#archlinux32 | Logs for 2018-04-15

[02:25:59] -!- davor has quit [Ping timeout: 260 seconds]
[02:32:45] -!- davor has joined #archlinux32
[04:11:57] -!- tigrmesh has quit [Ping timeout: 276 seconds]
[04:24:31] -!- eduardoeae has quit [Ping timeout: 256 seconds]
[04:31:02] -!- tigrmesh has joined #archlinux32
[06:52:38] <girls> tyzoid: https://bbs.archlinux32.org
[06:52:39] <phrik> Title:International forums / Pacman / Pacman Upgrades / Arch Linux 32 Forums (at bbs.archlinux32.org)
[06:53:15] <tyzoid> girls: Sure, though I won't be able to help much. IIRC you and abaumann both speak German?
[06:53:22] <girls> yes
[06:53:43] <girls> my question was more if you can / are willing to set up language specific sub-forums
[06:53:50] <girls> or rather: how it works in general
[06:54:03] <tyzoid> Can and willing, but I can't really do much in there. USA Education FTW
[06:55:47] <girls> I understand that - no problem
[06:55:47] <girls> I think, it's a little bold to register archlinux32.de, too
[06:55:47] <girls> so we'd need to go with bbs.archlinux32.org/de/... or de.bbs.archlinux32.org
[06:55:47] <girls> what do you prefer?
[06:58:07] <tyzoid> that's still going to be hard to differentiate
[06:58:17] <tyzoid> I'm thinking of just making a Language Specific category
[06:58:28] <girls> you would set up a second forum?
[06:58:34] <girls> or how should it work?
[06:58:42] <tyzoid> https://bbs.archlinux32.org
[06:58:43] <phrik> Title:Arch Linux 32 Forums (at bbs.archlinux32.org)
[06:58:53] <tyzoid> I added a "Language Specific" category
[06:59:04] <girls> ah, even better :-)
[06:59:49] <girls> can I move a thread somewhere?
[07:00:31] * girls found it :-)
[07:02:30] <girls> as soon as you add a russian category, we need to revise the spam filter :-D
[07:04:00] <tyzoid> lol
[07:05:50] <girls> or we just deactivate it for the russian forum, but then all the spam gets there ...
[07:07:28] <tyzoid> I don't think we need a russian forum
[07:07:29] <tyzoid> yet
[07:07:40] <tyzoid> We'll see. AFAIK, none of us know russian
[07:07:53] <tyzoid> And I don't know of any russian users.
[07:08:04] <girls> I can read the letters
[07:08:11] <girls> but that doesn't help much :-D
[07:08:14] <tyzoid> I'm sure there are, but I imagine they'd ask for support in English
[07:08:16] <tyzoid> yeah, lol
[07:13:20] <girls> one guy from russia asked in italian, but I think it was spam, though
[07:18:53] -!- tyzoid_ has joined #archlinux32
[07:19:08] <tyzoid> test
[07:20:01] <tyzoid_> test
[07:22:03] -!- deep42thought has joined #archlinux32
[07:22:23] <tyzoid> deep42thought / girls: ZNC migration is complete. Refresh / reconnect to get the new IP
[07:24:00] <tyzoid> test
[07:24:07] <deep42thought> ah, I need to re-register, too
[07:24:23] <tyzoid> perhaps, I think I have sasl enabled, so I don't need to
[07:24:38] -!- tyzoid has quit [Disconnected by services]
[07:24:49] tyzoid_ is now known as tyzoid
[07:25:24] -!- tyzoid_ has joined #archlinux32
[07:25:31] -!- mode/#archlinux32 [+imo tyzoid] by ChanServ
[07:25:58] -!- mode/#archlinux32 [-o tyzoid] by ChanServ
[07:28:06] -!- girls has quit [Disconnected by services]
[07:28:25] -!- mode/#archlinux32 [-i #archlinux32] by ChanServ
[07:28:42] -!- girls has joined #archlinux32
[07:29:49] -!- tyzoid_ has quit [Disconnected by services]
[07:30:02] -!- tyzoid_ has joined #archlinux32
[07:31:01] -!- tyzoid_ has quit [Client Quit]
[07:32:46] -!- mode/#archlinux32 [+o tyzoid] by ChanServ
[07:32:47] <tyzoid> test
[07:33:36] -!- mode/#archlinux32 [-m #archlinux32] by ChanServ
[07:33:45] <tyzoid> girls / deep42thought try now
[07:33:50] <deep42thought> test
[07:33:53] <deep42thought> \o/
[07:34:00] <girls> it works!
[07:34:05] <tyzoid> somehow both +i and +m got set :/
[07:34:58] <tyzoid> Ok. So mostly painless transition :P
[07:35:26] <tyzoid> anyway, I'm heading off to bed in a min here
[07:36:33] -!- mode/#archlinux32 [-o tyzoid] by ChanServ
[07:36:54] <girls> good night!
[07:37:10] <tyzoid> would say same, but it's morning for you, so good morning
[07:38:24] <girls> damn confusing time zones
[07:38:29] <girls> :-D
[07:40:31] -!- deep42thought has quit [Quit: Leaving.]
[07:56:46] -!- davor has quit [Ping timeout: 264 seconds]
[08:02:35] -!- davor has joined #archlinux32
[09:04:00] <eschwartz> girls: \o/ we go phrik
[09:04:14] <eschwartz> tigrmesh: thx
[09:23:50] -!- belanthor has joined #archlinux32
[10:01:22] -!- abaumann has joined #archlinux32
[10:08:10] <abaumann> mmh. Gnome/GDM is quite broken..
[10:08:18] <abaumann> https://bbs.archlinux32.org
[10:08:19] <phrik> Title:libQt5Core.so.5: No such file or directory after system update / Pacman / Pacman Upgrades / Arch Linux 32 Forums (at bbs.archlinux32.org)
[10:08:55] <abaumann> I assume, nobody of us is using Gnome or KDE as a desktop environemnt. :-)
[10:11:57] <abaumann> I rescheduled gnome, gnome-desktop, colord and mutter.
[10:35:56] <City-busz> abaumann: GNOME works fine from testing. Propably you just forgot to move all GNOME packages to stable together
[10:38:40] <City-busz> abaumann: e.g. it shouldn't happen that gnome-shell 3.28.0 is in stable, while mutter 3.28.0 is still in testing
[10:41:21] <abaumann> mmh. I'll test again on testing, but I also had problems there.
[10:42:05] <abaumann> ok. I'll try to stabilize some packages..
[10:42:41] <abaumann> buildermaster: why don't you stabilize gnome-desktop
[10:42:49] <abaumann> buildmaster: why don't you stabilize gnome-desktop
[10:42:50] <buildmaster> Sry, "why-dont-you stabilize" is unavailable, until someone recodes it to look into the database.
[10:42:56] <abaumann> ah. I remember. :-)
[10:44:17] <City-busz> I can't see any problems on GNOME using the testing repo in the last weeks
[10:44:38] <City-busz> I think you don't need to rebuild anything, just move the correct packages
[10:45:34] <abaumann> the buildmaster ignores my commands anyway, it seems. :-)
[10:46:50] <City-busz> what is the purpose of the testing repo? why don't you move all packages to stable? :)
[10:47:50] <abaumann> this is a good point. I also have the feeling that testing is more stable than stable.
[10:47:59] <abaumann> In theory things should be tested first.
[10:48:17] <abaumann> so we could catch i686-specific errors before releasing them.
[10:50:10] <City-busz> testing is only reasonable if the packages moved into stable together based on their origial build status
[10:50:36] <City-busz> otherwise you'll end up with a broken stable repository if you move packages selectively
[10:51:00] <abaumann> true. I would be more radical. I would move everything.
[10:51:56] <abaumann> SSE2 problems would maybe arise, but they are only catchable if I finish my opcode checker script..
[10:54:41] <abaumann> let me check. maybe my testing was still set to SSE2-disabled..
[11:01:16] <abaumann> ok. gdm/Gnome on testing works with SSE2-enabled, so another SSE2-thingy has crept in. *sigh*
[11:01:30] <abaumann> and buildmaster ignores my stabilize commands via email, no clue why..
[11:30:27] -!- eduardoeae_ has quit [Ping timeout: 240 seconds]
[11:53:27] -!- abaumann has quit [Quit: leaving]
[13:14:57] -!- deep42thought has joined #archlinux32
[13:15:42] <deep42thought> abaumann: I noticed fetchmail issues on the buildmaster before
[13:16:47] <deep42thought> and regarding moving-in-a-bunch: The information for that is now available in the database, but I didn't have a chance to implement it yet in db-update
[13:17:03] <deep42thought> purpose of testing was mostly to catch any stuff we broke by modifications
[13:18:18] <deep42thought> nevertheless, I'm strongly in favour of "repairing" db-update first, before decomissioning testing, because the same logic is applied when moving from staging to testing as from testing to stable
[13:18:44] <deep42thought> only difference is, that packages moved from testing to stable must be marked "tested", whereas staging packages do not
[13:27:59] <deep42thought> and I think, I also found a grip on the fetchmail problem: it looks like seed-build-list is extremely slow which hickups fetchmail which then marks the mails as "read" (when I killed id), so it does not read them again ...
[13:53:14] -!- abaumann has joined #archlinux32
[13:53:19] <deep42thought> Hi Andreas
[13:53:22] <abaumann> hi.
[13:53:37] <deep42thought> it looks, like there is something wrong with my locking :-/
[13:53:38] <abaumann> this is the killer argument, staging to testing migration has to happen as it happens now.
[13:53:58] <abaumann> and the gnome issue is again a SSE2 issue somewhere, I'm afraid..
[13:54:16] <deep42thought> yes, but the linking stuff has nothing to do with sse2
[13:54:37] <abaumann> I was puzzles, why testing was not working.. that was why.
[13:54:53] <abaumann> why gnome-shell managed to get published passed all dependencies, that's what worries me more.
[13:54:55] <deep42thought> and "move packages in one bunch only which were stabilized upstream in one bunch" is not yet implemented
[13:54:59] <deep42thought> ah ok
[13:55:19] <abaumann> it's rather complicated to define 'a bunch' exactly. :-)
[13:55:31] <deep42thought> no, not at all
[13:55:40] <deep42thought> we have upstream's commit time
[13:55:59] <abaumann> ah. and can device an order..
[13:56:06] <deep42thought> and I'd just say "|a.commit_time-b.commit_time| < 20 s => in one bunch"
[13:56:24] <abaumann> still, we should move packages when we have an empty build queue and there is nothing happening upstream..
[13:56:48] <deep42thought> the build queue is never empty
[13:57:05] <deep42thought> there's always firefox-developer-version lying around
[13:57:09] <deep42thought> and others ...
[13:57:24] <abaumann> ah. true. but I would actually stop building it after a while..
[13:57:40] <abaumann> ..and then try once a day if things recovered automagically.
[13:58:01] <deep42thought> yeah, we might want to automate that ...
[13:58:09] <deep42thought> btw: https://packages.archlinux32.org
[13:58:10] <phrik> Title:Arch Linux 32 - gnome-shell 3.28.0-1.0 (i686) (at packages.archlinux32.org)
[13:58:15] <deep42thought> the database _knows_ it's broken
[13:58:26] <abaumann> mmh. true.
[13:58:31] <deep42thought> I guess, I moved it by force and broke it :-/
[13:58:49] -!- eduardoeae_ has joined #archlinux32
[13:58:49] <deep42thought> !wtf libmutter-2.so.0
[13:58:52] <phrik> deep42thought: extra/mutter
[13:59:01] <deep42thought> buildmaster: wtf libmutter-2.so.0
[13:59:26] <deep42thought> ... time passes ...
[13:59:43] <buildmaster> [testing] mutter (3.28.0-1.0): /usr/lib/libmutter-2.so.0
[13:59:47] <abaumann> mutter-3.26.2+43+g77dd1bf63 in stable, testing/mutter-3.28.0-1.0 in testing.
[14:00:22] <deep42thought> let me move it ...
[14:01:38] <deep42thought> On the other hand, someone should _really_ review that function: https://github.com
[14:01:39] <phrik> Title:builder/bootstrap-mysql at da7eb1fc8e1ba94c2ba645d8b9ec2823ca0114c7 · archlinux32/builder · GitHub (at github.com)
[14:02:06] <deep42thought> it's only 196 lines O.o
[14:02:32] <abaumann> omg
[14:04:15] <deep42thought> buildmaster: wtf libgweather-3.so.15
[14:04:21] <buildmaster> [testing] libgweather (3.28.1-1.0): /usr/lib/libgweather-3.so.15
[14:04:36] <deep42thought> buildmaster: wtf libgnome-desktop-3.so.17
[14:04:39] <buildmaster> [testing] gnome-desktop (1:3.28.1-1.0): /usr/lib/libgnome-desktop-3.so.17
[14:08:52] <deep42thought> ... and the function will become even longer, when I implement the "bunch" part.
[14:26:54] <tyzoid> o.O
[14:27:06] <deep42thought> :-/
[14:27:51] <deep42thought> tyzoid: nothing like a Sunday morning looking at someone else's bash script generating a stored function for mysql?
[14:28:03] <tyzoid> That, but
[14:28:03] <tyzoid> yeah, that was unusually slow for the api
[14:28:18] <deep42thought> it was only the first
[14:28:23] <deep42thought> after that it got faster
[14:28:34] <deep42thought> might be the buildmaster, too, though
[14:29:24] <tyzoid> I may need to create my own index here, pkgfile is slower than I'd like
[14:29:31] <tyzoid> even though it's faster than pacman
[14:29:41] <tyzoid> *mysql fts*
[14:29:44] <tyzoid> ftw*
[14:29:57] * deep42thought was wondering what the "s" stands for :-D
[14:30:08] <tyzoid> for the save?
[14:30:09] <deep42thought> "mysql fucks the system"
[14:30:15] <tyzoid> lol
[14:30:29] <tyzoid> f*** that sh*t?
[14:30:34] <tyzoid> idk
[14:30:36] <deep42thought> :-D
[14:30:51] <tyzoid> I always like making bacronyms
[14:30:57] <tyzoid> like when the ssl -> tls rename happened
[14:31:04] <tyzoid> I would have preferred ssl stay ssl
[14:31:10] <tyzoid> and change to 'Ssl Security Layer'
[14:31:16] <deep42thought> lol
[14:32:07] <abaumann> /usr/lib/gnome-settings-daemon/gsd-power: error while loading shared libraries: libgnome-desktop-3.so.12: cannot open shared object file: No such file or directory
[14:32:33] <abaumann> after upgrading gnome-deskop to 1:3.28.1-1.0
[14:32:55] <deep42thought> this is in testing?
[14:32:59] <abaumann> no, stable.
[14:33:04] <deep42thought> that's fine
[14:33:05] <deep42thought> :-D
[14:33:10] <abaumann> ah. :-)
[14:33:12] <deep42thought> I'm moving most of gnome now to stable
[14:33:17] <deep42thought> but I'm not done yet
[14:33:17] <abaumann> ah.
[14:33:18] <abaumann> ok.
[14:33:27] <deep42thought> ... the buildmaster is somewhat slow, currently :-(
[14:33:28] <abaumann> so, I catch an inbetween-state. :-)
[14:33:30] <tyzoid> https://static01.nyt.com
[14:33:40] <abaumann> *catched
[14:33:47] <deep42thought> LOL
[14:33:48] <tyzoid> *caught
[14:33:49] <abaumann> tyzoid: :-)
[14:34:16] <tyzoid> Yeah, pretty much describes stable to a tee
[14:34:31] <tyzoid> anyway, I'm heading off, I'll be back in ~45min
[14:34:32] <deep42thought> maybe we should link that image on 500's?
[14:34:38] <deep42thought> k, cu later
[14:44:15] -!- tigrmesh has quit [Ping timeout: 255 seconds]
[14:47:14] -!- tigrmesh has joined #archlinux32
[14:51:19] <deep42thought> buildmaster: wtf libclang.so.6 libjsonrpc-glib-1.0.so.1 libvala-0.40.so.0 libdevhelp-3.so.5 libgspell-1.so.2 libgspell-1.so.2
[14:51:23] <buildmaster> Huh, I don't know that one.
[14:51:31] <deep42thought> buildmaster: wtf libclang.so.6
[14:51:36] <buildmaster> [testing] clang (6.0.0-1.0): /usr/lib/libclang.so.6
[15:06:15] <deep42thought> abaumann: you blacklisted fox-devel?
[15:06:52] <deep42thought> I'm just looking through the deletion-list ...
[15:07:14] <abaumann> aeh..
[15:07:29] <deep42thought> doesn't matter if it was you - it's on the black-list :-)
[15:07:51] <deep42thought> ooooh
[15:07:57] <deep42thought> archlinux removed hashcash :'-(
[15:08:38] <abaumann> my log says, fox-devel had big messups with int, long ints and atomics.
[15:08:52] <deep42thought> yes, I remember it was failing for a long time
[15:09:25] <deep42thought> https://packages.archlinux32.org
[15:09:27] <phrik> Title:List of packages to be deleted (at packages.archlinux32.org)
[15:09:45] <deep42thought> maybe, someone can have a brief glance at that, before I move on and remove them :-)
[15:11:53] <abaumann> ah, the old X-protos, they can go away
[15:11:54] <tyzoid> deep42thought: mailing list?
[15:12:02] <abaumann> yeah.
[15:12:02] <deep42thought> too slow
[15:12:11] <deep42thought> I'll just check agains my mirror :-)
[15:12:24] <tyzoid> why are we removing blender?
[15:13:53] <abaumann> libvirt?
[15:14:05] <abaumann> ok, maybe not that interesting to run on 32-bit.
[15:14:07] <deep42thought> https://ptpb.pw these are the ones, which are still available on x86_64
[15:14:51] <abaumann> dmd and other D-stuff, broken, mainly.
[15:15:03] <abaumann> qemu?
[15:15:53] <abaumann> ah: we have a Gnome on stable. Just tested.. :-)
[15:16:38] <deep42thought> :-)
[15:18:22] <tyzoid> lolrip
[15:18:30] <tyzoid> dvdrip gets removed
[15:18:50] <deep42thought> it's not in upstream
[15:18:55] <deep42thought> ... according to my mirror
[15:19:05] <deep42thought> https://www.archlinux.org
[15:19:05] <phrik> Title:Arch Linux - Package Search (at www.archlinux.org)
[15:21:23] <deep42thought> https://ptpb.pw
[15:21:25] <deep42thought> better?
[15:22:26] <tyzoid> why mustache-d?
[15:22:37] <deep42thought> because of d?
[15:22:39] <deep42thought> dunno
[15:22:43] <City-busz> deep42thought: why would you like to remove libvirt?
[15:23:11] <deep42thought> I think, I do not
[15:23:22] <deep42thought> but somehow the buildmaster came to the conclusion it should ...
[15:24:04] <tyzoid> oh, I didn't see the message about D being broken.
[15:24:17] <tyzoid> That said, dlang provides binary builds of the D compiler for i386...
[15:24:17] <deep42thought> well, dmd is broken
[15:24:20] <deep42thought> ldc work(ed)
[15:24:53] <tyzoid> https://dlang.org
[15:24:55] <phrik> Title:Downloads - D Programming Language (at dlang.org)
[15:24:56] <tyzoid> They provide binary builds of dmd
[15:25:09] <tyzoid> To me, that makes it seem like something in our env is broken
[15:25:20] <deep42thought> well ...
[15:25:29] <deep42thought> maybe :-)
[15:27:28] <deep42thought> https://ptpb.pw
[15:29:22] <tyzoid> Still think we should hold off removing dmd/dlang stuff
[15:30:14] <deep42thought> but then we should provide the binary package
[15:33:05] <tyzoid> collectd has i386 builds on debian for latest as well
[15:33:44] <deep42thought> https://issues.dlang.org
[15:33:47] <phrik> Title:17833 – compiling dmd on x86 linux fails (at issues.dlang.org)
[15:33:49] <City-busz> please keep virt-install, virt-manager and virt-viewer as well
[15:34:25] <deep42thought> City-busz: yes, I removed them from the list already: https://ptpb.pw
[15:35:00] <tyzoid> deep42thought: That bug says that if you bootstrap with an older version of the compiler, it works
[15:36:00] <deep42thought> so you want to keep dmd-2.067.1 lying around just to compile newer dmds?
[15:36:45] <tyzoid> didn't we pitch the idea of having a build-support repo?
[15:36:52] <deep42thought> we have one
[15:37:00] <tyzoid> lol, then throw it in there :P
[15:37:04] <tyzoid> I don't really see a downside here
[15:37:04] <deep42thought> but I still think, dmd should fix their stuff ...
[15:37:16] <tyzoid> They should, but at least we can pass the buck
[15:37:54] <deep42thought> I think I tried it back in November and there were issues, too
[15:37:59] <deep42thought> but let me retry again ...
[15:38:02] <tyzoid> This is all just my opinion, though. you and abaumann are doing all the work in regard to this
[15:38:16] <tyzoid> so I don't want to throw more work at ya, if it's not really useful
[15:38:34] <deep42thought> well: If you get dmd to compile (on whatever obscure route), I'm happy to put it back into the repositories
[15:40:53] <buildmaster> bazel is broken (says rechenknecht).
[15:47:15] <tyzoid> Ok, will let you know if I rig something
[15:50:49] <deep42thought> I think, I remember what the problem was:
[15:50:56] <deep42thought> dmd is compiled in multiple stages
[15:51:08] <deep42thought> and changing "the compiler" only affects stage1
[15:51:22] <deep42thought> e.g. if you compile with ldc, stage 1 compiles, but stage 2 crashes
[15:51:37] <deep42thought> and if you use dmd (newest), stage 1 crashes
[15:51:50] <deep42thought> ok
[15:52:06] <deep42thought> with dmd-2.067.1 I get the exact same error: /home/vagrant/dmd/src/dmd/generated/linux/release/*/dmd -conf= -c -o- -Isrc -Iimport -Hfimport/core/sync/barrier.di src/core/sync/barrier.d
[15:52:06] <deep42thought> make: *** [posix.mak:185: import/core/sync/barrier.di] Segmentation fault (core dumped)
[15:52:57] <tyzoid> Ok. I made a comment on the issue saying it's triggering the removal from arch32
[15:53:10] <deep42thought> ok, thanks :-)
[15:53:12] <tyzoid> I'd say we should be fine to remove it as is
[15:53:23] <tyzoid> We've got it in the archive if anyone needs the older version
[15:53:37] <deep42thought> ok
[15:54:50] <tyzoid> deep42thought: What's the reason for removing collectd?
[15:55:37] <deep42thought> I can't find one
[15:56:11] <tyzoid> Ok. Thought it was odd since debian provides builds for latest on i386
[15:56:16] <deep42thought> and the buildmaster is also not helpful
[15:56:28] <deep42thought> buildmaster: why dont you keep collectd
[15:56:29] <buildmaster> Sry, "why-dont-you keep" is unavailable, until someone recodes it to look into the database.
[15:56:36] <tyzoid> ah, lol
[15:56:55] <tyzoid> I'm going to ask in #manjaro and see if any manjaro32 users are in there to give opinions
[15:57:57] <deep42thought> I think, it's simply a mistake by the buildmaster ...
[15:58:00] <deep42thought> :-/
[15:58:33] <deep42thought> https://ptpb.pw
[15:59:48] <tyzoid> opera-ffmpeg-codecs seems odd, but I haven't looked it up
[16:00:04] <deep42thought> opera is black-listed
[16:00:14] <deep42thought> and opera-ffmpeg-codecs, too
[16:00:42] <deep42thought> gens looks wrong, too
[16:02:41] <tyzoid> Yeah, figured it was an opera issue
[16:03:36] <tyzoid> gogglesmm seems odd to remove, does it have an x86_64-only dep?
[16:03:45] <deep42thought> yes
[16:04:06] <deep42thought> fox-devel
[16:04:22] <deep42thought> it's a makedependency
[16:04:24] <deep42thought> though
[16:05:01] <deep42thought> oh, and it links against it O.o
[16:05:25] <tyzoid> :/
[16:05:42] <deep42thought> also: I found another inconsistency: https://packages.archlinux32.org
[16:08:41] <deep42thought> ah
[16:09:03] <deep42thought> I guess, the dependencies, I bootstrapped from the sources may differ sometimes
[16:09:16] <deep42thought> for example if I used a different revision of the source or something ...
[16:09:21] <deep42thought> still strange :-/
[16:11:54] -!- oaken-source has joined #archlinux32
[16:31:22] <tyzoid> deep42thought: oaken-source suggested we look at the fedore build procedure for dmd, if we wanted to see about getting it working
[16:34:47] <deep42thought> where do we find that?
[16:34:57] <deep42thought> https://src.fedoraproject.org does not bring anything up under "dmd"
[16:34:59] <phrik> Title:Home - Pagure (at src.fedoraproject.org)
[16:38:36] <tyzoid> yeah, I checked, and I only see 2.066
[16:39:33] <tyzoid> though tumbleweed has latest compiling on i586
[16:39:47] <tyzoid> https://pkgs.org
[16:39:47] <phrik> Title:Dmd Download (RPM, TXZ, XZ) (at pkgs.org)
[16:43:49] <deep42thought> how can archlinux32 get on that list?
[16:46:46] -!- isacdaavid has joined #archlinux32
[16:47:32] <tyzoid> not sure if they consider distro deravitives
[16:47:40] <tyzoid> they don't have manjaro / antergos
[16:47:49] <tyzoid> or even bigger projects, like solus / elementary
[16:47:49] <deep42thought> hmmm
[16:48:19] <tyzoid> no linux mint
[16:50:22] <tyzoid> deep42thought: Any reason not to hold the old version of dmd?
[16:50:30] <tyzoid> instead of deleting it, just blacklist?
[16:50:54] <deep42thought> it's a philosophical / fundamental decision
[16:51:11] <deep42thought> do we want to have software in the repos which we cannot update?
[16:51:24] <deep42thought> I thought, we decided on "no".
[16:51:36] <tyzoid> We already do, with the inclusion of vendor binary blobs
[16:51:50] <oaken-source> hi :)
[16:52:02] <tyzoid> hi
[16:52:10] <tyzoid> oaken-source suggested the hold-old-version idea
[16:52:28] <tyzoid> (he's from #parabola)
[16:52:30] <deep42thought> Hi oaken-source
[16:52:42] <tyzoid> or she. I don't know :P Power of the internet
[16:52:50] <deep42thought> doesn't matter
[16:53:37] <oaken-source> I've hit that in the risc-v port of parabola a couple of times that a package would not build on that arch in the last version, but an older version would build fine
[16:53:51] <oaken-source> nuch as librsvg, which started to require rust for the build post 2.40
[16:53:52] <deep42thought> IMHO there's only one case in which it's clever not to delete un-buildable packages: If it can be assumed, that the problem will get fixed (soon).
[16:54:12] <tyzoid> I guess the question here is what upstream does.
[16:54:22] <tyzoid> Does arch hold older versions of software?
[16:54:24] <deep42thought> "upstream" as in "dmd"?
[16:54:34] <tyzoid> upstream as arch
[16:54:41] <deep42thought> I think, they don't
[16:54:41] <tyzoid> i.e. the place we steal
[16:54:43] <tyzoid> erm borrow
[16:54:44] <tyzoid> pkgbuilds
[16:55:18] <deep42thought> ok, they do
[16:55:22] <oaken-source> arch has the luxury of building only for the primary platform of almost any piece of software
[16:55:23] <deep42thought> ... https://www.archlinux.org
[16:55:24] <phrik> Title:Arch Linux - Package Search (at www.archlinux.org)
[16:55:42] <deep42thought> they have flagged packages from 2016-10-25 O.o
[16:56:10] <oaken-source> imo, it's fine to package an older version of a piece of software, if a newer one has issues. it's just more difficult to maintain
[16:56:29] <oaken-source> backporting security fixes and such.
[16:56:31] <deep42thought> oaken-source: yes, but if upperstream decides to not support some architecture anymore, I don't think, it's a good idea to keep the package in that case
[16:57:01] <deep42thought> yeah, you can also go the debian-way
[16:57:09] <deep42thought> and backport all necessary changes
[16:57:14] <deep42thought> but I'm afraid _we_ can't
[16:57:20] <deep42thought> because we're too few people :-/
[16:57:33] <oaken-source> it's always a shame if upstream doesn't care about other architectures :/
[16:57:46] * oaken-source looks angrily at librsvg again
[16:58:25] <deep42thought> using a specific build system should _never_ be the reason why some package cannot be built for some specific environment
[16:58:33] <oaken-source> sometimes, there are drop-in replacements, or someone starts maintaining a fork of the last supported build for that architecture
[16:59:14] <oaken-source> wait, there is a build system that isn't arch agnostic?
[16:59:18] <oaken-source> and breaks for i686
[16:59:33] <deep42thought> no, that's not what I meant
[16:59:42] <deep42thought> but, e.g. we struggle to get dmd running correctly
[17:00:08] <deep42thought> and that means, everything that requires to be compiled by dmd (and does not build with ldc) is unavailable on archlinux32
[17:00:39] <oaken-source> the srpm might help
[17:00:44] <oaken-source> despite the older version
[17:00:58] <oaken-source> or is it some change that was introduced in a current version?
[17:01:01] <tyzoid> oaken-source: We've got a newer version in our repos than fedora has
[17:01:01] <deep42thought> like so: https://xkcd.com
[17:01:03] <phrik> Title:xkcd: All Adobe Updates (at xkcd.com)
[17:01:15] <oaken-source> (I'm not really in the loop what the actual problem is)
[17:01:19] <deep42thought> tyzoid: I've seen a newer .rpm than fedora's
[17:01:32] <tyzoid> Yeah, on tumbleweed
[17:01:35] <deep42thought> https://altlinux.pkgs.org
[17:01:36] <phrik> Title:dmd-2.076.0-alt1.S1.i586.rpm ALT Linux Sisyphus Download (at altlinux.pkgs.org)
[17:01:42] <tyzoid> that too
[17:02:32] <deep42thought> oaken-source: problem is, that compilation of dmd segfaults
[17:04:12] <tyzoid> deep42thought: do we still build on x86_64 and target i686? Or have we moved to building on i686 systems?
[17:04:30] <deep42thought> we do both
[17:04:35] <deep42thought> e.g. I build on x86_64
[17:04:41] <deep42thought> abaumann builds on i686 IIRC
[17:04:59] <deep42thought> and for tests, such as the one with dmd I build in the vagrant thingy you sent me
[17:05:12] <tyzoid> ah, lol
[17:05:21] <tyzoid> did you update the vagrant box before doing so?
[17:05:28] <deep42thought> sure :-)
[17:05:29] <tyzoid> that box is like 5 months out of date :P
[17:05:30] <oaken-source> dmd is another one of these awful things that need a working X compiler to compile an X compiler?
[17:05:36] <deep42thought> "update-me"
[17:05:41] <tyzoid> thanks for the reminder to fix my stuff
[17:05:55] <deep42thought> oaken-source: yes, but you can also compile it with ldc
[17:06:04] <deep42thought> ... then stage 1 finishes and stage 2 crashes
[17:06:28] <oaken-source> is that related to this thread? https://bbs.archlinux32.org
[17:06:30] <phrik> Title:community/dmd / Creating/Maintaining Packages / Arch Linux 32 Forums (at bbs.archlinux32.org)
[17:07:00] <deep42thought> yes
[17:07:04] <deep42thought> that's exactly it
[17:10:01] <oaken-source> do you have the pkgbuild somewhere for me to take a look at?
[17:10:39] <deep42thought> it's the official one from archlinux
[17:10:51] <deep42thought> https://git.archlinux.org
[17:10:51] <phrik> Title:trunk - svntogit/community.git - Git clone of the 'community' repository (at git.archlinux.org)
[17:11:17] <deep42thought> I installed old dmd and libphobos from the archive
[17:13:01] <oaken-source> what I'd do in create a dmd-bootstrap thing based on this https://github.com with a provides=('dmd') line, and try to do a clean bootstrap. I assume something as broken with the older dmd maybe
[17:13:02] <phrik> Title:GitHub - wilzbach/d-bootstrap: An example of boostraping D in any project (at github.com)
[17:13:37] <deep42thought> one can build dmd with ldc
[17:13:47] <deep42thought> s/can/should be able to/
[17:13:55] <oaken-source> :
[17:13:56] <oaken-source> :)
[17:13:59] <deep42thought> I tried that, too
[17:14:09] <deep42thought> and the result is simply, that dmd crashes at a later point
[17:14:20] <deep42thought> s/dmd/the build of dmd/
[17:14:26] <oaken-source> i see
[17:14:34] <tyzoid> I think the approache would be to put that in build-support and always bootstrap from that old version
[17:14:39] <tyzoid> approach*
[17:14:48] <tyzoid> somehow I started typing in french :/
[17:15:20] <tyzoid> lol
[17:15:55] <deep42thought> I think, it's spelled approaché
[17:16:03] <tyzoid> right
[17:16:06] <tyzoid> well, US keyboard and all
[17:16:09] <deep42thought> lo
[17:16:10] <deep42thought> l
[17:17:33] <deep42thought> tyzoid: I would do that _if_it_worked_
[17:17:51] <deep42thought> trying to compile dmd fails for me always in a segfault
[17:17:59] <tyzoid> regardless of compiler version?
[17:18:02] <tyzoid> weird
[17:18:02] <deep42thought> yes
[17:18:07] <deep42thought> and compiler type
[17:21:19] <tyzoid> deep42thought: Have we tried asking in #d ?
[17:21:25] <deep42thought> I do not
[17:21:44] <deep42thought> I'm currently trying to put "curl -fsS https://dlang.org | bash -s dmd" in prepare()
[17:21:45] <deep42thought> :-/
[17:22:07] <tyzoid> I'd ask, but I'm not a fan of acting like a relay
[17:22:24] <deep42thought> personally I'm tired of this compiler
[17:22:39] <tyzoid> same, though there's no alternate D compiler, right?
[17:22:45] <deep42thought> ldc
[17:23:00] <tyzoid> oh, nice
[17:24:04] <tyzoid> But then that should mean that we wouldn't need to remove the d packages, right?
[17:24:07] <oaken-source> I'm looking forward to doing the same stuff for risc-v >_>
[17:24:08] <tyzoid> I mean, other than dmd
[17:26:57] <deep42thought> tyzoid: might be - if we get them compiled by ldc
[17:28:37] <tyzoid> <BartoCH> i've seen cases where ldc would refuse to compile something, and dmd would just go an work.
[17:28:51] <tyzoid> from one of the archstrike devs
[17:29:03] <deep42thought> great, more good news
[17:31:46] <tyzoid> deep42thought: on a scale from 1 to erlang, how bad is this?
[17:31:58] <tyzoid> s/erlang/haslekk/gc
[17:32:03] <tyzoid> haskell*
[17:32:08] <tyzoid> man, I can't type today
[17:32:32] <deep42thought> it's orthogonal to haskell
[17:32:49] <tyzoid> *woosh*
[17:33:26] <tyzoid> it was a joke on haskell always being a problem
[17:33:45] <deep42thought> yeah, but haskell is partly my fault, I'm afraid
[17:33:58] <deep42thought> ... or rather yours, because you didn't check my crappy stored function yet ;-)
[17:34:17] <tyzoid> lol
[17:36:35] <deep42thought> whereas dmd is only "my fault", because I insist on deleting it and I'm fed up with that one :-D
[18:07:30] -!- belanthor has quit [Quit: Leaving]
[18:29:24] * buildmaster failed to execute a mysql query - can you have a look at "tmp.mysql-functions.query.stdin.2018-04-15T18:18:30.ynR6Py"?.
[19:31:57] -!- davor has quit [Ping timeout: 240 seconds]
[19:38:27] -!- davor has joined #archlinux32
[20:05:18] <tyzoid> Almost ready to bring srv0 down for reimaging
[20:05:29] <tyzoid> it'll be interesting to see if anything breaks when I take it offline :P
[20:08:27] <deep42thought> :-D
[20:08:57] <tyzoid> It's funny, some web crawlers have a dns cache measured on the order of weeks
[20:09:55] <tyzoid> I'm still getting requests to srv0 from some crawlers on the arch32 mirror
[20:09:57] <deep42thought> well, maybe they need to save some traffic ;-P
[20:10:28] <tyzoid> they don't save much traffic compared to GET requesting a bunch of 404s
[20:10:38] <deep42thought> I know
[20:10:42] <deep42thought> it was meant to be funny
[20:10:49] <deep42thought> ... german humor
[20:10:51] <tyzoid> oh, lol
[20:10:58] <oaken-source> german humor lol
[20:11:03] <tyzoid> no, that's my kind of humor too
[20:11:14] <tyzoid> just taking things too literally
[20:11:30] <deep42thought> yeah, that's the problem on irc
[20:11:37] <deep42thought> you never hear the "tone"
[20:12:03] <tyzoid> I use text to speech, so everything you say is in a yelling aussie's voice /s
[20:12:22] <deep42thought> ESPECIALLY ALL CAPS?
[20:12:32] <tyzoid> ow my ears are bleeding
[20:12:43] * deep42thought is sorry
[20:12:51] <tyzoid> you'll have to speak up, I can't hear you
[20:12:53] <oaken-source> I hate to interrupt :)
[20:13:03] <tyzoid> go ahead
[20:13:04] <oaken-source> but do you have documentation on your buildmaster setup?
[20:13:12] <oaken-source> I'd like to repurpose that for parabola builds mayb
[20:13:14] <tyzoid> should be on the forum and github, scattered
[20:13:25] <tyzoid> https://github.com
[20:13:27] <phrik> Title:GitHub - archlinux32/builder: tools for building 32-bit archlinux packages from archlinux.org's official, 64-bit tested PKGBUILDs et al. (at github.com)
[20:13:47] <deep42thought> database layout is not on github
[20:13:50] <tyzoid> https://bbs.archlinux32.org
[20:13:51] <phrik> Title:How to set up a build slave / Building / Arch Linux 32 Forums (at bbs.archlinux32.org)
[20:14:02] <deep42thought> https://buildmaster.archlinux32.org
[20:14:06] <tyzoid> ^
[20:14:29] <tyzoid> is buildmaster.archlinux32 down?
[20:14:36] <deep42thought> welll
[20:14:43] <deep42thought> technically: no
[20:14:55] <deep42thought> but it's reaallly slow, currently
[20:15:00] <deep42thought> especially disk i/o
[20:15:15] <oaken-source> how smart is the builder?
[20:15:33] <deep42thought> dumber than I
[20:15:35] <tyzoid> It thinks it's smarter than it is
[20:15:37] <deep42thought> and that means a lot
[20:15:51] <oaken-source> this reminds me of a text adventure I played a long time ago
[20:15:51] <deep42thought> oaken-source: what do you mean?
[20:16:01] <oaken-source> where you had to ask questions to a pair of scripted npc guards
[20:16:09] <tyzoid> lol
[20:16:25] <deep42thought> so what do you mean by "smart"?
[20:16:53] <tyzoid> I think he means how autonomously does it run
[20:16:55] <deep42thought> it pulls stuff from archlinux' svn2git, puts it into a database, hands out and takes back build assignments and moves packages between repositories
[20:16:58] <tyzoid> vs manual interaction
[20:17:01] <oaken-source> I'm looking for something that can deal with build-time dependencies, understand the madness that is the upstream arch pkgbuild repository, understand when things need to be rebuilt, etc.
[20:17:13] <tyzoid> ah, ours can do that
[20:17:17] <tyzoid> just not quite correctly
[20:17:31] <deep42thought> mainly, it moves stuff that should not be moved
[20:17:31] <oaken-source> that's a pain I know too well :)
[20:17:44] <deep42thought> but the dependencies are all there in the database
[20:17:46] <oaken-source> you know that svn2git is going away, right?
[20:17:51] <deep42thought> O.o
[20:17:57] <deep42thought> what's the substitute?
[20:18:00] <oaken-source> asp
[20:18:11] <deep42thought> asp checks out svn2git
[20:18:15] <deep42thought> AFAIK
[20:18:22] <oaken-source> I think they moved to git
[20:18:45] <deep42thought> well, git is fine
[20:19:00] <oaken-source> their git is a bit peculiar
[20:19:04] <oaken-source> they have a branch for each package
[20:19:10] <oaken-source> several, in fact
[20:19:17] <oaken-source> ah, you'll figure it out.
[20:19:22] <oaken-source> I'm not so sure myself, tbh
[20:19:51] <deep42thought> yeah
[20:19:57] <deep42thought> we're only looking at the master branch
[20:20:02] <deep42thought> but the others are there as well
[20:20:15] * deep42thought is not sure if the buildmaster actually pulls other branches, too
[20:20:58] <oaken-source> I've started pulling the release branches for the packages
[20:21:04] <oaken-source> because master is broken too much
[20:21:12] * oaken-source looks at openssh
[20:21:41] <deep42thought> in the master branch, there is $pkgbase/repos/$repo-$arch/
[20:21:47] <deep42thought> that's, what we're using
[20:22:00] <deep42thought> I think, it's identical to the content of the release branch
[20:34:43] <tyzoid> deep42thought: Any reason we can't move the rest of those resources off the buildmaster and onto packages?
[20:36:39] <tyzoid> i.e. things like the database-layout image
[20:38:07] <deep42thought> ummm
[20:38:30] <deep42thought> some of the html stuff is generated from bash
[20:38:38] <deep42thought> I wanted to get rid of that, actually
[20:38:53] <deep42thought> so: ssh log and email log won't work on packages
[20:40:55] <tyzoid> ssh log / email log can be thrown in mysql, right?
[20:40:59] <deep42thought> and more importantly: sanity-check status won't work
[20:41:07] <deep42thought> could
[20:41:17] <tyzoid> sanity check is the only thing that I can think of that can't be moved
[20:41:25] <tyzoid> but most users don't check that
[20:41:32] <tyzoid> vs we sometimes refer them to other resources
[20:42:01] <deep42thought> build-logs also cannot be moved off
[20:42:12] <tyzoid> rsync
[20:42:13] <deep42thought> and all that comes with it
[20:42:19] <deep42thought> hmmm
[20:42:35] <deep42thought> yeah, ok
[20:42:37] <tyzoid> you've got root + keys, you could make an auth'd user to push buildlogs
[20:43:04] <deep42thought> question is: what's the benefit?
[20:43:30] <tyzoid> quicker access to logs
[20:43:37] <tyzoid> doesn't hit disk on bm quite so hard
[20:43:54] <tyzoid> I mean, we already take backups, so it's not really a question of backups
[20:45:22] <deep42thought> hmm, yeah
[20:46:10] <deep42thought> ok, I'll start moving stuff over
[20:46:21] <deep42thought> with the goal of finally shutting down http(s) on the buildmaster
[20:46:28] <deep42thought> should be feasible
[20:47:22] <deep42thought> another thing I noticed: we should probably create / split off some "lib" for the website - e.g. stuff which is required on multiple places
[20:47:37] <deep42thought> I'm thinking of the *.css and the navbar entries
[20:47:52] <deep42thought> it bothers me, that we have to change like 5 locations for each thing we change
[20:50:13] <deep42thought> we could put sanity stuff into the database, too
[20:53:25] <deep42thought> I mean: it makes little difference _where_ the information about the sanity is stored
[20:54:07] <abaumann> we could store the insanity instead, it should be smaller than the sanity.. ;-)
[20:54:19] <deep42thought> _should_
[20:54:22] -!- oaken-source has quit [Ping timeout: 264 seconds]
[20:54:42] <abaumann> btw: I also had a look at the builder, and at plugbuild..
[20:54:46] <deep42thought> well, will the vm be heavier if it stores a "1" or a "0"?
[20:54:54] <tyzoid> heavier with a 1
[20:54:58] <tyzoid> more electrons
[20:55:15] <deep42thought> abaumann: and? should we switch?
[20:55:24] <abaumann> no.
[20:55:28] <deep42thought> *phew*
[20:55:42] <abaumann> but as Archlinux ARM is using it and I spotted some depenency mistakes there too.
[20:55:53] <abaumann> I was curious how they do things. :-)
[20:56:01] <tyzoid> so we're not the only ones getting it wrong
[20:56:08] <deep42thought> yay!
[20:56:20] <abaumann> no. the problem is tricky..
[20:56:30] <tyzoid> abaumann: We just need to find a third platform
[20:56:34] <tyzoid> and have them do majority vote
[20:56:38] <tyzoid> if they all vote to move, we move
[20:56:43] <tyzoid> or 2/3 vote
[20:56:51] <deep42thought> move to what?
[20:56:59] <deep42thought> ah, move the package
[20:57:02] <deep42thought> hmmm
[20:57:03] <tyzoid> yup :P
[20:57:09] <deep42thought> /dev/random?
[20:57:17] <abaumann> a, the package.. :-)
[20:57:21] <tyzoid> LGTM, ship it :P
[20:57:58] <deep42thought> nah, usually I'm a pedantic about such stuff
[20:58:12] <deep42thought> but if you make a mistake as a pedantic, it's still a mistake
[20:58:48] <tyzoid> deep42thought: I think the last important thing that's on srv0 is the 2222 backup system
[20:59:00] <deep42thought> ok
[20:59:09] <deep42thought> should I switch over to srv1, then?
[20:59:21] <tyzoid> Well I can't until I get power back
[20:59:27] <tyzoid> power outage over in my local area
[20:59:29] <deep42thought> btw: are you backing up the packages vm?
[20:59:39] <deep42thought> oh
[20:59:42] <tyzoid> Will be, haven't set that up yet
[20:59:49] <tyzoid> was planning on getting that today
[20:59:50] <deep42thought> well, at least it's not dark there yet
[20:59:53] <tyzoid> but power outage :P
[21:00:05] <tyzoid> Lol, local library has power
[21:00:14] <deep42thought> by burning books?
[21:00:23] <abaumann> lol
[21:00:45] <tyzoid> no, they do that at the Internet Archive (https://www.theverge.com/2013/11/7/5076166/the-internet-archive-seeks-donations-after-fire-destroys-equipment)
[21:00:45] <phrik> Title:The Internet Archive seeks donations after fire destroys $600,000 of equipment - The Verge (at www.theverge.com)
[21:01:38] <abaumann> who would have thought digital books burn too. ;-)
[21:01:43] <tyzoid> lol
[21:01:57] <deep42thought> !grab abaumann
[21:01:57] <phrik> deep42thought: 🎉
[21:02:04] <tyzoid> I think that's called cryptolocker
[21:07:27] <abaumann> $arch = "armv$arch";
[21:07:34] <deep42thought> O.o
[21:07:46] <abaumann> sometimes I wished a little more generaliziation would be there in plugbuild :-)
[21:07:49] <tyzoid> btw, deep42thought: I'm attempting to fix the packages.archlinux32 nginx config
[21:08:05] <deep42thought> let me know, what I did wrong
[21:08:25] <tyzoid> https://packages.archlinux32.org returns 404
[21:08:26] <deep42thought> because I'm using almost the same config on the buildmaster
[21:08:39] <tyzoid> it's the url prefixing that's causing problems somewhere
[21:08:41] <deep42thought> yeah, there is no index (yet)
[21:08:46] <tyzoid> there is, actually
[21:08:46] <abaumann> https://packages.archlinux32.org Error 404: Package Not Found In Sync Database, maybe we should point to the blacklist?
[21:08:49] <tyzoid> index.html exists
[21:08:52] <tyzoid> but it's not being loadeed
[21:08:54] <tyzoid> loaded*
[21:08:57] esotericnonsens_ is now known as esotericnonsense
[21:09:05] <deep42thought> abaumann: it's a WIP
[21:09:05] <tyzoid> You missed a /buildmaster/ non-php section
[21:09:12] <tyzoid> but it seems like adding it didn't help :/
[21:09:13] <deep42thought> oh
[21:09:15] <deep42thought> right
[21:09:37] <tyzoid> So mr. apache guy is attempting to fix that
[21:09:40] <deep42thought> I was assuming all static content is on the buildmaster itself
[21:09:50] <tyzoid> plus, get it so that the external IPs show up in the logs
[21:09:50] <deep42thought> but that will be wrong
[21:09:56] <tyzoid> since it's ignoring X-Forwarded-For
[21:10:03] <deep42thought> ok, thx
[21:15:31] <tyzoid> ok, 1 for 2
[21:15:36] <tyzoid> IPs are now coming through
[21:15:41] <tyzoid> still no static :/
[21:17:34] <eschwartz> Hmm, svntogit is madness apparently
[21:17:56] <deep42thought> eschwartz: in what respect?
[21:18:11] <eschwartz> According to oaken-source...
[21:18:20] <eschwartz> Or our something something
[21:18:33] <eschwartz> I mean, I don't exactly love using svn myself
[21:18:46] -!- oaken-source has joined #archlinux32
[21:18:58] <eschwartz> But svntogit is not really worse than svn itself
[21:19:02] <eschwartz> Ohai
[21:19:16] <eschwartz> oaken-source: we were just talking about you
[21:19:44] <eschwartz> How would you like me to fix our subversion repo for you...
[21:19:54] <tyzoid> deep42thought: Fixed the /buildmaster/ static endpoint
[21:19:55] <tyzoid> https://packages.archlinux32.org
[21:19:56] <phrik> Title:Archlinux32 packages (at packages.archlinux32.org)
[21:20:07] <tyzoid> that said, all the links have 'buildmaster' instead of an actual address
[21:20:08] <eschwartz> We do intend, somehow, to migrate to one git per package
[21:20:13] <tyzoid> I'm thinking they should be /buildmaster
[21:20:25] <eschwartz> I am technically in charge of implementing that
[21:20:43] <tyzoid> looking forward to that eschwartz
[21:20:50] <deep42thought> eschwartz: what is "one git per package" - one repository???
[21:20:51] <oaken-source> eschwartz: pardon?
[21:20:55] <abaumann> eschwartz: how would you handle notifications on package changes?
[21:21:15] <oaken-source> did I miss something game changing again :)
[21:21:32] <eschwartz> We have a feed for updated packages, it is not handled in the repo versioning of our PKGBUILDS
[21:22:01] <oaken-source> what kind of feed?
[21:22:14] <abaumann> but neither plugbuild or we are using it in the builder.. we do checkouts of huge git repos, which can be problem.
[21:22:17] <eschwartz> Or just look at git commits, does it matter whether that is in svntogit branches or separate repos?
[21:22:34] <eschwartz> You can do single branch currently
[21:22:49] <deep42thought> we can, but we don't
[21:22:53] <deep42thought> we use master
[21:22:58] <eschwartz> Soon, hopefully, there will just be 7000 different repos :D
[21:23:16] <oaken-source> I think what's important is to have an ordered list of repository events - package added, package removed, package updated; and a way to get the pkgfiles for that event
[21:23:45] <eschwartz> Well, we upload source packages, but only if the license requires it
[21:23:49] <abaumann> yes, this would avoid that everybody downstream implements his own mechanisms to build a build queue and push packages out.
[21:23:50] <oaken-source> I don't care whether that's from a feed or a series of git commits
[21:24:00] <eschwartz> Moving to git natively means we can tag things better
[21:24:24] <deep42thought> yes, but 7000 repos???
[21:24:26] <deep42thought> this is insane
[21:24:27] <abaumann> an have git hooks knitting a RSS feed or so.
[21:24:32] <eschwartz> So tagging would be good, instead of erratically shoving things into repos/
[21:24:42] <abaumann> every repos has a hook and calls a notification thingy..
[21:24:45] <abaumann> *repo
[21:24:53] <oaken-source> repo granularity is always a fun thing to figure out :)
[21:25:02] <eschwartz> deep42thought: this is how many packages we have, why do you dislike the idea?
[21:25:21] <deep42thought> I dislike the idea of having a separate repo for each package
[21:25:33] <deep42thought> we're talking about git repositories here, right?
[21:25:37] <oaken-source> eschwartz: it's kinda a bad thing if a package is removed - how long will you keep the repo around for packages that don't exist?
[21:25:45] <eschwartz> https://www.archlinux.org
[21:25:45] <oaken-source> in a monolithical repo, they are preserved in the history
[21:25:45] <phrik> Title:Arch Linux - RSS Feeds (at www.archlinux.org)
[21:26:01] <deep42thought> ^ exactly
[21:26:08] <eschwartz> I guess there is no reason to delete them
[21:26:17] <oaken-source> pkgbuilds are a treasured resource to us :)
[21:26:18] <eschwartz> We may end up restoring them after all
[21:26:43] <tyzoid> one repo with every package as a submodule?
[21:26:49] <abaumann> you can do some git-fu and split the big repo together with the history of every package into 7000 piecies.. in theory.
[21:26:53] <oaken-source> eww no submodules please
[21:26:56] <deep42thought> I'm in favour of one branch per package
[21:26:59] <eschwartz> Plus they'll be muchire trivial to mirror since currently
[21:27:08] <eschwartz> * much more
[21:27:08] <tyzoid> oaken-source: git submodule update --init --recursive
[21:27:21] <eschwartz> Currently svntogit simply prunes missing branches
[21:27:28] <abaumann> submodules. if yoy don't merge.. otherwise things get messy.
[21:27:39] <abaumann> *you
[21:27:39] <oaken-source> eschwartz: what is the benefit of having a single repo per package, instead of a file system structure in a single repo?
[21:27:40] <deep42thought> eschwartz: yes, but the history of master still has the content
[21:27:49] <eschwartz> Do we really need submodules?
[21:27:52] <tyzoid> oaken-source: faster checkouts
[21:28:03] <eschwartz> Faster checkouts, tagging...
[21:28:07] <oaken-source> tyzoid: not if you need a lot of packages :)
[21:28:08] <abaumann> submodules are the lost concept in git, IMHO.
[21:28:13] <abaumann> they don'y play well with the rest.
[21:28:15] <eschwartz> Submodules suck
[21:28:30] <abaumann> I would be curious about Linus' opinion ;-)
[21:28:38] <oaken-source> eschwartz: i can't believe I'm suggesting this, but did you look at google's repo tool?
[21:28:40] <eschwartz> Subtrees are marginally better but git was never meant for partial checkouts
[21:29:17] <oaken-source> that's what they use for the android repositories
[21:29:24] <abaumann> one repo for all packages is important if you want to some tagging perhaps? if they share a common history (also called releasing)..
[21:29:42] <abaumann> ..but archlinux is quite different in that regard.
[21:29:48] <abaumann> so also not a requirement.
[21:30:03] <eschwartz> We currently have communityco for checking out a svn subdir, we could migrate to some tiny git wrapper with a clone url
[21:30:30] <eschwartz> Not sure how important it is to update dozens of package checkouts at once
[21:30:39] <oaken-source> I think in the end the structure of the repository isn't that important - as long as there are adequate ways for downstream to subscribe to updates / import pkgbuilds / get pkgbuilds for a specific release version.
[21:31:01] <eschwartz> And this *will* make that easier, because tagging
[21:31:24] <eschwartz> It is impossible to do this with svntogit I think
[21:31:28] <deep42thought> how would we get notified about new packages (not new releases)?
[21:31:31] <oaken-source> update feed + asp with tags sounds like a cool thing to me
[21:31:40] <oaken-source> regardless of what lives underneath asp
[21:32:15] <eschwartz> deep42thought: git pull, checkout latest tag?
[21:32:26] <deep42thought> git pull of _what_?
[21:32:33] <deep42thought> there were 7000 repos
[21:32:35] <eschwartz> Follow the tip of community branch?
[21:32:37] <deep42thought> and now there are 7001
[21:32:53] <abaumann> I personally would put a hook on something like gitolite whenever a git repo is created.
[21:32:57] <eschwartz> Well, I guess we could have a submodules repo
[21:33:02] <oaken-source> gentoo is doing a pretty neat thing with overlay repositories
[21:33:14] <oaken-source> have one repo containing a meta structure with a list of repo urls
[21:33:22] <oaken-source> subscribe to updates on that meta repo
[21:33:26] <oaken-source> bob's your uncle.
[21:33:49] <abaumann> that's the same gitolite (ok, it's dated now) does: it has a top-level config file with all git repos in it.
[21:33:49] <eschwartz> We could also submit new packages to an atom feed, thenn have you subscribe to that
[21:34:03] <eschwartz> It doesn't need to be in git?
[21:34:19] -!- eduardoeae_ has quit [Quit: eduardoeae_]
[21:34:20] <abaumann> that's anyway a good idea. so you can notify independendly of the technology used underneath.
[21:34:22] <oaken-source> I think it's awesome that you're asking us before making the change :)
[21:34:31] <eschwartz> I think foxboron was trying to do something with gitolite though
[21:35:07] <eschwartz> oaken-source: do you follow the arch-projects mailing list
[21:35:18] <oaken-source> eschwartz: I had trouble with that list.
[21:35:30] <eschwartz> If so you will see whenever I get around to figuring this out
[21:35:42] <eschwartz> It's a big standard ML...
[21:35:53] <oaken-source> I should be subscribed, but I don't think I managed to do so, and I can't send messages to it, they keep being filtered out.
[21:36:23] <eschwartz> You do need to prefix the subject with the name of the thing you're discussing
[21:36:39] <eschwartz> [dbscripts] can we discuss foo
[21:36:42] <oaken-source> oh
[21:36:58] <oaken-source> still, I'm not getting mail from this list for some reason
[21:37:09] <eschwartz> The ML subject even mentions this now :D
[21:38:54] <oaken-source> I've had trouble with my email provider in the past :/
[21:39:03] * buildmaster resumes sanity.
[21:39:08] <oaken-source> their mail server is quite restrictive in the mails it accepts
[21:39:22] * oaken-source takes a look at the maillog
[21:39:48] <deep42thought> abaumann: the links on packages.archlinux32.org should now point in the correct direction
[21:40:03] <tyzoid> oaken-source: then you should look at migrating off of Microsoft's mail-servers, in that case
[21:40:13] <eschwartz> Lol
[21:40:35] * eschwartz uses archlinux.org mailservers
[21:41:14] <tyzoid> oh, deep42thought: ever hear back from the people over at SPI?
[21:41:18] <abaumann> * abaumann has his own mailserver, so at least if something doesn't work, he must yell at himself
[21:41:31] <deep42thought> yes, they had a question which I answered
[21:41:47] <deep42thought> abaumann: I know that feeling
[21:41:53] <abaumann> :-)
[21:42:05] <tyzoid> I think all of us do, just with slightly different things :P
[21:42:15] <oaken-source> I subscribed to the list april 4th
[21:42:25] <oaken-source> received 0 mails
[21:42:29] <oaken-source> nothing in the maillog
[21:42:35] <deep42thought> it's a nice feeling if everything you created works - but it is really annoying if anything broke
[21:43:02] <tyzoid> oaken-source: only one mail was sent in that period,
[21:43:07] <tyzoid> on apr 8th
[21:43:21] <tyzoid> and it was an auto-generated git email
[21:43:22] <deep42thought> tyzoid: I wrote back to them on 2018-04-03
[21:43:38] <tyzoid> deep42thought: Ok, sounds good
[21:44:00] <tyzoid> I imagine they'll take a while to handle inqueries/responses, all things considered
[21:44:03] <oaken-source> okay, then this list gets less traffic than I expected it to :)
[21:44:11] <tyzoid> yup
[21:44:22] <eschwartz> ... with code cleanups, little of importance
[21:44:29] -!- davor has quit [Ping timeout: 260 seconds]
[21:44:42] <eschwartz> Blame lukeshu
[21:45:09] <eschwartz> For the fact there is anything this month
[21:45:14] <deep42thought> I always like to watch you two fight ;-)
[21:45:28] <oaken-source> the beauty of the internet. sunday night, people arguing about vcs structures.
[21:46:09] <oaken-source> someone, somewhere, is doing this as a job, at normal office times, instead of burning their free time on it ^_?
[21:46:13] <oaken-source> *^_^
[21:46:52] * oaken-source notes this down as his personal life goal
[21:47:25] <deep42thought> The trick is not just to do what you like, but to find someone who pays you for it
[21:47:25] <tyzoid> to make money by arguing about vcs architecture?
[21:47:30] <tyzoid> ^^
[21:47:55] <eschwartz> Lol sounds great
[21:48:08] * eschwartz would apply
[21:48:09] <tyzoid> deep thoughts with deep42thought
[21:49:29] -!- davor has joined #archlinux32
[21:53:42] -!- deep42thought has quit [Quit: Leaving.]
[21:59:49] <tyzoid> lol
[21:59:59] <tyzoid> was just about to ask him a question :P
[22:00:48] -!- eduardoeae has joined #archlinux32
[22:00:57] <abaumann> well, it's late here.. I'm also thinking about stopping for today.
[22:01:12] <tyzoid> yeah, not unexpected
[22:01:14] <tyzoid> just funny
[22:01:16] <abaumann> read too much Perl for one day :-)
[22:01:19] <tyzoid> lol
[22:01:25] <tyzoid> If you're heading off, good night
[22:01:38] <abaumann> yep, bye.
[22:01:55] -!- abaumann has quit [Quit: leaving]
[23:28:27] -!- oaken-source has quit [Quit: sleep]