as well, but only downgrading the VM kernel helped
2022-08-05 10:42:50 <psykose> could be related to the zen2 mitigations
2022-08-05 10:57:38 <ikke> Talking about makefiles breaking DESTDIR semantics
2022-08-05 10:58:39 <ikke> I'm running exactly into the issue where building fails due to running CI without root and a Makefile touching things outside of DESTDIR
2022-08-05 10:59:13 <psykose> fun times
2022-08-05 10:59:20 <donoban> ouch, there is a new abuild-rootbld release Upgrading abuild-rootbld (3.9.0-r3 -> 3.9.0-r4)
2022-08-05 10:59:32 <donoban> but abuild still misses some merged commits
2022-08-05 10:59:45 <psykose> not a release, i just backported a shim fix into it
2022-08-05 10:59:58 <psykose> yes, i have had to patch abuild with the same changes like .. 14 times so far? 
2022-08-05 10:59:59 <psykose> :)
2022-08-05 11:00:21 <donoban> :(
2022-08-05 11:00:40 <donoban> https://gitlab.alpinelinux.org/alpine/abuild/-/commit/d9f03b387a02541f9d60384b6447a141765d342f
2022-08-05 11:00:49 <psykose> oh yeah
2022-08-05 11:00:50 <psykose> my favorite
2022-08-05 11:00:56 <psykose> i'm tempted to just go throw it in
2022-08-05 11:01:02 <ncopa> i guess i should tag a new release
2022-08-05 11:01:07 <ncopa> abuild release
2022-08-05 11:01:16 <psykose> well yeah
2022-08-05 11:01:21 <psykose> i did ask, but
2022-08-05 11:01:24 <donoban> great!
2022-08-05 11:01:31 <psykose> better late than never :)
2022-08-05 11:01:43 <psykose> ikke would have to go also update abuild.conf on builders for the changes
2022-08-05 11:01:50 <psykose> and you would have to drop the commit i added into aports
2022-08-05 11:01:55 <psykose> rest is alright
2022-08-05 11:01:57 <psykose> fine by me
2022-08-05 11:01:58 <ncopa> there are only 32 merge requests to review first...
2022-08-05 11:02:02 <psykose> you don't have to
2022-08-05 11:02:23 <psykose> i mean sure, go merge mine if you want, i even wrote a test for you, haha
2022-08-05 11:02:28 <donoban> can I try to review/check them?
2022-08-05 11:02:45 <psykose> but what i mean by don't have to is there are already plenty of fixes
2022-08-05 11:02:50 <psykose> and there's no limit on releases
2022-08-05 14:51:33 <ncopa> gnome people: can you please have a look at my suggestion of adding an automatic $pkgname-devhelp subpackage?
2022-08-05 14:51:38 <ncopa> https://gitlab.alpinelinux.org/alpine/abuild/-/merge_requests/87#note_253818
2022-08-05 14:53:23 <psykose> it makes sense (you also need the check for dir to warn it exists like the other ones warn)
2022-08-05 14:53:28 <psykose> there's a similar thing for apparmor.d
2022-08-05 14:53:36 <psykose> i guess ultimately they are harmless
2022-08-05 16:59:35 <handlerug> how does this package depend on itself? https://gitlab.alpinelinux.org/alpine/aports/-/blob/master/main/icu/APKBUILD#L11
2022-08-05 17:00:20 <psykose> depends_dev is used to set depends= inside default_dev, on the -dev subpackage
2022-08-05 17:00:27 <psykose> so what that means is, icu-dev depends on icu
2022-08-05 17:00:29 <handlerug> ah yeah indeed
2022-08-05 17:01:00 <psykose> that's commonly done when for some reason it's not traced; usually this is automatic because some library is in 'pkg' and the symlinks inside 'pkg-dev' depend on them, so both get in anyway
2022-08-05 17:01:12 <handlerug> kind of weird because _dev things are a special case so that you won't overwrite the default dev function I guess
2022-08-05 17:01:13 <psykose> it's generally intended to pull in everything on -dev so you see these forms around
2022-08-05 17:01:25 <psykose> yeah, it's just shorthand for adding depends= in dev
2022-08-05 17:02:03 <psykose> alternatively: dev() { default_dev; depends=".."; }
2022-08-05 17:02:15 <psykose> but that is literally the default more or less, that's all it does for depends= if you look at abuild
2022-08-05 17:07:06 <psykose> ikke: since you updates pkgs.a.o the actual required by/depends thing doesn't really seem to update. like for x86_64 it's said the same thing for 12+ hours
2022-08-05 17:07:28 <ikke> psykose: Is it updating at all?
2022-08-05 17:07:36 <psykose> it.. did once or twice before?
2022-08-05 17:07:40 <psykose> but now not in half a day +
2022-08-05 17:07:53 <psykose> same for other arches
2022-08-05 17:08:00 <psykose> it did for the first openssl3 batch and now stuck
2022-08-05 17:08:11 <psykose> and yes i double checked the real depends
2022-08-05 17:14:05 <ikke> The cron script was not executable
2022-08-05 17:14:07 <ikke> Didn'
2022-08-05 17:14:11 <ikke> Didn't touch that
2022-08-05 17:15:40 <psykose> must've been the cats again
2022-08-05 17:16:34 <ikke> I most likely manually fixed it on the host, and that got undone again once I updated the repo
2022-08-05 17:17:46 <psykose> mhm
2022-08-05 17:17:56 <ikke> Yup: from the stash I used to clean manual changes: https://tpaste.us/qaX4 
2022-08-05 17:21:03 <psykose> ones greatest enemy is oneself
2022-08-05 17:22:43 <ikke> It least I know made sure everything is properly committed
2022-08-05 17:22:48 <ikke> s/It/At/
2022-08-05 17:22:48 <alpine-meetbot> ikke meant to say: At least I know made sure everything is properly committed
2022-08-05 17:22:56 <ikke> s/know/now/
2022-08-05 17:22:56 <alpine-meetbot> ikke meant to say: At least I now made sure everything is properly committed
2022-08-05 17:23:33 <psykose> when are you gonna let me manage some of this stuff so you can blame me sometimes too :p
2022-08-05 17:24:04 <ikke> good question
2022-08-05 17:24:50 <psykose> also, private merge requests still hide the ci
2022-08-05 17:24:59 <psykose> something something i forgot (again, sorry) some option in new gitlab
2022-08-05 17:25:03 <psykose> was it fixed or nah
2022-08-05 17:26:06 <ikke> What, as far as I know, was fixed was that you could not rebase merge requests for internal projects
2022-08-05 17:26:30 <psykose> weh
2022-08-05 17:26:59 <psykose> i forget if it was possible to make it impossible to private-fork aports only
2022-08-05 17:28:36 <ikke> I don't think you can control what forks can do from the origin
2022-08-05 17:28:47 <ikke> You can do it globally, but not sure if we want to go that far
2022-08-05 17:29:14 <ikke> Perhaps it makes sense, not sure
2022-08-05 17:29:39 <psykose> yeah, for global i know it's possible
2022-08-05 17:29:45 <psykose> we did discuss it about 4 months ago, i think
2022-08-05 17:30:04 <psykose> overall i don't see a reason not to, if the policy for our gitlab is for working on alpine (it is)
2022-08-05 17:30:08 <psykose> there's nothing to hide there
2022-08-05 17:30:33 <ikke> Not permanently, but perhaps temporarily
2022-08-05 17:30:42 <handlerug> is this supposed to make icu effectively an alias to icu-libs? https://gitlab.alpinelinux.org/alpine/aports/-/blob/master/main/icu/APKBUILD#L106
2022-08-05 17:31:06 <psykose> no, it allows the -libs subpackage to overwrite files of `icu` on an upgrade
2022-08-05 17:31:14 <handlerug> ah, hm
2022-08-05 17:31:16 <psykose> historically it probably didn't exist
2022-08-05 17:31:20 <psykose> then, it was split
2022-08-05 17:31:24 <psykose> so it's possible on upgrade you get
2022-08-05 17:31:31 <psykose> installing icu-libs
2022-08-05 17:31:32 <psykose> ...
2022-08-05 17:31:40 <psykose> purging icu # rest of stuff not needed
2022-08-05 17:31:43 <psykose> but in that middle step
2022-08-05 17:31:49 <psykose> it would be overwriting the same 'owned' files
2022-08-05 17:31:54 <psykose> that's forbidden; you get an error
2022-08-05 17:31:59 <psykose> replaces allows that override
2022-08-05 17:32:13 <handlerug> I couldn't reproduce the bug with roughly the same packages but without replaces and apk worked it all out so that's why I am diggign
2022-08-05 17:32:17 <handlerug> but this seems to be unrelated
2022-08-05 17:32:23 <handlerug> digging*
2022-08-05 17:32:28 <psykose> no, definitely unrelated to these issues
2022-08-05 17:32:42 <handlerug> thanks for the explanation
2022-08-05 17:32:44 <psykose> provides=, and the metapackage-level stuff is where stuff would go wrong
2022-08-05 17:32:51 <psykose> replaces does not affect dependency resolution
2022-08-05 17:32:54 <psykose> (85% sure)
2022-08-05 17:32:59 <psykose> it's just file ownership
2022-08-05 17:33:12 <psykose> (feel free to fact check me via reading apk code if you wish, but it is C and takes a while)
2022-08-05 17:34:01 <ikke> psykose: seems pkgs.a.o is up-to-date again?
2022-08-05 17:34:27 <psykose> ikke: success
2022-08-05 17:38:55 <psykose> pound (not updated for like 5 years upstream), racket/rakudo (dlopen), php7(nosup, will drop), dotnet31(almost eol), openvpn3 (smol patch), openvpn (next release), odyssey(doesn't build anyway, probably better to drop), nyxt (dlopen), nimble (nim not support 3), kopano (not supported, php7), libneo4j (unupdated 5 years), manticoresearch (looking)
2022-08-05 17:39:03 <psykose> and that's everything left for openssl3 that directly specifies
2022-08-05 19:17:10 <eddsalkield> how do you specify dependencies for a custom-named subpackage?
2022-08-05 19:17:36 <eddsalkield> i'm building a package with a -tools subpackage, and these tools have some deps that the main package doesn't need
2022-08-05 19:18:27 <ikke> eddsalkield: you can set it in the split function
2022-08-05 19:18:35 <ikke> depends="..."
2022-08-05 19:18:49 <ikke> so in the tools() function in your case
2022-08-05 19:19:47 <psykose> and you probably want the same things in makedepends= too so the build order is correct
2022-08-05 19:22:05 <eddsalkield> since it's a python package these extra deps aren't needed when making, but this is useful to know - thanks
2022-08-05 19:23:37 <psykose> yes, but subdeps are invisible to the build order
2022-08-05 19:24:15 <ikke> psykose: but if it's not a build dependency, that should not matter
2022-08-05 19:24:29 <ikke> It should not matter if A is built before B or after B
2022-08-05 19:24:59 <psykose> it does not matter at the time it is added, yes
2022-08-05 19:25:21 <psykose> (because nothing can depend on it yet)
2022-08-05 19:39:54 <Guest7197> what is a unionfs? apparently i can set its size in the custom iso
2022-08-05 19:40:05 <Guest7197> and  i think i might need to set it, because adding more packages is breaking the system
2022-08-05 20:26:39 <Guest7197> i passed firefox to the $apks variable in mkimage and to /etc/apk/world, yet its not installed
2022-08-05 20:26:58 <Guest7197> also, the apk seems to be in the iso, just not installed. if i run apk add firefox, it installs
2022-08-05 20:27:14 <Guest7197> might have something to do with this 'unionfs' thing
2022-08-06 00:14:15 <Guest7208> increasing the 'unionfs' size fixed it
2022-08-06 01:10:15 <user2> does ash have something like .ashrc? where it runs everytimme the shell opens, instead of on login?
2022-08-06 01:12:27 <user2> can't set PS1 when it doesnt always run .profile
2022-08-06 09:27:14 <psykose> ikke: is there some easy way to see packages that haven't been rebuilt for like 2 years
2022-08-06 10:07:09 <handlerug> what a beautiful error message https://paste.sr.ht/~handlerug/70dfd3d0c2f8936f8d0d5f89c9048b4194068e22
2022-08-06 10:07:17 <handlerug> "i cant select this package :( idk why"
2022-08-06 10:07:35 <handlerug> (the package names are fake, they're from my own repositories)
2022-08-06 10:08:01 <psykose> what was the apkbuild for that
2022-08-06 10:10:20 <handlerug> here are the APKBUILDs in a tar file https://a.uguu.se/lzTvLXgM
2022-08-06 10:10:47 <handlerug> and the commands I ran https://paste.sr.ht/~handlerug/de085c52c4996eb9b3c3abad37f1dbb8ca33b1ec
2022-08-06 10:11:33 <handlerug> I wanted to test it on a different box, hence the 10.211.xx.xx IPs
2022-08-06 10:12:03 <psykose> that provides of h-unrelated has no version or provider_priority
2022-08-06 10:12:35 <psykose> it's technically accepted but breaks some things, the reason apk-tools was downgraded from .10 to .9 was because there is a commit in .10 that forbids that entirely
2022-08-06 10:12:46 <psykose> (and we hadn't fixed aports enough yet)
2022-08-06 10:12:54 <handlerug> https://gitlab.alpinelinux.org/alpine/aports/-/blob/master/main/postgresql14/APKBUILD#L44
2022-08-06 10:12:55 <psykose> so technically, that's an illegal apkbuild (assume it is for now)
2022-08-06 10:13:06 <psykose> that has a priority
2022-08-06 10:13:07 <psykose> line above
2022-08-06 10:13:12 <handlerug> ah, right
2022-08-06 10:13:22 <handlerug> I assumed because h-unrelated14 is the only provider it'd get picked anyway
2022-08-06 10:13:56 <psykose> maybe, i forget the semantics
2022-08-06 10:14:06 <psykose> but regardless these should have never been allowed, so pretend they're not
2022-08-06 10:14:10 <psykose> (and if you find any in aports, tell me)
2022-08-06 10:14:13 <handlerug> got it, thanks
2022-08-06 10:14:13 <psykose> i fixed 95% of them
2022-08-06 10:14:57 <psykose> i think now would be a good time to go back to .10 if anything, but the issue was the errors you get is an 'oops cant resolve constraints' on an upgrade if anything was invalid, with no more info
2022-08-06 10:15:02 <psykose> so it's hard to debug what even causes them
2022-08-06 10:15:20 <handlerug> yeah, not very descriptive
2022-08-06 10:15:40 <handlerug> that being said, the icu-data-{en,full} thingies do provides stuff too
2022-08-06 10:15:45 <handlerug> interesting
2022-08-06 10:16:19 <psykose> they seem to do it correctly
2022-08-06 10:16:27 <ikke> psykose: these are still left: https://tpaste.us/QNkO
2022-08-06 10:16:32 <psykose> the only thing that 'jumps out' at me, is that they have both a version and, and a priority
2022-08-06 10:16:43 <psykose> now, the versions will always match (they're the same), but..
2022-08-06 10:16:47 <psykose> i've never seen that anywhere else
2022-08-06 10:17:25 <psykose> usually the case is that things have only a version, or only a priority, *or* they have both, but the versions aren't the same (so the larger is picked, and the priority doesn't matter at all)
2022-08-06 10:17:34 <psykose> in this specific case.. i haven't seen it elsewhere
2022-08-06 10:18:15 <psykose> and the first thing i would try if you can reproduce the same scenario, is to remove the =$pkgver from both, and only that, and see if it fixes the scenario
2022-08-06 10:18:48 <psykose> and if so, i guess then you can inquire on apk-tools what the behaviour is meant to be, but if that's the sole cause of all this, it seems a bit buggy
2022-08-06 10:18:59 <psykose> (for once! not abuild)
2022-08-06 10:19:08 <psykose> but i'm just speculating :)
2022-08-06 10:19:42 <psykose> ikke: aha, okay, i'll fix them up
2022-08-06 10:20:21 <psykose> what did we want to do for lua?
2022-08-06 10:20:39 <psykose> could do priority= of the majorver, but.. what does luajit get?
2022-08-06 10:21:12 <psykose> (also thanks a lot for that list)
2022-08-06 10:21:22 <ikke> thanks to apkgquery ;-)
2022-08-06 10:21:28 <psykose> i should go install that
2022-08-06 10:21:35 <psykose> can you show me the commandline you typed
2022-08-06 10:21:43 <ikke> https://gitlab.alpinelinux.org/kdaudt/apkgquery/-/blob/master/README.md#user-content-provides-virtual-package-without-setting-a-provider_priority
2022-08-06 10:22:08 <psykose> thanks!
2022-08-06 10:22:35 <psykose> also as much as we like to complain about packaging, being able to `go install gitlab..@latest` or `cargo install ..` random stuff is quite fast
2022-08-06 10:23:17 <handlerug> it's kinda funny abuild fails to sign the packages properly on this VM if I run it in quick succession
2022-08-06 10:23:23 <handlerug> perhaps a problem with entropy
2022-08-06 10:23:29 <psykose> what error?
2022-08-06 10:23:50 <psykose> i assume you just run `abuild` and near the end you get a failure to sign, but on rerun touching nothing else it works?
2022-08-06 10:23:54 <handlerug> apk starts printing "ERROR: h-unrelated14-dev-1.0-r1: BAD signature" for random number of packages
2022-08-06 10:24:02 <psykose> hmm
2022-08-06 10:24:06 <psykose> that doesn't seem related to entropy
2022-08-06 10:24:10 <handlerug> after a while it just started printing that for every package
2022-08-06 10:24:59 <psykose> you did abuild-keygen correctly at first right
2022-08-06 10:25:13 <handlerug> what does correctly mean
2022-08-06 10:25:33 <psykose> abuild-keygen -ai
2022-08-06 10:25:39 <handlerug> well yeah
2022-08-06 10:25:46 <psykose> that's very strange then
2022-08-06 10:26:40 <handlerug> I waited for a while and rebuilt all packages again, without sleeping. now it doesn't print any errors. so it's truly random
2022-08-06 10:26:46 <handlerug> very strange
2022-08-06 10:27:26 <psykose> maybe you should go to sleep then? i think that changes one of the factors there, maybe it fails again
2022-08-06 10:27:27 <psykose> :3
2022-08-06 10:36:13 <ikke> psykose: not sure what we want for lua
2022-08-06 10:38:22 <psykose> so far from what i've seen it's latest anyway
2022-08-06 10:38:29 <psykose> and things specifically pick versions to pull
2022-08-06 10:38:32 <ikke> right
2022-08-06 10:38:42 <psykose> and anything on main/lua5.3 won't affect any install_if rules, etc
2022-08-06 10:38:46 <ikke> so I suppose luajit would have a low priority
2022-08-06 10:38:52 <psykose> so.. i guess i'll just add priority
2022-08-06 10:38:57 <psykose> but yeah what does luajit get
2022-08-06 10:39:06 <ikke> 0?
2022-08-06 10:39:10 <psykose> it's more or less equivalent to 5.1
2022-08-06 10:39:13 <ikke> yeah
2022-08-06 10:39:17 <psykose> i.e. it uses the 5.1 modules
2022-08-06 10:39:23 <ikke> but afaik, only the higher priority really matters, right/
2022-08-06 10:39:30 <ikke> highest*
2022-08-06 10:39:33 <psykose> yeah
2022-08-06 10:40:06 <psykose> also 5.1 already has '100'
2022-08-06 10:40:06 <psykose> lol
2022-08-06 10:43:41 <psykose> there's also the whole pile of sxmo weirdness
2022-08-06 10:44:09 <psykose> that i would rather remove from aports but what can i do
2022-08-06 10:44:20 <psykose> so, i guess the intention is to keep the same allowed override
2022-08-06 10:44:47 <psykose> so.. idk. they provide something that could upgrade at any time
2022-08-06 10:44:56 <ikke> Don't use a version
2022-08-06 10:45:08 <psykose> if sway becomes 1.8 while sxmo-sway provides sway, then it would sway anyway if 'sway' was in world
2022-08-06 10:45:19 <psykose> this never was going to work as intended, and the provides= at all is meaningless
2022-08-06 10:45:23 <psykose> you have to add the one you want specifically
2022-08-06 10:45:32 <ikke> Yeah, you cannot mix virtual and concrete packages
2022-08-06 10:45:56 <psykose> so.. any changes here that fix this are 'breaking' for whoever was using this because making it correct might actually suddenly swap them for whatever reason
2022-08-06 10:45:59 <psykose> (guessing)
2022-08-06 10:46:13 <psykose> egh, guess i'll create sway-virtual and the like
2022-08-06 10:46:25 <handlerug> "ERROR: h-deep-dev-1.0-r1: BAD signature" but "h-deep-dev-1.0-r1.apk: 0 - OK"
2022-08-06 10:46:28 <handlerug> okay I guess
2022-08-06 10:46:46 <ikke> I guess the idea was that things that have sway as dependency would also be satisfied by sxmo-sway
2022-08-06 10:47:05 <psykose> aha, right
2022-08-06 10:47:07 <psykose> forgot about that
2022-08-06 10:47:24 <psykose> ok, i'll check revs too
2022-08-06 10:47:30 <psykose> update accordingly to depends=sway-virtual
2022-08-06 10:47:43 <psykose> (just one there, but there's more sxmo packages)
2022-08-06 11:43:02 <psykose> ikke: your tool scans unmaintained/ too :p
2022-08-06 11:43:21 <ikke> psykose: sure :)
2022-08-06 11:43:37 <ikke> It just scans everything
2022-08-06 11:45:18 <psykose> my computer has been at 40 load for the past 10 minutes and i didn't even notice
2022-08-06 11:45:33 <psykose> scheduling do be good these days
2022-08-06 11:45:56 <nmeum> > ERROR: openssl1.1-compat-dev-1.1.1q-r2: trying to overwrite usr/lib/libcrypto.so owned by openssl-dev-3.0.5-r1.
2022-08-06 11:46:03 <psykose> yeah, that's normal
2022-08-06 11:46:04 <nmeum> I think openssl1.1-compat-dev and openssl-dev should conflict
2022-08-06 11:46:07 <psykose> but what pulls it
2022-08-06 11:46:16 <psykose> sure, i can make it explicit
2022-08-06 11:46:22 <psykose> depends_dev=!openssl-dev ?
2022-08-06 11:46:26 <nmeum> some makedependency pulls it
2022-08-06 11:46:36 <psykose> iirc only one thing in testing should be broken like this
2022-08-06 11:46:59 <psykose> (because basically everything is openssl3 now except very few things)
2022-08-06 11:47:07 <psykose> so nothing should pull 1.1-dev at all
2022-08-06 11:47:35 <nmeum> > World updated, but the following packages are not removed due to:
2022-08-06 11:47:37 <nmeum> >   openssl1.1-compat-dev: .makedepends-libgit2
2022-08-06 11:48:18 <panekj> git pull?
2022-08-06 11:48:19 <psykose> you have an old makedepends for it
2022-08-06 11:48:22 <nmeum> ah, that's probably just an issue with the virtual package
2022-08-06 11:48:24 <nmeum> yes, exactly
2022-08-06 11:48:30 <psykose> anyway, fixed
2022-08-06 11:48:35 <psykose> (for being explicit)
2022-08-06 11:48:59 <nmeum> that's all I was asking for, thanks :)
2022-08-06 11:52:33 <ikke> apk del '.*';
2022-08-06 11:52:37 <psykose> ikke: fixed them all
2022-08-06 11:52:41 <ikke> psykose: thanks, saw
2022-08-06 11:52:47 <ikke> including top-level exec :)
2022-08-06 11:52:48 <psykose> i think .10 should be safe now
2022-08-06 11:52:52 <psykose> yeah, did that by accident
2022-08-06 11:52:53 <psykose> haha
2022-08-06 11:52:57 <psykose> that tool is helpful
2022-08-06 11:53:04 <psykose> nice to see we are top-level exec clean
2022-08-06 11:53:40 <psykose> iirc the issues for people with apk .10 was only this? so now that they are all gone, it shouldn't cause the same breakage
2022-08-06 11:53:44 <psykose> i think we could go back
2022-08-06 14:00:35 <nmeum> is it just me or does gitlab automatically assign created merge requests to the MR author (i.e. myself) nowadays?
2022-08-06 14:00:59 <panekj> it does that now
2022-08-06 14:01:08 <nmeum> unfourtunate, it's a bit annoying as algibot does not automatically assign the MR to the package maintainer then
2022-08-06 14:01:35 <ikke> That change will be reverted
2022-08-06 14:02:06 <nmeum> nice! ty
2022-08-06 14:02:06 <panekj> gitlab does quite questionable things
2022-08-06 14:03:06 <psykose> (reverted by gitlab itself, next release)
2022-08-06 14:03:19 <psykose> nmeum: you can also unselect yourself
2022-08-06 14:03:22 <psykose> in the create screen
2022-08-06 14:03:28 <psykose> that's the only default afaik
2022-08-06 14:03:31 <psykose> just preselected
2022-08-06 14:03:37 <psykose> if you have the ability to assign
2022-08-06 14:03:38 <ikke> correct
2022-08-06 14:03:39 <nmeum> I know, but I always forget to do so :D
2022-08-06 14:03:42 <psykose> :)
2022-08-06 14:04:20 <nmeum> and if you unassign yourself after the MR was created then algibot does not revisit the MR
2022-08-06 14:04:28 <psykose> indeed
2022-08-06 14:10:05 <psykose> nmeum: for packages like that specifically i generally don't pay much attention since the maintainer hadn't done anything in years
2022-08-06 14:10:28 <psykose> also i had run into the same issue :D
2022-08-06 14:10:43 <psykose> forgot to fix it back then
2022-08-06 14:10:59 <psykose> i'm not sure if it's a cabal requirement, or a ghc requirement
2022-08-06 14:11:10 <nmeum> for cabal? yea, I know he isn't active anymore I was just reminded of the MR-assignment issue and wanted to complain ;)
2022-08-06 14:11:15 <psykose> :)
2022-08-06 14:11:32 <nmeum> oh, good point let me try with ghc
2022-08-06 14:11:55 <psykose> (i also don't know how to test with ghc, because you'd have to build a bunch of stuff manually and link it for the same kind of level)
2022-08-06 14:11:56 <nmeum> right, it's a ghc issue
2022-08-06 14:12:00 <psykose> ah, neat
2022-08-06 14:12:07 <nmeum> I will just add it to the $depends of ghc
2022-08-06 14:12:16 <nmeum> so I don't need to install libffi-dev manually everywhere
2022-08-06 14:12:22 <psykose> speaking of ghc also needs an update
2022-08-06 14:12:31 <panekj> are you updating cabal?
2022-08-06 14:12:57 <nmeum> no, I am just using it for a small software project of mine
2022-08-06 14:13:05 <panekj> ;\
2022-08-06 14:13:09 <ikke> I'll probably ugprade gitlab to 15.1 tonight
2022-08-06 14:13:16 <nmeum> but both ghc and cabal need to be updated yes. I might look into that soonish but it's just always very painful :D
2022-08-06 14:13:26 <psykose> i looked at it months ago
2022-08-06 14:13:31 <panekj> I know, last built in 2021
2022-08-06 14:13:35 <psykose> (the patch release absorbs all the backported patches)
2022-08-06 14:13:39 <psykose> but it failed for some other reasons
2022-08-06 14:14:28 <psykose> (not to the latest version - there's no point to update to like 9.2, just like 9.0.2 at the time)
2022-08-06 14:15:39 <ikke> nmeum: https://gitlab.com/gitlab-org/gitlab/-/issues/365891
2022-08-06 14:16:44 <panekj> > for the time being.
2022-08-06 14:17:00 <panekj> why do they think it's a desired behaviour by default at all
2022-08-06 14:17:56 <ikke> " It's also not clear what role assignees are playing in the merge request and review process for all teams"
2022-08-06 14:18:32 <psykose> wish gitlab had a better indication of where commits are in branches/tags
2022-08-06 14:18:37 <ikke> revert is in 15.2 apparently
2022-08-06 14:18:40 <psykose> in github the message just clearly shows it
2022-08-06 14:18:55 <panekj> i find whole gitlab ui confusing
2022-08-06 14:19:12 <psykose> e.g. https://img.ayaya.dev/Bw7ppsDYH6sH.png
2022-08-06 14:19:29 <psykose> you can see which mr it came from, what branch it went to, and also what tags it's in (if there's more than one, you see them all)
2022-08-06 14:19:33 <psykose> gitlab shows... uhh fuck all i guess
2022-08-06 14:19:44 <psykose> there's a 'tags' thing that fails to ever detect anything
2022-08-06 14:21:16 <ikke> https://i.imgur.com/xW7qulI.png
2022-08-06 14:21:58 <ptrc> psykose: it usually works quite well though?
2022-08-06 14:22:26 <ikke> https://i.imgur.com/OeTeZmT.png
2022-08-06 14:22:40 <panekj> I think the problem of gitlab is that it overloads you with information? and GitHub shows it very concise way
2022-08-06 14:22:48 <psykose> did they actually finally fix that after forever
2022-08-06 14:23:11 <ptrc> not sure if it was more broken before
2022-08-06 14:23:12 <psykose> https://img.ayaya.dev/hrnZWMzjQdXK.png
2022-08-06 14:23:13 <psykose> wow they did
2022-08-06 14:23:24 <ptrc> it was always okay
2022-08-06 14:23:40 <ptrc> s/$/ for me/
2022-08-06 14:23:47 <psykose> maybe it was fixed at the same time i stopped getting signed out of gitlab every time too
2022-08-06 14:23:51 <psykose> /s
2022-08-06 14:24:14 <panekj> just have less broken pc, ez
2022-08-06 14:24:18 <ikke> The signout is anoying
2022-08-06 14:24:39 <panekj> I was never signed out of gitlab_a.o
2022-08-06 14:25:10 <ikke> I am after x amount of time when the browser is restarted
2022-08-06 14:26:01 <psykose> oh you too?
2022-08-06 14:26:12 <ikke> yes
2022-08-06 14:26:13 <psykose> for me it was at any time i restarted my computer or crashed firefox
2022-08-06 14:26:16 <psykose> and only alpine gitlab
2022-08-06 14:26:17 <psykose> nowhere eelse
2022-08-06 14:26:23 <ikke> hmm, gitlab.com as wel
2022-08-06 14:26:36 <psykose> and now it stopped doing that
2022-08-06 14:26:45 <psykose> and every time it happened, the sessions page still had the previous session in it
2022-08-06 14:26:53 <psykose> so i had like 120 sessions in there
2022-08-06 14:27:25 <ikke> in devtools?
2022-08-06 14:27:50 <psykose> gitlab -> preferences -> active sessions
2022-08-06 14:27:57 <psykose> https://gitlab.alpinelinux.org/-/profile/active_sessions
2022-08-06 14:28:20 <psykose> https://img.ayaya.dev/mJeboTPZVLye.png
2022-08-06 14:28:23 <psykose> over and over and over again
2022-08-06 14:28:58 <psykose> the current one says 'signed in today' but actually i wasn't prompted for anything for like 3 days
2022-08-06 14:29:04 <psykose> so i'm not sure what any of those mean really
2022-08-06 14:29:29 <ikke> close to 100 sessions in there for me
2022-08-06 14:30:13 <psykose> i'm sure they're actually stored somewhere, imagine that multiplied by all the people that use it :D
2022-08-06 14:30:17 <psykose> maybe something is misconfigured
2022-08-06 14:33:10 <nmeum> haha there is not even a button to revoke all sessions
2022-08-06 14:33:16 <nmeum> I also have several hundered active sessions it seems
2022-08-06 14:36:12 <panekj> is gitlab still a normal page or is it already one of those fancy SPA that loads in background
2022-08-06 14:37:14 <ptrc> it's a "normal page" in a sense that every click sends you a new html page
2022-08-06 14:37:23 <ptrc> ah, no
2022-08-06 14:37:57 <ptrc> sometimes it redraws the same page, like when going between directories in one git repo
2022-08-06 14:38:45 <panekj> yeah, changing open/closed/all issues seem to load via JS and changing major stuff like clicking logo or repo it reloads
2022-08-06 14:39:32 <ptrc> it seems very inconsistent in my experience
2022-08-06 14:39:40 <ptrc> but i guess it might make sense to the gitlab people
2022-08-06 14:39:52 <ikke> they are just constantly in flux
2022-08-06 14:40:01 <ikke> without taking time to make everything consistent
2022-08-06 14:40:41 <ikke> although: https://about.gitlab.com/releases/2022/07/22/gitlab-15-2-released/#merge-request-reports-redesign
2022-08-06 14:44:08 <ikke> https://about.gitlab.com/releases/2022/07/22/gitlab-15-2-released/#improved-and-faster-file-browsing-and-syntax-highlighting
2022-08-06 14:44:20 <psykose> i wouldn't mind a spa design if it actually worked
2022-08-06 14:44:33 <psykose> half the pages never reload properly even when you tab back and forth spa-style inside the same page
2022-08-06 14:44:38 <psykose> and when you refresh it fixes itself
2022-08-06 14:45:08 <psykose> also not sure what dictates the self-updating policy, if i keep an mr page open and look at it i will see new comments on it, but if i tab away and come back after 3 hours... it will never reload
2022-08-06 14:45:59 <psykose> what version are we on
2022-08-06 14:46:27 <ikke> 15.0.x
2022-08-06 14:46:33 <psykose> maybe it's time for an upgrade :)
2022-08-06 14:46:41 <psykose> (if you have the time)
2022-08-06 14:46:48 <ikke> yes, will do it tonight
2022-08-06 14:46:51 <psykose> sure sure
2022-08-06 14:47:06 <ikke> Needed to fix some build issues
2022-08-06 14:47:18 <psykose> if you need some help i'm free
2022-08-06 14:47:24 <ikke> No, I managed that yesterday
2022-08-06 14:47:29 <psykose> might even set up the 10 repositories for that again
2022-08-06 14:47:30 <ikke> that DESTDIR thing was one of them
2022-08-06 14:47:32 <psykose> haha
2022-08-06 14:47:53 <ikke> https://gitlab-test.alpinelinux.org
2022-08-06 14:48:00 <ikke> it's running there already
2022-08-06 14:48:34 <psykose> appears to work
2022-08-06 14:48:34 <ikke> (15.1 ftr)
2022-08-06 14:51:10 <ptrc> ooh, on the topic of gitlab, apparently there's a few fun redirect loops like https://gitlab.alpinelinux.org/alpine/aports.git?service=git-upload-pack
2022-08-06 14:51:41 <ikke> Why are you navigating there? :P
2022-08-06 14:52:12 <panekj> ptrc: probably for git client?
2022-08-06 14:52:14 <ptrc> i tried to list refs to complain how much is wasted on stale 'patches/' branches :p
2022-08-06 14:52:26 <ikke> heh
2022-08-06 14:52:57 <ikke> https://gitlab.alpinelinux.org/alpine/aports/-/branches?state=all&sort=updated_desc&search=patches
2022-08-06 14:52:59 <ikke> here you go
2022-08-06 14:53:59 <ptrc> i meant it more like "here's how much data your client needs to pull with every `git fetch` that's stale and not needed"
2022-08-06 14:55:44 <ikke> 223 refs compared to ~400k in total
2022-08-06 14:55:54 <ikke> not sure that amounts to much
2022-08-06 14:56:08 <ptrc> oh
2022-08-06 14:56:50 <psykose> i wonder if pressing 'delete merged branches' breaks anything
2022-08-06 14:57:12 <ikke> 380k refs are refs/keep-around
2022-08-06 14:57:39 <ikke> 251 branches though, so 223 out of 251 is a larger amount
2022-08-06 14:57:55 <ikke> psykose: I'd rather not try
2022-08-06 14:58:06 <psykose> i would guess it's fine
2022-08-06 14:58:30 <ikke> Would be nice if it showed what it would delete
2022-08-06 14:58:38 <ikke> I could try on gitlab-test
2022-08-06 14:58:42 <ptrc> hm, i wonder what are those "keep-around" refs
2022-08-06 14:58:53 <ptrc> i wish there was better git internal docs than random irc logs
2022-08-06 14:59:13 <ikke> it's something gitlab specific
2022-08-06 14:59:51 <ikke> For things like merge requests 
2022-08-06 15:00:10 <ptrc> oh, apparently there's also a reference per pipeline
2022-08-06 15:00:18 <ikke> right
2022-08-06 15:03:14 <ikke> psykose: I don't think there are any branches considered merged
2022-08-06 15:03:34 <psykose> which makes it even more funny it's just a big red button that says nothing about what it will do
2022-08-06 15:05:15 <ikke> only 7 open MRs from mailinglist-bot
2022-08-06 15:06:38 <ptrc> there's a bunch of closed ones that still have a branch on the aports repo
2022-08-06 15:06:44 <ptrc> ah
2022-08-06 15:06:47 <ptrc> nvm
2022-08-06 15:08:20 <psykose> what if we just closed every mr?
2022-08-06 15:11:44 <ikke> https://gitlab.alpinelinux.org/alpine/aports/-/branches?state=all&search=patches
2022-08-06 15:12:42 <ptrc> wow, 2 pages of branches in total, nice :D
2022-08-06 15:19:55 <ikke> But does it make fetching faster?
2022-08-06 15:21:42 <ptrc> hard to say for sure, because it randomly takes >5 seconds sometimes
2022-08-06 15:22:08 <ptrc> but i think it's slightly faster?
2022-08-06 15:22:27 <ikke> possibly
2022-08-06 15:24:58 <psykose> you didn't measure, did you
2022-08-06 15:25:00 <psykose> :p
2022-08-06 15:25:41 <psykose> wish making clang faster didn't come with tradeoffs
2022-08-06 15:26:01 <ikke> psykose: too much noise for it to be easily measurable
2022-08-06 15:26:08 <psykose> yeah, i know
2022-08-06 17:07:49 <ikke> gitlab has been upgraded to 15.1 :-)
2022-08-06 17:09:43 <psykose> : )
2022-08-06 17:11:42 <ikke> Nothing special added in this version as far as I'm concerned
2022-08-06 17:15:01 <psykose> the close button on the 'has been upgraded to 15.1' doesnt work
2022-08-06 17:15:09 <psykose> i guess that got 'added' in this version onto my screen
2022-08-06 17:15:11 <psykose> :-)
2022-08-06 17:17:20 <ikke> :D
2022-08-06 17:17:29 <ikke> works for me ;-)
2022-08-06 17:18:17 <psykose> worked after a minute
2022-08-06 22:01:30 <psykose> one thing i forgot in the new gitlab 15 is that in the diff view there's no button for 'expand all context' anymore
2022-08-06 22:01:37 <psykose> so sometimes i have to click the arrow like 8 times instead of once
2022-08-06 22:01:42 <psykose> did they put that somewhere else or
2022-08-06 22:02:41 <panekj> three dots
2022-08-06 22:02:43 <panekj> upper right
2022-08-06 22:02:45 <panekj> show full file
2022-08-06 22:12:50 <psykose> thanks, guess that works
2022-08-06 22:12:54 <psykose> though it's two clicks instead of one
2022-08-06 22:13:08 <psykose> and shows the whole file instead of only in one direction from where you click
2022-08-06 22:13:47 <psykose> i love spacebar heating
2022-08-07 11:07:02 <Guest7304> where does the 'unionfs' from the live image allocate memory from?
2022-08-07 11:18:27 <ikke> Guest7304: From your RAM
2022-08-07 11:19:03 <Guest7304> but i give my VM 2G of RAM, and also 2G to the unionfs and still works
2022-08-07 11:19:38 <ikke> It doesn't reserve it
2022-08-07 11:19:49 <Guest7304> ooh, i see
2022-08-07 11:19:58 <Guest7304> thanks
2022-08-07 11:20:02 <ikke> It's just the max it can use
2022-08-07 11:24:33 <Guest7304> is it possible to make it use disk space as well? just in case i'm installing in a system with very little memory
2022-08-07 11:24:53 <ptrc> that would probably be just swap
2022-08-07 11:26:24 <Guest7304> yes, but if i burn the iso to the flash drive i dont think it will use swap, just RAM
2022-08-07 17:09:20 <psykose> Ariadne: the new pkgconf version breaks some versions in cmake
2022-08-07 17:09:55 <psykose> downgrading to 1.7 fixes it; with 1.8 you get something like: https://img.ayaya.dev/r3JOUl2H85DT
2022-08-07 17:10:07 <psykose> relevant cmake: https://img.ayaya.dev/MdgBUmeDFJjw
2022-08-07 17:10:23 <psykose> the _VERSION there gets set to multiple things
2022-08-07 17:10:28 <psykose> not sure why it fails
2022-08-07 17:24:31 <psykose> everything else more or less still works fine, so i guess it's just some weird edge case here
2022-08-07 17:25:40 <psykose> aha
2022-08-07 17:25:43 <psykose> `pkgconf --print-provides gobject-introspection-1.0`
2022-08-07 17:26:04 <psykose> 1.8 reports 1.72.0 -> https://img.ayaya.dev/uhVTsuVuiCy2
2022-08-07 17:26:24 <psykose> 1.9 reports the depends as well -> https://img.ayaya.dev/HVlp8mHlklyk
2022-08-07 17:26:52 <psykose> i'm assuming this is what cmake is using, or something similar, and it creates an array of versions into the _VERSION variable
2022-08-07 18:38:45 <psykose> (i mixed up the versions at first and i do mean 1.9)
2022-08-07 20:26:44 <Ariadne> i’ll cut a 1.9.1 with the fix
2022-08-07 20:28:38 <psykose> ah
2022-08-07 20:28:38 <psykose> thanks!
2022-08-08 00:32:19 <Ariadne> psykose: i have a set of regression tests for this now :)
2022-08-08 00:32:45 <Ariadne> psykose: what is happening is that --modversion and --print-provides are acting on the flattened graph, rather than the original one.
2022-08-08 00:32:53 <psykose> ahuh
2022-08-08 00:33:23 <psykose> regression tests are great
2022-08-08 00:45:35 <Ariadne> anyway, past me thought about that, but had a mitigation in an #ifdef XXX_NOTYET block
2022-08-08 00:54:15 <psykose> you're certainly very good at this graph stuff
2022-08-08 00:54:30 <Ariadne> no i'm not
2022-08-08 00:54:34 <psykose> are :)
2022-08-08 00:54:41 <psykose> also there's new binutils, you should yolopush that one
2022-08-08 00:54:43 <psykose> more fun for me
2022-08-08 00:55:44 <Ariadne> the flattener is not that exciting
2022-08-08 00:56:36 <psykose> maybe it expands sometimes? that would be of much excitement
2022-08-08 00:56:53 <psykose> 🥺
2022-08-08 00:59:16 <Newbyte> 👁️
2022-08-08 00:59:52 <psykose> see? newbyte loves looking at expanding things
2022-08-08 01:00:26 <Newbyte> yes, love me some expanding ... graphs
2022-08-08 01:02:47 <psykose> there must be some good inflation joke here but i can't quite come up with one
2022-08-08 01:04:32 <Ariadne> it just copies all the dependencies to a sequence and sorts the sequence by how many edges each dependency has
2022-08-08 01:05:55 <Ariadne> the previous solver did the same, but instead, reordered nodes in the sequence (pushing them back) as it visited each dependency edge
2022-08-08 01:07:38 <psykose> see, i don't know what any of that actually means, as i'm terrible with graphs
2022-08-08 01:07:43 <psykose> which makes it incredibly impressive
2022-08-08 01:07:54 <psykose> (i was also about to make an edging joke to highlight how little i understand)
2022-08-08 01:08:14 <Newbyte> this is the first time all the stuff I learnt about graphs in university has become useful
2022-08-08 01:08:49 <psykose> i dropped out before i got to the graph part
2022-08-08 01:08:51 <psykose> big sad
2022-08-08 02:27:57 <sam_> psykose: can you build rust with new pkgconf 
2022-08-08 02:27:59 <sam_> i get some horrible openssl thing
2022-08-08 02:28:09 <psykose> 1.9 or .9.1?
2022-08-08 02:28:12 <sam_> (like https://bugs.gentoo.org/864265 but for Rust itself too, I thought it was DT_RELR at first)
2022-08-08 02:28:17 <sam_> same with .1
2022-08-08 02:28:26 <psykose> hmm
2022-08-08 02:28:28 <psykose> ok, me check
2022-08-08 02:28:29 <psykose> moment
2022-08-08 02:28:50 <psykose> (by moment i mean i just started it, will see how long to fail, haha)
2022-08-08 02:28:59 <sam_> sure no rush
2022-08-08 02:29:00 <sam_> i need to go eat
2022-08-08 02:29:05 <psykose> but yeah might just be the new graph changes
2022-08-08 02:29:16 <psykose> ariadne is right here, will see if our toolchain fails the same so it's that for sure
2022-08-08 02:29:56 <sam_> so you have 2 minutes
2022-08-08 02:30:00 <sam_> fwiw with 1.9.0, I got a weird thing with pkg-config --modversion Qt5Core where the version repeated itself
2022-08-08 02:30:02 <sam_> but that went away with .1
2022-08-08 02:30:39 <psykose> yeah i reported that right above
2022-08-08 02:30:45 <psykose> also are you using system libgit2?
2022-08-08 02:30:47 <psykose> we don't
2022-08-08 02:30:48 <psykose> for rust
2022-08-08 02:31:21 <Ariadne> i pushed a patch that might help with static
2022-08-08 02:31:37 <psykose> aha, i see vendored-libgit2 in the logs there
2022-08-08 02:31:39 <psykose> guess not
2022-08-08 02:34:12 <psykose> will test the static patch if it fails
2022-08-08 02:34:34 <Ariadne> anyway if it doesn’t work i’ll dig into it tomorrow
2022-08-08 02:34:51 <Ariadne> i did describe the 1.9 branch as experimental :)
2022-08-08 02:36:48 <psykose> is there a reason you didn't name it 2.0_alpha or similar? since it's in preparation for 2
2022-08-08 02:37:07 <psykose> (just curious, makes no difference)
2022-08-08 02:46:33 <psykose> >I've just built my system with DT_RELR (-Wl,-z,pack-relative-relocs in LDFLAGS)
2022-08-08 02:46:40 <psykose> you really *are* a gentoo developer :p
2022-08-08 03:11:16 <psykose> same failure
2022-08-08 03:12:12 <psykose> now trying the extra patch
2022-08-08 03:13:58 <sam_> ah, ionen beat me to reporting it: https://github.com/pkgconf/pkgconf/issues/258
2022-08-08 03:14:00 <sam_> I suspect this is it
2022-08-08 03:14:02 <sam_> Ariadne: it's not unleashed on anyone, it's fine!
2022-08-08 03:14:04 <sam_> i'm not crazy, honest
2022-08-08 03:16:26 <psykose> aha
2022-08-08 03:16:27 <psykose> so it was
2022-08-08 04:35:05 <panekj> trying to compile rust bin statically on aarch64 but it fails when finding getauxval, even when passed -mno-outline-atomics, what could be the problem? https://tpaste.us/x5d1
2022-08-08 04:38:58 <psykose> there is an -lc, then an end group, then compiler-builtins
2022-08-08 04:39:08 <psykose> and no -lc after compiler-builtins
2022-08-08 04:39:16 <psykose> i bet my changes to 'fix static linking' are incomplete
2022-08-08 04:39:17 <psykose> blech
2022-08-08 04:39:36 <psykose> the issue is you need an -lc after the compiler-builtins (getauxval is a libc symbol), since it comes first it's discarded
2022-08-08 04:39:47 <panekj> not really, this happens for rustup rust as well
2022-08-08 04:39:51 <psykose> oh, huh
2022-08-08 04:39:54 <psykose> hm
2022-08-08 04:39:58 <psykose> well, i don't really have any idea
2022-08-08 04:40:14 <psykose> (although do you mean glibc rustup? because musl rustup rust is even more broken than ours, and not a good comparison)
2022-08-08 04:41:12 <panekj> glibc static compiling is even more cursed than all of this
2022-08-08 04:41:17 <panekj> yeah, no
2022-08-08 04:42:27 <panekj> I mean: aarch64-alpine-linux-musl has same issue as aarch64-unknown-linux-musl
2022-08-08 04:43:36 <panekj> I'm not sure how to pass -lc after it
2022-08-08 04:44:05 <psykose> you can't
2022-08-08 04:44:11 <panekj> :(
2022-08-08 04:46:52 <psykose> i mean, your mno-outline is there
2022-08-08 04:47:01 <psykose> you can try see if the same thing works i guess, but i doubt it
2022-08-08 04:47:27 <psykose> and it's a little beyond my knowledge
2022-08-08 05:03:10 <psykose> post what you're compiling
2022-08-08 05:34:53 <panekj> psykose: lapce
2022-08-08 05:35:06 <psykose> the flags and whole build
2022-08-08 05:35:56 <panekj> https://www.irccloud.com/pastebin/eUzmEBvc/
2022-08-08 05:36:20 <panekj> apk add -t lapce zlib-static freetype-static fontconfig-static libgit2-static libssh2-static openssl-libs-static libssl3 gtk+3.0-dev http-parser-dev cargo build-base openssl-dev git
2022-08-08 06:02:32 <psykose> 12mb, wtf
2022-08-08 06:03:13 <panekj> mhmm, tasty wasm
2022-08-08 06:28:48 <psykose> panekj: https://img.ayaya.dev/qYS2fpfv7WUh idk it builds
2022-08-08 06:28:52 <psykose> get good i guess
2022-08-08 06:29:07 <panekj> >:(
2022-08-08 06:29:16 <psykose> i only added lld
2022-08-08 06:33:02 <panekj> I think I saw that somewhere
2022-08-08 06:33:24 <panekj> but rust-lld
2022-08-08 06:37:17 <psykose> there is no rust-lld unless you are using some rustup stuff again
2022-08-08 06:38:20 <panekj> as a solution, on the interwebz
2022-08-08 06:38:50 <psykose> it's just link-arg=-fuse-ld=lld
2022-08-08 06:39:17 <psykose> rust-lld is the exact same, by default they build an llvm toolchain and shit renamed binaries
2022-08-08 06:39:22 <psykose> ship*
2022-08-08 06:39:38 <psykose> it's a patched llvm technically, there's a few patches on top
2022-08-08 06:39:42 <psykose> i think for 14 there's like three
2022-08-08 06:39:56 <panekj> I know it's llvm toolchain
2022-08-08 06:39:58 <psykose> maybe it's a good idea to even apply them to ours, but in practice i've never seen any reason
2022-08-08 06:40:13 <panekj> but how am I supposed to know what kind of shit they do to lld
2022-08-08 06:40:17 <psykose> nothing
2022-08-08 06:40:39 <panekj> void
2022-08-08 06:40:44 <psykose> void?
2022-08-08 06:40:54 <panekj> nothing
2022-08-08 06:40:58 <psykose> something
2022-08-08 07:09:23 <staceee> Do we package amdvlk ?
2022-08-08 07:09:31 <psykose> nope
2022-08-08 07:10:01 <staceee> is that intended ?
2022-08-08 07:12:11 <panekj> no one just packaged it
2022-08-08 07:13:44 <staceee> okay
2022-08-08 07:14:22 <psykose> i've built it before
2022-08-08 07:14:28 <psykose> i think
2022-08-08 07:14:43 <psykose> not really sure what it would even be used for
2022-08-08 07:15:03 <staceee> I dunno if it is necessary for steam flatpak
2022-08-08 07:15:18 <staceee> I'm investigating why dota 2 doesnt start
2022-08-08 07:15:40 <psykose> i don't think that would be related
2022-08-08 07:15:47 <psykose> the mesa radv one isn't worse or anything
2022-08-08 07:15:52 <psykose> if anything it's 2x faster half the time lmao
2022-08-08 07:16:37 <psykose> iirc i started dota2 fine.. i can check
2022-08-08 07:17:00 <psykose> (also i am judging your videogame choices /s)
2022-08-08 07:17:29 <panekj> > AMDVLK is built separate from Mesa. It's not using any Mesa/Gallium/NIR APIs or anything living within the Mesa tree. This is since this Vulkan driver is designed with cross-platform and other internal AMD APIs in mind.
2022-08-08 07:17:59 <psykose> yes, it's a separate vulkan driver
2022-08-08 07:18:13 <psykose> things use the vulkan loader to load a specific one, by default you get the mesa radv
2022-08-08 07:19:21 <psykose> gimme 30 minutes to even download dota lol
2022-08-08 07:19:41 <staceee> aha could be nice to double check yap !
2022-08-08 07:19:51 <staceee> It fail with proton-ge after loading screen
2022-08-08 07:19:54 <psykose> anything you have to pass specifically for vulkan or nah
2022-08-08 07:19:59 <psykose> iirc there was an arg
2022-08-08 07:20:06 <staceee> it fail instantly in linux linux runtime
2022-08-08 07:20:10 <psykose> but that was 5 years ago
2022-08-08 07:20:40 <staceee> it worked 7 monthes ago then stopped to works. Now some issues has been closes recently so I expect things to works now o_O
2022-08-08 07:21:30 <psykose> uhh, linux runtime and proton? isn't one for native games and the other for windows
2022-08-08 07:23:09 <staceee> linux games can still be toggled to windows mode if not working
2022-08-08 07:23:26 <staceee> funny but most of games works better on proton-ge
2022-08-08 07:23:27 <psykose> yes, but what i mean is that they're not at the same time?
2022-08-08 07:23:39 <psykose> how do you run with proton and crash in the linux runtime
2022-08-08 07:23:49 <psykose> unless i forgot how it works
2022-08-08 07:24:43 <staceee> I toggle the compatibility mode
2022-08-08 07:24:59 <staceee> not at the same time. Steam re-download the difference
2022-08-08 07:25:58 <psykose> yeah, but they're still separate when you press play
2022-08-08 07:26:21 <psykose> so i guess you mean it crashes either way in both
2022-08-08 07:26:39 <staceee> yap !
2022-08-08 07:27:03 <psykose> oki
2022-08-08 07:27:12 <psykose> loading into the game start or an actual game match
2022-08-08 07:27:50 <staceee> into the game
2022-08-08 07:27:59 <staceee> after the black background and logo screen
2022-08-08 07:28:14 <psykose> oki
2022-08-08 07:28:17 <psykose> will test when it downloads
2022-08-08 07:28:27 <psykose> before that, are you on edge
2022-08-08 07:28:51 <staceee> edge and got mesa-vulkan-ati
2022-08-08 07:28:56 <psykose> remove that :)
2022-08-08 07:29:14 <psykose> `apk add mesa-dri-gallium mesa-va-gallium mesa-vdpau-gallium`
2022-08-08 07:29:17 <psykose> and remove whatever else you had
2022-08-08 07:29:20 <psykose> it should upgrade 21->22
2022-08-08 07:29:29 <staceee> I also got that but I need vulkan ?
2022-08-08 07:29:45 <psykose> yeah, and -vulkan-ati
2022-08-08 07:31:02 <staceee> so now I get no /usr/share/vulkan/icd.d/
2022-08-08 07:31:11 <staceee> is that correct ? o_O
2022-08-08 07:31:31 <psykose> mesa-vulkan-ati has it
2022-08-08 07:31:45 <staceee> ah okay
2022-08-08 07:31:58 <staceee> I just added va- and vdpau- so
2022-08-08 07:36:54 <staceee> probably I should wait for the driver fixes to land on flatpak runtime environments
2022-08-08 07:46:32 <psykose> staceee: i can load into tutorial fine on native
2022-08-08 07:47:48 <psykose> https://img.ayaya.dev/b4cApZ6PZNKs
2022-08-08 07:48:06 <psykose> vulkan in the settings
2022-08-08 07:48:14 <psykose> there's also opengl to picl i guess lol
2022-08-08 07:50:44 <staceee> o_O
2022-08-08 07:50:58 <staceee> okay so something is wrong on my setup or hardware
2022-08-08 07:51:08 <staceee> psykose: so you mean steam linux runtime ?
2022-08-08 07:51:10 <psykose> ye
2022-08-08 07:51:28 <staceee> could you flatpak list ?
2022-08-08 07:51:36 <staceee> I'll investigate later
2022-08-08 07:52:31 <psykose> uhh it's just stable steam
2022-08-08 07:52:34 <psykose> nothing else is used
2022-08-08 07:52:46 <psykose> and by stable i mean the 'stable' flatpak
2022-08-08 07:52:49 <psykose> actual steam is beta
2022-08-08 07:53:05 <staceee> ah, I got stable
2022-08-08 07:55:01 <staceee> g2g I'll investigate later. Ty for the double check !
2022-08-08 08:42:55 <Ariadne> sam_: pkgconf 1.9.2 will land later today with further solver improvements for that issue.  basically libssl gets optimized out of the solution because of requires.private, so i am going to rework the solver a little to special case requires.private
2022-08-08 08:43:22 <Ariadne> 
ACTION mutters about requires.private

2022-08-08 08:59:40 <ncopa> Ariadne: congrats with the 1.9 release(es)!
2022-08-08 08:59:51 <Ariadne> not congrats yet
2022-08-08 09:00:06 <Ariadne> entirely new solver, have to work out the bugs :p
2022-08-08 09:00:27 <Ariadne> the good news is that pkgconf's regression suite is growing
2022-08-08 09:03:05 <ncopa> getting the release out is always a good thing
2022-08-08 09:03:19 <Ariadne> yeah, this seemed like a good time to start testing the new solver in the real world
2022-08-08 09:03:29 <ncopa> *nod*
2022-08-08 09:03:36 <Ariadne> there was no further work that could be done without actually putting it through its paces
2022-08-08 09:04:39 <ncopa> i am confident that whatever issues that pops up, will be fixed quickly
2022-08-08 09:05:11 <ncopa> that was at least the case in the past
2022-08-08 09:07:15 <Ariadne> the current plan is to shake out any remaining issues, and release pkgconf 2.0 with alpine 3.17 :)
2022-08-08 09:13:57 <panekj>   [77] Problem with the SSL CA cert (path? access rights?) (error setting certificate file: /etc/ssl/certs/ca-certificates.crt)
2022-08-08 09:14:30 <psykose> i would love some context
2022-08-08 09:14:31 <panekj> is ssl fine o rv64?
2022-08-08 09:14:35 <psykose> probably not
2022-08-08 09:14:38 <panekj> :c
2022-08-08 09:14:40 <psykose> none of community is built
2022-08-08 09:14:42 <psykose> check in 2 weeks
2022-08-08 09:14:58 <psykose> (i am serious with that number)
2022-08-08 09:15:17 <panekj> Is it that slow
2022-08-08 09:15:20 <psykose> rust will be blocked again, there are 300 more, it freezes randomly, and random stuff gets thrown into the pile all the time on restarts
2022-08-08 09:15:20 <psykose> yes
2022-08-08 09:15:44 <psykose> i question the sanity of whoever started a riscv port
2022-08-08 09:15:50 <psykose> or why i have spent countless hours fixing shit for it
2022-08-08 09:15:53 <psykose> but it is what it is
2022-08-08 09:16:22 <psykose> it's mostly the freezing really
2022-08-08 09:16:28 <psykose> it takes a long time to notice
2022-08-08 09:16:32 <panekj> is it proper rv64 board is it qemu
2022-08-08 09:16:39 <psykose> there are no boards, it's qemu
2022-08-08 09:16:56 <psykose> a real board would be slower, shrug
2022-08-08 09:17:06 <psykose> would stop the freezing though
2022-08-08 09:17:20 <psykose> in ~2 years there should be some slightly better hardware, but it's expensive in any case
2022-08-08 09:23:44 <ikke> panekj: qemu-user
2022-08-08 09:23:56 <panekj> sad
2022-08-08 09:24:03 <ikke> We do have some real boards, but they are low on specs, not suited as builders 
2022-08-08 09:24:17 <ikke> And not hosted anyway
2022-08-08 09:25:39 <psykose> and yet i don't have one 😠
2022-08-08 09:25:51 <psykose> nobody even brought it up yet did they
2022-08-08 09:28:18 <panekj> why would you need one
2022-08-08 09:30:56 <panekj> now I kinda wish we had crosscompilers
2022-08-08 09:32:45 <psykose> they're very easy to make
2022-08-08 09:33:17 <psykose> for clang you just need a sysroot; pass --target --sysroot and it works. for gcc you have to build one, but it's --target at build time +- some extras
2022-08-08 09:33:35 <psykose> then for linkers we have cross-binutils already
2022-08-08 09:33:51 <panekj> I'm currently using gcc from musl.cc
2022-08-08 09:33:57 <psykose> also works
2022-08-08 09:34:03 <panekj> but had to mirror that because it timedout in CI
2022-08-08 09:34:14 <psykose> well that's easy
2022-08-08 09:34:16 <psykose> stop running ci
2022-08-08 09:34:18 <psykose> ez
2022-08-08 09:34:27 <panekj> I'll try to get clang working
2022-08-08 09:41:56 <ddevault> I'd like to state for the record that the GitLab API is dumb
2022-08-08 09:42:29 <Ariadne>  i've never had the pleasure, but i'll take your assessment as accurate
2022-08-08 09:43:13 <ddevault> the GraphQL API and REST API have incompatible ID namespaces and you have to do string munging heuristics to convert between them
2022-08-08 09:43:23 <ddevault> and you do have to convert between them because you cannot configure webhooks via the GraphQL API
2022-08-08 09:43:34 <ddevault> and webhook payloads are undocumented so you just have to issue one and see what it gives you
2022-08-08 09:43:51 <ddevault> and there are at least three different permissions spaces, the rights afforded by which are undocumented and the interactions between which are undocumented
2022-08-08 09:47:50 <Ariadne> amazing
2022-08-08 09:59:51 <ncopa> psykose: what was the reason to switch back to openssl 1.1 for spdk?  333be493b7226eb839142a038d808651f2999b29
2022-08-08 10:00:03 <ncopa> can we do openssl3 again?
2022-08-08 10:00:19 <psykose> i'm pretty sure that is openssl3
2022-08-08 10:00:22 <psykose> and i wrote that wrong
2022-08-08 10:00:27 <psykose> because openssl-dev.. is openssl3
2022-08-08 10:00:48 <ncopa> well, in june i think it was 1.1
2022-08-08 10:00:49 <psykose> oor
2022-08-08 10:00:54 <psykose> oh
2022-08-08 10:00:56 <psykose> that's june
2022-08-08 10:00:58 <psykose> nvm
2022-08-08 10:01:16 <ncopa> i just wondered if there was known issue that it doesnt work with openssl3
2022-08-08 10:01:33 <ncopa> or other reason to not use openssl3
2022-08-08 10:01:57 <psykose> no, it was just june
2022-08-08 10:02:03 <psykose> it was the only thing in all of aports that used openssl3
2022-08-08 10:02:11 <psykose> for no reason, it was added as a mistake
2022-08-08 10:02:22 <psykose> so, idk, probably fine
2022-08-08 10:02:30 <ncopa> ok. good thanks!
2022-08-08 10:02:34 <psykose> it builds
2022-08-08 10:02:37 <ncopa> yeah
2022-08-08 10:02:39 <ncopa> i'll push it
2022-08-08 10:03:28 <psykose> accidentally looked that one over
2022-08-08 10:03:32 <psykose> as for everything else, they don't support 3
2022-08-08 10:04:15 <ncopa> im looking over the list stuff that still links to openssl 1.1
2022-08-08 10:04:31 <ncopa> but the list is short
2022-08-08 10:05:02 <psykose> none of them support it; i looked over racket/rakudo/nyxt since they are lisps and dlopen it and pretty sure they won't
2022-08-08 10:05:13 <Ariadne> rakudo is perl
2022-08-08 10:05:22 <psykose> same thing different brackets
2022-08-08 10:05:33 <Ariadne> no, one is line noise, the other is lisp
2022-08-08 10:05:44 <psykose> see? same thing, different brackets
2022-08-08 10:05:52 <ncopa> :)
2022-08-08 10:05:55 <psykose> :p
2022-08-08 10:06:00 <psykose> but yes, i'm tired
2022-08-08 10:06:19 <Ariadne> i have to say
2022-08-08 10:06:29 <Ariadne> this pkgconf stuff would be a lot easier if i was at my old job still
2022-08-08 10:06:33 <Ariadne> i could just rebuild
2022-08-08 10:06:35 <Ariadne> all of alpine
2022-08-08 10:06:38 <Ariadne> and see what breaks
2022-08-08 10:06:40 <Ariadne> in like 2 hours
2022-08-08 10:06:53 <Ariadne> alas, no supercomputing grid
2022-08-08 10:06:54 <Ariadne> :(
2022-08-08 10:07:03 <psykose> it's fine
2022-08-08 10:07:13 <psykose> the time spent isn't that different, is it? we just randomly tell you :)
2022-08-08 10:07:21 <psykose> just ignore it for a week and let it pile up first /s
2022-08-08 10:07:43 <Ariadne> i think 1.9.2 should provide a fairly stable solver
2022-08-08 10:07:57 <psykose> fingers crossed
2022-08-08 10:08:09 <ncopa> im curious why a new solver was needed?
2022-08-08 10:08:21 <Ariadne> you don't want to know
2022-08-08 10:08:28 <ncopa> :)
2022-08-08 10:08:39 <Ariadne> lets just say
2022-08-08 10:08:55 <Ariadne> there are serious crimes being committed with pkgconf
2022-08-08 10:09:03 <Ariadne> at a certain search monopoly
2022-08-08 10:09:28 <Ariadne> they took my advice about declaring *all* dependencies very seriously
2022-08-08 10:09:53 <Ariadne> which is good, but then they created these huge dependency graphs with millions of edges
2022-08-08 10:10:05 <Ariadne> because pkgconf's solver never did an optimization step
2022-08-08 10:10:11 <Ariadne> so it would gladly walk all those edges
2022-08-08 10:10:24 <ncopa> and 2 years later it completed....
2022-08-08 10:10:24 <psykose> how much did you get paid by google to fix that
2022-08-08 10:11:09 <Ariadne> hey i never said it was them :)))
2022-08-08 10:11:12 <Ariadne> maybe it was bing
2022-08-08 10:11:21 <panekj> lol
2022-08-08 10:11:22 <Ariadne> okay it wasn't bing
2022-08-08 10:11:44 <Ariadne> ncopa: not 2 years later, but several minutes
2022-08-08 10:12:06 <ncopa> yeah, i get the point
2022-08-08 10:12:25 <Ariadne> (apk-tools also does not do an optimization step :))
2022-08-08 10:12:28 <psykose> how could you do this to ask jeeves >:(
2022-08-08 10:13:11 <Ariadne> the good news is, now you can play sudoku with pkgconf
2022-08-08 10:13:17 <ncopa> lol
2022-08-08 10:14:19 <Ariadne> freedesktop pkg-config solves those google graphs very fast
2022-08-08 10:14:22 <Ariadne> ... by segfaulting
2022-08-08 10:14:29 <Ariadne> ... that counts as a solution, right?
2022-08-08 10:14:37 <ncopa> absolutely :)
2022-08-08 10:14:41 <psykose> it is
2022-08-08 10:15:07 <psykose> if they removed the memory protection in the kernel, and just didn't segfault, it would play chess
2022-08-08 10:15:19 <ncopa> so it took several minutes, how long does it take now?
2022-08-08 10:15:36 <Ariadne> I lie, technically, pkgconf 1.x did have an optimization pass (PKGCONF_PKG_PROPF_VISITED), but it was very naive
2022-08-08 10:15:56 <Ariadne> ncopa: 0.038 seconds :)
2022-08-08 10:16:14 <ncopa> very nice
2022-08-08 10:17:29 <Ariadne> psykose: they paid enough :)))
2022-08-08 10:17:48 <psykose> $5 and a stale donut?
2022-08-08 10:17:54 <Ariadne> some more zeros
2022-08-08 10:18:06 <psykose> $0.5 and zero stale donuts?
2022-08-08 10:18:10 <Ariadne> wow
2022-08-08 10:18:13 <psykose> :3
2022-08-08 10:18:54 <Ariadne> i dunno, they allowed me to take a year of paid sabbatical to just hack on alpine full time
2022-08-08 10:19:12 <psykose> yuge
2022-08-08 10:21:26 <Ariadne> anyway, the riskier stuff was introduced after i wrote the flattener
2022-08-08 10:21:58 <Ariadne> the stuff that is being a pain, that is, came from desires to improve the rigidity of the solver from the meson maintainers
2022-08-08 10:22:08 <Ariadne> especially relating to mesonpp
2022-08-08 10:22:58 <psykose> how is that alternate meson stuff coming along anyway
2022-08-08 10:23:02 <psykose> i don't keep up much
2022-08-08 10:23:31 <Ariadne> (the new solver has behavior which is can actually be specified without any UB)
2022-08-08 10:23:37 <Ariadne> -is
2022-08-08 10:24:01 <psykose> o yea let's throw asan at it
2022-08-08 10:25:31 <Ariadne> unfortunately, i have introduced some regressions in the ASan department :p
2022-08-08 10:25:37 <psykose> :)
2022-08-08 10:29:50 <Ariadne> sadly, alpine does not have ASan afaik
2022-08-08 10:30:04 <psykose> we do
2022-08-08 10:30:09 <psykose> clang -fsanitize=address
2022-08-08 10:30:12 <psykose> add compiler-rt
2022-08-08 10:30:14 <psykose> :)
2022-08-08 10:30:16 <psykose> it works
2022-08-08 10:30:34 <psykose> as does ubsan, and i think a few others
2022-08-08 10:31:03 <psykose> the libsanitizer gcc shit doesn't work though, yes
2022-08-08 10:31:12 <psykose> (doesn't build with the usual glibc missing stuff reasons)
2022-08-08 10:32:06 <Ariadne> ah yeah with clang it is possible :p
2022-08-08 10:32:26 <psykose> the better compiler as they say
2022-08-08 10:32:44 <psykose> i wonder how long i can keep being a heretic before someone bans me from this channel
2022-08-08 10:45:52 <Ariadne> i mean
2022-08-08 10:45:57 <Ariadne> i would like to switch to clang someday
2022-08-08 10:45:58 <Ariadne> :p
2022-08-08 10:47:41 <panekj> *chimera-linux*
2022-08-08 10:47:50 <panekj> :P
2022-08-08 10:49:31 <psykose> i've selectively enabled it in quite a few places
2022-08-08 10:49:43 <psykose> for some reason clang -dbg splits are 50% smaller in some cases
2022-08-08 10:49:46 <psykose> usually 20+
2022-08-08 11:04:37 <Ermine> Is anything wrong with gitlab? git pull upstream master just hangs up
2022-08-08 11:05:11 <skarnet> it’s just slow
2022-08-08 11:05:14 <panekj> if you are not pulling to local master branch, it might do that
2022-08-08 11:05:32 <skarnet> mine succeeded after like 20 minutes
2022-08-08 11:06:20 <panekj> I don't know why but whenever I had to push/pull something, it would hang if I didn't update my local/remote master branch
2022-08-08 11:07:27 <Ermine> My command is running for more than a minute
2022-08-08 11:07:51 <Ermine> and another returned 'fatal: the remote end hung up upon initial contact'
2022-08-08 11:08:13 <psykose> try without ipv6
2022-08-08 11:08:17 <psykose> aside from that, dunno
2022-08-08 11:08:22 <psykose> even ssh is slow
2022-08-08 11:08:29 <psykose> i usually pull master through https instead
2022-08-08 11:22:37 <Ermine> Ah, apparently my connection is slow
2022-08-08 13:01:29 <Ermine> should I report s390x issue in !37381 to upstream?
2022-08-08 13:05:25 <psykose> who on earth is using this on s390x
2022-08-08 13:06:01 <Ermine> idk but just in case
2022-08-08 15:07:31 <ncopa> i think it might be a good idea to report things upstream and let they decide if its worth fix or not
2022-08-08 15:24:51 <ikke> Ermine: fetching via ssh does not seem to be slow for me
2022-08-08 15:46:23 <Ermine> ncopa: can I send CI output to upstream?
2022-08-08 15:47:58 <ikke> Sure, why not?
2022-08-08 15:48:10 <ikke> Note that the log itself is kept for at most 30 days
2022-08-08 15:51:15 <Ermine> ok. Do builders have edge installed?
2022-08-08 15:51:45 <ikke> Each builder has packages for the version it builds
2022-08-08 15:52:17 <ikke> For CI, it up/downgrades to the the target version
2022-08-08 15:53:22 <ikke> Ermine: the builders only use packages that it built itself
2022-08-08 15:53:30 <ikke> (apart from bootstrapping)
2022-08-08 15:53:48 <Ermine> well
2022-08-08 15:53:57 <Ermine> thank you
2022-08-08 18:35:05 <nmeum> anybody have a better solution for !37399? I do not recall how we handled this previously when OpenSSL 1.1 and OpenSSL 3 co-existed but it's definitly broken for me right now
2022-08-08 18:38:42 <nmeum> https://gitlab.alpinelinux.org/alpine/aports/-/blob/832bbeb063dbb7ef6b6dd6487a9ed26f32666977/main/openssl1.1-compat/APKBUILD#L126-L127
2022-08-08 18:38:51 <nmeum> ah, previously the symlinks were provided directly by openssl but they aren't anymore now
2022-08-08 18:39:08 <nmeum> hm, I think it makes more sense to have those symlinks in ca-certificates, no? otherwise the files they point to don't exist
2022-08-08 18:58:39 <sam_> Ariadne: fantastic, thank you
2022-08-08 18:58:47 <sam_> i'll let you know ofc if anything else exciting happens
2022-08-08 19:04:04 <Ariadne> i think we are good… dcbaker is going to be annoyed that i reintroduced some load bearing memory leaks though.
2022-08-08 19:07:48 <Ermine> What's wrong here? https://pastesite.org/view/2d8c5e20 . In musl, stdout is defined to (stdout), and seems like c++ compiler does not like it
2022-08-08 19:12:32 <nmeum> probably expands the macro there and that causes a syntax error? I would just suggest renaming the enumerator to something else
2022-08-08 19:23:23 <Ermine> nmeum: good idea, thank you!
2022-08-08 19:28:21 <Ermine> It indeed helped
2022-08-08 19:59:44 <Ermine> psykose: did you just cut out execinfo support in packages?
2022-08-08 20:01:29 <nmeum> yes, libexecinfo never really worked
2022-08-08 22:09:32 <Guest7748> is there a way to make the iso automatically create a partition for swap when i 'dd' it?
2022-08-08 22:18:32 <Ermine> Guest7748: afaik no
2022-08-09 07:20:06 <psykose> Ermine: if by just you mean a week ago, yes
2022-08-09 07:20:11 <psykose> everything has it patched out
2022-08-09 07:20:24 <psykose> nmeum: yeah, that's a better way of doing it than before :) i guess we forgot in the upgrade
2022-08-09 07:20:41 <nmeum> np
2022-08-09 07:22:21 <psykose> i don't think that openconnect fix works
2022-08-09 07:22:38 <psykose> it depends on krb5-libs and those link openssl3
2022-08-09 07:22:49 <psykose> (as a start, i didn't follow the rest of the dep chain)
2022-08-09 07:28:44 <nmeum> openconnect itself is definitly linked against libcrypto.1 and libssl.1 with this patch, not sure how it interacts with kerberos though
2022-08-09 07:32:07 <nmeum> hm
2022-08-09 07:35:28 <psykose> > 9069B3F2667F0000:error:0A000152:SSL routines:final_renegotiate:unsafe legacy renegotiation disabled:ssl/statem/extensions.c:879: 
2022-08-09 07:35:37 <psykose> that looks like the website is tls 1.1 or something
2022-08-09 07:35:44 <psykose> openssl3 explicitly has that disabled
2022-08-09 07:35:47 <psykose> looks like it works fine to me
2022-08-09 07:36:56 <psykose> you can reproduce the same by curl'ing the domain
2022-08-09 07:43:07 <psykose> although the website /does/ open fine in a web browser, so i'm a little confused
2022-08-09 07:46:12 <ncopa> morning!
2022-08-09 07:46:25 <ncopa> i am about to tag new stable releases, for the zlib fix
2022-08-09 07:46:33 <ncopa> zlib fixes
2022-08-09 07:46:55 <ncopa> psykose: we have what we need for zlib now?
2022-08-09 07:47:08 <psykose> looks like it
2022-08-09 07:47:16 <psykose> backport was done before, additional bugfix i did just now
2022-08-09 07:47:23 <ncopa> i saw
2022-08-09 07:47:28 <ncopa> thank you for taking care of it
2022-08-09 07:47:48 <psykose> i take care of all the easy things i see on oss-security ^^
2022-08-09 08:01:13 <nmeum> psykose: oh, I didn't actually notice that he posted a link to the site. feel free to revert it then
2022-08-09 08:01:27 <psykose> i am not 100% sure on what is actually happening
2022-08-09 08:01:32 <psykose> i mean, the website is literally on 3DES
2022-08-09 08:01:40 <nmeum> haha wow
2022-08-09 08:01:43 <psykose> but i would like to know what openssl is doing specifically
2022-08-09 08:01:45 <psykose> that 1.1 didn't do
2022-08-09 08:01:55 <nmeum> I was just reminded of my isync issue and hence reverted back to openssl 1.1 since it looked very similar to what I was seeing with isync
2022-08-09 08:01:56 <psykose> does 1.1 just allow the non-rfc-5ksomething renegotiation?
2022-08-09 08:47:29 <ncopa> im tagging releases now, please hold pushing big things for an hour or so
2022-08-09 08:47:54 <skarnet> ncopa: um, wait, question about !14046
2022-08-09 08:48:00 <skarnet> !@#$%^
2022-08-09 08:48:04 <skarnet> about #14046
2022-08-09 08:48:28 <skarnet> apparently you just closed it, with a commit that has nothing to do with my alpine-conf modifications?
2022-08-09 08:48:50 <skarnet> idgi, what does testing over ssh have to do with hwdrivers?
2022-08-09 08:49:30 <ncopa> skarnet: sorry, i have to get the release out *now* so i fixed the hwdrivers thing *only* and will follow up the other stuff later
2022-08-09 08:49:38 <psykose> https://gitlab.alpinelinux.org/alpine/alpine-conf/-/merge_requests/102
2022-08-09 08:49:46 <ncopa> the testing over ssh has nothing to do with it
2022-08-09 08:50:11 <psykose> the linked commits are not about ssh
2022-08-09 08:51:19 <skarnet> psykose: I literally clicked on the links and ended up with something talking about ssh
2022-08-09 08:51:39 <panekj> because the commit comment mentions the issue
2022-08-09 08:51:45 <ncopa> skarnet: there are a few unresolved things i'd like clarification about: https://gitlab.alpinelinux.org/alpine/alpine-conf/-/merge_requests/100#note_253349
2022-08-09 08:51:55 <psykose> there's 4 links there
2022-08-09 08:51:58 <skarnet> ncopa: this literally doesn’t work if you switch from mdevd to udev
2022-08-09 08:52:11 <skarnet> it will only work if you switch from mdev to udev
2022-08-09 08:52:18 <psykose> the first two link to the commit in alpine-conf (and don't mention ssh)
2022-08-09 08:52:26 <psykose> the latter two are an upgrade for alpine conf (and also don't mention ssh)
2022-08-09 08:52:31 <psykose> i am curious what you are looking at
2022-08-09 08:52:39 <skarnet> this bug isn’t that simple to fix, which is part of the reason why I’m taking time to answer
2022-08-09 08:53:12 <panekj> psykose https://usercontent.irccloud-cdn.com/file/ihJFud9F/image.png
2022-08-09 08:53:23 <psykose> yeah that's a deleted patch file
2022-08-09 08:53:30 <psykose> (and so has nothing to do with ssh)
2022-08-09 08:53:39 <panekj> ah, that's from aports
2022-08-09 08:53:48 <skarnet> psykose: “@ncopa closed via commit f866235e 3 minutes ago”
2022-08-09 08:53:54 <panekj> yes, that's aports
2022-08-09 08:53:58 <psykose> yeah, and the commit is an upgrade in aports
2022-08-09 08:54:11 <psykose> the actual upgrade in alpine-conf has the change, and there's 4 links there
2022-08-09 08:54:20 <skarnet> ah, got bitten again by the alpine-conf double upgrade thing
2022-08-09 08:54:32 <skarnet> sigh, forget it
2022-08-09 08:54:42 <skarnet> someday I’ll get used to it
2022-08-09 08:54:49 <psykose> (i assume you clicked a link from email)
2022-08-09 08:55:51 <skarnet> ncopa: yes, I will provide a new change over your commit
2022-08-09 08:56:11 <ncopa> skarnet: yeah, separating upstream project from aports might be confusing when we have them on same gitlab
2022-08-09 08:56:29 <skarnet> days since I haven’t had to rebase stuff and merge conflicts: 0
2022-08-09 08:56:39 <skarnet> or since I’ve had, English is hard
2022-08-09 08:56:48 <ncopa> heh...
2022-08-09 08:57:09 <ncopa> be glad you dont do things lie openssl3 upgrades on a movin git master
2022-08-09 08:57:21 <ncopa> i eat merge conflicts as breakfast...
2022-08-09 08:58:11 <ikke> nom nom nom
2022-08-09 09:02:21 <panekj> that's nothing, I'm creating my own merge conflicts and then solve them
2022-08-09 09:02:46 <ncopa> :D
2022-08-09 09:03:10 <panekj> git discipline is hard ;\
2022-08-09 09:04:36 <ncopa> skarnet: i am actually sorry for messing up your merge request, and for creating conflict with it. I can try sort it out and resolve it next week
2022-08-09 09:04:49 <ncopa> but i am a bit in a hurry today :-/
2022-08-09 09:05:18 <skarnet> don’t worry about it, from your comments it needed a full rewrite anyway
2022-08-09 09:06:04 <skarnet> I’m more worried about the mdevd-to-udev case and I will fix it, so please don’t touch it anymore
2022-08-09 09:06:58 <skarnet> what I find weird is that stuff can remain in the queue for ages, and then one day you wake up and it’s panic time and you’re merging the world, and the things I had in my pending queue are now obsoleted
2022-08-09 09:07:31 <skarnet> I wonder what prompted the need to get something out NOW ;)
2022-08-09 09:08:51 <panekj> security releases do be like that
2022-08-09 09:10:08 <psykose> just the zlib cve i guess
2022-08-09 09:10:22 <skarnet> wait what? *another* one?
2022-08-09 09:10:43 <panekj> we should replace zlib for something secure and better (:
2022-08-09 09:11:39 <skarnet> honestly, with the bugs history it has, I’m starting to get sympathetic to that view
2022-08-09 09:15:45 <psykose> if zlib is considered bad i'm going to retire from computers
2022-08-09 09:15:52 <psykose> literally nothing is good at that point
2022-08-09 09:16:13 <skarnet> pretty close to the truth
2022-08-09 09:16:25 <psykose> i think i've seen more bugs in skaware in the past 6 months than anyone has ever reported for zlib
2022-08-09 09:16:56 <skarnet> sure, but consider the age of zlib
2022-08-09 09:17:11 <psykose> aged like a fine wine :p
2022-08-09 09:17:13 <skarnet> by now you’d think that all the security issues would have been worked out
2022-08-09 09:17:25 <psykose> they are incredibly banal issues though
2022-08-09 09:17:56 <skarnet> yeah, and the fact that there are several of them still found after decades tells me something’s wrong with the design
2022-08-09 09:18:09 <skarnet> I may be wrong though
2022-08-09 09:18:16 <panekj> > The recent commit to fix a gzip header extra field processing bug introduced the new bug fixed here.
2022-08-09 09:18:45 <skarnet> ha
2022-08-09 09:18:50 <skarnet> okay that happens
2022-08-09 09:19:01 <panekj> fixed 15 hours ago
2022-08-09 09:19:11 <panekj> the bug it fixes is from 9 days ago
2022-08-09 09:19:57 <panekj> although that bug also is buffer overflow soooooooooo.....
2022-08-09 09:23:00 <panekj> blame reveals that it's from 11 years ago
2022-08-09 09:37:50 <ncopa> yeah, the panic is due to zlib cve
2022-08-09 09:46:53 <ncopa> ok releases are done. only edge snapshot that failed doe to riscv64 not being done yet
2022-08-09 09:46:58 <ncopa> will take that later
2022-08-09 09:52:46 <psykose> if they need the builds to finish first, it will take weeks
2022-08-09 09:52:49 <psykose> don't think it matters
2022-08-09 10:17:31 <nmeum> psykose: thanks for following up on the openconnect thing
2022-08-09 10:17:41 <psykose> how come
2022-08-09 10:18:12 <nmeum> hm? I mean just for debugging it further and responding to nico again
2022-08-09 10:18:15 <psykose> ah
2022-08-09 10:18:27 <psykose> :)
2022-08-09 10:18:40 <psykose> usually when i get pinged it's for bad news
2022-08-09 10:26:05 <skarnet> ncopa: alpine-conf:!100 should be okay now, I pushed a new MR version rebased on your change and addressing your concerns
2022-08-09 10:26:13 <skarnet> algitbot, you’re an idiot
2022-08-09 10:26:22 <skarnet> (sometimes)
2022-08-09 14:29:21 <Ermine> Do I need to specify autoconf in makedepends?
2022-08-09 14:33:13 <ptrc> yeah, it's not a part of build-base
2022-08-09 14:47:16 <psykose> same for automake libtool
2022-08-09 15:23:06 <adhawkins> Not sure if I should ask here or in just #alpine-linux. Is there an issue with expat in Edge?
2022-08-09 15:23:33 <adhawkins> https://paste.debian.net/1249817/
2022-08-09 15:23:34 <panekj> there is many issues in edge, you have to be more specific
2022-08-09 15:23:41 <adhawkins> See the paste.
2022-08-09 15:23:52 <panekj> apk upgrade -a
2022-08-09 15:24:41 <adhawkins> https://paste.debian.net/1249818/
2022-08-09 15:24:47 <adhawkins> Seemed to install it, but still lists an error.
2022-08-09 15:24:55 <ptrc> `apk fix` should tell you more about the error
2022-08-09 15:25:00 <panekj> probably because error is from different package
2022-08-09 15:25:14 <ptrc> ^
2022-08-09 15:25:18 <panekj> ^
2022-08-09 15:25:28 <adhawkins> Yes, reinstalled ca-certificates-bundle when I ran 'fix' and now the error is gone.
2022-08-09 15:25:40 <adhawkins> Any way to get 'apk upgrade' give more details about what the error is?
2022-08-09 15:25:55 <panekj> normally it should do that when upgrading
2022-08-09 15:26:04 <panekj> when *actually* upgrading/installing
2022-08-09 15:26:07 <panekj> not after
2022-08-09 15:26:09 <ptrc> and it probably did, but you might have missed it
2022-08-09 15:26:23 <adhawkins> Oh I see, so the error occurred during a past upgrade, and I just missed it? Got it.
2022-08-09 15:26:38 <ptrc> subsequent transactions just say that there is an error, but don't try to fix it by themselves
2022-08-09 15:26:53 <adhawkins> Ok. apk fix seemed to do the right thing, which is good. I'm learning, slowly. Thanks.
2022-08-09 15:27:06 <panekj> if there is any error, apk fix will help
2022-08-09 15:27:13 <panekj> or at least tell you again what is wrong
2022-08-09 15:27:53 <adhawkins> Got it, will remember that in future. Was looking for an 'apk upgrade -v' or similar to give me more information.
2022-08-09 15:27:56 <adhawkins> Thanks again panekj and ptrc
2022-08-09 16:09:09 <psykose> i think rv is going to be stuck on rust every damn time huh
2022-08-09 16:09:23 <ikke> psykose: yeah, looks like it
2022-08-09 17:42:07 <Ermine> Is MPL-1.0 considered non-free?
2022-08-09 17:42:39 <Ermine> spdx.org says it is approved by OSI but not by FSF
2022-08-09 17:44:17 <psykose> no
2022-08-09 19:29:11 <psykose> Ermine: also where do you see that? https://wiki.gentoo.org/wiki/License_groups/FSF-APPROVED for instance references it, and the official gnu page has it, though admittedly only 1.1 and 2
2022-08-09 19:29:18 <psykose> i don't think anything differs though
2022-08-09 19:30:37 <psykose> 1.1 adds some commercial wording at a glance
2022-08-09 19:31:31 <Ariadne> Ermine: alpine itself has not made a determination either way
2022-08-09 20:24:58 <Ermine> psykose: at spdx.org/licenses
2022-08-09 20:25:21 <psykose> ah
2022-08-09 22:04:53 <user2> i just added the 'linux-firmware' apk to my custom iso, but it increased its size by a lot
2022-08-09 22:05:12 <user2> which means that it can no longer fit on a 2G unionfs properly
2022-08-09 22:06:00 <user2> is there really no way to save some stuff on the usb drive or something? more than 2G for the unionfs for such a lightweight system is very overkill
2022-08-09 22:06:18 <panekj> just don't add whole linux-firmware
2022-08-09 22:07:12 <user2> i guess thats an option too
2022-08-09 22:08:36 <psykose> hm
2022-08-09 22:08:39 <panekj> \o/
2022-08-09 22:08:44 <panekj> nice
2022-08-09 22:10:34 <psykose> >server closed connection
2022-08-09 22:10:37 <psykose> oftc.png
2022-08-09 22:10:44 <psykose> user2: you should find what firmware you use
2022-08-09 22:15:17 <psykose> algitbot: ping
2022-08-09 22:15:33 <psykose> algitbot: what
2022-08-09 22:16:15 <psykose> algitbot: how to
2022-08-09 22:17:27 <psykose> algitbot: give me panekj 
2022-08-09 22:17:52 <psykose> surprised that even works
2022-08-09 22:18:34 <psykose> google cats
2022-08-09 22:18:46 <psykose> ree
2022-08-09 22:19:05 <psykose> lmgtfy cats
2022-08-09 22:19:30 <panekj> he wish
2022-08-09 22:19:39 <panekj> algitbot: deny
2022-08-09 22:19:58 <psykose> algitbot: why panekj stinky
2022-08-09 22:20:21 <panekj> algitbot: how to make psykose go sleep
2022-08-09 22:20:33 <psykose> algitbot: hello
2022-08-09 22:24:11 <psykose> algitbot: hello there
2022-08-09 22:24:16 <psykose> there we go
2022-08-09 22:59:42 <user2> psykose: it already works on my hardware even without linux-firmware, but not on other laptops i tested
2022-08-09 23:00:04 <user2> the wifi card was not detected
2022-08-09 23:00:19 <user2> so i thought i would just install all the available firmware to avoid trouble
2022-08-09 23:00:28 <psykose> yeah, and that gives you all the firmware :)
2022-08-09 23:00:34 <psykose> you could figure out which have the actual wifi firmware
2022-08-09 23:00:49 <psykose> -other has a bunch of iwlwifi stuff, then the rtl* ones for a bunch of realtek
2022-08-09 23:01:49 <psykose> the -extended image by default has firmware in it, and you can see how huge it is; there's no way to make a portable 'wifi must work everywhere' kinda thing without ending up with a huge image
2022-08-09 23:01:59 <psykose> but i think if you find the right files it won't be 2gb big
2022-08-09 23:02:17 <user2> maybe i can just remove the heavy firmware that i wont use, since there's so much wifi firmware
2022-08-09 23:02:28 <psykose> shrug
2022-08-09 23:02:34 <user2> the image itself weighs 890mbs, but when it installs everything it doesnt fit in the unionfs
2022-08-09 23:02:34 <psykose> i'd just find the top 4 vendors or something
2022-08-09 23:02:38 <psykose> it depends on your usecase
2022-08-09 23:02:56 <user2> might just do that
2022-08-09 23:03:06 <minimal> user2: or add all the wifi firmware packages rather than all general firmware packages
2022-08-09 23:03:50 <user2> yeah, i think that's the only way probably
2022-08-09 23:04:05 <psykose> -rtlwifi -other -ath10k -ath11k -ath6k -ath9k_htc 
2022-08-09 23:04:26 <user2> should i be worried about installing gpu stuff from linux-firmware-* ? or does it just work by default
2022-08-09 23:04:36 <psykose> should work without
2022-08-09 23:04:40 <user2> nice
2022-08-09 23:04:56 <psykose> (for the random shitty screen, maybe not actual graphics)
2022-08-09 23:05:36 <user2> i guess i wont be able to find out, since i am using integrated gpu
2022-08-09 23:05:48 <user2> for now i wont install them
2022-08-09 23:25:09 <user2> wait a second, linux-lts depends on linux-firmware
2022-08-09 23:25:31 <user2> so all this firmware must already be there, and the wifi card is just not supported
2022-08-09 23:27:50 <user2> i dont understand though, if it's already a dependency then how is it getting installed again? making the iso bigger
2022-08-09 23:30:08 <psykose> it doesn't, it depends on -any
2022-08-09 23:30:18 <psykose> and then i guess something adds -none
2022-08-09 23:30:52 <psykose> or, hm
2022-08-09 23:31:26 <user2> https://pkgs.alpinelinux.org/package/edge/main/x86_64/linux-lts
2022-08-09 23:31:26 <psykose> maybe mkimg is magic
2022-08-09 23:31:32 <user2> yeah, i dont know
2022-08-09 23:31:39 <psykose> that page doesn't say what the actual dependencies are
2022-08-09 23:31:47 <user2> it says that it depends on linux-firmware and mkinitfs
2022-08-09 23:31:48 <user2> oh
2022-08-09 23:31:57 <psykose> apk info -R linux-lts would
2022-08-09 23:32:17 <user2> ooh
2022-08-09 23:32:19 <psykose> (the page gives the first satisfier of the list from that)
2022-08-09 23:32:28 <user2> yeah, now it says linux-firmware-any
2022-08-09 23:32:37 <psykose> because i guess nobody ever wrote the support for virtuals into it
2022-08-09 23:32:41 <psykose> but that would be really hard
2022-08-09 23:32:48 <psykose> since they don't exist
2022-08-09 23:33:07 <psykose> and it tracks packages that exist
2022-08-09 23:34:03 <user2> i will have to read how mkimage does things now
2022-08-09 23:34:50 <user2> even during the build it says "kernel x86_64 lts <some code> linux-lts linux-firmware wireless-regdb xtables-addons-lts"
2022-08-10 04:58:00 <Danct12> what happens when you do apk upgrade and one of the file in a package is completely missing?
2022-08-10 04:58:59 <Danct12> im currently working on a package (that won't be upstreamable) which rely on fw_helper_fallback to load a nvram firmware since it's nvram is stored in a nand partition
2022-08-10 05:02:48 <ikke> Danct12: missing where?
2022-08-10 05:03:34 <Danct12> ikke: i mean, intentionally missing as the user moved/deleted it
2022-08-10 05:03:47 <ikke> apk would restore it
2022-08-10 05:04:02 <ikke> if it's in the new package
2022-08-10 05:09:25 <Danct12> alright, got it
2022-08-10 05:09:52 <Danct12> im going to write a trigger file to delete it everytime it's restored
2022-08-10 11:18:05 <Ermine> I'm having the following problem with meson: "ERROR: Automatic wrap-based subproject downloading is disabled" . Will it be satisfied if I package and install relevant dependency? Or I need to patch meson.build so it looks for dependency?
2022-08-10 12:35:30 <psykose> Ermine: it means you are either missing the system dependency, or you have to submodule something
2022-08-10 12:44:09 <Ermine> psykose: the line which gives me an error calls subproject()
2022-08-10 12:44:18 <psykose> just post the file
2022-08-10 12:45:04 <Ermine> https://github.com/open5gs/open5gs/blob/main/lib/diameter/common/meson.build#L37
2022-08-10 12:45:50 <psykose> you have to submodule it
2022-08-10 12:46:13 <Ermine> ok
2022-08-10 12:53:14 <psykose> ikke: can you check the global git config on the builders
2022-08-10 12:54:31 <ikke> Later 
2022-08-10 12:58:04 <psykose> sure thing
2022-08-10 12:58:23 <psykose> (something probably added some stuff into there which is why the git-lfs thing is magically needed)
2022-08-10 12:58:34 <psykose> they should probably not be writable
2022-08-10 15:14:10 <psykose> i was just looking at a gdk-pixbuf update where one of the commits is to fix some jpeg-memory related stuff
2022-08-10 15:14:25 <psykose> and the commit for it on their gitlab.. has a test jpeg that allocates an infinite amount of memory
2022-08-10 15:14:38 <psykose> so if you don't have the updated gdk-pixbuf.. your memory usage goes to 100% if you open the commit
2022-08-10 15:14:39 <psykose> lmao
2022-08-10 15:15:11 <psykose> open this in firefox if you want to run out of memory (dos): https://gitlab.gnome.org/GNOME/gdk-pixbuf/-/commit/b659038e4296534c2e068de9bce8d9e17fbe58b4
2022-08-10 15:15:15 <psykose> hilarious
2022-08-10 15:16:29 <psykose> or maybe it's not even related to gdk-pixbuf for firefox reading the issue, but it's still funny
2022-08-10 15:19:34 <psykose> ah, no, it's libjpeg directly there
2022-08-10 16:45:47 <bl4ckb0ne> what happened to the dotet package?
2022-08-10 16:46:16 <panekj> it probably doesn't build
2022-08-10 16:46:30 <panekj> or is just in progress
2022-08-10 16:46:50 <bl4ckb0ne> where can i see the builder status?
2022-08-10 16:46:57 <bl4ckb0ne> i remember seeing a page for it somewhere
2022-08-10 16:47:02 <panekj> build.alpinelinux.org
2022-08-10 16:47:06 <panekj> https://build.alpinelinux.org/buildlogs/build-edge-x86_64/community/dotnet6-stage0/dotnet6-stage0-6.0.108-r0.log
2022-08-10 16:47:33 <bl4ckb0ne> yay, errors
2022-08-10 16:51:37 <psykose> it was disabled and will be back when it passes
2022-08-10 16:51:44 <psykose> usually takes 2 tries on x86_64
2022-08-10 16:54:47 <bl4ckb0ne> i pushed an upgrade for jellyfin, ill keep an eye on the package to restart my pipeline
2022-08-10 22:04:44 <ayakael> Sorry guys, had some *very* dotnet specific issues relating to RID being weird cause of a change in `/etc/os-release`. It should be up now on `edge`, although the hack I had to implement might have some consequences. Feel free to DM me if  the `jellyfin` update has some issues relating to an RID bug or something.
2022-08-10 22:34:02 <bl4ckb0ne> i already made the pull request for the bump, i think psykose rebased it to run the ci
2022-08-10 22:34:43 <bl4ckb0ne> ill try the package locally 
2022-08-10 23:06:25 <bl4ckb0ne> ayakael: ci fails https://gitlab.alpinelinux.org/alpine/aports/-/jobs/797608
2022-08-11 05:04:00 <ayakael> bl4ckb0ne: copy that, investigating further tomorrow 
2022-08-11 05:11:57 <ayakael> bl4ckb0ne: Ive reproduced the bug on my system. Off the bat, not quite sure if this is related to my mitigation for edge's RID bug as seen [here](https://github.com/dotnet/runtime/issues/73525)
2022-08-11 05:12:24 <ayakael> But I will look into this further - they are some other issues that need resolving tomorrow.
2022-08-11 07:54:34 <Ermine> Waypipe-0.8.3 test suite fails on alpine s390x but passes on debian s390x. May it be a musl issue?
2022-08-11 08:04:57 <bl4ckb0ne> whats the failurr
2022-08-11 09:16:37 <Ermine> bl4ckb0ne: https://gitlab.alpinelinux.org/Ermine/aports/-/jobs/795021
2022-08-11 09:39:34 <bl4ckb0ne> whack
2022-08-11 09:39:38 <bl4ckb0ne> maybe worth reporting it
2022-08-11 09:40:20 <Ermine> already reported it: https://gitlab.freedesktop.org/mstoeckl/waypipe/-/issues/65
2022-08-11 12:50:06 <psykose> Ariadne: think i have another pkgconf issue
2022-08-11 12:50:24 <psykose> (resolved by downgrading to <1.9)
2022-08-11 12:51:21 <psykose> https://img.ayaya.dev/t8EdbMbEMe5o
2022-08-11 12:52:10 <psykose> i assume only-I is what is used for cmake x_INCLUDE_DIRS, and that's what it's using from these and then the includes are missing so <> headers fail to import
2022-08-11 12:52:12 <psykose> etc etc
2022-08-11 12:52:46 <psykose> if you want a reproducing package, community/headlines
2022-08-11 12:53:21 <Ariadne> ok
2022-08-11 12:53:40 <Ariadne> will look into it at some point today
2022-08-11 12:53:42 <psykose> sure thing
2022-08-11 12:53:57 <Ariadne> i suspect this is underspecified .pc file though
2022-08-11 12:54:13 <psykose> it may be
2022-08-11 12:54:51 <psykose> i mean, the libadwaita one is indeed barren
2022-08-11 12:54:59 <psykose> funny how it somehow returned everything before though
2022-08-11 12:55:02 <Ariadne> every time i try to tighten the solver, we go through a cycle of this
2022-08-11 12:55:05 <psykose> aha
2022-08-11 12:55:10 <psykose> ok, i will check out the pcs too
2022-08-11 12:55:19 <Ariadne> use pkgconf --dot
2022-08-11 12:55:26 <Ariadne> or is it --digraph
2022-08-11 12:55:27 <Ariadne> i forget
2022-08-11 12:57:24 <psykose> hmm i think you're right
2022-08-11 12:57:31 <psykose> because the missing include is in the .private section
2022-08-11 12:57:41 <psykose> for some reason
2022-08-11 12:57:44 <psykose> very much not private
2022-08-11 12:58:41 <psykose> if you want the dots anyway: https://img.ayaya.dev/j2bMlKFZPFwH
2022-08-11 13:04:08 <Ariadne> hmm
2022-08-11 13:04:26 <Ariadne> okay
2022-08-11 13:05:10 <Ariadne> check if --cflags --fragment-filter=I works
2022-08-11 13:05:12 <Ariadne> :D
2022-08-11 13:09:01 <psykose> identical output to `--cflags-only-I`
2022-08-11 13:09:19 <Ariadne> okay
2022-08-11 13:09:25 <psykose> but yeah it's the .privates section having stuff that shouldn't be there :/
2022-08-11 13:09:27 <psykose> gtk+3.0 is the same
2022-08-11 13:09:40 <Ariadne> then most likely PKGCONF_PKGF_PKG_SEARCH_PRIVATE is not getting enabled before the solver actually runs
2022-08-11 13:09:40 <psykose> gio-unix is in privates so it gets skipped
2022-08-11 13:09:47 <psykose> yeah
2022-08-11 13:09:56 <Ariadne> pkgconf/cli/main.c
2022-08-11 13:10:03 <Ariadne> you're smart, i'm sure you can figure it out :D
2022-08-11 13:10:11 <Ariadne> also feel free to refactor that entire thing :D
2022-08-11 13:10:13 <psykose> am dumb
2022-08-11 13:11:04 <Ariadne> the problem is that Requires.private was designed by, well, people who don't know what they are doing
2022-08-11 13:11:43 <Ariadne> most of pkg-config design is bad, actually.  we are trying to create replacement functionality for stuff like Requires.private, which has inconsistent behavior
2022-08-11 13:12:11 <psykose> i never knew the difference so i assumed 'private' was just a list for static linking but not part of the public interface
2022-08-11 13:12:30 <psykose> so all of these are actually wrong? the headers are used even though they're in private
2022-08-11 13:13:06 <Ariadne> they are "correct" in terms of pkg-config definition
2022-08-11 13:13:17 <Ariadne> the feature is designed poorly, however
2022-08-11 13:13:25 <Ariadne> there should be
2022-08-11 13:13:30 <Ariadne> Requires.includes: ...
2022-08-11 13:13:33 <Ariadne> Requires.libs: ...
2022-08-11 13:13:39 <Ariadne> Requires.staticlibs: ...
2022-08-11 13:14:06 <psykose> yea
2022-08-11 13:14:25 <Ariadne> anyway, problem is most likely that PKGCONF_PKGF_PKG_SEARCH_PRIVATE is not being set before pkgconf_queue_solve() gets called
2022-08-11 13:14:30 <psykose> looking at the cli/main.c SEARCH_PRIVATE is only set for various want static
2022-08-11 13:14:38 <Ariadne> yep
2022-08-11 13:14:58 <psykose> but i don't think anything actually wants it to be static.. weh
2022-08-11 13:15:10 <Ariadne> change it to enable SEARCH_PRIVATE for the solution if cflags is requested, and then suppress that for dumping the libs
2022-08-11 13:16:29 <Ariadne> this worked in 1.8 because the solver just got invoked multiple times
2022-08-11 13:16:33 <Ariadne> first for cflags, then for libs
2022-08-11 13:16:54 <Ariadne> but 1.9 moves to solving the entire thing up front
2022-08-11 13:17:31 <Ariadne> otherwise, i'll do it in a couple hours
2022-08-11 13:18:59 <Ariadne> i think the latter is already done (~PKGCONF_PKGF_PKG_SEARCH_PRIVATE)
2022-08-11 13:20:11 <Ariadne> what would be nice is some API to set/unset individual flags
2022-08-11 13:20:57 <psykose> hmmm
2022-08-11 13:24:33 <psykose> i definitely hacked it to work
2022-08-11 13:24:36 <psykose> does not look very good though
2022-08-11 13:25:26 <psykose> https://img.ayaya.dev/OS90HWP9RqVp if this is what you mean
2022-08-11 13:25:32 <psykose> it does fix the --cflags-only-I output to include the missing stuff
2022-08-11 13:27:40 <psykose> (the cleanup for libs was already there as you noted)
2022-08-11 13:39:01 <Ariadne> psykose: should do it prior to pkgconf_queue_solve, and if CFLAGS are requested at all, but yes
2022-08-11 13:39:34 <Ariadne> there is `want_client_flags`
2022-08-11 13:40:44 <psykose> aha
2022-08-11 13:41:08 <Ariadne> pkgconf developer(tm) psykose to the rescue
2022-08-11 13:41:09 <Ariadne> 
ACTION ducks

2022-08-11 13:41:19 <psykose> this is bullying
2022-08-11 13:41:42 <Ariadne> the entire pkg-config ecosystem is bullying ;/
2022-08-11 13:42:12 <Ariadne> gaslight, gatekeep, pkg-config or something
2022-08-11 13:43:25 <psykose> https://img.ayaya.dev/82rNSQfO5wbq happy? :)
2022-08-11 13:43:34 <psykose> but yeah
2022-08-11 13:43:36 <psykose> cflags not I
2022-08-11 13:43:57 <psykose> https://img.ayaya.dev/3gZPU8OwTA6Q
2022-08-11 13:44:00 <psykose> now i have cleaning to do
2022-08-11 13:44:46 <Ariadne> yeah
2022-08-11 13:44:48 <Ariadne> this looks fine
2022-08-11 13:44:53 <Ariadne> feel free to do a pkgconf NMU
2022-08-11 13:45:05 <psykose> you want me to just throw that in or actually PR it to gitea first
2022-08-11 13:45:08 <Ariadne> i'll cut a 1.9.3 release later today or tomorrow with some SPDX stuff
2022-08-11 13:45:13 <Ariadne> whatever you wanna doooo
2022-08-11 13:45:17 <psykose> ite
2022-08-11 13:46:03 <Ariadne> i can review on gitea really quick
2022-08-11 13:46:24 <Ariadne> but i have a meeting in 15 minutes introducing ikke to some of my engineers who want to take his alpine/go project for a ride it will never forget
2022-08-11 13:46:57 <Ariadne> wait
2022-08-11 13:47:02 <Ariadne> maybe i shouldn't have put it that way :D
2022-08-11 13:47:21 <ikke> :D
2022-08-11 13:48:10 <psykose> https://gitea.treehouse.systems/ariadne/pkgconf/pulls/243
2022-08-11 13:51:04 <psykose> giga merge
2022-08-11 13:51:06 <Ariadne> release 1.9.3 probably tomorrow afternoon
2022-08-11 13:51:11 <Ariadne> feel free to NMU in meantime
2022-08-11 13:51:13 <psykose> back to roleplaying a housewife i guess
2022-08-11 13:51:14 <psykose> yea already did
2022-08-11 13:51:19 <psykose> along with 1000 boost rebuilds
2022-08-11 13:51:21 <psykose> 1.80 is here
2022-08-11 13:51:44 <Ariadne> (i have an SPDX meeting tomorrow and want some feedback on the SPDX work I am doing in pkgconf 1.9)
2022-08-11 13:52:15 <psykose> the bom stuff is very cool
2022-08-11 13:53:01 <Ariadne> we have better stuff in kubernetes ecosystem, but we will get there for C :)
2022-08-11 15:06:21 <Ariadne> psykose: https://gitea.treehouse.systems/ariadne/pkgconf/commit/68b5cab72d6529a665dcc9882a9197f3b586927b
2022-08-11 15:06:35 <psykose> heh
2022-08-11 15:07:37 <psykose> i did see that but there was too much boolean logic there so i just skipped it
2022-08-11 15:07:38 <psykose> : )
2022-08-11 15:07:49 <psykose> thanks bestie
2022-08-11 15:09:40 <ajhalili2006> Is there an API I can use to check build status of an package after it's merged to the main branch (aka master branch) and deployed to dl-cdn?
2022-08-11 15:10:11 <ikke> No, not in that sense
2022-08-11 15:10:18 <psykose> build.a.o says what the builders are currently doing, but there's no individual granularity
2022-08-11 15:10:20 <ikke> dl-cdn is just one mirror
2022-08-11 15:12:25 <ajhalili2006> psykose maybe i should poll wss://msg.alpinelinux.org/mqtt and parse mqtt using some tools? (i used DevTools on builds.a.o)
2022-08-11 15:12:32 <psykose> what for
2022-08-11 15:12:54 <Ariadne> a build being complete does not mean it is on dl-cdn
2022-08-11 15:13:06 <psykose> yeah, that's at the end of the repo finishing building only
2022-08-11 15:13:13 <ikke> and you don't get information about single packages being built
2022-08-11 15:13:24 <psykose> after all failures, so on retry it's still not there, has to actually finish
2022-08-11 15:13:37 <psykose> seeing 'x building' on mqtt and then finishing doesn't mean much
2022-08-11 15:13:51 <psykose> also, i'm curious if there's any downside to removing per-repo upload
2022-08-11 15:14:02 <psykose> instead of main/ community/ testing/, upload only when all three are done
2022-08-11 15:14:03 <Ariadne> even if a build is complete, dl-cdn is fed from a tier 1 mirror, not the master
2022-08-11 15:14:09 <psykose> it would avoid all the openssl-level breaks
2022-08-11 15:14:17 <psykose> and not be much of a delay 90% of the time
2022-08-11 15:14:29 <Ariadne> what if we just did not ever touch openssl :D
2022-08-11 15:14:36 <psykose> we can't be on 1.1 forever
2022-08-11 15:14:40 <Ariadne> sure we can
2022-08-11 15:14:43 <psykose> and the only reason anyone broke is that
2022-08-11 15:14:45 <ikke> psykose: if it's says master uploaded, it does mean master has been uploaded
2022-08-11 15:14:46 <psykose> that main uploads first
2022-08-11 15:15:00 <ikke> so any packages built until then for that repo should be available after sync
2022-08-11 15:15:37 <psykose> do you mean that if you "dont" upload, the self-upgrade to use main/ for subsequent builds doesn't work
2022-08-11 15:15:37 <ikke> Oh, you mean the upload itself, not the messages
2022-08-11 15:15:43 <psykose> i mean only the uploads, yes
2022-08-11 15:16:08 <psykose> generally there is no downside i can think of to uploading only after testing/, except some very special cases where there is something important in main/ and some shit has been failing in testing/ repeatedly
2022-08-11 15:16:11 <ikke> the self update would ofcourse still work, that does not depend on uploading
2022-08-11 15:16:22 <psykose> but as long as someone watches it that doesn't matter, it's easy to fix/disable things as needed
2022-08-11 15:16:52 <Ariadne> anyway, there is more stuff broken by being rebuilt with openssl 3, because they require stuff that is now in the legacy plugin thingy
2022-08-11 15:17:01 <Ariadne> so they just like segfault or worse
2022-08-11 15:17:05 <Ariadne> when you go to use them
2022-08-11 15:17:08 <psykose> if anything all the constant world-rebuilds for things in main/ are messy because of the split behaviour, and imo it's a lesser-of-two-evils to upload all combined
2022-08-11 15:17:22 <psykose> which legacy plugin specifically
2022-08-11 15:17:37 <psykose> most of the ones i've seen would just fail to start instead, or fail to rebuild since the apis are not there
2022-08-11 15:17:40 <Ariadne> https://www.openssl.org/docs/manmaster/man7/OSSL_PROVIDER-legacy.html
2022-08-11 15:17:51 <psykose> ah that
2022-08-11 15:18:06 <Ariadne> i cannot say i am a fan of any of this
2022-08-11 15:18:14 <psykose> can you name a broken package that uses that with openssl3 but segfaults instead of succeeds
2022-08-11 15:18:20 <Ariadne> openconnect
2022-08-11 15:18:26 <psykose> that doesn't segfault
2022-08-11 15:18:37 <Ariadne> it fails to work because it hasn't been updated to use the legacy plugin
2022-08-11 15:18:42 <psykose> it has
2022-08-11 15:18:45 <Ariadne> oh, has it now?
2022-08-11 15:18:47 <psykose> you have to pass the --insecure flag
2022-08-11 15:18:48 <psykose> yes
2022-08-11 15:18:55 <psykose> very intentionally mind you
2022-08-11 15:19:07 <psykose> openssl3 does not allow non-rfc-whatever tls renegotiation without legacy plugin
2022-08-11 15:19:14 <psykose> i spent a few hours debugging that one
2022-08-11 15:19:19 <psykose> and yeah, it works with the insecure flag
2022-08-11 15:19:31 <Ariadne> pass --insecure flag to what, though
2022-08-11 15:19:50 <psykose> https://img.ayaya.dev/m7ZqRPLW0ssh
2022-08-11 15:20:00 <psykose> see? works
2022-08-11 15:20:15 <psykose> as intended; that's a "breaking change" but i don't mind openssl3 being strict about it
2022-08-11 15:20:47 <psykose> if you can find actual segfault failures (there *has* been like 3) then report them, i'll fix them all
2022-08-11 15:21:16 <Ariadne> the last one i knew about was in apk-tools because they did not bother to even hook up MD_NULL at all
2022-08-11 15:21:32 <Ariadne> but they sure kept the API around!
2022-08-11 15:21:36 <Ariadne> very good footgun, that
2022-08-11 15:21:48 <psykose> yes, that one was a legitimate openssl3 bug (iirc it even got fixed after a few months of you reporting it)
2022-08-11 15:21:52 <psykose> we both know the api sucks :)
2022-08-11 15:21:59 <Ariadne> wow a whole few months!
2022-08-11 15:21:59 <psykose> but i'm just talking from the packaging perspective and aports
2022-08-11 15:22:14 <Ariadne> i am just saying i expect there to be additional fallout
2022-08-11 15:22:17 <psykose> there will be
2022-08-11 15:22:30 <Ariadne> and users demanding guillotines for alpine developers (rather than openssl developers)
2022-08-11 15:22:33 <psykose> there's a few things using dlopen() currently that want 1.1 only but don't depend on it, so it might be missing and fail
2022-08-11 15:22:42 <psykose> what even was that tweet lol, it was deleted
2022-08-11 15:22:53 <Ariadne> idk some pissed off docker user
2022-08-11 15:23:06 <ikke> due to opeensl?
2022-08-11 15:23:18 <psykose> yeah but i mean the wording, wtf were they saying
2022-08-11 15:23:21 <Ariadne> no, i think he was mad because he had to rebuild all his images due to zlib CVE
2022-08-11 15:23:25 <psykose> loool
2022-08-11 15:23:31 <psykose> yeah really sorry i fixed that :( 
2022-08-11 15:23:36 <psykose> better to never updated anything
2022-08-11 15:23:48 <Ariadne> that's the spirit
2022-08-11 15:23:50 <ikke> That's endenic to docker
2022-08-11 15:23:52 <psykose> :)
2022-08-11 15:23:58 <ikke> endemic if I can type
2022-08-11 15:24:02 <psykose> but yeah openssl3 is a mess, i don't disagree
2022-08-11 15:24:14 <psykose> i think """long term""" it's better to kinda migrate to it than be locked to 1.1 forever
2022-08-11 15:24:16 <Ariadne> ikke: yeah one of the key talking points we use with customers is that they should be rebuilding their images nightly
2022-08-11 15:24:19 <psykose> nobody has any other alternatives
2022-08-11 15:24:49 <ikke> Ariadne: I probably should switch the rebuilds for our infra images from weekly to nightly
2022-08-11 15:24:50 <psykose> there's that weird openssl-api-implementing microsoft ssl thing, but nobody has tested it at distro scale that i know of
2022-08-11 15:24:57 <psykose> ikke: yeah
2022-08-11 15:25:23 <psykose> also helps with the ci time for a few seconds, since there's less to upgrade
2022-08-11 15:25:24 <psykose> hah
2022-08-11 15:25:27 <Ariadne> (and we even have a product that, among other things, alerts you when you have images in production that are too old)
2022-08-11 15:25:48 <ajhalili2006> ikke are you talking about the headaches of reproducible containers/builds lately? (and yes, it's pain)
2022-08-11 15:26:01 <Ariadne> reproducible containers is easy if you use alpine
2022-08-11 15:26:09 <Ariadne> just apk add apko and go to town
2022-08-11 15:26:10 <psykose> ikke: but yeah what do you think of the idea of uploading after testing/?
2022-08-11 15:26:15 <ikke> ajhalili2006: just making sure they are up-to-date
2022-08-11 15:26:18 <psykose> i can't think of a downside worse than the current behaviour
2022-08-11 15:26:28 <psykose> imo it's a big +
2022-08-11 15:27:14 <Ariadne> downside: CVE update gets blocked because some package in testing has bitrotted
2022-08-11 15:27:20 <ikke> It probably would force us to disable a lot more packages
2022-08-11 15:27:46 <ikke> (and everything depending on it if you do it right)
2022-08-11 15:29:37 <Ermine> openssl again?
2022-08-11 15:29:56 <Ariadne> Ermine: i will continue to complain about openssl until it is yeeted from alpine
2022-08-11 15:30:22 <psykose> yeah, but that's extremely rare compared to the 'rebuild 100 things' rate
2022-08-11 15:30:25 <Ermine> you'll do it for a reason
2022-08-11 15:31:35 <psykose> (because most of the time failed things aren't left there; the queue is just empty, cve update is pushed, then it uploads)
2022-08-11 15:31:52 <ikke> Ariadne: We can yeet it once we have a proper replacement :P
2022-08-11 15:31:53 <psykose> the implication is only if it just so happens to have a bitrotted failure and also there is some cve update at the exact same time
2022-08-11 15:32:24 <ikke> psykose: those things do happen ;-)
2022-08-11 15:32:43 <psykose> at a much lower rate than the alternative of cross-repo rebuilds, yes
2022-08-11 15:33:20 <psykose> Ariadne: while you're here can you press merge on gcc12 :D
2022-08-11 15:33:37 <Ariadne> psykose: is alpine-gcc-patches rebased?
2022-08-11 15:33:59 <psykose> probably? nmeum's mr has been there forever
2022-08-11 15:34:24 <psykose> and the rebizassing is in https://gitlab.alpinelinux.org/ariadne/alpine-gcc-patches/-/merge_requests/12
2022-08-11 15:34:34 <psykose> that's the same target as aports afaik
2022-08-11 15:35:07 <psykose> then again it says conflicts so probably not
2022-08-11 15:35:15 <Ariadne> hmmph
2022-08-11 15:35:32 <psykose> there's also new binutils
2022-08-11 15:35:39 <psykose> but i can NMU that for you if you wish
2022-08-11 15:36:25 <ikke> yay, more bugs :P
2022-08-11 15:37:01 <psykose> new binutils seems to have jansson support for json package metadata, but we can opt for that later if wanted
2022-08-11 15:37:17 <Ariadne> i'm hesitant to merge gcc 12 if it means we lose gdc on 32-bit
2022-08-11 15:37:23 <Ariadne> if D is going, i want it to go 100%
2022-08-11 15:37:31 <Ariadne> and i want it to not come back, ever
2022-08-11 15:37:32 <psykose> that is very far away
2022-08-11 15:37:37 <psykose> like, it needs a month of discussion
2022-08-11 15:37:47 <psykose> currently, all the D targets are already x86_64/aarch64
2022-08-11 15:37:51 <psykose> because ldc is blocked to there
2022-08-11 15:37:53 <psykose> and everything uses ldc
2022-08-11 15:38:00 <Ariadne> hmm, okay
2022-08-11 15:38:11 <psykose> they don't support 32 themselves, haha
2022-08-11 15:38:23 <psykose> for actual D removal, there's only 2 aports and appstream-generator
2022-08-11 15:38:24 <Ariadne> sure, why not
2022-08-11 15:38:34 <psykose> appstream-generator is used by our infrastructure, that needs discussion with the gnome people
2022-08-11 15:38:35 <psykose> (afaik)
2022-08-11 15:38:40 <ikke> nod
2022-08-11 15:38:47 <psykose> but again, 64 only
2022-08-11 15:39:33 <psykose> and there's also the -O2 CXXFLAGS for gcc
2022-08-11 15:39:50 <Ariadne> ...
2022-08-11 15:39:56 <Ariadne> just NMU that
2022-08-11 15:39:59 <psykose> sure
2022-08-11 15:40:06 <psykose> oh and there's openssl3 KTLS
2022-08-11 15:40:08 <psykose> : )
2022-08-11 15:40:19 <psykose> and that's my entire backlog of asking you about stuff
2022-08-11 15:40:41 <psykose> ( !37294 )
2022-08-11 15:40:44 <Ariadne> (new binutils where?  the latest i see is 2.39, which we have)
2022-08-11 15:40:57 <ikke> psykose: oh btw, there is indeed lfs stuff in the config
2022-08-11 15:41:02 <psykose> we don't have 2.39
2022-08-11 15:41:06 <psykose> we have 2.38
2022-08-11 15:41:13 <psykose> .39 came out like a week ago
2022-08-11 15:41:32 <psykose> ikke: it should all be deleted (it's safe)
2022-08-11 15:41:43 <Ariadne> i thought we have 2.39
2022-08-11 15:41:43 <ikke> yes
2022-08-11 15:41:45 <psykose> though do you know of a way to forbid things being written to global git? any build can just call stuff
2022-08-11 15:42:03 <Ariadne> hmm, so we don't
2022-08-11 15:42:08 <Ariadne> did they fix the manpages this time
2022-08-11 15:42:12 <psykose> oh god
2022-08-11 15:42:38 <psykose> they look like they work
2022-08-11 15:42:48 <Ariadne> feel free to NMU that as well
2022-08-11 15:42:58 <psykose> alright
2022-08-11 15:43:02 <psykose> (and ktls?)
2022-08-11 15:43:10 <Ariadne> i pushed it
2022-08-11 15:43:16 <psykose> oki
2022-08-11 15:43:19 <psykose> me slow
2022-08-11 15:48:21 <ikke> any build / test that tries to modify the global git config now will face an immutable file
2022-08-11 15:49:37 <psykose> nice :)
2022-08-11 15:49:49 <psykose> i assume with chattr +i? does that persist cross reboot
2022-08-11 15:49:52 <psykose> i always forget
2022-08-11 15:50:54 <ikke> psykose: yes, it persists
2022-08-11 15:50:57 <ikke> just like file permisions
2022-08-11 15:50:59 <psykose> neat
2022-08-11 15:51:04 <ikke> they are stored in extended attributes
2022-08-11 15:55:31 <psykose> sam_: do you have a patch for the binutils 2.39 pwrite64/mmap64 stuff already
2022-08-11 15:56:08 <sam_> not yet! what's up (I've been trying out 2.39 but not on musl yet)
2022-08-11 15:56:50 <psykose> same old
2022-08-11 15:56:54 <psykose>    78 |   ssize_t (*pwrite64)();
2022-08-11 15:56:54 <psykose> ./../src/collector_module.h:78:13: error: duplicate member 'pwrite'
2022-08-11 15:56:58 <psykose> you know how it is
2022-08-11 15:57:37 <psykose> also i noticed gentoo elibc_musl passes 2M stacksize for musl but we don't
2022-08-11 15:57:44 <psykose> do you have the original report of where that was needed
2022-08-11 15:57:58 <psykose> (i've never noticed any bfd specific stack issues, but maybe they look funny)
2022-08-11 15:59:53 <psykose> also that failure is x86* and aarch64 only, lol
2022-08-11 16:00:49 <psykose> i guess gprofng isn't built on the others
2022-08-11 16:01:56 <sam_> gprofng is really brittle so far i think
2022-08-11 16:02:01 <psykose> ah, also RTLD_BINDING_MASK, and sigev._sigev_un._tid 
2022-08-11 16:02:07 <sam_> https://gitweb.gentoo.org/repo/gentoo.git/commit/sys-devel/binutils?id=a43643b92f154070540d0f7b84ca6e5a84ea84b3 but i don't have proof yet
2022-08-11 16:02:12 <sam_> if you remind me in a few hours or tomorrow ill go digging in irc logs
2022-08-11 16:02:13 <psykose> i remember fixing this months ago for binutils master
2022-08-11 16:02:14 <sam_> cause i wanna know too
2022-08-11 16:02:19 <Ariadne> just
2022-08-11 16:02:22 <Ariadne> pass it
2022-08-11 16:02:25 <Ariadne> 64MB stack
2022-08-11 16:02:26 <Ariadne> nbd
2022-08-11 16:02:30 <psykose> so true!!!!!!
2022-08-11 16:02:41 <Ariadne> 64GB stack
2022-08-11 16:02:44 <psykose> sam_: i would prefer the dig, because we've built huge things with default stack bfd
2022-08-11 16:02:55 <psykose> webkit, ceph, etc all used bfd with default for a long time
2022-08-11 16:03:03 <psykose> (former is now lld because i'm a weirdo)
2022-08-11 16:03:18 <Ariadne> what if we just used lld for everything
2022-08-11 16:03:26 <psykose> +1 from me but it fails on some things
2022-08-11 16:03:31 <sam_> ack
2022-08-11 16:03:33 <psykose> iirc it has a few more failures than gold
2022-08-11 16:04:07 <psykose> also it makes the default toolchain llvm so i just know you're trolling you silly goose
2022-08-11 16:04:12 <psykose> also there's no lld for s390x
2022-08-11 16:04:17 <psykose> and llvm is a bit immature for riscv64
2022-08-11 16:04:50 <psykose> but if we dropped those... hmmm....
2022-08-11 16:06:35 <psykose> int noload = RTLD_BINDING_MASK | RTLD_NOLOAD; //XXXX why RTLD_BINDING_MASK?
2022-08-11 16:06:36 <psykose> lol
2022-08-11 16:08:04 <Ariadne> what if we just used mold for everything
2022-08-11 16:11:13 <psykose> what if we hopped in the forest
2022-08-11 16:11:22 <Thermi> what if we all went on vacation
2022-08-11 16:11:26 <psykose> hmm, how does one port a DL_serinfo use
2022-08-11 16:11:40 <Hello71> i would rather have a moldy linux distro than an oxidized one
2022-08-11 16:19:11 <psykose> not sure how good an idea it is to just disable gprofng
2022-08-11 16:19:19 <Ariadne> i mean
2022-08-11 16:19:26 <Ariadne> it's next gen
2022-08-11 16:19:34 <Ariadne> can't drop it
2022-08-11 16:19:52 <psykose> >.>
2022-08-11 16:20:10 <psykose> it only builds on like 3 arches anyway lol, i assume the others are old gprof
2022-08-11 16:20:24 <Ariadne> tbh we can (and should) drop all of this crap :D
2022-08-11 16:20:33 <Ariadne> just use oldschool gprof
2022-08-11 16:20:37 <Ariadne> it works on musl
2022-08-11 16:20:51 <psykose> sure sure
2022-08-11 16:21:06 <psykose> then i just have to fix some rebuilds for libbfd
2022-08-11 16:21:28 <Ariadne> did you know libbfd is called that because RMS thought it was a BFD
2022-08-11 16:21:45 <psykose> is that the actual acronym
2022-08-11 16:21:48 <psykose> i never found what it means
2022-08-11 16:21:56 <Ariadne> the original acronym yes
2022-08-11 16:22:04 <Ariadne> they later came up with "binary format descriptor"
2022-08-11 16:22:20 <psykose> ld.binaryformatdescriptor
2022-08-11 16:22:20 <psykose> yes
2022-08-11 16:22:23 <psykose> i see
2022-08-11 17:53:27 <psykose> sam_: ah, right, since we're on gcc12 now- can you link the thing of things that fail with gcc12
2022-08-11 17:53:32 <psykose> all i remember is the zfs segfaults
2022-08-11 17:54:24 <sam_> https://bugs.gentoo.org/show_bug.cgi?id=gcc-12 (https://bugs.gentoo.org/showdependencytree.cgi?id=839906&hide_resolved=0 might be more useful, idk)
2022-08-11 17:54:34 <psykose> thanks! what i was looking for
2022-08-12 10:13:57 <pabloyoyoista> psykose, Ariadne, on the D/gdc problem for appstream-generator. gdc does not even compile in their upstream CI, only ldc works. And if D were to be removed, there is one alternative to appstream-generator, but is less conformant to the spec and less fitted in my opinion. Would be worst case if you have reasons to kill D
2022-08-12 10:15:14 <pabloyoyoista> Also, it's not just for gnome, but also used KDE discover :)
2022-08-12 10:15:29 <pabloyoyoista> s/used/used for/
2022-08-12 10:18:15 <Ariadne> but this is all so fragile
2022-08-12 16:21:59 <mlebd> Hi everyone. How can I submit a request to add my mirror mirror.lebedinets.ru ? alpine-mirrors mailing list does not work.
2022-08-12 18:47:33 <nmeum> psykose: do you know any aports which make heavy use of cgo from the top of your head?
2022-08-12 18:47:47 <psykose> hmmmmmm
2022-08-12 18:48:01 <nmeum> I was also wondering if we still need to force external link on ppc64le
2022-08-12 18:48:18 <nmeum> because they hopefully fixed the ppc64le linking issues in 1.19
2022-08-12 18:48:20 <psykose> that one should be fixed (and if it's not i'll repatch it)
2022-08-12 18:48:32 <nmeum> so maybe that's something we could also test by rebuilding some cgo aports on the ci
2022-08-12 18:49:36 <psykose> i'm not really sure what even counts as 'using cgo', maybe zabbix
2022-08-12 18:49:41 <psykose> (it does link openssl)
2022-08-12 18:50:13 <psykose> podman too
2022-08-12 18:53:13 <sam_> nmeum: hugo i think
2022-08-12 18:53:35 <nmeum> let's just bump the pkgrel of those and see what happens ':D
2022-08-12 18:53:56 <ikke> nmeum: zabbix (agent2)
2022-08-12 18:54:00 <ikke> not sure if that counts as heavy
2022-08-12 18:54:13 <ikke> oh lol
2022-08-12 18:54:17 <ikke> psykose: already mentioned that :D
2022-08-12 18:54:49 <nmeum> triggered a rebuild of hugo, zabbix and podman on my MR
2022-08-12 18:56:00 <nmeum> psykose: btw, the testsigfwd test case actually fails with `-lssp_nonshared` in LDFLAGS and without `-fno-stack-protector` in CFLAGS
2022-08-12 18:56:13 <psykose> same errors?
2022-08-12 18:58:02 <nmeum> https://tpaste.us/X0aD
2022-08-12 18:58:02 <nmeum> yes
2022-08-13 07:47:49 <ikke> Going to reboot gbr2, meaning things like pkgs.a.o and security.a.o will be unavailable for a bit
2022-08-13 07:55:19 <ikke> all done
2022-08-13 12:25:43 <ikke> https://gitlab.alpinelinux.org/alpine/aports/-/jobs/799872 default_dbg fails because it apparently tries to get sumbols from an s390x qemu image.
2022-08-13 12:26:11 <ikke> xen/usr/share/qemu-xen/qemu/s390-netboot.img
2022-08-13 12:26:17 <ikke> Not sure what to do about it
2022-08-13 12:29:04 <Ariadne> rm the s390-netboot.img
2022-08-13 12:29:11 <Ariadne> it is not remotely relevant to xen
2022-08-13 12:35:35 <ikke> ok
2022-08-13 12:36:29 <ikke> The package already does some custom stripping apparently
2022-08-13 12:36:33 <ikke> (ie, skipping /usr/share)
2022-08-13 13:58:40 <Ermine> psykose: I didnt squash commits!
2022-08-13 13:58:53 <psykose> xD
2022-08-13 13:59:13 <psykose> happens
2022-08-13 13:59:28 <panekj> ew
2022-08-13 16:24:37 <donoban> I have an error connecting to RDP with remmina since yesterday ( [19903:4124] [ERROR][com.freerdp.core] - transport_connect_tls:freerdp_set_last_error_ex ERRCONNECT_TLS_CONNECT_FAILED [0x00020008] )
2022-08-13 16:24:49 <donoban> is it likely to be related with openssl3?
2022-08-13 17:23:58 <psykose> probably, but that doesn't say the actual error
2022-08-13 23:17:52 <ptrc> donoban: this exact error happened to me way before when i couldn't connect to the server, so it might be unrelated to openssl3
2022-08-14 10:10:53 <omni> !37596 !37597 a test fail on s390x on edge but not on 3.16-stable
2022-08-14 13:22:30 <Hello71> does it work if you rerun it?
2022-08-14 13:29:06 <omni> no, I tried, and it also failed on the same test in !37539
2022-08-14 16:40:40 <Nulo> Anybody having issues pulling from gitlab.alpinelinux.org?
2022-08-14 16:40:52 <Nulo> error: RPC failed; HTTP 429 curl 22 The requested URL returned error: 429
2022-08-14 16:46:57 <ikke> 429 sounds ratelimit
2022-08-14 17:01:04 <ikke> Nulo: Is it intermittent, or continuously?
2022-08-14 17:01:37 <Nulo> ikke: continuously, and I haven't used it all day
2022-08-14 17:02:46 <ikke> I can at least clone via http
2022-08-14 17:02:58 <ikke> Nulo: what repository?
2022-08-14 17:03:53 <Nulo> Sorry I'm dumb, it wasn't gitlab, it was https://git.alpinelinux.org/aports/
2022-08-14 17:04:02 <Nulo> GitLab works fine
2022-08-14 17:04:11 <ikke> ooh
2022-08-14 17:05:42 <ikke> Nulo: What do you use to fetch?
2022-08-14 17:05:54 <Nulo> Uhh, git?
2022-08-14 17:05:58 <Nulo> `git fetch upstream`
2022-08-14 17:06:00 <ikke> ok
2022-08-14 17:06:29 <ikke> A long time ago we did block a specific user agent that kept hammering cgit
2022-08-14 17:06:39 <ikke> but that should not affact plain git
2022-08-14 17:07:36 <ikke> Nulo: can you try again, see if I can see something
2022-08-14 17:07:55 <Nulo> Works now
2022-08-14 17:08:12 <ikke> hmm, strange, didn't touch anything
2022-08-14 17:09:20 <Nulo> The worst kind of "works now" :P
2022-08-14 17:09:34 <Nulo> Or possibly the best depending on how you see it
2022-08-14 17:19:31 <ikke> Server is not busy or something, didn't find anything relevant in the logs either.
2022-08-14 17:22:55 <ptrc> it sometimes happens with cgit, we stumbled over it before but couldn't figure out why
2022-08-14 17:25:31 <psykose> the 'sometimes' was consistent
2022-08-14 17:26:02 <psykose> if i update the uwsgi cgit container from 3.14 to 3.16 it always 429's on every request
2022-08-14 17:27:31 <psykose> there's nothing in the changes between the uwsgi versions that seem meaningful and i never had any other ideas
2022-08-14 23:00:42 <psykose> is there something special about the image generation process? 3.16 linux-firmware-brcm has brcmfmac43436-sdio.bin in it, but the alpine-rpi-3.16.2-aarch64 generated image does not have it (while having the rest of the things from the -brcm package) in the modloop
2022-08-14 23:06:41 <psykose> ah, someone seems to have figured it out
2022-08-15 01:16:04 <psykose> ikke: can you check the riscv builder world set when you get a chance
2022-08-15 02:52:02 <psykose> and s390x
2022-08-15 02:52:07 <psykose> which has udev in world or whatever
2022-08-15 02:52:21 <psykose> and armhf
2022-08-15 02:57:11 <psykose> and ppc64le for pcsc-lite-dev
2022-08-15 03:12:14 <psykose> and even x86
2022-08-15 10:23:53 <ikke> psykose: what specifically on rv64?
2022-08-15 12:58:17 <omni> Ariadne: ny idea on why the test would fail on s390x in edge but not in 3.16-stable? !37597
2022-08-15 12:58:50 <Ariadne> omni: maybe gcc 12
2022-08-15 13:36:11 <bl4ckb0ne> psykose ayakael ill test jellyfin locally and then we can ship 10.8.4
2022-08-15 13:36:19 <bl4ckb0ne> im a bit suspicious about those runtimes
2022-08-15 13:43:34 <bl4ckb0ne> hm hitting some errors
2022-08-15 13:43:58 <bl4ckb0ne> yeah the skia dll is required
2022-08-15 13:44:08 <bl4ckb0ne> > System.NullReferenceException: Object reference not set to an instance of an object.
2022-08-15 13:44:13 <bl4ckb0ne> >    at BlurHashSharp.SkiaSharp.BlurHashEncoder.Encode(Int32 xComponent, Int32 yComponent, String filename, Int32 maxWidth, Int32 maxHeight)
2022-08-15 14:05:56 <ayakael> I created a skia sharp aport some time ago, I can send it your way so you can include it in the MR
2022-08-15 14:07:13 <ayakael> I would like us to avoid including runtimes. The `--self-contained` flag is to make the build portable through distros. Being in a packaging environment, we should aim for non-portable builds to save on space.
2022-08-15 14:12:10 <bl4ckb0ne> i think the self-contained flag does some stuff on the runtimes
2022-08-15 14:12:15 <bl4ckb0ne> also they seems to have moved it
2022-08-15 14:12:24 <bl4ckb0ne> latest rev works fine, no stacktrace here
2022-08-15 14:14:12 <ayakael> I've been able to run jellyfin fine without included runtimes with my skia-sharp package.
2022-08-15 14:15:16 <ayakael> Also, we do produce 'libe_sqlite3.so' and 'libSkiaSharp.so' in /usr/lib/jellyfin
2022-08-15 14:16:02 <bl4ckb0ne> yeah thats what i realised
2022-08-15 14:16:12 <ayakael> Although I had to delete my /var/lib/jellyfin folder when migrating from --self-contained to --no-self-contained yesterday
2022-08-15 14:16:33 <ayakael> When starting a completely new instance of jellyfin, those two libraries are picked up
2022-08-15 14:16:43 <bl4ckb0ne> hm
2022-08-15 14:16:45 <bl4ckb0ne> thats annoyinh
2022-08-15 14:17:26 <omni> Ariadne: ok, so.. would you say disable it on s390x for now, or disable the test on s390x and see if anyone running tor on s390x on edge report any issue?
2022-08-15 14:17:43 <Ariadne> whichever you want to do
2022-08-15 14:18:13 <ncopa> hi
2022-08-15 14:19:15 <ikke> o/
2022-08-15 14:22:23 <ikke> ncopa: did you see the xen issue? 
2022-08-15 14:22:59 <sam_> report it upstream first and hope it doesn't indicate something serious? 
2022-08-15 14:23:01 <sam_> a failure on an arch doesn't mean it's exclusive to that arch
2022-08-15 14:24:48 <ncopa> ikke: no i havent seen anything so far
2022-08-15 14:34:50 <ikke> #14090
2022-08-15 14:37:56 <Hello71> tor theoretically supports big-endian, but i suspect it's not very well tested. this function doesn't seem to do anything endian-specific though...
2022-08-15 14:42:10 <bl4ckb0ne> huh nice jellyfin-web stacktrace on tests for arm7
2022-08-15 14:58:37 <psykose> ikke: was just curious
2022-08-15 15:02:06 <ayakael> bl4ckb0ne well that's annoying
2022-08-15 15:02:45 <ayakael> my aport for skia-sharp is a PITA. Broken cause it needs python2, so been trying to adapt it to python3, but no ciggar yet
2022-08-15 15:03:13 <ayakael> I dont think we need it really, given that restarting the instance for jellyfin fixes the library detection issues
2022-08-15 15:04:49 <omni> Ariadne, Hello71: I chose to disable on s390x, for now, to try again later
2022-08-15 15:08:54 <bl4ckb0ne> i ran again with the flag and no issue
2022-08-15 15:09:22 <bl4ckb0ne> > FATAL ERROR: CALL_AND_RETRY_LAST Allocation failed - JavaScript heap out of memory
2022-08-15 15:09:32 <bl4ckb0ne> are we just unlucky or should we skip the tests on arm7
2022-08-15 15:10:20 <ayakael> Try rerunning the pipeline
2022-08-15 15:11:17 <ikke> psykose: anyway, I cleaned up the stray .makedepends
2022-08-15 15:11:32 <psykose> :)
2022-08-15 15:16:32 <ayakael> also to mitigate the detection bug, have you tried creating a symbolic link between ./runtimes/linux-musl-$_dotnet_arch/libe_sqlite3.so ./libe_sqlite3.so on your machine? Self-contained is to be avoided when packaging a dotnet application. It needlessly inflates the size of the package (39.4 mb vs 127.7 mb)
2022-08-15 15:32:35 <bl4ckb0ne> ayakael: pipeline rerun went good o/
2022-08-15 15:32:43 <ayakael> awesome :))
2022-08-15 15:32:52 <bl4ckb0ne> idk about the runtime stuff, it runs good now
2022-08-15 15:32:59 <bl4ckb0ne> i have been playing around a bit and no stacktrace
2022-08-15 15:34:22 <psykose> https://img.ayaya.dev/GEalNuos0jgj no stacktrace/
2022-08-15 15:34:24 <psykose> ?
2022-08-15 15:35:50 <bl4ckb0ne> yes
2022-08-15 15:35:52 <bl4ckb0ne> ship it
2022-08-15 15:36:25 <bl4ckb0ne> bonus it runs on arm (untested)
2022-08-15 15:43:59 <Ariadne> ikke: i bisected to 5.15.55 kernel with current xen 4.16.1-r1
2022-08-15 15:44:12 <Ariadne> will look into what has changed
2022-08-15 15:45:18 <ikke> Ariadne: good 
2022-08-15 15:49:30 <psykose> the car runs (didn't start the engine)
2022-08-15 15:49:52 <ncopa> so kernel 5.15.56 with xen 4.16.1-r1 fails?
2022-08-15 15:50:03 <ncopa> i mean we now know that it is a kernel issue?
2022-08-15 15:50:05 <Ariadne> 5.15.56 also passes
2022-08-15 15:50:29 <ncopa> 5.15.59 fails with same xen version?
2022-08-15 15:51:07 <Ariadne> yes
2022-08-15 15:51:14 <Ariadne> i suspect retbleed patches are to blame
2022-08-15 15:52:19 <Ariadne> building 5.15.57 vanilla and with 87e6270183a4062b46c21bf483539eea7854df6a and 96907c5d3927515029ad1f38a31d56dc04b8d288 reverted
2022-08-15 15:52:33 <psykose> it's possible the squashing was borked in some way for retbleed/xsa407 in !36330
2022-08-15 15:52:43 <Ariadne> maybe
2022-08-15 15:52:52 <Ariadne> but -r1 works fine
2022-08-15 15:53:11 <Ariadne> could be the combination, i guess
2022-08-15 15:53:27 <psykose> perhaps
2022-08-15 15:53:33 <psykose> i wish they made actual releases for half this shit
2022-08-15 15:57:11 <Ariadne> will try 5.15.59 + XSA-407 revert
2022-08-15 15:57:32 <Ariadne> though right now i have an all-hands meeting in ... 3 minutes
2022-08-15 15:57:37 <Ariadne> so, it will have to be after
2022-08-15 16:18:24 <omni> andyhhp in #xen too thought it was related to the retbleed changes in .57
2022-08-15 16:19:01 <omni> but I haven't seen anything more since last month
2022-08-15 16:19:21 <omni> but possibly someone too look at it together with
2022-08-15 17:02:04 <Ariadne> omni: with XSA-407 revert, 5.15.59 survives
2022-08-15 17:20:28 <nmeum> psykose: should we merge the go 1.19 upgrade in the hopes that it doesn't cause too much fallout on x86 or wait a bit? 🤔
2022-08-15 17:33:24 <psykose> if the rebuilds work all is good
2022-08-15 17:33:32 <psykose> (the ones you tested)
2022-08-15 17:35:30 <psykose> hugo fails a random test for some reason but doesn't seem relevant
2022-08-15 18:32:59 <ptrc> https://pkgs.alpinelinux.org/contents?file=*.swf&name=thelounge
2022-08-15 18:33:01 <ptrc> :)
2022-08-15 18:34:14 <ptrc> your typical node project
2022-08-15 18:34:16 <ikke> fun
2022-08-15 18:34:38 <ptrc> do npm install, get a random swf file bundled with the dependencies
2022-08-15 18:35:38 <psykose> pkgs.a.o seems to be glacially slow
2022-08-15 18:35:50 <ptrc> works just fine for me
2022-08-15 18:36:25 <ikke> same
2022-08-15 18:41:47 <psykose> nope, the ipv6 is just broken again
2022-08-15 18:41:49 <psykose> so indeed it's fucked
2022-08-15 18:42:04 <psykose> can you just disable ipv6 on it since it doesn't even work
2022-08-15 18:42:32 <ikke> psykose: I use ipv6, and it does work
2022-08-15 18:42:40 <psykose> i hate how having ''working'' ipv6 breaks half the alpine websites for me twice a month, hah
2022-08-15 18:42:45 <psykose> grr
2022-08-15 18:42:54 <ikke> So I wonder why it doesn't work for you
2022-08-15 18:43:01 <psykose> it could also be my routing
2022-08-15 18:43:04 <panekj> ipv6 works for me as well
2022-08-15 18:43:12 <psykose> half the pings get dropped even
2022-08-15 18:43:23 <psykose> exactly 50%
2022-08-15 18:43:39 <panekj> scream at isp
2022-08-15 18:44:00 <ikke> https://tpaste.us/0WbN 
2022-08-15 18:44:00 <psykose> don't even have an isp
2022-08-15 18:44:38 <psykose> ree
2022-08-15 18:44:41 <psykose> every time!!
2022-08-15 18:46:10 <ptrc> https://ptrc.gay/SXgaEoiQ
2022-08-15 18:46:15 <ptrc> this looks insanely broken imo
2022-08-15 18:46:32 <ptrc> no route from paris whatsoever, from stockholm it drops packets randomly
2022-08-15 18:46:57 <psykose> well blocking paris is good, have to keep the french out
2022-08-15 18:47:15 <psykose> traceroute is blank for me too, so i'm surprised half the pings work
2022-08-15 18:47:34 <ikke> is bgp.he.net working for you?
2022-08-15 18:48:27 <ikke> Now it loads
2022-08-15 18:49:04 <psykose> the website opens
2022-08-15 18:50:07 <psykose> my ipv6 is tunneled through HE, but i don't think that matters for the rest of the routing
2022-08-15 18:56:15 <psykose> also a heads up, i pushed py3-sphinx 5 + py3-docutils 0.19 and new setuptools (65) so any random python failures would most likely be caused by that
2022-08-15 18:56:22 <psykose> everything in main/ works, didn't check past that
2022-08-15 18:57:19 <sam_> setuptools breaks gobject-introspection
2022-08-15 18:57:21 <sam_> https://gitlab.gnome.org/GNOME/gobject-introspection/-/issues/438
2022-08-15 18:57:39 <psykose> ours doesn't even use setuptools
2022-08-15 18:57:42 <psykose> for some reason
2022-08-15 18:58:22 <psykose> ah, hm
2022-08-15 18:58:23 <psykose> let me see
2022-08-15 18:59:19 <sam_> try networkmanager just to be sure
2022-08-15 18:59:34 <psykose> yeah, got the same
2022-08-15 19:00:41 <psykose> gobject also fails if setuptools is in world at build time
2022-08-15 19:06:04 <psykose> a shim fix works for now, i guess
2022-08-15 19:06:05 <psykose> https://img.ayaya.dev/syfKzKtHJAp4
2022-08-15 19:06:11 <psykose> could also downgrade to 64 instead
2022-08-15 19:07:44 <ikke> when in doubt, comment it out
2022-08-15 19:08:13 <psykose> we don't have msvc so it should be safe
2022-08-15 19:08:45 <psykose> but now that i've ran into the 4th thing that needs setuptools 62+ i though i'd just go for it to the latest :)
2022-08-15 19:09:09 <psykose> sadly the unvendoring is a challenge for someone else
2022-08-15 19:09:29 <psykose> mere unvendoring is easy, moving 10 things to main less so
2022-08-15 19:09:33 <ikke> psykose: for some reason, my browser downloads it as a file instead of showing it btw
2022-08-15 19:09:43 <ikke> content-type
2022-08-15 19:09:45 <ikke> text/x-diff
2022-08-15 19:09:47 <psykose> it's a text/x-diff
2022-08-15 19:10:23 <ikke> my browser cannot show text/x-diff :)
2022-08-15 19:10:28 <umk3> good evening
2022-08-15 19:10:30 <ikke> Or does not
2022-08-15 19:10:39 <psykose> chromium displays it fine
2022-08-15 19:10:45 <psykose> the firefox developers are very stubborn
2022-08-15 19:10:52 <psykose> and seem to not want to display anything but text/plain
2022-08-15 19:10:57 <psykose> and download everything else
2022-08-15 19:11:01 <psykose> i spent way too much time on that one
2022-08-15 19:11:12 <panekj> firefox devs are very stubborn
2022-08-15 19:11:21 <psykose> i think there's an issue open for this
2022-08-15 19:11:23 <psykose> for maybe 12 years?
2022-08-15 19:11:32 <umk3> I still have problems trying to open links from flatpak applications in labwc / sway compositor :(
2022-08-15 19:11:37 <psykose> there's no way around it without some awful extensions
2022-08-15 19:11:58 <ptrc> https://bugzilla.mozilla.org/show_bug.cgi?id=1319262
2022-08-15 19:12:04 <psykose> ah, just 6
2022-08-15 19:12:07 <ptrc> there's one for text/markdown, open since forever
2022-08-15 19:12:29 <panekj> when will ff support srv records 
2022-08-15 19:12:29 <ptrc> https://bugzilla.mozilla.org/show_bug.cgi?id=57342
2022-08-15 19:12:35 <ptrc> psykose: here's 22 :)
2022-08-15 19:12:42 <psykose> looooooooool
2022-08-15 19:12:45 <psykose> 22 years!!!!!!!
2022-08-15 19:12:51 <psykose> my god can't they just add like
2022-08-15 19:12:52 <psykose> ONE toggle
2022-08-15 19:12:59 <psykose> for text/* -> display as text
2022-08-15 19:13:03 <psykose> would it be that hard
2022-08-15 19:13:19 <panekj> ooo, it beat the SRV bug
2022-08-15 19:13:32 <psykose> i'm tempted to just patch it myself or something
2022-08-15 19:13:33 <ptrc> umk3: do you have it all under one dbus session with DBUS_SESSION_BUS_ADDRESS set?
2022-08-15 19:14:09 <umk3> ptrc: a system dbus session is running and the compositor runs with `$ dbus-run-session labwc`
2022-08-15 19:14:28 <umk3> yes DBUS_SESSION_BUS_ADDRESS is present
2022-08-15 19:14:41 <ptrc> and what's the exact issue? is there any error, any pop-up?
2022-08-15 19:15:02 <umk3> let me get it, just a sec
2022-08-15 19:15:19 <ikke> "Sadly, this doesn't look like a trivial change."
2022-08-15 19:15:41 <ikke> Is the code so convoluted that this is not trivial?
2022-08-15 19:15:58 <ptrc> sadly, treating one mimetime like another one is just too hard for firefox devs
2022-08-15 19:16:13 <ptrc> the whole text view code is hardcoded for "text/plain", no way to change it
2022-08-15 19:16:17 <umk3> when xdg-desktop-portal-gtk is running I just get the GNOME looking xdg-desktop-portal-gtk window "Open With..." and tells me to use gnome-software to look for app
2022-08-15 19:17:03 <ikke> ptrc: do they have `if mime == "text/plain
2022-08-15 19:17:14 <ikke> "` sprinkled all over the codebase or something
2022-08-15 19:17:20 <ptrc> umk3: and you're trying to open just a regular http(s) link from some flatpak app, expecting it to open in your local browser, right?
2022-08-15 19:18:12 <ptrc> ikke: exactly how i expect it to work, based on these bugzilla discussions :p
2022-08-15 19:18:30 <panekj> looking at the patch, it seems so cursed
2022-08-15 19:19:39 <umk3> and if x-d-p-gtk is not running, I get this message Failed to call portal: GDBus.Error:org.freedesktop.DBus.Error.UnknownMethod: No such interface “org.freedesktop.portal.OpenURI”
2022-08-15 19:20:27 <umk3> ptrc: it should open another flatpak or at least follow the mime call I think
2022-08-15 19:21:14 <umk3> i have tested with opening url in telegram and many others
2022-08-15 19:21:16 <panekj> they have no issue making fancy feature to display JSON
2022-08-15 19:21:18 <sam_> omg this whole time it's been a firefox bug?
2022-08-15 19:21:20 <sam_> i just assumed i had something misconfigured lol
2022-08-15 19:21:23 <panekj> but displaying text is hard
2022-08-15 19:21:51 <ptrc> umk3: ah, so it's a custom schema url? i'm still not sure what you're trying to do
2022-08-15 19:22:16 <psykose> what is the uri
2022-08-15 19:22:34 <umk3> hmmm, there is a flatpak for games that opens custom schema, but there is telegram or other apps that cannot even open https://
2022-08-15 19:22:51 <psykose> needing -portal-gtk running implies the underlying program is gtk, the 'open with' thing showing up instead means you have no mime handler for the type or it can't find it for some reason
2022-08-15 19:23:06 <psykose> what happens when you click an https:// in telegram flatpak
2022-08-15 19:23:53 <umk3> can I show screenshot? It shows a xdg-desktop-portal-gtk very gnome looking window "Open WIth.."
2022-08-15 19:24:12 <umk3> and it goes nowhere, it tells me to use gnome-software to update
2022-08-15 19:24:52 <psykose> and if you `xdg-open https://google.com` i assume it works
2022-08-15 19:25:10 <umk3> from outside flatpak apps yes
2022-08-15 19:25:16 <ikke> sam_: I was wondering why nothing showed up on the page
2022-08-15 19:25:28 <umk3> but from inside not
2022-08-15 19:28:01 <psykose> and you have xdg-desktop-portal-wlr running as well right
2022-08-15 19:28:07 <umk3> yes
2022-08-15 19:28:24 <umk3> some people say *-gtk and *-wlr don't mess with each other and fallback
2022-08-15 19:28:53 <umk3> seems true because having one and the other uninstalled doesn't change a thing
2022-08-15 19:35:28 <umk3> psykose: if I run xdg-desktop-portal manually `$ exec /usr/libexec/xdg-desktop-portal -vr` I don't see OpenURI in the verbose
2022-08-15 19:35:51 <psykose> can you post your full `env`
2022-08-15 19:38:10 <umk3> https://p.teknik.io/CMxLC
2022-08-15 19:40:13 <umk3> I find weird that dbus tmp socket I cannot find it in /tmp
2022-08-15 19:43:03 <psykose> you might also need XDG_CURRENT_DESKTOP=sway , rest looks ok
2022-08-15 19:43:09 <psykose> the tmp thing is correct
2022-08-15 19:43:30 <psykose> if you use firefox you probably need MOZ_DBUS_REMOTE=1
2022-08-15 19:43:32 <umk3> yes I have tried with XDG_CURRENT_DESKTOP=sway and wlroots
2022-08-15 19:45:11 <umk3> psykose: you can find your DBUS_SESSION_BUS_ADDRESS dir in /tmp ?
2022-08-15 19:45:33 <psykose> no, it's abstract
2022-08-15 19:45:45 <umk3> oh ok
2022-08-15 19:53:21 <psykose> but yeah you need all of those envs set
2022-08-15 19:53:41 <psykose> and then have a `exec dbus-update-activation-environment --systemd DISPLAY MOZ_DBUS_REMOTE WAYLAND_DISPLAY XDG_CURRENT_DESKTOP` or whatnot in your compositor startup
2022-08-15 19:53:43 <psykose> and start the portals
2022-08-15 19:53:44 <psykose> and reboot
2022-08-15 19:53:51 <psykose> and then you should be able to open an https link to start
2022-08-15 19:54:02 <psykose> if you miss literally any of the above it doesn't work : )
2022-08-15 19:57:12 <umk3> psykose: im gonna try again those envs and see what comes up, many thanks!
2022-08-15 19:57:50 <umk3> also uninstalled pipewire just in case this time
2022-08-15 21:20:20 <psykose> go size update: everything got ~3% bigger with 1.19
2022-08-15 21:21:00 <ikke> oof
2022-08-15 21:37:35 <omni> are old packages available anywhere?
2022-08-15 21:37:56 <ikke> nope, sadly not
2022-08-15 21:38:31 <ikke> I assume you talk about xen / linux-lts?
2022-08-15 21:39:04 <omni> yes
2022-08-15 21:39:09 <omni> I do have old packages myself
2022-08-15 21:40:13 <Ariadne> from what i can tell, reverting the XSA-407 patches allows it to survive
2022-08-15 21:40:21 <Ariadne> on the xen package
2022-08-15 21:40:56 <ikke> so do we want to revert that?
2022-08-15 21:41:29 <Ariadne> i would like somebody to verify it
2022-08-15 21:41:36 <Ariadne> :P
2022-08-15 21:42:11 <ikke> We can build it in CI and ask someone in that issue if they want to test it
2022-08-15 21:42:53 <Ariadne> my test scenario was "i installed xen inside a VM, and put it under some synthetic load (a VM running prime95), and it hasn't died yet"
2022-08-15 21:43:29 <ikke> I think the issue happens when creating and destroying domU?
2022-08-15 21:43:58 <omni> can you download builds from the CI?
2022-08-15 21:44:20 <Ariadne> ikke: yes, i put the load in a domU
2022-08-15 21:44:33 <ikke> omni: certainly, they are in the artifacts of the jobs
2022-08-15 21:45:40 <omni> ah, uhm, guess I could try
2022-08-15 21:47:38 <omni> andyhhp asked me to send something to xen-devel@lists.xenproject.org about this, but I generally don't do email much at all
2022-08-15 21:48:41 <panekj> psykose: what are you doing with luajit? 🤔
2022-08-15 21:49:51 <psykose> was just checking some stuff
2022-08-15 21:49:55 <psykose> and nope still fails
2022-08-15 21:50:01 <psykose> sad
2022-08-15 21:50:11 <psykose> you should ask your friend with a ppc machine to fix it /s
2022-08-15 21:50:44 <panekj> heh
2022-08-15 21:50:52 <panekj> i hate gitlab ui
2022-08-15 21:51:45 <ikke> omni: !37709
2022-08-15 21:56:42 <omni> saw and commented
2022-08-15 21:57:52 <omni> xsa407-4.16-00.patch was probably bad naming on my part, it is not part of the XSA-407 patches but they wouldn't apply without it (as explained in the commit message) and I wanted it right before those patches
2022-08-15 21:58:26 <ikke> omni: but we should revert it anyway, right?
2022-08-15 21:58:46 <ikke> if we don't apply the XSA-407 patches, we don't need that support patch
2022-08-15 21:58:47 <omni> idk
2022-08-15 21:59:25 <omni> it worked up until the retbleed patches in linux,  one could just as well argue that those should be reverted :p
2022-08-15 21:59:36 <omni> or rather get this properly sorted
2022-08-15 22:00:02 <ikke> in any case, this is just to be able to test it
2022-08-15 22:00:22 <omni> sure
2022-08-15 22:00:43 <ikke> omni: https://gitlab.alpinelinux.org/alpine/aports/-/jobs/804271/artifacts/browse/packages/main/x86_64/
2022-08-15 22:03:32 <omni> <3
2022-08-15 22:04:45 <omni> ah, but I run my dom0 on edge..
2022-08-15 22:05:21 <ikke> oh
2022-08-15 22:05:27 <omni> I can open one for that
2022-08-15 22:05:35 <ikke> sure
2022-08-15 22:36:40 <omni> but it won't build and I'm too tired for this
2022-08-15 22:51:44 <ebb> they're errors in the ipxe build
2022-08-15 22:51:55 <ebb> whenever you revisit i suggest making the ipxe build skip :p
2022-08-15 22:54:33 <psykose> just looks like Werror and not any real errors
2022-08-15 22:55:35 <ebb> some fingers are being pointed at gcc rather than ipxe, on ipxe's issue tracker
2022-08-15 22:55:54 <ebb> actually skipping building ipxe is probably a terrible idea (in general)
2022-08-15 22:56:19 <psykose> well yes, it's how the finger pointing always goes
2022-08-15 22:56:27 <psykose> 1) add -Werror to release builds
2022-08-15 22:56:33 <psykose> 2) anyone updates any part of their toolchain
2022-08-15 22:56:35 <psykose> 3) build fails
2022-08-15 22:56:38 <psykose> 4) blame the compiler
2022-08-15 22:57:20 <psykose> but this specific one (array/string bounds) have been gcc-specific issues for the past 2 releases iirc
2022-08-15 22:58:07 <psykose> seems like there's a toggle for this one
2022-08-15 23:08:04 <psykose> omni: fixed for ya
2022-08-15 23:15:55 <omni> yeah, saw that ;)
2022-08-15 23:16:14 <omni> I guess I'm not tired enough yet...
2022-08-16 00:17:47 <skarnet> holy backlog batman
2022-08-16 08:17:43 <ncopa> morning! good job with the xen issue!
2022-08-16 08:22:54 <ncopa> did anyone report the issue upstream?
2022-08-16 08:33:35 <omni> moin
2022-08-16 08:33:47 <omni> I haven't
2022-08-16 08:34:52 <omni> other than writing about it in #xen on this network
2022-08-16 08:35:44 <ebb> moin
2022-08-16 08:35:45 <omni> and they seem to think that it's a "linux regression, not a Xen regression"
2022-08-16 08:36:00 <ebb> I had a stab at reproducing yesterday; does it happen with PVH guests only or PV too?
2022-08-16 08:36:13 <ebb> (i.e. i was not successful)
2022-08-16 08:38:26 <omni> for me the issue is with PV Dom0 running linux 5.15.57 or later, I have no issues with DomUs with this kernel
2022-08-16 08:39:15 <ebb> right now I have a pvh dom0 running... i'll see if I can try with pv a bit later
2022-08-16 08:39:21 <omni> PVH Dom0 with linux 5.15.57 works, but PCI passthrough does not seem to be available in that mode
2022-08-16 08:39:31 <ebb> from reading the issue I got the impression that the flavour of domU did matter
2022-08-16 08:40:25 <ncopa> so currently it can be worked around by either revert XSA-407 or use a kernel 5.15.56?
2022-08-16 08:40:30 <ebb> (I'd have thought PCI passthrough works fine in PVH, so long as your CPU has an IOMMU/VT-d)
2022-08-16 08:42:28 <omni> ncopa: for me, reverting XSA-407 did not help
2022-08-16 08:42:47 <ebb> so the XSA patch cannot be the cause
2022-08-16 08:44:54 <omni> I'm saying "for me" and "my issue" since there may be several issues here, and I suspect the linux retbleed changes (without having really looked at them, since I'm not sure I would follow)
2022-08-16 08:45:11 <ebb> ah, okay, my bad
2022-08-16 08:45:42 <ebb> So your steps to reproduce were run a PV Dom0, start a guest, have the guest reboot?
2022-08-16 08:47:06 <omni> no, for me it is just trying to boot a PV Dom0 with linux 5.15.57 or later, it will reboot before it even tries to start any DomUs
2022-08-16 08:47:12 <ebb> ah
2022-08-16 08:50:36 <omni> I think I would rather like to run Dom0 as PVH, it may be me who is missing something in how it should be setup to allow PCI passthrough, I haven't dug into that
2022-08-16 08:51:44 <omni> btw, I edited a comment in the issue where I had written "I got issues with 5.15.56" when I meant .57, sorry about that
2022-08-16 08:58:59 <ebb> oh hello
2022-08-16 08:59:04 <ebb> my Dom0 crashed
2022-08-16 08:59:12 <ebb> No logs
2022-08-16 08:59:24 <ikke> omni: and you seem to be on edge, while most others use 3.16?
2022-08-16 09:32:07 <omni> yes, at least my Dom0
2022-08-16 09:37:06 <omni> ebb: PVH?
2022-08-16 09:37:13 <ebb> This time it was PV
2022-08-16 09:39:05 <ebb> I was watching the serial console and nothing came out of dom0 or the hypervisor (apart from "(XEN) Hardware Dom0 crashed: 'noreboot' set - not rebooting.")
2022-08-16 09:40:02 <ebb> Xen seemed to notice this quite a few seconds after the dom0 console hung
2022-08-16 10:01:07 <ncopa> ok, so the issue also happens with creating a PV?
2022-08-16 10:02:53 <ncopa> using PV means we can troubleshoot it in qemu
2022-08-16 10:03:22 <ncopa> i guess we can do it with PVH too, but I have issues with nested virtualizaion in the past
2022-08-16 10:07:49 <PureTryOut> is somebody actively cancelling pipelines? For several merge requests now I've had CI for multiple arches being cancelled
2022-08-16 10:12:48 <psykose> me, i cancel things that don't usually apply
2022-08-16 10:17:26 <psykose> (unless there's something important about running rebuilds for 150 packages when only enabling them on one architecture somewhere)
2022-08-16 10:17:45 <psykose> there's only two runners for each and it's annoying when something holds ci for 6 hours
2022-08-16 10:27:57 <PureTryOut> Makes sense, but although my PR title might not say it, some arches keep getting cancelled that are actually relevant 😅
2022-08-16 10:28:24 <PureTryOut> Might be good to do some automatic detecting if arches are unaffected and then skip the CI. We do that at postmarketOS at least
2022-08-16 10:28:57 <ikke> PureTryOut: how do you do it in pmos?
2022-08-16 10:30:44 <PureTryOut> some fancy scripting, https://gitlab.com/postmarketOS/pmaports/-/blob/master/.ci/build_changed_aports.py#L72
2022-08-16 10:31:32 <PureTryOut> Although now I'm thinking about it, that does not do what I said it does
2022-08-16 10:32:26 <ikke> But abuild will already skip packages that are disabled for the current arch?
2022-08-16 10:32:39 <PureTryOut> Yeah I just noticed our scripts do the same, that's what I said last 😅
2022-08-16 10:32:41 <PureTryOut> so nvm
2022-08-16 10:33:17 <ikke> One idea I had is to use dynamic pipelines so that in advance it can be determined whether a job for a certain arch should be run
2022-08-16 10:33:49 <panekj> eval (grep 'options=') then diff :P
2022-08-16 10:35:08 <ikke> panekj: you mean arch=?
2022-08-16 10:35:29 <panekj> yes, my brain is not thinking today
2022-08-16 10:35:40 <ikke> The diff does not even matter
2022-08-16 10:35:57 <panekj> why not?
2022-08-16 10:36:13 <ptrc> imo the simplest way would be to just check if the pkgname-pkgver-pkgrel already exists on the mirror, then skip building
2022-08-16 10:37:05 <panekj> I'd prefer it to be self-contained to the change itself
2022-08-16 10:37:15 <panekj> don't need curl'ing the mirror :P
2022-08-16 10:37:54 <ikke> We don't want to only build packages for which the arch changes
2022-08-16 10:38:41 <ikke> You _could_ check if that's the only change and not the current arch though
2022-08-16 10:38:49 <panekj> wasn't that the point of topic
2022-08-16 10:43:29 <ikke> panekj: yes, I realize that now
2022-08-16 10:44:15 <ikke> I was thinking about not starting CI jobs where the arch has been disabled
2022-08-16 10:44:37 <panekj> ah, that's also a valid thing to implement (:
2022-08-16 10:45:08 <panekj> although not sure how that could be done
2022-08-16 10:45:13 <ikke> dynamic pipelines
2022-08-16 10:45:31 <panekj> you say gitlab things to a github person
2022-08-16 10:45:42 <ikke> panekj: Give me time to epxlain it :P
2022-08-16 10:46:07 <ikke> panekj: The idea is that you use one stage to generate a new ci.yml file, which then gets triggered as a child-pipeline
2022-08-16 10:46:10 <panekj> I mean, I'm pretty sure I know what it entails, but syntax example would be nice
2022-08-16 10:46:40 <ikke> https://gitlab.alpinelinux.org/alpine/aports/-/merge_requests/31987/diffs?commit_id=97376ffcd1628e0649576d032ed84e0bf3547e65
2022-08-16 10:47:23 <ikke> https://gitlab.alpinelinux.org/alpine/aports/-/blob/97376ffcd1628e0649576d032ed84e0bf3547e65/.gitlab-ci.yml might be a bit more clear
2022-08-16 10:47:31 <ebb> ncopa: in my case I was just typing commands into a PV Dom0 and it hung.  No domain creation going on at the time, even
2022-08-16 10:47:43 <ebb> the last thing I ran was `sysctl --help`
2022-08-16 10:47:46 <panekj> uhm, .jsonnet, I think I'll nope out
2022-08-16 10:47:59 <ikke> panekj: that's one option, you can juse anything to generate it
2022-08-16 10:48:03 <ikke> use*
2022-08-16 10:48:34 <ikke> If you feel like writing a rust program to generate a new ci build, be my guest :P
2022-08-16 10:48:48 <ebb> (and that didn't return, ^C didn't interrupt it, seemed like the Dom0 hung around there.  I don't think the command is related)
2022-08-16 10:48:49 <panekj> pfff, I don't write *everything* in rust
2022-08-16 10:49:09 <panekj> but you hit home too close, since I was thinking of making own CI system
2022-08-16 10:49:12 <ikke> ebb: do you feel like testing out the 3.16 packages without the XSA-407 related packages?
2022-08-16 10:49:25 <ikke> panekj: heh
2022-08-16 10:49:54 <ebb> Sure
2022-08-16 10:49:56 <panekj> I did work for about 2 years on reimplementing GH Actions runner in Go using Docker
2022-08-16 10:50:04 <panekj> (pain)
2022-08-16 10:50:22 <ikke> ebb: https://gitlab.alpinelinux.org/alpine/aports/-/jobs/804271/artifacts/browse/packages/main/x86_64/
2022-08-16 10:50:30 <ebb> It might be a little while; I didn't realise that xen's 'noreboot' option means the machine just sits there and doesn't respond to serial keys, so I can't reboot it until I get back home :P
2022-08-16 10:50:50 <ikke> The artifacts are valid for 11 hours more
2022-08-16 10:50:59 <ikke> after that I'd need to start a new job
2022-08-16 10:52:02 <ebb> thanks; have pulled them
2022-08-16 10:53:08 <ebb> I have a slightly special setup in that I'm netbooting this machine - guess I can just get the xen image out of xen-hypervisor.apk and use that?
2022-08-16 10:56:47 <omni> christiank, running linux-edge, reported that their problems were gone with linux 5.19.1
2022-08-16 14:33:24 <Habbie> is "J0WI" here?
2022-08-16 14:34:20 <Habbie> oh, !37742 was closed already, nevermiond :)
2022-08-16 14:48:22 <pabloyoyoista> Hey PureTryOut, psykose: why did GNOME get massively enabled for s390x? It's not really supported upstream and a bit of waste of resources. Similar for riscv64. Would I be allowed to massively disable it? 
2022-08-16 14:49:03 <PureTryOut> because there were no dependencies blocking it anymore. Alpine policy afaik to enable it for all architectures where possible
2022-08-16 14:49:46 <panekj> We discussed some time ago that it's pointless to enable graphical/desktop software on arches such as s390x, since, well, they can't be used at all
2022-08-16 14:50:11 <PureTryOut> That's new to me. There are tons of stuff that don't really make sense on s390x but are still enabled there
2022-08-16 14:50:32 <panekj> it was discussed only on irc, so don't take it as granted (:
2022-08-16 14:50:50 <PureTryOut> I was just working away the "<arch> blocked by <some Rust dependency that is now available>" comments in APKBUILDs, which yes included GNOME
2022-08-16 14:52:04 <pabloyoyoista> PureTryOut, that makes sense. I guess this might be something to be raised to the TSC?
2022-08-16 14:52:16 <PureTryOut> Yeah, sure
2022-08-16 14:52:23 <panekj> 2022-07-20 13:30:32 <Ariadne> ncopa: it is not even possible to use s390x as desktop, there is no GPU :D
2022-08-16 14:52:46 <panekj> and I just pinged people unnecessarily 
2022-08-16 14:52:55 <PureTryOut> Let's make it an official policy before we're going to revert all the time I just put in lol
2022-08-16 14:54:14 <panekj> it would be easier if we had some kind of build categories "type=desktop" so it would get disabled automatically on specific arches (:
2022-08-16 14:54:40 <PureTryOut> That'd be awesome tbh
2022-08-16 14:56:34 <skarnet> we have an environment variable for bootstrapping
2022-08-16 14:56:50 <skarnet> we could have another one for desktops or something
2022-08-16 14:56:57 <pabloyoyoista> PureTryOut, in the meanwhile, what would you like me to do with this? https://gitlab.alpinelinux.org/alpine/aports/-/merge_requests/37743 gnome-bluetooth (test) and mutter (build) seem to fail in s390x. Just disable them again?
2022-08-16 14:57:50 <PureTryOut> Yeah that's fine. Just add a comment explaining why it's disabled
2022-08-16 14:58:04 <PureTryOut> I have more packages that I was planning to enable but failed in other ways, I just updated the comment to reflect that new state
2022-08-16 14:59:42 <pabloyoyoista> Ok, thanks
2022-08-16 15:13:32 <PureTryOut> oh and of course, make sure you also disable packages that depend on those
2022-08-16 15:17:25 <pabloyoyoista> Well, now randomly gnome-bluetooth passed the pipeline... So I guess I'll keep it enabled for now. Grabbing some food and then coming to send the TSC issue and the "desktop" proposal ;)
2022-08-16 15:17:31 <PureTryOut> pabloyoyoista: seems retrying the CI has resolved those failures? I don't see you disabling those arches but CI has succeeded now
2022-08-16 15:18:04 <pabloyoyoista> mutter problem seemed like gnome-settings-daemon was still not in the mirror
2022-08-16 15:18:14 <pabloyoyoista> gnome-bluetooth, no clue xD
2022-08-16 15:19:04 <pabloyoyoista> Thanks anyway for looking at it :)
2022-08-16 15:23:36 <ikke> If they are flaky in CI, they're probably also going to be flatly on the builders 
2022-08-16 15:53:13 <ikke> https://gitlab.alpinelinux.org/alpine/tsc/-/issues/54
2022-08-16 15:53:32 <panekj> ❤
2022-08-16 15:54:14 <panekj> another benefit (since you mentioned fixing builds) is that we don't have to do many builds at all
2022-08-16 16:03:40 <PureTryOut> Thanks ikke!
2022-08-16 16:03:48 <ajhalili2006> I'm getting "depmod: WARNING: could not open modules.builtin.modinfo at /tmp/mkinitfs.nNOFHe/lib/modules/5.15.60-0-lts: No such file or directory" during the post-install process
2022-08-16 16:03:56 <ajhalili2006> while upgrading to edge from v3.15.x
2022-08-16 16:03:57 <ikke> ajhalili2006: you can ignore that
2022-08-16 16:04:28 <ikke> in the sense that it's a small issue that does not affect anything, and it will be fixed
2022-08-16 16:04:42 <panekj> warnings are meant to be ignored (:
2022-08-16 16:05:34 <ajhalili2006> How about these ca-certificate-bundle errors which caused the "apk upgrade" attempt to fail? (please let me paste the errors one by one)
2022-08-16 16:05:35 <ajhalili2006> ERROR: ca-certificates-bundle-20220614-r2: trying to overwrite etc/ssl1.1/cert.pem owned by libcrypto1.1-1.1.1q-r0.
2022-08-16 16:05:41 <ajhalili2006> ERROR: ca-certificates-bundle-20220614-r2: trying to overwrite etc/ssl1.1/certs owned by libcrypto1.1-1.1.1q-r0.
2022-08-16 16:06:04 <ikke> Did you run apk upgrade --update --available?
2022-08-16 16:06:05 <panekj> apk del libssl1.1 && apk fix
2022-08-16 16:06:24 <panekj> but also what ikke said
2022-08-16 16:07:16 <panekj> mhm, libssl depends on libcrypto
2022-08-16 16:08:58 <ajhalili2006> thanks ikke and panekj
2022-08-16 16:10:16 <skarnet> I'm selfishly waiting for people to smooth out openssl3 migration bumps and kinks before updating my edge
2022-08-16 16:10:29 <ikke> I already did :)
2022-08-16 16:10:31 <ikke> no issue
2022-08-16 16:11:47 <skarnet> excuse me for doubting a "no issue" by a lead dev, tech expert who eats apk for breakfast, when a user *just* reported they experienced a problem
2022-08-16 16:12:12 <panekj> it was pebkac
2022-08-16 16:12:14 <skarnet> I'll wait until the normal user experience is smooth, I don't want to engage my brain on this
2022-08-16 16:13:41 <ikke> I just ran apk ugprade -Ua
2022-08-16 16:14:04 <ikke> No brain cycles spent :)
2022-08-16 16:20:07 <skarnet> I'm going to regret this
2022-08-16 16:21:39 <ikke> skarnet: just FYI, I waited as well until all the repos were at least built :)
2022-08-16 16:25:24 <Hello71> skarnet: you will be happy to hear that gcc has also been upgraded to 12
2022-08-16 16:51:18 <ajhalili2006> How do I migrate between x86 to x86_64 since I accidentally installed the x86 version of Alpine Linux? (or maybe should i do reinstall)
2022-08-16 16:54:18 <ikke> ajhalili2006: Not sure how apk will handle it, but you'd start by downloading the static x86_64 version of apk-tools
2022-08-16 16:58:44 <pabloyoyoista> Thanks for the TSC issue, ikke!
2022-08-16 18:16:44 <ebb> Getting back to trying the Xen problem (still with the XSA-407 patches)
2022-08-16 18:16:57 <ebb> my PV Dom0 locked up and I got Xen to dump registers
2022-08-16 18:17:12 <ebb> All 4 of the vCPUs are on the same RIP, which is suspicious
2022-08-16 18:17:31 <ikke> ebb: someone mentioned a patch for linux lts, which I'm preparing an MR for
2022-08-16 18:19:32 <ebb> oh right, that someone is andyhhp
2022-08-16 18:19:50 <ikke> indeed
2022-08-16 18:20:03 <ikke> kernel is still building locally
2022-08-16 18:34:43 <PureTryOut> huh, fun. Abiword's website has an expired SSL certificate since the 12th of July...
2022-08-16 18:36:26 <ikke> lol
2022-08-16 18:39:14 <PureTryOut> So uh, what to do with that? We can't touch the package atm as it'll just fail to fetch the sources, and it'll definitely be a problem next Alpine release
2022-08-16 18:43:07 <ikke> PureTryOut: I can put the file on distfiles, then the builders will fetch that
2022-08-16 18:43:28 <PureTryOut> that'll help for the builders yes, but you still can't test the package locally
2022-08-16 18:43:35 <ikke> correct
2022-08-16 18:44:24 <ikke> PureTryOut: you should have access to dev.a.o, right?
2022-08-16 18:44:50 <PureTryOut> yes, but I don't have the sources locally to upload there atm 😛
2022-08-16 18:45:07 <ikke> PureTryOut: they should be on distfiles
2022-08-16 19:02:26 <PureTryOut> does the pkgrel need to bump just to change the source to dev.alpinelinux.org?
2022-08-16 19:02:59 <ikke> PureTryOut: If the checksum does not change, I'd say no
2022-08-16 19:03:10 <ikke> As in, there is no reason to trigger a rebuild
2022-08-16 19:03:35 <PureTryOut> ah yeah ok, good
2022-08-16 20:52:16 <Hello71> ebb: that's probably HLT then
2022-08-16 20:53:04 <ikke> sadly the linux patch that was suggested doesn't seem to help eitheer
2022-08-16 22:39:08 <omni> PureTryOut: can't you just download it over http instead? I think it's fine since it still has to match the checksum
2022-08-17 06:24:05 <nmeum> does anybody have an idea what is happening here? https://tpaste.us/NOpm iirc this code used to compile fine with gcc 11
2022-08-17 06:35:43 <psykose> looks like the same fortify-headers issue as before
2022-08-17 06:35:54 <psykose> i am surprised it now happens without LTO at all, previously that was a hard requirement
2022-08-17 06:36:05 <psykose> iirc eventually we decided the issue is in gcc, so they made it worse in 12 i guess
2022-08-17 06:36:15 <psykose> also elfutils eu-readelf segfaults reading anything
2022-08-17 06:36:49 <psykose> invalid call to printf in readelf.c:1821, where they wrap the format string in ngettext
2022-08-17 06:36:53 <psykose> i assume something returns garbage somewhere
2022-08-17 06:38:06 <psykose> ah
2022-08-17 06:38:09 <psykose> it segfaults on 3.16-stable too
2022-08-17 06:38:16 <psykose> i guess it was always broken and i just found it
2022-08-17 06:38:49 <nmeum> does the LTO fortify-headers issue also disappear when compiling with -march=native?
2022-08-17 06:39:20 <psykose> reproduced on 3.15, 3.14, 3.13 too
2022-08-17 06:39:25 <psykose> nice, just a thing nobody ever used
2022-08-17 06:39:31 <psykose> nmeum: don't think so
2022-08-17 06:39:43 <nmeum> might be distinct issues then
2022-08-17 06:40:01 <psykose> ah, i see now yours gets fixed with march
2022-08-17 06:40:02 <psykose> haha
2022-08-17 06:40:09 <psykose> that is some very strange gcc stuff then
2022-08-17 06:40:13 <nmeum> I think elfutils is primarly used as a library, not sure if anyone actually uses the executables over the implementations in binutils
2022-08-17 06:40:19 <psykose> yeah
2022-08-17 06:40:25 <psykose> just.. ran into a meson build
2022-08-17 06:40:32 <psykose> that checked eu-readelf and then readelf
2022-08-17 06:40:35 <psykose> and i had it installed by chance
2022-08-17 06:40:53 <psykose> given then patches to elfutils already, not surprised it's broken with printf
2022-08-17 06:41:15 <psykose> i would report the gcc thing upstream, if you know how to minimise the source file
2022-08-17 06:41:19 <psykose> i could never figure it out
2022-08-17 06:41:28 <psykose> i got some easy cases, but if i shortened it to a 10 line file they went away
2022-08-17 06:41:41 <psykose> and 'compile all off ccache' is not a very friendly reproduction
2022-08-17 06:48:17 <nmeum> I don't have a minimal file yet, I opened #14105 to remind myself of this issue and hopefully investigate it further soonish
2022-08-17 06:52:13 <psykose> same
2022-08-17 06:52:21 <psykose> not sure i'll find anything but it's nice to keep track
2022-08-17 07:06:58 <ajhalili2006> ikke: I got the static version of apk-tools, but I chose the USB liveboot option. Might check the manpages for apk-install on hints.
2022-08-17 07:09:36 <psykose> most likely you can change the arch in /etc/apk/arch and run apk fix --reinstall \*
2022-08-17 07:09:46 <psykose> since it's x86/x86_64 you wouldn't even need static, though it's just to be safe
2022-08-17 07:11:23 <ikke> psykose: afaik apk is hard-coded for a certain arch unless you initialize a new root 
2022-08-17 07:11:55 <psykose> aha
2022-08-17 07:42:13 <Asmadeus> nmeum: fwiw -march=native doesn't fix it for me here (now's edge, g++ 12.1.1_git20220630, x86_64) so I'm curious what string.h you pull in with -E or something
2022-08-17 07:44:05 <nmeum> Asmadeus: can you comment that in the issue? I will check after $dayjob then but I am also just using g++ 12.1.1_git20220630 on x86_64 only with the flags provided in the issue
2022-08-17 07:44:32 <nmeum> maybe it's cpu-specific due to -march=native? I am using an Amd Ryzen 7 5700
2022-08-17 08:13:47 <Asmadeus> could be cpu-specific yeah, old xeon here. Will post that in the issue
2022-08-17 08:51:18 <PureTryOut> how does the TSC thing work exactly? Are issues made in the repo for it picked up in weekly/monthly meetings or so?
2022-08-17 09:03:17 <clandmeter> its mostly biweekly meetings
2022-08-17 09:05:26 <clandmeter> we check the issues in the tracker and add them as agenda items
2022-08-17 09:12:47 <ericonr> psykose: did alpine update argparse to be from my repo? I remember readelf used to segfault t.
2022-08-17 09:13:13 <ericonr> .... When showing help output, at least, which I fixed 
2022-08-17 09:13:38 <psykose> ericonr: the one is from https://github.com/argp-standalone/argp-standalone/ which is forked from yours
2022-08-17 09:13:42 <psykose> not sure if that's the one you're thinking of
2022-08-17 09:13:54 <ericonr> Sounds good
2022-08-17 09:31:11 <q66> <nmeum> I think elfutils is primarly used as a library, not sure if anyone actually uses the executables over the implementations in binutils
2022-08-17 09:31:13 <q66> flatpak does
2022-08-17 09:32:04 <q66> it's one of things i'll have to patch to get flatpak on my distro because i have neither binutils by default nor elfutils
2022-08-17 10:17:27 <ikke> PureTryOut: https://gitlab.alpinelinux.org/alpine/tsc/-/blob/master/workflow.md
2022-08-17 10:25:35 <omni> upcoming mosh 1.4.0 release https://github.com/mobile-shell/mosh/releases
2022-08-17 10:28:53 <ikke> yes, saw it
2022-08-17 10:28:59 <ikke> they asked about people testing the rc
2022-08-17 10:29:10 <ikke> s/about/for/
2022-08-17 10:44:16 <psykose> q66: what tools does flatpak use that aren't provided by the llvm binutils?
2022-08-17 10:48:18 <PureTryOut> ikke: thanks for the link, good to know!
2022-08-17 10:48:48 <q66> psykose: actually i think it's not flatpak, it's one of its deps (ostree i think)
2022-08-17 10:48:56 <q66> and none, it just hardcodes the eu- version
2022-08-17 10:49:03 <psykose> aha
2022-08-17 10:49:10 <psykose> at least easy to patch, heh..
2022-08-17 10:49:14 <psykose> wonder why they prefer that one
2022-08-17 10:50:23 <q66> ah, it seems that's also wrong
2022-08-17 10:50:25 <q66> i looked it up now
2022-08-17 10:50:31 <q66> it seems it's flatpak-builder that has elfutils dep
2022-08-17 10:50:40 <q66> ostree calls eu-readelf but only in tests
2022-08-17 10:51:39 <q66> https://github.com/flatpak/flatpak-builder/blob/386976adb8ab97c612394fd12118c7c7e477dda9/src/builder-utils.c#L144
2022-08-17 10:51:50 <q66> for some reason it wraps both eu-strip and strip
2022-08-17 10:51:57 <q66> and also eu-elfcompress which is not present elsewhere
2022-08-17 10:52:09 <wener> Maybe this should check firmware exists or not https://github.com/alpinelinux/alpine-conf/blob/master/setup-disk.in#L474-L480 
2022-08-17 10:52:14 <bl4ckb0ne> psykose: fwiw there's a new sdk version for spirv-tools/headers
2022-08-17 10:52:20 <psykose> is there
2022-08-17 10:52:26 <bl4ckb0ne> 1.3.224
2022-08-17 10:52:32 <q66> though elfcompress seems optional
2022-08-17 10:52:36 <psykose> it's not tagged in any github repo
2022-08-17 10:52:39 <psykose> i check all the time
2022-08-17 10:52:50 <bl4ckb0ne> they dont tag the sdk rev
2022-08-17 10:52:56 <psykose> they do for every prior one?
2022-08-17 10:53:04 <psykose> https://github.com/KhronosGroup/SPIRV-Headers/tags
2022-08-17 10:53:10 <bl4ckb0ne> they did that for the yearly ones
2022-08-17 10:53:23 <bl4ckb0ne> https://github.com/KhronosGroup/SPIRV-Tools/tree/sdk-1.3.224
2022-08-17 10:53:42 <bl4ckb0ne> https://github.com/KhronosGroup/SPIRV-Headers/tree/sdk-1.3.224
2022-08-17 10:53:56 <psykose> so we are back to no tags and changing the version scheme again i guess
2022-08-17 10:54:02 <psykose> since they don't have a tag for the actual release
2022-08-17 10:54:03 <bl4ckb0ne> heck they didnt tag it indeed for spirv-headers
2022-08-17 10:54:03 <psykose> i love that
2022-08-17 10:54:26 <bl4ckb0ne> https://github.com/KhronosGroup/Vulkan-Headers/releases/tag/v1.3.224 vuklan is properly tagged tho
2022-08-17 10:54:34 <psykose> why can't they just stick to one damn consistent tagging scheme
2022-08-17 10:54:41 <bl4ckb0ne> khronos does a very shit job at releases
2022-08-17 10:54:47 <psykose> it makes me cry
2022-08-17 10:54:52 <bl4ckb0ne> shaderc is not khronos
2022-08-17 10:54:54 <bl4ckb0ne> its google
2022-08-17 10:54:55 <psykose> yeah i know
2022-08-17 10:55:03 <psykose> but it's closely related enough they build on the same shit
2022-08-17 10:56:10 <bl4ckb0ne> oh no https://github.com/KhronosGroup/glslang/releases/tag/11.11.0
2022-08-17 10:56:14 <bl4ckb0ne> tf is that tag
2022-08-17 10:56:39 <bl4ckb0ne> https://github.com/KhronosGroup/glslang/tree/sdk-1.3.216.0 they dont even have 1.3.224
2022-08-17 10:57:11 <psykose> probably just gonna give up with those and just pick whatever individual latest version i guess
2022-08-17 10:57:20 <psykose> matching 224 release
2022-08-17 10:57:38 <psykose> oh well, at least they managed to have it working for half a year
2022-08-17 10:57:41 <psykose> i'd say that was a good run
2022-08-17 10:57:49 <bl4ckb0ne> we should just drop it
2022-08-17 10:57:55 <psykose> all of vulkan? :p
2022-08-17 10:57:55 <bl4ckb0ne> and become a non graphical distro
2022-08-17 10:57:59 <psykose> yes!!!
2022-08-17 10:58:02 <bl4ckb0ne> all 3d
2022-08-17 10:58:05 <bl4ckb0ne> just use pixman
2022-08-17 10:58:19 <psykose> speaking of we could move vulkan to main/ and add zink to mesa
2022-08-17 10:58:34 <bl4ckb0ne> https://github.com/KhronosGroup/glslang/releases/tag/untagged-305dd8c44fedc0e7814b i dont think khronos knows how to tag
2022-08-17 10:58:43 <bl4ckb0ne> vulkan isnt in main?
2022-08-17 11:00:08 <psykose> i think we actually moved it out due to it being unneeded
2022-08-17 11:00:14 <psykose> bl4ckb0ne: also click the commit of that tag
2022-08-17 11:00:21 <psykose> >  The versioning scheme is being improved, 
2022-08-17 11:00:32 <bl4ckb0ne> kek
2022-08-17 11:01:15 <psykose> ah no, wasn't
2022-08-17 11:01:26 <psykose> but yeah could be done
2022-08-17 11:01:38 <psykose> though i don't know of a single person that would be using it since no one has said anythin g
2022-08-17 11:01:46 <psykose> funnily someone did request d3d9
2022-08-17 11:11:18 <q66> [12:54] <bl4ckb0ne> its google
2022-08-17 11:11:27 <q66> coincidentally shaderc is also shit 
2022-08-17 11:12:25 <q66> in classic google fashion, they don't know what ABI and soname mean
2022-08-17 11:13:12 <bl4ckb0ne> >surprised_pikachu.png
2022-08-17 11:13:25 <bl4ckb0ne> why do we need shaderc again?
2022-08-17 11:13:47 <q66> the whole shaderc+glslang+vulkan crap+spirv crap+whatever is a massive clusterfuck
2022-08-17 11:14:47 <bl4ckb0ne> mpv it seems
2022-08-17 11:14:51 <bl4ckb0ne> and only mpv
2022-08-17 11:15:38 <q66> idk about alpine but in other places also libplacebo and gst-plugins-bad
2022-08-17 11:15:59 <bl4ckb0ne> it seems its only an optional dep for libplacebo
2022-08-17 11:16:08 <q66> opti
2022-08-17 11:16:14 <q66> useful optional dep 
2022-08-17 11:16:44 <q66> something being optional doesn't mean one should build generic packages without it 
2022-08-17 11:17:02 <q66> it's optional for mpv too fwiw
2022-08-17 11:17:18 <bl4ckb0ne> hm
2022-08-17 11:17:26 <bl4ckb0ne> we could open an issue to just complain
2022-08-17 11:17:35 <q66> about what 
2022-08-17 11:17:38 <q66> to whom 
2022-08-17 11:17:49 <bl4ckb0ne> shaderc, about how bad they tag releases
2022-08-17 11:17:55 <q66> lmao
2022-08-17 11:18:06 <q66> google will either ignore you or laugh at you 
2022-08-17 11:18:16 <bl4ckb0ne> yeah i fear this isnt going anywhere
2022-08-17 11:18:38 <q66> remember https://github.com/google/shaderc/pull/463
2022-08-17 11:18:46 <ericonr> possibly both and laugh at you from their chair without responding to the issue
2022-08-17 11:18:55 <bl4ckb0ne> isnt that the libplacebo dev
2022-08-17 11:18:59 <q66> now distros carry this downstream
2022-08-17 11:19:52 <bl4ckb0ne> marvelous
2022-08-17 11:31:47 <psykose> i'm confused what issue that mr is trying to solve
2022-08-17 11:31:52 <psykose> the link order already had it specified twice
2022-08-17 12:58:47 <omni> ncopa: both some xen and retbleed changes https://cdn.kernel.org/pub/linux/kernel/v5.x/ChangeLog-5.15.61
2022-08-17 12:59:07 <omni> 
ACTION tries to sigh optimistically

2022-08-17 13:04:12 <ncopa>  ok. im on it...
2022-08-17 13:07:18 <ncopa> omni: you can test it from edge, right?
2022-08-17 13:10:13 <omni> yes
2022-08-17 13:16:59 <ebb> None of the xen or retbleed or IBPB commit messages look promising 
2022-08-17 13:29:54 <omni> that was my impression as well, but I don't understand enough what is going on and would still want to at least give it a half-optimistic try
2022-08-17 14:26:36 <skarnet> little bump: !37788
2022-08-17 14:56:02 <Ariadne> skarnet: done
2022-08-17 14:59:05 <skarnet> thanks!
2022-08-17 15:22:24 <omni> ncopa: unfortunately that didn't help for me
2022-08-17 15:34:39 <ncopa> omni: ok. thanks for trying
2022-08-17 15:35:11 <ebb> At this point I think someone needs to attack the kernel with a debugger
2022-08-17 15:35:53 <omni> could there be something here? https://lkml.kernel.org/lkml/f4b00553-4e77-84bc-e25c-01383d7e92dd@leemhuis.info/T/#t
2022-08-17 15:36:29 <omni> also, perhaps  I should try if it works for me with linux-edge-5.19.1
2022-08-17 15:45:17 <omni> but there's no zfs-edge...
2022-08-17 19:14:35 <nmeum> when a package needs libintl.h: when should gettext-dev be used and when musl-libintl?
2022-08-17 19:20:52 <nmeum> #10825 ah
2022-08-17 20:43:08 <ikke> pabloyoyoista: fyi, deployed latest appstream generator image
2022-08-17 23:54:27 <psykose> ikke: https://gitlab.alpinelinux.org/J0WI/aports/-/jobs/806658
2022-08-17 23:54:38 <psykose> only on downgrade and only for some arches for some reason, hah
2022-08-17 23:54:46 <psykose> >libcrypto1.1-1.1.1q-r0: trying to overwrite etc/ssl1.1/cert.pem owned by ca-certificates-bundle
2022-08-18 07:53:02 <skarnet> are there policies on when to remove obsolete post-install and/or post-upgrade scripts? there are cases where such a script is *clearly* unapplicable
2022-08-18 07:54:10 <skarnet> if a script was written to upgrade from version A to version B, but we're now in version C which is compatible with B, you can install B but we don't provide A installations anymore
2022-08-18 07:54:18 <skarnet> or similar situations
2022-08-18 09:28:14 <skarnet> 4<Ariadne> skarnet: as we are doing other work to decouple busybox, please coordinate with us :)
2022-08-18 09:28:36 <skarnet> I'm branching off the busybox-initscripts in main *now*
2022-08-18 09:28:52 <skarnet> renaming it to busybox-openrc and working from there
2022-08-18 09:30:00 <skarnet> currently checking with what you said earlier and what other people wanted from this; if all goes well the MR should land in the afternoon
2022-08-18 09:49:37 <Ariadne> that seems fine
2022-08-18 10:57:50 <pabloyoyoista> ikke: thanks! However, for some reason looks like it didn't run at night the 18th? I just tried deploying to my local setup and everything worked fine apart from the segmentation faults in the parallel build
2022-08-18 10:58:07 <ikke> let me check the logs
2022-08-18 10:59:23 <ikke> pabloyoyoista: ah, you mounted ./asgen-config.json locally to ./asgen-config.json.in remotely, but locally it was renamed to asgen-config.json.in as well
2022-08-18 11:00:11 <ikke> https://gitlab.alpinelinux.org/alpine/infra/compose/appstream-generator/-/blob/master/docker-compose.yml#L19
2022-08-18 11:11:53 <pabloyoyoista> Oh sorry, that's a stupid mistake. https://gitlab.alpinelinux.org/alpine/infra/compose/appstream-generator/-/merge_requests/7
2022-08-18 11:12:17 <pabloyoyoista> I do not use the compose, but other setup, thus I didn't realize it, sorry the annoyance
2022-08-18 11:16:05 <skarnet> there. Ariadne: !37834
2022-08-18 11:16:32 <Ariadne> ok will review some point today.  have a lot of meetings
2022-08-18 11:19:34 <skarnet> cool, thanks
2022-08-18 12:48:32 <omni> how does build-riscv64-emulated work in CI? is it manual just in the sense that you have to push the play button? (as I did, it's "pending" and I'm waiting for it to run)
2022-08-18 13:07:35 <Ariadne> note that riscv64 CI is very slow
2022-08-18 13:07:42 <Ariadne> it may take a while to get to running your job
2022-08-18 13:07:42 <Ariadne> :p
2022-08-18 13:27:36 <ikke> omni: and yes, you need to push the play button. The idea is that it does not deley the bulk of the MRs, but you can run it in case you want to know if it builds on rv64
2022-08-18 13:40:18 <omni> it's for !37833
2022-08-18 13:41:22 <ikke> omni: I think the rv64 builder is stuck again
2022-08-18 13:41:49 <ikke> Load average: 136.00 136.01 136.00
2022-08-18 13:45:03 <omni> when will we get our 256core rv64 w/ 1TB of ram?
2022-08-18 13:45:51 <omni> hello? mixter Future? where are you?
2022-08-18 13:52:46 <ikke> omni: it's running now
2022-08-18 13:55:30 <omni> oh! and there I pushed a new commit and it got cancelled..
2022-08-18 14:06:22 <omni> it's running now, very cool!
2022-08-18 14:43:35 <omni> psykose: meh, I wanted to squash those commits in that draft
2022-08-18 14:43:48 <psykose> looked ok as is :)
2022-08-18 14:45:12 <omni> well, the message for the second commit wasn't very good, to my "standard" =)
2022-08-18 14:46:03 <omni> was also curious to see if there were any particular test to disable for RV64
2022-08-18 14:49:00 <omni> but that's just nitpicking! it's good
2022-08-18 15:03:18 <bl4ckb0ne> psykose: i should complain to khronos for having such shitty tags/release process
2022-08-18 15:04:04 <psykose> omni: there usually are some tests to disable, but since they don't run it's a bit moot, and even when they do fail half the time it's a qemu/speed issue and not rv specifically (think armhf test timeouts again..)
2022-08-18 15:04:11 <psykose> bl4ckb0ne: you should :) wonder who else has
2022-08-18 15:07:11 <bl4ckb0ne> probably not arch, they are very late
2022-08-18 15:27:31 <omni> psykose: ah, you mean check() isn't run at all on rv64?
2022-08-18 15:30:05 <ikke> omni: no, we've globally disabled it for the time being
2022-08-18 15:40:25 <zekra> By the name of Allah most great all able MALIK ALMOLOK almofadda - Allah is the only true God - Sunni Islam is the right religion - By the Greatest authority of Allah most great all capable
2022-08-18 15:40:25 <zekra> It is requested to leave fighting islam - leave harming muslim sunnis - not allter and divert islamic sources and books and religious heritage - leave harming poor people - don't revenge after reading these messages - don't make white/black people trouble to make wars between them - under consequences of falling down hard from Allah the most great all capable addayan
2022-08-18 15:41:07 <psykose> alter*
2022-08-18 16:44:15 <PureTryOut> hmm fun, applications using `/usr/include/SDL2/SDL_syswm.h` can't compile as it can't find `# include <directfb.h>`. Makes sense because it's in `/usr/include/directfb/directfb.h`, which would require `#include <directfb/directfb.h>` no?
2022-08-18 16:44:44 <ikke> PureTryOut: or a -I directive is missing
2022-08-18 16:45:00 <PureTryOut> `-ldirectfb` or something?
2022-08-18 16:45:13 <ikke> -l is for libraries
2022-08-18 16:45:17 <ikke> -I is include path
2022-08-18 16:45:53 <PureTryOut> I... Do not see the difference. Is one a capital i and the other a l?
2022-08-18 16:46:01 <ikke> yes
2022-08-18 16:46:14 <PureTryOut> i being include path?
2022-08-18 16:46:17 <ikke> correct
2022-08-18 16:46:32 <ikke> -l is to tell the linker to link against that library
2022-08-18 16:46:37 <PureTryOut> ok. And how would I pass this to CMake?
2022-08-18 16:46:47 <psykose> no, #include <directfb.h> is correct
2022-08-18 16:47:01 <psykose> directfb.pc passes -I /usr/include/directfb
2022-08-18 16:47:14 <PureTryOut> ok then idk what's up here
2022-08-18 16:47:34 <ikke> can you verbosely print the command it invokes?
2022-08-18 16:47:39 <psykose> now if something doesn't see that, they either have their own broken detection or similar
2022-08-18 16:47:47 <psykose> tell me what fails and i'll look at it
2022-08-18 16:48:02 <PureTryOut> 
ACTION sent a code block: https://matrix.org/_matrix/media/r0/download/matrix.org/tjnRlpwdTumvAcLzxbEXSqhg

2022-08-18 16:48:05 <psykose> also with regards to the qt6 opengl thing, wasn't it the case that qt can only support one qt thing at once
2022-08-18 16:48:05 <PureTryOut> ^ that fails
2022-08-18 16:48:09 <psykose> one gl thing*
2022-08-18 16:48:20 <PureTryOut> psykose: it can yes, but afaik I had only OpenGLES enabled
2022-08-18 16:48:21 <psykose> for arm* you can only pick desktop or gles and all of qt has to target it
2022-08-18 16:48:31 <psykose> debian builds both versions separately for instance
2022-08-18 16:48:50 <psykose> aha, yuzu
2022-08-18 16:48:53 <psykose> alright, i'll go look 
2022-08-18 16:49:02 <PureTryOut> I don't have a special interest in keeping desktop OpenGL around for arm, almost everything needs GLES there anyway
2022-08-18 16:49:13 <psykose> mhm
2022-08-18 16:49:15 <PureTryOut> note that yuzu doesn't compile to that point without a different "fix"
2022-08-18 16:49:17 <ikke> I don't see -I/usr/include/directfb in there
2022-08-18 16:50:45 <psykose> can you paste the full error
2022-08-18 16:52:35 <PureTryOut> 
ACTION sent a code block: https://matrix.org/_matrix/media/r0/download/matrix.org/DDrHETZkNjXFLSGwwGNNXJma

2022-08-18 16:52:38 <PureTryOut> psykose: ^
2022-08-18 16:53:01 <psykose> did you pass YUZU_USE_EXTERNAL_SDL2 by chance
2022-08-18 16:53:56 <psykose> try YUZU_USE_BUNDLED_SDL2=OFF ENABLE_SDL2=ON YUZU_USE_EXTERNAL_SDL2=OFF instead, but i'll go try
2022-08-18 16:54:07 <psykose> (just reading through the cmake)
2022-08-18 16:58:15 <psykose> aha
2022-08-18 16:58:23 <psykose> for once it's the pkgconfig file that's not up to par
2022-08-18 16:59:30 <psykose> or well, no, it's the cmake sdl file, so worse
2022-08-18 16:59:33 <PureTryOut> Oh lol
2022-08-18 17:00:28 <PureTryOut> (and yes I did pass that, in fact you can read the WIP APKBUILD here https://gitlab.alpinelinux.org/PureTryOut/aports/-/blob/testing_yuzu/testing/yuzu/APKBUILD)
2022-08-18 17:01:13 <PureTryOut> oh I'm passing OFF to that 🤔
2022-08-18 17:01:18 <PureTryOut> should probably change that
2022-08-18 17:04:05 <PureTryOut> oh no OFF is the correct value there, nvm
2022-08-18 17:06:45 <psykose> https://img.ayaya.dev/hIw6XSVHg3wz that makes it build that part fine
2022-08-18 17:06:50 <psykose> rest is up to you :)
2022-08-18 17:08:26 <PureTryOut> oh fun, moving to pkgconfig lol
2022-08-18 17:08:32 <PureTryOut> I suppose upstream SDL2 needs fixing
2022-08-18 17:14:02 <psykose> yeah the cmake files don't seem to do anything with directfb
2022-08-18 17:14:09 <psykose> even though it needs an include path
2022-08-18 17:14:19 <psykose> didn't check more than that, but i don't expect to find anything special
2022-08-18 17:14:27 <psykose> maybe the next sdl release will fix it
2022-08-18 17:14:31 <PureTryOut> it indeed goes through now, thanks
2022-08-18 17:14:32 <psykose> heard there's lots of changes
2022-08-18 17:33:39 <PureTryOut> hmm it fails to link to SDL2 now
2022-08-18 17:34:52 <psykose> probably have to modify another line to add the correct links
2022-08-18 17:36:14 <psykose> it adds some link below but it's probably wrong with pkgconfig, so you need to modify the SDL2_LIBRARIES to be something else i would guess
2022-08-18 17:36:30 <psykose> you can message() random stuff around there and see what they are and which one would have -lSDL2
2022-08-18 17:36:44 <psykose> if it's already correct then the linking would have to be fixed elsewhere
2022-08-18 17:42:04 <bl4ckb0ne> PureTryOut: iirc it has been fixed
2022-08-18 17:43:29 <bl4ckb0ne> https://github.com/libsdl-org/SDL/issues/3930
2022-08-18 17:43:50 <bl4ckb0ne> should be in the next release
2022-08-18 17:44:53 <bl4ckb0ne> https://github.com/libsdl-org/SDL/releases there's a RC, maybe i should update the package
2022-08-18 17:45:17 <PureTryOut> oh I'd love to have that
2022-08-18 17:45:43 <bl4ckb0ne> should be in the 2.23.1
2022-08-18 17:45:52 <bl4ckb0ne> i can do it if you waht
2022-08-18 17:48:18 <PureTryOut> well for me sure, but we need to be sure we won't break other stuff because we update to a RC rather than a full release
2022-08-18 17:48:27 <bl4ckb0ne> meh its a prerelease, url is different
2022-08-18 17:48:55 <bl4ckb0ne> alright then, gotta wait
2022-08-18 17:49:20 <PureTryOut> hmm?
2022-08-18 17:49:53 <bl4ckb0ne> RC stuff
2022-08-18 18:05:05 <psykose> bl4ckb0ne: i assume https://github.com/libsdl-org/SDL/pull/5803/files fixes it
2022-08-18 18:05:09 <psykose> the change to sdlchecks.cmake
2022-08-18 18:05:12 <psykose> i'll happily accept that
2022-08-18 18:06:03 <psykose> (by pressing the green merge button if you backport it)
2022-08-18 18:07:12 <bl4ckb0ne> a tad busy at the moment but yes that's the fix
2022-08-18 18:07:31 <bl4ckb0ne> 2.23 or 2.24 should be right around the corner though
2022-08-18 18:07:37 <psykose> sure sure
2022-08-18 18:16:39 <PureTryOut> I mean, I can make that MR if you want
2022-08-18 18:17:42 <psykose> go for it
2022-08-18 18:38:35 <PureTryOut> psykose: https://gitlab.alpinelinux.org/alpine/aports/-/merge_requests/37864
2022-08-18 18:38:50 <PureTryOut> had to modify the patch slightly to apply with our existing patch and remove the Github stuff
2022-08-18 18:40:03 <bl4ckb0ne> you could've taken only the cmake stuff from sdlchecks.cmake
2022-08-18 18:40:24 <bl4ckb0ne> i dont think the rest is relevant in this context
2022-08-18 18:41:53 <PureTryOut> true I suppose
2022-08-18 18:42:01 <psykose> the existing patch actually tries to fix the same thing
2022-08-18 18:42:03 <psykose> but does it wrong
2022-08-18 18:42:05 <psykose> you can delete it
2022-08-18 18:42:10 <PureTryOut> ha ok
2022-08-18 18:42:16 <PureTryOut> I'll minimize this patch and remove the other then
2022-08-18 18:52:00 <psykose> did it fix all of that linking/includedir stuff locally
2022-08-18 18:53:28 <Ellenor> eph
2022-08-18 18:54:01 <PureTryOut> I was just testing that, doesn't seem like it no 😢
2022-08-18 18:54:59 <psykose> if it doesn't even fix the includedir then i guess it's incomplete
2022-08-18 18:55:04 <psykose> linking could be a separate issue
2022-08-18 18:56:19 <PureTryOut> it didn't fix either
2022-08-18 19:16:39 <PureTryOut> hmm even that pre-release of SDL2 doesn't help
2022-08-18 19:16:49 <PureTryOut> with or without my changes to yuzu's CMakeLists.txt
2022-08-18 19:18:00 <psykose> unlucky
2022-08-18 19:18:08 <psykose> also libksane-dev needs to depend on ksanecore-dev
2022-08-18 19:38:23 <PureTryOut> it does in my MR
2022-08-18 19:40:11 <psykose> no, it's in makedepends, which is not -dev
2022-08-18 19:40:27 <psykose> that and okular is missing some deps (yes i am backseating the mr)
2022-08-18 19:40:29 <psykose> : )
2022-08-18 19:40:32 <psykose> haha
2022-08-18 19:44:28 <PureTryOut> oh sorry, that way lol
2022-08-18 20:00:42 <PureTryOut> pretty sure only community/okular will fail, but waiting for CI to finish that is annoying. I might just merge it and package chmlib as a dep when that comes up on the builders
2022-08-18 20:02:16 <psykose> package it then just push it
2022-08-18 20:02:19 <psykose> i'll fix it anyway
2022-08-18 20:02:33 <psykose> i watch them most hours of the day and fix whatever comes up so i don't mind
2022-08-18 20:03:54 <PureTryOut> I don't really want to package chmlib tbh 😛 But I probably should, so let's do it I guess
2022-08-18 20:04:33 <psykose> what, not a fan of 2009 microsoft format abandonware? i'm shocked
2022-08-18 20:04:39 <psykose> it can also be skipped with a cmake flag
2022-08-18 20:04:42 <psykose> i assume the support is optional
2022-08-18 20:05:36 <PureTryOut> I know about the cmake flag, I was trying that out but can't get the format down
2022-08-18 20:05:57 <PureTryOut> upstream already uses it internally too, I don't think I want to completely overwrite it but rather append to it https://invent.kde.org/graphics/okular/-/blob/master/CMakeLists.txt#L19
2022-08-18 20:06:19 <PureTryOut> so I'm hoping just setting it in the CMake call appends too it rather than overwriting, but I don't think it does
2022-08-18 20:06:49 <psykose> oof, yeah
2022-08-18 20:06:55 <psykose> reusing the same name instead of adding an extra one
2022-08-18 20:07:08 <PureTryOut> yeah...
2022-08-18 20:07:16 <PureTryOut> so now I have to patch the thing, which is such a hack
2022-08-18 20:07:20 <psykose> it explicitly skips that list given the `if not`
2022-08-18 20:07:28 <PureTryOut> rather that than chmlib though...
2022-08-18 20:07:30 <psykose> but i guess a patch on that one line and adding ;thing isn't too bad
2022-08-18 20:07:53 <PureTryOut> yeah I'll just patch it
2022-08-18 20:17:15 <PureTryOut> oh fun it just ignores KF5KHtml in that list (it was already there from upstream)
2022-08-18 20:18:31 <PureTryOut> oh CHM was already there too actually
2022-08-18 20:22:24 <PureTryOut> CHM is actually not required, and I can just add KHTML to the deps, so I'll just do that and be done with it
2022-08-18 22:26:19 <psykose> nah, whole list is for android
2022-08-19 07:12:25 <PureTryOut> Oh I missed that fact...
2022-08-19 07:35:14 <PureTryOut> psykose bl4ckb0ne: regarding the SDL2 with DirectFB CMake issue we talked about yesterday, I reported upstream that the merged fix earlier didn't actually work, and now they made a PR which _does_ work. It however only applies to 2.23.2 (and later 2.24.0)
2022-08-19 07:54:33 <psykose> it applies with enough hacks, though yes
2022-08-19 07:54:37 <psykose> thanks for reporting it
2022-08-19 07:54:43 <psykose> i usually don't have enough energy
2022-08-19 07:55:07 <PureTryOut> I've updated https://gitlab.alpinelinux.org/alpine/aports/-/merge_requests/37864
2022-08-19 07:55:28 <PureTryOut> and np!
2022-08-19 09:33:39 <bl4ckb0ne> thanks for taking care of this PureTryOut 
2022-08-19 09:39:09 <PureTryOut> no problem!
2022-08-19 10:22:25 <ikke> psykose: seems like the illigal instructions in the rspamd test suite also appear in the program itself
2022-08-19 10:22:30 <ikke> #14114
2022-08-19 10:22:50 <psykose> no
2022-08-19 10:22:57 <psykose> that's luajit from the backtrace
2022-08-19 10:23:04 <psykose> which is build with -msse4.2
2022-08-19 10:23:09 <ikke> ok
2022-08-19 10:23:17 <psykose> and that is an sse4.1 instruction
2022-08-19 10:23:23 <psykose> go blame jirutka i told him 20 times
2022-08-19 10:23:33 <psykose> it doesn't even make sense to use these on a JIT that generates native instructions anyway
2022-08-19 10:24:02 <ikke> Assigned the issue to him
2022-08-19 10:24:10 <psykose> it would be probably the 4th issue about this exact same thing
2022-08-19 10:24:32 <psykose> this is the other #13688
2022-08-19 10:24:44 <psykose> then there's been maybe 2 ML posts
2022-08-19 10:25:41 <psykose> https://lists.alpinelinux.org/~alpine/users/%3CCADH9QNoHup814iMEzuALEOb2RzjoPun0FpJ86Aoyf%2B254yMHfg%40mail.gmail.com%3E
2022-08-19 10:26:37 <psykose> the rspamd stuff specifically is rspamd which does dynamic detection and vectorscan which i fixed with a high degree of certainty, just in case is that backtrace is wrong, but i doubt it
2022-08-19 10:26:42 <psykose> (could also both apply at once)
2022-08-19 10:26:55 <psykose> it's hard to test for sure without making a vm with the instructions disabled
2022-08-19 10:29:26 <psykose> last i tested a benchmark luajit even has identical performance with like anything thrown at it because the makefile throws half of it out
2022-08-19 10:29:33 <psykose> but any sse or native didn't really change anything
2022-08-19 10:29:43 <psykose> as a haver of a modern cpu even i don't see the point
2022-08-19 11:46:34 <ncopa> too bad the linux-edge and linux-lts kernel configs are not in sync. now we dont know if its related the -lts config or just kernel version that makes xen crash. https://gitlab.alpinelinux.org/alpine/aports/-/issues/14090#note_256587
2022-08-19 11:47:50 <psykose> fixing that doesn't sound that hard
2022-08-19 11:55:24 <ncopa> not hard, technically, just time consuming. and there are soft issues as well
2022-08-19 11:56:28 <psykose> i mean in the sense of maintaining an -edge that just follows -lts + next-changes + potentially-experimental changes isn't necessarily 'double effort' over just lts
2022-08-19 11:56:32 <psykose> the current edge is not an edge
2022-08-19 11:57:09 <psykose> if anything -lts is edge (how did simpledrm get there first?) and -edge is $randomconfigurationofeverythingdisabled
2022-08-19 11:57:56 <psykose> just so happens to also follow latest-stable instead of lts, but i am still to this day confused what the purpose is meant to be, as everyone has overlooked it for seemingly forever
2022-08-19 11:59:14 <psykose> i'm 50% towards just building my own kernels again, since -lts does not have the random hardware support i want, and edge has random stuff missing i run into all the time, like even zswap is disabled lol
2022-08-19 11:59:47 <psykose> and proot doesn't work, but i never figured out what option is missing for that
2022-08-19 13:22:28 <panekj> one could complain to mps? (:
2022-08-19 13:24:26 <psykose> if you told mps to 'sync the changes to lts' he would probably start another rant about how alpine is not "small, simple" anymore or something
2022-08-19 13:24:46 <psykose> one could merely make linux-actualedge and simply commit it
2022-08-19 13:43:40 <skarnet> he will start this rant no matter what you say, so ask him anyway :P
2022-08-19 13:47:57 <psykose> well yes, i have heard it at least.. 6 times now?
2022-08-19 13:50:44 <skarnet> and you haven't yet learned to tune it out? :D
2022-08-19 13:52:36 <psykose> it's tuned out beyond belief
2022-08-19 13:52:51 <psykose> but then you reach the real issue of needing to actually use the damn kernel and you have to open your ears a bit
2022-08-19 13:53:37 <psykose> or well, just enough to complain and not enough to do anything
2022-08-19 20:23:29 <PureTryOut> bl4ckb0ne: psykose SDL2 2.24.0 is out 😄
2022-08-19 20:23:34 <PureTryOut> I'll update my MR
2022-08-19 20:27:12 <PureTryOut> 
ACTION doesn't understand why projects choose to version so that odd minor numbers are pre-releases and even numbers are stable

2022-08-19 20:34:57 <ayakael> Pero porque libsdl
2022-08-19 20:37:55 <bl4ckb0ne> last rev scheme was worst
2022-08-20 04:17:36 <psykose> PureTryOut: saw :) it came out the second i went to bed
2022-08-20 06:53:06 <psykose> ikke: is there any chance i can get access to the riscv64 builder and you can teach me how to start/stop it so i can manually do some rust stuff
2022-08-20 06:54:06 <ikke> psykose: I'd have to verify that with clandmeter
2022-08-20 07:56:45 <psykose> currently losing my mind on why it randomly hangs with 80% chance in the same spot
2022-08-20 07:57:03 <psykose> always lzma-sys of the very first part
2022-08-20 08:28:23 <nmeum> psykose: what's the build issue on riscv64? can it be reproduced outside the builder environment? if so: I could debug it on my unleashed
2022-08-20 08:28:35 <psykose> i think it's qemu specific sadly
2022-08-20 08:28:38 <psykose> uhh it randomly hangs
2022-08-20 08:28:39 <psykose> that's all
2022-08-20 08:28:47 <psykose> for me, on a riscv container, the cpu is even at 0%
2022-08-20 08:28:56 <psykose> just forever stuck on specific steps at random
2022-08-20 08:29:15 <psykose> either lzma-sys during very start, or some internal crate on stage0, etc
2022-08-20 08:29:27 <psykose> aside from that, it also now segfaults trying to build 1.63 (when it doesn't get stuck)
2022-08-20 08:29:44 <psykose> i momentarily hacked it to pretend it's up to date
2022-08-20 08:29:54 <psykose> 274194fd279d7155218ffbfd7e8154d4e9883327
2022-08-20 08:29:55 <nmeum> ah so rustc build issues again meh :(
2022-08-20 08:30:00 <psykose> so we can at least build other things
2022-08-20 08:30:10 <psykose> (rust always builds almost first due to many dependents)
2022-08-20 08:30:52 <psykose> now i'm trying to build it with a pkgrel bump only (still 1.62.1, but a rebuild) so at least it also works (cargo needs the openssl3 rebuild on riscv)
2022-08-20 08:31:12 <psykose> and so far this is getting-randomly-stuck-fiesta
2022-08-20 08:31:17 <psykose> i think i'm at attempt 35
2022-08-20 08:31:55 <psykose> the dev container has no ptrace so i can't say much more than that, but last ikke checked it was a random infinite loop (with 100% cpu iirc), so i guess for me it's something else (no cpu)
2022-08-20 08:32:33 <psykose> if you have an unleashed doing nothing else, feel free to throw rust 1.63 at it and see if it works
2022-08-20 08:32:40 <nmeum> if you use qemu-user you can instruct it to do some tracing for you
2022-08-20 08:32:45 <psykose> it takes a long time but if it passes we can rule out entire classes of issues
2022-08-20 08:32:50 <psykose> from inside the container?
2022-08-20 08:32:54 <psykose> i can't actually 'run qemu user'
2022-08-20 08:33:01 <nmeum> no, need to do it outside the container, i.e. invoke qemu-user with specific flags
2022-08-20 08:33:07 <psykose> no access for that then
2022-08-20 08:33:15 <psykose> but yeah go give it a rebuild yourself
2022-08-20 08:33:30 <psykose> just revert the commit above to start and let it build
2022-08-20 08:33:42 <psykose> if it passes, we can confirm literally all the issues are just qemu
2022-08-20 08:34:10 <psykose> if it doesn't ever get stuck but does segfault, then we can confirm the getting stuck part is qemu, and the segfault is something else
2022-08-20 08:34:42 <nmeum> does the stuck/segfault happen early during the build or does it take hours to get there?
2022-08-20 08:34:55 <psykose> segfault is some stage1 part, stuck is random (as soon as the very first step)
2022-08-20 08:35:57 <nmeum> ok, I will try to find some time to look into it
2022-08-20 08:36:51 <psykose> thanks
2022-08-20 08:37:26 <psykose> i am also currently looking at ghc for aarch64, was there any historical writing on how that went
2022-08-20 08:41:08 <nmeum> not sure, I think someone on irc wanted to look into it many months ago
2022-08-20 08:42:18 <psykose> sure sure
2022-08-20 08:42:32 <ikke> psykose: I have a visionfive
2022-08-20 08:42:37 <psykose> the `sed -i -e 's/unknown-linux-gnueabihf/alpine-linux/g' llvm-targets` looks wrong
2022-08-20 08:42:40 <ikke> I can try to build rust there
2022-08-20 08:43:06 <nmeum> because of the triplet change we also can't use the ghc bindest for bootstrapping
2022-08-20 08:43:18 <nmeum> *bindist
2022-08-20 08:43:21 <psykose> yeah, but the triplet also has no -musl on the end
2022-08-20 08:43:28 <psykose> which does not match our llvm target as the comment says
2022-08-20 08:44:01 <psykose> shouldn't it be something like https://img.ayaya.dev/EMvBkOvxlMJh
2022-08-20 08:44:54 <psykose> ah
2022-08-20 08:44:57 <psykose> no, they also strip it
2022-08-20 08:45:27 <psykose> strange, cause the other things in the file have the musl suffix
2022-08-20 08:46:01 <psykose> and we pass --target=$CTARGET which as the suffix
2022-08-20 08:46:08 <psykose> but the build then says  Target: aarch64-alpine-linux
2022-08-20 08:46:31 <psykose> ikke: sure, go for it
2022-08-20 08:56:39 <ikke> rust: rustc-1.62.1-src.tar.xz is missing in checksums
2022-08-20 08:59:09 <psykose> revert the commit above
2022-08-20 08:59:21 <psykose> or the last commit on it from history
2022-08-20 08:59:31 <ikke> ok
2022-08-20 09:06:36 <ikke> just installing and unpacking already takes 5 minutes :P
2022-08-20 09:08:38 <psykose> haha
2022-08-20 09:08:42 <psykose> even slower than qemu-user
2022-08-20 09:08:58 <nmeum> yea, even the unleashed is slower than qemu-user
2022-08-20 09:09:03 <psykose> i assume you don't have any decent storage and it's just some sd-card
2022-08-20 09:09:38 <nmeum> I have an nvme ssd in my unleashed but I don't think that that's the bottleneck :D
2022-08-20 09:09:47 <ikke> psykose: correct
2022-08-20 09:09:53 <psykose> unlucky
2022-08-20 09:10:01 <ikke> I do have an m2 ssd in a usb case I could attach to it
2022-08-20 09:10:01 <psykose> nmeum: it's the bottleneck for unpacking most likely
2022-08-20 09:12:30 <ikke> I guess it's going to take days to compile
2022-08-20 09:12:43 <psykose> yeah
2022-08-20 09:13:03 <psykose> but if it compiles at all.. go upload the artifacts to the builder directly
2022-08-20 09:13:21 <ikke> I'm leaving for holiday tomorrow
2022-08-20 09:13:28 <psykose> for how long
2022-08-20 09:13:40 <ikke> 2 weeks
2022-08-20 09:13:41 <psykose> aw
2022-08-20 09:13:47 <psykose> who's going to fix all the broken things with you gone
2022-08-20 09:13:55 <psykose> i don't have access to anything
2022-08-20 09:14:00 <ikke> ncopa / clandmeter :)
2022-08-20 09:14:35 <ikke> I wonder if I should disable checks
2022-08-20 09:14:42 <ikke> I probably should
2022-08-20 09:19:03 <psykose> no, they pass
2022-08-20 09:19:08 <psykose> and take like 30 seconds
2022-08-20 09:19:15 <psykose> it's just a sanity check
2022-08-20 09:19:19 <psykose> if it fails those it doesn't even work
2022-08-20 09:19:19 <ikke> oh, ok
2022-08-20 09:28:32 <nmeum> fyi: i also started building it on the unleashed
2022-08-20 09:29:46 <ikke> Does that have a faster cpu?
2022-08-20 09:32:41 <nmeum> I think they both have a SiFive U740 but mine has an ssd at least, so maybe that help's a bit :D
2022-08-20 09:33:15 <ikke> heh
2022-08-20 09:33:17 <psykose> 4x U54 + 1x E51 instead of 2x U74
2022-08-20 09:33:51 <nmeum> psykose: https://tpaste.us/QNe0 segfaults when compiling the regex create
2022-08-20 09:34:08 <psykose> interesting
2022-08-20 09:34:11 <psykose> usually happens a bit later
2022-08-20 09:34:16 <psykose> but yeah it's broken
2022-08-20 09:34:25 <nmeum> I actually have the unmatched, not the unleashed (I keep mixing those up, sorry)
2022-08-20 09:34:51 <psykose> could you try https://img.ayaya.dev/OV42KYugfC46 without reverting the commit
2022-08-20 09:34:57 <nmeum> hm, I have the feelign that rust upstream doesn't really test rustc on riscv64 musl
2022-08-20 09:35:11 <nmeum> prior to 1.62 I also ran into segfaults with the boostrapped compiler
2022-08-20 09:35:32 <psykose> (that one is just a pkgrel bump of 1.62.1)
2022-08-20 09:37:10 <ikke> nmeum: seems to have passed for me
2022-08-20 09:37:20 <ikke> Compiling regex v1.5.5     
2022-08-20 09:37:27 <panekj> nmeum: it's tier 3 so I wouldn't expect them to do anything with rv on musl
2022-08-20 09:38:26 <psykose> i think we added the riscv port a few years too early if anything
2022-08-20 09:38:35 <panekj> > Tier 3 targets are those which the Rust codebase has support for, but which the Rust project does not build or test automatically, so they may or may not work. Official builds are not available.
2022-08-20 09:39:01 <psykose> also
2022-08-20 09:39:05 <psykose> i build aarch64 gcc successfully
2022-08-20 09:39:35 <psykose> ikke: if you let me i can upload it to aarch64 builders and we can enable it there
2022-08-20 09:39:43 <psykose> though i guess i should test actually compiling something with it first
2022-08-20 09:39:47 <psykose> ghc*
2022-08-20 09:40:06 <nmeum> powerpc64le-unknown-linux-musl is also a tier 3
2022-08-20 09:41:47 <psykose> lets see if it can build pandoc at least
2022-08-20 09:41:52 <psykose> if it does i guess we have ghc on aarch64
2022-08-20 09:42:07 <nmeum> ikke: yea, it might be racy
2022-08-20 09:42:33 <nmeum> pandoc might not build atm due to #14105
2022-08-20 09:43:18 <ikke> psykose: I think the main reason rust was pushed is due to py3-cryptography
2022-08-20 09:43:25 <psykose> pandoc built earlier
2022-08-20 09:43:28 <psykose> when i tested
2022-08-20 09:43:37 <psykose> ikke: i mean the entire riscv port, not rust specifically
2022-08-20 09:43:45 <ikke> ah
2022-08-20 09:43:49 <nmeum> except for rust the riscv port works fine for me so far
2022-08-20 09:44:10 <psykose> sigh, need to bootstrap cabal too
2022-08-20 09:44:11 <nmeum> go and other fancy new programming languages also work fine with it
2022-08-20 09:44:28 <psykose> sure, i guess the 2 people with a riscv board can use it
2022-08-20 09:44:29 <nmeum> psykose: that's easy because we have a stage0 for it, should basically work automatically
2022-08-20 09:44:43 <psykose> for me it's nothing but endless headaches of fixing things on it
2022-08-20 09:46:01 <psykose> stage0 cabal bootstrap fails on ghc-pkg describe rts-1.0
2022-08-20 09:46:07 <psykose> i assume it needs to be updated for the new ghc version
2022-08-20 09:46:21 <nmeum> I mean fixing stuff on architectures you don't use or have access to is always annoying
2022-08-20 09:46:32 <nmeum> same thing with s390x for example which is also fairly obscure
2022-08-20 09:46:54 <nmeum> that's also why I proposed https://gitlab.alpinelinux.org/alpine/tsc/-/issues/48
2022-08-20 09:47:10 <nmeum> psykose: I updated the stage0 for ghc 9.0.2
2022-08-20 09:47:16 <nmeum> it works on x86_64 at least
2022-08-20 09:48:01 <nmeum> psnszsn: with your patch and without the revert it builds the regex crate
2022-08-20 09:48:04 <nmeum> oops
2022-08-20 09:48:05 <nmeum> psykose: ^
2022-08-20 09:48:09 <psykose> progress
2022-08-20 09:48:20 <psykose> i forgot to git pull
2022-08-20 09:49:42 <psykose> builds now
2022-08-20 09:50:42 <nmeum> great!
2022-08-20 09:51:07 <psykose> hopefully it passes
2022-08-20 09:51:21 <psykose> ikke: so yeah what is the process after i verify ghc working on aarch64 :)
2022-08-20 09:51:33 <psykose> it's built via bootstrap.sh so it has to be manually copied into there
2022-08-20 09:55:18 <nmeum> psykose: what' the purpose of your rustc patch btw, it does still build 1.62 instead of 1.63, no?
2022-08-20 09:55:22 <nmeum> *what's
2022-08-20 09:58:30 <psykose> yes, just to see if it can rebuild at all
2022-08-20 09:58:52 <psykose> rust was never rebuilt after building the first time
2022-08-20 09:59:11 <psykose> it's also what makes it unusable on riscv now, as it never rebuilt against openssl3
2022-08-20 09:59:37 <psykose> we're going to be out of time for it to be buildable at all soon
2022-08-20 09:59:51 <psykose> rust can only build with previous version, and there is a new one every 6 weeks, and .63 just came out
2022-08-20 10:00:11 <psykose> once .64 comes out either we have to wait for riscv to get .63 or upgrading to .64 will need manual work to fix it
2022-08-20 10:00:23 <nmeum> or re-bootstrap
2022-08-20 10:00:28 <psykose> yes, which is a huge pain
2022-08-20 10:00:41 <psykose> if it can't even build itself without 200 restarts and fails then.. it just doesn't work, idk
2022-08-20 10:00:57 <psykose> the s390x one for comparison just works so far
2022-08-20 10:01:50 <psykose> and in the meantime, basically anything using rust is always going to be broken in between those bootstrap steps, etc. the riscv builder just endlessly falls behind as long as rust doesn't work
2022-08-20 10:02:50 <nmeum> I think we should just report the segfaults on riscv64 upstream, since it seems that they don't test it and are thus not aware of them
2022-08-20 10:03:11 <nmeum> I don't think we can fix those ourselves without knowledge of rustc internals
2022-08-20 10:03:15 <psykose> you can, but it's tier3, which has no artifacts
2022-08-20 10:03:33 <nmeum> if rust upstream is not committed to supporting riscv64 musl then there is nothing we can do about it
2022-08-20 10:03:42 <nmeum> s390x and ppc64le are also tier 3
2022-08-20 10:04:06 <psykose> yeah, and the moment they fail i don't expect anyone from upstream to spend 10 hours bootstrapping it to test the failures
2022-08-20 10:05:41 <psykose> every issue i can remember in general for s390x/ppc64le was resolved through the help of someone from IBM
2022-08-20 10:06:02 <psykose> riscv has no equivalent, on top of being much more not-mature, etc
2022-08-20 10:06:07 <panekj> you could try raise them to tier 2 but you would need someone (or yourself) to become that target maintainer
2022-08-20 10:06:24 <ikke> The issue is lack of suitable hw I suppose
2022-08-20 10:06:34 <ikke> and qemu-user is not ideal apparnetly
2022-08-20 10:06:37 <ikke> apparently*
2022-08-20 10:07:46 <psykose> yes, it makes more sense to cross compile for such targets
2022-08-20 10:09:41 <panekj> that being said, raising it to next tier will probably not be an easy task
2022-08-20 10:09:59 <panekj> https://doc.rust-lang.org/nightly/rustc/target-tier-policy.html#tier-2-target-policy
2022-08-20 10:14:49 <nmeum> to me this is more of issue with rust then issue with riscv64 (qemu-user, lack of hw etc doesn't help ofc), but idk I don't see how we solve this downstream
2022-08-20 10:15:33 <nmeum> I think Ariadne used to be in rust's internal zulip or something maybe we could ask them if there is an interest in improving the rust riscv64 musl port through this channel
2022-08-20 10:16:05 <psykose> yeah, it's a rust issue
2022-08-20 10:17:08 <panekj> did you open issue on rust repo?
2022-08-20 10:19:37 <nmeum> I did open one for the 1.60 ICEs I encountered after compiling rustc itself with a boostrapped stage2 but never received a response
2022-08-20 10:20:28 <nmeum> also upstreamed a few minor riscv64 patches for the libc create
2022-08-20 10:22:00 <panekj> it would be good to know who is target owner
2022-08-20 10:22:35 <ikke> rust is still continueing for be btw
2022-08-20 10:25:41 <nmeum> psykose: 1.16 failed to compile with your patch on the unmatched when building regex-syntax btw https://tpaste.us/Kebo
2022-08-20 10:26:01 <psykose> guess the random segfaults are not qemu then
2022-08-20 10:26:25 <psykose> i am in the middle of stage1 currently
2022-08-20 10:26:25 <nmeum> I would assume there is probably some memory corruption or something and hence it is racy and sometimes works (e.g. for ikke) and sometimes doesn't or something along those lines
2022-08-20 10:26:30 <psykose> yep
2022-08-20 10:26:43 <nmeum> panekj: how do I find out how the target owner is?
2022-08-20 10:26:47 <nmeum> s/how/who/
2022-08-20 10:27:04 <panekj> I'm currently trying to do that
2022-08-20 10:27:09 <nmeum> ty 🙏
2022-08-20 10:28:34 <panekj> Khem Raj added the specs
2022-08-20 10:29:17 <psykose> hear that name a lot
2022-08-20 10:29:29 <panekj> yocto (:
2022-08-20 10:29:48 <psykose> ikke: can you copy the ghc-bootstrap for me to the builder please 🥺
2022-08-20 10:30:39 <ikke> psykose: where is it located? In your build container?
2022-08-20 10:30:46 <psykose> my computer
2022-08-20 10:31:28 <ikke> So how would I copy them to the build container? 🙃
2022-08-20 10:31:57 <psykose> https://img.ayaya.dev/m4xFsAcraXkf .tar.gz 335mb
2022-08-20 10:32:07 <psykose> i dunno, i asked twice on how to do it and you didn't tell me ^^
2022-08-20 10:32:25 <nmeum> panekj: do you have an email address where Khem can be contacted?
2022-08-20 10:32:40 <panekj> https://github.com/kraj
2022-08-20 10:32:45 <panekj> email, twitter, etc.
2022-08-20 10:32:51 <ikke> psykose: Because I was distracted
2022-08-20 10:33:08 <nmeum> he is not a rust developer though, right?
2022-08-20 10:33:16 <panekj> he is musl guy
2022-08-20 10:33:16 <nmeum> he just added the riscv64 musl spec file I think
2022-08-20 10:33:50 <panekj> I'm unsure what is his experience with Rust but I can't find proper information who owns which target
2022-08-20 10:34:01 <ikke> panekj: can you upload it to dev.a.o?
2022-08-20 10:34:03 <ikke> psykose: &
2022-08-20 10:34:09 <psykose> sure
2022-08-20 10:34:11 <panekj> sure
2022-08-20 10:34:19 <ikke> lol
2022-08-20 10:34:20 <psykose> nmeum: ah, fails on last step with `ghc: could not execute: aarch64-alpine-linux-musl-ld`
2022-08-20 10:34:22 <panekj> (:
2022-08-20 10:34:33 <psykose> i guess i need to rebuild ghc with ghc first too
2022-08-20 10:34:41 <nmeum> likely
2022-08-20 10:34:52 <panekj> nmeum: he also works a lot with riscv so maybe he'll know something...
2022-08-20 10:35:53 <psykose> ikke: https://dev.alpinelinux.org/~psykose/ghc/
2022-08-20 10:36:24 <panekj> one could also look at OE patches for rust
2022-08-20 10:37:59 <panekj> they use 1.62.0
2022-08-20 10:38:31 <ikke> OE?
2022-08-20 10:38:37 <psykose> openembedded
2022-08-20 10:38:40 <panekj> openembedded/yocto
2022-08-20 10:39:29 <psykose> nmeum: mm, no, building ghc itself needs that ld (same error)
2022-08-20 10:39:31 <panekj> https://github.com/openembedded/openembedded-core/tree/master/meta/recipes-devtools/rust
2022-08-20 10:39:48 <psykose> wonder where that is set..
2022-08-20 10:43:14 <panekj> Khem Raj is part of risc-v team in rust
2022-08-20 10:43:30 <panekj> https://github.com/rust-lang/team/blob/master/teams/risc-v.toml
2022-08-20 10:44:26 <panekj> you could also try asking on t-compiler/risc-v on zulip?
2022-08-20 10:45:00 <ikke> nmeum: for some reason my board cannot read the external m2 disk (in a usb case)
2022-08-20 10:45:05 <ikke> works on my desktop
2022-08-20 10:49:02 <ikke> [sda] Media removed, stopped polling
2022-08-20 10:52:53 <nmeum> maybe the driver in their custom kernel is scuffed somehow
2022-08-20 11:36:53 <psykose> got all the way to the fabricate dist part before the first hang this time
2022-08-20 11:43:47 <psykose> think i figured out the ghc thing too
2022-08-20 12:58:20 <Guest642> does anyone know how can I make java not look horrible on alpine?
2022-08-20 12:58:31 <Guest642> everything with a java gui look 30 years old
2022-08-20 12:58:37 <panekj> don't use java
2022-08-20 12:58:50 <Guest642> i like that idea very much, but i didn't write the software i need
2022-08-20 12:58:55 <Guest642> e.g libreoffice
2022-08-20 14:41:38 <ikke> rust core dumped now
2022-08-20 14:42:08 <ikke> https://tpaste.us/jP9v 
2022-08-20 14:42:40 <psykose> yep, one of those spots where it can do that at random
2022-08-20 14:43:06 <ikke> so it's not qemu-user related
2022-08-20 14:43:10 <Ermine> rust was supposed to not segfault, was it?
2022-08-20 14:43:17 <psykose> indeed not
2022-08-20 14:43:20 <ikke> Ermine: I believe ot
2022-08-20 14:44:16 <ikke> psykose: that was on my visionfive
2022-08-20 14:44:23 <psykose> yeah
2022-08-20 14:44:40 <psykose> anyway, the aarch64 ghc works
2022-08-20 14:44:49 <psykose> i reupdated the dev artifacts
2022-08-20 14:44:52 <ikke> ok
2022-08-20 14:45:01 <psykose> if you copy them to the builder, then you can merge !37929
2022-08-20 14:45:15 <psykose> and then the pkgrel bump will rebuild it just to make sure
2022-08-20 14:45:23 <psykose> or you can also rebuild manually first after taking down the builder
2022-08-20 14:45:24 <psykose> up to you
2022-08-20 14:46:01 <psykose> i built cabal-stage0 with it, building just cabal now (but i guess it will work)
2022-08-20 14:46:09 <psykose> i did not build ghc with it, but i would also guess it's fine
2022-08-20 14:46:21 <psykose> (and i would rather not wait 20 hours to rebuild that on my hardware)
2022-08-20 14:58:44 <ikke> merged
2022-08-20 14:58:47 <ikke> but builder is busy
2022-08-20 14:58:57 <psykose> yeah
2022-08-20 14:58:59 <psykose> that's ok
2022-08-20 14:59:11 <psykose> as long as you copied the stuff (and re-signed them) it'll rebuild anyway
2022-08-20 15:40:30 <Guest642> i installed libreoffice-gnome and now it looks good again
2022-08-20 17:20:52 <ikke> psykose: it's building now
2022-08-20 17:22:26 <psykose> yea, saw
2022-08-20 17:22:26 <psykose> :)
2022-08-20 17:22:42 <psykose> hopefully it passes, some chance the tests fail (the disables are x86_64 only)
2022-08-20 18:03:22 <Ermine> Please take a look at https://gitlab.alpinelinux.org/alpine/mdev-conf/-/merge_requests/1
2022-08-20 18:57:01 <psykose> for people that use libreoffice, it would be nice if you could test the artifacts from https://gitlab.alpinelinux.org/alpine/aports/-/jobs/810315
2022-08-20 20:06:02 <ikke> psykose: I removed ghc from world now
2022-08-20 20:06:14 <psykose> it built, makes sense
2022-08-20 20:06:18 <psykose> :)
2022-08-20 20:06:25 <psykose> did fail over 100 tests, but what can ya do
2022-08-20 20:08:45 <ikke> We did disable a lot of tests for x86_64 as well, right?
2022-08-20 20:11:07 <psykose> nah it was like 3
2022-08-20 20:11:20 <psykose> out of thousands or so
2022-08-20 20:11:20 <ikke> oh ,ok
2022-08-20 20:55:32 <Guest677> Is GitHub limiting the tag/release amount to 3?
2022-08-20 20:55:56 <Guest677> the button for creating tags has disappeared for me
2022-08-20 20:57:23 <ikke> Not sure why you are asking that here
2022-08-20 20:57:50 <Guest677> well, i cant seem to find anything about it
2022-08-20 20:57:55 <Guest677> + this is the only irc server i have
2022-08-20 20:59:02 <psykose> no
2022-08-20 20:59:09 <psykose> also not sure why you would make tags from github
2022-08-20 20:59:17 <psykose> just `git tag ..` and `git push --tags`
2022-08-20 20:59:22 <psykose> as for releases, well, no idea
2022-08-20 20:59:50 <Guest677> i'll try that
2022-08-20 21:01:52 <psykose> dunno, you just go to releases and press release
2022-08-20 21:02:48 <Guest677> that worked
2022-08-20 21:03:16 <Guest677> i'm uploading my custom build of alpine, its 416MB
2022-08-20 21:03:55 <psykose> if i was github i would ban u
2022-08-20 21:04:11 <Guest677> lol
2022-08-20 21:05:52 <Guest677> done
2022-08-20 21:05:58 <Guest677> what you guys think? https://github.com/rdbo/sigma-linux
2022-08-20 21:06:21 <psykose> i don't think sorry
2022-08-20 21:06:24 <psykose> no thoughts, head empty
2022-08-20 21:07:16 <Guest677> is there even any other alpine based distro? theres no alpine option in distrowatch last i checked
2022-08-20 21:08:06 <psykose> postmarketos
2022-08-20 21:20:00 <Guest677> damn, thought i was the first
2022-08-20 21:20:02 <Guest677> looks cool
2022-08-20 21:20:20 <Guest677> i wonder if their build scripts do so much weird stuff as i'm doing
2022-08-20 21:21:46 <Newbyte> Guest677 (IRC) can't you use xdg-user-dirs to create Downloads, Documents et cetera instead of doing it manually?
2022-08-20 21:21:57 <Newbyte> oh, never mind
2022-08-20 21:22:18 <psykose> that's all you noticed? :p
2022-08-20 21:22:32 <Guest677> lol
2022-08-20 21:23:28 <Guest677> i had to pass a lot of the work to local.d cause it won't let me copy stuff to some directories
2022-08-20 21:23:55 <Newbyte> psykose well, it was the first thing I noticed
2022-08-20 21:24:38 <Newbyte> this seems more like a preconfigured Alpine image than a separate distributino
2022-08-20 21:24:39 <Newbyte> s/distributino/distribution/
2022-08-20 21:25:02 <Guest677> pretty much, i guess
2022-08-20 21:25:06 <Newbyte> but yeah, what happens on updates?
2022-08-20 21:25:20 <Guest677> but isn't that what ubuntu, linux mint, anything debian based really is?
2022-08-20 21:25:30 <Newbyte> they have their own repos I think
2022-08-20 21:25:43 <Guest677> updates is something that got me thinking a bit
2022-08-20 21:26:14 <Newbyte> like, isn't /etc/hostname owned by alpine-baselayout-data? or does /etc get special treatment?
2022-08-20 21:26:53 <Newbyte> anyway, Guest677 (IRC), I can assure you our (postmarketOS') build scripts are very different from this
2022-08-20 21:26:59 <Guest677> a
2022-08-20 21:27:22 <Guest677> - /etc is one of the only writable directories from the genapkovl stage aparently
2022-08-20 21:27:45 <Guest677> thats why i use local.d to do the dirty work for me
2022-08-20 21:27:48 <Guest677> and put everything in place
2022-08-20 21:28:06 <Guest677> i write to /etc/local.d, enable the local service and it runs the script on startup
2022-08-20 21:28:38 <Guest677> Newbyte: oh, you develop postmarketOS?
2022-08-20 21:28:47 <Newbyte> Newbyte I'm a maintainer
2022-08-20 21:28:50 <Newbyte> so yes, I think
2022-08-20 21:28:59 <Guest677> interesting
2022-08-20 21:29:05 <Guest677> i might take a look later
2022-08-20 21:29:15 <Newbyte> It's for phones and tablets primarily
2022-08-20 21:29:24 <Newbyte> So a bit different from what you're doing
2022-08-20 21:29:42 <Guest677> well, if you can build for that your scripts must be really good
2022-08-20 21:29:44 <Newbyte> I'm also on Alpine's GNOME Team, but that doesn't mean much
2022-08-20 21:29:46 <Guest677> maybe i'll learn something
2022-08-20 21:30:35 <Guest677> can it run on any phone? would be nice to carry a full linux os around
2022-08-20 21:31:28 <Newbyte> Guest677 (IRC) support varies. see our list: https://wiki.postmarketos.org/wiki/Devices
2022-08-20 21:32:28 <Guest677> wow
2022-08-20 21:32:42 <Newbyte> most phones have pretty poor support
2022-08-20 21:32:55 <Newbyte> but there are a growing list that are good
2022-08-20 21:33:36 <Newbyte> if you're looking to get one, I would suggest OnePlus 6/6T, Pocophone F1, or Shift6mq
2022-08-20 21:33:49 <Newbyte> maybe LG G7 soon
2022-08-20 21:34:43 <Guest677> it's an interesting project, maybe i'll contribute one day
2022-08-20 21:35:02 <Guest677> although my phone hacking skills are not all that good
2022-08-20 21:38:36 <Newbyte> I didn't know much when I got started either
2022-08-20 21:38:42 <Newbyte> questions are always welcome :)
2022-08-20 21:39:08 <Newbyte> though, this is not the place for those
2022-08-20 21:39:16 <Guest677> nice, thanks
2022-08-20 21:40:15 <Newbyte> anyway, I'm leaving for now. feel free to join our Matrix/IRC channels if you are interested
2022-08-20 21:42:41 <Guest677> alright, cya
2022-08-21 01:28:19 <ktprograms> psykose: Congrats on getting ghc on aarch64! I did write my steps to get it done a few months ago, but anyway that was cross-compiling from another aarch64 distro so it wouldn't have been useful here.
2022-08-21 01:28:43 <ktprograms> Just curious, was it a just run bootstrap.sh and it works or did you need a lot of patching?
2022-08-21 01:30:45 <psykose> just works
2022-08-21 01:31:00 <psykose> the only change at all was fixing ghclib/configure to remove the ctarget- prefix
2022-08-21 01:31:02 <psykose> well
2022-08-21 01:31:05 <psykose> i mean it doesn't actually work
2022-08-21 01:31:08 <psykose> fails over 100 tests
2022-08-21 01:31:19 <psykose> but works for most things
2022-08-21 01:31:23 <ktprograms> lol
2022-08-21 01:33:18 <ktprograms> Is that a musl thing or aarch64 thing?
2022-08-21 01:35:56 <sam_> maybe even a ghc thing
2022-08-21 01:42:03 <psykose> the first failure is https://gitlab.haskell.org/ghc/ghc/-/issues/13062
2022-08-21 02:41:21 <sam_> i'm biased btw because i hate ghc
2022-08-21 02:41:33 <sam_> it doesn't help that the llvm backend is only needed for certain arches
2022-08-21 02:50:30 <psykose> they have a native aarch64 one now i think
2022-08-21 02:50:33 <psykose> same tho
2022-08-21 02:54:16 <Newbyte> ghc on armv7 when
2022-08-21 03:14:08 <psykose> i'll do it for a fat payment
2022-08-21 03:22:25 <skarnet> lies
2022-08-21 03:23:18 <psykose> not just 2 hugs, three!
2022-08-21 03:24:10 <skarnet> you're so greedy!
2022-08-21 03:26:57 <psykose> bully!
2022-08-21 03:52:12 <ayakael> what did ghc ever do to you
2022-08-21 04:15:42 <Ermine> Are there limitations  on number of devices on which I can stay logged in gitlab?
2022-08-21 04:17:59 <ptrc> not really
2022-08-21 05:20:04 <sam_> re ghc sins: i have a bias against languages which require themselves to boostrap because it's a pain whenever upstream don't provide binaries for various other platforms
2022-08-21 05:20:22 <sam_> but also, like rust, more often than is ideal, there's bugs in the releases which they don't do new versions for
2022-08-21 05:27:36 <psykose> ghc's build systems make cargo look like it's from the future
2022-08-21 05:29:11 <ktprograms> _Well_ gcc is worse... It compiles C but has C++ source code
2022-08-21 05:30:58 <psykose> unlike ghc which has 0 c++ code
2022-08-21 05:37:56 <sam_> yeah but gcc is something we live with and i don't want to get into a bootstrapping c/c++ compilers issue
2022-08-21 05:38:03 <sam_> but it's at least tolerable because it is what it is and we're all used to it
2022-08-21 05:38:07 <sam_> adding more into the mix is not good though
2022-08-21 05:38:25 <sam_> (and I have the same complaints about gcc's new D-in-D)
2022-08-21 05:42:31 <psykose> gcc is also much easier to bootstrap because you can manually jump a few steps from like any platform to any other
2022-08-21 05:43:21 <psykose> and for better or worse C defines the base portability of basically every platform; some random language people use for shellcheck and pandoc though is quite different : )
2022-08-21 05:44:53 <sam_> i was struggling to articulate it
2022-08-21 05:44:55 <sam_> that's really it
2022-08-21 05:46:03 <psykose> C is also the only language seemingly that has (so far) cared about that large jump range
2022-08-21 05:46:17 <psykose> gcc10 needs a c++98 compiler (11+ bumped it to c++11)
2022-08-21 05:46:48 <psykose> half these languages need like 'one of the last 2 versions' or somethingm, it's very painful to set up a 'reproducible' kind of bootstrap; admittedly ghc at least seems less bad than a lot of others
2022-08-21 05:47:45 <psykose> dare i say that is the only stable unix ABI you get! the ability to build a modern C compiler in 3 steps!
2022-08-21 06:25:51 <orbea> psykose: just wait till you need a rust compiler to have linux or mesa
2022-08-21 06:27:04 <orbea> and the rust compiler will hopefully be in gcc, that will be fun to bootstrap :P
2022-08-21 06:27:27 <orbea> probably somehow better than currently
2022-08-21 06:28:49 <sam_> isnt it in c++ tho
2022-08-21 06:30:06 <orbea> hmm, i was assuming there is some rust in there, but i'll be happy if that is wrong
2022-08-21 09:22:04 <mercenary> wait until gcc gets rewritten in rust
2022-08-21 09:32:57 <ayakael> *shudders*
2022-08-21 09:35:06 <ayakael> fortunately it seems like the opposite is happening https://github.com/Rust-GCC/gccrs 
2022-08-21 09:37:09 <ayakael> although it'd be real funny if a rust compiler for gcc would happen just to close the loop. rust-gcc compiles gcc-rust to compile rust-gcc. An eternal loop that could compete against POW blockchains in wasted cpu cycles.
2022-08-21 09:39:43 <mercenary> Game-of-life. Either compiling one with the other leads to a stable state, one or more oscillating stable states, or an infinitely expanding stream of warning messages
2022-08-21 09:41:46 <ikke> See you all in 2 weeks 
2022-08-21 09:42:15 <mercenary> enjoy the holidays
2022-08-21 09:43:42 <psykose> hf
2022-08-21 09:44:29 <ayakael> Enjoy!!
2022-08-21 09:49:33 <Ermine> Enjoy!
2022-08-21 11:00:22 <skarnet> have a nice holiday ikke!
2022-08-21 11:16:46 <panekj> psykose: that's just spam
2022-08-21 11:16:52 <panekj> delete
2022-08-21 11:21:48 <psykose> is me
2022-08-21 11:22:12 <panekj> no, you're ham
2022-08-21 11:23:25 <psykose> amn't
2022-08-21 13:51:52 <psykose> hmm
2022-08-21 13:52:15 <psykose> for some reason random things go up in size like 30% on aarch64 with new gcc
2022-08-21 13:54:08 <psykose> or maybe it's binutils
2022-08-21 13:59:25 <orbea> i assume you checked that the stripping is woring correctly?
2022-08-21 14:00:19 <psykose> i didn't literally check the binaries, no, but it would be strange if binutils strip was only sometimes broken
2022-08-21 14:00:26 <psykose> also unstripped things are usually more than ~30%
2022-08-21 14:04:05 <psykose> https://img.ayaya.dev/qvm9ncruKu3w
2022-08-21 14:04:06 <psykose> weird
2022-08-21 14:06:10 <orbea> i stripped two arbitrary binaries, first was 10% and the second was nearly 50% :shrug:
2022-08-21 14:07:33 <psykose> yep, it's binutils
2022-08-21 14:07:36 <psykose> clang and same shit
2022-08-21 14:07:40 <psykose> -fuse-ld=lld and it's fixed
2022-08-21 14:07:56 <psykose> as always binutils is a pile of shit, who guessed
2022-08-21 14:08:06 <psykose> time to tredge the ML
2022-08-21 14:08:10 <psykose> hopefully someone already found this
2022-08-21 14:09:29 <psykose> sam_: have u seen this
2022-08-21 14:09:54 <psykose> seems to be a 2.39 regression
2022-08-21 14:38:54 <zv> suppose an upstream source for a package (e.g. foo.com/path/to/file.txt) is updated periodically but the name does not change. this will cause a checksum failure when it is updated. is there a sane workaround?
2022-08-21 14:39:27 <skarnet> the version isn't changing?
2022-08-21 14:39:39 <zv> there is no "version" for that file.
2022-08-21 14:40:29 <skarnet> depending on random external Web-provided volatile data without checksums doesn't sound safe :/
2022-08-21 14:40:32 <psykose> you have to rehost it somewhere stable
2022-08-21 14:40:39 <psykose> and manually rename
2022-08-21 14:41:09 <psykose> in abuild side you can rename (name::uri), but that only works to fix unversioned names, not things that actually change at the same URI
2022-08-21 14:41:10 <psykose> so sadly not
2022-08-21 14:42:12 <skarnet> I'd say "fortunately not"
2022-08-21 14:42:34 <skarnet> not being able to do that is likely very much intended
2022-08-21 14:43:19 <psykose> yeah, it is
2022-08-21 14:43:27 <psykose> so, the workaround is mirroring it anywhere
2022-08-21 14:43:34 <psykose> plenty of options
2022-08-21 14:44:05 <zv> agreed, and there's one more consideration. 'file.txt', if cached, will be at /var/cache/distfiles/file.txt
2022-08-21 14:44:29 <zv> it will not be at /var/cache/distfiles/$pkgname/$pkgver/file.txt
2022-08-21 14:44:46 <zv> so if the version of e.g. https://git.alpinelinux.org/aports/tree/community/unicode-character-database/APKBUILD is bumped, cache must be purged.
2022-08-21 14:45:15 <psykose> that's the same issue as before
2022-08-21 14:45:22 <psykose> the name you get into distfiles is the name that is also cached
2022-08-21 14:45:44 <psykose> so, you rename it (file-$pkgver.txt::), and that fixes that
2022-08-21 14:45:54 <zv> I see.
2022-08-21 14:45:58 <psykose> but if you're mirroring it to make it not change you can do that there as well
2022-08-21 14:57:42 <minimal> zv: so if there's no version for the file then how would the Alpine package be versioned?
2022-08-21 14:58:17 <psykose> 0_flowers_and_roses
2022-08-21 14:58:43 <minimal> psykose: isn't a rose a flower? ;-)
2022-08-21 14:59:22 <psykose> you got me!
2022-08-21 15:03:02 <minimal> zv: if this is the unicode character data then their website mentioned versioned (rather than "latest") URLs, e.g. https://www.unicode.org/Public/14.0.0/
2022-08-21 15:04:17 <psykose> yeah those are versioned
2022-08-21 15:04:52 <psykose> there's even a zip dist https://www.unicode.org/Public/zipped/14.0.0/UCD.zip
2022-08-21 15:05:09 <psykose> $pkgname-$pkgver.zip::https://www.unicode.org/Public/zipped/$pkgver/UCD.zip is an example of that
2022-08-21 15:05:12 <zv> the answer I needed was naming the files as file-$pkgver::$url or similar
2022-08-21 15:05:14 <zv> thanks
2022-08-21 15:05:55 <zv> I wonder if there's a good reason to have all the distfiles files in the same namespace though?
2022-08-21 15:06:00 <zv> (other than for historical reasons)
2022-08-21 15:06:14 <psykose> there's a namespace per release and per architecture actually
2022-08-21 15:06:20 <psykose> but not per version
2022-08-21 15:06:38 <zv> right; speaking only of package name and version.
2022-08-21 15:06:53 <psykose> dunno, don't see much point to it
2022-08-21 15:07:04 <psykose> quite a few places the same name is reused
2022-08-21 15:07:34 <zv> main/foo and main/bar might both have files called 'beans.txt' :P not to make a big deal of edge cases
2022-08-21 15:08:18 <psykose> 'beans.txt' is not something that would normally ever be cached
2022-08-21 15:09:20 <psykose> the pkgname namespace already 'exists' in the sense of the pkgname is unique and you can just prefix it, but random files like that are never present without at least a version too
2022-08-21 15:10:15 <psykose> there's a lot of cached junk soup obviously, random stuff gets merged :p but it's usually $sha1.patch and doesn't really cause an actual issue
2022-08-21 15:10:23 <psykose> a few cases of 'test.patch' but those are more funny than anything
2022-08-21 15:10:34 <psykose> the issues are never majorly disruptive or common
2022-08-21 15:11:33 <psykose> also i think i'm wrong about the architecture cache and it's just per release
2022-08-21 15:11:55 <psykose> you can see yourself https://distfiles.alpinelinux.org/distfiles/edge/
2022-08-21 15:12:11 <psykose> buncha bare numbers and shas, and then it goes normal
2022-08-21 15:14:39 <zv> then conflict is risked in these: http://ix.io/48gh
2022-08-21 15:15:14 <zv> (or whatever subset of those are pulled from external servers)
2022-08-21 15:15:56 <ptrc> they might be stale, because the distfiles rarely get cleaned up
2022-08-21 15:16:23 <psykose> yeah, those are misnamed
2022-08-21 15:18:59 <psykose> well, only some
2022-08-21 15:19:07 <psykose> the $pkgname- ones are not
2022-08-21 17:32:19 <Newbyte> 
ACTION uploaded an image: (526KiB) < https://matrix.org/_matrix/media/r0/download/matrix.org/iBhDUthdLaGERJODlOxLftPt/clipboard.png >

2022-08-21 17:32:29 <Newbyte> ncopa: is this actually your doing?
2022-08-21 17:32:46 <Newbyte> this is on Telegram
2022-08-21 17:32:59 <panekj> Newbyte: https://twitter.com/ptrcnull/status/1561404020937596928
2022-08-21 17:33:34 <ebb> >reserved under the MIT License
2022-08-21 17:33:51 <Newbyte> panekj: yeah, I definitely get that feeling
2022-08-21 17:34:02 <ptrc> this whole situation is so funny
2022-08-21 17:34:19 <panekj> this stinks like mckay so much
2022-08-21 17:34:48 <ptrc> especially that the style in later messages matches his
2022-08-21 17:35:01 <Newbyte> yeah
2022-08-21 17:35:51 <Newbyte> I asked "Can you somehow verify that you are the real ncopa?"
2022-08-21 17:35:59 <Newbyte> and I got the answer "If you don't believe me don't believe"
2022-08-21 17:36:25 <Newbyte> very funny situation
2022-08-21 17:37:20 <panekj> yes, it's mckay 100%
2022-08-21 17:45:23 <psykose> people use telegram?
2022-08-21 17:45:31 <ptrc> unfortunately
2022-08-21 17:46:01 <Habbie> indeed
2022-08-21 17:46:16 <Habbie> who is mckay?
2022-08-21 17:46:58 <psykose> exiled troll people keep giving attention to that keeps impersonating random alpine-related things
2022-08-21 17:47:03 <ptrc> admin of https://t.me/alpine_linux_english, a random guy from latam who's abusing the alpine logo/name
2022-08-21 17:47:13 <Habbie> ack, thanks
2022-08-21 17:49:47 <panekj> a.k.a piccoro
2022-08-21 17:50:28 <ptrc> ( is he still banned from the mailing lists btw? )
2022-08-21 17:50:31 <panekj> If you see any wiki page that's super confusing and written in broken english, you will most likely find his name in history
2022-08-21 18:23:59 <skarnet> I lost track after he got banned for having zero social and communication skills on the mailing-list (was deserved imo)
2022-08-21 18:24:35 <skarnet> so he went from loser to criminal, it seems? :(
2022-08-21 18:25:49 <panekj> although it might not be mckay, I guess(?), since mckay does way much more spelling mistakes
2022-08-21 18:26:49 <Newbyte> yeah, and the first guy who posted the channel wasn't Gerhard, and it's not clear how that guy found the channel
2022-08-21 18:27:37 <Newbyte> I also don't see how Gerhard would benefit from this
2022-08-21 18:30:27 <ptrc> imo it was suspicious how mckay was uncritically supporting all of it
2022-08-21 19:32:45 <Ermine> He isn't banned from wiki?
2022-08-21 19:35:01 <Ermine> Also, there was an anti-irc guy recently. It might be him
2022-08-21 20:01:44 <Guest749> do you guys know how do i set a local repository in alpine's mkimage script? i tried setting the same repository from abuild's REPODEST, REPODEST/<parent_name>, REPODEST/<parent_name>/<architecture>, and nothing works: 
2022-08-21 20:01:59 <Guest749> WARNING: Ignoring /app/profile/repo/apk/x86_64: No such file or directory
2022-08-21 20:02:07 <Guest749> APKINDEX.tar.gz          sigma-rootfs-2.0-r0.apk
2022-08-21 20:02:37 <Guest749> ls /app $ ls /app/profile/repo/apk/x86_64/ -> APKINDEX.tar.gz          sigma-rootfs-2.0-r0.apk
2022-08-21 20:03:11 <Guest749> using the --repository flag
2022-08-21 20:07:07 <panekj> what are you exactly running
2022-08-21 20:07:36 <Guest749> sh mkimage.sh \
2022-08-21 20:07:37 <Guest749>  --tag edge \
2022-08-21 20:07:39 <Guest749>  --outdir "$OUTDIR" \
2022-08-21 20:07:41 <Guest749>  --arch x86_64 \
2022-08-21 20:07:43 <Guest749>         --repository "$REPODIR/apk/x86_64" \
2022-08-21 20:07:45 <Guest749>  --repository http://dl-cdn.alpinelinux.org/alpine/edge/main \
2022-08-21 20:07:47 <Guest749>  --repository http://dl-cdn.alpinelinux.org/alpine/edge/testing \
2022-08-21 20:07:49 <Guest749>  --repository http://dl-cdn.alpinelinux.org/alpine/edge/community \
2022-08-21 20:07:51 <Guest749>  --profile "$PROFILENAME" \
2022-08-21 20:07:53 <Guest749>  --workdir "$CACHEDIR"
2022-08-21 20:08:31 <Guest749> i have 1 apk in the local repo which i need to install
2022-08-21 20:08:31 <panekj> why are you adding arch to repo
2022-08-21 20:08:46 <panekj> also you are supposed to use --extra-repository
2022-08-21 20:08:58 <Guest749> because i tried just $REPODIR, $REPODIR/apk, $REPODIR/apk/x86_64
2022-08-21 20:08:59 <Guest749> none worked
2022-08-21 20:09:17 <Guest749> --extra-repository is for backwards compatibility i think
2022-08-21 20:10:07 <Guest749> do i have to be running a local web server to make the repository work or something?
2022-08-21 20:10:39 <Guest749> cause in the live image, at /etc/apk/repositories there is a local directory /media/something
2022-08-21 20:10:48 <Guest749> so i thought i could do the same here
2022-08-21 20:16:25 <panekj> what's the error when you specify proper path
2022-08-21 20:17:02 <Guest749> it says
2022-08-21 20:17:04 <Guest749> >>> mkimage-x86_64: Building sigma
2022-08-21 20:17:05 <Guest749> WARNING: Ignoring /app/profile/repo/apk/x86_64: No such file or directory
2022-08-21 20:17:13 <Guest749> even though it's there
2022-08-21 20:17:36 <Guest749> APKINDEX.tar.gz          sigma-rootfs-2.0-r0.apk
2022-08-21 20:18:01 <Guest749>  /app $ ls /app/profile/repo/apk/x86_64/
2022-08-21 20:18:02 <Guest749> APKINDEX.tar.gz          sigma-rootfs-2.0-r0.apk
2022-08-21 20:24:29 <Guest749> /app/profile/repo/apk worked now, i think it was the cache or something
2022-08-21 20:27:07 <panekj> I think you kept pointing it to invalid path, it worked for me first try
2022-08-21 20:37:00 <Guest749> yeah, the x86_64 part should be there, although i had already tried without it (maybe i misstyped or didn't clean the cache?). i started a new docker instance and it worked
2022-08-21 20:37:01 <Guest749> thx
2022-08-22 01:41:34 <Guest759> my alpine iso weighs 416MB and it needs more memory than linux mint (2.3GB) on QEMU with KVM enabled. with 2GB, it can't even fully load my system, some things are missing (had to make it 2.5GB for everything to work). how is this possible? how can they fit so much stuff in the iso and use so little memory? is that achievable in alpine?
2022-08-22 01:43:36 <minimal> Guest759: what is using the memory? have you looked at top/htop or ps output?
2022-08-22 01:44:47 <Guest759> i will check, but it's most likely the unionfs where it saves the whole system in the RAM i think
2022-08-22 01:45:29 <minimal> ah, run-from-ram
2022-08-22 01:46:46 <minimal> try Sys mode (install-to-disk) instead?
2022-08-22 01:47:43 <Guest759> with 2.5G, it says that 83% of the tmpfs is full
2022-08-22 01:47:54 <Guest759> Sys mode? what would that be
2022-08-22 01:48:55 <Guest759> this iso is based from alpine-standard with some extra packages and configs basically
2022-08-22 01:50:03 <Guest759> oh, by sys mode you mean installing the iso into a drive?
2022-08-22 01:50:52 <Guest759> if so, yes, that makes the RAM problem disappear, but in order to boot the ISO i will need at least ~2.5G of RAM, which is overkill for a 416M system
2022-08-22 01:51:05 <Guest759> compared to linux mint, which weighs 2.3G and can run on under 2G of RAM systems
2022-08-22 01:51:17 <ptrc> what iso are you using? is that some custom stuff, or one of the downloads page?
2022-08-22 01:51:27 <Guest759> custom
2022-08-22 01:51:37 <Guest759> https://github.com/rdbo/sigma-linux
2022-08-22 01:51:51 <Guest759> the version i'm using is the latest on Releases
2022-08-22 01:55:09 <ptrc> well, can't reproduce on the official extended x86_64 iso, it might due to the difference in compressed .apk vs uncompressed onto tmpfs
2022-08-22 01:56:00 <Guest759> the official ISOs are really small, so theres probably a lot of room left in the tmpfs
2022-08-22 01:56:26 <Guest759> but still, how can linux mint do it? it weighs more than the RAM i gave it, and it still fully works
2022-08-22 01:56:53 <ptrc> yes, but while the ISO is 718MB, the booted RAM usage is only 50MB
2022-08-22 01:56:55 <Guest759> i want to add more packages and apps, but then it might just take 10G of RAM to boot
2022-08-22 01:58:15 <ptrc> "how can linux mint do it?" they put everything in a compressed squashfs, then mount it directly off the iso
2022-08-22 01:58:17 <Guest759> on mine it says that it's using 213MB of RAM, but the tmpfs is 83% used
2022-08-22 01:58:31 <Guest759> interesting
2022-08-22 01:58:40 <Guest759> is that possible in alpine?
2022-08-22 01:59:18 <ptrc> yeah, but i don't think you can do that with the current mkimg scripts
2022-08-22 01:59:26 <ptrc> you can always just do it yourself
2022-08-22 01:59:39 <Guest759> sounds like a lot of trouble
2022-08-22 01:59:53 <Guest759> has anyone done it? would be nice to have a starting point
2022-08-22 02:02:11 <ptrc> i don't recall anything like that, but i can try to make a quick example image
2022-08-22 02:03:49 <Guest759> i would be grateful
2022-08-22 02:20:08 <minimal> Guest759: Mint does not run-from-ram AFAIK amd so its memory usage would be less
2022-08-22 03:27:47 <ptrc> Guest759: something *vaguely* like this: https://gitlab.alpinelinux.org/-/snippets/265
2022-08-22 03:28:20 <ptrc> ( i'm bad at making bootable images, so this doesn't have video support for some reason, but serial in qemu works )
2022-08-22 03:37:19 <Jean[m]1234> Hi all - are there any notes on how to build Chromium for Alpine?
2022-08-22 03:38:35 <ptrc> see APKBUILD at https://gitlab.alpinelinux.org/alpine/aports/-/tree/master/community/chromium - the closest you're gonna get to "build instructions", i'm afraid
2022-08-22 03:39:02 <Jean[m]1234> I tried building it today, using pmbootstrap (postmarketOS bootstrap tool) which just wraps around apk for building packages. I tried cross compiling, but got some weird errors, one I can post in a sec.
2022-08-22 03:40:11 <ptrc> i.. don't think *cross-compiling* chromium is gonna work, unless you mean building it with qemu-user, not sure how pmbootstrap wraps it
2022-08-22 03:40:20 <Jean[m]1234> Right - I was checking that out. Needed to add a patch for squeekboard (mobile Linux keyboard used for Phosh)
2022-08-22 03:41:03 <sam_> it does actually work(!)
2022-08-22 03:41:16 <Jean[m]1234> Yes I imagine pmbootstrap uses qemu-user, is there a better emulation tool? :P
2022-08-22 03:41:43 <Jean[m]1234> sam_: pmbootstrap or cross compiling? :)
2022-08-22 03:41:52 <ptrc> Jean[m]1234: do you mean this one? https://source.puri.sm/Librem5/squeekboard/-/issues/244#note_156038
2022-08-22 03:42:14 <Jean[m]1234> ptrc: Bingo!
2022-08-22 03:42:34 <sam_> cross
2022-08-22 03:42:45 <Jean[m]1234> About that - I don't see a build log for chromium on alpine's page for chromium...
2022-08-22 03:43:06 <ptrc> instead of trying to build chromium yourself, you can open an issue in aports, because imo this would be okay to actually add for aports, if it doesn't break anything
2022-08-22 03:43:07 <Jean[m]1234> sam_: Nice. I wonder why it failed super early for me then.
2022-08-22 03:43:27 <ptrc> as for build logs, here's a bunch of them: https://build.alpinelinux.org/buildlogs/build-edge-x86_64/community/chromium/
2022-08-22 03:43:35 <Jean[m]1234> ptrc: Right, not sure about regressions, just wanted to test it out.
2022-08-22 03:43:58 <Jean[m]1234> ptrc: Oh, awesome! Thank you!
2022-08-22 03:53:09 <Jean[m]1234> sam_: anything special that you did to cross compile chromium?
2022-08-22 03:58:31 <sam_> take a look at the is_cross_compiler blocks in https://gitweb.gentoo.org/repo/gentoo.git/tree/www-client/chromium/chromium-106.0.5245.0.ebuild (gn needs some pushing in right direction)
2022-08-22 04:02:49 <Jean[m]1234> heh, ok. taking a look. thank you
2022-08-22 04:32:01 <ptrc> https://gitlab.alpinelinux.org/alpine/aports/-/commit/7a521ce68a52b73081dadbb46bb84fb75f8d1d6d
2022-08-22 04:32:33 <ptrc> added 2 years ago, nobody ever realized it's not enabled on any architecture.. :')
2022-08-22 05:05:37 <ptrc> aaaand i broke gitlab, apparently (!38025)
2022-08-22 05:06:41 <ptrc> https://ptrc.gay/TifWmEME ???
2022-08-22 05:17:17 <panekj> > I have to say I love GitLab features... GitHub is really behind.
2022-08-22 05:17:27 <panekj> about the feature that github had for years
2022-08-22 06:42:54 <psykose> Jean[m]1234: yeah i can just add the patch for you (i practically maintain chromium), but i would like to see it complete (as the link notes, emoji keys don't work)
2022-08-22 08:19:52 <Guest759> ptrc: thanks for the snippet, i will take a look later :)
2022-08-22 08:23:43 <zekraa> -
2022-08-22 08:23:43 <zekraa> Allah the Greatest (God of Islam religion) Rules and the power is in his hands , Sunni islam is the true religion of God , your bad deeds are getting back on you , as All your life u been following secular armies and tyrant systems those who rule in injust and commite crimes , and u r requested now to destroy the positions and personalities with you in following the tyrant regime you follow
2022-08-22 08:40:59 <panekj> Mashallah
2022-08-22 08:49:39 <psykose> smh telling people rust static is fixed
2022-08-22 08:49:44 <psykose> now i'm responsible!!
2022-08-22 08:50:10 <panekj> hehe
2022-08-22 08:50:26 <panekj> isn't it fixed?
2022-08-22 08:50:46 <psykose> you tell me, static user
2022-08-22 08:50:51 <psykose> (but also doesn't work on 32-bit arm)
2022-08-22 08:51:10 <panekj> *works on my machine*
2022-08-22 09:33:52 <DylanVanAssche> Is there a way to keep an empty directory created in APKBUILD? After a split up into separate packages, fwupd daemon crashes if no plugins installed.
2022-08-22 09:34:17 <DylanVanAssche> I tried `mkdir` in the `package()` function, but that directory is not kept after install
2022-08-22 09:36:06 <psykose> how do i run the fwupd daemon
2022-08-22 09:36:22 <DylanVanAssche> `/usr/libexec/fwupd/fwupd -v` (verbose mode)
2022-08-22 09:36:53 <DylanVanAssche> If I create the dir manually it also crashes when interacting with it, maybe this should be fixed upstream instead...
2022-08-22 09:40:19 <psykose> interacting in what way
2022-08-22 09:41:21 <psykose> i fixed the first part now i'm wondering what else to test
2022-08-22 09:41:34 <DylanVanAssche> psykose: fwupdmgr get-devices
2022-08-22 09:41:45 <DylanVanAssche> If you run the daemon in verbose, you will see an assert
2022-08-22 09:42:00 <psykose> doesn't crash but yeah
2022-08-22 09:42:01 <DylanVanAssche> The command will never return because the assert causes a crash
2022-08-22 09:42:05 <psykose> expects nonzero plugins
2022-08-22 09:42:25 <psykose> i would report it upstream i guess
2022-08-22 09:42:34 <DylanVanAssche> Well yeah, the daemon keeps running that's true, but that's all 😛
2022-08-22 09:42:43 <DylanVanAssche> Alright will do 🙂 
2022-08-22 09:43:26 <DylanVanAssche> psykose: You said 'fixed the first part' (dir). With an APKBUILD instruction or manually? (just out of curiosity to learn something)
2022-08-22 09:43:51 <psykose> mkdir works, it only gets removed because of the amove semantics for the plugin splits (the directory gets deleted once it becomes empty)
2022-08-22 09:44:39 <psykose> but plugin_all runs last, so you can mkdir it in there and it sticks
2022-08-22 09:48:08 <DylanVanAssche> Ah okay, good to know!
2022-08-22 09:48:10 <DylanVanAssche> Bug report: https://github.com/fwupd/fwupd/issues/4932
2022-08-22 09:49:16 <psykose> `Failed to open polkit agent: missing executable pkttyagent in PATH` being reported on every single invocation of anything is also incredibly annoying
2022-08-22 09:49:31 <psykose> it works and still prints it, holy shit shut up already
2022-08-22 09:49:32 <psykose> haha
2022-08-22 09:49:54 <psykose> but yeah, i guess i can make it depend on nvme or something benign by default
2022-08-22 09:50:22 <DylanVanAssche> psykose: Where? I don't see that
2022-08-22 09:50:34 <psykose> fwupdmgr <anything>
2022-08-22 09:51:28 <psykose> without having pkttyagent in path
2022-08-22 09:52:37 <DylanVanAssche> Ah I don't have that here, maybe because I have polkit running
2022-08-22 09:53:25 <psykose> you have the binary in path, as it says
2022-08-22 12:22:47 <ericonr> psykose: touch .keepdir or similar? :P
2022-08-22 12:23:06 <psykose> that's the other way of doing it, yeah
2022-08-22 12:23:39 <psykose> but then you have to remove .keepdir in the same thing that runs last where you can make it anyway, since i don't think there is any support for removing stale keepdirs
2022-08-22 12:24:41 <ericonr> not sure I get it; but I was thinking of shipping that in the package itself
2022-08-22 12:25:13 <ericonr> can possibly report to upstream to only warn about pkttyagent when it actually needs permission from something, but this likely comes from polkit library initialization so I doubt there's anything you can actually do
2022-08-22 12:27:42 <ericonr> psykose: unless you meant you don't think apk itself removes stale dirs?
2022-08-22 12:28:08 <psykose> abuild doesn't remove any dirs
2022-08-22 12:28:35 <psykose> as for pkttyagent, yes
2022-08-22 12:30:37 <psykose> did you have a think about that libstdc++ thing again
2022-08-22 12:32:43 <ericonr> I looked at that last line you sent with the casts and it seemed right to me; but I didn't investigate if the rest of the functions actually try to use __cloc to see if it makes sense to try and set it]
2022-08-22 12:33:22 <ericonr> otherwise it could be enough to just do your conditional verification on *.UTF-8 and C
2022-08-22 12:33:48 <psykose> just the conditional alone works for fixing the random fmt tests (same 'locale not supported, skipped')
2022-08-22 12:35:10 <psykose> and `LANG=pt_BR.UTF-8 easyeffects` still works, so i guess it's fine
2022-08-22 12:37:06 <ericonr> what I'm thinking is: you create a new locale with .UTF-8 properties, then call the C++ equivalent of uselocale() on it; now it's global program state and it's expected that stuff like regexp and whatever will be unicode aware
2022-08-22 12:38:16 <ericonr> is this a likely use case? probably not; does the current generic implementation in libstdc++ come close to even supporting it? probably not either
2022-08-22 12:38:26 <ericonr> but I'd like to be sure first
2022-08-22 12:40:14 <psykose> i assume the C <regex.h> functions become utf-8 aware in c++ the moment something in c++ calls std::setlocale, don't they
2022-08-22 12:42:45 <psykose> where's that quick test to find out
2022-08-22 12:55:03 <Ariadne> regex.h is always utf-8 aware on musl
2022-08-22 12:58:54 <psykose> guess i can't test it on musl then
2022-08-22 13:11:15 <ericonr> Ariadne: dalias had talked about setlocale being important for some libc behavior; what isn't utf-8 aware by default, then?
2022-08-22 13:11:43 <Ariadne> no idea :D
2022-08-22 13:12:28 <ericonr> This counts as a ping so hopefully we can learn soon 
2022-08-22 13:12:51 <Ariadne> but i am pretty sure that regex.h is always utf-8 aware, as TRE is
2022-08-22 13:12:59 <Ariadne> and regex.h is basically just TRE
2022-08-22 14:34:31 <dalias> all programs should have called at least setlocale(LC_CTYPE, "")
2022-08-22 14:35:15 <dalias> otherwise the posix interpretation about C locale being byte-based applies
2022-08-22 14:36:27 <dalias> ericonr, what is this about matching *.UTF-8?
2022-08-22 14:36:38 <dalias> musl does not want that convention; it's only supported for compat
2022-08-22 14:37:06 <dalias> LANG=en is expected to work (and is utf-8 based)
2022-08-22 14:38:00 <dalias> i saw that issue being discussed on the distros channel and the *.UTF-8 matching thing is completely a red herring
2022-08-22 14:38:29 <dalias> the problem is that the software's tests are trying to probe if a given locale exists in order to assume it contains the language content they expect
2022-08-22 14:38:43 <dalias> and running into the issue that all locales exist on musl
2022-08-22 14:38:54 <psykose> they don't all exist though
2022-08-22 14:38:58 <dalias> they do!
2022-08-22 14:39:32 <dalias> they don't have meaningfun content unless you provided it, but they do exist!
2022-08-22 14:40:04 <dalias> as locales with the untranslated C locale messages, date formats, etc. and UTF-8 encoding
2022-08-22 14:40:09 <psykose> so how does one test for ru_RU.CP1251 actually not doing anything and just returning the default english values
2022-08-22 14:40:48 <dalias> i guess you could set it then all nl_langinfo and strcmp
2022-08-22 14:41:33 <dalias> it's possible it would be a better behavior to make strings containing . followed by something non-UTF-8 fail or produce clones of the C locale rather than of C.UTF-8
2022-08-22 14:41:58 <dalias> but this is a question of which behavior is better, not of making tests pass
2022-08-22 14:42:33 <dalias> there's a whole thread on the pros and cons of the behavior of "all locales exist" that desperately needs attention from people who care
2022-08-22 14:42:56 <dalias> and reviving that and trying to figure it out would be a lot more productive than trying to make hacks to make tests pass
2022-08-22 14:44:04 <dalias> (this is why i hate tests, same as why i hate warnings, btw. everyone just tries to do random unmotivated things to make them pass rather than trying to understand if there's an issue at all and whether the underlying issue has anything to do with what the test/warning output claims)
2022-08-22 14:45:20 <psykose> this just sounds like kicking things down the road, "it exists but actually it doesnt and you need these 40 other things to find out it really exists"
2022-08-22 14:45:36 <psykose> no wonder everyone hates this language where 'i want to see if i can print things in x language' is actually like 78 function calls
2022-08-22 14:46:20 <dalias> the problem is "user has LANG=foo set but no translations installed for foo so now UTF-8 is broken" is necessarily the outcome if it does not exist
2022-08-22 14:46:28 <dalias> since setlocale has to fail if it doesn't exist
2022-08-22 14:47:27 <dalias> and "applications have a hard time probing something they're not meant to be probing anyway" is a much lesser fault than "multilingual text doesn't work at all because translation to the user's language is not available"
2022-08-22 14:47:35 <psykose> could nobody have figured out "user has LANG=doesnt exist but actually we defaulted to C.UTF-8 or whatever instead"
2022-08-22 14:52:13 <dalias> ?
2022-08-22 14:52:51 <dalias> suppose someone is running a musl binary on a glibc system with no musl runtime stuff installed much less musl locale files
2022-08-22 14:53:00 <dalias> and they have LANG=fr in their environment
2022-08-22 14:53:17 <dalias> you either require them to unset this to run the musl binary correctly
2022-08-22 14:53:29 <dalias> or the musl binary has to accept LANG=fr as a success condition for setlocale
2022-08-22 14:54:17 <dalias> otherwise the binary behaves in a broken way where the user can't access *data* in their own language properly just because of missing message translations
2022-08-22 14:55:23 <dalias> the email thread on the topic explores ways to improve this that still meet the interface contract requirements of setlocale
2022-08-22 14:58:15 <ericonr> I'm interested in fixing this, but understanding how everything interacts is quite complicated 
2022-08-22 14:58:57 <psykose> my conclusion is it's all terrible and everyone should learn english instead
2022-08-22 14:59:29 <ericonr> Why fmt wants to support locales other than utf-8 i don't get
2022-08-22 14:59:59 <psykose> it's actually funnier than that as that is the entire test
2022-08-22 15:01:04 <psykose> but i mean it would fail with the same test and any locale
2022-08-22 15:01:30 <psykose> it just calls the c++ locale thing and expects the output to have a translated string, so understandably that is never going to work
2022-08-22 15:02:59 <psykose> i guess since the point is for every locale to exist, there's no point for the shim fix either
2022-08-22 15:03:06 <ericonr> Well that feels rather useless
2022-08-22 15:03:20 <ericonr> As a test
2022-08-22 15:03:20 <psykose> nor would it work for things that did 'en' instead of 'en.UTF-8'
2022-08-22 15:03:33 <psykose> it does say 'test legacy locale', haha
2022-08-22 15:03:47 <ericonr> mhm
2022-08-22 15:04:01 <psykose> the best patch is the one you don't write at all
2022-08-22 15:15:41 <skarnet> <psykose> my conclusion is it's all terrible and everyone should learn english instead
2022-08-22 15:15:53 <skarnet> that is unironically one of the reasons why I stick to system software
2022-08-22 15:16:05 <skarnet> nobody's going to give me shit for printing English error messages
2022-08-22 15:16:29 <panekj> just change all output to emoji, ez
2022-08-22 15:16:37 <skarnet> anything more applicative, more user-oriented, and I'd have to open the lid of the can of laser-shark-worms that is i18n
2022-08-22 15:16:41 <psykose> well actually the language itself sounds like the easy part
2022-08-22 15:16:51 <psykose> the rest of localisation and all the other formatting...
2022-08-22 15:17:34 <skarnet> yeah but even the language... when you want i18n, the simplest error message needs to go into a gettext db or a catalog
2022-08-22 15:17:50 <skarnet> and that's an order of magnitude of complexity added to your software right away
2022-08-22 15:18:53 <dalias> "break m17n to slightly improve i18n ux" = fail :-p
2022-08-22 15:19:15 <dalias> psykose, what is the "shim fix" above?
2022-08-22 15:20:54 <skarnet> m17n.org looks very well maintained :D
2022-08-22 15:20:58 <psykose> the thing from the other channel
2022-08-22 15:21:13 <psykose> to make libstdc++ reject some locale strings
2022-08-22 15:21:33 <psykose> that would never work though, so existing behaviour is mostly fine
2022-08-22 15:22:02 <psykose> skarnet: for a german website that is top class
2022-08-22 15:27:59 <Jean[m]1234> <psykose> "Jean: yeah i can just add the..." <- alright - I'll keep trying to get this thing to compile first, see what I can figure out
2022-08-22 15:28:27 <psykose> they changed the names of the stuff, but i assume it's not that hard
2022-08-22 15:29:49 <Jean[m]1234> 🤞
2022-08-22 15:29:58 <dalias> bbiab
2022-08-22 15:39:08 <ericonr> dalias: the question for the locale is: if someone does newlocale+uselocale, what difference does it make inside libc and should we make sure that part works, since it's not (as of yet) implemented in libstdc++
2022-08-22 15:39:28 <Ermine> skarnet: this complexity is inherent to human-machine interfaces, unfortunately
2022-08-22 15:39:40 <ericonr> Maybe it doesn't affect regex, but it does affect wprintf...
2022-08-22 15:40:14 <ericonr> psykose: ^^ I think this is reason enough to try and make it work...
2022-08-22 15:41:17 <ericonr> The .UTF-8 part is interesting tho; having seen dalias's reasoning,  it's at least annoying to have a binary completely fail because of LANG= values 
2022-08-22 15:42:14 <psykose> yeah, but it sounds more like the fix is "fix all of libstdc++ and the corner cases with all of this"
2022-08-22 15:42:33 <psykose> if it's as simple as that __cloc +- two checks, i guess we are almost there, though the *.UTF-8 is wrong
2022-08-22 15:42:41 <psykose> aside from that, i don't think i have the energy
2022-08-22 15:43:24 <psykose> i wonder what edge cases are actually common in the current state of passing it through
2022-08-22 15:43:50 <psykose> the random test localisation issues (that all get skipped before due to the throw) aren't real issues, so i wonder what is actually affected
2022-08-22 15:48:33 <skarnet> Ermine: there are several possible levels of complexity, and I'm happy to work at the level where "writing English" is still an acceptable level
2022-08-22 17:19:55 <andypost[m]> ikke: Looks it can save resources https://docs.gitlab.com/ee/user/project/merge_requests/methods/index.html#rebase-without-cicd-pipeline
2022-08-22 18:44:52 <Ermine> ncopa: please take a look at https://gitlab.alpinelinux.org/alpine/mdev-conf/-/merge_requests/1
2022-08-22 18:52:40 <psykose> https://img.ayaya.dev/8HRMFsb70hcS alignment.png
2022-08-22 19:10:44 <Ermine> Oh, that's eternal
2022-08-22 21:11:40 <Guest838> ptrc: the code you send me to make the squashfs iso does make an iso, but when I try running it on QEMU it says, "Could not boot from CDROM (code 0009)" and after some more stuff "No bootable device."
2022-08-22 21:11:53 <ptrc> ah, you have to use efi
2022-08-22 21:12:19 <ptrc> apk add ovmf; qemu-system-x86_64 ... -bios /usr/share/OVMF/OVMF.fd
2022-08-22 21:12:38 <Guest838> thanks, i will try it
2022-08-22 21:13:47 <Guest838> it booted now, nice
2022-08-22 21:14:08 <Guest838> does it do anything after boot? i see just a black screen
2022-08-22 21:15:16 <Guest838> after the grub screen i mean
2022-08-22 21:16:01 <ptrc> that's what i meant before, display isn't quite working properly
2022-08-22 21:16:06 <ptrc> but serial works just fine
2022-08-22 21:16:35 <Guest838> oh, ok
2022-08-22 21:17:55 <Guest838> thanks for the script, i will see what i can do now
2022-08-22 22:43:58 <dalias> psykose, can you explain what bogus stuff libstdc++ is doing that's breaking?
2022-08-22 22:44:18 <dalias> it should just be doing portable things with the C locale API
2022-08-22 22:47:11 <psykose> by default std::locale("") throws
2022-08-22 22:49:31 <psykose> "by default" -> not in alpine edge
2022-08-22 22:49:35 <psykose> LANG=anything-but-c
2022-08-22 23:06:06 <dalias> it should just succeed not throw
2022-08-22 23:06:21 <dalias> by calling newlocale
2022-08-22 23:12:36 <psykose> https://github.com/gcc-mirror/gcc/blob/cc4fa7a210b638d6a46f14dab17f2361389d18e1/libstdc++-v3/config/locale/generic/c_locale.cc#L246
2022-08-23 00:05:18 <dalias> the generic backend is just crap...
2022-08-23 00:05:39 <dalias> probably someone should copy the dragonfly one, call it posix, and fix the few nonportable things in it
2022-08-23 02:16:53 <Guest838> does anyone know how to get the foreground console working instead of the serial console? i've been playing around with ptrc's script for making a squashfs iso (https://gitlab.alpinelinux.org/-/snippets/265) and reading the 'mkimage.*.sh' scripts in aports, but i haven't made it work yet. i don't even understand how exactly does it work, i thought it just showed up by default,
2022-08-23 02:16:55 <Guest838> instead what i have is a blank screen
2022-08-23 02:25:37 <psykose> in qemu? add qemu-ui-gtk
2022-08-23 02:26:01 <psykose> unless you mean it works without that for a regular iso (personally i can't get a normal display without adding random shit)
2022-08-23 03:03:35 <Jean[m]1234> hi - I'm using fedora x64 and trying to setup an aarch64 alpine chroot. I'm bind-mounting the chroot (/var/chroots/alpine), as well as /dev and /sys, and proc. with a non-root user, I'm getting suid errors (sudo: effective uid is not 0). any idea how to fix that?
2022-08-23 03:04:49 <Jean[m]1234> 
ACTION posted a file: alpine (0KiB) < https://matrix.org/_matrix/media/r0/download/4ray.co/sloRWPGfgVgLDlHnTikEwqvQ >

2022-08-23 03:05:49 <Jean[m]1234> that's my current <del>script</del> HACK to chroot into alpine after a bootstrap
2022-08-23 03:23:33 <psykose> the binary could be not suid
2022-08-23 03:23:45 <psykose> or alternatively the chroot mount stuff is with =nosuid in options
2022-08-23 03:23:51 <psykose> verify in `mount` output
2022-08-23 03:25:24 <psykose> =users also implies nosuid
2022-08-23 03:26:29 <psykose> so i would guess you could also have /var/chroots/alpine options=users as an entry in fstab to facilitate a user mounting it; this disables suid for the implied reason of not letting a user gain root for free, for an example of that one
2022-08-23 03:26:33 <psykose> those are my guesses :)
2022-08-23 08:59:06 <Guest876> psykose: what i meant is that there is no display/foreground terminal on the iso that ptrc's build script generates. and i'm wondering how to get it working, and i have no clue what it could be. it doesn't work on qemu and booting on hardware, so it must be something in the iso.
2022-08-23 08:59:20 <Guest876> the iso builds *really* fast if you want to try it out
2022-08-23 08:59:33 <Guest876> https://gitlab.alpinelinux.org/-/snippets/265
2022-08-23 09:44:33 <bl4ckb0ne> PureTryOut: new sdl release works like a charm, thanks
2022-08-23 09:50:47 <psykose> things working?? what??
2022-08-23 09:50:53 <psykose> must.. break..
2022-08-23 09:50:54 <bl4ckb0ne> yes
2022-08-23 09:51:04 <bl4ckb0ne> i even yeeted a few of my directfb hacks
2022-08-23 09:51:11 <bl4ckb0ne> and it... works
2022-08-23 09:51:15 <psykose> impressive
2022-08-23 09:51:19 <bl4ckb0ne> innit
2022-08-23 09:51:25 <psykose> the only thing i tried via directfb is the last thing using sdl1 in alpine
2022-08-23 09:51:26 <psykose> netsurf
2022-08-23 09:51:34 <psykose> and with directfb sdl12-compat sdl2 it segfaults
2022-08-23 09:51:41 <psykose> due to not loading anything, i think
2022-08-23 09:51:57 <psykose> fixing that would be nice, then i can delete sdl1
2022-08-23 09:52:03 <bl4ckb0ne> sauerbraten still fails tho
2022-08-23 09:52:07 <psykose> does it
2022-08-23 09:52:07 <bl4ckb0ne> but its because they do dumb shit
2022-08-23 09:52:24 <psykose> ah rite
2022-08-23 09:52:27 <bl4ckb0ne> https://paste.sr.ht/~bl4ckb0ne/b3d5b68a331a8a19ecf2d198c6eb86c075eb1d91
2022-08-23 09:52:30 <bl4ckb0ne> thats very dumb shit
2022-08-23 09:52:38 <psykose> strange
2022-08-23 09:52:43 <psykose> i don't see myself on that page
2022-08-23 09:53:25 <panekj> I'm thinking of dropping lite-xl
2022-08-23 09:53:40 <bl4ckb0ne> just drop directfb
2022-08-23 09:54:33 <psykose> you have a penance for editors with completely fucked fonts
2022-08-23 09:54:40 <psykose> lite-xl can't even use fontconfig
2022-08-23 09:54:50 <panekj> They can't even release new version properly
2022-08-23 09:55:00 <psykose> lapce can't even render above ~0.1fps with my fonts
2022-08-23 09:55:03 <panekj> They needed someone with macos to test, I provided them with such person
2022-08-23 09:55:05 <panekj> they tested
2022-08-23 09:55:12 <panekj> 3 months later, no new release
2022-08-23 09:56:58 <psykose> wheres ur new release
2022-08-23 09:57:32 <panekj> 
ACTION notices your release

2022-08-23 09:57:38 <panekj> uwu
2022-08-23 09:57:51 <psykose> whole aports with mold would work for prolly 95%
2022-08-23 09:57:54 <psykose> it's not very good though
2022-08-23 09:58:05 <panekj> Not saying it is
2022-08-23 09:58:14 <psykose> i tried it on chromium once and it was slower than lld, hah
2022-08-23 09:58:20 <psykose> on top of fucking up my disk io
2022-08-23 09:58:21 <psykose> wild ride
2022-08-23 09:58:24 <panekj> But if I can find bugs and report them, that's good news for mold dev?
2022-08-23 09:58:29 <psykose> is very good
2022-08-23 09:58:31 <psykose> but most of them are going to be
2022-08-23 09:58:37 <psykose> 'musl is horrific at anything using locks'
2022-08-23 09:58:40 <psykose> lld has the same issue
2022-08-23 09:58:47 <panekj> ¯\_(ツ)_/¯
2022-08-23 09:58:54 <psykose> ~40% of the cpu time is on lock/unlock/malloc/free
2022-08-23 09:58:58 <panekj> 
ACTION switches to glibc

2022-08-23 09:59:10 <panekj> mold uses mimalloc tho?
2022-08-23 09:59:22 <PureTryOut> bl4ckb0ne: ha, great!
2022-08-23 09:59:23 <psykose> mimalloc does not provide locks
2022-08-23 10:00:24 <panekj> also building whole aports with mold is just an excuse for my machine to do something instead of sitting idle
2022-08-23 10:00:57 <psykose> got a better idea
2022-08-23 10:01:04 <psykose> 1) i give you a shipping address
2022-08-23 10:01:07 <psykose> 2) ..
2022-08-23 10:01:09 <psykose> 3) profit
2022-08-23 10:01:21 <panekj> What would I do with shipping address?
2022-08-23 10:01:26 <bl4ckb0ne> ship
2022-08-23 10:01:27 <psykose> you send it to me : )
2022-08-23 10:01:55 <panekj> sure, let me just fly to Helsinki
2022-08-23 10:01:58 <panekj> to get it
2022-08-23 10:03:30 <psykose> that's even easier i just give you an ssh key
2022-08-23 10:03:56 <panekj> tbh, I wouldn't mind
2022-08-23 10:04:02 <psykose> ah depends on the specs
2022-08-23 10:04:13 <psykose> i have a 'big' machine already, but a) its in a container and b) the cores are slow
2022-08-23 10:04:21 <psykose> so if it improves on that it'd be fun
2022-08-23 10:04:31 <panekj> also, it's running Ubuntu
2022-08-23 10:04:42 <psykose> my own computer beats it 50% of the time just due to slow ass cores
2022-08-23 10:05:18 <panekj> AMD Ryzen™ 9 5950X, 128 GB DDR4 ECC, 2 x 3.84 TB NVMe
2022-08-23 10:05:30 <psykose> definitely beats it by far
2022-08-23 10:05:46 <panekj> beats almost everything, even my PC isn't so good
2022-08-23 10:06:01 <psykose> the alpine shared one is 2x https://ark.intel.com/content/www/us/en/ark/products/91767/intel-xeon-processor-e52650-v4-30m-cache-2-20-ghz.html
2022-08-23 10:06:12 <psykose> though with 252g ram
2022-08-23 10:06:36 <panekj> Would be great if I could install Alpine on it though
2022-08-23 10:06:40 <psykose> cant ye
2022-08-23 10:06:45 <psykose> what restrictions did they put on it
2022-08-23 10:06:58 <psykose> is it the hetzner root memes
2022-08-23 10:07:02 <panekj> none, it just doesn't 
2022-08-23 10:07:05 <panekj> boot
2022-08-23 10:07:08 <psykose> ah, hmm
2022-08-23 10:07:10 <psykose> sounds debuggable
2022-08-23 10:07:21 <psykose> i assume you have some remote shell shit
2022-08-23 10:07:34 <panekj> I can get 3h of KVM
2022-08-23 10:07:38 <psykose> oof ow
2022-08-23 10:07:53 <psykose> and i guess the simple .iso didnt work
2022-08-23 10:08:04 <panekj> it boots iso, it doesn't boot after install
2022-08-23 10:08:28 <psykose> what boot method is it
2022-08-23 10:08:28 <panekj> even with the nvme module shenanigans
2022-08-23 10:08:32 <panekj> uefi
2022-08-23 10:08:34 <psykose> and how does it fail
2022-08-23 10:08:48 <panekj> It just doesn't see system
2022-08-23 10:09:30 <psykose> the efi loader cant boot the image at all?
2022-08-23 10:09:36 <psykose> pre-everything from the motherboard
2022-08-23 10:18:47 <panekj> you can go to https://irclogs.alpinelinux.org/%23alpine-linux-2022-05.log and start reading from 2022-05-11 10:23:25
2022-08-23 10:18:53 <panekj> (:
2022-08-23 10:21:04 <psykose> is grub at esp/efi/boot/bootx64.efi
2022-08-23 10:21:58 <psykose> you never even load anything in the first place, so either 1) the efi loader has no entry and that default one is missing, or it's fucked, or the raid drives have some quirk with that is an esp partition at all (i don't remember how it works)
2022-08-23 10:22:10 <panekj> you know, if you asked me that in may 12 I might have told you
2022-08-23 10:22:26 <panekj> iirc, yes, I did have boot efi stuff
2022-08-23 10:23:22 <psykose> did you set up the partitions something like this https://www.linuxquestions.org/questions/slackware-14/how-to-manage-efi-esp-in-raid-setup-4175687630/
2022-08-23 10:23:54 <panekj> I believe I let setup-disk do all the work for me
2022-08-23 10:25:54 <psykose> my very light perusing of the alpine-conf test improvements see no tests for raid, let alone raid+efi
2022-08-23 10:26:05 <psykose> i would recomment trying again and not using setup-disk at all
2022-08-23 10:26:38 <psykose> (manual partition, then setup-disk /mnt/root to install with the esp mounted inside that and the disk mounted all by yourself first, with all the raid shenanigans)
2022-08-23 10:27:11 <psykose> but i can't guess as to why it would fail really
2022-08-23 10:29:45 <psykose> hmm
2022-08-23 10:29:48 <panekj> mhmm
2022-08-23 10:29:49 <psykose> firefox 104 actually got faster
2022-08-23 10:29:57 <psykose> ctrl+tab is like 2x faster
2022-08-23 10:29:59 <psykose> literally
2022-08-23 10:30:01 <panekj> wow
2022-08-23 10:30:08 <psykose> i hold that down every time i reopen/new release/randomly
2022-08-23 10:30:17 <panekj> but will they fix years long bugs and stop removing features?
2022-08-23 10:30:22 <psykose> quite literally just zonks the screen at 75 key repeat rate
2022-08-23 10:30:28 <psykose> definitely not
2022-08-23 10:30:43 <panekj> I guess, I'm still with the chromium based browsers then
2022-08-23 10:30:57 <psykose> aside from the shitty text/* bug i cant think of anything else meaningful
2022-08-23 10:31:02 <psykose> and i never use features
2022-08-23 10:31:28 <panekj> I need some kind of PWAs 
2022-08-23 10:31:38 <psykose> ah like that
2022-08-23 10:31:45 <psykose> i used to.. install pwas in chromium
2022-08-23 10:31:47 <panekj> I'm interested in running separate browser for each electron app
2022-08-23 10:31:50 <psykose> but there's a bug that kinda broke them
2022-08-23 10:32:03 <psykose> something with media keys and window focus
2022-08-23 10:32:07 <panekj> and so far only Microsoft gives shit about PWAs
2022-08-23 10:32:07 <psykose> forgot now
2022-08-23 10:32:49 <psykose> (and yes, only pwas, regular tab is fine, i forgot what the issue even is)
2022-08-23 10:33:08 <psykose> i still use chromium just to keep one tab open to play music
2022-08-23 10:33:20 <psykose> since you can't really open two browsers in a way i know of, hmm
2022-08-23 10:33:26 <psykose> i guess firefox with a different session
2022-08-23 10:33:29 <panekj> also given that text/* thingy, I'm afraid of looking into ff source code
2022-08-23 10:33:31 <psykose> but firefox sessions are ass to manage
2022-08-23 10:33:58 <panekj> "since you can't really open two browsers in a way i know of" wdym
2022-08-23 10:34:12 <panekj> like separate profiles?
2022-08-23 10:34:15 <psykose> you can open two windows but then things are fucked depending on how they close
2022-08-23 10:34:23 <psykose> i.e. usually i close, reopen, reopen previous session
2022-08-23 10:34:28 <panekj> true
2022-08-23 10:34:30 <psykose> if i had two windows on different displays
2022-08-23 10:34:37 <psykose> then if i closed them wrong.. i just lose the session lol
2022-08-23 10:34:41 <psykose> the profiles would work, but that shit sucks
2022-08-23 10:34:46 <psykose> easier to just use another browser
2022-08-23 10:35:14 <psykose> well, either lose or on reopen it starts opening multiple windows and shit
2022-08-23 10:35:14 <panekj> yeah, I have everything in a single window and all "app" stuff is forced to be pwa-like
2022-08-23 10:35:20 <psykose> dunno, we all have our spacebar heating
2022-08-23 10:36:24 <panekj> the only thing I fondly remember from ff is it could hold so many tabs opened
2022-08-23 10:36:49 <psykose> funnily ptrc can't
2022-08-23 10:37:00 <psykose> she opens firefox with ~30 tabs and it never deallocates anything and sits there with 8+gb
2022-08-23 10:37:06 <psykose> i can open 500 and they just take nothing
2022-08-23 10:37:12 <panekj> that was few years ago so might hav echannged
2022-08-23 10:37:27 <panekj> "works on my machine"
2022-08-23 10:37:28 <psykose> idk, when people tell me things don't work i don't even know where to start
2022-08-23 10:37:32 <psykose> literally everything always works for me
2022-08-23 10:37:37 <panekj> same
2022-08-23 10:37:45 <psykose> until it doesn't and then i get to dig out some really weird bugs
2022-08-23 10:37:53 <panekj> people complain about windows and I'm like: never me
2022-08-23 10:40:03 <panekj> stop the internet, delete everything and start from scratch
2022-08-23 10:41:03 <psykose> same
2022-08-23 11:10:48 <panekj> > It is dangerous to rely on assumptions about Git tags that correspond to crates.io or other releases. Even if everybody is well intentioned, it is easy to accidentally modify a tag so it points to the wrong commit, and it is easy to accidentally release something that isn't exactly what matches the tagged revision.
2022-08-23 11:10:50 <panekj> > This is actually the main reason I've always avoided the tagging--so that people don't make such dangerous assumptions. Not just for this library, but all libraries. Git tags are simply not a good tool for security purposes.
2022-08-23 11:11:46 <psykose> broke: git tag + sha
2022-08-23 11:11:53 <psykose> woke: no git tag, still a sha
2022-08-23 11:11:56 <psykose> <GIGABRAIN>
2022-08-23 11:12:30 <psykose> this makes about as much sense as labels and dev nodes all being useless and everything having to be UUID actually
2022-08-23 11:12:40 <psykose> 'well actually you see!!! they can change!!'
2022-08-23 11:45:18 <Habbie> somebody please remind me - CVE fixes for a community package go into master and 3.16, and not 3.15?
2022-08-23 11:45:47 <psykose> yeah, last release
2022-08-23 11:46:31 <psykose> necessarily they don't have to be cve; some nice fixes can also be proposed, some security ones don't get cves, etc
2022-08-23 11:46:38 <psykose> the main thing is trying for no breaking
2022-08-23 11:53:41 <Habbie> ack - that part i knew :)
2022-08-23 11:53:58 <Habbie> thanks
2022-08-23 13:21:26 <minimal> panekj: your ESP partition is a MD RAID partition? That's likely the problem if Grub is not being loaded - UEFI knows nothing about MD software raid andn so will not use such an ESP. You mentioned you used setup-disk - for MD it uses RAID1 with metadata version 0.9, this will not work - apparently RAID1 can be made to "work" if metadata version 1.0 is used (which writes MD data to the *end* of the partition, not the start)
2022-08-23 13:23:45 <panekj> I have no idea, it was months ago
2022-08-23 13:23:56 <psykose> ah, looks like the resident booting expert is here
2022-08-23 13:24:02 <psykose> you should go try again
2022-08-23 13:24:46 <panekj> yeah, let me just yeet out all my data to debug alpine
2022-08-23 13:25:56 <minimal> panekj: basically either don't RAID the ESP partition at all, or use the kludge for RAID1 with version 1.0 metadata but in either cases you'd still potentially have "sync" issues as UEFI knows nothing about RAID and so the UEFI could potentially change things (in one of the ESPs)
2022-08-23 13:26:33 <panekj> I love software
2022-08-23 13:28:19 <minimal> panekj: software doing exactly what it was designed to do - UEFI does not support RAID, it just sees a FAT filesystem, and is blissfully unaware if there's any (Linux) RAID
2022-08-23 13:29:43 <minimal> panekj: the linuxquestions articel that psykose linked to specifically mentioned the importance of using "--metadata=1.0" for the partition holding the ESP
2022-08-23 13:31:01 <minimal> I'll look at submitting a MR to alpine-conf for setup-disk to handle RAIDing a ESP filesystem differently that it does currently
2022-08-23 13:31:55 <panekj> error: reference to packed field is unaligned                                                                                                                                                                                                                                                                                                                                                                                                                 
2022-08-23 13:31:55 <panekj> --> ruby-structs/src/ruby_2_5_0.rs:2373:223
2022-08-23 13:32:59 <psykose> ?
2022-08-23 13:33:07 <panekj> rbspy fails to build
2022-08-23 13:34:41 <psykose> fixed in master
2022-08-23 13:36:10 <minimal> Jean[m]1234: for your aarch64 alpine chroot, you said you are trying to run thid on a Fedora x86_64 machine, correct? Then you need to set up binfmt/qemu-user-aarch64 to be able to run aarch64 binaries on a non-aarch64 machine
2022-08-23 13:36:42 <panekj> psykose: I don't see any change?
2022-08-23 13:36:50 <psykose> i would have assumed that would have failed ages ago due to an exec format error
2022-08-23 13:36:57 <psykose> panekj: https://github.com/rbspy/rbspy/commit/f5a8eecfbf2ad0b3ff9105115988478fb760d54d
2022-08-23 13:37:10 <psykose> https://github.com/rbspy/rbspy/issues/368
2022-08-23 13:37:12 <panekj> oh you mean upstream lol
2022-08-23 13:37:26 <panekj> someone should patch it then
2022-08-23 13:37:32 <panekj> <:
2022-08-23 13:37:35 <psykose> what for
2022-08-23 13:39:40 <Jean[m]1234> minimal: right, what psykose said - I'm using qemu-user successfully. the suid issue comes up when I switch to a non-root user and try to run sudo
2022-08-23 13:42:14 <Jean[m]1234> psykose: sudo is definitely suid; the mount itself doesn't show `nosuid` or `users`; I don't have a custom fstab for the chroot either, didn't really think that was a common thing to do from your host OS, heh.
2022-08-23 13:42:44 <Jean[m]1234> `/dev/mapper/luks-[uuid] on /var/chroots/alpine type btrfs (rw,relatime,seclabel,compress=zstd:1,ssd,space_cache,subvolid=256,subvol=/root)`
2022-08-23 13:43:02 <Jean[m]1234> I'm wondering if it could be related to `seclabel`
2022-08-23 13:50:46 <minimal> Jean[m]1234: not familiar with btrfs and whenever I use a Alpine chroot I use a loopback device to present a partitioned disk image to the chroot, I've not tried presenting device-mapper devices (i.e. LUKS or LVM) directly to the chroot
2022-08-23 13:54:22 <Jean[m]1234> hm I see
2022-08-23 13:54:59 <Jean[m]1234> so do you use e.g. a qcow2 sparse image? or an actual partition?
2022-08-23 13:58:34 <Jean[m]1234> and to clarify, I don't think btrfs/luks is the issue; I use luks+btrfs on 2 pinephones with no issues (natively, of course). the mount entry looks kinda funny because my host OS is also luks+btrfs, and the mount is a bind mount to itself. host OS is SELinux-enabled though, which may be breaking something subtly
2022-08-23 14:02:44 <minimal> Jean[m]1234: I create a simple "RAW" disk image, partition and create filesystems on it (plus LUKS and/or LVM) and then use a loopback device to present the partitions into the chroot
2022-08-23 14:05:30 <Jean[m]1234> nice. I'll give that a try.
2022-08-23 14:06:58 <minimal> Jean[m]1234: I have my own comprehensive/over-complicated shell script for doing all this ;-)
2022-08-23 14:07:36 <Jean[m]1234> ohh is it proprietary all rights reserved ©️ ?
2022-08-23 14:07:40 <Jean[m]1234> :P
2022-08-23 14:09:58 <minimal> Jean[m]1234: quite the opposite ;-)  https://github.com/dermotbradley/create-alpine-disk-image
2022-08-23 14:11:57 <Jean[m]1234> 🤘
2022-08-23 14:12:03 <Jean[m]1234> thank you :)
2022-08-23 14:13:17 <minimal> Jean[m]1234: are you setting this up for a VM or a physical device?
2022-08-23 14:13:41 <Jean[m]1234> vm/chroot alone
2022-08-23 14:14:06 <Jean[m]1234> just trying to get a good/stable environment for cross-compiling
2022-08-23 14:14:15 <Jean[m]1234> (mainly to compile chromium heh)
2022-08-23 14:17:50 <Jean[m]1234> probably gonna have to go the vm route :')
2022-08-23 14:42:24 <Hello71> https://github.com/multiarch/qemu-user-static/issues/17 says that you need to set binfmt_misc flags POC
2022-08-23 14:42:44 <Hello71> i vaguely recall that qemu-openrc already does this properly
2022-08-23 16:47:33 <Jean[m]1234> hmmmm
2022-08-23 16:47:54 <Jean[m]1234> very nice find, by the way.
2022-08-23 16:52:41 <Jean[m]1234> Hello71: yes! that did it!!
2022-08-23 16:54:35 <Jean[m]1234> added `C` to the end of the aarch64 binfmt_misc config (fedora splits up all configs for architectures into different files, so /usr/lib/binfmt.d/qemu-aarch64-static.conf), as per https://docs.kernel.org/admin-guide/binfmt-misc.html, and sudo works for my non-root user!
2022-08-23 16:54:36 <panekj> Jean[m]1234: if you want chroot alone it's better if you just use https://github.com/alpinelinux/alpine-chroot-install
2022-08-23 16:55:36 <Jean[m]1234> ah! awesome!
2022-08-23 16:55:52 <Jean[m]1234> 5 seconds!!
2022-08-23 16:55:59 <Jean[m]1234> :)
2022-08-23 19:55:53 <sam_> psykose: re binutils 2.39, not seen anything with strip yet, any news on it?
2022-08-23 21:54:19 <Guest917> ptrc: i didn't wanna have to bother you again since you already gave me that nice script, but do you have any clue on why the tty display wont work? a starting point for fixing it maybe? i've been researching and reading the mkimage scripts for a long time now and i still don't know what to do
2022-08-23 21:55:06 <Guest917> i thought of modifying the mkimage scripts to install the apks instead of just saving their overlay, but it doesn't seem to be that simple
2022-08-23 21:57:22 <ptrc> i.. could try debugging it later, maybe it's because of using grub-mkrescue instead of just installing a bootloader and making the iso with xorriso
2022-08-23 21:58:19 <Guest917> i will try to use xorriso and see what happens
2022-08-24 01:08:31 <minimal> Guest917: with the script you are using modify the "modprobe" on line 27 to add "simpledrm" to the module list - that will likely resolve your no-display problem
2022-08-24 01:12:10 <Guest917> oh my god, it worked!
2022-08-24 01:12:28 <Guest917> all these hours of research and the fix was a single word
2022-08-24 01:12:39 <Guest917> thx minimal
2022-08-24 01:13:04 <minimal> Guest917: Alpine 3.15, 3.16, and Edge use DRM/KMS instead of FBdev so you need to load simpledrm (and then later perhaps a specific DRM module for your graphics card)
2022-08-24 01:13:28 <Guest917> ooh, so thats why
2022-08-24 01:14:10 <minimal> Guest917: if you look at the "official" initramfs init (from mkinitfs package) its modprobe includes simpledrm
2022-08-24 02:27:38 <Jean[m]1234> minimal, panekj, psykose, thank you all kindly. I'm nearly halfway done compiling chromium in a chroot!
2022-08-24 05:40:17 <psykose> sam_: haven't found anything either
2022-08-24 08:35:59 <skarnet> ncopa: I'm going to write a long answer to your new TSC issue, but I need a functional brain for that, so it may have to wait until tomorrow or Friday, depending on when the workers outside my window stop waking me up at ungodly hours
2022-08-24 08:55:22 <ncopa> :+1: 
2022-08-24 08:56:04 <ncopa> i dont think we are in a hurry
2022-08-24 09:49:00 <panekj> if the script helped, thanks go to jirutka
2022-08-24 09:51:14 <psykose> script for what
2022-08-24 09:51:28 <psykose> ncopa: could you restart the 3.16 aarch64 builder since it has been stuck for a few days
2022-08-24 09:51:44 <panekj> psykose: why are you interested all of the sudden? (:
2022-08-24 09:51:51 <psykose> ?
2022-08-24 09:52:26 <panekj> https://github.com/alpinelinux/alpine-chroot-install
2022-08-24 09:55:03 <ncopa> psykose: i have restarted build-3-16-aarch64
2022-08-24 09:55:17 <psykose> thanks
2022-08-24 10:12:15 <wyk72> found a bug in the network setup I think
2022-08-24 10:12:23 <wyk72> bonding does not enslave interfaces
2022-08-24 10:12:30 <wyk72> as per alpine examples
2022-08-24 10:12:39 <wyk72> in the wiki
2022-08-24 10:12:53 <wyk72> latest Alpine does not enslave interfaces at all
2022-08-24 10:12:58 <wyk72> you have to do it manually
2022-08-24 10:13:53 <panekj> what wiki page
2022-08-24 10:16:57 <wyk72> https://wiki.alpinelinux.org/wiki/Bonding
2022-08-24 10:25:38 <wyk72> as a workaround I configure bonding via iproute2
2022-08-24 10:34:39 <bl4ckb0ne> PureTryOut: taking acrack at onnxruntime (moving discussion from #monado)
2022-08-24 10:47:19 <bl4ckb0ne> bleh theres all of that stuff https://github.com/microsoft/onnxruntime/blob/main/.gitmodules to package
2022-08-24 10:47:29 <bl4ckb0ne> giving up for lacks of fuck to give
2022-08-24 10:56:09 <psykose> they have a toggle for system libs for most of that
2022-08-24 10:57:52 <psykose> though i am curious how this would even be usable
2022-08-24 10:58:07 <psykose> i never found anything that works with any gpu ML stuff
2022-08-24 10:58:24 <psykose> amd is completely fucked and don't provide support for anything except their proprietary pro driver
2022-08-24 10:58:29 <psykose> nvidia works but only on glibc
2022-08-24 10:58:45 <psykose> intel would work in theory, and we even had the intel runtime thing packaged, but it was unmaintained and i removed it
2022-08-24 10:58:49 <psykose> aside from that it builds i guess
2022-08-24 10:59:21 <ericonr> the intel runtime is a pain because of llvm versions :(
2022-08-24 10:59:27 <psykose> it was easy to patch
2022-08-24 10:59:39 <psykose> just your usual llvm migration patch
2022-08-24 10:59:56 <psykose> they even had an llvm14 patch merged into main shortly around 14 being release, just needed like 3 touchups
2022-08-24 11:01:09 <ericonr> I had some weird issues, but I think I was on llvm12 back then
2022-08-24 11:01:14 <psykose> i did all the work for the whole toolchain (though i had no idea how to actually then test it), but the person maintaining it didn't think it was a good idea or something
2022-08-24 11:01:21 <psykose> then i removed it from aports since the person doesn't update it anyway
2022-08-24 11:01:44 <ericonr> ouch
2022-08-24 11:02:05 <psykose> (wasn't installable due to linking to something that didn't exist)
2022-08-24 11:02:10 <psykose> for over a year
2022-08-24 11:02:10 <ericonr> mhm
2022-08-24 11:02:18 <ericonr> I get removing it :P
2022-08-24 11:05:21 <bl4ckb0ne> psykose: surprisingly ive had good runs with onnxruntime
2022-08-24 11:05:26 <bl4ckb0ne> its used by monado for hand tracking
2022-08-24 11:05:41 <bl4ckb0ne> i have it installed locally from a build with the submodules
2022-08-24 11:05:43 <bl4ckb0ne> so far so good
2022-08-24 11:05:48 <bl4ckb0ne> but its a nightmare to package
2022-08-24 11:06:24 <psykose> ah, i see
2022-08-24 11:27:12 <PureTryOut> yeah 😢 I need it for something else, but would also love to have it in Monado
2022-08-24 11:29:28 <PureTryOut> bl4ckb0ne: do you configure it with CMake directly? I needed it for a python lib so was doing it through `setup.py`
2022-08-24 11:29:43 <bl4ckb0ne> yeah with cmake directly
2022-08-24 11:29:56 <bl4ckb0ne> unit tests are failing and their setup script does not disable them
2022-08-24 11:30:02 <PureTryOut> oh fun
2022-08-24 11:30:15 <PureTryOut> if it's ran by ctest then we can at least disable them ourselves without patching
2022-08-24 11:30:32 <bl4ckb0ne> https://github.com/microsoft/onnxruntime/issues/9155
2022-08-24 11:30:40 <bl4ckb0ne> its building them that fails
2022-08-24 11:36:00 <PureTryOut> ah... fun
2022-08-24 11:39:17 <psykose> this git repository is bigger than aports
2022-08-24 11:39:21 <psykose> the hell did they put in it
2022-08-24 11:40:20 <PureTryOut> It's Microsoft and it's machine learning, so quite a bunch of shit probably
2022-08-24 11:42:03 <bl4ckb0ne> maths
2022-08-24 11:42:27 <psykose> this repo isn't even that big if you link system protobuf/abseil
2022-08-24 11:42:28 <psykose> not bad
2022-08-24 11:42:35 <psykose> few hundred objects
2022-08-24 11:52:19 <psykose> do you know what they normally build against? the last time i saw that kind of error with ::sinf kind of things it was a libc++/libstdc++ issue
2022-08-24 11:52:33 <psykose> libstdc++ has some wrong std cpp definitions around those
2022-08-24 12:00:03 <bl4ckb0ne> yeah
2022-08-24 12:00:09 <bl4ckb0ne> didnt look around much
2022-08-24 12:01:00 <psykose> ah
2022-08-24 12:01:02 <psykose> think i got it
2022-08-24 12:06:03 <psykose> builds with lto and the tests enabled
2022-08-24 12:06:56 <psykose> https://img.ayaya.dev/xwuW2wCNYyWA
2022-08-24 12:08:17 <psykose> at least one of the tests segfaults entirely, have fun with that part
2022-08-24 12:14:14 <bl4ckb0ne> nice ty
2022-08-24 12:17:49 <adhawkins> Getting the following trying to build using the CI container. https://paste.debian.net/1251567/
2022-08-24 12:17:53 <adhawkins> Do I need to update the container?
2022-08-24 12:25:11 <psykose> yes
2022-08-24 12:25:59 <adhawkins> Thanks psykose
2022-08-24 12:31:45 <adhawkins> Yep, that's done it. Thanks again.
2022-08-24 13:09:50 <minimal> wyk72: isn't bonding handled via ifupdown-ng: https://github.com/ifupdown-ng/ifupdown-ng/blob/main/doc/interfaces-bond.scd
2022-08-24 13:32:18 <wyk72> minimal: you're right
2022-08-24 13:32:50 <wyk72> minimal: wiki has become very misleading
2022-08-24 13:33:11 <wyk72> mininal: the options are totally different
2022-08-24 13:34:19 <minimal> wyk72: any wiki article may have been accurate at the time it was written but perhap never been updated to reflect chages in Alpine since then
2022-08-24 14:01:29 <psykose> if it starts with 'alpine 2+' you know it's ancient
2022-08-24 14:05:07 <minimal> psykose: I propose "wikiops" where wiki articles get automatically written by the latest greatest AI/ML code parsing the revised docs of updates to software packages
2022-08-24 14:06:17 <psykose> i'm in
2022-08-24 14:06:35 <wyk72> minimal: still the examples from ifupdown-ng do not work out of the box
2022-08-24 14:07:14 <minimal> psykose: and the AI/ML "bot" could be named after everyone's favourite wiki editor ;-)
2022-08-24 14:08:12 <minimal> wyk72: so that would be something to open an Issue with ifupdown-ng upstream?
2022-08-24 14:09:59 <wyk72> nope 
2022-08-24 14:10:02 <wyk72> my bad
2022-08-24 14:10:13 <wyk72> it's working as expected
2022-08-24 14:12:44 <wyk72> the module bonding" should be in /etc/modules
2022-08-24 14:12:49 <wyk72> that was my fault
2022-08-24 14:16:32 <minimal> wyk72: I'd assumed ifupdown-ng would load any required modules automatically, I guess not
2022-08-24 14:20:06 <mercenar-> it could try. it also could try to apk install bonding if not installed. depends on how far the user's hand needs to be held.
2022-08-24 14:20:48 <wyk72> I prefer to use a manual script since I find it less obscure and more  reliable
2022-08-24 14:20:54 <wyk72> it's just 3 lines
2022-08-24 14:21:13 <wyk72> modprobe bonding mode=4 miimon=100 xmit_hash_policy=1 ifconfig bond0  up ip link set eth3 master bond0 ip link set eth4 master bond0
2022-08-24 14:21:59 <psykose> hey that's four lines
2022-08-24 14:22:25 <mercenar-> with appropriate use of && it's one line ;)
2022-08-24 14:22:37 <minimal> psykose: don't forget the shebang line as well ;-)
2022-08-24 14:24:06 <mercenar-> but, but, it spawns 4 subprocesses. That may delay boot time by 800us
2022-08-24 14:24:53 <minimal> mercenar-: I could tell you except I've never managed to get bootchart to work in Alpine...
2022-08-24 14:25:56 <wyk72> yeah it's four
2022-08-24 14:26:16 <wyk72> copy&pasted from shell
2022-08-24 14:26:33 <wyk72> it's into a local.d
2022-08-24 14:27:35 <wyk72> this stupid bonding took me 4 hours of tries, this netgear ancient switch sucks too
2022-08-24 14:27:49 <wyk72> now it's working, let's not touch it
2022-08-24 15:42:18 <ncopa> Anyone want to test if linux-lts-5.15.62-r2 fixes the nasty xen issue we have?
2022-08-24 16:46:10 <ncopa> seems like rustc segfaults on riscv64
2022-08-24 16:49:55 <psykose> at least 100 times
2022-08-24 16:49:56 <psykose> or hangs
2022-08-24 16:52:07 <bl4ckb0ne> its a rust feature
2022-08-24 17:18:40 <q66> i thought rust was memory safe so it couldn't segfault :^)
2022-08-24 17:21:19 <psykose> i think it's a feature of riscv that things must segfault at least 20% of the time
2022-08-24 17:21:25 <psykose> it's just what the architecture was designed to do
2022-08-24 17:34:02 <bl4ckb0ne> or is it rust that just pegs the vm
2022-08-24 18:22:14 <bl4ckb0ne> oh boy, cemu is finally open source and it looks like a mess to compile
2022-08-24 18:22:16 <bl4ckb0ne> uses vcpkg
2022-08-24 18:23:33 <orbea> oh fun...
2022-08-24 18:24:03 <orbea> i wonder if they are open to using a non-terrible build system
2022-08-24 18:25:09 <psykose> what would that be
2022-08-24 18:26:02 <orbea> not vcpkg,but seems they also use cmake
2022-08-24 18:26:27 <orbea> https://github.com/cemu-project/Cemu/blob/main/CMakeLists.txt
2022-08-24 18:26:34 <bl4ckb0ne> cmake is decent
2022-08-24 18:26:40 <bl4ckb0ne> has its flaws but oh hey it works
2022-08-24 18:26:48 <bl4ckb0ne> vcpkg looks like another M$ abomination
2022-08-24 18:26:53 <bl4ckb0ne> i cut it mid compiling boost
2022-08-24 18:28:41 <panekj> > I like it because it generally seems to be the most hassle free across platforms. But yes, ideally we would conditionally use system libraries if present. Will look into this
2022-08-24 18:31:06 <psykose> the funniest part here is they're using cmake wrong so you have to pass the compiler and make/ninja as an argument
2022-08-24 18:31:07 <psykose> haha what
2022-08-24 18:31:43 <bl4ckb0ne> *sigh*
2022-08-24 18:31:54 <bl4ckb0ne> good news is yeeting the dumbass vcpkg line works
2022-08-24 18:32:01 <bl4ckb0ne> bad news is that now it chokes on finding curl
2022-08-24 18:32:05 <bl4ckb0ne> 
ACTION curses at cmake

2022-08-24 18:33:04 <psykose> you can replace find_package with the usual pkg_check_modules thing, the former only works for things that ship .cmake
2022-08-24 18:33:15 <psykose> still need to get an imgui copy though
2022-08-24 18:33:29 <psykose> aside from that this uses wxwidgets so i already closed the page
2022-08-24 18:33:34 <psykose> worst toolkit known to man
2022-08-24 18:33:52 <bl4ckb0ne> agreed
2022-08-24 18:34:01 <psykose> 3.2 actually fixes wayland can you believe it
2022-08-24 18:34:04 <bl4ckb0ne> ive once worked on a project using both scons and wxwidget
2022-08-24 18:34:09 <bl4ckb0ne> but at least i was paid
2022-08-24 18:34:11 <orbea> actually surprising how few vendored deps cemu has
2022-08-24 18:34:14 <bl4ckb0ne> never touching that shit again
2022-08-24 18:34:24 <psykose> it has like 0 vendored deps, but i think it wont build anyway
2022-08-24 18:34:35 <orbea> https://github.com/cemu-project/Cemu/tree/main/dependencies
2022-08-24 18:34:42 <orbea> cubeb is one, maybe zarchive too
2022-08-24 18:34:43 <psykose> bl4ckb0ne: did you like the 20 seconds to rebuild a project because each time you run scons it reconfigures all of it
2022-08-24 18:35:02 <bl4ckb0ne> even worst, there was some link time monstrosity that took 20 mins to link
2022-08-24 18:35:04 <bl4ckb0ne> nojoke
2022-08-24 18:35:09 <bl4ckb0ne> https://github.com/cemu-project/Cemu/blob/main/CMakeLists.txt#L61 curiously there isnt much deps
2022-08-24 18:35:24 <bl4ckb0ne> still dunno why it chokes on finding curl
2022-08-24 18:35:32 <bl4ckb0ne> FindCURL is there
2022-08-24 18:35:32 <psykose> curl-dev has no .cmake
2022-08-24 18:36:34 <orbea> fwiw cmake ships FindCURL.cmake in gentoo
2022-08-24 18:37:13 <bl4ckb0ne> > Wouldn't EGL_MESA_gl_texture_cubemap_image just define an interaction with EXT_EGL_image_storage that adds cube maps to the allowed targets for dma-bufs if that is its intent?
2022-08-24 18:37:17 <bl4ckb0ne> gah
2022-08-24 18:37:21 <bl4ckb0ne> >/usr/share/cmake/Modules/FindCURL.cmake is owned by cmake-3.24.1-r1
2022-08-24 18:38:16 <bl4ckb0ne> so its there
2022-08-24 18:38:48 <psykose> ah, cmake has it itself
2022-08-24 18:38:50 <psykose> weird then
2022-08-24 18:39:00 <bl4ckb0ne> aye since 3.12 iirc
2022-08-24 18:40:17 <bl4ckb0ne> >If CURL was built using the CMake buildsystem then it provides its own CURLConfig.cmake file for use with the find_package() command's config mode. This module looks for this file and, if found, returns its results with no further action.
2022-08-24 18:40:19 <bl4ckb0ne> gah
2022-08-24 18:43:30 <orbea> it uses vcpkg during cmake...
2022-08-24 18:43:34 <bl4ckb0ne> yes
2022-08-24 18:43:36 <bl4ckb0ne> ghastly
2022-08-24 18:57:45 <bl4ckb0ne> welp, dunno why cmake doesnt want to use the FindCURL it possess
2022-08-24 18:58:48 <psykose> just find_package(PkgConfig); pkg_check_modules(CURL curl)
2022-08-24 18:58:50 <psykose> call it a day
2022-08-24 18:59:06 <psykose> not pictured: perhaps some alias to add too
2022-08-24 18:59:26 <psykose> maybe it's libcurl as well
2022-08-24 19:00:08 <bl4ckb0ne> >--   Package 'curl', required by 'virtual:world', not found
2022-08-24 19:00:10 <bl4ckb0ne> gj cmake
2022-08-24 19:00:32 <psykose> yea it's libcurl
2022-08-24 19:00:59 <bl4ckb0ne> aye
2022-08-24 19:01:01 <bl4ckb0ne> imgui now
2022-08-24 19:01:08 <bl4ckb0ne> welp
2022-08-24 19:01:21 <psykose> mhm
2022-08-24 19:02:32 <bl4ckb0ne> no cemu on alpine then
2022-08-24 19:02:48 <psykose> weak!
2022-08-24 19:02:55 <psykose> does it even run
2022-08-24 19:03:02 <psykose> all i expect from the new emulators is 4 fps
2022-08-24 19:03:07 <psykose> i can't even emulate ps2 games
2022-08-24 19:03:20 <bl4ckb0ne> gonna wait for vcpkg to finish
2022-08-24 19:06:38 <bl4ckb0ne> chokes on vk sdk
2022-08-24 19:06:42 <psykose> mhm
2022-08-24 19:06:44 <bl4ckb0ne> apparently it does so on ubuntu and arch
2022-08-24 19:06:48 <psykose> yep
2022-08-24 19:06:56 <bl4ckb0ne> 
ACTION dunks cemu in the trash

2022-08-24 19:07:01 <bl4ckb0ne> time to buy a real wii u
2022-08-24 19:07:20 <psykose> expensive
2022-08-24 19:07:55 <bl4ckb0ne> found one for 70$
2022-08-24 19:08:15 <bl4ckb0ne> better than the time lost fighting cmake crap
2022-08-24 19:09:24 <psykose> expensive!
2022-08-24 19:10:49 <bl4ckb0ne> why does everything costs money
2022-08-24 19:10:51 <bl4ckb0ne> ughhhh
2022-08-24 19:12:34 <Newbyte> bl4ckb0ne what do you mean by "chokes on"?
2022-08-24 19:13:22 <bl4ckb0ne> >Could not find Vulkan SDK.  Before continuing, please download and install
2022-08-24 19:13:51 <Newbyte> psykose do you know if there's any better way to get all the debug symbols for Xwayland than building all the dependencies with $pkgname-dbg added?
2022-08-24 19:14:03 <Newbyte> e.g. libepoxy has no -dbg package
2022-08-24 19:14:14 <psykose> post the stacktrace with only xwayland-dbg
2022-08-24 19:14:18 <psykose> and musl-dbg
2022-08-24 19:14:35 <psykose> and whatever is available
2022-08-24 19:14:37 <psykose> then we go from there
2022-08-24 19:15:40 <bl4ckb0ne> i think i got it
2022-08-24 19:15:43 <bl4ckb0ne> boy is this dumb
2022-08-24 19:15:48 <bl4ckb0ne> VULKAN_SDK=/usr
2022-08-24 19:15:53 <bl4ckb0ne> and it finds vulkan.h
2022-08-24 19:15:57 <bl4ckb0ne> and its happy
2022-08-24 19:16:00 <bl4ckb0ne> ughhhhhhhhh
2022-08-24 19:16:05 <Newbyte> sounds fine
2022-08-24 19:17:27 <psykose> vulkant
2022-08-24 19:19:47 <Newbyte> psykose do you have any idea why it's still ?? after I installed libepoxy-dbg?
2022-08-24 19:20:15 <psykose> test !38141
2022-08-24 19:20:24 <psykose> the ?? is for the parts that don't have symbols
2022-08-24 19:20:29 <psykose> and/or the stack is fucked there anyway
2022-08-24 19:20:45 <psykose> if you see a corrupt stack 90% of the time it's this anyway
2022-08-24 19:20:48 <Newbyte> yes, I just thought I would have the symbols now
2022-08-24 19:20:55 <Newbyte> anyway, I'll try that once it builds
2022-08-24 19:21:16 <psykose> if the stack is destroyed you can have ?? anywhere
2022-08-24 19:24:31 <Newbyte> psykose same result
2022-08-24 19:24:36 <Newbyte> like, exact same
2022-08-24 19:24:37 <psykose> interesting
2022-08-24 19:25:29 <Newbyte> yes very interesting
2022-08-24 19:27:44 <Guest1050> is the alpine 3.16.2 x86_64 iso working?
2022-08-24 19:27:59 <psykose> yes
2022-08-24 19:28:06 <Guest1050> i just downloaded it + did a checksum, and there are a lot of messages of 'busybox <some stuff> not found'
2022-08-24 19:28:11 <Guest1050> using qemu
2022-08-24 19:28:12 <psykose> some libxcb dbg symbols could perhaps be useful
2022-08-24 19:28:12 <psykose> hmm
2022-08-24 19:28:25 <Guest1050> qemu-system-x86_64 -enable-kvm -cdrom alpine-standard-3.16.2-x86_64.iso
2022-08-24 19:28:41 <psykose> i assume you can build the revdeps with dbg yourself? or i can make an mr for it
2022-08-24 19:31:01 <Newbyte> psykose either way is fine by me
2022-08-24 19:31:45 <Newbyte> why would libxcb be useful btw?
2022-08-24 19:32:08 <Newbyte> actually it would probably be preferable if you could make an MR given that I'm about to build Linux
2022-08-24 19:32:21 <Newbyte> which will peg my CPU
2022-08-24 19:32:35 <bl4ckb0ne> isnt that what CPUs are made for
2022-08-24 19:32:55 <Newbyte> yeah but building two things is slow
2022-08-24 19:34:39 <psykose> sure, added a few things
2022-08-24 19:34:58 <psykose> libxcb is used by the x stack, dunno
2022-08-24 19:35:33 <psykose> might also want mesa-dbg around
2022-08-24 21:02:14 <Guest1050> does anyone know why the login screen says "(none) login: " instead of "<hostname> login: "? i also cannot login at all, even though there is /etc/hostname, /etc/passwd, /etc/group and /etc/shadow
2022-08-24 21:05:04 <ebb> The hostname isn't set - at a wild guess, it sounds like this might be a login prompt on the initramfs?
2022-08-24 21:06:04 <Guest1050> it's from an iso made with grub-mkrescue, so i guess so
2022-08-24 21:08:06 <ebb> ah, I just saw your messages above
2022-08-24 21:08:34 <ebb> but logging in as root doesn't work then?
2022-08-24 21:08:40 <Guest1050> nope
2022-08-24 21:11:05 <Guest1050> this is the script i'm using right now for building the iso: https://github.com/rdbo/sigma-linux/blob/master/build-minirootfs.sh
2022-08-24 21:12:00 <Guest1050> it's pretty similar to the original one, but with the simpledrm fix and with some extra desktop packages
2022-08-25 01:13:47 <Guest1050> does anyone know why grub's background_image won't work? it says "error: bitmap file /boot/wallpaper.png is of unsupported format." i already tried png, jpg, and xpm, on 1920x1080, 1280x1024 and 640x480.
2022-08-25 01:15:39 <Guest1050> grub.cfg: https://github.com/rdbo/sigma-linux/blob/master/profile/grub.cfg
2022-08-25 01:16:00 <psykose> did you try bmp
2022-08-25 01:16:26 <Guest1050> not yet
2022-08-25 01:18:21 <Guest1050> same error with bmp
2022-08-25 01:18:53 <ptrc> bit depth maybe?
2022-08-25 01:23:39 <Guest1050> just found someone's post trying to do the same thing as me
2022-08-25 01:23:42 <Guest1050> 'insmod png'
2022-08-25 01:24:14 <psykose> simply use limine with a bitmap image instead, so easy
2022-08-25 01:24:26 <psykose> justworks.bmp
2022-08-25 01:24:34 <psykose> you may have a weeb boot image
2022-08-25 01:25:30 <Guest1050> after turning on 'gfxterm' the font got all glitchy
2022-08-25 01:27:49 <ptrc> fwiw, this doesn't really have anything to do with development of alpine itself, so #alpine-linux (or even -offtopic) would be a better channel for such discussions :p
2022-08-25 01:29:20 <Guest1050> oh ok
2022-08-25 06:27:40 <ncopa> finally fixed xen problem
2022-08-25 06:31:38 <panekj> \o/
2022-08-25 07:52:29 <Ermine> Can I make abuild use clang++ instead of g++?
2022-08-25 07:53:01 <panekj> set CXX?
2022-08-25 07:55:28 <Ermine> Yeah, that worked, thank you~
2022-08-25 07:55:34 <Ermine> s/~/!/
2022-08-25 07:57:14 <panekj> no problem~
2022-08-25 16:40:35 <psykose> cute~
2022-08-25 18:15:02 <Ermine> clang~~
2022-08-25 19:55:09 <skarnet> where's Alpine at with proprietary / firmware-requiring wifi drivers? I have a broadcom chip I'd like to make work
2022-08-25 19:55:45 <panekj> proprietary stuff was dropped months ago
2022-08-25 19:55:48 <skarnet> do we have an up-to-date firmware list and officially support it, or do i need to take the screwdriver and hack things?
2022-08-25 19:55:59 <skarnet> I guess it's the latter then
2022-08-25 19:56:20 <panekj> iirc, if it's not in linux-firmware, it's gone
2022-08-25 19:56:57 <c7s> you can git log --oneline -- non-free/b43-firmware and build it yourself
2022-08-25 19:57:56 <panekj> git checkout f7fff70bb1f1a2b9756ec19318dfc4c9cc3c5f1c~1
2022-08-25 19:58:19 <ebb> bcm43xx seems to have stuff in linux-firmware upstream
2022-08-25 19:58:28 <panekj> algitbot: you forgot about ~1 >:(
2022-08-25 19:58:55 <skarnet> fair enough
2022-08-25 19:58:56 <skarnet> thanks
2022-08-25 20:00:30 <skarnet> $ grep license non-free/b43-firmware/APKBUILD
2022-08-25 20:00:39 <skarnet> license="propietary"
2022-08-25 20:00:46 <skarnet> yeah that was definitely well-maintained :P
2022-08-25 20:01:02 <panekj> it advocates for pies
2022-08-25 20:01:28 <skarnet> it advocates for pietaries, which isn't the same thing at all
2022-08-25 21:06:18 <omni> ncopa: could we get https://gitlab.alpinelinux.org/alpine/mkinitfs/-/merge_requests/106 ?
2022-08-25 21:08:49 <psykose> yeah i'm a little tired of seeing that everywhere
2022-08-25 21:09:07 <minimal> me three :-)
2022-08-25 23:47:07 <Hello71> skarnet: some broadcom wifis are supported by brcm*mac
2022-08-25 23:48:27 <skarnet> are supported by what now
2022-08-25 23:49:10 <psykose> linux-firmware-brcm
2022-08-25 23:49:17 <psykose> which one did you have
2022-08-26 00:05:04 <skarnet> a 4352, but don't waste any time on it, I got 'br43-phy0 unsupported' at boot time and decided it was a lost cause, that box is gonna choose between having Linux and having wifi until I order a better card.
2022-08-26 00:05:47 <skarnet> that's what happens sometimes when you try to refurbish old corporate PCs
2022-08-26 00:09:45 <psykose> aha, yeah that one is fucked
2022-08-26 00:09:57 <psykose> afaik if you find the firmware b43-fwcutter can extract it
2022-08-26 00:10:02 <psykose> then you copy it somewhere
2022-08-26 00:28:32 <Hello71> according to internet, it needs wl
2022-08-26 00:29:02 <Hello71> which translates to "*technically* you might be able to, but just buy a new card"
2022-08-26 01:43:01 <skarnet> exactly
2022-08-26 03:15:34 <psykose> sam_: is there any way to get pax-utils manpages now without needing xmlto
2022-08-26 03:15:54 <psykose> not sure how one ships dist-headers with meson even, but aside from that there is zero chance i am adding xmlto to it
2022-08-26 03:16:47 <psykose> (it's in bootstrap path and xmlto adds like another 5 dependencies)
2022-08-26 03:20:38 <sam_> psykose: sorry, I thought we'd added it into the meson dist stuff
2022-08-26 03:20:40 <sam_> I'll get that sorted
2022-08-26 03:20:54 <psykose> ah, that already exists
2022-08-26 03:20:59 <psykose> sure, have a look :)
2022-08-26 03:21:18 <psykose> or maybe i just held it wrong
2022-08-26 10:18:00 <ddevault> be advised that an incoming meson change will fuck up half of aports https://github.com/mesonbuild/meson/pull/10701
2022-08-26 10:20:58 <psykose> why would it
2022-08-26 10:21:14 <psykose> everything using abuild-meson does `meson setup ..`
2022-08-26 10:21:20 <psykose> i fixed that months ago
2022-08-26 10:21:31 <ptrc> and even if something doesn't, it's just a warning
2022-08-26 10:21:34 <psykose> there's a stray or two to catch
2022-08-26 11:05:08 <ddevault> there's a lot of packages that don't use abuild-meson
2022-08-26 11:05:12 <ddevault> good to know that it does the right thing though
2022-08-26 11:06:44 <psykose> strangely it's been a while since i've ran into one
2022-08-26 11:06:54 <psykose> i'd guess most are stale in testing
2022-08-26 11:07:00 <psykose> but yeah i was going to comb through the whole list
2022-08-26 11:07:26 <panekj> just delete everything not touched in months in testing :P
2022-08-26 11:09:53 <psykose> i always appreciate cleanups if you want to spend time on random things
2022-08-26 11:10:43 <psykose> granted, once i removed mono and a gitlab engineer emailed me about it, so that was funny, remember :p
2022-08-26 11:20:20 <skarnet> "how to find useful stuff: delete things until shit breaks, then the last thing you deleted was useful"
2022-08-26 11:21:22 <skarnet> (or: the trade secrets of the docker-slim algorithm)
2022-08-26 11:22:34 <psykose> docker slim is amazingly good
2022-08-26 11:22:57 <skarnet> I'm pretty sure Ariadne has Opinions about it
2022-08-26 11:23:07 <panekj> same
2022-08-26 11:23:28 <bl4ckb0ne> im sure i can do better in a single shell script
2022-08-26 11:27:29 <psykose> i can't
2022-08-26 12:54:06 <skarnet> wall of text dropped, time for (late) lunch
2022-08-26 13:21:04 <mercenar-> skarnet: I see a difference between 'minority of users' and 'minority of instances'. I for one would use busybox-less for anything not potato-powered or ephemeral container.
2022-08-26 13:21:51 <mercenar-> (and the wall stops short of providing an install choice between 'small' and 'complete')
2022-08-26 13:22:41 <skarnet> maybe? I haven't checked *everything* that's in busybox, but since you'd have to install an alternative for every feature before you can remove it, I'm not sure if it would be beneficial to you
2022-08-26 13:23:39 <skarnet> you'd have coreutils, util-linux, e2fsprogs (or something of the kind), findutils, GNU sed, mawk, bc, etc. etc.
2022-08-26 13:23:50 <mercenar-> It would be beneficial not to have a 'oh F, this is busybox-X, doesn't support --foo option' every once in a while
2022-08-26 13:24:02 <skarnet> oh, I definitely agree
2022-08-26 13:24:38 <skarnet> but I doubt that going all the way and having full-featured alternatives for *everything* just to ragedelete busybox will be that popular :P
2022-08-26 13:27:07 <mercenar-> Dunno. I have not run into anything yet in my common use cases where there was not already a better alternative already packaged
2022-08-26 13:30:21 <skarnet> would you call bash a 'better' alternative? because as it stands, it looks like the only shell that can currently be a drop-in for bb ash. :P
2022-08-26 13:34:01 <mercenar-> Even against all the hate for bash that I see, for interactive use it is still a hack of a lot better than ash :P
2022-08-26 13:35:21 <skarnet> maybe, but it comes at a cost. As everything does.
2022-08-26 13:46:30 <mercenar-> As long as we get to pick our own poison. I prefer the odd major problem over the 1000 papercuts, other's preferences may be different
2022-08-26 13:49:47 <orbea> mercenar-: and yash is a far better interactive shell than bash :)
2022-08-26 14:06:26 <mercenar-> orbea: I'd have to spend a couple of hours trying to tame it, out-of-the-box it looks annoying (trying to second-guess what I want to type, colours, messing with font weights) but on the positive side it has vi and emacs modes.
2022-08-26 14:07:22 <orbea> yea, it took me a little to get used to, but now its hard for me to go back :P
2022-08-26 14:17:01 <orbea> its kind of like the transition to tiling wms, first two weeks it was awkward, but now I can never go back.
2022-08-26 14:39:13 <panekj> sounds like fish
2022-08-26 14:43:47 <orbea> fish is probably a much bigger leap, yash scripts still largely work with any portable shell.
2022-08-26 14:44:30 <panekj> fish has $() now
2022-08-26 14:44:41 <orbea> ah :)
2022-08-26 14:45:11 <panekj> it still has breaking stuff but at least I can copy-paste commands from internet
2022-08-26 14:46:04 <panekj> can't assign vars like other shells, have to use `set` but `export` works as well as this `VAR=val program`
2022-08-26 14:47:09 <orbea> the biggest gotcha I can think of with yash from the top of my head is that variables set in for loops are implicitly local (i.e. "$i" is local in: for i in $foo; do ...)
2022-08-26 14:47:38 <panekj> that seems like a good behaviour (:
2022-08-26 14:47:58 <orbea> i like it too, but it can be unexpected when trying to fix prexisting scripts that depend on it :P
2022-08-26 14:48:18 <Ariadne> mercenar-: you can already use things that are not busybox in alpine, and /bin/sh is not intended to be interactive
2022-08-26 14:48:49 <Ariadne> (though cleaning up hardcoded dependencies is a good thing)
2022-08-26 14:49:00 <Ariadne> busybox in its current form is actually bad for containers
2022-08-26 14:49:41 <Ariadne> it has hundreds of things in it that make no sense for a container, some of which can be used to compromise the integrity of the container
2022-08-26 14:50:39 <Newbyte> like what?
2022-08-26 16:05:44 <ajhalili2006> Guess1277 please fix your client
2022-08-26 16:06:57 <panekj> ?
2022-08-26 16:38:05 <Ariadne> Newbyte: `chroot`, `mount`, ...
2022-08-26 16:38:31 <Ariadne> a container without chroot is a container that is a lot harder to escape :)
2022-08-26 16:43:32 <Hello71> only if you also mount everything W^X
2022-08-26 17:14:34 <anjan> hi all, I am trying to package gnu jami for alpine. I was told to install qt6-wayland. Im not sure where I can find this package. Is this a compile time flag I have to set for qt6?
2022-08-26 17:16:32 <panekj> https://pkgs.alpinelinux.org/package/edge/community/x86_64/qt6-qtwayland
2022-08-26 19:12:11 <anjan> thank you panekj 
2022-08-26 19:16:52 <anjan> Ok, now Im getting 
2022-08-26 19:16:54 <anjan> Failed to load vulkan: Cannot load library vulkan: (Error loading shared library vulkan: No such file or directory)
2022-08-26 19:16:56 <anjan> initInstance: No Vulkan library available
2022-08-26 19:17:00 <anjan> Failed to create platform Vulkan instance
2022-08-26 19:17:05 <anjan> qt.webenginecontext: 
2022-08-26 19:17:45 <panekj> did you add vulkan?
2022-08-26 19:18:09 <anjan> yes
2022-08-26 19:18:16 <anjan> via apk search vulkan
2022-08-26 19:18:22 <anjan> I installed the ones for my gpu
2022-08-26 19:18:43 <panekj> you might need to patchelf the binary
2022-08-26 19:19:11 <anjan> Not sure what that is panekj 
2022-08-26 19:19:43 <panekj> patchelf --add-needed libvulkan.so.1 [binary]
2022-08-26 19:21:29 <anjan> [anjan@navi /usr/bin]$ doas patchelf --add-needed libvulkan.so.1 jami-qt 
2022-08-26 19:21:32 <anjan> same error
2022-08-27 06:12:44 <raspbeguy> https://gitlab.alpinelinux.org/alpine/aports/-/merge_requests/38256
2022-08-27 06:15:37 <ptrc> raspbeguy: what about it?
2022-08-27 06:24:46 <PureTryOut> https://git.alpinelinux.org/aports/commit/main/mariadb-connector-c?id=db9bc616dc3c5951c1f78bc13d09120f76d99d43 made mariadb-connector-c-doc conflict with mariadb-doc, but I'm not sure which package the man files would belong more too. It seems mariadb-doc to me, but then why were they added to mariadb-connector-c-doc?
2022-08-27 06:30:51 <raspbeguy> ptrc: nothing, each time I create a merge request I link it here
2022-08-27 06:31:02 <ptrc> ah, okay
2022-08-27 07:37:57 <raspbeguy> Thanks psykose
2022-08-27 10:21:53 <Guest1309> is it possible to make mkinitfs use another kernel/modules? my kernel is 5.15, but i installed linux-edge (5.19) on a minirootfs and want to make a initramfs for it. it says "/lib/modules/5.15.59-0-lts does not exist or is not a directory"
2022-08-27 10:22:13 <ptrc> `mkinitfs 5.15.59-0-lts`
2022-08-27 10:22:33 <ptrc> ah, no
2022-08-27 10:23:09 <ptrc> anyway it's just `mkinitfs <version>`
2022-08-27 10:23:39 <ptrc> where version is the thing from uname -r or /lib/modules
2022-08-27 10:23:55 <Guest1309> oh, thanks
2022-08-27 11:00:39 <Guest1338> what does "/dev/loop0: Can't open blockdev" mean? if i run "losetup /dev/loop0" it says "/dev/loop0: 0 /media/cdrom/profile.sfs", and if i run "ls /media/root-ro" (which is the place the squashfs is mounted) it shows a proper root directory. is this even an error or should i ignore it?
2022-08-27 11:02:37 <psykose> sounds like an error reported by grub when ran inside a chroot
2022-08-27 11:04:11 <Guest1338> it happens in the 'initramfs' and it's the only error-looking thing i could find
2022-08-27 11:05:55 <Guest1338> if i exit the initramfs shell after this error, it starts the system but the login screen is "(none) login: " so I thought it might be related to it
2022-08-27 11:06:09 <psykose> merely installing a kernel would generate an initramfs for it via the mkinitfs hook
2022-08-27 11:06:14 <psykose> you normally don't have to do anything manually
2022-08-27 11:07:06 <psykose> if you think for some reason something else happened, post a full log of everything you did and all the output
2022-08-27 11:07:16 <Guest1338> yeah, but the default initfs does not seem to use squashfs and it also tries to install the apk overlay, which i'm trying to avoid
2022-08-27 11:07:36 <Guest1338> i want to go easy on ram, so i need to use the squashfs
2022-08-27 11:07:43 <psykose> then it sounds like it has nothing to with alpine-devel
2022-08-27 11:08:08 <ptrc> someday there will be "not a support channel" in the topic :')
2022-08-27 11:08:38 <psykose> i'm fine with support questions if they have anything to actually do with alpine and not 'i am doing a bunch of custom things, how can i make it work'
2022-08-27 11:09:02 <Guest1338> i don't understand, what is this channel for?
2022-08-27 11:09:10 <Guest1338> technically, isn't this developing alpine linux?
2022-08-27 11:09:10 <panekj> it's for alpine development
2022-08-27 11:09:13 <panekj> no
2022-08-27 11:09:29 <psykose> which part of it is developing alpine linux (in your opinion)
2022-08-27 11:09:51 <ptrc> if by "developing alpine linux" you mean stuff that gets upstreamed to alpine / aports for the general public, then yes
2022-08-27 11:10:09 <ptrc> but if you mean "using alpine linux for development", then not really
2022-08-27 11:11:09 <panekj> or "I'm doing wacky niche things on my Alpine install", then also not really
2022-08-27 11:11:22 <Guest1338> it's not for upstream alpine linux, so i will go to the other channel
2022-08-27 11:31:02 <skarnet> I'm doing wacky niche things on my Alpine install, such as patching aports
2022-08-27 11:35:42 <Ermine> you're also fixing device management, which is a super wacky niche thing
2022-08-27 11:38:26 <skarnet> the wackiness is out of control
2022-08-27 16:08:14 <ddevault> has anyone else had issues with qemu-binfmt on aarch64 hosts?
2022-08-27 16:08:44 <ddevault> for an x86_64 guest
2022-08-27 16:11:10 <ddevault> ah, it's missing entirely
2022-08-27 16:24:51 <minimal> ddevault: are you using the qemu-openrc package with its /etc/init.d/qemu-binfmt?
2022-08-27 16:24:59 <ddevault> yes
2022-08-27 16:25:02 <ddevault> I filed a ticket with qemu-openrc
2022-08-27 19:41:07 <ngorthone> Is there a way to see a package build log? Where alpine packages are built?
2022-08-27 19:46:11 <panekj> on pkgs.a.o each package has link to build log
2022-08-27 19:46:16 <Hello71> https://build.alpinelinux.org/buildlogs/
2022-08-27 19:47:04 <panekj> algitbot: 🤨
2022-08-27 19:49:16 <psykose> panekj: ping
2022-08-28 12:18:52 <ddevault> I have a non-PIE program which needs suid, how can I squelch the "Found non-PIE files that has SUID" error?
2022-08-28 12:19:47 <ddevault> reading abuild.in, looks like there's no way to circumvent this
2022-08-28 12:19:57 <ddevault> would a suid-pie option be disagreeable?
2022-08-28 12:20:29 <ddevault> err, suid-nopie, that is
2022-08-28 12:21:12 <panekj> options="!suid" ?
2022-08-28 12:21:23 <ddevault> it needs to be suid, panekj
2022-08-28 12:21:33 <ddevault> I could chmod a+s in the post-install but that leaves something to be desired
2022-08-28 12:22:07 <panekj> s/!suid/suid
2022-08-28 12:22:17 <ddevault> options="suid" is present
2022-08-28 12:22:25 <ddevault> see abuild.in:827 or thereabouts
2022-08-28 12:22:48 <ddevault> proposed solution: https://paste.sr.ht/~sircmpwn/20ac7a4faba225ece20c6fb38e44cb74117754af
2022-08-28 12:27:52 <ddevault> submitted via https://gitlab.alpinelinux.org/alpine/abuild/-/merge_requests/161
2022-08-28 15:04:22 <omni> gl.a.o doesn't auto assign MRs to the aport maintainer anymore, right?
2022-08-28 15:05:17 <omni> or maybe it does, but they need to have an account
2022-08-28 16:08:41 <Ariadne> ddevault: what blocks that suid program from being PIE?
2022-08-28 16:08:49 <ikke> omni: it automatically assigns it to the MR author in the webif. When that happens, it's not reassigned to the maintainer. 
2022-08-28 16:08:59 <panekj> Ariadne: it's written in Hare
2022-08-28 16:09:39 <panekj> > Making PIE executables is a lot of backend work and it's not an unambiguously good thing.
2022-08-28 16:09:42 <Ariadne> but lagomorphs are constantly changing position
2022-08-28 16:09:54 <ddevault> the qbe backend does not support PIE and, though we may add it at some point, it's a lot of work which is not prioritized
2022-08-28 16:10:12 <ddevault> PIE here is just defense in depth, turning it off does not introduce a vulnerability
2022-08-28 16:10:19 <ddevault> I think it can be disabled if due care and attention is given
2022-08-28 16:10:42 <Ariadne> i think just downgrade to warning instead of adding a new option tbh
2022-08-28 16:10:55 <ddevault> can you say as much on gitlab?
2022-08-28 16:11:00 <ddevault> for posterity
2022-08-28 16:11:01 <Ariadne> yep no problem
2022-08-28 20:52:36 <GreyDev> Just an FYI, I raised a bug report for libreswan-openrc (https://gitlab.alpinelinux.org/alpine/aports/-/issues/14149). It's a trivial bug that breaks the package, and is very easily fixed. I've corrected the issue on my machine, but a fix in aports is in order.
2022-08-28 21:03:17 <GreyDev> psykose: that was a rapid fix, thanks!
2022-08-28 21:03:21 <psykose> :)
2022-08-28 21:09:15 <omni> bah, it took several minutes!
2022-08-29 06:55:18 <omni> !35339
2022-08-29 07:38:54 <panekj> lol at the TSC issue
2022-08-29 07:48:21 <PureTryOut> lol?
2022-08-29 07:58:15 <panekj> PureTryOut: "One of your devs sits in front of computer too much, you should do something about it"
2022-08-29 07:58:58 <PureTryOut> that was not the intent of the issue ;
2022-08-29 07:59:00 <PureTryOut> ;)
2022-08-29 07:59:14 <PureTryOut> it's more that the rest of them aren't enough in front of their computer lol
2022-08-29 07:59:17 <panekj> I'm aware of the intent
2022-08-29 07:59:21 <PureTryOut> (not really, but you get the joke)
2022-08-29 08:00:23 <panekj> There is always that one hyperactive person that does a lot of work in certain context and that issue points out that everyone else should start doing the same
2022-08-29 08:01:42 <psykose> big jake volunteering i see
2022-08-29 08:02:18 <panekj> if I had more time to spare, sure, I would have rss feed of opened PRs in aports
2022-08-29 08:03:02 <psykose> that doesn't sound that fun, who would want to do that
2022-08-29 08:03:10 <panekj> but as everyone else here, I've got other shit to do :P so I only review PRs when someone complains here about "My PR wasn't merged in 5 minutes halp question mark"
2022-08-29 08:04:23 <psykose> >goes on holiday
2022-08-29 08:04:30 <psykose> i actually have i just used my phone at the airport instead
2022-08-29 08:04:35 <psykose> gitlab web mobile ui is terrible
2022-08-29 08:05:19 <PureTryOut> to be fair a lot, if not most, mobile UI's are terrible. It's just not a great device to do such things
2022-08-29 08:05:40 <panekj> > gitlab web ui is terrible
2022-08-29 08:05:43 <panekj> there
2022-08-29 08:21:49 <omni> other devs should be more competitive and try harder to be there first to merge things?
2022-08-29 08:21:59 <omni> gitlab badges could be earned!
2022-08-29 08:22:16 <panekj> we need leaderboards
2022-08-29 08:22:27 <panekj> and top merger of the month
2022-08-29 08:24:41 <omni> aren't the release notes a bit like that? (most of the time)
2022-08-29 08:24:42 <mercenary> and -10 score for merging anything that breaks things :p
2022-08-29 08:25:49 <panekj> omni: it only shows commits
2022-08-29 08:26:00 <panekj> we need an incentive for devs to compete
2022-08-29 08:26:31 <omni> ah, right, silly me
2022-08-29 08:27:37 <psykose> the main release notes don't have commits
2022-08-29 08:27:59 <psykose> the .1 one had a shortlog
2022-08-29 08:28:05 <psykose> (anymore)
2022-08-29 08:35:12 <psykose> panekj: you mean leaderboards like these https://git.alpinelinux.org/aports/stats/?period=q&ofs=10
2022-08-29 08:35:47 <panekj> yes, except those are commits
2022-08-29 08:36:34 <ncopa> how do we avoid the reviewers reduce the quality of the reviewing work in favor of quantity?
2022-08-29 08:37:12 <panekj> have dedicated group of contributors that are trained
2022-08-29 08:37:19 <ncopa> how do we avoid that devs fight about doing the easy/trivial ones and everybody avoids the timeconsuming reviews?
2022-08-29 08:37:25 <panekj> iirc, ptrc has some gitlab group that can manage issues
2022-08-29 08:37:43 <panekj> reviewer or something like that
2022-08-29 08:39:14 <ptrc> yup
2022-08-29 08:39:24 <panekj> need more of those people :P
2022-08-29 08:40:21 <psykose> at least 10 timeconsuming reviews a week, or death!
2022-08-29 08:44:28 <panekj> psykose: i don't care, just give me faster machine
2022-08-29 08:44:40 <psykose> urs are better than mine
2022-08-29 08:44:40 <panekj> s/psykose/psykose be like:
2022-08-29 08:44:49 <panekj> brain damage
2022-08-29 08:52:30 <ncopa> psykose: as you may have discovered, ffmpeg does not build on 32 bit arm. I suspect it is due to 145df12fb0f864f316fdedfb850e78e8fdcf3847
2022-08-29 08:52:56 <psykose> tis what i'm testing
2022-08-29 08:52:58 <ncopa> it didnt bump pkgrel so it was not caught by the builders
2022-08-29 08:53:06 <psykose> indeed
2022-08-29 08:53:10 <psykose> it would be strange, but possible
2022-08-29 08:54:26 <psykose> the specific errors are more indicative of binutils than anything, but there was already a rebuild with 2.39
2022-08-29 08:55:57 <ncopa> LTO was enabled in 34121b006851ed22138b748a9d3443123fbe3ef3
2022-08-29 08:56:14 <psykose> it was not
2022-08-29 08:56:17 <psykose> ah
2022-08-29 08:56:18 <psykose> yes
2022-08-29 08:56:23 <psykose> (misread the message)
2022-08-29 08:56:36 <omni> ncopa: I beleive we've mostly been making lighthearted jokes about the issue here in the chat, without the intent of making fun of Oliver and I think it's good that concerns like this are brought up and thought of
2022-08-29 08:58:30 <ncopa> psykose: I guess this is related: https://trac.ffmpeg.org/ticket/7191
2022-08-29 08:59:58 <psykose> nothing in that thread seems relevant to 32-bit arm and multi-partition lto
2022-08-29 09:01:03 <ncopa> its just an issue about --enable-lto being slow
2022-08-29 09:01:20 <psykose> yeah
2022-08-29 09:01:39 <psykose> funnily, `-flto` even prints a gcc notice usually
2022-08-29 09:02:00 <psykose> processing 50 ltrans jobs with 1 worker or whatever, see the documentation for details
2022-08-29 09:02:03 <psykose> (hinting to add a number)
2022-08-29 09:02:40 <psykose> you should cut an abuild/mkinitfs release
2022-08-29 09:04:27 <ncopa> mkinitfs has only a single commit since 3.6.2. we can backport that commit
2022-08-29 09:04:50 <ncopa> abuild otoh...
2022-08-29 09:04:58 <psykose> sure, functionally it's identical
2022-08-29 09:16:24 <ncopa> lto-wrapper: warning: using serial compilation of 44 LTRANS jobs
2022-08-29 09:16:24 <ncopa> lto-wrapper: note: see the '-flto' option documentation for more information
2022-08-29 09:17:31 <psykose> passed on one
2022-08-29 09:17:32 <psykose> interesting
2022-08-29 09:18:00 <ncopa> we want use -flto=auto right?
2022-08-29 09:18:07 <ncopa> both cflags and ldflags?
2022-08-29 09:18:17 <ncopa> or is it enough to add it to ldflags?
2022-08-29 09:19:15 <psykose> =auto is just a number taken from make (so passing =24 or whatever is mostly the same if that is =cores)
2022-08-29 09:19:28 <psykose> you need it in both but generally things pass both on the link line so it doesn't matter
2022-08-29 09:21:45 <psykose> annoyingly clang does not use the same options
2022-08-29 09:23:00 <psykose> clang has -flto[=full] (implied, 1 job always) or -flto=thin and then -flto-jobs=n or -Wl,--thinlto-jobs=n for lld or similar for other linkers
2022-08-29 09:23:13 <psykose> but i guess it makes sense because clang is not doing anything, it's all in llvm
2022-08-29 09:24:59 <psykose> yeah, even =4 fails the same way
2022-08-29 09:25:01 <psykose> that's interesting
2022-08-29 09:25:15 <psykose> i guess it's some binutils bug with partitioned lto stuff
2022-08-29 09:30:28 <ncopa> I suggest this: https://tpaste.us/0WrN
2022-08-29 09:30:59 <ncopa> >>> Size difference for ffmpeg-libs: 15 MiB -> 12 MiB
2022-08-29 09:31:10 <ncopa> I think that comes from -O3 -> -O2
2022-08-29 09:31:44 <ncopa> the linking is also fast
2022-08-29 09:32:08 <psykose> it's the same thing as before for lto, it would fail on arm32
2022-08-29 09:32:20 <ncopa> no
2022-08-29 09:32:27 <ncopa> i just built it on armhf
2022-08-29 09:32:32 <ncopa> sorry. armv7
2022-08-29 09:32:51 <psykose> the only other difference is O2 instead of O3
2022-08-29 09:33:25 <psykose> the lto is the same
2022-08-29 09:34:05 <ncopa> i don't know how or where the configure script uses  the CFLAGS. I change it where --enable-lto sets it
2022-08-29 09:35:40 <ncopa> I also drop the -fno-fat-lto-objects
2022-08-29 09:35:58 <psykose> if you change nothing at all, can you build it on 32-bit arm
2022-08-29 09:37:01 <ncopa> not with  CFLAGS="$CFLAGS -flto=auto -fno-fat-lto-objects"
2022-08-29 09:37:28 <psykose> alright
2022-08-29 09:37:47 <psykose> now if you add only --optflags="-O3 -flto=auto" with no other changes
2022-08-29 09:40:04 <ncopa> I suppose it will work, but I an not sure in which order configure scripts add the cflags with --enable-lto and --optflags
2022-08-29 09:41:10 <ncopa> we may end up with `-O3 -flto=auto -flto`  or `-flto -O3 -flto=auto`
2022-08-29 09:41:24 <ncopa> nly --optflags="-O3 -flto=auto" with no other changes works too
2022-08-29 09:41:33 <ncopa> s/nly/only/
2022-08-29 09:41:36 <psykose> quite interesting
2022-08-29 09:41:50 <psykose> it would make sense if them being passed wrong would break something, but why only 32-bit arm
2022-08-29 09:41:52 <psykose> in any case, yeah
2022-08-29 09:41:55 <psykose> the patch is nice
2022-08-29 09:42:06 <psykose> not the -O2 (it's a codec)
2022-08-29 09:42:11 <psykose> the other cleanups also good
2022-08-29 09:42:33 <ncopa> so you want drop the -O2 and use the default -O3?
2022-08-29 09:42:50 <psykose> you can drop the optflags line (it defaults to O3)
2022-08-29 09:46:03 <ncopa> i'd like to use -O2 for ffmpeg
2022-08-29 09:46:25 <ncopa> https://tpaste.us/a0gM
2022-08-29 09:46:32 <psykose> do you want to benchmark the difference
2022-08-29 09:46:47 <ncopa> ah.. i need update the commit message
2022-08-29 09:47:01 <ncopa> do you have any suggestion on how to benchmark it?
2022-08-29 09:48:01 <psykose> ffmpeg -benchmark_all <anything using ffmpeg>
2022-08-29 09:48:40 <psykose> -i something-usin-h264 -c:v libx265 out.hevc is one example
2022-08-29 10:00:50 <psykose> i am not sure which ones really matter, as for all of those the actual encoding is not even in ffmpeg (libx265 for instance)
2022-08-29 10:11:15 <ncopa> decoding: https://tpaste.us/NO6m
2022-08-29 10:11:49 <ncopa> I tested: time ffmpeg -i ~/Downloads/bbb_sunflower_1080p_60fps_normal.mp4 -f null -
2022-08-29 10:12:28 <ncopa> file is from http://bbb3d.renderfarming.net/download.html
2022-08-29 10:20:41 <ncopa> but then again, it is probably only using some decoder lib and not ffmpeg libs
2022-08-29 10:21:19 <psykose> yeah
2022-08-29 10:30:47 <ncopa> I'm now scaling the video with: time ffmpeg -i ~/Downloads/bbb_sunflower_1080p_60fps_normal.mp4 -filter:v scale=720:-1 -c:a copy  -f null -
2022-08-29 10:33:14 <ncopa> https://tpaste.us/x5b1
2022-08-29 10:33:25 <ncopa> seems like -O3 is slower on armv7
2022-08-29 10:33:31 <ncopa> slightly slower
2022-08-29 10:33:55 <ncopa> but that might be that the arm machine is busy with CI or similar
2022-08-29 10:34:07 <ncopa> on my i9 desktop there was no significant change at all
2022-08-29 10:39:16 <ncopa> im gonna push -O2. It save a few megs in space and there are no evidence that -O3 makes it any faster.
2022-08-29 11:44:31 <omni> the good luck web ui sometimes updates things dynamagically and sometimes not...
2022-08-29 12:34:00 <omni> excuse my hypertextsensitivity
2022-08-29 13:01:46 <Ariadne> i don't know why that TSC issue was opened, but this is not an appropriate use of the the TSC.
2022-08-29 13:04:09 <panekj> maybe they are unaware of council repo
2022-08-29 13:05:23 <Ariadne> as far as i am concerned, that issue is just concern trolling
2022-08-29 13:07:00 <Ariadne> if they would like to propose a concrete solution other than “wah there is somebody who is eager to help get things in” then they can try again
2022-08-29 13:07:28 <PureTryOut> Although I wasn't the one submitting it, I can ensure you it's not meant as a troll at all and a serious concern. Although it's fine to close the issue as non-valid of whatever, I feel you could answer with a bit more than just a plain "no" which honestly sounds a bit rude
2022-08-29 13:08:13 <PureTryOut> panekj: I personally was unaware of a council repo. Would it be a more relevant place for such an issue?
2022-08-29 13:08:40 <panekj> IMO, it's short-sighted and selfish since it only cares about aports/+MRs while not overall health of Alpine Linux project
2022-08-29 13:08:46 <Ariadne> opening an issue to complain about a developer is not acceptable
2022-08-29 13:09:18 <PureTryOut> He wasn't complaining about a single developer? If anything he was complimenting a single developer
2022-08-29 13:09:22 <Ariadne> the only thing your project leader should be doing is publishing an apology
2022-08-29 13:09:27 <panekj> PureTryOut: I believe it is more appropriate than TSC
2022-08-29 13:09:48 <PureTryOut> Wow, I find this quite harsh of a response to be honest. The issue was opened with the best of intentions and not intended to blame or fight anyone
2022-08-29 13:10:13 <PureTryOut> panekj: ok thanks will do that when something comparable comes up (hopefully never)
2022-08-29 13:10:19 <panekj> it can be understood in two different ways
2022-08-29 13:10:35 <panekj> - that one developer doing exceptionally great
2022-08-29 13:10:41 <panekj> - every other dev not doing enough
2022-08-29 13:11:34 <panekj> which something I'm trying to get at, since the issue only cares about aports MRs, while other devs handle lots of work in other sections of alpine project 
2022-08-29 13:11:41 <Ariadne> PureTryOut: a project of volunteers (alpine) does not exist to meet performance metrics set by your project leader
2022-08-29 13:13:00 <Ariadne> and — to be clear — if your project continues with this behavior — then i will boycott any possibility of future collaboration in personal capacity
2022-08-29 13:14:47 <PureTryOut> that's one way of looking at it panekj but if there is a serious concern like that, should it not be mentioned because of being afraid to hurt someone? I agree that it only targets aports but that's the only place where this concern is relevant currently. Other repositories do not seem to be so "reliant" on a single person, so it's not worth mentioning. Imo anyway
2022-08-29 13:15:18 <PureTryOut> But seriously, this was not meant as an attack or blame to anyone and no one has to be so defensive about it. It's fine saying "we don't feel this is an actual issue so we're closing this", but this response is way too harsh
2022-08-29 13:15:33 <Ariadne> it’s not harsh enough apparently
2022-08-29 13:15:36 <panekj> but why do you think that if psykose went missing, other devs would not step up?
2022-08-29 13:16:24 <PureTryOut> also ollieparanoid isn't the project leader of pmOS, he is just part of the team like some other people including myself are. He was only sharing a concern that some people, including myself, had. It's fine to not agree with it and I'm in fact glad to hear you all don't share it
2022-08-29 13:16:44 <panekj> we have jirutka, mpolanski, ikke, and few other who (while not to such extent like current top merger) are active in aports and tend to project
2022-08-29 13:17:25 <panekj> we also have few non-devs who tend to aports as well
2022-08-29 13:17:52 <PureTryOut> panekj: I don't think _nothing_ would get merged anymore, but the pace would slow considerably. We've seen it before with Leo. Before Leo came it could take months sometimes to get things merged, and after they left the time increased noticeably again
2022-08-29 13:18:20 <PureTryOut> But I'm glad you seem to think otherwise
2022-08-29 13:18:26 <Ariadne> good.  maybe aports moves too fast then
2022-08-29 13:18:29 <panekj> maybe, but there were also problems with Leo
2022-08-29 13:18:33 <Ariadne> psykose stop approving shit
2022-08-29 13:18:44 <PureTryOut> Good 👍️
2022-08-29 13:19:02 <panekj> ports were moving so fast there were multiple mistakes happening
2022-08-29 13:19:05 <Ariadne> problem solved
2022-08-29 13:22:03 <PureTryOut> Like I said it's fine to disagree but the issue was opened with the best of intentions and not meant to attack anyone and while Ariadne you think the issue was insulting, to others your response was insulting. There is no need to be so defensive and such a response does nothing but stir up a fight
2022-08-29 13:22:17 <panekj> PureTryOut: obviously it's super nice when you get MRs merged in a single day or few, but I don't think it's ok to expect that as a given
2022-08-29 13:22:45 <Ariadne> the lack of efficiency in volunteer labor may result in occasional inconvenience for pmOS, as it does for others, but other people do not open an issue like that
2022-08-29 13:23:52 <PureTryOut> panekj: I suppose there is a difference between a few days and a few months. Again, if you guys think there is no issue then there is probably none. I'm glad you at least try to have a reasonable discussion about it, thanks for that
2022-08-29 13:24:32 <Ariadne> as far as i am concerned you’re the one that is playing defense, i am just sitting here saying that demanding an SLA from volunteers is ridiculous
2022-08-29 13:24:43 <Ariadne> and it is ridiculous
2022-08-29 13:25:44 <psykose> doesn't PureTryOut have merge privileges
2022-08-29 13:25:47 <psykose> (hi! it's me)
2022-08-29 13:25:55 <panekj> hi me, it's you
2022-08-29 13:26:08 <PureTryOut> psykose: I do, and I'm probably part of the problem that the issue was about
2022-08-29 13:26:22 <Ariadne> and frankly if preexisting distributions aren’t good enough for pmOS then they should build their own distribution from the ground up rather than complain
2022-08-29 13:26:45 <PureTryOut> Ariadne: then let's agree to disagree. The issue was not meant to insult or attack anyone and let's leave it at that
2022-08-29 13:26:58 <psykose> no, what i mean is like, what is the theoretical place where 'i dissappear' and things 'take months to merge'
2022-08-29 13:27:01 <psykose> you.. can do it
2022-08-29 13:27:08 <psykose> i don't do anything special
2022-08-29 13:27:18 <Ariadne> but no it’s impossible for pmOS to find other sponsors
2022-08-29 13:27:26 <Ariadne> that’s why there is an issue opened
2022-08-29 13:28:02 <psykose> it just so happens i don't mind the volunteering and happen to do the bulk of it usually by just being there, but my absence doesn't block anything
2022-08-29 13:28:26 <Ariadne> stop volunteering ^_^
2022-08-29 13:28:45 <Ariadne> let’s get that merge rate to 0 per day
2022-08-29 13:29:12 <psykose> the person doing the 'merging' is not where most of the work is to begin with
2022-08-29 13:29:13 <panekj> disable MRs, bring back mailinglistbot and set it to automerge
2022-08-29 13:29:20 <psykose> people making nontrivial changes is where the bulk of it is
2022-08-29 13:29:49 <psykose> i do a lot of that too; that would be gone. but it's not related to mere 'merge of mrs'
2022-08-29 13:30:24 <psykose> the place where things "slowed down after leo dissappeared" is the place where you had the ability to merge things yourself, with some time investment. i'm not sure why you didn't
2022-08-29 13:30:31 <psykose> it's just a regular volunteer project
2022-08-29 13:30:50 <Hello71> Ariadne: i think you're reading things into the message which were not said. for example, they did not "[demand] an SLA". they also did not complain about the previous merge time. i think another way to say it is "good work, how can we keep it up", which is a completely different message to "good work, make sure it doesn't get worse"
2022-08-29 13:33:02 <PureTryOut> psykose: lack of time and things, the usual that everyone has. I'm blaming myself but I'm not blaming anyone else in particular or at all. The issue didn't say "things _will_ slow down" but more "what would need to happen if it did". Like Hello71 said it should be read mainly as you doing great work
2022-08-29 13:33:40 <panekj> I read that as "how can we keep it up" but we just can't force people to review MRs and merge stuff
2022-08-29 13:34:30 <panekj> as ncopa said earlier, how do we make sure that people who review the stuff don't introduce more workload than if we had no one reviewing it, and how do we not work on same things at the same time when not needed
2022-08-29 13:36:23 <ollieparanoid> hey... this TSC issue was not about making sure my/pmOS MRs get merged ASAP. I wrote this because I was concerned of burn out. I read the situation wrong and I'm sorry, shouldn't have opened this. I think Alpine is great and you all know what you are doing. Sorry for the mess.
2022-08-29 13:37:10 <ddevault> ollieparanoid: you've done nothing wrong, it was a good issue and a valid problem to consider and Ariadne's response is entirely unwarranted
2022-08-29 13:37:28 <ddevault> this entire discussion has been extremely unprofessional and not conducive to a good collaborative project environment
2022-08-29 13:37:42 <ddevault> I think a step back to reconsider and regroup is warranted before reapproaching this discussion in good faith
2022-08-29 13:37:53 <ddevault> this is ridiculous
2022-08-29 13:38:20 <panekj> I think it's an invalid issue, instead of going full open one could ask on IRC if psykose herfself is fine with the pace of work and if there isn't any issue
2022-08-29 13:38:45 <ddevault> it is not reasonable to perceive this issue as any kind of slight on psykose
2022-08-29 13:38:54 <psykose> i don't view it as one
2022-08-29 13:38:55 <ddevault> bus factor is an entirely valid concern that we can and should plan around
2022-08-29 13:39:08 <bl4ckb0ne> how is it hanging psykose 
2022-08-29 13:39:18 <psykose> terrible, wbu
2022-08-29 13:39:20 <panekj> bus factor is only a concern if you don't have people that can do the work
2022-08-29 13:39:24 <bl4ckb0ne> 
ACTION hugs psykose

2022-08-29 13:39:25 <psykose> saw khronos fixed the sdk tags
2022-08-29 13:39:26 <psykose> that was fun
2022-08-29 13:39:32 <psykose> did you beat them for it
2022-08-29 13:39:38 <bl4ckb0ne> FUCKING FINALLY
2022-08-29 13:39:41 <psykose> HELL YEAH
2022-08-29 13:39:42 <bl4ckb0ne> did not
2022-08-29 13:39:44 <Ariadne> i did not perceive it that way, but do see it as a request that alpine move at the speed an external project wants
2022-08-29 13:39:48 <bl4ckb0ne> 
ACTION high fives psykose

2022-08-29 13:39:53 <Ariadne> and i don’t think that is great sorry
2022-08-29 13:39:56 <psykose> 
ACTION high fives bl4ckb0ne

2022-08-29 13:39:57 <bl4ckb0ne> i actually forgot to bully them 
2022-08-29 13:40:06 <psykose> maybe that's for the best
2022-08-29 13:40:10 <psykose> no nighttime guilt now
2022-08-29 13:40:11 <panekj> 
ACTION bullies bl4ckb0ne to not forget again

2022-08-29 13:40:12 <ddevault> there are no internal and external projects, there's just the FOSS ecosystem as a whole
2022-08-29 13:40:19 <ddevault> what's good for pmOS is good for alpine
2022-08-29 13:40:31 <ddevault> and moving at an adequate pace to meet alpine's goals is a legitimate concern
2022-08-29 13:40:35 <Ariadne> if pmOS wants development to move at a specific speed they should participate in reviews
2022-08-29 13:40:38 <ddevault> your response was out of line, Ariadne
2022-08-29 13:40:45 <bl4ckb0ne> i wont forget next time panekj uwu
2022-08-29 13:40:49 <panekj> :3
2022-08-29 13:41:11 <panekj> > what's good for pmOS is good for alpine
2022-08-29 13:41:12 <panekj> what
2022-08-29 13:41:13 <Ariadne> there are pmOS contributors with commit rights in alpine
2022-08-29 13:41:24 <Ariadne> if they are concerned for burnout why don’t they step up
2022-08-29 13:41:42 <ddevault> you're not arguing in good faith
2022-08-29 13:41:48 <Ariadne> i am
2022-08-29 13:41:56 <ddevault> you might be right that pmOS can "step up" and help with reviews
2022-08-29 13:42:11 <ddevault> but you need to be arguing from the good faith perspective that any lack of pace is not their fault, or anyone's fault in particular
2022-08-29 13:42:11 <Ariadne> i have advocated for several of them to become committers — so they can autonomously do their work
2022-08-29 13:42:22 <ddevault> and interpret them not as demands from outsiders but suggestions for improvement from within
2022-08-29 13:42:30 <Ariadne> if we need more committers from their side, then we should talk about that
2022-08-29 13:42:40 <ddevault> yes, and this issue is a forum for that kind of discussion
2022-08-29 13:42:56 <Ariadne> then they should say “we need commit rights for more of our contributors”
2022-08-29 13:43:09 <ddevault> you're still presenting this as an us vs them
2022-08-29 13:43:11 <ddevault> and I won't accept that
2022-08-29 13:43:22 <ddevault> I think you need to remove yourself from this discussion and allow it to continue with participants arguing from a position of good faith
2022-08-29 13:44:42 <PureTryOut> it's new to me that you'd like more pmOS contributors to become committers. Who would you like in particular?
2022-08-29 13:44:56 <Ariadne> new to you?  i have wanted this for years
2022-08-29 13:45:01 <PureTryOut> Yes it's new to me
2022-08-29 13:45:47 <ddevault> if vague wantiness expressed on IRC and random issue threads hasn't solved the problem "for years", a gitlab ticket for the TSC could very well be a good place to address the question once and for all
2022-08-29 13:46:02 <Ariadne> i would like to see anyone trusted by the pmOS team be fast tracked for commit rights
2022-08-29 13:46:32 <Ariadne> ddevault: sure, but the current ticket is defective, it does not propose any solution
2022-08-29 13:46:46 <ddevault> the requirement for a good ticket is to present a problem
2022-08-29 13:46:51 <ddevault> the requirement for a good merge request is to present a solution
2022-08-29 13:47:13 <Ariadne> the problem presented by the ticket is also defective
2022-08-29 13:47:36 <ddevault> you and I can discuss this further in private later
2022-08-29 13:47:45 <ddevault> for the time being I am asking you to reverse your course and step away from the discussion
2022-08-29 13:47:53 <ddevault> let another TSC member close the ticket should they feel similarly
2022-08-29 13:48:19 <Ariadne> any TSC member can reopen the ticket if they wish to do so
2022-08-29 13:48:28 <ddevault> I am asking /you/ to do it
2022-08-29 13:49:07 <PureTryOut> Ariadne: good to know. I'll bring it up in the next team meeting and see who could use the rights appropriately, I know of 2 in particular that could probably use them well
2022-08-29 13:49:12 <Ariadne> i will not until the ticket is revised to be something other than a complaint about perceived bus factor
2022-08-29 13:49:22 <psykose> how does one become a tsc member, asking for a friend from another dimension
2022-08-29 13:49:38 <panekj> how does one become mentor, asking for new friends
2022-08-29 13:49:41 <ddevault> that's not what it is
2022-08-29 13:53:54 <bl4ckb0ne> psykose: did you made the changes for the khronos tags stuff
2022-08-29 13:54:02 <psykose> yeah
2022-08-29 13:54:27 <psykose> no actual real changes so just went with it
2022-08-29 13:54:36 <bl4ckb0ne> ah you merged directly
2022-08-29 13:54:40 <psykose> the .1 bumped cmake minimum
2022-08-29 13:54:51 <Ariadne> PureTryOut, ollieparanoid: great.  would you like to use the preexisting ticket to discuss this further?
2022-08-29 13:54:59 <ikke> FYI, when Leo dropped, I stepped in until psykose took over (and allowed me to focus on infra) 
2022-08-29 13:55:00 <psykose> or nothing (same commit)
2022-08-29 13:55:08 <bl4ckb0ne> nice
2022-08-29 13:55:17 <bl4ckb0ne> let me know next time if you want me to take care of the khronos bump
2022-08-29 13:55:32 <psykose> sure thing!
2022-08-29 13:55:36 <psykose> they've been pretty simple
2022-08-29 13:56:10 <psykose> only one thing needed adjusting and it was just an upgrade for libplacebo
2022-08-29 13:56:25 <psykose> ikke: you do a real good job at that
2022-08-29 13:56:45 <bl4ckb0ne> > (1.3.224-r0 -> 1.3.224.1-r1)
2022-08-29 13:56:50 <psykose> hah
2022-08-29 13:56:55 <bl4ckb0ne> shouldnt it be 1.3.244.1-r0 ?
2022-08-29 13:57:00 <psykose> which package
2022-08-29 13:57:07 <Ariadne> probably but it doesn’t matter so much
2022-08-29 13:57:08 <bl4ckb0ne> spirv-*
2022-08-29 13:57:15 <bl4ckb0ne> but yeah prolly doesnt matter
2022-08-29 13:57:25 <psykose> for tools is also added lto after
2022-08-29 13:57:25 <bl4ckb0ne> also the vulkan-*
2022-08-29 13:58:11 <PureTryOut> Ariadne: uh sure I suppose, not sure what ollieparanoid  would want there. Since I learned off the council repo that might be more appropriate though?
2022-08-29 13:58:25 <Ariadne> the TSC does approve new committers
2022-08-29 14:01:36 <ollieparanoid> Ariadne: it's fine with me to keep the issue closed, we can open a new one when we have a candidate.
2022-08-29 14:01:51 <Ariadne> anyway, sorry for my harshness on this
2022-08-29 14:14:28 <Ariadne> 8:49 AM <psykose> how does one become a tsc member, asking for a friend from another dimension
2022-08-29 14:14:48 <Ariadne> i don’t know but i plan to step down from tsc after 3.17 release, so i am sure we will all find out together
2022-08-29 15:36:58 <skarnet> wow, I missed the drama
2022-08-29 15:37:22 <skarnet> the discussion about how to become a TSC member is also interesting to me. Asking for a friend.
2022-08-29 15:38:09 <skarnet> on the initial issue: maybe ollieparanoid is more used to the way that corporations work, or big distros doing things by committee such as Debian
2022-08-29 15:38:24 <skarnet> but in the world of "alternative" free software, smaller projects with more dedicated teams...
2022-08-29 15:38:47 <skarnet> bus factor is an overwhelming and ubiquitous issue, it's there, always, everywhere
2022-08-29 15:39:06 <skarnet> and being surprised at the fact that most of the heavy lifting is done by one single person at a time...
2022-08-29 15:39:12 <skarnet> just shows inexperience honestly
2022-08-29 15:39:50 <skarnet> good FOSS is running on coffee, luck, and the good will of a few people, rarely more than one at a time
2022-08-29 15:40:22 <skarnet> just like the Internet is running on queer furries - it's a meme, but when you really get down to it, you realize it's much closer to the truth than you would have thought
2022-08-29 15:41:14 <skarnet> you don't have to like it, but it's how it is, it's how it works, best effort everywhere, no guarantees
2022-08-29 15:41:31 <skarnet> and people who make it work and volunteer and contribute don't get enough credit
2022-08-29 15:42:13 <ddevault> there is enough effort going into alpine that we can probably muster up some additional contributors
2022-08-29 15:42:19 <Newbyte> skarnet: what is your point?
2022-08-29 15:42:22 <ddevault> at the very least we should be prepared to have such a discussion in good faith
2022-08-29 15:42:36 <ddevault> professionalism is, or at least should be, a goal of alpine
2022-08-29 15:43:24 <skarnet> Newbyte: my point is that making a surprised pikachu face at the fact that there's one person doing most of the merges is not constructive
2022-08-29 15:43:33 <skarnet> and pretty noobish if I'm being honest
2022-08-29 15:43:56 <ddevault> I really do not appreciate these bad faith paraphrasings of oliver's point
2022-08-29 15:44:01 <skarnet> ddevault: I agree, there's a real discussion to have about bus factor
2022-08-29 15:44:21 <skarnet> oh it's not bad faith, it's my honest-to-God reaction when reading the issue
2022-08-29 15:44:37 <skarnet> and if I react this way, I can understand why other people react this way too
2022-08-29 15:45:12 <skarnet> the question of how to incite more contributors to be active participants in the process of getting stuff done with Alpine is not new
2022-08-29 15:45:19 <skarnet> it's been around for as long as I've been here
2022-08-29 15:45:25 <skarnet> yes, it's a legit question
2022-08-29 15:45:31 <skarnet> lots of people are asking it
2022-08-29 15:45:47 <skarnet> have we made progress towards an answer?
2022-08-29 15:45:50 <skarnet> stuff has been tried
2022-08-29 15:45:58 <skarnet> with mixed success
2022-08-29 15:46:26 <ddevault> an attempt to address this issue with documented minutes and planning seems like a pretty good move
2022-08-29 15:46:37 <Newbyte> Ariadne: for what it's worth, much like PureTryOut I had no idea that you were interested in seeing one of us "step up". Has this been communicated somewhere?
2022-08-29 15:46:41 <ddevault> and this is within the TSC's purview
2022-08-29 15:47:19 <skarnet> ddevault: I agree, this is a legit TSC discussion to have
2022-08-29 15:48:14 <skarnet> look at issue 25
2022-08-29 15:48:33 <skarnet> look at issue 11
2022-08-29 15:48:40 <skarnet> they're still open
2022-08-29 15:49:34 <skarnet> what are your bets on an issue specifically aimed at solving bus factor making faster progress than these?
2022-08-29 15:50:02 <ddevault> ¯\_(ツ)_/¯
2022-08-29 15:50:13 <ddevault> if nothing else it suggests emphasizing a re-prioritization of these concerns
2022-08-29 15:50:25 <ddevault> and good faith issues should be met with a good faith response, even if it's "dupe of whatever"
2022-08-29 16:09:37 <Ariadne> Newbyte: i made it fairly clear to ollieparanoid a few years ago, that i would be happy to do whatever i can to onboard pmOS contributors.  i also noted last year during alpineconf that we count pmOS contribution experience as being equivalent to alpine contribution experience
2022-08-29 16:10:41 <Newbyte> I see, thanks for clarifying.
2022-08-29 16:11:47 <Ariadne> perhaps this TSC issue can be used to officially communicate this policy
2022-08-29 16:49:48 <Newbyte> Why was https://gitlab.alpinelinux.org/alpine/council/-/issues/15 removed?
2022-08-29 16:50:09 <ddevault> it was made confidential
2022-08-29 16:50:16 <Newbyte> Oh, I see
2022-08-29 17:05:38 <Ariadne> Newbyte: are there any pmOS contributors who would like alpine commit privileges?
2022-08-29 17:06:57 <skarnet> can I apply without being a pmOS contributor?
2022-08-29 17:07:48 <Ariadne> sure, i would be willing to sponsor that proposal, too :)
2022-08-29 17:08:43 <skarnet> I promise to be good and not merge things requiring deep system changes
2022-08-29 17:08:57 <skarnet> (not before 5 months have passed without any input from core devs)
2022-08-29 17:09:11 <Newbyte> Ariadne: I think the people currently in the GNOME team (so, me and Pablo) would be interested at least
2022-08-29 17:09:41 <Newbyte> but I can forward it to our team chat if it hasn't been brought up already
2022-08-29 17:09:42 <Ariadne> Newbyte: what is his gitlab username?
2022-08-29 17:09:54 <Newbyte> Ariadne: pabloyoyoista
2022-08-29 17:10:00 <panekj> can I also get permissions if everyone else is getting them?
2022-08-29 17:10:04 <Newbyte> he's on the Alpine GNOME team already
2022-08-29 17:10:24 <panekj> I promise I won't push systemd port
2022-08-29 17:10:28 <Newbyte> he is away/busy until September though
2022-08-29 17:10:42 <Newbyte> panekj: I was about to make a joke about that
2022-08-29 17:11:17 <Ariadne> i mean, if there are situations where having alpine commit privileges would help to unblock people, that is a conversation worth having
2022-08-29 17:11:35 <Ariadne> but i like to focus on tactical changes which have direct impact
2022-08-29 17:12:05 <panekj> I mean, I don't see anything wrong with current setup, we just need more (wo)manpower to sift through all the ancient shit that accumulated in issues/MRs backlog
2022-08-29 17:12:22 <skarnet> there's always tension there: it's good to speed up things by unblocking people who need stuff done, but it's also good to make sure the overall project keeps its vision and doesn't become chaos central
2022-08-29 17:12:48 <Newbyte> so far in the pmOS team room my name and Pablo have been brought up, and it's also been added to our meeting agenda
2022-08-29 17:12:48 <panekj> merging/pushing everything ASAP is not the way
2022-08-29 17:13:05 <skarnet> panekj: +1, the problem is that competent and dedicated manpower is hard to find
2022-08-29 17:13:28 <Ariadne> i plan to step back from the alpine TSC after 3.17 release
2022-08-29 17:14:13 <Ariadne> (that includes stepping down as its security officer)
2022-08-29 17:14:21 <skarnet> Ariadne: did you miss the part where we need *more* manpower, not less?
2022-08-29 17:14:54 <Ariadne> skarnet: yes, unfortunately i have other mountains of technical debt which need my attention more
2022-08-29 17:15:14 <skarnet> yeah :/
2022-08-29 17:15:21 <Ariadne> i believe that people who sit on alpine's TSC should be actively involved in alpine
2022-08-29 17:15:55 <Newbyte> Ariadne: like hopalong?
2022-08-29 17:15:59 <panekj> > (that includes stepping down as its security officer)
2022-08-29 17:16:00 <panekj> sad
2022-08-29 17:16:32 <Ariadne> Newbyte: no, but i probably should cut a new release
2022-08-29 17:16:37 <Newbyte> :(
2022-08-29 17:23:19 <ncopa> Newbyte: I made https://gitlab.alpinelinux.org/alpine/council/-/issues/15 confidential as it is a case regarding an individual. I think it is best handled in private
2022-08-29 17:26:00 <ncopa> I also spoke with Ariadne about it, and she acknowledges that it is wasn't a good behavior and she expressed regrets
2022-08-29 17:26:34 <ncopa> also: 16:01 <Ariadne> anyway, sorry for my harshness on this
2022-08-29 17:27:02 <Ariadne> also note that i am trying to actively solve pmOS's problem
2022-08-29 17:27:27 <ncopa> that is absolutely noted
2022-08-29 17:28:16 <Ariadne> panekj: not sad.  i have other projects that require a significant commitment, and will likely strengthen the overall apk ecosystem.
2022-08-29 17:29:09 <Ariadne> and besides, if i have anything to say about it, whoever replaces me will probably do an even better job :)
2022-08-29 17:29:38 <ncopa> And for the record, the behavior is not what we (Alpine project) stand for. We should be better than this.
2022-08-29 17:29:42 <ncopa> And I think we are
2022-08-29 17:29:45 <Ariadne> +1
2022-08-29 17:29:58 <Ariadne> that was definitely not my finest moment
2022-08-29 17:30:58 <skarnet> Ariadne: it's *difficult* to find people for a security officer position that would do a better job
2022-08-29 17:31:32 <ncopa> I'd say that is an understatement 
2022-08-29 17:31:45 <Ariadne> anyway, my intention to step down from TSC is not new, nor related to any of this
2022-08-29 17:58:58 <ncopa> ollieparanoid: I have reopened https://gitlab.alpinelinux.org/alpine/tsc/-/issues/57 I think it is a valid concern
2022-08-29 18:01:13 <ncopa> I'm not sure it is a technical problem (so not sure it is a problem for TSC), but I think it may be good to discuss it
2022-08-29 18:02:06 <ncopa> ollieparanoid: I would also like to make it clear that you did nothing wrong in opening the ticket
2022-08-29 18:03:10 <PureTryOut> ncopa: thanks for that statement. ollieparanoid and I both agreed on opening such an issue and if you feel there is a better place on Gitlab somewhere, you are free to move it around
2022-08-29 18:04:45 <skarnet> the issue is really one that's at the core of all FOSS projects... how do we attract and retain enough competent and dedicated contributors so that bus factor isn't a problem? it's a human one, a political one, an economic one, not a technical one, and definitely not Alpine-specific, although I'm not at all opposed to discussing it in the context of Alpine
2022-08-29 18:04:49 <ncopa> if it wasn't the right place then its on me (or council) or doc team or anyone but you and ollieparanoid :)
2022-08-29 18:05:24 <skarnet> and I'd like to contribute to that discussion wherever the right place is
2022-08-29 18:05:35 <ddevault> also curious if only as an observer
2022-08-29 18:06:54 <Ariadne> skarnet: i think alpine has failed to recruit talent from pools already known and available to it
2022-08-29 18:07:56 <ddevault> well, that's not really all there is to it
2022-08-29 18:08:09 <ddevault> alpine has also historically had problems in offering the appropriate rights to potential contributors already in its contribution pool
2022-08-29 18:08:16 <Ariadne> +1
2022-08-29 18:08:25 <ddevault> navigating alpine politics is like wading through a waist high lake of molasses
2022-08-29 18:08:35 <Ariadne> i can speak to experience that it has been frustrating
2022-08-29 18:09:41 <Ariadne> over the years, i have identified many talented contributors, and have pushed the core team to give them access, only to have the core team sit around and do *nothing*
2022-08-29 18:10:16 <Ariadne> with the move to gitlab access being defined as infrastructure-as-code however, it has been helpful, because that brings sunlight to the whole process
2022-08-29 18:11:13 <Ariadne> i think we are starting to do better with that, but there is still work to be done
2022-08-29 18:12:43 <ncopa> sorry for that. yeah, we need to improve, and I think we do, but probably not fast enough
2022-08-29 18:13:05 <Ariadne> personally -- i think that alpine should give developer rights to whoever wants them, and should design infrastructure around enabling that vision.  but building these kinds of structures at distribution scale is difficult
2022-08-29 18:13:36 <ddevault> radical take, I wouldn't go so far
2022-08-29 18:13:54 <ddevault> we can start solving problems with less radical improvements I think
2022-08-29 18:14:06 <Ariadne> you have to have a radical vision in order to refine it into something more reasonable i think
2022-08-29 18:14:07 <ddevault> we need to apply solutions until we reach the desired outcome (i.e. patch review bandwidth) and no further
2022-08-29 18:14:22 <ncopa> i think its difficult. we dont want the situation we had in wiki, where a very eager contributor wrote *alot* but no proper quality checking was done
2022-08-29 18:14:27 <ddevault> a free for all would reduce trust in alpine as a whole
2022-08-29 18:14:40 <Ariadne> to be clear, i don't think we should turn alpine into npm
2022-08-29 18:15:18 <Ariadne> but i do think that we should have some sort of limited developer access where package maintainers can maintain packages without having to wait for patch review
2022-08-29 18:15:32 <Ariadne> other distributions such as debian have debian-maintainers for this
2022-08-29 18:15:40 <ddevault> on the whole I think it's fairly obvious at an intuitive level who would do well given the appropriate access
2022-08-29 18:16:52 <ddevault> not an expectation of perfection, but that (1) the person is invested in alpine, (2) has a good track record as a contributor without rights, and (3) gets along well enough with others that adding them would not increase social friction
2022-08-29 18:18:36 <Ariadne> yes, i have a shortlist of people i would like to see
2022-08-29 18:19:47 <ddevault> quick note that the resolution to today's social conflicts were much better than any such attempted resolutions in Alpine's past, at least during my tenure here
2022-08-29 18:19:54 <ddevault> I hope that future incidents will be resolved with a similar level of grace
2022-08-29 18:19:58 <ddevault> thanks for mediating things, ncopa
2022-08-29 18:20:35 <Ariadne> i think that we should try to focus on improving equity in the project for active stakeholders
2022-08-29 18:20:40 <Ariadne> and, i was wrong to close #57
2022-08-29 18:20:47 <Ariadne> thank you algitbot
2022-08-29 18:20:51 <Ariadne> very helpful
2022-08-29 18:20:57 <ddevault> it tried its best
2022-08-29 18:21:09 <skarnet> it's technically correct, nothing is started correctly by openrc. :P
2022-08-29 18:21:25 <Ariadne> openrc is like a good friend, trying its best
2022-08-29 18:21:42 <skarnet> more like a dumb dog
2022-08-29 18:22:00 <skarnet> you have affection for it because it's trying its best, but God it's so dumb
2022-08-29 18:22:29 <Ariadne> a dog can be a good friend
2022-08-29 18:22:47 <skarnet> the best one
2022-08-29 18:26:14 <ncopa> ddevault: np. It may take some effort to keep this a community a friendly place (from all of us) but it is worth it
2022-08-29 18:26:47 <Ariadne> in fairness, this was one of the things that was at the top of my e-mail inbox when i was still waking up
2022-08-29 18:27:20 <Ariadne> that is not intended to be an excuse, but just providing context
2022-08-29 18:29:18 <Ariadne> anyway, i do think that there should be a conversation about equity in the project, including key downstream partners like pmOS, but also people who have contributed and whose contributions have not been appropriately recognized
2022-08-29 18:29:41 <Ariadne> including ddevault, who should have commit rights
2022-08-29 18:29:59 <Ariadne> and all of the pmOS guys who maintain packages in alpine
2022-08-29 18:30:23 <ddevault> for the record I anticipate scaling back my contributions to alpine as I prefer to use email over gitlab and no one seems interested in helping with the maintenance of the bridge
2022-08-29 18:30:51 <Ermine> anything wrong with the bridge?
2022-08-29 18:30:54 <Ariadne> sure, i am just recognizing that the system did fail you
2022-08-29 18:31:04 <ddevault> yes, it's broken following some breaking change of some sort on gitlab's side
2022-08-29 18:31:08 <ddevault> it needs to be rewritten
2022-08-29 18:34:59 <ncopa> i'd be happy to give anyone commit rights if we had a good quality assurance of some sort
2022-08-29 18:35:39 <ncopa> and had something protecting from stepping on each others toes
2022-08-29 18:35:56 <Ariadne> i don't think we need to turn alpine into npm
2022-08-29 18:36:22 <ncopa> i dont think anyone wants that
2022-08-29 18:40:36 <minimal> One thing that does wrankle with me is "drive-by package upgrades". If a package maintainer takes his responsibilities seriously then it can be annoying to see someone else submit a (non-security fix) package upgrade MR mere hours after the upstream release, especially where the maintainer actively works with upstream and is already aware a release was due and is likely *already* preparing a new package upgrade
2022-08-29 18:41:17 <ddevault> on the flip side, I appreciate seeing that on my packages
2022-08-29 18:41:21 <ddevault> saves me the work!
2022-08-29 18:41:41 <Ariadne> it depends on the maintainer, i think
2022-08-29 18:41:45 <ddevault> but I don't usually ship a new release immediately after upstream
2022-08-29 18:41:54 <ddevault> usually just within a few days
2022-08-29 18:42:00 <minimal> ddevault: doesn't save you the work if you're already in the process of doing the work
2022-08-29 18:42:05 <ddevault> at least when the mailing lists were still working, now I just don't update my packages
2022-08-29 18:42:48 <minimal> plus a package building doesn't necessarily mean the software itself still works
2022-08-29 18:43:14 <panekj> I've been doing simple bumps on some packages, but when one of the devs completely refactors my own port and doesn't event consult that with me, that's a different story
2022-08-29 18:44:31 <ddevault> that one's a bit tricky
2022-08-29 18:44:48 <ddevault> aports should be fairly uniform for packages fitting into any particular mold
2022-08-29 18:45:04 <ddevault> cleaning up nonconforming packages as you work is, in my view, beneficial
2022-08-29 18:46:28 <panekj> It's only beneficial if I get a heads up
2022-08-29 18:46:52 <ddevault> well, I also think that a package update should not be merged without approval from its maintainer
2022-08-29 18:46:55 <ddevault> so there's your head's up
2022-08-29 18:47:13 <panekj> thing is, I got mail about PR but the PR was instantly merged
2022-08-29 18:47:17 <Newbyte> what is the maintainer is absent?
2022-08-29 18:47:33 <panekj> if the maintainer is absent, then we can change maintainership of package
2022-08-29 18:47:44 <panekj> just as we did it before already
2022-08-29 18:47:47 <ddevault> yeah, after waiting some appropriate period of time it can be merged without approval imo
2022-08-29 18:47:56 <ddevault> panekj: then you should probably have complained about that and asked for it to not happena gain
2022-08-29 18:47:58 <ddevault> again*
2022-08-29 18:48:14 <panekj> you can only complain so much
2022-08-29 18:48:32 <ddevault> next step is to push for it to graduate from "good idea" to "policy"
2022-08-29 18:49:32 <panekj> If the maintainer is absent, the package should be orphaned if no new maintainer steps up, not merged with update
2022-08-29 18:50:11 <ddevault> we can hash out all of the details in the TSC issue you're about to open
2022-08-29 18:51:53 <panekj> I don't think I'll be opening any TSC issues ever, I'm almost out when it comes to contributing to alpine, my main concerns are abuild/aports and that's all I'll be trying to improve
2022-08-29 19:54:43 <skarnet> I think requiring maintainer approval with a reasonable timeout is a good idea. NMUs can be a life-saver, but they can also be infuriating (and badly done).
2022-08-29 20:42:21 <donoban> hi guys, just some comments/ideas about this... Many package upgrades are straigthforward, change pkgrel && update cehcksum... couldn't be that MR automated if they pass CI etc...? Also there is an approval feature that maybe could be used for incentive devs to check MR's without having commit rights 
2022-08-29 20:43:14 <donoban> e.g. if some MR passes CI and has more than X approvations and 0 rejections, merge it
2022-08-29 20:44:36 <donoban> this way unprivileged users could do MR review without putting quality/security on risk and also feel that they do something useful
2022-08-29 20:49:50 <donoban> on the other hand I feel that alpine misses some social events that encourage people to help and improve the relations between the community, something like "bug fix day" or similar
2022-08-29 20:54:49 <donoban> is there another AlpineConf scheculed?
2022-08-29 21:03:51 <panekj> opening MRs can be automated but reviewing can't, since automation cannot know what the upgrade does and sometimes anitya isn't 100% correct about version
2022-08-29 21:04:48 <panekj> > some social events that encourage people to help and improve the relations between the community, something like "bug fix day"
2022-08-29 21:05:47 <panekj> since hacktoberfest I'm very against such events as they usually attract people that want to gain something and not contribute
2022-08-29 21:07:01 <panekj> > this way unprivileged users could do MR review without putting quality/security on risk
2022-08-29 21:07:14 <panekj> that would compromise backporting
2022-08-29 21:07:51 <panekj> and edge isn't so important to have everything merged instantly
2022-08-29 21:08:07 <skarnet> bug fix day eh?
2022-08-29 21:08:12 <skarnet> that should be every day
2022-08-29 21:08:54 <panekj> that is every day
2022-08-29 21:09:28 <skarnet> yeah, but my point it, exceptional events give the perception that contributions are exceptional
2022-08-29 21:09:34 <ebb> days matching /[ns]day$/
2022-08-29 21:09:43 <ebb> (gotta take a couple off)
2022-08-29 21:10:11 <donoban> well, the approvation scheme is not about getting things faster, it's more for having a pool of people doing that task without being a single point of failure and also making that people feel that do something useful and are part of the project
2022-08-29 21:11:34 <donoban> skarnet: well it could be also the "make fun memes" day, it's more in the social side also
2022-08-29 21:11:57 <skarnet> same thing
2022-08-29 21:11:59 <panekj> it's easier to have more trained people with commit rights than having an consensus based MRs 
2022-08-29 21:12:05 <skarnet> if I can't make fun memes every day, what's the point
2022-08-29 21:12:42 <skarnet> events don't foster community unless they're weekly or something
2022-08-29 21:13:27 <panekj> "all MRs approved tuesday"
2022-08-29 21:13:43 <panekj> each tuesday all MRs get approved
2022-08-29 21:13:53 <panekj> (:
2022-08-29 21:14:11 <donoban> the point is to get that some people discover how things work and are done, meet other people, enjoy a while talking/helping/or just reading, and hopeful some % of that people contributes more in the future
2022-08-29 21:14:15 <skarnet> brb descheduling all my Monday events, making Monday Alpine contrib day
2022-08-29 21:14:50 <panekj> donoban: an event doesn't make that happen, it's usually when people need something in alpine that leads to this
2022-08-29 21:15:30 <panekj> e.g.: someone discovers my-wonderful-glibc package isn't in alpine, I'll go and contribute that package, then they join irc, ask questions, etc.
2022-08-29 21:15:48 <donoban> well, in my personal experience I was too much motivated to work in the alpineconf days, maybe was just a coincidence
2022-08-29 21:18:13 <Ariadne> yes!  alpine glibc!
2022-08-29 21:18:17 <Ariadne> GNU/Alpine
2022-08-29 21:19:00 <Ariadne> we can also adopt systemd
2022-08-29 21:19:01 <skarnet> that's why it's important to both bring in new people *and* make sure the project keeps vision
2022-08-29 21:19:32 <panekj> I'm sure we would have more people contributing if we had proper documentation
2022-08-29 21:19:38 <Ariadne> real question: what even is the alpine project vision at this point
2022-08-29 21:19:47 <Ariadne> i sit on the TSC and i haven't a clue
2022-08-29 21:20:21 <skarnet> panekj: it's the case everywhere, and in order to do that, you need to have people motivated to write documentation, which brings us back to the starting point
2022-08-29 21:20:21 <panekj> I assume it's size-distro in main and "let's try make a desktop linux 20XX" in community
2022-08-29 21:20:36 <Ariadne> panekj: 
2022-08-29 21:20:39 <Ariadne> er
2022-08-29 21:20:50 <Ariadne> panekj: my assumption is that it's "try to make the best musl distribution"
2022-08-29 21:21:12 <skarnet> "musl-based distro suitable for container images and also general use as much as possible"
2022-08-29 21:21:13 <martijnbraam> isn't the mission simply "continue being the best"
2022-08-29 21:21:23 <Ariadne> but i don't think there has been a specified vision for ages
2022-08-29 21:21:27 <skarnet> martijnbraam: no, that's me
2022-08-29 21:21:39 <ebb> like noone ever was?
2022-08-29 21:21:59 <panekj> Ariadne: ok, best musl distro, best how?
2022-08-29 21:22:06 <Ariadne> 
ACTION nominates psykose as CEO of alpine

2022-08-29 21:22:15 <panekj> is it the fastest, securest, user friendliest
2022-08-29 21:22:23 <skarnet> yes
2022-08-29 21:22:44 <Ariadne> panekj: yes, but only because there really isn't any notable competition
2022-08-29 21:23:33 <skarnet> competition has the exact same issues
2022-08-29 21:24:06 <panekj> being the best in everything doesn't usually work
2022-08-29 21:24:15 <skarnet> that's why vision is important
2022-08-29 21:24:19 <skarnet> vision means prioritization
2022-08-29 21:24:45 <Ariadne> somebody should figure this out
2022-08-29 21:24:50 <Ariadne> not me though
2022-08-29 21:24:58 <Ariadne> i already have a /different/ mountain of technical debt to babysit
2022-08-29 21:25:04 <skarnet> it should be you in addition to other people
2022-08-29 21:25:11 <Ariadne> definitely not me
2022-08-29 21:25:20 <skarnet> I'd like to be a part of these other people
2022-08-29 21:25:48 <panekj> I'll be an innocent bystander
2022-08-29 21:26:04 <skarnet> you wish you were innocent :P
2022-08-29 21:26:37 <Ariadne> all i can say is
2022-08-29 21:26:54 <Ariadne> the reason i started using/contributing to alpine in the first place was because i needed an OS that i could run on servers
2022-08-29 21:27:13 <skarnet> isn't it the case anymore?
2022-08-29 21:27:14 <Ariadne> and that run from ram xen hypervisor shit was good shit
2022-08-29 21:27:17 <Ariadne> hell no
2022-08-29 21:27:32 <Ariadne> these days i only have my own servers because i am nuts
2022-08-29 21:27:43 <panekj> I run Alpine as server OS
2022-08-29 21:27:51 <Ariadne> i mean it's a good server OS
2022-08-29 21:27:53 <martijnbraam> I run full stack alpine :)
2022-08-29 21:27:57 <Ariadne> but people don't like running servers anymore
2022-08-29 21:27:58 <panekj> I also run it as desktop OS (rarely)
2022-08-29 21:28:02 <Ariadne> actually i suspect they never did
2022-08-29 21:28:19 <panekj> but I think it shouldn't be a desktop OS
2022-08-29 21:28:29 <skarnet> desktop is its own can of worms
2022-08-29 21:28:48 <Ariadne> i think the reality of it is that alpine is the universal musl distribution
2022-08-29 21:29:01 <Ariadne> you have stuff like docker pulling it in container direction
2022-08-29 21:29:10 <Ariadne> you have stuff like pmOS pulling it in desktop direction
2022-08-29 21:29:11 <skarnet> yeah it's the current state, the question is what do we want to make of it
2022-08-29 21:29:24 <skarnet> also
2022-08-29 21:29:39 <panekj> I love alpine for being able to build muh static binaries
2022-08-29 21:30:05 <skarnet> I'm going on vacation in 9 days, timeout for my mdevd MR arrives shortly before that, and the MR led to a TSC issue that you're supposed to give input on, Ariadne
2022-08-29 21:30:25 <Ariadne> i prefer musl because i can understand it
2022-08-29 21:30:38 <Ariadne> i mean, i can understand glibc too, but it requires a lot more effort
2022-08-29 21:30:51 <panekj> if I could read C, i would prefer it because of that as well
2022-08-29 21:31:09 <skarnet> if my MR disappears in limbo for the second time because it times out because it opened a TSC issue that hanged in limbo
2022-08-29 21:31:14 <skarnet> I will be very unhappy
2022-08-29 21:31:32 <Ariadne> what is the MR
2022-08-29 21:31:49 <panekj> fortunately, I sleep fine while being unaware of *((***(*(*a*++*)) magic
2022-08-29 21:32:21 <skarnet> Ariadne: !37834
2022-08-29 21:33:01 <skarnet> which sparked a discussion about the long-term goals for busybox
2022-08-29 21:33:17 <skarnet> this discussion is good and interesting, but it can last for a long time
2022-08-29 21:33:49 <skarnet> I would like to be able to merge things without having to rediscuss the meaning of life every single time
2022-08-29 21:34:09 <skarnet> or, alternatively, we should figure out what the meaning of life is, but sooner rather than later
2022-08-29 21:34:30 <panekj> game of life
2022-08-29 21:37:59 <Ariadne> skarnet: i think that ncopa is right, the openrc scripts should live with busybox.
2022-08-29 21:38:27 <Ariadne> skarnet: if you do that, and nmeum (as busybox maintainer) agrees, we can merge it
2022-08-29 21:38:28 <skarnet> two things
2022-08-29 21:38:43 <Ariadne> it's consistent with established practice
2022-08-29 21:39:03 <skarnet> 1. that's the exact opposite of what was asked of me in the TSC issue 52
2022-08-29 21:39:21 <Ariadne> no?
2022-08-29 21:39:23 <skarnet> I did not write a hash symbol, algitbot, would you kindly stfu
2022-08-29 21:39:39 <panekj> algitbot has been kinda weird lately
2022-08-29 21:39:54 <skarnet> I was asked to split into bb-openrc, bb-extras-openrc, bb-mdev-openrc and it's exactly what I did
2022-08-29 21:39:54 <Ariadne> delete busybox-initscripts, replace with busybox-openrc
2022-08-29 21:40:12 <skarnet> okay so 3 things
2022-08-29 21:40:15 <Ariadne> yeah, basically what i am asking is to have it with the busybox aport as subpackages
2022-08-29 21:40:37 <skarnet> okay I can do that but have you read my comment on TSC 55
2022-08-29 21:40:47 <Ariadne> i did
2022-08-29 21:41:06 <skarnet> and *yes* it's not consistent with established practice, *that is the point*, have better practices
2022-08-29 21:41:23 <skarnet> established practices, in some cases, and definitely this one, are an obstacle to improvement
2022-08-29 21:41:43 <Ariadne> it is assumed that if we do splitting as described in TSC 55, that they too, would be subpackages
2022-08-29 21:42:07 <skarnet> you love your technical debt
2022-08-29 21:42:12 <skarnet> you all do
2022-08-29 21:42:20 <Ariadne> there is a reason for this, it allows us to have a single input (the busybox port), with all related outputs
2022-08-29 21:43:01 <Ariadne> i love technical debt so much i am building an entire distribution using the exact opposite of bazel ;)
2022-08-29 21:43:52 <skarnet> what can I say, putting everything under the busybox APKBUILD is an obstacle to later progress
2022-08-29 21:44:13 <Ariadne> incremental changes win the race in alpine
2022-08-29 21:45:05 <skarnet> to gain an inch of incremental change it feels like dragging my balls over hot coals, every single time
2022-08-29 21:45:22 <panekj> sounds like foss
2022-08-29 21:45:23 <Ariadne> some find that to be an enjoyable experience, i'm told
2022-08-29 21:45:53 <skarnet> "if your ideas are any good you'll have to shove them down people's throats" has never been more true
2022-08-29 21:46:11 <skarnet> and then ncopa doesn't like it because I'm pushy
2022-08-29 21:46:21 <Ariadne> yes, hence
2022-08-29 21:46:28 <Ariadne> 4:44 PM <Ariadne> incremental changes win the race in alpine
2022-08-29 21:46:35 <skarnet> if I'm not pushy I get NOTHING done
2022-08-29 21:46:49 <skarnet> even when I'm pushy af it takes ages to get half a MR merged
2022-08-29 21:48:11 <Ariadne> all i am saying is that doing one thing is reasonably normal (by alpine standards), and doing another thing involves changing policy
2022-08-29 21:48:28 <Ariadne> i might sit on the TSC, but i can't just change policy
2022-08-29 21:48:36 <skarnet> what is the point of opening TSC issues and discussing things (kinda? I wasn't even invited to the discussion) if it's not to change policy
2022-08-29 21:48:59 <skarnet> you told me that my initial MR needed a TSC issue because it was changing policy
2022-08-29 21:49:01 <skarnet> I said fine
2022-08-29 21:49:05 <skarnet> I opened a TSC issue
2022-08-29 21:49:13 <skarnet> and now you're backtracking
2022-08-29 21:49:29 <Ariadne> no, i am saying i will merge your MR if you make the initscripts subpackages
2022-08-29 21:49:45 <Ariadne> which i will do
2022-08-29 21:50:02 <skarnet> I will do this, and in a few months I will open another MR to split them again, and another TSC issue to explain why it would be a good change
2022-08-29 21:50:08 <skarnet> and maybe in 15 years we'll have nice things
2022-08-29 21:50:11 <Ariadne> now you're getting it ^_^
2022-08-29 21:51:03 <skarnet> can't help but feel the process could be more efficient
2022-08-29 21:51:12 <Ariadne> open a TSC issue :D
2022-08-29 21:51:47 <skarnet> that's a funny troll response, but you have to see how unviable the process is
2022-08-29 22:18:25 <Ariadne> skarnet: not really a troll response.  if you want the TSC to change its policies, starting a TSC issue seems appropriate
2022-08-30 07:19:20 <omni> good morning
2022-08-30 07:29:55 <Ermine> good morning omni
2022-08-30 08:40:55 <omni> could algitbot remove status:mr-stale when there's activity, like new commits, again?
2022-08-30 08:49:52 <ikke> omni: if someone implements it 
2022-08-30 09:09:43 <omni> ok =)
2022-08-30 09:37:13 <ebb> where's the sauce?
2022-08-30 09:49:09 <ikke> https://gitlab.alpinelinux.org/Cogitri/aports-qa-bot
2022-08-30 10:52:37 <Ermine> how do I do security upgrades to stable branches (namely I want to upgrade virtualbox). Do I need to write CVEs to APKBUILD?
2022-08-30 10:53:39 <Ermine> err, virtualbox-guest-additions
2022-08-30 12:48:44 <ncopa> Ermine: yes, we add a yaml formatted secfixes comment. you can see how we have don it in for example main/curl/APKBUILD
2022-08-30 12:52:33 <Ermine> ncopa: thank you. Do I need to mark this upgrade as security in commit message?
2022-08-30 12:54:30 <ncopa> I normally do so, and I think it is nice if you do
2022-08-30 12:54:57 <ncopa> if it is a single CVE then keep it in commit message header (frst line), other wise I normally do a list
2022-08-30 12:55:17 <ncopa> its useful so you can git log --grep $CVE
2022-08-30 12:57:48 <Ermine> also, should I add secfixes comment to edge for consistency, and should I bump pkgrel if yes?
2022-08-30 12:58:49 <psykose> yes and no
2022-08-30 12:59:55 <Ermine> ok, thank you
2022-08-30 13:03:07 <ikke> Secfixes are not stored in the package, so a rebuild is not required 
2022-08-30 13:03:57 <psykose> i'm still confused what the point of most of them is
2022-08-30 13:04:25 <psykose> >cve-1234-5678 comes out
2022-08-30 13:04:27 <psykose> >open cve website
2022-08-30 13:04:33 <psykose> >versions before 1.2.3 affected
2022-08-30 13:04:36 <psykose> >upgrade to 1.2.3
2022-08-30 13:04:43 <psykose> >explicitly mark 1234-5678 as fixed
2022-08-30 13:04:45 <psykose> >????????????????
2022-08-30 13:05:29 <psykose> it makes sense for manual patches, but in every other case i am confused what the data is for
2022-08-30 13:05:43 <psykose> the one time i even tested a scanner it does not flag things that are missing explicit secfixes
2022-08-30 13:06:02 <panekj> so people in their companies who just woke up and remembered about something called "security" can check a website otherwise they will spam every possible point of contact to get that information 
2022-08-30 13:06:14 <psykose> that was my conclusion too
2022-08-30 13:06:30 <psykose> given the random 'issues' made in aports of people posting excel spreadsheets of 'vulnerabilities' they found
2022-08-30 13:06:47 <psykose> all of this is a waste of time for the most part to appease some 5 people that don't know how to check for vulnerabilities
2022-08-30 13:07:21 <psykose> anything weirdly disputed or not marked as fixed in some version or manually backported via patch needs them
2022-08-30 13:07:30 <psykose> but in general? it's just a bunch of useless data
2022-08-30 13:07:56 <psykose> not fun to scroll through 600 lines of text in the web browser apkbuilds just because someone doesn't know how to check anything
2022-08-30 13:08:45 <panekj> it's ok in browser, it's not ok in terminal
2022-08-30 13:09:49 <psykose> i mean 'web browser apkbuilds' like literally firefox/chromium etc
2022-08-30 13:09:55 <psykose> every single release adds another 15 entries to the list
2022-08-30 13:09:57 <ikke> I'd say we should be able to expire (remove) secfixes at some point 
2022-08-30 13:10:33 <ikke> Also secfixes in master would not make sense in the next stable release 
2022-08-30 13:10:51 <panekj> sorry, brain bad
2022-08-30 13:11:37 <psykose> i dunno, i feel like they would be easier to manage if they were not part of the apkbuilds
2022-08-30 13:11:56 <psykose> the data just gets scraped into the other service in any case
2022-08-30 13:12:13 <psykose> if it's really a must to flag everything ever for no reason, i would prefer to just input it somewhere
2022-08-30 13:15:12 <Ermine> something like SECFIXES file in aports dir?
2022-08-30 13:15:47 <ikke> That has been proposed before 
2022-08-30 13:18:46 <psykose> no, that's the same thing but in two files instead
2022-08-30 13:19:41 <psykose> i mean just manually adding it into a webform or whatever
2022-08-30 13:20:30 <psykose> it's much faster that way, no useless lines in apkbuilds, anything with whatever dispute can just be added in under a minute with anyone with access (reporters or the like)
2022-08-30 13:20:36 <psykose> granted, managing that now brings a lot more issues
2022-08-30 13:20:39 <ikke> For security.a.o it's not necessary to add every CVE 
2022-08-30 13:21:46 <ikke> The atomicity of having it part of the commit that fixes is, is good to have 
2022-08-30 13:22:21 <psykose> if it's not necessary to add all of them, what is the 1000% correct policy on what should be added
2022-08-30 13:22:26 <psykose> so far my impression has been the opposite
2022-08-30 13:22:49 <psykose> the most benign of random shit is always added, even with very clear package upgrades for only a security fix where the sole diff in the upstream release is the 1 commit to fix it
2022-08-30 13:23:04 <psykose> which is as close to 'literally the upgrade is the secfix' as anything gets, and even that gets added to secfixes
2022-08-30 13:24:17 <psykose> (i'm not necessarily complaining about it, but what is the policy meant to be? because every time i've asked i've gotten a different response)
2022-08-30 13:24:24 <psykose> this is the opposite of last time
2022-08-30 13:24:47 <psykose> and based on the people posting excel spreadsheets... apparently it's all of them, dunno
2022-08-30 13:26:15 <ikke> With a separate form, I have a feeling it would be more easily overlooked 
2022-08-30 13:27:17 <psykose> yeah, that's one of the downsides
2022-08-30 13:55:06 <ikke> I suspect the idea was that if we provide a complete picture, we'd reduce false positives with security scanners 
2022-08-30 15:41:12 <minimal> psykose: security scanners are a broad spectrum with at one end tools that simply check the "upstream" version number and know nothing about Alpine-specifics through to those that do understand Alpine specifics and do (directly or indirectly) look for secfixes info in APKBUILD
2022-08-30 15:43:34 <minimal> the "spreadsheet" type complaints/questions are typically where someone sees that trendy-package version 1.2.3 upstream has a CVE and has no idea/doesn't care how Alpine builds its version of that package (i.e. whether the functionality in question is even enabled in the Alpine build)
2022-08-30 15:47:21 <psykose> yes, but in the specific instances i have seen, the reports weren't for 1.2.3 upstream, as the version was 1.2.3 (with the fix, in alpine, just not in secfixes). i have no idea what detection they were using
2022-08-30 15:47:40 <psykose> for the scanners that don't look at the alpine db in the first place there isn't really anything we can do
2022-08-30 15:47:57 <psykose> (wrt patch backports)
2022-08-30 15:48:31 <psykose> so if they /are/ looking there, and they /also/ can look at 1.2.3 upstream, then i just feel we don't need to always put a secfix for every single thing on release ever
2022-08-30 15:48:32 <psykose> not much else
2022-08-30 15:48:50 <psykose> but currently, we just add everything
2022-08-30 15:55:00 <minimal> psykose: some of these reports are likely occuring due to Alpine popular use as a container base image - tools like clair and aqua are being run with the output treated by someone as a tick-box exercise to deal with
2022-08-30 15:55:31 <psykose> i can't speculate aside from concluding the scanners can't detect anything at all
2022-08-30 15:55:39 <psykose> not even the version apparently
2022-08-30 15:55:56 <psykose> libxy exists? here's every cve for libxy ever issued
2022-08-30 16:33:01 <ncopa> psykose: originally we only added backported to our secfixes but security vendors didnt check the version of apk, only data in secfixes db
2022-08-30 16:33:18 <psykose> well.. perhaps they should?
2022-08-30 16:33:19 <ncopa> they are just lazy, so we add evertying to secfixes db
2022-08-30 16:34:21 <psykose> i know most of the databases aren't very complete to begin with (a lot of cves don't even state what version fixed what, so it's understandable), but cmon
2022-08-30 16:34:29 <psykose> /zero/ checking at all..
2022-08-30 16:35:54 <minimal> I suspect also that for some scanner tools people may be using old (i.e. local, not up-to-date) versions of databases
2022-08-30 17:57:33 <Ariadne> psykose: the scanners are so utterly useless
2022-08-30 17:58:03 <Ariadne> the best part is that they miss anywhere between 40-80% of the contents of the image
2022-08-30 17:58:22 <Ariadne> for example... if you scan the official node image with trivy.  it does not even know node is there.
2022-08-30 17:59:23 <ikke> As they're only aware of what the package manager says is installed 
2022-08-30 17:59:28 <Ariadne> yes
2022-08-30 17:59:44 <psykose> i just checked
2022-08-30 17:59:47 <psykose> it's true, hah
2022-08-30 18:00:11 <ikke> This happens for any image where things are built from source 
2022-08-30 18:00:37 <Ariadne> well, built with docker from source, anyway.
2022-08-30 18:00:47 <ikke> Yes, that's what I meant 
2022-08-30 18:01:07 <psykose> i then ran `trivy image node`.. and it still didn't find node
2022-08-30 18:01:15 <psykose> but i did get 8 terminal screens of cves in the other packages
2022-08-30 18:01:20 <Ariadne> if you use melange to generate an apk, or whatever, then it will show up in inventory ;)
2022-08-30 18:01:32 <Ariadne> which is why $dayjob is trying to make it easy to build packages
2022-08-30 18:01:41 <psykose> yeah, the assembly approach makes much more sense
2022-08-30 18:03:10 <ikke> https://twitter.com/lorenc_dan/status/1563862217992425478
2022-08-30 18:04:16 <Ariadne> haha yeah
2022-08-30 18:04:18 <psykose> some of these don't even seem correct
2022-08-30 18:04:29 <psykose> >m4 version: 1.4.18-5
2022-08-30 18:04:34 <psykose> >CVE-2008-1688
2022-08-30 18:04:38 <Ariadne> :D
2022-08-30 18:04:39 <psykose> >Unspecified vulnerability in GNU m4 before 1.4.11 ..
2022-08-30 18:04:40 <psykose> ???
2022-08-30 18:04:58 <psykose> how do you even get this wrong? does it choke on the debian pkgrel again?
2022-08-30 18:05:01 <psykose> sigh
2022-08-30 18:05:06 <Ariadne> we are going to release a tool soon to show you the things the scanner didn't actually scan
2022-08-30 18:05:13 <Ariadne> it's gonna be lit
2022-08-30 18:05:22 <psykose> it's gonna say 90% of the image is unscanned
2022-08-30 18:05:26 <Ariadne> yep
2022-08-30 18:06:29 <Ariadne> 100% coverage from scanners if you use the products $dayjob sells though
2022-08-30 18:06:37 <Ariadne> assuming you use them correctly, anyway
2022-08-30 18:06:44 <psykose> ah of course you sell the working solution ;)
2022-08-30 18:07:02 <Ariadne> well, the components are FOSS, of course
2022-08-30 18:07:31 <Ariadne> but yes, the pretty GUI where you click "i want to build this container with these packages from this source" costs $$$
2022-08-30 18:07:38 <Ariadne> have to make money somehow :)
2022-08-30 18:08:23 <Ariadne> (though we may eventually release that too in a self-hosted version under something like AGPL)
2022-08-30 18:08:44 <psykose> cute gui 🥺
2022-08-30 18:09:06 <bl4ckb0ne> can you make it cheaper with an ugly gui
2022-08-30 18:09:37 <Ariadne> we plan to offer a basic PPA service for alpine and other apk distributions for free :p
2022-08-30 18:09:51 <ikke> Lol, agpl
2022-08-30 18:09:56 <psykose> >nvd.nist.gov
2022-08-30 18:09:59 <psykose> >503 Service Unavailable
2022-08-30 18:10:04 <psykose> for gods sake
2022-08-30 18:10:14 <Ariadne> a problem is that docker was spending basically $200k per month on transfer charges
2022-08-30 18:10:20 <Ariadne> so, trying to figure that one out
2022-08-30 18:10:28 <ebb> www.cve.org nowadays, although it's a worse UI I feel
2022-08-30 18:10:40 <Ariadne> as tianon put it "a porsche 911 turbo's worth of money"
2022-08-30 18:11:56 <Ariadne> ikke: i mean, AGPL was designed for SaaS copyleft, it makes sense i think :)
2022-08-30 18:12:09 <ikke> Yes, I agree 
2022-08-30 18:12:48 <Ariadne> psykose: anyway, CVE-2008-1688 does not have any CPE data, so the version match isn't constrained
2022-08-30 18:12:54 <psykose> ah
2022-08-30 18:13:07 <Ariadne> security.alpinelinux.org is basically a scanner just like these commercial products, except it scans apk indexes rather than images
2022-08-30 18:22:15 <Ariadne> what is a very large apk in alpine's repos?
2022-08-30 18:22:59 <psykose> supertuxkart-data
2022-08-30 18:23:11 <bl4ckb0ne> sauerbraten is pretty chonky
2022-08-30 18:23:32 <psykose> clang-static
2022-08-30 18:23:52 <bl4ckb0ne> yikes 1gb
2022-08-30 18:23:54 <panekj> something -dbg?
2022-08-30 18:24:02 <psykose> and uncompressible as it's all game assets
2022-08-30 18:24:11 <psykose> realistically i'm not sure why we ever added any games into the repository
2022-08-30 18:24:25 <bl4ckb0ne> its free yo
2022-08-30 18:24:30 <Ermine> you don't want to play?
2022-08-30 18:25:47 <Ermine> eli5 what the purpose of those scanners if they don't actually work? Is it to make people kalm?
2022-08-30 18:25:58 <psykose> security marketing
2022-08-30 18:26:26 <ebb> "Compliance"
2022-08-30 18:26:29 <Ermine> "Look we are SeCuRe"?
2022-08-30 18:26:39 <panekj> as I said earlier, people wake up and are like "oh my god we forgot about security"
2022-08-30 18:27:37 <Ariadne> thanks, coworker is looking for a reasonable worst case scenario to benchmark his apk repository management code
2022-08-30 18:27:53 <ebb> panekj: or put tape over their eyes
2022-08-30 18:29:15 <ebb> Sometimes it's a blessing working on a(n actually) closed system for $dayjob because the cybersec risks are fewer
2022-08-30 18:30:08 <ebb> But it does mean when working on related stuff outside of that bubble, people forget about security
2022-08-30 18:30:23 <psykose> Ariadne: if it's related to handling reading packages (i was following the buffer reading go code in the Go repo with your coworker in it) it's probably best to not assume those and make a fake ~10GB+ apk anyway :)
2022-08-30 18:31:47 <Ariadne> yeah i did a call with him yesterday to walk through why we need to rethink that
2022-08-30 18:32:30 <Ariadne> at google they will just spawn a VM or container or whatever that has like 256GB of RAM
2022-08-30 18:32:34 <Ariadne> yolo :D
2022-08-30 18:33:35 <psykose> if you know the limits and the memory is there anyway afaik that's a bit faster regardless
2022-08-30 18:33:46 <psykose> at that scale there's other ways of approaching/composing stuff to begin with, no?
2022-08-30 18:34:21 <Ariadne> yeah
2022-08-30 18:34:23 <psykose> much more generic code in this case has to assume less memory than apk size just to be friendly
2022-08-30 18:34:33 <Ariadne> but the ioutil.ReadAll golang meme comes from that thought process
2022-08-30 18:34:39 <psykose> haha, yeah
2022-08-30 18:34:49 <Ariadne> "out of memory?  we'll just add more!"
2022-08-30 18:36:02 <psykose> wish more people shipped .zst tarballs of code
2022-08-30 18:37:51 <minimal> Ermine: its a tick-box exercise in most cases. A bit like ISO-9000 which everyone calls "Quality", its not, it's for "Reproducibility" - McDonalds is ISO-9000 certified, this doesn't mean their food is quality food, just that their food is "guaranteed" to be the same (i.e. *equally* crap) in all their stores
2022-08-30 18:42:53 <psykose> but the nuggies
2022-08-30 18:49:51 <Ariadne> lmao
2022-08-30 18:50:00 <Ariadne> https://security-tracker.debian.org/tracker/CVE-2008-1688
2022-08-30 18:50:10 <Ariadne> that is why the scanners say vuln
2022-08-30 18:50:26 <Ariadne> because nobody in debian ever triaged it i guess
2022-08-30 19:00:23 <psykose> haha
2022-08-30 19:05:34 <minimal> Ariadne: the trvy docs do state "Trivy doesn't support self-compiled packages/binaries, but official packages provided by vendors such as Red Hat and Debian."
2022-08-30 19:06:10 <Ariadne> minimal: yes, i know, but that disclaimer sure is doing a lot of heavy lifting
2022-08-30 19:06:35 <Ariadne> for the average dockerfile it means the tool is basically useless
2022-08-30 19:07:03 <minimal> Ariadne: I'm guessing Aqua's commercial offering may do more as they're getting paid for it
2022-08-30 19:07:13 <Ariadne> (it doesn't)
2022-08-30 19:09:37 <minimal> I seem to remember one of the commercial scanners being used at a contract I worked on 3+ years ago (think it was Aqua) where I built some Alpine containers, the problem however was the customer was using it in an isolated network and so there were issues with having up-to-date vendor's definitions/database
2022-08-30 19:11:38 <Ariadne> for what it is, trivy is a good scanner
2022-08-30 19:12:03 <Ariadne> to be clear, i am not saying it is a bad scanner, i am just saying that most people who use it are using it wrong :D
2022-08-30 21:16:17 <skarnet> Ariadne: going offline, but just fyi, I updated my busybox-openrc MR, everything is a subpackage of busybox now
2022-08-30 21:17:14 <Ariadne> skarnet: okay, i'll review it in a bit.  right now i am in "packaging python3 hell"
2022-08-30 21:18:11 <skarnet> good luck. unlile this morning, I'm going to try and get some sleep before the workers start tearing down my building again.
2022-08-30 21:18:21 <skarnet> unlike, even.
2022-08-30 21:18:30 <psykose> why does the alpine-base package change alpine-release -> alpine-keys
2022-08-30 22:00:07 <pineapple_cake> When I install alpine to my hard drive and boot on it, it does not recognize my disk because of some kernel modules missing (probably vmd). Is there a way to install alpine with this module ?
2022-08-30 22:01:21 <minimal> pineapple_cake: what exactly is vmd? is your drive not either PATA, SATA or SCSI?
2022-08-30 22:02:03 <psykose> probably needs kernel/drivers/pci/controller/vmd.ko in the initrd
2022-08-30 22:03:55 <pineapple_cake> How do I insert it to the initrd ?
2022-08-30 22:04:29 <minimal> ah Intel Volume Management Device
2022-08-30 22:04:38 <pineapple_cake> yep
2022-08-30 22:06:29 <minimal> hmm, a sort of Intel-proprietary version of NVME
2022-08-30 22:08:57 <pineapple_cake> I have the kernel object but I don't know how to put it in the alpine initrd and make alpine load it on boot
2022-08-30 22:11:06 <minimal> pineapple_cake: you'd have to add it to /etc/mkinitramfs/features.d/nvme.modules, ensure that "nvme" is listed in /etc/mkinitfs/mkinitfs.conf, then run mkinitfs to rebuild the initramfs
2022-08-30 22:11:18 <psykose> make a /etc/mkinitfs/features.d/vmd.modules , look at the other .modules files there to see the path for it, and then put vmd in /etc/mkinitfs/mkinitfs.conf
2022-08-30 22:11:50 <minimal> doing this ALL after chroot-ing into the place where you installed Alpine after running setup-alpine.
2022-08-30 22:12:00 <minimal> That's assuming you installed Alpine in Sys-mode
2022-08-30 22:14:43 <minimal> pineapple_cake: another option might just be to disable VMD in ~UEFI/BIOS, sounds like this can be done
2022-08-30 22:15:02 <minimal> I guess then the drive would appear to "normal" NVME
2022-08-30 22:15:44 <pineapple_cake> After chrooting to the place I installed Alpine & adding this module should I call mkinitramfs ?
2022-08-30 22:16:00 <pineapple_cake> I will look up on the BIOS if I can disable it
2022-08-30 22:17:45 <pineapple_cake> Ok there was just an option to disable it in the BIOS
2022-08-30 22:17:52 <psykose> `apk fix linux-lts` would rerun mkinitfs for you
2022-08-30 22:17:57 <psykose> if you were booting linux-lts
2022-08-30 22:18:11 <minimal> pineapple_cake: I assume the Alpine ISO you used loaded the vmd module, as otherwise how did you manage to install to that drive in the 1st place
2022-08-30 22:18:27 <pineapple_cake> yes
2022-08-30 22:19:02 <minimal> ok, sounds like the mkinitfs package should be modified to bundle vmd
2022-08-30 22:19:07 <pineapple_cake> it was in /lib/modules/<KERNEL_VERSION>/drivers/pci/controller/vmd
2022-08-30 22:19:26 <pineapple_cake> Thank you very much for the help
2022-08-30 22:19:37 <minimal> VMD appears to be for RAID use
2022-08-30 22:20:13 <panekj> it's probably that weird motherboard raid thingy
2022-08-30 22:20:48 <minimal> panekj: its a Xeno CPU "special" - it relies on functionality present in those CPUs
2022-08-30 22:20:58 <minimal> Xeon even lol
2022-08-30 22:21:44 <panekj> AMD has same thing
2022-08-30 22:22:14 <minimal> I gave away all my Xeons as I found a cheaper way to heat my living room lol
2022-08-30 22:24:08 <panekj> I generally recommend not using that stuff and disabling it
2022-08-30 23:13:11 <omni> I found this earlier https://lists.alpinelinux.org/~alpine/devel/%3Ctrinity-9f3437a2-9c6a-43d8-a5be-71264898e6fc-1499780412728%403capp-mailcom-lxa13%3E
2022-08-30 23:13:53 <omni> I think I was looking for something around co-maintainership
2022-08-30 23:14:15 <psykose> > As a good side-effect this would also make the maintainer of a package easier to parse.
2022-08-30 23:14:17 <psykose> huh?
2022-08-30 23:18:07 <omni> I think I also tried to find something about what Contributor is useful for, the best I've found so far is in one of the comments: "a nice way to give credit to people who contribute"
2022-08-30 23:20:31 <omni> but I'm not advocating removing it, just that I found that old thread and it has arguments for why there should only be a single (main) Maintainer of an aport
2022-08-30 23:22:13 <psykose> i never really saw much point of it and don't add it anywhere anymore
2022-08-30 23:22:38 <psykose> i guess originally i cared a bit more about what was "on the surface"
2022-08-30 23:22:39 <omni> but I beleive some way to define co-maintainers could be valuable, so that any of a set of people could approve of changes to an aport
2022-08-30 23:22:54 <psykose> but these days i know that if you want to know anything at all, you need to actually peruse the git history somewhat
2022-08-30 23:23:07 <psykose> reading the log and blame is much more useful than some comment in a file
2022-08-30 23:23:24 <omni> right
2022-08-30 23:23:41 <psykose> aside from top-level contributors (think a project with an AUTHORS file, usually added for copyright reasons) or CODEOWNERS files, it's meh
2022-08-30 23:24:15 <omni> I think I added myself as a Contributor to a few aports the first few changes I did in aports, but then stopped
2022-08-30 23:24:24 <psykose> the multiple-maintainers thing is completely unrelated to whether or not contributors are there
2022-08-30 23:24:36 <psykose> not sure why it keeps getting brought up every time someone discusses this same thing, haha
2022-08-30 23:24:37 <omni> yes
2022-08-30 23:25:36 <omni> and I guess I did the reverse? brought up Contributors when I wanted to talk about Maintainers and Co-Maintainers
2022-08-30 23:26:00 <psykose> multiple-maintainers doesn't feel terribly useful
2022-08-30 23:26:50 <psykose> in the same way that giving 500 people commit rights or whatever isn't really useful either; you don't actually gain much value from it, as where all the challenge and work lies is in the actual work and coordination
2022-08-30 23:27:15 <psykose> a line in a file doesn't really facilitate anything, unless it's some system with 100000 people working on it and a forced-approval from exactly the people listed
2022-08-30 23:27:47 <psykose> that would only be relevant if we suddenly had a structure where maintainers could push things themselves
2022-08-30 23:31:44 <psykose> dunno, most of the time it's just endless bikeshedding
2022-08-30 23:38:34 <omni> perhaps
2022-08-30 23:40:24 <omni> somewhat related https://gitlab.alpinelinux.org/alpine/tsc/-/issues/46
2022-08-30 23:40:41 <omni> well, nn
2022-08-30 23:48:15 <psykose> i mean, at the end of the day, i know everyone that does everything in aports
2022-08-30 23:48:40 <psykose> having a team tag somewhere for me personally doesn't accomplish anything, same for basically every other suggestion in existence
2022-08-30 23:48:43 <psykose> i am an outlier though
2022-08-31 06:54:05 <omni> moin
2022-08-31 07:00:10 <psykose> is moin
2022-08-31 07:00:27 <panekj> good moin
2022-08-31 07:21:30 <psykose> when a release image is built, is it made from the master scripts/mkimg.*.sh called with --branch per branch, or with the specific release-branch mkimg.sh per branch
2022-08-31 08:15:05 <ikke> I think from each respective branch 
2022-08-31 08:15:33 <ikke> Source is in aports-build
2022-08-31 08:20:56 <maribu> Does anyone here still have to work with subversion? To me, it seems that the subversion client shipped in Alpine is broken. It eats through CPU time without doing anything meaningful.
2022-08-31 08:23:12 <maribu> Even after removing `~/.subversion` and doing a new checkout, it doesn't do anything. I would expect it to at least ask for a username and a password. Opening the https URL via Firefox works fine (and Firefox does ask for a username and password).
2022-08-31 08:29:06 <maribu> A quick peek with GDB looks like serf may be the actual culprit
2022-08-31 08:29:15 <ikke> I haven't used svn for a long time
2022-08-31 08:30:22 <maribu> I am jealous :) I am dreaming of the day I can remove svn and never think of it again :)
2022-08-31 08:51:24 <maribu> It really seems to be that serf fails to perform a TLS handshake and just retries over and over again. I just tried `svn co <RANDOM_HTTPS_URL>` (since it is prior to any SVN related operation, any HTTP server is fine here). And either svn is hanging or yielding `svn: E120171: Error running context: An error occurred during SSL communication`.
2022-08-31 08:52:08 <maribu> E.g. `svn co https://alpinelinux.org` yields the error (consistently), but e.g. `svn co https://golem.de` yields the endless loop (consistently).
2022-08-31 08:53:30 <ikke> openssl 3.0 related? 
2022-08-31 09:27:10 <maribu> Could be
2022-08-31 09:43:44 <maribu> https://gitlab.alpinelinux.org/alpine/aports/-/merge_requests/38441
2022-08-31 12:22:32 <ikke> ncopa: could you take a look at !26263? It has been open for a while. 
2022-08-31 19:20:51 <Boxst> Best Free Nextgen Social/Dating Site - Southern Society - http://ssd.us.to
2022-08-31 20:05:54 <ine-one1> In the name of Elohim the one and only the King of kings the right the sustainer of existence the most powerful 1_ Monotheism Judaism is a special Branch of islam, it is the belief in Elohim the only one, no God except he, from prophet Moses way
2022-08-31 20:06:02 <ine-one1> The right belief is: there is no God except one; Elohim, and prophet Moses is the messenger of Elohim (God Rules and the power is in his hands) 2_ No body asked U to commite crimes AND kiLL innocent believers and poor people, SySTeMATicly which U r judged upon by God the redeemer
2022-08-31 20:06:11 <ine-one1> 3_ By the authority of Elohim the one the greatest and from Prophets Moses haron and Joshua way: 4_ People u r now appointed by us to achieve big achievements - its your bad acts and deeds and God the redeemer Big punishments getting back on u till u do whats requested well.
2022-08-31 20:06:20 <panekj> what about flying spaghetti monster
2022-08-31 20:06:23 <ine-one1> 5_ You are requested to eliminate personalities and positions in power and in the ruling current regimes inplace of what u did for this tyrant regimes of crimes until its end, under consequence of big punishments from Elohim the Greatest redeemer if u were late [Moses of Elohim. M]
2022-08-31 20:06:39 <panekj> ikke: 
2022-08-31 20:18:12 <psykose> Terrible and powerful Talos! We, your unworthy servants, give praise! 
2022-08-31 20:26:12 <omni> all hail the goddess Eris!
2022-08-31 20:45:16 <Ermine> All hail psykose !
2022-08-31 20:45:21 <psykose> no
2022-08-31 20:45:28 <Ermine> Y
2022-08-31 20:56:32 <lucidiot> 
ACTION bows before psykose

2022-08-31 21:26:02 <Misthios> *offers a piece of paper*
2022-08-31 22:09:54 <omni> psykose: do you follow things in the web ui or are you just wired in?
2022-08-31 22:10:02 <psykose> whadya mean by that
2022-08-31 22:10:06 <psykose> also i fixed the whole opencpn thing
2022-08-31 22:12:10 <omni> I mean, are you hooked into the mainframe, you ninja
2022-08-31 22:12:30 <omni> where did you fix what with it?
2022-08-31 22:12:45 <psykose> it builds
2022-08-31 22:12:49 <omni> with !38456?
2022-08-31 22:13:00 <psykose> yea
2022-08-31 22:13:11 <psykose> uhh i just follow everything
2022-08-31 22:13:21 <omni> ah, there, I see
2022-08-31 22:18:07 <psykose> ah, seems it didn't even need webview, hehe, was just confused a month ago
2022-08-31 22:20:03 <omni> I've been confused each time I've decided to bang my head against that, to then give up and leave it for a few months
2022-08-31 22:20:32 <omni> b g ninja
2022-08-31 22:23:17 <omni> I read in your changes when I squint my eyes
2022-08-31 22:24:52 <psykose> blue+green = turquoise
2022-08-31 22:35:57 <omni> yes