#archlinux32 | Logs for 2024-03-08

Back
[00:26:57] -!- zxrom has joined #archlinux32
[01:00:16] <KillerWasp> zxrom: o/
[01:03:33] <zxrom> KillerWasp, Hey!
[01:20:08] <bill-auger> if proper maintenance is neglected, rust spreads and corrodes everything it touches
[02:57:46] -!- mvchtz has quit [Ping timeout: 255 seconds]
[03:01:10] -!- mvchtz has joined #archlinux32
[03:31:47] -!- sunshavi has joined #archlinux32
[05:34:10] <KillerWasp> bill-auger: https://i.pinimg.com
[05:50:45] <deep42thought> !grab bill-auger
[05:50:46] <phrik> deep42thought: Error: I couldn't find a proper message to grab.
[05:50:51] <deep42thought> :(
[05:56:18] <KillerWasp> !grab bill-auger
[05:56:18] <phrik> KillerWasp: Error: I couldn't find a proper message to grab.
[05:56:27] <KillerWasp> !grab deep42thought
[05:56:27] <phrik> KillerWasp: Tada!
[05:56:34] <KillerWasp> mmmm....
[05:59:13] -!- dmc has quit [Ping timeout: 264 seconds]
[06:01:39] <bill-auger> try again - i quite fancy that one too
[06:02:03] <bill-auger> if proper maintenance is neglected, rust spreads and corrodes everything it touches
[06:19:22] -!- dmc has joined #archlinux32
[07:37:53] -!- abaumann has joined #archlinux32
[07:37:53] <buildmaster> Hi abaumann!
[07:37:53] <buildmaster> !rq abaumann
[07:37:54] <phrik> buildmaster: <abaumann> otherwise you end up in a big mess - and also with that rule you end up in a big mess.
[07:38:19] <abaumann> Python depends on Rust.. wouldn't be surprised but would hope it's just an optional dependency
[07:39:13] <abaumann> devtools now uses git-cliff which is a rust program.
[07:39:22] <abaumann> for the tag target, ok fine. should affect our builders..
[07:40:02] <abaumann> about rust and i486, I was never able to bootstrap it, so I would have to use a distro providing it (like t2sde) to bootstrap it.
[07:41:49] <abaumann> Linux makes the mistake to replace an old ecosystem with an unmature one. This will end well..
[07:45:17] <abaumann> where `ver` is the latest version of `clap` supporting rustc 1.69.0
[07:45:18] <abaumann> sure.
[08:57:01] <abaumann> uh. don't update devtools32 yet. I let it run on eurnuc and do some testing after merging with upstream..
[09:05:58] -!- ssserpent has joined #archlinux32
[09:09:25] <abaumann> core-staging-pentium4.conf contains Architecture=i486. There almost zero possibility this can go wrong in a sed in devtools
[09:11:12] -!- ssserpent has quit [Quit: WeeChat 4.2.1]
[09:12:32] -!- ssserpent has joined #archlinux32
[09:15:52] <abaumann> -rw-r--r-- 1 root root 0 Mar 8 09:44 extra-testing-pentium4.conf
[09:16:04] <abaumann> all this patching and sedding is really really hard to understand..
[09:19:47] <abaumann> yeah. the usual problem: GNU make wildcards and make -j N.
[09:20:04] <abaumann> This stuff breaks happilly in non-deterministic ways when not compiled with make -j1
[09:26:58] <KitsuWhooa> 09:38:19 <abaumann> Python depends on Rust.. wouldn't be surprised but would hope it's just an optional dependency <-- yeah I was wrong
[09:27:08] <abaumann> uff :-)
[09:27:20] <KitsuWhooa> 11:20:04 <abaumann> This stuff breaks happilly in non-deterministic ways when not compiled with make -j1 <-- I've mentioned it (and been bitten by it) a few times
[09:27:25] <KitsuWhooa> the makefile is just badly written I think
[09:27:40] <KitsuWhooa> make should be able to tell if it can parallelise something I think
[09:27:51] <KitsuWhooa> or, well, you tell it that it can/can't
[09:28:02] <abaumann> Can you force it in the Makefile itself?
[09:28:31] <abaumann> Of course I can do a make -j manually or in the PKGBUILD, but it would be nice if the Makefile "knows" itself that it doesn't work in parallel.
[09:28:33] <KitsuWhooa> yes
[09:28:54] <KitsuWhooa> Also, I can't get buildmaster to rebuild python-setuptools
[09:28:54] * abaumann plows trhough GNU snowland of documentation
[09:29:13] <KitsuWhooa> I'm pretty sure I fixed it, but it's not rebuilding it
[09:29:24] <abaumann> It's not scheduled?
[09:29:42] <KitsuWhooa> I ran schedule-for-rebuild
[09:29:49] <KitsuWhooa> and picked the right one
[09:29:50] <KitsuWhooa> 2 any python-setuptools 2c9c85d0f6f03651ef555e55f07bd453d2af1e64 f0e566a906693b7b18bb6b36c0ee57f8e1587bd9 extra
[09:30:10] <abaumann> ah. maybe there is still a bug which shows up only when built in the build infrastucture.
[09:30:21] <abaumann> pkgctl repo clone --protocol=https --arch32 python-setuptools
[09:30:25] <KitsuWhooa> it doesn't send it to the builders
[09:30:29] <abaumann> core-staging-pentium4-build
[09:30:56] <abaumann> My wish would be we could do a pkgctl publish push to buildmaster (db and mirror) of such a manually built package.
[09:31:02] <abaumann> This would speed up things considerably
[09:31:08] <KitsuWhooa> I still don't like that idea
[09:31:21] <abaumann> Or of course make the other way faster.
[09:31:33] <KitsuWhooa> it's actually acceptable with get-package-updates-ignore now
[09:31:41] <abaumann> python-setuptools builds just fine.
[09:31:48] <KitsuWhooa> yup
[09:31:57] <KitsuWhooa> so why is buildmaster not queueing it up
[09:32:31] <abaumann> arch=('any')
[09:32:32] <abaumann> ?
[09:32:38] <KitsuWhooa> yeah
[09:32:46] <KitsuWhooa> most of the python packages are arch any
[09:32:56] <abaumann> maybe we can patch it to pentium4 i686 and i484 and thus force native builds.
[09:32:59] <KitsuWhooa> why?
[09:33:08] <KitsuWhooa> as in
[09:33:10] <KitsuWhooa> why would this be needed?
[09:33:12] <abaumann> I vaguely remember a thing from the past where I did something like this.
[09:33:23] <KitsuWhooa> yeah, and I came across a package that did it, and it caused headaches :p
[09:33:26] <abaumann> I fear, I just worked around some issue.
[09:33:30] <KitsuWhooa> so I'd like to avoid it
[09:33:41] <KitsuWhooa> 11:33:23 <KitsuWhooa> yeah, and I came across a package that did it, and it caused headaches :p <-- while trying to bootstrap python
[09:33:51] <abaumann> speaking of fortune-telling and stuff on the channel, I would in this case add some programming voodoo to it.
[09:34:32] <KitsuWhooa> I don't suppose you have any idea why make_source_info fails on 3 any python-setuptools 9db23d6d7da66fb99875a1bc49caf56ececccf7a bf7d478537e662322e8045be5249f383dfe16d4c extra
[09:34:39] <KitsuWhooa> "make_source_info" did not return a "pkgbase" - eh, what?
[09:34:39] <KitsuWhooa> >curl: (22) The requested URL returned error: 500<
[09:34:54] <KitsuWhooa> (that is the wrong one, but I wonder if that failing is what got buildmaster stuck)
[09:36:26] <abaumann> python-setuptools-rust, mmh.
[09:36:47] <KitsuWhooa> python-setuptools-git is what depends on rust
[09:36:56] <KitsuWhooa> but it doesn't seem absolutely needed
[09:37:19] <abaumann> the pkginfo files look ok on the buildmaster
[09:37:32] <abaumann> but for python-setuptools=extra=9db23d6d7da66fb99875a1bc49caf56ececccf7a=bf7d478537e662322e8045be5249f383dfe16d4c'
[09:37:36] <abaumann> which contains curl: (22) The requested URL returned error: 500
[09:38:01] <abaumann> if that is the revision not building, then the buildmaster is not adding failed pkginfo data.
[09:38:15] <KitsuWhooa> well, it can't build it to mark it as failed
[09:38:17] <abaumann> most likely in this case it should say something like 'uh?? or ehh? pkginfo??'
[09:38:18] <KitsuWhooa> maybe we just found the bug
[09:38:47] <KitsuWhooa> https://tasossah.com
[09:38:54] <KitsuWhooa> I wonder if I need to remove that one from the build list
[09:39:35] <abaumann> worth a try.
[09:39:43] <KitsuWhooa> I'll let you know if everything blows up
[09:39:44] <abaumann> I clean the /srv/http/pkginfo directory
[09:39:50] <KitsuWhooa> sure
[09:41:00] <abaumann> which the ignoring get-package-info the pkginfo cache is not that important anymore
[09:41:09] <abaumann> but for the full run maybe from time to time
[09:41:38] <abaumann> .NOTPARALLEL:
[09:42:00] <abaumann> forgot about that. I moved to saner build systems than make if possible, so I lost a little bit my make-ish..
[09:42:10] <KitsuWhooa> port devtools32 to cmake? :p
[09:42:20] <abaumann> uh, aeh. I said. sane :-)
[09:42:24] <KitsuWhooa> I like cmake, okay? :p
[09:42:36] <KitsuWhooa> meson seems to be all the rage these days, but it's python, so
[09:42:40] <abaumann> matter of taste, I suppose :-)
[09:42:55] <abaumann> cmake is ok for standard tasks
[09:43:08] <abaumann> it has some trouble to do probing and autoconf-things in my opition.
[09:43:23] <abaumann> it can do it, but it lacks some standard out-of-the-box things.
[09:43:32] <abaumann> from a language design percpective cmake is just bad
[09:43:43] <abaumann> you can still use it, of course
[09:44:14] <abaumann> mmh. where do I have to sprinke .NOTPARALLEL: mow..?
[09:44:18] <abaumann> on which targets..
[09:44:33] <abaumann> ok. all of them..
[09:44:56] <abaumann> which is especially funny on things like $(BUILDDIR)/pacman.conf.d/%-i686.conf
[09:45:13] <KitsuWhooa> I'm sure some of them can still be done in parallel
[09:45:18] <KitsuWhooa> but with modern computers it shouldn't matter
[09:45:55] <abaumann> I prefer to have it run in 20 seconds in -j1 and correct instead of 1s in parallel and creating all kind of headaches. :-)
[09:46:29] <abaumann> $(BUILDDIR)/pacman.conf.d/%-i486.conf: config/pacman/%.conf
[09:46:49] <abaumann> uh. this runs havock. this is a dependency in itself
[09:47:08] <abaumann> wildcards is one of the worst features in GNU make, you can shoot yourself in your foot so easily (and not only the foot)
[09:47:25] <KitsuWhooa> I see :p
[09:47:33] <KitsuWhooa> I've written very few makefiles in my life
[09:47:39] <abaumann> happy you ;-)
[09:47:48] <KitsuWhooa> I've also never used autotools in a personal project
[09:48:07] <abaumann> I have done the whole hand-writte make in Vize Venemma stule up to autoconf/automake/libtool/m4 thingies. In a 500kloc project.
[09:48:14] <abaumann> fun, just fun.
[09:48:17] <KitsuWhooa> indeed
[09:48:27] <abaumann> to be fair, there was not much around 20 years ago to achieve the same thing
[09:49:43] <abaumann> appologies go out: Wietse Venema, I brutaly butchered his name, sorry. :-)
[09:50:03] <KitsuWhooa> ooooh
[09:50:09] <abaumann> postfix is one example of makefiles used the right way.
[09:52:11] <KitsuWhooa> well, either I made buildmaster insane
[09:52:23] <KitsuWhooa> or it was going to be insane and I just pushed it by running the sanity check
[09:52:25] <KitsuWhooa> let's see
[09:52:39] <KitsuWhooa> ah
[09:52:40] <KitsuWhooa> https://buildmaster.archlinux32.org
[09:52:40] <phrik> Title: result of archlinux32 build master's sanity checksanity of the buildmaster's mysql database (at buildmaster.archlinux32.org)
[09:53:06] <KitsuWhooa> looks like that package was in i486
[09:53:30] <KitsuWhooa> there's a newer one in i486/extra-staging though
[09:53:41] <abaumann> yes, any packages can end up on i486 build mvs, which is not ideal.
[09:53:51] <KitsuWhooa> I don't think that's the issue
[09:53:56] <KitsuWhooa> I think it's just that it's ancient
[09:54:09] <abaumann> ah
[09:54:28] <abaumann> 69.0.3 bs. 67.3.0
[09:55:00] <abaumann> mmh, but this sounds a little bit like a very old i486 version was removed in the buildmaster to make way for the now new one. and this failed
[09:55:15] <KitsuWhooa> it's the one I just removed manually from the db I think
[09:55:22] <KitsuWhooa> the one that would return HTTP 500
[09:55:37] <abaumann> inf pkginfo, you mean.
[09:55:41] <KitsuWhooa> yes
[09:56:16] <abaumann> the 500 happens when either upstream or arch32 repos don't contain the necessary scheduled commits for building.
[09:56:21] <abaumann> I think.
[09:56:26] <KitsuWhooa> in this case I think it's upstream
[09:56:31] <abaumann> though it shouldn't
[10:07:04] <KitsuWhooa> right
[10:07:05] <KitsuWhooa> it's sane again
[10:07:24] <abaumann> good. let's make it unsafe again ;-)
[10:07:41] <abaumann> aeh. I mean, let's test ptyhon-setuptools again
[10:07:56] <KitsuWhooa> on it :p
[10:11:43] <KitsuWhooa> abaumann: do you happen to know what the currently_blocking column in the DB table means?
[10:11:59] <KitsuWhooa> still doesn't seem to be getting queued and I wodner if that is at fault
[10:12:13] <KitsuWhooa> *wonder
[10:13:17] <abaumann> mmh.
[10:13:24] <KitsuWhooa> I think it's how many packages are blocked by this
[10:14:03] <KitsuWhooa> oh I have an idea
[10:14:16] <KitsuWhooa> I didn't unset checkdepends
[10:14:21] <KitsuWhooa> maybe doing that will get it to rebuild
[10:14:47] <KitsuWhooa> wait
[10:14:49] -!- texus has joined #archlinux32
[10:14:50] <KitsuWhooa> there's no checkdepends?
[10:14:59] <KitsuWhooa> oh there is
[10:17:19] <abaumann> mmh. it's a counter which tells you how many packages the current build assignment is blocking?
[10:17:28] <KitsuWhooa> looks like it
[10:17:35] <abaumann> so it is just relevant to be shown on the web page as "blocked"
[10:17:42] <KitsuWhooa> yup
[10:17:47] <abaumann> this is just, so that huge counts can be manually prioritized.
[10:18:13] <KitsuWhooa> yeah, so, why isn't it building
[10:18:20] <KitsuWhooa> do I have to bump the pkgrel or something
[10:18:33] <KitsuWhooa> I shouldn't need to
[10:18:59] <KitsuWhooa> I spent 5 hours fixing setuptools last night and I can't get it to build. It's so frustrating
[10:19:29] <KitsuWhooa> next step is to delete the mod, get-package-updates, revert it, get-package-updates
[10:19:31] <KitsuWhooa> and hope it does something
[10:21:26] <KitsuWhooa> I wonder if we can clear the build_assignments table
[10:22:23] <KitsuWhooa> I think it'd require everything to be told to rebuild manually, but at this point I'm not sure that's a bad thing
[10:22:34] -!- texus has quit [Quit: WeeChat 4.2.1]
[10:24:47] <abaumann> yaah. frustrating is the right term.
[10:25:40] <abaumann> I'm reading 5 seds for 2 hours and try to translate them into something more readable.
[10:25:59] <abaumann> The problem is, if at least I would see a comment somewhere what those things are supposed to do.
[10:26:09] <KitsuWhooa> yes :p
[10:26:14] <abaumann> I want pkgctl repo clone --arch32 to do the same as in the builder
[10:26:31] <KitsuWhooa> [master@buildmaster ~]$ prioritize-build-list -d -w <(printf '^python-setuptools$' )
[10:26:31] <KitsuWhooa> 38
[10:26:40] <abaumann> The problem is a little bit, it has no notion of a pkg_subrel (of course, becasue that's a stateful thing inside the buildmaster)
[10:26:45] <KitsuWhooa> yet in the db, priority = 0
[10:26:55] <abaumann> well' 38 is not quite 42, but I'll take it ;-)
[10:27:33] <KitsuWhooa> buildmaster went insane
[10:27:45] <KitsuWhooa> I didn't even touch the db
[10:28:07] <KitsuWhooa> oh this is bad
[10:28:15] <abaumann> again, the same pytonn-setuptools 486 thingy
[10:28:19] <KitsuWhooa> no
[10:28:25] <KitsuWhooa> it's all three architectures
[10:28:29] <KitsuWhooa> and it's the last version
[10:28:55] <KitsuWhooa> wtf happened
[10:28:58] <abaumann> it tries to publish the any package in all three pacman databases
[10:29:07] <KitsuWhooa> does that mean it built it?
[10:29:13] <abaumann> think so.
[10:29:17] <KitsuWhooa> none of my builders did it
[10:29:18] <abaumann> it's just missing in the db file
[10:29:22] <KitsuWhooa> so it must've been euronuc
[10:29:35] <abaumann> mine is in limbo and switched off
[10:29:44] <KitsuWhooa> oh
[10:29:50] <KitsuWhooa> so it didn't get built then
[10:29:54] <abaumann> no
[10:30:12] <abaumann> it's an older package which no got readded to the pacman db files and failed again.
[10:30:16] <abaumann> that's what I think.
[10:30:31] <KitsuWhooa> I'll revert the db then
[10:30:46] <abaumann> -rw-r--r-- 2 http mirror 789918 Feb 25 13:03 python-setuptools-1:69.0.3-4.0-any.pkg.tar.zst
[10:30:53] <KitsuWhooa> I am not removing the only working setuptools :p
[10:30:54] <abaumann> doesn't sound _that_ new
[10:30:57] <KitsuWhooa> yeah that's old
[10:31:01] <abaumann> no, definitely not.
[10:31:08] <abaumann> but then I was wrong. nothing new was built
[10:33:29] <KitsuWhooa> I copied the .old over the new db and nothing changed
[10:33:32] <KitsuWhooa> now what :p
[10:33:50] <KitsuWhooa> maybe I should nuke setuptools after all
[10:34:40] <KitsuWhooa> ...wait
[10:34:42] <KitsuWhooa> I just realised
[10:34:53] <KitsuWhooa> since setuptools is any, the issue started from when buildmaster didn't copy setuptools to i486
[10:34:56] <KitsuWhooa> I think
[10:35:26] <KitsuWhooa> meh, screw it, I'm deleting setuptools
[10:36:44] <KitsuWhooa> OH
[10:36:46] <KitsuWhooa> I see what happened
[10:36:50] <KitsuWhooa> it tried to move them from extra-staging to extra
[10:37:38] <abaumann> extra? not extra-testing?
[10:37:46] <KitsuWhooa> extra
[10:37:56] <KitsuWhooa> <font color="#FF0000">-pentium4/extra/python-setuptools-1:69.0.3-4.0-any.pkg.tar.zst</font><br>
[10:38:00] <KitsuWhooa> pentium4 is extra
[10:38:08] <KitsuWhooa> whereas i486 and i686 is staging
[10:38:13] <KitsuWhooa> or no
[10:38:19] <KitsuWhooa> only i486 is staging
[10:38:24] <KitsuWhooa> sorry I have no idea what is going on
[10:39:07] <abaumann> it's exactly the same version it tries to push into differently stable repos. This doesn't sound good (or familiar)
[10:43:15] <KitsuWhooa> right, sane again
[10:43:18] <KitsuWhooa> with less setuptools
[10:43:37] <KitsuWhooa> yeah setuptools is completely gone
[10:43:54] <KitsuWhooa> hopefully build support will save the day
[10:44:37] <KitsuWhooa> [master@buildmaster ~]$ prioritize-build-list -d -w <(printf '^python-setuptools$' )
[10:44:37] <KitsuWhooa> 1
[10:44:39] <KitsuWhooa> this looks right now
[10:44:51] <KitsuWhooa> and it got priority 38
[10:45:04] * abaumann crosses fingers
[10:45:56] <KitsuWhooa> nope
[10:46:03] <KitsuWhooa> 2024-03-08 10:45:48: building package "python-persistent" (revisions b7310a286a0e93320d06a96e1fd1ba9dadeca009 0000000000000000000000000000000000000000, repository extra, straw :mirrored_source_by_hash:) for i686 ...
[10:46:04] <KitsuWhooa> why?
[10:46:45] <KitsuWhooa> I beg you buildmaster, build the damn package
[10:46:54] <abaumann> should I try on euronuc (or should I rename it to euronuke first) ;-)
[10:47:01] <KitsuWhooa> I don't think it'll make a difference
[10:47:12] <abaumann> hopefully not :-)
[10:47:57] <KitsuWhooa> I'll try one more thing
[10:48:12] <KitsuWhooa> I'll delete it from the build_assignments and run the full get-package-updates
[10:48:17] <KitsuWhooa> I don't see how it will make a difference
[10:48:18] <KitsuWhooa> but who knows
[10:53:20] <abaumann> good luck
[10:53:25] <KitsuWhooa> thanks :p
[10:53:29] <KitsuWhooa> only luck can save me at this point
[10:59:16] <abaumann> I have to leave now, I let euronuc run, it seems to run fine on devtools32 - famous last words of course :-)
[10:59:25] <abaumann> AFK
[10:59:26] <KitsuWhooa> take care
[10:59:27] <KitsuWhooa> thanks
[10:59:36] <abaumann> thanks too
[11:00:22] -!- abaumann has quit [Quit: leaving]
[11:12:22] <KitsuWhooa> error: could not read db 'extra' (Damaged tar archive)
[11:12:24] <KitsuWhooa> what the fuck
[11:12:29] <KitsuWhooa> nonononono
[11:12:32] <KitsuWhooa> what is going on
[11:16:43] <KitsuWhooa> okay, I replaced it with the .old ones
[11:39:42] -!- sunshavi has quit [Remote host closed the connection]
[11:40:01] -!- sunshavi has joined #archlinux32
[11:47:03] <KitsuWhooa> okay, now I need to get the database and the repo in sync
[12:31:59] <KitsuWhooa> It's *still* not queueing up setuptools
[12:32:19] <KitsuWhooa> whatever, there's at least working setuptools from build support now
[12:49:40] -!- texus has joined #archlinux32
[13:15:09] -!- texus has quit [Quit: WeeChat 4.2.1]
[13:18:05] -!- ssserpent has quit [Quit: WeeChat 4.2.1]
[14:41:07] -!- morriset has joined #archlinux32
[16:19:11] -!- zxrom has quit [Quit: Leaving]
[18:33:25] -!- zxrom has joined #archlinux32