#archlinux-ports | Logs for 2017-11-24

Back
[00:03:59] -!- because has quit [Ping timeout: 248 seconds]
[00:05:44] -!- because has joined #archlinux-ports
[00:22:20] -!- because has quit [Ping timeout: 255 seconds]
[00:27:19] -!- because has joined #archlinux-ports
[01:23:48] -!- isacdaavid has joined #archlinux-ports
[01:38:22] <fsckd> !ops
[01:38:22] <phrik> fsckd: Error: no list of ops to ping configured for this channel.
[01:40:10] <fsckd> !ops
[01:40:10] <phrik> brtln, deep42thought, tyzoid
[01:40:16] <fsckd> err
[01:40:28] <fsckd> !ops
[01:40:28] <phrik> brtln, deep42thought, fsckd, tyzoid
[01:40:33] <fsckd> \o/
[01:40:52] <fsckd> so people needing help with trolls can just issue that
[03:59:58] <cypa[m]> snd-hda-intel model=auto
[04:00:32] <cypa[m]> without this no sound in headsets
[04:00:43] <cypa[m]> while Ubntu autoconfigured
[04:01:06] <cypa[m]> while Ubuntu autoconfigured
[04:16:24] <cypa[m]> nice, sound in headset disappeared again
[04:32:21] -!- T`aZ_ has quit [Ping timeout: 240 seconds]
[04:34:29] -!- T`aZ has joined #archlinux-ports
[04:35:41] -!- hringriin has quit [Quit: ^Cc12^CbSee ^Cb^Cc04^Cya^Cc!]
[04:41:19] -!- because has quit [Ping timeout: 248 seconds]
[04:43:22] -!- and has joined #archlinux-ports
[05:01:35] -!- and has quit [Ping timeout: 248 seconds]
[05:03:53] -!- because has joined #archlinux-ports
[05:13:42] -!- hringriin has joined #archlinux-ports
[05:58:56] -!- hringriin_ has joined #archlinux-ports
[06:02:06] -!- hringriin has quit [Ping timeout: 268 seconds]
[06:02:07] hringriin_ is now known as hringriin
[07:04:28] -!- deep42thought has joined #archlinux-ports
[07:28:07] -!- deep42thought has quit [Quit: Leaving.]
[08:25:21] -!- deep42thought has joined #archlinux-ports
[08:35:34] <cypa[m]> magic ... headset sound is back
[08:44:42] <deep42thought> strange
[09:00:04] -!- abaumann has joined #archlinux-ports
[10:20:01] -!- because has quit [Ping timeout: 248 seconds]
[10:20:44] -!- because has joined #archlinux-ports
[11:13:14] <deep42thought> crap, my sources cache is not reachable from the internet
[11:13:39] <deep42thought> actually, it's no surprise, but I haven't noticed before :-/
[11:15:43] <abaumann> didn't see any failed packages because of that
[11:16:18] <deep42thought> It just kicks in, when the source is unavailable and it tries to download it by its hash
[11:16:34] <deep42thought> and then, only if someone from "the internet" tries to build the package
[11:16:44] <abaumann> ah. ok.
[11:16:45] <deep42thought> e.g. not nlopc46, nlopc43 and rechenknecht
[11:17:17] <abaumann> https://bbs.archlinux32.org
[11:17:18] <phrik> Title: multiple slaves building the same package / Creating/Maintaining Packages / Arch Linux 32 Forums (at bbs.archlinux32.org)
[11:18:04] <abaumann> most likely not a problem, just inefficient. :-)
[11:19:16] <deep42thought> it indicates a problem
[11:19:48] <deep42thought> usually, that the build script on the slave is outdated and thus can't correctly communicate with the build master
[11:20:04] <deep42thought> then it orders strange stuff
[11:20:29] <abaumann> so it's crutial to keep the build slaves up to date
[11:20:33] <deep42thought> yes
[11:20:45] <deep42thought> at least currently
[11:20:52] <deep42thought> because I added the "sub-pkgrel"
[11:20:58] <abaumann> I had once a build system which fetched the current state via git pulls before doing anything..
[11:21:00] <deep42thought> and this changed the interface to the build master
[11:21:11] <deep42thought> yep, I'm doing this, too
[11:21:16] <deep42thought> but it's asynchronous
[11:21:29] <abaumann> true
[11:22:06] <deep42thought> ah, btw: this particulat post was intentional - as rechenknecht crashed and I wanted to accellerate the build
[11:22:40] <abaumann> ah. this explains it.
[12:08:12] <abaumann> Do you want to delete it? [Y/n] error: virtualbox-guest-dkms: signature from "Andreas Baumann (sign) <mail@andreasbaumann.cc>" is unknown trust
[12:08:20] <abaumann> mmh. I have to check again. still not signed.
[12:08:34] <deep42thought> hmm
[12:12:19] <abaumann> ah.. still missing a signature. :-)
[12:12:25] <deep42thought> huh?
[12:12:52] <abaumann> ballogyor never replied.
[12:13:08] <abaumann> https://pgp.mit.edu
[12:13:10] <phrik> Title: Search results for '0xc8e8f5a0af9ba7e7' (at pgp.mit.edu)
[12:13:35] <deep42thought> three signatures are sufficient
[12:13:47] <deep42thought> and I count 4
[12:13:55] <abaumann> right.
[12:13:57] <deep42thought> do you have the current keyring installed?
[12:14:21] <abaumann> I do. I build virtualbox iso and I'm currently building virtualbox itself.
[12:14:44] <abaumann> virtualbox-guest-dkms: :: Import PGP key 2048R/16194A82231E9EF823562181C8E8F5A0AF9BA7E7, "Andreas Baumann (sign) <mail@andreasbaumann.cc>", created: 2017-11-02? [Y/n]
[12:15:01] <abaumann> This sounds like the slave didn't press 'yes' ;-)
[12:15:12] <deep42thought> :-D
[12:15:22] <abaumann> This was not on my slave..
[12:15:24] <deep42thought> yes | staging-i686-build
[12:15:41] <deep42thought> hmmm
[12:15:56] <abaumann> can we find out, on which slave virtualbox-guest-dkms was built?
[12:16:04] <deep42thought> yes
[12:16:09] <deep42thought> if it's still broken
[12:16:22] <abaumann> it is. or I I build it on my slave..
[12:16:42] <deep42thought> nope, it's not
[12:16:52] <abaumann> right. mmh.
[12:17:19] <deep42thought> the *.broken file contains a list of slaves who built it and reported it as broken
[12:17:35] <deep42thought> but this file is gone as soon as the package is built successfully
[12:18:22] <abaumann> I'll retrigger the build and build it on my slave. After virtualbox itself is built..
[12:19:24] <deep42thought> where did you get that log from, actually?
[12:19:55] -!- wastl has joined #archlinux-ports
[12:20:14] <abaumann> https://buildmaster.archlinux32.org
[12:20:30] <abaumann> Now it's gone. :-)
[12:20:35] <deep42thought> ah, ok
[12:57:11] -!- p71 has quit [Read error: Connection reset by peer]
[13:15:20] -!- wastl has quit [Quit: Lost terminal]
[13:26:03] -!- wyre has quit [Remote host closed the connection]
[13:42:06] <deep42thought> abaumann: it looks, like makechrootpkg is missing a 'pacman-key --populate'
[13:42:16] <abaumann> oh.
[13:43:13] <deep42thought> or maybe pacstrap misses to execute that at some point ...
[13:44:02] <abaumann> ok. I'll update the builder and devtools32 workspace and I reinstall the slave's chroot, just to be safe.
[13:44:25] <deep42thought> no, this was on my slave
[13:44:32] <abaumann> ah.
[13:44:36] <deep42thought> I haven't solved the issue in devtools32, yet
[13:44:52] <abaumann> So, I'll just do a pacman-key --populate to be safe..
[13:45:03] <deep42thought> inside the chroots
[13:45:08] <abaumann> yes, right.
[13:45:56] <deep42thought> I just inserted "pacman-key --populate" before the "... makepkg ..." line in _chrootbuild() in makechrootpkg
[13:46:04] <deep42thought> that solved it for me, but seems rather unclean :-/
[13:46:29] <deep42thought> ah, and the key will be only activated in the working copy of the chroot ...
[13:46:38] <deep42thought> so the change is not permanent
[14:01:20] <deep42thought> hmm, some keyring is installed in /build/.gnupg, but I'm unable to update that one ...
[14:01:33] <deep42thought> Why is this all so complicated?
[14:01:42] <abaumann> Ask upstream. :-)
[14:01:56] <deep42thought> It was a rethorical question
[14:01:59] <abaumann> :-)
[14:03:28] <abaumann> mkchrootpk echo 'keyserver-options auto-key-retrieve' >> "$copydir/build/.gnupg/gpg.conf
[14:04:37] <deep42thought> I think, that was me
[14:06:15] <deep42thought> that was for retreiving signing keys of package sources
[14:06:18] <deep42thought> I think
[14:34:26] -!- woffs has joined #archlinux-ports
[14:47:01] -!- guys has quit [Ping timeout: 240 seconds]
[15:24:19] -!- woffs has parted #archlinux-ports
[15:25:45] -!- guys has joined #archlinux-ports
[15:29:34] -!- deep42thought has quit [Quit: Leaving.]
[15:59:44] -!- vvv has joined #archlinux-ports
[16:00:06] vvv is now known as Guest26966
[16:07:49] <Guest26966> Hello! Is anyone related to new 32-bit arch repos here?
[16:09:26] <fsckd> the devs of Arch Linux 32 are very active here. if you need to speak with one, they will be able to in a bit. if you have a specific question, you should go ahead and ask it.
[16:10:51] <Guest26966> Some time ago new repos started to upgrade software. I have cups-filters conflict so it's unable to upgrade via pacman.
[16:11:15] <Guest26966> :: cups-filters: installing poppler (0.61.1-1) breaks dependency 'poppler=0.60.1'
[16:14:15] <Guest26966> I just want to figure out which version of cups-filters is really working with poppler 0.61.1-1 to be able to upgrade
[16:15:35] <fsckd> this error is on Syu?
[16:17:01] <Guest26966> Yes, it is
[16:20:09] <abaumann> Hi, what is your version of cups-filters?
[16:20:36] <Guest26966> Currently the system has 1.17.9-2 installed. It depends on exactly poppler=0.60.1
[16:20:42] <abaumann> Both on testing and stable I have cups-filters-1.17.9-3 poppler-0.61.1-1
[16:21:07] <abaumann> I suspect, this is now a side effect of the package subversions.
[16:21:31] <Guest26966> Ok, that means it's Parabola maintainers fail =)
[16:21:40] <abaumann> So poppler=0.60.1 is too strict and 0.60.1-1 is a rebuild of the package breaking the requirement in cups-filters
[16:22:29] <abaumann> As upstream Archlinux doesn't do those subversions..
[16:22:42] <abaumann> deep42thought: I think, this one is for you. :-)
[16:23:40] <abaumann> on the other hand: why does it work for me on both stable and testing.
[16:24:14] <Guest26966> Because you are not using 'libre' repos obviously
[16:25:29] <abaumann> libre who?
[16:25:50] <Guest26966> Hope there are some Parabola people too because it's their responsibility field
[16:26:39] <abaumann> Ah. this is an arch-based distro. Sorry. There are so many out there.. :-)
[16:27:10] <Guest26966> They promised continuation of 32-bit support based on archlinux32 repos
[16:27:15] <abaumann> Yep. This is a Parabola problem then.. all I can say is that the Archlinux32 packages seem fine.
[16:28:47] <abaumann> Ok. So we have to see how Parabola is seeding from Archlinux32 exactly..
[16:29:42] <abaumann> Let me hop onto the parabola IRC channel and ask them about this..
[16:30:03] <Guest26966> Sadly they have lots of dependency problems time to time. Ok
[16:30:57] <abaumann> Dependencies are really hard some times.. ask deep42thought. :-)
[16:31:16] <Guest26966> Despite this I know that arch32 repos refused to upgrade sbcl
[16:31:28] <abaumann> yes.
[16:31:49] <abaumann> See https://bugs.archlinux32.org
[16:31:51] <phrik> Title: FS#5 : [sbcl] fails to compile (at bugs.archlinux32.org)
[16:32:16] <abaumann> and https://bbs.archlinux32.org
[16:32:17] <phrik> Title: sbcl / Creating/Maintaining Packages / Arch Linux 32 Forums (at bbs.archlinux32.org)
[16:41:23] <Guest26966> By the way, do arch32 guys have plans to setup web-interface for packages like official one in https://www.archlinux.org or not?
[16:41:24] <phrik> Title: Arch Linux - Package Search (at www.archlinux.org)
[16:41:50] <tyzoid> Guest26966: Possibly, there's a bunch of stuff we're looking to do
[16:42:22] <tyzoid> I'm currently battling scope creep on this bugtracker thing
[16:42:27] <abaumann> mmh. one problem would be that we don't have our own PKGBUILDs, but diffs.
[16:42:41] <abaumann> It would look a little bit differently, I think..
[16:43:08] <abaumann> What I'm more interested in is, how is Parabola feeding their libre repository?
[16:43:25] <abaumann> For ARM they take upstream I suppose, so for x86_64.
[16:43:40] <abaumann> If they maintain the 32-bit versions on their own, then good look. :-)
[16:43:49] <abaumann> s/look/luck/
[16:44:49] <abaumann> oups.. and it looks Spanish to me. :-)
[16:44:53] <tyzoid> abaumann: how hard is it to automatically apply the diffs?
[16:45:00] <abaumann> IRC chats are somewhat challenging.
[16:45:29] <abaumann> Not too hard: deep42thought had a post on IRC lately..
[16:45:46] <tyzoid> I'm thinking if we can auto-apply the patch, we can generate a pkgbuild for download
[16:46:06] <abaumann> Isn't the final PKGBUILD baked into the package?
[16:46:16] <tyzoid> IIRC that's the .SRCINFO
[16:46:22] <abaumann> right.
[16:46:25] <abaumann> (. path/to/original/PKGBUILD; . path/to/patched/PKGBUILD; declare -f prepare)
[16:46:39] <abaumann> but there are also diffs of the form source+=()
[16:46:49] <abaumann> patches in the PKGBUILD bash script body itself
[16:46:53] <abaumann> additions..
[16:46:55] <abaumann> mmh.
[16:47:29] <abaumann> ah. I was thinking in SRPMS. ;-)
[16:47:31] <tyzoid> If we can get some documentation from our downstream partners about what they need from us, it'd be useful to keep as guidelines
[16:47:38] <tyzoid> and/or feature requests
[16:48:52] <abaumann> For sure they need the final PKBGUILD and all patches which go along 32-bit..
[16:49:20] <tyzoid> abaumann: If we have a script that can do that, I can piece it into a web interface.
[16:49:37] <tyzoid> or, you know, we could just distribute the script :P
[16:49:59] <abaumann> Aeh.. :-)
[16:50:18] <tyzoid> My question is where the buildmaster stores package info.
[16:50:40] <tyzoid> since I'd want to query against that for the web interfaces.
[16:50:50] * tyzoid heads to github
[16:51:24] <abaumann> Well, on the slaves you get a PKGBUILD which is basically a cat of the original and the second one interpreted on top.
[16:51:50] <abaumann> Those are the same as you see on the build slaves..
[16:52:07] <abaumann> *starts to build a package to see how it looks like*
[16:58:25] <abaumann> aha: https://git.parabola.nu
[16:58:38] <abaumann> it breaks. So, they used gitsvn from upstream directly?
[17:00:24] <abaumann> tyzoid: showing the PKGBUILD is quite hard. You have to read the orignial PKGBUILD, then append the diff one from Archlinux32, then let bash interpret it and the convert back to a texrt file.
[17:00:39] <abaumann> But, the concatenated file is also a valid PKBGUILD file.
[17:01:09] <abaumann> So, I would show directly that one, though it looks a little bit funny with the eval/prepare, etc. at the end.
[17:01:33] <abaumann> Let's wait for deep42thought and what he has up in his sleeve :-)
[17:12:15] -!- because has quit [Ping timeout: 248 seconds]
[17:12:42] <tyzoid> abaumann: Theoretically, if I containerize that process, and don't accept user input, it should* be fine
[17:13:12] <abaumann> that's true
[17:13:47] -!- because has joined #archlinux-ports
[17:13:55] <tyzoid> So... the only information about where packages are is stored in the master mirror's repo database files?
[17:14:15] <tyzoid> abaumann: Can you let me know if I'm missing something from the ^ conclusion?
[17:15:08] <abaumann> I'm most likely the wrong guy to ask :-)
[17:15:17] <tyzoid> Yeah, but you've got a build-slave
[17:15:20] <tyzoid> I don't :P
[17:15:43] <abaumann> The slave is quite dumb and get's all his commands from the master.
[17:15:52] <tyzoid> figured.
[17:15:59] <abaumann> as it should be. :-)
[17:17:21] <abaumann> Sort of: gitsvn repos for packages and community form the master.
[17:17:46] <abaumann> then there is some trickery to remove blacklisted stuff (would for instance also be nice to see the buildmaster information in the web front)
[17:19:01] <abaumann> find_pkgbuilds is quite a central function in common-functions
[17:19:35] <tyzoid> There's always the possibility of just scraping https://buildmaster.archlinux32.org
[17:19:36] <phrik> Title: Buildmaster for Archlinux32 packages (at buildmaster.archlinux32.org)
[17:19:37] <tyzoid> but I'd rather not
[17:19:45] <abaumann> I hacked a test-build-package just doing a local build, completly isolated from the rest of the build system.
[17:19:45] <tyzoid> plus, there's got to be a better way to query info
[17:20:06] <abaumann> No scraping. There must be one central point of information.
[17:20:27] <tyzoid> abaumann: where's this find_pkgbuilds function?
[17:20:51] <abaumann> in the builder in common-functions
[17:22:01] <abaumann> The scheme for PKGBUILD_mod is simple: 'repository'/package/PKBUILD
[17:22:15] <abaumann> upstream is split in several places with some precendence
[17:22:57] <abaumann> I would actually ask deep42thought to provide a script generating such a package list..
[17:23:28] <abaumann> ..as far as I can see there is no such thing: there are just the build-lists and such being persisted into the 'work' subdir
[17:23:35] <tyzoid> I could use pacman to parse those repo-dbs
[17:23:56] <abaumann> yeah. but that's again derived data. :-)
[17:23:57] <tyzoid> that'd get the package list
[17:24:14] <abaumann> For a first test this should be fine.
[17:25:17] <tyzoid> :/
[17:25:28] <tyzoid> $ pacman -Ss 'community.*' only returns 4 packages on mainline
[17:25:57] <tyzoid> So it doesn't include the repo name in the regex
[17:26:36] <tyzoid> $ pacman -Sql community
[17:26:41] <tyzoid> ^ That's what I was looking for :)
[17:26:48] <abaumann> yep :-)
[17:27:13] <tyzoid> Alright. I'll rebuild my i686 vagrant, then throw that on my server
[17:28:43] -!- because has quit [Ping timeout: 250 seconds]
[17:29:34] -!- because has joined #archlinux-ports
[17:29:47] <tyzoid> And the vagrant box seems a bit big, but oh, well.
[17:30:04] <tyzoid> I might just end up remaking it, since I think I've got gui stuff on this one
[17:31:15] <abaumann> Well.. there is a good side effect of having a machine which has all packages installed. You can do tests on the packages..
[17:31:47] <abaumann> .. for instance, I dream of a script going through all binaries hunting "modern" Intel instructions from SSE2 upwards.
[17:36:48] <tyzoid> Isn't there a free GNU disassembler from x86 to asm?
[17:37:24] <abaumann> yeah. but it's not that simple.
[17:37:37] <abaumann> first you have to read ELF to see where the sections are.
[17:37:51] <abaumann> then there can be binaries with mixed instruction sets
[17:38:05] <tyzoid> :/
[17:38:41] <abaumann> I had a i386-binary today of vivaldi which was executing SSE2 moves on MMX registers.
[17:38:51] <abaumann> So, it failed on an old laptop of mine.
[17:41:08] <tyzoid> Can't virtualbox disable those extensions?
[17:41:37] <abaumann> Also can libvirt, yeah. That's what I'm usually doing: I'm emulating various sets of cpus to find those bugs..
[17:41:43] <tyzoid> Nice.
[17:41:46] <abaumann> ..but it's a little bit tedious.
[17:41:50] <tyzoid> Virtualbox doesn't look like it can
[17:42:11] <abaumann> in libvirt you have a config XML file, there you can set CPU properties.
[17:43:11] -!- because has quit [Ping timeout: 255 seconds]
[17:43:44] <tyzoid> nice
[17:43:57] -!- because has joined #archlinux-ports
[17:43:59] <tyzoid> abaumann: have you seen openQa?
[17:44:14] <tyzoid> It's the OpenSuse project they use for tumbleweed.
[17:44:46] <tyzoid> We've been looking at it for a while, but it seems like a bear to repackage, and we haven't had the time required to get an instance working on Arch.
[17:44:56] <abaumann> ah. no. didn't know.
[17:45:12] <abaumann> I used the Arch machines on the OpenbuildService.
[17:45:20] <abaumann> I would suspect, they use the same infrastructure.
[17:45:28] <tyzoid> Afaik, they do
[17:45:41] <tyzoid> We wanted to integrate that without OBS, to use our own build system
[17:45:51] <tyzoid> but we didn't even get openqa itself up and runnign
[17:47:12] -!- xes has joined #archlinux-ports
[17:48:35] <tyzoid> I think there's a thread about it somewhere on the forum
[17:50:57] <tyzoid> abaumann: https://bbs.archlinux32.org
[17:50:59] <phrik> Title: What to test? / Testing / Arch Linux 32 Forums (at bbs.archlinux32.org)
[17:51:05] <tyzoid> Not sure if you've seen that
[18:00:50] -!- shentey has joined #archlinux-ports
[18:11:12] -!- because has quit [Ping timeout: 260 seconds]
[18:12:20] -!- because has joined #archlinux-ports
[18:12:54] -!- because has quit [Excess Flood]
[18:13:05] -!- because has joined #archlinux-ports
[18:14:54] <tyzoid> Alright, once it finishes uploading the vagrant box will be available in the vagrant cloud here: https://app.vagrantup.com
[18:14:55] <phrik> Title: Vagrant box archlinux32/archlinux32 - Vagrant Cloud (at app.vagrantup.com)
[18:15:07] <tyzoid> That should simplify distribution
[18:15:28] <abaumann> cool :-)
[18:20:06] -!- Guest26966 has quit [Quit: Page closed]
[18:24:16] <tyzoid> and it's done
[18:27:45] <tyzoid> abaumann: At some point, I'll make a libvirt build
[18:27:56] <tyzoid> probably not until next month's iso, though
[18:38:39] -!- because has quit [Ping timeout: 248 seconds]
[18:39:42] -!- because has joined #archlinux-ports
[19:20:04] -!- guys has quit [Ping timeout: 268 seconds]
[19:31:41] -!- zaungast has joined #archlinux-ports
[20:05:37] -!- p71 has joined #archlinux-ports
[21:05:36] -!- abaumann has quit [Quit: ChatZilla 0.9.92 [SeaMonkey 2.49.1/20171107234153]]
[21:26:15] -!- shentey has quit [Read error: Connection reset by peer]
[21:27:37] -!- shentey has joined #archlinux-ports
[22:29:05] -!- shentey has quit [Quit: Konversation terminated!]
[22:43:54] -!- isacdaavid has quit [Quit: Leaving.]
[22:44:52] -!- because has quit [Ping timeout: 240 seconds]
[22:46:04] -!- because has joined #archlinux-ports
[23:04:05] -!- isacdaavid has joined #archlinux-ports