#archlinux32 | Logs for 2024-10-15
Back
[00:23:26] -!- sL1pKn07 has quit [Ping timeout: 252 seconds]
[00:27:45] -!- sL1pKn07 has joined #archlinux32
[01:54:09] -!- lithium_pt has joined #archlinux32
[01:55:06] -!- lithiumpt has quit [Ping timeout: 252 seconds]
[03:16:35] -!- T`aZ has quit [Ping timeout: 265 seconds]
[03:17:46] -!- drathir_tor has quit [Ping timeout: 260 seconds]
[03:19:24] -!- drathir_tor has joined #archlinux32
[03:23:10] -!- T`aZ has joined #archlinux32
[05:09:22] -!- bill-auger has joined #archlinux32
[07:58:12] -!- bill-auger_ has joined #archlinux32
[07:58:29] -!- bill-auger_ has quit [Client Quit]
[08:00:25] -!- bill-auger_ has joined #archlinux32
[08:00:50] -!- bill-auger_ has quit [Client Quit]
[14:04:00] -!- drathir_tor has quit [Remote host closed the connection]
[14:09:50] -!- drathir_tor has joined #archlinux32
[15:43:03] -!- drathir_tor has quit [Remote host closed the connection]
[15:48:44] -!- drathir_tor has joined #archlinux32
[16:06:01] -!- drathir_tor has quit [Ping timeout: 260 seconds]
[16:08:46] -!- drathir_tor has joined #archlinux32
[16:15:53] <girls> there's nothing to update in the keyring
[16:16:42] <KitsuWhooa> I thought abaumann updated the key
[16:16:53] <KitsuWhooa> I'm confused at what needs to be done anymore
[16:16:57] <girls> v20240924 is the latest
[16:17:05] <girls> is this already in the repos?
[16:17:20] <KitsuWhooa> 20240131-5.1
[16:17:24] <KitsuWhooa> the one in the repos is that
[16:17:24] <girls> O.O
[16:17:46] <girls> https://mirror.archlinux32.org
[16:17:55] <girls> ah, my bad
[16:18:02] <girls> the arch_32_ repos
[16:18:06] <KitsuWhooa> yeah
[16:18:14] <KitsuWhooa> those matter more than releng :p
[16:18:22] <girls> this just needs a kick in the ... pkgver
[16:18:35] <KitsuWhooa> I didn't realise everything was already done
[16:18:46] <KitsuWhooa> and just needed a build
[16:19:07] <girls> if it is on sources.archlinux32.org, then all that's left to do is: build for [releng] and build for arch32
[16:19:42] <girls> where "build for [releng]" is actually "build for x86_64/archlinuxewe" and then "copy from archlinuxewe to releng" - at least, when I do it.
[16:21:02] <girls> ok, should be in git, now -> wait for the builders :)
[16:21:09] <KitsuWhooa> thanks :)
[16:22:24] <girls> "Add libutempter dependency, switch deps to .so" - this explains, why screen 4.9.1 is currently broken on arch32 :(
[16:22:46] <girls> it misses that dependency - which will come with screen 5.0.0
[16:24:31] <KitsuWhooa> I need to get back to fixing librsvg at some point
[16:24:44] <KitsuWhooa> I found the reason it doesn't build on i686, but actually getting cargo to behave is a whole other story
[16:25:22] <KitsuWhooa> basically one of the dependencies started enabling simd by default and you can't disable features of a dependency's dependency without forking it
[16:25:41] <girls> O.o
[16:25:58] <girls> is it possible to tell cargo to use "local" dependencies?
[16:26:02] <KitsuWhooa> yeah
[16:26:09] <KitsuWhooa> but you need to know which version to download :p
[16:26:13] <girls> do we have the dependency itself in the packages?
[16:26:16] <KitsuWhooa> which requires parsing Cargo.lock
[16:26:18] <KitsuWhooa> nope
[16:26:22] <girls> :-(
[16:26:27] <KitsuWhooa> I don't think arch packages rust libraries
[16:26:30] <KitsuWhooa> as in, .rlib
[16:27:03] <KitsuWhooa> so the solution will be something along the lines of download librsvg source -> parse Cargo.lock -> download dependency -> patch librsvg to use it -> patch dependency to not require sse2
[16:27:10] <KitsuWhooa> in the pkgbuild
[16:27:46] <girls> I would prefer to adding 1k rust packages automatically and have them use arch32-packaged rust crates only
[16:28:05] <KitsuWhooa> I don't think it's viable long term
[16:28:10] <KitsuWhooa> given that we don't have the resources for it
[16:28:18] <girls> hmm
[16:28:24] <KitsuWhooa> what debian does is pick a version, package it, then force everything to use that version
[16:28:29] <KitsuWhooa> if it doesn't, they patch it so that it does
[16:28:40] <girls> ah, right, the versioning is the problem
[16:28:43] <KitsuWhooa> you can easily have multiple rust programs wanting different versions of libraries
[16:28:49] <girls> I'm too much used to the rolling-release nature of arch
[16:29:16] <girls> "You don't compile with the newest version of this library? That is your problem!
[16:29:18] <girls> "
[16:29:29] <KitsuWhooa> with how fast rust is moving
[16:29:43] <KitsuWhooa> I don't think you can expect everything to build against the latest version of a library
[16:29:57] <KitsuWhooa> it's worse than ffmpeg/libav* :p
[16:31:00] <KitsuWhooa> s/rust/the rust ecosystem/
[16:33:42] <girls> are the patches of debian reusable for us?
[16:33:57] <girls> we could package the same version and just hook onto their patches
[16:34:02] <KitsuWhooa> some probably are
[16:34:10] <KitsuWhooa> but again, to me this seems like it will require way more maintenance from our end
[16:34:22] <girls> yeah, probably
[16:34:33] <KitsuWhooa> but then... There are so many crates I need to patch for https://github.com
[16:34:34] <phrik> Title: assert!(cfg!(target_feature = "sse") && cfg!(target_feature = "sse2")) panic at build on i386, i586, FreeBSD, or many target_os=none x86 targets. · Issue #1999 · briansmith/ring · GitHub (at github.com)
[16:34:37] <girls> let's talk again, when we have 10s or 20s of packages, that need sse patched out ;)
[16:34:45] <KitsuWhooa> we already do
[16:35:10] <KitsuWhooa> We could probably try doing an experiment with ring
[16:35:48] <KitsuWhooa> I can't remember what, but something major stopped building due to ring
[16:36:54] <KitsuWhooa> oh right
[16:37:06] <KitsuWhooa> it's needed to build python-cryptography
[16:37:08] <KitsuWhooa> (indirectly)
[16:38:58] <girls> talking of rust: My package mirror implemented in rust is almost functional (i.e. it is in a usable, but not presentable state - it will probably linger in this state forever, now). However, when trying to get my Krümel running, I saw some issues that looked like virtualization issues to me - so I most probably can not use docker for build slaves :-(
[16:39:33] <KitsuWhooa> docker doesn't use vms, does it?
[16:39:41] <girls> it does not, I think
[16:40:07] <girls> makepkg runs flawlessly in docker, however the whole chroot/systemd/whatnot layer seems to fail
[16:40:46] <KitsuWhooa> well, if you decide to use lxc, I can tell you how to get it working :p
[16:40:57] <KitsuWhooa> although, have you seen if my patches help?
[16:41:18] <girls> oh, I forgot about those
[16:41:19] <KitsuWhooa> css broke again in https://git.archlinux32.org
[16:41:20] <phrik> Title: builder - Archlinux32 build system (at git.archlinux32.org)
[16:41:21] <KitsuWhooa> wtf
[16:41:51] <KitsuWhooa> it's like a ghost that goes and breaks it every time
[16:43:08] <KitsuWhooa> anyway
[16:43:10] <KitsuWhooa> https://git.archlinux32.org
[16:43:12] <phrik> Title: lxc_pacstrap.patch - devtools32 - Archlinux32 fork of devtools (at git.archlinux32.org)
[16:44:12] <KitsuWhooa> and the diff at the bottom https://git.archlinux32.org
[16:44:13] <phrik> Title: Hack to fix building inside LXC containers - devtools32 - Archlinux32 fork of devtools (at git.archlinux32.org)
[16:44:48] <KitsuWhooa> I wonder if I could mount /proc in pacstrap directly
[16:50:59] <girls> Failed to open bus: No such file or directory
[16:50:59] <girls> Attempted to remove disk file system under "/run/systemd/nspawn/propagate/arch-nspawn-2362", and we can't allow that.
[16:51:09] <KitsuWhooa> ah
[16:51:13] <KitsuWhooa> that's nothing like what I've seen
[16:51:27] <KitsuWhooa> the first one sounds like dbus
[16:51:39] <KitsuWhooa> maybe you could figure it out with strace
[16:52:18] <girls> yeah, OR I just switch to qemu runners
[16:52:24] <girls> at least, that was my last plan
[16:52:44] <KitsuWhooa> 32 bit?
[16:52:48] <girls> nah
[16:52:51] <KitsuWhooa> okay good :p
[16:53:00] <girls> just to separate the host x86_64 from the arch x86_64
[16:53:02] <KitsuWhooa> I'd say give lxc a try if you want
[16:53:20] <girls> haven't used that so far - but maybe, that's even more a reason to try it :D
[16:53:29] <KitsuWhooa> beware though
[16:53:42] <KitsuWhooa> lxd is not lxc, and the `lxc` command is part of lxd :p
[16:54:09] <KitsuWhooa> lxd is a wrapper of sorts for lxc if I am not mistaken
[16:54:20] <girls> ok, later :-)
[16:54:22] <girls> cu
[16:54:25] <KitsuWhooa> see ya
[17:02:41] -!- lupo_on_libera has joined #archlinux32
[17:41:47] mavicaway is now known as mavica
[18:04:26] -!- drathir_tor has quit [Ping timeout: 260 seconds]
[18:34:13] -!- lithiumpt has joined #archlinux32
[18:34:49] -!- lithium_pt has quit [Ping timeout: 260 seconds]
[20:01:49] -!- drathir_tor has joined #archlinux32
[20:05:51] -!- drathir_tor has quit [Remote host closed the connection]
[20:11:52] -!- drathir_tor has joined #archlinux32
[20:34:56] -!- drathir_tor has quit [Ping timeout: 260 seconds]
[20:39:53] -!- drathir_tor has joined #archlinux32
[20:41:41] -!- drathir_tor has quit [Remote host closed the connection]
[20:49:54] -!- drathir_tor has joined #archlinux32
[22:28:41] -!- drathir_tor has quit [Remote host closed the connection]
[22:34:05] -!- drathir_tor has joined #archlinux32
[23:42:52] mavica is now known as mavicaway