#archlinux32 | Logs for 2025-11-23

Back
[00:07:15] -!- sL1pKn07 has joined #archlinux32
[00:38:36] <bill-auger> girls: KitsuWhooa: andreaas package signing key has expired - last chat here september? - the consequences are dire for the distro o/c
[00:40:32] <KitsuWhooa> I don't think there's anything I cna do
[00:40:34] <KitsuWhooa> can
[00:40:38] <bill-auger> ie: of the keys installed per archlinux32-keyring package, only 2 are valid: tyzoid and deep42
[00:41:24] <bill-auger> yea was just hoping you were around m u have no related news?
[00:41:29] <KitsuWhooa> nope
[00:41:46] <KitsuWhooa> I don't want to say it, but it might be time to sunset a32
[00:42:17] <bill-auger> unrelated actually, but also the jeys.php website is not working anymore - maybe it is not needed anymore
[00:42:34] <KitsuWhooa> I've been super busy with a whole bunch of other things and it seems like everyone else has been busy with life
[00:42:52] <KitsuWhooa> and it's becoming a massive uphill battle trying to maintain a rolling release
[00:48:19] <bill-auger> similarly uphill for us maintaining the i686 port - the keyrings have always been especially troublesome
[00:48:35] <KitsuWhooa> well, I did my part by making my key not expire
[00:48:44] <KitsuWhooa> because I am tired of expiring keys
[00:50:09] <bill-auger> it would be sad to see arch32 cease, as parabola's i686 port depends on it - it amounts to 1/3 less workload for us, but still would rather retain it
[00:50:50] <KitsuWhooa> I agree, but there is just no interest in it
[00:51:04] <KitsuWhooa> and I am tired of spending hours and hours fixing things only for them to break next month
[00:51:42] <bill-auger> sry to be clear: 1/3 less workload to drop the port entirely - mega more workload to continue, by adopting all of arch32
[00:52:13] <KitsuWhooa> I understood what you meant, don't worry
[00:53:23] <bill-auger> lets just say the small diffs against arch conceal the massive workload relative to x86_64
[00:53:58] <KitsuWhooa> x86_64 is much, much easier to deal with
[00:54:14] <KitsuWhooa> especially when it comes to rust
[00:54:20] <KitsuWhooa> since now everything under the sun depends on rust
[00:54:25] <bill-auger> arm is too and archarm is not much healthier
[00:54:37] <KitsuWhooa> I can see armv7 also having the same fate
[00:55:41] <bill-auger> archarm often freezes a month with no communication from upstream
[00:56:54] <bill-auger> yes its the latter bit that worries ppl though - they are doing the work silently and as silently spew private staging when its ready
[00:57:25] <KitsuWhooa> there is nothing silent on arch32, I'll tell you that
[00:57:43] <KitsuWhooa> I wonder how viable an i686 Alpine port would be
[00:58:16] <bill-auger> np AFAIAC - it has never stalled for more than a month that i remember , but the no communicatin bothers ppl
[00:58:35] <KitsuWhooa> yup, I know
[01:02:07] <bill-auger> guix and slackware still support 32bit x86 and may always - those are the only ones i can think of which may always
[01:03:40] <KitsuWhooa> unfortunately I have no interest in running either of them :p
[01:07:02] -!- txtsd has quit [Ping timeout: 244 seconds]
[01:08:33] <bill-auger> oh yes i forgot about hyperbola - it is essentially parabola LTS, which should be familiar enough to you - they may support i686 indefintely
[01:08:57] <KitsuWhooa> that's the only way I can see this being viable
[01:08:59] <KitsuWhooa> LTS
[01:09:21] <KitsuWhooa> well, not necessarily LTS in the sense that you support it for 10 years
[01:09:36] <bill-auger> https://www.hyperbola.info
[01:09:37] <KitsuWhooa> but one release every 2 years which goes EOL after 6 months
[01:09:37] <phrik> Title: Hyperbola (at www.hyperbola.info)
[01:10:05] <bill-auger> no, actually their plan is to use a BSD kernel for the next major release
[01:10:15] <KitsuWhooa> oh lol
[01:10:18] <KitsuWhooa> no thanks
[01:10:30] -!- txtsd has joined #archlinux32
[01:10:54] <KitsuWhooa> > For our release 0.4.5 we will remove pacman and begin our development and implementation of hyperman as drop-in replacement.
[01:11:26] <KitsuWhooa> good luck to them
[01:11:33] <KitsuWhooa> but that is definitely not my kind of thing
[01:13:13] <KitsuWhooa> I am usually also not a subscriber to 100% libre linux, but it might be viable for ancient hardware
[01:13:19] <bill-auger> i meant but the distro is there - if someone wanted to replace all of arch32's PKGBUILDs with those from hyperbola, arch32 may become LTS and maintainable
[01:13:21] <KitsuWhooa> until I remember that the sound card on my pentium 3 laptop needs firmware
[01:13:55] <KitsuWhooa> I mean, maybe?
[01:14:02] <KitsuWhooa> but then it's not arch32 anymore
[01:14:32] <KitsuWhooa> I'd be fine with arch snapshots, but then you have security issues, which would need security backports, so you end up with debian
[01:14:36] <KitsuWhooa> so at that point just make a debian port
[01:14:39] <KitsuWhooa> minus the pamcan
[01:14:41] <KitsuWhooa> pacman
[01:15:36] <bill-auger> sure it is though - all of those PKGBUILDs were once current tracking arch - they simply froze and are no longer tracking anything
[01:15:47] <KitsuWhooa> so you have security issues
[01:16:24] <bill-auger> ie: yes, hyperbola maintains that by tracking and importing debian patches
[01:18:29] <bill-auger> but those patches will be gone soon too or become inapplicable :( - the future is bleak for those machines
[01:19:32] <KitsuWhooa> the most discouraging thing with arch32 is that things are immediately broken
[01:19:36] <KitsuWhooa> so it feels like your work has gone to waste
[01:19:58] <KitsuWhooa> that, and buildmaster being broken
[01:20:07] <KitsuWhooa> having to constantly fight it
[04:05:11] -!- finsternis has joined #archlinux32
[07:15:39] <socksinspace> KitsuWhooa: if that is not a typo: alpines x86 baseline should not be too hard to lower, it was only raised to require sse2 somewhat recently (mid 2023). Depending on how low you would want to go you'd need to deactivate all things that need rust, and modify a few packages that have added sse2 requirements, if i wasn't busy with more exotic old architectures (SuperH) i might give it a
[07:15:40] <socksinspace> try
[09:45:09] -!- abaumann has joined #archlinux32
[09:45:10] <buildmaster> Hi abaumann!
[09:45:10] <buildmaster> !rq abaumann
[09:45:11] <phrik> buildmaster: <abaumann> I see a real niche here: write a browser which doesn't completly suck.
[09:45:45] <abaumann> I'm aware that the keys expired or are expiring. I wanted to see, what the echo is, if somebody is still using Arch32.
[09:46:09] <abaumann> The matter of fact is, all build machines are stopped since 2 or 3 months, no more new packages are built.
[09:46:27] <abaumann> I'm sometimes doing a shim package so that some software doesn't break, but that's all.
[09:46:51] <abaumann> Real live is a little bit challenging at the moment and Arch32 is not on the top list anymore (at least on my top list).
[09:47:24] <abaumann> Sunsetting is something we have to discuss, at the moment, I just thought leaving it as is, is an option. But we definitely have to update the signing and the master keys..
[09:47:33] <abaumann> ..otherwise we cannot even install a package.
[09:47:47] <abaumann> The iso is also broken, there is too much stuff on it, that is currently not building
[09:48:36] <abaumann> I hardly manage to keep my keys up to date, so deep42thought was usually dealing with the signing from upstream, etc.
[10:12:02] -!- abaumann has quit [Quit: leaving]
[11:02:55] <socksinspace> the fact that, in upstream linux, 32-bit x86 has no dedicated maintainer, but is just a part of general x86, is a blessing and a curse, whereas other niche architectures, like m68k, alpha, and parisc have designated maintainers, 32-bit x86 support is just a legacy byproduct that the companies employing the x86 maintainers don't care about. On the one hand "we" could take a free ride for the longest time, on the other hand, now no one feels r
[11:02:55] <socksinspace> esponsible.
[11:04:39] <socksinspace> thank you irc for splitting that message, very cool
[11:15:37] <socksinspace> same in gcc, in a way, not sure how much 32-bit x86 support is seen as something that "gets in the way" of "advancing" x86_64
[11:17:08] <socksinspace> tldr: blame capitalism, duplication of efforts, and modern computing hype-cyles. Thank you for coming to my ted talk :)
[11:37:04] <KitsuWhooa> <abaumann> The matter of fact is, all build machines are stopped since 2 or 3 months, no more new packages are built.
[11:37:17] <KitsuWhooa> I also noticed you took your builders down a few months ago for what it's worth
[11:37:43] <KitsuWhooa> socksinspace: yeah, that Alpine. I know they required sse2 recently
[11:38:29] <KitsuWhooa> Deactivating all the things that need rust is a big nono
[11:38:56] <KitsuWhooa> It would mean building rust with no sse2, which can be done, but upstream tell you not to do
[11:39:07] <socksinspace> if the level of x86 you need supports rust you won't need to deactivate it :P
[11:40:35] <socksinspace> It's just that I'm trying to target SuperH at the moment, which does not have any level of upstream support for rust at the moment
[11:42:38] <KitsuWhooa> I am fairly aware of what is needed to get a distro running on i686 without sse2, given I've been doing it on arch32
[11:42:51] <KitsuWhooa> i486 is dead in the water though
[11:43:08] <KitsuWhooa> And yes, good luck with super h
[11:43:27] <KitsuWhooa> What machine do you have to test against, anyway? Dreamcast?
[11:43:35] <socksinspace> yeah, apparently someone tried i486 for alpine, seemed like they faced much the same issues you faced
[11:43:51] <KitsuWhooa> I've unfortunately given up on i486
[11:43:55] <socksinspace> hp jornada 680, and qemu
[11:44:03] <KitsuWhooa> Oh huh
[11:44:18] <KitsuWhooa> I found qemu not to be too reliable when it comes to these things
[11:44:30] <KitsuWhooa> It'll gladly execute sse2 if you tell it to emulate a Pentium 3
[11:44:32] <KitsuWhooa> :p
[11:44:58] <socksinspace> it's a trade-off between being reliable, and wanting to have more than 16 MB of RAM :D
[11:45:04] <KitsuWhooa> Mmmmhm
[11:46:10] <socksinspace> also, mostly working is good enough for me, not like there are any other options for sh3
[11:47:44] <socksinspace> i did want to do alpine for older x86 too, for my thinkpad 600x, but all my cpu cycles, and brain power, is occupied by sh3
[11:48:15] <socksinspace> maybe i still have a git branch for that
[11:48:46] <KitsuWhooa> My issue with Alpine is that I need systemd
[11:48:54] <KitsuWhooa> But I believe that's been mostly resolved
[11:49:28] <KitsuWhooa> So I might just try to start contributing towards sse2-less i686 over there
[11:49:36] <KitsuWhooa> Although idk how
[11:49:50] <KitsuWhooa> If anything it'd need to be unofficial
[11:50:23] <socksinspace> you would need to pull in some changes from postmarketOS still
[11:50:49] <socksinspace> yeah, alpine doesn't have anything like debian-ports
[11:51:27] <socksinspace> i should bother them to make an irc for such unsupported usecases
[11:51:49] <KitsuWhooa> Heh
[11:53:47] <KitsuWhooa> But yeah, pulling stuff from PmOS shouldn't be too bad
[11:54:21] <KitsuWhooa> I just don't know what level of work is required to basically keep an Alpine fork up
[11:57:04] <socksinspace> i would hope it is not too crazy for something fairly well supported like sse2-less x86, but since my experience is with extremely niche architectures, and i'm not yet in the "keeing it up" part of my port, idk
[11:57:45] <socksinspace> the patches i came up with so far have mostly been fairly tame
[11:59:12] <socksinspace> i almost ended up without working cmake, which would've complicated things a bit, but i *think* i have averted that problem ;P
[12:01:48] <KitsuWhooa> The real problem is no rust
[12:05:16] <socksinspace> I am blissfully unaware of the complexity involved in making rust work ;)
[12:08:17] <socksinspace> i just know that llvm takes *forever* to build
[12:08:54] <socksinspace> so i am quite happy to not have to deal with that too
[12:50:05] <KitsuWhooa> I guess you don't expect to get a desktop running
[13:19:04] <socksinspace> i'll be pleasantly surprised if i get X11 working at all
[13:30:02] Xe_ is now known as Xe
[14:13:52] -!- sL1pKn07 has quit [Quit: Client Excited...]
[14:14:05] -!- sL1pKn07 has joined #archlinux32
[14:25:55] -!- sL1pKn07 has quit [Ping timeout: 246 seconds]
[14:26:58] -!- sL1pKn07 has joined #archlinux32
[14:50:36] <KitsuWhooa> Yeah that's what I thought
[14:50:45] <KitsuWhooa> Which sucks tbh but oh well
[14:50:57] <KitsuWhooa> i486 is basically the same
[14:58:55] -!- dmc has quit [Ping timeout: 240 seconds]
[15:09:56] <socksinspace> it sucks a bit, but i feel like i should not be surprised that i can't run *everything*, idk if it would be more disappointing to compile software and not be able to use it because the machine is too slow, or to have the build process just fail
[15:12:36] -!- dmc has joined #archlinux32
[22:11:34] -!- h|weechat has quit [Ping timeout: 260 seconds]
[22:37:07] -!- h|weechat has joined #archlinux32
[23:10:13] -!- titus_livius has joined #archlinux32