#archlinux32 | Logs for 2024-03-14

Back
[08:18:59] -!- titus_livius has joined #archlinux32
[08:19:36] -!- abaumann has joined #archlinux32
[08:20:02] <abaumann> mmh. the irc loging is broken again.
[08:20:13] <abaumann> I miss 2 days of messages
[08:20:42] -!- buildmaster has quit [Read error: Connection reset by peer]
[08:23:42] -!- buildmaster has joined #archlinux32
[08:26:47] <abaumann> aeh, calling sanity-check updates package files? whut?
[08:59:53] <abaumann> logging is back.
[08:59:55] -!- abaumann has quit [Quit: leaving]
[09:47:07] -!- sunshavi has quit [Remote host closed the connection]
[09:47:52] -!- sunshavi has joined #archlinux32
[10:51:30] <KitsuWhooa> abaumann: The archlinux-package keyring is failing to build
[10:51:43] <KitsuWhooa> sq returns error: The subcommand 'toolbox' wasn't recognized
[12:26:15] <KitsuWhooa> sources.archlinux32.org is failing a lot
[12:26:20] <KitsuWhooa> as in, 404 not found
[14:39:17] -!- dmc has joined #archlinux32
[14:46:51] -!- dmc has quit [Quit: WeeChat 4.2.1]
[14:56:13] -!- dmc has joined #archlinux32
[15:23:16] -!- abaumann has joined #archlinux32
[15:23:16] <buildmaster> Hi abaumann!
[15:23:16] <buildmaster> !rq abaumann
[15:23:17] <phrik> buildmaster: <abaumann> doesn't ## uncomment the #? ;-)
[15:23:56] <KitsuWhooa> hah
[15:23:58] <abaumann> KitsuWhooa: nothing really needs the archlinux-keyring package, archlinux32-keyuring is enough. It's just there in case you want to do something with upstream arch packages.
[15:24:01] <KitsuWhooa> that's a nice quote
[15:24:07] <abaumann> sequoia is some rust thingy, right.
[15:24:08] <KitsuWhooa> abaumann: ah, I see
[15:24:15] <KitsuWhooa> I don't know
[15:24:21] <abaumann> So likely archlinux-keyring has not been rebuild for ages. :-)
[15:24:23] <KitsuWhooa> speaking of which, I can't build rust because llvm runs out of memory
[15:24:29] <KitsuWhooa> I've been fighting with it :p
[15:24:36] <abaumann> that's absolutely not surprising. :-)
[15:24:39] <KitsuWhooa> but hey, I got linux, linux-lts, and linux-pae all building with the same PKBGUILDs
[15:24:43] <KitsuWhooa> or, well, same patches
[15:24:50] <abaumann> ah. that's nice.
[15:25:03] <KitsuWhooa> also, I think you were asking about being able to build packages like they'd be built in a builder
[15:25:11] <abaumann> https://archlinux.org
[15:25:12] <phrik> Title: Arch Linux - sequoia-sq 0.34.0-1 (x86_64) (at archlinux.org)
[15:25:13] <abaumann> yep, needs rust
[15:25:36] <KitsuWhooa> this is what I do
[15:25:41] <KitsuWhooa> makechrootpkg -r /chroottest/i686 -c -n -C -d/var/cache/archbuild32:/var/cache/pacman/pkg -U builder3 -- -A
[15:25:57] <KitsuWhooa> with something like mkarchroot -C /pacman-i686-staging-manual.conf /chroottest/i686/root base-devel
[15:26:04] <KitsuWhooa> (you can pick any of the builder's configs)
[15:26:44] <abaumann> rust gets big when you enable debug symbols IIRC
[15:26:52] <abaumann> there are flags to counteract that
[15:26:53] <KitsuWhooa> It's not rust itself
[15:26:57] <abaumann> ah?
[15:26:59] <KitsuWhooa> it's ld when linking llvm
[15:27:07] <KitsuWhooa> I haven't tried gold yet, but that's the next step
[15:27:09] <abaumann> ld.gold, I suppose.
[15:27:11] <KitsuWhooa> yup
[15:27:22] <abaumann> is lto enabled?
[15:27:29] <KitsuWhooa> that might also be the case
[15:27:35] <abaumann> this requires evertything to be in memory, which might be too much.,
[15:27:39] <KitsuWhooa> I don't think we patch it out, so if it is upstream, it is here too
[15:27:41] <abaumann> so disabling lto is another option
[15:27:55] <KitsuWhooa> I've also been wondering about trying mold
[15:28:04] <abaumann> I just remember switching off LTO for certain packages with big linking artifacts
[15:28:09] <KitsuWhooa> yeah
[15:28:15] <abaumann> in the options, I fear.
[15:28:22] <abaumann> PKGBUILD options
[15:28:31] <abaumann> which is not nice at all to do.
[15:28:41] <KitsuWhooa> the config.toml is embedded in the PKGBUILD for some reason
[15:29:01] <abaumann> extra/clang/PKGBUILD: 1i options=(!lto)
[15:29:03] <abaumann> yep.
[15:29:10] <KitsuWhooa> ah, I see
[15:29:21] <KitsuWhooa> that explains why the llvm16 package built fine too I assume
[15:29:29] <KitsuWhooa> So I'll look into disabling LTO for rust's llvm then
[15:30:17] <abaumann> switching ld between ld.gold and ld.bfd is usually dangerous, as people tend to test only one version, then you end up in missing symbols or worse.
[15:30:27] <KitsuWhooa> interesting
[15:30:32] <abaumann> So, yes, I would try LTO disabling first
[15:30:35] <KitsuWhooa> there's also lld too :p
[15:30:39] <KitsuWhooa> (which I doubt works right now)
[15:30:40] <abaumann> and mold
[15:31:30] <abaumann> mold can use multiple cores for linking - which is nice - till you run out of memory even faster :-)
[15:31:51] <KitsuWhooa> That's what I thought was happening with llvm
[15:32:03] <KitsuWhooa> but no
[16:40:55] -!- ssserpent has quit [Quit: WeeChat 4.2.1]
[16:54:50] -!- bill-auger has quit [Remote host closed the connection]
[16:55:57] -!- bill-auger has joined #archlinux32
[18:12:47] -!- Foxboron has quit [Ping timeout: 260 seconds]
[18:13:52] -!- Foxboron has joined #archlinux32
[18:36:30] <bill-auger> major GUI web browsing problem on this OS - i tried firefox epiphany netsurf qutebrowser midori - none work
[18:36:48] <KitsuWhooa> patches welcome :p
[18:36:49] <bill-auger> links works great though
[18:37:29] <KitsuWhooa> I use w3m personally
[18:37:38] <KitsuWhooa> (on arch32)
[18:42:02] -!- abaumann has quit [Quit: leaving]