#archlinux-ports | Logs for 2026-07-18

Back
[00:00:01] -!- bertptrs has quit []
[00:00:41] -!- bertptrs has joined #archlinux-ports
[03:58:34] -!- hcmb_ has joined #archlinux-ports
[03:58:34] hcmb is now known as Guest5727
[03:58:34] hcmb_ is now known as hcmb
[03:59:03] -!- Guest5727 has quit [Ping timeout: 252 seconds]
[04:09:11] -!- m_a_r_k_ has joined #archlinux-ports
[04:11:07] -!- m_a_r_k has quit [Ping timeout: 264 seconds]
[04:11:08] m_a_r_k_ is now known as m_a_r_k
[04:52:14] -!- hch12907 has joined #archlinux-ports
[05:00:38] -!- hch12907 has quit [Quit: See ya!]
[05:02:03] -!- hch12907 has joined #archlinux-ports
[05:03:45] -!- hch12907 has quit [Client Quit]
[05:05:06] -!- hch12907 has joined #archlinux-ports
[05:06:54] -!- h|ignition has quit [Remote host closed the connection]
[05:07:19] -!- h|ignition has joined #archlinux-ports
[06:43:05] -!- hch12907 has quit [Quit: See ya!]
[07:56:50] -!- h|ignition has quit [Remote host closed the connection]
[07:57:14] -!- h|ignition has joined #archlinux-ports
[08:31:16] -!- filmroellchen has joined #archlinux-ports
[13:07:02] -!- hch12907 has joined #archlinux-ports
[13:16:39] <solskogen|M> I tried to contact them, but got no reply. We're far ahead of the work Valve did.
[13:17:15] <solskogen|M> d4s: Correct.
[13:52:54] <solskogen|M> Nothing, really. Too bad they didn't respond to the forum post I made in 2023 about this.
[13:53:57] <solskogen|M> One thing they did, that we didn't is that they've bootstrapped from x86_64. We "bootstrapped" from ALARM.
[14:15:59] <d4s> solskogen|M: thanks for the answer.
[14:22:53] <d4s> solskogen|M: I've joined to that project only late 2025, so can't say something about the post from 2023 ;) But for now one of our goals is to not duplicate the efforts if that's possible for sure -- we definitely have a bit different goals, like maintaining the stable branches, which we have to be able to rebuild literally anytime and acheive the same (ok, almost the same) binaries.
[15:04:01] <SpieringsAE_sff> Jeah the Holo stuff is not intended to be a port of actual Arch like this is I think. Though I do think there is a lot of common work required
[15:04:30] <SpieringsAE_sff> I just want an aarch64 steam client so it can manage all of the fex stuff with proton or something
[15:10:15] -!- h|ignition has quit [Ping timeout: 252 seconds]
[15:13:34] -!- h|ignition has joined #archlinux-ports
[15:29:17] -!- h|ignition has quit [Remote host closed the connection]
[15:29:55] -!- h|ignition has joined #archlinux-ports
[15:32:09] <d4s> "not intended to be a port of actual Arch", yes and no ;) That we're trying to do -- based on the `state` repo create a "replay" of those packages versions in CI. So technically it is a subset of ArchLinux from the Nov 2025 + patches/hacks/builds re-ordering.
[18:24:11] <Solskogen> d4s: yeah, we try to follow x86_64 as much as possible.
[18:25:29] <Solskogen> That said, I've managed to build more packages than Collabora / Valve :-) As long as they contribute to the PKGBUILDs I'm happy
[18:27:04] <d4s> And that's great tbh ;) For contributing we have to catch more recent version of ArchLinux otherwise we will end up with unneeded or outdated patches :(
[18:27:38] <d4s> Solskogen: I'm curious how do you restoring the correct build sequence in CI if you stuck with some package not rebuilding?
[18:28:56] <solskogen|M> If a package fails (for whever reason) none of the packages that are dependent on that will be built either.
[18:29:34] <solskogen|M> But there has been times where I had to step in manually and fix - and then rebuild.
[18:31:56] <solskogen|M> Also worth to note: We don't build -any packages. They are taken from upstream.
[18:33:09] <d4s> heh... we discover some issues trying to rebuild some of them ;) rtp should have details
[18:35:17] <d4s> solskogen|M: what if the package with version x.y-1 has failed to build, and the x.y-2 has been provided, do you skip -1 build or you have hard build sequence in your CI?
[18:35:24] <solskogen|M> When we started we built them as well, but I had an experiment with using the ones from upstream and to my amusement they worked. I haven't seen any problem with doing it that way as of yet (8ish months)
[18:36:27] <solskogen|M> then we just skip -1. I rely on core.db and extra.db (we don't have -testing and staging(yet!)) - not the state repo. I tried relying on the state repo but I found too many issues.
[18:42:31] <d4s> right, we are relying to state tbh and trying to restore the correct sequence. The initial attempt relying to DB failed for us, since we discover some packages required for other packages rebuilds are disappeared too fast to became published in the repo.
[18:46:15] <d4s> just cureious how are you dealing with sonames changing, like icu, gpgme and similar? I mean -- in case when you need to use the new soname to link with for the package but tooling is still linked with a previous one
[18:57:36] <solskogen|M> good old hacks. That part has not been automated.
[18:58:04] <solskogen|M> icu has been pretty much straight forward, but gpgme (and his pal libassuan) was one very hacky thing.
[18:58:51] <solskogen|M> which is the reason why we are ahead when it comes to libassuan, compared to x86_64 - we got caught right in the middle of it. and I didn't want to hack it again.
[19:08:56] <d4s> right, we have the similar issue and have to re-order build sequence. And even provide additional deps/replaces couple of times.
[19:09:34] <d4s> btw the recent `state` is mostly usable in the sense it is containing all packages, but ordering sometimes fails, especially if all changes pushed in alphabetical order in a bulk way, or the order is incorrect due to other reason. For instance, kde, haskell, gcc and clang chains
[19:15:58] -!- h|ignition has quit [Remote host closed the connection]
[19:16:24] -!- h|ignition has joined #archlinux-ports
[19:18:39] <solskogen|M> kde has been ok. clang as well. haskell was a problem, but I think I've ironed all of them out now. gcc and friends needs manual work.
[19:21:35] <d4s> solskogen|M: sounds very interesting. May I ask you to point to CI sources? Sorry I think it is faster to ask here instead of searching :(
[19:22:20] -!- h|ignition has quit [Remote host closed the connection]
[19:22:21] <solskogen|M> also, some packages have moved from any to x86_64 (or the other way around) - for those kind of issues I have a script that I run manually that tries to find those kind of things.
[19:22:44] <solskogen|M> https://github.com
[19:22:45] <phrik> Title: GitHub - solskogen/archlinux-aarch64-builder · GitHub (at github.com)
[19:22:45] -!- h|ignition has joined #archlinux-ports
[19:23:23] <solskogen|M> 99% of that is written by AI, so look out for both dragons and drakoons.
[19:26:24] <d4s> oh! I see
[19:26:48] <d4s> impressed it works
[19:26:54] <d4s> Thanks for the link
[19:32:38] <solskogen|M> since I build multiple packages at the same time, you will find some strangs hacks for things like firefox. I can't build firefox and firefox-developer-editition at the same time.
[19:33:14] -!- h|ignition has quit [Remote host closed the connection]
[19:33:39] -!- h|ignition has joined #archlinux-ports
[21:09:49] -!- marmis has quit [Quit: Bye!]
[21:11:21] -!- marmis has joined #archlinux-ports
[21:37:45] -!- filmroellchen has quit [Quit: filmroellchen]
[22:07:31] -!- h|ignition has quit [Ping timeout: 252 seconds]
[22:07:59] -!- h|ignition has joined #archlinux-ports
[22:10:13] -!- titus_livius has joined #archlinux-ports
[22:20:31] -!- maxjrh3 has quit [Quit: The Lounge - https://thelounge.chat]
[22:21:18] -!- maxjrh3 has joined #archlinux-ports
[22:23:18] -!- maxjrh3 has quit [Client Quit]
[22:27:14] -!- maxjrh3 has joined #archlinux-ports
[23:07:49] -!- h|ignition has quit [Remote host closed the connection]
[23:08:36] -!- h|ignition has joined #archlinux-ports