2023-02-01 02:34:54 <psykose> ikke: don't suppose you have any ideas for how to debug the hanging rust in CI only for armhf, since you tried it by hand and it did not..
2023-02-01 05:44:20 <ikke> psykose: perhaps strace the process while the job is running
2023-02-01 05:45:08 <psykose> sure, you can start any ci job for anything rust and connect to running for that
2023-02-01 08:51:26 <PabloCorreaGomez[m]> psykose: regarding evince. Yes, that happens because of this patch: https://gitlab.alpinelinux.org/alpine/aports/-/blob/master/community/evince/window-Port-it-to-phones.patch Which changes the pane for a flap
2023-02-01 08:51:49 <psykose> ah
2023-02-01 08:51:56 <PabloCorreaGomez[m]> No idea how to fix it more than removing the said patch, to be fair
2023-02-01 08:52:03 <psykose> so we've had a broken evince this whole time :p
2023-02-01 08:52:58 <PabloCorreaGomez[m]> But if you think that should be done, then do it. I was hoping to be able to discuss with the maintainer at FOSDEM. But it's not gonna be there... So not sure what is the way forwards
2023-02-01 08:53:03 <PabloCorreaGomez[m]> forward*
2023-02-01 08:54:00 <psykose> that just breaks it for mobile instead does it not
2023-02-01 08:54:40 <PabloCorreaGomez[m]> Yes. Although it only breaks the side pane. But since you're the upstream, you choose :P
2023-02-01 08:55:58 <PabloCorreaGomez[m]> For me this has been a very frustrating process with the evince maintainership. There's me an another person spending a huge amount of effort to cleanup code, move the port to GTK4 forward, and hopefully make it better. But they don't even have time to review simple patches
2023-02-01 08:56:44 <PabloCorreaGomez[m]> So not really sure how shall be proceed. Now, there's not really a good solution. Hopefully I manage to find people at FOSDEM from GNOME that can help
2023-02-01 09:01:41 <psykose> i'm ok with waiting
2023-02-01 09:58:44 <PabloCorreaGomez[m]> Thanks :)
2023-02-01 11:42:01 <ddevault> I'm going to be making a lot of noise in !43750 while I try to patch around this bug without being able to reproduce it locally
2023-02-01 11:42:03 <ddevault> be aware
2023-02-01 11:43:40 <ddevault> really wish I knew what was going on with the alpine aarch64 builders to produce this bug given that that it works fine on all of my aarch64 machines
2023-02-01 11:46:28 <psykose> or the others that reproduced it :-)
2023-02-01 11:46:32 <psykose> it is really weird though indeed
2023-02-01 11:47:02 <psykose> go nuts though, there's two parallel jobs for ci so you can't hog all of it alone in one merge request
2023-02-01 13:17:44 <staceee> hey psykose, I would like to update my maintainer comment line on every aports I maintain. Should I do one commit, or split this on one commit per package ?
2023-02-01 13:18:33 <psykose> your preference, though i'd say 1 */*: update contact for xyz makes more sense
2023-02-01 13:20:26 <staceee> psykose: "xyz" would be ?
2023-02-01 13:22:44 <staceee> I guess my username
2023-02-01 13:22:59 <psykose> sure
2023-02-01 13:23:07 <psykose> also needs pkgrel
2023-02-01 13:25:30 <staceee> ah right
2023-02-01 13:38:31 <psykose> hmm
2023-02-01 13:38:33 <psykose> https://build.alpinelinux.org/buildlogs/build-edge-x86_64/community/bonsai/bonsai-1.0.0-r0.log weird
2023-02-01 13:39:24 <psykose> ah was just parallelism
2023-02-01 13:41:13 <staceee> psykose: ah ya it blinked also on my terminal
2023-02-01 13:41:24 <staceee> dunno why o_O
2023-02-01 13:42:44 <staceee> psykose: what are the rules in regard of testing/ ? how long a recipe should live here before moving to community/ ? What justifies a move to community/ ?
2023-02-01 13:51:38 <psykose> no time limits whenever you want
2023-02-01 13:51:48 <psykose> but really just add it there then later add it for the other thing
2023-02-01 14:48:11 <lattis>  
2023-02-01 16:54:39 <bl4ckb0ne> anybody here attempted to package tracy?
2023-02-01 16:58:42 <psykose> https://github.com/wolfpld/tracy ?
2023-02-01 17:06:41 <psykose> sure here u go https://img.ayaya.dev/a2IDUFgfsyaa
2023-02-01 17:13:40 <bl4ckb0ne> seems ti be suffering from the half melted pottery build system syndrome
2023-02-01 17:13:53 <psykose> it is the easiest shit i have ever used
2023-02-01 17:14:07 <bl4ckb0ne> https://github.com/wolfpld/tracy/tree/master/profiler/build/unix
2023-02-01 17:14:13 <bl4ckb0ne> check again
2023-02-01 17:14:20 <bl4ckb0ne> also check the aur build
2023-02-01 17:14:37 <psykose> ah, frontend
2023-02-01 17:14:55 <bl4ckb0ne> yeah
2023-02-01 17:16:35 <psykose> ok i built it
2023-02-01 17:16:36 <psykose> that was ez
2023-02-01 17:17:03 <bl4ckb0ne> no issue with ft2build?
2023-02-01 17:17:12 <psykose> nope
2023-02-01 17:17:21 <ikke> https://www.youtube.com/watch?v=OynKbfEGV_A
2023-02-01 17:17:21 <bl4ckb0ne> mst be on my end then
2023-02-01 17:17:22 <bl4ckb0ne> ty
2023-02-01 17:17:34 <psykose> there is a freetype2-dev you most likely hath forgotten
2023-02-01 17:17:52 <bl4ckb0ne> no i got it
2023-02-01 17:18:03 <bl4ckb0ne> something fails at the dependency resolve mess
2023-02-01 17:18:33 <bl4ckb0ne> https://github.com/wolfpld/tracy/blob/master/common/unix.mk#L46
2023-02-01 17:18:38 <bl4ckb0ne> cloned master?
2023-02-01 17:20:49 <psykose> https://img.ayaya.dev/1PKKRNWOcbyg
2023-02-01 17:20:50 <psykose> yeah
2023-02-01 17:26:09 <bl4ckb0ne> welp
2023-02-01 17:26:16 <bl4ckb0ne> i just suck
2023-02-01 17:27:35 <psykose> it's half melted but enjoy https://img.ayaya.dev/AJyYH0G8cgVV
2023-02-01 17:29:28 <bl4ckb0ne> thx fam
2023-02-01 18:10:34 <bl4ckb0ne> huh, was missing capstone-dev
2023-02-01 18:10:45 <bl4ckb0ne> but it was yelling about freetype
2023-02-01 18:15:46 <psykose> if you miss any it skips every -I and freetype is the first failure as a result
2023-02-01 18:16:04 <psykose> but right at the very top it tells you what it missed, so
2023-02-01 18:16:20 <psykose> it's a trainwreck but like 1.5/10 difficulty beginner course one
2023-02-01 18:16:26 <psykose> freshman "building stuff" exam
2023-02-01 18:16:55 <psykose> you've earned an E, see me after class.
2023-02-01 18:17:37 <psykose> also in that example the wayland/mesa are unused but they are used on master, i just added them for both :-)
2023-02-01 18:17:42 <psykose> if you want i can actually commit it
2023-02-01 18:40:10 <bl4ckb0ne> have I failed aports class?
2023-02-01 18:50:04 <psykose> i'm trying to think of a cool zoomer response but i'm too flu'd out
2023-02-01 18:51:07 <bl4ckb0ne> i can try to do a tik tok dance to make you feel better
2023-02-01 18:51:28 <bl4ckb0ne> but my knees arent what they used to be
2023-02-01 18:51:54 <psykose> i've never even been anywhere near tiktok tbh
2023-02-01 18:59:34 <bl4ckb0ne> havent been near that botnet too but zoomers seems to like it too much
2023-02-01 19:05:46 <invoked> feel better psykose. physician, heal thyself
2023-02-01 19:07:29 <psykose> thanks :-)
2023-02-01 19:43:20 <EvTheFuture> If anyone have the time to checkout !43697 to give feedback/merge it would be fantastic :)
2023-02-01 20:52:47 <ikke> Go 1.20 released: https://go.dev/doc/go1.20
2023-02-01 21:27:11 <elibrokeit> bl4ckb0ne: I love how I've named an entirely new class of software woes w.r.t. pottery buildsystems
2023-02-01 21:34:37 <bl4ckb0ne> gz
2023-02-01 21:34:49 <bl4ckb0ne> it shall be known across the distros now
2023-02-01 21:35:08 <bl4ckb0ne> we should carry the word around
2023-02-01 21:35:10 <bl4ckb0ne> godspeed!
2023-02-01 21:46:04 <staceee> someone got a doc about how to crosscompile with abuild ? I want to build a package from x86_64 to aarch64
2023-02-01 21:48:33 <psykose> generally you cannot and you don't
2023-02-01 21:48:51 <psykose> realistically i think it's something like CHOST=aarch64 abuild
2023-02-01 21:48:56 <psykose> almost nothing will work with that though
2023-02-01 21:49:44 <bl4ckb0ne> why does it even exist
2023-02-01 21:49:58 <staceee> psykose: yap but it ask for build-base-aarch64
2023-02-01 21:50:09 <staceee> which, looks like a ghost, a souvenir
2023-02-01 21:50:20 <psykose> because you need a cross version of every single revdep
2023-02-01 21:50:26 <psykose> that's how cross works in general
2023-02-01 21:50:27 <clandmeter> staceee: why not use qemu-user?
2023-02-01 21:50:35 <psykose> the easiest way is to get a minirootfs
2023-02-01 21:50:37 <psykose> unpack it
2023-02-01 21:50:45 <staceee> your the second person that asked about qemu-user. Is that a secte ?
2023-02-01 21:50:45 <psykose> (of the chost arch, i.e. aarch64)
2023-02-01 21:50:48 <psykose> set up qemu-user
2023-02-01 21:50:53 <psykose> and chroot into the minirootfs 
2023-02-01 21:50:55 <psykose> and do stuff in it
2023-02-01 21:51:17 <psykose> bl4ckb0ne: bootstrapping to a new arch
2023-02-01 21:51:22 <clandmeter> i think rootbld should work with qemu-user
2023-02-01 21:51:40 <staceee> mhh, minirootfs, qemu-user. Is there a doc somewhere ?
2023-02-01 21:51:41 <psykose> it also gets re-ran here and there and keeps the base bootstrap clean, since it's not far off from the base cross support
2023-02-01 21:52:12 <psykose> don't think there's docs but if you've used chroot() and you know how qemu-user works there is nothing more to it
2023-02-01 21:52:29 <psykose> it's like any other cross-arch root
2023-02-01 21:52:33 <clandmeter> it should be as simple as setting up rootbld and set the arch to something else, iirc.
2023-02-01 21:52:54 <psykose> e.g. you can chroot to the rootfs sdcard of an aarch64 sbc plugged into your x86_64 computer, if you set up qemu-user
2023-02-01 21:52:58 <clandmeter> and or course setup qemu
2023-02-01 21:53:04 <staceee> I know chroots, and I think I understand qemu enough
2023-02-01 21:53:04 <psykose> done that plenty of times for recovery
2023-02-01 21:53:08 <psykose> to setup up user, you do
2023-02-01 21:53:15 <psykose> apk add qemu-aarch64 qemu-openrc
2023-02-01 21:53:21 <staceee> that is done
2023-02-01 21:53:21 <psykose> service qemu-binfmt start
2023-02-01 21:53:23 <psykose> more or less
2023-02-01 21:53:29 <staceee> ah right, this black magic
2023-02-01 21:53:49 <psykose> if just writes to the binfmt sys thing to register qemu
2023-02-01 21:54:24 <staceee> so running an aarch64 binary and qemu-bindfmt should do its trick
2023-02-01 21:54:27 <psykose> and then chroot should work, but remember to mount stuff (i.e. dev, you know, fullchroot)
2023-02-01 21:54:30 <psykose> yeah
2023-02-01 21:54:32 <staceee> so you say I just chroot /bin/sh and hop
2023-02-01 21:54:40 <psykose> pretty much
2023-02-01 21:55:03 <staceee> it would be perfect if abuild could do all of this by itself
2023-02-01 21:55:24 <clandmeter> it can
2023-02-01 21:56:16 <staceee> rootbld ? okay there is doc about it. Nice
2023-02-01 21:57:14 <clandmeter> not sure there is a doc how to run it with qemu-user, but you can always rtfs
2023-02-01 21:58:13 <psykose> warning: rtfs in question is known to the state of california to cause ..
2023-02-01 21:59:21 <staceee> ah I understood where to find mini root fs images. By bad
2023-02-01 22:02:04 <psykose> you can also use that small indie portable minirootfs project
2023-02-01 22:02:09 <psykose> i think docker it was called
2023-02-01 22:02:11 <psykose> :-)
2023-02-01 22:02:22 <psykose> (works with qemu-user all the same)
2023-02-01 22:03:01 <staceee> aha !
2023-02-01 22:03:49 <staceee> badaboom, chrooted as aarch64
2023-02-01 22:04:58 <clandmeter> with abuild-rootbld qemu-openrc and qemu-aarch64 installed it should be as simple as: CBUILD=aarch64 abuild rootbld
2023-02-01 22:05:45 <staceee> wow
2023-02-01 22:06:00 <psykose> isn't it CHOST? CBUILD is just the current machine
2023-02-01 22:06:34 <clandmeter> i just checked my history as i used it long long time ago
2023-02-01 22:06:35 <staceee> > rootbld: set CBUILD=aarch64-alpine-linux-musl to build for aarch64-alpine-linux-musl
2023-02-01 22:07:05 <psykose> first line of rootbld does [ $CBUILD = $CHOST ] || die "rootbld: set CBUILD=$CHOST to build for $CHOST", which i guess makes sense for qemu
2023-02-01 22:07:41 <psykose> so here it makes sense, in cross compiling cbuild=not-your-computer doesn't really make sense though
2023-02-01 22:08:09 <psykose> not that it's invalid if you have a qemu binfmt, just weird
2023-02-01 22:08:10 <clandmeter> and it works as i just build zip for aarch64 on our AMD EPYC
2023-02-01 22:10:22 <clandmeter> yes it says it want to the whole triplet, but i think abuild will resolve the right one for just setting aarch64
2023-02-01 22:10:39 <psykose> yeah, and make cbuild=chost since it's qemu, etc
2023-02-01 22:10:47 <psykose> i just forgot it was qemu initially
2023-02-01 22:11:19 <clandmeter> its really cool this works so simple 
2023-02-01 22:12:11 <staceee> mhh okay seems to works. I built pnc with this
2023-02-01 22:13:14 <psykose> it fails if the arch dir doesn't exist it seems
2023-02-01 22:13:20 <clandmeter> yep
2023-02-01 22:13:25 <clandmeter> not sure why that is
2023-02-01 22:13:32 <staceee> So, abuild rootbld, it make a sysfs somewhere, re-install every dependency in it, chroot and build inside. qemu-bindfmt make this fluid
2023-02-01 22:13:33 <clandmeter> but creating it fixes it
2023-02-01 22:14:16 <staceee> where is the fs ?
2023-02-01 22:14:21 <clandmeter> rootbld uses bubblewrap
2023-02-01 22:15:28 <psykose> var/tmp/abuild.xxxx
2023-02-01 22:15:35 <clandmeter> it should be documented in the source :)
2023-02-01 22:15:59 <clandmeter> thats the advantage of using alpine
2023-02-01 22:16:01 <clandmeter> no need for man
2023-02-01 22:16:09 <staceee> okay all of this is very sexy
2023-02-01 22:16:25 <psykose> people usually run screaming
2023-02-01 22:16:37 <psykose> maybe you're a fetishist of some kind?
2023-02-01 22:16:39 <staceee> so "CBUILD=aarch64 abuild rootbld" actually is the command line I dreamed for
2023-02-01 22:16:45 <psykose> seems so
2023-02-01 22:16:45 <staceee> I am
2023-02-01 22:16:47 <clandmeter> only stubborn ppl stay here
2023-02-01 22:16:48 <psykose> i forgot it existed
2023-02-01 22:26:15 <staceee> Okay now I want to know how easy it is to use this in builds.sr.ht. But that's another story!
2023-02-01 22:26:16 <staceee> Thanks both of you for the help !
2023-02-01 22:26:30 <staceee> minor question. which repos abuild rootbld is using ?
2023-02-01 22:29:25 <clandmeter> the ones you have in repos i guess
2023-02-01 22:31:37 <staceee> oh right, I thought there was scoped to one arch
2023-02-02 07:53:43 <PureTryOut> psykose: that package build failure with SDL I was talking about is https://gitlab.alpinelinux.org/alpine/aports/-/merge_requests/42718
2023-02-02 15:06:14 <psykose> PureTryOut: if you read the logs you will see it's ambiguous because `u64` is defined as `typedef uint64_t u64;` in directfb and as `typedef long long unsigned int irr::u64` in supertuxkart source
2023-02-02 15:06:25 <psykose> this is not a directfb or sdl issue
2023-02-02 15:08:03 <PureTryOut> ah. I'm not that versed into C sorry I didn't realize stk defined it itself as well
2023-02-02 15:08:13 <PureTryOut> (well irrlicht then)
2023-02-02 15:32:09 <bl4ckb0ne> cursed
2023-02-02 15:32:18 <bl4ckb0ne> when are we yeeting directfb psykose 
2023-02-02 15:32:25 <psykose> probably never at this rate
2023-02-02 15:35:14 <bl4ckb0ne> https://github.com/libsdl-org/SDL/commit/3aea865cdf4c0802ba045bc095bb6084302c0186 tho
2023-02-02 15:38:00 <psykose> that's not sdl2
2023-02-02 15:39:28 <bl4ckb0ne> wdym?
2023-02-02 15:40:08 <bl4ckb0ne> oh sdl3
2023-02-02 16:54:28 <bl4ckb0ne> is there a reason why elfutils doesnt package debuginfod.h?
2023-02-02 16:57:50 <psykose> debuginfod is disabled so the header is also missing
2023-02-02 16:59:19 <bl4ckb0ne> why is it disabled?
2023-02-02 17:00:43 <psykose> because it's not enabled
2023-02-02 17:00:45 <psykose> what do you need it for
2023-02-02 17:04:18 <q66> i wonder how much it would take to implement debuginfod in elftoolchain
2023-02-02 17:04:33 <psykose> there's an llvm-debuginfod anyway
2023-02-02 17:05:01 <psykose> unless you wanna avoid that :p
2023-02-02 17:06:56 <q66> well probably not 
2023-02-02 17:07:02 <q66> i did not know that existed 
2023-02-02 17:07:06 <q66> it seems it's not in 15
2023-02-02 17:07:37 <q66> i guess there is no need for the elftoolchain tools or for libelf itself to implement it 
2023-02-02 17:08:00 <q66> the debugger has to have the client part and the server has to have the server part 
2023-02-02 17:08:25 <q66> i guess the llvm one is integrated with lldb?
2023-02-02 17:10:39 <q66> it seems lldb uses libdebuginfod from elfutils if available 
2023-02-02 17:10:44 <q66> so yeah, no
2023-02-02 17:10:57 <q66> would need a client library in elftoolchain anyway 
2023-02-02 17:11:03 <psykose> it is in llvm15
2023-02-02 17:11:34 <q66> it only implements the server 
2023-02-02 17:11:44 <psykose> ah, sure
2023-02-02 17:12:00 <q66> well, another thing to look into someday 
2023-02-02 17:12:31 <q66> it's not super important so not today or tomorrow or in near future 
2023-02-02 17:12:33 <q66> but someday 
2023-02-02 17:39:45 <vkrishn> any plans to upgrade wiki.a.o ?
2023-02-02 17:40:27 <ikke> upgrade what?
2023-02-02 17:41:31 <vkrishn> wiki.alpinelinux.org
2023-02-02 17:41:42 <psykose> to what
2023-02-02 17:41:43 <pj> but what
2023-02-02 17:41:43 <ikke> I know, but what do you think should be upgraded
2023-02-02 17:41:46 <ikke> :P
2023-02-02 17:41:52 <pj> upgrade content /s
2023-02-02 17:42:10 <vkrishn> new mobile friendly pages, like wikipedia.org
2023-02-02 17:42:19 <ikke> vkrishn: you can set the theme yourself
2023-02-02 17:42:23 <psykose> the existing pages work on mobile
2023-02-02 17:42:30 <vkrishn> how ?
2023-02-02 17:42:46 <ikke> Preferences
2023-02-02 17:42:55 <ikke> Appearance
2023-02-02 17:42:57 <ikke> Vector (2022)
2023-02-02 17:42:58 <vkrishn> ah,  think i know
2023-02-02 17:43:00 <vkrishn> ok
2023-02-02 17:44:54 <pj> those old icons definitely look great with new theme
2023-02-02 17:45:14 <ikke> I know, right
2023-02-02 17:46:05 <psykose> to be fair it still looks terrible on mobile even with vector-2022
2023-02-02 17:46:11 <psykose> but it is what it is
2023-02-02 17:46:19 <psykose> if you know of what theme to install feel free to tell me
2023-02-02 17:48:56 <ptrc> psykose: do we have minerva neue on wiki.a.o?
2023-02-02 17:49:07 <ikke> Is that a font?
2023-02-02 17:49:14 <ptrc> https://github.com/wikimedia/mediawiki-skins-MinervaNeue
2023-02-02 17:49:17 <ptrc> a skin
2023-02-02 17:49:19 <psykose> we have minerva neue at home
2023-02-02 17:49:24 <psykose> minerva neue at home:
2023-02-02 17:50:33 <ptrc> oooh
2023-02-02 17:50:54 <ptrc> the citizen theme looks kinda nice
2023-02-02 17:51:38 <psykose> i added that as a shitpost i think
2023-02-02 17:51:41 <psykose> and yeah minerva is there now
2023-02-02 17:52:56 <psykose> minerva is actually usable on mobile
2023-02-02 17:53:02 <psykose> maybe is should default to it
2023-02-02 17:53:15 <ptrc> that's what m.wikipedia.org defaulted to for a long while
2023-02-02 17:53:21 <ptrc> not sure if that still works these days
2023-02-02 17:53:35 <psykose> it's kinda bad on desktop tho
2023-02-02 18:05:26 <vkrishn> there is a adjutable dark-mode extension for firefox, kinda cool
2023-02-02 18:06:54 <bl4ckb0ne> psykose: tracy test
2023-02-02 18:07:18 <bl4ckb0ne> i havent tried to disable it (lunchtime) but ill see how far i can go
2023-02-02 18:09:55 <psykose> tracy test?
2023-02-02 18:10:56 <bl4ckb0ne> yeah
2023-02-02 18:11:09 <bl4ckb0ne> trying to improve your half melted pottery of an APKBUILD :D
2023-02-02 18:13:21 <bl4ckb0ne> eh it just hangs
2023-02-02 18:13:32 <bl4ckb0ne> 
ACTION slaps a !check in the options

2023-02-02 18:13:39 <psykose> no, works fine
2023-02-02 18:13:51 <psykose> https://img.ayaya.dev/RW73ybZWhPPH
2023-02-02 18:13:54 <vkrishn> psykose: al logo on wiki.a.o ?
2023-02-02 18:13:57 <psykose> the tests are broken though
2023-02-02 18:15:40 <bl4ckb0ne> ugh cpp linking time
2023-02-02 18:16:02 <bl4ckb0ne> works on your end? whut
2023-02-02 18:17:25 <psykose> it's just make -C test
2023-02-02 18:20:01 <bl4ckb0ne> yeah it builds
2023-02-02 18:20:20 <psykose> and fails to run
2023-02-02 18:20:23 <bl4ckb0ne> but tracy_test fails
2023-02-02 18:20:35 <psykose> i would try master but i don't care too much
2023-02-02 18:20:49 <psykose> their own alignment check assertion fails
2023-02-02 18:22:38 <bl4ckb0ne> half melted software smh
2023-02-02 18:25:33 <psykose> vkrishn: it's set, the default theme just doesn't display logos
2023-02-02 18:25:37 <psykose> i don't care about it either
2023-02-02 18:29:08 <vkrishn> ok with me, maybe it needs smaller one
2023-02-02 18:38:48 <bl4ckb0ne> psykose: you're saying that renderdoc 1.22 crash?
2023-02-02 18:38:56 <psykose> before the upgrade it didn't
2023-02-02 18:38:58 <psykose> i.e. current
2023-02-02 18:38:59 <psykose> afaik
2023-02-02 18:39:07 <psykose> vkrishn: smaller one? idk what you mean
2023-02-02 18:39:14 <psykose> if you change the theme you'll see a logo
2023-02-02 18:39:18 <psykose> there is an svg version i added
2023-02-02 18:39:23 <psykose> ptrc cropped it into just the logo
2023-02-02 18:39:27 <psykose> but the default theme does not show it
2023-02-02 18:39:30 <bl4ckb0ne> 1.22 crashes too
2023-02-02 18:41:06 <psykose> typical 'simon zeni' not testing anything
2023-02-02 18:41:09 <vkrishn> idk, would try to install the theme locally and resize to png, haven't played to mediawiki for yrs
2023-02-02 18:41:17 <psykose> what?
2023-02-02 18:41:25 <vkrishn> played with^
2023-02-02 18:41:33 <psykose> what is 'resize to png'?
2023-02-02 18:41:34 <bl4ckb0ne> D:
2023-02-02 18:41:36 <psykose> svgs can be any size
2023-02-02 18:41:39 <bl4ckb0ne> be kind, im redoing my aports class
2023-02-02 18:41:44 <vkrishn> convert to png
2023-02-02 18:41:48 <psykose> there is a png version too
2023-02-02 18:41:55 <vkrishn> ok
2023-02-02 18:41:56 <psykose> nor does that do anything
2023-02-02 18:42:01 <psykose> the theme does not display the icon
2023-02-02 18:42:06 <psykose> please read what i am saying
2023-02-02 18:42:37 <psykose> png icons are usually just fallbacks for things that can't render svg
2023-02-02 18:43:18 <psykose> (now if you know how to configure vector to display an icon, feel free to tell me)
2023-02-02 18:43:24 <psykose> personally i don't mind if the icon is there or not
2023-02-02 18:47:53 <vkrishn> difference, https://tpaste.us/ggm5 , theme probably needs settings changes
2023-02-02 18:50:10 <psykose> as i said, if you know what to change tell me
2023-02-02 18:50:25 <psykose> if you don't i don't care what the defaults are
2023-02-02 18:50:37 <vkrishn> ok
2023-02-02 18:50:44 <psykose> i think it looks of without the logo
2023-02-02 18:50:51 <psykose> it is just a small nicety :-)
2023-02-02 18:50:55 <psykose> looks ok*
2023-02-02 18:51:16 <psykose> there might be a config somewhere that does not require patching the theme, as i'd prefer to not patch anything
2023-02-02 18:55:20 <bl4ckb0ne> psykose: trying vulkan 1.3.240
2023-02-02 18:55:33 <psykose> i would be a couple million that is not the issue
2023-02-02 18:55:37 <psykose> bet*
2023-02-02 18:55:53 <bl4ckb0ne> i suspect the loader
2023-02-02 18:56:32 <bl4ckb0ne> https://paste.sr.ht/~bl4ckb0ne/14c64f46f34cef246214d7eda564338092680538 stacktrace
2023-02-02 18:57:28 <psykose> forgot the debug symbols
2023-02-02 18:57:59 <bl4ckb0ne> > 0x00007fffe50a98fc in linux_read_sorted_physical_devices (inst=inst@entry=0x7fffe313c730, icd_count=icd_count@entry=4, icd_devices=icd_devices@entry=0x7fffe526d190, phys_dev_count=phys_dev_count@entry=4, sorted_device_term=sorted_device_term@entry=0x7fffe1d36ab0) at /home/buildozer/aports/main/vulkan-loader/src/Vulkan-Loader-sdk-1.3.239.0/loader/loader_linux.c:282
2023-02-02 18:58:03 <bl4ckb0ne> told you
2023-02-02 18:59:49 <psykose> that doesn't point to vulkan specifically
2023-02-02 19:02:30 <bl4ckb0ne> cmake build type is RelWithDebInfo or MinSizeRel?
2023-02-02 19:02:47 <psykose> of what where
2023-02-02 19:02:54 <bl4ckb0ne> generally
2023-02-02 19:03:05 <psykose> generally it's None unless it doesn't work
2023-02-02 19:03:10 <psykose> the build types don't mean anything
2023-02-02 19:03:12 <bl4ckb0ne> vulkan-* has None and Release and MinSizeRel
2023-02-02 19:03:36 <psykose> except for what they do mean, but i got tired of explaining the difference
2023-02-02 19:03:39 <psykose> not relevant to anything
2023-02-02 19:03:53 <bl4ckb0ne> 
ACTION shrugs

2023-02-02 19:03:53 <alpine-meetbot> ¯\_(ツ)_/¯
2023-02-02 19:08:32 <bl4ckb0ne> soooo
2023-02-02 19:08:39 <bl4ckb0ne> bumping vk doesnt magically fix everything
2023-02-02 19:08:40 <bl4ckb0ne> :D
2023-02-02 19:09:23 <bl4ckb0ne> you still want the vulkan-* bump?
2023-02-02 19:09:48 <psykose> there is no new sdk
2023-02-02 19:09:59 <bl4ckb0ne> no its just a random bump
2023-02-02 19:10:26 <psykose> my "there is no new sdk" statement is getting a lot of questions answered by it :p
2023-02-02 19:11:19 <psykose> the next one is probably in march
2023-02-02 19:32:48 <q66> build types only affect default (c(xx)|ld)flags, which cmake injects after any other flags including env-supplied one in a very galaxy brain way 
2023-02-02 19:33:23 <bl4ckb0ne> but half of the dev override this anyway
2023-02-02 19:33:45 <q66> if you use none then it won't use any
2023-02-02 19:33:58 <q66> so pretty much everybody should always use none
2023-02-02 19:34:03 <q66> because build types suck 
2023-02-02 19:34:39 <q66> though there are projects which explicitly check "allowed" build types and do shit like put the build type name in library names 
2023-02-02 19:34:50 <q66> a move even more galaxy brain than their base existence 
2023-02-02 19:35:15 <bl4ckb0ne> revert to make
2023-02-02 19:35:15 <q66> which is why none sometimes does not work
2023-02-02 22:56:58 <elly> hm, our clang's -fsanitize=cfi does not work - it requires the presence of /usr/lib/clang/15.0.7/share/cfi_ignorelist.txt which is missing
2023-02-02 22:57:21 <elly> (try: clang -fsanitize=cfi -flto -fvisibility=default foo.c)
2023-02-02 23:01:50 <psykose> compiler-rt
2023-02-02 23:02:06 <psykose> i.e. apk add compiler-rt
2023-02-02 23:02:16 <psykose> all the sanitizers are there (and for cfi that file is)
2023-02-02 23:02:23 <elly> aha, thanks
2023-02-02 23:02:27 <elly> I didn't realize there was a separate package
2023-02-02 23:02:32 <psykose> for reference https://pkgs.alpinelinux.org/contents?file=cfi_ignorelist.txt&path=&name=&branch=edge
2023-02-02 23:03:02 <psykose> they implemented all the stuff in a separate projecet (this is llvm-runtimes), so it is how it is :)
2023-02-02 23:03:32 <psykose> aside from that it should work, i've tested most of the sanitizers
2023-02-02 23:03:43 <psykose> on x86_64 anyway
2023-02-02 23:03:53 <elly> and now /usr/lib/LLVMgold.so is missing, sigh :)
2023-02-02 23:03:56 <elly> secret deps abound
2023-02-02 23:04:02 <psykose> -fuse-ld=lld
2023-02-02 23:04:22 <elly> "invalid linker name in argument"
2023-02-02 23:04:28 <psykose> apk add lld :p
2023-02-02 23:04:47 <psykose> llvmgold is from `llvm` but it will just shim to binutils-gold iirc
2023-02-02 23:04:57 <psykose> you can't use clang -flto with binutils bfd
2023-02-02 23:05:23 <psykose> (and yes, the -flto is a requirement of cfi, i know)
2023-02-02 23:06:59 <elly> yeah, it is
2023-02-02 23:07:00 <elly> ugh
2023-02-02 23:07:28 <psykose> but basically just `apk add compiler-rt clang llvm lld` and clang -fuse-ld=lld -flto=thin .. works for everything
2023-02-02 23:07:36 <psykose> i use clang for everything in general myself
2023-02-02 23:08:40 <psykose> the only issues you might now run into is some 'unrecognised symbol in archive' for which you need AR=llvm-ar NM=llvm-nm RANLIB=llvm-ranlib in various build systems to invoke binutils that can handle the outputs
2023-02-02 23:08:45 <psykose> binutils doesn't handle them
2023-02-02 23:10:42 <q66> llvmgold is only compiled if binutils is present during llvm build
2023-02-02 23:10:52 <q66> without it lto won't work with bfd and gold
2023-02-02 23:11:16 <psykose> lto doesn't work with bfd anyway
2023-02-02 23:11:19 <psykose> thin at least
2023-02-02 23:11:42 <q66> full does
2023-02-02 23:11:48 <psykose> :-)
2023-02-02 23:11:50 <q66> thin might
2023-02-02 23:12:35 <psykose> works for the world of hello
2023-02-02 23:12:38 <psykose> weird
2023-02-02 23:13:12 <q66> as long as the compiler is clang, thin vs full does not matter at all for the linker
2023-02-02 23:13:31 <q66> the plugin just lets the linker call into the compiler
2023-02-02 23:15:44 <psykose> i am probably remembering this a little backwards as handling gcc -flto in lld :p
2023-02-02 23:15:51 <psykose> which is the thing that doesn't work
2023-02-02 23:16:32 <q66> i never tested gcc wtih lld
2023-02-02 23:16:42 <q66> they did not even allow it at all until recently
2023-02-02 23:16:57 <q66> like, gcc hardcodes allowed options for -fuse-ld
2023-02-02 23:17:02 <q66> and lld is the enemy so
2023-02-02 23:17:09 <psykose> it's been there for a while
2023-02-02 23:17:25 <q66> yeah they budged eventually
2023-02-02 23:17:37 <psykose> even mold is in the hardcoded list since 12
2023-02-02 23:17:45 <psykose> aside from that there's always the shitty -B workaround
2023-02-02 23:18:01 <psykose> or just ole ln -sf ... /usr/bin/ld
2023-02-02 23:18:17 <q66> well yes but you get the point
2023-02-02 23:18:38 <psykose> definitely not
2023-02-02 23:18:47 <psykose> i will get the point for one (1) serving of food
2023-02-02 23:20:33 <q66> should have come to brussels
2023-02-02 23:20:38 <q66> they got food
2023-02-02 23:20:53 <q66> you could be in a cramped train cabin with 5 strangers like me
2023-02-02 23:21:28 <q66> i feel like i will sleep badly on this train
2023-02-02 23:21:31 <q66> this pillow is too thin
2023-02-02 23:27:05 <psykose> cramped cabin with 5 strangers sounds miserable
2023-02-02 23:27:19 <psykose> outside food is also expensive
2023-02-02 23:27:32 <psykose> and usually not vegan so i don't even bother
2023-02-02 23:28:09 <q66> 1st class is probably comfier but
2023-02-02 23:28:12 <q66> it's expensive
2023-02-03 01:09:13 <omni> psykose: you have a typo in re2 pkgver..
2023-02-03 01:09:27 <psykose> happens
2023-02-03 01:10:16 <psykose> the luck they had a release exactly a year ago
2023-02-03 01:16:39 <omni> (=
2023-02-03 08:02:20 <maribu> I totally missed this: https://lwn.net/SubscriberLink/921787/a480ee63dfc54c3e/ I bet hell broke loose here when this happened...
2023-02-03 08:16:11 <maribu> My takeaway is that it might in the end be inevitable to somehow coupe with archives being not stable, as both git said that git archive has explicitly no guarantee for stable archives and Github also said they would guarantee stable archives. Also e.g. for sr.ht unstable archives previously were the default (I think it used the timestamp of compression instead of the timestamp of the last commit).
2023-02-03 08:18:52 <maribu> Maybe there is already a tool that can yield a single checksum covering the file names, the file permissions, and the file contents in an archive in a stable way? E.g. even if the timestamp, compression rate, compression algorithm, order the files are compressed, etc. are changed? I personally have never heard of such thing, but that might be a good fit.
2023-02-03 08:25:27 <psykose> only for a brief moment, re hell loose
2023-02-03 08:25:41 <psykose> i think i updated like 6 checksums (that i now will have to revert eventually haha)
2023-02-03 08:25:48 <psykose> they fixed it quite fast (like 12 hours i guess)
2023-02-03 08:26:57 <psykose> re stable checksum, there is (nar sums, etc, touched on in some thread somewhere), but they're not methods for security, just reproducibility
2023-02-03 08:27:28 <psykose> this thread talks about something like that https://github.com/orgs/community/discussions/45830#discussioncomment-4825608
2023-02-03 08:27:37 <psykose> regarding git archives, the current talk is to indeed make them stable
2023-02-03 08:27:59 <psykose> https://public-inbox.org/git/df7b0b43-efa2-ea04-dc5b-9515e7f1d86f@gmail.com/T/
2023-02-03 08:28:32 <psykose> personally i dislike the nar hash as well purely because it is slow as fuck
2023-02-03 08:29:10 <psykose> i don't think measures of doing things that are 20x slower as an alternative are ever good, and that's before we get to that part that it's not even the same security guarantaee
2023-02-03 08:29:50 <psykose> nor is it better in any way- the checksum guarantees the same thing :-) it's just less likely to break for being reproducible despite that changing
2023-02-03 08:41:19 <sam_> talking about this makes me want to have a stroke tbh
2023-02-03 08:41:21 <sam_> the nar thing
2023-02-03 08:41:42 <sam_> also oh god that thread has exploded since I last looked
2023-02-03 08:42:56 <maribu> Maybe when there is more awareness that not providing stable checksums results in huge costs, there is more motivation to actually provide them. At least in case of sr.ht persistence did pay out and so far have stable checksums from there.
2023-02-03 08:43:38 <psykose> didn't look at what they do specifically but maybe it's also just git-archive
2023-02-03 08:45:13 <psykose> https://git.sr.ht/~sircmpwn/git.sr.ht/tree/master/item/gitsrht/blueprints/repo.py#L423
2023-02-03 08:45:37 <psykose> :-)
2023-02-03 08:47:30 <psykose> i think if you do git -c tar.gz.command="gzip -9" .. it then becomes the old behaviour on 2.38+ or whatever, but eh
2023-02-03 09:03:49 <maribu> re sr.ht https://lists.sr.ht/~sircmpwn/sr.ht-dev/patches/15870 was the fix back then
2023-02-03 10:39:48 <ikke> But I think sr.ht will still be sensitive to changes in output from tar and gzip, like what happened with github 
2023-02-03 14:04:45 <staceee> is there a simple existing way to abuild a directory, and it would build packages newest than known rel-rev ?
2023-02-03 14:05:24 <staceee> would be very usefull and help me building a personal aports with auto-builds
2023-02-03 14:25:29 <ikke> buildrepo from lua-aports
2023-02-03 14:26:06 <ikke> That's what the builders use 
2023-02-03 14:31:11 <staceee> looks like what I need ! thanks
2023-02-03 14:43:23 <staceee> ikke: do you know if it allows to lazy build ? could it ask apk the last known version of the packages ?
2023-02-03 14:43:55 <staceee> looks tweakable using some "prerepo" plugins
2023-02-03 14:43:57 <ikke> It would compare it to the local package repo
2023-02-03 14:44:30 <staceee> it then means my CI job should have local access to the full repo
2023-02-03 15:15:06 <ikke> yeah, that's currently the assumption
2023-02-03 17:42:32 <omni> the last one !43827
2023-02-03 18:14:45 <omni> so I brought !42677 up in #iwd and some rather suggest the !42140 path and... I feel I don't care enough to deal with this
2023-02-03 20:09:37 <caskd> Is there a keyserver which has 293ACD0907D9495A (signing key for alpine releases) available with the ID?
2023-02-03 20:12:38 <caskd> If not, could ncopa verify the key on keys.openpgp.org?
2023-02-04 10:03:45 <PabloCorreaGomez[m]> ikke: I'm currently at FOSDEM with a IP6-only network, and for some reason https://appstream.alpinelinux.org is not reachable. I can ping, but not connect with HTTP. Other websites likes build.alpinelinux.org do work. Do you have some idea about where to look?
2023-02-04 10:04:23 <PabloCorreaGomez[m]> With the dual-stack network it works without issues
2023-02-04 10:37:44 <ikke> PabloCorreaGomez[m]: I'd have to check later. gbr2-dev1, which hosts appstream, should support ipv6
2023-02-04 10:43:58 <PabloCorreaGomez[m]> Ping works, so might it be something higher in the stack? I was a bit confused, that's why I asked.
2023-02-04 10:43:58 <PabloCorreaGomez[m]> And thanks a lot!!
2023-02-04 11:38:25 <fcolista> psykose, you around?
2023-02-04 11:43:53 <fcolista> There's something I'm missing with mr 43799, while reading openrc's service-script-guide. 
2023-02-04 11:44:14 <fcolista> gvmd set the PID path in the build().
2023-02-04 11:44:14 <fcolista> The pid file should be owned by root, according with that doc, and the gvmd daemon should be able to drop the privileges.
2023-02-04 11:44:14 <fcolista> Now...what I see is:
2023-02-04 11:44:14 <fcolista> Otherwise there's a upstream issue (https://github.com/OpenRC/openrc/blob/master/service-script-guide.md#pid-files-should-be-writable-only-by-root)
2023-02-04 11:44:21 <fcolista> libgvm base:CRITICAL:2023-02-04 11h31.01 UTC:5300: pidfile_create: failed to open pidfile: Permission denied
2023-02-04 11:44:30 <fcolista> If I use jsut the /run dir (so the pidfile will be /run/gvmd.pid).
2023-02-04 11:44:30 <fcolista> But if I use /run/gvmd, and gvmd is owned by gvm user, I can't do that too.
2023-02-04 11:44:30 <fcolista> Of course, this is due to gvm user unable to write on /run dir.
2023-02-04 11:44:30 <fcolista> So, what can I do?
2023-02-04 11:47:56 <AmyMalik> wind turbine
2023-02-04 12:05:01 <fcolista> ok, found 
2023-02-04 12:41:36 <ikke> PabloCorreaGomez[m]: appstream CNAME points to the wrong record
2023-02-04 15:39:02 <ikke> PabloCorreaGomez[m]: The record has been updated, I expect it to start working soon
2023-02-04 15:39:28 <ikke> It should be 2a01:7e00:e000:7eb:1::2
2023-02-04 15:57:00 <PabloCorreaGomez[m]> ikke: nameservers here seems to have it cached for now. I'll check and report later, thanks a lot!
2023-02-04 17:19:04 <PabloCorreaGomez[m]> ikke: it's working now, thanks!!
2023-02-04 17:33:05 <ikke> PabloCorreaGomez[m]: great
2023-02-04 19:50:18 <psykose> fcolista: if you run the daemon --foreground it probably doesn't write anything
2023-02-04 19:50:56 <psykose> if there's no way for it to not write it, then there is no way to do it correctly
2023-02-05 10:28:16 <nmeum> psykose: do you have a suggestion on how to fix the alex/happy ghc dependency circle? ghc no longer seems to support an in-tree build of these dependency so we need to package them separately but we can't do that without depending on ghc
2023-02-05 10:28:47 <psykose> i actually can't think of one :/
2023-02-05 10:30:04 <psykose> my only lucky guess is that it would be fine as is, but i think it does break the ordering for building e.g. git-annex and alex or whatever in the same transaction
2023-02-05 10:30:18 <psykose> it would just shuffle randomly every time because of the loop
2023-02-05 10:30:21 <nmeum> I suppose we could alex and happy them as part of the ghc package using ghc-bootstrap but then the ghc APKBUILD would depend on cabal again
2023-02-05 10:30:29 <nmeum> *could build
2023-02-05 10:30:46 <psykose> put cabal inside ghc too, nice and easy
2023-02-05 10:30:54 <nmeum> haha
2023-02-05 10:30:58 <psykose> (maybe that's actually easy? ..)
2023-02-05 10:31:09 <psykose> it's a mess but at the same time..
2023-02-05 10:31:38 <psykose> sometimes the centralisation makes it somewhat easier to manage later, especially since i don't think they will improve this bootstrapping anytime soon
2023-02-05 10:31:53 <nmeum> we could also start packaging the upstream ghc bindist as ghc-stage0 and then use that to build alex and happy
2023-02-05 10:32:02 <nmeum> but the ghc bindist is only available for x86_64 musl and there is also this triplet mismatch issue
2023-02-05 10:32:09 <psykose> aye
2023-02-05 10:32:33 <psykose> i guess it depends on what we want to achieve
2023-02-05 10:32:46 <psykose> there is "keeping aports clean" and there is "being able to build stuff"
2023-02-05 10:33:09 <psykose> it's not too challenging to make our own bindist (read: take the current package and stick it in a .tar.gz in distfiles)
2023-02-05 10:33:18 <psykose> and you can indeed make your own stage0 with that and everything is easy
2023-02-05 10:33:45 <psykose> the harder part is what to do on any upgrade, since it probably needs manual rebindisting every time
2023-02-05 10:34:15 <nmeum> hmhm, right that would also be an option
2023-02-05 10:34:32 <psykose> but, all this gets us is removing that loop
2023-02-05 10:34:51 <psykose> and said loop only really breaks the order afaik for git-annex on repo bootstrap for new releases more or less
2023-02-05 10:34:53 <nmeum> we could also ask ghc upstream to support the in-tree build of alex and happy again (as they did for 9.0.2) I think that would potentially be the best/easiest solution
2023-02-05 10:34:58 <psykose> afaik just retrying 10x will get it to pass
2023-02-05 10:35:23 <psykose> yeah, that also fixes it
2023-02-05 10:35:31 <nmeum> I will look into it :)
2023-02-05 10:35:40 <psykose> wonder what reason they had for removing it
2023-02-05 10:36:09 <psykose> nmeum: don't run out of energy before gcc13 though :p
2023-02-05 10:36:20 <nmeum> haha
2023-02-05 10:36:42 <nmeum> I also have the go 1.20 upgrade no my list
2023-02-05 10:36:51 <nmeum> though these are usually less painful
2023-02-05 10:37:05 <psykose> it looks fine, but i didn't check how the non-built stdlib affects packaging
2023-02-05 10:37:13 <psykose> aside from slowing down every build because we don't keep the cache
2023-02-05 10:37:45 <psykose> the rest of the packages (bumping quic-go and unsafe-moving-gc i'll just fix myself once you push the rebuilds)
2023-02-05 10:38:17 <nmeum> there is also the upcoming problem that go will now bump the bootstrap version every year and gcc-go is still at 1.18 so with go 1.22 our bootstrap path will break :S
2023-02-05 10:38:29 <psykose> that is not great
2023-02-05 10:38:41 <nmeum> I hope they bump gccgo to 1.20 in time
2023-02-05 10:39:17 <nmeum> https://github.com/golang/go/issues/54265#issuecomment-1416695006
2023-02-05 10:41:35 <sam_> thanks, I'd missed that
2023-02-05 10:42:27 <nmeum> psykose: it seems that ghc 9.6 will support building with alex/happy again https://github.com/ghc/ghc/commit/cac8c7bb098002ddce0ef7c1e4429940d22d8cc2 maybe we can backport that
2023-02-05 10:42:36 <nmeum> s/with/without/
2023-02-05 10:42:37 <alpine-meetbot> nmeum meant to say: psykose: it seems that ghc 9.6 will support building without alex/happy again https://github.com/ghc/ghc/commit/cac8c7bb098002ddce0ef7c1e4429940d22d8cc2 maybe we can backport that
2023-02-05 10:42:51 <psykose> hmm
2023-02-05 10:42:55 <psykose> that looks like a small patch
2023-02-05 10:43:24 <psykose> >This fixes two bugs which were adding dependencies on alex/happy
2023-02-05 10:43:32 <psykose> ..so it was never intentional
2023-02-05 10:43:55 <nmeum> yea, looks like it
2023-02-05 10:44:28 <psykose> if i find some pocket energy maybe i'll look at that on existing  9.4 and break the loop with it
2023-02-05 10:44:57 <psykose> unintentional -> fixed; is very different from hacking around stuff, mentally :p
2023-02-05 10:51:41 <nmeum> oh! it seems that this commit was backported to ghc 9.4.4 already and this was only an issue with ghc <=9.4.3 (to which I upgraded the package initially)
2023-02-05 10:51:53 <psykose> :)
2023-02-05 10:53:38 <psykose> the _bootstrapver seems broken (so it depends on 9.4 and 9.0 at the same time and fails)
2023-02-05 10:53:50 <psykose> ah
2023-02-05 10:53:51 <psykose> you fixed i
2023-02-05 10:54:18 <nmeum> you need a different hadrian bootstrap depending on the GHC version you use
2023-02-05 10:54:33 <nmeum> so originally, ghc 9.4.4 was build with ghc 9.0.2 and hence the bootstrapver is currently set to 9.0.2
2023-02-05 10:54:45 <nmeum> now that 9.4.4 is in the repos we can bump _bootstrapver to 9.4.4 as well
2023-02-05 10:55:35 <psykose> yep
2023-02-05 10:55:50 <nmeum> removed the alex/happy dependency in my MR as well, let's see if it builds…
2023-02-05 11:40:28 <nmeum> passed on the ci, pushed the removal of the cyclic alex/happy dependency to aports
2023-02-05 11:41:38 <psykose> happy indeed
2023-02-05 12:38:23 <ikke> Hmm, interesting, building docker images on 32-bits arches fails because it complains about buildx missing or broken
2023-02-05 12:38:29 <ikke> https://gitlab.alpinelinux.org/alpine/infra/docker/unbound/-/jobs/966652
2023-02-05 12:38:56 <ikke> (with buildkit)
2023-02-05 12:40:41 <ikke> I guess related to docker 23.0
2023-02-05 12:43:23 <psykose> does it get fixed by just adding docker-cli-buildx ?
2023-02-05 12:43:23 <psykose> :p
2023-02-05 12:44:37 <ikke> yes, checking that
2023-02-05 12:45:55 <ikke> I just wonder why only 32-bits arches are affected, but maybe to do with providing --platform
2023-02-05 12:47:39 <psykose> perhaps
2023-02-05 13:34:15 <ikke> Doesn't seem to have to do with providing --platform
2023-02-05 13:34:33 <psykose> i tried grepping for it but it's somewhere in one of 500 go modules
2023-02-05 13:43:07 <ikke> What I'm also puzzled about is that the issue linked from the release notes says DOCKER_BUILDKIT=1 should be default now and you have to explicitly enable it, but without setting it it still seems to use the old front-end
2023-02-05 13:43:28 <ikke> ie, it works if I don't export DOCKER_BUIDKIT=1
2023-02-05 14:57:09 <ikke> tomalok: Would it make sense to bundle docker-cli-buildx now it's considered the default front-end?
2023-02-05 14:57:20 <ikke> bundle docker-cli-buildx with docker-cli I mean
2023-02-05 15:31:03 <hjaekel> a question to the c++ experts: what does it mean when the compiler complains about error: conversion from 'long long unsigned int' to 'std::size_t' {aka 'unsigned int'} may change value [-Werror=conversion]
2023-02-05 15:31:09 <hjaekel> the line is: return std::stoull(fields[1]) * 4096;
2023-02-05 15:32:03 <hjaekel> does the 4096 have to be converted to something else?
2023-02-05 15:32:55 <hjaekel> it happens only on 32 bit. 
2023-02-05 15:33:43 <psykose> stoull returns ull
2023-02-05 15:33:48 <psykose> the function returns size_t
2023-02-05 15:34:07 <psykose> size_t is usually just unsigned int
2023-02-05 15:34:54 <psykose> and ull is usually just always uint64_t
2023-02-05 15:35:44 <psykose> er, size_t is unsigned long
2023-02-05 15:36:35 <hjaekel> so it is a mismatch between the declared return type of the function and the actual return type
2023-02-05 15:36:48 <psykose> so on 64-bit it's uint64_t fn() { return uint64_t; }
2023-02-05 15:37:06 <psykose> and on 32-bit it's uint32_t fn() { return uint64_t; }
2023-02-05 15:45:32 <hjaekel> so can I just declare the return type as std::uint64_t and it will work on both platforms?
2023-02-05 15:46:28 <psykose> i can't tell you unless google pays me $500k
2023-02-05 15:46:40 <psykose> thems the rules
2023-02-05 15:47:00 <hjaekel> haha
2023-02-05 15:47:04 <hjaekel> I will ask them
2023-02-05 15:47:34 <hjaekel> 500k seems to be fair
2023-02-05 16:20:29 <invoked> would 500k tanzanian shillings be ok
2023-02-05 16:21:19 <psykose> ye
2023-02-05 16:28:50 <hjaekel> psykose what is this want_check thing for? 
2023-02-05 16:29:04 <psykose> does not build tests with !check
2023-02-05 16:30:02 <hjaekel> you never want that :)
2023-02-05 16:45:13 <psykose> want what
2023-02-05 16:46:05 <psykose> it's specifically to just skip building a few hundred objects when !check is ever defined so check() doesn't get called
2023-02-05 16:46:12 <psykose> not to disable the tests
2023-02-05 20:44:14 <tomalok> ikke: thought about it -- but i haven't had a chance to take a closer look at that... however, in theory if you don't build your own images, you wouldn't need to have buildx installed 
2023-02-06 11:13:55 <ikke> tomalok: true, though it could be unexpected a built-in command (docker build) is not working
2023-02-06 14:48:38 <bl4ckb0ne> x86 runner seems borked https://gitlab.alpinelinux.org/bl4ckb0ne/aports/-/jobs/967217
2023-02-06 14:54:09 <dekedro> +1, https://gitlab.alpinelinux.org/Dekedro/aports/-/jobs/967189
2023-02-06 15:03:49 <nmeum> yep
2023-02-06 16:25:34 <ikke> should be fixed now
2023-02-06 16:28:11 <bl4ckb0ne> ty
2023-02-06 16:29:36 <bl4ckb0ne> pipeline fixed indeed
2023-02-06 16:29:50 <ikke> killed some leech
2023-02-06 17:13:54 <nangel> Hi, Once a MR passes, is there a process to notify someone with access to apply to master? Thanks.
2023-02-06 17:29:57 <skarnet> nangel: pinging people here usually works for me, but I agree that it would be nice to have a more formal process for people who aren't as online
2023-02-06 17:31:08 <nangel> Thanks!... Ok, I can add "ping #alpine-devel" to the wiki - even if it is not perfect, at least it is documented.
2023-02-06 17:32:14 <skarnet> nangel: maybe wait for an official dev answer before modifying the wiki, they might not want an influx of pings :)
2023-02-06 17:32:54 <nangel> :) understood.
2023-02-06 17:34:46 <nangel> Let me test though, to make sure I have it right.
2023-02-06 17:35:58 <nangel> MR 43879 (gpsd 3.25 version bump) passes - could a committer check it please?
2023-02-06 17:36:29 <skarnet> shortcut: !43879
2023-02-06 17:36:39 <nangel> skarnet: excellent!
2023-02-06 17:37:57 <nangel> oh... one more thing (from the wiki) my understanding is the subitting patches via mailing list is deprecated.  correct?
2023-02-06 17:38:41 <ncopa> i think in practice it is, yes
2023-02-06 17:39:03 <skarnet> depends on whom you ask, but let's say that the reaction time for patches via ML is much, much longer than the reaction time for gitlab MRs
2023-02-06 17:39:14 <ncopa> iirc there is an email gateway that will create a gitlab MR, but I don't know if it works
2023-02-06 17:39:46 <skarnet> ncopa: welcome back, I hope you enjoyed your vacation!
2023-02-06 17:40:00 <ncopa> im still on vacation, officially...
2023-02-06 17:40:09 <skarnet> oh. ^^'
2023-02-06 17:40:16 <ncopa> but yeah, i enjoyed it so far
2023-02-06 17:41:26 <ncopa> im trying to figure out how to do network boot with aarch64
2023-02-06 17:41:29 <ncopa> in qemu
2023-02-06 18:08:29 <minimal> ncopa: via UEFI HTTP boot?
2023-02-06 18:29:04 <nmeum> psykose: could you remind me why libssp_nonshared.a (which provides __stack_chk_fail_local etc.) behaves different on x86 compared to other architectures when it comes to linking?
2023-02-06 18:30:20 <q66> it also behaves that way on 32-bit ppc amd mips
2023-02-06 18:30:36 <q66> i do not remember why though
2023-02-06 19:04:58 <ncopa> minimal: via anything
2023-02-06 19:07:41 <minimal> ncopa: well PXE is a Microsoft/Intel standard so I believe it only applies to x86/x86_64. UEFI defines an (optional?) HTTP boot method, so if you use UEFI with aarch64 that might work. If you're using aarch64 without UEFI I don't think there is any network boot mechanism in general
2023-02-06 19:09:38 <minimal> ncopa: this is likely relevant to QEMU: https://edk2-docs.gitbook.io/getting-started-with-uefi-https-boot-on-edk-ii/introduction
2023-02-06 19:19:10 <ncopa> thank you sir!
2023-02-06 19:32:02 <psykose> nmeum: https://www.openwall.com/lists/musl/2018/09/11/2
2023-02-06 19:32:22 <psykose> gcc emits stuff differently there
2023-02-06 19:32:41 <psykose> or rather not gcc but some compiling stuff in general i guess
2023-02-06 19:32:43 <ikke> gcc: ommit different
2023-02-06 19:32:54 <alpine-meetbot> ikke meant to say: gcc: emit different
2023-02-06 19:32:54 <ikke> s/ommit/emit
2023-02-06 19:47:47 <psykose> ikke: you could probably do pkgdir=/some/path in https://gitlab.alpinelinux.org/alpine/infra/docker/apkbuild-lint-tools/-/blob/master/overlay/usr/share/abuild/APKBUILD_SHIM so it silences the $pkgdir/usr unset warning
2023-02-06 19:48:22 <ikke> psykose: right
2023-02-06 19:48:29 <ikke> Feel free to make an MR
2023-02-06 19:48:51 <ikke> there is pkgusers=
2023-02-06 19:49:02 <ikke> pkgdir= is already there
2023-02-06 19:51:43 <psykose> not entirely sure it works even
2023-02-06 19:51:54 <psykose> pkgusers? what do you mean?
2023-02-06 19:52:25 <ikke> https://gitlab.alpinelinux.org/alpine/infra/docker/apkbuild-lint-tools/-/blob/master/overlay/usr/share/abuild/APKBUILD_SHIM#L32
2023-02-06 19:52:49 <psykose> what about it
2023-02-06 19:53:10 <ikke> oh, you want to actually set it to a value
2023-02-06 19:53:19 <psykose> i'm not sure if it works, but yes
2023-02-06 19:53:21 <ikke> This did work when I set it up
2023-02-06 19:53:40 <ikke> can you show an example where it complains?
2023-02-06 19:54:07 <ikke> set but empty should satisfy shellcheck
2023-02-06 19:54:28 <ikke> Unless they changed something
2023-02-06 19:54:45 <psykose> it does not
2023-02-06 19:54:46 <psykose> oh well
2023-02-06 19:55:05 <psykose> https://img.ayaya.dev/3SMEEbCCtnB1
2023-02-06 19:55:35 <ikke> oh, a different warning
2023-02-06 19:56:08 <ikke> I think it gives that regardless of whether it has a value
2023-02-06 19:56:37 <ikke> It's to prevent bumblebee style errors
2023-02-06 19:57:14 <ikke> I believe it happened to steam as well
2023-02-06 19:58:39 <ikke> psykose: the wiki even refers to steam :D
2023-02-06 19:59:24 <psykose> yes
2023-02-07 05:17:19 <tomalok> ikke: life on the edge?  probably warrants being called out in 3.18's notes, and some better way to respond if "docker build" is attempted but the (new) default to use buildx is not installed.
2023-02-07 05:35:28 <ikke> tomalok: the docker-cli image we use to build other docker images use edge :)
2023-02-07 05:36:25 <ikke> At least mentioning it in the release notes would be good
2023-02-07 16:30:10 <elly> do we need openssl -> 3.0.8? https://www.openssl.org/news/secadv/20230207.txt
2023-02-07 16:30:35 <ikke> elly: psykose is way ahead of you :p
2023-02-07 16:30:46 <elly> impressive
2023-02-07 16:30:52 <ikke> (like 11 minutes)
2023-02-07 16:31:44 <ikke> does mean we need new releases as well
2023-02-07 16:31:50 <minimal> I did mention its release 25 mins ago in alpine-security ;-)
2023-02-07 16:31:58 <ikke> minimal: talk is cheap :P
2023-02-07 16:32:18 <minimal> nothing wrong with cheap dates lol
2023-02-07 16:33:16 <minimal> ikke: besides I like to avoid staring into the abyss (oops, openssl source)
2023-02-07 16:33:48 <ikke> sane policy
2023-02-07 17:33:23 <psykose> time for the giga go 1.20 rebuild of death
2023-02-07 17:34:06 <psykose> hopefully you don't end up having to clean the diskspace ikke :p
2023-02-07 17:40:51 <ikke> we'll see
2023-02-07 17:45:08 <skarnet> rebuild? isn't the point of dynamic linking to avoid rebuilds when a library needs a security upgrade?
2023-02-07 17:45:25 <ikke> skarnet: this is not about openssl
2023-02-07 17:45:33 <skarnet> oh
2023-02-07 17:45:34 <ikke> and go builds statically
2023-02-07 17:45:43 <skarnet> oh, go.
2023-02-07 17:46:01 <skarnet> I interpreted "giga go 1.20" as "1.20 GB"
2023-02-07 17:46:27 <skarnet> ("Go" is awful naming for a programming language)
2023-02-07 17:51:39 <q66_> s/naming for a //
2023-02-07 17:51:43 <AmyMalik> go = gigaoctet
2023-02-07 18:11:30 <bl4ckb0ne> Go 2 should be rename To to mess up french people
2023-02-07 19:34:49 <deathmist2> it appears Waydroid internal storage access is broken for some people (https://t.me/WayDroid/133686) with waydroid-1.3.3-r2 on Alpine v3.17 (postmarketOS v22.12) and just a "sudo waydroid upgrade -o" fixed it which I added to run in post-install (like it should've always been as per upstream debian packaging too) as a part of
2023-02-07 19:34:51 <deathmist2> https://gitlab.alpinelinux.org/alpine/aports/-/merge_requests/42450, is there any chance that could be backported to stable from edge?
2023-02-07 19:37:25 <psykose> something something open the merge request
2023-02-07 21:11:34 <EvTheFuture> Just want to check if there are anyone else that have comments on !43697 and if not if it's possible to get this merged?
2023-02-07 21:14:53 <psykose> personally can't review it but if minimal says it's good then i'm fine with it
2023-02-07 21:52:55 <deathmist2> psykose: https://gitlab.alpinelinux.org/alpine/aports/-/merge_requests/43982
2023-02-08 00:54:07 <DavidHeidelberg[m]> psykose: I have SSE2 fix for the LTO on Alpine; It's not enough to enable it by default without further testing, but "it's something" :P https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21180 tested in pmOS MR here https://gitlab.com/alpine-mobile/pmaports/-/pipelines/770476019
2023-02-08 00:55:23 <DavidHeidelberg[m]> (to be precise, that's 22.3.4 with backport of this patch, but the outcome should be identical)
2023-02-08 00:57:49 <psykose> exciting :)
2023-02-08 00:58:32 <psykose> to be clear i disabled it anyway in 224e17455de5ae3fcc1e43347b63f1e8e8f09acb
2023-02-08 00:58:44 <psykose> when someone with savant brain kills 12 hours on it maybe i'll turn it back on
2023-02-08 01:00:14 <DavidHeidelberg[m]> yeah, I'm aware, I was thinking testing on pmOS (testing with volunteers) and after initial testing maybe enabling for aarch64/armv7 
2023-02-08 01:00:55 <DavidHeidelberg[m]> because it has huge benefit of reducing cpu-bound operations and saving some RAM
2023-02-08 01:01:10 <DavidHeidelberg[m]> which is scarce resources on low-end phones/tablets
2023-02-08 01:01:54 <psykose> for armv7 specifically fixing the -mtls-dialect=gnu2 would also be useful, but that points more to a binutils bug
2023-02-08 01:04:31 <DavidHeidelberg[m]> I seen the option in build-system; one day I'll brave enough to go figure what it does
2023-02-08 12:45:16 <Nulo[m]> Hey! I believe that bumping the dart version (as done so here: https://gitlab.alpinelinux.org/alpine/aports/-/commit/118700752ba2a87db24d9da48f6e97e478d6bfb8) requires rebuilding the dart packages we currently have. For example, testing/dart-sass-embedded broke: Can't load Kernel binary: Invalid kernel binary format version.
2023-02-08 12:46:41 <Nulo[m]> Also, some dart-sass person is making it possible to build native musl binaries (instead of "kernel binaries" which requires the huge dart runtime): https://github.com/dart-musl/dart / https://github.com/dart-musl/dart-sass-embedded 
2023-02-08 12:48:32 <Nulo[m]> https://github.com/dart-lang/sdk/pull/51044 / https://github.com/dart-lang/sdk/pull/51057
2023-02-08 13:11:09 <nangel> nangel
2023-02-08 13:24:18 <ptrc> Nulo[m]: we already have working executables for dat-sass though
2023-02-08 13:24:21 <ptrc> dart*
2023-02-08 13:29:36 <ptrc> for dart-sass-embedded, it looks like you'd have to change `grinder pkg-compile-snapshot` to.. `grinder build`, i think?
2023-02-08 13:32:32 <Nulo[m]> <ptrc> "Nulo: we already have working..." <- dart-sass-embedded is a different thing
2023-02-08 13:32:45 <ptrc> yeah, i'm looking at it right now
2023-02-08 13:33:03 <ptrc> currently trying to build it with pkg-compile-native
2023-02-08 13:33:58 <Nulo[m]> ptrc: Actually getting working musl binaries is hard, and it's not as easy as changing to griner build I believe. (source: I spent hours trying to figure it out and failed.) I think this is the only thing that actually got it working: https://github.com/dart-musl/dart-sass-embedded/blob/main/.github/workflows/build.yml but I haven't tried porting it yet
2023-02-08 13:35:34 <ptrc> i tried too, a while ago it was broken beyond repair, but recently all it took was to replace `dart compile kernel` with `dart compile exe` - see 6dff747bf2379331e4e0408c34e751b37a21bd9b
2023-02-08 13:37:32 <ptrc> that's a fun error though: https://tpaste.us/Xy6E
2023-02-08 13:43:01 <ptrc> fwiw running https://github.com/google/grinder.dart/blob/1a8b809f/lib/grinder_sdk.dart#L39 by itself with `dart run` returns the correct thing
2023-02-08 13:43:11 <ptrc> so it's either some grinder cursedness, or.. idk
2023-02-08 13:46:58 <Nulo[m]> realistically we (Sutty) don't really need the dart-sass-embedded package as the rubygem now comes with a static musl binary (because of ntkme's work). combining that with me not knowing anything about Dart, I probably won't bother turning it into a static binary. I copied the "dart run" approach from the dart package (dart2js)
2023-02-08 14:42:49 <bl4ckb0ne> is there a rebuild going on? got some conflicts with libjxl and ffmpeg
2023-02-08 14:49:35 <ptrc> what kind of conflicts? there's nothing building at the moment
2023-02-08 14:49:49 <ptrc> ah, no, riscv64 is building something
2023-02-08 14:49:53 <ptrc> but all the other architectures are idle
2023-02-08 15:02:19 <ptrc> oh god, okay
2023-02-08 15:02:19 <ptrc> https://tpaste.us/LR66
2023-02-08 15:02:26 <ptrc> yeah, i can see what kind of conflicts
2023-02-08 15:05:55 <bl4ckb0ne> yep those
2023-02-08 15:06:11 <ptrc> ffmpeg-libs doesn't exist on edge anymore
2023-02-08 15:06:22 <ptrc> it was split in eb9e0629c3
2023-02-08 15:06:29 <ptrc> ( oh come on, algitbot )
2023-02-08 15:06:56 <bl4ckb0ne> you have too many digits for  short hash
2023-02-08 15:07:17 <ptrc> welp, that's what `git log --oneline` spit out for me
2023-02-08 15:07:52 <ptrc> whatevs, anyway, it looks like apk doesn't want to pick the new splits for some reason
2023-02-08 15:08:06 <bl4ckb0ne> test eb9e0629
2023-02-08 15:08:23 <ptrc> `apk add ffmpeg-libavcodec` seems to do the job
2023-02-08 15:08:23 <bl4ckb0ne> and eb9e0629c3bbeea84bf92c6681c540775f799b51  ?
2023-02-08 15:08:47 <bl4ckb0ne> smh psykose breaking my computer again
2023-02-08 15:08:51 <ptrc> bl4ckb0ne: yeah, it needs to have 40 characters - https://gitlab.alpinelinux.org/alpine/infra/compose/algitbot/-/blob/master/sircbot/scripts/sircbot.lua#L164
2023-02-08 15:09:50 <bl4ckb0ne> works indeed, thanks ptrc 
2023-02-08 15:40:43 <invoked> https://github.com/golang/go/discussions/58409
2023-02-08 16:24:52 <Oyo> https://vlang.io/ 
2023-02-08 16:26:48 <invoked> interesting.
2023-02-08 16:30:23 <Oyo> https://github.com/vlang/v/blob/master/doc/docs.md
2023-02-08 16:31:24 <skarnet> oh great, $language_of_the_year just dropped
2023-02-08 16:33:25 <minimal> skarnet: language of the month you mean? ;-)
2023-02-08 16:34:40 <skarnet> I wanted to be polite
2023-02-08 16:34:48 <ptrc> ...people treat vlang as if it were a serious programming language?
2023-02-08 16:35:16 <mdekstrand> what is the procedure for dropping maintenance of a package (in testing)? or indicating that further maintenance is unlikely? MR to move it to 'unmaintained', or just leave it there & move later if issues come up?
2023-02-08 16:35:45 <mdekstrand> i've been maintaining librespot but am no longer using it myself. if i start using it again in the future would be happy to maintain again.
2023-02-08 16:35:58 <ptrc> mdekstrand: if it's likely someone will take over, you can just open a MR removing yourself and leaving just `Maintainer:`
2023-02-08 16:36:51 <mdekstrand> ptrc: thanks! i'll do that.
2023-02-08 16:38:17 <Oyo> pdf book https://libgen.li/ads.php?md5=509ad558e93f3cc0ca8988f4f0086ee8
2023-02-08 16:38:56 <ptrc> not sure how it's related to alpine :p it might be better to continue this in #alpine-offtopic
2023-02-08 16:38:58 <skarnet> risky click of the day
2023-02-08 18:30:39 <ncopa> i think i need to do releases with the openssl fix
2023-02-08 19:09:09 <ncopa> drats, aarch64 builder failed to create the release for some reason I cannot log in to it from here
2023-02-08 19:20:21 <ncopa> oh, it actually passed. it was just slow
2023-02-08 19:32:53 <donoban> algitbot: has empathy IA? :O
2023-02-08 19:33:48 <ptrc> algitbot: what exactly are you?
2023-02-08 19:33:57 <ptrc> :p
2023-02-08 20:29:19 <ddevault> reopening this discussion: https://gitlab.alpinelinux.org/alpine/aports/-/merge_requests/26774
2023-02-08 20:38:13 <ikke> related to the telemetry discussion from RSC?
2023-02-08 20:39:55 <ikke> right, separate issue for that
2023-02-08 20:55:42 <ddevault> yeah
2023-02-08 20:56:29 <ddevault> I had been meaning to re-open the other discussion as well since I found out that fedora disables it
2023-02-08 20:56:39 <ddevault> the new telemetry stuff prompted me to address both issues
2023-02-08 20:56:59 <ddevault> (the separate issue: https://gitlab.alpinelinux.org/alpine/aports/-/issues/14618)
2023-02-09 02:19:11 <sam_> psykose: one option wrt python externally managed is to package pipx, which supposedly works like pip, but automagically handles a venv, but we gave up on packaging it for now as tests were hell
2023-02-09 02:25:59 <psykose> that doesn't sound like an improvement
2023-02-09 02:26:29 <psykose> the issue is not "have to handle venv"
2023-02-09 02:26:45 <psykose> the issue is "literally anything changed, cry cry cry, i have to fix my pipelines, please revert"
2023-02-09 02:26:56 <psykose> s|pip|pipx| is the same thing
2023-02-09 02:26:56 <alpine-meetbot> psykose meant to say: the issue is "literally anything changed, cry cry cry, i have to fix my pipxelines, please revert"
2023-02-09 02:27:07 <psykose> thanks sed bot
2023-02-09 02:33:09 <sam_> yes, fair enough
2023-02-09 02:40:04 <psykose> (which is unfortunate but there's no other way to do it i guess)
2023-02-09 02:45:51 <skarnet> fix your pipxelines, psykose
2023-02-09 02:46:27 <psykose> i would
2023-02-09 02:46:37 <psykose> someone found it easier to write a whole essay instead though
2023-02-09 02:49:57 <psykose> sam_: could you ping someone in llvm to generate their tarballs with xz -T0 :p
2023-02-09 02:52:02 <sam_> psykose: https://github.com/llvm/llvm-project/commit/009048810ac635a7ad6c5f788d537172418b6054 & https://github.com/llvm/llvm-project/commit/70165c55dc8cbb412d396897c2a588879b78ad1a should cover it i think
2023-02-09 02:52:11 <sam_> so I think it's already done 
2023-02-09 02:52:12 <sam_> or is it not
2023-02-09 02:52:59 <psykose> hmm
2023-02-09 02:53:07 <psykose> the first one is for test-release and is not in rc
2023-02-09 02:53:16 <psykose> i guess.. the release ones do but the rc's don't? haha
2023-02-09 02:53:20 <psykose> the present -rc2 one isn't
2023-02-09 02:53:39 <sam_> ok lemme see
2023-02-09 02:53:43 <psykose> (yes i am doing some rc testing because i'm a loser)
2023-02-09 02:53:51 <sam_> i know who to ask anyway
2023-02-09 02:54:22 <psykose> or maybe i just can't read so feel free to double check the tar directly first
2023-02-09 02:55:11 <sam_> do you have some way of checking if it's using multiblocks or did you just try decompress it and see the lamenes
2023-02-09 02:55:13 <sam_> s
2023-02-09 02:56:09 <psykose> decompressed
2023-02-09 02:57:09 <sam_> ack
2023-02-09 03:04:12 <sam_> with my also very scientific approach it doesn't look like it either, but all the relevant scripts seem to DTRT, huh
2023-02-09 03:05:14 <sam_> well it goes above 100% cpu briefly but not for long
2023-02-09 03:07:11 <psykose> i would imagine that first patch which is not in any release is responsible
2023-02-09 03:07:22 <psykose> that or we are just smoking some drugs
2023-02-09 03:11:13 <sam_> let's see what happens w/ https://github.com/llvm/llvm-project/issues/60617
2023-02-09 03:14:31 <psykose> sam_: have you seen a undefined reference to `rk_getpwnam_r' with heimdal
2023-02-09 03:14:40 <psykose> tryna backport the latest cve hotness but it does not even build anymore
2023-02-09 03:14:44 <psykose> weird
2023-02-09 03:19:16 <sam_> looking now
2023-02-09 03:19:18 <sam_> btw libreoffice is xzing it up as well
2023-02-09 03:19:20 <sam_> https://gerrit.libreoffice.org/c/core/+/146105
2023-02-09 03:21:12 <sam_> (re llvm: https://github.com/llvm/llvm-project/issues/60617#issuecomment-1423566467 hmm)
2023-02-09 03:24:04 <psykose> libreoffice can use the same improvement year
2023-02-09 03:24:06 <psykose> yeah*
2023-02-09 03:25:13 <sam_> any other big bois we're forgetting?
2023-02-09 03:25:15 <sam_> .. haskell maybe? idk
2023-02-09 03:25:28 <sam_> i already persuaded the gcc people but i don't know if they backported it, probably not
2023-02-09 03:29:14 <sam_> for heimdal, was that in the built libraries, or did it fail to compile, or what
2023-02-09 03:29:35 <psykose> fails 2 build
2023-02-09 03:35:13 <sam_> can't hit it :(
2023-02-09 03:35:36 <sam_> rk_getpwnam_r is their cheesy replacement, so i imagine the configure tests (there's 2) for system getpwnam_r fail for you
2023-02-09 03:37:11 <sam_> https://github.com/heimdal/heimdal/issues/106 huh
2023-02-09 03:40:14 <psykose> hmmmmmm
2023-02-09 03:45:59 <psykose> removing my Werrors fixed it
2023-02-09 03:46:06 <psykose> time to compare config logs
2023-02-09 03:46:07 <psykose> fun
2023-02-09 03:50:13 <psykose> checking for socket.. has an implicit decl but i have no idea where that is from
2023-02-09 03:50:34 <psykose> eh there's like 5 more
2023-02-09 03:51:00 <psykose> too many implicits
2023-02-09 03:53:14 <sam_> lmao
2023-02-09 03:53:16 <sam_> high q software
2023-02-09 03:53:20 <sam_> i see some fixes in git for those but also still open bugs (https://github.com/heimdal/heimdal/issues/790)
2023-02-09 03:55:37 <psykose> yeah
2023-02-09 07:10:03 <psykose> ikke: for some reason i can't pull/push to gitlab
2023-02-09 07:10:05 <psykose> weird
2023-02-09 07:16:52 <ikke> Hmm I still could just before
2023-02-09 07:16:58 <ikke> What happens? 
2023-02-09 07:17:50 <psykose> just hangs forever
2023-02-09 07:19:14 <ikke> Ipv4 or ipv6 
2023-02-09 07:19:27 <psykose> both
2023-02-09 07:19:31 <psykose> and now it finally worked
2023-02-09 14:40:29 <ncopa> i have noticed that gitlab pull/clone can be slow at times
2023-02-09 14:56:57 <ncopa> it is certainly slow here, at 60 KiB/s
2023-02-09 15:06:57 <elly> oh I thought that was just me
2023-02-09 15:07:04 <psykose> are you perchance cloning via ipv6 and git clone -4 fixes it
2023-02-09 15:07:32 <ncopa> likely ipv6 here yes
2023-02-09 15:08:41 <psykose> (i've seen that many times, idk what causes it either)
2023-02-09 15:10:11 <ncopa> heh... i started the linux-lts build a few minutes after i started the git clone. it built one kernel already and git clone is at 38%...
2023-02-09 19:29:37 <craftyguy> I noticed this circular dependency between libfreetypee and harfbuzz... https://bpa.st/raw/26XIK 
2023-02-09 19:30:55 <craftyguy> I'm assuming that's not intentional, but also not sure how to resolve it
2023-02-09 19:31:16 <ptrc> it's not an issue aports-wise
2023-02-09 19:31:40 <ptrc> there's harfbuzz-stage0 built without freetype
2023-02-09 19:31:41 <psykose> it's not circular
2023-02-09 19:31:48 <psykose> i'm curious how you even found that
2023-02-09 19:31:57 <ptrc> then freetype is built with that, and harfbuzz proper with freetype
2023-02-09 19:32:17 <craftyguy>         libharfbuzz.so.0 => !!! circular loop !!!
2023-02-09 19:32:22 <craftyguy> not circular? :P
2023-02-09 19:32:38 <psykose> not circular in packaging bootstrap
2023-02-09 19:32:46 <psykose> the actual dt_needed's are
2023-02-09 19:32:47 <craftyguy> ok, let me investigate further. something probably needs to depend on the harfbuzz-stage0 then
2023-02-09 19:32:57 <psykose> what is the real issue
2023-02-09 19:33:12 <psykose> nobody looks at lddtree output and reports random things for no reason
2023-02-09 19:33:39 <craftyguy> the real issue is an app that iterated dependencies for a binary got into a circular loop when listing the libsdl2 deps
2023-02-09 19:34:28 <psykose> ah
2023-02-09 19:34:30 <ptrc> oddly enough, my local libfreetype doesn't have harfbuzz in needed
2023-02-09 19:34:34 <psykose> not fun, but you do have to handle that :p
2023-02-09 19:34:37 <psykose> ptrc: i changed that today
2023-02-09 19:34:50 <ptrc> ah, makes sense :p
2023-02-09 19:34:52 <craftyguy> psykose: yep :)
2023-02-09 19:35:05 <psykose> craftyguy: for reference, it's like arch now https://img.ayaya.dev/BnrJdZj1rZRI
2023-02-09 19:35:07 <craftyguy> but I saw that output from lddtree and commented here, thanks for clarifying it
2023-02-09 19:35:13 <psykose> i didn't do it up until now because it was a 3-stage bootstrap
2023-02-09 19:35:27 <psykose> 2.13 has a commit that drops the harfbuzz dep from hb-ft to just hb
2023-02-09 19:35:38 <psykose> so it's hb-nothing -> freetype+hb -> hb+ft
2023-02-09 19:35:48 <psykose> before it was.. uhh no idea
2023-02-09 19:35:49 <craftyguy> psykose: that still shows         libfreetype.so.6 => !!! circular loop !!!
2023-02-09 19:35:53 <psykose> yeah
2023-02-09 19:35:55 <psykose> that's what i mean
2023-02-09 19:35:57 <psykose> it's not unique
2023-02-09 19:36:28 <craftyguy> ahhh ok I misunderstood
2023-02-09 19:36:40 <psykose> i am surprised a little though, is this the first thing you found like this? i'd imagine it's not that rare :p
2023-02-09 19:36:59 <craftyguy> yeah, it doesn't deal with many binaries/libs though
2023-02-09 19:37:22 <craftyguy> but ya, the real bug is I'm not handling circular deps :D
2023-02-09 19:38:41 <ncopa> do we need harfbuzz with freetype?
2023-02-09 19:46:13 <ncopa> upstream discussion: https://github.com/harfbuzz/harfbuzz/issues/2524
2023-02-09 19:55:40 <skarnet> circular dependencies are a real big brain moment
2023-02-09 19:55:59 <skarnet> I wonder what goes through the head of people when they decide it's a good idea
2023-02-09 19:57:38 <craftyguy> is that typically done intentionally?
2023-02-09 19:59:02 <psykose> if you exclude bootstrap, no
2023-02-09 19:59:25 <ikke> It grows organically
2023-02-09 20:03:05 <abby> i think void found that cycle with cairo in the mix
2023-02-09 20:03:44 <skarnet> ikke: of course it grows organically, but when it happens, you *see* it
2023-02-09 20:03:49 <psykose> abby: the bootstrap harfbuzz has to have everything disabled for there to be no circle
2023-02-09 20:03:58 <abby> yeah
2023-02-09 20:04:06 <skarnet> what makes people think "oh, it's okay, let's add the dependency anyway"
2023-02-09 20:04:11 <psykose> it also only works with new freetype, the old one depends on harfbuzz+ft so i am not sure how it's possible without copying manual headers around
2023-02-09 20:04:31 <psykose> now it's easy though so i added it
2023-02-09 20:05:20 <abby> skarnet: just link everything statically /s
2023-02-09 20:08:20 <skarnet> abby: that's entirely orthogonal to circular dependencies
2023-02-09 20:09:18 <psykose> yes yes we get it circle everything software delete rm -rf bad linux linux bad devs brain issue circle circle egg
2023-02-09 20:10:19 <skarnet> ... it may be past your bedtime
2023-02-10 01:31:43 <Ariadne> we had a customer open a ticket about python3 not having FORTIFY or stack canaries.  he mentioned also that alpine's python3 package has the same problem.  i think it is relating to python's build system doing weird things when LTO is enabled :)
2023-02-10 01:59:59 <Ariadne> oh, this is really interesting.  it's because in python 3.11, python3 is just a stub which jumps straight into libpython, without even having a main()
2023-02-10 02:01:55 <elly> huh, why is it like that?
2023-02-10 02:04:35 <Ariadne> when you run checksec against libpython, stack canaries are definitely present :)
2023-02-10 02:06:18 <elly> oh, but python3 the binary doesn't have any
2023-02-10 02:06:19 <elly> got it
2023-02-10 02:32:03 <psykose> so.. no issue?
2023-02-10 04:19:49 <sam_> someone came to me screaming about this before
2023-02-10 04:21:45 <psykose> corporate checklisters dot png
2023-02-10 13:07:22 <nangel> psykose: Thanks for taking care of the gpsd thing.  I ended up having a whole discussion with the gpsd developers on the x86 issue, and got lost in the weeds.
2023-02-10 16:26:14 <ncopa> i am about to tag releases for alpine 3.1[4-7]
2023-02-10 16:59:24 <ikke> 👍
2023-02-10 17:10:02 <ncopa> do you prefer one release notes per release or a single doc with them all
2023-02-10 17:10:33 <ncopa> eg https://alpinelinux.org/posts/Alpine-3.17.1-released.html vs https://alpinelinux.org/posts/Alpine-3.13.12-3.14.8-3.15.6-3.16.2-released.html
2023-02-10 17:18:33 <elly> psykose: apart from continuing to upstream patches as I have time to do so, is there anything else that would make chromium easier to package
2023-02-10 19:14:35 <ncopa> Alpine Linux distribution development | release 3.17.2 is out!
2023-02-10 19:15:41 <ikke> now with /topic :P
2023-02-10 19:15:50 <ncopa> yeah
2023-02-10 19:15:55 <ncopa> Cannot perform action: You're not a channel operator.
2023-02-10 19:16:20 <ikke> ncopa: you can probably ask chanserv to make you op
2023-02-10 19:16:25 <ikke> not sure why you are not auto-opped
2023-02-10 19:16:32 <ncopa> and i didnt bring my notes on how to do that
2023-02-10 19:16:34 <ikke> are you authetnitcated?
2023-02-10 19:16:37 <ncopa> yeah
2023-02-10 19:17:13 <ikke> I have this as alias: "/msg chanserv op $channel $nick"
2023-02-10 19:17:57 <ncopa> yeah i figured it out with /msg chanserv help
2023-02-10 19:41:29 <ncopa> ikke can you please post the announcement(s) on mastodon?
2023-02-10 19:42:08 <ikke> I already did :p
2023-02-10 19:43:33 <ikke> https://fosstodon.org/@alpinelinux/109842084226405924
2023-02-10 19:45:16 <ncopa> thank you sir! have a good weekend!
2023-02-10 19:45:35 <ikke> You too
2023-02-11 00:37:13 <ptrc> losing my mind trying to decode apk3 format
2023-02-11 00:37:21 <ptrc> has anyone tried doing anything with it yet?
2023-02-11 00:37:32 <ptrc> any notes on it, anything?
2023-02-11 00:38:44 <minimal> ptrc: didn't someone post notes on it about 1-2 weeks ago on here?
2023-02-11 00:40:00 <ptrc> tried looking in the logs, but 'apk3' yields no results and i'm not sure what else i could search for
2023-02-11 00:44:02 <minimal> https://gitlab.alpinelinux.org/alpine/apk-tools/-/merge_requests/120
2023-02-11 00:44:43 <minimal> mentioned here on 5th Jan at 05:30
2023-02-11 00:45:08 <ptrc> thank you!
2023-02-11 00:45:44 <ptrc> and thanks elly for writing some actual docs :)
2023-02-11 00:48:53 <elly> I should get that submitted
2023-02-11 00:48:54 <elly> sigh :)
2023-02-11 02:26:36 <psykose> elly: i guess the list of prepare() magic, but that is just more "upstreaming"..
2023-02-11 02:45:38 <elly> ah well
2023-02-11 02:45:43 <elly> that stuff might be harder to fix
2023-02-11 02:46:45 <elly> I just discovered something distressing: fuchsia uses musl (yay!) but for some reason has hacked the resolv.h to report a different version in __RES than musl normally does, which means my newest chromium patch needs an ugly fuchsia-specific special case
2023-02-11 02:46:58 <elly> in what is otherwise an extremely clean #if defined(__RES) && __RES >= 19991006
2023-02-11 03:01:10 <psykose> fuschia started from a musl fork but it's not that anymore
2023-02-11 03:02:09 <psykose> also unless i can't find it https://cs.opensource.google/search?q=__RES&sq=&ss=fuchsia%2Ffuchsia:zircon%2F
2023-02-11 03:02:13 <psykose> it is 19991006 there too
2023-02-11 03:02:25 <psykose> ah
2023-02-11 03:02:28 <psykose> not 1996
2023-02-11 03:03:05 <psykose> musl c8fdcfe550997243daba0814a95f42890dc60e45 is probably interesting to you then
2023-02-11 03:03:08 <psykose> it was actually changed
2023-02-11 03:03:21 <psykose> from exactly that, in 2016
2023-02-11 03:03:22 <bl4ckb0ne> psykose: wasnt vk validation dropped?
2023-02-11 03:03:28 <psykose> it is on master wasn't it
2023-02-11 03:03:28 <bl4ckb0ne> i think i yot it
2023-02-11 03:03:45 <psykose> i know i can use master wlroots and stuff and patch it out even but ree
2023-02-11 03:04:06 <psykose> (i followed all the upstream issues anyway and i know you did all that)
2023-02-11 03:04:20 <bl4ckb0ne> <3
2023-02-11 03:05:41 <bl4ckb0ne> can probably patch the package with https://gitlab.freedesktop.org/wlroots/wlroots/-/commit/f3ba14e491333d6bbba8c60328c4dbfd20571182
2023-02-11 03:07:27 <psykose> nyo
2023-02-11 03:09:01 <bl4ckb0ne> or i can pressure emersion to do 0.16.3 :p
2023-02-11 03:10:19 <psykose> i don't think he would agree with dropping that in a patch on stable
2023-02-11 11:01:01 <chereskata> hey there, i just updated my edge and noticed the following error: ERROR: imath-3.1.6-r1: trying to overwrite usr/lib/libImath-3_1.so.29 owned by openexr-3.1.5-r4.
2023-02-11 11:01:24 <chereskata> full log:
2023-02-11 11:01:28 <chereskata> https://paste.gnome.org/j2Bi3taLK
2023-02-11 11:06:16 <psykose> fixed
2023-02-11 12:11:32 <wolf> https://pastebin.com/zCLcgNQG
2023-02-11 12:11:44 <wolf> Good Moning
2023-02-11 12:12:58 <wolf> I'm compiling anjuta which is the last gtkpod dependency, but at the end of this error, I couldn't understand.
2023-02-11 12:14:34 <psykose> gtkpod, an ipod client, depends on an entire abandoned IDE to run?
2023-02-11 12:14:36 <psykose> lol
2023-02-11 12:15:31 <ikke> wolf: the errors mean that abuild expects a $pkgname-doc subpackages to be defined 
2023-02-11 12:15:32 <psykose> you are missing a $pkgname-doc in subpackages= i guess
2023-02-11 12:15:36 <ikke> (abuild itself would do the splitting)
2023-02-11 12:24:12 <wolf> Thank you ikke, that was it. Big hug.
2023-02-11 12:24:51 <ikke> wolf: it's basically a policy check, it wants man pages to be in the -doc subpackage instead of in the main package
2023-02-11 12:28:33 <wolf> Yes. Now I understand.
2023-02-11 12:29:18 <wolf> But it wasn't this one. haha. It's libanjuta. LOL
2023-02-11 12:29:41 <wolf> But it's good. I'm close.
2023-02-11 12:59:17 <wolf> The libanjuta worked. Another problem now.
2023-02-11 12:59:26 <wolf> https://pastebin.com/yfYiSBGq
2023-02-11 13:00:10 <ikke> Perhaps due to -fcommon?
2023-02-11 13:02:02 <ikke> (or rather -fno-common)
2023-02-11 13:02:04 <psykose> yeah
2023-02-11 13:02:26 <psykose> have to add CFLAGS="$CFLAGS -fcommon" CXXFLAGS="$CXXFLAGS -fcommon" for it
2023-02-11 13:34:46 <wolf> add where?
2023-02-11 13:34:58 <wolf> shell ?
2023-02-11 13:35:33 <psykose> probably `make CFLAGS= .....`
2023-02-11 13:35:58 <ikke> in the APKBUILD
2023-02-11 13:39:10 <wolf> Same mistake.
2023-02-11 13:39:18 <wolf> https://pastebin.com/tgYz8nxS
2023-02-11 13:41:03 <wolf> make CFLAGS="$CFLAGS -fno-common" CXXFLAGS="$CXXFLAGS -fno-common"
2023-02-11 13:41:17 <wolf> Same mistake
2023-02-11 13:43:22 <psykose> i typed -fcommon
2023-02-11 13:43:41 <ikke> the paste has -fcommon
2023-02-11 13:43:42 <psykose> ah
2023-02-11 13:43:56 <psykose> you have to make clean first
2023-02-11 13:50:50 <wolf> https://pastebin.com/N29Ln9re
2023-02-11 13:51:03 <wolf> cleaned
2023-02-11 13:57:32 <wolf> make CFLAGS="$CFLAGS -fcommon" CXXFLAGS="$CXXFLAGS -fcommon"
2023-02-11 13:57:42 <wolf> Same mistake
2023-02-11 13:58:03 <wolf> cleaned
2023-02-11 13:58:06 <ikke> s/mistake/error ;-)
2023-02-11 14:03:54 <psykose> you can try https://img.ayaya.dev/qBQYuF5p6YUH i guess
2023-02-11 14:11:54 <wolf> I didn't understand 
2023-02-11 14:15:24 <psykose> it's a patch to apply to the file
2023-02-11 14:15:41 <psykose> you can probably type it by hand since it's like 9 characters
2023-02-11 17:58:51 <umk3> hi! just upgraded to the new xdg-desktop-portal-gtk and it installed lots of gnome-desktop dependencies
2023-02-11 17:59:25 <umk3> is it correct this apk build lists gnome dependencies, or should be distributed independently
2023-02-11 18:01:51 <umk3> i don't see what is wrong by having xdg-gtk completely apart gnome
2023-02-11 18:03:07 <psykose> from what i can see the extra dependency is just gnome-desktop
2023-02-11 18:03:43 <umk3> it comes along with 47 other packages
2023-02-11 18:04:09 <psykose> which ones
2023-02-11 18:04:22 <psykose> post the log so i can see
2023-02-11 18:05:19 <umk3> ok
2023-02-11 18:07:10 <umk3> https://haste.tchncs.de/najutisuca.properties weird URL name lmao
2023-02-11 18:07:29 <umk3> results of apk add xdg-desktop-portal-gtk
2023-02-11 18:15:08 <psykose> is this an actual machine or a random empty system
2023-02-11 18:15:15 <psykose> because if it's the latter it's not what i asked
2023-02-11 18:15:30 <psykose> i mean specifically on the upgrade you did and nothing else :p
2023-02-11 18:15:32 <umk3> it is my daily pc
2023-02-11 18:15:41 <psykose> alrightie
2023-02-11 18:16:08 <umk3> running on not so many packages, just labwc waybar and a few packages
2023-02-11 18:16:55 <umk3> thank you psykose 
2023-02-11 18:17:02 <psykose> what does it look like now
2023-02-11 18:17:38 <umk3> good
2023-02-11 18:17:45 <psykose> show show
2023-02-11 18:17:49 <umk3> only installs along libgnome-desktop-3
2023-02-11 18:17:58 <umk3> ok
2023-02-11 18:19:04 <umk3> https://haste.tchncs.de/opejifivac.yaml
2023-02-11 18:19:10 <psykose> neat
2023-02-11 18:19:14 <psykose> the only difference is the gtk4.0
2023-02-11 18:19:27 <psykose> if you had any gtk4 applications, it would've been identical
2023-02-11 18:19:43 <psykose> good opportunity to split more stuff anyway, so thanks :)
2023-02-11 18:19:47 <umk3> true
2023-02-11 18:19:52 <umk3> i dont have any gtk4 
2023-02-11 18:20:09 <umk3> thank you :)
2023-02-11 18:23:10 <umk3> the commit looks really good
2023-02-11 18:25:12 <psykose> :)
2023-02-11 18:25:28 <psykose> like 3rd thing today someone mentions something within 15 minutes of it being built :p
2023-02-11 18:25:40 <umk3> hah
2023-02-11 18:25:45 <umk3> coincidence
2023-02-11 18:25:48 <psykose> indeed
2023-02-11 20:31:45 <lgehr> hey! I just submited a v2 patch (add testing/minisatip) to the aports mailing list. It seems that this is not forwared to gitlab. Is there anything else I need to do?
2023-02-11 20:32:15 <ikke> lgehr: It's something I do manually
2023-02-11 20:32:31 <ikke> I just didn't get to it yet
2023-02-11 20:32:40 <ikke> fyi, the pkgrel bump is not necessary for CI
2023-02-11 20:32:47 <ikke> it only matters for the actual builders
2023-02-11 20:33:04 <ikke> (I can skip it from the current patch)
2023-02-11 20:33:46 <lgehr> good to know. no worries then, take your time. I just didn't know :D. I figured but i guess its better to be on the safe side. 
2023-02-11 20:48:46 <ikke> lgehr: fyi, I just updated the MR with your latest patch
2023-02-11 21:02:17 <lgehr> ikke: thx
2023-02-11 22:27:56 <lgehr> psykose: regarding testing/minisatip: should I just remove the sed-comments and remove check() or should I move them to check()? I don't quite understand what you want me to do.
2023-02-11 22:36:29 <skarnet> psykose: remind me what will go into -libs and what will go into -dev again?
2023-02-11 22:37:14 <psykose> <- .so.* -> .so/include
2023-02-11 22:37:17 <psykose> lgehr: did for you
2023-02-11 22:37:19 <ptrc> lgehr: it was fixed by ikke and merged; the point was to use a .patch file instead a sed
2023-02-11 22:37:28 <psykose> no
2023-02-11 22:37:34 <skarnet> so -libs is runtime and -dev is build-time. gotcha.
2023-02-11 22:37:52 <psykose> in some fashion
2023-02-11 22:38:21 <skarnet> can you update s6-networking at the same time? something about depends and depends_dev needing to be adjusted, if you make that change
2023-02-11 22:38:39 <skarnet> or maybe abuild is extra clever and does The Right Thing automagically
2023-02-11 22:39:18 <psykose> nothing has to be adjusted :)
2023-02-11 22:39:27 <skarnet> awesomesauce.
2023-02-11 22:39:38 <ptrc> psykose: wdym by 'no'..?
2023-02-11 22:39:45 <skarnet> 10/10 ship it
2023-02-11 22:40:02 <psykose> ptrc: neither a patch nor the correct fixer of the fixes
2023-02-11 22:40:05 <psykose> :p
2023-02-11 22:40:51 <ptrc> gitlab shows "Kevin Daudt changed this line in version 2 of the diff", so i assumed that's correct 🤔
2023-02-11 22:41:44 <ptrc> ohhh, are we talking about different seds from the file
2023-02-11 22:41:46 <ptrc> nevermind then
2023-02-11 22:42:56 <psykose> alas you have been bamboozled, for there was no such "merge request", it was but a ye-olde patch to the "do not send patches here" mailing-list'e
2023-02-11 22:44:54 <lgehr> psykose: thx!
2023-02-11 22:45:53 <lgehr> Also: git send-email is just more confy then the 'do not send patches' warning is scary :D
2023-02-11 22:47:15 <psykose> can indeed be comfy
2023-02-11 22:47:32 <psykose> and yes scary warning does not be keeping out the senders :p
2023-02-11 22:48:51 <psykose> i don't mind, i just don't really put in the effort to interact with it myself
2023-02-11 22:53:07 <lgehr> ...and it did work in the end :D. I will be gone now, thanks again to everybody helping me get this patch commited
2023-02-11 22:53:43 <psykose> take care :)
2023-02-12 03:40:04 <psykose> donoban: did you ever get somewhere with that zsh overflow int segfault getting fixed
2023-02-12 07:39:08 <donoban> no, I didn't followed it. I suppose that is not something exploitable. I can ask again on #zsh and open an issue if stills exist
2023-02-12 07:51:46 <psykose> don't care much about exploitable, it just is a crash :p
2023-02-12 07:55:45 <donoban> hehe yes but a crash after putting a uultraaa biiiig command input
2023-02-12 07:58:35 <donoban> I tried to understand that code and have the feeling that the word list is not used later for anything, so probaby I didn't understand it enough :\
2023-02-12 08:01:18 <donoban> maybe just changuing the  *short to a properly size type is ok for avoid it crashing, if the string is too big it will start getting words from the begin instead crashing
2023-02-12 08:02:48 <psykose> i mean
2023-02-12 08:02:55 <psykose> hitting tab with * somewhere in there isn't that rare
2023-02-12 08:03:16 <psykose> it's also always wrong to crash, only the impact is the variance, but it's always above zero in badness there :p
2023-02-12 10:14:53 <ddevault> ERROR: openfst-1.8.2-r0: BAD signature
2023-02-12 10:16:46 <ikke> can confirm
2023-02-12 10:17:59 <ikke> probably due to https://git.alpinelinux.org/aports/commit/?id=366f6981ae95 without bumping pkgrel
2023-02-12 10:21:01 <ikke> ddevault: fixed
2023-02-12 10:23:42 <ikke> repo-tools fastly purge pkg --repo testing --filename openfst-1.8.2-r0.apk
2023-02-12 10:24:59 <psykose> indeed, forgot
2023-02-12 10:25:01 <psykose> fixed
2023-02-12 10:47:16 <ddevault> ikke: thanks
2023-02-12 14:06:22 <ikke> hmm, seems like the ipcalc package does not override ipcalc from busybox
2023-02-12 14:06:28 <ikke> ipcalc package installs in /usr/bin/ipcalc
2023-02-12 14:06:32 <ikke> bb provides /bin/ipcalc
2023-02-12 14:23:19 <psykose> something something usrmerge
2023-02-12 14:23:25 <ikke> ahuh
2023-02-12 14:24:28 <psykose> in better news practically everything works with ffmpeg6
2023-02-12 14:25:57 <Ermine> Is it released?
2023-02-12 14:26:45 <psykose> next week irrc
2023-02-12 14:27:08 <Ermine> ah
2023-02-12 17:44:51 <donoban> does anyone know what means this python optimization levels? https://tpaste.us/YYve what should be more according to alpine?
2023-02-12 18:20:17 <donoban> well, let's optimize size? https://tpaste.us/5XY7 :P
2023-02-12 18:30:02 <donoban> psykose: I just see your MR
2023-02-12 18:42:13 <donoban> meh I pushed a commit on my master https://gitlab.alpinelinux.org/donoban/aports and now I can't force push my rebased local version
2023-02-12 18:42:22 <donoban>  ! [remote rejected]       master -> master (pre-receive hook declined)
2023-02-12 19:32:46 <ptrc> donoban: try going to repository settings, your master branch is probably "protected"
2023-02-12 19:37:34 <bl4ckb0ne> how come sway 1.8.1 came out one hour ago and alpine edge isnt up to date :O
2023-02-12 19:39:28 <craftyguy> 🤦
2023-02-12 19:42:03 <donoban> ptrc: thanks but do you have some clue about where is that option?
2023-02-12 19:42:15 <donoban> there are tons of them :D
2023-02-12 19:42:41 <ptrc> there's a section of settings named "repository"
2023-02-12 19:42:47 <ptrc> and there should be a table of branches
2023-02-12 19:43:31 <donoban> Protected branches
2023-02-12 19:43:32 <donoban>  ?
2023-02-12 19:43:42 <donoban> ah
2023-02-12 19:43:42 <ptrc> yup, there should be a toggle for force-pushing
2023-02-12 19:43:44 <donoban> I see
2023-02-12 19:43:46 <donoban> there is master
2023-02-12 19:43:52 <donoban> yeah, let's try gain
2023-02-12 19:44:04 <donoban>  + bffe920b20...92f46528ea master -> master (forced update)
2023-02-12 19:44:07 <donoban> great! thanks
2023-02-13 06:46:20 <psykose> donoban: it's just the .pyc at different opt levels by default
2023-02-13 06:46:52 <psykose> by default when you run a .py it creates a .pyc in memory, unless it was already there, the -opt variants are for when you run python -O or -OO
2023-02-13 06:46:58 <psykose> the optimisation doesn't do anything at all though
2023-02-13 06:47:07 <psykose> it's just a nice way to use more disk space
2023-02-13 07:08:03 <elibrokeit> well, it sort of does something.
2023-02-13 07:08:42 <elibrokeit> e.g. strips out documentation, more compact bytecode with unused development variables not taking up memory
2023-02-13 07:09:21 <elibrokeit> the thing is that no one ever ever ever actually runs python with -O or -OO
2023-02-13 07:12:01 <lucidiot> both remove assert statements, which can be fun when they are used to verify important things like `assert user.password = input` or something
2023-02-13 07:12:52 <lucidiot> i definitely did not find this out on an admin page in production
2023-02-13 07:36:49 <psykose> :D
2023-02-13 07:37:26 <psykose> i think the intent was sound (asserts should... not be used for that), but i guess it's just how people ended up writing python and relying on the assertion failures and catching them
2023-02-13 07:39:12 <psykose> and the trade off between "strip some heredocs and asserts that break the program" and "literally double the size of the bytecode on disk as an extra" is funny
2023-02-13 07:39:22 <psykose> if it was dynamically generate eh, go nuts
2023-02-13 08:12:01 <lucidiot> i should try to run pytest with -OO, basically volkswagen mode
2023-02-13 08:12:06 <lucidiot> ignore all assertions, make all the tests pass
2023-02-13 08:23:22 <psykose> fixed the failing tests ! time for lunch
2023-02-13 08:25:15 <lucidiot> optimizing both disk space and development time!
2023-02-13 09:06:24 <psykose> imagine if every source file you opened
2023-02-13 09:06:31 <psykose> you got 23 lines of documentation for the actual file
2023-02-13 09:06:36 <psykose> and not 23 lines of licence header
2023-02-13 09:06:44 <psykose> i'm dreaming of too much i guess
2023-02-13 09:20:52 <lucidiot> amend DMCA to only apply copyright to documented source code
2023-02-13 09:21:14 <lucidiot> 23 lines of license header and 23 lines of documentation, everyone is happy
2023-02-13 09:33:59 <hcs> This might be an RTFM kind of question, but I've looked everywhere. How do I get an openrc service to pass environment variables to a program? I get that there's better ways to configure programs, but microsoft dotnet REQUIRES that I set a few things in env.
2023-02-13 09:41:46 <psykose> export XYZ=123
2023-02-13 09:41:58 <psykose> works in conf.d as much as anywhere
2023-02-13 09:42:47 <hcs> I've set XYZ=123 in /etc/environment and export XYZ=123 in /etc/profile.d/dotnet.sh. I'm gonna try /etc/conf.d/something now
2023-02-13 09:52:51 <psykose> etc/environment isn't read by openrc or anything afaik
2023-02-13 09:53:01 <psykose> that's probably some pam read thing on user login iirc
2023-02-13 09:53:08 <ikke> psykose: I'm annoyed by that too (license header VS documentation) 
2023-02-13 09:53:26 <psykose> and profile.d is only used for login shells (or shells that source it whenever i guess), so nothing in openrc
2023-02-13 09:53:45 <psykose> or well, you source /etc/profile, which sources the .d for you
2023-02-13 09:53:54 <psykose> ikke: :)
2023-02-13 09:54:01 <psykose> also gitlab seems mighty slow today
2023-02-13 09:55:02 <psykose> hcs: so yeah, if it's an openrc service, you need `export` in the conf.d (or init.d if you're just hardcoding some stuff, doesn't matter)
2023-02-13 09:55:26 <psykose> openrc doesn't read anything else
2023-02-13 09:58:32 <hcs> Thanks, I'm trying it out now. None of the other /etc/conf.d files have export in front of their declarations, do you know why that is?
2023-02-13 09:58:57 <psykose> they're not env vars
2023-02-13 09:59:15 <psykose> just regular shell variables used in the init.d
2023-02-13 09:59:28 <hcs> Alright, fair enough.
2023-02-13 09:59:33 <psykose> i.e. you do xyz=123, then in the init.d you'll see `something $xyz`
2023-02-13 09:59:45 <hcs> Gotcha, thanks
2023-02-13 10:00:04 <psykose> openrc was made before anyone really used "configure with env vars" as a pattern, so there's nothing "exported" into the process by default
2023-02-13 10:00:26 <psykose> nor should any process manager export random stuff you don't specify- but i guess most people haven't realised they need to type export themselves :)
2023-02-13 10:00:59 <psykose> in systemd it's explicit as Environment= in the files so everyone got more used to it, no implicit shell-like stuff
2023-02-13 10:21:59 <hcs> Thanks psykose you've fixed my issue. Even though this was expected behaviour to you and many others, it seems to be completely undocumented. It's not in the script writing guide, and it's not on the arch/gentoo/alpine wikis.
2023-02-13 10:22:29 <psykose> it could be documented in openrc itself (you can open an issue in https://github.com/OpenRC/openrc/)
2023-02-13 10:22:50 <psykose> to me personally it's documented by my niche knowledge that it's all just a shell anyway and it doesn't clean the environment
2023-02-13 10:23:45 <hcs> I've seen a post on stackoverflow/exchange that states that the program "service" explicitly cleans the outside env aside from PATH and a few others.
2023-02-13 10:26:03 <psykose> it cleans it in the sense that if you start a service (which you do from your shell), it ignores your entire shell environment
2023-02-13 10:26:26 <psykose> the start() implicit function does not ignore the environment of the init.d file itself, or the conf.d sourced in it
2023-02-13 10:26:53 <psykose> (and so if you define your own start() in there and export inside it that should also work, in the sense of that when you call start-stop-daemon in there it will not clean the env)
2023-02-13 10:27:44 <psykose> i.e. rc-service cleans the env before invoking any functions- but the conf.d is after all that
2023-02-13 10:27:56 <psykose> it also doesn't clean the whole env
2023-02-13 10:28:12 <psykose> there is an exception for TERM :p
2023-02-13 10:28:13 <psykose> https://github.com/OpenRC/openrc/issues/582
2023-02-13 10:28:23 <psykose> which gets inherited from anywhere you call rc-service from
2023-02-13 10:28:28 <psykose> tl;dr it's all a mess
2023-02-13 11:26:04 <ikke> psykose: seems like gitlab is getting more requests as usual
2023-02-13 11:26:08 <psykose> hm
2023-02-13 11:26:13 <ikke> Not excessive, just more
2023-02-13 11:27:15 <ikke> https://i.imgur.com/uhbprPg.png
2023-02-13 11:38:20 <donoban> Hi, is there some problem with 
2023-02-13 11:38:23 <donoban> !42949 
2023-02-13 11:41:17 <psykose> ikke: huh, weird
2023-02-13 11:41:21 <psykose> why is it suddenly double today
2023-02-13 11:41:23 <psykose> what are they for
2023-02-13 11:41:29 <ikke> trying to figure out
2023-02-13 11:42:26 <ikke> someone is spamming https://gitlab.alpinelinux.org/viktorp2p/p2p-dohod/-/issues/new
2023-02-13 11:43:44 <psykose> what the fuck
2023-02-13 11:43:48 <psykose> lmao
2023-02-13 11:43:53 <psykose> can nuke ?
2023-02-13 11:44:02 <psykose> or you wanna investigate it more
2023-02-13 11:45:31 <ikke> -> iptables DROP :)
2023-02-13 11:45:53 <psykose> hehe
2023-02-13 11:46:02 <psykose> sure, i deleted it just to clear the huge amount of issues too
2023-02-13 11:46:04 <ikke> oof, 50k issues
2023-02-13 11:46:07 <ikke> yes
2023-02-13 11:46:08 <psykose> there's a second repo with 100k
2023-02-13 11:46:14 <donoban> lol
2023-02-13 11:46:25 <psykose> bearp2p/profit-crypto
2023-02-13 11:46:28 <psykose> but i nuked it already
2023-02-13 11:46:31 <ikke> thanks
2023-02-13 11:46:32 <psykose> 131k
2023-02-13 11:47:17 <psykose> https://gitlab.alpinelinux.org/rian/watch-hd/-/issues/6
2023-02-13 11:47:19 <psykose> nice link hosting
2023-02-13 11:47:52 <psykose> https://gitlab.alpinelinux.org/Sepinm/apk-noob/-/tree/main lol
2023-02-13 11:48:08 <psykose> not sure if this is very on topic even if it technically is
2023-02-13 11:50:25 <ikke> psykose: fyi: https://zabbix.alpinelinux.org/zabbix.php?action=dashboard.view&dashboardid=2
2023-02-13 11:50:33 <psykose> whee
2023-02-13 11:51:00 <psykose> that dashboard list is very cool, thanks
2023-02-13 11:51:04 <psykose> somehow i could not find that before
2023-02-13 11:51:28 <psykose> also love the pink/red load graphs :p
2023-02-13 11:51:47 <ikke> :)
2023-02-13 11:52:28 <psykose> spam seems to be dead
2023-02-13 11:53:53 <Ermine> how does it look like?
2023-02-13 11:54:03 <ikke> Russian
2023-02-13 11:55:11 <psykose> Ermine: you can see the picture above
2023-02-13 11:55:16 <psykose> https://i.imgur.com/uhbprPg.png
2023-02-13 11:55:16 <psykose> hehe
2023-02-13 11:55:29 <psykose> and then the 50k/130k issues in two repos
2023-02-13 11:56:08 <psykose> hm
2023-02-13 11:56:11 <psykose> there's more spam somewhere
2023-02-13 11:56:43 <psykose> aports is 13k issues, but there's 172k total on the instance
2023-02-13 11:56:58 <psykose> ah
2023-02-13 11:57:00 <psykose> busy deletion
2023-02-13 11:57:15 <psykose> or is it
2023-02-13 11:57:25 <psykose> yeah
2023-02-13 11:57:35 <psykose> ikke: look how long this takes https://gitlab.alpinelinux.org/admin/background_jobs
2023-02-13 11:57:44 <psykose> in busy
2023-02-13 11:58:44 <psykose> it deletes like 100 a second
2023-02-13 11:58:56 <psykose> slow, slow..
2023-02-13 11:59:42 <donoban> it should have a limit rate for creating too :P
2023-02-13 12:00:01 <ikke> it has, I regularly saw 429
2023-02-13 12:00:14 <ikke> but not on amount of issues
2023-02-13 12:00:32 <psykose> averaged out it was like.. 3 open issues a second for both?
2023-02-13 12:00:40 <psykose> for 10 hours
2023-02-13 12:01:05 <psykose> fuzzy math a bit, but not 100/s or 0.001/s
2023-02-13 12:17:18 <psykose> ikke: don't think you ever ran the note deleter, did you
2023-02-13 12:17:25 <psykose> given the 300k spam notes :p
2023-02-13 12:17:42 <psykose> or was that a snippet deleter
2023-02-13 12:17:45 <psykose> hm, what are notes
2023-02-13 12:38:01 <lucidiot> aren't notes comments?
2023-02-13 12:59:13 <psykose> so that's what it is
2023-02-13 12:59:19 <psykose> nice and consistent naming
2023-02-13 12:59:28 <psykose> if you google "gitlab notes" all you find is like admin notes
2023-02-13 13:07:23 <lucidiot> first result on DDG is https://docs.gitlab.com/ee/api/notes.html
2023-02-13 13:07:39 <lucidiot> google wants to annoy gitlab users
2023-02-13 13:08:35 <lucidiot> that's definitely not going to be confused with `git notes` either
2023-02-13 13:11:22 <psykose> indeed
2023-02-13 13:11:29 <psykose> recently can't find shit on google
2023-02-13 13:11:48 <psykose> if you want to have fun, try finding literally anything on the D programming language for specific queries related to the api and versions
2023-02-13 13:26:14 <donoban> psykose: lol, I just found your bubblejail MR on my SPAM folder, there are hundreds of messages from gitlab
2023-02-13 13:26:56 <donoban> anyway I'm gonna disable this mail notifications
2023-02-13 13:27:21 <psykose> sure, you get notified for things you are subscribed to (like things you comment on)
2023-02-13 13:29:15 <donoban> I'm subscribed to whole aports, definitevely too much notifications for me :\
2023-02-13 13:30:13 <psykose> lol haha
2023-02-13 13:30:17 <psykose> don't do that
2023-02-13 13:30:30 <psykose> even i don't do that (i subscribe to everything except aports)
2023-02-13 13:30:35 <hcs> Where do I see which packages are included on the extended iso?
2023-02-13 13:30:45 <psykose> scripts/mkimg.standard.sh has an extended list
2023-02-13 13:30:52 <psykose> (in aports)
2023-02-13 13:31:03 <psykose> https://git.alpinelinux.org/aports/tree/scripts/mkimg.standard.sh#n29
2023-02-13 13:31:27 <hcs> Wonderful, thanks. I'm blind.
2023-02-13 13:31:35 <psykose> nah
2023-02-13 13:31:38 <psykose> hard to find this stuff really
2023-02-13 13:33:07 <hcs> Is there any chance I could get a relatively small package added to the extended list? I know you're trying to keep everything small, but we're talking less than a megabyte and no deps
2023-02-13 13:33:23 <psykose> which one
2023-02-13 13:33:26 <hcs> lz4
2023-02-13 13:33:36 <hcs> The kernel has support but userspace does not
2023-02-13 13:34:38 <psykose> hmm
2023-02-13 13:35:04 <psykose> usually the intent is mostly that the isos are just to be able to install and get an internet connection going
2023-02-13 13:35:21 <psykose> (extended has stuff that would let wifi work, etc)
2023-02-13 13:35:32 <psykose> so i'm not sure what support is needed inside an iso for anything related to compression
2023-02-13 13:35:48 <psykose> (not that there isn't a bunch of garbage not for that in extended, i'm just curious what the point is)
2023-02-13 13:36:06 <psykose> you're not meant to boot them as images to run anything on
2023-02-13 13:36:07 <hcs> Our specific usecase is decompressing images before writing to disk with dd
2023-02-13 13:36:46 <hcs> We're actually using the alpine iso for installation in this case, we're just not installing alpine using the installer.
2023-02-13 13:37:01 <psykose> i.e. you boot extended, there is another image attached somewhere, you dd the attached image into the disk, and there is no networking?
2023-02-13 13:37:02 <hcs> (we're writing prebuilt images instead)
2023-02-13 13:38:04 <hcs> Yes curl http://somewhere/image.bin.lz4 | lz4 -dc | dd of=/dev/sda or something like that. It's more complicated but that's the gist
2023-02-13 13:39:07 <hcs> We often use a usb stick for the image aswell, that's why I want lz4 in the extended iso
2023-02-13 13:39:27 <hcs> lz4 is our sole reason for wanting networking when flashing images from usb
2023-02-13 13:43:54 <Ermine> But you also need networking to load image.bin.lz4
2023-02-13 13:44:29 <maribu> psykose: It looks like that there are issues with KiCad and GLEW :/ Hardware acceleration fails due to "Unknown Error". Quite descriptive :/
2023-02-13 13:44:43 <psykose> which kind of hardware accel
2023-02-13 13:44:55 <maribu> OpenGL on AMD
2023-02-13 13:44:56 <psykose> did you do anything other than open it? what window manager are you in
2023-02-13 13:45:04 <psykose> it works on sway for me
2023-02-13 13:45:45 <maribu> swaywm and I opened the schematics editor. (I'm not entirely sure where 3D acceleration comes into play on the schematics editor, though. But by default anti-aliasing is only enabled with hw accel on.)
2023-02-13 13:45:45 <psykose> (it's definitely my fault regardless and i have some stuff you can test, but perhaps we can narrow it down a bit)
2023-02-13 13:46:31 <psykose> https://img.ayaya.dev/23G0J3iflMzM this? good catch
2023-02-13 13:46:48 <maribu> Exactly
2023-02-13 13:47:24 <psykose> ok fixed it
2023-02-13 13:47:28 <psykose> i did not read something properly :p
2023-02-13 13:50:06 <maribu> That was quick :) I still haven't even figured out what the issues was :D
2023-02-13 13:50:07 <psykose> next glew should work
2023-02-13 13:50:22 <psykose> i applied a busted patch to glew for egl since i needed to get eglew.h installed anyway
2023-02-13 13:50:39 <psykose> but everything seems to work anyway, patch is probably fixed for libOpenGL and not something for regular egl
2023-02-13 13:50:52 <psykose> i did test it with some games that used glew but apparently this is different
2023-02-13 13:51:16 <psykose> since probably the others don't use egl.. funny that.. cause if they did they couldn't have until now because the header was missing.. smart me..
2023-02-13 13:52:49 <psykose> hcs: if you can "curl somewhere" you can also "apk add lz4", unless this is some weird internal only networking
2023-02-13 13:53:15 <hcs> psykose: Ermine: it just occurred to me that I could apk fetch the package and then just install it offline to fix my issue. Having it baked in would be more nifty but this works aswell
2023-02-13 13:54:04 <psykose> if you had said you want zstd in the image i would've been more convinced :p
2023-02-13 13:54:09 <hcs> Regarding "curl somewhere", we also write images residing on a usb stick I'm trying to make the entire procedure work completely offline though
2023-02-13 13:54:53 <psykose> right, the other case i mentioned of "another attached image to write"
2023-02-13 13:54:54 <hcs> Ehh I don't why zstd would matter more, they're both commonly used compression algorithms
2023-02-13 13:56:57 <psykose> (offtopic) i don't think lz4 makes much sense for the specific usecase
2023-02-13 13:57:22 <psykose> it's mostly favored for very fast compression/decompression in realtime and similar, not an image you want as small as possible to then (also perhaps) quickly write to something
2023-02-13 13:57:25 <hcs> Oh I completely agree
2023-02-13 13:58:15 <psykose> for which zstd is actually perfect- the ratio is top tier and you still get very fast decode (beaten by lz4, but for writing one image in the range of 100-1000MB they are both near instant to decompress, unless this is microcontrollers)
2023-02-13 13:58:22 <psykose> but sure, let me add some compression utils to extended
2023-02-13 13:59:30 <hcs> I agree with using something else for the images, but that's not my choice unfortunately. We have a bunch of other systems building this stuff
2023-02-13 13:59:41 <hcs> Thanks for adding the compression tools, that sounds good
2023-02-13 14:02:09 <psykose> b0b993207b5c84f3b20e81ce1e7df216d3e867fd
2023-02-13 14:02:17 <psykose> i won't backport it, but i guess it'll be in 3.18 isos
2023-02-13 14:04:46 <hcs> Thanks, that sounds good!
2023-02-13 14:07:02 <psykose> (also i typod to not add xz in there, then added it, then removed it because i forgot busybox unxz exists..)
2023-02-13 14:14:33 <skarnet> honestly xz is fine and for now I see it used more often than zstd
2023-02-13 14:15:17 <skarnet> my stance is "I'll use xz until a majority of people have shifted to zstd and zstd tooling is everywhere, at which point I can switch to zstd as well"
2023-02-13 14:15:28 <skarnet> which is obviously not helping making the change
2023-02-13 14:27:36 <psykose> xz is great if you love things being slow as fuck yeah
2023-02-13 14:28:18 <maribu> psykose: Updating to glew r5 didn't fix the issue for me. Still the same error message and no output to either stdout or stderr. Have you checked that you are indeed using hardware acceleration? It has to be re-enabled in the settings prior to opening the schematics editor. A failure in glew will disable it permanently until acceleration is manually re-enabled.
2023-02-13 14:28:25 <psykose> hmm
2023-02-13 14:28:28 <psykose> ah
2023-02-13 14:28:33 <psykose> didn't know it stays disabled
2023-02-13 14:29:58 <psykose> where is this option..
2023-02-13 14:31:49 <maribu> In the main window's menu: preferences -> preferences. In the new preferences window select the tab "Schematics Editor ->Display Options" in the left panel. In the top left corner under "Rendering Options" it can be re-enabled. 
2023-02-13 14:36:31 <psykose> indeed broken
2023-02-13 14:36:33 <psykose> ok, let me see
2023-02-13 15:00:31 <psykose> guess back to vendored it is, i'll debug it some other time
2023-02-13 15:00:36 <psykose> works with that one i think
2023-02-13 16:05:41 <maribu> psykose: Thanks, the vendored in version works fine for me. I think Alpine was about the first distro to package the new release. Let's see how the others are handling this. Maybe someone find a solution to let KiCad behave well with upstreams glew.
2023-02-13 16:06:07 <psykose> it's probably more us-specific, idk
2023-02-13 16:06:20 <psykose> GL stuff is a bit of a mess with everyone using glvnd for the base layer but us :p
2023-02-13 16:06:52 <psykose> which isn't a bad thing, but i don't know enough about all of it to get it to play nice
2023-02-13 16:07:02 <psykose> then again, glew/glfw/.. are just always a mess
2023-02-13 16:16:12 <skarnet> "gl ew" sounds accurate
2023-02-13 16:20:29 <invoked> (preface: i know little about the gl stuff) what is/are the reason(s) for not using libglvnd on alpine
2023-02-13 16:22:17 <psykose> the point of glvnd is to provide a fully abi compatible libGL, libEGL, etc that then dynamically dispatches to one of your choice at runtime
2023-02-13 16:23:11 <psykose> (there is also libOpenGL from it, that's something else i guess, since normally libGL requires linking to libX11 to be usable for glx or some shit, but with libOpenGL you can have a "clean" one..)
2023-02-13 16:23:15 <psykose> ((i don't know if that's accurate))
2023-02-13 16:23:17 <psykose> now the issue is
2023-02-13 16:23:32 <psykose> who are these "choices" "of your choice"?
2023-02-13 16:23:51 <psykose> mesa provides libGL for instance- and it doesn't provide *multiple*, it's just "mesa"
2023-02-13 16:23:57 <psykose> so it's a vendor other than mesa you could use
2023-02-13 16:24:11 <psykose> i.e... you guessed it! nvidia!
2023-02-13 16:24:29 <invoked> is libglvnd what happened after nvidia's kernel shim was rejected?
2023-02-13 16:24:29 <psykose> you can have drop-in side by side nvidia libgl in some folder, and mesa in some folder, and dynamically load one per-application with some env var
2023-02-13 16:24:39 <psykose> as you can gues
2023-02-13 16:24:45 <psykose> this is completely fucking useless on alpine
2023-02-13 16:24:57 <psykose> because every other "vendor" is a proprietary glibc driver
2023-02-13 16:25:26 <Newbyte> (and the "amber" branch of Mesa)
2023-02-13 16:25:27 <psykose> the only benefit is the libOpenGL- because things do use that nowadays, and you have to sed it to libEGL to get it to build
2023-02-13 16:25:29 <invoked> ok gotcha.
2023-02-13 16:25:35 <psykose> and it comes with issues like https://gitlab.freedesktop.org/glvnd/libglvnd/-/issues/222
2023-02-13 16:25:52 <psykose> and yes, as i was going to get to- the only use for it would be amber
2023-02-13 16:26:07 <invoked> thanks for the explanation
2023-02-13 16:26:08 <psykose> which is just an old mesa branch for some old intel cpus
2023-02-13 16:26:40 <Newbyte> I thought it included AMD drivers too
2023-02-13 16:26:48 <Newbyte> or perhaps, ATI
2023-02-13 16:27:09 <Ermine> What about i.e. replacing radeonsi with zink?
2023-02-13 16:27:12 <psykose> the full list has like 6 vendors and some amd drivers too etc
2023-02-13 16:27:21 <psykose> Ermine: that you can just do
2023-02-13 16:27:38 <Ermine> without glvnd? Okay
2023-02-13 16:27:49 <psykose> MESA_LOADER_DRIVER_OVERRIDE=zink
2023-02-13 16:27:55 <psykose> and have the vulkan ones installed
2023-02-13 16:28:17 <psykose> but if you have a working gl driver there is no point for that
2023-02-13 16:28:24 <Ermine> I mean, this is implemented without glvnd?
2023-02-13 16:28:26 <psykose> radeonsi will just always be faster (unless one day..)
2023-02-13 16:28:32 <psykose> it's not related to glvnd
2023-02-13 16:28:35 <psykose> so yes
2023-02-13 16:28:48 <Newbyte> Ermine: the reason the "amber" Mesa is a problem is because it's a separate codebase that needs to be built independently from the regular Mesa
2023-02-13 16:29:03 <psykose> for reference https://docs.mesa3d.org/amber.html
2023-02-13 16:29:28 <psykose> r200/radeon is like truly ancient amd gpus though, even older than i965 iirc
2023-02-13 16:29:44 <psykose> and i965/i915 have gallium counterparts (crocus/i915g)
2023-02-13 16:29:58 <psykose> they don't always work (i saw someone that didn't have a working gpu on alpine anymore in dri-devel)
2023-02-13 16:30:54 <psykose> which is unfortunate but having everything pay the cost of glvnd for some 20 year old gpus via amber is a little ridiculous
2023-02-13 16:34:41 <Ermine> It's the choice of tradeoffs: speed vs hardware coverage
2023-02-13 16:35:01 <Ermine> Arguably some people may want to use Alpine for ancient machines which have ancient gpus
2023-02-13 16:36:31 <psykose> in some sense yes
2023-02-13 16:39:22 <elly> yeah but trading speed on the vast majority of hardware for compat with certain old hardware is not necessarily a good idea
2023-02-13 16:43:48 <psykose> i don't get to fix things on 32-bit archs all the time for someone to tell me i don't care about old hardware smh
2023-02-13 16:46:54 <bl4ckb0ne> how come abuild-meson needs meson in makedeps
2023-02-13 16:47:31 <ikke> abuild-meson is a helper that comes with abuild, but that does not mean all packages depend on meson
2023-02-13 16:47:45 <bl4ckb0ne> 
ACTION hides his commit amendment

2023-02-13 16:48:24 <psykose> abuild-meson comes with meson
2023-02-13 16:48:53 <psykose> so i'm confused what makedeps and where
2023-02-13 16:49:16 <bl4ckb0ne> in my fresh new apkbuild
2023-02-13 16:49:33 <bl4ckb0ne> i forgot to add meson in hopes that it was automatically there since im using abuild-meson
2023-02-13 16:49:38 <psykose> lol
2023-02-13 16:49:40 <psykose> nah
2023-02-13 16:49:43 <bl4ckb0ne> only to have my failure thrown at my face by the CI in my mail
2023-02-13 16:49:55 <bl4ckb0ne> i brought dishonor to my family
2023-02-13 16:50:03 <psykose> this shit literally came out 10 minutes ago
2023-02-13 16:50:14 <bl4ckb0ne> fastest apkbuild in the west
2023-02-13 16:50:29 <psykose> you forgot every subpackage
2023-02-13 16:50:34 <bl4ckb0ne> i dont want to fix the aliasing issues in my codebase
2023-02-13 16:51:54 <bl4ckb0ne> did I
2023-02-13 16:52:05 <ikke> psykose: and yes, I only deleted snippets, not notes
2023-02-13 17:44:51 <psykose> elly: is the jitter in the room with us right now?
2023-02-13 17:45:14 <elly> lol
2023-02-13 17:45:19 <elly> don't be mean :P
2023-02-13 17:45:32 <psykose> good point
2023-02-13 17:45:42 <psykose> this tuna thing does look cool though
2023-02-13 17:46:05 <psykose> https://git.kernel.org/pub/scm/utils/tuna/tuna.git/
2023-02-13 17:47:14 <minimal> psykose: is it dolphin-friendly though? ;-)
2023-02-13 17:47:29 <psykose> hopefully :(
2023-02-13 17:57:23 <elly> gold ethernet cable energy
2023-02-13 17:58:55 <skarnet> ah yes, a GUI, which eats up considerable resources, to set thread and irq affinity, which are a tweak of two or three orders of magnitude of precision
2023-02-13 17:59:16 <psykose> typical swedes
2023-02-13 17:59:40 <alpine-meetbot> skarnet meant to say: ah yes, a GUI, which eats up considerable resources, to set thread and irq affinity, which are a tweak of two or three more orders of magnitude of precision
2023-02-13 17:59:40 <skarnet> s/two or three/two or three more/
2023-02-13 18:59:00 <Piraty> where are packages after `abuild rootbld` ?
2023-02-13 18:59:44 <ptrc> you mean built .apks? in the same place as usua
2023-02-13 18:59:47 <ptrc> usual*
2023-02-13 19:00:13 <Piraty> and what is "usual", if you'd be so kind to tell me
2023-02-13 19:01:07 <psykose> home/packages/$repo
2023-02-13 19:01:30 <psykose> repo being the folder the package folder is in
2023-02-13 19:01:41 <psykose> so main/package goes to home/packages/main
2023-02-13 19:01:53 <psykose> there is a var to put it elsewhere iirc
2023-02-13 19:02:03 <psykose> REPODEST
2023-02-13 19:02:12 <psykose> then it's repodest/$repo
2023-02-13 19:02:13 <Piraty> $HOME/packages is what i wasn't aware of
2023-02-13 19:02:18 <Piraty> thank you
2023-02-13 19:02:42 <psykose> np :)
2023-02-13 19:02:47 <Piraty> not you ptrc
2023-02-13 19:04:38 <Piraty> must be arch linux user
2023-02-13 19:07:19 <ptrc> ...hm?
2023-02-13 19:07:26 <Piraty> i have built some patched packages from my aports checkout. pkgrel is incremented, can i make a system upgrade and have apk take the built apk into account in the same run?
2023-02-13 19:08:29 <psykose> you have to put that repo path in /etc/apk/repositories
2023-02-13 19:08:35 <psykose> i.e. /home/you/packages/$repo
2023-02-13 19:08:42 <psykose> then yes, apk upgrade and you'll get it
2023-02-13 19:10:01 <psykose> also don't be rude, void linux user
2023-02-13 19:10:10 <Piraty> great. thanks again
2023-02-13 19:10:44 <Piraty> if you think calling someone an arch user is rude, you are odd. and it makes me wonder what you think of void linux
2023-02-13 19:11:25 <psykose> it's a great line to say seriously if you are a child, yes
2023-02-13 19:12:16 <ikke> Please stop calling eachother names please
2023-02-13 19:12:25 <ikke> and be nice
2023-02-13 19:13:09 <psykose> that also
2023-02-13 19:14:06 <skarnet> bunch of distro users
2023-02-13 19:14:27 <ikke> Now that's rude :)
2023-02-13 19:17:33 <Piraty> i think i need apk to trust my key
2023-02-13 19:18:01 <Piraty> at least warns about untrusted, and doesn't list my *.apk
2023-02-13 19:18:06 <psykose> abuild-keygen -ai would have installed it
2023-02-13 19:18:26 <psykose> otherwise, copy ~/.abuild/*.pub to /etc/apk/keys/
2023-02-13 19:18:29 <psykose> (don't rename the file)
2023-02-13 19:19:02 <psykose> (the filename is part of the signature)
2023-02-13 19:21:45 <Piraty> yeah it works now. thanks again
2023-02-13 19:37:53 <minimal> skarnet: I didn't realise the collective term for distros was "bunch" ;-)
2023-02-13 19:39:23 <skarnet> "a murder of distro users" would be ambiguous, for good reason
2023-02-13 20:07:16 <bl4ckb0ne> Piraty: i use --allow-untrusted when its my own packages
2023-02-13 20:07:28 <psykose> it takes 4 seconds to install the key
2023-02-13 20:08:34 <bl4ckb0ne> never bothered
2023-02-13 20:09:33 <Piraty> i won't perform system upgrade involving remote repositories with that flag
2023-02-13 20:10:35 <psykose> it takes as long to type --allow-untrusted once in the command line as copying the key
2023-02-13 20:11:29 <bl4ckb0ne> i just ctrl-r to fill the line 
2023-02-13 22:25:41 <elibrokeit> bl4ckb0ne: I wanna see a patch for abuild that detects whenever people run meson or cmake in the APKBUILD and adds the relevant build dependency for free
2023-02-13 22:25:56 <elibrokeit> do it for cargo too
2023-02-13 22:31:55 <psykose> nah
2023-02-13 22:32:34 <psykose> magic introspection all the time read the file to save someone typing 5 characters in makedepends is the definition of awful feature
2023-02-13 22:33:03 <psykose> you'd even have to parse the entire file in a special syntax-detecting way, because one can type cmake in comments/strings
2023-02-13 22:33:27 <psykose> that is strongest nack i could possibly give it
2023-02-13 22:34:09 <psykose> people usually take the opposite approach of build_style=cmake and it autofills the very standard build/install though :p
2023-02-13 22:34:15 <skarnet> who cares about comments, just grep and let heuristics heurist! if someone mentions cmake in an autotools project that's on them!
2023-02-13 22:34:29 <skarnet> -- random engineer, probably
2023-02-13 22:35:24 <psykose> in some sense it would've been nice to have some build style thing for autotools/meson (not cmake) i guess, but the longer i work on ""build recipes"" the less i care for such things
2023-02-13 22:36:05 <psykose> they're very thin frequently broken abstractions that are hard to comprehend for people not versed in the Specific Flavor They Suddenly Have To Look At that only get worse the more magic they do
2023-02-13 22:36:32 <skarnet> one day people will understand that magic is bad
2023-02-13 22:36:35 <skarnet> one day
2023-02-13 22:36:48 <psykose> sometimes you think "yeah would've been nice if i could've edited some args without sed'ing files" but then you have to consider all the headaches you didn't have too
2023-02-13 22:37:03 <psykose> gentoo is the best example really :p
2023-02-13 22:37:33 <skarnet> so you're saying gentoo is actually the best example of *something* ? o.O
2023-02-13 22:37:39 <psykose> because not only do you have pages of abstraction, you still update it all the damn time as you modernise templates for new versions instead.. so it's still doing all the work..
2023-02-13 22:37:54 <psykose> obviously a bit apples to oranges, but
2023-02-13 22:38:23 <sam_> elibrokeit: or, whisper it, eclasses
2023-02-13 22:38:26 <sam_> i mean it works fine for us
2023-02-13 22:38:27 <psykose> a little magic is okay, like the minimal $distro-meson thing people ship, because those are indeed super generic and practically always applicable args
2023-02-13 22:38:37 <psykose> sam_: look no hard feelings bud
2023-02-13 22:38:40 <psykose> but i used gentoo for years
2023-02-13 22:38:47 <psykose> i still had no clue what the fuck was happening in there
2023-02-13 22:38:57 <sam_> for what tho
2023-02-13 22:38:59 <psykose> :p
2023-02-13 22:39:09 <psykose> you mean using it? primary machine
2023-02-13 22:39:16 <sam_> no sorry, I mean in there = in ???
2023-02-13 22:39:26 <sam_> some of the eclasses are definitely stupidly complicated
2023-02-13 22:39:44 <psykose> understanding how what you typed translated into a package
2023-02-13 22:40:02 <psykose> tbf it had its nice parts, like how i just typed some arg=var and it built a whole cmake project with git submodules and all
2023-02-13 22:40:40 <psykose> but when anything doesn't work suddenly having to figure out how to debug one of the 9 emerge build steps and where in what someplace something broke.. for a beginner it wasn't that fun
2023-02-13 22:40:45 <psykose> nowadays it would be fine
2023-02-13 22:40:49 <skarnet> shows where the priorities are
2023-02-13 22:40:56 <psykose> i got better at the head->wall banging part
2023-02-13 22:41:09 <skarnet> the wall->head part is still eluding you?
2023-02-13 22:41:15 <c7s> so that's why they call it bash
2023-02-13 22:41:16 <psykose> yes :(
2023-02-13 22:41:16 <sam_> i think a good example of something problematic is defaulting to BUILD_SHARED_LIBS on
2023-02-13 22:41:21 <sam_> because it's actually a good idea most of the time
2023-02-13 22:41:26 <sam_> but also it's fucking confusing if it turns out your thing doesn't build
2023-02-13 22:41:46 <psykose> ugh build_shared_libs is like one of the most annoying parts of cmake ecosystem
2023-02-13 22:41:50 <psykose> because meson does that shit like
2023-02-13 22:41:53 <psykose> literally 500000000000000000000x better
2023-02-13 22:41:59 <psykose> for one you can output as shared+static by default
2023-02-13 22:42:07 <psykose> for another it doesn't do wildly different things per project
2023-02-13 22:42:14 <psykose> or constantly install project-internal private libs as .so
2023-02-13 22:42:25 <sam_> so i do sympathise / get it
2023-02-13 22:42:29 <sam_> god i hate that so much
2023-02-13 22:42:31 <sam_> + with fucked rpath always
2023-02-13 22:42:33 <psykose> (you usually never want it for non-libs)
2023-02-13 22:42:37 <psykose> haha yeah
2023-02-13 22:43:11 <psykose> cmake needing weird special stuff everywhere is why it happens so much
2023-02-13 22:43:19 <psykose> you come across cmake_magic() that does what you want
2023-02-13 22:43:41 <psykose> ..but you forget the 50 args to get it to work (EXCLUDE_FROM_ALL is very common), so everything breaks for distro packaging when you want it clean
2023-02-13 22:43:56 <psykose> or just passing STATIC to submodules (that gets broken with shared above), etc
2023-02-13 22:44:26 <skarnet> I honestly wonder how cmake got out of the trash can
2023-02-13 22:44:32 <psykose> gov fundment
2023-02-13 22:44:34 <psykose> windows support
2023-02-13 22:44:48 <skarnet> so, financial power eh
2023-02-13 22:44:57 <psykose> (there literally is like nothing on windows at all, only meson also exists now but it came way later)
2023-02-13 22:45:13 <psykose> (yes, mingw doesn't count, nobody cares)
2023-02-13 22:45:34 <psykose> ((also a bit apples/oranges for that one))
2023-02-13 22:46:24 <sam_> mingw lol
2023-02-13 22:46:26 <sam_> even the name is so icky
2023-02-13 22:46:30 <psykose> :3
2023-02-13 22:58:48 <elly> it's literally windows support that carried it I think
2023-02-13 23:15:37 <bl4ckb0ne> elibrokeit: ill do it but not for cargo
2023-02-13 23:28:03 <elibrokeit> bl4ckb0ne: hold on did you legitimately take me seriously
2023-02-13 23:30:56 <bl4ckb0ne> did you?
2023-02-13 23:32:10 <psykose> bl4ckb0ne: wen mesa 9999 fps dirt rally ?
2023-02-13 23:32:43 <elibrokeit> bl4ckb0ne: I can't be sure, the channel got a lot more serious about it than I expected...
2023-02-13 23:32:54 <psykose> i just sniped any preemptive discussion
2023-02-13 23:33:17 <psykose> read: got baited
2023-02-13 23:33:42 <psykose> (your irc life would go more better with an /ignore on me :p)
2023-02-13 23:34:12 <bl4ckb0ne> btw the chan gets a more lot serious if you /ignore psykose 
2023-02-13 23:34:13 <elibrokeit> but it would be less fun
2023-02-13 23:36:55 <psykose> bl4ckb0ne: lmao you played that perfectly :p
2023-02-13 23:39:29 <skarnet> damn I was afk and missed the chance
2023-02-13 23:39:37 <skarnet> grats for grabbing it, bl4ckb0ne
2023-02-13 23:40:15 <bl4ckb0ne> hehe
2023-02-13 23:45:47 <elibrokeit> I was just sitting here panicking, like "why are 4 different people getting into a very intense, serious discussion about that joke"
2023-02-13 23:45:48 <elibrokeit> help
2023-02-13 23:47:21 <psykose> welcome to psykose sleep deprivation sniping hour, we have snacks and refreshments near the entrance
2023-02-13 23:53:12 <bl4ckb0ne> same but snacks and refreahments cost 5 bucks but the ambiant music is better
2023-02-14 02:10:08 <elly> is there a list somewhere of how people might help out with Alpine development?
2023-02-14 02:16:41 <psykose> don't think so
2023-02-14 02:16:48 <psykose> hmm
2023-02-14 02:17:07 <psykose> uh there's https://wiki.alpinelinux.org/wiki/Contribute
2023-02-14 02:17:12 <psykose> not sure if it's up to date or not
2023-02-14 02:18:08 <psykose> personally i like bug reports, testing stuff, and (developers only) fixing difficult issues in whatever
2023-02-14 02:18:14 <psykose> that page pretty much covers every generic base
2023-02-14 02:18:31 <psykose> docs are of course nice, we did have https://docs.alpinelinux.org/ but currently i don't think anyone is writing towards it
2023-02-14 02:19:05 <elly> aha, cool :)
2023-02-14 02:19:48 <psykose> since you like non-wiki docs and writing, you can write things in that style if you wish, the repo is https://gitlab.alpinelinux.org/alpine/docs/user-handbook
2023-02-14 02:20:46 <elly> I do
2023-02-14 02:20:54 <elly> is there a list of like, most wanted bugfixes?
2023-02-14 02:21:01 <psykose> hrm
2023-02-14 02:21:04 <psykose> they are all upstream fyi
2023-02-14 02:21:14 <psykose> i.e. a legit thing to fix in code
2023-02-14 02:21:24 <psykose> (i.e. my mental list)
2023-02-14 02:21:27 <psykose> so, there is uhh
2023-02-14 02:21:51 <psykose> #13735
2023-02-14 02:21:52 <psykose> #14105
2023-02-14 02:21:54 <psykose> probably the same issue
2023-02-14 02:22:07 <psykose> gcc, upstream maybe is linked bugzilla
2023-02-14 02:22:32 <psykose> https://sourceware.org/bugzilla/show_bug.cgi?id=25694
2023-02-14 02:22:39 <psykose> binutils, this causes every riscv64 "textrels" issue
2023-02-14 02:22:49 <psykose> this isn't even musl specific or anything lol
2023-02-14 02:23:23 <psykose> https://gitlab.alpinelinux.org/alpine/aports/-/issues/13962
2023-02-14 02:23:37 <psykose> this requires a complete rework of gcc libstdc++ locale handling (the "posix" backend)
2023-02-14 02:23:52 <psykose> i.e. a port from the dragonflybsd one (maybe) or a full rewrite-ish
2023-02-14 02:24:04 <psykose> (glibc doesn't use it, they have the gnu one)
2023-02-14 02:24:15 <psykose> (in libstdc++ that is)
2023-02-14 02:24:39 <psykose> that a nice short list or do you want more stuff to pick from
2023-02-14 02:24:46 <psykose> maybe i should make a personal webpage
2023-02-14 02:25:20 <elly> wiki page :)
2023-02-14 02:25:34 <psykose> hah
2023-02-14 02:25:40 <psykose> Bounty Board $$$
2023-02-14 02:29:43 <elly> I should get back to my apk documentation
2023-02-14 02:29:51 <psykose> perhaps
2023-02-14 02:30:58 <elly> I find the gitlab MR workflow pretty uncomfortable, I wish I was mailing patches
2023-02-14 02:32:07 <psykose> it's harder to set up in general
2023-02-14 02:32:23 <psykose> but personally i find it much easier to edit stuff via this one when i work with it all the time
2023-02-14 02:32:27 <psykose> that and editing is not cringe
2023-02-14 02:32:34 <psykose> nobody wants to see patch v50
2023-02-14 02:33:44 <psykose> there's also musl docs? i don't know how that is contributed
2023-02-14 02:34:11 <psykose> https://wiki.musl-libc.org/
2023-02-14 02:34:18 <psykose> https://github.com/somasis/musl-wiki
2023-02-14 02:34:26 <elly> how does one actually edit patches in gitlab?
2023-02-14 02:34:41 <psykose> git commit --amend && git push -f
2023-02-14 02:34:46 <elly> ah, okay, that's what I do
2023-02-14 02:34:49 <psykose> remote branch is state
2023-02-14 02:35:19 <psykose> admittedly that is faster than --amend && git send-email ..--.. <editor stuff, typing> as a default
2023-02-14 02:35:31 <psykose> as a counter nobody documents any v's, but i find this is not needed for most work at all
2023-02-14 02:35:38 <psykose> how does one work with patches in the google house
2023-02-14 02:37:58 <psykose> reading those musl commits seems there are a ton of distros huh
2023-02-14 02:38:09 <psykose> i forget there are these small few-dozen-people distros sometimes
2023-02-14 02:38:12 <psykose> https://merelinux.org/ for one
2023-02-14 02:38:52 <psykose> ah that isn't even started
2023-02-14 02:38:53 <psykose> hehe
2023-02-14 02:40:31 <psykose> ah, was in the wrong repo..
2023-02-14 02:41:32 <elly> for patches we just all use gerrit
2023-02-14 02:41:36 <elly> at least for chromium
2023-02-14 02:42:09 <psykose> yeah, but what does "use gerrit" mean (i don't personally know))
2023-02-14 02:42:16 <psykose> i know the reviewing part if it's just the ui
2023-02-14 02:42:52 <elly> ah! the workflow is basically: hack hack hack; git commit; git cl upload (uploads to gerrit for review); get comments; hack hack hack; git commit --amend; git cl upload; ...
2023-02-14 02:43:08 <psykose> haha
2023-02-14 02:43:27 <psykose> well, that's basically gitlab once you remember what the push state corresponds do, just push -f instead of cl wrapper
2023-02-14 02:43:42 <elly> yeah, I don't remember where to push -f to
2023-02-14 02:43:51 <psykose> no target, default
2023-02-14 02:43:54 <elly> once again I have created a merge request against *my own fork* of apk-tools
2023-02-14 02:43:59 <elly> rather than the upstream one
2023-02-14 02:43:59 <elly> sigh
2023-02-14 02:44:02 <psykose> that's just a webui issue :p
2023-02-14 02:44:10 <psykose> if your fork is internal/private it defaults to self
2023-02-14 02:44:17 <psykose> if you make it public, it defaults to upstream
2023-02-14 02:44:34 <psykose> also for "no target", i guess it depends how you manage your branches
2023-02-14 02:44:36 <elly> aha, that's the problem
2023-02-14 02:44:39 <elly> I made it private
2023-02-14 02:44:41 <psykose> people do all sorts of fancy git stuff
2023-02-14 02:44:53 <psykose> i do 1 branch per mr or local wip stuff
2023-02-14 02:45:00 <psykose> which is very easy to keep track of
2023-02-14 02:45:09 <psykose> need to change mr -> checkout, amend, push -f
2023-02-14 02:45:11 <psykose> onto the next
2023-02-14 02:45:29 <psykose> this is admittedly the most obvious use of git, so nothing new there
2023-02-14 02:45:29 <elly> yeah
2023-02-14 02:45:34 <elly> that's what I do normally as well
2023-02-14 02:45:39 <psykose> some people to branchless or something, i guess it really depends
2023-02-14 02:45:45 <ptrc> fwiw, in modern git if you have multiple remotes and do `git push <remote>` on a fresh branch, it associates the branch with the remote
2023-02-14 02:45:53 <ptrc> really handy when working with forks
2023-02-14 02:46:04 <bl4ckb0ne> you need to enable it in the config
2023-02-14 02:46:14 <ptrc> oh? it worked for me by default without a config
2023-02-14 02:46:52 <ptrc> oooor nevermind, you're right
2023-02-14 02:46:54 <psykose> yeah, and you can also do stuff like git push remote HEAD:branch, to just push detached heads around without anything local
2023-02-14 02:47:05 <psykose> there's 50 million ways to update everything
2023-02-14 02:47:07 <ptrc> `git config --global push.autoSetupRemote true`
2023-02-14 02:47:36 <psykose> now my config is 144 lines
2023-02-14 02:47:37 <psykose> ty ptrc
2023-02-14 02:47:38 <ptrc> psykose: pushing detached heads around sounds extremely cursed both in the literal meaning and in the git one
2023-02-14 02:48:42 <psykose> i end up in it sometimes because people never name their branches anything meaningful
2023-02-14 02:48:46 <psykose> they just name it 'package'
2023-02-14 02:48:53 <psykose> 3 weeks later they name the new update 'package'
2023-02-14 02:49:08 <psykose> but i want to just amend something quickly myself.. but i had the branch already
2023-02-14 02:49:20 <psykose> so i can't just check out the branch as the same name with auto implicit same-name remote branch
2023-02-14 02:49:25 <psykose> so it gets detached
2023-02-14 02:49:30 <psykose> and you know exactly who i am talking about
2023-02-14 02:49:43 <ptrc> o-okay mom i'll name my branches pkgname-pkgver
2023-02-14 02:50:23 <ptrc> ( but also, running just `git rebase` after fetching usually updates your branch to what remote has )
2023-02-14 03:04:23 <bl4ckb0ne> i just use pkgname and rebase the branch to reuse it
2023-02-14 03:04:43 <psykose> don't like that
2023-02-14 03:04:58 <psykose> prefer no stale branches of merged stuff
2023-02-14 03:05:07 <psykose> no real change -> no branch
2023-02-14 03:05:34 <psykose> detecting ""merged"" is quite hard however so even with "cleaning scripts" that fell off a truck you'll always keep a few
2023-02-14 03:07:42 <bl4ckb0ne> you're not my mum :p
2023-02-14 03:12:30 <psykose> so many bad responses to that and so few good ones
2023-02-14 03:28:53 <umk3[m]> I'm kind of glad uinput is disabled by default in Alpine, it's been bugging me for months when I enabled it for Steam
2023-02-14 03:31:15 <umk3[m]> I learned by force that games should read from sdl2 or whatever is directly talking
2023-02-14 03:46:08 <psykose> i don't think anyone even thought about it, more like "the kernel disables it by default"
2023-02-14 03:46:23 <psykose> god i love portability macros
2023-02-14 03:46:58 <psykose> #if defined(__APPLE__ || MACOS || ANY_BSD|| WINDOWS || MINIX || PLAYSTATION0.2 || THEUNIVERSE
2023-02-14 03:47:05 <psykose> portable_thing()
2023-02-14 03:47:11 <psykose> #else glibc_thing()
2023-02-14 03:47:30 <elly> dislike
2023-02-14 03:48:46 <psykose> was gonna just report that to dolphin devs for strerror_r
2023-02-14 03:48:52 <psykose> but they don't even have an issue tracker apparently
2023-02-14 03:49:31 <psykose> are you kidding
2023-02-14 03:49:33 <psykose> they literally wrote
2023-02-14 03:49:36 <psykose> the correct version in 2017
2023-02-14 03:49:43 <psykose> https://github.com/dolphin-emu/dolphin/pull/6255/files
2023-02-14 03:49:58 <elly> minix portability macros take me back to openssl
2023-02-14 03:50:04 <elly> __DJGPP__ and such
2023-02-14 03:50:10 <psykose> ah
2023-02-14 03:50:12 <psykose> different path
2023-02-14 03:50:28 <psykose> https://github.com/dolphin-emu/dolphin/blob/master/Source/Core/Common/Network.cpp#L556
2023-02-14 03:50:55 <psykose> they have the correct one in one place and the wrong one in another
2023-02-14 03:56:46 <sam_> whats up
2023-02-14 03:56:50 <sam_> oh wow
2023-02-14 03:56:52 <sam_> https://github.com/dolphin-emu/dolphin/commit/cdd3e636e1f4b06ca3e2f3c279c420900b4db0b4 love macro hell
2023-02-14 04:01:49 <AmyMalik> good fuckness.
2023-02-14 04:05:17 <orbea> psykose: they have an issue tracker
2023-02-14 04:05:36 <orbea> https://bugs.dolphin-emu.org/issues
2023-02-14 04:07:58 <psykose> there we go
2023-02-14 04:08:02 <psykose> did not find it in two minutes
2023-02-14 04:08:14 <psykose> but now i am lazy sam_ you do it for me ty ty ^
2023-02-14 04:08:50 <sam_> oops goodbye
2023-02-14 04:11:34 <psykose> :p
2023-02-14 04:13:26 <elly> what does this error message indicate?
2023-02-14 04:13:27 <elly> >>> ERROR: qoi: rootbld: /home/elly/p/aports/testing/qoi/testing/.rootbld-repositories does not exist
2023-02-14 04:14:07 <psykose> ah the classic
2023-02-14 04:14:26 <psykose> uhh the default way to get repo root path is wrong so rootbld fails to find stuff, sometimes
2023-02-14 04:14:40 <psykose> it only works for like
2023-02-14 04:14:41 <psykose> ~/aports
2023-02-14 04:14:42 <elly> I have builddir set explicitly though
2023-02-14 04:14:46 <elly> wait, really?
2023-02-14 04:14:55 <psykose> so you need to do APORTSDIR="$HOME/p/aports"
2023-02-14 04:14:58 <psykose> yeah, and i forgot the other
2023-02-14 04:15:06 <elly> ew, okay
2023-02-14 04:15:16 <psykose> https://gitlab.alpinelinux.org/alpine/abuild/-/issues/10041
2023-02-14 04:15:18 <psykose> https://gitlab.alpinelinux.org/alpine/abuild/-/issues/10069
2023-02-14 04:15:31 <psykose> it checks your remotes or something too
2023-02-14 04:16:05 <psykose> so if it's not named origin and stuff also doesn't work
2023-02-14 04:17:15 <psykose> as a not-stuff-in-home haver it's also inconvenient for me :p
2023-02-14 04:19:30 <elly> yeah
2023-02-14 04:19:32 <elly> oh well, thank you
2023-02-14 10:54:39 <Piraty> opening a MR on gitlab for wider discussion that is in itself not ready to me merged should be prefixed 'Draft: ' ?
2023-02-14 10:54:47 <Piraty> or WIP or sth?
2023-02-14 10:55:56 <Piraty> well here we go
2023-02-14 11:05:58 <EvTheFuture> I'm trying to find out why the package ttf-font-awesome is no longer available after 3.16. It seems like the font itself is avaiable in the apk baculum-common as a theme component for baculum... Does anyone know the story behind this?
2023-02-14 11:10:17 <EvTheFuture> Piraty: My first thought is to use Draft: but I'm not sure
2023-02-14 11:11:08 <Piraty> that's at least consistent with the others
2023-02-14 11:15:45 <ptrc> iirc gitlab handles "Draft: " or "WIP: " to mark the MR as not ready to merge yet
2023-02-14 11:17:16 <Piraty> not sure what's to handle by gitlab other than being a hint for maintainers not to merge yet
2023-02-14 11:17:29 <ptrc> most use "Draft: ", as it's the thing gitlab automatically appends when you click "Mark as draft" in the UI
2023-02-14 11:17:58 <ptrc> Piraty: it makes the MR unmergeable by normal means
2023-02-14 11:18:23 <ptrc> so someone cannot just slip their finger and accidentally merge a thing that's still work in progress
2023-02-14 11:18:27 <Piraty> yolo-merge-protection
2023-02-14 11:19:10 <Piraty> 
ACTION prefers mob branch

2023-02-14 15:35:44 <durrendal> elly: thanks for the MR on Fennel. Would you mind also enabling lua 5.1 support? 
2023-02-14 15:37:34 <elly> durrendal: hm, I guess I could - to be honest I would like to drop 5.2 and 5.3 :P
2023-02-14 15:37:41 <elly> (but also I don't own the package)
2023-02-14 15:38:59 <durrendal> yeah I get that, but it's pretty trivial to support all of the recent lua releases now that the bug that qas preventing 5.1 from working was patched 
2023-02-14 15:39:16 <elly> ya
2023-02-14 15:39:23 <elly> well, I am happy to put up such an MR at any rate
2023-02-14 15:39:40 <elly> not in the same change as the 1.3.0 upgrade though
2023-02-14 15:40:06 <durrendal> yeah that's fair, it should be a seperate enable lua 5.1 support MR
2023-02-14 15:40:22 <elly> nobody at all uses 5.2 and 5.3, right?
2023-02-14 15:40:39 <elly> I legit do not know
2023-02-14 15:41:09 <durrendal> I use 5.3 personally. I should migrate to 5.4 but havent had a chance to make that change 
2023-02-14 15:42:03 <durrendal> there's a difference in the way 5.3 ans 5.4 handles with-open, at least there was with fennel 1.2, that may have changed with the 1.3 release
2023-02-14 15:42:30 <elly> ah, okay
2023-02-14 15:42:32 <elly> hmb
2023-02-14 15:42:45 <elly> well, I'll do an MR to support 5.1, but we do already support luajit on x86 as well
2023-02-14 15:46:34 <durrendal> I appreciate it! I think theres a couple of arches where jit doesnt work, like s390x if Im remembering correctly
2023-02-14 15:46:48 <elly> yup
2023-02-14 15:47:03 <elly> willingly writing a jit compiler targeting s390x is a niche hobby for sure
2023-02-14 16:01:46 <skarnet> anything computer-related that is actually fun is a niche hobby :(
2023-02-14 16:05:17 <bl4ckb0ne> i heard init service system are fun 
2023-02-14 16:08:06 <elly> skarnet: that is kind of a depressing perspective, I disagree with that :P
2023-02-14 16:09:20 <skarnet> bl4ckb0ne: and it is a very niche hobby :(
2023-02-14 16:10:26 <skarnet> elly: I value positivity and hope, but when they're at odds, I value reality even more :(
2023-02-14 16:11:12 <elly> which things are fun or not is subjective though, so I don't know if you can describe that as "reality"
2023-02-14 16:13:19 <skarnet> it is - to a certain extent - objective to say what is a niche hobby and what isn't: for instance, the things that get eeeendlessly talked about on various tech media platforms *cough* orange site *cough* deserve to be qualified as mainstream, not niche
2023-02-14 16:13:36 <skarnet> and I've never found one of these things to be *fun*
2023-02-14 16:14:51 <elly> I see
2023-02-14 16:15:17 <elly> fun is the subjective part though :P I write a lot of C++ for work and I almost always enjoy doing so, even though it's not really very niche
2023-02-14 16:15:56 <skarnet> yeah that's *definitely* subjective :D
2023-02-14 16:16:27 <skarnet> (or maybe it's not, and you're just *wrong* for enjoying C++. It's one or the other.)
2023-02-14 16:19:29 <elly> harsh
2023-02-14 16:19:55 <elly> the thing with writing C++ is, you get all these nice little breaks in your workday while it compiles :P
2023-02-14 16:20:22 <skarnet> obligatory https://xkcd.com/303/
2023-02-14 16:20:23 <alpine-meetbot> [xkcd] Compiling | Alt-text: 'Are you stealing those LCDs?' 'Yeah, but I'm doing it while my code compiles.'
2023-02-14 16:20:27 <elly> precisely
2023-02-14 16:21:34 <orbea> ccache/distcc, you dont need to wait :P
2023-02-14 16:22:09 <orbea> skarnet: c++ is far from the worst offender sadly
2023-02-14 16:22:38 <skarnet> orbea: races to the bottom are never interesting, because it's always possible to do worse
2023-02-14 16:22:52 <orbea> yep...
2023-02-14 16:23:29 <orbea> the suffering developers never realize they are forcing upon maintainers...
2023-02-14 16:23:40 <elly> orbea: we have an enormous build farm, it doesn't help with the link steps :(
2023-02-14 16:24:14 <skarnet> elly: yall still linking all your executables with the world?
2023-02-14 16:24:17 <elly> yes
2023-02-14 16:24:21 <skarnet> sigh
2023-02-14 16:24:27 <elly> I wish we weren't!
2023-02-14 16:24:32 <elly> it'd be very cool
2023-02-14 16:24:34 <skarnet> yeah
2023-02-14 16:25:51 <elly> the final link step for me takes... 48 seconds
2023-02-14 16:25:52 <orbea> doesn't lld support using multiple cores or am I mistaken?
2023-02-14 16:26:12 <elly> I believe it does
2023-02-14 16:26:16 <elly> the cores aren't the problem, this is:
2023-02-14 16:26:35 <elly> ls -alh out/rel/chrome | cut -d' ' -f5-
2023-02-14 16:26:35 <elly> 3.5G Feb 14 16:25 out/rel/chrome
2023-02-14 16:26:42 <elly> it's so... big
2023-02-14 16:26:53 <orbea> ah yea, chrome....
2023-02-14 16:26:59 <elly> and it made that by linking together like dozens of gigabytes of Stuff
2023-02-14 16:28:00 <ikke> I wish some people would be forced to use dev pc's with 4G ram and 20G disks
2023-02-14 16:28:20 <skarnet> s/people/companies/
2023-02-14 16:28:37 <skarnet> individuals are rarely the problem, processes are
2023-02-14 16:29:19 <elly> ikke: we used to have a bunch of wifi networks, back when I was in a real office, that emulated various types of network
2023-02-14 16:29:38 <elly> like there was a special wifi network you could connect to that simulated the performance of 3g, or 2g, or whatever
2023-02-14 16:30:08 <elly> which was pretty neat! but did not really turn into better performance on low end connections
2023-02-14 16:30:26 <skarnet> did you have a wifi network simulating the performance of ethernet? 🤡
2023-02-14 16:30:41 <elly> we just had ethernet for that :P
2023-02-14 16:30:45 <skarnet> fair
2023-02-14 16:30:52 <ikke> elly: but I assume that was opt-in?
2023-02-14 16:31:19 <elly> yes
2023-02-14 16:31:33 <elly> but, the incentives for engineers *never* lined up to actually do anything with it
2023-02-14 16:31:40 <elly> still don't, really
2023-02-14 16:31:57 <elly> same deal for compile/link time or binary size, there's no particular incentives to reduce them or care about not increasing them
2023-02-14 16:32:26 <orbea> and we end up with monsters like chrome
2023-02-14 16:32:29 <elly> yes
2023-02-14 16:32:35 <elly> or, indeed, much of the modern software ecosystem
2023-02-14 16:32:48 <elly> adding more code has an obvious benefit to the person doing it, deleting code rarely does
2023-02-14 16:33:46 <elly> that's why "software expands to fill all available space", as they say
2023-02-14 16:33:51 <ikke> yup
2023-02-14 16:33:59 <orbea> deleting code, making it more concise or improving the logic become important when you maintain something for a long time
2023-02-14 16:34:39 <elly> orbea: most industry software engineers do not stay on one project long enough to benefit from doing that, or to suffer the consequences of not doing it
2023-02-14 16:34:55 <skarnet> hence why companies, not people, are the problem
2023-02-14 16:35:21 <orbea> i can agree with that
2023-02-14 16:35:44 <skarnet> when the code isn't owned by anyone in particular, nobody's invested enough to put in the effort of reducing debt and generally improving not-immediate metrics
2023-02-14 16:36:19 <elly> imo it is fundamentally a failure of engineering managers
2023-02-14 16:36:53 <elly> it is our (collective) job to say things like "We need to slow feature work so we can rework this design." and stuff, but we (again collectively) often don't do that
2023-02-14 16:37:36 <skarnet> at this level of complexity that has to be a super tough decision to make, because that requires serious resource commitment
2023-02-14 16:38:09 <elly> yes
2023-02-14 16:38:23 <elly> there is a good reason why people don't do this, and it is not that they are stupid or lazy :P
2023-02-14 16:39:06 <skarnet> it's rarely that... most of the time it's an incentives issue
2023-02-14 16:39:36 <elly> yeah
2023-02-14 16:39:44 <elly> we're way off the alpine topic really but I have a lot of thoughts about this
2023-02-14 16:39:47 <elly> -> offtopic
2023-02-14 19:07:14 <dsankouski[m]> Hi guys! Would you assept [extract-dtb package](https://gitlab.com/postmarketOS/pmaports/-/blob/master/main/extract-dtb/APKBUILD)?
2023-02-14 19:07:49 <ikke> probably
2023-02-14 21:17:19 <bl4ckb0ne> psykose did you broke clang again? build with sanitizer fails
2023-02-14 21:17:47 <bl4ckb0ne> missing -lasan and libasan_preinit.o
2023-02-15 04:47:02 <psykose> that tells me nothing except you held it wrong
2023-02-15 04:48:23 <psykose> and forgot to install compiler-rt or something
2023-02-15 04:58:29 <psykose> EvTheFuture: it's just font-awesome (or font-fontawesome-4 for the older version..)
2023-02-15 04:58:33 <psykose> every font is font- now
2023-02-15 06:59:04 <EvTheFuture> @psykos: yes I got the name change, bit the content of the packages is not the same anymore. I searched for the file fa-brands-400.ttf which is available in /use/share/fonts/TTF in 3.16 but I don't find any package in 3.17 where that file is available under /use/share/fonts/*
2023-02-15 06:59:48 <psykose> you probably want Font Awesome 6 Brands-Regular-400.otf from font-awesome-brands
2023-02-15 07:11:59 <psykose> yeah
2023-02-15 07:11:59 <psykose> https://img.ayaya.dev/4lzRBZGB4qIc
2023-02-15 10:48:43 <EvTheFuture> @psykose: Yes I solved my problem. I just was a bit confused about what had happend t the TTF version.
2023-02-15 10:52:20 <EvTheFuture> Does anyone know if there are any known bluetooth problems with Alpine 3.17 and Pipewire? After fresh install I can not connect any of my bluetooth headsets and I find some issues that apperantly has been solved in pipewire 0.3.65 that existed in 0.3.60, however I can not confirm if that is the same problem I have, so I'm just reaching out just in case anyone can confirm it
2023-02-15 10:52:22 <EvTheFuture> works just fine for you or if anyone know of any known problems?
2023-02-15 10:53:24 <psykose> worked for me on every pipewire version for the past 50
2023-02-15 10:53:38 <psykose> so it probably was indeed specific to your device or controller
2023-02-15 10:59:34 <EvTheFuture> @psykose: I guess I have missed something then because It's the same hardware now as before on Alpine 3.16. I just switched SSD and made a fresh install of Alpine 3.17.
2023-02-15 11:00:24 <psykose> probably indeed some regression somewhere then
2023-02-15 11:01:58 <EvTheFuture> Most likely something really easy an stupid (as usual) :)
2023-02-15 11:06:08 <EvTheFuture> @psykose: are you using wireplumber in you rsetup?
2023-02-15 11:06:15 <psykose> ye
2023-02-15 11:06:19 <EvTheFuture> k
2023-02-15 11:21:56 <EvTheFuture> I constantly now get "Failed to connect: org.bluez.Error.Failed br-connection-profile-unavailable" when trying to connect to any of my headphones... I evern reverted back to stock configuration files for Pipewire and Wireplumber
2023-02-15 13:06:05 <EvTheFuture> @psykose: Have you disabled D-bus support?
2023-02-15 13:06:25 <psykose> yeah i removed dbus from all of alpine in everything
2023-02-15 13:06:35 <psykose> (no, and idk in what you're referring to)
2023-02-15 13:06:42 <psykose> pipewire didn't change
2023-02-15 13:08:24 <bl4ckb0ne> psykose: conpiler-rt is there
2023-02-15 13:08:30 <bl4ckb0ne> compiler-rt*
2023-02-15 13:08:35 <psykose> post more info then
2023-02-15 13:08:41 <psykose> like everything you ran and the whole console
2023-02-15 13:08:46 <psykose> can't read your mind
2023-02-15 13:09:19 <EvTheFuture> @psykose: Yeah I understand, just didn't want to spam this channel with my logs and stuff :)
2023-02-15 13:09:38 <psykose> you can pastebin all the logs you want
2023-02-15 13:09:39 <psykose> no issue there
2023-02-15 13:10:04 <psykose> that specific error just looks like wireplumber being broken or not running at all in my experience
2023-02-15 13:12:47 <EvTheFuture> @psykose: Yeah, but it is running, and I've now reverted all my config to stock. I'll collect info and post a pastbin
2023-02-15 13:44:03 <maribu> > The wxPython library was compiled against wxWidgets 3.2.2.1 but KiCad is using 3.2.2.  Python plugins will not be available.
2023-02-15 13:44:22 <maribu> I start to wonder if we should just patch that check out :/
2023-02-15 13:45:56 <psykose> yeah that is just wrong
2023-02-15 13:46:04 <psykose> and i did even rebuild it
2023-02-15 13:47:31 <psykose> hmm
2023-02-15 13:47:40 <psykose> something is funny here
2023-02-15 13:47:43 <psykose> smells like.. python..
2023-02-15 13:48:15 <psykose> https://img.ayaya.dev/Q52M9RPwhCHa
2023-02-15 13:48:17 <psykose> the classic
2023-02-15 13:48:21 <psykose> hardcoded dep scanner
2023-02-15 13:48:22 <psykose> nice
2023-02-15 13:49:28 <psykose> https://gitlab.com/kicad/code/kicad/-/issues/13887
2023-02-15 13:53:06 <maribu> I honestly wonder if wxWidgets is doing something terrible wrong in their APIs, or if that check is just annoying nonsese. My expectation is that an update from 3.2.2 to 3.2.2.1 or even to 3.2.9 doesn't change the API and ABI in a way that would matter, only implementation internal stuff gets changed. And hence, KiCad shouldn't worry about wxPython being build for the different version.
2023-02-15 13:53:39 <psykose> would have to look at why the check was added and the reasoning
2023-02-15 13:54:00 <psykose> this specific thing is not that check though- it's a bug in the check itself, as the versions do match :D
2023-02-15 13:54:37 <psykose> rebuilding it to check, probably fixed in a moment
2023-02-15 14:04:46 <psykose> so yeah, fixed on rebuild
2023-02-15 14:04:52 <psykose> sorry for the bother marian :)
2023-02-15 14:11:10 <maribu> No problem. Thx for the fast fix :)
2023-02-15 14:15:06 <EvTheFuture> I've collected info about my bluetooth problem: https://tpaste.us/KEdB
2023-02-15 14:19:09 <EvTheFuture> Here is a list of the installed aports (if I have missed something obvious): https://tpaste.us/NB7y
2023-02-15 14:24:21 <psykose> hmm
2023-02-15 14:24:22 <psykose> yeah weird
2023-02-15 14:24:34 <psykose> nothing jumps out at me
2023-02-15 14:25:02 <psykose> i've definitely seen the same error, it was just usually not having wireplumber at all, oor one other thing i am forgetting
2023-02-15 14:29:22 <EvTheFuture> I'm thinking of backporting 0.3.65 from edge to 3.17 just to see if I somehow hit this problem: https://gitlab.freedesktop.org/pipewire/pipewire/-/issues/2829
2023-02-15 14:29:45 <EvTheFuture> I had the same behaviour at first, but after unpairing and re-pairing the headphones, I got this problem instead...
2023-02-15 14:31:12 <psykose> if 0.3.65 works you can bisect the versions
2023-02-15 14:33:30 <EvTheFuture> I'm still convinced though that it's just some stupid mistake from my side which made me lose half a work day for this :)
2023-02-15 15:37:41 <EvTheFuture> Hm... Pipewire dependencies does not go through on 3.17... I backported the missing liblc3 but after going through installing the dependencied when running abuild it complain of missing roc dependency https://tpaste.us/ogQL
2023-02-15 15:37:59 <EvTheFuture> I thought it was the roc-toolkit but that is included in the build depends
2023-02-15 15:38:22 <psykose> version
2023-02-15 15:38:43 <psykose> just disable it
2023-02-15 15:39:16 <EvTheFuture> Yes, I see now quite a big jump in versoin. Thanks
2023-02-15 15:39:22 <psykose> -Droc=disabled
2023-02-15 15:40:29 <EvTheFuture> Thanks, that worked.
2023-02-15 18:00:45 <EvTheFuture> Didn't help to upgrade pipewire, hae not been able to build wireplumber yet, but that is the next move. If that doesn't work I plan on intalling 3.16 to make sure everything works on a fresh install of 3.16 (everything worked on 3.16 before replaving the SSD with a fresh install
2023-02-15 18:05:03 <psykose> very unfortunate :/
2023-02-15 19:14:53 <nmeum> psykose: about to do go rebuilds again: why does grafana depend on the pkgrel of grafana-frontend? can we not just use ~= instead of = for the version constraint, thus only enforcing the same $pkgver but not the same $pkgrel?
2023-02-15 19:15:07 <psykose> not sure, it was just wrotten that way
2023-02-15 19:15:19 <psykose> ~= would work i think
2023-02-15 19:15:50 <ikke> check the grafana-frontend APKBUILD
2023-02-15 19:15:52 <ikke> it has more details
2023-02-15 19:16:32 <ikke> But I guess that's mostly about pkgver
2023-02-15 19:17:35 <ikke> @kaey maintains it
2023-02-15 19:17:49 <ikke> kaey[m]: ^
2023-02-15 19:17:53 <nmeum> I will just send an MR with a version constraint change, then they can review it when they have the time :)
2023-02-15 19:18:25 <psykose> it is correct, the pkgrel doesn't matter
2023-02-15 19:18:43 <psykose> =$pkgver just doesn't work :p and i guess ~= is more niche
2023-02-15 19:18:54 <psykose> or rather just ~
2023-02-15 19:18:56 <psykose> same thing
2023-02-15 19:19:15 <ikke> apk compares the complete version including pkgrel
2023-02-15 19:19:21 <ikke> so =$pkgver indeed does not work
2023-02-15 19:21:16 <nmeum> !44263
2023-02-15 19:26:36 <psykose> aha, so it does
2023-02-15 19:26:58 <psykose> what is the difference between =$pkgver ~$pkgver then
2023-02-15 19:27:04 <psykose> i guess the latter allows .1?
2023-02-15 19:27:08 <psykose> the former is equal
2023-02-15 19:27:14 <ikke> ~ is prefix match
2023-02-15 19:27:22 <ikke> (roughly)
2023-02-15 19:27:22 <psykose> ah exactly like shell?
2023-02-15 19:27:25 <psykose> yea
2023-02-15 19:27:26 <psykose> okay
2023-02-15 19:27:33 <psykose> =$pkgver is the correct form then
2023-02-15 19:27:57 <psykose> nmeum: :D
2023-02-15 19:28:54 <nmeum> wait. I thought = always matches the whole string includnig the pkgrel?
2023-02-15 19:29:07 <ikke> =$pkgver does not work
2023-02-15 19:29:10 <ikke> nmeum: yes
2023-02-15 19:29:42 <ikke> ~ allows the version components to vary that you do not specify
2023-02-15 19:29:59 <nmeum> exactly
2023-02-15 19:30:32 <nmeum> I assume it was specifically designed for this use case (matching $pkgver without $pkgrel)
2023-02-15 19:30:47 <ikke> probably one of the reasons
2023-02-15 19:30:58 <ikke> You can also use it constrain major versions
2023-02-15 19:31:09 <ikke> apk add pkg~1 allows 1.x, but not 2.x
2023-02-15 19:32:07 <ikke> nfpm-2.26.0-r0:
2023-02-15 19:32:10 <ikke>     breaks: world[nfpm=2.26.0]
2023-02-15 19:35:55 <nmeum> yep
2023-02-15 19:36:36 <psykose> i totally didn't just read "does not work" as does work
2023-02-15 19:36:40 <psykose> not at all
2023-02-15 19:57:30 <nmeum> do I need to wait with pushing go rebuilds until all builders have build go 1.20.1? the riscv64 builder is currently still building webkit…
2023-02-15 19:57:43 <psykose> no
2023-02-15 19:57:50 <psykose> the order is correct regardless and go comes first
2023-02-15 19:57:57 <nmeum> even if the go build itself fails?
2023-02-15 19:57:59 <psykose> if the order is not correct, that means there's a dependency loop
2023-02-15 19:58:00 <psykose> yes
2023-02-15 19:58:03 <psykose> it just retries into go each time
2023-02-15 19:58:11 <nmeum> ok! sweet
2023-02-15 19:58:23 <psykose> tbf, the dep loop does happen sometimes
2023-02-15 19:58:39 <psykose> lemme check
2023-02-15 19:58:46 <psykose> clear, no circles
2023-02-15 19:59:30 <psykose> for reference https://gitlab.alpinelinux.org/ptrcnull/apkcircledep/
2023-02-15 19:59:39 <psykose> yes, it does work, and it saved me hours on 3.17 :D
2023-02-15 20:00:08 <psykose> not much of an apkcircledep and more of a aportsapcircledep but
2023-02-15 20:41:50 <EvTheFuture> Hmmm... I'm trying to build wireplumber from edge for Alpine 3.17 with backported pipewire (0.3.65), roc-toolkit (0.2.1) and liblc3 (1.0.1)
2023-02-15 20:43:58 <EvTheFuture> It compiles just fine, but on test number 3 out of 27 it gets a SIGSEGV https://tpaste.us/j4DV
2023-02-15 20:44:31 <EvTheFuture> Any ideas? I checked the buildlog for wireplumber currently in edge and it's built against pipewire 0.3.62
2023-02-15 20:44:41 <psykose> just skip the tests
2023-02-15 20:45:34 <EvTheFuture> I thought so first, but then I started windering if the tests get SIGSEGV will ireplumber even work at all. Well I'll try
2023-02-15 20:45:58 <EvTheFuture> wondering (gosh, I'm getting tired and can't type) :)
2023-02-15 20:52:39 <EvTheFuture> seems to run, but same result.
2023-02-15 20:53:08 <EvTheFuture> @psykose: Do you use stock config files for pipewire, wireplumber and bluez ?
2023-02-15 20:53:15 <psykose> ye
2023-02-15 20:53:21 <EvTheFuture> Weird
2023-02-15 20:54:13 <EvTheFuture> It feels like some module or something is missing somewhere
2023-02-15 20:55:38 <psykose> pipewire-spa-bluez?
2023-02-15 20:55:53 <psykose> you have it
2023-02-15 20:58:19 <EvTheFuture> pipewire-spa-bluez-0.3.65-r2 x86_64 {pipewire} (LGPL-2.1-or-later) [installed]
2023-02-15 20:58:21 <EvTheFuture> yes
2023-02-15 20:59:19 <EvTheFuture> Maybe it's how I start everything... I start sway when I log in as a user: dbus-launch --exit-with-session seatd-launch sway
2023-02-15 20:59:45 <EvTheFuture> Then sway start pipewire which in turn stat pipewire -c pipewire-pulse.conf and wireplumber
2023-02-15 20:59:57 <elly> my gosh I dislike depot_tools / gclient / etc
2023-02-15 21:00:02 <elly> this whole stack just doesn't work on alpine at all
2023-02-15 21:00:15 <elly> depot_tools has a vendored copy of python3.8 that doesn't run
2023-02-15 21:01:36 <ptrc> there's a "rewrite-it-in-rust" version of depot_tools in aports: https://pkgs.alpinelinux.org/package/edge/testing/x86_64/teapot-tools
2023-02-15 21:01:45 <elly> intriguing
2023-02-15 21:02:03 <ptrc> also, iirc there was a way to use system python instead of the vendored one
2023-02-15 21:02:08 <EvTheFuture> @psykose: Do you run dbus system wide or like I do?
2023-02-15 21:02:22 <psykose> that's a requirement to run the user session at all
2023-02-15 21:02:37 <psykose> i.e you have the system dbus, then the user session
2023-02-15 21:02:46 <psykose> i do dbus-run-session -- instead but it's the same shit
2023-02-15 21:03:01 <psykose> elly: you want uhh
2023-02-15 21:03:10 <psykose> 	export VPYTHON_BYPASS="manually managed python not supported by chrome operations"
2023-02-15 21:03:11 <psykose> there
2023-02-15 21:03:12 <psykose> enjoy
2023-02-15 21:03:18 <psykose> uses system python
2023-02-15 21:03:28 <elly> hmmmm
2023-02-15 21:03:30 <elly> I can try that
2023-02-15 21:03:36 <psykose> the vpython script is literally
2023-02-15 21:03:37 <ptrc> beat me to it, i was about to post a link to aports with that line :p
2023-02-15 21:03:43 <psykose> if $that; python
2023-02-15 21:03:50 <psykose> else; some/path/downloaded/python
2023-02-15 21:05:59 <elly> I see, how annoying
2023-02-15 21:06:04 <elly> thanks
2023-02-15 21:06:15 <psykose> the teapot tools don't implement triggers/hooks/scripts/whatever iirc so it's useful for one use and nothing else only really
2023-02-15 21:06:54 <EvTheFuture> @psykose: Yeah, this is really getting on my nerves now... I guess the next thing would be to revert back to 3.16 again but Pipewire is a bit to buggy and make carla crash when connecting and disconnecting bt devices or changing profiles to much.
2023-02-15 21:07:04 <psykose> the other packager for electron specifically however is quite good, and practically replaces electron-builder
2023-02-15 21:07:12 <psykose> but electron is electron so it's a waste of effort really :p
2023-02-15 21:08:33 <elly> I don't need triggers or hooks I think, I just don't want to run a random 3p program if I can avoid it
2023-02-15 21:09:10 <psykose> for chromium itself you cannot get a checkout without them pretty sure
2023-02-15 21:09:28 <psykose> i.e. it clones stuff then fixups things/downloads artifacts via scripts
2023-02-15 21:09:39 <psykose> buut maybe i am just wrong
2023-02-15 21:09:56 <psykose> for electron it definitely requires it, and that's where i only use it really
2023-02-15 21:10:03 <elly> I'll see if the vpython bypass works for me
2023-02-15 21:10:06 <elly> I just want to build crashpad
2023-02-15 21:10:09 <psykose> it should
2023-02-15 21:10:18 <psykose> on this topic annoyingly chrome broke their tarballs again
2023-02-15 21:10:49 <psykose> https://ci.chromium.org/p/infra/builders/cron/publish_tarball
2023-02-15 21:10:50 <psykose> annoyge
2023-02-15 21:10:56 <elly> really? ugh
2023-02-15 21:10:58 <psykose> so there is no .100 for us unless i manually make it
2023-02-15 21:10:59 <psykose> ye
2023-02-15 21:11:13 <psykose> i would've thought someone can retry it, but i think it's like last time
2023-02-15 21:11:15 <psykose> fails = they give up
2023-02-15 21:11:28 <elly> it has another build running right now though?
2023-02-15 21:11:43 <elly> oh but that's for 97
2023-02-15 21:11:44 <elly> hm
2023-02-15 21:11:49 <psykose> 97 was the previous
2023-02-15 21:12:18 <psykose> and it's from 1 minute ago, i'm pretty sure it just dies in a loop :D
2023-02-15 21:12:20 <elly> wait, which specific build failure are you looking at?
2023-02-15 21:12:40 <psykose> idk, that page
2023-02-15 21:13:21 <elly> you want a tarball for 110.0.5481.100 specifically right?
2023-02-15 21:13:26 <psykose> yeah pretty much
2023-02-15 21:15:11 <psykose> i mean i know how to make them and there's already a script :D
2023-02-15 21:15:18 <psykose> but do i really wanna spend 30 minutes on it
2023-02-15 21:15:49 <elly> I'll check what's up with them
2023-02-15 21:16:15 <elly> https://groups.google.com/a/chromium.org/g/chromium-packagers/c/ZevDhKpO680 looks like that whole system is having a normal one
2023-02-15 21:17:08 <psykose> that is poggers
2023-02-15 22:31:39 <elly> well, one more patch down at least
2023-02-15 22:31:41 <elly> slow going
2023-02-15 22:32:20 <psykose> better than my zero
2023-02-15 22:32:35 <elly> https://chromium-review.googlesource.com/c/crashpad/crashpad/+/4255370 crashpad now builds with musl
2023-02-15 22:33:45 <psykose> does it even do anything useful however
2023-02-15 22:34:41 <elly> that's debateable
2023-02-15 22:34:55 <elly> I'm not sure if the tests even run
2023-02-15 22:35:24 <psykose> i skip some stuff in chromium base that crashes related to this somewhat
2023-02-15 22:36:22 <psykose> i guess readelfbuildid is close
2023-02-15 22:40:20 <elly> I can't even get a crashpad checkout onto my laptop yet so who knows
2023-02-15 22:41:49 <psykose> aw
2023-02-15 22:41:51 <psykose> how come
2023-02-15 22:41:58 <EvTheFuture> @psykose: Now this is interesting... Since i now use zfs i created a second root and just copied everything over from the previous SSD and just tweaked fstab initramfs etc. After removing the headset and repairing it, everything just works.
2023-02-15 22:42:09 <psykose> lol
2023-02-15 22:42:12 <EvTheFuture> I have to compare what differs
2023-02-15 22:42:21 <psykose> well, readding bluetooth device working suddenly is not unusual really
2023-02-15 22:42:31 <EvTheFuture> And also take a snapshot and then upgrade to 3.17 and see if it stops working
2023-02-15 22:42:46 <elly> even with system python, it tries to invoke cipd during fetch
2023-02-15 22:42:49 <elly> and I don't have a working cipd
2023-02-15 22:56:50 <psykose> they have a 300 line script to wrap that shit
2023-02-15 22:57:04 <psykose> oof
2023-02-15 22:57:16 <elly> the maddening thing is that git submodules already exist
2023-02-15 23:02:17 <ikke> and google also has repo
2023-02-15 23:02:32 <ikke> although, not sure if they use it for anything else but aosp
2023-02-15 23:02:47 <ikke> (are they even still using it?)
2023-02-15 23:05:59 <elly> no idea
2023-02-16 00:25:47 <sam_> oh nice re crashpad
2023-02-16 00:25:58 <sam_> if nothing else, I care about that working because then it's less hassle if something bundles it and it's hard to disable or whatever
2023-02-16 00:26:22 <sam_> thank you folks for working on all the chromium upstreaming
2023-02-16 00:26:59 <psykose> tis but one singular person all alone in the weeds
2023-02-16 00:41:46 <elly> two singular people :P
2023-02-16 00:42:03 <psykose> just you
2023-02-16 00:42:20 <psykose> or is there some partner in crime there
2023-02-16 00:55:26 <elly> I meant you!
2023-02-16 00:55:30 <elly> you do like half the work
2023-02-16 01:00:27 <elly> $ grep -F '[ ]' p/musl | wc -l ; grep -F '[x]' p/musl | wc -l
2023-02-16 01:00:28 <elly> 18 15
2023-02-16 01:00:33 <elly> woah oh, we're a little under halfway there
2023-02-16 01:00:40 <elly> woah oh, compiling chromium on a laptop
2023-02-16 01:01:02 <elly> (Bon Jovi would not be proud)
2023-02-16 04:26:42 <elly> psykose: apparently the pipeline that makes tarballs is having a bit of trouble, there ought to be one for .100 tomorrow early
2023-02-16 04:26:45 <elly> (us time)
2023-02-16 04:26:54 <psykose> oh wow
2023-02-16 04:27:03 <psykose> didn't expect one would get made :)
2023-02-16 04:27:08 <psykose> that's nice, thanks for giving it a look
2023-02-16 04:27:12 <elly> sure thing
2023-02-16 04:30:13 <psykose> ^.^
2023-02-16 04:47:13 <elly> I wish every the qoi author a very answer my question on github
2023-02-16 04:47:51 <psykose> it's whatever, you did your job
2023-02-16 04:47:53 <psykose> :p
2023-02-16 04:48:01 <psykose> you can just type in the rest and it's all good
2023-02-16 06:13:15 <ikke> anyone else noticing vim syntax highlighting acting up on APKBUILD files?
2023-02-16 06:14:51 <elly> yes
2023-02-16 06:15:06 <elly> it gets very confused by lines starting with _, it seems
2023-02-16 06:15:38 <elly> https://elly.town/t/apkbuild.png ends up looking like this
2023-02-16 06:15:57 <elly> actually, no, it's not the _
2023-02-16 06:16:13 <ikke> it also happens on multiline strings
2023-02-16 06:16:25 <elly> not only those though
2023-02-16 06:16:37 <ikke> no
2023-02-16 06:16:58 <elly> something odd is going on... if the line is: _testfonts=336e
2023-02-16 06:17:06 <elly> then the highlighter breaks, but if it's: _testfonts=336
2023-02-16 06:17:08 <elly> then it works fine
2023-02-16 06:18:07 <ikke> even just a function
2023-02-16 06:18:11 <ikke> in my case
2023-02-16 06:18:32 <ikke> first function is highlighted correctly, everything else is almost all the same color
2023-02-16 06:19:24 <ikke> https://i.imgur.com/9xrd8Pw.png
2023-02-16 06:19:32 <elly> I cannot really get it to happen in other sh files though
2023-02-16 06:20:23 <elly> in fact, if I rename APKBUILD to apkbuild.sh, it goes away
2023-02-16 06:20:27 <elly> suspect
2023-02-16 06:21:22 <ikke> yes
2023-02-16 06:21:44 <elly> oh man
2023-02-16 06:22:20 <elly> our /etc/vim/vimrc has a special case for APKBUILD specifically that sets filetype=sh during BufRead
2023-02-16 06:22:26 <ikke> au BufNewFile,BufRead .bashrc,bashrc,bash.bashrc,.bash[_-]profile,.bash[_-]logout,.bash[_-]aliases,bash-fc[-.],*.ebuild,*.bash,*.eclass,PKGBUILD,APKBUILD call dist#ft#SetFileTypeSH("bash")
2023-02-16 06:24:20 <elly> I wonder if ft=sh is confused somehow
2023-02-16 06:24:44 <ikke> I had that commented out
2023-02-16 06:24:49 <ikke> and still does it for me
2023-02-16 06:25:13 <elly> it's just very strange that it doesn't happen if I rename the file
2023-02-16 06:25:32 <ikke> Yeah, that's why I was searching for APKBUILD in /usr/share/vim
2023-02-16 06:27:51 <ikke> commenting out those lines doesn't fix it
2023-02-16 06:28:01 <elly> yeah, indeed
2023-02-16 06:28:14 <elly> it is, unfortunately, time to try to read /usr/share/vim/vim90/syntax/sh.vim
2023-02-16 06:31:30 <elly> aha, I have a lead
2023-02-16 06:31:38 <elly> sh.vim thinks that APKBUILD is using bash syntax
2023-02-16 06:31:47 <elly> rename APKBUILD -> APKBUILD.sh: no problem
2023-02-16 06:31:53 <elly> add #!/bin/sh at the top: no problem
2023-02-16 06:31:57 <elly> add #!/bin/bash at the top: problem
2023-02-16 06:32:19 <ikke> hmm
2023-02-16 06:36:51 <psykose> bash isn't really wrong for apkbuild since there's bash features in use anyway, and i don't think the triggering _ are like special and very different in bash syntax parsing
2023-02-16 06:37:14 <elly> yeah, I don't either, but I think the weird highlighting is a bug in the bash mode of sh.vim specifically
2023-02-16 06:37:20 <psykose> ye
2023-02-16 06:37:36 <psykose> also keep in mind vim doesn't really do releases so aports is just.. basically git master of random commits
2023-02-16 06:37:44 <psykose> i bump it every few days
2023-02-16 06:38:22 <ikke> let's see what 1313 does
2023-02-16 06:38:25 <psykose> i'm mostly a vimmer for a long time so it works out ok in terms of testing, but ofc stuff like this can really be "the commit is wrong"
2023-02-16 06:38:44 <psykose> in this case the diff doesn't have anything HL relevant
2023-02-16 06:38:51 <ikke> ok
2023-02-16 06:39:20 <psykose> the reverse is easier to test- go back 300 commits
2023-02-16 06:39:48 <ikke> 1004?
2023-02-16 06:39:56 <psykose> arbitrary :p
2023-02-16 06:39:58 <ikke> heh
2023-02-16 06:42:13 <ikke> .1000 works
2023-02-16 06:42:21 <ikke> now 1150 I guess
2023-02-16 06:42:37 <psykose> there we go a whole bisect going
2023-02-16 06:42:49 <ikke> ahuh
2023-02-16 06:43:24 <ikke> 1150 fails to build
2023-02-16 06:43:28 <elly> this will probably not surprise you, but sh.vim has no case at all to handle when /bin/sh resolves to busybox
2023-02-16 06:43:37 <ikke> yeah, I noticed
2023-02-16 06:44:31 <elly> good luck and we're all counting on you
2023-02-16 06:45:40 <ikke> 1151 good
2023-02-16 06:47:23 <ikke> 1225 good
2023-02-16 06:47:39 <psykose> this looks similar https://github.com/vim/vim/issues/11977
2023-02-16 06:49:07 <ikke> 1275 bad
2023-02-16 06:50:40 <ikke> 1250 good
2023-02-16 06:52:07 <ikke> 1262 good
2023-02-16 06:53:44 <ikke> 1269 good
2023-02-16 06:54:05 <psykose> yeah it's gonna be https://github.com/vim/vim/commit/be4e01637e71c8d5095c33b9861fd70b41476732 from that issue for sure
2023-02-16 06:54:37 <ikke> which patch is that?
2023-02-16 06:54:43 <psykose> 1275
2023-02-16 06:54:49 <psykose> 1274 should be good
2023-02-16 06:56:30 <psykose> ok, new sh.vim does fix it
2023-02-16 06:56:32 <ikke> yup
2023-02-16 06:56:39 <psykose> so $soon or i can just backport it
2023-02-16 06:56:55 <psykose> if you put https://www.drchip.org/astronaut/vim/syntax/sh.vim.gz in /usr/share/vim/vim90/syntax/sh.vim you will see it fixed
2023-02-16 06:57:04 <psykose> it's one version above 1275+
2023-02-16 06:58:25 <ikke> we can wait I suppose
2023-02-16 06:58:30 <ikke> I can remain in 1274 for now
2023-02-16 07:02:41 <psykose> unfixed bugs are sad bugs
2023-02-16 07:02:43 <psykose> 69e308aa808818f851995fc8c8ef8c4b3e71a781
2023-02-16 09:58:08 <staceee> hey! is it possible to review a last time, and merge eventually 38362?
2023-02-16 10:00:14 <Newbyte> !38362
2023-02-16 10:00:15 <Newbyte> hm ...
2023-02-16 16:37:39 <elly> psykose: I just chatted with the desktop platforms lead and there is some interest on their part in supporting musl upstream :)
2023-02-16 16:37:56 <elly> (for chromium I mean)
2023-02-16 17:30:30 <bl4ckb0ne> what's the packaging rule for stuff that depends on a database
2023-02-16 17:30:58 <bl4ckb0ne> provide a setup for all DBs or let the user do it
2023-02-16 17:33:15 <pj> if it's just the config you can provide all you want
2023-02-16 17:33:49 <pj> for service you might want to select most popular one or probably make subpackage for each
2023-02-16 17:34:31 <ikke> In general, let the user choose
2023-02-16 17:35:27 <ikke> And don't assume the db will run on the same host as the application
2023-02-16 17:36:25 <bl4ckb0ne> so just package the stuff and let the user do the rest
2023-02-16 17:36:38 <bl4ckb0ne> stuff in question is https://photoview.github.io/en/docs/installation-manual/
2023-02-16 17:37:28 <ikke> yes
2023-02-16 17:37:55 <bl4ckb0ne> perfect
2023-02-16 20:33:53 <bl4ckb0ne> psykose: solved my asan issue, seems it was tracy acting like a little bitch again
2023-02-17 04:21:36 <psykose> elly: that is weirdly interesting
2023-02-17 04:21:49 <elly> yeah?
2023-02-17 04:22:58 <psykose> i dunno! definitely didn't expect any such thing
2023-02-17 04:23:08 <elly> ahh
2023-02-17 04:23:19 <elly> yeah, he was surprisingly interested by the idea of cross-libc portability
2023-02-17 04:23:23 <elly> I was not really expecting that
2023-02-17 04:23:46 <elly> and so the idea of explicitly having a CI target for musl is appealing
2023-02-17 04:23:48 <elly> we'll see!
2023-02-17 04:24:36 <sam_> nice
2023-02-17 08:11:19 <clandmeter> i thought chromium interest in musl wasn't new
2023-02-17 13:03:24 <umk3[m]> psykose:  steam-devices package didn't reload the installed udev rules until I rebooted
2023-02-17 13:03:32 <psykose> sounds normal to me
2023-02-17 13:03:55 <psykose> probably udevadm trigger
2023-02-17 13:03:59 <umk3[m]> ok :P good to know
2023-02-17 13:04:16 <umk3[m]> i did udevadm control -R and didn't work, absolutely forgot about udevadm trigger
2023-02-17 13:04:30 <umk3[m]> but no worries
2023-02-17 13:04:52 <umk3[m]> thank you
2023-02-17 13:48:48 <dsankouski[m]> What's the Alpine way to trigger rebuild python packages, when python version updates? 
2023-02-17 13:50:09 <ikke> Bump pkgrel
2023-02-17 13:50:42 <dsankouski[m]> ikke: Is it done manually each time python version upgraded?
2023-02-17 13:50:56 <Piraty> psykose: oO https://gitlab.alpinelinux.org/alpine/aports/-/merge_requests/44204/
2023-02-17 13:51:10 <psykose> dsankouski[m]: yes
2023-02-17 13:52:19 <Piraty> ok
2023-02-17 13:53:06 <psykose> Piraty: what's the oO
2023-02-17 13:53:23 <Piraty> two o's
2023-02-17 13:53:28 <Piraty> sorry for confusion
2023-02-17 14:41:15 <elly> clandmeter: there may be other interest elsewhere in the team, we're pretty big and I don't know everyone / everything going on
2023-02-17 15:49:45 <skarnet> wheeeeeee !44314 whenever the pipeline passes
2023-02-17 15:58:59 <skarnet> 6.5 seconds, thanks psykose
2023-02-17 16:25:36 <elly> what's new in this release?
2023-02-17 16:26:43 <skarnet> elly: https://skarnet.org/lists/skaware/1783.html
2023-02-17 16:27:30 <elly> that's an impressive size gain
2023-02-17 16:28:03 <ikke> heh
2023-02-17 16:28:08 <ikke> was about to comment the same
2023-02-17 16:28:12 <skarnet> yeah. I was very much surprised.
2023-02-17 16:28:37 <elly> I guess multicall binary really should be a standard technique for "many small programs" packages
2023-02-17 16:29:00 <skarnet> It's *really annoying* to maintain
2023-02-17 16:29:04 <elly> how come?
2023-02-17 16:29:45 <skarnet> basically, you forbid yourself to use any kind of global data in all your applets
2023-02-17 16:30:52 <skarnet> also, it doesn't show here, but the way I did it was basically cat *.c > multicall.c, a stupid hack that happens to work
2023-02-17 16:31:08 <skarnet> but that enforces constraints on naming and on #defines
2023-02-17 16:31:14 <elly> right
2023-02-17 16:31:55 <skarnet> the right way would be to have a foo_main() in foo.c for applet foo, and link everything together, but that would be a significantly bigger change to the build system
2023-02-17 16:32:07 <elly> yup
2023-02-17 16:32:29 <ikke> like apk does
2023-02-17 16:32:43 <ikke> though it does not support separate binaries afaik
2023-02-17 16:32:52 <elly> still, if you designed something from the start to be multicalled, it wouldn't be all that bad
2023-02-17 16:32:59 <skarnet> but that's the point
2023-02-17 16:33:01 <elly> only the global constraint would be irritating I think
2023-02-17 16:33:05 <skarnet> it's *not* a good target to have
2023-02-17 16:33:47 <skarnet> it's just like Solaris developers who forbid themselves to use fork() because Solaris fork() is slow af
2023-02-17 16:34:15 <skarnet> and because of that we ended up with multithreaded horrors and other monstrosities like nsswitch
2023-02-17 16:35:11 <elly> haha
2023-02-17 16:35:23 <skarnet> if you adapt your programming style to idiosyncrasies of a system (here: "muh small disk space"), you will adopt harmful habits
2023-02-17 16:35:41 <elly> I am familiar with this
2023-02-17 16:35:58 <elly> I was originally taught to write C without using dynamic allocation and ended up with quite a few bad habits as a result
2023-02-17 16:36:13 <skarnet> exactly
2023-02-17 16:36:46 <psykose> to be fair
2023-02-17 16:36:53 <psykose> a lot of the "no dynamic" habits are good
2023-02-17 16:37:22 <skarnet> people who crititize busybox should understand vda's priorities and the toll it takes to maintain such a massively multicall thing
2023-02-17 16:38:02 <elly> psykose: some of them definitely are, but some of them ("panic if the user tries to read more than 1024 lines" or similar) are not :P
2023-02-17 16:38:09 <psykose> :)
2023-02-17 16:38:11 <skarnet> yes, and a lot of the "no globals" habits are good as well, it's just... what elly said
2023-02-17 16:38:21 <ikke> who needs more then 1024 lines? :P
2023-02-17 16:38:30 <skarnet> or 640 kB
2023-02-17 16:38:57 <skarnet> that said, I love to play the "look ma, no heap!" game
2023-02-17 16:39:01 <elly> hehe
2023-02-17 16:39:20 <skarnet> but that's mostly golfing for my own enjoyment
2023-02-17 16:39:23 <elly> I remember ages ago I was reading the source for rio (the plan 9 wm) and came across, like, Window *windows[100];
2023-02-17 16:39:30 <skarnet> brb
2023-02-17 16:39:37 <elly> and it was like "wait, what happens if you open 101 windows?" "well, the WM crashes, so don't do that"
2023-02-17 17:19:20 <skarnet> ha. kwality programming.
2023-02-17 17:41:36 <donoban> https://winaero.com/high-cpu-use-by-taskhost-exe-in-windows-8-1-when-user-name-contains-user/
2023-02-17 17:41:45 <donoban> "To resolve the issue, do not create a user account contains the string "user" on the computer."
2023-02-17 17:43:10 <psykose> god i hate windows articles
2023-02-17 17:43:24 <psykose> when you read this on linux you get a 40 page blog post on why it happens
2023-02-17 17:43:34 <psykose> and here you just get the title repeated 9 times and the article ends
2023-02-17 17:44:06 <invoked> or an embedded youtube video
2023-02-17 17:44:26 <psykose> hey buddy, i think you got the wrong door, the offtopic club is two blocks down
2023-02-17 17:45:02 <donoban> I remeber the official bug entry on microsoft.com, but it seems removed now
2023-02-17 17:45:16 <donoban> it was like a bug entry with the descrption and that solution :)
2023-02-17 17:47:35 <donoban> well, I just remember it due 16:39 <elly> and it was like "wait, what happens if you open 101 windows?" "well, the WM crashes, so don't do that"
2023-02-17 17:47:43 <donoban> sorry for the noise :|
2023-02-17 17:48:21 <psykose> read it more as humor and less aggression :D
2023-02-17 17:48:50 <psykose> love the guy with "disabled updates and my cpu okay"
2023-02-17 17:48:57 <donoban> ohh ok :)
2023-02-17 19:52:47 <psykose> how many damned people can ping the flagged page about "aws cli v2" in one day
2023-02-17 19:52:55 <ikke> heh
2023-02-17 19:53:14 <psykose> it is literally in the repositories
2023-02-17 19:53:37 <psykose> starting to think just disabling this feature would be nice, it is like 90% noise
2023-02-17 19:54:00 <psykose> and 10% actual bug reports, so..
2023-02-17 19:54:17 <ikke> for me it's 95% accurate :P
2023-02-17 19:54:28 <psykose> well
2023-02-17 19:54:36 <psykose> ok sure i kinda glance over the package upgrade ones
2023-02-17 19:54:43 <psykose> (and i am not referring to anitya at all here)
2023-02-17 19:54:54 <psykose> but aside from anitya it's really just
2023-02-17 19:54:55 <ikke> yeah, most of the notifications I get is through anitya
2023-02-17 19:55:04 <psykose> 40% update package (ok)
2023-02-17 19:55:13 <psykose> 55% spam (xxxxx msgh zzzzaaa)
2023-02-17 19:55:21 <psykose> 1% bug reports
2023-02-17 19:55:31 <psykose> 4% "user error request"
2023-02-17 19:56:10 <psykose> the anitya is nice, i've added so much stuff it's almost a large page per day
2023-02-17 19:56:27 <ikke> to be honest, I never look at the flagged page
2023-02-17 19:56:51 <psykose> i use it for anitya because i update stuff
2023-02-17 19:57:03 <psykose> so i get to see the funnee posts daily :D
2023-02-17 19:57:24 <ikke> heh
2023-02-17 19:57:32 <psykose> ok fine, i'll replace aws-cli with -v2
2023-02-17 19:57:37 <psykose> how fast before someone complains v1 is gone?
2023-02-17 19:57:43 <psykose> i give it 2 days?
2023-02-17 19:57:50 <ikke> I would :P
2023-02-17 19:57:54 <psykose> how come
2023-02-17 19:57:55 <ikke> Kidding
2023-02-17 19:57:57 <psykose> grr
2023-02-17 19:58:07 <psykose> ok fine i'll fess up
2023-02-17 19:58:15 <psykose> i only packaged it because i got emailed like 3 fuckin times
2023-02-17 19:58:20 <psykose> i don't use aws and i have no clue if it works
2023-02-17 19:58:21 <ikke> Because only v1 was available, I've been uding v1
2023-02-17 19:58:22 <psykose> plz test
2023-02-17 19:58:32 <ikke> I can probably test it next week
2023-02-17 19:58:36 <psykose> noooo
2023-02-17 19:58:39 <psykose> that is *forever*
2023-02-17 19:58:54 <psykose> also the whole doc/help/completion stuff is kinda broken in both
2023-02-17 19:59:05 <psykose> `aws` kinda doesn't function at all without the help
2023-02-17 19:59:18 <psykose> you either know the exact commands or... no output can't open file
2023-02-17 19:59:23 <ikke> yeah
2023-02-17 19:59:28 <ikke> I had to install aws-cli-doc
2023-02-17 19:59:43 <psykose> i suppose that's ok? for v2 i at least left the main `aws help` page so it's less bad
2023-02-17 21:08:06 <omni_> could .onion addresses be added to the mirrors page? I know of these two: http://dotsrccccbidkzg7oc7oj4ugxrlfbt64qebyunxbrgqhxiwj3nl6vcad.onion/ and http://lysator7eknrfl47rlyxvgeamrv7ucefgrrlhk7rouv3sna25asetwid.onion/
2023-02-17 21:09:50 <ikke> I have no way to verify what's on there
2023-02-17 21:10:35 <ikke> And it would be nice if the people who host these would request them to be added
2023-02-17 21:10:46 <psykose> there has been a few
2023-02-17 21:10:54 <psykose> i.e. as part of the urls they gave
2023-02-17 21:11:13 <ikke> lysator
2023-02-17 21:11:28 <psykose> i think i've seen like three
2023-02-17 21:11:36 <omni> they're linke to from https://mirrors.dotsrc.org/alpine/ and https://ftp.lysator.liu.se/pub/alpine/
2023-02-17 21:12:49 <omni> ikke: how do you mean verify? that you'd need to use tor to fetch from them?
2023-02-17 21:12:59 <ikke> Need to check how the mirror page handles htem
2023-02-17 21:13:02 <ikke> them
2023-02-17 21:17:13 <ikke> It would show as an extra http url
2023-02-17 21:17:16 <ikke> I think
2023-02-17 21:18:02 <omni> is it infra/mirror-status or what is it?
2023-02-17 21:18:47 <ikke> yes
2023-02-17 21:18:54 <ikke> https://gitlab.alpinelinux.org/alpine/infra/mirror-status/-/blob/master/generate-html.lua#L139
2023-02-17 21:21:28 <omni> I'll have a look, an opportunity to practice some Lua =)
2023-02-17 21:22:19 <omni> because it wouldn't add much if it wasn't easy to see that it was an onion link, that you could search the page for "onion" and find a mirror
2023-02-17 21:27:40 <ikke> agreed
2023-02-17 21:28:51 <ikke> One question is if we want them to be filtered out from https://mirrors.alpinelinux.org/mirrors.txt, which is what the installer uses to select mirrors
2023-02-17 21:29:10 <ikke> And mostly because the feature to test the fastest mirror will fail on them
2023-02-17 21:30:35 <elly> I think you would want to have that feature fast-reject them instead of even trying
2023-02-17 23:26:49 <EvTheFuture> @psykose: JFYI, I upgraded the working Alpine 3.16 that I copied from my old SSD into the new SSD and upgraded it to Apline 3.17 and Bluetooth still works. Super weird. Thanks for the support trying to figure out what was wrong with my fresh install of 3.17.
2023-02-17 23:27:54 <psykose> well at least something works
2023-02-18 01:25:33 <minimal> psykose: what is/where is "the flagged page"? I'm aware only of the "Flag out of date" button
2023-02-18 01:25:42 <psykose> it is that
2023-02-18 01:25:55 <psykose> https://pkgs.alpinelinux.org/flagged
2023-02-18 01:27:37 <minimal> thanks
2023-02-18 07:09:58 <wener> Hi, is there any place I can find all pkgs flaged as outdated ?
2023-02-18 07:10:28 <abby> https://pkgs.alpinelinux.org/flagged
2023-02-18 07:10:54 <wener> Thanks, any json ?
2023-02-18 07:16:58 <wener> Working on a script to upgrade outdated pkgs https://github.com/wenerme/wode/blob/main/packages/alpine/src/aports/upgrade.ts , want to pull all outdated pkg to test.
2023-02-18 08:53:36 <ikke> wener: there is no API atm
2023-02-18 08:54:42 <wener> Thanks, I will grep html
2023-02-18 09:24:34 <omni> you can use regex
2023-02-18 09:44:04 <ikke> heh
2023-02-18 10:18:24 <omni> for the uninitiated: https://stackoverflow.com/a/1732454
2023-02-18 15:50:52 <caskd> someone broke g+++
2023-02-18 15:51:03 <caskd> g++* signature
2023-02-18 15:51:26 <caskd> ERROR: g++-12.2.1_git20220924-r9: BAD signature
2023-02-18 15:53:35 <ikke> caskd: what mirror? dl-cdn?
2023-02-18 15:53:54 <caskd> yep
2023-02-18 15:54:16 <caskd> on edge
2023-02-18 15:54:59 <ikke> no issue here..
2023-02-18 15:55:08 <ikke> (5/5) Installing g++ (12.2.1_git20220924-r9)
2023-02-18 15:55:13 <ikke> OK: 234 MiB in 30 packages
2023-02-18 15:55:19 <caskd> uh
2023-02-18 15:55:27 <caskd> might be my package cache i guess
2023-02-18 15:55:28 <caskd> sec
2023-02-18 15:56:01 <caskd> oh yeah
2023-02-18 15:56:03 <caskd> nvm
2023-02-18 15:56:10 <caskd> something went wrong there
2023-02-18 15:56:10 <ikke> ok
2023-02-18 15:56:22 <ikke> I didn't see anything that would normally cause these kinds of bad signatures
2023-02-18 15:56:33 <ikke> (ie, reverting a package with the same pkgrel as before)
2023-02-18 16:43:12 <minimal> wener: perhaps the htmlq package would be useful for parsing HTML?
2023-02-18 18:31:42 <staceee> Is it possible to disable checksum checking in APKBUILD? It is for a "nightly" self hosted repo
2023-02-18 19:00:24 <ikke> I don't see an option 
2023-02-18 22:35:14 <Cogitri> staceee: You could fetch the source yourself in fetch()
2023-02-18 22:48:39 <kreyren> Piraty, https://gitlab.alpinelinux.org/alpine/aports/-/commit/f1b1fcf8300a733f01a86355b9853ec360670493 mesa seems to have pushed a patch for the A64
2023-02-18 22:49:06 <kreyren> oh it's alpine patch O.o
2023-02-19 02:58:59 <elibrokeit> psykose: the zstd APKBUILD comments about +30% speed refer to the PGO use, right?
2023-02-19 02:59:12 <psykose> seemingly
2023-02-19 02:59:18 <psykose> just some zstd -b's i ran
2023-02-19 02:59:44 <elibrokeit> pretty sweet
2023-02-19 03:04:22 <psykose> maybe it was also something else tbf, idk
2023-02-19 09:27:34 <wener> minimal: parsed by jsdom https://apis.wener.me/api/alpine/pkg/flagged but the flagged page is slow, can not fetch all in 10s
2023-02-19 09:28:36 <wener> setup-interfaces can not setup for multi eth, only set the last one
2023-02-19 09:28:53 <wener> s/setup-interfaces/setup-interfaces -a/
2023-02-19 09:28:53 <alpine-meetbot> wener meant to say: setup-interfaces -a can not setup for multi eth, only set the last one
2023-02-19 09:31:07 <psykose> yeah it picks one
2023-02-19 09:31:20 <psykose> see auto_setup() in setup-interfaces
2023-02-19 09:31:23 <psykose> doesn't configure all
2023-02-19 09:32:13 <wener> should I create a issue for this ? I expected to setup all, make auto setup easier
2023-02-19 09:32:24 <psykose> sure, an alpine-conf feature request i guess
2023-02-19 09:32:36 <psykose> also a script to automatically update things from the flagged page doesn't sound very useful
2023-02-19 09:33:32 <psykose> most of the work of updating something is not incrementing the version and updating the checksums for it, so it's not very useful even for that simplest cast
2023-02-19 09:33:34 <psykose> case*
2023-02-19 09:33:43 <psykose> and for every other case it would just be wrong
2023-02-19 09:35:31 <wener> yes, only pick those hosted on github, will verify patch
2023-02-19 09:37:29 <psykose> not sure how it's relevant where it's hosted
2023-02-19 10:31:15 <wener> I mean the src is github.com
2023-02-19 10:33:31 <psykose> yeah, i have no idea how that is relevant to anything
2023-02-19 10:40:12 <staceee> someone know an effective way to cleanup world? Research packages already in the dependency graph, lib I installed once that no one rely on anymore, etc
2023-02-19 10:40:44 <staceee> I'm trying apk-dot but graphviz seems to struggle with the large output
2023-02-19 10:43:34 <psykose> i just read the world file
2023-02-19 10:44:05 <psykose> since you generally don't manually install libs and everything is purged on uninstall if you didn't have it in world, there's not much to usually clean
2023-02-19 10:44:11 <psykose> aside from a bunch of -dev i forgot about
2023-02-19 10:46:52 <pogduhog> Why use a modloop instead of individual kmod packages ala openwrt? Been looking into the boot process and I can't see a benefit so I'm curious as to this design decision
2023-02-19 10:48:08 <psykose> those two things seem at different levels of abstraction and not related to eachother
2023-02-19 10:48:45 <psykose> but for the latter, because having individual module packages is.. annoying and nobody has the time to figure out which modules they need, i guess
2023-02-19 10:50:29 <pogduhog> Perhaps I am misunderstanding then. To me it seems all kernel modules are packaged into a modloop whether or not they will be needed so individual packages makes more sense. Just a curiosity really
2023-02-19 10:51:18 <pogduhog> Yes it is a pain to mind the right modules, but a lot of this is already done with the initramfs feature files 
2023-02-19 10:52:13 <pogduhog> And you can always just look at the openwrt kmod package for inspiration 
2023-02-19 10:53:12 <pogduhog> As I said, not a critism, just a curiosity and wondering if there was an inherent benefit to a modloop I didn't understand
2023-02-19 10:54:56 <ikke> For an installation medium you want to support most systems, so it would need to have all modules available anyway
2023-02-19 10:56:53 <ikke> The modloop contains not only modules but firmware as well
2023-02-19 10:58:47 <pogduhog> Yes, it makes more sense for general consuption to have all modules available. 
2023-02-19 10:59:44 <ikke> and the idea of the modloop is to save memory on run-from-ram systems
2023-02-19 11:00:44 <pogduhog> I see, yes that makes sense
2023-02-19 11:02:09 <pogduhog> For my case it won't make a difference though as I am booting entirely over the network so the modloop will be loaded into memory anyway. In this situation it actually uses more ram since I have to have them all in memory
2023-02-19 11:03:15 <ikke> yeah, modloop only makes sense when you have storage
2023-02-19 11:04:26 <pogduhog> Gotcha, I understand 
2023-02-19 11:05:53 <ikke> did openwrt split up every module in a dedicated package?
2023-02-19 11:10:00 <psykose> it makes sense there because routers can get pretty small
2023-02-19 11:10:06 <psykose> what like 32MB of ram or less? :-)
2023-02-19 11:10:20 <ikke> https://openwrt.org/packages/index_owrt18_6/kernel-modules
2023-02-19 11:11:39 <ikke> Is the idea that you manually install any module you need, or are there some presets?
2023-02-19 11:32:10 <pogduhog> With openwrt everything needed to boot is compiled in. Their kmod packages are similar to the mkinitfs features but perhaps a bit more granular
2023-02-19 11:35:00 <psykose> i'd imagine it's not like the features, because the features are purely for the initfs, while not having modules installed is not having modules installed :)
2023-02-19 11:35:15 <psykose> most of the modules i use are not in the initfs because they can just be loaded later for example
2023-02-19 11:35:42 <psykose> but if you mean the definition, sure, it's just defining a "set of things you use" in general
2023-02-19 11:38:15 <pogduhog> That's fair
2023-02-19 20:03:36 <ikke> Hmm, never realized, but apparently ash prevents the ssh escape character from working: https://superuser.com/a/985448/4134
2023-02-19 20:07:43 <psykose> >If you press Enter then ~ and it will not appear the character ~ on the prompt that escape will be act in the ssh. 
2023-02-19 20:07:55 <psykose> my head exploded
2023-02-19 20:09:05 <ikke> You were not aware of ~ in ssh?
2023-02-19 20:09:30 <ikke> Stuck ssh session -> <enter>~.
2023-02-19 20:11:04 <minimal> you can change the escape character with "-e" or EscapeChar in ssh_config
2023-02-19 20:11:17 <ikke> yes, but the issue is not the escape character itself
2023-02-19 20:11:39 <minimal> don't think I ever remember using ssh escape character though (do remember using telnet escape sequence)
2023-02-19 20:11:46 <ikke> but rather that the ash prompt sends escape characters that occur between the <enter> and ~<x>
2023-02-19 20:12:05 <ikke> I use it to dynamically change forwards
2023-02-19 20:12:27 <ikke> <enter>~c -> -L 1234:1.2.3.4:123
2023-02-19 20:12:45 <psykose> if i try type that it just.. enter ~ c
2023-02-19 20:12:50 <psykose> no escape or anything
2023-02-19 20:12:55 <psykose> how do you input that
2023-02-19 20:12:57 <ikke> cat
2023-02-19 20:13:03 <ikke> A workaround
2023-02-19 20:13:16 <psykose> i typed cat
2023-02-19 20:13:20 <ikke> enter
2023-02-19 20:13:23 <ikke> ~c
2023-02-19 20:13:38 <ikke> sorry, capital C
2023-02-19 20:13:50 <psykose> >escape not available to multiplexed sessions
2023-02-19 20:13:51 <psykose> heh
2023-02-19 20:13:53 <psykose> ok sure
2023-02-19 20:14:02 <psykose> thanks
2023-02-19 20:14:45 <lucidiot> >commandline disabled
2023-02-19 20:25:30 <minimal> ikke: deja vu, I think this came up before some time ago
2023-02-20 11:23:01 <minimal> ikke: re ash and ssh escape, seems like disabling FEATURE_EDITING_ASK_TERMINAL in Busybox config may prevent this behaviour, though not sure if there are implications of doing so...
2023-02-20 11:23:36 <ikke> me neither
2023-02-20 11:24:26 <ikke> "it tells the terminal emulator to send back another ANSI escape code, which tells the shell where the cursor (text insertion point) is located in the terminal window."
2023-02-20 11:25:35 <minimal> "If you are not sure whether your terminals respond to this code correctly, or want to save on code size (about 400 bytes), then do not turn on this option on."
2023-02-20 11:26:32 <ikke> What would the shell use the cursor location for?
2023-02-20 11:32:04 <minimal> ikke: dunno, that code is inside read_line_input which is used by more than just shell in Busybox, e.g. by hexedit, ed, and fdisk
2023-02-20 11:38:52 <Piraty> what's jirutka's irc nick?
2023-02-20 11:38:59 <ikke> Piraty: he's not on irc
2023-02-20 11:39:47 <Piraty> thank you
2023-02-20 16:12:32 <bl4ckb0ne> is there a python update in progress?
2023-02-20 16:12:49 <bl4ckb0ne> wlroots alpine CI chugs on ncurses
2023-02-20 16:13:36 <ptrc> s390x is in progress, other than that all's idle
2023-02-20 16:13:52 <bl4ckb0ne> https://gitlab.freedesktop.org/wlroots/wlroots/-/jobs/36714190
2023-02-20 16:14:14 <bl4ckb0ne> whack
2023-02-20 16:14:36 <bl4ckb0ne> ncurses was updated 7 hoursa go
2023-02-20 16:14:52 <ptrc> weird, ncurses-libs-6.4_p20230211 literally provides those libraries
2023-02-20 16:15:00 <ptrc> ah
2023-02-20 16:19:34 <bl4ckb0ne> missing rebuild?
2023-02-20 16:20:20 <ptrc> it looks like your local ncurses-libs needs to be updated to 20230218 to match the -dev you're trying to install
2023-02-20 16:26:36 <psykose> don't think that's the isssue
2023-02-20 16:34:07 <omni> anyone know of this "DNS randomly not working"? https://lobste.rs/s/bozzog/praise_alpine_apk#c_zpvyhl
2023-02-20 16:35:13 <omni> can't recall seeing anything like that being mentioned before
2023-02-20 16:35:49 <omni> in the context of it being an alpine issue
2023-02-20 16:37:06 <Ermine> Maybe it is related to old musl getaddrinfo implementation?
2023-02-20 16:40:30 <elly> the thing where it doesn't fall back to TCP for too-long replies?
2023-02-20 16:44:05 <elly> I vaguely remember this from some months ago, someone had a hostname that musl's resolver couldn't resolve, but nslookup could
2023-02-20 16:46:14 <invoked> tcp pipelining
2023-02-20 16:46:49 <invoked> felker made some changes and posted about it recently
2023-02-20 16:46:53 <elly> ooh
2023-02-20 16:48:45 <invoked> it's been talked about for a long time, eg here  https://twitter.com/richfelker/status/994629795551031296
2023-02-20 16:49:22 <invoked> so musl recently changed to do tcp fallback, but i don't remember the specifics
2023-02-20 16:49:26 <invoked> i can't find it, giving up
2023-02-20 16:50:35 <invoked> long story short, some stuff just wouldn't work on musl without it, so it forced the change
2023-02-20 16:51:16 <psykose> yeah, next release has working tcp fallback
2023-02-20 16:51:19 <psykose> i tested it somewhat
2023-02-20 16:53:56 <invoked> iirc recursive queries that return long (too many long aws names, or whatever) exceed what udp can handle which necessitates tcp
2023-02-20 16:56:25 <minimal> could also be related to musl's parallel lookups with multiple DNS servers vs glibc sequential lookups, where people have the 1st DNS server for "local" lookups (i.e. Consul or similar)
2023-02-20 16:58:36 <psykose> which also isn't valid usage
2023-02-20 17:01:27 <minimal> yeah, it's something that crops up occasionally though
2023-02-20 18:06:56 <skarnet> I've written a fix for that, specifically for Alpine, 2 years ago, and it has never been integrated
2023-02-20 18:06:59 <skarnet> so *shrug*
2023-02-20 18:48:14 <omni> where's that?
2023-02-20 18:52:08 <skarnet> it's called dnsfunnel. it's packaged. it required some integration for the way it handles /etc/resolv.conf, so coordination with Alpine core devs to do that. It never happened.
2023-02-20 18:57:56 <omni> it needs testing by anyone who can simulate spurious NXDOMAIN and confirm that it is working as intended to motivate it being moved from testing?
2023-02-20 18:58:18 <skarnet> that would be great, yes
2023-02-20 19:00:01 <skarnet> and beyond that in order for it to be useful for all users it would need some deeper integration with the system, and that's a, hmmm, delicate subject with Alpine
2023-02-20 19:00:25 <skarnet> so I'm kinda resigned on musl TCP getting it first
2023-02-20 19:00:31 <skarnet> oh well
2023-02-20 19:00:45 <ikke> The tricky part is that it requires a daemon to be running
2023-02-20 19:01:03 <skarnet> daemons are so scary
2023-02-20 19:01:09 <ikke> They are not
2023-02-20 19:01:20 <skarnet> they're so tricky
2023-02-20 19:01:23 <ikke> But most users who complain about this are often using docker
2023-02-20 19:01:52 <skarnet> yeah, how many of these users keep all the gettys in inittab, in docker
2023-02-20 19:02:01 <skarnet> that's 6 daemons they could do without
2023-02-20 19:02:20 <skarnet> there's no argument I've heard about all this that was not ridiculous
2023-02-20 19:02:53 <ikke> there are not gettys running in docker?
2023-02-20 19:02:56 <ikke> no*
2023-02-20 19:03:15 <skarnet> fine, let me have a look
2023-02-20 19:03:31 <ikke> You just start a single process in docker by default
2023-02-20 19:03:44 <ikke> that process then could spawn other processes
2023-02-20 19:04:57 <ikke> https://tpaste.us/d1k8 
2023-02-20 19:06:10 <skarnet> yeah, so they're not running Alpine in docker, they're running empty containers with Alpine binaries, that's very different
2023-02-20 19:06:55 <skarnet> and what you're saying is that they want everything to magically work with zero system processes
2023-02-20 19:07:06 <skarnet> so, pure client-side modifications
2023-02-20 19:08:15 <psykose> that's how the majority of containers work yeah
2023-02-20 19:08:22 <psykose> just one process
2023-02-20 19:08:41 <skarnet> no, that's how Docker says the majority of containers should work, but they've been wrong from the start
2023-02-20 19:08:55 <skarnet> it's not a reasonable or sustainable model
2023-02-20 19:09:29 <skarnet> and *of course* there's gonna be issues for people who do that
2023-02-20 19:09:36 <psykose> i'm just telling you how it is
2023-02-20 19:09:43 <psykose> in k8s land they came up with ""pods"" as an alternative
2023-02-20 19:09:49 <psykose> which is.. a grouping of single process containers
2023-02-20 19:09:50 <psykose> very funny
2023-02-20 19:09:59 <ikke> Yeah
2023-02-20 19:10:07 <skarnet> I know you're just telling how it is, but as I said:
2023-02-20 19:10:07 <skarnet> <19skarnet> there's no argument I've heard about all this that was not ridiculous
2023-02-20 19:10:21 <ikke> ridiculous to you
2023-02-20 19:10:56 <psykose> it's not ridiculous because dnsfunnel was always usable
2023-02-20 19:11:01 <psykose> i even used it myself
2023-02-20 19:11:11 <psykose> you merely install it and start it, nothing more needed
2023-02-20 19:11:20 <skarnet> yup
2023-02-20 19:11:34 <skarnet> so what do you call it when people won't do that?
2023-02-20 19:11:36 <psykose> the issue is that you cannot do exactly that in the way everyone uses containers, and there's nothing alpine can do about that
2023-02-20 19:11:42 <psykose> what are we supposed to do
2023-02-20 19:11:52 <psykose> "ah yes lads, please change the way you run containers"
2023-02-20 19:11:59 <psykose> x1000000 for everyone that did it that way
2023-02-20 19:12:06 <skarnet> it's all I have
2023-02-20 19:12:08 <psykose> what is the request or special integration to be done exactly
2023-02-20 19:12:21 <skarnet> I don't have any magic powers to fix everything process-less by snapping my fingers
2023-02-20 19:12:23 <skarnet> sorry
2023-02-20 19:12:29 <skarnet> or maybe
2023-02-20 19:12:32 <skarnet> I should write a kernel module
2023-02-20 19:12:39 <skarnet> that way, no process, weeeee
2023-02-20 19:12:42 <psykose> i don't think anyone asked you to fix it
2023-02-20 19:12:47 <psykose> or maybe they did, i wasn't there
2023-02-20 19:13:04 <skarnet> Ariadne was talking about the problem
2023-02-20 19:13:04 <Newbyte> kthreads are obviously superior to processes
2023-02-20 19:13:09 <skarnet> I said I would fix it
2023-02-20 19:13:10 <skarnet> and I did
2023-02-20 19:13:25 <psykose> you presented an option people could use to fix it, yes
2023-02-20 19:13:25 <Ariadne> ?
2023-02-20 19:13:32 <ikke> Ariadne: dnsfunnel
2023-02-20 19:13:40 <psykose> nothing else to be done from that point
2023-02-20 19:13:53 <skarnet> it appears the solution cannot be used because people are using containers wrong
2023-02-20 19:13:56 <ikke> truncated dns
2023-02-20 19:14:02 <psykose> sure, and that's their problem
2023-02-20 19:14:06 <skarnet> so, as I said, *shrug*, and you can't make me say it's not ridiculous
2023-02-20 19:14:22 <psykose> i'm not sure why you are upset about it, unless you want to personally argue with literally everyone running containers
2023-02-20 19:14:28 <psykose> nobody here controls how people choose to use them
2023-02-20 19:14:37 <skarnet> I'm doing that every day with s6-overlay, dear
2023-02-20 19:14:45 <psykose> yeah, and it's nice
2023-02-20 19:14:52 <Ariadne> technically, kubernetes could solve this, and docker generally does not have this problem
2023-02-20 19:15:23 <skarnet> psykose: I'm not upset about this, I'm just very tired because of the accumulation over 20+ years
2023-02-20 19:15:32 <psykose> it certainly is tiring
2023-02-20 19:15:42 <psykose> but i mean, anyone that knew how to solve it, solved it
2023-02-20 19:15:52 <skarnet> so now I think I've stopped caring about feefees and will say something is ridiculous when it is
2023-02-20 19:15:59 <psykose> it's not an impossible to fix problem in principle with the right knowledge of the right part of the stack, as a person that runs into it at some company
2023-02-20 19:16:22 <psykose> tcp fallback merely just solves it now for everyone else by default, and that's all there is to it
2023-02-20 19:16:37 <psykose> dnsfunnel presented yet another tool someone in the former group could pick from
2023-02-20 19:16:55 <skarnet> Magic Dalias will save the day this time, but I certainly wouldn't want to be in his place
2023-02-20 19:17:07 <skarnet> the shit people ask of musl is even more egregiously stupid
2023-02-20 19:17:33 <skarnet> and it has the exact same limitations: you have to do everything in libc, you can't have a daemon process
2023-02-20 19:17:36 <skarnet> everything client-side
2023-02-20 19:17:44 <skarnet> so of course there are things it cannot do
2023-02-20 19:18:09 <ikke> The issue is that people compare it to glibc
2023-02-20 19:18:18 <ikke> things are working in glibc and thus musl must be broken
2023-02-20 19:18:29 <skarnet> things appear to be working in glibc :)
2023-02-20 19:19:22 <ikke> skarnet: I hope you understood this was from the point of someone complaining about musl
2023-02-20 19:19:32 <skarnet> yes
2023-02-20 19:22:01 <ikke> And I think Cormack had a nice point: if you have specific requirements for dns, you should use a specialized library
2023-02-20 19:22:36 <ikke> "The point at the end that if you want total control over every detail you should use a specialist library is a very good one, and to be encouraged."
2023-02-20 19:28:05 <skarnet> absolutely
2023-02-20 19:29:13 <Ermine> name resolution shouldn't have appeared in libc in the first place
2023-02-20 19:43:20 <skarnet> yeah :/
2023-02-20 19:45:30 <bl4ckb0ne> ha, found the issue ncurses issue in 3e22458e9716a4a341144baa7ba4815861227b55
2023-02-20 19:47:34 <bl4ckb0ne> emacs still tracks ncurses-libs
2023-02-20 19:56:31 <psykose> that came way after whatever you ran into
2023-02-20 19:56:38 <psykose> if you look at the dates
2023-02-20 19:57:27 <bl4ckb0ne> ah true
2023-02-20 19:57:33 <bl4ckb0ne> 
ACTION grumbles after the computer

2023-02-20 19:59:09 <psykose> the output from your issue looks kinda funny because usually there is another half
2023-02-20 19:59:10 <psykose> strange
2023-02-20 19:59:56 <bl4ckb0ne> its easily fixed by running an upgrade on the CI runner but each new run fails because it wipes everything
2023-02-20 20:00:17 <psykose> i can't reproduce that myself by just doing it manually
2023-02-20 20:00:19 <bl4ckb0ne> havent been able to find what drags it down
2023-02-20 20:00:24 <bl4ckb0ne> yeah same on my system
2023-02-20 20:00:32 <psykose> and i couldn't find what the actual ci does
2023-02-20 20:00:40 <bl4ckb0ne> https://git.sr.ht/~sircmpwn/builds.sr.ht/tree/master/item/images/alpine/functions
2023-02-20 20:00:47 <bl4ckb0ne> update & add all my shit
2023-02-20 20:01:54 <psykose> sure but what does it start with
2023-02-20 20:02:56 <bl4ckb0ne> what do you mean
2023-02-20 20:03:37 <psykose> ci jobs start from somewhere with something
2023-02-20 20:03:41 <psykose> then something runs
2023-02-20 20:04:09 <psykose> idk how any of these parts work
2023-02-20 20:04:21 <psykose> or what they're doing in your case
2023-02-20 20:04:35 <bl4ckb0ne> it has an image generated daily (dunno when)
2023-02-20 20:04:44 <bl4ckb0ne> then the runner boots that image and adds my shit
2023-02-20 20:05:01 <psykose> ok think i found it
2023-02-20 20:05:29 <bl4ckb0ne> one of the dep is stuck on ncurses-terminfo-base, i reproduced the error on the runner
2023-02-20 20:05:37 <psykose> link
2023-02-20 20:05:42 <ikke> http://dup.pw/alpine/aports/584e86ab3a45
2023-02-20 20:05:43 <psykose> and paste logs
2023-02-20 20:06:21 <bl4ckb0ne> https://paste.sr.ht/~bl4ckb0ne/236283365786bf5d3bb1a6334aa885b347da38f6
2023-02-20 20:06:45 <bl4ckb0ne> thats run every day https://git.sr.ht/~sircmpwn/builds.sr.ht/tree/master/item/images/alpine/genimg
2023-02-20 20:06:57 <bl4ckb0ne> and then for each runner it runs the script i linked earlier
2023-02-20 20:08:07 <psykose> could you give me ssh to that right before it fails
2023-02-20 20:08:45 <bl4ckb0ne> cant, its locked to my ssh key
2023-02-20 20:08:57 <psykose> hm
2023-02-20 20:09:21 <psykose> i dunno it's weird
2023-02-20 20:09:31 <psykose> it does see the new versions so it would normally just upgrade
2023-02-20 20:09:34 <psykose> as it does everywhere else
2023-02-20 20:10:33 <psykose> hm, is that image published somewhere so i could just run it
2023-02-20 20:12:09 <ikke> I think I saw this sort of issue before with curl
2023-02-20 20:12:57 <ikke> No, that was different. There libcurl was not updated when a newer curl version as installed
2023-02-20 20:13:05 <ikke> was*
2023-02-20 20:14:03 <psykose> that was just a runtime issue
2023-02-20 20:14:21 <psykose> in any case i think this is some bug in the new minor changes in apk probably, it looks a little too weird
2023-02-20 20:14:35 <psykose> haven't reproduced it by hand but i've mostly pieced the state there
2023-02-20 20:14:59 <psykose> image starts, has the old version, runs apk update, runs apk add bunchofstuff that pulls an extra thing with the inbetween deps in the ncurses thing
2023-02-20 20:15:06 <psykose> and for some reason it doesn't upgrade with the add
2023-02-20 20:15:13 <psykose> on top of missing half the error output
2023-02-20 20:15:17 <bl4ckb0ne> no answer yet for the source, i also asked to redo an image gen
2023-02-20 20:15:35 <bl4ckb0ne> i tried a few thigns but nothing outputs more
2023-02-20 20:15:46 <psykose> i know you see everything
2023-02-20 20:15:51 <psykose> i just mean normally you'll see
2023-02-20 20:15:53 <psykose> - that block
2023-02-20 20:16:03 <bl4ckb0ne> i did `apk add vim` and it worked fine
2023-02-20 20:16:09 <psykose> - another block with ncurses-terminfo-base-$newversion-r0: and a list of things blocking it
2023-02-20 20:16:13 <psykose> hm
2023-02-20 20:16:38 <psykose> and if you just upgrade it.. upgrades?
2023-02-20 20:16:58 <bl4ckb0ne> xwayland-dev!
2023-02-20 20:17:22 <bl4ckb0ne> new image gens in 15 mins so should be good
2023-02-20 20:18:06 <psykose> if u just run upgrade and then add stuff does it work
2023-02-20 20:18:14 <bl4ckb0ne> yeah
2023-02-20 20:19:40 <bl4ckb0ne> always X breaking my shit
2023-02-20 20:19:46 <bl4ckb0ne> why cant we have a wayland utopia
2023-02-20 20:20:16 <psykose> well, that's still a bug then
2023-02-20 20:20:24 <psykose> the add should work by itself and upgrade the things to match
2023-02-20 20:20:30 <psykose> oor
2023-02-20 20:20:30 <psykose> wait
2023-02-20 20:22:29 <ikke> 
ACTION is hanging on

2023-02-20 20:23:38 <ikke> psykose: is https://gitlab.alpinelinux.org/alpine/aports/-/issues/14088#note_292387 still relevant?
2023-02-20 20:23:49 <ikke> Sorry, I mean the entire issue, not that comment
2023-02-20 20:24:06 <ikke> I think rust builds fine now, right?
2023-02-20 20:24:50 <psykose> it seems related to https://gitlab.alpinelinux.org/alpine/apk-tools/-/commit/067c17312418937b6fc81cbba78349ce2566f942
2023-02-20 20:24:58 <psykose> ikke: nah, fixed
2023-02-20 20:25:04 <ikke> right
2023-02-20 20:25:08 <psykose> well, kinda sometimes fails
2023-02-20 20:25:11 <psykose> and there's other issues
2023-02-20 20:25:13 <psykose> but not that one!
2023-02-20 20:26:46 <ikke> I missed you already commented on it and closed it :P
2023-02-20 20:49:47 <bl4ckb0ne> seems to be running fine now
2023-02-20 20:49:52 <bl4ckb0ne> thanks for poking around psykose 
2023-02-20 20:52:40 <psykose> unfortunate
2023-02-20 21:15:09 <psykose> bl4ckb0ne: beat you
2023-02-20 21:15:14 <bl4ckb0ne> :<
2023-02-20 21:15:23 <bl4ckb0ne> beat you because there's also a gamescope fix and CI breaks
2023-02-20 21:15:44 <bl4ckb0ne> https://gitlab.alpinelinux.org/bl4ckb0ne/aports/-/pipelines/153870
2023-02-20 21:15:47 <bl4ckb0ne> checkmate
2023-02-20 21:16:34 <psykose> mine passes
2023-02-20 21:16:36 <psykose> beat you x2
2023-02-20 21:17:07 <ikke> lol
2023-02-20 21:17:15 <bl4ckb0ne> how
2023-02-20 21:18:32 <bl4ckb0ne> did you just turn of werror you naughty dev
2023-02-20 21:18:55 <psykose> hey if immersion stopped putting werror into every project i wouldn't have to :p
2023-02-20 21:18:58 <ikke> werror is for upstream :P
2023-02-20 21:19:25 <psykose> the actual fail for sign-compare looks legit as a 32-bit failure i guess
2023-02-20 21:19:35 <bl4ckb0ne> i linked him the errors already, should have a rev soon
2023-02-20 21:19:39 <psykose> damn
2023-02-20 21:19:42 <psykose> speedy
2023-02-20 21:19:46 <bl4ckb0ne> 2fast4u
2023-02-20 21:20:06 <psykose> bet i'd beat you in a race old man
2023-02-20 21:20:18 <bl4ckb0ne> 
ACTION shakes fist at youngster

2023-02-20 21:20:39 <psykose> how old are you anyway
2023-02-20 21:20:39 <bl4ckb0ne> with the state of my knees you can beat me at many things
2023-02-20 21:20:42 <psykose> my guess is... 35
2023-02-20 21:20:47 <bl4ckb0ne> 
ACTION checks cal

2023-02-20 21:20:53 <bl4ckb0ne> id say about 27 
2023-02-20 21:20:55 <psykose> damn
2023-02-20 21:20:56 <psykose> tiny
2023-02-20 21:20:57 <psykose> small
2023-02-20 21:21:15 <bl4ckb0ne> and already worned out by computers
2023-02-20 21:21:27 <bl4ckb0ne> my dream is to raise goats in the mountains and craft my own furnitures
2023-02-20 21:21:31 <psykose> worn out by computers is just default state these days
2023-02-20 21:21:32 <bl4ckb0ne> away from those evil rocks
2023-02-20 21:21:53 <psykose> sadly the goats don't know much about build systems
2023-02-20 21:22:01 <psykose> but they did invent autotools! so there's that
2023-02-20 21:24:12 <Newbyte> are gnus goats?
2023-02-20 21:24:54 <bl4ckb0ne> i bet a goat can beat autotools in many ways
2023-02-20 21:25:02 <bl4ckb0ne> can a goat do milk? yes
2023-02-20 21:25:09 <bl4ckb0ne> canautotools do milk? no
2023-02-20 21:25:14 <bl4ckb0ne> checkmate RMS
2023-02-20 21:26:41 <Newbyte> can the goat make you curse though?
2023-02-20 21:26:43 <elly> I would rather pet a goat than autotools
2023-02-20 21:26:49 <Newbyte> not likely I say
2023-02-20 21:26:54 <pj> <skarnet> "no, that's how Docker says the..." <- old man shouts at clouds
2023-02-20 21:28:11 <Newbyte> pj: oh no, you used a matrix reply! the IRC users will burn you at the stake
2023-02-20 21:29:44 <pj> Newbyte: why would they?
2023-02-20 21:30:28 <Newbyte> pj: I have been cursed at for doing it. Maybe they improved how the bridge displays replies since.
2023-02-20 21:31:08 <Newbyte> I haven't used IRC for years. don't know what it looks like on their end
2023-02-20 21:31:16 <pj> last time I checked it does truncated message and reply after it
2023-02-20 21:32:34 <pj> Newbyte: yeah, I can't be bothered to use IRC client and I stopped using irccloud
2023-02-20 21:32:57 <pj1> well, it seems it doesn't even mention the reply anymore lol
2023-02-20 21:40:26 <bl4ckb0ne> another pros for goats: they dont use matrix
2023-02-20 21:42:37 <psykose> oh boy weekly chat protocol discussion, how exciting
2023-02-20 21:45:44 <psykose> they're all great in their own ways and people can use whatever
2023-02-20 21:45:44 <bl4ckb0ne> goats dont chat
2023-02-20 21:45:47 <psykose> that should cover it
2023-02-20 21:46:07 <bl4ckb0ne> but which one is better and why
2023-02-20 21:46:49 <invoked> roll d20 for initiative
2023-02-20 21:47:51 <ptrc> psykose: nope, all chat platforms are bad and should be abolished
2023-02-20 21:48:03 <psykose> why do you keep messaging me then
2023-02-20 21:48:04 <psykose> checkmate
2023-02-20 21:48:11 <psykose> log off queen
2023-02-20 21:48:28 <ptrc> we should improve society somehow dot png
2023-02-20 21:49:00 <ptrc> /s
2023-02-20 21:50:30 <Ermine> Cocogoats where
2023-02-20 21:57:24 <Newbyte> invoked: I attack goat no. 6 with my sledgehammer
2023-02-20 21:57:57 <invoked> your party disbanded
2023-02-20 22:24:18 <bl4ckb0ne> goat 6 counter attacks with autotools
2023-02-20 22:35:27 <Newbyte> I'm defeated
2023-02-21 02:02:49 <skarnet> pj: free software is made by people who are discontent with the way bad tech is enforced on us by corporate overlords, and want to make things better. Who knows, maybe someday you'll join us.
2023-02-21 02:44:00 <elly> not all free software is made that way, definitely :P
2023-02-21 02:44:07 <elly> some of us get paid to do it by our corporate overlords
2023-02-21 02:45:31 <psykose> free software is made by people tired of companies telling them what to do so now they tell everyone else what to do in the one true correct way only via their free software
2023-02-21 02:45:34 <psykose> am i doing it correctly
2023-02-21 02:46:32 <valerius> free software is made by rich people who stopped caring about making money writing software
2023-02-21 02:47:40 <valerius> and for option 4, all of the above
2023-02-21 02:49:06 <skarnet> you people are exhausting
2023-02-21 02:49:19 <skarnet> how about "historically"
2023-02-21 02:51:40 <psykose> poor in money but rich in soul
2023-02-21 02:51:49 <psykose> so i guess you could say they are the richest of all
2023-02-21 02:53:35 <skarnet> thinking that FOSS is made by people who don't care about making money is one of the most harmful misconceptions one can have about the community
2023-02-21 02:54:27 <skarnet> same energy as "we want passionatem, driven people" (small print: so they'll take a pay cut without complaints)
2023-02-21 02:55:07 <bl4ckb0ne> "we want rockstars"
2023-02-21 02:55:39 <valerius> lol, I'm just describing myself in that context
2023-02-21 02:55:57 <valerius> the truth is that all sorts of people are doing this for all sorts of reasons
2023-02-21 02:56:08 <skarnet> valerius: then you should learn your value and get properly compensated :P
2023-02-21 02:56:11 <abby> code ninjas
2023-02-21 03:05:07 <psykose> prerequisite to getting paid is doing something useful
2023-02-21 03:06:21 <skarnet> ... have you ever met a salesman? XD
2023-02-21 03:08:08 <psykose> yeah
2023-02-21 03:08:13 <psykose> i didn't pay them! just saying
2023-02-21 03:08:52 <valerius> ultimately what matters is whether you enjoy doing it or not
2023-02-21 03:09:41 <valerius> if it stops being enjoyable, it is time to consider a simpler life in the countryside with chickens and your favourite crops
2023-02-21 03:10:09 <skarnet> psykose: what I meant is, there are a lot of useless jobs, fulfilled by useless people who get a salary anyway. Prerequisite to getting paid is pretty orthogonal to doing something useful, unfortunately.
2023-02-21 03:10:34 <valerius> hey, don't be so mean to scrum masters and project managers :p
2023-02-21 03:10:50 <minimal> project manglers? ;-)
2023-02-21 03:11:08 <skarnet> valerius: I think all the best people I know in FOSS are always on the verge of leaving everything to go raise goats
2023-02-21 03:11:25 <skarnet> and it actually happened to at least two
2023-02-21 03:11:42 <valerius> lol... well, I have to admit I have been looking at more rural locations where I can raise animals lately :p
2023-02-21 03:11:50 <minimal> why goats specifically?
2023-02-21 03:12:04 <valerius> any time I hear about kernel modules written in Rust it causes me to inch closer to the agrarian life
2023-02-21 03:12:32 <skarnet> minimal: because goats are adorable, but sheep, alpaca or cows are also an option
2023-02-21 03:14:06 <minimal> I tried to leave my last boss, but I couldn't lol
2023-02-21 03:14:51 <minimal> <looks in mirror> He's still here... (well he's no longer my boss)
2023-02-21 03:16:53 <skarnet> yeah I was thinking about bad tech, but being at odds with corporate culture and bosses is another classic reason for burnout
2023-02-21 03:17:15 <skarnet> being an independent helps, tremendously, but not everyone can do that, especially not young people
2023-02-21 03:19:35 <minimal> yeah in theory you can pick and choose which work to take on/avoid, however in many cases the true nature of the work (and any unfortunate design decisions already made etc) does not become apparent until you're several weeks/months into it
2023-02-21 03:21:49 <skarnet> it doesn't make the projects better, but (IME) it makes the day-to-day drudgery better, knowing that it's not your project, that these guys are not your bosses, and if one day you're fed up you can flip them the bird and send them all packing. Not that it would be a wise thing to do too often, but just the fact you *can* helps.
2023-02-21 03:25:58 <minimal> the last employment I had I left after 9 months due to the corporate culture/management, my direct manager/team leader gave notice, a senior manager came over to try and change his mind, she met the team and 2 more of us resigned the same day lol
2023-02-21 03:27:25 <skarnet> and then she got promoted
2023-02-21 03:28:00 <minimal> oh no, she was senior management, coasting towards her retirement in the near future
2023-02-21 03:30:05 <skarnet> yeah, but I meant, this kind of corporate culture usually rewards incompetent management and careerism
2023-02-21 03:30:37 <bl4ckb0ne> smh everybody stealing my goats idea
2023-02-21 03:31:56 <minimal> bl4ckb0ne: not me, goats don't interest me, I retired to work on FOSS lol
2023-02-21 03:32:55 <bl4ckb0ne> but goats 
2023-02-21 03:34:09 <valerius> chickens are less effort per gram of protein produced
2023-02-21 03:35:27 <skarnet> not as cute tho
2023-02-21 03:38:23 <minimal> PECTA rather than PETA? lol
2023-02-21 04:07:29 <bl4ckb0ne> cant milk chicken
2023-02-21 04:11:47 <skarnet> but it's harder to collect goat eggs
2023-02-21 06:12:14 <psykose> just came across a configure test
2023-02-21 06:12:19 <psykose> that failed to detect nodejs
2023-02-21 06:12:30 <psykose> because a random non-executable 0-byte `node` was in /usr/local/bin
2023-02-21 06:12:51 <psykose> literally running "node" gives you node, how do you fail that
2023-02-21 06:12:54 <psykose> sifg
2023-02-21 06:16:42 <ikke> are they manually checking some paths for a file called node?
2023-02-21 06:16:46 <ikke> I assume it was not executable
2023-02-21 06:40:18 <psykose> isn't indeed
2023-02-21 06:54:58 <skarnet> sounds just like your average configure script, nothing to see here
2023-02-21 07:05:41 <psykose> welcome 2 qmake
2023-02-21 07:22:23 <skarnet> is this the little new one with the unwashed hair and the ugly smirk?
2023-02-21 07:27:27 <psykose> worse than autotools
2023-02-21 07:27:29 <psykose> gg go next
2023-02-21 09:41:42 <pj> skarnet: free software is also full of people with their heads up their asses, so I'll gladly deny joining your club
2023-02-21 09:43:23 <skarnet> are you hanging out in the development channel of a free software distribution just to be a dick to people here, or?
2023-02-21 09:44:26 <pj> according to you, yes
2023-02-21 09:45:00 <skarnet> it's certainly what it feels like from this side, and I'm not particularly sensitive
2023-02-21 09:45:37 <skarnet> so whatever kind of weird beef you have with me, I suggest you drop it now
2023-02-21 09:47:17 <pj> so far hostility has been coming from you but sure, I'll "drop the beef"
2023-02-21 09:49:33 <psykose> no, it's you
2023-02-21 09:49:44 <psykose> most of what you say is random hostility
2023-02-21 09:50:57 <pj> you forgot to mention that I'm polish
2023-02-21 09:51:10 <pj> and that it invalidates all my arguments
2023-02-21 09:51:40 <skarnet> you certainly lack polish
2023-02-21 15:28:22 <bl4ckb0ne> psykose: stop touching ncurses *shakes fist in the air*
2023-02-21 15:28:56 <psykose> well it does sound like an actual bug :p
2023-02-21 15:29:03 <bl4ckb0ne> leave those alone
2023-02-21 15:29:09 <bl4ckb0ne> bug dont break my ci
2023-02-21 15:29:37 <psykose> you could also fix your ci by just upgrading on start
2023-02-21 15:29:47 <bl4ckb0ne> its easier to complain on irc
2023-02-21 15:30:00 <psykose> which is really just what you should do anyway, though of course the bug still exists
2023-02-21 15:31:01 <bl4ckb0ne> 
ACTION closes the CI tab

2023-02-21 15:31:24 <bl4ckb0ne> rest is green anyway, gonna wait for the daily image gen
2023-02-21 15:31:40 <psykose> alpine upgrades aren't really different from an arch -Syu, just there are more protections :-)
2023-02-21 15:31:43 <psykose> and now of course this bug
2023-02-21 16:01:56 <Nulo[m]> i'm making a pile of scripts that make a squashfs rootfs image. does it make sense to just skip mkinitfs and make it load my image directly? of course I would need to reimplement the important mkinitfs in my image
2023-02-21 16:17:16 <Nulo[m]> okay, so this isn't possible because squashfs is built as a module
2023-02-21 16:29:09 <Nulo[m]> would a patch to change the compression level of zstd be accepted in mkinitfs? or an option to do so? https://gitlab.alpinelinux.org/alpine/mkinitfs/-/blob/master/mkinitfs.in#L304
2023-02-21 16:29:47 <psykose> 19 is fine though
2023-02-21 16:29:47 <Nulo[m]> there's a big time difference but a small size one (image incoming):
2023-02-21 16:29:56 <psykose> it's missing a -T0 but aside from that it's whatever
2023-02-21 16:30:13 <psykose> you do it once, it could take 5 minutes :p
2023-02-21 16:30:22 <Nulo[m]> 
ACTION uploaded an image: (32KiB) < https://matrix.org/_matrix/media/v3/download/tchncs.de/322a130e69cabd2984c9345814c9c12a8baa6eac/Screenshot%20from%202023-02-21%2013-30-00.png >

2023-02-21 16:30:46 <Nulo[m]> i'm doing it constantly as I'm building a whole alpine rootfs again and again (reasons :P)
2023-02-21 16:31:08 <psykose> i would patch it to -T0 in your building fiesta
2023-02-21 16:31:16 <Nulo[m]> oooh -T0
2023-02-21 16:31:45 <psykose> will probably go to like 3-4s there (the size is a parallelism limitation of course)
2023-02-21 16:31:53 <psykose> test and find out
2023-02-21 16:32:17 <psykose> i did open an mr to just pass -T0 by default already, maybe one day it'll just be there
2023-02-21 16:32:19 <Nulo[m]> interestingly -T0 doesn't change anything for -3 which I'm already using 🤷
2023-02-21 16:32:29 <psykose> you don't want -3 for the initfs
2023-02-21 16:33:05 <Nulo[m]> why not? it only increases by about 200KB from the gzip -9 default
2023-02-21 16:33:32 <Nulo[m]> yes this is microoptimization i know
2023-02-21 16:33:37 <psykose> because the point is to make it small, the time is the quite far off secondary factor
2023-02-21 16:33:49 <psykose> it only seems slightly inconvenient now because you're sitting there watching it
2023-02-21 16:34:11 <psykose> in reality you aim for smallest and start at -19, and then go down a bit if you have more realistic time constraints
2023-02-21 16:34:19 <psykose> (they all have the same decompression requirements unless you go above 19)
2023-02-21 16:34:29 <Nulo[m]> does it stay in RAM a
2023-02-21 16:34:30 <Nulo[m]> after boot?
2023-02-21 16:34:41 <psykose> the initfs itself? no
2023-02-21 16:34:54 <psykose> but you are writing it somewhere, and it has to be decompressed, etc
2023-02-21 16:35:13 <psykose> now you could say inside this real of megabytes all of this is a microoptimisation, but
2023-02-21 16:35:17 <psykose> realm*
2023-02-21 16:36:16 <Nulo[m]> i think it's fine for my usecase? i'm writing a much bigger 200MB+ image anyway, and this saves like 20% of total build time and only increases size 200KB
2023-02-21 16:36:38 <psykose> the difference between -3 and -19 for a 200mb+ image would be quite huge
2023-02-21 16:37:03 <psykose> again, unless you are making some "create hundreds of images per minute" pipeline, the time doesn't matter :p
2023-02-21 16:37:07 <psykose> you just want a small image
2023-02-21 16:37:09 <Nulo[m]> i'm talking about the difference in initramfs, it's 200KB vs about half the time
2023-02-21 16:37:23 <Nulo[m]> and 20% off the total build time
2023-02-21 16:38:09 <Nulo[m]> eh this conversation is going in circles anyway :)
2023-02-21 16:38:11 <psykose> we've both repeated the same thing two times to eachother, so in the end you can pick what works for you
2023-02-21 16:38:15 <psykose> i'm not going to backseat it :p
2023-02-21 16:38:26 <skarnet> in usual situations, build time is machine time so it shouldn't be an important factor
2023-02-21 16:38:39 <skarnet> now, if you're going to be using it interactively, of course that changes the assessment
2023-02-21 16:38:41 <Nulo[m]> yeah exactly
2023-02-21 16:40:14 <Nulo[m]> as a reference, for the bigger image (it's built with mksquashfs which afaik doesn't have an option for -T0, and not sure if it uses multiple threads or not) -1 yields a 217M image in 1.3s and -19 yields a 184M image in 59s
2023-02-21 16:40:49 <psykose> mksquashfs defaults to all cpus afaik
2023-02-21 16:41:23 <psykose> that reference there depends on how you use it
2023-02-21 16:41:44 <psykose> from the general way (i) use images, that metric is "one gives a 217mb image and the other gives a 184 mb image"
2023-02-21 16:41:53 <psykose> so it is also faster to decompress for the latter and to send across the network
2023-02-21 16:42:12 <psykose> i don't even look at the time unless the latter is 30 minutes
2023-02-21 16:42:37 <Nulo[m]> yeah of course there's advantages in some use cases. but a much better use of my time would be optimizing the go programs inside the image with an absurd amount of unused dependencies :)
2023-02-21 16:43:02 <psykose> that is just a different and unrelated subject entirely
2023-02-21 16:44:05 <Nulo[m]> what i mean is that it would reduce the image size much more than tweaking compression parameters. but in my case image size is not a big problem so it's fine
2023-02-21 16:44:29 <Nulo[m]> overall i'm glad i can just patch mkinitfs as it's a shell script xd
2023-02-21 16:44:31 <psykose> sure, and that is another subject with other steps :)
2023-02-21 16:44:42 <psykose> you can think about.. both things, and do both
2023-02-21 16:46:28 <Nulo[m]> ofc. either way i'm glad i can build a useful alpine rootfs in <6s. i'm experimenting with building an immutable image mainly for having everything defined in code (i know there's other tools to do similar things)
2023-02-21 17:15:16 <nmeum> psykose: btw: the native compiler backend for ghc is only supported on aarch64-darwin. unfourtunately, it is presently not supported for aarch64-linux (looked into this again after having to wait forever for cabal to build on aarch64)
2023-02-21 17:17:11 <psykose> aha
2023-02-21 17:17:17 <psykose> how terribly inconvenient
2023-02-21 17:17:22 <psykose> thanks for tracking it down
2023-02-21 17:17:33 <nmeum> np
2023-02-21 21:47:59 <craftyguy> is there a good example of a package implementing option #3 here? https://wiki.alpinelinux.org/wiki/APKBUILD_examples:Go
2023-02-21 21:48:30 <craftyguy> re: vendoring dependencies in Go for abuild
2023-02-21 22:43:47 <ikke> I don't know any package where we do option #3
2023-02-21 22:43:59 <ikke> Note that that page is outdated 
2023-02-21 22:48:39 <craftyguy> oh ok. is there a new example somewhere?
2023-02-21 22:49:07 <ikke> Most projects will use go modules nowadays 
2023-02-21 22:49:33 <ikke> Which is reproducible regarding dependencies 
2023-02-21 22:50:34 <ikke> nfpm is an example 
2023-02-21 22:51:22 <ikke> craftyguy: what do you want to package? 
2023-02-21 22:53:40 <craftyguy> interesting, that seems to be using GoReleaser to basically do "option 3", from what I can tell?
2023-02-21 22:54:17 <craftyguy> oh there's a few packages in aports that build Go apps, that I maintain, and was wanting to try to reduce network usage at build time
2023-02-21 22:55:35 <craftyguy> these are apps that I also maintain upstream, so i can change how releases are done to help facilitate this. I am just not sure what the current "best way" to do that is. maybe it's using this GoReleaser thing
2023-02-21 22:58:25 <craftyguy> oh wow, I hope there's something better than GoReleaser... it does some other questionable things on this list here: https://goreleaser.com
2023-02-22 02:53:20 <EvTheFuture> @psykose: I spoke to early regarding Alpine 3.17 and Bluetooth.... Yes it worked after I upgraded from 3.16 to 3.17, however as soon as I pair a new device I can not connect to that device. However if I reboot and boot into Alpine 3.16 instead the newly paired device works. Have you tried pairing a new device after upgrade to 3.17?
2023-02-22 03:15:47 <elibrokeit> craftyguy: I feel like the real solution is probably to invent an alternative compiler that just uses the go language, but not the package management :p
2023-02-22 03:16:30 <craftyguy> heh. I think just vendoring dep source code in the project is "fine", i'll see how that works out
2023-02-22 05:38:31 <ikke> Maybe you are looking for go mod download
2023-02-22 05:38:35 <ikke> craftyguy: ^
2023-02-22 05:38:56 <ikke> You can download all dependencies in advance without needing to vendor all your dependencies
2023-02-22 08:02:37 <wej> Hi! I would like to submit a merge request for a new package that depends on another new package. What would be the right way to do this, so that the CI pipeline doesn't fail?
2023-02-22 09:31:13 <ikke> wej: you can add them both in the same MR 
2023-02-22 09:34:32 <wej> ikke: alright, thanks :)
2023-02-22 10:06:02 <maribu> Is anyone here using claws-mail as email client? Is it just me, or did an update a few days ago break the spell checking?
2023-02-22 12:20:38 <wej> ikke: should i do it in a single commit or via two commits, one for each package in that one MR?
2023-02-22 12:20:52 <ikke> One commit for each package 
2023-02-22 12:22:07 <wej> okay
2023-02-22 13:54:33 <psykose> EvTheFuture: always worked for me
2023-02-22 13:55:09 <psykose> maribu: i don't use it but i just tried and it works, maybe you don't have the aspell dictionary or something other than `en` broke
2023-02-22 14:02:26 <ncopa> psykose: thanks for fixing libmagick-static -> libmagic
2023-02-22 14:04:08 <psykose> :)
2023-02-22 14:37:02 <wej> ikke: i created the merge request, but unfortunately the pipeline failed because it couldn't find the dependency. i probably did something wrong. any ideas? https://gitlab.alpinelinux.org/alpine/aports/-/merge_requests/44459
2023-02-22 14:43:53 <ikke> wej: the getdns package failed, so the package is still missing 
2023-02-22 14:46:42 <wej> oh, i'll check that, thanks. i'll also go over psykose's comments and include those changes
2023-02-22 14:48:10 <wej> ah, bummer, one of the tests failed. those tests talk to the root name server, perhaps some issue connecting
2023-02-22 14:58:39 <maribu> OK, claws-mail requires enchant2 to be installed for spellchecking to work
2023-02-22 15:02:41 <psykose> i'd guess the /usr/lib/enchant-2/ plugins are loaded by the lib
2023-02-22 15:02:47 <psykose> since there's one for aspell, nuspell, etc
2023-02-22 15:04:04 <maribu> I think it is actually calling the command line tools, as just installilng enchant2 without rebuilding claws-mail fixed the issue. !44461
2023-02-22 15:04:15 <psykose> no, it's not
2023-02-22 15:04:35 <psykose> rebuilding doesn't matter either
2023-02-22 15:04:44 <psykose> and those plugins are in enchant2 so that's why it fixes it
2023-02-22 15:08:47 <psykose> that change doesn't fix it either, because if you don't have a language dictionary installed it will give you the same error
2023-02-22 15:10:07 <psykose> claws-mail just inits enchant2 via the library for whatever current language
2023-02-22 15:10:22 <psykose> enchant2 loads one of its backends, and then tries to load a language for the backend
2023-02-22 15:10:44 <psykose> if anything is missing you will get that error, so the only way to fix this is to install literally every language in the universe and every provider
2023-02-22 15:11:02 <psykose> i.e. there is nothing to fix
2023-02-22 15:16:30 <psykose> i split the plugins better so at least if you have `aspell` installed now it will also install the enchant plugin by default (no changes), but installing the actual languages or whatever is still manual
2023-02-22 15:43:58 <wej> about those failed tests with the testing/getdns package, it appears that the dns lookups time out ("All queries for the name timed out"). does the ci host have internet connectivity while running the tests? i don't really like that those tests require internet connectivity in the first place. not sure what to do..
2023-02-22 15:44:28 <ikke> They do 
2023-02-22 15:44:50 <wej> hm
2023-02-22 15:45:07 <wej> both ipv4 and ipv6?
2023-02-22 15:47:09 <ikke> They should, but I have to verify 
2023-02-22 15:47:26 <ikke> Some things got moved 
2023-02-22 15:48:04 <ikke> Though, s390x and ppc64le might not have ipv6
2023-02-22 15:50:29 <minimal> doesn't "net" need to be specified in APKBUILD's option list if network connectivity required?
2023-02-22 15:51:08 <psykose> no, that's just for rootbld
2023-02-23 09:49:43 <kreyren> Piraty, https://gitlab.alpinelinux.org/alpine/aports/-/commit/f1b1fcf8300a733f01a86355b9853ec360670493 did this fixed the rendering for you?
2023-02-23 09:50:02 <kreyren> 
ACTION is working on archlinux port for the OLIMEX Teres-A64 and is checking if they have the fix

2023-02-23 10:42:14 <j400[m]> Hi! Does launching `spot` work for someone on edge? It didn't work for me, even after building it locally. I had to update Cargo.lock for it to work locally.
2023-02-23 10:46:28 <j400[m]>  * Hi! Does launching `spot` work for someone on edge? It didn't work for me, even after building it locally.
2023-02-23 10:50:05 <j400[m]>  * Hi! Does launching `spot` work for someone on edge? It didn't work for me, even after building it locally. (Updating cargo.lock to pull the latest 0.4 gtk4rs version doesn't work either, seems like something needs to be adjusted)
2023-02-23 10:50:42 <j400[m]>  * Hi! Does launching `spot` work for someone on edge? It didn't work for me, even after building it locally. (Updating cargo.lock to pull the latest 0.4 gtk4rs version doesn't work either, seems like something needs to be adjusted. I'm just using the latest commit from git now.)
2023-02-23 10:56:30 <pj> how does it not work
2023-02-23 11:00:56 <Piraty> kreyren: i guess it did. i'm still using my own patches mesa from https://gitlab.alpinelinux.org/alpine/aports/-/merge_requests/44204 though, didn't bother to downgrade and upgrade to alpine's package (as psykose cherry-picked the backported patch from upstream). skimmed quickly, they should be semantically equivalent
2023-02-23 11:01:15 <Piraty> didn't notice the segfault ever sinc
2023-02-23 11:17:11 <kreyren> Piraty, thanku for info
2023-02-23 11:17:12 <kreyren> <3
2023-02-23 11:51:40 <j400[m]> <pj> "how does it not work" <- It asserts on launch on my desktop and pmOS-device.
2023-02-23 12:02:38 <graywolf> Hi, I got an email regarding `Alpine aport bazel5 has been flagged out of date'. The message is `abump bazel5-6.0.0`, is it expected that I should upgrade bazel5 package to contain bazel 6.0.0?
2023-02-23 12:04:53 <quintasan> I come here from https://github.com/utdemir/ghc-musl/issues/15 which suggest there's a problem with Alpine's postgresql package. To be very precise my problem is that `pg_strcasecmp` is missing from /usr/lib/libpq.a provided by Alpine as seen in https://github.com/utdemir/ghc-musl/issues/15#issuecomment-1441463722. Any ideas on how to fix that? I looked at APKBUILD for postgresql15 but it doesn't look like it's stripping anything
2023-02-23 12:13:00 <sam_> j400[m]: asserts how?
2023-02-23 12:13:03 <sam_> full output?
2023-02-23 12:22:42 <j400[m]> thread 'main' panicked at 'assertion failed: !self.ptr.is_null()', /home/buildozer/.cargo/registry/src/github.com-1ecc6299db9ec823/gtk4-0.4.6/src/subclass/widget.rs:1310:13
2023-02-23 12:22:43 <j400[m]> note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace
2023-02-23 12:22:43 <j400[m]> [1]    4270 abort      spot
2023-02-23 12:25:15 <j400[m]> And for some reason we are missing community/libappindicator in edge but it's in 3.17.
2023-02-23 12:25:15 <j400[m]> [v3.17](https://pkgs.alpinelinux.org/packages?name=libappindicator*&branch=v3.17&repo=&arch=&maintainer=)
2023-02-23 12:25:15 <j400[m]> [edge](https://pkgs.alpinelinux.org/packages?name=libappindicator*&branch=edge&repo=&arch=&maintainer=)
2023-02-23 12:32:27 <j400[m]>  * ~~And for some reason we are missing community/libappindicator in edge but it's in 3.17.... (full message at <https://matrix.org/_matrix/media/v3/download/matrix.org/wSoXfIdxCNwoXrDNgKxBQzmS>)
2023-02-23 12:36:53 <ikke> graywolf: The idea is that there is some sort of bazel package that provides 6.x
2023-02-23 17:05:15 <oldes> Hi there... I have a question.. I am a maintainer of Rebol language sources: https://github.com/Oldes/Rebol3 and I would like to create a package for it.
2023-02-23 17:05:57 <oldes> Can I use pre compiled binaries? 
2023-02-23 17:06:54 <elly> not generally
2023-02-23 17:07:54 <elly> your APKBUILD specifies how to do a source build, and some automation executes that build to make binary packages for us
2023-02-23 17:08:11 <oldes> So I must setup the APKBUILD to make all arch versions?
2023-02-23 17:08:58 <elly> you can generally say like... arch="list of arches" or even arch="all"
2023-02-23 17:09:06 <oldes> What if I need a binary tool (other then GCC) to make the binaries?
2023-02-23 17:09:15 <elly> oh this is a compiler that need a bootstrap? hmmm
2023-02-23 17:09:39 <elly> I do not know the answer to this, better wait for a real expert :)
2023-02-23 17:10:54 <oldes> The thing is, that to build Rebol, one needs already compiled version of Rebol to generate some of its own sources... like may be visible here: https://github.com/Oldes/Rebol3/actions/runs/4240269694/jobs/7369124210
2023-02-23 17:11:22 <elly> yeah
2023-02-23 17:11:51 <elly> I went to look at how the ghc package (community/ghc) does it, and the answer is that it depends on a special bootstrap package called ghc-bootstrap
2023-02-23 17:12:49 <elly> and it does... some complicated stuff
2023-02-23 17:12:51 <ikke> Is there a rebol binary that runs with musl?
2023-02-23 17:13:09 <ikke> The latest standard is to define an explicit stage0 package
2023-02-23 17:13:26 <ikke> to make it more transparent how the package is being bootstrapped from scratch
2023-02-23 17:13:35 <elly> I see arm64-musl binaries, but for x86-64 there is only a glibc binary: https://github.com/Oldes/Rebol3/releases
2023-02-23 17:15:07 <psykose> just rewrite it so it's not self hosting
2023-02-23 17:15:10 <psykose> that would fix all the issues
2023-02-23 17:15:11 <psykose> yw
2023-02-23 17:15:25 <ikke> lol
2023-02-23 17:15:40 <oldes> @elly https://github.com/Oldes/Rebol3/releases/download/3.10.3/rebol3-bulk-linux-x64-musl.gz
2023-02-23 17:16:10 <elly> oh, tricked by github's UI
2023-02-23 17:16:44 <oldes> here are Docker images for Alpine: https://hub.docker.com/repository/docker/rebol/alpine/general
2023-02-23 17:17:36 <oldes> Actually just one so far under this name
2023-02-23 17:18:04 <oldes> But that is not an issue... I am have it also for other archs
2023-02-23 17:22:15 <elly> psykose: it kinda sucks either way because as a compiler author you either have to 1) keep your compiler written in some other language, 2) have two compilers, or 3) require prebuilt binaries a lot
2023-02-23 17:22:55 <oldes> Here is how I build binaries for Alpine so far: https://github.com/Oldes/Rebol-Docker/actions/runs/4243686197
2023-02-23 18:07:21 <psykose> elly: yeah, certainly a game of tradeoffs
2023-02-23 18:07:53 <psykose> i like the zig approach kinda, which is 2)
2023-02-23 18:08:04 <psykose> it's self-bootstrapping but ships the c++ bootstrap with itself
2023-02-23 18:08:57 <psykose> i feel like 3) is the easiest for the developers ofc but much more miserable for everyone else
2023-02-23 18:10:13 <elly> I like the (2) approach as well, especially because it lets you test the compilers against each other
2023-02-23 18:10:18 <elly> on the downside, you then maintain two compilers :(
2023-02-23 18:11:26 <psykose> well
2023-02-23 18:11:37 <psykose> compiler engineers are also compiler nerds
2023-02-23 18:11:42 <psykose> so they obv love maintaining 2 compilers
2023-02-23 18:11:43 <psykose> easy sell
2023-02-23 18:11:47 <c7s> didn't zig do some wasm bootstrapping thing recently
2023-02-23 18:11:55 <psykose> even Go does that, since they maintain gccgo /s
2023-02-23 18:12:21 <psykose> c7s: you mean https://ziglang.org/news/goodbye-cpp/ ?
2023-02-23 18:12:28 <psykose> ah
2023-02-23 18:12:32 <psykose> i guess they did change it
2023-02-23 18:18:20 <elly> that's slick, I like that design
2023-02-23 18:18:43 <elly> so they still ship "a binary" but that binary is actually for a small VM, and they have a portable implementation of just the VM
2023-02-23 18:19:16 <psykose> yeah, i'm a lot more partial to that kind of design tbh
2023-02-23 18:24:47 <elly> oh I missed that they are actually doing wasm2c in here
2023-02-23 18:24:52 <elly> instead of running a wasm VM
2023-02-23 18:30:07 <elly> so it's: ship a wasm binary, use system cc to compile wasm2c, translate wasm binary to c, use system cc to compile that c, run the resulting zig compiler to compile the zig compiler
2023-02-23 18:44:30 <skarnet> self-hosting is such a masturbatory exercise, it makes things difficult for everyone for no practical reason
2023-02-23 18:46:27 <skarnet> "let's force every distro to either use binaries they'll have to trust or go through a cpu-intensive and likely super-broken bootstrapping process, so we can pretend our awesome new language doesn't depend on anything else! (and obviously hasn't learned anything from its predecessors, either)"
2023-02-23 19:12:17 <PureTryOut> psykose: you made https://gitlab.alpinelinux.org/alpine/aports/-/commit/cd45e68e23fedc63352a51294d6fef0c95261781 which now causes CI to fail if something pulls in pipewire. Could you take a look at that? Also please make merge requests when you're touching a package that is not yours, I had no idea this change was made to the package I'm listed as maintainer for
2023-02-23 19:13:35 <psykose> indeed, was missing an exit 0
2023-02-23 19:13:40 <psykose> fixed
2023-02-23 19:17:15 <PureTryOut> Thanks!
2023-02-23 19:19:42 <skarnet> !44478 when the pipeline passes. Nothing urgent though - it's bugfixes that Alpine never hits (because they involve glibc or locales)
2023-02-23 19:20:58 <psykose> you downgraded execline in that instead of making it .1 by accident
2023-02-23 19:21:13 <skarnet> wait what
2023-02-23 19:21:15 <skarnet> checking
2023-02-23 19:21:21 <psykose> 2.9.2.0 -> 2.9.1.0
2023-02-23 19:21:25 <psykose> instead of 2.9.2.1
2023-02-23 19:21:40 <skarnet> yeesh
2023-02-23 19:21:53 <skarnet> thanks for double-checking, fixing
2023-02-23 19:23:13 <skarnet> abump is practical but if you typo your command line, there's no second chance
2023-02-23 19:23:30 <skarnet> pushed another abump on top of it, nobody saw anything
2023-02-23 19:23:40 <psykose> i did see it
2023-02-23 19:23:43 <psykose> but i squashed it for you
2023-02-23 19:24:13 <skarnet> hush, no you didn't see anything, and on a completely unrelated note I'll send you cookies
2023-02-23 19:25:03 <skarnet> ... and you already merged before I could click the rebase button XD thanks!
2023-02-23 19:25:12 <psykose> you never have to click it yourself :p
2023-02-23 19:25:40 <skarnet> when gitlab says "merge blocked" I become paranoid
2023-02-23 19:25:50 <psykose> yes yes
2023-02-23 19:26:09 <psykose> personally i always fix conflicts myself though
2023-02-23 19:26:11 <psykose> amongst other things
2023-02-23 19:27:03 <skarnet> you do, but there was a time, not long ago, when if I wanted something merged in less than 6 months, I had to make it real smooth and real easy and turnkey
2023-02-23 19:27:15 <skarnet> and even then that wasn't a guarantee
2023-02-23 20:14:39 <psykose> Newbyte: feel free to run the usual on !43183
2023-02-23 20:14:44 <psykose> and see if everything still works before .1
2023-02-23 20:32:52 <Nulo[m]> is there hooks into initramfs-init? so ways to change behaviour without having to fork the entire 800 line script
2023-02-23 20:33:07 <Nulo[m]> i want to mount a squashfs inside an ext4 partition as root and not sure how to approach that, it seems editing initramfs-init is the only way
2023-02-23 20:36:34 <psykose> pretty sure no
2023-02-23 21:31:19 <Nulo[m]> is there a way to live change the mount in /? like, switching from a squashfs image to a different one
2023-02-23 21:36:42 <psykose> you mean aside from pivot_root?
2023-02-23 21:38:03 <psykose> or switch_root
2023-02-23 21:38:32 <psykose> initrd usually uses the latter i think
2023-02-23 21:39:21 <Nulo[m]> what I intended is for it to just change the current root filesystem, like  a file modification if it were a r/w fs
2023-02-23 21:40:24 <Nulo[m]> but it seems it isn't possible (I guess I would have to do a chroot and reexec init?)
2023-02-23 21:40:25 <Nulo[m]> is this the right channel to ask this sort of stuff?
2023-02-23 21:41:29 <psykose> well, what you're describind does just sound like switch_root
2023-02-23 21:41:43 <psykose> i'm not sure how different it could be
2023-02-23 21:42:28 <Nulo[m]> what I wanted to do is to just mount a new squashfs rootfs and restart affected services but it doesn't seem very possible
2023-02-23 21:43:14 <psykose> well.. they're all affected
2023-02-23 21:43:34 <psykose> so.. they all have to be restarted
2023-02-23 21:43:49 <psykose> which is why you reexec init in the new root
2023-02-23 21:43:54 <psykose> it's not like it's slow or whatever
2023-02-23 21:44:44 <Nulo[m]> but it will need to kill all services, even if i were to use something like inetd it would be killed
2023-02-23 21:45:40 <Nulo[m]> they are affected in the sense that everything needs to be switched into the rootfs, but some services might not have any changes
2023-02-23 21:46:31 <Newbyte> psykose: sure, thanks
2023-02-23 21:47:00 <psykose> they're all affected by some fundamental property of how the root works pretty sure
2023-02-23 21:47:26 <psykose> perhaps skarnet knows why you can't just change the root with a running inetd :p
2023-02-23 21:47:26 <invoked> i didn't read all of this, but isn't this something you'd use ignition for
2023-02-23 21:47:36 <psykose> what is ignition
2023-02-23 21:48:04 <invoked> finding a link, one sec
2023-02-23 21:48:18 <psykose> the name vaguely resembles some cloud garbage
2023-02-23 21:48:22 <invoked> https://coreos.github.io/ignition/
2023-02-23 21:48:24 <psykose> hah
2023-02-23 21:48:32 <Nulo[m]> switch_root says PID must be 1 which checks out
2023-02-23 21:48:33 <invoked> it is, but you could peel it away from cloud-init i think
2023-02-23 21:49:16 <psykose> yeah i guess i see what is meant then
2023-02-23 21:49:34 <Nulo[m]> i don't think this can help me? it seems to be a config util for initramfs
2023-02-23 21:49:35 <skarnet> yeah, who would have thought changing the root of the filesystem, thus the whole fs infrastructure, would be a heavy operation forcing you to restart anything that depends on the filesystem hierarchy!
2023-02-23 21:50:01 <skarnet> that's why when people need this, they do their stuff in an initramfs, then change rootfs, then exec into init
2023-02-23 21:50:16 <Nulo[m]> what I imagined is that I would be able to remount it into a new fs but I guess not lol
2023-02-23 21:50:17 <skarnet> and once init is running, well, there's no more changing the rootfs
2023-02-23 21:50:25 <skarnet> well you *could*, technically
2023-02-23 21:50:25 <Nulo[m]> there goes my PoC i worked on for so many weeks
2023-02-23 21:50:28 <psykose> unless init itself wants to do it
2023-02-23 21:50:32 <psykose> Nulo[m]: what /do/ you want to do
2023-02-23 21:50:41 <skarnet> but since every process is basically relying on the filesystem, you'd have to restart every process
2023-02-23 21:50:53 <skarnet> so, rebooting is just as simple
2023-02-23 21:51:30 <Nulo[m]> my intention is to build a pile of scripts (or something nicer) that builds a rootfs that gets used as ro when running. here i am trying to add a mechanism to update (change the running image into another)
2023-02-23 21:51:40 <Nulo[m]> i may be able to achieve something similar with btrfs snapshots though
2023-02-23 21:51:45 <skarnet> it's extremely rare that you need to change the rootfs though... it's fundational to a system and apart from loading a kernel module or something, which is done early on, you generally don't need that
2023-02-23 21:52:18 <psykose> that just sounds like diskless, and you want to make an update mechanism?
2023-02-23 21:52:33 <psykose> you won't be able to update without a reboot of sorts no
2023-02-23 21:52:48 <psykose> i guess kexec would be faster than a full one but that's another can of worms
2023-02-23 21:53:12 <skarnet> there's a reason why firmware updates require a reboot on every device you can possibly imagine
2023-02-23 21:53:14 <Nulo[m]> it is somewhat like diskless, but i want to be able to make it completely defined in code
2023-02-23 21:53:16 <Nulo[m]> yeah i was thinking this switch_root on an already running system sounded a lot like kexec
2023-02-23 21:53:25 <skarnet> firmware updates are exactly that: switching to a new ro rootfs
2023-02-23 21:53:35 <psykose> haha nope :)
2023-02-23 21:56:24 <Nulo[m]> well this sucks! i made this thing over some time assuming this would just work lol
2023-02-23 21:56:25 <Nulo[m]> i'll figure something out with btrfs snapshots probably
2023-02-23 21:58:09 <Nulo[m]> these people do something similar: https://github.com/system-transparency/system-transparency/ but the approach to updating is just rebooting lol
2023-02-23 21:58:36 <Nulo[m]> but the use case for them is things like VPN servers, small amount of software that doesn't change much
2023-02-23 22:00:29 <psykose> well yeah
2023-02-23 22:00:38 <psykose> just rebooting lol is a pretty good strategy
2023-02-23 22:01:46 <Nulo[m]> it's not when it's too often and it's a single node without redundancy
2023-02-23 22:03:26 <invoked> even rebooting doesn't work on some systems eg where usb doesn't full reinit and usb devices are stuck in their previous state on boot
2023-02-23 22:04:53 <invoked> true for some gpus as well
2023-02-23 22:11:34 <Nulo[m]> ugh it isn't even possible with btrfs subvols even though subvols are just mount options https://unix.stackexchange.com/a/55463
2023-02-23 22:17:38 <psykose> for the same reason as without btrfs
2023-02-23 22:17:47 <psykose> you cannot change the root, that is changing it
2023-02-23 22:57:00 <psykose> bl4ckb0ne: hmm
2023-02-23 22:57:13 <psykose> i'm doing some xwayland 23 testing, and it seems to be broken with vulkan wlr in sway
2023-02-23 22:57:24 <bl4ckb0ne> how
2023-02-23 22:57:33 <psykose> any app,
2023-02-23 22:57:39 <bl4ckb0ne> yes
2023-02-23 22:57:41 <psykose> open wayland -> works
2023-02-23 22:57:51 <psykose> open x11 with xwayland 22 opengl -> works
2023-02-23 22:58:03 <psykose> open x11 with xwayland 22 vulkan -> works
2023-02-23 22:58:12 <psykose> open x11 with xwayland 23 opengl -> works
2023-02-23 22:58:20 <psykose> open x11 with xwayland 23 vulkan -> hangs forever
2023-02-23 22:58:25 <psykose> no output or anything
2023-02-23 22:58:34 <psykose> just GDK_BACKEND=x11 some-gtk-app and it sits there
2023-02-23 22:58:43 <bl4ckb0ne> id suggest to uninstall xwayland and forget X ever existed
2023-02-23 22:58:51 <psykose> yeah but i mean it sounds like something to debuug
2023-02-23 22:58:53 <psykose> idk where to start
2023-02-23 22:59:25 <vyivel> is it https://gitlab.freedesktop.org/wlroots/wlroots/-/issues/3569?
2023-02-23 22:59:39 <psykose> that looks very likely
2023-02-23 22:59:46 <psykose> since new xwayland is dmabuf v4
2023-02-23 23:00:23 <psykose> let me try a meme
2023-02-23 23:00:31 <psykose> where i revert only the v4 support in xwayland and retest
2023-02-23 23:01:25 <psykose> vyivel: do you have like some of these words on ping or something, how did you see this smh
2023-02-23 23:01:52 <vyivel> randomly checked what this chan's up to
2023-02-23 23:02:07 <psykose> mfw not bed
2023-02-23 23:02:15 <vyivel> i am in bed actually
2023-02-23 23:02:23 <psykose> yeah but
2023-02-23 23:02:24 <psykose> sleepge
2023-02-23 23:02:31 <vyivel> right
2023-02-23 23:02:32 <vyivel> gn
2023-02-23 23:06:59 <psykose> sleepy well
2023-02-23 23:07:00 <psykose> and yep
2023-02-23 23:07:06 <psykose> reverting that series fixes it, so it's that issue
2023-02-23 23:19:47 <bl4ckb0ne> i live in an xless cloud and ive never been happier
2023-02-23 23:26:09 <skarnet> I live on Earth and it's not too bad either
2023-02-23 23:26:22 <bl4ckb0ne> but theres X11
2023-02-23 23:26:29 <psykose> steam without x11 omegalol
2023-02-23 23:27:33 <bl4ckb0ne> thanks sdl2
2023-02-23 23:27:41 <skarnet> life sounds painful on the client side
2023-02-23 23:32:25 <psykose> sdl2 works great with wayland
2023-02-23 23:32:42 <Newbyte> can't tell if you're sarcastic
2023-02-23 23:32:53 <psykose> you can even preload host sdl2 with a bunch of steam games to add in the wayland support they didn't compile with and it works
2023-02-23 23:33:07 <psykose> not being sarcastic
2023-02-23 23:33:12 <psykose> works better than gtk3 wayland
2023-02-23 23:33:16 <Newbyte> but there's also all those that don't work
2023-02-23 23:33:17 <Newbyte> not SDL 2's fault of course
2023-02-23 23:33:25 <psykose> that case is a hack
2023-02-23 23:33:28 <Newbyte> excuse me
2023-02-23 23:33:33 <psykose> so anything working at all is pretty good
2023-02-23 23:33:57 <psykose> hell, it works better than gnome wayland
2023-02-23 23:54:44 <bl4ckb0ne> sdl works great without xcb libs installed
2023-02-24 01:47:18 <Nulo[m]> update regarding my Experiment: it seems i have to look into ostree
2023-02-24 03:06:18 <Nulo[m]> update: i got it working.. by symlinking /usr, /etc and some other directories, and then just replacing those symlinks
2023-02-24 12:52:59 <oldes> Hi.. I am trying to cross-compile aarch64 on x86_64 Alpine... where I can get the C library for the target architecture? Because now I have this error:
2023-02-24 12:53:17 <oldes>  /usr/lib/gcc/aarch64-none-elf/12.2.0/../../../../aarch64-none-elf/bin/ld: cannot find crt0.o: No such file or directory
2023-02-24 12:53:27 <oldes>  /usr/lib/gcc/aarch64-none-elf/12.2.0/../../../../aarch64-none-elf/bin/ld: cannot find -lc: No such file or directory
2023-02-24 12:54:57 <psykose> you have to make a sysroot yourself
2023-02-24 12:55:48 <psykose> i.e. there is 'no' way to 'do it' in some recommended way
2023-02-24 12:55:53 <psykose> it's just standard cross compiling
2023-02-24 12:56:25 <oldes> Is there any useful tutorial?
2023-02-24 12:56:40 <psykose> not that i know of
2023-02-24 12:57:29 <oldes> Hm. And how are made the apkbuilds for foreign architectures?
2023-02-24 12:57:39 <psykose> they're not
2023-02-24 12:58:29 <oldes> So when I do the apkbuild, I test just the one of the host`s architecture?
2023-02-24 12:59:33 <psykose> i don't know what you mean
2023-02-24 12:59:36 <ikke> Our CI builds for all our architectures 
2023-02-24 12:59:53 <ikke> And the APKs are provided 
2023-02-24 13:01:31 <oldes> I mean that I want to setup a new package for multiple archs. Now when I run `apkbuild -r`, only the host's arch is used
2023-02-24 13:02:06 <oldes> So the other architectures will be done in CI?
2023-02-24 13:02:18 <psykose> all of them will be
2023-02-24 13:02:23 <psykose> it doesn't matter what you build yourself
2023-02-24 13:02:24 <ikke> We don't provide means for cross compilation 
2023-02-24 13:03:36 <oldes> Ok... so I will just my local package works for the current architecture and than hope that the other will pass the tests on CI, right?
2023-02-24 13:03:56 <ikke> Yes
2023-02-24 13:04:07 <ikke> That's what most people will do 
2023-02-24 13:05:31 <oldes> So on CI there will be different values for $CARCH, right?
2023-02-24 13:06:27 <psykose> yes
2023-02-24 14:20:30 <Piraty> i assume that some foreign sysroot magic could easily be used with / around abuild tooling
2023-02-24 14:20:54 <Piraty> like xbps-src does
2023-02-24 14:21:35 <psykose> not really
2023-02-24 18:07:24 <oldes> When making a package using the apkbuild, is it ok to do in the package() section something like:
2023-02-24 18:07:25 <oldes> mv "Rebol3-$pkgver"/build/rebol3-bulk-linux-"$_reb_arch" "$pkgdir"/usr/bin/rebol 
2023-02-24 18:07:37 <oldes> ?
2023-02-24 18:09:06 <ikke> oldes: you mean manually moving files in $pkgdir?
2023-02-24 18:09:19 <oldes> I am not using any makefile
2023-02-24 18:09:24 <ikke> sure, that is no problem
2023-02-24 18:10:23 <oldes> And do I understand it well, that what is in the `$pkgdir`, that would be installed using the `apk add`, right?
2023-02-24 18:10:35 <ikke> yes
2023-02-24 18:11:27 <oldes> I should probably name the resulting binary `rebol3`, although it is very unlikely that someone would add good old rebol2
2023-02-24 18:18:51 <ptrc> oldes: usually when installing executables by hand, `install -Dm755 path "$pkgdir"/usr/bin/name` is used; this ensures correct permissions and that $pkgdir/usr/bin is created
2023-02-24 18:21:33 <oldes> Ah... good to know, thanks.
2023-02-24 18:22:52 <oldes> One more question... if I am able to produce multiple versions, which differs just with included functionalities.. like rebol-base, rebol-core, rebol-bulk... what would be the best strategy?
2023-02-24 18:23:49 <psykose> just don't
2023-02-24 18:24:02 <oldes> So I should include the most complete?
2023-02-24 18:55:38 <skarnet> include what users will expect
2023-02-24 20:11:11 <oldes> Is there any real difference between Alpine (docker) on linux/arm/v6 and linux/arm/v7? It looks that both are identified by `armv7l` using `uname -m` and having the same musl libc (armhf)
2023-02-24 20:11:44 <ikke> oldes: armv6 with hardfloat vs armv6
2023-02-24 20:12:06 <psykose> isn't v7.. v7
2023-02-24 20:14:04 <oldes> Is there any way how to find out, if my binary is v7 or the other v7?
2023-02-24 20:14:32 <ikke> v6 vs v7
2023-02-24 20:14:39 <ikke> they are 2 different architectures
2023-02-24 20:15:37 <oldes> But my docker images seems to have just one:/
2023-02-24 20:18:05 <psykose> what is /etc/apk/arch in the v7 one
2023-02-24 20:20:30 <psykose> (that is the only output that matters)
2023-02-24 20:21:25 <oldes> Ah... you are right.. they differ in this location.
2023-02-24 20:22:45 <ikke> apk --print-arch
2023-02-24 20:22:47 <ikke> (same)
2023-02-24 20:22:54 <psykose> the musl loader is named `armhf` for both v6+hf and v7+hf
2023-02-24 20:23:02 <psykose> mostly the arm names are a mess, as everyone knew already
2023-02-24 20:25:03 <oldes> is it better to use armv6 or armhf when naming my builds?
2023-02-24 20:41:58 <ikke> oldes: if you want to match alpine, you'd use armhf
2023-02-24 20:42:08 <ikke> but armhf means something different in debian
2023-02-24 20:42:15 <ikke> (which uses it for armv7)
2023-02-24 20:42:51 <psykose> it is better to use armv6 and at least specify if it's hardfloat
2023-02-24 20:47:19 <skarnet> it's better to throw arm into the trash can
2023-02-24 20:47:30 <ikke> with what?
2023-02-24 20:47:55 <skarnet> an accessibility aid if you have to
2023-02-24 20:47:56 <psykose> s390 and the ppc arches
2023-02-24 21:06:05 <clandmeter> we are going to turn of our arm builders and dev containers
2023-02-24 21:06:24 <clandmeter> and try to migrate to another machine
2023-02-24 22:08:34 <clandmeter> the builders are online again, but they dont announce its activity on build.a.o. 
2023-02-25 13:50:41 <elly> hm, I wonder if I can get the "Upgrade" UI chip thing to show up in alpine chromium, if apk has overwritten it on disk with a new binary
2023-02-25 13:51:12 <elly> and/or the "About Chromium" page to show the "Relaunch to update" button...
2023-02-25 13:53:45 <ikke> What is required for that?
2023-02-25 14:06:15 <elly> well, currently all that stuff tries to talk to Omaha (the Google updater) and if it's not present it does nothing
2023-02-25 14:06:30 <elly> but I wonder if I can give it a fallback path where it just watches its own binary for replacement on disk
2023-02-25 14:10:26 <elly> hmm, it doesn't look like it would be THAT bad to do
2023-02-25 14:10:33 <elly> next week maybe
2023-02-25 14:19:19 <JohannesJ[m]> do package names differ much from other distro's? or can i export a packagelist on arch and use that on alpine?
2023-02-25 14:21:06 <ikke> There are differences
2023-02-25 14:21:15 <JohannesJ[m]> oh hmz
2023-02-25 14:22:57 <ikke> A project like repology tries to normalize them
2023-02-25 14:25:00 <JohannesJ[m]> i'm still strugling with networkmanager and KDE on Alpine :) so i got this briljant idea of looking up the packages in manjaro (where it works) and see if i miss any package
2023-02-25 14:25:20 <JohannesJ[m]> next step is copying the working config :|
2023-02-25 14:33:27 <psykose> elly: it works by default
2023-02-25 14:33:42 <psykose> if you replace the binary it changes something somewhere and prompts to restart
2023-02-25 14:34:08 <psykose> hm or does it
2023-02-25 14:34:18 <psykose> i swear i've seen that
2023-02-25 15:11:34 <ikke> I know firefox has a mechanism like that
2023-02-25 15:51:04 <elly> psykose: it doesn't work for me
2023-02-25 15:54:40 <elly> it is possible it used to and broke
2023-02-25 15:59:56 <psykose> yeah, i swear i remembered it did
2023-02-25 16:00:02 <psykose> even on 107 first release
2023-02-25 16:00:12 <psykose> when you apk upgraded it had some greenish light in the ui
2023-02-25 16:00:49 <elly> I'll have a look next week and see if I can fix it
2023-02-25 16:01:07 <elly> the true polished experience would be for chromium to ask apk if there's a newer version available
2023-02-25 16:10:12 <psykose> maybe, but i don't like those
2023-02-25 16:10:18 <psykose> it's too distro specific and the like
2023-02-25 16:10:27 <psykose> just "binary replaced on disk, please restart" is good enough
2023-02-25 16:10:49 <psykose> (in a more friendly output way of course)
2023-02-25 18:48:19 <umk3[m]> psykose: installed udevil and saw that "devmon" is calling "busybox's ps" rather than "ps" from procps package
2023-02-25 18:48:32 <umk3[m]> maybe procps should be a dependency for devmon? no idea
2023-02-25 18:51:33 <psykose> and bash
2023-02-25 18:51:40 <psykose> but looking at it it makes more sense to just remove it
2023-02-25 18:52:05 <umk3[m]> is ok
2023-02-25 18:52:17 <umk3[m]> i just tried it out of curiosity
2023-02-25 18:53:37 <psykose> personally i just type mount
2023-02-25 18:54:34 <umk3[m]> now personally I feel bad for anyone using udevil lmao
2023-02-25 18:55:22 <psykose> less suid = better xd
2023-02-25 18:55:31 <umk3[m]> good
2023-02-25 18:55:36 <umk3[m]> thank you psykose
2023-02-25 19:38:14 <skarnet> it's called udevil and it's not even related to udev
2023-02-25 19:38:29 <skarnet> naming is hard but there are limits
2023-02-25 19:38:48 <skarnet> I wouldn't trust someone who's that bad at naming :P
2023-02-25 19:39:46 <orbea> its a udisks replacement and doesn't depend on polkit
2023-02-25 19:56:31 <skarnet> and udevil is still a bad name for this
2023-02-25 20:00:34 <orbea> not going to argue its perfect, but i dont see any good options
2023-02-25 20:00:52 <orbea> and the udevil dev is long since inactive
2023-02-25 20:01:02 <orbea> but its fairly complete
2023-02-25 20:42:50 <skarnet> udisks-replacement-that-does-not-depend-on-polkit would be a better name than udevil
2023-02-25 20:42:54 <skarnet> shows you how low the bar is
2023-02-25 20:43:09 <orbea> heh
2023-02-25 21:12:43 <minimal> u-dont-polka ? ;-)
2023-02-25 21:44:28 <Newbyte> what's wrong with polkit?
2023-02-25 21:44:28 <Newbyte> dbus?
2023-02-25 22:43:23 <Hello71> mostly dbus and javascript
2023-02-25 22:43:31 <orbea> my reason used to be not to build spidermonkey, but it seems it uses something called duktape now which I never heard of before
2023-02-25 22:46:57 <skarnet> duktape is good, if you ignore the fact it's about running javascript
2023-02-25 22:47:08 <orbea> good to know
2023-02-26 17:29:05 <ptrc> https://ptrc.gay/DGQceWyV
2023-02-26 17:29:17 <ptrc> i wonder, is that an apk-tools bug, or am i missing something obvious
2023-02-26 17:31:04 <psykose> report it and find out
2023-02-26 17:41:59 <Newbyte> psykose: find out where?
2023-02-26 17:43:31 <psykose> ?
2023-02-26 18:16:34 <ptrc> psykose: did
2023-02-26 18:16:34 <ptrc> https://gitlab.alpinelinux.org/alpine/apk-tools/-/issues/10880
2023-02-26 18:16:53 <Newbyte> psykose: dumb joke
2023-02-27 16:37:19 <psykose> ddevault: hey! i don't know to where exactly to post any email, but could you change the `update` on https://git.sr.ht/~sircmpwn/builds.sr.ht/tree/master/item/images/alpine/functions#L17 to `upgrade -U`?
2023-02-27 16:37:34 <psykose> it would ensure the image upgrades before proceeding which while not necessarily required, does sidestep a bug
2023-02-27 16:37:49 <psykose> (though generally upgrading is good anyway on the run, unless you intentionally avoided that)
2023-02-27 16:38:04 <psykose> the bug is: https://gitlab.alpinelinux.org/alpine/apk-tools/-/issues/10881
2023-02-27 18:54:27 <Newbyte> psykose: did you intend to merge !44567?
2023-02-27 18:54:40 <psykose> did u
2023-02-27 18:54:50 <Newbyte> no, I can't merge it
2023-02-27 18:54:57 <Newbyte> otherwise I would have
2023-02-27 19:09:40 <bl4ckb0ne> kek the libbsd stuff
2023-02-27 19:12:21 <bl4ckb0ne> psykose: hit that ncurses ci bug again
2023-02-27 19:13:07 <psykose> as mentioned 9 lines above
2023-02-27 19:13:08 <psykose> : )
2023-02-27 19:13:42 <bl4ckb0ne> yeah i saw it, but i need to complain you know
2023-02-27 19:13:44 <bl4ckb0ne> its in my blood
2023-02-27 20:10:25 <ddevault> psykose: can you email me about this? sir@cmpwn.com (or send a patch, ~sircmpwn/sr.ht-devel@lists.sr.ht)
2023-02-27 20:16:03 <psykose> sent
2023-02-28 07:00:13 <Asmadeus> psykose: since you merged !44618 I assume you agree there's no way of precreating devices like /dev/fuse and other devices with a modalias? (I brought it up yesterday on #alpine-linux but was a quiet hour)
2023-02-28 10:45:10 <psykose> it might be possible, it even sounds like some openrc service that should already exist
2023-02-28 10:45:30 <psykose> i feel like it reminds me of some vague discussion in #s6 and things not working correctly with that, but it's been too long
2023-02-28 10:45:43 <psykose> i merged it because it's correct :p aside from that you can look into the rest if you wish
2023-02-28 11:22:00 <superjamie> hello, are there debug symbol packages for the kernels? i can see other packages with suffix like -dbg but not for the linux-* packages
2023-02-28 11:24:26 <psykose> dont think so, no
2023-02-28 11:27:05 <ncopa> morning
2023-02-28 11:27:08 <superjamie> no worries, thank you for the quick reply!
2023-02-28 11:27:47 <ncopa> 632 package updated with apk upgrade, after being away for a month
2023-02-28 11:27:52 <ncopa> ERROR: libtraceevent-plugins-1.7.1-r1: trying to overwrite usr/lib/traceevent/plugins/plugin_cfg80211.so owned by perf-6.1.6-r0.
2023-02-28 11:30:48 <psykose> i added a replaces for that
2023-02-28 11:30:53 <psykose> i think
2023-02-28 11:31:29 <psykose> perf 6.2 removed the plugins to that project
2023-02-28 11:31:55 <psykose> could you fix it for me if i missed it? i'm on a train
2023-02-28 11:35:29 <psykose> ah, i used the wrong version
2023-02-28 11:35:45 <psykose> should be 6.2 not 6.1.6
2023-02-28 11:35:51 <psykose> in the replaces
2023-02-28 11:45:39 <ikke> psykose: because it should replace 6.1.6 as well, right?
2023-02-28 11:46:53 <ncopa> i suppose  libtraceevent-plugins shoudl have a replaces=perf
2023-02-28 11:47:01 <ncopa> i will fix it
2023-02-28 11:49:20 <ikke> ncopa: I already have a commit
2023-02-28 11:49:27 <ikke> https://tpaste.us/8MXV
2023-02-28 11:49:34 <ncopa> ok. push it. thanks!
2023-02-28 11:52:11 <omni> https://gitlab.alpinelinux.org/-/snippets/804
2023-02-28 11:55:06 <ncopa> omni: do you mix repos?
2023-02-28 12:01:39 <psykose> thanks ikke
2023-02-28 12:02:43 <psykose> omni: should be fine after rebuilds, but you have some local packages that use old libvpx
2023-02-28 12:03:14 <psykose> and libjxl
2023-02-28 12:05:29 <psykose> hm
2023-02-28 12:05:43 <psykose> that output makes no sense
2023-02-28 12:06:43 <psykose> it says the firefox just built is satisfied by the .so.7 libvpx but it has been built against .8 for weeks
2023-02-28 12:08:09 <psykose> could you apk info -R firefox
2023-02-28 12:08:23 <psykose> omni:
2023-02-28 12:10:02 <psykose> and apk list -d so:libvpx.so.7
2023-02-28 12:21:26 <ptrc> psykose: here's 3.17
2023-02-28 12:21:26 <ptrc> https://ptrc.gay/dYwByLWV
2023-02-28 12:22:08 <ptrc> i assume it's not about edge, because edge doesn't even have these versions of packages currently
2023-02-28 12:22:24 <ptrc> ah, but
2023-02-28 12:23:11 <ptrc> yeah that sounds like either a bad case of mixed branches to me or apk-tools being high and making stuff up (not unusual)
2023-02-28 12:23:36 <ncopa> yeah. i dont think there is a problem in current git
2023-02-28 12:23:48 <skarnet> psykose: re precreating devices: I don't remember such a discussion, but if there is an actual uevent emitted then it should work, I suppose
2023-02-28 12:26:42 <psykose> ah yeah
2023-02-28 12:26:43 <psykose> i
2023-02-28 12:26:52 <psykose> confused 109
2023-02-28 12:27:11 <psykose> skarnet: grep for kmod or something
2023-02-28 12:29:58 <omni> ncopa: no, edge, although it could be a valid question when it comes to me :D
2023-02-28 12:30:08 <omni> https://gitlab.alpinelinux.org/-/snippets/805
2023-02-28 12:30:28 <skarnet> psykose: I have no archive, but ask the channel if they remember that
2023-02-28 12:31:45 <omni> https://gitlab.alpinelinux.org/-/snippets/806
2023-02-28 12:31:57 <omni> psykose: ^^
2023-02-28 12:32:08 <omni> I use dotsrc mirrors
2023-02-28 12:34:07 <omni> I have a vague memory of maybe mixing repos with alpine, but if I did then that was some years ago
2023-02-28 12:34:37 <psykose> why is that firefox smelly
2023-02-28 12:35:31 <psykose> the imagemagick is old too
2023-02-28 12:35:33 <psykose> somegow
2023-02-28 12:35:41 <psykose> should be .62 or somethin
2023-02-28 12:35:56 <ptrc> the r2 qtwebengine sounds a lot like 3.17
2023-02-28 12:36:10 <psykose> it is yeah
2023-02-28 12:36:41 <ptrc> omni: can you get this? grep 'P:firefox' -B1 -A11 /lib/apk/db/installed
2023-02-28 12:36:42 <omni> it does look weird to me too and I'm pretty sure that I've upgraded past it... what gives..
2023-02-28 12:38:02 <psykose> post world and repositories files i guess
2023-02-28 12:38:36 <omni> ptrc: https://gitlab.alpinelinux.org/-/snippets/807
2023-02-28 12:39:00 <ptrc> that's.. not 3.17, that's for sure
2023-02-28 12:39:22 <ptrc> algitbot: what's 4b6467152210bbaa5e285d5795e3405392cfffa1
2023-02-28 12:40:00 <ptrc> huh, that's edge
2023-02-28 12:40:02 <ptrc> ah
2023-02-28 12:40:12 <ptrc> some random package is probably holding back updates for you
2023-02-28 12:40:39 <ncopa> apk version
2023-02-28 12:41:00 <psykose> it's probably ffmpeg-libs related
2023-02-28 12:41:06 <psykose> just post world file
2023-02-28 12:41:58 <psykose> i wish apk had a feature to have dependency tags that couldn'tbe 
2023-02-28 12:42:04 <ptrc> ncopa: oh, thanks! didn't know apk could do that
2023-02-28 12:42:09 <psykose>   in world
2023-02-28 12:42:19 <ptrc> found some broken things in my own world too
2023-02-28 12:42:56 <psykose> keep forgetting that exists :p
2023-02-28 12:51:30 <omni> looks like I have more out-of-date packages than I thought, seems like I haven't been paying attention enough this year...
2023-02-28 12:52:05 <psykose> just post apk version and world so wec
2023-02-28 12:52:12 <omni> I just do 'apk -aU upgrade' from time to time (usually almost every day and sometimes several times a day)
2023-02-28 12:52:12 <psykose>  an rule out some stuff
2023-02-28 12:53:22 <omni> https://gitlab.alpinelinux.org/-/snippets/809
2023-02-28 12:54:29 <ikke> @testing 
2023-02-28 12:55:27 <omni> oh, I hope it's not due to lab, that is the one I intentionally hold back
2023-02-28 12:56:51 <ikke> lab only depends on git 
2023-02-28 12:56:51 <omni> ikke: yes, is that a problem? https://gitlab.alpinelinux.org/-/snippets/808
2023-02-28 12:57:05 <omni> @testing
2023-02-28 12:58:05 <omni> I also tried 'apk -aU upgrade' with dotsrc and fastly https mirrors, fwiw
2023-02-28 12:59:12 <psykose> and the world
2023-02-28 13:01:42 <omni> psykose: https://gitlab.alpinelinux.org/-/snippets/810
2023-02-28 13:01:57 <omni> now I'm starting to feel a bit naked here...
2023-02-28 13:02:30 <omni> do you need my wallet too?
2023-02-28 13:03:39 <psykose> hm
2023-02-28 13:04:09 <psykose> you should check if everything with @testing is actually in testing and didnt move
2023-02-28 13:04:29 <psykose> aside from that one of those might just not exist, causes a forever hold
2023-02-28 13:04:36 <psykose> sadly im on a phone
2023-02-28 13:05:49 <psykose> if anything moved its the same as a package that got removed when you use tags like that i think
2023-02-28 13:06:09 <psykose> like mono
2023-02-28 13:06:22 <omni> I check from time to time if things have moved from testing and remove the @testing, but I don't check if things have been removed completely
2023-02-28 13:07:25 <psykose> ah no mono is testing
2023-02-28 13:07:35 <psykose> not sure, someone would have to go thru
2023-02-28 13:07:44 <psykose> the one time im on a plane
2023-02-28 13:08:57 <omni> euclidean?
2023-02-28 13:09:17 <psykose> try and find out :)
2023-02-28 13:09:44 <psykose> one easy way is to remove every @tag i guess, but you probably wouldnt want to do that
2023-02-28 13:09:59 <skarnet> <ptrc> found some broken things in my own world too
2023-02-28 13:10:04 <skarnet> every day, ptrc, every day
2023-02-28 13:10:11 <ptrc> too relatable
2023-02-28 13:10:16 <psykose> its me
2023-02-28 13:10:40 <skarnet> now *that* is a wonderful insult
2023-02-28 13:10:44 <skarnet> "you are the broken thing in my world"
2023-02-28 13:10:51 <psykose> taking off
2023-02-28 13:10:55 <psykose> love eachother
2023-02-28 13:11:18 <psykose> enjoy ffmpeg 6
2023-02-28 13:11:51 <skarnet> have a nice flight. can do love, maybe, but enjoying ffmpeg, that's a lot to ask
2023-02-28 13:16:21 <omni> when was kubectl removed? not easy to find... (not that I use it, so I'll remove it from my world)
2023-02-28 13:17:22 <ikke> omni: it's in testing 
2023-02-28 13:18:36 <omni> ikke: ah, it's a subpackage of kubernetes, of course...
2023-02-28 13:19:23 <omni> I have installed too many things over the past few years...
2023-02-28 13:20:09 <omni> but had this idea to keep a lot in order to notice early when things break, but not like this...
2023-02-28 13:25:16 <omni> removing lyx solved it!
2023-02-28 13:30:28 <ikke> What did get removed along with it? 
2023-02-28 13:37:06 <omni> ikke: https://gitlab.alpinelinux.org/-/snippets/811
2023-02-28 13:37:30 <omni> still a few out-of-date, as listed at the bottom of that snippet
2023-02-28 13:37:45 <omni> really need to go out for a bit now, bbl
2023-02-28 13:37:54 <omni> thanks for the help so far everyone!
2023-02-28 16:21:55 <eddsalkield> I'm packaging https://github.com/Zettlr/Zettlr (which is unfortunately an electron app) so looking at the `rocketchat-desktop` package for inspiration
2023-02-28 16:22:01 <eddsalkield> Following the build instructions from the Zettlr readme gives an executable depending on symbols not in Alpine
2023-02-28 16:22:37 <eddsalkield> I noticed that rocketchat uses electron-tasje instead, which runs with exit status 0 but produces no output
2023-02-28 16:23:25 <eddsalkield> will tasje help in producing a build that doesn't rely on these missing symbols, or is there something else I should do here?
2023-02-28 16:25:03 <psykose> every day i wonder why i packaged electron
2023-02-28 16:25:12 <psykose> omni: glad you solved it :)
2023-02-28 16:28:06 <eddsalkield> I mean I'd rather it not be electron as well
2023-02-28 17:24:34 <omni> psykose: I assumed you did it for the lulz
2023-02-28 17:34:39 <omni> now I just wonder why docs and mandoc are held back...
2023-02-28 17:56:24 <ikke> omni: You can try to explicitly add the latest version
2023-02-28 19:57:31 <aeroraptor> hey all - I'm building my own ISO (just a standard image w/ serial console enabled) using this guide: https://wiki.alpinelinux.org/wiki/How_to_make_a_custom_ISO_image_with_mkimage
2023-02-28 19:57:47 <aeroraptor> does anyone know if it's possible to build a BIOS-only image, and forgo UEFI support entirely?
2023-02-28 20:28:19 <Habbie> i don't, but i'm curious why?
2023-02-28 21:19:43 <aeroraptor> oh, I just figured that since I was cooking up a custom image in the first place for a machine that doesn't support UEFI - I might as well leave out UEFI entirely, right?
2023-02-28 21:20:23 <aeroraptor> I ended up building a dual mode ISO, but that wiki article probably needs to be updated 
2023-02-28 21:21:06 <aeroraptor> because there is a blurb at the top saying "For efi you should add the following:" - but there is nothing else in the article about *not* making an efi image
2023-02-28 23:24:40 <Asmadeus> psykose: (kmod static-nodes) yeah, it feels like something an openrc service like devfs should provide.. But it's actually not trivial to know which nodes are required. kmod's depmod will build /lib/modules/uname/modules.devname from each module's modinfo alias (devname:<name>, char-major-<x>-<y>, block-major-<z>-*...), but busybox's depmod doesn't gather this info so a service wouldn't know what 
2023-02-28 23:24:46 <Asmadeus> to create
2023-02-28 23:26:10 <Asmadeus> I can kludge something static for our kernel easily, but a proper solution would need to start by having busybox depmod create that modules.devname, then have some service use that -- plus some of the devnodes need permissions fixed according to what's in eudev rules which wouldn't automatically trigger on simple mknod... can of worms
2023-02-28 23:26:39 <Asmadeus> I guess it's probably best to document it somewhere and will stop poking at it