#archlinux32 | Logs for 2018-06-08

Back
[00:04:07] -!- Cthulu201 has quit [Ping timeout: 256 seconds]
[00:05:16] -!- Cthulu201 has joined #archlinux32
[00:15:22] -!- bill-auger has quit [Ping timeout: 264 seconds]
[00:56:03] -!- bill-auger has joined #archlinux32
[03:21:30] -!- NoobAlice has quit [Ping timeout: 260 seconds]
[03:21:37] -!- NoobAlice1 has joined #archlinux32
[06:21:46] -!- brosame has joined #archlinux32
[06:23:11] -!- brosame has parted #archlinux32
[06:33:31] -!- NoobAlice1 has quit [Quit: Leaving.]
[06:56:15] <buildmaster> girls, my database is dirty again ...
[07:08:42] -!- eduardoeae has quit [Ping timeout: 260 seconds]
[07:58:43] -!- oaken-source has joined #archlinux32
[08:08:53] -!- oaken-source has quit [Ping timeout: 265 seconds]
[08:23:05] -!- oaken-source has joined #archlinux32
[08:29:47] -!- oaken-source has quit [Ping timeout: 245 seconds]
[08:34:32] <abaumann> deep42thought: tsan is not that important, only a plugin detecting some thread races..
[08:34:40] -!- deep42thought has joined #archlinux32
[08:34:41] <buildmaster> Hi deep42thought!
[08:34:51] <deep42thought> abaumann: ok, so we just skip that in package()?
[08:35:06] <abaumann> I would actually like to see, why it fails to build..
[08:35:17] <deep42thought> ok
[08:36:17] <abaumann> in haskell: is simply not testing an option?
[08:36:34] <abaumann> I mean, in those modules forming a cycle
[08:36:54] <deep42thought> yes, that would be an option
[08:37:15] <deep42thought> but since the build differs for --enable-tests and --disable-tests, it may cause problems for users needing the tests
[08:37:26] <abaumann> yeah. this is poor ABI design.
[08:37:34] <abaumann> test code affecting the final artifacts.
[08:37:58] <deep42thought> who knows if --disable-tests does not introduce some bugs?
[08:38:10] <abaumann> I understand too little of haskell to judge.
[08:44:40] * buildmaster resumes sanity.
[08:48:11] <deep42thought> but anyway, disabling tests for some modules might be the best solution, I think
[10:30:20] <abaumann> 8.1.1+20180531-1: install lsan and tsan libraries: ok, disabling them for now is a good option. Some tests may break if they use the sanitizers to check the code.
[10:30:26] <abaumann> I'll change accordingly.
[10:30:59] <deep42thought> ok
[10:31:09] <deep42thought> this might pull the plug from the clogged build list :-)
[10:31:18] <abaumann> exactly.
[10:32:16] <deep42thought> btw: I'm about to fork pacman to accept different architectures for installation
[10:32:24] <deep42thought> e.g. i486 on i686
[10:32:32] <deep42thought> but more importantly i686 on pentium3
[10:32:56] <abaumann> C coding. fun. :-)
[10:33:08] <deep42thought> that's at least something I (partially) understand
[10:33:20] <deep42thought> ... it's close enough to my native language, pascal
[10:33:56] <abaumann> ui. don't tell that any Pascal/Modula/Oberon guy. :-)
[10:34:32] <deep42thought> well, it's definitely closer than haskell, python, php, mysql, ...
[10:38:17] <abaumann> mmh: in our PKGBUILD there is a 'sed s@libsanitizer/{a,l,ub,t}san@libsanitizer/{a,l,ub}san@'
[10:38:47] <deep42thought> maybe it was there for i686 upstream before?
[10:38:50] <deep42thought> let me check
[10:39:32] <abaumann> ah. it removes tsan
[10:39:38] <deep42thought> yes
[10:39:42] <deep42thought> question is: why
[10:40:36] <deep42thought> yep: tsan was not there for i686 upstream either
[10:40:44] <deep42thought> if [[ $CARCH == "x86_64" ]]; then
[10:40:44] <deep42thought> make -C $CHOST/libsanitizer/tsan DESTDIR="$pkgdir" install-toolexeclibLTLIBRARIES
[10:40:44] <deep42thought> fi
[10:40:49] <abaumann> "propegate the emptydir option, libtsan.a is x86_64 only"
[10:41:29] <abaumann> then it's safe to remove it.
[10:42:00] <deep42thought> maybe we can add a comment
[10:43:55] <deep42thought> man, the upload from the build master to the master mirror takes like forever O.o
[11:00:18] <buildmaster> girls, my database is dirty again ...
[11:14:26] * buildmaster resumes sanity.
[11:16:36] <buildmaster> deep42thought: Your buildslave "rechenknecht" builds some outdated package.
[11:17:32] <deep42thought> thx, buildmaster
[11:17:32] <buildmaster> np
[11:46:36] -!- eduardoeae has joined #archlinux32
[11:56:29] <abaumann> deep42thought: it seems more build slaves get into an endless loop, so it would not be the upload.
[11:56:32] <abaumann> running namcap ... ok.
[11:56:34] <abaumann> smoothen namcap log ... ok.
[11:56:37] <abaumann> tar: Cowardly refusing to create an empty archive
[11:56:40] <abaumann> Try 'tar --help' or 'tar --usage' for more information.
[11:56:48] <deep42thought> umm
[11:56:53] <deep42thought> that might be a different issue
[11:56:57] <deep42thought> which package is that?
[11:57:03] <deep42thought> or which build slave?
[11:57:33] <abaumann> mine: bison/eurobuild3
[12:00:41] <deep42thought> I'll have a look at this after lunch
[12:01:07] <abaumann> yeah. ditto. enjoy. :-)
[12:01:10] <deep42thought> did you update the build scripts on your slave?
[12:01:24] <deep42thought> the slaves now get told what packages to send to the master
[12:01:46] <abaumann> aha.
[12:01:48] <abaumann> I'll try.
[12:02:06] <buildmaster> bison is broken (says eurobuild3).
[12:05:30] <abaumann> no, I was not up to date, so let's see now :-)
[12:12:58] -!- Alina-malina has quit [Ping timeout: 265 seconds]
[12:19:21] <elibrokeit> deep42thought: you're adding new pacman features to do what?
[12:20:37] <elibrokeit> because pacman looks at the architecture value in pacman.conf and doesn't care what the host is, so just pacstrap with something other than "auto" (looks up the uname -m)
[12:30:34] -!- eworm has joined #archlinux32
[12:31:36] <eworm> We have broken dependencies in the repositories.
[12:32:00] <eworm> Please move xf86-video-amdgpu with new video ABI for xorg-server 1.20.0
[12:32:49] <abaumann> hi. xorg-server 1.20 is not yet in stable. there are good reasons for that. :-)
[12:33:14] <eworm> The website and repositories tell another story
[12:33:16] <abaumann> I get everything from sporadic segfaults to black screen or broken video driver.
[12:33:23] <abaumann> huh, really?
[12:33:26] <abaumann> let me check..
[12:33:43] <abaumann> https://news.archlinux32.org
[12:33:45] <phrik> Title:Archlinux 32 News (at news.archlinux32.org)
[12:34:13] <abaumann> https://packages.archlinux32.org
[12:34:13] <eworm> https://packages.archlinux32.org
[12:34:14] <phrik> Title:Arch Linux 32 - Package Search (at packages.archlinux32.org)
[12:34:15] <phrik> Title:Arch Linux 32 - Package Search (at packages.archlinux32.org)
[12:34:20] <abaumann> ah yes. that one shows staging.
[12:35:05] <eworm> xorg-server 1.20.0-5.0 moved at 2018-06-08 08:22:15
[12:35:41] <abaumann> oh.
[12:35:49] <abaumann> I see.
[12:36:27] <abaumann> So they hopped this morning.
[12:37:09] <eworm> but only the xorg-server packages
[12:37:19] <abaumann> and not the videos, yep.
[12:37:22] <eworm> driver packages with new ABI are still in testing
[12:37:44] <abaumann> so something is blocking their move.. let's ask deep43thought, he can trigger them manually.
[12:38:42] <abaumann> *deep42thought = deep43thought-1
[12:39:51] <abaumann> buildmaster: why don't you stabilize xf86-video-amdgpu?
[12:39:52] <buildmaster> Sry, "why-dont-you stabilize" is unavailable, until someone recodes it to look into the database.
[12:39:56] <deep42thought> sry
[12:40:00] <abaumann> hehe :-)
[12:40:03] <abaumann> I remember..
[12:40:20] <deep42thought> elibrokeit: I want to be able to install i486 packages on a i686 system, where otherwise i686 packages are installed
[12:40:34] <deep42thought> sry, abaumann, I moved packages >1week in testing to stable
[12:40:44] <deep42thought> db-update must have missed some packages in the process O.o
[12:40:46] <deep42thought> (bug)
[12:41:05] <abaumann> no problem. we just have to move all xorg stuff now.
[12:41:20] <abaumann> I'm more worried what will all not work with new xorg 1.20
[12:42:12] <deep42thought> :-/
[12:44:16] <deep42thought> elibrokeit: there is currently logic in pacman to allow install of "any" packages on any architecture
[12:44:28] <elibrokeit> is this bad???
[12:44:31] <deep42thought> no
[12:44:35] <deep42thought> I just want to extend it
[12:44:52] <elibrokeit> to?
[12:45:09] <deep42thought> we're planning for 3 architectures: i486 < i686 < pentium3
[12:45:26] <deep42thought> and I plan not to recompile packages for each architecture if it does not make a difference actually
[12:45:35] <abaumann> :: removing xorg-server breaks dependency 'xorg-server' required by lxdm
[12:45:42] <abaumann> really funny
[12:45:43] <deep42thought> abaumann: O.o
[12:45:55] <deep42thought> give the script some time, I'm moving stuff, currently
[12:45:58] <deep42thought> (slow as hell)
[12:46:01] <abaumann> ok.
[12:46:11] <abaumann> so it's a temporary glitch.
[12:46:23] <elibrokeit> deep42thought: then this allows installing packages which are genuinely incompatible, without being able to distinguish between them
[12:46:23] <deep42thought> you'll see more :-)
[12:46:45] <elibrokeit> I'd say just recompile them...
[12:46:45] <deep42thought> elibrokeit: how so?
[12:47:06] <deep42thought> ah, I get it
[12:47:07] <deep42thought> gmm
[12:47:09] <deep42thought> *hmmm
[12:47:44] <deep42thought> you mean, I can then install the i486 libs and the i686 binary on a system, where the i686 binary is actually linked against i686 libs, right?
[12:47:51] <elibrokeit> pacman -U https://repo.foo.com
[12:48:36] <elibrokeit> just because you cherry-pick what goes in the sync database, does not mean you actually control which architecture gets installed for which package
[12:48:50] <elibrokeit> these programmatic restrictions have reasons.
[12:49:53] <deep42thought> ok, I see
[12:50:06] <abaumann> I have a trace of my slave tar error:
[12:50:11] <abaumann> + find /data/arch32/builder/bin/../work/tmp.build-packages.FJIVHS/tar-content.TVE3xF/ -maxdepth 1 '(' -name '*.pkg.tar.xz-namcap.log' -o -name '*.pkg.tar.xz.so.needs' -o -name '*.pkg.tar.xz.so.provides' ')' -execdir gzip '{}' ';'
[12:50:15] <abaumann> ++ find /data/arch32/builder/bin/../work/tmp.build-packages.FJIVHS/tar-content.TVE3xF/ -maxdepth 1 '(' -name '*.pkg.tar.xz' -o -name '*.pkg.tar.xz.sig' -o -name '*.pkg.tar.xz-namcap.log.gz' -o -name '*.pkg.tar.xz.so.needs.gz' -o -name '*.pkg.tar.xz.so.provides.gz' ')' -printf '%f\n'
[12:50:20] <abaumann> + tar -cf package.tar -C /data/arch32/builder/bin/../work/tmp.build-packages.FJIVHS/tar-content.TVE3xF --
[12:50:23] <abaumann> tar: Cowardly refusing to create an empty archive
[12:50:25] <abaumann> Try 'tar --help' or 'tar --usage' for more information.
[12:50:28] <abaumann> + bail_out
[12:50:39] <abaumann> I'm on 63953bf1520b423fdc9048d8b2a2559bd45613d9 of builder
[12:50:44] <deep42thought> I see the problem
[12:50:53] <deep42thought> (on my end :-D)
[12:50:59] <abaumann> ah. good. :-)
[12:54:09] <deep42thought> get-package-update adds binary_packages with "sub_pkgrel_omitted", which leads to get-assignment believing, the sub_pkgrel will be omitted for the built package - and thus no matching packages are found
[12:54:44] <deep42thought> "`sub_pkgrel_omitted` bit(1) NOT NULL DEFAULT b'1'"
[12:54:51] <deep42thought> hmmm, which idiot dit write this ???
[12:57:11] <deep42thought> ok, should be fixed
[12:57:53] <abaumann> ok, I'm updating and testing
[12:59:22] <abaumann> ah. it's a change in the db.
[13:00:24] <deep42thought> elibrokeit: what do you think of this change: https://git.archlinux32.org
[13:00:25] <phrik> Title:scripts/makepkg.sh.in: install all dependencies simultanously, if possible - this avoids package conflicts in some cases · 54f10cc594 - Archlinux32 Gitea (at git.archlinux32.org)
[13:00:46] <deep42thought> I wrote this once on the bug tracker or mailing list (can't remember), but there was no reaction
[13:04:31] <elibrokeit> I'm trying to figure out how this error handing is intended here, to see if there's reasons, but I've got to go for a bit so I'll get back to you on that
[13:04:57] <deep42thought> ok, thx
[13:05:32] <deep42thought> actually there is no change to the error handling intended, only that $depends and $make/checkdepends are installed at the same time if possible
[13:15:59] <abaumann> ok. bison built. :-)
[13:16:04] <deep42thought> :-)
[13:16:34] <deep42thought> xorg moving is over
[13:16:39] <abaumann> ah. ok.
[13:16:46] <abaumann> so let's update some machines and test.
[13:16:54] * buildmaster failed to execute a mysql query - can you have a look at "tmp.mysql-functions.query.stdin.2018-06-08T13:16:53.EZawUv"?.
[13:17:16] <abaumann> "return-assignment" was running already.
[13:17:16] <abaumann> I could not complete a mysql query!
[13:17:16] <abaumann> I cannot determine, where this package (or some part of it)
[13:17:16] <abaumann> should be published:
[13:17:16] <abaumann> package-was-sent: bison-3.0.5-1.0-i686.pkg.tar.xz
[13:17:18] <abaumann> This is some internal error and not (necessarily) your fault.
[13:17:21] <abaumann> "return-assignment" was running already.
[13:17:23] <abaumann> oo.
[13:17:34] <abaumann> But.. not necessarily my fault. :-)
[13:17:35] <deep42thought> this is related to the last failure
[13:17:48] <deep42thought> well, it depends if the owner of the buildslave is deep42thought
[13:18:50] <deep42thought> ah, I misspelled a column in that query
[13:19:04] <deep42thought> I _was_ wondering if it was really called that way, when I wrote it :-/
[13:19:36] <deep42thought> should work, now
[13:20:25] <abaumann> ok.
[13:24:58] -!- abaumann has quit [Ping timeout: 264 seconds]
[13:33:39] -!- abaumann has joined #archlinux32
[13:33:39] <buildmaster> Hi abaumann!
[13:33:45] <deep42thought> Hi abaumann, all good?
[13:34:16] <abaumann> yes. I got kicked from the chatroom.
[13:34:17] <buildmaster> girls, my database is dirty again ...
[13:34:19] <abaumann> no clue why?
[13:34:33] <deep42thought> did you deauthenticate?
[13:34:53] <abaumann> nope.
[13:35:00] <abaumann> maybe a network glitch on my side.
[13:36:08] <deep42thought> can you pastebin me the upload log from your build slave, please?
[13:36:25] <deep42thought> there seems to be some bug during the upload of the package into staging
[13:37:35] <deep42thought> oh, I see it
[13:39:44] * buildmaster resumes sanity.
[13:50:58] <deep42thought> Vollzornbrot: your mirror is down, which breaks all of your build slaves
[13:53:47] <elibrokeit> deep42thought: "intended", I like how you say that. ;)
[13:55:19] <deep42thought> :-D
[13:57:37] <elibrokeit> The logic is broken
[13:57:43] <abaumann> huh. my slave?
[13:57:57] <deep42thought> my logic?
[13:58:39] <elibrokeit> deep42thought: in order to handle -sri, you must calculate the original deplist after runtime depends are installed
[13:59:03] <deep42thought> yes
[13:59:10] <deep42thought> this is, why it is in the first if clause
[13:59:22] <elibrokeit> you'll need a deeper refactor in order to get the right deplist every time, but only running it once
[13:59:42] <elibrokeit> hmm
[13:59:48] * elibrokeit checks again
[13:59:59] <deep42thought> I know, that with -i it won't install all deps at the same time
[14:00:20] <deep42thought> but this is not really feasible IMHO without refactoring all the dependency checking in pacman
[14:00:46] <elibrokeit> aha
[14:02:36] <elibrokeit> okay, yes, the patch looks right and makes sense
[14:02:42] <deep42thought> \o/
[14:04:35] <deep42thought> so: can/will you upstream this?
[14:06:54] <elibrokeit> "makepkg: install all dependencies simultanously, if possible
[14:06:54] <elibrokeit> when using rmdeps but not install, runtime dependencies do not
[14:06:54] <elibrokeit> need special handling for (non)removal, and can be installed in the
[14:06:54] <elibrokeit> same transaction. This allows pacman to intelligently resolve some
[14:06:54] <elibrokeit> things which would otherwise be conflicts."
[14:07:20] <elibrokeit> do you not want to submit this yourself?
[14:07:35] <deep42thought> how?
[14:07:40] <deep42thought> I'm a noob to git email ...
[14:08:52] <elibrokeit> git config sendemail.*
[14:09:34] <deep42thought> where should patches be sent to?
[14:09:46] <deep42thought> the arch-projects mailing list?
[14:09:50] <elibrokeit> smtpserver = mail.myserver.org
[14:09:50] <elibrokeit> smtpuser = username@myserver.org
[14:09:50] <elibrokeit> smtpserverport = 587
[14:10:07] <elibrokeit> to = pacman-dev@archlinux.org
[14:10:37] <elibrokeit> the "to" key should prolly be git config --local ;)
[14:10:56] <deep42thought> :-D
[14:11:24] <elibrokeit> also please use format.signoff = true; format.subjectprefix = pacman-dev] [PATCH
[14:13:29] <elibrokeit> this all configures git send-email to act as an email client, you can then do git send-email -1 with optional additional headers like --cc=<address> --in-reply-to=<message-id> --to=<some-other-list-instead>
[14:14:50] <elibrokeit> Consider stealing all or part of the alternative commit message I proposed
[14:15:00] <deep42thought> yes, will do
[14:15:02] <deep42thought> :-)
[14:20:41] -!- oaken-source has joined #archlinux32
[14:21:23] <abaumann> mmh. two machines with no X: one has a black screen (a MACH64 card), the other one green funny artifacts on the screen (Alix AMD Geode).
[14:21:36] <deep42thought> :-/
[14:21:57] <abaumann> Should I really update my eeepc with i915 now?
[14:33:49] * eworm has another broken embedded device with AMD Geode
[14:34:02] <deep42thought> :'-(
[14:34:25] <abaumann> I'm really afraid, this is upstream (xorg) breaking things.
[14:34:44] <abaumann> If at least VESA would work.. but it doensn't
[14:35:54] <abaumann> xorg 1.20 and VIA is also known to cause trouble: https://forum.manjaro.org
[14:35:55] <phrik> Title:[x32] After update to Xorg 1.20 my VIA PN800 boots to black screen - Drivers - Manjaro Linux Forum (at forum.manjaro.org)
[14:41:46] -!- Alina-malina has joined #archlinux32
[14:43:00] <deep42thought> man, I hate configuring this stupid email stuff
[14:43:12] <deep42thought> msmtp connects, but does not try to authenticate
[14:43:14] <deep42thought> O.o
[14:46:22] <abaumann> Xorg and alix: using VESA and "Option "NoDRI"" works :-)
[14:46:40] <abaumann> Also some module section with Load "shadowfb" is needed, the dependency system is borked..
[14:47:45] <abaumann> ok, read-only CF mode with Xorg and systemd sessions are also a nogo.
[14:48:14] <abaumann> the xauth cookie is in the home, the home is read-only. :-)
[14:48:45] <abaumann> rumours are intel drivers are working in their very newest version (which we have) :-)
[14:48:52] <abaumann> so I'll update and see what happens..
[14:49:18] <abaumann> my IIRC and email clients are text-mode, so I can work even in the case of a disaster in X.
[14:49:51] <deep42thought> iirc client?
[14:49:52] <deep42thought> ;-)
[14:50:06] <abaumann> aeh.
[14:50:10] <abaumann> :-)
[14:50:16] <abaumann> abbreviations..
[14:55:04] <deep42thought> elibrokeit: I can't set all the configuration you posted, I get: "error: invalid key: sendemail.format.signoff=true"
[14:56:31] <elibrokeit> why did you strip out the spaces around the =
[14:56:41] <deep42thought> because I always did
[14:56:43] <deep42thought> :-(
[14:56:45] <elibrokeit> It's standard git config format for setting keys
[14:57:02] <elibrokeit> hmm, and also it's format.signoff, not sendemail.format.signoff
[14:57:20] <deep42thought> "error: invalid key: format.signoff = true"
[14:57:23] <elibrokeit> git config options are always section.option, not section.subsection.option
[14:57:38] <deep42thought> ah, no quotes :-)
[14:57:52] <deep42thought> and no "="
[14:57:53] <deep42thought> ...
[14:58:01] <elibrokeit> $ git config fakesection.foonothing = bleep
[14:58:02] <elibrokeit> works
[14:58:31] <elibrokeit> oh, that too, yes
[14:58:44] <elibrokeit> $ git config --get fakesection.foonothing
[14:58:44] <elibrokeit> =
[14:58:57] * elibrokeit scratches head, considers going back to sleep
[14:59:40] <elibrokeit> I should stop mixing config file idioms and commandline config setter idioms
[15:07:22] <abaumann> * abaumann boots an updated machine with i915GM chipset ...
[15:07:22] -!- abaumann has quit [Quit: leaving]
[15:10:21] -!- abaumann has joined #archlinux32
[15:10:22] <buildmaster> Hi abaumann!
[15:10:32] <abaumann> * abaumann is back in graphics mode and is extremly happy. :-)
[15:12:37] <elibrokeit> deep42thought: why is it all one line?
[15:13:06] <deep42thought> dunno, it was multiple lines in the commit message O.o
[15:13:17] <deep42thought> ... at least, I thought so
[15:14:05] <deep42thought> I hope this does not prevent it from being accepted :-/
[15:14:45] <elibrokeit> no, it's just potentially embarrassing ;)
[15:14:59] <deep42thought> well, nothing to lose on my end, then
[15:15:00] <deep42thought> :-D
[15:15:10] <elibrokeit> allan is certainly capable of reflowing the commit message text...
[15:15:14] <elibrokeit> lol
[15:44:31] -!- deep42thought has quit [Quit: Leaving.]
[15:55:52] -!- eworm has quit [Quit: Leaving.]
[17:08:05] -!- bill-auger has quit [Ping timeout: 240 seconds]
[17:08:10] -!- oaken-source has quit [Ping timeout: 264 seconds]
[17:40:31] elibrokeit is now known as noteli
[17:41:47] noteli is now known as elibrokeit
[18:19:46] -!- bill-auger has joined #archlinux32
[18:51:50] <elibrokeit> deep42thought: hmm, did you make a blank line between the first line (title) and later lines (body)?
[20:00:32] -!- becoming_i has joined #archlinux32
[20:00:52] -!- becoming_i has parted #archlinux32
[20:09:22] -!- Lard has joined #archlinux32
[20:09:29] <abaumann> n/quit
[20:09:31] <abaumann> exit
[20:09:35] -!- abaumann has quit [Quit: leaving]
[20:52:44] -!- oaken-source has joined #archlinux32
[22:45:57] -!- NoobAlice has joined #archlinux32
[23:06:27] -!- oaken-source has quit [Ping timeout: 245 seconds]