#archlinux-ports | Logs for 2018-03-08

[00:34:55] -!- yans has quit [Ping timeout: 256 seconds]
[00:51:52] -!- |ast| has joined #archlinux-ports
[05:26:14] -!- hringriin_ has joined #archlinux-ports
[05:29:45] -!- hringriin has quit [Ping timeout: 264 seconds]
[05:29:46] hringriin_ is now known as hringriin
[06:20:02] -!- deep42thought has joined #archlinux-ports
[06:33:08] -!- |ast| has quit [Remote host closed the connection]
[06:43:37] <buildmaster> delve is broken (says buildknecht3).
[07:21:07] -!- deep42thought has quit [Quit: Leaving.]
[07:49:06] -!- abaumann has joined #archlinux-ports
[07:49:45] <abaumann> deep42thought: mysql dropped a transaction because of a deadlock and is telling you that one transaction was selected as victim and has to be redone.
[07:50:43] <abaumann> usually this happens because there is a cycle in the dependencies of the transactions because for instance two transaction access two records in reversed order, so A waits for B, B waits for A
[08:16:20] <buildmaster> firefox-developer-edition is broken (says buildknecht).
[08:18:39] -!- Vollzornbrot has quit [Ping timeout: 256 seconds]
[08:18:56] -!- Vollzornbrot has joined #archlinux-ports
[08:26:43] <abaumann> guys: possible bootstrapping issues in openldap: package_openldap, for dir in clients servers doc/man/man{1,5,8};
[08:26:54] <abaumann> this installs the client and server part, not libldap.
[08:27:14] -!- deep42thought has joined #archlinux-ports
[08:27:23] <abaumann> Now, libtool wants to relink the binaries to the new libldap libraries in pkgdir, but they are missing.
[08:27:39] <abaumann> I get link errors like "libtool: install: error: relink `accesslog.la' with the above command before installing it
[08:28:11] <deep42thought> abaumann: actually, there was my >8 minutes query running and blocking the transaction
[08:28:26] <abaumann> my proposed solution is to always install everything in such split packages (at least if they use libtool) and then for instance remove pkgdir/usr/include, pkgdir/usr/lib and pkgdir/usr/share/man3
[08:28:58] <abaumann> I suspect it usually works, because libtool links against -lldap_r against /usr/lib/libldap_r.so!
[08:29:13] <abaumann> I saw similar problems in the toolchain around libquadmath and fortran.
[08:29:40] <abaumann> deep42thought: ah. the other reason for selecting a victim. (timeout) :-)
[08:30:34] <deep42thought> there is currently no error handling for mysql: if running "mysql" fails, I have no idea what to do, so nothing is done :-/
[08:31:15] <abaumann> yeah. a little bit risky, isn't it? ;-)
[08:31:31] <deep42thought> well, the buildmaster goes insane and then everything stops :-D
[08:31:41] <abaumann> but error handling is more than just checking the exit code, I'm afraid.
[08:31:49] <abaumann> defensive programming. :-)
[08:32:00] <deep42thought> :-D
[08:32:00] <abaumann> so at least there is no (or only little) data corruption.
[08:32:37] <abaumann> I start to find nice bootstrapping bugs in Archlinux(32). :-)
[08:32:49] <deep42thought> probably I could/should save the failed query and try to run it later a few times - if that fails, too, then I should mark the buildmaster as insane and resort to manual correction
[08:33:22] <deep42thought> what distinguishes a nice from a bad bootstrapping bug?
[08:33:32] <abaumann> aeh.. :-)
[08:33:55] <abaumann> nice, because it's a real bug and not just for instance a bug in the toolchain or libtool causing a bug in a package.
[08:34:02] <abaumann> the later ones I really hate.
[08:34:20] <deep42thought> ah, ok
[08:41:29] <deep42thought> crap, I pushed to the wrong branch
[08:57:12] <deep42thought> If a query times out: Are all statements from that query dropped or only the one which cannot get its lock?
[09:03:14] <abaumann> the whole transaction
[09:03:21] <deep42thought> ok, thanks
[09:17:24] <Vollzornbrot> hello guys
[09:17:29] <deep42thought> Hi Vollzornbrot
[09:17:30] <abaumann> hi
[09:17:39] <Vollzornbrot> all okay?
[09:17:49] -!- Vollzornbrot has quit [Changing host]
[09:17:49] -!- Vollzornbrot has joined #archlinux-ports
[09:17:52] <abaumann> yeah. pretty much.
[09:18:00] <abaumann> some insanity in the buildmaster. ;-)
[09:18:04] <deep42thought> :-D
[09:18:22] <Vollzornbrot> oh okay :D
[09:18:24] <deep42thought> I'm still fighting with the database
[09:18:32] -!- Vollzornbrot has quit [Quit: WeeChat 1.6]
[09:18:37] <abaumann> guys: I tried to condense my earlier rambling in a bug report: https://bugs.archlinux.org
[09:18:37] <phrik> Title: FS#57749 : [openldap] libtool relink errors during install (at bugs.archlinux.org)
[09:19:20] <deep42thought> hmm, I get an idea, why the database operations are becoming slow: we have 259784 recorded dependencies :-/
[09:19:29] -!- heradon has joined #archlinux-ports
[09:19:41] <abaumann> ui. and it's a graph.
[09:20:04] <deep42thought> each "dependency" records a connection between a package and an install target
[09:20:12] <deep42thought> so this number is already the number of edges
[09:20:15] <heradon> i am at the moment on a samba problem xD
[09:20:24] <deep42thought> but you get multiple providers for install targets sometimes
[09:22:12] <abaumann> give mysql enough memory, maybe it gets acceptable then.
[09:22:44] <abaumann> heradon: samba problem?
[09:22:45] <deep42thought> there is still some room for optimization in the queries
[09:23:17] <heradon> abaumann: i need to change the user passsword from samba but without ssh login
[09:23:30] <abaumann> oh. :-)
[09:23:47] <deep42thought> you mean the password of the system user "samba"?
[09:23:55] <heradon> no for the samba user
[09:24:07] <abaumann> the pdbedit thingy normally?
[09:24:19] <heradon> pdbedit?
[09:24:41] <abaumann> ok. depends on your samba backend for storing user credentials.
[09:24:55] <heradon> yes i have pdbedit
[09:24:55] <abaumann> that's for the standalone samba server
[09:25:08] <heradon> yes is standalone
[09:25:13] <deep42thought> that's what I thought, too: you just need ldap access ;-P
[09:25:16] <abaumann> but you have no ssh access.
[09:25:40] <abaumann> maybe smbclient has a remote change password function
[09:26:32] <heradon> yes i test this
[09:28:13] <abaumann> smbpasswd is for local users only, I think.
[09:28:30] <abaumann> smbpasswd -r -U, aha.
[09:28:35] <abaumann> maybe they work.
[09:31:12] <abaumann> https://bbs.archlinux32.org
[09:31:14] <phrik> Title: libQt5Core.so.5: No such file or directory after system update / Pacman / Pacman Upgrades / Arch Linux 32 Forums (at bbs.archlinux32.org)
[09:31:31] <abaumann> Do we have a Qt problem? I fail to reproduce that on testing or on stable.
[09:33:43] <deep42thought> /usr/lib/libQt5Core.so.5 is provided by qt5-base-5.10.0-4.0-i686.pkg.tar.xz, currently being in extra
[09:33:56] <deep42thought> so there should be no problem with stable, either
[09:34:53] <deep42thought> ah, you already wrote that (I read "staging")
[09:39:04] <abaumann> yes. especially because it's not clear to me, if that is really an Archlinux32.
[09:40:53] <deep42thought> abaumann: What do you think of the idea to put the hashes of the binary packages into the database?
[09:41:29] <deep42thought> https://bbs.archlinux32.org
[09:41:31] <phrik> Title: automatic testing via manual installation(s) / Testing / Arch Linux 32 Forums (at bbs.archlinux32.org)
[09:49:13] <abaumann> this would avoid to stabilize the wrong packages?
[09:49:32] <deep42thought> this would help with reporting about tested packages
[09:50:02] <deep42thought> e.g. if manjaro32 wants to help in testing, we would not falsely mark their packages in our database as tested
[09:51:44] <deep42thought> I'm currently rewriting the db-update to use the database, and meanwhile changing the condition on when to move packages: https://github.com
[09:51:45] <phrik> Title: builder/db-update at db-update · archlinux32/builder · GitHub (at github.com)
[09:52:48] <deep42thought> mainly, this should avoid these huge deadlocks we had in the past
[09:54:03] <deep42thought> the hashes are separate from that
[10:01:58] <abaumann> yeah, good idea.
[11:23:05] <abaumann> AFW, off for lunch.
[11:23:08] -!- abaumann has quit [Quit: leaving]
[11:34:48] * buildmaster failed to execute a mysql query - can you have a look at "tmp.mysql-functions.query.2018-03-08T10:28:02.DebpWr"?.
[12:21:45] <deep42thought> at least, the report works :-)
[12:22:04] <deep42thought> "1 row in set (42 min 45.42 sec)" ... and it's still executing :-/
[12:36:16] * buildmaster resumes sanity.
[13:01:17] -!- oaken-source has joined #archlinux-ports
[13:35:28] * buildmaster failed to execute a mysql query - can you have a look at "tmp.mysql-functions.query.2018-03-08T12:35:28.bHUHbP"?.
[13:35:56] <deep42thought> my bad
[13:39:34] * buildmaster resumes sanity.
[13:42:21] <buildmaster> deep42thought: my database is dirty again ...
[13:42:28] <deep42thought> gimme a sec ...
[13:45:51] * buildmaster resumes sanity.
[14:17:27] * buildmaster failed to execute a mysql query - can you have a look at "tmp.mysql-functions.query.2018-03-08T13:17:15.AHNVU1"?.
[14:19:27] * buildmaster resumes sanity.
[14:21:27] -!- buildmaster has quit [Remote host closed the connection]
[14:21:32] <deep42thought> O.o
[14:22:13] -!- buildmaster has joined #archlinux-ports
[14:26:03] * buildmaster goes insane.
[14:26:29] <deep42thought> ah, my test is too restrict ...
[14:26:57] -!- guys has quit [Ping timeout: 240 seconds]
[14:36:18] * buildmaster resumes sanity.
[14:44:32] -!- abaumann has joined #archlinux-ports
[14:44:54] <deep42thought> abaumann: that was a long lunch break ;-)
[14:46:32] <abaumann> yes. :-)
[14:46:40] <abaumann> business lunch, this is called.
[14:47:30] <abaumann> the mysql error messages are nice in the buildmaster :-)
[14:48:02] <deep42thought> yeah, but I made the mistake to also let it complain, when I enter a wrong query manually :-D
[15:07:29] <tyzoid> lol
[15:35:49] <deep42thought> ok, beware: I'm going to try the new db-update - most probably I'll break everything now :-D
[15:41:12] <abaumann> make a backup of the universe before.. ;-)
[15:44:00] <tyzoid> abaumann: I've got a backup of the buildmaster from this morning
[15:44:16] <deep42thought> then we're save :-)
[15:44:38] <tyzoid> :)
[15:47:24] <deep42thought> yay, typo - it dit nothing :-)
[15:47:58] <tyzoid> lol, the intensity
[15:58:34] -!- guys has joined #archlinux-ports
[16:03:52] -!- deep42thought has quit [Quit: Leaving.]
[16:07:39] -!- Polichronucci has quit [Ping timeout: 256 seconds]
[16:08:34] -!- Polichronucci has joined #archlinux-ports
[16:08:35] -!- abaumann has quit [Quit: leaving]
[16:24:52] -!- abaumann has joined #archlinux-ports
[16:24:56] <abaumann> https://bugs.archlinux.org
[16:24:56] <phrik> Title: FS#57749 : [openldap] libtool relink errors during install (at bugs.archlinux.org)
[16:25:02] <abaumann> seriously?
[16:25:50] <abaumann> It's not enough that a package builds somehow on your laptop.. *sigh*
[16:26:30] <abaumann> * abaumann: makes a mental note to himself not to write 'this is possibly not a bug' in a bug report.. :-)
[16:26:39] -!- abaumann has quit [Client Quit]
[17:01:05] -!- abaumann has joined #archlinux-ports
[17:05:22] -!- abaumann has quit [Client Quit]
[17:23:40] -!- deep42thought has joined #archlinux-ports
[17:29:00] <deep42thought> abaumann: look at the bright side: now the issue is archived for future generations stumbling over the same while bootstrapping on 128 bit ;-)
[17:56:43] -!- abaumann has joined #archlinux-ports
[17:57:08] <abaumann> deep42thought: actually, an ABI bump will be enough. :->
[18:15:51] <deep42thought> then you need to manually untar the old package?
[18:16:05] <deep42thought> like we had to do with libicu for recompiling gcc?
[18:30:02] -!- abaumann_ has joined #archlinux-ports
[18:30:36] <abaumann_> deep42thought: yes, I think so.
[18:32:06] -!- abaumann_ has quit [Client Quit]
[18:32:12] -!- abaumann has quit [Quit: leaving]
[18:33:13] -!- abaumann has joined #archlinux-ports
[18:54:01] <deep42thought> another advantage of having the hashes in the database: we do not need to pull the packages from the master mirror just do update the package database files
[18:54:27] <deep42thought> (that's what's taking so long, right now, on the build master)
[19:02:26] -!- deep42thought has quit [Quit: Leaving.]
[19:08:38] <abaumann> ah. yes. that's handy. :-)
[19:10:38] -!- |ast| has joined #archlinux-ports
[19:22:27] -!- shentey has joined #archlinux-ports
[19:27:47] -!- shentey has quit [Quit: Konversation terminated!]
[19:35:00] <buildmaster> deep42thought: my database is dirty again ...
[19:45:18] <abaumann> https://bbs.archlinux32.org
[19:45:19] <phrik> Title: libQt5Core.so.5: No such file or directory after system update / Pacman / Pacman Upgrades / Arch Linux 32 Forums (at bbs.archlinux32.org)
[19:45:28] <abaumann> Archlinux32 is truly multi-arch. :-)
[19:46:46] <tyzoid> wow
[19:47:38] <abaumann> https://bbs.archlinux32.org
[19:47:39] <phrik> Title: кот саймон св¤зать крючком - Правдивая история Кота в сапогах - JT / Creating/Maintaining Packages / Arch Linux 32 Forums (at bbs.archlinux32.org)
[19:47:42] <abaumann> Spam or no spam.
[19:47:50] <abaumann> I actually like 'Simon the Cat' :-)
[19:49:46] -!- shentey has joined #archlinux-ports
[20:25:08] <guys> abaumann: um, how *exactly* could you end up with asciidoc and not libxslt?
[20:31:10] -!- sunshavi has quit [Remote host closed the connection]
[20:32:06] <abaumann> guys: in which package?
[20:32:24] -!- sunshavi has joined #archlinux-ports
[20:32:27] <guys> ca-certificates
[20:33:00] <abaumann> makedepends=(asciidoc p11-kit)
[20:33:12] <abaumann> asciidoc.py in build.
[20:33:17] <abaumann> xsltproc in build
[20:33:17] <guys> please check the depends of the makedepends, and don't open bugs for things that actually work fine unless you're pretty sure there is a legitimate problem...
[20:33:35] <guys> yes, and asciidoc is a frontend for xsltproc...
[20:33:42] <guys> see a2x
[20:33:45] <abaumann> aha.
[20:33:48] <abaumann> sorry.
[20:33:55] <abaumann> trasnsitive dependency.
[20:34:11] -!- sunshavi has quit [Remote host closed the connection]
[20:34:14] -!- subalpine has quit [Ping timeout: 260 seconds]
[20:34:14] <guys> given, that it works in a clean chroot build, why file a bug? What convinced you there was an issue?
[20:34:58] <abaumann> no, I forgot to check the dependencies of asciidoc, and there is docbook-xsl which needs libxslt.
[20:35:01] <abaumann> my mistake.
[20:35:12] <abaumann> close the bug. :-)
[20:35:30] -!- sunshavi has joined #archlinux-ports
[20:35:32] <abaumann> but the other one about libtool and openldap is not just for closing..
[20:36:24] <abaumann> FS#57749
[20:36:48] <guys> abaumann: I'm just trying to understand your process for reporting a bug, presumably you reported that because the package failed to build, but that should be impossible since if you have asciidoc installed you have libxslt installed...
[20:37:05] <guys> did your chroot not have libxslt installed?
[20:37:54] <abaumann> I had asciidoc patched out (because I'm bootstrapping in stage 3 without documentation tools). So makedepends had asciidoc patched out and asciidoc.py..
[20:38:05] <abaumann> ..I forgot about xsltproc and then it failed.
[20:38:40] <abaumann> building in the chroot.. I'm not using Archlinux chroots, I'm bootstrapping on a real/VM using a remote SSH makepkg
[20:39:00] <abaumann> so.. the build process various quite a lot.
[20:39:48] <guys> abaumann: this build process varying causes a fair number of not-bugs, please double-check next time...
[20:40:11] <abaumann> ..yes. sorry. and they obviously also trigger corner cases which are only present when bootstrapping.
[20:40:18] <guys> Surely it is not hard to look at "wait, why did the maintainer think this would work"
[20:40:19] <abaumann> I'll collect those somewhere else.
[20:40:49] <abaumann> If you are not interested in boostrapping Archlinux, just say so.
[20:41:03] <abaumann> I think, there are some bad issues lurking there..
[20:41:11] <guys> I mean, if you are going to collect bootstrapping bugs then at least wait and sit on them before reporting them.
[20:41:37] <abaumann> Should I use another category for the bug? Feature request or something similar?
[20:41:58] <guys> When you end up with things like "I patched out asciidoc but forgot to patch out xsltproc" this is sort of not helpful
[20:42:45] <abaumann> Is there a possibility, that I can delete my own bugs?
[20:42:53] <abaumann> In case I misfile a bug?
[20:43:02] <guys> No, that takes a bug wrangler
[20:43:02] <abaumann> Or close it myself..
[20:43:07] <abaumann> ah :-)
[20:43:17] <abaumann> sorry for creating too much work.. :-)
[20:43:24] <guys> you can request it be closed, but we still need to do the closing
[20:44:00] <guys> don't get me wrong, you've discovered some real issues in the past, but you've also mixed in some not issues as well...
[20:44:27] <abaumann> yep. I'm famous for that also in my job.. shoot first, ask questions later.. :-)
[20:44:56] <abaumann> sorry.. I'll make a checklist for bugreporting.. like check first with the standard tools, check first on 64-bit.. etc.
[20:45:04] <guys> yes, thanks.
[20:45:22] <abaumann> mmh "General Gripe" ?
[20:45:38] <abaumann> fits to bootstrapping stuff :-)
[20:46:28] <guys> See for example how deep42thought usually reports issues: https://bugs.archlinux.org
[20:46:29] <phrik> Title: FS#56883 : [python-paste] check() fails (at bugs.archlinux.org)
[20:46:39] <guys> >
[20:46:41] <guys> Steps to reproduce:
[20:46:43] <guys> > git checkout c0cfd0cf0156c5149f70c3a8f891c57de781eb08
[20:46:45] <guys> > cd python-paste/repos/community-any
[20:46:47] <guys> > extra-x86_64-build
[20:46:52] <guys> >> log attached
[20:47:02] <abaumann> yes. that's hard for bootstrapping stuff..
[20:47:41] <guys> if it cannot be reproduced that way, then you should double-check to see *why* the bug happens, not just that it does.
[20:48:35] <abaumann> Well, all bootstrapping bugs are not reportable in this case, because I cannot expect you to go through 3 steps of bootstrapping.
[20:48:51] <abaumann> So.. those bugs I have to collect myself.. I think, github issues for bootstrap32
[20:48:55] <abaumann> is a good place..
[20:49:33] <abaumann> ok.. thanks.. all fine with me.. :-)
[20:52:02] <guys> abaumann: sure, collect issues there and if after thinking about it you still think there is an issue that we should fix, at least you will be more prepared to explain why it needs fixing
[21:00:24] -!- abaumann has quit [Quit: leaving]
[21:30:30] -!- shentey has quit [Quit: Konversation terminated!]
[21:59:22] -!- bill-auger has quit [Read error: Connection reset by peer]
[21:59:43] -!- bill-auger has joined #archlinux-ports
[22:13:46] -!- isacdaavid has joined #archlinux-ports
[22:22:15] -!- |ast| has quit [Remote host closed the connection]
[22:23:27] -!- deep42thought has joined #archlinux-ports
[23:04:45] * buildmaster resumes sanity.
[23:11:06] -!- deep42thought has quit [Quit: Leaving.]
[23:15:42] <tyzoid> hey deep42thought: At some point, we should perform a test restore of the Buildmaster to verify everything is as it should be.