#archlinux32 | Logs for 2022-02-20

Back
[00:28:08] -!- buildmaster has quit [Remote host closed the connection]
[00:33:06] -!- buildmaster has joined #archlinux32
[01:40:17] -!- T`aZ has quit [Ping timeout: 272 seconds]
[01:41:45] -!- T`aZ has joined #archlinux32
[04:50:03] -!- T`aZ has quit [Ping timeout: 256 seconds]
[04:51:57] -!- T`aZ has joined #archlinux32
[05:04:13] -!- T`aZ has quit [Ping timeout: 256 seconds]
[05:06:14] -!- T`aZ has joined #archlinux32
[08:58:32] -!- T`aZ has quit [Ping timeout: 256 seconds]
[09:00:04] -!- T`aZ has joined #archlinux32
[09:35:05] -!- T`aZ has quit [Ping timeout: 256 seconds]
[09:36:45] -!- T`aZ has joined #archlinux32
[10:20:11] -!- abaumann has joined #archlinux32
[10:20:41] <buildmaster> Hi abaumann!
[10:20:41] <buildmaster> !rq abaumann
[10:20:42] <phrik> buildmaster: <abaumann> vvrng: np, we are all noobs, it depends on the environment we are in. :-)
[10:21:26] <abaumann> deep42thought: ah, ok then, pyopenssl, if it is upstream, no problem. It seems just to pop up very fast on build slaves, other builds are not affected by it. :-)
[10:26:42] <abaumann> So, toolchain rebuilding is ok, I'll push the toolchain to testing and this time I check on real hardware first, whether glibc is still working.. :->
[10:32:53] -!- T`aZ has quit [Ping timeout: 256 seconds]
[10:34:43] -!- T`aZ has joined #archlinux32
[10:38:13] <abaumann> "Your buildslave "eurobuild6-5" uploaded a package with a not fully-trusted signature" why that now?
[10:39:07] <abaumann> sub rsa2048 2017-11-02 [E] [expires: 2023-01-31]
[10:39:20] <abaumann> sorry, I simply don't get this PGP stuff.. it's all weird..
[10:40:27] <abaumann> my signing key is _not_ expired..
[10:40:43] <abaumann> ..maybe my master key?
[10:41:16] <abaumann> sub rsa2048 2017-08-29 [E] [expires: 2022-09-02]
[10:41:17] <abaumann> nope
[10:41:42] <abaumann> I start to understand why OBS has a dedicated sign-server..
[10:43:44] <abaumann> Is the buildmaster checking signatures with old keys?
[11:02:59] <abaumann> there was a gcc10-libs still around upstream, but gcc10 has been removed?
[11:07:24] <abaumann> "You do not build anything currently - abort whatever you are doing", yeah..
[11:08:05] <abaumann> It's really hard sometimes just to test something, like "are the signing keys ok in a build slave"
[11:09:08] <abaumann> mmh. per default the "MATE terminal" is black on black, completely unreadable.. :-)
[11:14:25] <abaumann> so, no more signing errors..
[11:18:19] <abaumann> and back again to signature errors.. *grmbl*
[11:20:02] <abaumann> ok, I don't know what to do: I prolongued my master and my package signing key, I published both to keyservers, I manually imported those keys, I refreshed GPG on the buildmaster (master user) and on all slaves..
[11:21:10] <abaumann> singing should not be part of the build process, I would rather use a dedicated signer, this also allows to resign all packages without rebuilding them.. we currently still have packages with wrong signatures, I believe.
[11:21:57] <abaumann> ..especially as more and more packages refuse to rebuild easily, but are still installable and runable.
[11:23:12] <abaumann> building with build-package works, not in a systemd service.. mmh..
[11:58:20] <girls> Hi abaumann, all good?
[12:07:19] <abaumann> hi.
[12:07:29] <abaumann> not with my keys, no clue, really..
[12:07:39] <abaumann> I'm building outside systemd for now..
[12:12:16] <abaumann> just testing the new toolchain on new hardware, painfully slow.. :-)
[12:12:30] <abaumann> sed s/new/real/
[12:12:53] <abaumann> I don't want to fix ISA errors for the next weeks..
[12:16:11] <abaumann> weird gnome-shell complains about libcanberra-pulse=0.30+2+gc0620e4 but there is a package with that number in the pool..
[12:24:26] <abaumann> seems like an alias for provides libcanberra-pulse is not working, gnome-shell just builds fine when using libcanberra as dependency directly
[12:36:33] <girls> be careful: the provides=() alias also can have a pkgver
[12:36:47] <girls> if that is absent, and the depends has a pkgver, it will *not* match
[12:45:00] -!- T`aZ has quit [Ping timeout: 240 seconds]
[12:46:57] -!- T`aZ has joined #archlinux32
[13:48:36] -!- volki57 has quit [Ping timeout: 240 seconds]
[13:50:40] -!- volki57 has joined #archlinux32
[14:16:43] <abaumann> ah, Gnome, *sigh*
[16:08:15] <abaumann> "Your buildslave "eurobuild6-1" uploaded a package with a not fully-trusted signature", ah, now it happens manually too..
[16:15:34] <abaumann> it's amazing, this sounds random, almost like a time synchronization issue..
[16:16:16] <abaumann> ..I also hat ssh_key issues yesterday, with random failures when logging in into the buildmaster or when script were trying to connect to it.
[16:17:19] <abaumann> let's put a proper NTP onto the buildmaster :-)
[16:21:38] <abaumann> "Signature(s) is/are not fully trusted", ok.
[16:21:49] <abaumann> must be something else
[16:22:35] <abaumann> 16194A82231E9EF823562181C8E8F5A0AF9BA7E7 is the signing key, rsa2048 2017-11-02 [E] [expires: 2023-01-31]
[16:28:06] <abaumann> --homedir /etc/pacman.d/gnupg
[16:28:07] <abaumann> mmh.
[16:28:31] <abaumann> archlinux32-keyring 20220209-1
[16:29:12] <abaumann> [ full ] Andreas Baumann (sign) <mail@andreasbaumann.cc> 2017-11-02 [E] [expires: 2023-01-31]
[16:29:19] <abaumann> this starts to get really annoying..
[16:31:11] <abaumann> return-assignment swallows all output, this is not handy :-(
[17:15:16] -!- drathir_tor has quit [Ping timeout: 240 seconds]
[17:25:24] -!- ejjdhfjsu has quit [Ping timeout: 240 seconds]
[17:30:21] -!- drathir_tor has joined #archlinux32
[17:36:42] <abaumann> this is simply madening, I can not make my keys work again..
[17:37:00] <abaumann> uid [ultimate] Andreas Baumann (sign) <mail@andreasbaumann.cc>
[17:37:13] <abaumann> On the buildmaster, where it checks the key, in /etc/pacman.d/gnupg
[17:37:20] <abaumann> even that doesn't help
[17:40:59] -!- drathir_tor has quit [Remote host closed the connection]
[17:41:09] <abaumann> So, some empathy is necessary, if people are complaining about keys in the forums.. the GPG key system is old, hard to use, hard to unserstand and - I would argue - fails completely to be intuitive or easy to use..
[17:41:35] <abaumann> my slaves are down till I understand what's going wrong here..
[17:41:36] -!- abaumann has quit [Quit: leaving]
[17:41:47] -!- drathir_tor has joined #archlinux32
[17:49:55] -!- abaumann has joined #archlinux32
[17:49:55] <buildmaster> Hi abaumann!
[17:49:55] <buildmaster> !rq abaumann
[17:49:56] <phrik> buildmaster: <abaumann> who would have thought digital books burn too. ;-)
[17:50:56] -!- volki57 has quit [Ping timeout: 240 seconds]
[17:53:50] -!- volki57 has joined #archlinux32
[18:08:36] -!- drathir_tor has quit [Ping timeout: 240 seconds]
[18:21:14] <abaumann> tar: Cowardly refusing to create an empty archive
[18:21:14] <abaumann> Try 'tar --help' or 'tar --usage' for more information.
[18:21:15] <abaumann> /home/slave1/builder/work/tmp.build-packages.Ehbve3/tar-content.lWsRsS/
[18:21:15] <abaumann> drwxr-xr-x 4,734,976 2022/02/20 13:42:24 .
[18:21:15] <abaumann> -rw-r--r-- 4,008,726 2022/02/20 04:07:02 aarch64-linux-gnu-binutils-2.38-1.0-pentium4.pkg.tar.zst
[18:21:18] <abaumann> -rw-r--r-- 310 2022/02/20 04:07:02 aarch64-linux-gnu-binutils-2.38-1.0-pentium4.pkg.tar.zst.sig
[18:22:01] <abaumann> The signatures are empty because the script is deleting the package before signing?
[18:23:19] <abaumann> removing unexpected build artifact "expat-2.4.4-1.0-i686.pkg.tar.zst"
[18:23:33] <abaumann> so, yeah, but why do I _get_ expat then as a build assignment?
[18:26:48] -!- drathir_tor has joined #archlinux32
[18:27:15] <abaumann> This doesn't make any sense to m/quit
[18:27:16] -!- abaumann has quit [Quit: leaving]
[18:40:45] -!- buildmaster has quit [Excess Flood]
[18:41:02] -!- buildmaster has joined #archlinux32
[18:46:12] -!- abaumann has joined #archlinux32
[18:46:12] <buildmaster> Hi abaumann!
[18:46:13] <buildmaster> !rq abaumann
[18:46:13] <phrik> buildmaster: <abaumann> It's generally more important to not interrupt building of packages than to produce working packages.. :-)
[18:46:27] <abaumann> Now eurobuild6-1 is producing working packages again..
[18:46:36] <abaumann> ..I simply don't understand this system..
[18:46:46] <abaumann> ..anyway: I pushed the new toolchain to master.. have fun. :-)
[18:47:04] <abaumann> It didn't break and I compiled test programs on realy hardware on i486, i686 and pentium4.
[18:48:39] -!- abaumann has quit [Client Quit]
[18:51:31] -!- abaumann has joined #archlinux32
[19:08:36] -!- volki57 has quit [Ping timeout: 240 seconds]
[19:08:36] -!- GNUtoo has quit [Ping timeout: 240 seconds]
[19:08:56] -!- drathir_tor has quit [Ping timeout: 240 seconds]
[19:09:16] -!- jonathon has quit [Ping timeout: 240 seconds]
[19:10:00] -!- ejjdhfjsu has joined #archlinux32
[19:24:49] -!- volki57 has joined #archlinux32
[19:24:55] -!- jonathon has joined #archlinux32
[19:25:43] -!- GNUtoo has joined #archlinux32
[19:29:44] -!- drathir_tor has joined #archlinux32
[20:07:01] <girls> you probably need(ed) to update the archlinux32-keyring on the buildmaster
[20:11:59] <abaumann> yep. exactly. :-)
[20:40:35] <girls> if it removes the built expat, the reason is, that the buildmaster expects something else (e.g. a different pkgname, a different version or a different architectures)
[20:40:51] <girls> The only way to investigate this, is to have a look into the buildmaster's database
[20:40:56] <girls> ... I think
[20:45:24] <abaumann> The funny thing is, that it disappeared afterwards..
[20:46:49] <girls> I added some verbosity to the script
[20:46:59] <girls> it should now tell, what package *was* expected
[20:47:16] <girls> probably, the buildmaster had a different idea than what was in the PKGBUILD
[20:47:35] <girls> i.e. `makepkg --printsrcinfo` "lied"
[20:49:45] <abaumann> ah, that's a neat extension :-)
[20:59:00] -!- eschwartz has quit [Ping timeout: 250 seconds]
[21:20:52] <abaumann> this with architecture any starts to get a problem. gnome-video-effects has a working gstreamer on pentium4, but a broken one on i686. So it's now random, whether this package is actually being built for which any..
[21:22:16] <girls> any should be built on pentium4, IIRC
[21:22:24] <abaumann> ah.
[21:22:25] <abaumann> warning: cannot resolve "svt-hevc", a dependency of "gst-plugins-bad"
[21:22:41] <abaumann> this I get also on pentium. though gstreamer provides a gst-plugins-bad
[21:22:53] <abaumann> I start to think something is broken in this provides-mechanism..
[21:22:57] <girls> but we could consider "building" any by copying the upstream package
[21:23:47] <abaumann> but then the artifacts have to be "binary" equivalent.
[21:24:10] <girls> they should be, if it's "any"
[21:24:12] <abaumann> not building them violates a little bit reproducability of build?
[21:24:37] <girls> we could build them on x86_64, yeah
[21:24:47] <abaumann> what if you have a header file with "#ifdef _pentium4_cpu_" or so and this produces an error only under certain cpus?
[21:25:00] <abaumann> I don't agree with upstream to declare header-only libraries as 'any'
[21:25:40] <girls> they do that?
[21:26:07] <abaumann> if a header library contains only header files (this is called header-only-library, I now, even the name is silly, but it is a thing in C++)
[21:26:17] <abaumann> this is considered any, not x86_64
[21:26:34] <abaumann> there was a bug..
[21:28:40] <abaumann> ..libcuckoo.. 73732
[21:28:44] <abaumann> https://bugs.archlinux.org
[21:28:45] <phrik> Title: FS#73732 : [libcuckoo] any architecture (at bugs.archlinux.org)
[21:35:05] -!- abaumann has quit [Quit: leaving]
[21:39:28] -!- eschwartz has joined #archlinux32
[23:28:56] -!- drathir_tor has quit [Ping timeout: 240 seconds]
[23:34:53] -!- drathir_tor has joined #archlinux32