#archlinux32 | Logs for 2018-08-16

Back
[00:05:57] <buildmaster> pifpaf is broken (says nlopc46).
[00:06:39] <buildmaster> scrapy is broken (says nlopc46).
[00:19:29] <buildmaster> gqrx is broken (says nlopc43).
[01:17:20] -!- eduardoeae__ has quit [Ping timeout: 244 seconds]
[01:19:21] -!- eduardoeae__ has joined #archlinux32
[02:42:48] <buildmaster> python-cinderclient is broken (says nlopc43).
[03:49:11] <buildmaster> bluegriffon is broken (says buildknecht2).
[04:06:15] <buildmaster> abiword is broken (says rechenknecht).
[04:11:26] <buildmaster> xfce4-screenshooter is broken (says rechenknecht).
[06:03:46] -!- deep42thought has joined #archlinux32
[06:03:47] <buildmaster> Hi deep42thought!
[06:04:08] <deep42thought> elibrokeit: pool.mirror.archlinux32.org is a dns round-robin - not really practical with https
[06:12:52] <elibrokeit> Well, the problem is it is also HSTS...
[06:13:09] <elibrokeit> and you use that link on the packages index for the download link
[06:13:48] <deep42thought> the dns server uses curl to check if the mirror is really available on http before adding it to the dns
[06:14:04] <deep42thought> so I think, tyzoid needs to add a redirect on http
[06:14:40] <deep42thought> or disable hsts for pool.mirror.archlinux32.org
[06:15:35] <deep42thought> ah, no - I use "-L" with curl
[06:15:36] <deep42thought> my bad
[06:27:55] <deep42thought> ok, _and_ there is no redirect, currently
[06:46:46] <elibrokeit> also, fixed a couple checksum issues
[06:46:56] <elibrokeit> looking at pipenv but it's all like wtf
[06:47:39] <elibrokeit> deep42thought: consider this :p https://github.com
[06:47:40] <phrik> Title:setup: don't setup_requires on things used to run development tasks · eli-schwartz/pipenv@29c76d5 · GitHub (at github.com)
[06:48:43] <elibrokeit> it adds setup.py dependencies in order to ensure you have the right deps for running a script that modifies the __version__.py file and tags the new release :/
[06:51:55] <deep42thought> elibrokeit: I don't quite understand?
[06:52:02] <deep42thought> for _what_ should I consider this?
[06:52:36] <elibrokeit> just if you want to see insane things and understand how truly weird the python packaging authority is
[06:52:57] * elibrokeit is sleepy and possibly not completely coherent ATM
[06:53:00] <deep42thought> uh, ok - I always want to see this :-)
[06:53:59] <elibrokeit> *and* I need to finish reviewing more of your list of things downloading things in setup_requires
[06:58:58] -!- oaken-source has joined #archlinux32
[07:03:23] * buildmaster failed to execute a mysql query - can you have a look at "tmp.mysql-functions.query.stdin.2018-08-16T07:03:23.dFzrd6"?.
[07:04:01] <deep42thought> mariadb ate all /tmp again ... :-(
[07:11:53] * buildmaster resumes sanity.
[07:16:59] -!- NoobAlice has parted #archlinux32
[08:20:05] -!- eduardoeae has quit [Ping timeout: 240 seconds]
[08:31:46] -!- deep42thought has quit [Quit: Leaving.]
[08:32:12] -!- oaken-source has quit [Ping timeout: 256 seconds]
[08:44:22] -!- abaumann has joined #archlinux32
[08:44:22] <buildmaster> Hi abaumann!
[08:56:14] <abaumann> mmh. cuda is on the blacklist, cudnn depends on cuda, so why is cudnn built?
[09:23:57] -!- abaumann has quit [Quit: leaving]
[09:24:46] <buildmaster> qt4 is broken (says nlopc46).
[10:11:26] -!- z3ntu has quit [Remote host closed the connection]
[10:13:38] -!- Alina-malina has quit [Ping timeout: 260 seconds]
[10:20:11] -!- Alina-malina has joined #archlinux32
[10:20:56] -!- z3ntu has joined #archlinux32
[10:52:51] <buildmaster> mplayer is broken (says nlopc43).
[10:53:10] <buildmaster> opal is broken (says nlopc46).
[10:53:28] <buildmaster> handbrake is broken (says nlopc46).
[10:53:46] <buildmaster> libquicktime is broken (says nlopc46).
[11:01:56] <buildmaster> gst-plugins-ugly is broken (says rechenknecht).
[11:14:23] <buildmaster> cinelerra-cv is broken (says nlopc43).
[11:38:09] -!- deep42thought has joined #archlinux32
[11:38:10] <buildmaster> Hi deep42thought!
[11:38:24] <deep42thought> abaumann: good question, I'll have a look
[11:47:19] -!- eduardoeae has joined #archlinux32
[12:36:05] <deep42thought> abaumann: I found the problem: blacklisting on build-list packages currently only works for "any" architecture - I'll try to resolve this bug :-)
[12:36:51] -!- abaumann has joined #archlinux32
[12:36:52] <buildmaster> Hi abaumann!
[12:37:08] <abaumann> mmh several packages like qemu, xfce4-screenshooter fail with ": CommandLine Error: Option 'asm-instrumentation' registered more than once!
[12:37:11] <abaumann> LLVM ERROR: inconsistency in registered CommandLine options
[12:37:13] <abaumann> "
[12:37:52] <abaumann> When using normal gcc ad binutils, this should not happen. And I don't really trust LLVM/clang/Gold LD at all to produce good code on 32-bit Intel.
[12:37:55] <deep42thought> maybe we hard-coded something, upstream does not?
[12:38:20] <abaumann> There was a post of a guy who said: "things work when I delete the gold linker".
[12:38:41] <deep42thought> we did this once, too
[12:38:53] <deep42thought> I think it was with firefox or something
[12:38:58] <abaumann> http://archlinux32.andreasbaumann.cc
[12:38:58] <phrik> Title:#archlinux32 | Logs for 2018-08-13 (at archlinux32.andreasbaumann.cc)
[12:39:01] <deep42thought> but merely due to oom
[12:39:11] <abaumann> it was with valgrind-git
[12:39:40] <abaumann> mmh. yes. He removed a bfd plugin for LLVM gold.
[12:39:52] <abaumann> this can hardly be the solution for our build chroots.
[12:40:24] <abaumann> Thogh personally I would switch all those "donations" to bintuils like LTO, Gold, etc..
[12:40:32] <abaumann> .. OFF :->
[12:40:39] <deep42thought> cu
[12:40:49] <deep42thought> ah, sry, misread
[12:40:50] <abaumann> ok. I'll investigate further.
[12:40:51] <abaumann> cu
[12:40:52] <deep42thought> :-D
[12:40:55] <abaumann> ah.
[12:40:56] <deep42thought> I'm not going afk
[12:40:57] <abaumann> :-)
[12:40:57] <abaumann> LOL
[12:41:24] <deep42thought> but you have a history of /quit'ing fast after saying "cu", so I reacted instinctly
[12:42:02] <abaumann> true. true.
[12:42:46] <deep42thought> hmm, why not disable those "extras"?
[12:42:51] <deep42thought> what are they good for?
[12:43:02] * deep42thought understands horribly few about compilers and linkers
[12:43:14] <abaumann> LTO optimized your gode. Gold LD uses insane amounts of memory for better linking results.
[12:43:46] <abaumann> I would prefer to keep them in and switch them off. Firs, I have to find out what those things are acutally doing :-)
[12:44:25] <abaumann> /usr/bin/meinproc5: error while loading shared libraries: libicui18n.so.61: cannot open shared object file: No such file or directory
[12:44:40] <abaumann> ou. so, I thouhg new ICU 62 went through?
[12:44:44] <deep42thought> is the new icu out already?
[12:44:44] <abaumann> &thought
[12:44:51] <abaumann> should be.
[12:44:54] <abaumann> lemme check.
[12:45:19] <abaumann> I get icu 62.1 on staging.
[12:46:15] <abaumann> yep. it's in tsaging. so why is my build-slave running against 61?
[12:47:43] <abaumann> /usr/bin/meinproc5: error while loading shared libraries: libicui18n.so.61: cannot open shared object file: No such file or directory
[12:47:45] <deep42thought> !wtf meinproc5
[12:47:52] <phrik> deep42thought: extra/kdoctools
[12:47:52] <abaumann> aha. so part of qt is linked against icu 61
[12:48:01] <abaumann> yep
[12:48:08] <deep42thought> part of -> probably all of
[12:49:09] <deep42thought> let's check if the auto-rescheduler does what I want it to do :-D
[12:51:05] <abaumann> maybe it's temporary till kdoctools is rebuild because of icu 62?
[12:51:12] <deep42thought> yes
[12:51:15] <abaumann> *rebuilt
[12:51:23] <deep42thought> question is if it gets rebuilt automatically
[12:51:23] <abaumann> then everyhthing is peachy :-)
[12:51:25] <deep42thought> (it should)
[12:53:41] <deep42thought> kdoctools themself do not link agains libicu
[12:55:08] <deep42thought> qt5-base does
[12:55:48] <abaumann> ah.
[12:55:58] <deep42thought> and this one is already on the build-list
[12:56:17] <abaumann> good. :-)
[12:56:44] <deep42thought> tyzoid: it looks like ssh on git.archlinux32.org is not reachable by ipv6
[13:09:49] -!- abaumann has quit [Ping timeout: 244 seconds]
[13:14:51] -!- abaumann has joined #archlinux32
[13:14:51] <buildmaster> Hi abaumann!
[13:15:01] <abaumann> systemd-journald effectively killed my system.
[13:15:05] <abaumann> on 64-bit arch.
[13:15:53] <deep42thought> abaumann: blacklisting should work now (more) properly
[13:16:08] <deep42thought> how did the journal kill your box?
[13:16:25] <abaumann> don't know.. the reason would be in the journal, I guess. :->
[13:16:36] <deep42thought> lol
[13:16:56] <abaumann> yep. my log is pretty much empty.. *sigh*
[13:17:12] <abaumann> coredumps in the journal is a famous one.
[13:17:21] <abaumann> and storage of the journal is in-memory.
[13:17:49] <deep42thought> freeze your memory and read it out "manually" ;-)
[13:18:06] <abaumann> freeze as in freezer.. ;-)
[13:18:12] <deep42thought> yes, exactly
[13:18:23] <abaumann> nothing freezez ATM with this wather..
[13:18:27] <abaumann> *weather
[13:18:45] <deep42thought> liquid nitrogen
[13:19:48] <abaumann> I suppose, that's something you have in abundance :-)
[13:20:08] <deep42thought> well ... not at home, where I am right now
[13:20:18] <abaumann> ah. lol :-)
[13:42:31] <abaumann> mmh. happens when 'ar-ing' a static library.. When I remove /usr/lib/bfd-plugins/LLVMgold.so from llvm-libs, then I can link.
[13:42:38] <abaumann> so llvm 6 is borked for 32-bit?
[13:43:09] <abaumann> I really wonder, why this is drawn in as dependency? clang/mesa using them?
[13:43:41] <abaumann> What I really hate, those plugins just start magically to do something in the background. I want some control over this..
[13:44:12] <abaumann> mrkiko: thanks for bringing this up with valgrind-git.
[13:44:16] <deep42thought> automagic is good as long as it works - but if it breaks ...
[13:45:48] <abaumann> maybe we have to add --disable-gold now in all autotooled projects.
[13:46:23] <abaumann> uff. now. it breaks also this way.
[13:54:54] <abaumann> https://bugs.archlinux.org
[13:54:54] <phrik> Title:FS#59631 : [binutils] CommandLine Error: Option 'asm-instrumentation' registered more than once! (at bugs.archlinux.org)
[13:55:17] <abaumann> I suspect we are missing the latest patch for 6.0.1-3
[13:56:40] <deep42thought> "llvm-libs-6.0.1-4.0 [build-list] (i686)"
[13:56:44] <abaumann> I rescheduled llvm.. let's see.
[13:56:53] <deep42thought> it's still on the build-list
[13:56:57] <abaumann> ah.
[13:57:14] <abaumann> good. but this means, it be dissapear after the llvm rebuild..
[13:57:39] <abaumann> What I really don't like, why is LLVM stuff running when I explicitely use GNU ld.
[13:58:05] <abaumann> and llvm-libs has to be installed fore mesa at least, so it cannot be removed.
[13:58:31] <abaumann> this leads to a dsitribution using two compiler toolchains, I don't feel comfortable with that..
[13:58:42] <deep42thought> can selinux block access to certain libraries for certain programs?
[13:58:42] <deep42thought> ;-)
[13:58:57] <abaumann> yes. but it was hardly invented for that. ;-)
[14:04:00] -!- sahasatvik__ has quit [Quit: ZNC 1.8.x-git-99-eec5c5a - https://znc.in]
[14:06:06] <abaumann> aha. This plugin is needed for LTO linking, so it's symlinked. See binutils PKGBUILD.
[14:06:50] <abaumann> So the final linking is still done with GNU ld (ld.bfd), but the code analyzer for doing link time optimizations needs LLVM to inspect the code.
[14:07:03] <abaumann> * abaumann hopes, he got it more or less correctly now. :-)
[14:15:06] <abaumann> cu later
[14:15:12] <abaumann> (now for real) :-)
[14:15:14] -!- abaumann has quit [Quit: leaving]
[14:32:14] <buildmaster> python-snowboy is broken (says nlopc46).
[14:50:11] -!- Alina-malina has quit [Ping timeout: 272 seconds]
[14:56:19] -!- davor_ has joined #archlinux32
[14:56:33] -!- davor has quit [Ping timeout: 248 seconds]
[14:56:33] davor_ is now known as davor
[15:10:05] <zer0def> um, quick question - was there supposed to be a signature with 2018.08.01-dual.iso?
[15:12:42] <deep42thought> there is supposed to be one, but tyzoid forgot it and apparently didn't see my reminder either
[15:12:58] <tyzoid> Yeah, one was signed, the other wasn't yet
[15:13:21] <tyzoid> main issue is that the file wasn't on the system with the signing key
[15:13:42] <deep42thought> ah, hi tyzoid
[15:13:49] <tyzoid> hi deep42thought
[15:14:03] <deep42thought> you saw my message about ssh://git.archlinux32.org on ipv6?
[15:14:14] <tyzoid> what's up with it?
[15:14:25] <tyzoid> (I did not)
[15:14:26] <deep42thought> it looks like I can't reach it
[15:14:30] <tyzoid> hmm
[15:15:16] <tyzoid> try connecting to 2607:5600:dc::3 directly
[15:16:16] <deep42thought> connection refused
[15:16:20] <tyzoid> zer0def: I'll have that sig up in ~10 minutes, btw
[15:16:50] <zer0def> well, i didn't mean to pester, but thanks :)
[15:17:10] <tyzoid> zer0def: nah, I meant to do it, but hit a minor snag and forgot
[15:24:02] <tyzoid> zer0def: The sig should now be there.
[15:24:11] <tyzoid> And the hashes should now be in the sums file as well (oops)
[15:24:30] <deep42thought> tyzoid: should I fix that?
[15:24:32] <zer0def> thanks a bunch
[15:24:38] <tyzoid> deep42thought: Fix what?
[15:24:52] <deep42thought> I thought for a sec, the hashes would not belong there
[15:25:04] <deep42thought> sry, I should think faster and type/speak slower
[15:25:05] <tyzoid> no, I forgot to put the hashes in the file
[15:25:26] <tyzoid> I had the hashes already done sitting in a file, just forgot to upload them
[15:25:45] <deep42thought> yeah, I understand that now :-) all good :-D
[15:57:43] <buildmaster> riscv64-linux-gnu-glibc is broken (says nlopc46).
[15:58:20] <buildmaster> gst-plugins-ugly is broken (says nlopc46).
[15:58:36] <buildmaster> python-setuptools is broken (says nlopc46).
[16:33:50] -!- Alina-malina has joined #archlinux32
[16:35:13] <elibrokeit> deep42thought: packer-io is still in the broken-packages list, but it was renamed to packer
[16:36:21] <deep42thought> when was the rename?
[16:36:33] <deep42thought> get-package-updates should pick it up and remove the build order
[16:37:14] <elibrokeit> 47 hours ago
[16:37:19] <deep42thought> oh
[16:37:44] <elibrokeit> why does spring complain a patch file checked into svntogit, is failing the checksums, when it works for me?
[16:38:20] <deep42thought> O.o
[16:39:19] <deep42thought> elibrokeit: you're talking about the archlinux32 build?
[16:39:26] <deep42thought> there is some patch applied on our side
[16:40:35] <elibrokeit> um
[16:40:57] <elibrokeit> why does it overwrite the checksum index 1, instead of appending a new one?
[16:41:17] <deep42thought> because it's overwriting the goldlinker.patch file
[16:41:20] <deep42thought> which was at position 1
[16:42:09] <elibrokeit> which is not
[16:42:14] <deep42thought> I see
[16:42:17] <deep42thought> but it _was_
[16:42:54] <elibrokeit> hmm
[16:43:24] <elibrokeit> why is the unused patch still there...
[16:44:22] <elibrokeit> deep42thought: I could fix this, by removing the patch for you :)
[16:44:30] <deep42thought> ok
[16:45:22] <deep42thought> this would be great, then I'll just prepare for that and append the source and checksum :-)
[16:46:21] <elibrokeit> wait, what is the difference between the current patch in our svn, and in your git
[16:46:29] <deep42thought> umm
[16:46:59] <elibrokeit> oh, upstream modified quotes somewhere
[16:54:52] <elibrokeit> okay I just committed this removal of two patches, give it a few minutes for svntogit to sync changes
[16:55:06] <elibrokeit> and again, please remember to ping me for things like this :)
[16:55:46] <elibrokeit> there's no reason to modify indexes of our PKGBUILDs just because there are unmaintained, unused patches overwriting your namespace :p
[17:04:25] <elibrokeit> deep42thought: position-independent patches, the new wave of the future :D https://git.archlinux32.org
[17:04:31] <phrik> Title:community/spring: make patch position-independent · f88a10a63c - Archlinux32 Gitea (at git.archlinux32.org)
[17:04:38] <deep42thought> yeah
[17:05:28] <deep42thought> you as PKGBUILD sanity guard should complain, that "i" is a bad name for a local variable in a PKGBUILD, because it does not start with a "_" ;-)
[17:06:58] <elibrokeit> well, fortunately it is removed ;)
[17:08:03] <deep42thought> :-D
[17:28:46] -!- eduardoeae has quit [Ping timeout: 260 seconds]
[17:32:24] -!- Cthulu201 has quit [Quit: Nowhere special. I always wanted to go there.]
[17:37:23] -!- abaumann has joined #archlinux32
[17:37:24] <buildmaster> Hi abaumann!
[17:38:24] <abaumann> just wanted to sync this one: readdir_r -> readdir is fine, if DIR * is given by an opendir in the same thread. So it's the C library maintainers business to deal with thread issues? And nobody is crazy enough to pass around DIR * of the SAME directory between threads, I hope.
[17:39:04] <tyzoid> I wouldn't put it past anyone
[17:40:11] <abaumann> It's just, because I have seen many fixes out there just rewriting readdir_r as readdir and dropping the self-allocated buffer..
[17:41:06] <abaumann> IIRC the "same thread DIR * opendir, readdir safety assumption" (what a horrid name) is true in glibc, OSX libc at least.
[17:41:52] <abaumann> http://elliotth.blogspot.com
[17:41:53] <phrik> Title:elliotth's blog: How (not) to use readdir_r(3) (at elliotth.blogspot.com)
[17:42:10] <abaumann> Actually a nice blog of a POSIX function just going wrong over time. :-)
[17:42:31] <abaumann> so, have to leave, sorry (famous CU from my point)
[17:42:32] -!- abaumann has quit [Quit: leaving]
[17:42:38] <deep42thought> cu
[17:42:39] <deep42thought> ^^
[17:43:39] <deep42thought> btw: I found the bug in our deletion-list logic - as far as I can tell, we did not delete packages that were removed upstream for quite some time (2-4 weeks)
[17:57:37] <deep42thought> tyzoid: let me know when ssh on git.archlinux32.org works over ipv6 again and I'll kill the ipv4-only entries on the build master and my box
[17:58:47] <tyzoid> yeah, issue is something I need to fix
[17:59:03] <deep42thought> probably some port forwarding or a faulty dns entry
[17:59:05] <deep42thought> but no hurry
[17:59:17] <deep42thought> all the ipv6 boxes have ipv4 legacy internet, too ;-P
[18:00:35] <tyzoid> I see what the issue is
[18:00:45] <tyzoid> on the old server, I had a direct ipv6 connection to it
[18:00:49] <tyzoid> but now that's not the case
[18:00:57] <tyzoid> I'll just add a forwarding rule here
[18:00:59] <tyzoid> and that should fix it
[18:02:23] <deep42thought> or you just add that logo and keep it as-is: https://lh3.googleusercontent.com
[18:02:23] <tyzoid> deep42thought: Try now
[18:02:41] <tyzoid> lol
[18:02:44] <deep42thought> do I need to flush dns?
[18:02:48] <tyzoid> shoudn't
[18:03:10] <tyzoid> hmm, maybe
[18:03:12] <deep42thought> "ssh git@2607:5600:dc::3" works
[18:03:15] <deep42thought> the other don't
[18:03:49] <deep42thought> is 2607:5600:dc:200b::1 the right address?
[18:04:02] <deep42thought> (I get this via dns)
[18:04:09] <tyzoid> that's the direct, but the vm doesn't have that address
[18:04:13] <tyzoid> one sec, while I add that
[18:04:57] <tyzoid> huh
[18:05:05] <tyzoid> deep42thought: Any opposition to a reboot of the git server?
[18:05:11] <deep42thought> no
[18:05:21] <deep42thought> if that's necessary
[18:06:00] <tyzoid> I think the systemd nondeterminism patch may have hit this box
[18:06:47] <deep42thought> network device names change based on moon phase?
[18:10:19] <tyzoid> no, systemd updates cause lxc containers to fail to get an ip address
[18:11:00] <tyzoid> that and network devices change based on the alignment of jupiter and io
[18:11:09] <tyzoid> anyway, try now
[18:11:36] <deep42thought> works :-)
[18:11:45] <deep42thought> thank you
[18:14:20] <tyzoid> deep42thought: You can thank systemd for breaking that
[18:27:52] <buildmaster> python-oslo-config is broken (says nlopc46).
[18:45:33] -!- Cthulu201 has joined #archlinux32
[18:57:02] -!- AndrevS has joined #archlinux32
[19:17:05] -!- oaken-source has joined #archlinux32
[20:32:42] -!- eduardoeae has joined #archlinux32
[20:51:45] -!- deep42thought has quit [Quit: Leaving.]
[20:58:21] -!- oaken-source has quit [Quit: Lost terminal]
[20:58:45] -!- oaken-source has joined #archlinux32
[21:03:24] -!- oaken-source has quit [Client Quit]
[21:03:50] -!- oaken-source has joined #archlinux32
[22:09:58] -!- oaken-source has quit [Ping timeout: 260 seconds]
[22:26:53] -!- Alina-malina has quit [Ping timeout: 260 seconds]
[22:34:59] -!- AndrevS has quit [Remote host closed the connection]
[22:36:43] -!- Alina-malina has joined #archlinux32
[23:05:09] <buildmaster> man-pages is broken (says nlopc46).