#archlinux32 | Logs for 2024-05-16

Back
[00:03:29] -!- zxrom has quit [Quit: Leaving]
[01:07:49] -!- AtleoS has joined #archlinux32
[01:25:14] -!- bdju has joined #archlinux32
[04:29:51] -!- ssserpent has joined #archlinux32
[04:34:22] -!- lithiumpt has joined #archlinux32
[04:35:25] -!- lithium_pt has quit [Ping timeout: 255 seconds]
[05:32:36] -!- abaumann has joined #archlinux32
[05:32:37] <buildmaster> Hi abaumann!
[05:32:37] <buildmaster> !rq abaumann
[05:32:38] <phrik> buildmaster: <abaumann> you should start every morning with a *-zen practice.. ;-)
[05:32:51] <abaumann> icu was a forced move
[05:33:31] <abaumann> now some packages required the new icu, some others the old one. the shim packages icu73 etc. are exactly for that purpose.
[05:33:49] <abaumann> I also forced libxml2 (because it depends on icu).
[05:33:59] <abaumann> The situation is not ideal, I know..
[05:34:11] <abaumann> mympd causes insanity, ok, will fix.
[05:35:50] <abaumann> also, intensions should only be executed with . ../lib/mysql-functions _and_ . ../lib/common-functions sources before into the active shell before calling them. I forgot and some rsync and copy scripts were not executed, most likely causing inconsistencies on the mirror and/or archive
[05:37:07] <KitsuWhooa> abaumann: there's no reason for shims when new packages are in staging
[05:37:20] <KitsuWhooa> We just need something that pushes evreything that depends on it
[05:37:32] <abaumann> the buildmaster? ;-)
[05:37:35] <KitsuWhooa> yes
[05:37:38] <abaumann> it does.
[05:37:44] <KitsuWhooa> I mean, I can write a bodge
[05:37:53] <abaumann> but so many things don't build, that it decedices not to push icu forever.
[05:37:54] <KitsuWhooa> and graft it on top of your pushall :p
[05:38:07] <KitsuWhooa> yeah
[05:38:09] <abaumann> and then yoy rebuild a package against a new version of icu in staging (like samba)
[05:38:09] <KitsuWhooa> I understand
[05:38:14] <abaumann> and push it to stable
[05:38:15] <KitsuWhooa> yup
[05:38:17] <abaumann> and it doesn't work.
[05:38:17] <KitsuWhooa> and it doesn't work
[05:38:18] <abaumann> :-)
[05:38:30] <abaumann> then you pick the lesser of two evils.
[05:38:42] <abaumann> I should have written it into the chat, the fact that I pushed icu
[05:39:01] <KitsuWhooa> Yeah if you don't mind, please let me know when you push things
[05:39:17] <KitsuWhooa> I only pushed qt5-base, too
[05:39:20] <KitsuWhooa> I suspect there's way more
[05:39:20] <abaumann> exactly because we cannot distinguish between manual issues and issues in the buildmaster
[05:40:33] <abaumann> /usr/lib/libQt5Core.so.5.15.13 depends directoly on /usr/lib/libicui18n.so.74
[05:40:41] <KitsuWhooa> tasos@nightopia:~$ audacious
[05:40:41] <KitsuWhooa> Cannot mix incompatible Qt library (5.15.10) with this library (5.15.13)
[05:40:47] <KitsuWhooa> yeah we need to push the rest of Qt5 now
[05:40:53] <abaumann> I see no other way than to do a lddtree on some typical qt5 libraries and see, what has to be pushed.
[05:41:02] <KitsuWhooa> I was thinking an SQL query on the buildmaster
[05:41:10] <KitsuWhooa> it keeps track of dependencies
[05:41:11] <abaumann> and there are QT5 ABI markers which might croak then, so it's usually all of qt5-*
[05:41:16] <KitsuWhooa> yeah
[05:42:00] <abaumann> "IIRC pushing all of qt5-* and then see what breaks" was the way of the past IIRC. :-)
[05:42:11] <KitsuWhooa> okay sounds good
[05:42:27] <KitsuWhooa> but how do you do it without manually specifying every single package?
[05:43:36] <abaumann> I usually did a selection on the packages.archlinux32.org list, then some printf trickery to get a list of packages to db-update (in pushall script)
[05:43:42] <KitsuWhooa> ah
[05:44:04] <abaumann> semi-auto-magically ;-)
[05:45:08] <KitsuWhooa> I'll try to make SQL that resolves dependencies, at least on the first level
[05:45:19] <KitsuWhooa> so that we can do pushall python and have it push everything that depends on it
[05:45:26] <abaumann> sounds good
[05:45:56] <KitsuWhooa> easier said than done but I'll try(tm)
[05:47:30] <KitsuWhooa> but yes I need to fix audacious because I need my music :p
[05:47:40] <KitsuWhooa> are you going to pushall Qt5 or do you want me to do it
[05:48:30] <abaumann> just do it.
[05:48:40] <abaumann> but there is an insanity around mympd
[05:48:56] <KitsuWhooa> ah, okay
[05:49:03] <KitsuWhooa> honestly I'm glad it's not python-snappy again
[05:55:48] <abaumann> There is something wrong with the database: mympd-14.1.2-1.0-i686.pkg.tar.zst
[05:55:59] <abaumann> no clue now, what the problem is?
[05:56:10] <KitsuWhooa> want me to handle it?
[05:56:17] <abaumann> it's only in the pool/mympd-14.1.2-1.0-i686.pkg.tar.zst
[05:56:24] <abaumann> probably missing in extra-staging or so?
[05:56:31] <abaumann> let me give another try. :-)
[05:56:33] <KitsuWhooa> sure
[05:58:47] <KitsuWhooa> tasos@nightopia:~$ qalc
[05:58:47] <KitsuWhooa> qalc: /usr/lib/libstdc++.so.6: version `GLIBCXX_3.4.32' not found (required by qalc)
[05:58:47] <KitsuWhooa> qalc: /usr/lib/libstdc++.so.6: version `GLIBCXX_3.4.32' not found (required by /usr/lib/libqalculate.so.23
[05:58:52] <KitsuWhooa> ...soooo, are we pushing glibc
[05:59:05] <abaumann> urgh.
[05:59:07] <abaumann> yeah.
[06:02:31] <abaumann> The following packages from the master mirror are not tracked in the database or vice versa:
[06:02:34] <abaumann> package-file i686/extra-testing/mympd-14.1.2-1.0-i686.pkg.tar.zst
[06:02:42] <abaumann> I start to think, the entry in the mysql database is missing.
[06:02:57] <KitsuWhooa> is it supposed to be in testing?
[06:03:01] <abaumann> no clue
[06:03:42] * abaumann boldly removes the symlink
[06:04:10] <KitsuWhooa> there's no version 14 anywhere in the database
[06:04:25] <KitsuWhooa> 15 is in staging
[06:05:23] <abaumann> aha, so it complains about an orphaned package in pool itself (14.1.2)
[06:05:33] <abaumann> I'll move that one to the archive
[06:06:00] <abaumann> full sanity restored.. of the buildmaster.. and of myself ;-)
[06:06:15] <abaumann> so, go ahead with whatever you want to destroy now ;-)
[06:06:23] <KitsuWhooa> hahaha
[06:06:26] <KitsuWhooa> thanks :)
[06:06:53] <abaumann> you're welcome :-)
[06:10:39] <KitsuWhooa> MariaDB [buildmaster]> SELECT UNIQUE pkgname INTO OUTFILE '/tmp/pkgs.csv' from binary_packages WHERE pkgname LIKE 'qt5%';
[06:10:39] <KitsuWhooa> Query OK, 51 rows affected (0.001 sec)
[06:10:39] <KitsuWhooa> [master@buildmaster ~]$ cat /tmp/pkgs.csv
[06:10:40] <KitsuWhooa> cat: /tmp/pkgs.csv: No such file or directory
[06:10:42] <KitsuWhooa> ?????
[06:11:07] <abaumann> permissions?
[06:11:14] <KitsuWhooa> maybe, but it said OK
[06:11:22] <abaumann> it's mysql, remember
[06:11:53] <KitsuWhooa> maybe systemd does tmp isolation?
[06:13:10] <KitsuWhooa> nope. nothing in there
[06:13:22] <KitsuWhooa> okay I have no idea how one runs a query and stores it to a file
[06:13:29] <KitsuWhooa> I always used HeidiSQL
[06:15:44] <abaumann> lol, that's a thing.. :-) heidisql
[06:16:01] <KitsuWhooa> there's also meowsql which I've been meaning to try
[06:16:06] <KitsuWhooa> but so far I just ran heidi in qine
[06:16:08] <KitsuWhooa> *wine
[06:16:39] <KitsuWhooa> oaky I'll do something else, I'll just ls in the mirror and parse that
[06:16:42] <KitsuWhooa> *okay
[06:16:47] <abaumann> if you start mysql as master and wrote to /home/master/work?
[06:17:17] <KitsuWhooa> it's not worth the effort honestly :p
[06:17:20] <abaumann> urgh, building pacman-static with LTO creates internal gcc errors, nice.
[06:17:32] <KitsuWhooa> out of memory?
[06:17:41] <KitsuWhooa> try options+=(!debug)
[06:17:47] <abaumann> no, more like gcc people should test their own optimizers..
[06:17:50] <abaumann> !lto rather
[06:17:50] <phrik> abaumann: Eh?
[06:17:51] <KitsuWhooa> oh :p
[06:18:08] <KitsuWhooa> well, I've gotten internal gcc errors before because the linker ran out of memory due to debug symbols
[06:18:36] <abaumann> lemme see which magic flags help here. might well be, I need both. :-)
[06:20:11] <KitsuWhooa> always start with !debug when something doesn't build
[06:20:21] <abaumann> oh, it built with !lto
[06:20:25] <abaumann> fine.
[06:20:39] <abaumann> compiler/optimizer bug in gcc 14
[06:20:42] <KitsuWhooa> I'm just saying that it's more important to have lto than debug symbols
[06:21:11] <KitsuWhooa> which get thrown out anyway on arch32
[06:21:17] <abaumann> Here I disagree. LTO goes into huge build times, out of memory etc. for what, some percents of faster runtime?
[06:21:27] <abaumann> debug symbols are really handy if things don't work :-)
[06:21:33] <KitsuWhooa> but we throw them out!
[06:21:37] <abaumann> yes
[06:21:50] <KitsuWhooa> I wouldn't say it's really long build times
[06:21:51] <abaumann> but on x86_64 I would expect lto to just work
[06:22:03] <KitsuWhooa> oh this is on amd64?
[06:22:05] <KitsuWhooa> yeah that's broken then
[06:22:26] <abaumann> all those 20 lto1 processes I often see when I log in to build machines tell me, that LTO is way inefficient.
[06:22:27] <KitsuWhooa> but LTO is also important because it catches bugs in the code that the compiler would've otherwise not seen
[06:22:51] <abaumann> You can achieve the same with amalgamation (cat *.c | gcc)
[06:23:07] <abaumann> I have nothing per se against LTO as an idea.
[06:23:18] <abaumann> I just think, the current implementation has ways for improvement..
[06:23:28] <KitsuWhooa> amalgamation is a pain tbh
[06:23:42] <KitsuWhooa> in fact I'd say it's slower
[06:23:46] <KitsuWhooa> but anyway
[06:23:48] <KitsuWhooa> here's my hack
[06:23:48] <abaumann> It is also way simpler to do something like LTO with a nice object format and a sane language. C and ELF are not exactly ideal for that. :-)
[06:23:54] <KitsuWhooa> for now
[06:23:55] <KitsuWhooa> [mirror@buildmaster extra-staging]$ ls | grep -o "qt5-.*[a-z]-" | sed s/-$// | uniq | tr '\n' ' '
[06:23:55] <KitsuWhooa> qt5-3d qt5-charts qt5-connectivity qt5-datavis3d qt5-declarative qt5-feedback qt5-gamepad qt5-graphicaleffects qt5-imageformats qt5-location qt5-lottie qt5-multimedia qt5-networkauth qt5-purchasing qt5-quick3d qt5-quickcontrols qt5-quickcontrols-nemo qt5-quicktimeline qt5-remoteobjects qt5-script qt5-scxml qt5-sensors qt5-serialbus qt5-serialport qt5-speech qt5-svg qt5-systems qt5-tools
[06:23:55] <KitsuWhooa> qt5-virtualkeyboard qt5-wayland qt5-webchannel qt5-webengine qt5-webglplugin qt5-websockets qt5-webview qt5-x11extras qt5-xmlpatterns
[06:24:06] <abaumann> sounds good
[06:24:08] <KitsuWhooa> time to pushall those
[06:24:16] <KitsuWhooa> I wonder if we have to do the same with qt6
[06:24:21] <KitsuWhooa> I don't have any qt6 programs to test
[06:24:25] <abaumann> no worries, qt6 hardly builds..
[06:24:32] <KitsuWhooa> it does build
[06:24:33] <KitsuWhooa> :p
[06:24:36] <abaumann> wow
[06:24:56] <KitsuWhooa> there's a few issues with some modules but I'd say it's very usable
[06:30:24] <KitsuWhooa> so for glibc, do I push it on its own, or do I push something else with it?
[06:31:07] <abaumann> libc should be backwards compatible, libc++ might break some things.
[06:31:16] <abaumann> but that's hard to tell in advance
[06:31:20] <KitsuWhooa> I see, okay
[06:31:33] <KitsuWhooa> I was worried about pushing glibc and then suddenly the system not booting
[06:31:48] <abaumann> aeh.
[06:31:50] <abaumann> hold on.
[06:31:59] <KitsuWhooa> I'm still waiting for qt5 to be pushed so I'm not doing anything
[06:32:05] <abaumann> did you say libstdc++ had a ABI error or glibc?
[06:32:17] <KitsuWhooa> glibc
[06:32:25] <KitsuWhooa> wait no
[06:32:27] <KitsuWhooa> glibcxx
[06:32:29] <abaumann> /usr/lib/libstdc++.so.6: version `GLIBCXX_3.4.32'
[06:32:30] <abaumann> ah
[06:32:35] <KitsuWhooa> qalc: /usr/lib/libstdc++.so.6: version `GLIBCXX_3.4.32' not found (required by /usr/lib/libqalculate.so.23)
[06:32:49] <abaumann> that's gcc-libs
[06:32:59] <abaumann> so you have to push gcc
[06:33:15] <abaumann> and gcc-libs
[06:33:18] <KitsuWhooa> welp
[06:33:30] <KitsuWhooa> nice catch
[06:33:36] <KitsuWhooa> I didn't notice it
[06:33:37] <abaumann> yeah, the GLIBCXX is a highly missleading clue here. :-)
[06:34:03] <abaumann> well glibc as said is highly backward compatible, nothing too serious would have happened. :-)
[06:34:21] <abaumann> but it wouldn't have solved the problem.
[06:34:22] <KitsuWhooa> I see
[06:34:31] <KitsuWhooa> doing gcc and gcc-libs now
[06:34:51] <KitsuWhooa> audacious is up and running with the new qt5, nice
[06:34:56] <abaumann> sometimes they depend on the specific binutils version, or the whole toolchain as a matter of fact.
[06:35:01] <abaumann> nice
[06:36:26] <KitsuWhooa> right, updating to gcc14 now
[06:36:27] <KitsuWhooa> fingers crossed
[06:36:30] <abaumann> actually, after changing gcc/binutils/glibc/clang one would have to rebuild all native packages, this is the only guarantee to catch everything
[06:36:37] <abaumann> but that's hardly feasable.
[06:36:42] <KitsuWhooa> yeah it's not
[06:37:19] <abaumann> people building kernels with source code modules might be unhappy about gcc 14 in stable now.
[06:37:26] <abaumann> but let's see
[06:38:04] <KitsuWhooa> I was going to update xorg-server (and try to add a patch workaround for my ddx) but it doesn't build under gcc14
[06:38:15] <KitsuWhooa> so now I need to see if anyone fixed it, and if not, fix it upstream
[06:38:17] <abaumann> uh
[06:40:04] <KitsuWhooa> also the patch we have no longer applies as it's been applied upstream, so I'll remove that too
[06:41:05] <KitsuWhooa> tasos@nightopia:~$ qalc
[06:41:05] <KitsuWhooa> > 1+1
[06:41:05] <KitsuWhooa> 1 + 1 = 2
[06:41:06] <KitsuWhooa> nice nice
[06:41:55] <abaumann> advanced math is working, nice ;-)
[06:42:11] <KitsuWhooa> at this point I don't even remember why I wanted qalc yesterday
[06:42:25] <KitsuWhooa> such is the life of a distro maintainer I guess :p
[06:44:18] <KitsuWhooa> also, question
[06:44:24] <KitsuWhooa> why do you have pushall write to a file and then source it?
[06:45:21] <KitsuWhooa> gmv
[06:45:26] <KitsuWhooa> oops, wrong terminal
[06:53:32] -!- buildmaster has quit [Read error: Connection reset by peer]
[06:54:53] <KitsuWhooa> uhhh
[06:54:55] <KitsuWhooa> ssh: connect to host buildmaster.archlinux32.org port 22: Connection refused
[06:54:55] <KitsuWhooa> ERROR: Unknown exit code 255 from 'get-assignment'.
[06:54:59] <KitsuWhooa> what happened?
[06:55:51] <KitsuWhooa> oh, someone rebooted it
[06:56:19] <KitsuWhooa> ...while pushall was running
[06:57:02] <KitsuWhooa> abaumann: was it you?
[06:59:18] <KitsuWhooa> anyway, yeah libQt6Core.so.6 => /usr/lib/libQt6Core.so.6
[06:59:18] <KitsuWhooa> libicui18n.so.73 => None
[07:00:03] <abaumann> nope
[07:00:17] <KitsuWhooa> does it reboot on its own?
[07:00:26] <abaumann> when a shutdownasap was running..
[07:00:42] <abaumann> but it checks whether people are still logged in and whether the buildmaster is doing something important.
[07:00:47] <abaumann> so, it shouldn't have booted.
[07:00:59] <abaumann> [root@buildmaster ~]# reboot
[07:01:02] <abaumann> wrong termintal
[07:01:04] <abaumann> darn.
[07:01:06] <abaumann> so it was me.
[07:01:07] <KitsuWhooa> welp :p
[07:01:16] <abaumann> darn.
[07:01:19] <KitsuWhooa> I just hope the db didn't get messed up
[07:01:36] <abaumann> let's hope not.
[07:02:00] <abaumann> I had some issues with a KVM switch, then typed reboot on a alix 1e machine.
[07:02:05] <abaumann> nothing happened there.
[07:02:15] <abaumann> so now I know where this 'reboot' went. :-(
[07:02:24] <KitsuWhooa> hahaha
[07:02:31] <abaumann> no sorry. really.
[07:02:37] <abaumann> that was completely unnecessary..
[07:02:39] <KitsuWhooa> it's okay, don't worry about it
[07:02:50] <KitsuWhooa> no need to apologise
[07:03:05] <abaumann> running a sanity check on the build master. fingers crossed.
[07:03:32] <abaumann> actually, that is what those intension logs in work are for..
[07:03:43] <KitsuWhooa> yup
[07:03:47] <abaumann> ..they protocoll every db and file system operation..
[07:04:30] <abaumann> it could have been a 'poweroff'
[07:04:53] <abaumann> that would be way worse.. then I would have to find out how to bring this machine up again via remote console. :-)
[07:05:48] <abaumann> sanity looks ok
[07:06:10] <KitsuWhooa> awesome
[07:06:38] <KitsuWhooa> as logn as you have access to the hetzner web UI it should be a matter of pressing a button
[07:06:40] <KitsuWhooa> but now I wonder
[07:06:47] <KitsuWhooa> should we somehow mask poweroff/shutdown operations?
[07:07:05] <abaumann> I have a console at Hetzener
[07:07:10] <abaumann> I just didn't use it in ages. :-)
[07:07:13] <KitsuWhooa> ah :p
[07:07:38] <KitsuWhooa> okay, pushall is running for qt6
[07:08:00] <abaumann> I logged out from the buildmaster :-)
[07:12:27] <KitsuWhooa> I masked systemd-poweroff
[07:12:30] <KitsuWhooa> so now poweroff just complains
[07:12:59] <abaumann> should I test? ;-)
[07:13:43] <KitsuWhooa> sure :p
[07:14:06] <KitsuWhooa> (I did the same as a test on my computer locally and it worked)
[07:14:37] <KitsuWhooa> tasos@nightopia:~$ LANGUAGE=en poweroff
[07:14:37] <KitsuWhooa> Call to PowerOff failed: Unit systemd-poweroff.service is masked.
[07:14:37] <KitsuWhooa> Failed to talk to init daemon: Permission denied
[07:14:49] <abaumann> ok. sounds good.
[07:18:52] -!- zxrom has joined #archlinux32
[07:19:06] <KitsuWhooa> qt6 works too
[07:19:08] <KitsuWhooa> at least xpdf does
[07:20:04] <KitsuWhooa> ah, looks like e89edec497bac581ca9b614fb00c25365580f045 needs to be backported for X
[07:24:55] <KitsuWhooa> yup that fixed it
[07:38:53] -!- zxrom has quit [Quit: Leaving]
[07:41:05] -!- zxrom has joined #archlinux32
[08:17:35] <abaumann> so, have to catch a train.. cu. :-)
[08:17:37] -!- abaumann has quit [Quit: leaving]
[08:21:42] <KitsuWhooa> see ya
[08:21:46] <KitsuWhooa> DB awaits
[08:21:47] <KitsuWhooa> :p
[08:41:01] <KitsuWhooa> I'm going to force a rebuild of all xf86-*
[08:41:16] <KitsuWhooa> also, what do you think of adding the xf86- drivers that upstream dropped?
[08:41:33] <KitsuWhooa> building xf86-video-savage every time is getting annoying
[09:30:09] <KitsuWhooa> aaand buildmaster went insane
[09:31:44] <KitsuWhooa> Oh whoops. It's because I ^Ced a script
[09:37:44] -!- drathir_tor has quit [Remote host closed the connection]
[10:19:23] -!- lithiumpt has quit [Quit: ZNC 1.9.0 - https://znc.in]
[10:19:44] -!- lithiumpt has joined #archlinux32
[10:23:03] -!- drathir_tor has joined #archlinux32
[10:47:11] -!- abaumann has joined #archlinux32
[10:47:38] <abaumann> KitsuWhooa: yes, we can do that, adding the xf86-X drivers I mean. To extra. We did that for pcmciautils.
[10:47:49] -!- abaumann has quit [Client Quit]
[10:58:32] -!- zxrom has quit [Ping timeout: 260 seconds]
[11:01:05] <KitsuWhooa> Great
[11:24:38] -!- zxrom has joined #archlinux32
[13:51:02] -!- drathir_tor has quit [Remote host closed the connection]
[13:51:38] -!- drathir_tor has joined #archlinux32
[16:28:49] -!- mvchtz has quit [Quit: WeeChat 3.5]
[16:52:06] -!- ssserpent has quit [Quit: WeeChat 4.2.2]
[16:58:38] -!- mvchtz has joined #archlinux32
[18:06:27] -!- kiska31 has joined #archlinux32
[18:08:05] -!- kiska3 has quit [Ping timeout: 240 seconds]
[18:08:05] kiska31 is now known as kiska3
[22:22:34] -!- zxrom has quit [Quit: Leaving]
[22:35:22] -!- AtleoS has quit [Ping timeout: 256 seconds]
[22:35:40] -!- AtleoS has joined #archlinux32