IRC Logs for #crux-devel Friday, 2012-01-06

Romsterhmm some logic fail there still though needs more work.00:33
juegood morning02:47
jueRomster: 296.09 is beta, running it here since some days, works flawless02:50
Romstermorning jue yeah i'm gonna bump mine and test02:52
Romsterjaeger, pretty much does the job.02:56
*** Romster has quit IRC03:15
*** Romster has joined #crux-devel03:25
*** Romster has joined #crux-devel03:25
Romsterfile /usr/lib/xorg/modules/extensions/libdri.so03:34
Romster/usr/lib/xorg/modules/extensions/ ELF 32-bit LSB shared object, Intel 80386, version 1 (SYSV), dynamically linked, stripped03:34
Romsterok that's strange.03:34
Romsterfile /usr/lib/xorg/modules/extensions/libdri2.so03:35
Romster/usr/lib/xorg/modules/extensions/ ELF 64-bit LSB shared object, x86-64, version 1 (SYSV), dynamically linked, stripped03:35
Romsterrest of the files are ok.03:35
Romstertaking the dogs for a walk bbs03:36
Romster i think gl-select has an issue or my system does.05:07
Romster yeah problem with gl-select on multilib05:10
Romsterok fixed it gl-select use xorg ; rm /usr/lib/xorg/modules/extensions/libGLcore_so ; prt-get update xorg-server ; gl-select use nvidia05:18
Romsterold stuff from my 32bit.05:18
Romsteri would of expected gl-select to not mv the same file back and forth on me05:19
*** Romster has quit IRC05:24
*** Romster has joined #crux-devel05:24
*** Romster has joined #crux-devel05:24
frinnstso, what do i need to do to set up a 64bit multilib? havent really followed the development that close :)07:22
frinnst64bit install + ports ?07:22
juefrinnst: a good start should be to install the tools-chain, ports and pkgutils from jaeger's iso07:38
frinnsturl? i always lose it :)07:38
jueand glibc-32 if must be able to build 32bit stuff07:38
Romsterdon't forget to rejmerge and and like me double check your GL stuff.07:46
frinnstits just in a vm07:50
Romsterah k07:56
juefrinnst, jaeger: currently I'm looking at the diffs between core-mutlilib and core-x86_64, I've uploaded the diffs for ports that looks mostly identical to me to
juefrinnst: could you try the multilib-unzip port on your x86_64, please?08:15
frinnstyeah, hang on08:15
frinnstbuilds ok atleast08:17
juehmm, ok08:17
jueI'm asking because x86_64-unzip didn't build for me on multilib08:18
frinnstok, i'll use jaegers unzip then..08:19
Romsterdo we even need unzip in core?08:21
juethe only diff to x86 is linux_noasm instead of linux, shouldn't we chance x8608:21
jueoops, too late ;)08:21
jueRomster: not really, but if we move zip/unzip to opt we need a maintainer for it ;)08:22
Romsterah i see.08:23
jueRomster: there's a ticket for it in FS08:24
Romsteryeah i'm aware08:24
Romsteri could maintain them they don't change much?08:27
juejaeger: looks like that you can savely remove binutils,libgmp, unzip and util-linux-ng from core-multilib08:28
jueRomster: no, mostly a static port08:28
frinnstlinux_noasm as a target for i686 too? should be pretty safe08:28
Romsterhmm what about the --host name?08:28
Romsteris that going to change?08:28
jueyou mean the diff in libgmp08:29
frinnstunknown seems pretty set in stone08:29
frinnstso why the fuck am i the one with -pc-linux-gnu? :)08:30
Romsteri don't know why jaeger got the unknown i bet that was off LFS08:31
Romsterfrinnst, looks better.08:31
frinnstyeah but all other gnu-tools use unknown08:31
Romsteri really wonder why though.08:32
Romsteranyways it's jsut a silly name as long as it works right.08:33
frinnstits because someone is stubborn iirc08:33
juefrinnst: done08:34
RomsterFYI -flto is painless once you get over the rebuilding of broken libs that revdep seems to miss. so another task to look into if revdep can be more reliable at detecting broken libs.08:36
juewhat's the problem with revdep08:36
Romsterseems to not detect all the broken libs not sure exactly perhaps symbol are in another location due to LTO, i'm not certain but it missed about half the libs that needed rebuilding after enabling that.08:37
jueyou should run it as root, because not all libs are world readable08:38
Romsteri did.08:38
Romsterrevdep said port was ok but configure of another port said can not find -lfoo08:40
jaegerthe noasm target would be safe for x86 unzip, yes, but it's likely a lot slower08:40
jaegerthe only reason I chose it for multilib is that it wouldn't build otherwise08:40
Romsterhmm a quick configure script to choose the right define for the makefile? i could knock one up i'm nearly at that stage of making a few of my own.08:41
Romstersince you don't like the logic in the Pkgfile itself.08:42
juejaeger: IMO we should care too much about unzip, I've already changed core/unzip to use linux_noasm ;)08:42
jaegerI see that now08:43
jaegerfine by me, not too concerned with speed, just wanted to make sure that was pointed out08:43
jueyeah, thanks08:44
Romsteractually not many things use zip these days?08:44
frinnstls: cannot access *.zip: No such file or directory08:45
Romsteri'm sure i got a few floating around but it's like 3% of all the terabytes of files08:46
jueit's more or less windows-only nowadays08:46
Romsterugly format anyways08:46
jaegerFirst archive format to which I was ever exposed08:49
jaegergood ol' pkzip08:49
jaeger$ grep -i \.zip /usr/ports/*/*/Pkgfile | grep source= | wc -l08:50
jaegerthat was ghetto and would miss multiline sources and stuff but I just wanted a quick glance :)08:51
juefrinnst: wrt nfs-utils, chown /var/lib/nfs to nobody seems to work fine for me, at least I cannot see any side effects08:51
frinnstyeah, me neither08:52
jueok, I'll commit it, thanks for the hint08:52
jaegerfrinnst, Romster: the -unknown- bit is simply the most common so I went with it. I doubt that changing it to -pc- harms anything but I tend not to mess with defaults unless required08:53
frinnstyeah, i already pushed it to core_x86-6408:53
frinnstso you can remove the port if you want08:53
jaegerI saw, just explaining the reasoning08:53
Romsteri see08:54
frinnsti *though* it was my port that used unknown at first, so it was probably something i overlooked when i was messing around a year or so ago :D08:54
Romsterit's the coffee frinnst08:55
jaegerThat bit of the triplet often gets changed in cross-compile toolchains, for what that's worth08:55
jaegerI think it's used as the vendor string spot08:55
Romsterdoesn't seem to be that many differences between frinnst's 64bit pure and jaeger's multilib08:55
jaegeranyway, I'll remove binutils, libgmp, unzip, and util-linux-ng08:55
Romsterother than the extra *-32 stuff08:55
Romsterno expected breakage?08:56
jaegerwell, I'm going to test it on my system before I push but I trust that jue has thought about it08:57
juejaeger: you've seen the diffs?08:57
jaegeryes, I read them a couple minutes ago08:57
jaegerno problems building any of them again and no linking issues09:05
*** mavrick61 has quit IRC21:31
*** mavrick61 has joined #crux-devel21:32

Generated by 2.11.0 by Marius Gedminas - find it at!