2022-01-01 00:16:42 <boomanaiden154> minimal: Yes, the base image is basically just the minirootfs. I believe it gets built everytime a new release is tagged.
2022-01-01 00:16:54 <boomanaiden154> As for how often it gets rebuilt for edge, I'm not sure.
2022-01-01 00:17:00 <boomanaiden154> https://github.com/alpinelinux/docker-alpine/blob/master/prepare-branch.sh
2022-01-01 02:19:48 <boomanaiden154> Shoot. I missed that Ikke answered the same exact question.
2022-01-01 02:32:32 <Hello71> also they aren't here?
2022-01-01 03:35:39 <boomanaiden154> That definitely doesn't help either.
2022-01-01 07:04:39 <ktprograms> lol the MR for removing aports-qa-bot stale messages has _just_ been marked as stale by aports-qa-bot.
2022-01-01 07:09:32 <ikke> That MR itself is not really doing what it's intended though
2022-01-01 07:11:42 <ktprograms> ikke: Why isn't it?
2022-01-01 07:12:11 <ikke> Just removing the configuration does not disable it
2022-01-01 07:12:28 <ikke> It just overrides the default config
2022-01-01 07:16:49 <ktprograms> ikke: Oh, I see
2022-01-01 07:17:26 <ikke> So it does not really make sense to merge it, but right now it kind-of acts as an issue
2022-01-01 07:48:30 <oneinsect> happt new year fellas
2022-01-01 07:48:48 <oneinsect> :-D
2022-01-01 16:36:49 <psykose> ikke: could you merge !28854
2022-01-01 16:52:58 <ikke> done
2022-01-01 16:57:28 <psykose> thankies
2022-01-01 17:41:43 <omni> I'm havnig trouble reaching both gitlab and the main page over IPv6
2022-01-01 17:45:06 <psykose> it is being fixed™️
2022-01-01 18:29:08 <exprez135> I've noticed that running `apk list --upgradable` doesn't work, while its counterpart `apk list -u` does. But using "--upgradeable" does work.
2022-01-01 18:31:43 <mps> Ariadne: ^ need change in helo
2022-01-01 18:31:57 <mps> fabled: ^
2022-01-01 18:32:11 <mps> s/helo/help/
2022-01-01 18:32:11 <alpine-meetbot> mps meant to say: Ariadne: ^ need change in help
2022-01-01 18:34:56 <Ermine> Where can I find sources for alpine-meetbot?
2022-01-01 18:46:01 <fabled> mps: fixed in apk 2.12.8+. Need to upgrade apk in edge...
2022-01-01 18:46:52 <mps> fabled: ok, thanks for fast info
2022-01-01 18:47:23 <ikke> Ermine: It's sopel
2022-01-01 18:47:35 <ikke> https://sopel.chat/
2022-01-01 18:48:10 <Ermine> ikke: thank you!
2022-01-01 20:01:04 <mps> what is diff for cmake option CMAKE_BUILD_TYPE MinSizeRel and None?
2022-01-01 20:01:34 <ikke> None regularl triggers debug builds
2022-01-01 20:01:40 <ikke> regularly*
2022-01-01 20:01:42 <psykose> depends on the cmakefiles of the project
2022-01-01 20:02:07 <psykose> tiny projects with minimal cmake files? identical, except minsizerel will also have -DNDEBUG on the flag line
2022-01-01 20:02:13 <psykose> anything else, project dependent on what they are doing
2022-01-01 20:03:41 <mps> I'm asking for telegram-desktop in particular
2022-01-01 20:03:44 <psykose> anything but the simplest cmake projects usually start doing various combinations off adding special flags to only non-none types, defaulting to debug if it's not a release type, having debug code gated behind #ifndef NDEBUG, etc
2022-01-01 20:04:03 <mps> just built it with None and it works
2022-01-01 20:04:22 <mps> and rebuilt with qt6
2022-01-01 20:04:52 <ikke> It's not a matter of working vs not working
2022-01-01 20:04:58 <mps> ok, I will create draft MR so other people can comment
2022-01-01 20:05:27 <mps> ikke: iiuc your advice is to use MinSizeRel
2022-01-01 20:05:53 <ikke> mps: mariadb was built with debug flags in production because we defaulted to None
2022-01-01 20:06:00 <psykose> for telegram desktop it makes no difference
2022-01-01 20:06:12 <psykose> aside from the implied -DNDEBUG the generation is the same
2022-01-01 20:06:28 <mps> psykose: what is your advice then
2022-01-01 20:06:44 <ikke> MinSizeRel corresponds with use providing -Os
2022-01-01 20:06:46 <psykose> my advice is to default to MinSizeRel and add exceptions for things that prefer something else
2022-01-01 20:06:59 <ikke> us*
2022-01-01 20:07:10 <psykose> for many things i am changing it is in fact identical for both variations, but i prefer being proactive
2022-01-01 20:07:36 <mps> ok, thank you, I will keep MinSizeRel then
2022-01-01 20:07:37 <psykose> there are probably some dozen or more things left that have slightly degraded builds or none that we just don't know about
2022-01-01 20:08:07 <psykose> every cmake project sadly needs a magnifier glass to decide what is correct, it's just minsizerel is the most correct default you can pick
2022-01-01 20:08:12 <mps> lets test on armv7
2022-01-01 20:08:30 <mps> yes, cmake is bad
2022-01-01 20:08:33 <psykose> or well, release has even fewer issues, but release is -O3 as a default without project overrides
2022-01-01 20:08:40 <psykose> and everything goes up 50% in size :p
2022-01-01 20:11:50 <mps> oh no, we don't have msgsl on armv7
2022-01-01 20:23:20 <Nulo> I've sadly ended up in a conflict about Alpine. The CoC suggests to "work through the conflict using a neutral third party in a transparent manner", how would that work? Thanks
2022-01-01 20:23:55 <mps> use common sense
2022-01-01 20:24:45 <bl4ckb0ne> anybody tried to package signal-desktop?
2022-01-01 20:25:47 <mps> when I joined alpine I'm told by 'elders' to use common sense as best thing
2022-01-01 20:26:06 <Nulo> bl4ckb0ne, I can't find any issue or merge request so I guess not
2022-01-01 20:26:19 <bl4ckb0ne> yeah, thats what i went to check first 
2022-01-01 20:26:28 <jvoisin> bl4ckb0ne: it's horrible iirc
2022-01-01 20:26:34 <jvoisin> the Debian people tried at some point
2022-01-01 20:26:36 <bl4ckb0ne> wanted to know before i started it if it was a definite no-go or nobody bothered
2022-01-01 20:26:44 <bl4ckb0ne> i tried locally but hit typescript issues quickly
2022-01-01 20:27:01 <bl4ckb0ne> the arch package has a bunch of sed in the config
2022-01-01 20:27:03 <jvoisin> more like "welcome to hell, here are a ton of horrible dependencies, some are outdated, glhf"
2022-01-01 20:27:11 <bl4ckb0ne> innit
2022-01-01 20:27:27 <Nulo> https://gitlab.alpinelinux.org/alpine/aports/-/issues/8634 
2022-01-01 20:27:31 <Nulo> > Sadly electron upstream isn't interested in supporting musl, so this would be too much of a maintenance effort for us.
2022-01-01 20:27:38 <Nulo> And https://gitlab.alpinelinux.org/alpine/aports/-/merge_requests/25028
2022-01-01 20:27:44 <Nulo> !25028
2022-01-01 20:27:48 <Nulo> #8634
2022-01-01 20:29:11 <bl4ckb0ne> so loads of fun ahead
2022-01-01 20:29:42 <psykose> afaik electron isn't many more patches than supporting chromium
2022-01-01 20:30:02 <psykose> but i do not have a computer capable of actually building chromium, so i don't care much about doing it :)
2022-01-01 20:32:15 <psykose> Nulo: pick a neutral third party developer and message them i guess
2022-01-01 20:33:15 <Nulo> psykose, well, I guess I don't know enough people here to know who would be up to something like that, that's why I asked here.
2022-01-01 20:33:31 <mps> psykose: I don't see any reason to have chromium in alpine
2022-01-01 20:33:35 <Nulo> I don't want to randomly message people and ask "do you want to take part in a ~conflict~"?
2022-01-01 20:33:43 <Nulo> s/"?/?"
2022-01-01 20:33:43 <alpine-meetbot> Nulo meant to say: I don't want to randomly message people and ask "do you want to take part in a ~conflict~?"
2022-01-01 20:34:20 <psykose> Nulo: ah, but you should
2022-01-01 20:34:40 <psykose> or just open an issue with details in it
2022-01-01 20:34:49 <Nulo> Ah, that could work. I'll do that, thanks
2022-01-01 20:34:49 <mps> simple, create issue report
2022-01-01 20:35:01 <mps> psykose: heh
2022-01-01 20:38:02 <skarnet> psykose: do you want to take part in a conflict?
2022-01-01 20:38:17 <psykose> only if it's with you on my side
2022-01-01 20:38:25 <skarnet> Nulo: that's how you do it
2022-01-01 20:38:50 <Nulo> :P
2022-01-01 20:55:51 <Ariadne> hi
2022-01-01 20:58:28 <psykose> hello
2022-01-01 22:18:47 <omni> gitlab CI is half way though rebuilding community aports for !29060, when it's done I'll fix things and rebase, please tell me if you catch something I missed
2022-01-01 22:19:53 <omni> then I'll open a separate MR for aports in testing and add community aports in need of rebuild for !29061
2022-01-01 23:26:21 <omni> nmeum: android-tools fail to build
2022-01-01 23:32:59 <psykose> needs a patch in vendor/boringssl/CMakeLists.txt for werror
2022-01-01 23:35:11 <psykose> https://img.ayaya.dev/T0np4YK4DwfL
2022-01-02 00:46:52 <omni> thanks
2022-01-02 08:48:06 <anjan> just tried out jami
2022-01-02 08:48:20 <anjan> cause jitsi has not worked for the last 3 nights Ive tried to video call
2022-01-02 08:48:40 <anjan> and jami is super good. Probably the best video calling program Ive used (including proprietary video calling)
2022-01-02 08:48:52 <anjan> gonna see if I can get it packaged for alpine hehe
2022-01-02 08:54:11 <z3ntu> Can someone please review (and merge?) https://gitlab.alpinelinux.org/alpine/aports/-/merge_requests/27956 Thanks
2022-01-02 08:55:39 <ikke> Will check in a bit
2022-01-02 08:58:55 <ikke> z3ntu: can you rebase it? Apparently there is a conflict
2022-01-02 08:59:07 <ikke> "Merge blocked: fast-forward merge is not possible. To merge this request, first rebase locally. "
2022-01-02 09:02:25 <z3ntu> ikke: done
2022-01-02 09:02:51 <z3ntu> git took around 2.5 minutes to push it, no clue why it's so slow lately
2022-01-02 09:41:52 <ddevault> an unhinged rant landed in my inbox, which can be mostly ignored
2022-01-02 09:42:02 <ddevault> but one thing which it drew my attention to was the fact that the alpine wiki has an unusual copyright policy
2022-01-02 09:42:06 <ddevault> https://wiki.alpinelinux.org/wiki/Alpine_Linux:Privacy_policy#Copyright
2022-01-02 09:42:12 <anjan> whats alpine's replacement for systemd-libs?
2022-01-02 09:42:13 <ddevault> we should consider moving to creative commons or the GNU FDL
2022-01-02 09:42:26 <anjan> I keep getting 
2022-01-02 09:42:28 <anjan> configure: error: Missing libudev development files
2022-01-02 09:43:54 <ikke> ddevault: The question is whether we can change it after the fact
2022-01-02 09:44:06 <ddevault> aye, this presents a problem, but one which has been solved before
2022-01-02 09:44:19 <ddevault> I think the best approach is to set the policy that all edits after $DATE use a new license
2022-01-02 09:44:32 <ddevault> then collect voluntary agreements from copyright holders to retroactively relicense their contributions
2022-01-02 09:44:51 <ddevault> then the wiki becomes better licensed over time and the copyright status of any particular page can be ascertained should someone need to
2022-01-02 09:45:09 <anjan> ay, the pmOS website has a similar issue
2022-01-02 09:45:24 <anjan> Thats a good recommendation. Ill add that to the gitlab issue
2022-01-02 10:02:51 <ddevault> moved to the ML
2022-01-02 10:04:02 <anjan> I figured out the systemd-libs thing
2022-01-02 15:09:39 <omni> can I fetch CI or builder cached tarballs somewhere?
2022-01-02 15:10:07 <omni> upstream containerd tarball seem to have changed and I want to compare
2022-01-02 15:10:51 <omni> hmm.. it's probably not cached by CI since checksum fail there
2022-01-02 15:16:31 <ikke> omni: https://distfiles.alpinelinux.org/distfiles/
2022-01-02 15:44:15 <omni> thanks!
2022-01-02 16:00:22 <omni> https://tpaste.us/Qrgz
2022-01-02 16:06:53 <omni> will the builders prefer the already cached tarball?
2022-01-02 16:07:46 <ikke> yes
2022-01-02 16:09:09 <omni> so no need of updating the sha512sum for containerd in !29060 then
2022-01-02 16:09:21 <omni> it changed like 10-15 hours ago
2022-01-02 16:14:46 <omni> z3ntu: pushing and pulling is really slow for me too
2022-01-02 16:30:40 <ikke> it would be good to have the hash match what upstream has
2022-01-02 16:30:59 <ikke> otherwise it would fail for others who try to build it
2022-01-02 16:31:37 <ikke> I can remove it from distfiles
2022-01-02 17:21:08 <omni> Ok, let's do it like that instead
2022-01-02 17:21:45 <omni> no idea why that was changed upstream though
2022-01-02 17:27:08 <mla> few wks old, but these are ready: https://gitlab.alpinelinux.org/alpine/aports/-/merge_requests/28432 , https://gitlab.alpinelinux.org/alpine/aports/-/merge_requests/28432 , https://gitlab.alpinelinux.org/alpine/aports/-/merge_requests/28308
2022-01-02 17:27:30 <mla> wonder if someone w merge privs could review / add feedback?
2022-01-02 17:29:37 <mla> https://gitlab.alpinelinux.org/alpine/aports/-/merge_requests/28309
2022-01-02 17:34:05 <omni> I just waited 7 minutes for a 'git pull' to say "Already up to date." on the master branch
2022-01-02 17:34:44 <ikke> huh
2022-01-02 17:35:04 <omni> just saying since someone else also mentioned things being slow
2022-01-02 17:35:12 <ikke> I just ran git pull without issue
2022-01-02 17:37:44 <omni> could be unrelated issue and something at my end, but pulling from github and kernel.org feel reasonable
2022-01-02 17:37:58 <ikke> Do you have ipv6?
2022-01-02 17:39:07 <omni> I am specifically not using IPv6 for gitlab.alpinelinux.org and alpinelinux.org since it wasn't working
2022-01-02 17:53:32 <psykose> ipv4 actions are fine for me but ipv6 is still busted for me
2022-01-02 18:03:04 <ikke> yeah, that is known, need to fix ssh on ipv6
2022-01-02 18:14:33 <omni> I have issues over ipv6 with both gitlab.alpinelinux.org and alpinelinux.org on port 443
2022-01-02 20:23:21 <z3ntu> https pulling works fine on my end, but ssh with ipv6 is super slow, 1 commit pushes take about 2-3 minutes
2022-01-02 20:23:30 <ikke> strange
2022-01-02 20:23:48 <z3ntu> `debug1: Connecting to gitlab.alpinelinux.org [2a01:7e01:e001:15a:1::2] port 22.` < definitely using ipv6
2022-01-02 20:24:01 <ikke> well, ssh over ipv6 does not work yet
2022-01-02 20:24:26 <ikke> Does it work if you set AddressFamily inet for gitlab.alpinelinux.org in .ssh/config?
2022-01-02 20:25:07 <z3ntu> aaah ssh times out on ipv6 and retries on ipv4 after a while
2022-01-02 20:25:17 <z3ntu> debug1: connect to address 2a01:7e01:e001:15a:1::2 port 22: Connection timed out
2022-01-02 20:25:17 <z3ntu> debug1: Connecting to gitlab.alpinelinux.org [172.105.69.85] port 22.
2022-01-02 20:26:17 <ikke> I have a solution, just need to implement it :P
2022-01-02 20:26:31 <z3ntu> and yeah "AddressFamily inet" seems to force ipv4 and it works instantly with that
2022-01-02 20:28:06 <dalias> :/
2022-01-02 20:28:19 <ikke> dalias: what?
2022-01-02 20:30:53 <dalias> ipv6 being broken
2022-01-02 20:31:39 <ikke> Well, it's step one of enabling ipv6 in the first place
2022-01-02 20:32:43 <dalias> :)
2022-01-02 20:57:18 <minimal> ikke: like AA? step one is admitting you have a problem? :-)
2022-01-02 20:58:20 <minimal> step 2 is enabling IPv6, step 3 is realising you now have multiple problems (SLAAC, DHCPv6, RA etc) lol
2022-01-02 20:58:56 <tobikoch> Uhm, where is it broken?
2022-01-02 21:00:37 <ikke> The ipv6 endpoint for http(s) is different from ssh
2022-01-02 21:02:00 <tobikoch> Ah
2022-01-02 21:32:01 <psykose> x86 builder seems stuck
2022-01-02 21:47:25 <ikke> restarted
2022-01-02 21:48:19 <psykose> hmm
2022-01-02 21:48:27 <psykose> that was definitely not something that ever failed on ci
2022-01-02 21:48:28 <psykose> classic
2022-01-02 23:12:16 <anjan> psykose: I need to vendorize pjproject
2022-01-02 23:12:25 <anjan> with jami's fork
2022-01-02 23:12:30 <anjan> they have a bunch of patches
2022-01-02 23:13:04 <anjan> but it it seems that the build system cant find it even though I followed arch's PKGBUILD: https://github.com/archlinux/svntogit-community/blob/packages/jami-daemon/trunk/PKGBUILD
2022-01-02 23:13:08 <anjan> any help?
2022-01-02 23:17:54 <psykose> looking at it
2022-01-02 23:18:02 <psykose> but you are missing all those dependencies as well
2022-01-02 23:20:03 <anjan> Ya, I want to get pjproject to load first psykose 
2022-01-02 23:20:12 <anjan> the build system cant find it
2022-01-02 23:20:37 <anjan> and if I load alpine's pjproject-dev I get a build error because it needs a jami patch
2022-01-02 23:23:41 <psykose> there is nothing i can see in the output that indicates it can't find it, just a missing include somewhere
2022-01-02 23:23:56 <psykose> and since this is autotools mess for all i know it's caused by missing dependencies somewhere
2022-01-03 00:39:27 <tomalok> something weird going on with the website?   seems to take a long time to get https://alpinelinux.org/releases.json from python urlopen()...
2022-01-03 00:42:06 <boomanaiden154> tomalok: Loading fine for me.
2022-01-03 00:51:07 <tomalok> via browser it works fine, just not from a python script
2022-01-03 00:55:39 <psykose> it works with ipv6 disabled for some reason
2022-01-03 00:56:20 <boomanaiden154> psykose: Why wouldn't it work with ipv6 disabled? All of the Alpine infrastructure fully supports ipv4 as far as I'm aware.
2022-01-03 00:56:59 <psykose> huh?
2022-01-03 00:57:07 <psykose> i mean it works fine if you specify ipv4
2022-01-03 00:57:20 <psykose> so it hangs it hangs in urlopen() since it probably tries ipv6 first
2022-01-03 00:57:34 <psykose> manually doing curl -6 also hangs
2022-01-03 00:57:38 <boomanaiden154> Oh. I read that wrong. Sorry.
2022-01-03 00:57:52 <boomanaiden154> Seems likely that ipv6 would be the issue then.
2022-01-03 00:58:55 <psykose> yes, seems there are more ipv6 issues than just ssh
2022-01-03 00:59:38 <omni> but they surfaced a bit later
2022-01-03 00:59:42 <psykose> can't even ping the addresses
2022-01-03 01:13:33 <tomalok> it tries IPv6 first, times out, then tries IPv4...  gave urlopen() a timeout= parameter, so at least it's not hanging for an extended period of time 
2022-01-03 01:15:52 <boomanaiden154> tomalok: You could probably just use the ipv4 address instead of the domain and set the host header manually and then there shouldn’t be any timeout.
2022-01-03 01:16:50 <minimal> tomalok: perhaps related to this apparently not yet addressed? https://github.com/urllib3/urllib3/pull/1000
2022-01-03 01:18:07 <Hello71> boomanaiden154: no because https
2022-01-03 01:19:47 <boomanaiden154> Didn’t even think about that.
2022-01-03 02:34:49 <an3223> i'm packaging ccv ( https://libccv.org/ ). the last release is from 2016, but a lot of development has happened since then. should i make a normal package and a _git package?
2022-01-03 02:40:18 <ktprograms> an3223: I think the general practice is that if the current code (meaning git master) is stable and has no major bugs, then jsut use that and there's no need for a package built from an old release
2022-01-03 02:43:19 <an3223> ktprograms: okay thanks, i will do that
2022-01-03 08:47:03 <nmeum> Ariadne: any idea how we could fix #13369? looks to me like we need a custom gcc/config/t-linux which pins MULTILIB_OSDIRNAMES to lib? I am asking because you already patched these files for other architectures so I am hoping you might have some clues on this
2022-01-03 08:49:34 <psykose> am i blind or is the x86_64 one also :/lib/../lib/:/usr/lib/../lib/:, doing lib before usr/lib just like in the riscv line
2022-01-03 08:50:04 <psykose> (since you mention usr/lib before lib)
2022-01-03 08:51:02 <psykose> oh, the huge amount of .. on the one before is /usr/lib
2022-01-03 08:51:10 <psykose> god, why do these paths even exist
2022-01-03 08:51:32 <nmeum> yes, /usr/lib/gcc/x86_64-alpine-linux-musl/11.2.1/../../../../lib/ is /usr/lib
2022-01-03 09:16:57 <Ariadne> nmeum: yes that seems right
2022-01-03 09:19:17 <Ariadne> nmeum: although i have not patched gcc/config/t-linux for any architecture :)
2022-01-03 09:19:23 <psykose> another more funny solution is to replace -static with -staticdev, and duplicate the dev files for it :p
2022-01-03 09:19:42 <psykose> that way you can force static linking for this instead of needing a solid searchpath
2022-01-03 09:21:21 <Ariadne> its gcc/config/riscv64/t-linux that needs to be fixed
2022-01-03 09:22:43 <nmeum> yes, we need to fix that file it also causes other issues, e.g. gcc-go is broken on riscv64 for the same reason currently
2022-01-03 09:29:31 <nmeum> I guess we can just take config/riscv/t-linux-multilib and just pin MULTILIB_DIRNAMES/MULTILIB_OSDIRNAMES to ../lib, use this for config/riscv/t-linux and we are done?
2022-01-03 09:30:27 <nmeum> though the gcc <-> musl multilib stuff is slightly beyond me
2022-01-03 09:43:21 <ddevault> python is the actual worst
2022-01-03 09:43:33 <Ariadne> nmeum: try https://tpaste.us/Bgy9
2022-01-03 09:44:04 <psykose> what's broken in python land this time
2022-01-03 09:44:28 <Ariadne> weasyprint broke, that wasn't fun
2022-01-03 09:44:35 <ddevault> some packages I depend on pin py3-setuptools to a very narrow range of versions ahead of the alpine package in order to support a godawful hack they "need"
2022-01-03 09:44:40 <ddevault> pinned*
2022-01-03 09:44:56 <Ariadne> ugh
2022-01-03 09:45:00 <nmeum> Ariadne: tried something similar already, the problem is that MULTILIB_OPTIONS is empty when building with --disable-multilib so you need to set that as well otherwise gcc won't pick up MULTILIB_OSDIRNAMES
2022-01-03 09:45:08 <nmeum> I am currently trying https://tpaste.us/NM1k
2022-01-03 09:45:46 <Ariadne> nmeum: yes, i'm not sure how these options map to risc-v
2022-01-03 09:46:02 <nmeum> it just takes ages to rebuild gcc on riscv64 and since this is a build system config file you can't really do incremental builds so I need to wait half a day to test changes :S
2022-01-03 09:46:24 <Ariadne> yocto has something really weird for riscv
2022-01-03 09:46:37 <Ariadne> https://git.yoctoproject.org/poky/tree/meta/recipes-devtools/gcc/gcc/0004-64-bit-multilib-hack.patch
2022-01-03 09:46:42 <Ariadne> i'm not sure what to make of it
2022-01-03 09:46:49 <nmeum> that looks fun
2022-01-03 09:46:54 <Ariadne> it seems totally wrong for our toolchain
2022-01-03 09:47:21 <nmeum> yes
2022-01-03 09:48:13 <psykose> maybe it's time to bump setuptools then :^)
2022-01-03 09:48:16 <Ariadne> the original patch that kills multilib, we got it from gentoo patch series
2022-01-03 09:48:32 <Ariadne> before riscv was added
2022-01-03 09:48:36 <psykose> it would bring quite the chaos on random python packages as a result as they get bumped later thouggh
2022-01-03 09:48:40 <Ariadne> and gentoo does not have the patch anymore
2022-01-03 09:48:46 <Ariadne> they "support" multilib now, i guess
2022-01-03 09:49:00 <nmeum> Ariadne: are you referring to 0024-use-pure-64-bit-configuration-where-appropriate.patch?
2022-01-03 09:49:04 <Ariadne> yes
2022-01-03 09:49:11 <Ariadne> we've had that one forever
2022-01-03 09:50:03 <psykose> and our tag of setuptools /is/ a year old, i guess eventually it has to be done anyway breakage or not
2022-01-03 09:50:05 <Ariadne> most of our patches are a mix of debian and gentoo patches that i've rebased
2022-01-03 09:50:39 <nmeum> there is also an gcc upstream issue regarding musl multilib handling but upstream doesn't seem to be actively working on this https://gcc.gnu.org/bugzilla/show_bug.cgi?id=90077
2022-01-03 10:09:52 <psykose> i can seemingly rebuild all of main/ with setuptools 60.2.0, arch seems to use 57.4.0 with 59.1.0 in testing, maybe it's somehow doable to bump it to 57.4.0 at least
2022-01-03 10:27:11 <ddevault> celery pinned setuptools to >=59.1.1,<59.7.0, which is a range of releases spanning 28 days
2022-01-03 10:28:34 <psykose> that is funny
2022-01-03 10:28:40 <ddevault> no
2022-01-03 10:28:42 <ddevault> it is tragic
2022-01-03 10:29:04 <psykose> if you.. wink wink nudge nudge modify the constraint does it actually fail to build
2022-01-03 10:29:14 <ddevault> no, and that's exactly what I did
2022-01-03 10:29:19 <psykose> phenomenal
2022-01-03 10:29:28 <ddevault> they did this for the sake of a very very stupid, shitty hack
2022-01-03 10:29:34 <ddevault> which is totally supported by non-hacky means and which they chose not to fix
2022-01-03 10:33:00 <ddevault> I cannot get python out of my dependency tree soon enough
2022-01-03 13:46:29 <omni> https://lwn.net/ml/linux-kernel/YdIfz+LMewetSaEB@gmail.com/
2022-01-03 13:47:20 <jvoisin> almost as if clean code had benefits beyond readability
2022-01-03 14:04:00 <skarnet> who would have thought
2022-01-03 14:16:32 <skarnet> the most hilarious part is where he says precompiled headers only brought a 1-2% improvement, compared to 78% and up for clean header rewriting
2022-01-03 14:17:55 <ktprograms> skarnet: How actually can headers be precompiled?
2022-01-03 14:22:01 <jvoisin> ktprograms: https://en.wikipedia.org/wiki/Precompiled_header
2022-01-03 14:22:02 <alpine-meetbot> [WIKIPEDIA] Precompiled header | "In computer programming, a precompiled header (PCH) is a (C or C++) header file that is compiled into an intermediate form that is faster to process for the compiler. Usage of precompiled headers may significantly reduce compilation time, especially when applied to large header files, header files that..."
2022-01-03 14:23:16 <skarnet> it's yet another gcc mutant feature
2022-01-03 14:26:36 <ktprograms> skarnet: mutant?
2022-01-03 14:26:45 <ktprograms> Thanks for the link jvoisin 
2022-01-03 14:27:36 <skarnet> ktprograms: you know, uncontrolled unethical experiments. :)
2022-01-03 14:28:23 <ktprograms> skarnet: lol I see.
2022-01-03 14:31:58 <Hello71> but clang and msvc also have it
2022-01-03 15:04:43 <skarnet> still mutant
2022-01-03 15:05:26 <skarnet> (and, as the fast-headers experiment shows, ridiculously ineffective)
2022-01-03 15:11:19 <andypost[m]> hi, I'm getting slow pull/fetch from gitlab because ipv6, meantime fixed it by using git pull -4 
2022-01-03 16:06:34 <ikke> andypost[m]: yes, will fix that this week
2022-01-03 18:22:19 <ddevault> any tips for alpine installs on RAM-constrained systems?
2022-01-03 18:29:53 <jvoisin> I'm running Alpine on 32M systems, so far so good
2022-01-03 18:30:08 <jvoisin> apk doesn't like it very much iirc, for upgrades
2022-01-03 18:34:34 <ddevault> attempted to use setup-alpine to get it started
2022-01-03 18:34:36 <ddevault> but it ran out of space
2022-01-03 18:34:40 <ddevault> since it works in RAM
2022-01-03 18:43:00 <omni> ddevault: you could try zstd compressed ramdisk for swap with zram-init
2022-01-03 18:44:28 <omni> (or other compression algorithm)
2022-01-03 18:45:06 <ddevault> I think it would be best to set up the install media on a different machine
2022-01-03 18:45:10 <ddevault> possibly via magic/hacks
2022-01-03 19:01:32 <skarnet> ddevault: sanity check: you have your rootfs on a real disk, not a tmpfs with modloop and all the related (RAM-eating) hacks?
2022-01-03 19:02:47 <ddevault> just following the raspberry pi installation guide on the wiki, but on the original pi hardware
2022-01-03 19:03:00 <ddevault> which has 256M RAM (shared with the GPU)
2022-01-03 19:03:45 <ddevault> I'm sure I can figure it out by doing it all manually from a second machine
2022-01-03 19:04:00 <skarnet> the /-on-tmpfs setup sacrifices ram to save disk, so if your ram is constrained that's where I'd start
2022-01-03 19:05:18 <ddevault> yeah
2022-01-03 19:05:37 <a16bitsysop> If it’s a pi you can try this: https://gitlab.com/raspi-alpine/builder it doesn’t run from ram, but rootfs in readonly.  With a bit of tweaking you can make root r/w
2022-01-03 19:05:37 <alpine-meetbot> No such subreddit.
2022-01-03 19:06:00 <ddevault> ...is it really necessary to have a reddit bot
2022-01-03 19:06:07 <ddevault> thanks a16bitsysop 
2022-01-03 19:07:06 <omni> r/testing-alpine-meetbot
2022-01-03 19:07:06 <alpine-meetbot> No such subreddit.
2022-01-03 19:43:26 <donoban> Hi, I tried to fix the problem with mkusers without copying the host passwd/group to the chroot, but it was difficult due write permissions and really it doesn't matters very much. It seems enough with https://gitlab.alpinelinux.org/donoban/abuild/-/merge_requests/1
2022-01-03 20:22:11 <ikke> --cache-dir /etc/apk/cache seems redundant to me. Doesn't it default to /etc/apk/cache already? Or does it follow the symlink and use that somehow in the chroot?
2022-01-03 20:30:57 <mps> I lost two days trying to boot alpine from internal nvme on apple M1 machine with grub and 'found' solution using grub efi from debian
2022-01-03 20:31:42 <mps> not sure is 'our' aarch64 grub is buggy or I used it incorrectly
2022-01-03 20:33:19 <mps> I don't have much experience with grub (last time I used and tweaked it was 15 years ago) so could anyone who use grub on nvme on alpine could shed some light
2022-01-03 20:34:25 <minimal> mps: what exactly was the problem/error?
2022-01-03 20:35:07 <mps> minimal: with alpine grub only what I get is 'grub-rescue' prompt
2022-01-03 20:35:52 <minimal> "alpine grub only"? have have to install the grub-efi package as well, its got the UEFI support files
2022-01-03 20:36:02 <mps> minimal: strange thing is that it works 'out-of-the box' when I boot from usb disk
2022-01-03 20:36:23 <mps> minimal: trust me, I installed everythig
2022-01-03 20:36:51 <mps> problem is only with internal nvme
2022-01-03 20:36:56 <minimal> so what about the generated grub.cfg file, did it look correct?
2022-01-03 20:37:35 <mps> yes, because grub.cfg worked fine when I installed files from debian install iso
2022-01-03 20:40:01 <minimal> mps: confused - using Grub to boot an ISO would involve different contents for grub.cfg than when using Grub to boot from NVME
2022-01-03 20:40:32 <mps> minimal: no, I didn't boot iso, just copied needed files from it to ESP
2022-01-03 20:41:09 <mps> and M1 doesn't have efi firmware, it uses this what u-boot 'have'
2022-01-03 20:43:06 <minimal> mps: ok, so its *NOT* EFI then why are you trying to use Grub? Grub supports BIOS/MBR and UEFI booting
2022-01-03 20:43:57 <mps> minimal: well, not so, machine could be boot only with efi provided by u-boot
2022-01-03 20:44:03 <mps> efi payload
2022-01-03 20:45:12 <minimal> ok, so how is the grub.cfg file created?
2022-01-03 20:45:39 <mps> minimal: by hand
2022-01-03 20:46:15 <mps> not really, by copiyng from usb disk and adapted by hands
2022-01-03 20:46:19 <minimal> ah - that will be fun :-)
2022-01-03 20:46:32 <minimal> normally grub-probe/grub-mkconfig etc take care of that for you
2022-01-03 20:46:45 <minimal> inserting things like UUIDs in the config etc
2022-01-03 20:48:09 <mps> minimal: I forgot fine points of grub but I remember basic things like UUID, PARTUUID and similar :)
2022-01-03 20:48:31 <mps> it is not these kind of problem
2022-01-03 20:48:58 <minimal> so you'll need to work out which Grub modules you need to load in your config - typically for UEFI this would be things like "insmod efi_gop" and "insmod efi_uga" for graphics as, unlike BIOS, UEFI has no text console
2022-01-03 20:49:04 <mps> it can't find grub.cfg on rootFS or ESP with alpine grub
2022-01-03 20:49:53 <mps> minimal: I didn't changed grub.cfg at all to get it boot with debian grub
2022-01-03 20:50:25 <mps> I suspect 'our' grub have problem but have no idea how to debug it
2022-01-03 20:51:43 <mps> (hmm, 'our' now sounds strange to me)
2022-01-03 20:51:47 <minimal> does the u-boot efi payload support nvme?
2022-01-03 20:52:10 <mps> minimal: sure, else how it will boot debian grub
2022-01-03 20:52:11 <minimal> can you show the grub.cfg contents?
2022-01-03 20:52:35 <mps> minimal: I can, but it will be useless to look at it
2022-01-03 20:52:46 <minimal> why?
2022-01-03 20:53:11 <mps> https://tpaste.us/ovWY
2022-01-03 20:53:47 <mps> minimal: because same grub.cfg work with debian but not with alpine grub
2022-01-03 20:54:02 <mps> it is on above url
2022-01-03 20:54:55 <mps> minimal: anyway I think you understand this a lot better than me, so forgive me my 'bad wording'
2022-01-03 20:57:58 <donoban> ikke: without it, it doesn't use the cache. it runs outside the chroot, when abuild starts inside the chroot all deps are already installed
2022-01-03 20:57:59 <mps> minimal: one other person reported that also Arch linux grub works
2022-01-03 20:59:11 <mps> minimal: this could shed light https://wiki.debian.org/UEFI#Force_grub-efi_installation_to_the_removable_media_path
2022-01-03 20:59:13 <minimal> there was a discussion a couple of days ago where the nvme kernel module wasn't being loaded, think adding "modules=nvme" to the boot cmdline was a workaround
2022-01-03 20:59:44 <mps> minimal: nmve is compiled in kernel, not as module
2022-01-03 21:00:52 <donoban> I also tested if removing the '/etc/apk/cache' symlink it tries to create a new folder or something but it just ignores it, so it just uses the cache if the user has already is using it
2022-01-03 21:00:57 <minimal> mps: and "a5889c8c-7e35-460b-a7f4-3866c192d46f" is the correct UUID for boot/root fs?
2022-01-03 21:01:48 <mps> minimal: for now I boot only first option, other doesn't have kernels
2022-01-03 21:02:12 <mps> i.e. root=/dev/nvme0n1p5
2022-01-03 21:02:57 <donoban> lol, I did the MR against my own repository
2022-01-03 21:03:59 <mps> minimal: also here are short notes from another person related to grub and debian https://tg.st/u/asahi-uboot-quickstart.txt
2022-01-03 21:04:43 <mps> this 'cp /boot/efi/EFI/debian/grubaa64.efi /boot/efi/EFI/boot/bootaa64.efi' was 'main trick'
2022-01-03 21:05:48 <mps> s/main/master/
2022-01-03 21:05:48 <alpine-meetbot> mps meant to say: this 'cp /boot/efi/EFI/debian/grubaa64.efi /boot/efi/EFI/boot/bootaa64.efi' was 'master trick'
2022-01-03 21:05:53 <mps> :)
2022-01-03 21:08:21 <mps> minimal: this is not urgent to solve for me, just wanted to ask if someone else 'hit' this problem on aarch64 with nvme to know do 'we' have bug in grub
2022-01-03 21:08:54 <donoban> ikke: maybe abuild-apk doesn't use /etc/apk/cache as default cache?
2022-01-03 21:09:55 <donoban> uhM it's just a symlink to abuild-sudo
2022-01-03 21:22:34 <minimal> mps: (was called away) I was referring to the "search --no-floppy --fs-uuid --set=root a5889c8c-7e35-460b-a7f4-3866c192d46f" earlier in the grub.cfg file
2022-01-03 21:23:30 <minimal> mps: the cp to bootaa64.efi is so that it uses UEFI fallback mode, i.e. does not rely on UEFI variables being set
2022-01-03 21:28:33 <ikke> donoban: abuild-apk should just be a wrapper around abuild
2022-01-03 21:32:13 <mps> minimal: something like this, which I'm not sure I understand fully
2022-01-03 21:32:16 <donoban> so maybe the --root option makes the default '/etc/apk/cache' points to the chroot/etc/apk/cache
2022-01-03 21:32:26 <donoban> which doesn't exist
2022-01-03 21:33:12 <psykose> that sounds likely
2022-01-03 21:33:26 <mps> minimal: I think grub.cfg content is not problem but that grub didn't found it
2022-01-03 21:34:38 <mps> minimal: but I'm too tired now to think anything seriously
2022-01-03 21:35:02 <mps> minimal: and it is not urgent as I told above
2022-01-03 21:37:04 <mps> good thing is that I got u-boot, grub, kernel and alpine working fine on M1 (even patched keyboard driver to have Fkeys on machine which have touchbar emulating them with number keys and Fn key)
2022-01-03 21:37:19 <mps> minimal: thank you for help
2022-01-03 21:38:01 <mps> (now is the time for red wine and later for bed)
2022-01-03 21:38:40 <donoban> hehe it sounds great mps (the M1 and alpine working ;) )
2022-01-03 21:39:52 <mps> donoban: yes, also wifi works fine, only audio is not fixed yet
2022-01-03 23:22:02 <psykose> could someone with merge access press the magic button on !29151
2022-01-03 23:23:27 <HRio> psykose: done
2022-01-03 23:24:28 <psykose> thanks :)
2022-01-04 01:42:27 <bl4ckb0ne> is it normal that alpine-base depends on dbus-openrc?
2022-01-04 01:42:35 <bl4ckb0ne> dep isnt listed with apk info -R
2022-01-04 01:44:23 <psykose> it says that but it's not an actual dependency it seems
2022-01-04 01:44:28 <psykose> if you remove `dbus` it goes away
2022-01-04 10:09:17 <Ariadne> bl4ckb0ne: the -openrc packages are automatically installed if you use openrc (which alpine-base does) and you have the parent package (e.g. dbus) installed too
2022-01-04 13:09:53 <bl4ckb0ne> all of them?
2022-01-04 13:23:23 <psykose> no, it's per package
2022-01-04 13:23:34 <psykose> if you pull something in and it has an -openrc, that is also pulled in if you have openrc installed
2022-01-04 13:23:40 <psykose> if you have a container with no init it's not
2022-01-04 13:23:49 <psykose> or if you hack your system into runit or s6 or something
2022-01-04 14:55:07 <adhawkins> ikke: Has any progress been made on !28865. Do you need me to do anything?
2022-01-04 14:55:15 <adhawkins> I clicked 'approve' on the MR, is that enough to show my support?
2022-01-04 14:56:36 <psykose> it's waiting for the maintainer that has been afk for a while to approve the move to community of py3-build
2022-01-04 14:57:09 <ikke> It's waiting on Fab indeed
2022-01-04 14:57:20 <mps> I created script to install alpine on usb disk (here https://tpaste.us/WE8N) but got this as error https://tpaste.us/gBoP
2022-01-04 14:58:13 <mps> is there a 'trick' to install grub in such case
2022-01-04 14:58:48 <adhawkins> ikke, psykose: Understood. Hopefully it'll get resolved before a new release of my package is required. They've been relatively infrequent so shouldn't be an issue with a bit of luck.
2022-01-04 14:59:20 <adhawkins> Just wanted to double check my 'approval' was enough of a signal from me that I accept the changes to my package.
2022-01-04 14:59:29 <psykose> it is
2022-01-04 15:02:21 <adhawkins> Good stuff.
2022-01-04 15:10:11 <ikke> I'll try to contact fab 
2022-01-04 16:05:53 <ddevault> I got alpine running on this RAM-constrained raspberry pi after some manual work
2022-01-04 16:06:38 <ddevault> what I ended up doing is extracting the boot tarball from alpinelinux.org/downloads to the first (vfat) partition, then extracting mini root filesystem onto a second ext4 partition, updating cmdline.txt manually, then using qemu-user from another system to install alpine-base and a kernel and suchwhat
2022-01-04 16:08:44 <skarnet> ddevault: nice. Did you identify the limiting factors? would it at all be possible to make a boot tarball that works under your constraints?
2022-01-04 16:09:09 <ddevault> I don't think that the standard alpine installation setup is suitable on systems without a lot of RAM
2022-01-04 16:09:14 <ddevault> and it would require significant updates to become so
2022-01-04 16:09:27 <ddevault> s/without a lot of RAM/with very little RAM/
2022-01-04 16:09:27 <alpine-meetbot> ddevault meant to say: I don't think that the standard alpine installation setup is suitable on systems with very little RAM
2022-01-04 16:09:32 <mps> https://pbot.rmdir.de/HWrGOrau34RW28Vdg2XaMA
2022-01-04 16:09:39 <mps> uh, sorry
2022-01-04 16:10:00 <mps> https://arvanta.net/alpine/flash-rpi-zero/
2022-01-04 16:11:06 <ddevault> this is essentially what I did, but setup-alpine does not work with this little RAM
2022-01-04 16:11:23 <ddevault> the zero has 512M which I suppose is enough, but this one has <200M by default
2022-01-04 16:11:48 <mps> I forgot, thought it was 256M ram
2022-01-04 16:11:57 <ddevault> it's shared with the GPU
2022-01-04 16:12:13 <ddevault> so you don't get to use all 256
2022-01-04 16:12:50 <mps> couldn't it be set somewhere in cmdline.txt
2022-01-04 16:13:01 <ddevault> it can be configured in config.txt, but you still aren't getting away with much RAM
2022-01-04 16:13:30 <ddevault> the margins are pretty thin here if we're splitting hairs over 50M of RAM
2022-01-04 16:14:14 <mps> hmm, though I don't understand why setup-alpine require so much RAM
2022-01-04 16:14:39 <skarnet> neither do I tbh and profiling it would be interesting
2022-01-04 16:14:40 <ddevault> it's not setup-alpine per-se, it's the overlay that alpine's install system uses
2022-01-04 16:14:56 <ddevault> setup-alpine just causes enough disk space to be used (e.g. by downloading packages) for it to exceed available storage
2022-01-04 16:15:02 <mps> ah, yes, overlay
2022-01-04 17:35:32 <an3223> can abuild rootbld be made to use the host's /etc/apk/cache/ ? 
2022-01-04 17:37:56 <psykose> add --cache-dir=/etc/apk/cache (or whatever cache dir) to like like 2426 in the apk add invocation in abuild :)
2022-01-04 17:38:16 <psykose> like line *
2022-01-04 17:40:32 <an3223> gross, thanks lol
2022-01-04 18:18:47 <jvoisin> > .githooks/pre-commit: 13: testing/hardened-malloc/APKBUILD: Bad substitution
2022-01-04 18:18:49 <jvoisin> mh.
2022-01-04 18:18:55 <jvoisin> (when I'm trying to do a `git commit -a`
2022-01-04 18:18:57 <jvoisin> )
2022-01-04 18:20:09 <ikke> Is there a conflict in that file?
2022-01-04 18:21:17 <ikke> that package does not exist in aports
2022-01-04 18:21:24 <jvoisin> nope, I'm adding it
2022-01-04 18:21:28 <jvoisin> at least I'm trying ot
2022-01-04 18:21:30 <ikke> ok
2022-01-04 18:34:02 <ikke> I'm puzzled by this function: https://gitlab.alpinelinux.org/alpine/apk-tools/-/blob/master/src/app_update.c#L17
2022-01-04 18:34:16 <ikke> I have no clue where the actual updating is done
2022-01-04 18:36:36 <ikke> I guess that's implicit by .forced_force = APK_FORCE_REFRESH
2022-01-04 19:48:09 <Ariadne> ikke: correct, apk runs with APK_FORCE_REFRESH, which forces the indices to be redownloaded.  all the `update` applet does is tell you how many packages it sees in the indices.
2022-01-04 20:13:04 <mps> algitbot: retry master
2022-01-04 20:20:17 <mps> tree fails to build because of wrong source url
2022-01-04 20:20:49 <psykose> i can fetch it locally
2022-01-04 20:20:57 <psykose> it's probably just some legendary routing
2022-01-04 20:22:06 <mps> psykose: yes, also I fetched it
2022-01-04 20:22:06 <psykose> it also updated on the other arches
2022-01-04 20:22:29 <mps> algitbot: retry master
2022-01-04 20:23:03 <mps> algitbot: what
2022-01-04 20:23:23 <mps> hmm, looks like it sleeps
2022-01-04 20:23:33 <ikke> I have a feeling they block us
2022-01-04 20:23:51 <psykose> arch uses debian as a mirror for tree lol
2022-01-04 20:23:56 <ikke> It's already happening for a long time
2022-01-04 20:24:03 <psykose> they blocked only the x86 host?
2022-01-04 20:24:09 <ikke> I suppose
2022-01-04 20:24:14 <psykose> unlucky
2022-01-04 20:24:32 <mps> ikke: I fetched from x86_64 lxc
2022-01-04 20:24:49 <ikke> PING mama.indstate.edu (139.102.70.201) 56(84) bytes of data.
2022-01-04 20:24:51 <ikke> no answer yet for icmp_seq=1
2022-01-04 20:24:53 <ikke> mps: different host
2022-01-04 20:24:59 <mps> it is not at same net as x86 builders
2022-01-04 20:25:05 <ikke> correct
2022-01-04 20:25:12 <mps> that is
2022-01-04 20:25:16 <mps> hm
2022-01-04 20:26:10 <psykose> just manually copy it into distfile
2022-01-04 20:26:36 <ikke> yes, that's what I've been doing for the longest time
2022-01-04 20:26:46 <psykose> ugh
2022-01-04 20:27:02 <psykose> i guess it's actually blocked then
2022-01-04 20:29:20 <ikke> algitbot: retry master
2022-01-04 20:32:50 <ikke> Ok, 'fixed' for now
2022-01-04 21:22:00 <adhawkins> ikke: There's just been a new release of parametrize_from_file. Typical :)
2022-01-04 21:27:14 <ikke> adhawkins: heh
2022-01-04 21:27:22 <ikke> we can incorporat that in the MR
2022-01-04 21:44:15 <adhawkins> Ok, if you're happy to do that. Let me take a look at the changes to see if there's anything significant.
2022-01-04 21:47:37 <adhawkins> No, looks like it should be a relatively simple update ikke. Let me know if you need any assistance.
2022-01-04 21:47:51 <ikke> Alright
2022-01-04 21:50:22 <ikke> 0.10.0
2022-01-04 21:52:23 <ikke> https://gitlab.alpinelinux.org/alpine/aports/-/merge_requests/28865/diffs?commit_id=ffa4b974b588b177ba602212a554b4fe17d8d04f
2022-01-04 21:57:16 <psykose> ikke: could you press the shiny on !29166
2022-01-04 22:10:49 <Thermi> Hello hello, somebody please merge https://gitlab.alpinelinux.org/alpine/aports/-/merge_requests/29114
2022-01-04 22:12:17 <Thermi> tyvm!
2022-01-04 22:15:43 <ikke> Both have been merged apparently
2022-01-04 22:16:53 <psykose> hah
2022-01-04 22:17:17 <psykose> michał got my back :)
2022-01-04 22:21:07 <psykose> algitbot: retry master
2022-01-04 22:24:58 <adhawkins> Looks good to me ikke.
2022-01-04 23:04:55 <mps> about my incorrect conclusion today and yesterday that alpine grub-efi maybe have bug, with minimal (Dermot) help I found solution for my problem on M1
2022-01-04 23:05:24 <mps> so, alpine grub-efi is ok, though grub is mess anyway
2022-01-04 23:16:39 <psykose> xapian-core seems to have a single failing test on x86_64 every time
2022-01-04 23:30:28 <psykose> algitbot: retry master
2022-01-04 23:58:43 <omni> what's so special with my Core(TM)2 Duo L7700? is it because of it that I get "illegal instruction" when I try to edit a file with neovim? (regular vim and busybox vi is fine)
2022-01-04 23:59:12 <omni> I installed 3.15 on an old laptop
2022-01-05 00:07:22 <psykose> that sounds like some i386/i686 level stuff, i forgot what the minimum was
2022-01-05 00:07:33 <psykose> if you can pick out what the specific instruction is that would be nice
2022-01-05 00:20:26 <Hello71> did you use x86_64
2022-01-05 00:22:36 <Hello71> psykose: Intel Core is about a decade newer than 80686, and about two decades newer than 80386
2022-01-05 00:23:12 <psykose> yes, i misassumed
2022-01-05 00:23:17 <psykose> but it's the same issue anyway
2022-01-05 00:23:29 <Hello71> not really
2022-01-05 00:23:35 <Hello71> probably
2022-01-05 00:24:57 <Hello71> x86-32 is only an issue because it's so old it doesn't have proper ieee 754 or locking. x86-64 is new enough that it doesn't have these issues (sse2 is mandatory)
2022-01-05 00:25:31 <Hello71> if you get illegal instruction on x86-64 it's more likely a hardware issue, or on alpine, a fortify_source violation
2022-01-05 00:25:52 <Hello71> _FORTIFY_SOURCE
2022-01-05 00:25:54 <psykose> ah, right, those as well
2022-01-05 00:27:09 <Hello71> it's highly unlikely that illegal instruction on x86-64 is because of an isa extension unless you're compiling programs yourself or using live migration
2022-01-05 00:31:51 <adamplumb> so I was looking into upgrading libadwaita to the final 1.0.1 release (our latest package is 1.0.0-alpha4).  The final 1.0 release requires gtk 4.5+ and alpine edge is on 4.4.1.  I started looking into upgrading the gtk4.0 package but 4.5.1 requires a new pango (and possibly so on)
2022-01-05 00:32:19 <adamplumb> should i just go down the rabbithole here or is there any guidance on what I should do for managing libadwaita?
2022-01-05 00:34:21 <Hello71> i guess this all needs to be upgraded eventually
2022-01-05 00:35:22 <Hello71> but knowing gnome, probably several days into this adventure you will discover that a dozen applications need fixes which are not yet written because the api changes were just released yesterday or whatever
2022-01-05 00:38:16 <adamplumb> yeah that sounds about right
2022-01-05 00:55:48 <ktprograms> Hi, if a package has a license file, should it be put in the package or the -doc subpackage?
2022-01-05 01:01:18 <omni> psykose: I'm not that used to debugging, really, do I look at an strace?
2022-01-05 01:18:29 <omni> I can open files but on any keystroke I'm immediately thrown out with "illegal instruction"
2022-01-05 01:23:03 <ktprograms> omni: Probably run it with GDB, then either use 'bt full' or 'layout asm', the latter to show the actual instructions.
2022-01-05 01:23:36 <ktprograms> strace probably wouldn't be helpful here since that only logs syscalls and not instructions
2022-01-05 01:24:56 <ktprograms> omni: ignore what I said about 'bt full', just do 'layout asm' and see if you see any instructions that wouldn't be supported by your cpu
2022-01-05 01:25:54 <psykose> ktprograms: /usr/share/licenses/pkgname/LICENSE , and it will be in -doc
2022-01-05 01:32:49 <ktprograms> psykose: Eaglemode only seems to have a License.html in its installation directory, which is a HTML version of the GPLv3
2022-01-05 01:33:22 <psykose> that's fine
2022-01-05 01:33:38 <ktprograms> so no need to put anything in /usr/share/licenses?
2022-01-05 01:33:44 <psykose> you can put exactly that there
2022-01-05 01:34:03 <psykose> personally i never bother
2022-01-05 01:34:19 <psykose> i am sure people can go and find out what license something has if they have the internet to download it
2022-01-05 01:35:19 <ktprograms> psykose: Ah I see. BTW is it a good idea to package something if I'm not going to be actively using it, or should I just put the APKBUILD somewhere for future reference?
2022-01-05 01:35:39 <psykose> most of my packaging is things i don't use
2022-01-05 01:35:47 <psykose> because somebody else asked me for it
2022-01-05 01:35:57 <psykose> i do the smoke tests, but that's about it
2022-01-05 01:36:26 <ktprograms> But in this case I'm pretty sure *no one* is using eaglemode
2022-01-05 01:36:35 <psykose> it is quite the niche program
2022-01-05 01:44:11 <boomanaiden154> I believe apk info -a also shows the license when run against a package.
2022-01-05 01:44:33 <psykose> technically the licence contains more
2022-01-05 01:44:38 <psykose> like who actually holds the copyright
2022-01-05 01:44:39 <psykose> etc
2022-01-05 01:44:53 <psykose> this is really quite pointless though
2022-01-05 01:45:10 <psykose> though it is 'required' by some licences pretty sure, that the licence comes with the packaging
2022-01-05 01:45:41 <boomanaiden154> True to your first point. But I think most people look to just see which license it is to check for compliance and what not.
2022-01-05 01:45:54 <psykose> yep
2022-01-05 01:46:08 <psykose> i wish there were fewer licences
2022-01-05 01:47:11 <boomanaiden154> It’s not a bad thing to have quite a few. Gives people more options based on the criteria with which they want to release their software since it is very difficult to create a new license without being a lawyer.
2022-01-05 01:48:55 <psykose> it's not 'bad', it's just slightly annoying to come across more or less the exact same thing yet again written slightly different in 50000 variants
2022-01-05 01:49:03 <psykose> the greater issue lies with the need for them in the first place
2022-01-05 01:51:34 <ktprograms> psykose: Do you think having a package in aports that *no one* is using is ok, or should I just put it in my custom-aports repo which is also on the alpine gitlab?
2022-01-05 01:51:44 <psykose> that is up to you
2022-01-05 01:52:41 <ktprograms> I think I'll not put it in aports, also because upstream has no release notifications
2022-01-05 01:53:04 <ktprograms> I'll keep the APKBUILD around so if it's ever requested it can be added to aports.
2022-01-05 02:49:45 <omni> Hello71: yes, x86_64
2022-01-05 02:51:24 <omni> ktprograms: thanks, but that's a lot to figure out what to look for, probably for another day
2022-01-05 02:52:12 <psykose> gdb nvim; layout asm; run; crash it, or some shit like that
2022-01-05 02:52:28 <psykose> i think since it's a tui program you might want to remote it, but not sure
2022-01-05 02:52:29 <omni> ok, I'll try =)
2022-01-05 02:55:12 <ktprograms> psykose: The layout asm should be after the crash
2022-01-05 02:55:26 <psykose> the order doesn't matter i don't think
2022-01-05 02:55:31 <psykose> it just changes the display type
2022-01-05 02:56:11 <ktprograms> psykose: I see
2022-01-05 02:56:26 <psykose> i didn't test.. don't exactly have a bad instruction exe on hand
2022-01-05 02:56:43 <ktprograms> psykose: How would remote debugging work? Because I also had a bit of problem when debugging vim that it messes up the display
2022-01-05 02:57:40 <omni> ah, pmaxud %xmm2,%xmm0
2022-01-05 02:57:47 <omni> so lacking sse4 then?
2022-01-05 02:58:26 <psykose> gdbserver :9999 - program; gdb; target remote localhost:9999
2022-01-05 02:58:28 <psykose> or something
2022-01-05 02:58:44 <psykose> i don't really remember but it basically just passes the stuff to another gdb so it's a separate display
2022-01-05 02:59:47 <psykose> omni: hmm.. nothing in the neovim build indicates to toggle that
2022-01-05 03:05:44 <Hello71> that sounds overly complicated for vim, just attach gdb afterwards
2022-01-05 03:06:41 <Hello71> also x/i $pc is simpler
2022-01-05 03:09:23 <psykose> what is x/i $pc
2022-01-05 03:10:50 <psykose> ah, print instruction from current program counter
2022-01-05 03:19:50 <Hello71> i mean instead of layout asm which works but is overkill here
2022-01-05 03:35:49 <omni> still, seem to be that pmaxud instruction that came with sse4 that my cpu doesn't have on this old laptop
2022-01-05 03:36:21 <omni> I can live without neovim on it, but I'm still curious
2022-01-05 03:36:47 <omni> and it's fun to finally try something with gdb,thanks for hints!
2022-01-05 03:41:46 <psykose> i think if you compile it yourself you might be able to get around that.. maybe
2022-01-05 03:41:59 <psykose> assuming it doesn't force it somewhere that i can't find
2022-01-05 03:42:08 <psykose> it could also be one of the dependencies
2022-01-05 03:46:14 <ktprograms> omni: Now's where you want to use 'bt full' to see which part of the code is using the sse4 instruction
2022-01-05 03:57:20 <omni> /usr/lib/libluajit-5.1.so.2?
2022-01-05 03:57:56 <psykose> it's explicitly compiled with sse4.2
2022-01-05 03:58:05 <psykose> for x86_64, see the apkbuild
2022-01-05 03:58:52 <psykose> if i'm being honest i would probably not change it
2022-01-05 04:24:01 <omni> nah
2022-01-05 07:27:50 <nmeum> Ariadne: I have a patch for aligning the default GCC LIBRARY_PATH on riscv64 with what we have for existing arches !29185 it's a bit hacky but not sure if there is a better way to do this without introducing lib64 and lib32 library directories (they are hardcoded in 2-3 places in gcc for riscv)
2022-01-05 07:28:44 <Ariadne> i was planning on updating the snapshot soon
2022-01-05 07:29:37 <nmeum> the patch also applies cleanly with gcc git master
2022-01-05 07:29:51 <nmeum> i.e. it seems they haven't changed any of this in gcc 12
2022-01-05 07:42:29 <mps> Ariadne: what keeps you to upgrade gcc to 12
2022-01-05 07:45:54 <Ariadne> mps: have you seen the output of the GCC developers lately
2022-01-05 07:46:29 <mps> no, tbh
2022-01-05 07:46:36 <Ariadne> mps: the first release (which hasnt even happened for GCC 12 yet) is usually quite buggy
2022-01-05 07:46:55 <mps> ok
2022-01-05 07:47:02 <Ariadne> waiting for 11.2 was intentional
2022-01-05 07:47:41 <mps> I've only seen that mold is 'added' there
2022-01-05 07:50:14 <nmeum> ah, so by "updating the snapshot" you just mean rebasing the patches but not upgrading to gcc 12?
2022-01-05 08:40:56 <mps> ncopa: !29199 !29057 !29168 !29201
2022-01-05 08:42:03 <mps> I don't dare to do NMU merge anymore, risking to be called by TSC :)
2022-01-05 09:22:39 <ncopa> morning
2022-01-05 09:22:52 <ncopa> mps: I'll have a look at those, thanks
2022-01-05 09:35:45 <omni> moin
2022-01-05 09:42:46 <Ariadne> nmeum: yes, moving to another 11.2 snapshot
2022-01-05 09:45:39 <nmeum> ah ok, I created patches for your alpine-gcc-patches repo with the recent patches for main/gcc could include those with the rebase
2022-01-05 09:46:15 <nmeum> there is also !29096 which replaces the two libucontext related patches on ppc with a simpler one which just includes asm/ptrace.h on ppc
2022-01-05 09:46:56 <Ariadne> i'd prefer to roll all of those changes up into the rebase.  though adding patches to aports is fine, too, i suppose
2022-01-05 09:47:47 <nmeum> yeah, rolling them with the rebase is fine with me
2022-01-05 10:44:08 <Ariadne> ok will work on it today
2022-01-05 10:51:00 <nmeum> sweet! feel free to ping me if any issues show up (:
2022-01-05 19:03:03 <bl4ckb0ne> should the version of a package absolutely use dots?
2022-01-05 19:03:30 <bl4ckb0ne> i am packaging something that is versioned with yyyy_mm_dd
2022-01-05 19:05:07 <ikke> _ is not a valid separator
2022-01-05 19:05:18 <ikke> but it does not have to contain a .
2022-01-05 19:05:45 <bl4ckb0ne> so I can put yyyy.mm.dd as pkgver and leave the hardcoded yyyy_mm_dd in the source url?
2022-01-05 19:06:09 <ikke> either that, or yyyymmdd
2022-01-05 19:06:29 <ikke> but yes, you could use ${pkgver/./_} to replace it
2022-01-05 19:08:24 <bl4ckb0ne> didnt know i could do that, thanks!
2022-01-05 19:10:29 <bl4ckb0ne> huh it replaces only the first dot
2022-01-05 19:10:39 <bl4ckb0ne> i end up with yyyy_mm.d in the url
2022-01-05 19:11:30 <Hello71> apply more /
2022-01-05 19:14:02 <bl4ckb0ne> concatenated?
2022-01-05 19:14:51 <ikke> /./_
2022-01-05 19:14:54 <ikke> \//./_
2022-01-05 19:15:15 <ikke> so ${pkgver//./_}
2022-01-05 19:15:25 <ikke> https://tldp.org/LDP/abs/html/string-manipulation.html
2022-01-05 19:15:57 <bl4ckb0ne> thanks
2022-01-05 23:20:04 <Newbyte> Is testing/agg actually used for anything?
2022-01-05 23:33:12 <psykose> it's not a dependency of anything, why
2022-01-05 23:33:47 <psykose> i can see it was dropped from fedora 4 years ago
2022-01-05 23:34:06 <psykose> for bad upstream reasons
2022-01-05 23:36:17 <Newbyte> psykose: it's the only package that hasn't been switched from using sdl-dev to sdl12-compat-dev
2022-01-05 23:37:06 <Newbyte> I have an MR for that, but I don't particularly know how to test it: !28184
2022-01-05 23:40:34 <psykose> yeah, i am aware of the sdl context
2022-01-05 23:40:40 <psykose> sadly i don't know anything about agg itself
2022-01-05 23:40:57 <psykose> good work by the way- i'm sure it's taken forever to work on every sdl1 package
2022-01-06 00:55:55 <Newbyte> thanks!
2022-01-06 02:44:48 <Sheila> anyone else experiencing significant lag interacting with gitlab via git/ssh?
2022-01-06 02:45:12 <Sheila> checked mtr and it's not a networking issue for v4 or v6.
2022-01-06 02:45:47 <boomanaiden154> There have been reports recently, but only related to failover from IPv6 to IPv4, at least as far as I’m aware.
2022-01-06 02:48:34 <Sheila> took about 150s for me to connect in order to pull just now, is why I ask.
2022-01-06 02:52:24 <ktprograms> boomanaiden154: Also takes long for me to `git pull upstream master`, no change with using `-4` flag (anyway my network doesn't have IPv6)
2022-01-06 02:52:44 <ktprograms> But git push is pretty instant
2022-01-06 02:53:20 <Sheila> having same issue with push, actually.
2022-01-06 02:53:33 <psykose> for me it was definitely an ipv6 issue
2022-01-06 02:53:46 <psykose> but if you are sure it's not ipv6 (you can sysctl disable it) then it would be yet another one
2022-01-06 02:53:48 <psykose> unlucky
2022-01-06 02:53:58 <Sheila> this is from my dedi in hetzner.
2022-01-06 02:54:55 <psykose> could you try `sysctl net.ipv6.conf.all.disable_ipv6=0` to definitely rule out ipv6 (assuming that doesn't break anything) or the -4 flags (less destructive)
2022-01-06 02:56:02 <boomanaiden154> I just ran a pull against aports and pushed to my remote fork and everything seemed to work fine, so it seems like the issue isn't super likely to be on the infrastructure end.
2022-01-06 03:05:50 <Ariadne> i've noticed gitlab is inaccessible via ipv6
2022-01-06 03:06:02 <Ariadne> GIT_SSH_COMMAND="ssh -4" helped for me
2022-01-06 03:11:19 <psykose> you can also add the .ssh/config option
2022-01-06 03:11:42 <psykose> `AddressFamily inet` under a `Host gitlab.alpinelinux.org` block
2022-01-06 03:13:21 <Sheila> yeah, v6 issue; $GIT_SSH_COMMAND worked.
2022-01-06 03:14:26 <psykose> i wonder how ikke is doing on fixing awall for it
2022-01-06 03:24:09 <Ariadne> Hello71: you said you would like to change the GCC specs, can you link me to the aports issue for it
2022-01-06 03:26:34 <Ariadne> nevermind, found it :)
2022-01-06 03:29:38 <Ariadne> Hello71: you want -fstack-clash-protection and -fcf-protection enabled on those archs, right?
2022-01-06 03:30:43 <Hello71> i don't remember what i said
2022-01-06 03:33:10 <Ariadne> https://gitlab.alpinelinux.org/alpine/aports/-/issues/13246#note_195905
2022-01-06 03:33:27 <Ariadne> i'm also curious as to why -z relro can be removed from specs
2022-01-06 03:33:32 <Hello71> yes, i still haven't looked into _GLIBCXX_ASSERTIONS (i know what it does but not sure about actual costs). for gcc specs, i think remove -z relro; -f* i think is better with Init(1) rather than specs; i'm not sure it works in specs, and also i'm not sure it gives the right -v
2022-01-06 03:33:40 <Hello71> because binutils is built with --enable-relro
2022-01-06 03:33:51 <Ariadne> which does what? :p
2022-01-06 03:34:09 <Hello71> uh... i was under the impression that it effectively enables -z relro by default
2022-01-06 03:34:53 <Hello71> according to the ld and gold configure output, --enable-relro          enable -z relro in ELF linker by default
2022-01-06 03:35:04 <Hello71> so my conclusion is that gcc specs don't need to write -z relro
2022-01-06 03:35:52 <Hello71> and probably -z now should be enabled in gcc upstream for musl targets. not sure how that affects shared libraries though; afaik DT_BIND_NOW causes RTLD_NOW or somesuch
2022-01-06 03:37:11 <Hello71> i did not specifically test whether --enable-relro works as documented, in general or on alpine
2022-01-06 03:38:54 <Hello71> i would be fairly surprised if it didn't work as documented; binutils is bloated but not usually that broken
2022-01-06 03:39:41 <Ariadne> ok
2022-01-06 03:45:26 <Ariadne> Hello71: i revised the patch so that it only does -Wl,-z,now
2022-01-06 03:45:44 <Ariadne> worst case, we put it back in the specs :)
2022-01-06 03:57:37 <Ariadne> Hello71: arch seems to build with --enable-cet=auto, but that only seems to enable -fcf-protection for GCC and runtime libs, it doesn't put it in Init(1)
2022-01-06 04:10:53 <Ariadne> i'm mostly curious as to how we can do Init(1) for specific architectures, i dont see anyone doing it
2022-01-06 04:27:42 <Ariadne> god, gitlab is being so slow
2022-01-06 04:36:05 <psykose> yup
2022-01-06 04:52:47 <Hello71> allegedly turning off ipv6 fixes it?
2022-01-06 04:53:12 <Hello71> for the specific arches, I was thinking about selective patching
2022-01-06 04:53:32 <Hello71> but I'm not really happy about that
2022-01-06 04:54:17 <Hello71> doesn't alpine also already --enable-cet
2022-01-06 10:06:07 <smcavoy> could !26763 get looked at?
2022-01-06 10:49:09 <adhawkins> ikke: !28865 is currently assigned to me. As far as I can see there is nothing for me to do now. Should it be assigned to fab?
2022-01-06 10:50:18 <psykose> just waiting for fab
2022-01-06 10:50:34 <psykose> the assignees are just a usually-useful abstract thing that doesn't really matter here
2022-01-06 10:50:42 <adhawkins> Ah ok, thanks psykose.
2022-01-06 10:51:09 <psykose> maybe if they could auto assign based on every changed aport it would be a tad more useful
2022-01-06 10:52:03 <adhawkins> Perhaps. Will just leave as is and wait for fab's input.
2022-01-06 10:52:09 <psykose> it's probably possible
2022-01-06 10:52:14 <psykose> maybe i should look at it sometime
2022-01-06 10:52:19 <adhawkins> :)
2022-01-06 10:52:26 <adhawkins> Do you need a round tuit? :)
2022-01-06 10:52:33 <psykose> round tuit?
2022-01-06 10:52:47 <psykose> not familiar with that word
2022-01-06 10:52:58 <adhawkins> Apologies. Poor attempt at humout.
2022-01-06 10:53:05 <adhawkins> s/humout/humour/
2022-01-06 10:53:05 <alpine-meetbot> adhawkins meant to say: Apologies. Poor attempt at humour.
2022-01-06 10:53:20 <psykose> that must be a very obscure pun for me to not get it
2022-01-06 10:53:23 <psykose> or i am too tired again
2022-01-06 10:53:24 <adhawkins> "When I get around to it" is a popular English phrase relating to procrastination.
2022-01-06 10:53:26 <psykose> oh
2022-01-06 10:53:36 <psykose> right, that makes sense
2022-01-06 10:53:37 <psykose> thanks :)
2022-01-06 10:54:07 <adhawkins> https://www.zazzle.co.uk/a_round_tuit_6_cm_round_badge-145861117369111267?rf=238405708775511253&tc=Cj0KCQiAw9qOBhC-ARIsAG-rdn6EPEDwvYdQLCeGiJWlebRgZXxWIpQMwc1Dhp1-pimVmWH5o8rsKZcaAnhWEALw_wcB&utm_source=google&utm_medium=cpc&utm_campaign=uk_shopping&utm_term=&ca_chid=2001810&ca_source=gaw&ca_ace=&ca_nw=g&ca_dev=c&ca_pl=&ca_pos=&ca_cid=571874159952&
2022-01-06 10:54:07 <adhawkins> ca_agid=132412632713&ca_caid=15528779968&ca_adid=571874159952&ca_kwt=&ca_mt=&ca_fid=&ca_tid=pla-1517409312853&ca_lp=9045592&ca_li=&ca_devm=&ca_plt=&gncclsrc=aw.ds&gclid=Cj0KCQiAw9qOBhC-ARIsAG-rdn6EPEDwvYdQLCeGiJWlebRgZXxWIpQMwc1Dhp1-pimVmWH5o8rsKZcaAnhWEALw_wcB
2022-01-06 10:54:14 <adhawkins> Wow, that's a long URL...
2022-01-06 10:54:39 <psykose> just need the first part before `?`
2022-01-06 10:54:43 <adhawkins> https://www.zazzle.co.uk/a_round_tuit_6_cm_round_badge-145861117369111267
2022-01-06 10:54:45 <psykose> classic tracking urls
2022-01-06 10:54:49 <adhawkins> Yep! :D
2022-01-06 10:55:05 <psykose> this is definitely very british
2022-01-06 10:55:27 <adhawkins> Yes, should never assume locality on the internet!
2022-01-06 10:55:46 <psykose> i lived in scotland for 3 years
2022-01-06 10:55:50 <psykose> just never heard that specific one :p
2022-01-06 10:55:55 <adhawkins> Heh, ok :)
2022-01-06 11:18:37 <AntoniAloyTorrens[m]> ncopa: Could you take a look at !26184 ? algitbot marked it as stale, but it should be ready to merge anyways
2022-01-06 11:24:30 <psykose> you can also replace the python `mv` with amove if you wanna fully 'modernise' it for fun :p
2022-01-06 11:34:06 <AntoniAloyTorrens[m]> <psykose> "you can also replace the python..." <- Done! It's ready now :)
2022-01-06 11:34:29 <psykose> awesome :)
2022-01-06 12:28:13 <ncopa> AntoniAloyTorrens[m]: done. thanks!
2022-01-06 13:04:05 <MaximKarasev[m]> Hi. I had nothing to do and therefore accidentally wrote a couple of APKBUILDs. However, I don't think I'll be a good maintainer cause I generally don't use this software. Should I put them to unmaintained?
2022-01-06 13:36:31 <skarnet> "accidentally wrote a couple of APKBUILDs". Yeah, I hate it when that happens.
2022-01-06 13:38:31 <psykose> i mean
2022-01-06 13:38:41 <psykose> i have written maybe 25 i never submitted
2022-01-06 13:38:56 <psykose> accidentally is correct
2022-01-06 13:53:44 <minimal> adhawkins: yeah I always intended to buy one of those Round Tuits to keep on my desk at work.......but I never got around to it :-)
2022-01-06 13:55:13 <mps> minimal: hi! I solved problem with grub-efi on M1 with help of your very good explanations and help
2022-01-06 13:55:45 <minimal> mps: cool, so what was the fix/workaround you ended up with?
2022-01-06 13:56:41 <mps> minimal: simple, apk add --no-script grub-efi
2022-01-06 13:56:46 <mps> and 'grub-install --target=arm64-efi --efi-directory=/boot/efi'
2022-01-06 13:57:17 <jvoisin> MaximKarasev[m]: ask the maintainer of depending packages to take care of them?
2022-01-06 13:57:38 <minimal> mps: oh, that's pretty fundamental for getting EFI working :-)
2022-01-06 13:58:27 <mps> minimal: yes ofc. but your hint that grub-efi embeds partition in bootaa64.efi was the 'light'
2022-01-06 14:03:31 <minimal> mps: so you've now got an easy-to-boot installation procedure working?
2022-01-06 14:09:45 <mps> minimal: not finished yet, but this is one important step
2022-01-06 14:28:22 <MaximKarasev[m]> <jvoisin> "Maxim Karasev: ask the maintaine..." <- This packages aren't aleady in aports, and it's end-user software, therefore nothing may depend on it
2022-01-06 14:55:55 <ddevault> I would advise a more urgent approach to relicensing the wiki
2022-01-06 15:02:00 <psykose> i really can't understand what this email is about
2022-01-06 15:03:31 <psykose> i checked some of these newbie pages and i am very confused by like, this discussion page https://wiki.alpinelinux.org/wiki/Talk:Alpine_newbie
2022-01-06 15:03:49 <psykose> i guess one of them was moved to deletion?
2022-01-06 15:06:21 <ddevault> this account should probably be removed from the wiki
2022-01-06 15:07:10 <psykose> what for
2022-01-06 15:07:58 <ddevault> this is not a productive use of the wiki
2022-01-06 15:08:36 <ddevault> I don't think the wiki should be taking a position on "lammers" or who may or may not be the best wiki editors in siloed off sections no one is allowed to touch under threat of copyright rug-pulls
2022-01-06 15:09:34 <psykose> i guess yyyy is the same person
2022-01-06 15:09:46 <psykose> which kind of just looks like trolling
2022-01-06 15:10:24 <psykose> yes, i guess i agree
2022-01-06 15:13:17 <bl4ckb0ne> the wiki isnt 4chan, this has nothing to do on it
2022-01-06 15:26:13 <adhawkins> I really need to start a doc mapping people's GitLab names to IRC ones. Who is Michal Polanski on here?
2022-01-06 15:27:28 <psykose> not here to my knowledge
2022-01-06 15:27:37 <psykose> never seen em
2022-01-06 15:28:03 <psykose> did you need to ask something
2022-01-06 15:28:16 <adhawkins> Ah ok. They've added a comment to a couple of my recent MRs pointing to the GitHub release of the software.
2022-01-06 15:28:23 <adhawkins> !29233 for example
2022-01-06 15:28:30 <adhawkins> Just wondering should I do that as a matter of course?
2022-01-06 15:28:39 <psykose> generally it's considered a good thing
2022-01-06 15:28:56 <psykose> i forget most of the time and i don't get that thrown on lol, but if you want to add it it's more good than bad
2022-01-06 15:29:01 <psykose> saves someone a minute
2022-01-06 15:29:06 <adhawkins> Ok. If I make that an additional line in the commit for the upgrade, would that be acceptable?
2022-01-06 15:29:18 <psykose> in the commit message?
2022-01-06 15:29:20 <psykose> you can if you want
2022-01-06 15:29:23 <adhawkins> That's what I meant, yes.
2022-01-06 15:29:29 <adhawkins> Or should it just be in the text of the MR?
2022-01-06 15:31:20 <psykose> the vast majority of upgrades don't add it, but you can if you want
2022-01-06 15:31:29 <psykose> these things are not very strictly enforced
2022-01-06 16:02:03 <ddevault> oh, I remember this wiki guy
2022-01-06 16:02:06 <ddevault> he is a troll
2022-01-06 16:02:10 <ddevault> was banned from another mailing list I used to frequent
2022-01-06 16:06:28 <ikke> I think we also banned him from ~alpine/users
2022-01-06 16:06:34 <ikke> if it's the same person
2022-01-06 16:06:58 <psykose> a troll with a blog
2022-01-06 16:21:28 <ddevault> yeah, he has been
2022-01-06 16:21:40 <ddevault> about a year ago
2022-01-06 16:21:50 <ddevault> I will copy the ban to the other list as well
2022-01-06 16:51:32 <sam_> Ariadne: we've started doing CET for amd64 in Gentoo: https://gitweb.gentoo.org/proj/gcc-patches.git/tree/11.3.0/gentoo/26_all_enable-cet.patch. any feedback welcome
2022-01-06 16:52:25 <sam_> (then passing EXTRA_OPTIONS_CF for now in the ebuild for certain USE)
2022-01-06 16:52:27 <sam_> commit message is slightly outdated there
2022-01-06 16:53:31 <sam_> (I did play with a CET_UNSET state but it was overly complicated)
2022-01-06 16:58:36 <Hello71> is there any issue with if use amd64; then eapply enable-cet.patch; fi where enable-cet.patch does -...cf-protection...Init(0)... +...cf-protection...Init(1)... other than "it looks ugly"? i agree it looks ugly but it feels less bad than this way
2022-01-06 16:59:21 <sam_> i'm wondering about that too. if it were non-amd64, i would be worried about it not applying / becoming stale and nobody noticing, but with amd64, that's not really an issue
2022-01-06 16:59:41 <Hello71> the stack clash patch has the issue that -Q --help=common prints the wrong value. i think this patch does not have that issue
2022-01-06 17:34:37 <psykose> is it intended that using liburing requires -D_GNU_SOURCE? merely including the header fails to compile, as it uses cpu_set_t that it gets from <sched.h> that flags it under gnu source
2022-01-06 17:36:12 <psykose> ah, yes https://github.com/axboe/liburing/commit/43b7ec8d17888df0debccda27dd58f4d1b90245e
2022-01-06 17:48:28 <dalias> uhg why would they use cpu_set_t in public signature?
2022-01-06 17:48:35 <dalias> couldn't they just take void* or something?
2022-01-06 17:56:10 <psykose> was just wondering why a random dependency got failed in a cmake conf check and the conf check does that literal basic test
2022-01-06 17:56:45 <ddevault> it would probably be good to mention this on the mailing list, dalias, I've run into it too
2022-01-06 17:56:48 <ddevault> I think they would be receptive
2022-01-06 17:57:01 <ddevault> but I investigated it a while ago and it's not entirely clear if there's a better place to source this type from
2022-01-06 17:58:34 <dalias> there's not. the problem is using the type unconditionally in a public api
2022-01-06 17:58:41 <ddevault> agreed
2022-01-06 17:58:46 <ddevault> but still, take it to the list
2022-01-06 17:58:57 <ddevault> they are fairly easy to work with in my experience
2022-01-06 17:58:58 <dalias> when it's a very obscure specialized functionality most users will not want or need
2022-01-06 17:59:04 <dalias> which list?
2022-01-06 17:59:07 <ddevault> the io_uring list
2022-01-06 17:59:16 <ddevault> check MAINTAINERS
2022-01-06 18:01:06 <psykose> didn't think my mere rubberducking would spawn some upstream fixes :p
2022-01-06 18:01:29 <ddevault> better to deal with problems than to lament them
2022-01-06 18:01:36 <psykose> for sure
2022-01-06 18:02:42 <mps> yes, these people on liburing ML are quite fine
2022-01-06 18:02:57 <Hello71> isn't it safe to forward-declare cpu_set_t here?
2022-01-06 18:03:24 <ddevault> not exactly, no
2022-01-06 18:04:00 <Hello71> ah, because it's a macro, not a struct
2022-01-06 18:04:03 <Hello71> stupid posix
2022-01-06 18:05:49 <Hello71> yet another reason why typedefed structs suck i guess
2022-01-06 18:06:33 <Hello71> s/macro/typedef/
2022-01-06 18:06:33 <alpine-meetbot> Hello71 meant to say: ah, because it's a typedef, not a struct
2022-01-06 18:43:55 <ikke> Can you test ssh over ipv6 to gitlab?
2022-01-06 18:44:03 <ikke> Can someone*
2022-01-06 18:44:09 <ikke> For me it's working now
2022-01-06 18:44:37 <ikke> zeta alpine » ssh -6 -T git@gitlab.alpinelinux.org
2022-01-06 18:44:39 <ikke> Welcome to GitLab, @kdaudt!
2022-01-06 18:48:11 <Hello71> works for me
2022-01-06 18:50:13 <psykose> connects
2022-01-06 18:50:20 <psykose> minus the 401
2022-01-06 18:50:22 <psykose> :p
2022-01-06 18:50:32 <ikke> 401?
2022-01-06 18:50:39 <psykose> think it's due to the restart
2022-01-06 18:50:46 <ikke> That was 50 minutes ago
2022-01-06 18:51:00 <psykose> uhh
2022-01-06 18:51:05 <psykose> why do i still get unauthorized then
2022-01-06 18:51:08 <ikke> hmm
2022-01-06 18:51:09 <psykose> another person in an mr said the same
2022-01-06 18:51:12 <psykose> not just me
2022-01-06 18:51:18 <ikke> hmm
2022-01-06 18:51:45 <psykose> just threw a bunch more in if you log them
2022-01-06 18:52:17 <psykose> https://img.ayaya.dev/9V4541mHRUrs
2022-01-06 18:53:17 <ikke> hmm, I see
2022-01-06 18:55:44 <ikke> Ok fixed, needed to restart gitaly
2022-01-06 18:56:07 <ikke> (it was not restarted, only gitlab and gitlab-shell
2022-01-06 18:56:30 <psykose> ah
2022-01-06 18:59:31 <psykose> all fixed, thanks
2022-01-06 19:00:13 <ikke> good 😌
2022-01-06 19:00:22 <psykose> :)
2022-01-06 20:48:27 <nmeum> Ariadne: just FYI: docker/moby has whitelisted the ppc swapcontext syscall in their default seccomp profile so there maybe no need to rewrite this part in libucontext
2022-01-06 21:24:46 <psykose> algitbot: retry master
2022-01-06 21:28:12 <psykose> now this is a bizarre checksum change, because both the upstream source and checksums match, but the cache was different
2022-01-06 21:28:25 <psykose> and the files do have a bunch of mostly ci changes
2022-01-06 21:28:42 <ikke> psykose: I'm talking with andypost[m] about it in #alpine-commits
2022-01-06 21:29:08 <psykose> should i perhaps join that
2022-01-06 21:29:23 <andypost[m]> NP, I'm here too)
2022-01-06 21:29:27 <ikke> https://tpaste.us/zlv5 
2022-01-06 21:29:35 <ikke> That's the diff
2022-01-06 21:29:54 <psykose> it's the mtime diff
2022-01-06 21:29:59 <psykose> the actual changes are much less
2022-01-06 21:30:00 <ikke> there is more
2022-01-06 21:30:02 <psykose> hm
2022-01-06 21:30:04 <ikke> there is also content differences
2022-01-06 21:30:23 <psykose> yes, the contents are different
2022-01-06 21:30:28 <psykose> but it's not as extreme as that looks
2022-01-06 21:30:56 <psykose> but the upstream tarball is identical to shasum, i wonder how the cache got this version
2022-01-06 21:32:05 <andypost[m]> ikke: can you please check 3.15 distfiles - is it different from master?
2022-01-06 21:32:11 <ikke> c4cde07dc0e2ff2965362e00ab0ec4d6ed5bc3b9
2022-01-06 21:32:27 <ikke> Leo adjusted the checksum
2022-01-06 21:32:57 <ikke> The version on edge was built earlier most likely and stil has the old archive
2022-01-06 21:33:51 <psykose> i looked over the full diff and it is some style changes, and macos
2022-01-06 21:34:06 <psykose> and a single #ifdef win
2022-01-06 21:34:08 <ikke> algitbot: retry master
2022-01-06 21:34:08 <psykose> looks fine to me
2022-01-06 21:34:34 <ikke> oh, move the wrong version
2022-01-06 21:34:59 <ikke> algitbot: retry master
2022-01-06 21:35:31 <ikke> psykose: btw, that's output from diffoscope :)
2022-01-06 21:35:37 <psykose> i am using the same
2022-01-06 21:35:47 <psykose> usually with exclude metadata after run one
2022-01-06 21:35:53 <psykose> because otherwise the output is 50k lines of mtime
2022-01-06 21:36:20 <psykose> very neat tool overall
2022-01-06 21:36:25 <psykose> thanks for the rec :)
2022-01-06 21:37:03 <ikke> yeah, helps a lot
2022-01-06 21:38:47 <ikke> Now it was successfully build
2022-01-06 21:38:55 <psykose> i love successful builds
2022-01-06 21:39:04 <ikke> Don't we all?
2022-01-06 21:39:24 <andypost[m]> ikke: how did you fix it?
2022-01-06 21:39:31 <ikke> I moved the file on distfiles
2022-01-06 21:39:38 <ikke> so that the builders will fetch it from upstream
2022-01-06 21:39:43 <andypost[m]> ah, thanks a lot!
2022-01-06 21:41:24 <psykose> i wonder why the log output claims a retry will purge them
2022-01-06 21:41:35 <ikke> That's locally on the builder
2022-01-06 21:41:37 <psykose> ah
2022-01-07 05:50:49 <tomalok> addressing a CVE in community -- remind me how far back i should be backporting?  3.15-stable is a given, 3.14 too?
2022-01-07 05:51:32 <ktprograms> tomalok: community is only supported for one release I think
2022-01-07 05:53:16 <ikke> Yes, correct
2022-01-07 05:55:12 <tomalok> cool, thx.  should have the update for containerd vs 3.15-stable before lights-out here.
2022-01-07 05:56:07 <ikke> thanks
2022-01-07 09:32:43 <maribu> Hi, the currently shipped linux-lts kernel in edge was built with an older version of gcc than currently shipped, which causes my akms hooks to fail. Has anyone already bumped pkgrel?
2022-01-07 09:34:46 <mps> maribu: hmm, this shouldn't be problem for akms
2022-01-07 09:35:06 <mps> if it is then this should be investigated
2022-01-07 09:37:38 <maribu> it is, the gcc plugins used to build the kernel need to match the version uses to build kernel modules
2022-01-07 09:39:40 <mps> ah, gcc plugins. this is why disabled them in linux-edge
2022-01-07 14:49:39 <Ariadne> i think that it might be a good idea to organize non-english community groups, as in the vacuum that presently exists, are problematic alternatives
2022-01-07 14:51:10 <Ariadne> for example, in retaliation for the thread about the wiki on alpine-devel, the "leader" of one of these "alternative" groups is advocating his members dogpile on ddevault in his telegram chat
2022-01-07 14:51:42 <Ariadne> which is, uhh, not great
2022-01-07 15:14:17 <indy> hi all, any rust-a-man here? :)
2022-01-07 15:19:08 <aparcar> indy: what is a rust-a-man here?
2022-01-07 15:21:06 <indy> aparcar, somebody experienced with rust on alpine
2022-01-07 15:21:43 <aparcar> indy: ty
2022-01-07 15:22:19 <indy> aparcar, i'm trying to compile influxdb_iox (rust based new generation of influxdb engine)
2022-01-07 15:23:05 <indy> when i use rustup it ends with error on linking against libclang.so.12
2022-01-07 15:24:18 <indy> when i use alpine's cargo and rust, it ends with missing component rustfmt
2022-01-07 15:55:26 <aparcar> Ariadne: by the way, thank you again for all the help around APK!
2022-01-07 16:29:11 <ericonr> indy: you shouldn't use rustup rust for packages, and rustfmt shouldn't be needed to *build* things 
2022-01-07 16:31:23 <ericonr> Are you using cargo or a build script they provide?
2022-01-07 18:16:20 <indy> ericonr, cargo
2022-01-07 18:18:47 <indy> ericonr, https://github.com/influxdata/influxdb_iox cargo build, something calls cargo fmt / rustfmt
2022-01-07 18:20:21 <indy> and in rust-toolschain.toml is rustfmt and clippy which are not in alpine
2022-01-07 18:21:29 <ericonr> indy: rm that file, and be happy
2022-01-07 23:00:36 <smcavoy> can anyone take a look at !28258 ?
2022-01-07 23:09:24 <Ariadne> aparcar: thank you for being a guinea pig for APKv3 so we don't have to in alpine :D
2022-01-08 07:02:06 <sosodank> hello there alpine friends
2022-01-08 07:03:22 <sosodank> i noticed today that my package runs noticeably more slowly on alpine than it does elsewhere, and i was looking into how it's compiled. i noticed that in alpine i'm passing -DCMAKE_BUILD_TYPE=None
2022-01-08 07:03:44 <sosodank> which surprises me, as i don't see "None" as an option on https://cmake.org/cmake/help/latest/variable/CMAKE_BUILD_TYPE.html
2022-01-08 07:04:15 <sosodank> when i tested locally, building with "None" generated slow code indeed, but perhaps there's some common definition on Alpine i'm inheriting?
2022-01-08 07:04:19 <psykose> which package
2022-01-08 07:04:29 <sosodank> notcurses
2022-01-08 07:04:39 <sosodank> i use CMAKE_BUILD_TYPE=RelWithDbgInfo on most distros
2022-01-08 07:04:50 <psykose> oh, it's you
2022-01-08 07:04:51 <psykose> right so
2022-01-08 07:04:54 <psykose> we had a None policy
2022-01-08 07:04:59 <sosodank> it is me!
2022-01-08 07:05:00 <psykose> but that is now changed
2022-01-08 07:05:11 <psykose> to MinSizeRel as a default, with some exceptions depending on what the package needs
2022-01-08 07:05:18 <sosodank> i'm down with MinSizeRel
2022-01-08 07:05:20 <psykose> if you are not doing funny shit in the cmake files
2022-01-08 07:05:27 <psykose> then MinSizeRel is basically the same as None
2022-01-08 07:05:43 <psykose> since the only difference is MinSizeRel also passes -DNDEBUG (does this affect nc much?)
2022-01-08 07:05:59 <psykose> aside from that, i think the RelWithDbgInfo default is like -O2 or maybe -O3
2022-01-08 07:06:24 <psykose> could you run some benches with all three and then set it to that type? i can also 'eventually' get to it in the mass rebuilds that i am doing for cmake
2022-01-08 07:06:27 <psykose> but that will take a while
2022-01-08 07:06:50 <psykose> all three being MinSizeRel, RelWithDebInfo, Release
2022-01-08 07:07:24 <sosodank> running some quick tests locally
2022-01-08 07:07:32 <psykose> withdebinfo isn't explicitly needed (unless you are doing something special for it) since adding a -dbg subpackage adds the debug split for abuild, and it does the debugging extra itself
2022-01-08 07:07:48 <psykose> so it should be identical to release if they pass the same -O
2022-01-08 07:08:03 <psykose> afaik it literally just adds a -g by default
2022-01-08 07:08:18 <sosodank> ok *on my local system* (which is Debian and thus perhaps differs from Alpine, i ought bring up the latter in its VM)
2022-01-08 07:08:25 <sosodank> =None does not perform any optimization
2022-01-08 07:08:29 <sosodank> explaining that mystery
2022-01-08 07:08:33 <psykose> you need CFLAGS
2022-01-08 07:08:42 <psykose> on alpine this is -Os -fomit-frame-pointer
2022-01-08 07:08:50 <psykose> the type itself will pass nothing, but those flags will be there from env
2022-01-08 07:09:15 <sosodank> Release is using O3
2022-01-08 07:09:19 <psykose> mhm
2022-01-08 07:09:31 <sosodank> RelWithDebInfo is using O2
2022-01-08 07:09:56 <sosodank> MinSizeRel is using -Os
2022-01-08 07:10:23 <psykose> tracks about right
2022-01-08 07:10:31 <sosodank> and for the test is was running
2022-01-08 07:10:35 <sosodank> larger number is better
2022-01-08 07:10:57 <sosodank> that results in 111, 237, 280, 286 for None, SizeMinRel, RelWithDebInfo, and Release
2022-01-08 07:11:13 <psykose> was None with CFLAGS
2022-01-08 07:11:27 <sosodank> so a noticeable drop for SizeMinRel, almost surely due to -Os vs -O2 and i would wager largely due to autovectorization differences
2022-01-08 07:11:42 <psykose> gcc doesn't have autovec on O2
2022-01-08 07:11:57 <sosodank> well i'm full of shit then
2022-01-08 07:11:58 <psykose> it will be the case in gcc12
2022-01-08 07:12:14 <psykose> clang does at any non -Oz opt level though
2022-01-08 07:12:17 <psykose> hm, let me compare the final size
2022-01-08 07:13:09 <sosodank> that was not my understanding re gcc but i will trust you. ahhh, perhaps i was confused by clang.
2022-01-08 07:13:34 <psykose> did you set -Os in cflags for the None test? if you did, then that entire difference would probably be caused just by debugging
2022-01-08 07:13:47 <sosodank> i set no CFLAGS for None
2022-01-08 07:13:52 <sosodank> in my local testing
2022-01-08 07:14:08 <sosodank> was one beiong set in aports?
2022-01-08 07:14:36 <psykose> it's global, everything has -Os -fomit-frame-pointer
2022-01-08 07:14:40 <sosodank> oh hrmmm
2022-01-08 07:14:42 <psykose> from the default abuild.conf
2022-01-08 07:14:47 <sosodank> ok let me retry locally with those CFLAGS
2022-01-08 07:17:23 <sosodank> confirmed that with -DCMAKE_C_FLAGS="-Os -fomit-frame-pointer" and -DCMAKE_CXX_FLAGS as same, perf of None comes up to MinSizeRel
2022-01-08 07:17:38 <sosodank> well not all the way, actually, but pretty close
2022-01-08 07:18:29 <sosodank> so that's explained
2022-01-08 07:19:13 <sosodank> but -Os remains about .70% the performance of -O2. not world-ending but not great
2022-01-08 07:19:45 <psykose> -O3 has roughly +30% size vs what we had, -O2 is +20%
2022-01-08 07:19:48 <psykose> but i think it is fine to change
2022-01-08 07:19:56 <psykose> i guess i will set it to the RelWithDebInfo type
2022-01-08 07:20:26 <sosodank> or i can do so, but i'll happily let you take it. yeah i'd prefer -O2 with debug info, RelWithDebInfo is what i use everywhere elsx
2022-01-08 07:20:33 <sosodank> thanks a lot for helping me look into it!
2022-01-08 07:21:48 <psykose> the debug info is there with any type, as the same -g gets thrown on due to the $pkgname-dbg subpackage
2022-01-08 07:21:52 <psykose> and then split out into -db
2022-01-08 07:21:55 <psykose> -dbg*
2022-01-08 07:22:00 <psykose> !29312
2022-01-08 07:22:45 <sosodank> approved, thanks a lot, always a pleasure to work with alpine
2022-01-08 07:22:50 <sosodank> see y'all
2022-01-08 08:06:47 <mps> so, cmake and all these modern build systems are horrible
2022-01-08 08:07:10 <mps> s/modern/'smart'/
2022-01-08 08:07:10 <alpine-meetbot> mps meant to say: so, cmake and all these 'smart' build systems are horrible
2022-01-08 08:09:31 <psykose> this would have the same issue with any build system because the problem was that it is -Os
2022-01-08 08:12:39 <mps> I mean, horrible because they 'hide' -O (and not only this) compiler and linker flags in some 'smart' defs
2022-01-08 08:12:55 <psykose> nothing was hidden here
2022-01-08 08:13:05 <psykose> alpine sets -Os... it was -Os
2022-01-08 08:13:29 <mps> it is, I prefer to set these myself
2022-01-08 08:13:45 <psykose> it was set by 'yourself', from abuild.conf, like every other package
2022-01-08 08:14:09 <mps> no, set it in APKBUILD
2022-01-08 08:15:16 <psykose> i await the proposal to remove cflags from abuild.conf and copy them into every apkbuild thenm
2022-01-08 08:15:49 <mps> not bad idea
2022-01-08 08:21:00 <mps> psykose: anyway, good 'review' of cmake flags above
2022-01-08 08:21:28 <mps> I learned something about this reading backlog
2022-01-08 08:27:21 <psykose> there is more context in the cmake rebuild mr and the following stuff from it
2022-01-08 08:31:24 <mps> yes, and my conclusion is for some pkgs change from Nono is but not sure it is ok for all
2022-01-08 08:32:03 <mps> for example can-tools could stay with None because speed is not so important for it
2022-01-08 08:32:14 <ikke> I'm asking all maintainers to approve the changes so that they can decide whether it makes sense
2022-01-08 08:32:51 <ikke> mps: but MinSizeRel is -Os, which is our default anyway
2022-01-08 08:33:12 <psykose> a good idea, as much as i am frustrated by the fact this will take like 4 months at this pace
2022-01-08 08:33:27 <mps> ikke: ah, so None and MinSizeRel is nearly same?
2022-01-08 08:33:29 <psykose> no issue for me though, i get to feel cool doing rebases all the time :p
2022-01-08 08:33:41 <ikke> mps: yes
2022-01-08 08:33:58 <mps> ikke: then I'm ok with change
2022-01-08 08:34:30 <ikke> Maintainers seem to be happy about the pings
2022-01-08 08:34:54 <psykose> incidentally can-utils has an upgrade for v2021.08.0
2022-01-08 08:36:09 <mps> psykose: thanks for info
2022-01-08 08:39:30 <ikke> mps: can you mention the approval in the MR?
2022-01-08 08:41:21 <mps> ikke: will do somewhat later, I'm working on something unrelated to software now
2022-01-08 08:41:26 <ikke> sure, np
2022-01-08 09:52:49 <nmeum> psykose: does MinSizRel pass -DNDEBUG by default?
2022-01-08 09:52:54 <psykose> yes
2022-01-08 09:53:29 <psykose> which sounds like a mildly serious change at first, except _everything_ is used to that as everything is using the release targets generally in production
2022-01-08 09:54:01 <ikke> What does NDEBUG signify?
2022-01-08 09:54:09 <ikke> No debug?
2022-01-08 09:54:09 <nmeum> makes assert(3) a no-op
2022-01-08 09:54:12 <ikke> ok
2022-01-08 09:54:13 <psykose> aside from applications that ifdef it, asser()
2022-01-08 09:54:14 <psykose> ye
2022-01-08 09:54:16 <nmeum> See https://lists.alpinelinux.org/~alpine/devel/%3C1ZU8S8ERRPKR9.35CUIBSS8QET7%408pit.net%3E
2022-01-08 09:54:43 <psykose> it would be nice to add it to global cflags too, but note this does imply breakage for some things
2022-01-08 09:54:56 <nmeum> I don't think it should be enabled globally
2022-01-08 09:55:23 <psykose> me either
2022-01-08 09:55:23 <nmeum> if cmake does that in release mode as well upstream software will likely expect that but that isn't the case for most other packages
2022-01-08 09:55:28 <psykose> yep
2022-01-08 09:56:03 <psykose> it does it for Release/MinSizeRel, and projects themselves use NDEBUG sometimes with the same intention
2022-01-08 09:56:19 <psykose> there might be a singular thing out there that expects RelWithDebInfo always and afaik that one doesn't
2022-01-08 09:57:04 <psykose> but unlike straight up custom targets, or only release, or or or, i have never ever seen that
2022-01-08 09:57:43 <psykose> i have spent too much time in cmake hell as can be seen so far
2022-01-08 11:52:50 <mps> nmeum: any objection to !29315
2022-01-08 11:53:13 <nmeum> no, looks good
2022-01-08 11:53:24 <mps> thanks
2022-01-08 12:50:26 <indy> ericonr, so problem is that generated_types in influxdb_iox generates some code from proto files using protoc and after using rustfmt
2022-01-08 12:51:28 <indy> ericonr, problem is that rustfmt in crates is 0.10.0 but it has different options and commands compared to latest stable
2022-01-08 12:52:26 <indy> ericonr, what is needed is rustfmt-nightly, but it depends on nightly version of rustc
2022-01-08 12:53:25 <psykose> it can be built without 'nightly' by building it as part of rust
2022-01-08 12:53:36 <psykose> and i am working on this currently, to add rustfmt to packaged rust
2022-01-08 12:53:43 <psykose> along with wasm32 and clippy
2022-01-08 12:53:54 <indy> psykose, you are taking words from my keyboard :)
2022-01-08 12:54:06 <indy> psykose, count me in for testing :)
2022-01-08 12:54:13 <psykose> the hack there is rust itself builds itself with nightly :p
2022-01-08 12:55:32 <indy> psykose, yes that was conclusion to build it as part of rustc
2022-01-08 12:55:35 <psykose> i could split it, but the wasm support needs a wasm sysroot (that i already packaged in !28735 ), so it would depend on that anyway
2022-01-08 12:56:20 <psykose> or i guess wasm32 needs only wasi-libc, but i think to actually use it you need the full root
2022-01-08 12:56:42 <indy> psykose, do you have somewhere git with your work?
2022-01-08 12:57:10 <indy> psykose, what about also adding rls and other components downloaded using rustup
2022-01-08 12:57:15 <psykose> when i can run it locally i will open an mr, the last pass all cleared except for one split function cause a path changed
2022-01-08 12:57:22 <psykose> i can add rls too sure
2022-01-08 12:57:42 <psykose> not sure how much of these will actually build
2022-01-08 12:57:49 <psykose> outside of x86_64
2022-01-08 12:59:56 <psykose> if you want to use my current changes here you go https://gitlab.alpinelinux.org/psykose/aports/-/commits/rust , though it takes an hour to build each time so maybe it fails :p
2022-01-08 13:00:09 <psykose> have fun with it, i changed it from configure to the new format, and i am unsure if the cross config is correct as well
2022-01-08 13:01:10 <psykose> the entire wasi-sdk is in the other mr, but you can just remove the wasm32 targets and -wasm subpackage, and then it won't matter, for testing the other parts
2022-01-08 13:02:29 <psykose> indy: ^
2022-01-08 13:03:29 <indy> psykose, thanks, i know, yesterday i was rebuilding rust and in took quite long time
2022-01-08 13:57:32 <ericonr> indy: wow, that's messy. Shouldn't they make rustfmt releases for every compiler version or something?
2022-01-08 13:58:09 <ericonr> psykose: afaik rls is deprecated in favor of rust-analyzer, they just haven't made it official yet
2022-01-08 13:58:15 <psykose> sure, makes sense
2022-01-08 13:58:46 <psykose> rustfmt is tricky because nightly has technically different syntax.. so you would need to maintain a nightly fork just for nightly use (unless i am very wrong)
2022-01-08 13:59:15 <psykose> makes more sense to have 'only' the nightly version that also formats stable fine, but i did not look into what the actual reason is
2022-01-08 14:01:44 <ericonr> psykose: if you need nightly, you won't be using the distribution's rust toolchain though, right?
2022-01-08 14:01:55 <psykose> yep
2022-01-08 14:02:33 <ericonr> Could do the same for rustfmt, I guess 
2022-01-08 14:02:48 <psykose> do what the same
2022-01-08 14:02:57 <ericonr> Install with rustup if doing nightly stuff, package only the stable one
2022-01-08 14:03:04 <psykose> there is no stable one
2022-01-08 14:03:14 <psykose> the one you would be getting is the nightly one on stable
2022-01-08 14:03:20 <ericonr> Ooh
2022-01-08 14:03:30 <psykose> like everyone else that rustup add whatever rustfmt while one stable
2022-01-08 14:03:48 <ericonr> I hadn't understood that, thanks
2022-01-08 14:04:09 <psykose> it's just a compile-time requirement, and the rustc process is built with nightly technically
2022-01-08 18:16:38 <andypost[m]> Where I can fix link for pkgs.a.o the same way as https://gitlab.alpinelinux.org/alpine/infra/build-server-status/-/merge_requests/4
2022-01-08 18:19:10 <ikke> https://gitlab.alpinelinux.org/alpine/infra/aports-turbo
2022-01-08 18:25:55 <andypost[m]> ikke: thank you! filed https://gitlab.alpinelinux.org/alpine/infra/aports-turbo/-/merge_requests/47
2022-01-08 19:27:49 <indy> psykose, rust apkbuild from your branch failed on aarch64
2022-01-08 19:29:07 <indy> psykose, https://dpaste.org/1z5V
2022-01-08 22:24:35 <omni> I might be able to look at !29060 and !29061 again tomorrow but if anyone else want to take over, please do
2022-01-08 22:25:38 <omni> also, as nmeum pointed out, go 1.17.6 is released
2022-01-08 22:29:31 <omni> btw, I love this distro/project and everyone working on it :*
2022-01-08 22:41:07 <mps> omni: even me :D
2022-01-08 22:42:15 <mps> I don't like me on this project ;)
2022-01-08 22:51:07 <minimal> omni: you saw the issue I raised on the Nomad github? and their reply?
2022-01-08 23:00:42 <omni> mps: yes, you too <3
2022-01-08 23:01:04 <omni> I may not agree with everyone on everything, but I don't want to live in such a world either
2022-01-08 23:01:17 <omni> minimal: not yet, I don't think, I'll have a look tomorrow
2022-01-08 23:02:05 <mps> omni: ofc, also I disagree with some people but most people here are nice ones
2022-01-08 23:02:44 <mps> omni: and you are very cooperative though we disagree sometimes
2022-01-09 04:23:57 <psykose> indy: no idea what would cause that, but i will work more on it today
2022-01-09 04:57:52 <andypost[m]> omni: I did upgrade your MR to 1.17.6 and rebased (hope nothing lost)
2022-01-09 06:26:18 <ktprograms> Would it be possible to have flightgear's data files be part of the flightgear package (as a -data subpackage)? It would drastically improve the end user experience because currently it has to be download from sourceforge's very slow servers (around 500 kB per second). If it can be downloaded once to distfiles and packaged, it can then be installed from the Alpine mirrors which are generally much 
2022-01-09 06:26:24 <ktprograms> faster. The data tar is about 1.6 GB but I'm not sure what size it'll end up as when extracted and packaged.
2022-01-09 06:27:48 <psykose> 1.6gb is a little up there, though there are packages in that range anyway
2022-01-09 06:28:02 <psykose> does nobody but sourceforge mirror it already
2022-01-09 06:28:53 <dalias> what existing packages are in that range??
2022-01-09 06:29:50 <psykose> widelands is 1gb
2022-01-09 06:30:02 <psykose> a few other games too
2022-01-09 06:30:05 <psykose> some fonts
2022-01-09 06:30:38 <dalias> ah so another case of packaging game data
2022-01-09 06:30:52 <dalias> fonts that big?!
2022-01-09 06:31:03 <psykose> all the droid fonts are pretty huge
2022-01-09 06:31:14 <psykose> or was it noto
2022-01-09 06:31:17 <psykose> that big google one
2022-01-09 06:32:09 <ktprograms> psykose: I can't find any other download link
2022-01-09 06:32:24 <psykose> unlucky
2022-01-09 06:34:48 <ktprograms> Yes, takes almost an hour to complete
2022-01-09 06:38:15 <dalias> noto is so awful
2022-01-09 06:38:33 <dalias> i avoid it by having ubo block fonts for sites that use it
2022-01-09 06:38:45 <dalias> but having it installed would probably break that :/
2022-01-09 06:43:30 <ktprograms> PureTryOut: any opinion on this? (Since you're the maintainer)
2022-01-09 06:55:27 <psykose> i have firefox just block site fonts, though that is much more extreme
2022-01-09 06:57:59 <dalias> :)
2022-01-09 06:58:14 <dalias> yeah i like it per-site
2022-01-09 08:09:39 <PureTryOut> ktprograms: sorry, maintainer of what package?
2022-01-09 08:09:53 <psykose> flightgear
2022-01-09 08:10:14 <psykose> or you aren't actually, so no idea
2022-01-09 08:10:20 <PureTryOut> Yeah I'm not the maintainer of that lol
2022-01-09 08:10:44 <psykose> hope your morning is going well
2022-01-09 08:11:04 <PureTryOut> just woke up 😅
2022-01-09 08:11:07 <psykose> :3
2022-01-09 08:11:30 <boomanaiden154> The maintainer of flightgear is Thomas Kienlen. Not sure if they’re on IRC.
2022-01-09 08:13:39 <psykose> in the other channel
2022-01-09 08:13:40 <psykose> kmmndr
2022-01-09 10:51:28 <nmeum> Ariadne: could you elaborate a bit on what would need to be done to package rustc for riscv64? it seems to me that upstream has offical support tier 3 support for riscv64gc-unknown-linux-musl, does it simply take too long to bootstrap it in a qemu-user environment?
2022-01-09 10:52:15 <Ariadne> i mean, i have no idea, have not tried to bootstrap it, as i would assume it will take a long time
2022-01-09 10:52:43 <Ariadne> we need to add riscv64gc-alpine-linux-musl target, which does the things that aren't switched over upstream yet
2022-01-09 10:55:47 <nmeum> but that shouldn't be too complicated, no? at least the modifications made for existing tier 3 architectures (like ppc64le) in alpine-target.patch seem simple enough
2022-01-09 10:56:59 <psykose> alpine-target.patch only adds the triple doesn't it
2022-01-09 10:57:51 <psykose> none of the patches are arch specific
2022-01-09 10:58:00 <psykose> just musl fixes
2022-01-09 10:58:20 <nmeum> looks like it (I am looking at the rust aport for the first time)
2022-01-09 10:58:21 <psykose> so.. i would guess it kinda works, and maybe ends up needing one patch
2022-01-09 10:58:30 <psykose> i have spent many hours on the aport today and yesterday
2022-01-09 10:58:35 <Ariadne> yeah should only need the one patch
2022-01-09 10:58:49 <Ariadne> psykose: i want to move to building rust from git
2022-01-09 10:58:54 <nmeum> how did we bootstrap rust for existing architectures? cross-compilation from x86_64?
2022-01-09 10:59:00 <psykose> think so
2022-01-09 10:59:05 <Ariadne> psykose: so we can just replace the libc crate with patched one
2022-01-09 10:59:09 <psykose> hmmm
2022-01-09 10:59:26 <Ariadne> nmeum: yes, the bootstrap.sh will do it
2022-01-09 10:59:34 <Ariadne> nmeum: if it works
2022-01-09 10:59:36 <nmeum> great
2022-01-09 10:59:39 <Ariadne> nmeum: i was able to get it to do it for aarch64 anyway
2022-01-09 10:59:40 <psykose> i am presently quite drunk, what do you mean by 'from git', a specific git commit like for gcc?
2022-01-09 10:59:52 <Ariadne> psykose: i mean, i want to take our own git snapshots of rust
2022-01-09 10:59:55 <psykose> right
2022-01-09 10:59:57 <Ariadne> psykose: from our own git tree of rust
2022-01-09 10:59:59 <psykose> that makes sense
2022-01-09 11:00:16 <psykose> sure, that's orthogonal to what i'm doing
2022-01-09 11:00:24 <psykose> but i am not very sure how well this config.toml business works
2022-01-09 11:00:26 <Ariadne> psykose: which have the submodules swapped for ones that have our patches
2022-01-09 11:01:12 <psykose> i guess you mean patching the vendor/ folder
2022-01-09 11:01:23 <psykose> though, that can kind of be done anyway, but yes, it's easier with our own snapshot
2022-01-09 11:01:38 <psykose> no complaints from me
2022-01-09 11:02:13 <psykose> the existing apkbuild clears the checksums for that reason
2022-01-09 11:03:18 <mps> heh, I tried this 2 or 3 years ago (to bootstrap rust on aarch64 and armv7) but not using bootstrap.sh, requires effort
2022-01-09 11:03:43 <mps> actually I made it but it didn't accepted on alpine
2022-01-09 11:03:49 <nmeum> so…should we first build rust from git before we start adding support for riscv64?
2022-01-09 11:04:02 <mps> so I stop to waste time at such things
2022-01-09 11:04:50 <psykose> nmeum: i would say everything is orthogonal
2022-01-09 11:05:01 <psykose> but it is slightly easier to first get git going
2022-01-09 11:05:40 <psykose> afaik cogitri is mostly busy with other things currently, so you can take the lead on it if you want
2022-01-09 11:06:23 <psykose> don't think anyone would complain about some work being done ahead of time
2022-01-09 11:06:39 <mps> no for sure
2022-01-09 11:06:44 <nmeum> true I am just trying to evaluate what would need to be done first
2022-01-09 11:07:22 <psykose> and the issues of bootstrapping rust through qemu-user are mostly going to be the same regardless of the underlying apkbuild
2022-01-09 11:07:34 <psykose> so any headache of finding the compat patches, is going to just be copied in afterward
2022-01-09 11:27:32 <nmeum> https://rustc-dev-guide.rust-lang.org/img/rustc_stages.svg
2022-01-09 11:27:35 <nmeum> this looks………fun
2022-01-09 11:28:08 <Cogitri[m]> psykose: Yup, currently preparing my bachelor thesis so I can’t really get to bootstrapping rust currently
2022-01-09 11:28:20 <psykose> :)
2022-01-09 11:28:27 <psykose> back to work on the thesis
2022-01-09 11:28:53 <psykose> nmeum: that is automatic from rusts own build system
2022-01-09 11:29:09 <psykose> for the most part
2022-01-09 11:31:03 <nmeum> yes, for the most part the x.py should (hopefully) automate this
2022-01-09 13:40:50 <ktprograms> PureTryOut: Whoops, sorry, was looking at the wrong package for the maintainer
2022-01-09 13:41:36 <PureTryOut> Ah np
2022-01-09 15:16:33 <aparcar> Ariadne: thanks for the zlib patch
2022-01-09 15:16:39 <Ariadne> :)
2022-01-09 16:29:28 <aparcar> Ariadne: thoughts on libfetch vs uclient?
2022-01-09 20:31:54 <Ariadne> aparcar: i like uclient better, but i need to package it for alpine
2022-01-09 23:05:44 <exprez135> Ariadne: will apk support for macOS be announced to the apk-tools mailing list once it's all finished up?
2022-01-09 23:21:26 <aparcar> Ariadne: is the packaging process complicated? If not I might have a look
2022-01-10 01:32:25 <Ariadne> aparcar: not particularly.  if you can muddle through the openwrt build system, APKBUILDs are much easier :)
2022-01-10 01:33:01 <Ariadne> exprez135: it is for all intents and purposes finished
2022-01-10 01:41:47 <Ariadne> aparcar: some more thoughts on uclient: i would rather support libtls for the TLS API, so that we can swap the TLS implementation easily
2022-01-10 07:26:38 <ddevault> chrononicles of Python
2022-01-10 07:26:57 <ddevault> let's add a style linter to our test suite, then ship releases with style errors that fail our test suite
2022-01-10 07:27:07 <ddevault> I have fixed this problem in a dozen or more python packages
2022-01-10 08:10:33 <ikke> FUN™
2022-01-10 08:13:08 <aparcar> ddevault: which linter do you use/recommend?
2022-01-10 08:13:33 <ddevault> I recommend not putting a linter in your production test suite
2022-01-10 08:13:38 <ddevault> put it in your CI if you must
2022-01-10 08:14:06 <aparcar> Ariadne: talking to Daniel that's his plan, too. Isn't uclient already supporting that? I see us having three different TLS libs enabled by default
2022-01-10 08:14:09 <ddevault> whatever your take is on tabs or spaces, it doesn't actually change the behavior of the code and does not need to be part of the release acceptance criteria
2022-01-10 08:17:52 <Ariadne> aparcar: i didn’t see any support for libtls
2022-01-10 08:18:24 <aparcar> Ariadne: I'll check-in with daniel
2022-01-10 08:18:44 <aparcar> do you want to support anything outside openssl/mbedtls/wolfssl?
2022-01-10 08:19:13 <Ariadne> yes i want to support libtls, which will allow us to support bearssl via libtls-bearssl :)
2022-01-10 08:19:29 <aparcar> ack
2022-01-10 08:20:37 <Ariadne> libtls is an abstract API that can be implemented by any TLS implementation
2022-01-10 08:21:49 <aparcar> yea I get the idea. I wonder if they avoi
2022-01-10 08:21:54 <aparcar> did it for size reasons
2022-01-10 08:33:40 <ktprograms> Hi, can someone please review !29348 and !29376 ? Thanks!
2022-01-10 09:20:13 <mps> PureTryOut: good morning! https://krita.org/en/item/krita-5-0-2/
2022-01-10 09:20:46 <mps> 5.0 crashes often
2022-01-10 09:25:59 <PureTryOut> Good morning! Thanks for the news, I'll update the package
2022-01-10 09:30:14 <mps> tia
2022-01-10 12:46:00 <aparcar> Ariadne: I'm seeing a segfault with the new ABI field if the ABI version is 0, are integers not allowed to be 0? https://gitlab.alpinelinux.org/alpine/apk-tools/-/merge_requests/92
2022-01-10 13:03:08 <aparcar> Ariadne: never mind NULL just got "beautified" to 0 and that crashes the indexing process. I converted it to string and it works fine now
2022-01-10 13:18:24 <mighty17> ping!
2022-01-10 13:22:29 <mighty17> https://paste.debian.net/1226598/ any ideas why this isnt working
2022-01-10 13:23:06 <mighty17> alpine doesnt support lazy loading, so what can i do now? :(
2022-01-10 13:30:58 <mighty17> https://github.com/MightyM17/xf86-video-omap/ source
2022-01-10 13:38:26 <ddevault> what would be a reasonable next step to getting the wiki relicensing done
2022-01-10 14:07:50 <Ariadne> ddevault: i can ask the TSC what they think at tomorrow's meeting, but presumably we could just post a statement that all contributions past date X are CC-BY-SA
2022-01-10 15:04:33 <mps> PureTryOut: just tested upgraded krita, it works now without crashing. thank you
2022-01-10 15:05:20 <ddevault> Ariadne: I would have already done this if not for the fact that the copyright page is edit protected
2022-01-10 15:05:46 <Ariadne> ikke: can you help with this?
2022-01-10 15:06:03 <Ariadne> ikke: or do you want to spend a minute or two formally discussing it tomorrow
2022-01-10 15:12:59 <PureTryOut> mps: np, thanks for reporting!
2022-01-10 15:13:12 <mps> anyone know how to extract patch from pipermail malining list web 
2022-01-10 15:15:44 <mps> hm, download gzip mbox and run mutt -f $archive
2022-01-10 15:42:48 <zv> Ariadne: this commit seems to cause "implicit declaration of strlcpy" https://git.alpinelinux.org/apk-tools/commit/?id=2c3cef8787198455f70fb530157ba413b77a2b1e
2022-01-10 15:56:05 <ikke> Ariadne: We shouldn't just change the license without at least a notication for people to see. I think we can discuss tomorrow what further steps to take for already existing content.
2022-01-10 15:56:23 <ikke> We also somehow need a way to track the status of existing content
2022-01-10 16:06:37 <ikke> So I say, it would be good to have a concrete plan first
2022-01-10 17:22:09 <earboxer> What does it mean to be listed as a contributor in an APKBUILD?
2022-01-10 17:23:40 <earboxer> 1) having contributed something substantive to the buildscript or 2) being responsible/a stakeholder for the package (but less than maintainer)
2022-01-10 17:24:03 <ikke> The former
2022-01-10 17:24:15 <ikke> It does not really have a semantic meaning
2022-01-10 17:25:09 <mighty17> <mighty17> "alpine doesnt support lazy..." <- anyone? :(
2022-01-10 17:26:52 <mps> ikke: isn't removal of Contributor field already proposed
2022-01-10 17:27:49 <mps> to me more sense has to have two Maintainer fields
2022-01-10 17:28:16 <mps> yes, I know limitation of apk
2022-01-10 17:33:58 <ikke> https://lists.alpinelinux.org/~alpine/devel/%3Ctrinity-9f3437a2-9c6a-43d8-a5be-71264898e6fc-1499780412728%403capp-mailcom-lxa13%3E
2022-01-10 17:48:40 <mps> ikke: yes, I reread it now
2022-01-10 17:52:06 <staceee> hey there ! could someone merge 29322 please ? It is just a sxmo swayidle backport to bring a fix
2022-01-10 17:53:18 <ikke> !29322
2022-01-10 17:53:49 <ikke> staceee: s/allign/align
2022-01-10 17:53:57 <staceee> mhh as I think about it, maybe I should instead backport swayidle itself no ?
2022-01-10 19:38:55 <Xe> with community packages, can i bump older releases of alpine to have newer versions of packages that would otherwise change the minor version of them? IE can i bump tailscale in 3.14 to the most recent version?
2022-01-10 19:44:55 <ikke> Xe: we try to refrain from that, unless there is a compelling reason
2022-01-10 19:46:15 <Xe> fair enough, thanks ikke!
2022-01-10 19:47:40 <ikke> It's a delicate thing. The more we start backporting those changes, the more people come to expect it
2022-01-10 20:09:01 <mps> yes, set in stone rule: backport only sec and bug fixes
2022-01-10 20:22:22 <PJ[m]> Xe: are you planning on keeping tailscale up2date?
2022-01-10 20:23:21 <Xe> PJ[m]: tis the plan when i can remember to (disclaimer: i work there)
2022-01-10 20:24:30 <PJ[m]> I know you work there :P I was asking if you're going to do it consistently to not collide with each other (duplicate PRs)
2022-01-10 20:34:28 <ikke> Xe: You could provide your own repositories with packages built against each stable release
2022-01-10 20:39:09 <PJ[m]> it was decided that for now they won't provide them
2022-01-10 20:39:52 <PJ[m]> https://github.com/tailscale/tailscale/issues/230#event-5593823968
2022-01-10 20:40:24 <Xe> ikke: yeah i'm looking into doing that, but i probably won't push for it until the package format change to make our lives easier :)
2022-01-10 20:40:41 <ikke> You mean apkv3?
2022-01-10 20:40:45 <Xe> yeah
2022-01-10 20:40:59 <ikke> ah ok. Out of curiosity, how does that help?
2022-01-10 20:41:23 <Xe> reduces having to serve two package formats, unless apkv2 packages work with apkv3 natively
2022-01-10 20:45:48 <Xe> (i admit i haven't done much research there, i only really know that there's about to be a change and until there is it's probably less effort to bump things in community when we need to)
2022-01-10 20:46:59 <ikke> I don't think it will be switched for 3.16
2022-01-10 20:47:15 <ikke> so maybe 3.17, so at least one year out
2022-01-10 20:47:30 <ikke> (though, I could be wrong)
2022-01-10 20:48:12 <mps> ikke: need your help, u-boot developer want to check build on alpine and I don't know how to point to correct docker file
2022-01-10 20:48:51 <ikke> alpinelinux/build-base for example
2022-01-10 20:49:26 <ikke> It has alpine-sdk installed and a normal user setup with doas enabled
2022-01-10 20:50:31 <mps> ikke: many thanks
2022-01-10 21:37:21 <earboxer> My merge request !23376 is marked as stale, but it's no longer stale, it's ready to be reviewed.
2022-01-10 21:40:39 <Newbyte> Is it possible to tell apk to ignore a repository for one command?
2022-01-10 21:41:10 <ikke> Not that I'm aware of. What are you trying to achieve?
2022-01-10 21:41:32 <Newbyte> I have a repository of local packages and sometimes I want to ignore that when installing things
2022-01-10 21:42:03 <Newbyte> So far the best option seems to be to just comment it out
2022-01-10 21:42:25 <PJ[m]> you could have 2 repositories files and use --repositories-file
2022-01-10 21:42:46 <Newbyte> oh, yes. that is also an option
2022-01-10 21:43:27 <PJ[m]> other options are tagging a repository or doing a constraint on package that doesn't match that repo
2022-01-10 22:40:54 <Ariadne> zv: cause this on what?
2022-01-10 23:18:21 <zv> Ariadne: when compiling apk-tools from source at that commit
2022-01-10 23:18:31 <Ariadne> compiling how, and on what
2022-01-10 23:18:50 <Ariadne> the `makefile` based build system is deprecated, and assumes a musl-based system
2022-01-10 23:19:10 <zv> using the Makefile yes
2022-01-10 23:19:12 <zv> and on glibc
2022-01-10 23:19:24 <Ariadne> not supported
2022-01-10 23:19:28 <Ariadne> use meson
2022-01-10 23:19:58 <Ariadne> when apk3 is released, the makefile build system will be gone
2022-01-10 23:20:13 <Ariadne> it remains for running the testsuite basically
2022-01-10 23:20:27 <zv> ok. would it be rewritten to wrap meson at that point?
2022-01-10 23:20:54 <Ariadne> no
2022-01-11 03:05:29 <craftyguy> mps: why does the linux-edge kernel have a (very) different config than the linux-lts kernel?
2022-01-11 08:18:13 <mps> craftyguy: linux-edge is tweaked to be fast and use less RAM as possible while still trying to have similar config as -lts where it could
2022-01-11 08:19:08 <mps> craftyguy: and new and experimental options and drivers where makes sense
2022-01-11 17:42:30 <captainchris> hi everybody 
2022-01-11 17:43:20 <captainchris> I want to learning to create package, I created succesfully a package for libplist 
2022-01-11 17:44:25 <captainchris> and I wanted to create a package for wyeb browser. But there are not release on github 
2022-01-11 17:44:36 <captainchris> archlinux uses git://github.com/jun7/wyeb.git#branch=$_branch with branch master 
2022-01-11 17:45:08 <captainchris> but abuild says wyeb#branch=master is missing in checksums
2022-01-11 17:45:37 <ikke> abuild does not support that
2022-01-11 17:51:38 <zv> captainchris: https://github.com/jun7/wyeb/archive/refs/heads/master.tar.gz
2022-01-11 17:54:49 <ikke> for apkbuilds, you should use a fixed source.
2022-01-11 17:55:03 <ikke> PureTryOut already mentioned in #alpine-linux what you can use
2022-01-11 17:55:54 <captainchris> zv, How did you find the link ?? 
2022-01-11 17:59:26 <zv> captainchris: (1) any github or gitlab repo can give you .tar.gz (or .tar.bz2, .zip) links if you find the download button; sometimes formats are hidden from view. (2) you could become the upstream for that package; fork it and offer tagged releases.
2022-01-11 18:07:36 <PureTryOut> captainchris: are you ignoring me? I already told you how to get the link you need earlier today
2022-01-11 18:08:01 <PureTryOut> the link zv gave you is not the one you should use, as it's not a fixed source
2022-01-11 18:13:08 <mps> how we solve this https://gitlab.alpinelinux.org/alpine/aports/-/jobs/590651#L152
2022-01-11 18:13:26 <mps> it failed only on aarch64 and ppc64le
2022-01-11 18:13:41 <mps> configure: error: cannot guess build type; you must specify one
2022-01-11 18:13:43 <ikke> update_config_guess in prepare()
2022-01-11 18:13:53 <ikke> or update_config_sub, forgot which one
2022-01-11 18:14:03 <mps> ikke: thanks
2022-01-11 18:15:37 <mps> looks like it is update_config_sub
2022-01-11 18:20:23 <mps> and it didn't worked
2022-01-11 18:20:27 <mps> :(
2022-01-11 18:24:06 <mps> ikke: looks like both used and then it works
2022-01-11 18:24:26 <ikke> right, that's also an option :)
2022-01-11 18:26:27 <mps> btw, does this description is correct 'tools for manipulate PSF fonts'
2022-01-11 18:26:45 <mps> word manipulate or something else
2022-01-11 18:27:11 <psykose> manipulating
2022-01-11 18:28:29 <mps> psykose: first line from docs 'The PSFTOOLS are designed to manipulate fixed-width bitmap fonts,'
2022-01-11 18:28:51 <psykose> yes, that is correct
2022-01-11 18:29:18 <psykose> 'to manipulate' and 'for manipulating' are written differently
2022-01-11 18:29:31 <mps> aha, thanks
2022-01-11 18:29:51 <psykose> you could also write 'PSF font manipulation tools'
2022-01-11 18:29:55 <mps> will try to use 'to' then
2022-01-11 18:30:11 <psykose> or 'tools to manipulate.. '
2022-01-11 18:30:46 <mps> will rewrite as 'tools to manipulate fixed-width bitmap fonts'
2022-01-11 18:30:59 <mps> to be more in line with their doc
2022-01-11 18:31:04 <psykose> psf is more accurate as there are more than psf for fixed-width
2022-01-11 18:31:53 <mps> they added tools for some other fixed width fonts
2022-01-11 18:32:20 <mps> http://www.seasip.info/Unix/PSF/
2022-01-11 18:32:50 <mps> hehe, sinclair spectrum fonts also
2022-01-11 18:33:11 <psykose> hah
2022-01-11 18:33:38 <mps> psykose: thanks for help
2022-01-11 18:33:44 <psykose> anytime
2022-01-11 18:34:04 <mps> are you native english speaker
2022-01-11 18:41:45 <psykose> nope
2022-01-11 18:44:04 <mps> nor I am
2022-01-11 19:22:06 <captainchris> PuretryOut, No I don't ignore you.
2022-01-11 19:22:35 <captainchris> PureTryOut: But i don't understand how can search the commit on github 
2022-01-11 19:22:45 <captainchris> I didn't find the right way
2022-01-11 19:23:39 <PureTryOut> You don't understand how to get a commit hash from Github? 🤔 That's like the major function of Git, hosting commits lol
2022-01-11 19:23:43 <PureTryOut> *of Github
2022-01-11 19:23:44 <captainchris> I mapped all jun7 github to find the way to understand how search for the correct URL
2022-01-11 19:24:55 <ikke> captainchris: there is a green code button
2022-01-11 19:25:03 <ikke> and you can also guess the urls
2022-01-11 19:25:27 <ikke> github.com/name/project/archive/<hash>.tar.gz
2022-01-11 19:26:01 <captainchris> Ok the clone button down as ZIP
2022-01-11 19:27:24 <ikke> Yes, and then you change .zip to .tar.gz
2022-01-11 19:32:51 <captainchris> Ok
2022-01-11 19:33:46 <ikke> As a package maintainer, you sometimes need to figure out how to get things done a specific way, even if it's not immediately obvious or directly offered by upstreams
2022-01-11 19:45:49 <zv> captainchris: another cool thing, due to how branches and tags and commits work, https://github.com/jun7/wyeb/archive/daed8b86d15922f9ceabab7337f51472c91c5a3b.tar.gz
2022-01-11 20:55:33 <craftyguy> mps: hmm ok. I was expecting that the only real difference would be the kernel version and not primarily in the config differences. I'm currently building the linux-edge kernel with the kconfig from linux-lts, because I want the latest released kernel with a sane(-ish) config
2022-01-11 21:03:41 <psykose> the -edge config is more sane for desktop use
2022-01-11 21:04:13 <psykose> as someone that spends a lot of their time at load>nproc anyway
2022-01-11 21:05:10 <psykose> if you're building your own you also gain the benefit of being able to drop 90% of the modules, if it's what you wish for
2022-01-11 21:05:12 <mps> craftyguy: soon you will not build -edge with -lts config
2022-01-11 21:05:52 <craftyguy> ?
2022-01-11 21:06:00 <mps> and as psykose say -edge is more performance oriented
2022-01-11 21:06:19 <mps> craftyguy: 5.16 will be -edge in a few days
2022-01-11 21:06:22 <psykose> i wouldn't say the performance is very different (though i didn't check)
2022-01-11 21:06:37 <mps> psykose: it is
2022-01-11 21:06:37 <psykose> the RT changes in edge are just more responsive for things with higher nicelevel
2022-01-11 21:06:38 <craftyguy> mps: I just built 5.16 using the -lts kconfig... what's your point?
2022-01-11 21:06:40 <skarnet> living on the -edge
2022-01-11 21:07:08 <mps> craftyguy: some subsystems removed some added
2022-01-11 21:07:17 <craftyguy> `make oldconfig` ... ?
2022-01-11 21:07:37 <mps> and you don't have -lts confg
2022-01-11 21:07:51 <craftyguy> it's obviously based on it
2022-01-11 21:08:05 <mps> craftyguy: /o\
2022-01-11 21:08:09 <craftyguy> i.e., the functionality you disabled in -edge that exists in -lts, is still enabled
2022-01-11 21:08:45 <mps> craftyguy: try to run -lts armv7 in qemu and you will see
2022-01-11 21:08:53 <craftyguy> my point it, it seems weird that there are two kernels in alpine with completely differnet configs, but the names of those kernels only 'implies' the difference in versions
2022-01-11 21:09:21 <craftyguy> mps: red herring
2022-01-11 21:09:33 <mps> craftyguy: point is actually right this, else why we would have two kernels if they are nearly same
2022-01-11 21:09:44 <psykose> to have a different version
2022-01-11 21:09:47 <craftyguy> ^^
2022-01-11 21:09:55 <mps> psykose: no
2022-01-11 21:09:59 <psykose> yes
2022-01-11 21:10:15 <mps> no, I know why I'm added it
2022-01-11 21:10:39 <psykose> i'm not disputing your adding it or the reasons for anything in it
2022-01-11 21:10:50 <mps> and argued with ncopa more than 8 months about this :)
2022-01-11 21:11:03 <psykose> merely that having two near identical kconfigs but for multiple kernel versions is... very common
2022-01-11 21:11:29 <craftyguy> some folks want to run the latest release kernel (e.g. 5.16) because it includes newer drivers for their laptops, etc. but they don't want some riced out kernel config that may break things they use
2022-01-11 21:11:39 <psykose> and that people expecting the same kconfig for latest stable and latest lts is actually not rare
2022-01-11 21:11:48 <psykose> aside from that i don't really care at all what is in the configs or how they differ
2022-01-11 21:11:57 <psykose> just explaining the reasoning
2022-01-11 21:12:07 <craftyguy> psykose: thank you, you nailed it in your explanation
2022-01-11 21:12:07 <mps> psykose: we (or I) wanted something different to test new features and option tweaks
2022-01-11 21:12:20 <craftyguy> mps: maybe make a 'linux-mps' kernel then?
2022-01-11 21:12:47 <mps> craftyguy: and you will linux-craftyguy :)
2022-01-11 21:12:52 <craftyguy> or, would folks be opposed if I introduced a 'linux-mainline' or something that is as closely based on the -lts kconfig as possible, but tracks the latest released kernel version?
2022-01-11 21:12:55 <craftyguy> heh
2022-01-11 21:13:06 <psykose> but more to your point craftyguy , have you found anything actually broken on -edge or are you just scared by the difference
2022-01-11 21:13:14 <psykose> because ime it is just a usually better config
2022-01-11 21:13:18 <craftyguy> no I have, and I can't find my notes about it now :/
2022-01-11 21:13:23 <psykose> ah, unlucky
2022-01-11 21:13:32 <psykose> usually it was only a difference in enabled modules for drivers/subsys
2022-01-11 21:13:41 <psykose> from all the reports on issues and in these channels
2022-01-11 21:13:55 <mps> psykose: forced preempt is main diff
2022-01-11 21:14:01 <craftyguy> yeah IIRC it was something to do with namespaces? but yeah I don't remember exactly. I have requested things added to -edge in the past but was shot down by mps  :)
2022-01-11 21:14:13 <mps> and disabled useless options 
2022-01-11 21:14:19 <craftyguy> (the things I requested were in -lts already)
2022-01-11 21:14:45 <mps> craftyguy: I can't remember, sorry
2022-01-11 21:14:53 <mps> what you asked for?
2022-01-11 21:15:29 <craftyguy> yeah me neither, heh. I'm trying to pull up my IRC logs
2022-01-11 21:15:59 <mps> I mostly accept all reasonable requests, denied just two iirc
2022-01-11 21:17:07 <mps> well, I deny debugging options always
2022-01-11 21:17:47 <craftyguy> yeah I'm sure you're reasonable in what you accept, etc. this isn't about you. it's about how alpine is an exception with two kernels with different configs (what psykose said earlier)
2022-01-11 21:18:37 <mps> someone asked for RT scheduler and I denied it because it is not yet stable
2022-01-11 21:19:23 <psykose> there is also linux-gru linux-elm for some chromebooks, but those are specific
2022-01-11 21:19:37 <psykose> mps: iirc in 5.15 (or maybe it is 5.16 upcoming?) a lot of RT changes are now mainline
2022-01-11 21:19:45 <mps> craftyguy: psykose: as specially tailored kernels you can look at linux-elm and linux-gru in testing
2022-01-11 21:20:01 <craftyguy> mps: ah the thing I requested was debugfs
2022-01-11 21:20:12 <psykose> i would love to have a kernel with more of the -rt stuff, but i guess i can just build it myself
2022-01-11 21:20:22 <mps> psykose: RT is mainlined in 5.12 iirc from head
2022-01-11 21:20:26 <psykose> mhm
2022-01-11 21:20:28 <psykose> something like that
2022-01-11 21:20:42 <mps> psykose: but is not yet safe to use
2022-01-11 21:20:48 <craftyguy> is there any quantitative data on how having debugfs enabled impacts performance?
2022-01-11 21:21:31 <craftyguy> anyways, thanks for explaining your rationale wrt the -edge kernel
2022-01-11 21:22:05 <mps> craftyguy: there are some but I judged from my personal experience, building and using linux kernels for 28 (or 29) years
2022-01-11 21:22:47 <mps> and for a lot of different boards, arches and use cases
2022-01-11 21:22:50 <craftyguy> I mean, debugfs from the 90s may not be a good indication of debugfs in 2020s ;)
2022-01-11 21:23:26 <mps> craftyguy: yes, it is better now but still not enough
2022-01-11 21:23:29 <psykose> it's 16 years old
2022-01-11 21:23:37 <craftyguy> :P
2022-01-11 21:23:41 <psykose> though i have no idea what it does without mounting debugfs
2022-01-11 21:23:43 <mps> psykose: who?
2022-01-11 21:23:55 <psykose> debugfs
2022-01-11 21:24:30 <craftyguy> seems like there'd be no impact if it wasn't mounted, but I really have no idea how it works
2022-01-11 21:24:30 <mps> psykose: ah, debugfs, yes. I mean debugging in general
2022-01-11 21:24:41 <psykose> debugging i would never have on
2022-01-11 21:25:15 <psykose> this discussion slightly makes me miss gentoo
2022-01-11 21:25:23 <psykose> ..very slightly
2022-01-11 21:25:33 <mps> those who need/want debugging should build kernels for self
2022-01-11 21:26:04 <mps> I want small, simple and fast
2022-01-11 21:26:06 <psykose> did you know you can build the kernel with lto and clang? some fantastic thoughts for 5.16
2022-01-11 21:26:15 <psykose> :p
2022-01-11 21:26:31 <mps> psykose: I hate lto
2022-01-11 21:27:05 <mps> beside simple, small I forgot 'correctness'
2022-01-11 21:27:05 <PJ[m]> hey psykose, are you still working on enabling rust components like rustfmt?
2022-01-11 21:27:14 <psykose> i have them working
2022-01-11 21:27:32 <PJ[m]> but?
2022-01-11 21:27:42 <craftyguy> mps: what's incorrect about debugfs? I also don't know what you mean by 'correctness' with respect to the linux kernel, can you explain?
2022-01-11 21:28:09 <psykose> but i am too depressed to actually work on anything so i have not cleaned it up or made it work outside of x86_64/tested at all, with the other changes
2022-01-11 21:28:10 <mps> craftyguy: correctness is related to lto
2022-01-11 21:28:14 <psykose> clippy and wasm32 also works
2022-01-11 21:28:18 <craftyguy> ah
2022-01-11 21:28:37 <psykose> there is... some amount of file diff in the changes, in suspicious .rlibs that seems like a bunch of things are missing, even though i can build whatever after
2022-01-11 21:28:43 <psykose> shrug
2022-01-11 21:28:50 <psykose> the commit in the branch is there if you want to continue
2022-01-11 21:28:50 <PJ[m]> I'm in my emotional hole as well
2022-01-11 21:28:56 <psykose> i really have no energy for anything
2022-01-11 21:30:26 <mps> craftyguy: ok, I'm now busy with M1 and u-boot but will reconsider debugfs when I got some free time
2022-01-11 21:31:11 <psykose> PJ[m]: if you care to continue https://gitlab.alpinelinux.org/psykose/aports/-/tree/rust
2022-01-11 21:31:43 <psykose> note the hack in package()
2022-01-11 21:32:06 <PJ[m]> yeah, I read whole commit earlier today :>
2022-01-11 21:32:29 <psykose> there was a minor bug in it
2022-01-11 21:32:55 <psykose> set a depend to pkg-pkgver instead of pkg=pkgver
2022-01-11 21:33:06 <psykose> anyways, have fun with it
2022-01-11 21:33:10 <psykose> back to my hole
2022-01-11 21:34:59 <craftyguy> mps: thanks :)
2022-01-11 21:35:28 <craftyguy> but I also wonder if we should have a kernel that shares a config with the lts kernel. or maybe the -lts kernel should share config with the -edge kernel?
2022-01-11 21:36:28 <mps> craftyguy: I didn't 'set in stone' anything but not easy to do things without thinking, reading and listening first
2022-01-11 21:36:54 <psykose> we should just follow the issue to template standardize configs, mostly set -lts to follow -edge with adding in some modules from the difference, and then have a linux-crazyshit for whatever testing stuff
2022-01-11 21:37:04 <craftyguy> yeah :) thanks for listening!
2022-01-11 21:37:27 <mps> psykose: in some parts yes, I'm not against, but not in some important parts
2022-01-11 21:37:57 <mps> linux-lts and linux-edge are different things
2022-01-11 21:38:32 <mps> linux-edge is more like as testbed for linux-lts
2022-01-11 21:39:21 <mps> linux-lts is alpine official kernel
2022-01-11 21:39:57 <mps> linux-edge is well, edge
2022-01-11 21:46:25 <craftyguy> that analogy breaks down when you consider that the lts config is not at all like the edge config though. with alpine edge, a snapshot of that becomes 'frozen' as a stable release
2022-01-11 21:46:59 <mps> well
2022-01-11 21:47:47 <craftyguy> so right now, the 5.15 kernel in linux-edge is a different kernel than the 5.15 kenrel in linux-lts
2022-01-11 21:48:10 <mps> intentionally I'll say
2022-01-11 21:48:17 <minimal> mps: I started on a document for virt kernel settings as a way to pre-justify the MR I've been meaning to raise to disable some virt kernel settings. Must put the doc up on Github somewhere
2022-01-11 21:48:50 <craftyguy> mps: how is it a test bed if they're so different?
2022-01-11 21:49:23 <mps> craftyguy: new features and drivers
2022-01-11 21:50:10 <mps> minimal: yes, we need to work on -virt to unify where it makes sense and to clean some things
2022-01-11 21:51:52 <minimal> mps: yeah, that's what I start doing a few weeks ago when I started that document - to define what is needed for virt kernel, but also what is not needed and should be disabled.
2022-01-11 21:54:18 <mps> minimal: post me a note when you make something
2022-01-11 22:24:14 <minimal> mps: here's what I've got so far: https://github.com/dermotbradley/alpine-misc-notes/blob/main/virt-kernel-drivers.md
2022-01-11 22:24:55 <minimal> basically trying to decide what should be built-in, or module, or neither for linux-virt
2022-01-11 22:29:34 <mps> minimal: this is based on linux-virt not linux-edge4virt?
2022-01-11 23:02:09 <minimal> mps: well its more of a general list of what config I think should apply to a virt kernel, and then afterwards to see what the diffs are in relation to actual current kernel config
2022-01-12 00:40:21 <ktprograms> mps: Just wondering, did you try enabling CONFIG_ARM64_16K_PAGES for your M1 kernel configs? Is there a noticable performance improvement? (Found out about it from https://github.com/evansm7/vftool/discussions/9)
2022-01-12 01:09:16 <mla> could someone who can merge take a look: https://gitlab.alpinelinux.org/alpine/aports/-/merge_requests/28432
2022-01-12 01:09:16 <mla> need this for my map application to be used in postmarketos
2022-01-12 06:15:52 <psykose> was just wondering why pam login didn't work for the past few weeks.. and it's because shadow login got split to shadow-login...
2022-01-12 06:15:53 <psykose> classic
2022-01-12 06:18:03 <ikke> e2b4e2093e6bcc1bf6acb8f8d6d6b174605fc25b
2022-01-12 06:18:38 <psykose> yep, i read the commits when they were made
2022-01-12 06:18:45 <psykose> and seemingly it did not trigger something in my tiny head
2022-01-12 06:24:09 <ikke> not a clear explenation as to why
2022-01-12 06:24:31 <ikke> oh, the commit before does that
2022-01-12 06:24:54 <ikke> ce0392393120a7077e99d50dec6f8574efe7f09a
2022-01-12 07:14:50 <Ariadne> yes, goal is to start deprecating the shadow utils as util-linux implements replacements
2022-01-12 07:35:10 <psykose> aye, swapped to util-linux-login now
2022-01-12 07:56:24 <smcavoy>  for !29452 py3-build is required. It exists is in testing is it ok to create a MR to move to community?
2022-01-12 07:58:12 <ikke> smcavoy: I allreayd have an MR to do that
2022-01-12 07:58:31 <ikke> !28865
2022-01-12 07:59:48 <smcavoy> ikke: great!
2022-01-12 08:13:34 <Asmadeus> Hello, I'm not sure what's the polite delay before asking, but I usually get some feedback faster than a week. Would someone have time to check !29223 please? (distcc segfault fix when cross-building, and some ipv6-related fixes I sneaked in -- happy to remove the ipv6 part as it's workaroundable for people with ipv4, but I'm quite stuck on the segfault and would like to backport this to 3.15 once 
2022-01-12 08:13:40 <Asmadeus> it's accepted)
2022-01-12 08:15:07 <Asmadeus> happy to do something else in exchange if that helps :)
2022-01-12 10:50:53 <mps> good morning
2022-01-12 10:51:03 <mps> ncopa: !29453
2022-01-12 10:51:29 <mps> this version needs libcap-dev and libax25-dev
2022-01-12 10:52:42 <mps> libax25 is in testing and we/I should move it to main, but libax25 don't have much development in last years
2022-01-12 10:53:58 <mps> though it is quite stable in aprx package which I use
2022-01-12 10:54:30 <mps> ncopa: what is your opinion about this?
2022-01-12 10:57:57 <ncopa> hi! i think we can move it to main
2022-01-12 11:01:26 <mps> ok
2022-01-12 14:42:45 <Ariadne> fabled: what do you think about the MR !91 now
2022-01-12 14:42:49 <Ariadne> ...
2022-01-12 14:42:55 <Ariadne> apk-tools!91
2022-01-12 14:43:00 <Ariadne> ...
2022-01-12 14:43:10 <Ariadne> 
ACTION looks at algitbot with disapproval

2022-01-12 15:17:01 <Hello71> Ariadne: arrr/gh
2022-01-12 15:17:11 <Hello71> hm.
2022-01-12 15:20:11 <Ariadne> hm?
2022-01-12 16:41:45 <minimal> Ariadne: re shadow changes, I assume the 3.16 Release Notes will cover these sort of changes as otherwise someone upgrading to 3.16 and relying on pam-based login (e.g. using 2FA such as TOTP via pam) risks being locked out of machines
2022-01-12 16:42:12 <Ariadne> open an issue for it
2022-01-12 16:42:23 <Ariadne> so it does not get overlooked
2022-01-12 16:47:25 <aparcar> Ariadne: could you please elaborate on the "namespaces" regarding ABI versions and APK?
2022-01-12 16:48:47 <Ariadne> not ABI versions per se, but in alpine, packages provide all sort of semantic data in `provides` field
2022-01-12 16:49:00 <Ariadne> such as cmd:bash (package provides the "bash" command)
2022-01-12 16:49:32 <Ariadne> or pc:gtk+-3.0 (package provides the "gtk+-3.0" pkg-config module)
2022-01-12 16:50:11 <Ariadne> it seems like you could tag the abi this way, and also use it as a dependency
2022-01-12 16:52:14 <aparcar> so libgcc6 would be tagged as libgcc, I guess?
2022-01-12 16:52:53 <aparcar> point is currently the upgrade mechanism which tells a backend all installed packages. This tends to break with ABI versions being attached to the package name
2022-01-12 16:53:34 <aparcar> something like ${PROVIDES:-$NAME} could be used if there is only a single provides or the _abi less_ name comes first
2022-01-12 16:56:14 <Ariadne> oh
2022-01-12 16:56:23 <Ariadne> alpine solves that with soname providers
2022-01-12 16:56:30 <Ariadne> so:libgcc.so.6 or whatever
2022-01-12 17:01:09 <aparcar> I'm assuming there is some downstream tooling that fiddles those provides variables apart?
2022-01-12 17:03:57 <Ariadne> yes, abuild does it
2022-01-12 17:05:36 <Ariadne> aparcar: https://gitlab.alpinelinux.org/alpine/abuild/-/blob/master/abuild.in#L1412-1424
2022-01-12 18:45:40 <aparcar> Ariadne: thank you I'll check If that's a better solution than storing the ABI as an extra field
2022-01-12 18:46:15 <aparcar> Ariadne: Ideally packages would allow some more info fields like the passwd stuff "alternatives"
2022-01-12 18:46:30 <aparcar> currently I'm awkwardly storing it in an invented folder called /lib/apk/packages/
2022-01-12 23:25:41 <Ariadne> aparcar: i’m not sure how to design an alternatives system.  we can use the debian one, but it has a lot of design flaws
2022-01-12 23:25:54 <Ariadne> maybe having an ADB for alternatives makes sense
2022-01-13 00:08:06 <skarnet> disjunctions in dependencies are super hard
2022-01-13 02:31:18 <eschwartz> due to a random discussion in another channel, I happened to be looking at the meson APKBUILD, I wanted to see if it runs the meson testsuite. Turns out it doesn't. However, it is running something truly bizarre instead.
2022-01-13 02:31:34 <eschwartz> Does anyone know why alpine runs `python3 setup.py check` ?
2022-01-13 02:40:11 <zv> it's the python equivalent of 'make check'
2022-01-13 02:49:54 <eschwartz> no, it is not
2022-01-13 02:50:37 <eschwartz> make check is a target autotools habitually makes. python's testsuite runner, if properly configured, creates a target called *test*, not check
2022-01-13 02:51:01 <eschwartz> but you probably want to run pytest directly instead
2022-01-13 03:01:03 <Hello71> i think check is a mistake. i won't get into setup.py test vs pytest
2022-01-13 03:01:25 <Hello71> but i don't think that's a policy (yet)
2022-01-13 03:02:14 <eschwartz> is it cargo-culted or what?
2022-01-13 03:03:08 <eschwartz> because 'check' is a distutils command that never exits with non-zero status (unless you pass --strict) and the only thing it does anyway is check if the metadata has all three required fields name, version, url
2022-01-13 03:04:22 <eschwartz> as packaging validation goes, this is precisely as useful as a check function that runs `sleep 1 && echo "everything is probably fine" && return 0` (third part intentionally left in there specifically because it is wrong)
2022-01-13 03:04:27 <eschwartz> anyway
2022-01-13 03:04:36 <eschwartz> it would be neat to run the meson testsuite regardless :p
2022-01-13 04:22:56 <Ariadne> Hello71: check as in `setup.py check`?
2022-01-13 04:23:04 <Ariadne> check() {} is not a mistake :)
2022-01-13 04:26:47 <Ariadne> skarnet: well, the solver understands alternatives.  i'm talking about the /usr/bin/foo diverted to /usr/bin/foo.pkgname thing that debian does
2022-01-13 04:27:14 <Ariadne> Hello71: jirutka and i have an idea for abuild, that i was wondering if you had input about
2022-01-13 04:27:49 <Ariadne> Hello71: we want to propose an `$optpolicy` variable, which could set an optimization profile, which sets up a set of sane CFLAGS/CXXFLAGS/DFLAGS/GOFLAGS/etc
2022-01-13 04:28:38 <Ariadne> this way, we can choose to better balance size vs performance where it makes sense without having to copy around a bunch of CFLAGS :P
2022-01-13 04:33:49 <c7s> speaking of CFLAGS, is it considered a bug if an APKBUILD doesn't use the CFLAGS that apk sets for it (e.g. overrides -Os in a Makefile)?
2022-01-13 04:40:40 <boomanaiden154> c7s: There are a couple aports that do that already.
2022-01-13 04:41:14 <boomanaiden154> I had one program that I was packaging that completely crashed when compiled with space savings optimizations, so I just overrode the default variables in the build step.
2022-01-13 04:41:25 <boomanaiden154> I don’t believe it is generally encouraged without good reason though.
2022-01-13 04:43:09 <c7s> ah I was unclear, I meant the build system being weird and using something else
2022-01-13 04:43:14 <c7s> e.g. kakoune does this
2022-01-13 04:45:36 <Ariadne> c7s: hte build system should be fixed to respect ${CFLAGS}
2022-01-13 04:46:26 <Ariadne> boomanaiden154: if a program crashes when compiled with CFLAGS other than -O2, it usually means there is a bug in the program
2022-01-13 04:46:35 <Ariadne> go get that CVE :)
2022-01-13 05:05:11 <boomanaiden154> Ariadne: Yes. I need to go back and do some more debugging at some point. For some reason the test suite would throw a bunch of memory access violations when compiled with -Os which was pretty odd.
2022-01-13 05:05:21 <boomanaiden154> Also, I thought CVEs were only for security issues?
2022-01-13 05:05:54 <psykose> failing with -Os but not -O2 implies there are some security issues to find
2022-01-13 05:06:36 <Ariadne> :)
2022-01-13 05:06:40 <psykose> :)
2022-01-13 05:06:58 <Ariadne> same the other way around btw
2022-01-13 05:07:18 <Ariadne> if you are hitting undefined behaviour
2022-01-13 05:07:26 <Ariadne> that is by definition a security issue
2022-01-13 05:12:46 <boomanaiden154> Good points. Guess I’ll have to put some time into this. Would be pretty cool to have a CVE to my name.
2022-01-13 08:00:49 <aparcar> Ariadne: morning
2022-01-13 08:01:04 <aparcar> Ariadne: my current shell implementation is fairly simple and does the trick for openwrt
2022-01-13 08:01:12 <aparcar> could you elaborate on the issues with Debian?
2022-01-13 08:12:14 <Ariadne> dpkg-divert isn't properly tracked by apt
2022-01-13 08:12:26 <Ariadne> or at least wasn't 10 years ago before i stopped caring about the debian ecosystem
2022-01-13 09:07:06 <aparcar> Ariadne: how about we store alternatives as an array in the database and add a list function to dump all known alternatives?
2022-01-13 09:07:33 <Ariadne> alternatives i think could be implemented similarly to install_if
2022-01-13 09:07:34 <aparcar>  This way we already have the metadata in store and can alter on create a script for it
2022-01-13 09:08:11 <Ariadne> but want to get fabled's thoughts on the design before i start coding it
2022-01-13 09:08:48 <aparcar> I'd just make it an array for now but sure I'm fine waiting on fabled feedback
2022-01-13 09:09:24 <Ariadne> well, we want to store the command name, and its priority
2022-01-13 09:10:07 <Ariadne> so, /usr/bin/nc.netcat-openbsd should be stored at {"path":"/usr/bin/nc","priority":"100}
2022-01-13 09:10:12 <Ariadne> or whatever
2022-01-13 09:10:43 <aparcar> we need source and target and prio
2022-01-13 09:11:18 <aparcar> each package may contain multiple binaries so you need the specific paths
2022-01-13 09:12:25 <Ariadne> the target can be derived
2022-01-13 09:12:41 <Ariadne> path + '.' + pkgname
2022-01-13 09:21:50 <aparcar> Ariadne: what if netcat offers two binaries? say /sbin/nc and /sbin/nc_foo, which path would the alternative link to?
2022-01-13 09:22:26 <aparcar> gotta run, I'll think a bit about it and get back to you later today
2022-01-13 09:23:38 <Ariadne> why would a single package offer two binaries for the same command
2022-01-13 09:23:58 <Ariadne> this is a hypothetical situation that doesn't make any sense
2022-01-13 09:24:02 <Ariadne> bbl
2022-01-13 09:29:04 <fcolista> Ariadne, a CVE for jenkins has been relased and edge has been patched. I think i've to backport it, but the 3.15-stable does not have, like edge, the LTS version.
2022-01-13 09:29:29 <fcolista> Do you advice to move 3.15-stable jenkins from the weekly upgrade to LTS as well?
2022-01-13 09:29:34 <Ariadne> yes
2022-01-13 09:29:58 <fcolista> ok. I think I need to go back with that until the supported alpine
2022-01-13 12:05:50 <sicelo> Hi. I am enquiring about the i3blocks package, https://pkgs.alpinelinux.org/packages?name=i3blocks&branch=edge
2022-01-13 12:08:14 <sicelo> What could be done to get it in a stable release? It works very well in postmarketOS
2022-01-13 12:13:19 <psykose> ask the maintainer to move it to community
2022-01-13 12:14:16 <sicelo> Okay 
2022-01-13 12:36:16 <lmarz> Since the maintainer hasn't reacted within 2 weeks, i think !28943 is ready to be merged
2022-01-13 12:37:03 <mps> I like this `And no, we don't add "benchmark tuning Kconfig questions" to the kernel. We leave those kinds of games to companies that need to fake their benchmark numbers.`
2022-01-13 12:37:09 <mps> https://lwn.net/ml/linux-kernel/CAHk-=wjc3vpQfSTzJ8DwCVfmj6Mhq16d375cMPOgFB5A1EL2tw@mail.gmail.com/
2022-01-13 13:27:33 <Ariadne> mps: agreed, knobs for stuff like this should be minimal
2022-01-13 13:34:23 <Hello71> aren't the rcu tuning questions still in there
2022-01-13 13:34:46 <Hello71> also the klog length but at least the effect of those are obvious
2022-01-13 13:38:48 <mps> Hello71: yes rcu is there still
2022-01-13 13:44:53 <Ariadne> Hello71: what are your thoughts on https://gitlab.alpinelinux.org/alpine/abuild/-/issues/10057 ?
2022-01-13 13:45:22 <Hello71> not sure
2022-01-13 14:14:02 <ktprograms> FYI it seems that dl-cdn.a.o doesn't work with IPv6 currently (On the Woodpecker (General) Matrix channel there was a message about 'apk add' hanging while trying to fetch APKINDEX.tar.gz from dl-cdn that was fixed by disabling IPv6)
2022-01-13 14:16:51 <psykose> i seem to reach it fine with ipv6, but maybe it's a cache difference since it's cdn'd
2022-01-13 14:20:25 <ktprograms> Possible. Also it seems it was being used in a CI in a way that is basically docker-in-docker so I guess if it's working for you this can be ignored.
2022-01-13 14:20:51 <psykose> that doesn't make a difference unless something along the stack broke ipv6
2022-01-13 14:20:59 <psykose> but the easiest check is to do the same request on the host
2022-01-13 14:21:05 <psykose> two layers up
2022-01-13 14:21:16 <psykose> it's also possible the host does not have ipv6 configured
2022-01-13 14:21:33 <psykose> or the routing for the host, or the hosts isp
2022-01-13 14:22:10 <psykose> i've had broken ipv6 via misconfigured router before
2022-01-13 14:22:26 <psykose> try reaching known-good ipv6 addresses
2022-01-13 14:22:56 <ktprograms> It wasn't me who had the problem and I don't particularly feel like bothering to ask the person with the problem
2022-01-13 14:23:11 <psykose> you should
2022-01-13 14:25:47 <ktprograms> Ok I've asked
2022-01-13 16:04:21 <mps> here are new config for kernel 5.16 stable https://tpaste.us/lyRn
2022-01-13 16:05:03 <mps> anyone want/need some of these to be enabled
2022-01-13 16:09:36 <psykose> hid_nintendo=m i guess
2022-01-13 16:12:32 <mps> psykose: I thought about it but 'our' policy is to enable driver/option if someone need it. do you have this device
2022-01-13 16:12:55 <psykose> i have a controller that pairs as a switch controller, yes
2022-01-13 16:13:13 <psykose> but sure, leave it =n i guess
2022-01-13 16:13:25 <mps> ok, then I will enable it
2022-01-13 16:13:36 <mps> 'm' ofc
2022-01-13 18:03:43 <minimal> mps: CONFIG_ALIBABA_ENI_VDPA=m for virt kernel?
2022-01-13 18:04:39 <mps> minimal: I had a hope you will write 'remove virt kernel flavor' :)
2022-01-13 18:05:09 <mps> minimal: sorry, flavour ;)
2022-01-13 18:05:18 <minimal> mps: don't you mean "create separate cloud kernel flavour?" ;-)
2022-01-13 18:05:47 <mps> hehe
2022-01-13 18:06:29 <mps> minimal: ok, will add CONFIG_ALIBABA_ENI_VDPA=m to notes
2022-01-13 18:06:50 <psykose> i should take more notes
2022-01-13 18:07:15 <minimal> psykose: $1 or $100 notes? :-)
2022-01-13 18:07:25 <PureTryOut> oh 5.16 has Joycon (Switch controller) support? 😱
2022-01-13 18:07:28 <psykose> yes
2022-01-13 18:07:28 <PureTryOut> I need to get one now
2022-01-13 18:07:52 <psykose> minimal: $0 for me
2022-01-13 18:08:13 <PureTryOut> 
ACTION searches how expensive Joycons are

2022-01-13 18:08:15 <minimal> mps: I have updated my virt kernel settings doc locally, I'll ping you later when I push new version to Github
2022-01-13 18:08:24 <minimal> psykose: /bin/sh ? lol
2022-01-13 18:08:30 <ikke> psykose: I'm still playing with chunking CI builds, see if it's feasible
2022-01-13 18:08:31 <PureTryOut> Oh damn €70
2022-01-13 18:08:41 <psykose> minimal: you are insufferable
2022-01-13 18:09:03 <ikke> :D
2022-01-13 18:09:17 <psykose> ikke: glad to see it :) thought you forgot about me
2022-01-13 18:09:31 <ikke> You nerdsniped me :D
2022-01-13 18:09:42 <psykose> this isn't the first time is it
2022-01-13 18:10:41 <ikke> probably
2022-01-13 18:11:16 <psykose> glad to be of service
2022-01-13 18:25:46 <danieli> feel free to chuck !29353 at the CI when you have a moment
2022-01-13 18:27:12 <ikke> will be merged
2022-01-13 18:27:16 <danieli> thanks
2022-01-13 18:36:19 <mps> minimal: what about CONFIG_SCHED_CLUSTER, new on 5.16 and default is yes 
2022-01-13 18:36:35 <psykose> that should definitely be y
2022-01-13 18:37:27 <mps> that is my opinion also but want to hear you all
2022-01-13 18:37:48 <mps> also CONFIG_NETFILTER_EGRESS=m
2022-01-13 18:39:49 <ikke> https://lkml.org/lkml/2021/10/15/401
2022-01-13 18:39:58 <minimal> mps: I'd assume y for SCHED_CLUSTER for both phy and virt x86/x86_64 and aarch64 kernels for that also. Not sure whether y or n would make sense for other archs
2022-01-13 18:40:11 <mps> CONFIG_NVIDIA_WMI_EC_BACKLIGHT=m for equipped with NVIDIA hybrid graphics and drive LCD backlight
2022-01-13 18:42:41 <mps> minimal: I can't remember did the make oldconfig asked me to enable SCHED_CLUSTER when tested build on armv7, but for sure I will repeat and see
2022-01-13 18:43:15 <mps> are there NUMA on armv7? I doubt
2022-01-13 18:45:09 <mps> above new config is on x86_64, on aarch64 and armv7 some things missing but some other are added
2022-01-13 18:45:35 <mps> and didn't yet tried on riscv64
2022-01-13 18:47:33 <psykose> cluster seems distinct from numa proper
2022-01-13 18:47:42 <psykose> though i do doubt it for arm*
2022-01-13 18:52:45 <minimal> the cluster stuff depends on SMP, aren't SMP and NUMA 2 distinct things?
2022-01-13 18:54:01 <Hello71> logically numa is impossible without smp
2022-01-13 18:54:02 <minimal> so a multi-core processor is SMP and then if you have multi-socket motherboard NUMA can apply across the CPU chips
2022-01-13 18:56:18 <ikke> Hello71: why is that?
2022-01-13 18:58:06 <psykose> aren't the big arm boards with 2x cpus numa but not smp
2022-01-13 19:04:31 <minimal> psykose: https://elinux.org/images/0/00/Clement-smp-bring-up-on-arm-soc.pdf
2022-01-13 19:08:37 <psykose> do the ampere altra cpus have smp
2022-01-13 19:08:53 <psykose> these slides tell me nothing i care about
2022-01-13 19:10:04 <minimal> psykose: well they indicated "All the new ARM CPUs are now SMP capable"
2022-01-13 19:10:12 <psykose> this says absolutely nothing
2022-01-13 19:10:30 <psykose> the rpi4b next to me is most definitely not, unless i smoked some drugs
2022-01-13 19:10:52 <ikke> psykose: our Altera's cpuinfo mentions SMP
2022-01-13 19:11:05 <ikke> Linux usa9-dev1 5.10.27-0-lts #1-Alpine SMP
2022-01-13 19:11:05 <minimal> it doesn't say whether or not Arm boards are SMP capable, just the CPUs
2022-01-13 19:11:20 <ikke> I mean uname -a
2022-01-13 19:11:27 <psykose> is that SMP not just the kernel being built with smp support
2022-01-13 19:11:34 <psykose> does lscpu show more than one thread per core
2022-01-13 19:11:48 <ikke> ah no, 1 thread per core
2022-01-13 19:11:59 <ikke> but it is NUMA
2022-01-13 19:12:03 <psykose> yep
2022-01-13 19:12:12 <psykose> unless i am also forgetting what smp is
2022-01-13 19:12:16 <psykose> this is quite a weird feeling
2022-01-13 19:12:26 <ikke> I always understood it as hyperthreading
2022-01-13 19:12:30 <psykose> me too
2022-01-13 19:12:48 <minimal> uname on a RPI3 running 4.x kernel indicates "SMP" also
2022-01-13 19:12:58 <psykose> well, hyperthreading is just the name intel gave to it
2022-01-13 19:13:17 <psykose> smp in uname means kernel support
2022-01-13 19:13:25 <psykose> same for PREEMPT or RT
2022-01-13 19:13:30 <minimal> ikke: HT is something different isn't it - its a way to make e.g. 4 real cores appear to be 8 "virtual" cores
2022-01-13 19:13:55 <minimal> whereas SMP is to do with co-ordinating cores (whether "real" or "virtual"), right?
2022-01-13 19:14:53 <psykose> no, it's just the name of smp for intel branding, but there is some extra stuff that i have no idea what it is as anything talking about HT is just talking about smt
2022-01-13 19:15:10 <minimal> according to Wikilies (tm): "For each processor core that is physically present, the operating system addresses two virtual (logical) cores and shares the workload between them when possible"
2022-01-13 19:15:24 <ikke> Sounds like hyperthreading to me
2022-01-13 19:15:51 <minimal> ikke: that was from Wikilies' HT article
2022-01-13 19:16:20 <psykose> > In 2013, Intel dropped SMT in favor of out-of-order execution for its Silvermont processor cores
2022-01-13 19:16:25 <psykose> ok, maybe it is 'radically' different
2022-01-13 19:17:11 <minimal> whereas re: SMP Wikilies says: "In the case of multi-core processors, the SMP architecture applies to the cores, treating them as separate processors"
2022-01-13 19:17:19 <mps> SMP - Simetrical MultiProcessor (system)
2022-01-13 19:17:40 <minimal> so you can have SMP whether or not you have HT or non-HT CPUs
2022-01-13 19:17:52 <mps> minimal: right
2022-01-13 19:18:16 <mps> old intel machines where I started with this were SMP only
2022-01-13 19:18:41 <mps> in these times cores and HT didn't existed
2022-01-13 19:18:47 <mps> nor SMT
2022-01-13 19:19:09 <psykose> cores didn't exist? did you run computations on a microwave?
2022-01-13 19:19:11 <psykose> :p
2022-01-13 19:19:24 <mps> psykose: hehe
2022-01-13 19:19:37 <mps> in this time term core is not used
2022-01-13 19:19:47 <ikke> I suppose a cpu was just a single execution unit
2022-01-13 19:19:47 <mps> it was 'CPU'
2022-01-13 19:20:23 <mps> yes, imagine 4 486 cpus at around 40MHz
2022-01-13 19:20:55 <mps> 4 physically separate on one board
2022-01-13 19:21:05 <mps> code name 'sitka'
2022-01-13 19:21:15 <aparcar> Ariadne: well from my understanding it is possible for packages to contain all kinds of things. It also installs a binary to some random place and therefore you always need two (source, target) path. as an example, dropbear would install /usr/bin/dropbear which should be linked to /bin/ssh
2022-01-13 19:21:55 <minimal> mps: multi-socket motherboard
2022-01-13 19:22:35 <mps> minimal: yes
2022-01-13 19:26:25 <minimal> "lscpu -e" shows CPU, core, and socket info
2022-01-13 19:29:42 <mps> minimal: cpufetch -s legacy | tpaste => https://tpaste.us/41Qr
2022-01-13 19:29:46 <mps> :)
2022-01-13 19:30:33 <psykose> i think the new version has more info
2022-01-13 19:30:36 <psykose> there is a 1.01
2022-01-13 19:30:42 <psykose> for m1
2022-01-13 19:31:59 <mps> psykose: I looked few days ago but didn't saw 1.01 release tag
2022-01-13 19:32:21 <psykose> eh
2022-01-13 19:32:24 <psykose> the releases are just builds
2022-01-13 19:32:36 <psykose> the source code is the same, i don't think they override the tarballs on releases
2022-01-13 19:32:57 <psykose> i stopped caring about the average 'release' on github/lab a long time ago as very often they are neglected
2022-01-13 19:33:34 <psykose> and the default `Source code (tar.gz)` is the same as for a tag
2022-01-13 19:34:20 <mps> I don't like to upgrade till pkg is offiicially released
2022-01-13 19:36:02 <psykose> a tag is an official release
2022-01-13 19:36:16 <psykose> there are things with 3 years worth of tags above the release
2022-01-13 19:37:02 <mps> ah, yes
2022-01-13 19:37:19 <mps> I see, could upgrade it
2022-01-13 19:37:52 <psykose> also just a theory, maybe i heard wrong
2022-01-13 19:37:59 <psykose> buuuut if i'm right that paste should be more pretty
2022-01-13 19:38:05 <minimal> one of the projects that I maintain packages for often have RCs and releases triggered by changes being made only to their gitlab CI config and other files, I think once they went through 3 or 4 releases where the actual source code did not change
2022-01-13 19:39:21 <mps> modern times, as they say
2022-01-13 19:49:31 <ikke> any simpler way to do n=$(echo "$foo"
2022-01-13 19:49:45 <ikke> any simpler way to do n=$(echo "$foo" | xargs -n1 echo | wc -l)
2022-01-13 19:50:41 <Hello71> wc -w?
2022-01-13 19:50:46 <jvoisin> ^
2022-01-13 19:51:00 <ikke> oh, forgot about -w
2022-01-13 19:51:07 <Hello71> "word count"
2022-01-13 19:58:44 <ikke> thanks!
2022-01-13 20:29:57 <mps> psykose: https://tpaste.us/MznX with cpufetch 1.01
2022-01-13 20:30:50 <psykose> ahh, looks like i was wrong
2022-01-13 20:49:40 <ikke> With an offset of 1, would you expect something to start at element 1 or element 2?
2022-01-13 20:50:08 <ikke> (the first element or second element)
2022-01-13 20:51:34 <psykose> if it has to be offset by 1 and that is the only case then i would prefer it to be 2
2022-01-13 20:51:46 <psykose> really in any case aside from literal arrays i would prefer the 1-indexing
2022-01-13 20:52:06 <psykose> as in 1st element being 1, and so forth
2022-01-13 20:52:23 <psykose> with an offset, that would make it start at.. 2? which is fine, but idk the context
2022-01-13 20:52:34 <ikke> You do know the context :P
2022-01-13 20:52:41 <psykose> uhh
2022-01-13 20:52:58 <ikke> But that's what I was assuming as well, just wanted to verify if that was logical
2022-01-13 20:53:02 <psykose> 
ACTION appears confused

2022-01-13 20:53:10 <ikke> Try to guess the context
2022-01-13 20:53:17 <psykose> something something cute
2022-01-13 20:53:25 <psykose> (ci splitting?)
2022-01-13 20:53:29 <ikke> yes
2022-01-13 20:53:34 <psykose> but i'd want to see an example
2022-01-13 20:53:43 <ikke> I have a CI variable offset
2022-01-13 20:53:52 <ikke> for offset*
2022-01-13 20:54:02 <psykose> uhh just post the code
2022-01-13 20:56:22 <ikke> https://gitlab.alpinelinux.org/alpine/infra/docker/alpine-gitlab-ci/-/merge_requests/8
2022-01-13 20:59:00 <psykose> mm
2022-01-13 20:59:36 <psykose> yeah starting at 0 seems fine for the offset
2022-01-13 21:00:03 <ikke> It's a bit more challenging because each repo is built separately
2022-01-13 21:00:38 <psykose> yeah
2022-01-13 21:11:16 <bl4ckb0ne> can I have a review for 
2022-01-13 21:11:18 <bl4ckb0ne> !29216
2022-01-13 21:11:20 <bl4ckb0ne> please
2022-01-13 21:14:51 <mps> psykose: OT but fyi https://tpaste.us/EM8w this is from lscpu
2022-01-13 21:15:06 <ikke> psykose: CI is passing now :)
2022-01-13 21:15:13 <ikke> (how meta)
2022-01-13 21:15:36 <psykose> everything: not affected is nice to see for once
2022-01-13 21:15:47 <psykose> ikke: :) happy nerdsnipe
2022-01-13 21:15:55 <psykose> how does this function from the user perspective
2022-01-13 21:16:03 <ikke> Well, this is just step one
2022-01-13 21:16:21 <ikke> The next step would be to look at dynamic pipelines
2022-01-13 21:16:38 <ikke> The idea is to generate multiple build stages, each with x amount of packages
2022-01-13 21:17:24 <psykose> oh yeah
2022-01-13 21:17:26 <psykose> that sounds ideal
2022-01-13 21:18:08 <ikke> And by using the cache, the built packages should be available in the next stage
2022-01-13 21:18:57 <MY-R> bl4ckb0ne: typo here: https://gitlab.alpinelinux.org/alpine/aports/-/blob/bd6b2f75b29431ba20e3d81b048d1415411578fc/testing/sauerbraten/APKBUILD#L29
2022-01-13 21:19:56 <bl4ckb0ne> i dont see it
2022-01-13 21:20:00 <ikke> Me neither
2022-01-13 21:20:03 <jvoisin> is sauerbraten still maintained?
2022-01-13 21:20:36 <bl4ckb0ne> it is
2022-01-13 21:20:39 <MY-R> bl4ckb0ne: "usb"
2022-01-13 21:20:47 <bl4ckb0ne> ahhh yes thanks
2022-01-13 21:20:49 <ikke> d'oh :D
2022-01-13 21:20:59 <bl4ckb0ne> jvoisin: last release was 2021-12-29
2022-01-13 21:21:01 <psykose> does `install -Dm755 "$srcdir"/sauerbraten-server "$pkgdir"/usr/bin` not overwrite usr/bin with the thing
2022-01-13 21:21:12 <psykose> you need a -t or somethin
2022-01-13 21:21:22 <ikke> yea
2022-01-13 21:21:47 <psykose> also confused why there is both sauer_server and sauerbraten-server
2022-01-13 21:21:52 <psykose> but maybe that's package specific
2022-01-13 21:22:25 <bl4ckb0ne> yes there's a note  about that
2022-01-13 21:22:39 <jvoisin> bl4ckb0ne: https://sourceforge.net/projects/sauerbraten/files/ 2020
2022-01-13 21:23:12 <jvoisin> but you're right, it's still active: https://sourceforge.net/p/sauerbraten/code/commit_browser
2022-01-13 21:23:15 <jvoisin> thanks
2022-01-13 21:24:15 <MY-R> rugby server is 3/4 of a day full
2022-01-13 21:24:45 <bl4ckb0ne> psykose: they suggest packages to make their own script to launch the game
2022-01-13 21:24:48 <bl4ckb0ne> arch does the same
2022-01-13 21:24:51 <psykose> ah right
2022-01-13 21:24:53 <psykose> i missed that part
2022-01-13 21:25:02 <psykose> sure, looks good, just have to install it correctly
2022-01-13 21:25:11 <bl4ckb0ne> cant find it on the sourceforge page
2022-01-13 21:25:16 <bl4ckb0ne> i have it in the source
2022-01-13 21:25:36 <bl4ckb0ne> im also surprised that it compiled flawlessly on armv7 and s390x
2022-01-13 21:25:37 <jvoisin> sigh, now I'm pondering about packaging OpenMW
2022-01-13 21:26:01 <bl4ckb0ne> client ran on my side (x86_64)
2022-01-13 21:47:35 <bl4ckb0ne> ha found it psykose https://sourceforge.net/p/sauerbraten/code/HEAD/tree/bin_unix/readme.txt#l39
2022-01-13 21:47:42 <psykose> awesome
2022-01-13 21:48:16 <bl4ckb0ne> im open to naming bikeshed
2022-01-13 21:48:54 <psykose> it sounds perfectly fine to me
2022-01-13 21:49:22 <ikke> psykose: that's not bikeshedding :P
2022-01-13 21:49:41 <psykose> are you bikeshedding my bikeshedding feedback?
2022-01-13 21:49:45 <psykose> it's bedtime, ikke
2022-01-13 21:49:45 <ikke> :D
2022-01-13 21:49:51 <psykose> drink your warm milk
2022-01-13 21:50:03 <bl4ckb0ne> sleep tight, dont let the bikeshed bite
2022-01-13 21:50:07 <psykose> lmao
2022-01-13 22:01:49 <smcavoy> can anyone take a look at !28258 ? 
2022-01-13 22:02:40 <psykose> needing py3-pip in makedepends usually means you are missing some python modules
2022-01-13 22:02:48 <psykose> you can see the failed request it makes for what to add
2022-01-13 22:03:03 <psykose> though maybe this specifically explicitly calls pip for some reason, didn't check
2022-01-13 22:06:36 <smcavoy> thanks, will take a peek
2022-01-13 22:09:09 <smcavoy> superfluous inclusion of py3-pip :| removed.. 
2022-01-13 22:16:16 <ikke> smcavoy: psykose was right
2022-01-13 22:16:25 <psykose> with what
2022-01-13 22:16:42 <ikke> it does try to download packages
2022-01-13 22:16:44 <psykose> yeah
2022-01-13 22:16:46 <ikke> https://gitlab.alpinelinux.org/smcavoy/aports/-/jobs/593202
2022-01-13 22:17:00 <ikke> You miss pbr
2022-01-13 22:17:09 <psykose> mhm
2022-01-13 22:22:31 <smcavoy> ikke: indeed. fixing that now
2022-01-13 23:01:17 <Hello71> i wonder if apk should support a "repository announcement" message. e.g. apk update -X .../v3.14/community would print "WARNING: Alpine Linux 3.14 community repository is no longer supported."
2022-01-13 23:02:35 <Hello71> this wouldn't warn for using old ISOs. maybe that's good?
2022-01-13 23:03:12 <psykose> i am for it, but i'm sure this will get quite the bikeshed
2022-01-13 23:03:27 <psykose> maybe even mountain bikes will be inside the shed
2022-01-13 23:06:33 <Hello71> other ideas were alpine-base pulling in a new package "eol-warning" which prints warnings on... boot? apk? not sure; or a date-based mechanism (after 2022-01-01 this repository is out-of-date)
2022-01-13 23:09:45 <Hello71> also every release email should have a line "Per the Alpine Linux support schedule, Alpine Linux 3.14 community repo and Alpine Linux 3.11 all repos are now EOL."
2022-01-14 01:03:42 <zv> the APKBUILD reference mentions "internal dependencies"; what are these exactly?
2022-01-14 01:09:58 <psykose> the manpage does not have 'internal' in it, could you link to where
2022-01-14 01:17:37 <zv> psykose: https://wiki.alpinelinux.org/wiki/APKBUILD_Reference look for 'internal dependency'
2022-01-14 01:18:06 <psykose> package submodules
2022-01-14 01:18:14 <psykose> separate snapshot per submodule
2022-01-14 01:18:44 <psykose> i.e. the same thing that outside of snapshot() you usually just add to source with a git commit link for projects that don't put their submodules into the archive
2022-01-14 01:18:52 <psykose> but since it's snapshot these are uploaded to a different place
2022-01-14 01:19:13 <psykose> generally almost nothing needs snapshot()
2022-01-14 02:13:47 <anjan> It would be nice if the algitbot would automatically add tags for security upgrades by reading the diff
2022-01-14 02:13:57 <anjan> https://gitlab.alpinelinux.org/alpine/aports/-/merge_requests/29529
2022-01-14 02:14:13 <anjan> and tell the user to add a commit changing community/ if the package is in community
2022-01-14 02:14:31 <psykose> would require a parse of secfixes
2022-01-14 02:14:32 <psykose> doable
2022-01-14 02:14:45 <anjan> ok sweet, Ill submit an issue =D
2022-01-14 02:15:03 <anjan> btw, for that package, the version in stable is 2 versions behind
2022-01-14 02:15:22 <anjan> it's ok if I just upgrade to the latest in stable right?
2022-01-14 02:15:27 <anjan> it's all minor releases anyways
2022-01-14 02:15:32 <psykose> yes
2022-01-14 02:15:35 <psykose> depends on package
2022-01-14 02:15:42 <psykose> no breaking/funny stuff yea
2022-01-14 02:15:47 <psykose> tons of stuff gets forgotten like this, but thats ok
2022-01-14 02:15:54 <anjan> eh, prosody is pretty minimal so I wouldnt worry about it
2022-01-14 02:15:57 <anjan> thanks
2022-01-14 02:19:21 <anjan> Also, algitbot doesnt have a license
2022-01-14 02:19:30 <anjan> might be worthwhile to fix that asap
2022-01-14 02:20:14 <psykose> where is the repo
2022-01-14 02:23:01 <anjan> psykose: https://gitlab.alpinelinux.org/alpine/infra/compose/algitbot
2022-01-14 02:23:11 <anjan> or maybe thats not the repo. it's under the compose group
2022-01-14 02:24:46 <psykose> this is the irc bot
2022-01-14 02:24:56 <psykose> and is just using other components, not much to licence here really as it's some configs
2022-01-14 02:25:24 <anjan> oh derp
2022-01-14 02:25:36 <anjan> I guess I openned the issue to add the secfixes in the wrong repo
2022-01-14 02:25:42 <anjan> do you have the correct repo?
2022-01-14 02:25:59 <psykose> i've looked and not found it before
2022-01-14 02:26:06 <psykose> hence why i was curious
2022-01-14 02:26:59 <anjan> okayyy, let me know guys, Ill be happy to open the issue again/contribute a fix
2022-01-14 02:27:35 <anjan> I like how seriously alpine takes security - there's room for improvement but I like how easy it is for me to send patches fixes packages
2022-01-14 02:30:45 <ktprograms> If you're refering to the algitbot that issues stale warnings on MRs, I think it's just a Json config file: https://gitlab.alpinelinux.org/alpine/aports/-/blob/master/.gitlab/aports-qa-bot.json
2022-01-14 02:32:13 <anjan> ktprograms: where is the code that adds "aports:upgrade" to merge requests https://gitlab.alpinelinux.org/alpine/aports/-/merge_requests/29529
2022-01-14 02:35:18 <Asmadeus> Hello, could I have a review for !29223 please ? (trying at a different time of the day since nobody replied two days ago, and things look lively here :-D happy to look at something else in exchange or anything)
2022-01-14 02:41:36 <ktprograms> anjan: I have no idea, sorry
2022-01-14 02:41:50 <anjan> ok no worries. thank you ktprograms 
2022-01-14 04:05:12 <Ariadne> aparcar: ah, ok, that dropbear example makes more sense
2022-01-14 05:35:07 <ikke> I already have go code to do that
2022-01-14 05:35:16 <ikke> sorry, no context
2022-01-14 05:56:52 <Asmadeus> ikke: thanks!
2022-01-14 05:57:02 <ikke> you're welcome
2022-01-14 08:07:58 <ddevault> Ariadne: update on the wiki relicensing?
2022-01-14 08:08:22 <Ariadne> ddevault: implemented
2022-01-14 08:08:27 <ddevault> yeah, I just saw
2022-01-14 08:08:30 <ddevault> missing the relicensing thread
2022-01-14 08:08:41 <ddevault> I will write something up for the mailing list
2022-01-14 08:09:07 <Ariadne> the plan is to nuke the wiki and retain only the pages after the date listed
2022-01-14 08:09:19 <Ariadne> we want also to focus on the handbook when possible
2022-01-14 08:09:46 <ddevault> that's a good idea
2022-01-14 08:09:54 <ddevault> but we should not throw this trove of data off a cliff
2022-01-14 08:10:01 <ddevault> we should still ask for re-licensing for posterity's sake
2022-01-14 08:10:10 <Ariadne> no, we're not going to throw the data off the cliff, per se
2022-01-14 08:10:38 <Ariadne> the idea is to take the CC-BY-SA content from this wiki and eventually relaunch the wiki with only that content
2022-01-14 08:11:16 <Ariadne> a lot of wiki content is in desparate need of copyediting anyway, so i suspect this will solve itself
2022-01-14 08:11:36 <ddevault> the vast majority of the content is not CC-BY-SA
2022-01-14 08:11:42 <ddevault> so if you want to repurpose it, we need to relicense it
2022-01-14 08:11:56 <ddevault> draft: https://l.sr.ht/VHij.png
2022-01-14 08:12:44 <Ariadne> yes, this seems fine.  thanks for taking the initiative
2022-01-14 08:12:55 <Ariadne> or perhaps, we can have a wiki page for the relicensing?
2022-01-14 08:13:04 <ddevault> too late now
2022-01-14 08:13:08 <Ariadne> :D
2022-01-14 08:13:13 <ddevault> can you update the copyright page with a link to the mailing list archive
2022-01-14 08:13:16 <ddevault> and the notice on the home page as well
2022-01-14 08:13:29 <Ariadne> ikke: ^
2022-01-14 08:14:46 <ikke> Will do 
2022-01-14 08:15:03 <ddevault> erp, sent it to lists.sr.ht again
2022-01-14 08:15:05 <ddevault> I do that every time
2022-01-14 08:15:12 <Ariadne> oh
2022-01-14 08:15:17 <Ariadne> i just noticed that in your screenshot
2022-01-14 08:15:19 <Ariadne> haha
2022-01-14 08:16:04 <ddevault> https://lists.alpinelinux.org/~alpine/devel/%3CCH58XEZPYFYT.1MV854Y4P2Q2O%40taiga%3E
2022-01-14 08:19:25 <ikke> Maybe add note people should include their wiki username 
2022-01-14 08:20:17 <ddevault> I did
2022-01-14 08:21:11 <ikke> Sorry, missed that 
2022-01-14 08:21:15 <ddevault> np
2022-01-14 08:21:37 <ktprograms> As a side effect this will link everyone's wiki account with their email address lol...
2022-01-14 08:22:46 <ddevault> I don't see that as a problem, helps to better establish accountability for the copyright change
2022-01-14 08:22:59 <Ariadne> that information is already easily deducible anyway
2022-01-14 08:23:05 <ktprograms> Yeah sure, I also don't see any problem.
2022-01-14 08:23:10 <ktprograms> Ariadne: How?
2022-01-14 08:23:26 <Ariadne> well, first of all, you have to supply an email to sign up for the wiki.
2022-01-14 08:23:34 <ktprograms> Ariadne: it's optional
2022-01-14 08:23:36 <Ariadne> secondly, most people who have a wiki, also have a gitlab.
2022-01-14 08:23:49 <ikke> How about guest users 
2022-01-14 08:23:56 <Ariadne> and gitlab is leaky
2022-01-14 08:24:05 <ddevault> the wiki allows guest edits? bad idea imo
2022-01-14 08:25:10 <Ariadne> i dont think it does anymore, due to the spam issue
2022-01-14 08:51:56 <ikke> There are guest users 
2022-01-14 08:56:24 <Ermine> Do minor edits require consent for re-licensing?
2022-01-14 09:02:43 <adhawkins> Regarding this comment: https://gitlab.alpinelinux.org/alpine/aports/-/merge_requests/29497#note_206024 should I stop putting the changelog in the commit (it means it automatically gets put in the merge request, so seemed the easiest way to do it and not forget!)
2022-01-14 09:08:58 <mps> nowadays docs should be in git repo and every text and contribution should be sent as patch with 'Signed-By: name <mail-address>'
2022-01-14 09:37:50 <adhawkins> Also, is there an equivalent of the python '~=' operator for dependencies in an APKBUILD file?
2022-01-14 09:37:57 <adhawkins> https://www.python.org/dev/peps/pep-0440/#compatible-release
2022-01-14 09:38:22 <ikke> That works with apk as well
2022-01-14 09:38:32 <adhawkins> Cool, thanks ikke.
2022-01-14 09:46:21 <adhawkins> Doesn't seem to work the same way ikke. I've specified 'python3~=3.6', and get the following: https://paste.debian.net/1227076/
2022-01-14 09:47:20 <Ariadne> uhh
2022-01-14 09:47:29 <Ariadne> yeah, it does work the same way
2022-01-14 09:47:47 <Ariadne> what that is telling you is that there is no python3 3.6 package that is installable
2022-01-14 09:48:02 <Ariadne> do *not* create a frankenalpine to get one
2022-01-14 09:49:42 <adhawkins> If I'm reading the PEP document correctly, '~=3.6' means 'any version 3.x, as long as x is > 6'?
2022-01-14 09:49:53 <adhawkins> Sorry, >=6.
2022-01-14 09:50:29 <adhawkins> For a given release identifier V.N, the compatible release clause is approximately equivalent to the pair of comparison clauses: '>= V.N, == V.*'
2022-01-14 09:56:52 <Ariadne> then you want 'python3>=3.6'
2022-01-14 09:57:01 <Ariadne> ~=3.6 means 3.6.y
2022-01-14 09:57:27 <adhawkins> That's not what the PEP document says Ariadne, unless I'm reading it wrong...
2022-01-14 09:57:38 <Ariadne> i'm just telling you
2022-01-14 09:57:40 <adhawkins> They give specific example for x.y and x.y.z
2022-01-14 09:57:40 <Ariadne> what you want
2022-01-14 09:58:14 <Ariadne> if you want 'any version 3.x, as long as x is > 6', you want 'python3>3.6'
2022-01-14 09:58:41 <adhawkins> Then it appears that the '~' operator doesn't work the same way in Python as it does in apk...
2022-01-14 09:58:46 <adhawkins> Or am I being dense? Apologies if I am.
2022-01-14 09:59:05 <Ariadne> no, it does not
2022-01-14 09:59:14 <Ariadne> the '~' operator matches ruby
2022-01-14 09:59:43 <adhawkins> Ah ok, that was where I started from and ikke suggested they work the same way. Now I know they don't, I'll adjust the apk appropriately.
2022-01-14 09:59:46 <adhawkins> Thanks for clarifying Ariadne.
2022-01-14 10:00:03 <Ariadne> sorry, i did not know that python ~ worked differently than ruby
2022-01-14 10:00:13 <adhawkins> That's Ok, we got there in the end :)
2022-01-14 10:02:02 <ikke> Yeah, I was assuming ruby semantics as well 
2022-01-14 10:02:20 <adhawkins> No worries, thanks as ever for the assistance.
2022-01-14 10:02:58 <ikke> Nice thing about standards, you have so many of them 😑
2022-01-14 10:28:07 <aparcar> Ariadne: so a step could be to add an array in that format
2022-01-14 10:28:22 <aparcar> and implement a apk list --alternatives function
2022-01-14 10:28:31 <aparcar> some post install scripts can then iterate over that
2022-01-14 10:28:34 <Ariadne> i think we should open a gitlab
2022-01-14 10:28:38 <Ariadne> issue
2022-01-14 10:28:45 <aparcar> bueno
2022-01-14 10:28:45 <Ariadne> and spec this out properly
2022-01-14 10:28:56 <Ariadne> the implementation is the easy part
2022-01-14 10:29:19 <aparcar> https://gitlab.alpinelinux.org/alpine/apk-tools/-/issues/10798
2022-01-14 10:29:24 <aparcar> This was already opened
2022-01-14 10:30:07 <Ariadne> yeah reopen it
2022-01-14 10:32:22 <aparcar> Ariadne: please join the fun
2022-01-14 10:32:30 <Ariadne> :)
2022-01-14 10:32:36 <Ariadne> i will dump my thoughts there
2022-01-14 10:32:58 <Ariadne> right now, i am in the land of "cursed glibc-based alpine variant supporting CUDA"
2022-01-14 10:33:27 <Ariadne> which is the opposite of embedded i guess :P
2022-01-14 10:34:55 <Ariadne> aparcar: i'll probably package up ustreams/uclient today, and start testing apk with uclient-based backend
2022-01-14 10:35:16 <Ariadne> look for: feature/die-libfetch-die branch on my apk-tools tree
2022-01-14 10:35:17 <Ariadne> :D
2022-01-14 10:35:25 <aparcar> heh
2022-01-14 10:35:55 <aparcar> I'm glad both projects can offer some nice code to bring this forward
2022-01-14 10:37:51 <Ariadne> libfetch was the least offensive option at the time we picked it
2022-01-14 10:37:59 <Ariadne> but we're not married to it
2022-01-14 11:21:58 <ikke> ddevault: Ariadne https://wiki.alpinelinux.org/wiki/Alpine_Linux:Privacy_policy
2022-01-14 11:22:36 <ddevault> thanks ikke, can you update the red notice on the home page also?
2022-01-14 13:13:03 <Newbyte> could someone look into merging !29282? mirage is completely broken on Edge without this
2022-01-14 13:41:43 <Ariadne> Newbyte: done
2022-01-14 13:43:24 <Newbyte> thanks!
2022-01-14 14:26:19 <Ariadne> aparcar: it would be nice if we could have tagged releases of uclient, ubox, etc
2022-01-14 15:40:27 <ncopa> Ariadne: I updated qemu to 6.2 and rebased the ppc patch. I wonder if you could have a quick review? https://gitlab.alpinelinux.org/alpine/aports/-/blob/master/community/qemu/fix-ppc.patch
2022-01-14 15:40:41 <ncopa> this commit: https://gitlab.alpinelinux.org/alpine/aports/-/commit/fa3020b1d7207293a42f4260ed18f38d90c7ce9b
2022-01-14 15:40:59 <Ariadne> ncopa: we can actually simplify the patch
2022-01-14 15:41:07 <ncopa> ok?
2022-01-14 15:41:11 <Ariadne> using the same approach we used to fix gcc-go
2022-01-14 15:41:32 <Ariadne> other than that, this looks fine
2022-01-14 15:43:14 <ncopa> you mean simply #include <asm/ptrace.h>
2022-01-14 15:44:14 <Ariadne> yes, it should make the underlying register structure available if you build with _GNU_SOURCE
2022-01-14 15:44:41 <ncopa> let me try that
2022-01-14 16:05:12 <ncopa> seems to work. will use that instead. Thanks!
2022-01-14 16:17:47 <Ariadne> :)
2022-01-14 16:24:55 <Ariadne> nmeum: http://lists.busybox.net/pipermail/busybox/2022-January/089414.html
2022-01-14 16:47:13 <nmeum> ArMor[m]: nice, thanks!
2022-01-14 16:47:26 <nmeum> *Ariadne
2022-01-14 17:15:42 <z3ntu> Can someone please merge !29239?
2022-01-14 17:16:26 <z3ntu> ah they probably need a pkgrel bump
2022-01-14 17:16:57 <Ariadne> oh
2022-01-14 17:16:58 <Ariadne> yes
2022-01-14 17:16:59 <Ariadne> they do
2022-01-14 17:17:03 <Ariadne> i'll take care of it
2022-01-14 17:18:00 <z3ntu> cheers
2022-01-14 17:21:37 <Ariadne> mps: why does ncurses-terminfo and ncurses-terminfo-base install to different places
2022-01-14 18:13:58 <aparcar> Ariadne: what kind of tags do you want?
2022-01-14 18:14:01 <aparcar> of the release?
2022-01-14 18:14:13 <Ariadne> like v0.1
2022-01-14 18:14:16 <Ariadne> or whatever
2022-01-14 18:14:30 <Ariadne> so i can use URIs to download an archive :)
2022-01-14 18:14:48 <aparcar> fair
2022-01-14 18:14:59 <aparcar> I'll talk to the other devs
2022-01-14 18:17:45 <mps> Ariadne: I remember that we talked about this but forgot who decide this with ncurses
2022-01-14 18:18:01 <mps> I didn't
2022-01-14 18:46:56 <aparcar> Ariadne: dangole hello
2022-01-14 18:47:24 <ikke> o/
2022-01-14 18:49:24 <dangole> aparcar, Ariadne: is adding tags (instead of just referencing the commit hash and using the date as version as we do in our repo) a strict requirement for packaging the various OpenWrt sub-projects in Alpine?
2022-01-14 18:50:00 <Ariadne> dangole: the key thing is having a stable URI.  `abuild` is not really capable of using git to fetch sources yet.
2022-01-14 18:51:12 <dangole> Ariadne: we got source tarballs which our build-system auto-generates, see for example http://sources.openwrt.org/ubus-2021-08-09-a72457b6.tar.xz
2022-01-14 18:51:29 <Ariadne> oh, that will work then :)
2022-01-14 18:55:19 <dangole> Ariadne, aparcar: nice, so lets just use those then. adding tags would of course also be possible. i guess the snapshot downloads from gitweb are also reproducible, so that would work as well: https://git.openwrt.org/?p=project/ubus.git;a=snapshot;h=a72457b61df045d3c499a6211362b751710590d7;sf=tgz is not the same file, but also works (and is independent from that commit ever having been built by our buildbots)
2022-01-14 18:56:48 <Ariadne> alright, sounds good.  i did not know about sources.openwrt.org :)
2022-01-14 19:22:03 <aparcar> dangole: thanks for this quick solution
2022-01-14 19:31:20 <mps> we don't know if anyone uses arm kirkwood or mvebu boards with our u-boot, right?
2022-01-14 19:31:35 <mps> maybe pmOS?
2022-01-14 19:32:17 <minimal> mps: binned my SheevaPlug a long time ago, have a couple of PogoPlugs which I'm pretty sure are Kirkwood but never tried Alpine on them
2022-01-14 19:32:58 <mps> minimal: aren't these armv5?
2022-01-14 19:35:15 <minimal> mps: yeah ARMv5TE: http://www.swissdutch.ch/Marvell-88F6281.pdf
2022-01-14 19:36:07 <minimal> they came with u-boot 1.1.4 according to this: http://en.techinfodepot.shoutwiki.com/wiki/Cloud_Engines_Pogoplug_E02
2022-01-14 19:38:33 <mps> kirkwood and mvebu makes problem on new u-boot-tools to build, I'm asking to patch them out, we didn't build u-boot for them in last 3 years at last
2022-01-14 19:39:50 <mps> I think we don't need kwboot in u-boot-tools, we didn't had it in few previous releases
2022-01-14 19:46:12 <minimal> mps: does armhf even work for armv5? it thought that was v6 and above
2022-01-14 19:48:09 <ddevault> does boot hanging when mounting efivars on u-boot ring any bells
2022-01-14 19:48:19 <ddevault> doing some riscv64 stuff
2022-01-14 20:07:22 <mps> minimal: no, alpine armhf doesn't work on armv5
2022-01-14 20:08:08 <mps> and I doubt it works on armv6 except rpi zero
2022-01-14 20:08:20 <minimal> right, and Alpine only have armhf, armv7, and aarch64 - none of which support armv5 (Kirkwood) so why have armv5 support in Alpine u-boot?
2022-01-14 20:08:29 <mps> have no info that anyone tested it on anything else
2022-01-14 20:09:45 <mps> minimal: new changes in u-boot-tools want to build kwboot for mvebu and kirkwood and fails because musl don't have c_ospeed type
2022-01-14 20:10:23 <mps> hacking this will require time and thinking
2022-01-14 20:10:59 <mps> though dalias told that should be done for musl
2022-01-14 20:11:23 <mps> for now I will patch u-boot to skip this
2022-01-14 20:12:40 <minimal> mvebu seem to be armv7 & v8 cores so still should be supported
2022-01-14 20:14:43 <mps> minimal: uh, you ruined my 'solution' :_
2022-01-14 20:15:01 <mps> s/:_/:)/
2022-01-14 20:15:01 <alpine-meetbot> mps meant to say: minimal: uh, you ruined my 'solution' :)
2022-01-14 20:16:13 <minimal> doh!
2022-01-14 20:21:54 <mps> minimal: this must be of newest mvebu board
2022-01-14 20:22:29 <minimal> mps: some relevant info: https://01.org/linuxgraphics/gfx-docs/drm/arm/marvel.html
2022-01-14 20:23:26 <mps> we didn't had kwboot in previous u-boot-tools so lets see will anyone will ask for it
2022-01-14 20:23:39 <minimal> "EBU Armada family" and "EBU Armada family ARMv8"
2022-01-14 20:23:55 <ikke> KeyError: 'entry_points'
2022-01-14 20:23:57 <ikke> sigh
2022-01-14 20:32:01 <ddevault> efivars issue went away on its own
2022-01-14 20:32:02 <ddevault> ¯\_(ツ)_/¯
2022-01-14 20:49:31 <ddevault> did anyone make progress with rust on rv64?
2022-01-14 20:49:37 <ddevault> Ariadne maybe?
2022-01-14 20:53:05 <mps> is it reasonable to try with qemu-user on x86_64? I have doubt but maybe
2022-01-14 20:53:33 <ddevault> sure
2022-01-14 20:53:40 <ddevault> in fact it's more reasonable to do that than to use real hardware
2022-01-14 20:55:03 <mps> if I find time I could try then
2022-01-14 20:55:18 <ddevault> nice
2022-01-14 20:55:34 <ddevault> I'm working on bringing up a rv64 desktop
2022-01-14 20:55:44 <mps> maybe this could be faster on aarch64 than x86_64
2022-01-14 20:55:45 <ddevault> a surprisingly large number of things work, but no browser because firefox/rust
2022-01-14 20:56:34 <mps> iirc nmeum told he plan to bootstrap rust on rv64
2022-01-14 20:56:56 <ddevault> I tried a while ago, it's a huge pain in the ass
2022-01-14 20:57:05 <ddevault> but I think that some improvements were made in how alpine bootstraps rust in the meanwhile
2022-01-14 20:57:49 <ddevault> visurf works though :D
2022-01-14 20:58:11 <mps> nice
2022-01-14 20:58:13 <nmeum> I was just looking into it
2022-01-14 20:58:18 <nmeum> but I also figured that it's a huge pain in the ass
2022-01-14 20:58:36 <nmeum> and hence haven't done any substantial work
2022-01-14 20:58:37 <nmeum> on it
2022-01-14 20:58:40 <ddevault> oh :(
2022-01-14 20:59:04 <mps>  /o\
2022-01-14 20:59:35 <minimal> It'll be more "fun" once the kernel starts to add Rust modules
2022-01-14 21:00:34 <nmeum> I am sure it is possible but I wasn't too motivated since I don't use rust personally
2022-01-14 21:00:40 <mps> oh, especially on mips32 :D
2022-01-14 21:02:18 <ddevault> I don't use rust, but I do use firefox
2022-01-14 21:02:43 <mps> same
2022-01-14 23:13:27 <orbea> regretfully same ^
2022-01-14 23:57:19 <earboxer> Where is the source for pkgs.alpinelinux.org?
2022-01-15 00:22:45 <ptrc> IIRC it's aports-turbo on gitlab
2022-01-15 00:23:54 <ptrc> you might take a look at pmOS' reimplementation here: https://gitlab.com/postmarketOS/apkbrowser
2022-01-15 00:24:36 <ptrc> mostly because aports-turbo is notoriously hard to set up and actually get working on modern alpine
2022-01-15 02:04:16 <Ariadne> i can try to bootstrap rust on riscv64 next week
2022-01-15 02:04:34 <Ariadne> i worry it will take a lot of time though
2022-01-15 02:25:14 <Hello71> for a second i thought you said mips64
2022-01-15 07:59:43 <ddevault> re: !29575
2022-01-15 07:59:56 <ddevault> I cannot view this issue because I am on RISC-V and the fanciest web browser that works is NetSurf
2022-01-15 08:00:02 <ddevault> which is something I said would happen when moving to gitlab -_-
2022-01-15 08:00:14 <ddevault> anyway, someone else needs to review this, not me
2022-01-15 08:02:16 <boomanaiden154> What RISC-V hardware are you on?
2022-01-15 08:02:54 <ddevault> hifive unmatched
2022-01-15 08:09:47 <boomanaiden154> Wish I could get my hands on some RISC-V hardware like that.
2022-01-15 08:10:00 <boomanaiden154> What would it take to get something like Firefox working?
2022-01-15 08:10:11 <ddevault> someone willing to invest their blood, sweat, and tears into getting rust working
2022-01-15 08:10:28 <ddevault> I gave it a full week of my time several months ago before giving up
2022-01-15 09:00:26 <Ariadne> ddevault: i looked at it a little already, i need to make some minor fixes to bootstrap.sh, but should be able to do the bootstrap over the weekend
2022-01-15 09:00:52 <ddevault> rust bootstrap?
2022-01-15 09:00:56 <Ariadne> yes
2022-01-15 09:00:59 <ddevault> great
2022-01-15 09:01:01 <ddevault> good luck
2022-01-15 09:01:20 <Ariadne> assuming libc and std crates are properly ported anyway
2022-01-15 09:01:28 <Ariadne> i found out the hard way that std is still broken on s390x
2022-01-15 09:01:29 <ddevault> I think they are
2022-01-15 09:01:46 <ddevault> I complained about it several months ago and I recall my github issue being closed, though not sure if it's because they fixed it or because they thought I was annoying
2022-01-15 09:02:04 <ddevault> RISC-V and musl? why would anyone use that?
2022-01-15 09:02:10 <psykose> could be either
2022-01-15 09:02:20 <ddevault> could also be both
2022-01-15 09:02:22 <ddevault> ¯\_(ツ)_/¯
2022-01-15 09:02:46 <ddevault> let me know if you need someone to test or bounce ideas off of, Ariadne, at least for as long as my patience lasts
2022-01-15 09:03:11 <Ariadne> let me look at the libc crate sources
2022-01-15 09:03:17 <Ariadne> but there's stuff in std too, i think
2022-01-15 09:04:51 <Ariadne> looks like s390x might be fixed in 1.57 too
2022-01-15 09:04:55 <Ariadne> my lucky day
2022-01-15 09:05:46 <psykose> oh, that's current
2022-01-15 09:05:49 <psykose> even more work for you then
2022-01-15 09:07:07 <Ariadne> well
2022-01-15 09:07:17 <Ariadne> yes/no
2022-01-15 09:08:00 <Ariadne> harrumph
2022-01-15 09:08:32 <Ariadne> i want this to be reasonably fast, so i am thinking to rent some bare metal server from packet or whatever
2022-01-15 09:10:43 <ddevault> in other news, my new language can be fully bootstrapped on this RISC-V machine, including building and running its test suite, in 272 seconds
2022-01-15 09:11:19 <ddevault> about 10 seconds on x86_64
2022-01-15 09:11:20 <psykose> is your own s390x hardware not fast enough
2022-01-15 09:11:51 <psykose> ..or does equinix metal offer some fast rv64 i am not avare of
2022-01-15 09:12:11 <psykose> ddevault: that means it must not be a real language... it has to take 10 hours
2022-01-15 09:12:18 <psykose> because reasons!
2022-01-15 09:12:25 <ddevault> 
ACTION adds a 'sleep' command at the end of the makefile

2022-01-15 09:12:32 <psykose> that'll do
2022-01-15 09:18:18 <Ariadne> psykose: its the qemu-riscv64 that i am wanting fast x86 hardware for
2022-01-15 09:18:26 <Ariadne> qemu-user with s390x host is a shitshow
2022-01-15 09:18:27 <psykose> ahh, that's what you mean
2022-01-15 09:18:28 <Ariadne> totally broken
2022-01-15 09:18:35 <psykose> yes, good idea
2022-01-15 09:23:31 <ikke> Ariadne: you could get an rv64 container on nld5 if you want
2022-01-15 09:23:42 <Ariadne> what is nld5
2022-01-15 09:24:06 <ikke> our developer host from equinix
2022-01-15 09:24:16 <Ariadne> i meant what specs are nld5
2022-01-15 09:24:47 <ikke> 48 core, 256G mem
2022-01-15 09:24:52 <Ariadne> arm or x86
2022-01-15 09:25:12 <ikke> x86
2022-01-15 09:25:20 <Ariadne> hmm, might be good enough
2022-01-15 09:25:58 <Ariadne> i think we should try s390x bootstrap first
2022-01-15 09:33:02 <mps> https://blog.rust-lang.org/2022/01/13/Rust-1.58.0.html
2022-01-15 09:33:46 <mps> 
ACTION drinking morning coffee and reading news noticed this

2022-01-15 09:33:59 <Ariadne> ok
2022-01-15 09:34:03 <Ariadne> guess we upgrade to 1.58 :D
2022-01-15 09:34:14 <Ariadne> i'll do that first
2022-01-15 09:35:25 <psykose> this is the smallest rust changelog i've seen
2022-01-15 09:35:27 <psykose> finally
2022-01-15 09:35:33 <psykose> they have stopped adding 50 features every release
2022-01-15 09:36:30 <mps> nmeum: !29565 if you find time and know we need something there for rv64
2022-01-15 09:49:24 <Ariadne> ok rust 1.58 is building on my x86 machine
2022-01-15 09:49:39 <Ariadne> assuming it produces a sane binary i'll push that in an hour or so when its done
2022-01-15 10:49:10 <ikke> Ariadne: so should I create an rv64 container for you?
2022-01-15 10:49:32 <Ariadne> yeah
2022-01-15 10:49:37 <Ariadne> we can go that route
2022-01-15 10:49:42 <ikke> ok
2022-01-15 11:17:48 <psykose> Ariadne: is the rebase of link-musl-dynamically.patch not now incorrect? before it changed static to static-nobundle
2022-01-15 11:18:11 <Ariadne> psykose: no
2022-01-15 11:18:15 <psykose> now it's still static
2022-01-15 11:18:15 <psykose> hm
2022-01-15 11:18:37 <ikke> Ariadne: did you have vpn access?
2022-01-15 11:18:43 <Ariadne> ikke: no :)
2022-01-15 11:18:56 <Ariadne> i think i did a long time ago
2022-01-15 11:19:04 <ikke> We do have a wg bridge now
2022-01-15 11:19:16 <Ariadne> wireguard is probably easier yes
2022-01-15 11:29:14 <ikke> For some reason, dhcp is not working on rv64
2022-01-15 11:29:28 <ikke> For now, we set a static address
2022-01-15 11:30:35 <Ariadne> i shall create a wireguard pubkey when back at computer
2022-01-15 11:33:21 <mps> Ariadne: I created wiki somewhere on gl.a.o about this
2022-01-15 11:33:29 <mps> gitlab.a.o
2022-01-15 11:33:35 <Ariadne> about what
2022-01-15 11:33:51 <mps> using wg for alpine developers
2022-01-15 11:34:45 <mps> I think this is only wiki page we have on gitlab.a.o
2022-01-15 11:36:41 <mps> but where it is in this 'jungle' I wonder
2022-01-15 11:37:28 <ikke> mps: what command did I need to run again on the bridge to load the new config?
2022-01-15 11:40:19 <mps> ikke: wg setconf wg0 /etc/wireguard/wg0.conf
2022-01-15 11:41:14 <ikke> mps: thanks
2022-01-15 11:41:18 <Ariadne> hmm, i should write an ifreload(8) command for this
2022-01-15 11:42:34 <mps> 'ifdown wg0 && ifup wg0' maybe
2022-01-15 11:42:56 <mps> but for wg interface 'wg' command is fine
2022-01-15 11:43:33 <ikke> wg syncconhf?
2022-01-15 11:43:51 <mps> heh
2022-01-15 11:44:38 <mps> ikke: there is add-wg-user.sh in /root/ dir, my first try to make script to add new users to config
2022-01-15 11:44:46 <ikke> yes, I saw it
2022-01-15 11:45:06 <ikke> And users would need to receive the server public key, correct?
2022-01-15 11:45:26 <mps> users should copy it from wiki page
2022-01-15 11:45:32 <ikke> right
2022-01-15 11:45:48 <mps> how to find this wiki on gl, I'm lost
2022-01-15 11:46:17 <ikke> https://gitlab.alpinelinux.org/alpine/infra
2022-01-15 11:46:54 <ikke> https://gitlab.alpinelinux.org/alpine/infra/infra
2022-01-15 11:46:56 <ikke> that one
2022-01-15 11:47:57 <mps> https://gitlab.alpinelinux.org/alpine/infra/infra/-/wikis/Alpine-wireguard-VPN
2022-01-15 11:48:53 <mps> note: hub public key is: hm9zrdHJU2fJsrKnRqo4xXlqbsdNC0CfTySwOzdBwm8=
2022-01-15 11:49:29 <Ariadne> mps:  if you are using ifupdown-ng, you can just use `use wireguard`
2022-01-15 11:49:35 <Ariadne> instead of all that post-up stuff
2022-01-15 11:50:49 <mps> I'm not sure did I installed ifupdown-ng on hub
2022-01-15 11:51:28 <mps> and I didn't yet found time to test and learn ifupdown-ng to use it in production systems
2022-01-15 11:51:38 <ikke> it is installed
2022-01-15 11:52:48 <mps> ah I see
2022-01-15 12:13:47 <ikke> syncconf does not disrupt existing connections
2022-01-15 12:15:51 <mps> ikke: you mean setconf
2022-01-15 12:16:24 <ikke> "This is much less efficient than setconf, but has the benefit of not disrupting current peer sessions. "
2022-01-15 12:16:44 <mps> ikke: right
2022-01-15 12:17:08 <mps> I need another coffee, obviously
2022-01-15 12:18:10 <mps> just finished adding terminus 12x24 bold fonts to u-boot and tested it
2022-01-15 12:19:07 <ikke> mps: can we upgrade deu7 (wg bridge) to alpine 3.15?
2022-01-15 12:19:28 <ikke> it's now 3.14
2022-01-15 12:19:29 <mps> very nice for machine with high resolution displays, no need to search for loupe 
2022-01-15 12:19:51 <mps> ikke: sure
2022-01-15 12:19:58 <psykose> i think there was a 16x32 one too
2022-01-15 12:20:12 <mps> psykose: not yet merged
2022-01-15 12:20:14 <psykose> ah
2022-01-15 12:20:29 <mps> 16x32 is in kernel
2022-01-15 12:20:52 <mps> and 16x32 is too big imo
2022-01-15 12:21:43 <mps> tomorrow I will try to add 12x24 to kernel, if something doesn't IRQ me
2022-01-15 12:23:16 <mps> and I started to add this 4-5 years ago to kernel but never finished, someone added 16x32 in meantime and I lost incetive
2022-01-15 12:27:39 <ikke> ok, done
2022-01-15 12:27:52 <ikke> hmm
2022-01-15 12:28:20 <mps> wg0 is not up
2022-01-15 12:28:34 <ikke> yeah
2022-01-15 12:28:56 <mps> it is removed from interfaces
2022-01-15 12:29:10 <ikke> Oh, that is what linode does by default
2022-01-15 12:29:27 <mps> ah yes
2022-01-15 12:29:59 <mps> there is something to set so then they will not remove anything
2022-01-15 12:30:03 <ikke> Yes
2022-01-15 12:31:06 <ikke> disabled it
2022-01-15 12:31:38 <ikke> but now we need to restore the wg config
2022-01-15 12:32:02 <mps> do you want to do this or I?
2022-01-15 12:32:41 <ikke> Oh, found the backup
2022-01-15 12:32:57 <mps> nice
2022-01-15 12:33:47 <mps> where  is it?
2022-01-15 12:34:00 <mps> would be good note this
2022-01-15 12:34:37 <ikke> /etc/network/.interfaces.linode-last
2022-01-15 12:35:09 <mps> ah, hidden :)
2022-01-15 12:35:19 <ikke> dmvpn is still down though
2022-01-15 12:37:32 <ikke> ok, restarting zebra service fixed it
2022-01-15 12:40:19 <mps> cat /proc/sys/net/ipv4/ip_forward => 0
2022-01-15 12:40:35 <mps> ikke: ^
2022-01-15 12:40:40 <ikke> right
2022-01-15 12:40:59 <Ariadne> ifupdown-ng has options for configuring that on a per interface basis too
2022-01-15 12:41:45 <mps> we need this set on ifup
2022-01-15 12:41:45 <ikke> forward-ipv4
2022-01-15 12:45:35 <ikke> fordwarding is enabled now
2022-01-15 12:46:07 <mps> i did it already with echo
2022-01-15 12:46:18 <ikke> ok
2022-01-15 12:49:02 <ikke> mps: everything working again?
2022-01-15 12:49:17 <mps> I think so
2022-01-15 12:50:01 <mps> are you editing /etc/network/interfaces
2022-01-15 12:50:11 <ikke> I had it still open
2022-01-15 12:51:02 <mps> for now we can add 'post-up /bin/echo 1 > /proc/sys/net/ipv4/ip_forward'
2022-01-15 12:51:13 <ikke> I have already added forward-ipv4 to yes
2022-01-15 12:51:33 <mps> aha
2022-01-15 12:51:41 <mps> we should test this
2022-01-15 12:51:45 <ikke> I tried :P
2022-01-15 12:51:50 <ikke> but you already set it to 1
2022-01-15 12:51:52 <mps> good
2022-01-15 12:52:04 <ikke> so we should it to 0 again
2022-01-15 12:52:11 <mps> I'll reset now
2022-01-15 12:52:19 <ikke> ok
2022-01-15 12:52:53 <ikke> I'll ifdown and ifup eth0 now
2022-01-15 12:53:12 <ikke> hmm, it remains 0
2022-01-15 12:53:18 <mps> ok, just wanted to ask you to do this
2022-01-15 12:53:21 <ikke> oh, not for eth0
2022-01-15 12:53:26 <ikke> eth0 is 1 now
2022-01-15 12:53:32 <ikke> cat /proc/sys/net/ipv4/conf/eth0/forwarding
2022-01-15 12:53:53 <mps> we need it also for wg0
2022-01-15 12:53:57 <ikke> ok
2022-01-15 12:54:38 <ikke> done
2022-01-15 12:55:44 <mps> looks like we still global ip_forward=1
2022-01-15 12:55:59 <mps> need*
2022-01-15 12:56:53 <Hello71> https://www.kernel.org/doc/Documentation/networking/ip-sysctl.txt "ip_forward: This variable is special, its change resets all configuration parameters to their default state (RFC1122 for hosts, RFC1812 for routers)"
2022-01-15 12:58:09 <mps> ikke: try this also for gre1 interface
2022-01-15 12:58:11 <ikke> yes
2022-01-15 12:58:13 <ikke> was just doing trhat
2022-01-15 12:58:47 <ikke> done
2022-01-15 12:58:56 <mps> that is, works now
2022-01-15 12:59:04 <ikke> good
2022-01-15 12:59:04 <Hello71> ddevault: https://gitlab.alpinelinux.org/alpine/aports/-/merge_requests/29575.patch
2022-01-15 12:59:59 <ikke> mps: I don't think we even need it for eth0, do we
2022-01-15 13:00:12 <ddevault> thanks Hello71
2022-01-15 13:00:17 <ikke> It's only routing between wg0 and gre1 iirc
2022-01-15 13:00:17 <ddevault> +1 from me
2022-01-15 13:00:42 <mps> ikke: maybe if we want to ssh over wg to hub
2022-01-15 13:00:59 <mps> uh
2022-01-15 13:01:05 <ikke> Then you'd just use the hub address
2022-01-15 13:01:09 <mps> no, we don't need it 
2022-01-15 13:01:49 <mps> yes, but as control freak I want everything enabled from my deepest room in brain :)
2022-01-15 13:02:13 <mps> *instinct*
2022-01-15 13:02:31 <ikke> We can disable it, see if it's necessary, and enable it again in that case
2022-01-15 13:02:46 <mps> we don't need it actually because we can ssh to public address
2022-01-15 13:03:16 <ikke> disabled it manually now
2022-01-15 13:04:01 <mps> ok
2022-01-15 13:05:09 <ikke> Let me see if I can setup wg myself
2022-01-15 13:06:08 <mps> you can, I have no doubt :)
2022-01-15 13:10:23 <ikke> Does it work behind NAT?
2022-01-15 13:11:49 <mps> yes
2022-01-15 13:17:59 <ikke> Oh, I should reserve your macbook address in netbox
2022-01-15 13:18:23 <ikke> I only had 2 and 3 reserved
2022-01-15 13:23:17 <mps> uh, I also forgot to do this
2022-01-15 13:23:42 <ikke> https://netbox.alpin.pw/ipam/prefixes/44/ip-addresses/
2022-01-15 13:25:58 <ikke> cool, it's working now
2022-01-15 13:26:35 <ikke> disabled openvpn and enabled wg
2022-01-15 13:28:18 <mps> \o/
2022-01-15 13:41:45 <Ariadne> working on s390x bootstrap
2022-01-15 13:41:49 <Ariadne> cause like
2022-01-15 13:41:55 <Ariadne> i can do that easily
2022-01-15 13:44:44 <ikke> Ariadne: I have a container ready btw. Should I use your ssh keys from gitlab?
2022-01-15 13:44:56 <Ariadne> yes
2022-01-15 13:46:06 <ikke> Done, you can access it once you've set wg setup
2022-01-15 15:44:07 <aparcar> Ariadne: please let me know about the ports of uclient 
2022-01-15 15:44:19 <Ariadne> i'll get to it in a bit
2022-01-15 15:44:24 <Ariadne> trying to get this rust stuff underway
2022-01-15 15:52:46 <Ermine> Sorry for interrupting for such an unimportant thing, but it seems like this line https://gitlab.alpinelinux.org/alpine/aports/-/blob/master/main/openrc/modules.initd#L12 is unused. If I'm correct, I'll open a MR removing it
2022-01-15 15:53:08 <Ariadne> Ermine: it is used by `veinfo`
2022-01-15 15:53:25 <Ermine> Ah ok
2022-01-15 15:53:51 <Ariadne> :)
2022-01-15 15:54:04 <psykose> does it implicitly export to veinfo
2022-01-15 15:54:19 <Ariadne> no idea
2022-01-15 20:03:28 <Ariadne> moving on to riscv64 for now
2022-01-15 20:03:41 <Ariadne> i'm going to have to figure out how to selectively lobotomize part of the rust build system for s390x
2022-01-15 20:03:52 <Ariadne> it is *very* insistent that i want llvm-libunwind
2022-01-15 20:04:01 <Ariadne> which, seems reasonable normally
2022-01-15 20:04:19 <ikke> Do you need the container?
2022-01-15 20:04:20 <Ariadne> but... llvm-libunwind doesnt support s390x
2022-01-15 20:04:25 <Ariadne> ikke: not yet :)
2022-01-15 20:04:28 <Ariadne> but soon
2022-01-15 20:04:30 <ikke> ok
2022-01-15 20:04:34 <Ariadne> lets get the wireguard up
2022-01-15 20:08:05 <ikke> Fun, the nlplug-findfs issue we have with one of our CI hosts is still not solved
2022-01-15 20:18:34 <ikke> It works if I run nlplug-finds manually and continue  ¯\_(ツ)_/¯
2022-01-15 20:26:03 <__abby__> is the maintainer of the openjdk packages in this channel? I'm working on similar packages for void and had a couple questions about some issues I'm having
2022-01-15 20:26:25 <ikke> bratkartoffel
2022-01-15 21:05:36 <Ariadne> welp
2022-01-15 21:07:04 <Ariadne> ikke: im almost to having a workable bootstrap rust compiler on s390x, i think starting there makes most sense, since we don't have to fight qemu-user
2022-01-15 21:13:37 <ikke> ok
2022-01-15 21:41:10 <ikke> so rust at least at times only uses 2 threads?
2022-01-15 21:41:59 <psykose> in the build? yes
2022-01-15 21:42:30 <ikke> Like watching paint dry
2022-01-15 21:42:32 <psykose> building things with rust is mostly the same as well, thread per crate
2022-01-15 21:54:43 <Ariadne> bbl
2022-01-15 21:54:47 <Ariadne> will pick this up in the morning
2022-01-15 21:54:58 <Ariadne> need to make a new gcc snapshot to make forward progress on riscv64 cross
2022-01-15 21:55:09 <ikke> o/
2022-01-15 22:01:02 <mps> I've got terminus 12x24 bold font in kernel on console, so will not do this tomorrow as I told earlier. looks good on 1920x1080 display
2022-01-15 22:09:31 <psykose> ikke: https://gitlab.alpinelinux.org/tim-devel/aports/-/jobs/595374 what is happening here
2022-01-15 22:21:57 <a16bitsysop> getting a strange in octave !29612, changed to just pkgrel bump and it cant detect changed aport on 32bit platforms https://gitlab.alpinelinux.org/a16bitsysop/aports/-/jobs/595459#L72
2022-01-15 22:23:35 <psykose> same as above then
2022-01-15 22:23:37 <psykose> broken ci
2022-01-15 22:25:26 <a16bitsysop> okay will change back and wait for fix
2022-01-16 04:34:15 <__abby__> does alpine run check() on aports' ci?
2022-01-16 04:34:41 <psykose> yes
2022-01-16 04:34:57 <psykose> unless you have options="!check"
2022-01-16 04:35:19 <psykose> and the default check() if not defined in apkbuild just prints a warning of no tests
2022-01-16 04:38:12 <__abby__> interesting, because I'm looking at the openjdk{12..17} package templates and they don't have !check, but the CI on recent updates finished in minutes (less than it takes on my machine!)
2022-01-16 04:38:47 <__abby__> also the check make target doesn't work for me so i've been trying to figure out if it's working for y'all
2022-01-16 04:39:15 <psykose> it manually execs java stuff
2022-01-16 04:39:20 <psykose> and make test-hotspot-gtest
2022-01-16 04:39:32 <psykose> as for ci speed, dunno
2022-01-16 04:39:36 <psykose> the builders are a little fat
2022-01-16 04:39:54 <psykose> somewhere above 40 threads
2022-01-16 04:39:57 <__abby__> yeah test-hotspot-gtest is broken for me on 15..17
2022-01-16 04:40:01 <psykose> mm
2022-01-16 04:40:02 <__abby__> ok that makes more sense then
2022-01-16 04:40:15 <psykose> have you tried running the exact invocation
2022-01-16 04:40:19 <psykose> is this for alpine or some other stuff
2022-01-16 04:40:22 <__abby__> void
2022-01-16 04:40:31 <psykose> uhh idk then
2022-01-16 04:40:41 <psykose> it could be basically anything in the dependency chain that magically diverges
2022-01-16 04:40:46 <psykose> or the 6 patches
2022-01-16 04:40:49 <psykose> or anythin
2022-01-16 04:41:58 <__abby__> I don't think it's in a dep chain, it's like the makefile itself doesn't understand the test set to run but it hasn't changed since jdk14
2022-01-16 04:44:22 <psykose> idk, i can't really guess without seeing all the output and stuff
2022-01-16 04:44:37 <psykose> make stuff is usually not super bad to trace in the makefile with some searching
2022-01-16 07:20:43 <ikke> psykose: I'm not sure
2022-01-16 07:21:17 <ikke> psykose: I mean, it has to do with the recent changes, but I cannot reproduce that error
2022-01-16 07:21:23 <psykose> it happens on multiple mrs
2022-01-16 07:21:39 <psykose> here too https://gitlab.alpinelinux.org/a16bitsysop/aports/-/pipelines/106758
2022-01-16 07:21:54 <ikke> I checked the test suite uses bb sed as well
2022-01-16 07:22:13 <ikke> I'd need to reproduce those MRs
2022-01-16 07:22:20 <psykose> here as well https://gitlab.alpinelinux.org/otlabs/aports/-/pipelines/106777
2022-01-16 07:22:26 <psykose> but yeah i think that is plenty examples
2022-01-16 07:23:54 <psykose> strange how it's arch specific
2022-01-16 07:24:29 <ikke> Well, I pulled the new image on a single CI host
2022-01-16 07:25:05 <ikke> Only on x86
2022-01-16 07:27:03 <ikke> https://gitlab.alpinelinux.org/alpine/infra/docker/alpine-gitlab-ci/-/blob/master/overlay/usr/local/bin/build.sh#L135
2022-01-16 07:27:28 <ikke> But I cannot get sed to print that message
2022-01-16 07:28:18 <ikke> `sed 1` -> missing command
2022-01-16 07:31:01 <ikke> Hmm, I see now what you mean, seems to be only 32-bits arches
2022-01-16 07:31:23 <psykose> and s390x on the last one
2022-01-16 07:36:51 <ikke> I can at least reproduce when trying to build an actual mr
2022-01-16 07:37:51 <ikke> ooooh
2022-01-16 07:37:56 <ikke> -1
2022-01-16 07:54:23 <ikke> Ok, found the 2 issues that compounded it
2022-01-16 07:55:52 <psykose> :)
2022-01-16 07:57:52 <ddevault> is the sloccount package broken?
2022-01-16 07:58:05 <ikke> ddevault: you tell us :-)
2022-01-16 07:58:12 <ddevault> sloccount appears to be broken
2022-01-16 07:58:17 <ddevault> please install it and see if you can reproduce
2022-01-16 07:58:29 <psykose> it is
2022-01-16 07:59:39 <ddevault> no build logs available
2022-01-16 07:59:42 <ddevault> can someone queue it for a rebuild?
2022-01-16 08:00:28 <psykose> i can't even rebuild it locally
2022-01-16 08:00:39 <ddevault> it's not very good
2022-01-16 08:00:42 <ddevault> we could just get rid of it
2022-01-16 08:00:57 <psykose> sadly i can't look at it more now
2022-01-16 08:01:01 <psykose> but i can try fix it later
2022-01-16 08:01:15 <ddevault> I'll file an aports ticket for now
2022-01-16 08:38:04 <fraolt> can somebody please merge !29054 ?
2022-01-16 12:27:08 <nmeum> mps: regarding !29565 I think the hifive-unmatched-ramdisk.patch file can be removed with the upgrade to 2022.01 but I haven't yet tested it without the patch, see the patch description for details
2022-01-16 12:29:57 <nmeum> hm, maybe nott the patch series which fixes this problem seems to not be included in the 2022.01 release
2022-01-16 12:44:45 <mps> nmeum: ok, I have to add another patch received from one upstream developer and test, and also to report yet another bug, so 'we' shouldn't hurry to merge this MR
2022-01-16 12:45:09 <mps> nmeum: thanks for review
2022-01-16 12:45:54 <nmeum> np
2022-01-16 13:04:49 <donoban> Hi, I'm comparing lsmod between ubuntu and alpine for trying to discover why sound doesn't work properly on the second
2022-01-16 13:05:26 <donoban> there is a module on ubuntu, modprobe: FATAL: Module snd_sof_intel_byt not found in directory /lib/modules/5.15.12-0-lts
2022-01-16 13:05:44 <donoban> which alpine misses, but I don't see what option should enable it
2022-01-16 13:07:41 <ktprograms> https://www.kernelconfig.io/CONFIG_SND_SOC_SOF_INTEL_ATOM_HIFI_EP?q=sof_intel_byt&kernelversion=5.11.20&arch=x86
2022-01-16 13:08:09 <donoban> uhM
2022-01-16 13:08:16 <donoban> This option is not user-selectable but automagically handled by
2022-01-16 13:08:17 <donoban> 'select' statements at a higher level.
2022-01-16 13:08:21 <ktprograms> Selected by https://www.kernelconfig.io/CONFIG_SND_SOC_INTEL_SOF_PCM512X_MACH?kernelversion=5.11.20
2022-01-16 13:09:06 <ktprograms> It's `=m` in the lts config https://gitlab.alpinelinux.org/alpine/aports/-/blob/master/main/linux-lts/config-lts.x86_64#L6291
2022-01-16 13:09:40 <donoban> I see
2022-01-16 13:10:01 <ktprograms> So I don't know why it
2022-01-16 13:10:14 <ktprograms> it's not there
2022-01-16 13:10:50 <ktprograms> Is snd-sof-intel-byt.ko in /lib/modules ?
2022-01-16 13:11:25 <donoban> no
2022-01-16 13:11:56 <donoban> only, intel-hda-common, intel-atom, intel-ipc and inte-hda
2022-01-16 13:12:29 <donoban> same for linux-edge
2022-01-16 13:13:30 <donoban> I'm also running a hostpad here. It gets like 40Mbps with linux-lts and ~120Mpbs with linux-edge
2022-01-16 13:13:52 <donoban> but sound fails with both
2022-01-16 13:15:16 <ktprograms> Hmm no idea why the module isn't there
2022-01-16 13:18:17 <donoban> maybe this? https://www.kernelconfig.io/CONFIG_SND_INTEL_BYT_PREFER_SOF?q=CONFIG_SND_INTEL_BYT_PREFER_SOF&kernelversion=5.11.20&arch=x86
2022-01-16 13:25:42 <ikke> psykose: https://gitlab.alpinelinux.org/alpine/infra/docker/alpine-gitlab-ci/-/merge_requests/9
2022-01-16 13:43:48 <ikke> That fixes it
2022-01-16 13:51:34 <mps> donoban: create issue/bug report with some explanation and mention me there
2022-01-16 13:53:46 <donoban> ok mps, thanks
2022-01-16 13:54:47 <donoban> in the meantine I tried to force sof driver with options snd_intel_dspcfg dsp_driver=3
2022-01-16 13:55:11 <donoban> and ended without sound card hehe
2022-01-16 14:19:13 <pabloyoyoista> Could anybody have a look at !29456 and !29163? The second one fixes a segfault due to a missing dependency, and needs the first one to build with newest meson
2022-01-16 14:24:46 <donoban> if I am going to upload many text files too gitlab is better to tar/gzip them?
2022-01-16 14:25:30 <donoban> posting an issue
2022-01-16 14:31:12 <ikke> What is the goal?
2022-01-16 14:31:29 <ikke> for these files
2022-01-16 14:31:36 <donoban> there are, dmesg/lsmod/aplay output from ubuntu and alpine
2022-01-16 14:32:09 <ikke> How many is many?
2022-01-16 14:32:17 <donoban> 7 files
2022-01-16 14:32:57 <ikke> I would just attach them invidually then
2022-01-16 14:33:10 <donoban> ok great
2022-01-16 14:39:37 <mps> donoban: who will read big files
2022-01-16 14:39:46 <mps> I will not for sure
2022-01-16 14:40:04 <mps> explain issue in few words, sounds a lot better
2022-01-16 16:19:24 <Ariadne> dalias: hypothetical question for shits and giggles
2022-01-16 16:19:35 <Ariadne> what would musl dlopen do if you tried to
2022-01-16 16:19:41 <Ariadne> open a .so for another arch
2022-01-16 16:41:21 <Hello71> huh, i thought it would check e_machine
2022-01-16 16:47:47 <Ariadne> ldd seems to reject it
2022-01-16 16:47:49 <Ariadne> but i duno
2022-01-16 16:48:07 <Ariadne> i am wondering if rustc is somehow succeeding in dlopening plugins for the wrong arch
2022-01-16 18:35:00 <ikke> psykose: !27089, there is already an MR for ICU 
2022-01-16 18:37:52 <Ariadne> https://github.com/rust-lang/rust/issues/92976
2022-01-16 18:38:03 <Ariadne> we're stuck on this for now
2022-01-16 18:38:20 <ikke> sadface
2022-01-16 18:39:08 <xpufx> Hey guys. I attempted to fix a broken init script. Tried a MR using the gitlab gui. Phenomenal failure. Now I want to learn how to do it properly. Is there a github hub like tool that needs to be used? I already set up git and cloned aports.
2022-01-16 18:39:32 <ikke> There is a tool you can use (glab), but certainly not required
2022-01-16 18:39:45 <ikke> xpufx: What failed the first time?
2022-01-16 18:40:10 <xpufx> I didn't know there was a required format for the MR title and commit message. Then I didn't know how to revert it using the gitlab gui.
2022-01-16 18:40:29 <ikke> The gitlab interface can only create new commits, not adjust existing ones
2022-01-16 18:40:31 <xpufx> So i deleted the MR to come back to it when I had some time.
2022-01-16 18:40:47 <xpufx> Also apparently I have to update some things in the package build.
2022-01-16 18:40:55 <ikke> You can fetch the branch using git
2022-01-16 18:41:07 <ikke> and then update that branch, and then force push to that branch
2022-01-16 18:41:19 <xpufx> the branch with my MR right?
2022-01-16 18:41:22 <ikke> yes
2022-01-16 18:41:44 <xpufx> Anything against starting over? And letting that MR stay closed?
2022-01-16 18:42:05 <ikke> Loosing context of the MR (ie, any comments that were left behind)
2022-01-16 18:42:29 <xpufx> Right. It's already closed though. Will it reopen when I force push?
2022-01-16 18:42:49 <ikke> No, but you can reopen it yourself
2022-01-16 18:43:42 <xpufx> OK. I have created a new lxc alpine with the latest available iso (my prox host needs the template downloaded and doesn't have one for edge) then updated the repositories to use edge. Is this enough of an environment for testing edge?
2022-01-16 18:43:56 <xpufx>  /etc/alpine-release doesn't say edge
2022-01-16 18:44:10 <ikke> yes, should work
2022-01-16 18:44:31 <ikke> Just keep your installed packages clean, or use rootbld (apk add abuild-rootbld)
2022-01-16 18:44:49 <xpufx> Yeah I don't need this host for anything else.
2022-01-16 18:45:19 <xpufx> and mounted /home/myuser separately in case i need to quickly wipe it
2022-01-16 18:46:08 <xpufx> I think before doing anything. I will test if that init script is really/still broken in edge.
2022-01-16 18:46:43 <minimal> xpufx: which init script?
2022-01-16 18:46:48 <xpufx> dcc
2022-01-16 18:46:56 <ikke> Someone is already working on dcc
2022-01-16 18:47:19 <ikke> !29459
2022-01-16 18:48:07 <xpufx> Duncan Bellamy?
2022-01-16 18:48:11 <ikke> yes
2022-01-16 18:48:28 <xpufx> He told me to send an MR. But he might have worked on it afterwards. I will check.
2022-01-16 18:48:54 <xpufx> 4 days ago is pretty recent.
2022-01-16 18:49:16 <xpufx> openldap had some issues when I tried it a few days ago too.
2022-01-16 18:49:35 <xpufx> it's looking for module.la files but installing module.so files
2022-01-16 18:51:11 <Ariadne> ugh
2022-01-16 18:51:25 <Ariadne> libltdl is a crime against humanity
2022-01-16 18:51:40 <ikke> What is libltdl?
2022-01-16 18:52:06 <orbea> a crime against humanity
2022-01-16 18:52:40 <Ariadne> ikke: dlopen, but with extra libtool bullshit layered on top
2022-01-16 18:52:45 <orbea> https://www.gnu.org/software/libtool/manual/html_node/Libltdl-interface.html
2022-01-16 18:52:55 <Ariadne> it lets you dlopen .la files
2022-01-16 18:52:57 <Ariadne> basically
2022-01-16 18:53:07 <xpufx> ok so Duncan is working on it already but it's not merged yet. I'm sure he'll fix it better than I can.
2022-01-16 18:54:20 <xpufx> it's not installing .la files but the config is referring to .la files. I don't know why.
2022-01-16 18:57:26 <xpufx>  /etc/init.d/slapd start  -> lt_dlopenext failed: (back_mdb.la) file not found
2022-01-16 18:59:09 <xpufx> that module does not exist when doing 'apk add openldap' anyway
2022-01-16 18:59:14 <xpufx> .la or .so
2022-01-16 19:01:14 <xpufx> I should probably just post a bug report right? In case there's a maintainer who'd be better equipped to handle it. I mean in general for things I find.
2022-01-16 19:02:37 <ikke> there are many subpackages for openldap
2022-01-16 19:02:42 <ikke> https://pkgs.alpinelinux.org/contents?file=*.so&path=&name=openldap*&branch=edge&arch=x86_64
2022-01-16 19:05:15 <xpufx> sorry. hit the wrong button there.
2022-01-16 19:06:23 <xpufx> there are indeed a lot of packages. I suppose a back database module is necessary too. But the default config (/etc/openldap/slapd.conf) has this line: moduleload      back_mdb.la
2022-01-16 19:06:23 <xpufx>  
2022-01-16 19:07:07 <xpufx> when that modules is not installed by default, and when you install openldap-back-mdb   it's installed as usr/lib/openldap/back_mdb.so   not .la
2022-01-16 19:08:36 <xpufx> pretty easy fix when you need to use it. I suppose this is why nobody posted a bug report.
2022-01-16 19:09:22 <ikke>  
2022-01-16 19:10:15 <minimal> xpufx: yes mdb is the standard/default/recommended backend database for openldap
2022-01-16 19:10:37 <xpufx> Perhaps it should be installed as a dependency then?
2022-01-16 19:11:45 <minimal> xpufx: apk has a "depends" but not "recommends" or "suggest" fields - so some people may want to use another backend without the mdb backend being installed, I guess that's why the maintainer did not make it a hrd dependancy
2022-01-16 19:11:55 <minimal> s/hrd/hard/
2022-01-16 19:11:55 <alpine-meetbot> minimal meant to say: xpufx: apk has a "depends" but not "recommends" or "suggest" fields - so some people may want to use another backend without the mdb backend being installed, I guess that's why the maintainer did not make it a hard dependancy
2022-01-16 19:13:48 <minimal> I guess the assumption is that when you install openldap you know you need at least one backend and you either also install openldap-backend-all or the specific backend that you intend to use
2022-01-16 19:14:18 <Ariadne> apk3 has recommends
2022-01-16 19:14:34 <minimal> Ariadne: we're all living in the future now? ;-)
2022-01-16 19:15:13 <mps> Ariadne: bad news
2022-01-16 19:15:19 <xpufx> OK. I don't know the alpine way of handling this. Is it normal for a package to be installed but not be able to started?
2022-01-16 19:16:05 <minimal> xpufx: I'd expect the package's README to have a general "to get this working" section
2022-01-16 19:16:34 <Ariadne> mps: ? :)
2022-01-16 19:16:37 <minimal> then again when I wrote such a README for one of the packages I maintain I found that people generally did not read it
2022-01-16 19:16:48 <Ariadne> mps: i did not say alpine has recommends :)
2022-01-16 19:16:54 <Ariadne> just that apk3 has them :P
2022-01-16 19:17:17 <mps> Ariadne: then not so bad news ;)
2022-01-16 19:17:22 <minimal> Ariadne: time to add a mps-recommends field also? :-)
2022-01-16 19:17:24 <xpufx> where would that README be?
2022-01-16 19:17:54 <Ariadne> xpufx: /usr/share/doc/$pkgname/README.Alpine
2022-01-16 19:17:58 <mps> minimal: it should be called mps-prohibited :D
2022-01-16 19:18:20 <xpufx> I don't have /usr/share/doc
2022-01-16 19:18:38 <minimal> xpufx: for openldap there does not appear to be an Alpine specific Readme
2022-01-16 19:19:01 <xpufx> it has a schema.README but that seems to be from openldap itself.
2022-01-16 19:19:38 <minimal> there is also no "setup-openldap" either
2022-01-16 19:20:09 <minimal> so I guess it is left to the user to figure out how things work. This is LDAP we're talking about after all :-)
2022-01-16 19:20:50 <zombiei> Hey good afternoon all
2022-01-16 19:20:58 <xpufx> They'd get frustrated and start looking at FreeIPA or something. Not a path that leads to less pain. :)
2022-01-16 19:21:25 <zombiei> Is anyone able to trigger an update on the squid cache package to use latest upstream 5.3 ?
2022-01-16 19:21:52 <zombiei> edge and main repo are still on 5.2-r0 for some months now
2022-01-16 19:24:36 <mps> zombiei: is 5.3 stable release?
2022-01-16 19:26:14 <mps> looks like it is
2022-01-16 19:28:27 <zombiei> sorry internet dropped
2022-01-16 19:28:42 <zombiei> @mps: yes it's a stable production release: http://www.squid-cache.org/Versions/
2022-01-16 19:29:39 <mps> zombiei: then it make sense to be upgraded on edge
2022-01-16 19:30:28 <zombiei> @mps is there a way for me to trigger/request that ?
2022-01-16 19:31:24 <mps> zombiei: you can create issue/wishlist report on gitlab.a.o
2022-01-16 19:31:44 <mps> in aports subproject
2022-01-16 19:34:21 <minimal> xpufx: I've used Openldap on Alpine to set up multi-master replication in the past, a bit of learning involved but worked fine once figured out
2022-01-16 19:47:40 <ikke> Someone asking for squid to be upgraded to 5.3
2022-01-16 19:47:46 <MY-R> is it normal when building kernel it "stuck" on "awk -f /home/builder/aports/main/linux-lts/src/linux-5.15/arch/x86/tools/objdump_reformat.awk" with 100% one thread usage for around 15-20 minutes?
2022-01-16 19:48:15 <mps> ikke: yes
2022-01-16 19:49:00 <MY-R> building Alpine kernels is basically 3 times longer than on any other distro for me :\
2022-01-16 19:55:40 <xpufx> !13441  it's not my problem anymore :)
2022-01-16 19:55:50 <xpufx> sorry. i meant to link to an issue.
2022-01-16 19:55:55 <ikke> use #
2022-01-16 19:55:55 <xpufx> https://gitlab.alpinelinux.org/alpine/aports/-/issues/13441
2022-01-16 19:56:13 <xpufx> ikke thanks
2022-01-16 20:04:42 <Hello71> MY-R: you can increase awk speed ~100x (estimate) by installing gawk
2022-01-16 20:05:14 <Hello71> busybox awk has a bad interaction with mallocng, i haven't had time to dig into it yet
2022-01-16 20:09:00 <MY-R> Hello71: yeah that what I was thinking that maybe BB awk "suck", thanks, will try it next time
2022-01-16 20:09:45 <MY-R> but at some point Im starting replace every BB tool...
2022-01-16 20:20:07 <mps> zombiei: !29639
2022-01-16 20:21:12 <PJ[m]> is there any way to create manpages out of markdown files? I was suggested pandoc but it seems it's x64 only currently
2022-01-16 20:21:27 <ikke> scdoc is close
2022-01-16 20:31:18 <PJ[m]> thanks
2022-01-16 20:52:59 <xpufx> cheers guys. i'll be back if i find another package i can tackle. Thanks for the help.
2022-01-16 20:53:35 <Habbie> PJ[m], we (powerdns) make manpages from .rst with sphinx; i don't know if it also does markdown
2022-01-16 20:55:34 <mps> what are minimum number of packages to install to run docker? 'apk add docker' only?
2022-01-16 20:55:58 <ikke> yes
2022-01-16 20:57:03 <ikke> install the package, run the service, and it should be ready
2022-01-16 20:57:34 <mps> ikke: ok, I need to run cgroup daemon also
2022-01-16 21:01:28 <mps> I've got dockerfile from u-boot developer to test something, lets hope I will manage to do this
2022-01-16 21:01:43 <mps> 'docker build -t .' gives me 'invalid argument "." for "-t, --tag" flag: invalid reference format'
2022-01-16 21:02:34 <boomanaiden154> You need to specify the name of the image after -t
2022-01-16 21:02:51 <boomanaiden154> You should be able to just omit it altogether if you just want the autoassigned name.
2022-01-16 21:04:41 <xpufx> hello again guys :)
2022-01-16 21:04:45 <mps> looks like 'docker build .' works
2022-01-16 21:05:20 <xpufx> # /etc/init.d/slapd start
2022-01-16 21:05:26 <xpufx> unable to open file "/var/lib/openldap/run/slapd.pid": 2 (No such file or directory)
2022-01-16 21:05:38 <boomanaiden154> mps: Yes. It won’t name the image though so if you want to run it later you need to find the reference that is outputted by docker build.
2022-01-16 21:05:40 <xpufx>  /var/lib/openldap/run does not exist.
2022-01-16 21:06:26 <mps> boomanaiden154: thanks
2022-01-16 21:06:43 <xpufx> I am not sure if this is a bug since the config provided does not work anyway
2022-01-16 21:06:48 <ikke> xpufx: pid files should be placed under /run, not /var/lib/run
2022-01-16 21:06:55 <xpufx> That's in slapd.conf though. 
2022-01-16 21:07:20 <xpufx> I think it's best if the maintainer just called the config slapd.conf.sample
2022-01-16 22:17:47 <MY-R> when Im doing "apk add or upgrade" do I need to use "--update" arg if already have "--no-cache" ?
2022-01-16 22:18:22 <MY-R> --no-cache will always trigger update right? :P
2022-01-16 22:38:39 <PJ[m]> you don't need --update
2022-01-16 22:39:53 <MY-R> ok!
2022-01-17 00:13:34 <MY-R> minimal: 20 minutes... after "apk add gawk" building linux-lts took 20 minutes less, something is terribly wrong with BB awk
2022-01-17 00:14:07 <Hello71> well i'm pretty sure it is related to mallocng too
2022-01-17 00:14:30 <MY-R> or more like linux kernel tools expect something more from awk
2022-01-17 00:14:36 <Hello71> with glibc maybe it takes say 1 minute longer with busybox awk compared to gawk
2022-01-17 00:15:23 <MY-R> hmm
2022-01-17 00:16:30 <Hello71> i think problem is busybox awk is calling malloc and free very often for some linux kernel scripts, and mallocng is repeatedly requesting and returning memory from/to kernel
2022-01-17 00:16:32 <MY-R> for now Im installing gawk on every alpine host/docker/vm
2022-01-17 00:16:45 <Hello71> whereas glibc just keeps all ram allocated
2022-01-17 00:17:04 <Hello71> not sure though, guessing based on strace
2022-01-17 00:18:05 <MY-R> Hello71: dunno, I only noticed BB awk stuck for 20 minutes with some kernel script abusing one core 100% :\
2022-01-17 00:19:23 <MY-R> with gawk same process done just in 1 second
2022-01-17 00:21:17 <Hello71> yes, i had same experience
2022-01-17 00:22:05 <MY-R> :\
2022-01-17 00:22:54 <Hello71> filed #13442
2022-01-17 00:24:18 <MY-R> Hello71: oh, nice
2022-01-17 00:25:02 <MY-R> Im wondering how many packages are affected by this
2022-01-17 00:25:25 <Hello71> probably nobody really notices because afaik it only affects fat distro kernel, if only compile a few small modules it doesn't use too much time
2022-01-17 00:25:55 <Hello71> pretty sure only linux, gcc, and maybe binutils use awk heavily
2022-01-17 00:26:34 <MY-R> and what about Alpine builders? Looking at compile kernel times it seems that gawk isnt installed
2022-01-17 00:37:45 <MY-R> https://build.alpinelinux.org/buildlogs/build-edge-x86_64/main/linux-lts/linux-lts-5.15.14-r0.log
2022-01-17 00:37:53 <MY-R> it stuck when doing this: arch/x86/tools/insn_decoder_test: success: Decoded and checked 5912135 instructions
2022-01-17 05:43:26 <ikke> MY-R: correct, we only have minimal dependencies installed
2022-01-17 05:50:22 <psykose> ikke: ah nice, didn't see
2022-01-17 05:50:42 <ikke> No one seems to be continuing with it apparently
2022-01-17 08:14:10 <ikke> Morning 
2022-01-17 09:31:05 <ncopa> morning
2022-01-17 09:34:21 <mps> goede morgen
2022-01-17 09:36:04 <Misthios> gm
2022-01-17 09:37:19 <mps> today I especially salute good Dutch people
2022-01-17 09:44:19 <Misthios> lol
2022-01-17 09:55:20 <mps> which if these lines 7 or 8 are maintainer comment here https://git.alpinelinux.org/aports/tree/community/telegram-desktop/APKBUILD#n7
2022-01-17 09:55:27 <mps> s/if/of/
2022-01-17 09:55:28 <alpine-meetbot> mps meant to say: which of these lines 7 or 8 are maintainer comment here https://git.alpinelinux.org/aports/tree/community/telegram-desktop/APKBUILD#n7
2022-01-17 09:56:01 <mps> I think I will be attacked again if I fix this ;)
2022-01-17 10:05:14 <Misthios> 2 version?
2022-01-17 10:07:10 <mps> or maintainer comment
2022-01-17 10:07:52 <mps> jk, ofc
2022-01-17 10:14:13 <psykose> that just looks like an oversight with two pkgvers
2022-01-17 10:14:37 <psykose> not sure why you are acting like a child again, but there is no issue with deleting an obviously duplicated pkgver
2022-01-17 10:14:48 <ikke> Probably badly resolved conflict 
2022-01-17 10:15:11 <psykose> yeah
2022-01-17 10:18:21 <mps> psykose: please be careful with your words
2022-01-17 10:20:08 <psykose> i am
2022-01-17 10:20:51 <mps> don't talk with me like this!
2022-01-17 11:28:36 <ikke> fabled: did something change in apk that it now mentions 'masked in: cache' when it cannot install a certain package?
2022-01-17 11:28:45 <ikke> for example: https://build.alpinelinux.org/buildlogs/build-edge-ppc64le/testing/way-displays/way-displays-1.2.0-r0.log
2022-01-17 11:28:52 <ikke> Don't recall seeing that before
2022-01-17 11:33:40 <fabled> ikke: that error has existed a long time, but not shown in this case before. Likely side effect of https://gitlab.alpinelinux.org/alpine/apk-tools/-/commit/74f67ab81c646b9c1fb1136776cf05a949de2526
2022-01-17 11:34:30 <ikke> aha, ok
2022-01-17 11:36:01 <ikke> What does it mean that it's masked in cache?
2022-01-17 11:38:26 <ikke> /etc/apk/cache does not exist on the ppc64le builder
2022-01-17 11:42:02 <fabled> Basically it's package not present in any index 
2022-01-17 11:44:39 <ikke> ok
2022-01-17 16:03:00 <markand> bonjour
2022-01-17 16:03:23 <mps> bona sera
2022-01-17 16:07:00 <Hello71> so apparently the reason why gawk is much faster than busybox awk when compiling linux is that in some circumstances, busybox awk recompiles regexes on every search
2022-01-17 16:07:59 <Hello71> and with new musl, every regcomp calls mmap and munmap like four times
2022-01-17 16:08:04 <mps> perl is already deps for kernel so I wonder why they use awk at all
2022-01-17 16:08:15 <Hello71> perl is only recent dep
2022-01-17 16:08:33 <mps> yes, recent for few years
2022-01-17 16:08:42 <Hello71> right but these awk scripts are very old
2022-01-17 16:09:05 <mps> I know, just wonder why not rewritten in perl
2022-01-17 16:09:19 <ikke> Rewriting introduces new bugs
2022-01-17 16:09:31 <mps> and new fixes
2022-01-17 16:10:10 <mps> rewriting awk to perl shouldn't be complicated
2022-01-17 16:10:48 <Hello71> mps: i await your patches
2022-01-17 16:11:16 <mps> Hello71: I fear you will wait toooooo long
2022-01-17 16:11:32 <ikke> And that's why it has not been rewritten yet :P
2022-01-17 16:11:50 <mps> going through kernel ML is in easy job
2022-01-17 16:12:04 <mps> s/in/no/
2022-01-17 16:12:04 <alpine-meetbot> mps meant to say: gonog through kernel ML is in easy job
2022-01-17 16:12:21 <mps> s/in /no/
2022-01-17 16:12:21 <alpine-meetbot> mps meant to say: gonog through kernel ML is noeasy job
2022-01-17 16:12:30 <mps> meh :)
2022-01-17 16:17:25 <mps> ncopa: !29639
2022-01-17 16:18:18 <mps> I think this is latest stable squid, but announce is not clear about this
2022-01-17 16:18:37 <ikke> mps: I checked it yesterday, according to the website, it is
2022-01-17 16:18:47 <ikke> http://www.squid-cache.org/Versions/
2022-01-17 16:19:10 <ikke> 5.3 is listed as the latest stable version
2022-01-17 16:19:21 <mps> ikke: yes, I looked also there but 'language' is somewhat uncomprehensible to me
2022-01-17 16:19:37 <ikke> What part?>
2022-01-17 16:19:53 <Ariadne> before we push new kernel
2022-01-17 16:20:02 <Ariadne> can i update the gcc snapshot to fix bootstrap :D
2022-01-17 16:20:02 <mps> there is no simple '5.3 is latest stable release'
2022-01-17 16:20:17 <mps> Ariadne: sure
2022-01-17 16:20:25 <ikke> 5.3 is under 'stable versions'
2022-01-17 16:20:32 <Ariadne> mps: for squid, 5.3 is the latest stable
2022-01-17 16:20:54 <mps> I think I wil start linux-edge 5.16.1 tomorrow to work on it
2022-01-17 16:21:10 <mps> ok, then I will merge squid now
2022-01-17 16:21:59 <mps> ikke: btw, could we somehow tell gitlab to not retry CI when we rebase pkg
2022-01-17 16:23:16 <ikke> mps: I wish
2022-01-17 16:23:35 <mps> heh
2022-01-17 16:24:05 <mps> about !29661
2022-01-17 16:24:53 <mps> WEP is now disabled by default, should we enable it explicitly in config
2022-01-17 16:25:35 <mps> '* removed WEP support from the default build (CONFIG_WEP=y can be used to enable it, if really needed)'
2022-01-17 16:25:48 <mps> https://w1.fi/cgit/hostap/plain/wpa_supplicant/ChangeLog
2022-01-17 16:30:51 <ikke> I really hope no one is still using WEP
2022-01-17 16:31:08 <jvoisin> I think it's ok in 2022 to remove WEP support.
2022-01-17 16:32:27 <jvoisin> speaking of security, can someone merge https://gitlab.alpinelinux.org/alpine/aports/-/merge_requests/29182 ? The hardened_malloc/GrapheneOS people are impatient :D
2022-01-17 16:32:55 <jvoisin> s/impatient/excited/
2022-01-17 16:32:55 <alpine-meetbot> jvoisin meant to say: speaking of security, can someone merge https://gitlab.alpinelinux.org/alpine/aports/-/merge_requests/29182 ? The hardened_malloc/GrapheneOS people are excited :D
2022-01-17 16:33:35 <psykose> i am confused why something seemingly written for security has so many segfaulting tests
2022-01-17 16:33:37 <mps> ikke: jvoisin: also I think same but who knows what users expect
2022-01-17 16:33:41 <psykose> this doesn't strike confidence
2022-01-17 16:33:52 <jvoisin> psykose: it's the whole point
2022-01-17 16:33:57 <psykose> segfaults are the point?
2022-01-17 16:34:01 <jvoisin> yes
2022-01-17 16:34:04 <psykose> i don't follow
2022-01-17 16:34:10 <jvoisin> segfault and crash in case of memory corruptions
2022-01-17 16:34:23 <psykose> that sounds very wrong
2022-01-17 16:34:34 <jvoisin> it's the expected behaviour: instacrash in case of corruption
2022-01-17 16:34:40 <jvoisin> instead of silently continuing the execution
2022-01-17 16:39:12 <Ariadne> jvoisin: even still, i am not in favor of merging this until check() is fixed
2022-01-17 16:39:20 <jvoisin> ack
2022-01-17 16:39:46 <Ariadne> if the tests are supposed to segfault, there should be a runner to catch those segfaults and turn them into traditional pass/fail results
2022-01-17 16:40:16 <psykose> yeah this seems extremely orthogonal from the intentional case
2022-01-17 16:41:22 <Hello71> seems like switching linux-lts from busybox awk to mawk speeds up build by 12 minutes (about 33% less time) on x86_64 CI
2022-01-17 16:42:28 <Hello71> it would be less gain on a percentage basis on a less-parallel machine but 12 minutes is still a huge savings
2022-01-17 16:42:44 <Hello71>  s/savings/saving/
2022-01-17 16:47:09 <MY-R> with my haswell with 3.4 GHz turbo BB awk can stuck for 20 minutes, now wondering how long it was with rpi4 :P
2022-01-17 16:50:12 <MY-R> Hello71: if I will add mawk/gawk to "makedepends" in APKBUILD then it will use it or no?
2022-01-17 16:50:23 <Hello71> gawk yes, mawk no
2022-01-17 16:50:28 <Hello71> only gawk installs /usr/bin/awk
2022-01-17 16:50:32 <MY-R> ah ye
2022-01-17 16:50:35 <MY-R> ok, thanks!
2022-01-17 16:50:53 <Ariadne> i wonder if i can turn the ISC reference DNS resolver code into something that can be LD_PRELOAD into an app
2022-01-17 16:50:57 <Hello71> see !28218
2022-01-17 16:51:06 <Hello71> need to add mawk and set AWK=mawk
2022-01-17 16:51:13 <Ariadne> for use cases requiring bug compatibility with the ISC resolver code
2022-01-17 16:52:57 <MY-R> thank you Hello71!
2022-01-17 16:53:21 <Hello71> note that kbuild needs make AWK=mawk, not AWK=mawk make
2022-01-17 16:53:44 <Hello71> i still don't know why it is written this way
2022-01-17 16:54:52 <MY-R> I will probably stay with gnu awk :P
2022-01-17 16:55:46 <Hello71> mawk is still a lot faster than gawk; say 10 seconds with mawk, 30 seconds with gawk, 20 minutes with busybox awk
2022-01-17 16:56:04 <Hello71> i already closed my terminal so don't have exact numbers but roughly along those lines
2022-01-17 16:58:10 <MY-R> still, BB awk shouldnt be used or at least fixed, 10-20 minutes delay sounds wrong, not to mention other less powerful systems
2022-01-17 16:58:59 <Ariadne> if the issue is musl malloc-ng, that is what should be fixed
2022-01-17 16:59:23 <Ariadne> one possibility might be to LD_PRELOAD jemalloc and see if that improves performance
2022-01-17 17:00:20 <psykose> or mimalloc2
2022-01-17 17:00:32 <Ariadne> point is
2022-01-17 17:00:39 <Ariadne> we can verify it easily
2022-01-17 17:01:12 <Ariadne> (for that matter, i would like to see an /etc/ld.preload.conf)
2022-01-17 17:02:28 <Ariadne> this would be nice because then people could make decisions on what malloc, DNS implementation, etc they would like to use
2022-01-17 17:02:39 <Ariadne> i am sure dalias will hate the concept
2022-01-17 17:03:42 <Hello71> well it's roughly what i expected: busybox awk calls regcomp excessively, which with mallocng results in excessive mmap and munmap. the proper solution is to fix busybox awk to not call regcomp excessively, but even then it will still be slower than gawk and much slower than mawk
2022-01-17 17:05:05 <Hello71> possibly the only adjustment to mallocng would be to cache memory slightly more aggressively, but i think musl regcomp will still take substantial time even if malloc is free
2022-01-17 17:06:43 <Ariadne> musl regcomp is just TRE
2022-01-17 17:06:48 <Ariadne> i can fix that aspect of busybox tho
2022-01-17 17:06:49 <Hello71> mmhmm
2022-01-17 17:07:22 <psykose> i would like to preload mimalloc indeed
2022-01-17 17:07:23 <psykose> sometimes
2022-01-17 17:07:25 <Hello71> the best solution is to use mawk for intensive awk scripts, such as compiling linux. i highly doubt that busybox will accept patches to make awk faster at significant size expense
2022-01-17 22:18:28 <adhawkins> This is moot now as I've just noticed my MR has been merged. However, the owner of a python package I maintain has just released two versions within hours of each other. Is there any benefit to creating and merging both versions?
2022-01-17 22:21:27 <psykose> as an upgrade? sure
2022-01-17 22:37:20 <Hello71> i assume they mean sequential versions of the same package
2022-01-17 22:37:27 <Hello71> in which case, no
2022-01-17 22:43:23 <adhawkins> Hello71: Yes. It was on 1.11.1, and today he release 1.12.0 and 1.13.0. Wondering if I should have packaged both?
2022-01-17 22:43:38 <Hello71> well what does "packaged both" mean exactly
2022-01-17 22:44:28 <adhawkins> Generate a MR request for 1.12.0, wait for it to be merged, and then do another for 1.13.0
2022-01-17 22:48:20 <psykose> no, just do latest
2022-01-17 22:59:46 <adhawkins> YEah, that's what I did in the end.
2022-01-17 23:06:11 <adhawkins> Thanks for the advice psykose, Hello71.
2022-01-18 00:13:21 <ktprograms> MY-R: Interestingly, arm64 kernel builds don't seem to be stuck with awk for a long time, although it's possible that my config excludes building parts that require awk.
2022-01-18 00:20:17 <Hello71> i think mainly x86 heavily uses awk
2022-01-18 00:21:14 <Hello71> the specific script at issue is x86-only
2022-01-18 00:21:47 <MY-R> ktprograms: ye, it is x86 tool, in aarch64 dont see it in use
2022-01-18 00:26:55 <ktprograms> I see
2022-01-18 09:41:59 <maribu> PureTryOut: Did you also run into this: I cannot set the default sink via pulse audio clients (e.g. pactl and pulsemixer) with pipewire-pulse any more. The command seems to succeed (pactl exits without an error and exit code 0), but the default sink remains unchanged.
2022-01-18 09:42:48 <PureTryOut> default sink == default output device, right? If so then no I did not run into that
2022-01-18 09:43:49 <maribu> OK. I'll try to check if there is something wrong with my setup. (I still can move streams to a different output and apparently apps remember their last sink. But it is a bit annoying, still.)
2022-01-18 09:46:38 <PureTryOut> if I change my default output device all streams move to it as expected
2022-01-18 11:51:25 <mps> PureTryOut: 'Nintendo Joy-Con and Pro Controller support (HID_NINTENDO) [N/m/?] (NEW)' for kernel 5.16.x ?
2022-01-18 11:51:46 <mps> iirc you mentioned something like this
2022-01-18 11:51:51 <PureTryOut> I don't have the controller myself (yet, hopefully) but sure.
2022-01-18 11:51:59 <mps> ok
2022-01-18 11:52:35 <PureTryOut> Strangely enough another developer of a program of mine had a Joy-Con controller work just fine before 5.16, so not sure what's new about this and why it's required
2022-01-18 11:54:20 <mps> have no idea even what is it
2022-01-18 12:17:50 <mps> donoban: I can't find any reference in kernel about #13439
2022-01-18 12:18:11 <mps> I mean, in configs
2022-01-18 12:20:40 <donoban> thanks mps, I couldn't find any option which is supposed to build it as module
2022-01-18 12:20:42 <psykose> snd-sof-intel-byt.ko is CONFIG_SND_SOC_SOF_INTEL_ATOM_HIFI_EP  and enabled
2022-01-18 12:20:55 <psykose> everything with snd_soc_sof is enabled in both except for debug/acpi
2022-01-18 12:22:35 <mps> psykose: I looked to sources but didn't concluded this, thanks
2022-01-18 12:26:19 <ktprograms> mps: FYI I find kernelconfig.io quite useful for this stuff because it shows `is selected by` and `selects`, and the equilvalent for depends
2022-01-18 12:27:53 <psykose> cateee is the less out of date version
2022-01-18 12:28:38 <mps> ktprograms: I looked there but found it is not useful for me
2022-01-18 12:46:16 <adamplumb> I've been working on upgrading to libadwaita 1.0 which requires gtk 4.5+ and a new pango.  Unfortunately some change in pango fails tests on alpine (due to libc differences it looks like).  Ticket here: https://gitlab.gnome.org/GNOME/pango/-/issues/659#note_1358882
2022-01-18 12:47:03 <adamplumb> He suggests filing a ticket with glib, so I can go in that direction, but just wanted to document where I'm at with this and see if there's any advice
2022-01-18 12:53:51 <adamplumb> Looks like someone beat me to it: https://gitlab.gnome.org/GNOME/glib/-/issues/2552
2022-01-18 12:56:45 <psykose> there were also some tests that fail due to a missing font i think, there is an open mr for pango
2022-01-18 13:02:20 <psykose> and i think for new gtk you need some other new things too
2022-01-18 13:15:16 <psykose> though i wonder why the gtk site still says 4.4.1 is latest stable when 4.6.0 is released, given the normal version naming
2022-01-18 13:16:16 <adamplumb> yeah i noticed that too.  In my aports branch I was looking at upgrading to 4.5.1, but have been stuck on the pango issue
2022-01-18 13:17:58 <donoban> I'm experimenting an strange problem with bubblejail, some % of the time it starts bwrap before the dbus proxy socket is available altough it should already exist
2022-01-18 13:19:05 <donoban> bubblejail authors can't reproduce it altough he uses pyton 3.10 and tmpfs for /run too , https://github.com/igo95862/bubblejail/issues/13
2022-01-18 13:19:26 <donoban> aparrentely adding just a sleep(1) or a os.sync() "fixes" it
2022-01-18 13:21:07 <donoban> anyone has some experience with tmpfs and sync problems?
2022-01-18 13:22:49 <psykose> adamplumb: i don't recall if uneven releases were 'unstable' in gtk land, like in some other nearby projects (pango has this for instance)
2022-01-18 13:22:53 <psykose> i would go to 4.6 probably
2022-01-18 13:23:19 <psykose> also it seems gtk4 has an optionally enablable vulkan thing and colord for cups
2022-01-18 13:28:41 <adamplumb> yeah i think i just followed the homepage since it still gives either 4.4.1 or 4.5.1 as the options.  but i'm happy to give 4.6 a try
2022-01-18 13:30:04 <psykose> if you want to do testing, just install new pango with !check to build new gtk and do local stuff
2022-01-18 14:19:04 <Newbyte> Could someone explain why wlroots-phosh (yes, unfortunately) doesn't get built before phoc: https://gitlab.alpinelinux.org/Newbyte/aports/-/jobs/598201
2022-01-18 14:22:31 <psykose> it does
2022-01-18 14:22:48 <psykose> the wlroots-phosh build fails because of wrong $builddir
2022-01-18 14:22:55 <psykose> then it goes to phoc and that fails cause the former failed
2022-01-18 14:23:24 <psykose> https://img.ayaya.dev/4dnw2p2I5sVC.png left-to-right order is correct
2022-01-18 14:23:33 <psykose> https://img.ayaya.dev/ZaS9m7YguS0p.png
2022-01-18 14:24:15 <psykose> you probably need builddir="$srcdir/wlroots-$pkgver"
2022-01-18 14:24:31 <psykose> since pkgname is no longer accurate to the actual folder
2022-01-18 14:31:38 <Newbyte> oh I see, thanks!
2022-01-18 17:16:10 <mps> donoban: I upgraded linux-edge to 5.16.1, please check your sound card with it so we could investigate further what is needed if it doesn't work
2022-01-18 17:17:13 <donoban> thanks mps, I will check in a few minutes
2022-01-18 17:19:46 <mps> I found that we need sound baytrail, will look at it
2022-01-18 17:21:31 <mps> there are a lot of option related to this
2022-01-18 17:25:20 <donoban> ok, let me know if I can help testing something
2022-01-18 17:25:38 <donoban> it doesn't seem on mirrors yet
2022-01-18 17:25:57 <psykose> it is on dl-cdn
2022-01-18 17:26:14 <donoban> ok ty
2022-01-18 17:27:47 <psykose> boots with no issues for me, but that is expected :)
2022-01-18 17:28:16 <mps> SND_SOC_SOF_ACPI
2022-01-18 17:28:48 <mps> donoban: that is what i this is missing
2022-01-18 17:31:03 <donoban> maybe is not really needed
2022-01-18 17:32:54 <donoban> setup-apkrepos doesn't work proplery
2022-01-18 17:33:01 <donoban> s/proplery/properly
2022-01-18 17:33:01 <alpine-meetbot> donoban meant to say: setup-apkrepos doesn't work properly
2022-01-18 17:33:48 <donoban> (1/1) Upgrading linux-edge@edge (5.15.13-r0 -> 5.16.1-r0)
2022-01-18 17:35:16 <mps> SND_INTEL_BYT_PREFER_SOF -> The kernel has 2 drivers for the Low Power Engine audio-block on Bay- and Cherry-Trail SoCs. The old SST driver and the new SOF driver. If both drivers are enabled then the kernel will default to using the old SST driver, unless told otherwise through the snd_intel_dspcfg.dsp_driver module-parameter
2022-01-18 17:35:22 <mps> donoban: ^
2022-01-18 17:35:48 <mps> does this looks as it should be enabled also for your case
2022-01-18 17:35:49 <donoban> I tried to force that via modprobe.d and it was worse
2022-01-18 17:35:59 <donoban> but I maybe it's better with this new kernel
2022-01-18 17:39:01 <mps> donoban: here is diff which I think could solve your problem but not 100% sure https://tpaste.us/8ElW
2022-01-18 17:40:25 <donoban> well, with last kernel I get the same errors
2022-01-18 17:40:44 <donoban> it seems that it's using the old driver:  intel_sst_acpi 808622A8:00: FW Version 01.0b.02.02
2022-01-18 17:42:41 <mps> I can push upgrade though not sure if it will fix your problem, if I'm sure I would push it for sure
2022-01-18 17:43:30 <mps> but on next upgrade it will be there for sure (if you can wait few days)
2022-01-18 17:43:56 <mps> donoban: that is on x86_64?
2022-01-18 17:43:57 <donoban> no worries, it's not urgent
2022-01-18 17:44:10 <donoban> yes
2022-01-18 17:44:17 <donoban> I think that I could try to build it with my laptop
2022-01-18 17:44:28 <donoban> or maybe even on this device
2022-01-18 17:45:47 <mps> diff to config is above, I think this could work
2022-01-18 17:45:56 <donoban> ok I will try it
2022-01-18 17:46:08 <donoban> I forced sof driver and no sound card is detected
2022-01-18 17:46:13 <donoban> ALSA lib confmisc.c:855:(parse_card) cannot find card '0'
2022-01-18 17:47:28 <donoban> I will ping you if the new build helps, thanks mps 
2022-01-18 18:07:36 <maribu> PureTryOut: Regarding setting the default sink on pipewire: It works on the same notebook PC at home. Maybe my USB connected headphones at work trigger a bug.
2022-01-18 19:03:44 <pabloyoyoista_> ncopa: Is it safe to remove this patch from glib? https://gitlab.alpinelinux.org/alpine/aports/-/blob/master/main/glib/0001-gquark-fix-initialization-with-c-constructors.patch It seems to me that one of your original proposals has been accepted https://gitlab.gnome.org/GNOME/glib/-/issues/1177#note_454483 and issue in glibmm fixed
2022-01-18 19:55:05 <donoban> mps: just testing the diff config, stills failing
2022-01-18 20:11:43 <minimal> donoban: installed the sof-firmware package?
2022-01-18 20:12:08 <donoban> uhm
2022-01-18 20:12:44 <donoban> (1/1) Installing sof-firmware@testing (1.9.2-r0)
2022-01-18 20:12:47 <donoban> let's try
2022-01-18 20:12:54 <minimal> and possibly sof-tools as well
2022-01-18 20:15:39 <donoban> [    3.097780] sof-audio-acpi-intel-byt 808622A8:00: Direct firmware load for intel/sof/sof-cht.ri failed with e
2022-01-18 20:16:26 <psykose> yes
2022-01-18 20:17:14 <psykose> though i think the sof-firmware package is not packaged correctly as that path doesn't exist in it, instead it's intel/sof-v1.9.2/sof-cht.ri
2022-01-18 20:17:26 <psykose> the ver has to be missing, otherwise you need to move it yourself
2022-01-18 20:17:54 <donoban> or make a symlink?
2022-01-18 20:18:57 <psykose> yes
2022-01-18 20:19:04 <mps> modinfo module_name | grep firmware
2022-01-18 20:20:15 <donoban> hmm...
2022-01-18 20:20:42 <mps> and then you will know where it expect it, iirc
2022-01-18 20:20:59 <donoban> the symlink already exists sof -> sof-v1.9.2
2022-01-18 20:21:05 <donoban> ok, let's check it
2022-01-18 20:21:52 <psykose> ah
2022-01-18 20:21:52 <donoban> ouch
2022-01-18 20:21:54 <psykose> neat
2022-01-18 20:21:59 <donoban> modinfo: ERROR: Module sof-audio-acpi-intel-byt not found.
2022-01-18 20:22:20 <mps> donoban: `modinfo `lsmod | grep -v Module | awk -e '{ print $1 }'` | grep ^firmware`
2022-01-18 20:22:51 <mps> try with this, you will get list of all firmware in modules which need firmware
2022-01-18 20:24:39 <donoban> uhM
2022-01-18 20:24:59 <donoban> https://tpaste.us/PrlJ
2022-01-18 20:25:15 <donoban> it seems that there is nothing about firmware
2022-01-18 20:25:43 <psykose> one of the depends will have it in there
2022-01-18 20:26:06 <donoban> I'm gonna check mps command
2022-01-18 20:26:10 <mps> donoban: try above line I posted
2022-01-18 20:26:14 <donoban> ok ok
2022-01-18 20:26:22 <mps> ok ok :)
2022-01-18 20:27:07 <donoban> uhM
2022-01-18 20:28:03 <donoban> modinfo `lsmod | grep -v snd_sof_acpi_intel_byt | awk -e '{ print $1 }'` | grep ^firmware -> https://tpaste.us/6Wl9
2022-01-18 20:28:14 <donoban> this?
2022-01-18 20:28:45 <mps> no
2022-01-18 20:28:58 <mps> `modinfo `lsmod | grep -v Module | awk -e '{ print $1 }'` | grep ^firmware`
2022-01-18 20:29:24 <donoban> ahh
2022-01-18 20:29:26 <donoban> lol
2022-01-18 20:29:27 <mps> or `modinfo `lsmod | grep -v Module | awk -e '{ print $1 }'` | grep ^firmware | tpaste` :)
2022-01-18 20:29:43 <donoban> sorry I didn't understand the grep intention
2022-01-18 20:30:38 <donoban> https://tpaste.us/mqrv
2022-01-18 20:31:01 <donoban> it seems the same
2022-01-18 20:31:22 <minimal> is the relevant sof module loaded?
2022-01-18 20:31:47 <mps> yes, and looks like firmware is not needed for this (whatever it is) module
2022-01-18 20:31:55 <psykose> does the module not fail to load if the firmware is missing
2022-01-18 20:32:03 <psykose> in which case using lsmod to get missing firmware wouldn't work
2022-01-18 20:32:17 <mps> modinfo will show if external firmware is needed
2022-01-18 20:32:24 <psykose> yes but you are using modinfo on lsmod output
2022-01-18 20:32:28 <psykose> lsmod is only loaded modules
2022-01-18 20:32:34 <mps> right
2022-01-18 20:32:43 <psykose> so if they didn't load it would not be in the output
2022-01-18 20:32:53 <minimal> yupe, that's why I asked :-)
2022-01-18 20:32:55 <psykose> so it doesn't tell you 'you are missing some firmware' 
2022-01-18 20:33:03 <mps> kernel should detect needed module on boot and try to load it
2022-01-18 20:33:06 <psykose> yes
2022-01-18 20:33:12 <psykose> that is what dmesg is for and says as much above
2022-01-18 20:33:16 <psykose> lsmod is not really useful for this
2022-01-18 20:33:21 <mps> then dmesg | less and look
2022-01-18 20:33:58 <minimal> just do "modinfo /lib/modules/<kernel version>/path to the module file | grep -i firmware"
2022-01-18 20:34:08 <mps> lsmod is not 100% safe but in most cases it is useful in above chain
2022-01-18 20:34:19 <donoban> wellyes the module which failed to load the firmware is sof-audio-acpi-intel-byt
2022-01-18 20:34:23 <donoban> which doesn't appear on lsmod
2022-01-18 20:34:29 <mps> because some module stays loaded even if firmware is not found
2022-01-18 20:34:47 <minimal> donoban: yupe, so try the command I just posted
2022-01-18 20:34:56 <donoban> uhM, it doesn't seem a module
2022-01-18 20:35:03 <donoban> maybe it's internal built?
2022-01-18 20:35:17 <donoban> minimal: I don't found a module file for it
2022-01-18 20:36:47 <donoban> https://tpaste.us/REl1 this is the error on dmesg
2022-01-18 20:36:51 <mps> snd-sof-intel-atom.ko.gz
2022-01-18 20:37:28 <ericonr> donoban: if it's built-in, you need firmware in initramfs, usually
2022-01-18 20:37:30 <donoban> apparentely it doesn't need any firemware
2022-01-18 20:38:14 <donoban> hmm..
2022-01-18 20:38:52 <psykose> snd-sof-intel-byt.ko should be the name
2022-01-18 20:39:17 <minimal> source code has refs to multiple firmware files - sound/soc/intel/soc-acpi-intel-byt-match.c:
2022-01-18 20:39:23 <donoban> maybe is CONFIG_SND_SOC_INTEL_MACH=y ?
2022-01-18 20:40:18 <mps> huh, let me rebuild and look what modules are added
2022-01-18 20:40:56 <minimal> different files for different makes of device - each refs files ending in ..bin, .ri, and .tplg extensions
2022-01-18 20:41:21 <mps> minimal: yes, this is mess
2022-01-18 20:41:39 <donoban> uhM
2022-01-18 20:41:41 <minimal> so I assume soc-firmware provides all those files
2022-01-18 20:41:49 <mps> but I don't have intel machine to test this, just shooting in dark
2022-01-18 20:41:53 <donoban> I noticed that soc firmware is in /usr/lib/firmware
2022-01-18 20:41:58 <donoban> instead in /lib/firmware
2022-01-18 20:42:00 <donoban> is that ok?
2022-01-18 20:42:09 <mps> usualy no
2022-01-18 20:42:19 <donoban> localhost:/usr/lib/firmware$ ls
2022-01-18 20:42:21 <donoban> intel
2022-01-18 20:42:23 <donoban> it's alone
2022-01-18 20:42:46 <mps> this is probably copied from some other distro without checking
2022-01-18 20:42:56 <donoban> I'm gonna try manual moving
2022-01-18 20:43:29 <mps> these distro have symlink of /lib/firmware to /usr/lib/firmware
2022-01-18 20:43:29 <psykose> ah
2022-01-18 20:43:33 <psykose> that does look wrong
2022-01-18 20:44:02 <mps> 'smart' idea (pun intended) of /usr merge :)
2022-01-18 20:44:07 <psykose> Hello71: should sof-firmware not be /lib/firmware instead
2022-01-18 20:44:33 <donoban> [    3.110444] sof-audio-acpi-intel-byt 808622A8:00: Firmware info: version 1:9:2-e096c
2022-01-18 20:44:42 <donoban> it loaded the firmware now
2022-01-18 20:44:51 <donoban> localhost:~$ aplay alarm.wav
2022-01-18 20:44:53 <donoban> Playing WAVE 'alarm.wav' : Unsigned 8 bit, Rate 8000 Hz, Mono
2022-01-18 20:44:55 <donoban> ohh
2022-01-18 20:44:57 <donoban> I need to plug in a cable :)
2022-01-18 20:45:21 <mps> donoban: so I can stop building?
2022-01-18 20:46:04 <donoban> hmm I'm not sure
2022-01-18 20:46:10 <donoban> no sound yet, playing with alsamixer
2022-01-18 20:46:23 <mps> ok
2022-01-18 20:46:39 <Hello71> psykose: yes, i already have mr open for that
2022-01-18 20:46:39 <minimal> mps: I saw a reference on Phoronix that AMD are starting to use SOF also, so it guess its the modern replacement for everyone using "intel-hda"
2022-01-18 20:46:45 <psykose> ah
2022-01-18 20:46:52 <minimal> donoban: checked if your output is software muted?
2022-01-18 20:47:14 <donoban> hmm, not sure
2022-01-18 20:47:21 <donoban> I only see that there are tons of controls on alsamixer
2022-01-18 20:47:25 <donoban> many are just mute/unmute
2022-01-18 20:47:52 <mps> finding proper alsa switches and options is cumbersome nowadays
2022-01-18 20:49:30 <donoban> ouch, no sound yet
2022-01-18 20:50:49 <donoban> maybe do I need to do something with sof-tools?
2022-01-18 20:53:17 <mps> if alsamixer works then the card is properly initialized, I think. problem solved on my side, donoban?
2022-01-18 20:54:02 <donoban> hehe probably
2022-01-18 20:54:57 <donoban> I discovered a control
2022-01-18 20:55:07 <donoban> "OUT MIXL INL1"
2022-01-18 20:55:20 <donoban> that does some uggly sound when switching
2022-01-18 20:55:53 <donoban> localhost:/home/donoban# aplay alarm.wav
2022-01-18 20:55:55 <donoban> aplay: main:831: audio open error: Resource busy
2022-01-18 20:56:04 <donoban> uhM.., maybe is because spotifyd is playing
2022-01-18 21:00:39 <donoban> omg
2022-01-18 21:00:55 <donoban> now it sounds only left speakaer with some strange noise
2022-01-18 21:02:18 <donoban> yeah it works
2022-01-18 21:02:20 <donoban> thanks guys
2022-01-18 21:12:39 <donoban>  * Storing ALSA Mixer Levels ...                                                                                                                          [ ok ]
2022-01-18 21:13:47 <donoban> lol, it's unstable. It was playing fine and suddenly just "beeeeeeeeeeeeeeeeeeeeeeeeeeeeeep"
2022-01-18 21:13:51 <donoban> hehe
2022-01-18 21:34:57 <PureTryOut> <maribu> "PureTryOut: Regarding setting..." <- Yeah sounds like a bug specific to your hardware 🤷‍♂️
2022-01-19 03:48:11 <bweiss> I created a new aport for softether. The configure script failed on s390x and ppc64le. I have created a patch to correct the failure, and refactored the apkbuild. I am unsure how to proceed.
2022-01-19 03:52:10 <ktprograms> bweiss: Are you getting an error?
2022-01-19 03:53:50 <psykose> is this published anywhere
2022-01-19 03:54:05 <psykose> nobody can help you if you can't post some output and what you're doing
2022-01-19 03:55:30 <bweiss> here is the MR https://gitlab.alpinelinux.org/alpine/aports/-/merge_requests/29564
2022-01-19 03:55:50 <bweiss> the original one that is
2022-01-19 03:55:51 <psykose> well, to make changes you need to send a new patch
2022-01-19 03:56:18 <bweiss> right, but do I send the entire thing again or just the changes?
2022-01-19 03:57:02 <psykose> afaik entire
2022-01-19 03:57:05 <psykose> it makes a new mr every time
2022-01-19 03:58:06 <bweiss> so is the old mr considered dead/closed/whatever?
2022-01-19 03:58:08 <psykose> yes
2022-01-19 03:58:49 <bweiss> ok. So no need to call this new patch "v2"?
2022-01-19 03:59:49 <psykose> not really no
2022-01-19 04:04:09 <bweiss> ok, I'
2022-01-19 04:05:06 <bweiss> I'll send it back through. Thanks for your help- my first contribution, and I don't want to clutter up the system
2022-01-19 08:00:07 <staceee> Hey guys, I got this mr for a long time now. It is about loading /etc/profile.d/*.sh on non login bash sessions (terminal emulators by example). Could you check it ? !25253
2022-01-19 09:13:25 <ikke> ncopa: opinion on ^?
2022-01-19 10:04:52 <mps> gitlab web is slow today
2022-01-19 10:10:27 <clandmeter> mps: seems ok for me
2022-01-19 10:11:24 <mps> maybe my link is slow today then
2022-01-19 10:41:14 <PJ[m]> its been slow for me as well
2022-01-19 10:42:51 <ikke> Seems fast for me as well
2022-01-19 11:58:51 <donoban> my problem with bubblejail/tmpfs now seems more likely be related to bubblewrap and musl
2022-01-19 11:59:49 <donoban> bubblewrap tries to bind_mount() a recently created socket and fails with "No such file or directory"
2022-01-19 12:00:43 <donoban> if I add sleep(1), os.sync() or os.system("ls <that_dir>") it works
2022-01-19 12:01:05 <donoban> I feel that it is similar to https://github.com/containers/bubblewrap/issues/387
2022-01-19 12:04:33 <mps> how we package github pkg which does have submodules
2022-01-19 12:04:48 <ericonr> donoban: bind mounting should only be attempted when the directory is created. I wouldn't be surprised if adding the syscalls is delay enough to sync the operations, rather than a specific effect of the syscalls
2022-01-19 12:05:06 <mps> I'm preparing m1n1 (apple Silicon) bootloader and debug tool
2022-01-19 12:05:28 <psykose> add the submodules to source as gitrev checkouts
2022-01-19 12:05:37 <psykose> like when you download a commit targz from github
2022-01-19 12:05:50 <psykose> then mv them into the location
2022-01-19 12:06:02 <mps> any example
2022-01-19 12:06:24 <mps> https://tpaste.us/Zjvz that is where I'm now
2022-01-19 12:06:50 <mps> any example pkg in aports?
2022-01-19 12:07:11 <donoban> ericonr: according to 'ls' the directory and the sockets exist before launching bwrap. And running it as it is works like 50% of the time
2022-01-19 12:07:48 <psykose> main/luajit has a separate thing for tests
2022-01-19 12:08:03 <psykose> now imagine that but you copy the actual module to the folder where it's supposed to be
2022-01-19 12:08:16 <psykose> !28662 is another
2022-01-19 12:10:42 <mps> psykose: ok, will look if I can get idea, thanks
2022-01-19 12:11:02 <psykose> here it copies to external/ 
2022-01-19 12:11:02 <ericonr> donoban: my point is that running ls might be enough delay (what with fork+exec) to make the data race not manifest itself
2022-01-19 12:11:10 <psykose> if you checj .gitmodules you see where they are supposed to be
2022-01-19 12:11:14 <psykose> and it should 'just work'
2022-01-19 12:11:40 <donoban> ericonr: maybe, but I don't believe that upstream will accept it as solution
2022-01-19 12:11:57 <donoban> specially since they can't reproduce it
2022-01-19 12:12:19 <donoban> should I just add a patch for alpine?
2022-01-19 12:12:58 <ericonr> donoban: if upstream doesn't accept fixing a race condition, I wouldn't like to package their security software :P but I'm not an alpine dev
2022-01-19 12:14:22 <donoban> well, adding forced delays or forced fs sync are not good way for fixing race conditions
2022-01-19 12:14:33 <ericonr> which is why you need dependency ordering
2022-01-19 12:14:39 <donoban> it's probably something related with bwrap and musl, not bubblejail
2022-01-19 12:16:52 <ericonr> > <socket.socket fd=14, family=AddressFamily.AF_UNIX, type=SocketKind.SOCK_STREAM, proto=0, laddr=/run/user/1000/bubblejail/qutetor/dbus_system_proxy>
2022-01-19 12:17:08 <ericonr> are both of the messages printed to the same fd? (so stdout or stderr)
2022-01-19 12:17:42 <mps> psykose: I found that I can download tar.gz of submodule, untar and move it to proper dir, then build works
2022-01-19 12:17:45 <donoban> are just print()'s that I've added for debugging so stdout
2022-01-19 12:17:50 <psykose> yep, correct
2022-01-19 12:22:44 <ericonr> donoban: I see. that would suggest it happens before bwrap is called, but idk if that's true. idk how things are called, but could you make the bwrap command only be called after the function that creates the sockets has returned?
2022-01-19 12:23:06 <ericonr> looking at a strace would also be helpful
2022-01-19 12:23:26 <donoban> the LOL message is the return of that function
2022-01-19 12:23:41 <donoban> yeah I will try to strace it
2022-01-19 12:25:15 <donoban> hehe, the author accepts to add some hack if it runs on musl system 'I might add this hack in case the bubblejail gets launched on system with Musl.'
2022-01-19 12:26:37 <ericonr> that's horrifying
2022-01-19 12:27:01 <donoban> well, better that nothing
2022-01-19 12:27:25 <donoban> but hopefully we could find the root of the problem
2022-01-19 12:29:55 <ericonr> I left a comment there
2022-01-19 12:30:00 <donoban> hm..
2022-01-19 12:30:27 <donoban> it seems that using strace is not reproducible :(
2022-01-19 12:31:04 <donoban> running without strace, 3/3 failed
2022-01-19 12:31:35 <donoban> ouch
2022-01-19 12:31:42 <ericonr> I'd call that a race condition :P
2022-01-19 12:31:54 <ericonr> honestly I don't even get why it's doing async lol
2022-01-19 12:32:13 <donoban> yes, I thought the same too
2022-01-19 12:32:24 <donoban> but it seems that doing it sequentially won't fix this
2022-01-19 12:32:50 <ericonr> donoban: why?
2022-01-19 12:32:51 <donoban> using strace it works 100% of the time
2022-01-19 12:33:19 <ericonr> that it works under strace but not without it just makes me think "race condition" harder
2022-01-19 12:33:27 <ericonr> strace makes syscalls slower
2022-01-19 12:33:31 <ericonr> (afaik)
2022-01-19 12:34:06 <donoban> because it seems that the python execution order is correct
2022-01-19 12:34:33 <donoban> the problem is lauching bwrap too fast, it has some problem finding the bind source
2022-01-19 12:35:20 <ericonr> donoban: *this problem shouldn't exist*
2022-01-19 12:35:41 <ericonr> if I linearly create a socket and then a launch a program that uses that socket, it should always work
2022-01-19 12:35:50 <ericonr> if it doesn't, that's a horrible OS bug
2022-01-19 12:36:40 <ericonr> however, if my async code doesn't guarantee ordering between those two operations, it's my async code to blame
2022-01-19 12:37:50 <donoban> I think that the python order should be fine
2022-01-19 12:38:40 <ericonr> I don't trust it, because they didn't say "this shouldn't happen because my code guarantees it doesn't", they said "I have never seen this happen"
2022-01-19 12:38:41 <donoban> otherwise it should be failing in many systems
2022-01-19 12:39:25 <donoban> uhm yep
2022-01-19 12:55:00 <donoban> ericonr: !
2022-01-19 12:55:23 <ericonr> donoban: ?
2022-01-19 12:55:52 <donoban> I'm printing time() when socket connection ends and bwrap is launched
2022-01-19 12:56:22 <donoban> http://ix.io/3MLx
2022-01-19 12:57:24 <donoban> theorically it should work right?
2022-01-19 12:57:48 <donoban> this is a working diff:
2022-01-19 12:57:50 <donoban> 1642597047.756474
2022-01-19 12:57:52 <donoban> 1642597047.756535
2022-01-19 12:58:46 <mps> !29738 anyone have time to review, first time I tried such 'hacks' in aports
2022-01-19 12:59:47 <ericonr> donoban: it's still slightly racy debugging. I think you need some way to specify which operation each timestamp belongs to
2022-01-19 13:00:04 <Ariadne> mps: looks ok to me
2022-01-19 13:00:04 <donoban> uhM
2022-01-19 13:00:15 <mps> Ariadne: thanks
2022-01-19 13:01:08 <ericonr> donoban: just append one letter to each to be sure :)
2022-01-19 13:01:19 <donoban> hehe
2022-01-19 13:02:27 <donoban> http://ix.io/3MLE
2022-01-19 13:09:00 <ericonr> donoban: time() is called before launching bwrap, correct?
2022-01-19 13:11:06 <donoban> just before calling "            bwrap_process = await create_subprocess_exec(
2022-01-19 13:13:18 <donoban> I'm trying to speed up strace in the hope of reproduce the race condition
2022-01-19 13:13:49 <donoban> maybe doing some filtering
2022-01-19 13:14:52 <ericonr> idk enough about how strace works to know if that works
2022-01-19 13:17:35 <donoban> it doesn't seem reproducible with it
2022-01-19 13:17:39 <donoban> even filtering 99% of the syscalls
2022-01-19 13:20:18 <donoban> hmm.. I'm looking a working log and there are things like
2022-01-19 13:20:24 <donoban> 19577 readlink("/run/user/1000/bubblejail/qutetor/dbus_session_proxy", 0x7ffcb5a2236f, 4096) = -1 EINVAL (Invalid argument)
2022-01-19 13:24:13 <donoban> ahm, it's just checking if the mounpoint is a symlink
2022-01-19 13:40:04 <skarnet> cursed pathname
2022-01-19 13:41:51 <psykose> what's cursed about it
2022-01-19 13:41:56 <skarnet> what's not
2022-01-19 13:42:03 <psykose> it's just xdg_runtime_dir
2022-01-19 13:43:40 <skarnet> /run/user -> ohey that stupid fdo tempdir. qutetor -> ??? are we doing automatically generated pathnames now? dbus_session_proxy -> Lord get me out of here.
2022-01-19 13:44:51 <donoban> qutetor is the name of the bubblejail instance
2022-01-19 13:45:27 <skarnet> did you specify that or was it automatically generated?
2022-01-19 13:45:33 <donoban> I specify it
2022-01-19 13:45:40 <donoban> it's just qutebrowser over tor
2022-01-19 13:46:08 <skarnet> oh, that makes sense
2022-01-19 13:46:29 <donoban> bubblejail fakes the system and session dbus with a proxy which filters the messages over them
2022-01-19 13:46:42 <Ariadne> skarnet: fd.o is the UNIX vendor mentality centralized into a single org
2022-01-19 13:47:01 <skarnet> it's much worse than that
2022-01-19 15:26:01 <bl4ckb0ne> can I have a review for 
2022-01-19 15:26:03 <bl4ckb0ne> !29216
2022-01-19 15:26:05 <bl4ckb0ne> please
2022-01-19 15:32:47 <bl4ckb0ne> is somebody working on wine 7? I saw !29545 in the open requests
2022-01-19 15:33:10 <psykose> it's 6.23 on edge
2022-01-19 15:35:12 <bl4ckb0ne> ah yes, 29545 targets 3.15
2022-01-19 15:35:16 <bl4ckb0ne> guess ill do it
2022-01-19 15:56:08 <PureTryOut> does anybody actually use Wine on Alpine? Not sure how useful it is without multilib
2022-01-19 15:58:12 <MaximKarasev[m]> PureTryOut: I use it inside of podman container with Arch :D
2022-01-19 15:58:15 <PureTryOut> bl4ckb0ne: !29216 seems largely fine, left 2 small comments
2022-01-19 15:58:33 <PureTryOut> Maxim Karasev: interesting, what for? Are you running a pure 32-bit or 64-bit program?
2022-01-19 15:59:02 <bl4ckb0ne> PureTryOut: thanks
2022-01-19 15:59:14 <MaximKarasev[m]> PureTryOut: Pure 32-bit; I had nothing to do and wanted to play an old game
2022-01-19 15:59:17 <bl4ckb0ne> wine 7 has a neat feature to run 32bits on 64
2022-01-19 15:59:25 <PureTryOut> Ah cool
2022-01-19 15:59:30 <PureTryOut> bl4ckb0ne: yeah I just read that, interesting
2022-01-19 15:59:34 <PureTryOut> Makes it so much more useful on Alpine
2022-01-19 15:59:36 <bl4ckb0ne> https://www.winehq.org/announce/7.0
2022-01-19 15:59:38 <bl4ckb0ne> yep
2022-01-19 15:59:45 <MaximKarasev[m]> bl4ckb0ne: How full is that support?
2022-01-19 15:59:47 <bl4ckb0ne> i have d2 sitting in a folder waiting for this 
2022-01-19 16:00:16 <bl4ckb0ne> havent had the time to test the package locally, but if it's there it should be good
2022-01-19 16:04:46 <Hello71> 15:59 <bl4ckb0ne> wine 7 has a neat feature to run 32bits on 64
2022-01-19 16:05:02 <Hello71> the sentence "Once the remaining modules are converted to PE, this will make it possible to run 32-bit applications without installing 32-bit Unix libraries." implies that this feature is not complete
2022-01-19 16:08:51 <bl4ckb0ne> dammit, no d2 
2022-01-19 16:09:46 <PureTryOut> I'll stick to my Lutris Flatpak anyway, it's easy to install scripts are too good to miss
2022-01-19 16:09:59 <PureTryOut> *its
2022-01-19 16:21:13 <bl4ckb0ne> doesnt being the maintainer of a package also implies contributor?
2022-01-19 16:23:26 <ikke> bl4ckb0ne: yes
2022-01-19 16:27:13 <Ariadne> MaximKarasev[m]: it is fully functional, i use crossover on my M1 mac which is using the same code as that in upstream wine
2022-01-19 16:44:30 <PureTryOut> I'm not sure why the contributor line exists in the first place, but let's not omit it while we use it 🤷
2022-01-19 16:44:54 <bl4ckb0ne> true
2022-01-19 16:45:34 <ikke> There have been discussions for ages on whether to remove it or not
2022-01-19 16:56:16 <bl4ckb0ne> PureTryOut: CI fixed for sauerbraten
2022-01-19 18:07:54 <ikke> ncopa: jool-modules-lts is failing on s390x, complains about different compiler versions and "incompatible gcc/plugin versions"
2022-01-20 02:18:47 <asdflkj> is a github release tarball download link the best newapkbuild source url to use for a package hosted only there?
2022-01-20 02:29:49 <bl4ckb0ne> yes
2022-01-20 02:30:37 <asdflkj> thanks
2022-01-20 05:28:38 <ikke> mips64 is back :D
2022-01-20 05:52:05 <boomanaiden154> Guess it would’ve been nice if a bunch of APKBUILDs hadn’t been changed regarding mips64. 😛
2022-01-20 05:53:47 <ikke> I think we're going to remove edge anyway
2022-01-20 05:55:42 <boomanaiden154> Is there a point at which it could be reenabled for edge? Or is the plan currently to only support already existing releases?
2022-01-20 05:59:19 <ikke> boomanaiden154: the hardware is EOL. As soon as the current hardware is broken, we have nothing anymore
2022-01-20 05:59:48 <ikke> so we decided to not continue with the port, as probably no one is ever going to use it, but at least keep the stable releases for the time being
2022-01-20 06:00:46 <ikke> I mean, the entire architecture is EOL
2022-01-20 06:01:53 <boomanaiden154> Makes sense. I totally blanked on MIPS64 being EOL.
2022-01-20 07:38:14 <ncopa> good morning
2022-01-20 07:39:09 <ncopa> ikke: I'll have a look at the s390x thingy. my guess is that kernel was built before gcc got updated
2022-01-20 11:08:52 <Ariadne> ikke: i don't even know, somebody at BISDN keeps making it work again lol
2022-01-20 11:13:21 <psykose> tell them to kill it already
2022-01-20 11:17:55 <pabloyoyoista_> ikke: Hi, I'm Pablo and I've been working on updating alpine's infra for the appstream-generator.  Cogitri has already merged all the changes to the docker image and the compose. Would it be possible to redeploy? Also, I am not sure how would migration work from the last generator, as there were some non-upstream patches. So might it be possible to remove the current volume?
2022-01-20 11:19:46 <pabloyoyoista_> I have had the generator running for some weeks already in a personal server, and it is working alright so far
2022-01-20 11:19:56 <pabloyoyoista_> Thanks a lot!
2022-01-20 11:20:27 <ikke> pabloyoyoista_: ok, will look into it
2022-01-20 11:20:50 <ikke> ncopa: Ariadne would it make sense to let the builder upload built packages for stable releases?
2022-01-20 11:20:58 <pabloyoyoista_> Amazing, thanks! 
2022-01-20 11:21:20 <Ariadne> ikke: yes, i think so.  presumably somebody keeps fixing it because they are using the port, i guess
2022-01-20 11:21:22 <ikke> So it's not a problem to loose all current content?
2022-01-20 11:21:44 <Ariadne> i mean there is literally a label on the box
2022-01-20 11:21:50 <Ariadne> that says "alpine mips64 builder"
2022-01-20 11:21:56 <Ariadne> its not too hard to figure out what it is :P
2022-01-20 11:22:08 <ikke> Ariadne: heh
2022-01-20 11:22:22 <Ariadne> though imo, with openwrt adopting apk, there is not much value in us offering a port
2022-01-20 11:35:26 <ikke> pabloyoyoista_: not sure why, but I get Error loading shared library libappstream-compose.so.0: No such file or directory (needed by /usr/local/bin/appstream-generator) when trying to run the update job
2022-01-20 11:35:54 <psykose> perhaps it needs appstream-compose installed
2022-01-20 11:36:49 <ikke> pabloyoyoista_: the last build job failed
2022-01-20 11:37:45 <pabloyoyoista_> ikke: Adding appstream-compose was one of the updates I did to the container
2022-01-20 11:38:02 <ikke> Yes, I see it in the merge request, but the job to build that image failed
2022-01-20 11:38:10 <ikke> armhf is failing
2022-01-20 11:38:13 <ikke> https://gitlab.alpinelinux.org/alpine/infra/docker/appstream-generator/-/jobs/599963
2022-01-20 11:38:25 <pabloyoyoista_> Yes, it's not expected to run on armhf
2022-01-20 11:38:45 <pabloyoyoista_> I believe neither the compiler, neither the generator can run on armhf
2022-01-20 11:38:53 <pabloyoyoista_> Only aarch64 and x86_64
2022-01-20 11:38:56 <ikke> Understood, it's just that our default docker image CI builds on all arches
2022-01-20 11:39:41 <pabloyoyoista_> Ok, should I send a MR adding a variable here then? version of the ubiquity slideshow html files and was wondering whether the corresponding screenshots should 
2022-01-20 11:39:51 <pabloyoyoista_> https://gitlab.alpinelinux.org/alpine/infra/docker/appstream-generator/-/blob/master/.gitlab-ci.yml#L5
2022-01-20 11:40:19 <ikke> pabloyoyoista_: I'll push it directly
2022-01-20 11:40:42 <pabloyoyoista_> Ok, thanks a lot. I didn't know it mattered. I will have it in mind for next time
2022-01-20 11:41:43 <ikke> https://gitlab.alpinelinux.org/alpine/infra/docker/appstream-generator/-/pipelines/107295
2022-01-20 11:45:41 <ikke> pabloyoyoista_: the generator is running now
2022-01-20 11:49:30 <pabloyoyoista_> ikke: Big, big thank you!
2022-01-20 12:46:10 <ktprograms> Hi, how do I reply to the wiki relicensing consent thread using aerc? (I'm not subscribed to the ~alpine/devel email list so I don't have the thread in my history)
2022-01-20 12:47:58 <psykose> subscribe to it then just send a message to the ml
2022-01-20 12:49:24 <ktprograms> psykose: How do I make sure it stays in the thread? (Trying to follow best practices as I'm new to it)
2022-01-20 12:50:03 <psykose> if you look at https://lists.alpinelinux.org/~alpine/devel/%3CCH58XEZPYFYT.1MV854Y4P2Q2O%40taiga%3E there is a reply at the bottom
2022-01-20 12:50:05 <psykose> that is a mailto:
2022-01-20 12:50:38 <psykose> if you open mailto: with /usr/bin/aerc it should work
2022-01-20 12:50:52 <psykose> like literally `aerc mailto:...` from that copied
2022-01-20 12:51:41 <ktprograms> psykose: I see, I'll give it a try now
2022-01-20 12:55:24 <ktprograms> How do mailing lists work that the message will be cc'd to the previous person?
2022-01-20 12:56:46 <psykose> you can remove that persons cc if you want, it was added because the person added it themselves i think
2022-01-20 13:11:50 <ktprograms> psykose: Ok, I _think_ that worked. Thanks
2022-01-20 13:22:05 <psykose> it did
2022-01-20 13:22:28 <ktprograms> Anything I did wrong?
2022-01-20 13:23:08 <psykose> nope
2022-01-20 13:23:20 <ktprograms> Ok, thanks for your help
2022-01-20 15:05:00 <ptrc> if a package has empty "Maintainer:" line, is the only contributor also the package's maintainer?
2022-01-20 15:06:56 <Hello71> no
2022-01-20 15:11:59 <psykose> no, it means it doesn't have a maintainer
2022-01-20 15:12:02 <psykose> you can usually take it
2022-01-20 15:12:23 <psykose> or ask someone to
2022-01-20 15:13:43 <ptrc> ok, thanks
2022-01-20 16:56:20 <ikke> pabloyoyoista_: seems like the updater errored out
2022-01-20 16:56:41 <ikke> https://tpaste.us/jNwq 
2022-01-20 20:47:03 <nmeum> how do I correctly merge a GitLab MR without rebasing it constantly and waiting for the tests to pass again?
2022-01-20 20:47:25 <nmeum> I usually do `curl -L <MR url>.patch | git am` and close the MR manually
2022-01-20 20:47:37 <nmeum> but if I recall correctly there was also a better way of doing this
2022-01-20 20:47:59 <ikke> I use glab
2022-01-20 20:48:12 <ikke> merge is an alias for glab mr merge --when-pipeline-succeeds=false -yr
2022-01-20 20:48:38 <ikke> The only thing you need to do manually afterwards is stop the new pipeline
2022-01-20 20:49:01 <ikke> You _can_ also do that via the interface btw
2022-01-20 20:49:21 <ikke> The 'merge when pipeline succeeds' button also has a 'merge immediately' option
2022-01-20 20:49:45 <nmeum> oh
2022-01-20 20:49:46 <nmeum> ah!
2022-01-20 20:50:28 <nmeum> that seems to work like a charm, thanks
2022-01-20 20:50:44 <ikke> Please do stop those redundant pipelines though
2022-01-20 20:50:49 <ikke> Keeps CI usuable
2022-01-20 20:53:21 <nmeum> ah oh, they are not stopped automatically
2022-01-20 20:53:23 <nmeum> will do, yes
2022-01-20 20:53:39 <ikke> usable* :)
2022-01-20 21:02:47 <ikke> andypost[m]: you said libuv fails on s390x due to ipv6?
2022-01-20 21:16:31 <andypost[m]> ikke: yes, only 1 test
2022-01-20 21:17:09 <ikke> Do you have an idea why it fails?
2022-01-20 21:17:41 <andypost[m]> As I see it now fails because of udp
2022-01-20 21:18:38 <andypost[m]> I commented on https://gitlab.alpinelinux.org/alpine/aports/-/merge_requests/29735#note_207733
2022-01-20 21:20:58 <ikke> Yes, -90 for nread sounds suspicious
2022-01-20 21:30:25 <Hello71> EMSGSIZE 90 Message too long
2022-01-20 21:31:10 <Hello71> strace would help
2022-01-20 21:31:29 <ikke> today is your lucky day
2022-01-20 21:31:40 <ikke> I just ran it with strace
2022-01-20 21:32:04 <ikke> https://tpaste.us/BglY
2022-01-20 21:36:17 <mps> ikke: clandmeter: ncopa: you don't need to discuss about me on council, I will leave alpine. case closed
2022-01-20 21:39:00 <Ariadne> ugh
2022-01-20 21:39:59 <Ariadne> the way this has gone down is unfortunate.  i wish i had provided guidance to Nulo on how to open an issue with the council when they expressed interest in still pursuing their complaint as a CoC issue
2022-01-20 21:40:12 <psykose> i don't think that would change the mps response
2022-01-20 21:40:20 <Nulo> ^
2022-01-20 21:40:30 <Ariadne> no, but i don't like that he was slagged on the mailing list.
2022-01-20 21:40:36 <psykose> eh
2022-01-20 21:40:37 <psykose> words are words
2022-01-20 21:40:47 <psykose> placing them into a gitlab issue page is not very different from the ml
2022-01-20 21:40:52 <psykose> the things said would be identical
2022-01-20 21:41:03 <Ariadne> does not matter.  if it happens again *I* will leave alpine
2022-01-20 21:41:16 <psykose> because someone placed the words in the wrong location?
2022-01-20 21:41:30 <psykose> i don't mean to say it's bad to place them in the correct one
2022-01-20 21:41:39 <psykose> it just technically does not change much of the outcome
2022-01-20 21:42:07 <Ariadne> I do not think it is appropriate to use the alpine-devel list in this way.  I think that is a CoC violation.
2022-01-20 21:42:29 <psykose> on paper i agree, but we are really just discussing abstractions
2022-01-20 21:42:33 <psykose> the feelings really don't change
2022-01-20 21:42:43 <Ariadne> no, we are not discussing just abstractions
2022-01-20 21:43:39 <Ariadne> the request for a CoC inquiry was not appropriate to be posted on alpine-devel, in my opinion, and the end result was that people started complaining about mps on the thread.
2022-01-20 21:44:15 <psykose> ddevault would post the same thing in the council thread too
2022-01-20 21:44:31 <Ariadne> if he were aware of it
2022-01-20 21:44:56 <Ariadne> the request for a CoC inquiry should have been filed as a private bug, until the council made a decision.
2022-01-20 21:45:04 <psykose> sure, fair enough
2022-01-20 21:45:19 <Ariadne> at which point, the decision should have been posted, and bug locked
2022-01-20 21:45:34 <Ariadne> which is literally how every other project does this
2022-01-20 21:45:39 <Ariadne> but instead we have "death by email"
2022-01-20 21:45:57 <Ariadne> while I agree Nulo's grievances have merit, I also think that "death by email" or "death by gitlab thread" is not okay either
2022-01-20 21:46:11 <Nulo> I understand, however: the CoC asks for things to be resolved transparently (which I agree with, unless victims or certain actors don't want to) and I didn't know the Council even existed.
2022-01-20 21:46:14 <skarnet> and people look at me funny when I say that pinging maintainers over public e-mail isn't a good idea
2022-01-20 21:47:15 <Ariadne> Nulo: it is because I failed to advise you of this, is why I am not pushing this on you.  it was my fault: I should have advised you of your escalation path.
2022-01-20 21:48:01 <Ariadne> i am neck deep in alpine governance, so i just kind of assume everyone knows everything to know about governance :P
2022-01-20 21:48:56 <Ariadne> mps: even if you choose to leave (i can hardly blame you after this), I do think it is important for the council to follow through with the process, if only so we can identify bugs in the process.
2022-01-20 21:49:20 <Nulo> I'm happy to add something in top of the post if it's of interest to avoid backlash or similar
2022-01-20 21:49:32 <Ariadne> what is done is done
2022-01-20 21:50:08 <Ariadne> as i said on the list, your grievances have merit, and it is unfortunate that our documentation does not explain how to resolve them effectively
2022-01-20 21:50:29 <Ariadne> I am willing to believe that if we had properly documented this, that you would have followed the appropriate governance structure
2022-01-20 21:50:53 <Ariadne> but, at the same time, slagging people on the mailing lists takes me back to 2018 and that doesn't make me feel happy
2022-01-20 21:51:06 <Ariadne> so, can we please not do that in the future :)
2022-01-20 21:51:18 <jvoisin> ^
2022-01-20 21:51:33 <Ariadne> *I* do not want to be in a project where we are airing "dirty laundry" like this
2022-01-20 21:52:56 <Nulo> I didn't mean it that way; my intention is to address the issue of CoC handling and harmful behavior publicly. I'm sorry for it looking that way, I will keep it in mind, but most importantly: I hope this makes these mechanisms available 
2022-01-20 21:53:31 <Ariadne> I have proposed a work item for the council to update the CoC from the draft I wrote in 2018 after the struggles we had in 2018
2022-01-20 21:53:37 <Ariadne> including redress
2022-01-20 21:53:58 <Nulo> Yes, I saw it. I think this is the most important outcome of this
2022-01-20 21:54:07 <Ariadne> but I am not on the council, nor wish to be on the council, nor wish to be on the CoC team, as I have limited time I can contribute to governance as it is
2022-01-20 21:58:46 <Hello71> ikke: hm, i don't understand. can you run again with strace -k (and musl-dbg installed)?
2022-01-20 22:00:26 <Ariadne> 47569 recvmmsg(10, 0x3ffc937ad50, 4, 0, NULL) = -1 EMSGSIZE (Message too large)
2022-01-20 22:00:46 <Ariadne> i think the test is probably flawed on s390x, perhaps a buffer needs to be increased
2022-01-20 22:00:49 <Hello71> right but why? why does the kernel return that? and why doesn't strace print the struct?
2022-01-20 22:01:29 <Hello71> my unlikely guess now is that the address is not mapped at all and for some reason kernel is returning EMSGSIZE instead of EFAULT
2022-01-20 22:02:44 <Hello71> also -e raw=recvmmsg might be helpful if it's easy to run
2022-01-20 22:03:07 <Hello71> maybe that one first then -y if it's easier not to install musl-dbg
2022-01-20 22:05:32 <ikke> Will do in a bit, am occupied now 
2022-01-20 22:07:07 <Hello71> ok
2022-01-20 22:10:20 <Hello71> ah, wait, this was fixed already https://github.com/libuv/libuv/commit/e9cb18484ecdc10e005d306db53d099e06ade9f4
2022-01-20 22:12:02 <Hello71> i am suspicious of the explanation though
2022-01-20 22:16:27 <Hello71> oh wait they're using raw syscalls with the libc msghdr
2022-01-20 22:17:22 <Ariadne> im sure libuv is doing all sorts of fucko shit
2022-01-20 22:17:31 <Nulo> Yeah that sounds wacky
2022-01-20 22:17:38 <Ariadne> every time i have gone spelunking through libuv source
2022-01-20 22:17:46 <Ariadne> i have not liked the experience
2022-01-20 22:18:29 <Hello71> and strace annoyingly gives up on printing it instead of printing what it can
2022-01-20 22:18:36 <Ariadne> Hello71: the explanation makes sense, on s390x structs can have somewhat exotic packing
2022-01-20 22:18:50 <Ariadne> i can do a test run in a few minutes
2022-01-20 22:19:41 <Hello71> well i thought they were using libc recvmmsg
2022-01-20 22:19:56 <Hello71> but they're using raw syscall with libc msghdr which of course doesn't work properly
2022-01-20 22:20:17 <Ariadne> yes, this is not really sane
2022-01-20 22:35:10 <Ariadne> Hello71: that fixes it
2022-01-20 22:46:45 <Ariadne> though the build environment needs to be cleaned up :P
2022-01-20 22:46:48 <Ariadne> bbl
2022-01-21 00:06:42 <Asmadeus> Hello71: oh you added -k to strace in 3.15! Awesome, I was still recompiling it locally when I needed it, thanks :)
2022-01-21 00:16:34 <Hello71> :)
2022-01-21 10:51:05 <MY-R> Hello71: I see #13442 closed but shouldnt be there some commit with exporting awk to use mawk? with gawk it would work withou any changes like you said before
2022-01-21 12:00:50 <donoban> Hi, I was checking security tracker and noticed that some CVE's like https://security.alpinelinux.org/vuln/CVE-2022-0156 doesn't affect alpine. Should they be fixed in the APKBUILD? is there some convention for it?
2022-01-21 12:30:12 <Hello71> MY-R: yes
2022-01-21 13:28:11 <Ariadne> donoban: https://gitlab.alpinelinux.org/ariadne/security-rejections
2022-01-21 13:31:34 <donoban> ok ty
2022-01-21 13:44:01 <donoban> uhM, maybe it's better just upgrade vim. I'm not sure about this releasing scheme. On github they have tons of tags (last v8.2.4171), edge currently has 8.2.3779-r1 and according to vim.org last stable version (for ms-windows) is  is gvim_8.2.2825.exe.
2022-01-21 14:17:33 <orbea> donoban: i think they make a tag for every commit...
2022-01-21 14:19:16 <donoban> yes, that seems
2022-01-21 14:25:37 <donoban> https://www.openwall.com/lists/oss-security/2022/01/15/1 , that is near the exact list that appears in https://security.alpinelinux.org/srcpkg/vim
2022-01-21 14:28:17 <donoban> I feel that most of them there are unexplotaible on alpine due gcc (some poc examples where built with clang) and fortify. I can't reproduce any of them. However it seems simple&safer to just upgrade the package and mark as solved
2022-01-21 15:27:35 <Ariadne> donoban: well, if you find yourself wanting to update the security-rejections database, just send an MR with the updates and explain why.  i will check them, of course, and if they seem correct, i will merge them into the DB
2022-01-21 15:28:19 <Ariadne> it is also possible to reject the CVE in the APKBUILD, by asserting it was fixed by version 0
2022-01-21 15:49:54 <donoban> well, if any overflow/uaf/similar is considered a security risk regardless compiler mitigations most of them should considered ok, and also it seems more worth to just upgrade it and mark as fixed. The problem is since they not have clear versioning scheme, should bump the version to master or maybe the last tag that fixed a security bug?
2022-01-21 15:56:07 <Ariadne> if there is a CVE that impacts Linux (regardless of whether we have a mitigation for it), it should be patched
2022-01-21 15:56:14 <Ariadne> and that CVE should be recorded
2022-01-21 15:56:47 <Ariadne> ultimately: what people care about is the security scanner being green
2022-01-21 16:02:17 <donoban> ok, so vim just should be upgraded https://security.alpinelinux.org/srcpkg/vim
2022-01-21 16:02:51 <donoban> maybe is better to ask ncopa what he thinks about upgrading to last tag or last that fixed some security bug
2022-01-21 16:02:57 <donoban> specially for backporting
2022-01-21 16:31:19 <Ariadne> in general, security team will back any NMU to fix a security bug as long as it is from the same stable branch
2022-01-21 16:31:28 <Ariadne> (sorry i was in a meeting)
2022-01-21 16:36:23 <donoban> Ariadne: no worries, I pushed this !29813
2022-01-21 16:36:56 <Ariadne> historically, the security team has consideres 8.2.x versions to all be on the same branch
2022-01-21 16:36:57 <donoban> the #secfixes versions refers to alpine versions or upstream?
2022-01-21 16:37:03 <Ariadne> alpine versions
2022-01-21 16:37:16 <donoban> ok
2022-01-21 16:37:59 <donoban> so will be ok to backport v8.2.4172 to 3.12-15?
2022-01-21 16:40:02 <Ariadne> yes
2022-01-21 16:40:10 <donoban> ok nice
2022-01-21 16:40:16 <Ariadne> but if you do this, please include any missing CVEs in the backports
2022-01-21 16:40:22 <Ariadne> so our secfixes db is accurate
2022-01-21 16:40:39 <donoban> should it identical to edge?
2022-01-21 16:40:45 <donoban> s/it/it be
2022-01-21 16:40:45 <alpine-meetbot> donoban meant to say: should it be identical to edge?
2022-01-21 16:42:48 <Ariadne> no, each branch should record which version in that branch addressed which CVE.
2022-01-21 16:43:18 <donoban> ah gotcha
2022-01-21 22:45:21 <tuzelko> Hello! Are there any plans to update the "nginx unit" package? I would like to use the current version of PHP, for example 8 or 8.1.
2022-01-22 01:57:27 <boomanaiden154> Anyone have any idea why this (https://tpaste.us/VE1e) python test runs fine locally inside a docker container (created from alpinelinux/build-base) but not on the gitlab CI?
2022-01-22 04:16:57 <PJ[m]> do you allocate tty for your container?
2022-01-22 04:39:49 <andypost[m]> @tuzelko please file issue with details what's need changes
2022-01-22 04:41:39 <andypost[m]> Btw there's only few aports depending on php7, I wanna move php7 to testing before unsupported before 3.16
2022-01-22 04:57:32 <boomanaiden154> PJ[m]: what exactly do you mean by that? The container has a shell that is used by the gitlab runner. There are also three other test cases that use the shell that pass. It only seems to be the tests that call the get_pid() function.
2022-01-22 04:57:49 <boomanaiden154> For some reason echo $$ doesn't print the PID of the process but rather just $$.
2022-01-22 04:58:00 <boomanaiden154> Sorry not mentioning build logs earlier. They're on !29824.
2022-01-22 05:19:01 <ktprograms> boomanaiden154: I think what PJ[m] meant by 'allocate tty', is did you add the `-t` (--tty) flag to your docker run command?
2022-01-22 05:27:19 <boomanaiden154> Yes. All the tests passed when I ran it locally and I did specify -t on the docked command.
2022-01-22 05:34:06 <ktprograms> boomanaiden154: For some reason, in the stdout, there's an extra \r\r\n in the echo: https://gitlab.alpinelinux.org/boomanaiden154/aports/-/jobs/601529#L5005
2022-01-22 05:45:27 <boomanaiden154> Interesting. It definitely shouldn’t be doing that. Guess it’s a point to start investigating though.
2022-01-22 13:56:31 <ktprograms> Hi, just wondering if https://wiki.alpinelinux.org/wiki/Alpine_Linux:Mailing_lists#Subscribing is still valid? When I sent an empty email to ~alpine/devel+subscribe@lists.alpinelinux.org, I got a reply from mailer@lists.alpinelinux.org with this content: https://tpaste.us/RED1. So it seems like there is no longer a second confirmation step.
2022-01-22 13:57:23 <ikke> ktprograms: most likely. We used to have a different mailinglist software
2022-01-22 14:00:01 <ktprograms> ikke: Ok, I'll update it when I get the time
2022-01-22 14:03:34 <ikke> ktprograms: not really extensive, but here is the documentation: https://man.sr.ht/lists.sr.ht/
2022-01-22 14:23:03 <PJ[m]> I mentioned that yesterday in alpine-docs :P
2022-01-22 14:34:38 <ktprograms> PJ[m]: I don't join #alpine-docs so I didn't see it
2022-01-22 18:38:44 <donoban> anyone has idea about this error https://gitlab.alpinelinux.org/donoban/aports/-/jobs/602317#L895 ? how should I test this marbling.c:181:5: note: use '-flax-vector-conversions' to permit conversions between vectors with differing element types or numbers of subparts
2022-01-22 18:39:01 <donoban> it builds fine on all archs except aarch64
2022-01-22 18:42:44 <aparcar> Ariadne: can you explain me the difference between apk dependencies being <> or ==?
2022-01-22 19:38:28 <sam_> i think you should instead report the issue upstream
2022-01-22 19:38:41 <psykose> yes
2022-01-22 19:39:28 <sam_> supposedly someone did already, fwiw: https://bugs.gentoo.org/830914#c4, but I'll follow up on it
2022-01-22 19:41:50 <donoban> ty
2022-01-22 19:43:19 <donoban> well, so until fixed let's add a patch
2022-01-22 19:44:32 <sam_> i'm not sure it's a good idea
2022-01-22 19:44:49 <sam_> although someone in the gentoo bug does claim that it seems to work okay at runtime with the flag, i still wouldn't
2022-01-22 19:48:52 <donoban> uhM, and disable it for aarch64?
2022-01-22 19:49:05 <donoban> is not very dangerous but current version is affected by a CVE
2022-01-22 19:53:59 <donoban> it would break lumina-desktop
2022-01-22 19:59:45 <sam_> afaik 6.01 was OK for the CVE but maybe not
2022-01-22 20:13:51 <Ariadne> aparcar: what do you mean ?
2022-01-22 20:22:40 <donoban> sam_: but it currently is 5.45-r0, I can try to upgrade to 6.01
2022-01-22 20:23:31 <sam_> donoban: i'll let an alpine expert help on that, just wanted to comment as I'd seen the build failure elsewhere :)
2022-01-22 20:39:49 <ikke> Would there be a reason to do `DESTDIR=foo make` instead of `make DESTDIR=foo`?
2022-01-22 20:41:43 <psykose> i can't think of any, the former sometimes doesn't work and the latter always does (with some exception for override afaik)
2022-01-22 20:42:15 <ikke> For some reason, J0WI changes this in a vim MR
2022-01-22 20:42:29 <psykose> probably an unrelated style change that has no point
2022-01-22 20:42:37 <ikke> yeah
2022-01-22 20:42:50 <ikke> That was my thought as well
2022-01-22 20:42:52 <psykose> the output should be the same anyway- if the former wasn't read correctly it would be profoundly broken
2022-01-22 20:43:00 <ikke> yup
2022-01-22 22:14:45 <Ariadne> sometimes there is an important difference.  makefiles sometimes disregard environment and instead take a variable as a command line argument
2022-01-22 22:15:07 <ikke> psykose already mentioned that
2022-01-22 22:15:13 <ikke> And I was aware of it as well
2022-01-22 22:17:32 <sam_> DESTDIR=foo make looks odd anyway
2022-01-22 22:17:34 <sam_> even if it is style, it's unconventional
2022-01-22 23:38:12 <skarnet> both should work because no Makefile should define a default DESTDIR anyway
2022-01-22 23:38:22 <skarnet> but traditionally it's a make variable, not an environment variable
2022-01-23 01:41:04 <ktprograms> Hi, I packaged qt5-qtpositioning because I needed it for building something I wanted to test (but not package, GPXSee), mainly modified the APKBUILD from the other qt5-* packages. I'm not really interested in maintaining it, so what should I do? (I just have the APKBUILD on my local system, and I can confirm that it works)
2022-01-23 01:52:13 <Ariadne> you could contribute it to testing with the Maintainer field blanked if you want
2022-01-23 01:52:41 <Ariadne> just say in the MR, "since i packaged it for testing, might as well contribute it, even though i'm not interested in being maintainer" or such
2022-01-23 01:54:14 <ktprograms> Ariadne: Ok, and should I also ask PureTryOut if he wants to maintain it, seeing that all the other qt5-* packages are maintained by him?
2022-01-23 01:54:23 <Ariadne> sure
2022-01-23 01:54:35 <ktprograms> ok thanks!
2022-01-23 01:58:39 <ktprograms> Ariadne: Is '(LGPL-2.0-or-later OR GPL-3.0-only OR GPL-2.0-or-later) AND Qt-GPL-exception-1.0' the correct license for all QT packages? That seems to be what I find on the other qt5-* packages
2022-01-23 01:58:48 <Ariadne> i mean
2022-01-23 01:58:50 <Ariadne> idk
2022-01-23 01:58:52 <Ariadne> probably
2022-01-23 01:59:43 <ktprograms> ok
2022-01-23 02:02:06 <omni> !28535 was approved a week ago but not yet merged
2022-01-23 09:06:54 <PureTryOut> ktprograms: note that qt5-qtpositioning is qt5-qtlocation. Only in Qt6 did it get renamed to qt6-qtpositioning. Confusing I know, but just use the existing qt5-qtlocation package
2022-01-23 09:22:23 <andypost[m]> Looks dup.pw domain has gone and links from commits channel are broken
2022-01-23 09:36:54 <ikke> oof
2022-01-23 09:37:20 <ikke> I've notified ncopa
2022-01-23 09:54:13 <ikke> ncopa renewed the domain, should be back shortly
2022-01-23 10:03:54 <ikke> for me it's working again
2022-01-23 10:06:33 <psykose> me too
2022-01-23 10:36:31 <andypost[m]> Thanks, it works
2022-01-23 11:22:29 <donoban> Hi, I would like to enable -flax-vector-conversions only for aarch64 (https://gitlab.alpinelinux.org/alpine/aports/-/merge_requests/29851/diffs#b378c09c690dd6d60f76bfa865a177808510e5d9_34_38) should work if I edit CFLAGS outside the build() func?
2022-01-23 11:28:25 <psykose> does a regular export not work
2022-01-23 11:28:52 <donoban> I'm gona try defining it before calling ./configure
2022-01-23 11:29:14 <psykose> https://img.ayaya.dev/thNezK3ColLw like this
2022-01-23 11:32:10 <PureTryOut> what is the policy on packaging FOSS launchers for proprietary applications? As example I'd like to package https://github.com/PolyMC/PolyMC which is FOSS but launches the proprietary game Minecraft
2022-01-23 11:34:55 <donoban> psykose: can I use just [ "$CARCH" = aarch64" ] && export .... ?
2022-01-23 11:35:09 <psykose> sure, you can do whatever you want
2022-01-23 11:35:13 <donoban> ok
2022-01-23 11:35:34 <psykose> i just personally prefer multiline :p
2022-01-23 12:59:21 <psykose> PureTryOut: is there a way to get pipewire-jack to not conflict with jack-dev that is sane? it gets a little tiring needing to remove every jack-linked application and the former to pull in jack-dev any time one builds something that needs jack-dev
2022-01-23 13:29:17 <ktprograms> PureTryOut: (re qt5-qtpositioning) I see, thanks for informing me.
2022-01-23 13:42:27 <fraolt> Novice maintainer here. When someone adds the label "status:mr-changes-requested" to one of my MR's, what is the expectation after I resolved all issues? Do I remove the label?
2022-01-23 13:42:39 <ikke> which MR?
2022-01-23 13:43:16 <omni> is there any helper script to bump pkgrel on all aports that, for example, makedepends on ${aport}-dev?
2022-01-23 13:43:56 <ikke> ap revdep + apkgrel
2022-01-23 13:45:10 <psykose> it does miss recursions though, if they matter, and for that there is recursive-revdeps, and then you have to filter dupes
2022-01-23 13:46:06 <fraolt> ikke: !29054
2022-01-23 13:47:05 <fraolt> you did the review, so you'll probably get a mail. My question was more of a general wondering. :-)
2022-01-23 13:54:20 <nmeum> psykose: if you use abuild rootbld, i.e. use a clean chroot to build stuff that needs jack-dev you won't run into this but not sure if that's what you are looking for
2022-01-23 13:54:38 <psykose> that i already know of, but it comes with it's own set of issues
2022-01-23 13:54:44 <nmeum> yes, true
2022-01-23 13:55:34 <psykose> installing all the packages takes longer, i have to patch abuild to get it to pass the apk cache, it would need even more patches to pass sccache specifically as well, options=net is needed for quite a few things,,, etc
2022-01-23 13:55:41 <psykose> but i do use it for verifying quite a few things
2022-01-23 13:58:28 <nmeum> I am always adding options=net whenever I encounter issues with it, there is also a long-time plan to migrate the builders to rootbld eventually at tsc/33
2022-01-23 13:59:01 <nmeum> we do have support for using ccache with abuild rootbld via abuild.conf, maybe something similar could also be added for sccache (though I am not familiar with it)
2022-01-23 14:00:43 <psykose> it's mostly the same, though i've had to make the hardlinks manually, and i'm fine with it being unsupported
2022-01-23 14:00:53 <psykose> as for the net builder stuff i've followed through that so far
2022-01-23 14:02:38 <nmeum> *nod*
2022-01-23 14:07:04 <ikke> fraolt: non-contributors (as role in gitlab) cannot change labels
2022-01-23 14:07:36 <ikke> fraolt: so generally you just need to wait, or gently ping someone if it's taking longer
2022-01-23 14:08:50 <PureTryOut> psykose: uh if you build something that needs jack-dev, it just shouldn't pull in pipewire-jack
2022-01-23 14:09:04 <PureTryOut> I tried quite hard to already make that the case
2022-01-23 14:09:10 <psykose> PureTryOut: it doesn't pull it in itself, i just mean that on a desktop i have the latter installed since.. well, jack support
2022-01-23 14:09:36 <PureTryOut> and you build things without rootbld I'm assuming?
2022-01-23 14:09:43 <PureTryOut> start using rootbld, it's awesome 😄
2022-01-23 14:09:56 <fraolt> ikke, thanks for explaining. :-)
2022-01-23 14:10:03 <psykose> i use both, rootbld is just a slower workflow so i run it once at the end
2022-01-23 14:10:03 <PureTryOut> that way whatever package you build builds in a clean chroot and isn't influenced by your host
2022-01-23 14:10:31 <PureTryOut> I see no proper way to have both pipewire-jack and jack-dev installed sorry, I advise just to use rootbld for those packages
2022-01-23 14:10:40 <psykose> sure, no worries :)
2022-01-23 14:10:42 <psykose> thanks for the input
2022-01-23 14:11:28 <PureTryOut> np
2022-01-23 14:11:36 <ikke> or use a chroot
2022-01-23 14:12:01 <ikke> I build my packages in an lxc container
2022-01-23 14:12:05 <ikke> assisted with rootbld
2022-01-23 14:15:44 <ikke> lol, a pipeline that goes green _after_ merging (after manually verifying it builds in lxc)
2022-01-23 14:16:31 <psykose> my favorite
2022-01-23 14:17:12 <Hello71> gentoo has a "jack-sdk"
2022-01-23 14:22:45 <PureTryOut> what is the policy on packaging FOSS launchers for proprietary applications? As example I'd like to package https://github.com/PolyMC/PolyMC which is FOSS but launches the proprietary game Minecraft
2022-01-23 14:23:49 <ikke> There is a pragmatic and a dogmatic answer to that
2022-01-23 14:26:41 <ikke> The policy is that the things we ship should have an appropriate license
2022-01-23 14:27:15 <Hello71> ah, i see, jack-sdk builds programs against pipewire-jack, not regular jack
2022-01-23 14:27:56 <psykose> technically it can build against either and be ran with either as the abi is the same, or something
2022-01-23 14:27:59 <psykose> from what i hear
2022-01-23 14:31:14 <PureTryOut> ikke: and what is the actual "we allow this" or "we do not allow this" answer to it? 😛 Because the package itself is FOSS, it'll just download a proprietary program and run that
2022-01-23 14:38:08 <ikke> PureTryOut: afaik, that should not be an issue
2022-01-23 14:38:25 <ikke> as long as we do not distribute the propriatary files
2022-01-23 14:38:33 <PureTryOut> That is not the case
2022-01-23 14:38:49 <PureTryOut> Ok thanks, will add the package then. I'll remove it if it becomes a policy after all to ban that later on
2022-01-23 14:39:24 <Ariadne> we only care about the license
2022-01-23 14:39:37 <Hello71> ah, i misread it, i thought pipewire_jack was libpipewire-jack.so but it's just libjack.so
2022-01-23 14:44:08 <aparcar> Ariadne: for reference, in APK a = means "exact version", == doesn't exists, I remembered it wrong and `<>` means specific checksum of a package.
2022-01-23 14:51:41 <Ariadne> yes, i just did not know what you were asking because the ==
2022-01-23 16:49:49 <markand> one of my package has been upgraded with my approval, again
2022-01-23 16:50:48 <jvoisin> with or without?
2022-01-23 16:50:55 <markand> without 
2022-01-23 16:50:58 <jvoisin> either way, if it's only an upgrade, why do you care?
2022-01-23 16:51:07 <ikke> markand: which one?
2022-01-23 16:51:09 <markand> so what's the purpose of a maintainer?
2022-01-23 16:51:25 <markand> https://gitlab.alpinelinux.org/alpine/aports/-/merge_requests/29796
2022-01-23 16:51:35 <markand> upgrading a library can have side effects
2022-01-23 16:56:10 <ikke> andypost[m]: please verify merge requests with maintainers if it's not your own package
2022-01-23 17:38:23 <telmich> I've to complain - alpine is too easy. I find myself too often in "just taking alpine + a few commands" = new container instead of re-using existing (more complex) containers
2022-01-23 17:38:54 <ikke> telmich: You have my appologies
2022-01-23 17:44:34 <ikke> markand: https://gitlab.alpinelinux.org/alpine/tsc/-/issues/22
2022-01-23 17:45:47 <ikke> markand: And I've now also created https://gitlab.alpinelinux.org/alpine/tsc/-/issues/36
2022-01-23 17:51:48 <Ariadne> telmich: ok we will make it harder
2022-01-23 17:52:02 <psykose> lewd
2022-01-23 17:52:07 <Ariadne> i will open TSC plan to switch alpine to "boot to kubernetes plus helm by default"
2022-01-23 17:52:10 <Ariadne> :D
2022-01-23 17:52:14 <ikke> heh
2022-01-23 17:52:26 <ikke> and selinux setenforce
2022-01-23 17:52:29 <Ariadne> yes
2022-01-23 17:52:34 <Ariadne> selinux for everyone!
2022-01-23 17:54:34 <psykose> also bring back grsec, straight back to 4.x
2022-01-23 17:56:25 <telmich> Ariadne: That is actually how we use Alpine a lot
2022-01-23 17:56:33 <telmich> (not joking)
2022-01-23 17:56:40 <Ariadne> psykose: we actually made grsec work well
2022-01-23 17:56:53 <Ariadne> it is a shame that spender decided to take it private
2022-01-23 17:57:12 <psykose> mhm
2022-01-23 17:57:44 <Ariadne> he loves the GPL so much he runs a company that uses GPLv2 enforcement maliciously
2022-01-23 17:57:49 <Ariadne> he said so on twitter
2022-01-23 17:57:56 <psykose> i've read the thread lol
2022-01-23 17:58:00 <Ariadne> i guess he loves the GPL like Microsoft loves Linux
2022-01-23 17:58:31 <Ariadne> telmich: my apologies
2022-01-23 17:58:45 <Ariadne> someday you will get to retire
2022-01-23 17:58:46 <markand> ikke, thanks
2022-01-23 17:58:53 <Ariadne> and you can just pet sheep for a living from then on
2022-01-23 17:58:56 <Ariadne> no more kubernetes
2022-01-23 17:59:10 <telmich> Actually, there is some former staff of ungleich that +/- does that ...
2022-01-23 17:59:17 <telmich> ... and that *is* actually related to k8s ...
2022-01-23 17:59:20 <ikke> hah
2022-01-23 17:59:32 <ikke> Must be nice to herd animals in the alps
2022-01-23 17:59:57 <telmich> My k8s TL;DR is: after you spent ca. 1y working time, it totally makes sense and is incredibly fast.
2022-01-23 18:00:33 <telmich> If you stop anywhere before that, likelihood of becoming a farmer are magnitudes higher
2022-01-23 18:00:52 <telmich> ikke: Feel free to visit, I can introduce you - summer stays are also possible
2022-01-23 18:02:09 <ikke> thanks :)
2022-01-23 18:02:16 <Ariadne> kubernetes was once called "borg"
2022-01-23 18:02:22 <Ariadne> sounds like it takes 1 year to be "assimilated"
2022-01-23 18:02:33 <ikke> Ariadne: isn't google still using borg internally?
2022-01-23 18:02:40 <Ariadne> idk
2022-01-23 18:03:02 <boomanaiden154> I thought Kuburnetes was a complete rewrite only loosely based on borg?
2022-01-23 18:03:07 <ikke> yeah
2022-01-23 18:03:08 <Ariadne> look
2022-01-23 18:03:10 <Ariadne> it was a joke
2022-01-23 18:03:12 <ikke> heh
2022-01-23 18:03:15 <Ariadne> you guys are overthinking it
2022-01-23 18:03:20 <telmich> Welcome to k8s
2022-01-23 18:03:25 <psykose> what are jokes
2022-01-23 18:03:26 <ikke> :D
2022-01-23 18:18:59 <omni> jokes are part of the control plane
2022-01-23 18:20:13 <ikke> I'm just a mere pod
2022-01-23 18:43:37 <telmich> I assume you are aware that k8s supports CRDs (custom resource definitions) and that you could easily define a joke resource
2022-01-23 18:45:38 <Ariadne> LALALALA
2022-01-23 18:45:46 <Ariadne> i am NOT listening to this kubernetes propaganda
2022-01-23 18:48:55 <aparcar> Ariadne: news on uclient etc? I'm keen to try it
2022-01-23 18:49:04 <Ariadne> soon(tm)
2022-01-23 18:49:06 <aparcar> getting the last bits of OpenWrt stuff together
2022-01-23 18:49:14 <aparcar> all right :)
2022-01-23 18:49:15 <Ariadne> probably on monday or tuesday
2022-01-23 18:49:21 <Ariadne> i need to wrap up some openbgpd stuff too
2022-01-23 18:49:23 <jvoisin> psykose: there are 5.1 grsec patches floating around :)
2022-01-23 18:49:36 <psykose> i'm sure there are
2022-01-23 18:49:50 <Ariadne> i think the main reason nobody wants to touch grsec at this point
2022-01-23 18:49:54 <Ariadne> is because they are afraid
2022-01-23 18:49:57 <Ariadne> spender will sue
2022-01-23 18:50:03 <Ariadne> and/or just show up at their house with a gun
2022-01-23 18:50:09 <jvoisin> it's GPLv2, E_DONTCARE
2022-01-23 18:50:48 <Ariadne> GPLv2 does not protect you from "second amendment remedies"
2022-01-23 19:17:44 <elibrokeit> honestly, the fact that your *security patches* are version-specific bootleg stuff sourced from "it was floating around the internet" is probably sufficient grounds for not wanting to touch it, too.
2022-01-23 19:18:35 <sam_> right, and "it applies" is also not "it does what it should"
2022-01-23 19:23:57 <elibrokeit> at best, I would say it is interesting to kernel developers interested in analyzing it, pulling apart the changes, and putting them in the review queue for mainline or using it as inspiration for newly coded changes of their own. But entirely too much of grsec regularly turns out to be not-upstreamable
2022-01-23 19:39:44 <jvoisin> elibrokeit: I mean, with the gpg signature, it's pretty legit :)
2022-01-23 19:45:37 <Garb0> Ariadne: `kubernetes propaganda`
2022-01-23 19:45:45 <Garb0> Are the agendaposters here? fire the alarms
2022-01-23 19:46:01 <Ariadne> i was kidding :)
2022-01-23 19:46:32 <Garb0> that did make me lol, though.
2022-01-23 19:49:33 <elibrokeit> jvoisin: I hadn't investigated since I have little interest in grsec, but that's a lot better than I feared.
2022-01-23 19:51:05 <elibrokeit> although I would not put it entirely beyond spender to publish false patchsets signed with a valid pgp signature and hope no one with the real patchset calls him on it
2022-01-23 19:54:42 <Garb0> very devestating.
2022-01-23 20:12:44 <andypost[m]> ikke: thanks, commented
2022-01-23 22:08:45 <genkitj> I'm kinda out of loop, but why does sway pull sxmo-wlroots?
2022-01-23 22:10:27 <genkitj> isn't sxmo some kind of mobile/pmOS only thing?
2022-01-23 22:10:47 <psykose> it is a bug
2022-01-23 22:10:52 <psykose> they both provide the same .so
2022-01-23 22:11:51 <psykose> you can `apk add !sxmo-wlroots` in the meantime
2022-01-23 22:12:58 <genkitj> thanks
2022-01-24 02:30:56 <ktprograms> Is it bad that setup-keymap allows setting any keymap, even the ones that cannot type ascii? There was a user of a virtualization software that was having problems installing alpine because they selected am (Armenian) for the keyboard layout (by mistake, not actually having an Armenian keyboard), and therefore (a) had white squares showing up when typing anything, and (b) couldn't select any 
2022-01-24 02:31:02 <ktprograms> options in the later parts of setup-alpine because it doesn't match (for example the disk name). I also checked, and it seems that if selecting am, but using a us keyboard, there is no way to press Ctrl-C to even exit.
2022-01-24 07:13:31 <ncopa> good morning
2022-01-24 07:45:17 <markand> ohai
2022-01-24 07:52:20 <andypost[m]> ncopa: please check #11528 there's 3 MRs which needs approval/commit
2022-01-24 10:03:38 <ncopa> andypost[m]: thanks. will have a look at it in a bit
2022-01-24 10:28:10 <omni> not sure if this is releated to the qt5-qtwayland upgrade the other day, but in qutebrowser I now have tiny text in rolldown menus, for example at pkgs.alpinelinux.org
2022-01-24 10:30:39 <ktprograms> omni: Does qutebrowser actually work for you? I'm getting 'Renderer process crashed (status 11)' whenever I open _anything_, even `:version`. (It was working for a while after the qtwebengine stack size patch, but now it isn't)
2022-01-24 10:31:24 <psykose> does for me
2022-01-24 10:31:56 <ktprograms> psykose: Hmm weird. I'll do more debugging when I get the time
2022-01-24 10:32:54 <omni> ktprograms: it does, but I too have crashes sometimes, I've been thinking of doing an update with the latest changes in https://invent.kde.org/qt/qt/qtwebengine-chromium/-/commits/87-based/ and see if that helps
2022-01-24 10:40:40 <omni> ktprograms: I'm using qutebrowser from edge btw, are you running on 3.15-stable?
2022-01-24 10:47:35 <ktprograms> omni: edge
2022-01-24 10:55:44 <omni> PureTryOut: ^^
2022-01-24 10:56:09 <omni> opened !29906
2022-01-24 16:04:32 <omni> btw, I agree that all my wiki contributions under my username omni are re-licensed under CC-BY-SA
2022-01-24 16:04:54 <omni> do I also need to send an email? I don't do much emailing these day
2022-01-24 16:05:35 <bl4ckb0ne> could somebody merge !29894 pls
2022-01-24 16:07:16 <ikke> omni: an email would be nice, as we have something on record in a central location 
2022-01-24 16:24:35 <omni> done
2022-01-24 16:36:12 <ikke> Thx! 
2022-01-24 17:57:48 <aparcar> Ariadne: if a packages "provides" something via "provides", is it considered to provide both the "name" and the list of "provides" or from then on only the "provides" list?
2022-01-24 17:58:10 <Ariadne> both
2022-01-24 17:58:16 <aparcar> excellent thank you
2022-01-24 17:58:24 <aparcar> I'm seeing wired issues that's why I'm asking
2022-01-24 17:59:00 <aparcar> https://paste.sr.ht/~aparcar/d4d708c5b33e21e32bc9f6eea753931152944e47
2022-01-24 17:59:54 <aparcar> it's odd since both packages are provided
2022-01-24 18:18:17 <oneinsect> ikke:
2022-01-24 18:18:20 <oneinsect> namaste
2022-01-24 18:57:51 <Ariadne> aparcar: it cant find a package in the repos which satisfy those checksums
2022-01-24 19:51:50 <Thermi> Hello hello my people, can somebody please merge https://gitlab.alpinelinux.org/alpine/aports/-/merge_requests/28781? Thank you!
2022-01-24 20:05:14 <Thermi> I believe the armhf, armv7, and s390x builders are stuck
2022-01-24 20:05:51 <ikke> Thermi: I just left a comment
2022-01-24 20:06:31 <ikke> Thermi: why do you think those builders are stuck?
2022-01-24 20:06:55 <ikke> You mean the CI runners?
2022-01-24 20:07:50 <Thermi> Yes
2022-01-24 20:08:00 <Thermi> Because there's no output for a couple of minutes already
2022-01-24 20:08:12 <Thermi> *no more
2022-01-24 20:19:27 <Thermi> ikke, it still hangs
2022-01-24 20:20:05 <ikke> I think it is continuing
2022-01-24 20:20:15 <ikke> Keeping an eye on it
2022-01-24 20:22:00 <ikke> Just finished
2022-01-24 20:35:06 <Thermi> Great, thank you
2022-01-24 20:51:47 <ikke> There are 2 long-running x86 jobs that keep the rest waiting 🤐
2022-01-24 21:08:28 <anjan> Hi all, a program I use just released a fix for a memory leak that was introduced in a security release
2022-01-24 21:08:39 <anjan> I obviously backported the security release
2022-01-24 21:08:47 <anjan> should I backport the memory leak fix too?
2022-01-24 21:09:07 <ikke> Yeah, makes sense
2022-01-24 21:12:31 <Thermi> Thank you ikke
2022-01-24 21:12:37 <Thermi> I will be off for now. See you next time. :)
2022-01-24 21:13:01 <ikke> o/
2022-01-25 09:30:35 <ikke> There is fakeroot-tcp in testing. Would it make sense to build it as a subpacke of fakeroot? 
2022-01-25 10:37:41 <psykose> sounds like it
2022-01-25 12:18:50 <mps> nmeum: you are zig contributor and iirc you told earlier that you are interested in keeping in good shape, would you mind to consider taking maintainership of it (I think it will be in good hands)
2022-01-25 12:20:16 <nmeum> mps: I am not a zig contributor, but I do play around with it occassionally for embedded stuff. I could take over maintainership if don't want it anymore
2022-01-25 12:21:18 <mps> nmeum: hmm, # Contributor: Sören Tempel <soeren+alpine@soeren-tempel.net>, aports/testing/zig/APKBUILD
2022-01-25 12:21:57 <nmeum> ah, in the aport you mean!
2022-01-25 12:22:02 <mps> yes
2022-01-25 12:22:03 <nmeum> I thought you were refering to upstream :D
2022-01-25 12:22:09 <mps> :)
2022-01-25 15:08:14 <aparcar> Ariadne: ping
2022-01-25 15:22:38 <skarnet> aparcar: don't hold your breath, she isn't well atm
2022-01-25 15:23:02 <aparcar> skarnet: I'm sorry to hear that
2022-01-25 18:40:33 <ikke> psykose: Enjoy your new status as developer
2022-01-25 18:42:10 <psykose> thank you!
2022-01-25 18:42:44 <psykose> is there anything in particular i need to join/do that i might miss
2022-01-25 18:45:14 <ikke> Can't think of anything particular at the moment
2022-01-25 18:47:32 <jvoisin> psykose: dink a tall glass of the beverage of your choice to celebrate I would say
2022-01-25 18:48:20 <psykose> perhaps i will get some tea then
2022-01-25 18:48:28 <omni> cheers!
2022-01-25 18:49:09 <psykose> ikke: there's the -private -team channels here but i am not sure if those are meant to be joined
2022-01-25 18:49:42 <ikke> psykose: #-team is used for the TSC
2022-01-25 18:49:48 <psykose> ah
2022-01-25 18:50:18 <ikke> -private is for, well private matters :)
2022-01-25 18:50:51 <ikke> (which we try to avoid as much as possible)
2022-01-25 18:51:57 <psykose> of course :)
2022-01-25 18:54:53 <omni> remember: with great power comes great current squeared times resistance
2022-01-25 18:55:17 <Saijin_Naib[m]> psykose: Congrats! Well-deserved for such a helpful person!
2022-01-25 18:55:29 <jvoisin> also, psykose, congratz!
2022-01-25 18:55:48 <psykose> aww :)
2022-01-25 18:55:53 <psykose> thank you all so much
2022-01-25 18:58:42 <bl4ckb0ne> gz psykose 
2022-01-25 19:00:10 <psykose> :3
2022-01-25 19:20:23 <Ariadne> skarnet: feeling like shit, dealing with a family shitshow, but still dazed and confused and trying to continue :)
2022-01-25 19:24:19 <psykose> ikke: do you see any issues with !29172
2022-01-25 19:25:28 <ikke> psykose: checking
2022-01-25 19:31:03 <PureTryOut> psykose: oh you're an official Alpine developer now? Welcome to the club, congratz! 😄
2022-01-25 19:32:53 <ikke> psykose: in general, we try to avoid enabling -dbg subpackage any everything pre-emptively. Sadly, our mirror diskspace is not infinite
2022-01-25 19:33:04 <psykose> sure, i can drop the -dbg
2022-01-25 19:33:22 <psykose> i initially added it there for debugging from ci artifact on the actual platform, fixed a sigsegv upstream with it
2022-01-25 19:33:25 <psykose> think it's not needed now
2022-01-25 19:33:31 <ikke> alright
2022-01-25 19:34:11 <ikke> Size difference for trunk-recorder-dbg:  -> 116 MiB Size difference for trunk-recorder:  -> 4528 KiB 🤐
2022-01-25 19:35:26 <ikke> 100MB of debug symbols for a 5MB package
2022-01-25 19:36:13 <ikke> Not sure what this means:
2022-01-25 19:36:15 <ikke> scanelf: rpath_security_checks(): Maybe? sec problem with DT_RPATH='lib/trunk-recorder' in /builds/alpine/aports/testing/trunk-recorder/pkg/trunk-recorder/usr/lib/trunk-recorder/libgnuradio-op25_repeater.so
2022-01-25 19:36:46 <psykose> i have no idea either
2022-01-25 19:39:33 <ericonr> ikke: maybe that the rpath isn't absolute?
2022-01-25 19:39:35 <ikke> psykose: oh, one request: after rebasing + merging a MR (if you merge before the pipeline finished): please cancel the redundant pipelines
2022-01-25 19:39:50 <psykose> yes, read about that before :)
2022-01-25 19:39:54 <ikke> ok good
2022-01-25 19:40:54 <ikke> maybe a good idea to do chrpath -r on that file I guess
2022-01-25 19:41:34 <psykose> with what other path? i'm not really sure what that rpath setting does
2022-01-25 19:41:46 <psykose> it can also probably be stripped via cmake install option
2022-01-25 19:42:08 <ikke> psykose: I believe it tells where to load libraries from
2022-01-25 19:42:43 <ericonr> it looks like it's missing /usr/, actually
2022-01-25 19:43:09 <ericonr> unless non absolute rpaths work some way I'm not aware of
2022-01-25 19:43:55 <psykose> i would assume relative to file and so pointing to nothing, but really no idea
2022-01-25 19:44:33 <ikke> According to https://stackoverflow.com/a/38058216/20261 it does not make sense
2022-01-25 19:44:37 <ikke> it would be relative to PWD
2022-01-25 19:44:54 <ikke> https://dev.my-gate.net/2021/08/04/understanding-rpath-with-cmake/
2022-01-25 19:45:50 <psykose> ok, will just strip it then
2022-01-25 19:50:40 <ikke> psykose: for the rest it looks alright
2022-01-25 19:51:00 <psykose> sure, i'll see which magic cmake option fixes that and merge it then
2022-01-25 19:51:02 <psykose> thanks!
2022-01-25 20:13:16 <eletrotupi> can someone take a look at !28986, please?
2022-01-25 20:28:24 <Saijin_Naib[m]> Is this the right venue to ask about the possibility of an linux-oldlts which tracks one LTS release prior to what we currently ship? So 5.10.x for this time period
2022-01-25 20:28:52 <Saijin_Naib[m]> Would that be appropriate for Edge even?
2022-01-25 20:32:15 <Ariadne> the kernel team is running ragged as it is
2022-01-25 20:33:01 <ikke> Saijin_Naib[m]: the kernel is one of the few things which is fairly safe to use from other stable releases
2022-01-25 20:34:48 <Saijin_Naib[m]> Ariadne: Absolutely understood. It's almost entirely just Nataneal, right?
2022-01-25 20:35:00 <Ariadne> yes
2022-01-25 20:35:04 <Saijin_Naib[m]> Since this is a selfish ask, if I "maintained" it in Edge, would it be unwelcome?
2022-01-25 20:35:16 <Saijin_Naib[m]> I can't use my laptop with anything newer than 5.11.22 :\
2022-01-25 20:35:26 <Ariadne> ...why not?
2022-01-25 20:36:39 <Saijin_Naib[m]> Ariadne: I wish I actually understood what is going on. It seems like something in the C-State/P-State handling changed in 5.12.x and above that breaks my laptop if it isn't on AC power the entire time or if I dont' completely disable C-State support.
2022-01-25 20:36:51 <Saijin_Naib[m]> It'll hard-freeze between 0 and 3min of boot
2022-01-25 20:37:04 <Ariadne> have you opened a bug at bugzilla.kernel.org
2022-01-25 20:37:11 <Saijin_Naib[m]> ikke: That's an interesting solution! Just grab an older old lts... Thanks for the tip
2022-01-25 20:37:30 <ikke> Saijin_Naib[m]: It's something I have done on a system where newer kernels would result in boot issues
2022-01-25 20:37:35 <Saijin_Naib[m]> Ariadne: I have not 😬 To be honest, I think I'm too ignorant to be helpful there after opening the issue
2022-01-25 20:38:01 <Saijin_Naib[m]> I would need major hand-holding to debug it
2022-01-25 20:38:17 <Ariadne> you could do a git bisect
2022-01-25 20:38:21 <Ariadne> between 5.11 and 5.12
2022-01-25 20:39:12 <Saijin_Naib[m]> And just try to intuit what might be impacting it?
2022-01-25 20:39:29 <Ariadne> well, the way a git bisect works is
2022-01-25 20:39:38 <Ariadne> it will walk you though the process
2022-01-25 20:39:51 <Ariadne> basically it will drop you half way between 5.11 and 5.12
2022-01-25 20:39:53 <ikke> didn't ncopa have a post about it?
2022-01-25 20:39:59 <ikke> not sure where
2022-01-25 20:40:06 <Saijin_Naib[m]> https://gitlab.alpinelinux.org/alpine/aports/-/issues/13163
2022-01-25 20:40:10 <Ariadne> and if that kernel works
2022-01-25 20:40:19 <Ariadne> then you go on
2022-01-25 20:40:23 <Ariadne> and it will drop you 75% of the way
2022-01-25 20:40:26 <Saijin_Naib[m]> This is my issue which I've been trying to track anything salient I can find out on my own
2022-01-25 20:40:29 <Ariadne> or if it fails, then it drops you 25%
2022-01-25 20:41:02 <Saijin_Naib[m]> Ahh, that's a really interesting process. I'll have to try and learn how to use that. I at least have a working abuild setup thanks to psykose and others, so that part is easy
2022-01-25 20:41:03 <Ariadne> https://wiki.ubuntu.com/Kernel/KernelBisection
2022-01-25 20:41:36 <Ariadne> this is ubuntu specific, but it gives a basic idea of the process
2022-01-25 20:42:27 <Saijin_Naib[m]> Thank you. This looks great
2022-01-25 20:42:34 <Ariadne> basically
2022-01-25 20:42:44 <Ariadne> git clone git@git.kernel.org:.../torvalds/linux
2022-01-25 20:42:53 <Ariadne> git bisect v5.12 v5.11
2022-01-25 20:43:01 <Ariadne> er
2022-01-25 20:43:04 <Ariadne> git bisect start v5.12 v5.11
2022-01-25 20:43:08 <Ariadne> and then build kernels
2022-01-25 20:43:17 <Ariadne> for each kernel, if it is good
2022-01-25 20:43:21 <Ariadne> git bisect good
2022-01-25 20:43:23 <Ariadne> if it is bad
2022-01-25 20:43:25 <Ariadne> git bisect bad
2022-01-25 20:43:37 <Ariadne> eventually, it will narrow down to the exact commit that introduced the regression
2022-01-25 20:43:45 <Saijin_Naib[m]> That's freaking awesome
2022-01-25 20:43:49 <Ariadne> note that this will likely take a day's worth of work to do
2022-01-25 20:43:53 <Saijin_Naib[m]> Thank you!
2022-01-25 20:44:31 <psykose> since it's binary search it takes like what
2022-01-25 20:44:36 <psykose> 10-12 attempts or something
2022-01-25 20:44:43 <Ariadne> yes, most likely
2022-01-25 20:44:44 <ikke> O(log n)
2022-01-25 20:44:48 <psykose> and you have to boot/test each one
2022-01-25 20:44:51 <psykose> mhm
2022-01-25 20:44:51 <Ariadne> it depends on how many commits there are
2022-01-25 20:44:52 <Ariadne> :)
2022-01-25 20:44:56 <psykose> the scaling is very little
2022-01-25 20:44:58 <Saijin_Naib[m]> With something like 30min-1hr per kernel build, depending upon how much ccache gits hit... Hmm
2022-01-25 20:45:08 <Ariadne> note that you don't have to make clean each time
2022-01-25 20:45:09 <psykose> ccache should get severely hit
2022-01-25 20:45:12 <Ariadne> most likely
2022-01-25 20:45:39 <Saijin_Naib[m]> Awesome. Project for when I need to keep the house warmer 🤣
2022-01-25 20:45:44 <Saijin_Naib[m]> Thank you all
2022-01-25 20:46:03 <Ariadne> but basically, it will try commits
2022-01-25 20:46:06 <Ariadne> and then you go through
2022-01-25 20:46:13 <ikke> I know another project that can help you keep you house warm :D
2022-01-25 20:46:14 <Ariadne> and if it works, it works, and if it fails, it fails
2022-01-25 20:46:46 <Saijin_Naib[m]> ikke: Oh no... Do I even want to know?
2022-01-25 20:46:53 <ikke> Saijin_Naib[m]: probably not
2022-01-25 20:47:19 <Saijin_Naib[m]> Ariadne: And the one that passes means the next commit is the one that breaks it. Then I use my judgement to figure out of this needs to go upstream to linux-kernel team vs Alpine, right?
2022-01-25 20:47:36 <Saijin_Naib[m]> And if this is a platform-specific "quirk", is it still valid?
2022-01-25 20:50:58 <skarnet> psykose.gz
2022-01-25 20:51:08 <psykose> at least make it .zst
2022-01-25 20:51:34 <skarnet> I just typed "gz psykose", short for "congrats", and that was the output
2022-01-25 20:51:35 <telmich> skarnet: why do I have the feeling it has been years since I saw you? Just realised you are in this channel, too!
2022-01-25 20:51:49 <psykose> Saijin_Naib[m]: platform specific is probably valid depending on what is happening, but this sounds like an upstream bug for sure
2022-01-25 20:52:15 <skarnet> telmich: because this channel is noisy and my interventions in it are few and far between? ;)
2022-01-25 20:52:56 <psykose> and most of them are also noise :p
2022-01-25 20:53:28 <skarnet> I wouldn't want to deviate too much from the channel spirit
2022-01-25 21:01:43 <Ariadne> Saijin_Naib[m]: you'll eventually get to a point where `git bisect good|bad` does not return any more commits to test, at which point you'll be on the one that broke it
2022-01-25 21:08:30 <andypost[m]> I'm looking for help to fix include for !29959
2022-01-25 21:12:25 <telmich> I assume we are also vulnerable by https://www.qualys.com/2022/01/25/cve-2021-4034/pwnkit.txt ?
2022-01-25 21:13:33 <ikke> `options="suid..` Yeah probably
2022-01-25 21:23:21 <ikke> This is the patch: https://gitlab.freedesktop.org/polkit/polkit/-/commit/a2bf5c9c83b6ae46cbd5c779d3055bff81ded683
2022-01-25 21:23:56 <ikke> hmm, 'argc < 1' ..
2022-01-25 21:27:50 <Saijin_Naib[m]> psykose: noted! Ariadne Perfect, thank you.
2022-01-25 21:34:22 <Ariadne> telmich: i consider polkit to be a vulnerability just by way of existence
2022-01-25 21:35:13 <ikke> :D
2022-01-25 21:35:32 <skarnet> agreed
2022-01-25 21:38:39 <ikke> in any case !29965
2022-01-25 21:45:54 <ikke> "This vulnerability is one of our most beautiful discoveries; to honor
2022-01-25 21:45:56 <ikke> its memory, we recommend listening to DJ Pone's "Falken's Maze" (double
2022-01-25 21:45:58 <ikke> pun intended) while reading this advisory. Thank you very much!"
2022-01-25 21:46:00 <ikke> :D
2022-01-25 22:24:25 <telmich> Ariadne: I cannot disagree with you on that one, even if I wanted to
2022-01-25 22:24:47 <telmich> It's so broken on so many levels.. and I wonder why I actually have it installed...
2022-01-25 22:25:39 <telmich> fwupd-plugin-all that is the one
2022-01-25 22:26:06 <telmich> fwupd does not run without it, afair
2022-01-25 22:27:48 <Ariadne> doesn't polkit send like
2022-01-25 22:27:52 <Ariadne> unencrypted passphrase
2022-01-25 22:27:54 <Ariadne> over dbus
2022-01-25 22:28:05 <Ariadne> or something crazy
2022-01-25 22:30:20 <telmich> That would totally make sense
2022-01-25 23:25:31 <Ariadne> i have a kernel patch that makes execv[e](2) fail when argc < 1
2022-01-25 23:26:36 <psykose> add it to linux-edge and i will tell you if anything explodes
2022-01-25 23:34:54 <Ariadne> https://tpaste.us/5nyE
2022-01-25 23:35:00 <Ariadne> the patch is very simple
2022-01-26 02:12:59 <Ariadne> looks like the proof of concept works but i need to refine it a little to throw -EFAULT as POSIX expects
2022-01-26 02:13:17 <Ariadne> i will do it in an hour or so
2022-01-26 02:25:02 <Ariadne> new patch: https://tpaste.us/DXgO
2022-01-26 04:51:55 <Ariadne> https://lore.kernel.org/lkml/20220126043947.10058-1-ariadne@dereferenced.org/T/
2022-01-26 05:16:31 <ericonr> Ariadne: the reading in musl had been that POSIX didn't mandate that, I think
2022-01-26 05:18:14 <ericonr> s/musl/#musl/
2022-01-26 05:18:14 <alpine-meetbot> ericonr meant to say: Ariadne: the reading in #musl had been that POSIX didn't mandate that, I think
2022-01-26 06:34:22 <DylanVanAssche> I did a while ago a lot of work to revive the PBAP support in BlueZ (the Bluetooth thingy to sync your address book with your car for example)
2022-01-26 06:34:26 <DylanVanAssche> The backend I used is libebook library provided by the evolution-data-server. However, the evolution-data-server is in community and BlueZ fails to build on main because of it
2022-01-26 06:35:07 <DylanVanAssche> Moving evolution-data-server seems impossible as a lot of dependencies of must be moved. Any suggestions on how to proceed here? !29955
2022-01-26 08:23:05 <ddevault> can I get this patch merged? https://gitlab.alpinelinux.org/alpine/aports/-/merge_requests/29913
2022-01-26 08:23:24 <ddevault> I will send a follow-up with 0.11.5 but only once this lands
2022-01-26 08:44:40 <ktprograms> omni: I ran qutebrowser in gdb, and the segfault seems to be in Thread 46 "Chrome_InProcRe", 0x0000ffffec642ef8 in base::internal::PartitionBucket<true>::SlowPathAlloc(base::PartitionRoot<true>*, int, unsigned long, bool*) () from /usr/lib/libQt5WebEngineCore.so.5.
2022-01-26 08:45:45 <ktprograms> The full backtrace is https://tpaste.us/BgpW, seems too short to be correct, probably means some kind of memory corruption happened.
2022-01-26 08:49:54 <Habbie> iirc, alpine tried to move to openssl3 and eventually rolled that back. this week I got reports from two powerdns downstreams (fedora and ubuntu) that they had openssl 3-related build failures, so we fixed those. Ubuntu 22.04 will apparently ship with openssl 3, so I expect the open source landscape to adjust pretty quickly now
2022-01-26 08:52:30 <ikke> Does mariadb already support it? 
2022-01-26 08:52:51 <ikke> And we also ran into a bug with apk-tools
2022-01-26 09:05:11 <ddevault> thanks ncopa!
2022-01-26 09:05:16 <ncopa> ktprograms: might be that it runs out of thread stack space
2022-01-26 09:05:19 <ncopa> ddevault: np
2022-01-26 09:06:15 <ktprograms> ncopa: This is with it just trying to open start.duckduckgo.com
2022-01-26 09:07:17 <ddevault> fucking gitlab
2022-01-26 09:07:25 <ddevault> ncopa: can you apply my 0.11.5 patch from the aports list instead of gitlab?
2022-01-26 09:07:46 <ddevault> https://lists.alpinelinux.org/~alpine/aports/patches/3905
2022-01-26 09:08:15 <ddevault> curl -s https://lists.alpinelinux.org/~alpine/aports/patches/3905/mbox | git am -3; git push
2022-01-26 10:08:10 <bratkartoffel> can someone please merge my openjdk (security at least) upgrades? Thanks! https://gitlab.alpinelinux.org/alpine/aports/-/merge_requests?scope=all&state=opened&author_username=bratkartoffel&search=openjdk
2022-01-26 11:53:44 <ikke> bratkartoffel: done
2022-01-26 11:53:48 <bratkartoffel> thanks ikke :)
2022-01-26 11:55:07 <Ariadne> patch v2: https://lore.kernel.org/lkml/20220126114447.25776-1-ariadne@dereferenced.org/T/
2022-01-26 11:55:46 <ikke> Hopefuly it lands
2022-01-26 11:56:21 <Ariadne> we should carry it in alpine regardless imo
2022-01-26 11:56:45 <Ariadne> there is no legitimate scenario where you execve(2) a program with argc < 1
2022-01-26 11:57:11 <Ariadne> debian codesearch is basically badly written tests & rootkit examples
2022-01-26 12:46:10 <Hello71> ktprograms: ime gdb's "corrupt stack" guess is almost never accurate
2022-01-26 12:46:34 <Hello71> possibly it was more accurate in mandatory frame pointer days
2022-01-26 12:49:21 <ktprograms> Hello71: That's true, and also thinking about it, for a non-main thread, a 3 entry long backtrace is believable. Any idea what could be the problem then?
2022-01-26 12:49:43 <Hello71> p $_siginfo, disas
2022-01-26 13:03:09 <ktprograms> Should i be using disas or x/i?
2022-01-26 13:04:29 <ktprograms> Output of p $_siginfo: https://tpaste.us/1EMB
2022-01-26 13:08:10 <Hello71> depends how long the function is
2022-01-26 13:08:45 <ktprograms> At least 200 lines of assembly gets output
2022-01-26 16:54:13 <ptrc> how does setup-disk handle a situation when alpine is booted from EFI, but getting installed to disk with MBR partition table?
2022-01-26 16:55:49 <ptrc> i'm not sure if i read that correctly, but in setup_grub it seems to only check if USE_EFI is not empty
2022-01-26 17:19:25 <ncopa> good question
2022-01-26 17:20:12 <ncopa> actually, i think if the disk is already partitioned, it will not try to install bootloader, but im not sure
2022-01-26 18:28:01 <ptrc> tried that on a vm, it looks like it tries installing grub-efi, fails with "grub-install: error: /mnt/boot/efi doesn't look like an EFI partition"
2022-01-26 18:31:33 <ptrc> it works fine when erasing the whole disk, because it uses gpt in that case; the issue is with installing alongside other OSes
2022-01-26 18:50:42 <Ariadne> ncopa: any objection to adding gawk as a makedepends for the kernel?  busybox awk is very slow
2022-01-26 18:51:10 <minimal> I thought Hello71 was recommending mawk?
2022-01-26 18:51:23 <Ariadne> gawk drops in without having to do anything
2022-01-26 18:51:30 <Hello71> but mawk is faster
2022-01-26 18:51:46 <Ariadne> i mean both are orders of magnitude faster than busybox awk
2022-01-26 18:52:27 <Hello71> i have explained this several times including at https://gitlab.alpinelinux.org/alpine/aports/-/issues/13442#note_208013 and preceding comments
2022-01-26 18:54:07 <Ariadne> i could try to upstream the objdump_reformat.awk stuff too
2022-01-26 18:55:39 <Hello71> busybox awk will still be slow, but then i think it will not be 100x slower but instead only, say, 5x slower. mawk is still 3x faster than gawk though, for this script
2022-01-26 18:55:58 <Hello71> interestingly perl is roughly the same speed as mawk
2022-01-26 18:56:39 <Hello71> i suspect that regular users are not really supposed to enable CONFIG_X86_DECODER_SELFTEST
2022-01-26 18:59:42 <Ariadne> probably not, no :)
2022-01-26 19:05:59 <psykose> linux-lts already uses mawk does it not
2022-01-26 19:06:19 <psykose> think that was patched a few days ago
2022-01-26 19:07:15 <psykose> ah, it's in makedepends but not passed as AWK= , i think that was needed
2022-01-26 19:07:16 <Ariadne> it does not actually use it, it needs AWK=mawk i guess
2022-01-26 19:07:22 <psykose> yes
2022-01-26 19:07:40 <psykose> well, simple enough fix
2022-01-26 19:08:16 <psykose> ncopa already intended to add it with that commit so i think it's fine
2022-01-26 19:14:14 <Ariadne> i was just not aware of the mawk change
2022-01-26 19:14:24 <Ariadne> i've been building a lot of kernels lately
2022-01-26 19:14:40 <Ariadne> you see, i have a thesis that pkexec is not the only vulnerable SUID program in alpine for this issue
2022-01-26 19:14:45 <Ariadne> so, i would like to fix them all at once
2022-01-26 19:15:20 <Ariadne> and so installing gawk was a "improve kernel build times with this one weird trick" thing :)
2022-01-26 19:16:03 <psykose> mhm
2022-01-26 19:16:06 <psykose> makes sense :)
2022-01-26 19:16:12 <psykose> anyway, +1 for the nmu from me :p
2022-01-26 19:16:32 <psykose> also afaik it was outlined the majority of the speedup is on x86 as well
2022-01-26 19:16:45 <psykose> the other arches don't do nearly as much awk hell
2022-01-26 19:21:49 <jvoisin> https://dustri.org/b/grsecurity-versus-cve-2021-4034.html ← or just forward-port some grsec-fu :P
2022-01-26 19:34:18 <Ariadne> i mean, i would prefer to deny exec on argc < 1, which seems a very grsec way of doing it
2022-01-26 19:37:17 <ikke> I suppose grsec cares less about breaking userspace
2022-01-26 19:38:56 <Ariadne> i don't think i care about breaking userspace for this
2022-01-26 19:38:59 <skarnet> I don't understand what legit userspace uses argc==0
2022-01-26 19:39:04 <Ariadne> ^
2022-01-26 19:39:23 <ikke> Me neither
2022-01-26 19:39:44 <Ariadne> well, you see, it is allowed in theory, so we must allow it
2022-01-26 19:39:46 <Ariadne> or something
2022-01-26 19:40:15 <skarnet> which is a valid pov but then at least the libc should crash in __libc_start_main
2022-01-26 19:40:37 <Ariadne> dalias seems to agree with that for SUID case at least
2022-01-26 19:40:38 <skarnet> and my opinion is that it's better to fail execve() than crash the new process after a successful execve()
2022-01-26 19:40:41 <Ariadne> yes
2022-01-26 19:41:00 <Ariadne> i don't really know how i feel about this argc == 1 {"", NULL} idea
2022-01-26 19:41:10 <Ariadne> it does "fix" the problem, but only some of the time
2022-01-26 19:41:14 <ikke> hyrums law
2022-01-26 19:41:19 <skarnet> argc= what?
2022-01-26 19:41:31 <skarnet> empty argv[0] is a totally different problem
2022-01-26 19:41:40 <skarnet> from NULL argv[0]
2022-01-26 19:41:50 <Ariadne> yes, the idea is to turn argc == 0 scenario into argc == 1 {"", NULL}
2022-01-26 19:42:11 <skarnet> that doesn't sound right at all
2022-01-26 19:42:27 <Ariadne> feel free to say as such on LKML :)
2022-01-26 19:42:43 <skarnet> I'm not getting paid enough to post on lkml
2022-01-26 19:42:48 <Ariadne> for the pkexec case, it does mitigate it
2022-01-26 19:42:57 <Ariadne> but i'm more interested in taking away the gadget forever
2022-01-26 19:43:12 <Ariadne> like, more umm, enlightened kernels have done
2022-01-26 19:43:20 <Ariadne> 
ACTION points at OpenBSD as an example, also QNX

2022-01-26 19:43:30 <skarnet> I'm more interested in my programs not crashing because of a mitigation for a fucking noobish pkexec idiocy
2022-01-26 19:43:44 <skarnet> because assuming argv[0] is not empty is valid
2022-01-26 19:44:09 <skarnet> and if Linux can now return empty argv[0]s then I'm not happy
2022-01-26 19:44:11 <Ariadne> yes, i agree with you that it should just be denied -EINVAL
2022-01-26 19:44:50 <Ariadne> POSIX says argv[0] should be present
2022-01-26 19:45:14 <skarnet> even when argc == 0 ?
2022-01-26 19:45:38 <Ariadne> POSIX does not discuss argc == 0
2022-01-26 19:45:40 <Ariadne> that is the problem
2022-01-26 19:45:57 <Ariadne> personally i think argc < 1 is not a valid execution state
2022-01-26 19:46:08 <Ariadne> i have not seen any proof that it should be considered one
2022-01-26 19:46:45 <skarnet> I agree with that
2022-01-26 19:47:09 <Ariadne> dalias does not :(
2022-01-26 19:47:59 <ikke> Just because POSIX is not explicit about it?
2022-01-26 19:48:02 <skarnet> even if one is for Linux being able to push empty stack of arguments, the libc should give something valid to main()
2022-01-26 19:48:45 <skarnet> never have I heard anywhere that main(0, {NULL}) was supported
2022-01-26 19:49:03 <Ariadne> lets ask dalias in those terms
2022-01-26 19:53:30 <ikke> https://xkcd.com/1172/ also comes to mind :P
2022-01-26 19:53:32 <alpine-meetbot> Workflow | Alt-text: There are probably children out there holding down spacebar to stay warm in the winter! YOUR UPDATE MURDERS CHILDREN.
2022-01-26 22:52:46 <omni> don't think I'll be able to look at this again before later tomorrow !29061
2022-01-27 00:13:17 <Ariadne> ncopa: i want to integrate https://lore.kernel.org/lkml/20220127000724.15106-1-ariadne@dereferenced.org/T/#u tomorrow, but after updating GCC snapshot
2022-01-27 00:18:20 <Ariadne> Hello71: i see you on github :P
2022-01-27 00:18:27 <Hello71> mhm
2022-01-27 00:19:27 <Hello71> Ariadne: https://gitlab.alpinelinux.org/ariadne/alpine-gcc-patches/-/merge_requests/9 and https://gitlab.alpinelinux.org/alpine/aports/-/merge_requests/26078 for gcc
2022-01-27 00:19:54 <Ariadne> yes, also nmeum has some things to change
2022-01-27 00:19:59 <Ariadne> i'll sort it tonight
2022-01-27 00:28:14 <psykose> if you want to run something post-upgrade and post-install, do you just duplicate the script
2022-01-27 00:28:28 <Ariadne> yes
2022-01-27 00:28:41 <psykose> thanks
2022-01-27 00:31:47 <Hello71> technically i think abuild might accept a symlink
2022-01-27 00:31:58 <Hello71> but i think it will be duplicated in the built package
2022-01-27 00:34:06 <psykose> eh, it's one line anyway
2022-01-27 00:35:37 <Hello71> mmhmm
2022-01-27 00:43:09 <psykose> also fix your highway mr so i can actually merge it
2022-01-27 00:46:29 <Ariadne> . o O (what if Hello71 could merge his own MRs)
2022-01-27 00:46:38 <Ariadne> thinking intensifies
2022-01-27 00:48:38 <Hello71> oh i accidentally archived that email, forgot about it
2022-01-27 01:24:43 <psykose> more friendos
2022-01-27 08:17:04 <bolzerr> Hi, i want to use alpine as a docker host system All good but i fail to setup a macvlan, the kernel modules are missing: veth and macvlan. I am using the virtialisation distro with this kernel: ALPINE1 5.15.16-0-virt. Can someone please help me how to either obtain the modules or compile them? 
2022-01-27 08:25:01 <bolzerr> Although it seems to be configured as a module: https://gitlab.alpinelinux.org/alpine/aports/-/blob/master/main/linux-lts/config-virt.x86_64#L2261
2022-01-27 08:26:06 <ktprograms> ncopa: ^ ? It only seems to be built for the rpi kernels: https://pkgs.alpinelinux.org/contents?file=macvlan.ko&path=&name=&branch=edge
2022-01-27 08:29:03 <bolzerr> You are right. I was assuming the gitlab refelcts the current builds. 
2022-01-27 08:29:34 <ktprograms> bolzerr: It _should_.
2022-01-27 08:30:24 <ktprograms> in the build log, macvlan seems to be properly built and installed.
2022-01-27 08:31:35 <bolzerr> Strange
2022-01-27 08:33:50 <bolzerr> i already tried to build some modules, but i ran into an erro when running make: ln: unrecognized option: target-directory=. BusyBox v1.34.1 (2021-11-23 00:57:35 UTC) multi-call binary.
2022-01-27 08:34:19 <bolzerr> I am new to alpine but used to build on debian/ ubuntu. 
2022-01-27 08:35:00 <bolzerr> Is this something usual and can be easily fixed? It is breaking because of the muissing option for ln
2022-01-27 08:35:25 <ktprograms> bolzerr: Install coreutils, it will replace the busybox version
2022-01-27 08:35:51 <ktprograms> You can see what package you need to install in order to replace/get a command with `apk search cmd:<cmd>`
2022-01-27 08:36:22 <bolzerr> Thank you, very appreciatewd. Will try it now
2022-01-27 08:36:59 <bolzerr> And it did the trick :-)
2022-01-27 08:49:00 <bolzerr> Weird, after i reboot i was loading the veth and macvlan modules using modprobe. I did a minor kernel update to the latest version.
2022-01-27 08:51:13 <ktprograms> What do you mean?
2022-01-27 08:51:57 <bolzerr> macvlan is working now - without the need to compile a module. before the reboot modprobe was telling me the macvlan would not exist
2022-01-27 08:52:20 <ktprograms> Hmm interesting
2022-01-27 08:53:11 <ktprograms> Wierd because CONFIG_MACVLAN has been in the config for 2 years according to the blame
2022-01-27 08:54:26 <bolzerr> Maybe i i did a mistake when trying before
2022-01-27 08:54:36 <bolzerr> sorry if i caused confusion
2022-01-27 09:11:37 <bolzerr> One more question, when i try to build "media_build" for my dvb card i am running into following issue: error: 'const struct regmap_config' has no member named 'reg_bits'. Is there a build dependency i need to add? Could not find anything using apk search
2022-01-27 09:27:23 <ncopa> morning
2022-01-27 09:27:25 <ncopa> ncopa-desktop:~$ doas modprobe macvlan && lsmod | grep macvlan
2022-01-27 09:27:25 <ncopa> macvlan                28672  0
2022-01-27 09:27:39 <omni> moin
2022-01-27 09:27:42 <PJ[m]> mornin
2022-01-27 10:37:21 <omni> ktprograms: how about the latest qt5-qtwebengine then?
2022-01-27 10:59:26 <bolzerr> macvlan was resolved - i guess i misstyped or something. It is working fine now. Any idea about this "regmap" issue?
2022-01-27 11:05:56 <ktprograms> omni: Yes version 5.15.3_git20220121-r0. The full valgrind log is https://tpaste.us/r6V1. Seems to be strlen/memcpy that's doing the invalid reads.
2022-01-27 11:06:58 <ncopa> Ariadne: do you have a `git format-patch -1 --stdout | tpaste` handy for https://lore.kernel.org/lkml/20220127000724.15106-1-ariadne@dereferenced.org/T/#u ?
2022-01-27 11:10:56 <ncopa> nm, i got it
2022-01-27 12:00:28 <aparcar> Ariadne: Am I missing something about the APK provides flag? If I set a package to "provide" itself without the ABI suffix, it conflicts with itself
2022-01-27 12:01:01 <aparcar> https://paste.sr.ht/~aparcar/9858c7f164670c63431e2e118cfad932c56dd233
2022-01-27 12:37:51 <aparcar> Ariadne: okay the issue was most likely that the package is provided without a version and therefore considered a "virtual" package
2022-01-27 12:59:32 <Hello71> ktprograms: https://pkgs.alpinelinux.org/contents?file=macvlan.ko*&path=&name=&branch=edge
2022-01-27 13:00:12 <ktprograms> Hello71: And...?
2022-01-27 13:00:34 <ktprograms> Oh you mean search for *.gz?
2022-01-27 13:00:44 <Hello71> yes
2022-01-27 13:01:25 <ktprograms> Right thanks
2022-01-27 13:03:21 <minimal> ktprograms: I'm assuming he updated his kernel and had not yet rebooted and that's why he couldn't load the macvlan module
2022-01-27 13:03:36 <ktprograms> minimal: Possible
2022-01-27 13:03:51 <ktprograms> And that's why you always reboot after an upgrade
2022-01-27 13:15:49 <omni> ktprograms: I feel it's a bit beyond me, PureTryOut?
2022-01-27 13:16:14 <PureTryOut> what?
2022-01-27 13:17:29 <omni> PureTryOut: the paste URL a bit earlier
2022-01-27 13:24:02 <ktprograms> PureTryOut: https://tpaste.us/r6V1 qutebrowser segfaulting just loading start.duckduckgo.com or anything else
2022-01-27 13:26:31 <omni> on aarch64, I only run on x86_64 myself
2022-01-27 13:29:00 <PureTryOut> Uhm, yeah that's beyond me as well. It would help if you had debug symbols of re2 installed
2022-01-27 13:29:20 <ktprograms> Which package is re2 in?
2022-01-27 13:30:13 <PureTryOut> re2 😛
2022-01-27 13:30:15 <PureTryOut> That is the package
2022-01-27 13:30:45 <ktprograms> PureTryOut: Ah I see. So look slike I need to build it with a -dbg subpackage myself?
2022-01-27 13:32:21 <PureTryOut> Yeah, it'd give more information at least
2022-01-27 13:32:32 <Hello71> i doubt it's related to re2
2022-01-27 13:33:04 <PureTryOut> probably not but it's causing a lot of ???
2022-01-27 13:33:07 <Hello71> it might help to increase the stack trace Kenneth
2022-01-27 13:33:10 <Hello71> length
2022-01-27 13:34:21 <skarnet> I told Kenneth several times to increase the stack trace but he just won't listen
2022-01-27 13:35:36 <Hello71> i usually ignore uninitialized value valgrind warnings. i think they're false positives but not sure
2022-01-27 13:45:41 <ktprograms> I built and installed re2-dbg, but it doesn't seem to be showing in the valgrind log stack traces. Anyway I think it's more likely to be the Invalid reads, which seem to be happening in strlen/memcpy
2022-01-27 13:48:35 <Hello71> unless the build is 100% reproducible you also need to install your new re2
2022-01-27 13:48:56 <ktprograms> Hello71: k
2022-01-27 13:50:23 <Hello71> did you try increasing the stack trace size
2022-01-27 13:50:31 <ktprograms> Hello71: How?
2022-01-27 13:50:34 <Hello71> dunno
2022-01-27 13:51:01 <ktprograms> The log with re2 debug symbols installed is https://tpaste.us/mqJx
2022-01-27 13:56:34 <ktprograms> gtg, you can message about any ideas/solutions re this problem and I'll read it in the irclogs. Thanks!
2022-01-27 13:57:14 <ncopa> that looks like a use-after-free
2022-01-27 14:03:38 <Hello71> mmhmm
2022-01-27 14:04:26 <Hello71> but like i said, there is no useful information about what was used after free except that it's a string
2022-01-27 14:42:14 <wyk72> hello everyone
2022-01-27 14:42:26 <wyk72> need some help with init
2022-01-27 14:42:40 <wyk72> I have a some machines that runs from ram
2022-01-27 14:42:51 <wyk72> I use to run the update-kernel script
2022-01-27 14:42:58 <wyk72> in the latest 3.15 Alpine
2022-01-27 14:43:07 <wyk72> it always fails at init
2022-01-27 14:43:15 <wyk72> with "mounting boot media: failed"
2022-01-27 14:43:35 <wyk72> the USB keyboard also gets locked
2022-01-27 14:44:02 <wyk72> I'm stuck at this and I have no clue because no clue is given
2022-01-27 14:44:31 <wyk72> maybe it has something to do with zpaq compression? squashfs ? modloop ?
2022-01-27 14:44:36 <wyk72> can you point me somewhere ?
2022-01-27 14:44:47 <wyk72> zstd compression
2022-01-27 14:45:19 <wyk72> kernel boots and fails to find /
2022-01-27 14:45:24 <wyk72> to mount
2022-01-27 14:51:48 <wyk72> maybe it fails to mount .modloop?
2022-01-27 14:51:58 <wyk72> there is a debug somewhere?
2022-01-27 14:55:29 <ncopa> hi, i think this is due to the change we did to /dev/usbdisk symlink
2022-01-27 14:56:17 <ncopa> previously it would create a symlink /dev/usbdisk which would be the first disk on usb bus with fat partition
2022-01-27 14:56:32 <ncopa> that got stored in your /etc/fstab i guess
2022-01-27 14:59:45 <wyk72> this is not an usb disk: it's a sd disk (/dev/sda)
2022-01-27 15:04:17 <ncopa> how does your /etc/fstab look like?
2022-01-27 15:05:07 <ncopa> how much RAM do you have?
2022-01-27 15:05:26 <wyk72> 20Gb
2022-01-27 15:05:49 <wyk72> the initram gets built correctly
2022-01-27 15:06:08 <wyk72> I'll check fstab
2022-01-27 15:07:20 <wyk72> it's in the overlay I guess
2022-01-27 15:07:39 <wyk72> since I can't boot the system
2022-01-27 15:10:26 <wyk72> /dev/sda1 /media/sda1 vfat ro,relatime,fmask=0022,dmask=0022,errors=remount-ro 0 0
2022-01-27 15:10:33 <wyk72> that's my /etc/fstab
2022-01-27 15:10:56 <wyk72> inside the overlay file
2022-01-27 15:11:31 <wyk72> I don't even know if it gets checked/mounted correctly
2022-01-27 15:11:49 <wyk72> it's a fat32 partition
2022-01-27 15:13:42 <wyk72> those are the grub entries (it's an EFI boot)
2022-01-27 15:13:45 <wyk72> initrd  /boot/intel-ucode.img
2022-01-27 15:13:45 <wyk72> linux   /boot/vmlinuz-lts modules=loop,squashfs,sd-mod,usb-storage iommu=pt intel_iommu=on amd_iommu=on  mitigations=off  pci=noaer
2022-01-27 15:13:45 <wyk72> initrd  /boot/initramfs-lts
2022-01-27 15:13:45 <wyk72> menuentry "Linux lts" {
2022-01-27 15:13:45 <wyk72> }
2022-01-27 15:14:12 <wyk72> drwxr-xr-x 6 root root 4.0K Jan 27 15:10 ./
2022-01-27 15:14:12 <wyk72> -rwxr-xr-x 1 root root 228K Jan 27 15:00 config-lts*
2022-01-27 15:14:13 <wyk72> -rwxr-xr-x 1 root root 4.1M Jan 27 15:00 System.map-lts*
2022-01-27 15:14:13 <wyk72> drwxr-xr-x 6 root root 4.0K Jan  1  1970 ../
2022-01-27 15:14:13 <wyk72> drwxr-xr-x 3 root root 4.0K Jan 22 10:37 dtbs/
2022-01-27 15:14:14 <wyk72> drwxr-xr-x 2 root root 4.0K Jan 22 10:22 dtbs-lts/
2022-01-27 15:14:15 <wyk72> drwxr-xr-x 2 root root 4.0K Jan 22 10:22 grub/
2022-01-27 15:14:17 <wyk72> -rwxr-xr-x 1 root root  43M Jan 27 15:20 initramfs-lts*
2022-01-27 15:14:19 <wyk72> -rwxr-xr-x 1 root root  43M Jan 27 15:00 initramfs-lts.old*
2022-01-27 15:14:21 <wyk72> -rwxr-xr-x 1 root root 4.6M Jan 22 10:46 intel-ucode.img*
2022-01-27 15:14:23 <wyk72> -rwxr-xr-x 1 root root 119M Jan 27 15:00 modloop-lts*
2022-01-27 15:14:25 <wyk72> drwxr-xr-x 2 root root 4.0K Jan 22 10:22 syslinux/
2022-01-27 15:14:27 <wyk72> -rwxr-xr-x 1 root root 7.5M Jan 27 15:00 vmlinuz-lts*
2022-01-27 15:14:29 <wyk72> root@imm-install-8GB:/mnt/boot#
2022-01-27 15:14:31 <wyk72> that's what I have in the FAT32 (sda1) partition
2022-01-27 15:16:14 <minimal> wyk72: which partition is your EFI ESP partition?
2022-01-27 15:17:21 <wyk72> which is the EFI ESP itself
2022-01-27 15:17:53 <minimal> wky72: was that an "ls -l /boot" or "ls -l /" ?
2022-01-27 15:19:18 <wyk72> ls -l /boot
2022-01-27 15:19:46 <wyk72> total 13M
2022-01-27 15:19:46 <wyk72> drwxr-xr-x 27 root root 4.0K Jan  6 16:42 ../
2022-01-27 15:19:46 <wyk72> drwxr-xr-x  6 root root 4.0K Jan  1  1970 ./
2022-01-27 15:19:46 <wyk72> -rwxr-xr-x  1 root root   30 Jan 22 10:22 .alpine-release*
2022-01-27 15:19:46 <wyk72> drwxr-xr-x  3 root root 4.0K Jan 22 10:22 apks/
2022-01-27 15:19:48 <wyk72> drwxr-xr-x  6 root root 4.0K Jan 27 15:10 boot/
2022-01-27 15:19:49 <wyk72> drwxr-xr-x  2 root root  20K Jan 27 14:16 cache/
2022-01-27 15:19:51 <wyk72> drwxr-xr-x  3 root root 4.0K Jan 22 10:22 efi/
2022-01-27 15:19:53 <wyk72> -r-xr-xr-x  1 root root 114K Jan 22 10:22 ldlinux.c32*
2022-01-27 15:19:55 <wyk72> -r-xr-xr-x  1 root root  68K Jan 22 10:22 ldlinux.sys*
2022-01-27 15:19:57 <wyk72> -rwxr-xr-x  1 root root  13M Jan 27 15:02 optplx7010.apkovl.tar.gz*
2022-01-27 15:19:59 <wyk72> root@imm-install-8GB:/mnt#
2022-01-27 15:20:01 <wyk72> that's what I have in /
2022-01-27 15:21:53 <wyk72> maybe it's the FTDIR parameter I'm missing?
2022-01-27 15:22:06 <wyk72> FDTDIR
2022-01-27 15:25:43 <wyk72> the machine used to boot correctly
2022-01-27 15:25:55 <wyk72> it hanged when I launched the update-kernel script
2022-01-27 15:26:05 <wyk72> machine built with setup-bootable script
2022-01-27 15:26:38 <wyk72> setup-bootable http://<ALPINE ISO> /dev/sda
2022-01-27 15:26:51 <wyk72> used to work perfectly
2022-01-27 15:26:55 <wyk72> with LBU set and all
2022-01-27 15:27:05 <wyk72> broke after update-kernel
2022-01-27 15:27:22 <wyk72> which was already launched and working 
2022-01-27 15:27:45 <wyk72> with kernel 5.15.15-0
2022-01-27 15:28:01 <wyk72> updated to 5.15.16-0 and init broke 
2022-01-27 15:34:23 <wyk72> I wonder why the recovery initram-fs shell is not working too
2022-01-27 15:34:27 <wyk72> i.e. no keyboard
2022-01-27 15:34:42 <wyk72> modloop not loaded ?
2022-01-27 15:35:05 <wyk72> I can see the kernel booting by removing the "quiet" line in grub-cfg
2022-01-27 15:35:40 <wyk72> video loaded, drm sets video mode
2022-01-27 15:35:56 <wyk72> then at second 6.5 : Mounting boot media:failed.
2022-01-27 15:36:06 <minimal> wyk72: are said you are using Grub to boot? the ldlinux.c32 and ldlinux.sys files in / look like part of Syslinux
2022-01-27 15:37:04 <wyk72> that's how Alpine builds it with the setup-bootable script
2022-01-27 15:37:05 <minimal> if the keyboard is not working I'm guessing its because the USB drivers are not loaded
2022-01-27 15:37:14 <wyk72> both grub & syslinux gets installed
2022-01-27 15:38:09 <wyk72> I suspect as much, but no clue is given on what is happening
2022-01-27 15:39:03 <wyk72> I can try with an ancient ps/2 keyboard
2022-01-27 15:39:19 <wyk72> this pc has a serial port too, but attaching a null-modem cable seems like 1980 to me
2022-01-27 15:43:51 <minimal> wyk72: is the Grub menu keyboard accessible? you could modify the cmdline modules entry to load usb drivers
2022-01-27 15:44:00 <wyk72> yes it is
2022-01-27 15:44:57 <wyk72> minimal: how can I accomplish this ?
2022-01-27 15:45:07 <wyk72> the grub line reads
2022-01-27 15:45:25 <wyk72> linux   /boot/vmlinuz-lts modules=loop,squashfs,sd-mod,usb-storage iommu=pt intel_iommu=on amd_iommu=on  mitigations=off  pci=noaer
2022-01-27 15:46:31 <minimal> for the "modules=" part append ",usbhid,ohci_hcd" (replace "ohci" with "uhci" if its an Intel chipset)
2022-01-27 15:47:16 <wyk72> oops
2022-01-27 15:47:22 <wyk72> I see the mess I've made
2022-01-27 15:47:43 <wyk72> I removed usb from /etc/mkinitfs/mkinitfs.conf
2022-01-27 15:48:09 <wyk72> that's why usb drivers etc  do not get into the image
2022-01-27 15:48:37 <minimal> ah, in that case add the usbhid stuff to cmdline will have to effect - the modules need to be in inintramfs in the 1st place
2022-01-27 15:49:10 <wyk72> I have to put them in manually 
2022-01-27 15:49:21 <minimal> does that mean usb-storage is also missing from initramfs? (are you booting off USB?)
2022-01-27 15:49:23 <wyk72> and do a depmod inside
2022-01-27 15:49:43 <wyk72> I am booting from a SATA (AHCI) disk
2022-01-27 15:49:45 <wyk72> not usb
2022-01-27 15:49:51 <wyk72> used to work fine btw
2022-01-27 15:49:59 <wyk72> broke after update-jkernel�
2022-01-27 15:50:48 <minimal> a kernel update would trigger a initramfs rebuild - so if you remove usb from mkinitfs.conf but never actually rebuilt it until then that would explain change in keyboard behaviour
2022-01-27 15:52:26 <wyk72> yeah I made a mess
2022-01-27 15:53:06 <wyk72> I've unrolled the initram into a directory, will add the files and do a depmod
2022-01-27 15:55:43 <wyk72> Oh I solved the mistery
2022-01-27 15:55:50 <wyk72> nwo I get it
2022-01-27 15:55:52 <wyk72> now
2022-01-27 15:56:16 <omni> !30053
2022-01-27 15:56:35 <wyk72> there is an inconsistency in the features.d
2022-01-27 15:56:49 <wyk72> for reasons unknown to me
2022-01-27 15:56:55 <wyk72> root@imm-install-8GB:/etc/mkinitfs/features.d# cat usb.modules
2022-01-27 15:56:55 <wyk72> kernel/drivers/hid/usbhid
2022-01-27 15:56:55 <wyk72> kernel/drivers/usb/storage
2022-01-27 15:56:55 <wyk72> kernel/drivers/usb/host
2022-01-27 15:56:55 <wyk72> kernel/drivers/hid/hid-generic.ko*
2022-01-27 15:56:56 <wyk72> kernel/drivers/hid/hid-cherry.ko*
2022-01-27 15:56:58 <wyk72> kernel/drivers/hid/hid-apple.ko*
2022-01-27 15:57:00 <wyk72> kernel/fs/fat
2022-01-27 15:57:02 <wyk72> kernel/fs/nls
2022-01-27 15:57:33 <wyk72> those are the modules that mkinitfs puts inside initram-lts with "feaure usb"
2022-01-27 15:57:49 <wyk72> kernel/fs/fat
2022-01-27 15:57:50 <wyk72> is there
2022-01-27 15:58:22 <wyk72> so IF u don' put usb into mkinitfs.conf
2022-01-27 15:58:27 <wyk72> vfat does not make into the image
2022-01-27 15:58:52 <wyk72> so when booting, ALpine fails to load vfat
2022-01-27 15:58:58 <wyk72> and can't find /
2022-01-27 15:59:16 <wyk72> damn
2022-01-27 15:59:28 <wyk72> isn't it bad for all UEFI stuff?
2022-01-27 15:59:36 <wyk72> ESP is almost always vfat
2022-01-27 16:00:37 <wyk72> what's the point in adding kernel/fs/fat into USB features ??
2022-01-27 16:20:30 <minimal> wyk72: it would make more sense for vfat to be in base.modules
2022-01-27 16:21:08 <minimal> then again I create/tailor my own feature files :-)
2022-01-27 16:26:39 <ncopa> actually good point. fat is needed for efi
2022-01-27 16:27:11 <ncopa> makes sense to have it in base
2022-01-27 16:34:59 <minimal> ncopa: also I think there have been several issues recently regarding NVME booting, I don't remember if nvme is in base.modules but it should be
2022-01-27 16:34:59 <wyk72> I managed to rebuild the initramfs-lts by hand into a dir
2022-01-27 16:35:06 <wyk72> added the files
2022-01-27 16:35:15 <wyk72> the keyboard now works
2022-01-27 16:35:23 <wyk72> but / does not mount
2022-01-27 16:35:36 <wyk72> it fails with "codepage cp437 not found"
2022-01-27 16:36:18 <wyk72> nvme is not in base.modules
2022-01-27 16:36:23 <minimal> ncopa: btw !29869
2022-01-27 16:36:56 <minimal> wyk72: hmm, one of the features files has a nls wildcard that should bring in codepage files
2022-01-27 16:37:27 <wyk72> I forgot nls
2022-01-27 16:37:36 <wyk72> stupid me
2022-01-27 16:37:57 <Hello71> i think usb keyboard has been broken for several years
2022-01-27 16:39:04 <Ariadne> minimal: i would prefer we just rm the binaries in package()
2022-01-27 16:39:14 <Hello71> #9588, #9089
2022-01-27 16:39:22 <Ariadne> minimal: i don't want to maintain superfluous diffs with upstream
2022-01-27 16:40:11 <minimal> Ariadne: ok, it just seems "strange" to have a halfworking openrc init ratheer than either removed or fully functional
2022-01-27 16:40:33 <Ariadne> i agree.  can you update it to just kill those binaries in package() ?
2022-01-27 16:40:37 <minimal> wyk52: nls files are brought in by usb.modules
2022-01-27 16:41:18 <minimal> Ariadne: oh, "rm", sure... thought it made more sense not to build rather than delete after.....but ok, I'll revise the MR for that
2022-01-27 16:41:25 <Ariadne> as the long term plan is to switch to s6, i don't really think we should adopt more openrc things
2022-01-27 16:41:33 <Ariadne> so +1 to rm the binaries
2022-01-27 16:41:46 <Ariadne> (also a lot of openrc things are... half-baked)
2022-01-27 16:41:55 <Ariadne> (looking at you, supervise-daemon(8))
2022-01-27 16:43:16 <wyk72> it finally woirked by adding nls 
2022-01-27 16:43:22 <wyk72> omg
2022-01-27 16:43:37 <wyk72> such a pain I've inflicted to myself
2022-01-27 16:47:58 <wyk72> painful porocedure was: booting from another Alpine USB key, unroll initramfs-lts into a dir, rsync the right dirs from /lib/modules/kernel/<...>, do a depmod --basedir=., recomrpess the initramfs & replace the one
2022-01-27 16:48:26 <wyk72> morale: NEVER remove usb feature from mkinitfs.conf
2022-01-27 16:48:31 <wyk72> never ever
2022-01-27 16:50:33 <Ariadne> ;0
2022-01-27 16:50:36 <Ariadne> er
2022-01-27 16:50:37 <Ariadne> :)
2022-01-27 16:59:58 <donoban> ncopa: I'm just testing 5.15.17-r0 on a Beelink SoC #13439 and sound doesn't work "aplay: device_list:274: no soundcards found..."
2022-01-27 17:00:37 <donoban> I can try to find a working config, I've found one based on linux-edge. Do you want that I upload a working diff if works?
2022-01-27 17:09:24 <Ariadne> sure
2022-01-27 17:50:23 <donoban> ok
2022-01-27 17:52:36 <donoban> I'm getting this errors when upgrading http://ix.io/3NHr , mirror problem?
2022-01-27 17:53:43 <psykose> yes
2022-01-27 17:53:48 <psykose> only -r7 is present on the mirror
2022-01-27 17:53:57 <psykose> for -libs for instance
2022-01-27 17:54:05 <psykose> i think they imported the new apkindex but not the rest
2022-01-27 17:54:21 <donoban> it started some hours ago
2022-01-27 17:54:48 <donoban> should be already fixed if it's just waiting for sync?
2022-01-27 17:54:58 <psykose> i can't tell you why the mirror files are behind
2022-01-27 17:55:41 <donoban> ok, I will wait a little more and change it if stills failing
2022-01-27 20:10:50 <ikke> psykose: oh, one of the perks of being a developer is that you can request build containers for the various arches we have
2022-01-27 20:11:07 <psykose> how does that work
2022-01-27 20:12:24 <psykose> but yeah, think getting an rv64 one would be nice to test things on preemptively myself too
2022-01-27 20:12:27 <ikke> You'd ask me or clandmeter for a container
2022-01-27 20:12:37 <ikke> and we setup something where you can ssh into
2022-01-27 20:12:47 <psykose> sure
2022-01-27 20:13:11 <psykose> same key as on gitlab: ssh-ed25519 AAAAC3NzaC1lZDI1NTE5AAAAIJ1GNaR5BlDlMvqtJYJPRZIvyBiBwNpRp9d2p8ES3ErY
2022-01-27 20:13:24 <ikke> 👍
2022-01-27 20:14:05 <ikke> Do you prefer this username or your gitlab username?
2022-01-27 20:14:21 <ikke> Well, it's more like as in the container name
2022-01-27 20:14:30 <psykose> this one
2022-01-27 20:14:32 <ikke> ok
2022-01-27 20:15:25 <psykose> thanks :)
2022-01-27 20:20:04 <ikke> psykose: fyi, on gitlab, you have an rsa key
2022-01-27 20:28:40 <psykose> ikke: are you sure? there is only one on the account and it's the same one
2022-01-27 20:29:12 <ikke> Oh, yes, my bad
2022-01-27 21:04:38 <omni> psykose: I was just about to ask if kit-ty-kate wanted to take over maintainership of opam! :D
2022-01-27 21:04:55 <omni> since they seem to work on it upstream
2022-01-27 21:05:32 <omni> I just took over ownership since it was orphaned, I had a lot of time on my hands and wanted to improve the overall ocaml situation in alpine
2022-01-27 21:06:24 <psykose> afaik she is an ocaml developer
2022-01-27 21:06:28 <psykose> sure, feel free to ask
2022-01-27 21:06:58 <omni> I meant, I was about to do that when I saw that it was already merged =)
2022-01-27 21:07:28 <omni> is she on IRC?
2022-01-27 21:07:34 <psykose> not to my knowledge
2022-01-27 21:08:28 <omni> can you DM someone on gitlab?
2022-01-27 21:08:34 <psykose> not sure..
2022-01-27 21:08:48 <psykose> there's always email
2022-01-27 21:08:55 <omni> true
2022-01-27 21:08:57 <psykose> or just ping in the same mr
2022-01-27 21:09:10 <omni> yeah, I'll start by asking there
2022-01-27 21:09:13 <psykose> or twitter, or or or
2022-01-27 21:17:57 <genkitj> Ariadne: are we waiting for s6-rc v1.0? Or are there any other utilities that require some work?
2022-01-27 21:18:25 <Ariadne> skarnet is reworking the user experience a little bit
2022-01-27 21:18:31 <Ariadne> we are waiting for that
2022-01-27 21:18:49 <Ariadne> we also need a story for making sure people's openrc stuff continues working
2022-01-27 21:46:00 <skarnet> oh, it will
2022-01-27 21:46:39 <skarnet> if openrc is installed, s6-rc will run it as one of its services
2022-01-27 21:47:00 <skarnet> not the early runlevels, of course, but the user ones
2022-01-27 21:47:24 <skarnet> so stuff that comes with -openrc scripts will run
2022-01-27 21:48:17 <psykose> what if.. we wrote s6 scripts in openrc syntax
2022-01-27 21:48:19 <psykose> unless..
2022-01-27 21:48:21 <psykose> 😳
2022-01-27 21:49:00 <skarnet> we want the opposite: openrc driven by s6, not s6 driven by openrc
2022-01-27 21:49:09 <skarnet> because openrc doesn't root a supervision tree in pid 1
2022-01-27 21:55:55 <MY-R> damn, today I found https://wiki.gentoo.org/wiki/S6
2022-01-27 21:56:07 <MY-R> who wrote all this!? :D
2022-01-27 21:58:30 <Hello71> https://wiki.gentoo.org/index.php?title=S6&action=history
2022-01-27 21:58:56 <psykose> guillermodh for most of it
2022-01-27 21:58:57 <psykose> :p
2022-01-27 21:59:20 <psykose> not updated since 2019
2022-01-27 22:04:34 <skarnet> I'm worried about him, haven't heard of him in a while and the gentoo skaware packages (which he maintains) are rotting though he's usually very reactive
2022-01-27 22:05:23 <psykose> has he been gone for years
2022-01-27 22:05:47 <Ermine> What about running openrc from s6-linux-init and then instructing openrc to interact with s6 (like described here: https://wiki.alpinelinux.org/wiki/Writing_Init_Scripts#Services_supervised_by_s6)
2022-01-27 22:06:47 <skarnet> Ermine: the "s6 integration" existing in openrc has a lot of problems
2022-01-27 22:07:35 <skarnet> but as a transitional step, it's definitely possible to boot on s6-l-i and run openrc in stage 2
2022-01-27 22:07:41 <skarnet> (that's what Adélie does)
2022-01-27 22:08:54 <MY-R> where alpine s6 wiki howto!? 
2022-01-27 22:09:00 <Ermine> In this case we have supervision tree rooted at pid 1. But, of course, this cannot be considered as permanent solution
2022-01-27 22:09:01 <skarnet> psykose: years, maybe not, but months, definitely
2022-01-27 22:09:35 <skarnet> Ermine: if the end goal is to replace openrc, keeping openrc isn't a permanent solution indeed :D
2022-01-27 22:09:50 <skarnet> MY-R: when we get to it, which will still be a while
2022-01-27 22:10:14 <psykose> ah
2022-01-27 22:10:17 <MY-R> :/
2022-01-27 22:12:10 <skarnet> MY-R: currently when you install the s6 package you get a supervision tree you can run with "service s6 start" (or add the s6 service to your main runlevel), it's not rooted in pid 1 but it's there, with a catch-all logger and all, so you can run your services there if you want
2022-01-27 22:16:42 <MY-R> skarnet: does it start early enough to catch all logs early enough?
2022-01-27 22:17:41 <MY-R> with openrc any logging daemon start so late that I see maybe few lines of logs from many services... it is annoying
2022-01-27 22:18:33 <MY-R> and since alpine started use those simpledrm weird stuff even my "dmesg" output and kernel logs are trimmed off
2022-01-27 22:21:06 <MY-R> thought systemd journal will be nice with logs but after seen that somehow it is not abble to catch simple error logs from daemon... then I will never look at this again
2022-01-27 22:22:11 <Ermine> to write to systemd-journald, you need to use special function from their library
2022-01-27 22:22:26 <skarnet> MY-R: it's a normal service so it will only catch the logs of whatever you put in the supervision tree, but it will catch everything from there
2022-01-27 22:22:31 <MY-R> I was trying system-joiurnal on different system
2022-01-27 22:23:34 <MY-R> stupid systemd journal couldnt log that Im trying start dns service and port was busy BEACUSE I didnt know that systemd by default start some own shit dns cache service....
2022-01-27 22:25:35 <MY-R> so ye, debuging systemd is like stop all services and run them one by one manualy to get more info which suposed to land in journald...
2022-01-27 22:26:56 <MY-R> ok, thanks skarnet!
2022-01-27 22:37:12 <Hello71> Ermine: no. the whole point of systemd-journald is to suck in whole system logs, klog, syslog, stdout, stderr, coredump, EVERYTHING must be consumed
2022-01-27 22:37:27 <Hello71> insert zalgo here
2022-01-27 22:37:42 <MY-R> Hello71: it stil doesnt print all logs :D
2022-01-27 22:38:20 <Hello71> vaguely sounds like you don't have permissions
2022-01-27 22:38:48 <MY-R> :D
2022-01-27 23:49:12 <minimal> MY-R: would setting CONFIG_EARLY_PRINTK in kernel config make any diff to the dmesg output?
2022-01-28 00:32:13 <MY-R> minimal: thanks, I will try it next time when compile kernel and if will work then create issue
2022-01-28 01:31:30 <ktprograms> Hello71: Any way to continue debugging the uaf? Also why doesn't my valgrind log have line numbers? Is it because qt5-qtwebengine was built with optimization?
2022-01-28 02:52:45 <boomanaiden154> Does anyone mind looking at !29824?
2022-01-28 03:44:21 <Hello71> ktprograms: increase --num-callers?
2022-01-28 03:44:47 <Hello71> or is it  --unw-stack-scan-frames
2022-01-28 03:44:49 <Hello71> that would explain 5
2022-01-28 03:57:54 <ktprograms> Hello71: https://tpaste.us/XKxj, with --track-origins=yes --num-callers=500 --unw-stack-scan-frames=32, but that still doesn't show line numbers in Qt5WebEngineCore
2022-01-28 07:26:05 <aparcar> how does alpine manage kernen modules with APK? I'm trying to add a hash (kernel magic) to the version field but it's not recognized, the apk version --check always fails
2022-01-28 07:26:11 <aparcar> is there a way to add a hash?
2022-01-28 07:30:21 <aparcar> Ariadne: ideas?
2022-01-28 11:37:01 <ikke> aparcar: I don't think the apk version scheme supports arbitry tokens
2022-01-28 11:37:33 <aparcar> ikke: all right, I'll try to figure something out
2022-01-28 11:37:55 <aparcar> how does alpine make sure kmods fit?
2022-01-28 11:38:24 <ikke> The modules are part of the kernel package
2022-01-28 11:39:46 <ikke> There is no strong binding between out-of-tree modules and the kernel
2022-01-28 11:40:53 <ikke> One consequence of this is that you loose the kernel modules for the current running kernel if you upgrade the kernel
2022-01-28 11:40:56 <aparcar> in alpine it's possible to upgrade the kernel right? How to make sure that all kmods are upgraded too?
2022-01-28 11:41:19 <aparcar> oh so that's an open issue?
2022-01-28 11:41:23 <aparcar> or is that by design?
2022-01-28 11:41:44 <ikke> More or less an open issue
2022-01-28 11:42:10 <ikke> There have been discussions about it, but did not lead to a satisfying solution (or at least, nothing was implemented)
2022-01-28 11:42:13 <aparcar> over at openwrt we solve this by adding the kernel magic hash to the required version
2022-01-28 11:42:24 <ikke> What does that add?
2022-01-28 11:42:57 <aparcar> if you'd upgrade the kernel, it'd upgrade modules too
2022-01-28 11:43:05 <aparcar> it stays consistent 
2022-01-28 11:43:11 <ikke> That's not an issue in ALpine
2022-01-28 11:43:18 <ikke> because the modules are part of the kernel package
2022-01-28 11:43:34 <ikke> And, wouldn't it be possible to do that by version either?
2022-01-28 11:43:59 <aparcar> we often enough change kernel parameters so the kernel magic changes
2022-01-28 11:44:18 <aparcar> but sure, kernel modules could depend on something which isn't a specific kernel version
2022-01-28 11:45:02 <ikke> aparcar: I assume that would mean the pkgrel is bumped (when you change kernel parameters)?
2022-01-28 11:45:18 <ikke> aparcar: and do you ship modules in separate packages?
2022-01-28 11:45:31 <aparcar> yea modules are it's own packages which can be installed individually
2022-01-28 11:45:36 <ikke> ok
2022-01-28 11:45:40 <aparcar> kmod-usb3 etc
2022-01-28 11:46:01 <ikke> So that's like the out-of-tree modules that we have in alpine
2022-01-28 11:46:54 <ikke> you could do something like `depends="linux-foo=5.15.1-r2"`
2022-01-28 11:47:14 <ikke> It does require you to know the kernel version, though
2022-01-28 11:47:56 <ikke> ncopa has a script that keeps them synchronized and pushes these out-of-tree modules at the same time as a kernel upgrade
2022-01-28 11:49:04 <aparcar> I thought of the linux=5.10 thing but again, I can't attach the hash 
2022-01-28 11:49:39 <ikke> Just help me understand, why can't the revision be a good proxy of the hash?
2022-01-28 11:50:57 <aparcar> maybe that's the solution I'll think about it. It's just different from what we use in our build system right now and the revision would need to be detected automagically
2022-01-28 11:50:59 <aparcar>  --info "depends:kernel=5.10.92-1-c11166738fcd8a63b0c129b71f78fdc1"
2022-01-28 11:51:03 <aparcar> this is what I hoped to use
2022-01-28 11:51:18 <ikke> yes, understood, just not something apk-tools supports at the moment
2022-01-28 11:51:56 <ikke> in alpine, every time we change a package that requires a rebuild, we bump the pkgrel
2022-01-28 11:51:57 <aparcar> I guess I was mostly confused since the `version` field itself is allowed to contain whatever arbitrary version you like
2022-01-28 11:52:33 <aparcar> https://gitlab.alpinelinux.org/alpine/apk-tools/-/issues/10807
2022-01-28 11:53:53 <ikke> aparcar: how should apk-tools determine which version is newer/
2022-01-28 11:55:39 <aparcar> you're right, the kernel itself needs a specific version that increases. Currently it's not possible to upgrade the kernel  via a package manager in openwrt. Instead you'd refresh the image. Dependency wise I just want to glue kernel + mods together
2022-01-28 11:56:07 <aparcar> I wonder if the <> function would do the trick since it allows dependence on a package hash if I understand that correctly
2022-01-28 11:56:28 <ikke> That I'm less familiar with
2022-01-28 11:57:27 <aparcar> ikke: anyway, thanks for your pointers so far, that's helpful!
2022-01-28 11:57:34 <ikke> no problem
2022-01-28 11:57:36 <aparcar> any plans For APK to switch to v3?
2022-01-28 11:58:02 <ikke> Nothing concrete yet, afaik. ariadne mentioned v3.16, but not sure if there is enough time left for such a transition
2022-01-28 11:58:51 <ikke> edge would need to be completely rebuilt afaik
2022-01-28 12:03:52 <donoban> it sounds fun :P
2022-01-28 12:04:06 <ncopa> aparcar: we add a depend=linux-lts=$_kernver-r$_kernelpkgrel to each module package, so it depends on exact kernel version
2022-01-28 12:04:25 <ncopa> it means that kernel modules needs to be rebuilt at the same time as the kernel
2022-01-28 12:12:50 <aparcar> ncopa: yea following this approach I just need to automatically increase the kernelpkgrel on every change :)
2022-01-28 12:16:04 <ncopa> $ tpaste < ~/abump-kmods 
2022-01-28 12:16:04 <ncopa> https://tpaste.us/ZjYe
2022-01-28 12:16:38 <ncopa> i do: abump linux-lts-$newversion && sh ~/abump-kmods main/linux-lts
2022-01-28 12:18:02 <ncopa> even if we used kernel magic or a hash, we would need to rebuild the kernel modules
2022-01-28 12:20:46 <aparcar> yea it's not about rebuilding, it's about having a unique dependency between them. I just realized p5 or whatever wouldn't work since those are not necessarily unique 
2022-01-28 12:22:51 <ncopa> right, if you mix release branches that is relevant
2022-01-28 12:23:34 <ncopa> would be nice to have a 'use exact version of X as dep'
2022-01-28 12:23:40 <ncopa> which would be the hash
2022-01-28 12:24:48 <aparcar> yea I'm looking into that right now
2022-01-28 12:24:49 <aparcar> https://gitlab.alpinelinux.org/alpine/apk-tools/-/blob/master/src/version.c#L158
2022-01-28 12:25:02 <ncopa> the downside with that is that the 3rd partyu modules need to be rebuilt for every build to be happy. Now I can control it by intentionally bump (or not bump) pkgrel
2022-01-28 12:25:05 <aparcar> all mods can just depend on that checksum
2022-01-28 12:26:03 <ncopa> which means that I can for example rebuild a kernel locally, which only adds a module for example, and test it without first rebuilding 3rd party modules that I use
2022-01-28 12:26:17 <ncopa> do you split your kernel package into sub packages?
2022-01-28 12:26:50 <ncopa> i personally think that just depending on exact $pkgver-r$pkgrel is good enough
2022-01-28 12:28:12 <ncopa> depmask checksum is used when installing local file, eg apk add /path/to/file.apk
2022-01-28 12:29:21 <ncopa> ...
2022-01-28 12:29:21 <ncopa> logger><Q1D1Pvo5Z5ZsbF8IOE7rM+2gyvnPs=
2022-01-28 12:29:21 <ncopa> $ doas apk add -u  ~/packages/main/x86_64/logger-2.37.2-r2.apk
2022-01-28 12:29:21 <ncopa> $ grep logger /etc/apk/world 
2022-01-28 12:31:43 <aparcar> yea so depending on that directly could do the trick right?
2022-01-28 12:31:47 <aparcar> for our case at openwrt
2022-01-28 12:38:42 <ncopa> yes
2022-01-28 12:39:14 <aparcar> I'll check it out thank you!
2022-01-28 12:44:03 <donoban> is there a strong policy on alpine for place fish completions in '/usr/share/fish/completions' instead '/usr/share/fish/vendor_completions.d'?
2022-01-28 12:47:03 <ncopa> not that I am aware of
2022-01-28 12:49:53 <donoban> there are only 2 packages https://pkgs.alpinelinux.org/contents?file=&path=%2Fusr%2Fshare%2Ffish%2Fvendor_completions.d&name=&branch=edge using the second
2022-01-28 12:50:43 <donoban> I wonder because I patched one package for using '/usr/share/fish/completions' and usptream suggest me to remove the patch
2022-01-28 12:51:09 <aparcar> ncopa: this checksum is part of the index or also stored in the APK file itself?
2022-01-28 12:52:52 <aparcar> I figure it's the unique-id in the v3 format
2022-01-28 13:11:57 <PJ[m]> I think fish completions should be fixed and put in vendor directory
2022-01-28 13:12:07 <PJ[m]> donoban: which package?
2022-01-28 13:12:38 <PJ[m]> Ubuntu also has mix of vendor/fish dir in some packages
2022-01-28 13:14:07 <donoban> bubblejail
2022-01-28 13:16:42 <donoban> in alpine most packages put in fish dir (https://pkgs.alpinelinux.org/contents?file=&path=%2Fusr%2Fshare%2Ffish%2Fcompletions&name=&branch=edge&arch=x86_64) so I wonder if its a distro policy
2022-01-28 13:22:04 <DylanVanAssche> ncopa: Regarding !29955: moving the needed dependency around to main or move BlueZ to community won't be possible I think.
2022-01-28 13:22:04 <DylanVanAssche> My MR only applies to the separate `obexd` not the main Bluetooth daemon. Would it be an option to make `obexd` a separate package and build it in community? That would avoid this dependency hell as well and keep BlueZ itself in community.
2022-01-28 13:24:22 <DylanVanAssche> *keep BlueZ itself in main.
2022-01-28 13:24:58 <PJ[m]> now that I remember, I also dealt with fish completions for github-cli package as it was installing completions to fish dir and fish was already shipping own completions
2022-01-28 13:26:39 <donoban> well, if there is no position about that, better just let upstream decision?
2022-01-28 13:38:00 <skarnet> !30104 can merge, but since minimal was in charge of the previous package I'd like him to take a look first, whenever he's here ;)
2022-01-28 14:06:58 <Hello71> ktprograms: actually do you even have qt5-qtwebengine-dbg installed
2022-01-28 14:08:25 <ktprograms> Hello71: Yes
2022-01-28 14:08:36 <Hello71> hm.
2022-01-28 14:08:56 <Hello71> apk upgrade -a?
2022-01-28 14:11:38 <MY-R> is good to run "apk version" from time to time, sometimes there can be packages which upgrade doesnt wanna upgrade without any notice
2022-01-28 14:12:16 <ktprograms> Hello71: MY-R: No change with upgrading, apk version doesn't list anything
2022-01-28 14:12:35 <Hello71> try vgdb then
2022-01-28 14:12:42 <ktprograms> How do I use it?
2022-01-28 14:12:54 <Hello71> what happened with regular gdb again?
2022-01-28 14:13:09 <ktprograms> 3 item long stack trace
2022-01-28 14:13:26 <ktprograms> https://tpaste.us/BgpW
2022-01-28 14:15:11 <Hello71> bt full?
2022-01-28 14:15:51 <Hello71> and what is the rest of the gdb output
2022-01-28 14:17:46 <ktprograms> I'm in the middle of getting it, but what's interesting is that, when I downgrade the VM software I'm using, it still crashes, but instead of saying status 11, it says status 0.
2022-01-28 14:18:14 <ktprograms> The only major thing I can think of that changed is that the virgl renderer string changed.
2022-01-28 14:20:09 <ktprograms> bt full: https://tpaste.us/dP7J
2022-01-28 14:23:34 <ktprograms> nvm ignore the status 0 part, it's back to saying status 11
2022-01-28 14:25:47 <ktprograms> gtg, I'll read messages directed to me in the irclog
2022-01-28 15:13:43 <Ariadne> ikke: apk allows a mix of v2 and v3 :)
2022-01-28 15:14:39 <ncopa> DylanVanAssche: sounds like a reasonable workaround
2022-01-28 15:15:22 <DylanVanAssche> ncopa: I will try this idea out over the next week or so and if it works, update the MR. Thanks!
2022-01-28 15:15:56 <ncopa> might be enough to only build obexd as I understand?
2022-01-28 15:16:35 <DylanVanAssche> ncopa: Yes, the evolution-data-server dependency is only needed to build the obexd with PBAP support. 
2022-01-28 16:13:39 <ikke> Ariadne: ah, didn't know. Nice 
2022-01-28 16:14:12 <Ariadne> yes, i think the plan is that we will introduce apk-tools 3 to edge, and then switch over to v3 indices after a while
2022-01-28 16:14:26 <Ariadne> and then switch over to v3 packages after a longer while
2022-01-28 16:14:39 <Ariadne> but a v3 index can describe v2 packages, and vice versa
2022-01-28 16:15:08 <Ariadne> (the valgrind test suite issue fixed itself upstream, and they found out the test was actually just wrong in general)
2022-01-28 16:16:05 <ikke> aha, I suppose that was ragerding argc == 0, right?
2022-01-28 16:20:02 <ikke> Ariadne: how would abuild compattibility work? 
2022-01-28 16:20:21 <Ariadne> i have some patches
2022-01-28 16:20:26 <Ariadne> that make abuild generate v3 apks
2022-01-28 16:20:45 <ikke> ok
2022-01-28 16:20:47 <Ariadne> but i'm going to redo them, because we're adding support for abuild's manifest format to apk mkpkg
2022-01-28 16:21:13 <Ariadne> (.PKGINFO)
2022-01-28 16:42:23 <ikke> ncopa: p7zip fails on armhf
2022-01-28 16:47:13 <ikke> bus error
2022-01-28 16:55:46 <Ariadne> most likely an unaligned access
2022-01-28 16:56:08 <genkitj> Does linux-edge have a maintainer currently?
2022-01-28 16:56:56 <Ariadne> no idea
2022-01-28 17:02:35 <ikke> the package still lists a maintainer: https://pkgs.alpinelinux.org/packages?name=linux-edge&branch=edge
2022-01-28 17:03:00 <genkitj> Yea, but actually...
2022-01-28 17:04:12 <genkitj> Haven't mps been removed from the project?
2022-01-28 17:04:16 <ikke> No
2022-01-28 17:04:20 <donoban> mps upgrated some packages today
2022-01-28 17:04:54 <donoban> I suppose that he will ask someone to take his pacakges if he don't want to maintain them
2022-01-28 17:05:23 <ikke> genkitj: we gave him a warning, and he takes some time to overthing things
2022-01-28 17:26:31 <genkitj> I didn't follow the story, but I would say some of the packages he maintained are pretty significant. I hope "overthinking" leads to positive results and he returns.
2022-01-28 17:53:58 <omni> I hope it's "think over" rather than "overthink" =)
2022-01-28 17:54:21 <ikke> omni: yes :)
2022-01-28 17:54:36 <ikke> That's my dutch comming through
2022-01-28 17:54:49 <ikke> coming*
2022-01-28 18:01:15 <Misthios> When u translate it literally
2022-01-28 18:08:49 <skarnet> current goal: being able to think things over without overthinking them
2022-01-28 18:09:07 <ikke> A noble goal
2022-01-28 18:10:49 <omni> skarnet: +1
2022-01-28 18:13:33 <Saijin_Naib[m]> Are the out of date package/flagged package emails sent out on a schedule or something? I flagged one of my own to test the workflow and I still have not received it
2022-01-28 18:39:10 <Newbyte> What are -static subpackages? 
2022-01-28 18:39:53 <Newbyte> Saijin_Naib: yeah, I'm not sure what's up with that. I have received an email like that exactly one time.
2022-01-28 18:40:03 <Newbyte> despite my packages being flagged more times than that
2022-01-28 18:40:21 <Saijin_Naib[m]> Newbyte: Thanks. I'll just make sure I watch out more often then.
2022-01-28 18:48:11 <ikke> Saijin_Naib[m]: You can also look at something like repology
2022-01-28 18:51:23 <Saijin_Naib[m]> ikke: Just seeing it now. Looks neat!
2022-01-28 18:51:42 <Saijin_Naib[m]> Thankfully I'm only responsible for like, 3 packages so the cognitive load is small right now 🤣
2022-01-28 18:51:47 <ikke> heh
2022-01-28 18:51:56 <ikke> If it's hosted on github, you can also subscribe to new releases
2022-01-28 18:54:01 <PJ[m]> I'm monitoring releases and other stuff myself, get to quickly test new releases if they break anything ;)
2022-01-29 04:11:46 <Saijin_Naib[m]> What's the proper process/channel for a package I think might be misnamed?
2022-01-29 04:12:02 <Saijin_Naib[m]> https://pkgs.alpinelinux.org/contents?file=&path=&name=rstcheck&branch=edge&repo=community&arch=x86_64
2022-01-29 04:12:13 <Saijin_Naib[m]> From what I know, this should be py3-rstcheck
2022-01-29 04:12:42 <psykose> it's not, it has a binary
2022-01-29 04:12:57 <psykose> py3 programs you just run by name don't get py3- prefixed
2022-01-29 04:13:07 <Saijin_Naib[m]> Ahhh okay
2022-01-29 04:13:23 <psykose> 'binary' just meaning something you run to use i guess
2022-01-29 04:13:24 <Saijin_Naib[m]> Thanks 🙂
2022-01-29 04:13:28 <psykose> any libs get py3- ofc
2022-01-29 04:13:33 <psykose> :)
2022-01-29 04:14:08 <Saijin_Naib[m]> Makes sense. py3- is for pure python packages then, right?
2022-01-29 04:14:20 <psykose> this is also pure python pretty sure
2022-01-29 04:14:24 <Saijin_Naib[m]> I was just thinking anything that shipped on pypi was a py3- candidate
2022-01-29 04:14:26 <psykose> but really anythin that is lib/ only
2022-01-29 04:14:32 <psykose> mm
2022-01-29 04:14:34 <psykose> not sure
2022-01-29 04:14:57 <Saijin_Naib[m]> Makes sense! Glad we have it already
2022-01-29 08:11:24 <psykose> eschwartz: i remember you mentioning the meson testsuite was not being run, do you have some insight into how to disable/fix certain things in it? e.g. in !29489
2022-01-29 08:11:31 <psykose> some of the tests are not really feasible to run properly
2022-01-29 08:12:12 <psykose> and some fail for unknown reasons :)
2022-01-29 09:55:51 <z3ntu> Could I please get a review for this MR? https://gitlab.alpinelinux.org/alpine/aports/-/merge_requests/29589 It's been open since nearly two weeks, but I know that 15 new packages isn't super trivial to review :)
2022-01-29 09:56:52 <ikke> !29589
2022-01-29 09:58:24 <PureTryOut> z3ntu: oh shit I promised to do that earlier. Will do in an hour or so 🙈
2022-01-29 09:59:22 <ikke> psykose also started reviewing it already
2022-01-29 12:54:05 <donoban> hmm.. it seems that problem with non-suid sway is related with XKB_DEFAULT_VARIANT
2022-01-29 12:57:55 <donoban> http://ix.io/3NRx
2022-01-29 12:58:22 <donoban> if I unset XKB_DEFAULT_VARIANT it works
2022-01-29 13:10:30 <donoban> could anyone refresh me what package easy the setup for getting coredumps?
2022-01-29 13:14:53 <ikke> donoban: uname -c unlimited
2022-01-29 13:16:32 <donoban> I did http://ix.io/3NREulimit -S -c unlimited
2022-01-29 13:16:42 <donoban> wops, anyway I think that I've already get it directly with gdb
2022-01-29 13:18:53 <ktprograms> ikke: It's ulimit, not uname
2022-01-29 13:30:06 <donoban> on #sway say that this is not their fault (http://ix.io/3NRx http://ix.io/3NRE)
2022-01-29 13:30:52 <donoban> they say that my env are wrong but they are available via setup-keymap
2022-01-29 13:31:08 <ikke> ktprograms: yes, thanks
2022-01-29 13:32:27 <Hello71> pretty sure setup-keymap is for kbd, not skb
2022-01-29 13:32:30 <Hello71> s/skb/xkb/
2022-01-29 13:32:30 <alpine-meetbot> Hello71 meant to say: pretty sure setup-keymap is for kbd, not xkb
2022-01-29 13:33:16 <donoban> do you where XKB_DEFAULT_VARIANT is defined?
2022-01-29 13:35:12 <donoban> lol, it works fine with XKB_DEFAULT_VARIANT=cat
2022-01-29 14:01:48 <omni> I saw that unrar has moved to non-free, but how do I install it then?
2022-01-29 14:02:16 <omni> or what alternties are there for rar archives?
2022-01-29 14:02:22 <Misthios> I think it's manually building it
2022-01-29 14:04:09 <ikke> omni: 7zip, bsdtar should be able to extract them
2022-01-29 14:06:19 <omni> Misthios: aha
2022-01-29 14:06:39 <omni> ikke: I tried with p7zip, but it said "unknown suffix", I'll try with bsdtar
2022-01-29 14:06:46 <psykose> you need uhh
2022-01-29 14:06:50 <psykose> 7zr x
2022-01-29 14:06:54 <psykose> or 7za x
2022-01-29 14:06:55 <psykose> or some shit
2022-01-29 14:06:56 <psykose> not just 7z
2022-01-29 14:06:58 <psykose> i forget
2022-01-29 14:07:10 <Hello71> p7zip doesn't have it
2022-01-29 14:07:11 <psykose> but if it doesn't work then bsdtar probably does
2022-01-29 14:07:31 <ikke> man 7z / 7za do not list it
2022-01-29 14:07:33 <ikke> so bsdtar then
2022-01-29 14:07:38 <Hello71> and the 7zip unrar code is non-free anyways
2022-01-29 14:07:49 <ikke> aha
2022-01-29 14:08:21 <Hello71> the bsdtar unrar code is... dubious but not clearly non-free
2022-01-29 14:08:56 <ikke> (part of libarchive-tools)
2022-01-29 14:10:31 <psykose> armhf and s390x still need to have p7zip fixed incidentally
2022-01-29 14:11:13 <omni> bsdtar: Error opening archive: Failed to open '/dev/st0'
2022-01-29 14:11:38 <omni> I found https://gitlab.com/bgermann/unrar-free perhaps that's something to package
2022-01-29 14:12:24 <psykose> https://unix.stackexchange.com/questions/131010/why-am-i-getting-an-error-failed-to-open-dev-st0-when-extracting-an-archiv
2022-01-29 14:12:41 <psykose> u need an -f
2022-01-29 14:13:52 <omni> ah, thanks, sloppy reading of the --help output on my part
2022-01-29 14:14:33 <omni> bsdtar: RAR solid archive support unavailable.
2022-01-29 14:17:13 <omni> "RAR archive data, v4, os: Win32, flags: Solid" is what file(1) has to say about this archive
2022-01-29 14:17:32 <ericonr> omni: unrar-free seems to be only a frontend for libarchive, so libarchive still needs to have that code built into it
2022-01-29 14:17:36 <ericonr> Ands
2022-01-29 14:17:39 <ericonr> And some rat 
2022-01-29 14:17:58 <ikke> "Solid archive is an archive packed with a special compression method, which treats several or all files within the archive as one continuous data stream."
2022-01-29 14:17:58 <ericonr> .... And some rar files can't be read by libarchive either 
2022-01-29 14:23:14 <omni> ok, I just added v3.14 main as a tagged repo and installed unrar from there
2022-01-29 14:23:26 <omni> thanks everyone!
2022-01-29 14:25:48 <psykose> you can also build it
2022-01-29 14:50:06 <donoban> my XKB_DEFAULT_VARIANT=es-cat comes from 00locale.sh on profile.d, I don't know how it ended there maybe I edited it when installed alpine
2022-01-29 14:52:27 <ikke> It's not something that Alpine itself placed there on instal
2022-01-29 14:52:47 <ikke> It comes from  musl-locales
2022-01-29 14:52:52 <donoban> yes I see that some older snapshots doesn't have it
2022-01-29 14:53:03 <donoban> currently I don't have musl-locales installed
2022-01-29 14:53:26 <donoban> maybe it come from some gnome that I intalled for test something and uninstalled later
2022-01-29 14:53:41 <donoban> but if I manual edited it, it wasn't removed later
2022-01-29 14:53:56 <donoban> well I'm gonna try removing it
2022-01-29 14:54:33 <donoban> lol, /mnt/btrfs/@ROOT-old/.snapshots/before-gnome here it isn't 
2022-01-29 14:55:36 <donoban> yes it appears on next snapshot but only with export MUSL_LOCPATH="/usr/share/i18n/locales/musl"
2022-01-29 14:55:51 <donoban> probably I just added the other later trying to fix some keyboard problem
2022-01-29 14:57:51 <donoban> I feel a pretty hard that sway segfaults just for having this wrong configured but they seem ok with it
2022-01-29 14:59:23 <psykose> i mean the segfault is in the xkbcommon so
2022-01-29 15:01:06 <ikke> the wezterm buildlog is filled with nul-bytes
2022-01-29 15:04:56 <psykose> that is.. bizarre
2022-01-29 15:05:00 <psykose> why did it pass on ci
2022-01-29 15:05:06 <psykose> PJ[m]: ^
2022-01-29 15:05:12 <ikke> now x86_64 as welll
2022-01-29 15:05:37 <PJ[m]> why you ask me
2022-01-29 15:05:57 <PJ[m]> it works fine for me in docker and locally as well
2022-01-29 15:06:56 <psykose> cause it's your package silly :p
2022-01-29 15:07:23 <psykose> but yes, should test it on the same infra by hand
2022-01-29 15:07:26 <PJ[m]> weird
2022-01-29 15:19:24 <PJ[m]> psykose: should I assume you are going to try? :)
2022-01-29 15:19:45 <psykose> that's effort
2022-01-29 15:19:50 <psykose> groan
2022-01-29 15:20:11 <PJ[m]> I think I know where in the code it happens
2022-01-29 15:20:16 <psykose> ikke: sure, give me an x86_64 container too ^~^
2022-01-29 15:20:46 <PJ[m]> I'm running it now and hopefully catch those pesky bytes
2022-01-29 15:21:08 <psykose> i'm just not sure why the entire log is empty
2022-01-29 15:21:11 <PJ[m]> so you can wait with that for me
2022-01-29 15:21:26 <PJ[m]> entire?
2022-01-29 15:21:32 <psykose> oh
2022-01-29 15:21:33 <psykose> it's not
2022-01-29 15:22:17 <psykose> check fails but none of the tests do
2022-01-29 15:22:22 <psykose> i think your killall command fails
2022-01-29 15:22:30 <psykose> just do || true
2022-01-29 15:22:43 <PJ[m]> eh...
2022-01-29 15:22:46 <psykose> the same hang does not occur on builders
2022-01-29 15:22:52 <psykose> and empty killall exits nonzero
2022-01-29 15:23:32 <psykose> i guess i will
2022-01-29 15:24:13 <psykose> done
2022-01-29 15:24:31 <PJ[m]> well
2022-01-29 15:24:35 <PJ[m]> i still had few things to amend
2022-01-29 15:24:41 <PJ[m]> :P
2022-01-29 15:24:43 <psykose> what few things
2022-01-29 15:24:46 <psykose> feel free to mr them
2022-01-29 15:25:07 <PJ[m]> i know, but I also want to figure out the null bytes
2022-01-29 15:25:15 <psykose> ah, still fails
2022-01-29 15:26:05 <psykose> maybe that stream of nuls that truncates the log is the failing test
2022-01-29 15:26:20 <psykose> otherwise it looks.. fine
2022-01-29 15:42:20 <ikke> psykose: port 22047 :)
2022-01-29 15:42:25 <ikke> same host
2022-01-29 15:42:25 <psykose> thanks!
2022-01-29 16:22:39 <PJ[m]> I guess that I don't have to bump pkgrel if it didn't even build?
2022-01-29 16:23:11 <ikke> correct
2022-01-29 16:27:20 <psykose> builds over ssh fine
2022-01-29 16:27:21 <psykose> classic
2022-01-29 16:27:42 <psykose> ikke: is there a reason the actual log would be so truncated
2022-01-29 16:29:00 <PJ[m]> building it locally with nul bytes fixed just to be sure, will push mr in a bit
2022-01-29 16:29:40 <ikke> Can you try with setsid + nohup! 
2022-01-29 16:29:44 <ikke> ? 
2022-01-29 16:30:19 <psykose> like `setsid nohup cargo ..`
2022-01-29 16:31:22 <psykose> sure
2022-01-29 16:31:24 <psykose> let's see..
2022-01-29 16:34:56 <psykose> wait
2022-01-29 16:35:07 <psykose> it passed on the builders
2022-01-29 16:35:10 <psykose> ?????
2022-01-29 16:35:25 <psykose> lol
2022-01-29 16:35:36 <PJ[m]> oh yeah, aarch64
2022-01-29 16:35:40 <psykose> both
2022-01-29 16:36:07 <PJ[m]> i only see aarch64
2022-01-29 16:36:15 <psykose> (uploads are just delay batched and aarch was first, you're probably looking at pkgs.a.o)
2022-01-29 16:36:23 <PJ[m]> i do
2022-01-29 16:36:29 <PJ[m]> still confused
2022-01-29 16:36:34 <psykose> https://build.alpinelinux.org/
2022-01-29 16:36:38 <psykose> or the -commits channel
2022-01-29 16:36:43 <PJ[m]> yes, I have it opened
2022-01-29 16:38:36 <PJ[m]> well, as you said, magic memes
2022-01-29 16:38:56 <psykose> sigh
2022-01-29 16:39:23 <psykose> well, have fun debugging it some more
2022-01-29 16:39:31 <PJ[m]> :D
2022-01-29 16:39:40 <psykose> i am still seeing what that invocation above will change
2022-01-29 16:56:13 <psykose> builds fine with nohup setsid
2022-01-29 16:56:26 <psykose> and unset TERM just in case xterm-256color did something
2022-01-29 16:56:35 <psykose> shrug
2022-01-29 17:33:38 <psykose> p7zip still stuck on s390x
2022-01-29 17:35:08 <ikke> Seems like all threads are deadlocked
2022-01-29 17:35:23 <ikke> main/p7zip/src/p7zip-17.04/check/../bin/7za a -m0=lizard 7za433_7zip_lizard.7z 7za433_7zip_lzma
2022-01-29 17:36:52 <ikke> hanging on brotli tests
2022-01-29 17:37:18 <ikke> at least, I think it is
2022-01-29 17:37:34 <ikke> https://tpaste.us/NMNv
2022-01-29 17:57:23 <Hello71> maybe just package new 7-zip and drop p7zip
2022-01-29 17:57:33 <Hello71> it's still dubious but not more than the old one i think
2022-01-29 18:02:27 <Hello71> as a bonus, there are no tests to fail
2022-01-29 18:02:59 <psykose> why does every compression thing like this feel like it fell off the back of a truck
2022-01-29 18:06:16 <Hello71> well the other option is "big tech" zstd
2022-01-29 18:06:37 <psykose> usually a better choiuce
2022-01-29 18:06:42 <psykose> choice*
2022-01-29 18:07:27 <Hello71> in general "big tech" is the antithesis of "small and fast" which compressors must be
2022-01-29 18:08:15 <psykose> zstd is certainly one of the slightly faster ones
2022-01-29 18:08:17 <psykose> :p
2022-01-29 18:08:17 <Hello71> or more precisely, compressor authors need to constantly be thinking of "small and fast". there's no point in making compressor with chromium development methodology and results
2022-01-30 02:20:38 <bdju> boomanaiden154: Hey. Do you know if freecad could have support for aarch64 enabled?
2022-01-30 02:55:13 <artok> bdju: I still recommend flatpak for it, for the sake of it's dependencies
2022-01-30 02:55:48 <artok> not that it would be great to have it, since tomorrow I'm getting my 3d printer =D
2022-01-30 02:55:50 <bdju> I have kinda avoided flatpak and things like it up to this point and will probably try to continue avoiding that sort of thing
2022-01-30 02:56:33 <ktprograms> artok: The flatpak is only available on x86_64
2022-01-30 02:56:37 <ktprograms> https://github.com/flathub/org.freecadweb.FreeCAD/blob/master/flathub.json
2022-01-30 02:56:38 <artok> I'm in that great position that my regular working machine is macOS
2022-01-30 02:56:46 <artok> oh damn
2022-01-30 02:57:06 <artok> my bad there then
2022-01-30 02:57:13 <ktprograms> artok: Yea but it seems that FreeCAD doesn't sign their dmg releases
2022-01-30 02:57:26 <ktprograms> I guess $99 a year is quite a lot for a FOSS project
2022-01-30 02:57:48 <artok> they don't so additioonal keypress for the first start is needed
2022-01-30 02:58:13 <artok> blender is signed as it has Apple supporting it ;)
2022-01-30 02:58:13 <ktprograms> That's what I meant
2022-01-30 02:58:26 <ktprograms> really? Didn't know that
2022-01-30 02:58:32 <artok> I know what you meant
2022-01-30 02:58:51 <ktprograms> Although I suppose that's why blender had support for the M1 GPU soon after release
2022-01-30 02:58:57 <artok> I was compiling official blender dmg 2+ years ago
2022-01-30 02:59:08 <ktprograms> artok: I mean that yes, because the dmgs aren't signed, you need to manually allow it
2022-01-30 03:00:55 <artok> I did some openmp patches to allow macos to run simulations on all cores
2022-01-30 03:01:59 <artok> still loving the software after I just dropped out of that project with silence
2022-01-30 03:03:34 <artok> covid kicked hard on me so my time allocation shifted a lot
2022-01-30 03:50:54 <boomanaiden154> bdju: potentially. There was an MR a while ago by someone else to enable it on more architectures but I believe there were a couple tests failing. I can look into it more probably tomorrow if you need it on aarch64.
2022-01-30 03:51:22 <ktprograms> boomanaiden154: I'm currently building it on aarch64 to see how the tests fail
2022-01-30 03:52:46 <boomanaiden154> Okay. If you could send over some logs that would be great.
2022-01-30 03:52:57 <boomanaiden154> The best course of action will probably be to just patch the tests out.
2022-01-30 03:53:19 <boomanaiden154> I had to do it for four on x86_64 due to some incompatibilities with Python 3.10.
2022-01-30 03:53:50 <ktprograms> boomanaiden154: I'll see what I can do. Problem is it takes about half an hour to build even on a VM on a M1 with all 8 cores
2022-01-30 03:55:44 <boomanaiden154> Yeah. It takes like 20-30 minutes to build on my 16 core x86_64 system.
2022-01-30 03:57:33 <ktprograms> boomanaiden154: It seems that when I add "sh" to the check() function and try to manually run the commands used there (so that I can do debugging of it without having to wait for it to build again), it doesn't seem to be in the right dir, because (for example) the "ln" line references ../lib, which isn't a dir that exists
2022-01-30 04:01:35 <boomanaiden154> You can just use abuild check instead of running the commands separately.
2022-01-30 04:01:59 <boomanaiden154> You need to delete the test install folder each time though or at least delete the generated symlinks.
2022-01-30 04:03:28 <ktprograms> boomanaiden154: The problem is I'm using rootbld
2022-01-30 04:03:44 <ktprograms> I'm just wondering _why_ the dir when I use sh doesn't seem to be correct
2022-01-30 04:08:07 <boomanaiden154> I’m not sure why either.
2022-01-30 04:08:21 <boomanaiden154> I haven’t tried it with rootbld before.
2022-01-30 04:08:44 <boomanaiden154> Maybe try using full paths instead of relative paths for all the symlinks.
2022-01-30 04:46:10 <ktprograms> boomanaiden154: After adding COIN_GL_NO_CURRENT_CONTEXT_CHECK=1, then there's a segfault in PathTests.TestPathHelix.TestPathHelix (src/Mod/Path/PathTests/TestPathHelix.py:103)
2022-01-30 05:48:31 <ktprograms> I've tried running abuild check a few times, and it seems to segfault in different tests each time
2022-01-30 06:11:05 <ktprograms> The backtrace is https://tpaste.us/QrYr
2022-01-30 06:30:38 <ktprograms> boomanaiden154: Backtrace with coin-dbg built and installed: https://tpaste.us/KMLe
2022-01-30 06:44:42 <ktprograms> boomanaiden154: I found https://www.mail-archive.com/debian-bugs-dist@lists.debian.org/msg1839559.html and tried building coin with the patch from https://github.com/coin3d/coin/pull/404/files, and it prevents the segfault, but as said in a later message in the thread, I see black triangles on the window.
2022-01-30 14:54:12 <JuniorJPDJ> o/
2022-01-30 14:54:31 <JuniorJPDJ> I'm having weird issue on !30209 - could I ask someone to take a look here?
2022-01-30 14:56:16 <JuniorJPDJ> oh nvm.. I've not updated checksum
2022-01-30 15:07:01 <JuniorJPDJ> anyway - it works on all platforms ;D
2022-01-30 19:58:11 <maxzor> Hello, are cmake gnuinstalldirs fine for Alpine?  I am having discussions on ROCm installation layout for Debian https://lists.debian.org/debian-ai/2022/01/msg00103.html so you are welcome to chime in and voice your issues
2022-01-30 20:00:01 <DylanVanAssche> ncopa: I applied the discussed 'workaround' of moving OBEXD out of main (actually not moving, but providing a more enhanced version as opt-in).
2022-01-30 20:00:01 <DylanVanAssche> This works fine on my devices with PyOBEX PBAP client. !29955
2022-01-31 01:14:23 <ktprograms> boomanaiden154: Any ideas re freecad?
2022-01-31 01:40:04 <boomanaiden154> Not really anything that’s going to add full functionality. There seem to be quite a few issues with upstream packages that don’t support ARM.
2022-01-31 01:40:38 <ktprograms> Do you think it's an ARM problem?
2022-01-31 01:41:41 <boomanaiden154> Not totally sure, but pretty much everything is fully functional on x86_64 while it isn’t on aarch64.
2022-01-31 03:22:00 <boomanaiden154> ktprograms: I just opened up !30249 to start working on aarch64 for FreeCAD. Was the seg fault that you hit yesterday immediately after the check started?
2022-01-31 03:22:24 <boomanaiden154> Also, did you get a warning from testBooleanCommonCase about not being able to set up the GL context?
2022-01-31 03:22:27 <boomanaiden154> Relevant build log: https://gitlab.alpinelinux.org/boomanaiden154/aports/-/jobs/612985
2022-01-31 03:24:42 <ktprograms> boomanaiden154: The segfault at random points in the tests, but it happens consistently if you run the binary in the test_install dir and press Ctrl+N to create a new file
2022-01-31 03:25:11 <ktprograms> I did also get the GL context warning from coin
2022-01-31 03:26:08 <ktprograms> As I said earlier, I was able to fix the segfault by adding the patch from https://github.com/coin3d/coin/pull/404/files, but then that causes black triangles to appear over the window
2022-01-31 03:26:24 <ktprograms> See https://www.mail-archive.com/debian-bugs-dist@lists.debian.org/msg1839559.html
2022-01-31 03:26:33 <boomanaiden154> I was about to try and apply that patch.
2022-01-31 03:26:43 <boomanaiden154> Did it fix the flaky segfaults in the tests at all?
2022-01-31 03:26:52 <ktprograms> Yes, no segfault at all
2022-01-31 03:27:10 <ktprograms> But I think the tests failed for a different reason. Can't remember exactly
2022-01-31 03:27:28 <boomanaiden154> They do tend to randomly segfault sometimes.
2022-01-31 03:27:39 <ktprograms> Also I don't think that would be a good solution. We should try to figure out why there isn't any current GL context at that point.
2022-01-31 03:30:09 <boomanaiden154> Yes. It would be nice if the coin tests would catch this instead of the error manifesting in FreeCAD.
2022-01-31 03:31:33 <boomanaiden154> I don't really have any aarch64 hardware at the moment though other than an RPi3, so I'm not sure how much I'm going to be able to do.
2022-01-31 03:31:56 <boomanaiden154> I've been meaning to try and experiment with qemu for some time now though, so maybe I'll see what I can get done with that.
2022-01-31 03:32:07 <ktprograms> boomanaiden154: If you have any ideas, I can help to build/test, I've got an M1 Air
2022-01-31 09:21:31 <omni> ncopa: you approved !28797 but then I managed to stall it being unsure of licenses
2022-01-31 09:22:45 <omni> !29072 would -DNO_LIBUDEV be preferred?
2022-01-31 09:23:21 <omni> !30053 should we wait for an upstream release?
2022-01-31 15:29:59 <clandmeter> nmeum: which kernel are you running on your unmatched?
2022-01-31 15:30:44 <nmeum> clandmeter: linux-edge
2022-01-31 15:31:03 <clandmeter> ok
2022-01-31 15:31:10 <clandmeter> looks like reboot should work on 5.17
2022-01-31 15:31:20 <nmeum> oh, glad to here that
2022-01-31 15:31:27 <nmeum> we don't package it yet though
2022-01-31 15:31:32 <clandmeter> https://www.phoronix.com/scan.php?page=news_item&px=Linux-5.17-RISC-V
2022-01-31 15:32:38 <Hello71> well 5.17 isn't even out and probably won't be for at least a month
2022-01-31 15:34:14 <clandmeter> we can wait, but ill need to update with im in the office :)
2022-01-31 15:34:15 <nmeum> ah, I just can't keep up with kernel releases these days
2022-01-31 16:27:59 <skarnet> could someone please merge !30130 ? thanks!
2022-01-31 16:29:51 <ikke> done
2022-01-31 16:30:12 <Ermine> Yay
2022-01-31 16:38:13 <skarnet> thanks!
2022-01-31 16:39:27 <minimal> can some please merge !29869
2022-01-31 16:48:01 <Hello71> i think some people use openrc-init?
2022-01-31 16:48:54 <minimal> without the OpenRC shutdown/halt stuff being built?
2022-01-31 16:49:48 <ikke> minimal: I'd like ncopa to approve the MR as maintainer
2022-01-31 16:50:30 <minimal> ikke: np, its just been sitting around for a while now...
2022-01-31 17:19:48 <Ermine> Should I fix this? https://gitlab.alpinelinux.org/alpine/aports/-/issues/13488
2022-01-31 17:22:21 <omni> as I read this https://gcc.gnu.org/onlinedocs/gcc/Optimize-Options.html and also 'gcc -Os -Q --help=optimizers | grep fomit', -fomit-frame-pointer is enabled but we also define it in our CFLAGS https://gitlab.alpinelinux.org/alpine/abuild/-/blob/master/abuild.conf
2022-01-31 17:28:46 <ikke> Ermine: that one is tricky 
2022-01-31 17:29:19 <ikke> Subpackages with a different package version are not supported at the moment 
2022-01-31 17:42:00 <Hello71> omni: it is redundant except on aarch64
2022-01-31 17:42:08 <Hello71> the manual is wrong
2022-01-31 17:53:29 <Ermine> ikke: okay, I'll add a comment to this issue
2022-01-31 18:04:04 <omni> Hello71: ah, look at that, thanks
2022-01-31 18:04:06 <ikke> Ermine: not sure if it makes sense, but if it's something that has a different version, it might make sense to create a separate package for it
2022-01-31 18:05:09 <Hello71> also gcc -Q --help is often inaccurate
2022-01-31 18:07:19 <omni> unfortunate
2022-01-31 19:22:54 <omni> ncopa: how about this? !29072
2022-01-31 19:47:39 <Ermine> ikke: if I convert this subpackage to package, it version would be lower than current subpackage's version. Is it fine?
2022-01-31 19:47:57 <ikke> Ermine: not for users who already have a newer version
2022-01-31 19:48:09 <ikke> unless they use -a / --available when upgrading
2022-01-31 20:13:41 <donoban> "ERROR: curl-fish-completion-7.81.0-r1: trying to overwrite usr/share/fish/completions/curl.fish owned by fish-3.3.1-r2.
2022-01-31 20:14:05 <donoban> maybe is this related with I asked about fish path?
2022-01-31 20:19:48 <donoban> mv: can't rename '/home/donoban/aports/community/bubblejail/pkg/bubblejail/usr/share/fish/completions': No such file or directory
2022-01-31 20:20:17 <donoban> it seems that abuild tries to look for that path
2022-01-31 20:20:31 <psykose> no, someone just added the completions but forgot to remove them from fish
2022-01-31 20:20:36 <psykose> since fish also ships them
2022-01-31 20:20:45 <donoban> well, I mean on my package
2022-01-31 20:20:57 <donoban> I had a patch for put them inside fish/competions instead vendor.d
2022-01-31 20:21:03 <donoban> I removed it and now the build fails
2022-01-31 20:21:35 <donoban> in fact, abuild expect them to be in     if [ -d "$pkgdir"/usr/share/fish/completions ]; then
2022-01-31 20:22:24 <psykose> yes
2022-01-31 20:22:37 <psykose> though the first error you posted is something else and broken for everyone
2022-01-31 20:24:04 <donoban> hmm..
2022-01-31 20:24:56 <donoban> I don't know what to do, just restore the patch for now?
2022-01-31 20:26:09 <psykose> probably
2022-01-31 20:26:09 <donoban> I can try to "fix" abuild but maybe it needs some discusion about the alpine policy on this
2022-01-31 20:26:21 <donoban> maybe open an issue?
2022-01-31 20:26:21 <psykose> by default the -fish-comp split func looks there
2022-01-31 20:26:24 <psykose> if it's empty then it fails
2022-01-31 20:26:29 <psykose> this is not an abuild issue
2022-01-31 20:26:41 <donoban> well, it's contrary to the fish policy
2022-01-31 20:26:52 <psykose> usr/share/fish/completions is alpine policy
2022-01-31 20:26:52 <psykose> shrug
2022-01-31 20:26:58 <donoban> fish wants only upstream completions on /usr/share/fish/completions
2022-01-31 20:26:58 <psykose> i don't know why it is that way either
2022-01-31 20:27:04 <donoban> and all externals on vendor.d
2022-01-31 20:27:07 <eschwartz> psykose: I just noticed your ping about the meson testsuite, let me take a look
2022-01-31 20:27:12 <psykose> yes, so perhaps uhh
2022-01-31 20:27:19 <psykose> back when this was decided somebody didn't realise that
2022-01-31 20:27:22 <psykose> and we should amend abuild
2022-01-31 20:27:26 <psykose> to look in *both* places
2022-01-31 20:27:29 <psykose> you should file an issue
2022-01-31 20:27:35 <psykose> in the meantime for that package, !30280
2022-01-31 20:27:55 <psykose> eschwartz: thank you! let me know of what should be done
2022-01-31 20:27:55 <donoban> https://fishshell.com/docs/current/completions.html#where-to-put-completions
2022-01-31 20:28:43 <donoban> maybe, for backwards compatilibty
2022-01-31 20:28:47 <donoban> abuild could use both path's
2022-01-31 20:29:54 <psykose> yes
2022-01-31 20:30:02 <psykose> just file an issue and get some discussion going
2022-01-31 20:30:10 <psykose> then it should be as simple as amending abuild to look in two places
2022-01-31 20:30:20 <psykose> and just not doing the force moving in the future
2022-01-31 20:30:20 <donoban> on abuild repositroy or better on aports?
2022-01-31 20:30:38 <psykose> either, but i would say aports is more likely to not be ignored
2022-01-31 20:30:43 <ikke> donoban: abuild repo
2022-01-31 20:30:51 <donoban> ok
2022-01-31 20:31:30 <ikke> ncopa typically looks at abuild MRs before releases
2022-01-31 20:31:38 <psykose> ah
2022-01-31 20:31:39 <psykose> makes sense
2022-01-31 20:31:53 <ikke> donoban: make sure to include tests
2022-01-31 20:32:17 <eschwartz> psykose: the lasan issue is a known issue, see https://github.com/mesonbuild/meson/issues/8283 -- you will want to skip test_pch_with_address_sanitizer on musl for now
2022-01-31 20:32:23 <donoban> hM, do you mean a MR for fixing this on abuild?
2022-01-31 20:32:31 <ikke> yes
2022-01-31 20:32:43 <psykose> eschwartz: how does one actually skip any of the tests? i couldn't find an -arg for it
2022-01-31 20:32:47 <donoban> I can try to do it but I feel that I tried to send a MR to it on the past and I didn't see any button for it
2022-01-31 20:32:58 <psykose> in the ~short time i looked
2022-01-31 20:33:03 <psykose> my apologies if it's really simple 
2022-01-31 20:33:13 <ikke> donoban: just like you would make MRs for aports
2022-01-31 20:33:21 <ikke> fork the repo, push a branch to it, make an MR
2022-01-31 20:33:34 <donoban> e.g.
2022-01-31 20:33:37 <donoban> https://gitlab.alpinelinux.org/donoban/abuild/-/tree/addcache
2022-01-31 20:33:57 <donoban> I don't see any MR button
2022-01-31 20:33:59 <ikke> donoban: go to the merge reqeusts page
2022-01-31 20:34:04 <ikke> https://gitlab.alpinelinux.org/donoban/abuild/-/merge_requests
2022-01-31 20:34:08 <ikke> There is a blue button on the right
2022-01-31 20:34:18 <donoban> ah
2022-01-31 20:34:20 <donoban> wait
2022-01-31 20:34:28 <donoban> I have a MR to my own repository
2022-01-31 20:34:30 <donoban> hehe
2022-01-31 20:34:33 <ikke> oh haha
2022-01-31 20:34:36 <ikke> yeah, that does not help
2022-01-31 20:34:45 <eschwartz> you will also want for complete test coverage, a wide range of checkdepends on boost, vala, rust, mono, qt5-base, cmake, cuda, ldc, cython...
2022-01-31 20:35:08 <eschwartz> some of these may be marked for skipping when dependencies are not found, so you can ignore those in an MVP
2022-01-31 20:35:12 <donoban> I think that on aports the interface is different
2022-01-31 20:35:23 <donoban> maybe due this? "This merge request is from a private project to a public project.
2022-01-31 20:35:24 <donoban> Review the target project before submitting to avoid exposing private changes."
2022-01-31 20:35:31 <ikke> donoban: I just made the fork public
2022-01-31 20:35:51 <ikke> for aports we have something that does that automatically for forks
2022-01-31 20:35:54 <donoban> well, I'm gonna try to open the other MR
2022-01-31 20:36:05 <eschwartz> seems like the cmake and LLVM tests aren't skippable at least, there's a bunch of issues in the pipeline I just peeked at
2022-01-31 20:36:19 <donoban> ah ok , I thought that I just doesn't have some privilege or permission
2022-01-31 20:37:06 <eschwartz> psykose: py3-pytest is NOT needed, you can run the testsuite with the stdlib unittest module, but... you get better output if pytest is installed. That being said, you need pytest-xdist too, or it will just use unittest anyway (and ignore pytest)
2022-01-31 20:37:39 <donoban> https://gitlab.alpinelinux.org/alpine/abuild/-/merge_requests/133
2022-01-31 20:38:31 <donoban> ok, I will check the fish-completion issue
2022-01-31 20:38:44 <omni> https://pkgs.alpinelinux.org/contents?file=curl.fish&path=%2Fusr%2Fshare%2Ffish%2Fcompletions&name=&branch=edge
2022-01-31 20:39:29 <psykose> eschwartz:  ah, makes sense. sure, made a few amendments. -xdist is not possible to integrate as it's not also in main/, and a few of the language suites as well for either the same reason or because of circular dependencies, but i enabled a few more
2022-01-31 20:39:39 <psykose> how would i skip test_pch_with_address_sanitizer from inside the check() invocation
2022-01-31 20:42:29 <omni> ah, it's known !30280
2022-01-31 20:45:34 <eschwartz> psykose: with pytest it is easy, ./run_unittests.py 'not test_pch_with_address_sanitizer', but sadly I don't think unittest *has* a way to do exclusions (that is one of the reasons pytest is so awesome)
2022-01-31 20:45:46 <eschwartz> so, you could add a patch to remove it from the file
2022-01-31 20:46:01 <psykose> sure thing
2022-01-31 20:46:04 <eschwartz> unittests/linuxliketests.py
2022-01-31 20:46:17 <psykose> i'll do either that or see if i can move the dependencies up into main/
2022-01-31 20:46:33 <psykose> thanks!
2022-01-31 20:47:32 <eschwartz> no problem -- once the current pipelines finish I'll take another look
2022-01-31 20:49:12 <donoban> how can I pass a custom option for abuild-meson?
2022-01-31 20:49:24 <psykose> -Dsomething=abc ?
2022-01-31 20:49:39 <psykose> you can look at a lot of the existing uses as an example
2022-01-31 20:49:52 <donoban> ok let me check
2022-01-31 20:50:22 <eschwartz> psykose: ModuleNotFoundError: No module named 'jsonschema'
2022-01-31 20:50:32 <donoban> ah I see, thanks psykose 
2022-01-31 20:50:36 <eschwartz> this should be a skipTest, unless CI=... in the env
2022-01-31 20:51:08 <psykose> it.. might be
2022-01-31 20:51:09 <psykose> not sure
2022-01-31 20:51:11 <eschwartz> in theory if $CI is set, we assume it is running the @mesonbuild CI on a push/PR to meson itself
2022-01-31 20:51:22 <ikke> gitlab sets CI
2022-01-31 20:51:28 <psykose> and can't add py3-jsonsschema due to it being in community
2022-01-31 20:51:32 <psykose> alright, one more patch
2022-01-31 20:51:36 <psykose> when i get to it
2022-01-31 20:51:40 <psykose> i am working on 4 broken things at once
2022-01-31 20:51:40 <psykose> sigh
2022-01-31 20:51:52 <eschwartz> this particular test is checking that we didn't break our own schema :D it serves no practical value for packagers to check that too
2022-01-31 20:51:54 <ikke> psykose: don't get overworked :-)
2022-01-31 20:52:02 <psykose> :p
2022-01-31 20:52:47 <eschwartz> although... I think we have MESON_CI_JOBNAME always set in our CI workers, so hmm
2022-01-31 20:54:25 <eschwartz> you know what, let me change that
2022-01-31 21:05:04 <eschwartz> psykose: https://github.com/mesonbuild/meson/pull/9919 here, I saved you the effort of fixing this one
2022-01-31 21:07:04 <psykose> thanks!
2022-01-31 22:37:24 <omni> !29576 could be closed
2022-01-31 22:38:56 <psykose> done
2022-01-31 23:39:06 <omni> these too !20157 !24941 !27080 
2022-01-31 23:43:07 <psykose> done
2022-01-31 23:44:22 <eschwartz> psykose: huh
2022-01-31 23:44:24 <eschwartz> AssertionError: 'ninja: nothing to do' not found in ['ninja: no work to do.', 'samu: nothing to do']
2022-01-31 23:44:42 <eschwartz> so samurai seems to be printing `ninja: <msg>` ?
2022-01-31 23:44:55 <eschwartz> intriguing...
2022-01-31 23:45:23 <psykose> yes, afaik it's identical in output
2022-01-31 23:45:28 <eschwartz> also the next test failure I discovered is for something that needs python3-dev installed
2022-01-31 23:45:29 <psykose> oh
2022-01-31 23:45:30 <psykose> right
2022-01-31 23:45:31 <psykose> not identical
2022-01-31 23:45:35 <psykose> hm
2022-01-31 23:45:42 <psykose> this sounds like a samurai bug
2022-01-31 23:45:57 <psykose> sure, i will add py3-dev
2022-01-31 23:46:10 <eschwartz> ah right, you have /usr/bin/ninja == samurai, but I guess samurai prints argv[0] there rather than 'samu'
2022-01-31 23:46:35 <psykose> that makes sense
2022-01-31 23:46:39 <psykose> (in why it does it)
2022-01-31 23:46:51 <psykose> yeah, we have just have it named as ninja for most uses
2022-01-31 23:47:12 <psykose> but i do think it should be printing 'no work to do.' to match upstream here
2022-01-31 23:47:18 <psykose> or well '''upstream'''
2022-01-31 23:47:50 <eschwartz> do you have /usr/bin/samu as a link to ninja?
2022-01-31 23:48:09 <psykose> ninja -> samu
2022-01-31 23:48:12 <eschwartz> because if so, meson's unittests can use NINJA=/usr/bin/samu and it will run that directly
2022-01-31 23:48:15 <psykose> sure
2022-01-31 23:48:16 <psykose> i can add that
2022-01-31 23:48:21 <eschwartz> instead of finding /usr/bin/ninja first :p
2022-01-31 23:48:41 <psykose> amended
2022-01-31 23:49:41 <psykose> one by one the little failures get fixed
2022-01-31 23:49:42 <psykose> ^~^
2022-01-31 23:59:28 <omni> !26695 already in by another merge