IRC Logs for #crux-devel Thursday, 2014-09-04

*** sepen has joined #crux-devel00:36
*** sepen has quit IRC00:49
*** dwts has quit IRC02:17
*** dwts has joined #crux-devel02:23
*** mavrick61 has quit IRC02:30
*** mavrick61 has joined #crux-devel02:31
*** dwts has quit IRC02:39
*** dwts has joined #crux-devel02:40
*** mike_k has joined #crux-devel07:16
*** sepen has joined #crux-devel07:20
*** mike_k has quit IRC08:03
*** mike_k has joined #crux-devel08:20
*** sepen has quit IRC09:35
*** sepen has joined #crux-devel10:11
juehmm, the glib-schema crap isn't solved at all, I came around the following within my i686 VM11:25
jue1. not schemas installed, thus running glib-schema-bla brings nothing11:26
jue2. if I start firefox as root I get the following -> (firefox:1757): GLib-GIO-ERROR **: No schemas found. Please make sure they are compiled!11:27
jue3. works as normal user, so WTF :)11:28
*** prologic_ has joined #crux-devel11:50
*** erdic_ has joined #crux-devel11:50
*** horrorSt1uck has joined #crux-devel11:50
*** erdic has quit IRC11:50
*** horrorStruck has quit IRC11:50
*** prologic has quit IRC11:50
juecorrection: 3. is wrong. With a new created user it's the same: cannot run firefox, it stops with the error above11:59
Romsterwhy all of a sudden does this schema cause issues? and what does it actually cache?12:26
Romsterits like a type of registry thing?12:27
Romsterfor settings.12:27
Romstergsettings list-schemas12:27
Romsterhonestly freedesktop/gnome stuff...12:28
*** sepen has quit IRC12:37
frinnstgeez12:42
frinnstfeel free to revert the patch then. Im at work for a quite a while longer so12:42
frinnstnwe reported it to me last night too, but i've not had a chance to look at it yet12:43
frinnstyeah projects that rely on gnome stuff ought to be worried12:43
frinnstjue: procmail issue: https://bugzilla.redhat.com/show_bug.cgi?id=112129912:56
frinnstCVE-2014-361812:56
*** sepen has joined #crux-devel13:03
juefrinnst: thanks13:28
frinnsthavent seen a patch yet though13:28
Romsterwish gnome could keep out of glib and gtk stuff from upstream13:29
jaegerthat ship sailed a while back, heh13:31
Romsterthey have dconf for hevens sake why do they also need gsettings13:32
juebtw, procmail is one of the few ports I don't use anymore, replaced it by maildrop some years ago13:33
frinnstah. I still use it. Dont mind maintaining it13:34
jaegerdconf is a gsettings backend13:34
jaegergconf'13:35
juefrinnst: great, feel free to adopt it, thanks :)13:35
jaegers successor, more or less13:35
nrxtxfrinnst: really glib relies on a null pointer passing around without being checked?13:35
nrxtxRomster: it now appears more because the patch i supplied to frinnst stops when passing the null pointer further after initialization13:36
nrxtxinitialization is done in a void (void) function13:36
nrxtxsimply changing a global boolean to true to done without checking if there were errors13:37
Romsterah stops it than just segfaulting?13:37
nrxtxRomster: yes13:37
juefrinnst: here's the debian bug report -> https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=760443 and a patch -> http://www.openwall.com/lists/oss-security/2014/09/03/813:37
nrxtxmaybe firefox will not segfault normally because they havent used schemas13:37
nrxtxwhich is as i now see also a way13:38
nrxtxwhen a program is only using glib without using schemas it will not access the null pointer13:38
nrxtxdue to my patch it will also be terminated13:39
nrxtxeven if it might not crash13:39
Romsterah that makes sense.13:41
nrxtxRomster: but how to solve it finally? :D13:47
Romsterdon't know.13:48
Romsteri doubt anything outside of gnome even uses gsettings.13:49
jaegerinstall gnome and systemd, it's the only way13:49
nrxtxi think the patch has to go down deeper to work fine, keeping the null pointer, patching it in the functions using it13:49
jaegerI wonder if we could get gnome integrated into systemd13:49
Romsterlol13:49
nrxtxjaeger: you just killed my eyes13:50
nrxtxjust for writing that down13:50
Romsteryeah why don't they add dconf to systemd13:50
jaeger:D13:50
Romsterbe the registry in a binary format.13:50
nrxtxfrinnst: if i get home i will move the null pointer check to schema initialization, so only the programs using schemas will be terminated before using the null pointer13:51
nrxtxall others can safely pass the null pointer around as long as they are not using it13:52
frinnst"if i get home"? that sounds scary :)13:52
nrxtx*when13:53
frinnsthehe13:54
*** sepen has quit IRC13:54
*** heroux_ has joined #crux-devel14:10
*** sepen has joined #crux-devel14:11
*** mike_k has quit IRC14:17
*** heroux has quit IRC14:17
*** dwts has quit IRC14:19
*** mike_k has joined #crux-devel14:21
jaegerincidentally this glib schemas chatter has been pretty funny for me. Since I use MATE as my DE I have a lot of schemas installed anyway so I haven't seen any problems14:22
jaegerweird that having a big DE installed has avoided a problem rather than creating one14:23
*** dwts has joined #crux-devel14:28
*** 17SAA4G1P has joined #crux-devel14:33
*** 17SAA4G1P has quit IRC14:33
*** sepen has quit IRC14:46
*** sepen has joined #crux-devel14:48
c0xcore/eudev depends on core/kernel-headers16:36
frinnstwat?16:37
frinnstyou mean glibc?16:38
c0xinclude/linux/*.h16:39
frinnstyeah, we bundle those with glibc16:39
c0xok, this is superset16:40
c0xkernel-headers subset, just for build eudev16:40
c0xand OK, core/eudev depends on core/glibc )16:41
frinnstyep, everything depends on glibc :)16:41
frinnstwe never list glibc in the dependency16:42
frinnstdid this cause a problem for you?16:42
c0xyep16:42
frinnsthow? during build?16:44
c0xhttp://pastebin.com/wxMtae0g16:45
c0xrebuild from scratch on 3.016:46
c0xeudev missing include/linux/btrfs.h16:47
c0xduring configure/build16:48
jaegerglibc in 3.0 might not have had that include16:49
jaegerI think I ran into something similar in a 2.8 -> 3.1 build16:49
jaegerdon't remember for sure, though16:49
jaegerhttp://crux.nu/gitweb/?p=ports/core.git;a=blob;f=glibc/.footprint;h=40174d96631307816b3aa1741f487946ee9a637b;hb=refs/heads/3.0 seems to back this up16:50
c0xok, but btrfs.h missing16:53
jaegermay have to revert the eudev version in that case16:54
c0xjust add to core/eudev/Pkgfile "# Depends on:  kmod glibc" or "# Depends on:  kmod kernel-headers"16:55
jaegerI think you're missing my point - btrfs.h doesn't exist in 3.0's glibc16:56
jaegerchanging the depends line will not fix that problem.16:57
c0x3.1/ports/core/eudev/Pkgfile16:57
frinnstthere is no "kernel-headers" package16:58
c0xok, glibc )16:58
jaegerAre you building 3.0 or 3.1?16:59
c0x3.1 on 3.0 from scratch17:01
jaegerAh. Then you'll need to upgrade glibc to build eudev17:01
c0xmy script scan deps and do build accordingly17:02
c0xok, never mind17:02
jaegerglibc is an implicit dep, we don't list it explicitly17:02
jaegerwith that said, how can you even build eudev if glibc isn't installed?17:02
jaegerhow can you build *anything* like that?17:02
c0xfor eudev is enough kernel-headers, the rest builds without problem17:05
c0xon 3.017:05
jaegerdid you make some custom kernel-headers port?17:05
jaegerI don't understand, obviously, since building anything without a c library won't work too well, heh17:05
jaegerIs your script installing packages into a chroot or something like that without glibc?17:06
c0xpass one: build 3.1 packages on 3.0, pass two: rebuild all in sane 3.1 chroot env17:07
c0xI done it all now17:07
c0xwell how still to tell these infidels, what they don't understand a paradoxicality of the Russian soul?17:11
c0xbut pass two - overkill, imho17:15
c0x'core' metaport ) http://pastebin.com/RRHfMhd017:20
jaegerc0x: for what it's worth the ISO bootstrap process builds all packages 3 times total18:15
*** sepen has quit IRC18:37
*** sepen has joined #crux-devel18:53
*** sepen has quit IRC18:58
nrxtxfrinnst: got it19:14
nrxtxif you reopen the ticket i'll attach a better patch19:15
frinnstack19:49
*** mike_k has quit IRC20:51
nrxtxfrinnst: that hopefully fixes it correctly for all users21:24
nrxtxplease test21:24
frinnstthanks! will do21:28
nrxtxi think noone of them is still using gsettings tool directly21:33
nrxtxat least if you look at the code21:44

Generated by irclog2html.py 2.11.0 by Marius Gedminas - find it at mg.pov.lt!