#archlinux32 | Logs for 2024-05-21

Back
[01:33:17] <KitsuWhooa> pacman-static was built on 2021-10-31 10:01:53
[01:33:18] <KitsuWhooa> uh
[01:34:22] <KitsuWhooa> I mean, sure, it's static, but I get the feeling we'll have issues at some point
[01:37:20] <KitsuWhooa> I also don't yet understand how pacman-static is even supposed to work. Do we add an explicit makdepends to it?
[01:46:51] <KitsuWhooa> yeah even with pacman-static in the makedepends, it fails way too soon
[02:16:14] <KitsuWhooa> I can't even make a new chroot now?
[02:16:17] <KitsuWhooa> (115/115) checking package integrity [#####################################] 100%:: File /var/cache/pacman/pkg/linux-api-headers-6.7-1.0-any.pkg.tar.zst is corrupted (invalid or corrupted package (checksum)).
[02:22:10] <KitsuWhooa> Anyway, build support doesn't work because the moment the chroot is created, it picks a broken pacman
[02:22:45] <KitsuWhooa> only way I can think of fixing this is by building a static pacman called simply `pacman` and adding it to build-support manual
[07:30:08] -!- abaumann has joined #archlinux32
[07:30:38] <abaumann> I got the feeling that pacman-static gets into play too late: devtools must probably also call the static versions of pacman?
[07:43:44] <abaumann> detected dubious ownership in repository at '/srv/git/packages.git
[07:43:52] <abaumann> this is the stupidest feature I have seen
[07:46:24] <abaumann> the git user has a HOME of /, nice
[07:50:05] <abaumann> And systemd git-daemon sets no environment, so that .gitconfig can be found.
[07:50:15] <abaumann> Nobody uses git:// obviously. :-0
[07:55:49] <abaumann> archbuild has a arch-nspawn for pacman (which is linked against icu74), now, this can be changed to pacman-static and before install pacman-static package into /var/lib/archbuild/xxx/root.
[07:56:49] <abaumann> The only problem is: I don't have a pentium4/i686/i486 version around of pacman-static, and I have to build one with.. guess what.. :-)
[07:57:15] <abaumann> ah. maybe we can just install icu74 into the archbuilds the same way and make pacman happy again..
[07:58:04] <abaumann> but then I have to build an icu74 package :-)
[08:05:40] <abaumann> bsdtar: error while loading shared libraries: libicuuc.so.74: cannot open shared object file: No such file or directory
[08:05:43] <abaumann> next problem
[08:05:50] <abaumann> so pacman-static doens't help
[08:06:17] <abaumann> manually crafting an icu74 from an old icu in the archive..
[08:06:30] <abaumann> there is a icu74 shim package, but I cannot build it.
[08:35:49] <KitsuWhooa> abaumann: renaming pacman static to pacman and putting it in build support manual works
[08:36:02] <KitsuWhooa> It'll install the right once when it recreates the chroot
[08:36:03] <KitsuWhooa> Well
[08:36:07] <KitsuWhooa> Needs a symlink too
[08:36:12] <KitsuWhooa> I haven't had time to look into it
[08:36:59] <KitsuWhooa> Did you get anywhere with forcing the builders to use pacman static or should I go back into doing that?
[08:37:30] <KitsuWhooa> I feel like always using pacman static in build support isn't such a bad idea
[08:37:39] <KitsuWhooa> As long as it's relatively up to date
[08:46:59] <KitsuWhooa> Sometimes I wonder how bad of an idea it'd be to make Arch32 less rolling release :p
[08:47:06] <KitsuWhooa> it's clear we do not have enough resources to keep up with upstream
[08:49:21] <abaumann> I'm currently trying to prepare some build slaves to have an icu74 in them
[08:49:26] <abaumann> That's easier.
[08:50:01] <abaumann> the problem is: pacman-static is only fixing the pacman/icu74 problem.
[08:50:14] <abaumann> bsdtar is also linked against icu74 (and what else who knows?)
[08:50:38] -!- zxrom has joined #archlinux32
[08:50:48] <KitsuWhooa> we don't need to use bsdtar right?
[08:51:27] -!- zxrom has parted #archlinux32
[08:51:44] <abaumann> something needs it
[08:51:54] <abaumann> I got bsdtar icu74 linkage errors..
[08:51:57] <abaumann> when build which
[08:52:05] <abaumann> which has not a whole lot of dependencies :-)
[08:53:01] <abaumann> so, core-staging-<arch>-build are working on my euronuc build slaves
[08:53:13] <abaumann> I force a libarchive (bsdtar) and a pacman now..
[08:57:09] <abaumann> ah, it build jumpy now.
[08:57:23] <KitsuWhooa> make sure to start the builders again if you haven't
[08:57:31] <abaumann> certain packages just get rescheduled over again for now good reason.
[08:57:32] <KitsuWhooa> surely there has to be a way to automate this
[08:57:40] <KitsuWhooa> yeah, been meaning to look into it
[08:57:42] <KitsuWhooa> but then qt6 broke
[08:57:49] <KitsuWhooa> well, gcc14 (presumably) broke qt6
[08:58:48] <abaumann> I have seen big breakages in KDE qt6 in testing
[09:12:58] <abaumann> so, libarchive,pacman are rebuild
[09:14:32] <abaumann> bsdtar: error while loading shared libraries: libicuuc.so.74: cannot open shared object file: No such file or directory
[09:14:35] <abaumann> grmpf
[09:15:14] <abaumann> isn't that part of libarchive?
[09:15:37] <abaumann> or it linked again against icu74
[09:19:52] <abaumann> of course. libarchive cannot be rebuild this way.
[09:21:48] <KitsuWhooa> lddtree time? :p
[09:22:37] <abaumann> yeah. libxml2
[09:22:52] <KitsuWhooa> I hate icu with a passion :p
[09:23:37] <KitsuWhooa> I wonder if we can make our PKGBUILD build both the latest icu and the previous version automatically
[09:23:43] <KitsuWhooa> that way systems will have both the new and the old one
[09:23:47] <KitsuWhooa> and not have to deal with this headache
[09:26:23] <abaumann> pacman-conf: error while loading shared libraries: libicuuc.so.74: cannot open shared object file: No such file or directory
[09:26:29] <abaumann> sweet.
[09:26:48] <abaumann> /usr/share/makepkg/source/git.sh: line 95: git: command not found
[09:27:06] <abaumann> bugs everywhere. :-)
[09:29:31] <abaumann> I have to rethink this. It's basically a nightmare which is hard to automatize.
[09:29:59] <KitsuWhooa> there's a pacman-conf-static, right?
[09:30:04] <KitsuWhooa> or is it pacman-static-conf
[09:30:06] <KitsuWhooa> one of the two
[09:30:29] <abaumann> yes, but adapting all of devtools and makepkg is not nice
[09:32:47] <abaumann> i686 and i486 build, pentium4 not, why..
[09:33:37] <abaumann> in the past I had scripts executed every second injecting missing libraries into /var/lib/archbuild
[09:33:41] <KitsuWhooa> oucfh
[09:33:43] <KitsuWhooa> *ouch
[09:35:22] <abaumann> all static build tools and core..
[09:35:33] <KitsuWhooa> well I'm building libxml2
[09:35:34] <abaumann> but that would get big
[09:35:48] <KitsuWhooa> for build support only maybe
[09:36:19] <KitsuWhooa> have a static pacman that replaces the normal pacman and a static bsdtar that replaces the normal one
[09:38:56] <KitsuWhooa> or maybe
[09:39:12] <KitsuWhooa> have devtools run ldd on pacman, and if it finds missing libraries switch to the static one
[09:39:34] <KitsuWhooa> that will at least solve the broken pacman
[09:39:41] <KitsuWhooa> having a static bsdtar should be easy to do
[09:44:43] <KitsuWhooa> qt5 rebuild now I see
[09:49:16] <KitsuWhooa> which failed because of this again pacman-conf: error while loading shared libraries: libicuuc.so.74: cannot open shared object file: No such file or directory
[09:51:01] <KitsuWhooa> I stopped all my builders
[09:58:12] <KitsuWhooa> in the meantime, I did something I'm sure I'll regret https://github.com
[09:58:28] <KitsuWhooa> next step, either getting mesa 7 to build on a modern compiler, or porting the old drivers to mesa amber
[10:02:42] <abaumann> oh, yes. forking everything they throw out upstream. that's fun :-)
[10:02:55] * abaumann cooks a red sauce for Spaghetti in parallel
[10:03:57] <KitsuWhooa> sounds nice
[10:06:36] <KitsuWhooa> these old drivers were dropped from mesa in 2011
[10:06:42] <KitsuWhooa> so it's going to be quite the challenge
[11:17:02] <KitsuWhooa> abaumann: any luck?
[11:25:30] mavica is now known as mavicaway
[12:03:30] -!- ssserpent has joined #archlinux32
[12:05:23] -!- AtleoS has quit [Ping timeout: 264 seconds]
[12:10:09] <abaumann> nope.
[12:11:07] <abaumann> not for pentium4
[12:11:10] <abaumann> for some reason
[12:15:08] <KitsuWhooa> alright
[12:15:21] <KitsuWhooa> please stop the builders if you give up
[12:15:28] <abaumann> yup
[12:15:29] <KitsuWhooa> so that it doesn't mark packages as broken for no reason
[12:16:02] <abaumann> I'm running them manually anyway ATM during those experiments. :-)
[12:26:22] -!- abaumann has quit [Quit: leaving]
[13:15:16] -!- abaumann has joined #archlinux32
[13:17:05] <abaumann> now that I found the bug in the build slave now I have a sanity error..
[13:31:54] <KitsuWhooa> oh come on
[13:32:41] <KitsuWhooa> Murphy's law :p
[13:39:16] <abaumann> EOI (end of insanity)
[13:40:24] <KitsuWhooa> nice :p
[13:40:32] <KitsuWhooa> it better not have been python-snappy again :p
[13:40:44] <KitsuWhooa> oh also, if you don't mind being interrupted, since you're around
[13:40:52] <KitsuWhooa> is there a way to check what prevents a package from being moved to testing?
[13:41:04] <KitsuWhooa> for example I want to see what else needs to be built for python to move to testing on its own
[13:41:19] <KitsuWhooa> (actually I force moved python to testing by accident but the question stands)
[13:45:31] <abaumann> mmh. I don't know whether you can prevent a move, I think after some time the move will happen.
[13:45:39] <abaumann> libxml2 rebuilt.
[13:48:04] <abaumann> !wtf python-snappy
[13:48:06] <phrik> abaumann: What the hell is python-snappy?
[13:48:32] <abaumann> we had once some questions like this "what blocks X" in the buildmaster, I think.
[13:49:55] <abaumann> maybe the bot could ask those questions?
[13:50:23] <abaumann> let me force and push icu74
[13:50:48] <abaumann> ah, it fails again..
[13:51:05] <KitsuWhooa> oh
[13:51:09] <abaumann> hang on. maybe my archbuild directories need a cleanup..
[13:51:23] <abaumann> ok, I seem to be able to build which
[13:51:32] <abaumann> with a new chroot
[13:52:10] <abaumann> mmh. ok. I force rebuilds also of libarchive, libxml2 and pacman without icu74 in place now
[13:52:23] <abaumann> then we should be in normal operation mode again
[14:02:03] <KitsuWhooa> awesome
[14:02:05] <KitsuWhooa> thank you for taking care of it
[14:02:16] <abaumann> no problem.
[14:20:57] <KitsuWhooa> > QTBUG-123374 Linux on Arm fails with '-headersclean' in global/qfloat16.h
[14:20:57] <KitsuWhooa> hm
[14:25:55] <KitsuWhooa> that seems unrelated
[14:43:47] -!- abaumann has quit [Ping timeout: 264 seconds]
[14:52:55] -!- abaumann has joined #archlinux32
[15:13:00] -!- abaumann has quit [Quit: leaving]
[15:38:59] -!- ssserpent has quit [Ping timeout: 264 seconds]
[15:43:36] -!- ssserpent has joined #archlinux32
[15:46:19] -!- ssserpent has quit [Client Quit]
[19:18:18] <girls> KitsuWhooa: you can ask buildmaster, why something is not moved (but I see, it is no longer around)
[19:18:43] <girls> but since you have ssh access, you can still ask: there's a "why-dont-you" executable
[19:19:03] <girls> why-dont-you unstage $pkg
[19:19:19] <girls> but be warned: it takes some time :(
[20:58:10] <KitsuWhooa> Ooooh
[20:58:15] <KitsuWhooa> That's what that does p
[20:58:17] <KitsuWhooa> Thank you