#archlinux32 | Logs for 2018-07-19

Back
[00:00:26] -!- elibrokeit has quit [Quit: A random quit message]
[00:00:27] -!- City-busz has quit [Quit: http://quassel-irc.org - Csevegjen k�nyelmesen. B�rhol.]
[00:15:50] -!- ztrawhcse has joined #archlinux32
[00:17:56] ztrawhcse is now known as elibrokeit
[00:58:15] -!- prurigro has quit [Ping timeout: 268 seconds]
[01:04:33] -!- oaken-source has quit [Ping timeout: 264 seconds]
[01:18:27] -!- eduardoeae has quit [Ping timeout: 240 seconds]
[01:21:58] -!- eduardoeae has joined #archlinux32
[01:30:37] -!- thePiGrepper has joined #archlinux32
[01:31:08] -!- thePiGrepper has quit [Client Quit]
[01:50:48] -!- bill-auger has quit [Quit: No Ping reply in 60 seconds.]
[01:51:39] -!- bill-auger has joined #archlinux32
[03:22:08] -!- NoobAlice has joined #archlinux32
[04:00:41] <buildmaster> firefox-developer-edition is broken (says buildknecht2).
[04:00:49] -!- bill-auger has quit [Quit: No Ping reply in 60 seconds.]
[04:01:37] -!- bill-auger has joined #archlinux32
[04:05:53] -!- bill-auger has quit [Client Quit]
[04:07:28] -!- bill-auger has joined #archlinux32
[04:12:46] <elibrokeit> How much fun will you all have, bootstrapping python 3.7 :D
[04:13:02] <elibrokeit> Because we're getting closer to completion
[04:20:11] -!- bill-auger has quit [Quit: No Ping reply in 60 seconds.]
[04:21:12] -!- bill-auger has joined #archlinux32
[04:29:12] -!- bill-auger has quit [Quit: No Ping reply in 60 seconds.]
[04:30:01] -!- bill-auger has joined #archlinux32
[05:46:20] -!- prurigro has joined #archlinux32
[05:55:35] -!- NoobAlice has quit [Ping timeout: 276 seconds]
[06:27:00] -!- NoobAlice1 has joined #archlinux32
[06:30:55] <girls> elibrokeit: anything special with python3.7 - does it not build with python 3.6 or what?
[06:31:56] <elibrokeit> in the super-sekrit Trusted Users channel we're having a *lot* of fun with cyclic dep trees, packages which simply don't build, and so on
[06:32:41] <elibrokeit> *some* things are solved by skipping testsuites, but not all... and the testsuites are vital, because we caught lots of things that need python37 compat patches
[06:33:11] <elibrokeit> luckily we've applied many patches to svn so you don't need to do that too :p
[06:33:40] <girls> yeah, we let you guys do all the stuff that really requires work :-D
[06:33:44] <elibrokeit> and filed upstream bugs in yet more cases
[06:33:55] <elibrokeit> but it's definitely "fun"
[06:34:00] <elibrokeit> those dep cycles..
[06:34:31] <girls> well, better than cycles in haskell, I guess :-)
[06:34:38] <elibrokeit> lolololol
[06:35:03] <girls> anyway, time for breakfast :-)
[06:35:04] <elibrokeit> indeed, python dep cycles still build the same freaking code whether tests are skipped or not
[06:35:44] <elibrokeit> for some reason zope's own testrunner is completely broken on namespace packages... but zope is a collection of namespace packages...
[06:35:59] <elibrokeit> the solution is to || warning "tests failed"
[06:36:12] <elibrokeit> then run the tests inside the chroot with the package installed -_-
[06:36:27] <elibrokeit> ARGHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHH
[07:05:59] -!- abaumann has joined #archlinux32
[07:06:00] <buildmaster> Hi abaumann!
[07:06:11] -!- isacdaavid has quit [Quit: Leaving.]
[07:06:22] <abaumann> I'm just glad I bootstrapped i486 with python 3.6 and not 3.7 :->
[07:07:08] <abaumann> tyzoid: varnish config, nice. :-)
[07:10:46] <tyzoid> abaumann: Yeah. Let me know how the nl.32arlm.cdn.tyzoid.com one performs.
[07:11:16] <tyzoid> it has to backhaul traffic all the way from the US, so it'll be representative of what yours will perform like
[07:18:44] -!- bill-auger has quit [Ping timeout: 255 seconds]
[07:23:06] -!- bill-auger has joined #archlinux32
[07:26:09] <abaumann> mmh. I can only test up to 10MB/s, but this is my limitation. :-)
[07:26:37] <abaumann> let me set up a VM somewhere with faster network..
[07:31:08] -!- elibrokeit has quit [Ping timeout: 276 seconds]
[07:31:09] -!- lukeshu has quit [Ping timeout: 276 seconds]
[07:34:46] -!- ztrawhcse has joined #archlinux32
[07:35:24] -!- lukeshu has joined #archlinux32
[07:40:15] <abaumann> tyzoid: getting around 25MB/s.. 30MB/s now. This looks nice. :-)
[07:47:36] <abaumann> now when getting to older ISOS I drop to 2/3MB/s..
[07:47:45] <abaumann> so I guess, those are not as cached..
[07:49:42] -!- NoobAlice1 has quit [Quit: Leaving.]
[07:51:09] <tyzoid> yup
[07:51:32] <tyzoid> It's really just an accelerator to limit bandwidth usage
[07:59:55] -!- deep42thought has joined #archlinux32
[07:59:56] <buildmaster> Hi deep42thought!
[08:00:02] <deep42thought> Good morning, all!
[08:13:15] ztrawhcse is now known as elibrokeit
[08:13:40] -!- elibrokeit has quit [Quit: A random quit message]
[08:13:56] <deep42thought> abaumann: I still don't understand what is going on with your linux build - or were you talking about i486?
[08:15:13] -!- ztrawhcse has joined #archlinux32
[08:16:51] ztrawhcse is now known as elibrokeit
[08:25:49] <abaumann> deep42thought: if some variables are missing in 'config' then building the package asks for values of those variables.
[08:25:58] <abaumann> maybe we have different versions of the kernel?
[08:26:11] <deep42thought> yeah, maybe
[08:26:19] <deep42thought> what git revision is your package source?
[08:27:40] <abaumann> b09589512e9ba1f7b28a94ab82aa9b420719d5b3 on i486, same on my eurobuild3 build slave
[08:28:13] <abaumann> d299e4f0ee4303f6184169b7c55c2fbf54d158f2 in the diff PKGBUILD?
[08:29:22] <deep42thought> thanks, but I meant "git rev-parse HEAD" of the source :-)
[08:29:41] <deep42thought> e.g. what commit the modification PKGBUILD is in
[08:30:26] <abaumann> d516581ef278b49536a43d8de664d50197280ead
[08:30:29] <abaumann> ups. sorry. :-)
[08:30:50] <deep42thought> ok, that's exactly, what I have got here
[08:31:06] <deep42thought> onto which upstream revision are you applying this?
[08:31:31] <abaumann> c4fd4b802ff6b6f65c77e76bebc33857ad0b45b5
[08:31:53] <deep42thought> same here
[08:31:59] <deep42thought> and you're building for i686?
[08:32:02] <deep42thought> or i486?
[08:32:05] <abaumann> Yes, both.
[08:32:18] <abaumann> though on i486 I have another config file.
[08:32:37] <deep42thought> and you're sure the error / necessity for "make oldconfig" is on i686, too
[08:32:53] <abaumann> not anymore.
[08:32:59] <deep42thought> :-D
[08:33:01] <abaumann> I did that and checked in a new version.
[08:33:10] <abaumann> commit: d516581ef278b49536a43d8de664d50197280ead
[08:33:12] <deep42thought> you're not sure anymore or it's not necessary anymore?
[08:33:24] <abaumann> I built it and it didn't hang anymore
[08:33:48] <abaumann> git diff d516581ef278b49536a43d8de664d50197280ead..f5730724bc6eb0f4d190cb59599ee65201ffe20c
[08:33:55] <abaumann> -# CONFIG_CRYPTO_SALSA20_586 is not set
[08:33:57] <abaumann> -# Linux/x86 4.17.6-1 Kernel Configuration
[08:33:58] <abaumann> +# Linux/x86 4.17.8-1 Kernel Configuration
[08:35:20] <deep42thought> ah, now I see
[08:35:29] <deep42thought> umm
[08:35:39] <deep42thought> you're sure, you build the 4.17.8 kernel?
[08:35:45] <deep42thought> not the 4.17.6 one
[08:35:48] <deep42thought> s/build/built/
[08:35:54] <deep42thought> the config says something different :-/
[08:36:05] * deep42thought is confused
[08:36:40] <abaumann> ah. i486 built the 4.17.6 version
[08:36:49] <abaumann> but it has it's own config.i486
[08:36:56] <deep42thought> still
[08:37:08] <deep42thought> the config for i686 says 4.17.6 (in a comment)
[08:37:16] <abaumann> huh?
[08:37:53] <abaumann> I don't understand which versions are used when and from where.
[08:38:14] <abaumann> I see things fixed upstream and in the git versions of the repos I see some other versions.
[08:39:08] <abaumann> and sometimes build slaves still insist of building old versions, though a new git version is pushed.
[08:39:14] <abaumann> *on
[08:39:28] <deep42thought> ^ that might be a bug we have just solved
[08:39:36] <abaumann> ah. :-)
[08:39:39] <deep42thought> e.g. if a commit is unavailable, it was ignored in the past
[08:39:42] <deep42thought> but now it fails
[08:39:49] <abaumann> good.
[08:40:02] * deep42thought should add proper erroring
[08:43:06] <deep42thought> "I see things fixed upstream and in the git versions of the repos I see some other versions." - I cannot follow
[08:43:11] <deep42thought> what differs from what?
[08:44:07] <abaumann> https://git.archlinux.org
[08:44:09] <phrik> Title:PKGBUILD\trunk - svntogit/packages.git - Git clone of the 'packages' repository (at git.archlinux.org)
[08:44:21] <abaumann> we stil get the broken ftp download links downstream
[08:44:47] <abaumann> * abaumann remembers a discussion about trunk and stuff in git..
[08:45:15] <abaumann> ah. now it's there..
[08:45:18] <abaumann> *puzzle*
[08:45:18] <deep42thought> exactly: we do not build from trunk
[08:45:37] <deep42thought> we build from repos/{core,extra,community}-{x86_64,any}/
[08:46:10] <abaumann> ah. I have to monitor, when things get pushed there, so I can remove locally applied patches again from our PKGBUILD
[08:46:27] <deep42thought> well, the builds will simply fail :-)
[08:46:28] <abaumann> like: 089b5230767e3d52dfe93f2f550a23f0eca0a7a5 ( core/kbd: we don't have the correct download links...)
[08:46:33] <deep42thought> and then you can remove patches
[08:46:50] <abaumann> this one not, as it just overwrites variables with another set of identical variables :-)
[08:47:12] <deep42thought> yeah
[08:47:29] <deep42thought> such stuff I usually pack into an "if" clause
[08:47:31] <abaumann> make: functions/wildcard breaks with segfaults.. this is refreshing..
[08:47:39] <abaumann> yep. better.
[08:48:05] <deep42thought> if [ "${source[0]}" = '...' ]; then
[08:48:06] <deep42thought> ...
[08:48:08] <deep42thought> else
[08:48:11] <deep42thought> unset source
[08:48:11] <deep42thought> fi
[08:48:32] <deep42thought> just to make sure it really breaks if the patch moved from trunk to repos/..../
[08:48:43] <abaumann> good idea. as for linux, linux-lts
[08:48:55] <deep42thought> it's a little different there
[08:48:59] <deep42thought> but a similar idea
[08:49:00] <deep42thought> yeah
[08:49:36] <abaumann> make: alloca, how I like GNUism in C.. :->
[09:03:30] -!- City-busz has joined #archlinux32
[09:13:55] <abaumann> make: segfault in tests with globs. O my..
[09:14:14] <abaumann> we speack 64-bit here. :->
[09:14:31] <deep42thought> how can globs be 64 bit?
[09:14:51] <abaumann> there is some ABI mismatch using certain glob funcions in glibc, as I understood.
[09:15:11] <deep42thought> :-/
[09:15:19] <deep42thought> sounds like you have a lot of fun ;-)
[09:15:34] <abaumann> yeah. I just wanted to rebuild make, but hey... :-)
[09:16:18] <abaumann> it just tells me: you have to rebuild the toolchain after glibc updates, as the code is sometimes a little bit interveawed..
[09:16:28] <abaumann> *interweaved
[09:18:36] <abaumann> glob and glob64, indeed :-)
[09:21:18] <deep42thought> why exactly are we rebuilding make?
[09:23:44] <abaumann> I am rebuilding make because I try to integrate bootstrap32 into packages as i486 branch.
[09:24:00] <abaumann> but some patches are not really i486 related. make just doesn't build currently anywhere
[09:24:16] <deep42thought> ah, ok
[09:24:22] <deep42thought> shouldn't we report upstream, then?
[09:24:29] <abaumann> it is reported upstream: https://bugs.archlinux.org
[09:24:30] <phrik> Title:FS#58674 : [make] build fails with glibc 2.27 (at bugs.archlinux.org)
[09:24:34] <deep42thought> or is "anywhere" = "anywhere except x86_64"?
[09:24:44] <abaumann> no anywhere is anywhere.
[09:24:51] <abaumann> for sure the alloca thingy.
[09:25:04] <abaumann> I'm currently testing if the glob/glob64 thing is just a 32-bit thing.
[09:25:15] <abaumann> if not, there will be another bug report upstream.
[09:26:04] <deep42thought> :-)
[09:26:09] <abaumann> "anywhere is anywhere for all values of anywhere"
[09:26:32] <deep42thought> !grab abaumann
[09:26:32] <phrik> deep42thought: Bazinga!
[09:26:49] <abaumann> hehe :-)
[09:30:09] <deep42thought> I'd prefer if you would try to commit only working stuff to the master branch - or at least if you would not reschedule working packages (which do not need to be rescheduled) which then cannot be built
[09:30:28] <deep42thought> because make is currently blocking all other builds :-)
[09:36:12] <abaumann> yep. too late. the reason was that the build slave refuses to build stuff from a working git repo, so I committed it.
[09:36:44] <abaumann> what's really bad, make cannot be built with the current glibc, so we were building with a 'stale' but working make.
[09:37:19] <abaumann> I fear, there is only the options 'a) old make into build-support' or 'b) go ahead and try to fix the glob issue'
[09:37:47] <deep42thought> I could kick the build order off the build-list
[09:38:02] <deep42thought> or is there something wrong with the current make (besides being stale)
[09:38:40] <abaumann> hard to say. I thing globs with symlinks is not a too often scenario, so also the new make could be used (with a make check || true that is).
[09:39:14] <deep42thought> no, I meant the compiled make we currently have in our repositories
[09:39:52] <abaumann> ah. as it passed building, it is still ok. but it was built against an older glibc, so some segfaults may also occur, as glibc has been updated meanwhile.
[09:39:54] <buildmaster> make is broken (says nlopc43).
[09:40:08] <abaumann> so far I didn't see a segfault around globbing, so we are most likely safe.
[09:40:14] <abaumann> ..as safe as upstream is.. :-)
[09:41:30] -!- oaken-source has joined #archlinux32
[09:41:39] <abaumann> ah.. the globbing sefault think is fixed in 'make', it just has to be backported to 4.2.1
[09:41:46] <abaumann> *segfault thing
[09:42:03] <abaumann> https://sourceware.org
[09:42:09] <abaumann> what a patch monster. :-)
[09:43:10] <deep42thought> lol, noscript emitted a xss-warning on opening this :-D
[09:43:25] <abaumann> * abaumann coughs nervously
[09:44:16] <abaumann> I have to go now, be back in 1/2 hour.. then I try to fix this segfault..
[09:44:26] <deep42thought> ok, thx
[09:44:27] <deep42thought> cu
[09:44:29] <abaumann> cu
[09:44:29] -!- abaumann has quit [Quit: leaving]
[09:49:27] -!- oaken-source has quit [Ping timeout: 240 seconds]
[10:08:51] -!- abaumann has joined #archlinux32
[10:08:56] <buildmaster> Hi abaumann!
[10:09:11] <deep42thought> Hi abaumann!
[10:09:17] <deep42thought> I removed make from the build list for now
[10:09:29] <abaumann> good.
[10:09:32] <deep42thought> the next commit to the patch-source will put it there again :-)
[10:09:41] <deep42thought> that is, commit to master branch
[10:10:09] <abaumann> so, I should be carefull what I commit next.. ;-)
[10:10:25] <deep42thought> you always should be ;-p
[10:16:17] <deep42thought> ah, the buildmaster accumulates load again - during "git gc" :-D
[10:28:53] -!- oaken-source has joined #archlinux32
[11:23:56] <abaumann> so, a fix for make is underway. At least locally it builds and checks work fine. :-)
[11:24:03] <deep42thought> \o/
[11:26:19] <abaumann> so, now I continue to build make on i486.. :-)
[11:26:24] <deep42thought> :-D
[11:27:06] <abaumann> hope eli is not overly angried by my frantic error reporting upstream..
[11:27:31] <deep42thought> ... probably not overly - just the regular amount ;-)
[11:27:50] <abaumann> he seems to pull out quite some hairs with python 3.7 :->
[11:30:58] <abaumann> vim: python37.patch LOL
[11:31:14] <abaumann> one hell of fun :-)
[12:20:33] <buildmaster> girls, my database is dirty again ...
[12:28:01] -!- oaken-source has quit [Ping timeout: 256 seconds]
[12:49:22] <abaumann> mmh. is 'make' now blocked?
[13:08:05] -!- oaken-source has joined #archlinux32
[13:19:50] * buildmaster resumes sanity.
[13:24:25] <deep42thought> abaumann: buildmaster's insanity was the problem :-)
[13:38:35] <abaumann> ah. bad insanity. :-)
[14:12:24] -!- T`aZ has joined #archlinux32
[14:12:28] <T`aZ> hi
[14:12:35] <deep42thought> Hi T`aZ
[14:14:09] <T`aZ> first i'd like to say thanks for archlinux32 !
[14:14:23] <T`aZ> now here comes a question/bugreport :p
[14:15:14] <T`aZ> it seems that the package version provided is not the same as the one that should be generated
[14:15:50] <T`aZ> for example, if you build gzip manually, i'll create the package gzip-1.9-1 , however gzip-1.9-1.0 is provided
[14:16:02] <deep42thought> that's called sub_pkgrel
[14:16:08] <deep42thought> it's a feature
[14:16:09] <T`aZ> s/i'll/it'll/
[14:16:34] <deep42thought> we have this decimal in the pkgrel to indicate if we needed to rebuild some package
[14:16:54] <deep42thought> the PKGBUILD gets mangled, before the build system builds it
[14:17:14] <deep42thought> andas one step in the mangling, we add the sub_pkgrel
[14:17:29] <deep42thought> to unambiguously enumerate all builds of a package
[14:18:18] <T`aZ> but how can i keep manually built packages with that feature
[14:18:34] <deep42thought> just append ".0" to pkgrel
[14:18:35] <T`aZ> because now everytime it will ask to update it
[14:18:39] <deep42thought> but it's not neccessary, really
[14:18:44] <deep42thought> ah, ok
[14:18:52] <deep42thought> yeah, then just add ".0" to your pkgrel
[14:19:13] <deep42thought> or: rename your package (should be best practice anyway) and add a proper "provides=..." entry
[14:19:49] <deep42thought> like so: https://git.eckner.net
[14:19:50] <phrik> Title:PKGBUILD\moc-nosort - Erich/archlinuxewe - Package sources for [archlinuxewe] repository for Arch Linux and Arch Linux 32 (at git.eckner.net)
[14:20:58] <T`aZ> hmmm, thanks for the tip, i'll tro to do that
[14:21:26] <T`aZ> my use case is a bit different, im using yaourt+customizepkg to rebuild only some packages automatically
[14:22:00] <T`aZ> i can probably add a sed line to add that ".0"
[14:22:11] <deep42thought> I'd recommend a higher number
[14:22:28] <deep42thought> otherwise, you'll be asked to upgrade as soon as we rebuild that package and have sub_pkgrel=1
[14:22:59] <tyzoid> Another option is to tell pacman to ignore the custom version
[14:27:52] <T`aZ> but i want to update it when a new version is available ofc :)
[14:28:08] <deep42thought> update or rebuild?
[14:28:21] <T`aZ> rebuild indeed, sorry
[14:28:41] <deep42thought> in that case, I'd recommend a separate monitoring solution
[14:28:55] <deep42thought> like I do with the moc-nosort I showed you
[14:29:05] <deep42thought> it's a more complex solution, but IMHO cleaner
[14:56:13] <T`aZ> i went with the ugly hack to add ".99" suffix , it should be good enough for me
[14:56:20] <deep42thought> :-)
[14:56:32] <T`aZ> thank you guys for the help !
[14:56:33] <deep42thought> depends on how often we need to rebuild packages
[14:56:35] <deep42thought> np!
[15:00:34] <abaumann> We should really not build a package more than 99 times. ;-)
[15:00:42] <deep42thought> maybe haskell?
[15:00:47] <T`aZ> i hope so ;)
[15:00:48] <deep42thought> !grab abaumann
[15:00:50] <phrik> deep42thought: Bazinga!
[15:01:25] <abaumann> May I ask just out of curiosity: why are you rebuilding packages?
[15:01:39] <T`aZ> -O3 -march=native
[15:01:46] <T`aZ> for some selected libs
[15:01:51] <abaumann> ah. A former Gentoo user ;-)
[15:01:52] <T`aZ> that i use often
[15:01:59] <deep42thought> lol
[15:02:18] <T`aZ> s/gentoo/LFS/ :D
[15:02:25] <abaumann> :)
[15:03:58] <abaumann> yeah. that's a good idea.
[16:24:47] <abaumann> ah. that was my 3rd Microsoft scamming call today.. fun :-)
[16:25:09] <deep42thought> you scam microsoft?
[16:25:23] <abaumann> very funny. should I try?
[16:25:24] <abaumann> :-)
[16:26:05] <deep42thought> if you do, I guess, you'll earn a few non-scam calls from lawyers ;-)
[16:26:18] <abaumann> now, the newest trick they have: they have a local phone number from Switzerland to look more serious, the problem is they call me in Indian English, so.. :-)
[16:26:28] <deep42thought> LOL
[16:27:12] <deep42thought> they couldn't scam me in indian english - even if I wanted to, it would be too hard for me to understand what they want :-/
[16:27:29] <abaumann> :-)
[16:27:57] <abaumann> one guy I started to read my windows 8.1 key. to prove, I'm a Microsoft customer. He hung up on me..
[16:28:20] <deep42thought> now he has a valid key
[16:28:31] <abaumann> no, it's not the real key. :->
[16:28:36] <abaumann> just in the same scheme..
[16:28:37] <deep42thought> hehe
[16:29:03] <abaumann> I should really once set up a vm and let the guy onto the machine and see what they try to do..
[16:29:27] <abaumann> there are tons of youtube videos around..
[16:29:27] <deep42thought> if he's in, just pull the plug and catch him forever :-)
[16:29:38] <abaumann> :-)
[16:29:55] <abaumann> how sad is it to have to work for such a company..
[16:31:30] <deep42thought> my friend was in india for a year and he said one thing to me: one thing, which is hard to imagine, but true, is, that whatever possible job exists in india, it's done by someone - e.g. if there is a queue somewhere, theres also someone selling tea to the people waiting
[16:32:05] <abaumann> sounds reasonable. tea selling in a queue is a honest job.
[16:32:24] <deep42thought> yeah, but this principle applies to _all_ possible jobs
[16:32:27] <abaumann> ..I would prefer calls of people selling me some tea :-)
[16:32:33] <deep42thought> LOL
[16:32:34] <abaumann> oups.
[16:33:32] <abaumann> ..on the other hand.. I had once an Italian guy trying to sell me some prixy Italian food
[16:33:36] <abaumann> *pricy
[16:33:52] <abaumann> that ended in some written letters to stop being bugged.
[16:34:00] <abaumann> the record was 151 missed calls on my phone.
[16:34:05] <abaumann> per day!
[16:34:07] <deep42thought> O.o
[16:34:30] <abaumann> though the olive oil looked genuine ;-)
[16:34:43] * deep42thought takes distance from the thought of calling abaumann and selling him tea
[16:35:24] <abaumann> (or olive oil) :-)
[16:46:32] -!- deep42thought has quit [Quit: Leaving.]
[16:48:27] -!- abaumann has quit [Ping timeout: 240 seconds]
[17:04:03] <buildmaster> firefox-developer-edition is broken (says buildknecht).
[17:26:22] <tyzoid> abaumann: Windows 93 is awesome for that sort of thing: https://www.windows93.net
[17:26:25] <phrik> Title:WINDOWS93 (at www.windows93.net)
[17:39:08] -!- abaumann has joined #archlinux32
[17:39:09] <buildmaster> Hi abaumann!
[17:39:21] <abaumann> tyzoid: windows93, didn't know about that one.. :-)
[17:39:47] <tyzoid> yeah, windows93 fullscreen in a vm, and they'll never know
[17:40:06] <tyzoid> better yet if you make a custom linux box that boots to chromium fullscreen with that
[17:52:27] <buildmaster> firefox is broken (says rechenknecht).
[18:00:55] <tyzoid> abaumann: If you're interested, I have a youtube walkthrough where I do just that :)
[18:01:25] <abaumann> you do? fun? yeah. I'm interested.. :-)
[18:04:12] <tyzoid> abaumann: https://youtube.com
[18:04:14] <phrik> Title:Arch Linux Kiosk Installation Walkthrough - With sound - YouTube (at youtube.com)
[18:04:44] <abaumann> ...and now you have the attention of all spamers.. ;-)
[18:06:07] <tyzoid> lol
[18:06:20] <tyzoid> Only difference is at 11min, put windows93 instead of netflix xD
[18:12:32] <abaumann> oh. I just missed phone call number 4.. *sniff*
[18:24:12] -!- deep42thought has joined #archlinux32
[18:24:12] <buildmaster> Hi deep42thought!
[18:24:16] <tyzoid> wb
[18:24:47] <abaumann> hi
[18:25:38] <deep42thought> abaumann: maybe you should call back :-)
[18:25:57] <abaumann> deep42thought: sure, it's a local call in the end.. ;-)
[18:26:19] <abaumann> ui. what's the fix for things like;
[18:26:20] <abaumann> if [[ $(find "$pkgdir/usr/lib/perl5/" -name "*.so") ]]; then
[18:26:20] <abaumann> _perlver_min=$(perl -e '$v = $^V->{version}; print $v->[0].".".($v->[1]);')
[18:26:20] <abaumann> _perlver_max=$(perl -e '$v = $^V->{version}; print $v->[0].".".($v->[1]+1);')
[18:26:23] <abaumann> depends+=("perl>=$_perlver_min" "perl<$_perlver_max")
[18:26:26] <abaumann> fi
[18:26:34] <abaumann> https://bugs.archlinux.org
[18:26:35] <phrik> Title:FS#59393 : [perl-xml-parser] pkgver in depends is not allowed to be empty. (at bugs.archlinux.org)
[18:27:13] <deep42thought> ah, that is due to our makepkg
[18:27:19] <deep42thought> because it's too recent
[18:27:22] <deep42thought> or too late
[18:27:23] <abaumann> no, I also get it on 64-bit
[18:27:28] <deep42thought> anyway - I'm afk for dinner
[18:27:32] <abaumann> with staging-x86_64-build
[18:27:40] <abaumann> anyways, eli will be happy and see it. :-)
[18:27:47] <abaumann> have a nice dinner. :-)
[18:33:43] -!- isacdaavid has joined #archlinux32
[18:50:04] -!- abaumann has quit [Quit: leaving]
[18:52:27] <tyzoid> abaumann / deep42thought: Planned downtime tonight on packages.archlinux32, bbs.archlinux32, bugs.archlinux32, znc.tyzoid, and my mirrors, between 0200z and 0400z.
[18:52:52] <tyzoid> Total estimated downtime is ~15 minutes, but I wanted to have a larger window in case stuff happens.
[18:57:28] <deep42thought> thx, abaumann, dinner was ok (we got screamed at for several minutes, because my wife cut the nectarines once too often)
[18:57:36] <deep42thought> tyzoid: thx for the notice :-)
[18:58:51] <deep42thought> what timezone is "z"?
[18:59:13] <deep42thought> +26?
[19:00:18] <tyzoid> deep42thought: zulu time (UTC)
[19:01:10] <deep42thought> ah, ok
[19:01:36] <tyzoid> If I'm going to do maintaince, might as well do it while y'all are asleep :)
[19:12:28] <deep42thought> btw: I thought a little about putting report-installed-packages and co. into i686/releng, and I think it's not such a brilliant idea
[19:12:56] <deep42thought> because it would increase maintenance effort and they're not really official packages
[19:13:56] <deep42thought> tyzoid: what are your current plans with sso, currently?
[19:14:45] <tyzoid> deep42thought: To finish it, at some point
[19:15:57] <deep42thought> do you want to make bbs authenticate to accounts - or do you want to keep the direction as-is?
[19:16:27] -!- AndrevS has joined #archlinux32
[19:17:54] <tyzoid> Ideally, make bbs auth to accounts
[19:26:04] <deep42thought> ah, ok, so this is still the plan :-)
[19:40:59] <tyzoid> mainly just a matter of finding the time to do it
[19:41:35] <deep42thought> :-)
[20:23:26] <elibrokeit> abaumann, deep42thought: the makepkg thing is because makepkg now lints packages for invalid metadata, but for things that are set inside the package() function we locate it using grep :p
[20:23:57] <elibrokeit> The fix is that people should really stop doing this... It makes srcinfo generation quite derpy for example
[20:24:14] <deep42thought> is this different from the "makepkg --printsrcinfo does not undestand all bash" issue?
[20:24:31] <tyzoid> abaumann / deep42thought: if you can think of any tests you guys want to run on the varnish system, let's get that done within the next two days, since I'll be disabling that sometime saturday
[20:24:37] <elibrokeit> It's the exact same issue manifesting slightly differently
[20:25:01] <deep42thought> tyzoid: we could put varnish in front of the master mirror :-D
[20:25:09] <elibrokeit> In this case, if you try using depends=($var) but $var is not set yet
[20:25:14] <deep42thought> but otoh this would only be useful if it was a permanent solution
[20:26:25] <tyzoid> Yeah, though it could still be useful for testing.
[20:26:34] <tyzoid> It's only $5/mo for the node in the netherlands
[20:26:39] <elibrokeit> So again the green is stupid and cannot detect the preceding lines any more than it could detect true &&
[20:26:45] <elibrokeit> *grep
[20:26:52] <tyzoid> I'd be willing to put that up for a few months
[20:27:02] <tyzoid> if we wanted an extended test
[20:27:45] <deep42thought> elibrokeit: yeah, and one cannot reasonably expect grep to execute perl to get the current version, either :-)
[20:28:09] <deep42thought> tyzoid: I don't need (to test) varnish, currently
[20:28:22] <deep42thought> it's a nice toy, but I don't see, that we really _need_ it :-)
[20:28:41] <tyzoid> deep42thought: Wouldn't it fix some bandwidth issues with the master mirror / build systems?
[20:28:51] <deep42thought> not really
[20:29:00] <deep42thought> the bottleneck is the buildmaster currently
[20:29:09] <deep42thought> and there, caching does not make sense
[20:29:14] <tyzoid> right
[20:29:22] <deep42thought> or rather: I circumvent the buildmaster already if possible
[20:29:39] <deep42thought> packages are uploaded directly into some transition directory on the master mirror
[20:29:49] <deep42thought> and the buildmaster uses rsync to copy the packages over
[20:29:56] <tyzoid> or maybe a vps as a staging server?
[20:30:01] <deep42thought> and allows the master mirror to look into the transition directory
[20:30:12] <deep42thought> staging what?
[20:30:26] <tyzoid> for the build slaves to upload their packages to, instead of to the buildmaster
[20:30:38] <deep42thought> they upload to the master mirror
[20:30:52] <deep42thought> we could replace the buildmaster
[20:31:04] <deep42thought> but I don't see much tasks which could be split off
[20:31:19] <deep42thought> running the database, maybe
[20:31:21] <tyzoid> I think we're heading for a separate discussion at this point
[20:31:26] <deep42thought> oh
[20:31:27] <deep42thought> sry
[20:31:35] <tyzoid> No, it's all good
[20:31:56] <tyzoid> Just wanted to delineate the part where we're speculating, and the part where I spend money
[20:32:19] <tyzoid> keep in mind the slave db on my system in the US too
[20:32:28] <deep42thought> and on my developement box
[20:32:42] <deep42thought> and on polichronucci's server, once he gets php+mysql running there
[20:33:10] <deep42thought> but those are all read-only-slaves
[20:36:17] <elibrokeit> deep42thought: well, it does get eval'ed
[20:36:58] <deep42thought> elibrokeit: so "depends=(perl=$(perl --version ...)) works?
[20:37:06] * deep42thought forgot a quote
[20:37:57] <elibrokeit> deep42thought: yes
[20:38:06] <elibrokeit> assuming perl is installed
[20:38:34] <elibrokeit> which, since this runs before --syncdeps, is not guaranteed...
[20:39:17] <elibrokeit> IMO code like this should behave more like find_libdepends, and have a "base" state and something that gets updated during package()
[21:49:46] -!- AndrevS has quit [Quit: umount /dev/irc]
[23:01:06] -!- deep42thought has quit [Quit: Leaving.]