IRC Logs for #crux-devel Saturday, 2012-06-16

Romsterhmm glibc seems broken not sure if pkg-config did this or not01:31
Romstermake[2]: *** No rule to make target `/usr/ports/work/glibc/pkg/lib/', needed by `install-lib-nosubdir'.  Stop.01:31
juegood morning01:54
jueRomster: probably not, but will run a build here01:55
RomsterUsed internally by Glibc as a gross hack to get broken programs (e.g., some Motif applications) running. See comments in glibc-2.15/locale/broken_cur_max.c for more information01:55
Romsterseems hacky01:56
Romsteroff LFS site01:56
Romsteri had an idea on pkg-config why doesn't the dev just rely on the glibc source being pressent at build time much like how gcc can use mpfr, mpc01:58
Romsterbuilding with -j1 to see if it works or shows me something i missed.01:59
Romstereverything else besides glib built fine (missing python expected to fail)01:59
Romstersame error02:00
jueworks for me02:01
Romsterthis is in a chroot i did put your pkg-config in built it updated that to system, then cd /usr/ports/core ; prt-get update *02:02
Romsteri remvoed all built packages before that so everything got rebuilt02:02
Romsterwhy would it fail for me i'm using default cflags that crux supplies.02:03
Romsterunless it is ccache this does not like02:05
jueRomster: sorry, no idea, will try it on my laptop too, but this will take some time02:07
Romstertesting with CCACHE_DISABLE=102:07
Romsterfound that but i can't read it...02:10
Romsterstill failed.02:11
Romsterah fuck i know what it is02:13
Romsterand i did ask for glibc Pkgfile to unset V02:13
jueRomster: wrt your pkg-config idea, TBH I don't understand what you mean ;)02:13
Romsterat some point andi totally forgot about it02:14
Romsterexport MAKEFLAGS='V=1 VERBOSE=1 -j 4'02:14
Romsterin my chroot02:14
Romsterglibc is the only project that uses V for something else than verbosity in make output...02:14
Romstercould you unset V in glibc Pkgfile ? or should I just worka round that some other way?02:15
juedo it another way02:15
juewhy do you set V?02:16
Romsteri hate running into that issue i should post to the glibc mailing list. nto the first time i've hit this and wasted time figuring it out again.02:16
jueIMO it's better to have less output on the screen, because errors and warning are much better to see02:18
Romsterhmm but then i can't see if it's missing a linker or include on the compile line.02:18
Romsteri used to pipe it into pretty make that made the output nice and neat and coloured.02:19
Romstermight set that up again.02:19
Romsterit also removed duplicate stuff on each line.02:19
juewell, if you run into a problem you can do something different02:19
Romsterit's just a chroot build for testing compiles.02:20
Romsteri don't set V on my system.02:20
Romstersorry for the noise.02:20
Romsterack it still failed...02:20
Romsterprt-get update -fr binutils gcc then i'll retest.02:21
Romsterglibc has a few patches in it..02:21
jueFWIW, glibc build just succeeded on my laptop02:45
Romsteri'm going to say eff hlibc it's fine but wont build in my chroot for some strange reason02:48
*** mike_k has joined #crux-devel02:49
juewould be nice to know the reason, though02:50
Romsteri'll make a new chroot and test again02:51
*** c0x` has joined #crux-devel03:55
*** c0x has quit IRC03:56
*** c0x has joined #crux-devel04:14
*** c0x` has quit IRC04:15
*** c0x` has joined #crux-devel05:44
*** c0x has quit IRC05:46
Romsterjue, i mised another variable i had set VERBOSE=1 and V=1 only breaks glibc nothing else...05:49
Romsterok testing moved glib and libffi to opt and using jues pkg-config06:12
*** joe9 has quit IRC07:16
*** joe9 has joined #crux-devel07:22
*** pedja has quit IRC07:53
Romsterjaeger, you can remove the combined ports when you like i've removed them now.08:01
*** deus_ex has joined #crux-devel08:13
Romsterjue, i wonder why libffi has it's includes in a funky location /usr/lib/libffi-3.0.11/include/ffitarget.h08:43
jaegerRomster: ok, thanks08:54
Romsteralso do you want to or me to update glib-32 add libffi-32 i've just bumped them here already08:55
Romsteronly in my chroot atm08:55
jaegerI will do it if you prefer but either's fine with me08:56
*** c0x` has quit IRC08:57
Romsterjue, i'm happy with pkg-config, i moved libffi and glib to opt in my chroot added python to glib dependencies everything builds ok.08:57
Romsterjaeger, i can add them there already built here. not sure what else needs bumping i haven't looked at multilib ports in a while.08:57
jaegerI usually look at the stuff I have installed about once a week, bumped a few this week already08:58
jaeger 08:59:18 up 512 days, 14:20,  7 users,  load average: 0.22, 1.29, 1.8208:59
jaegerGonna have to reboot that server sometime soon, heh08:59
jaegerwow, another firefox update09:00
jaegerI shouldn't be surprised by it, I guess09:00
Romsterwhens the last kernel update on that box :D09:03
Romsteri never did try that hot swapping kernel on live system09:04
*** c0x has joined #crux-devel09:09
jaegerIt's running 2.6.2609:14
jaegernot sure what patchlevel, it's a debian install09:14
jaegerwhy did you move glib-32 to opt?09:44
jaegerglib and libffi are both in core09:44
Romsterjue's going to put pkg-config he has in jue/pkg-config that has it's own builtin glib, then glib and libffi that glib depends on can be moved back to opt. ans since we don't have pkg-config-32 at all there is no reason for glib-32 to be in core. if this is all in error i'll move them to core.09:50
Romsterelse python would have to be moved to core as well as glib depends on python too.09:50
Romsterelse python would have to be moved to core as well, as glib depends on python too.09:51
Romstersorry missing comma for clarification.09:51
jaegerIf they're going to move to opt in x86, too, that's fine09:51
Romsteri'm pretty sure they will when jue pushes jue/pkg-config to core. a few of us have been testing for breakages before it's pushed and i have not found any09:52
Romsterjust trying to save double handling.09:53
jaegerMy aim is to mirror the official crux as much as possible so if this version ever becomes an official one it'll be less of a pain to keep things in sync09:54
Romsteryeah i was expecting that and also anticipating jues next move. i'm pretty certain that's what's going to happen, if not i can move them back to core in a few days time or when jue says otherwise.09:56
jaegerWorks for me, just wondered09:59
Romsteri'm hoping multilib tool chain and glibc-32 goes into x86_64 and that becomes official.10:09
jaegerI would like that but I don't really think that's how it'll go. We'll see10:12
Romsterif it doesn't ill keep using multilib anyways10:35
juejaeger: do you have read by chance the logs about the glib issue?11:05
juejaeger: to summarize: current core/glib depends at run- and buildtime on opt/python, what isn't acceptable. The reason why we moved glib from opt to core was pkg-config, because with the latest release the internal glib sources were removed. That's indeed nonsense because they introduced a circular dependency between glib and pkg-config, but well, shit happens. Now fortunately upstream has reconsidered his decision and readded glib sources to pkg-co11:22
juemy plan is to use that new pkg-config port and move glib back to opt and add python as a dependency. Any objections?11:25
jueRomster: that's indeed a bit strange location for the includes. However, if other stuff is using pkg-config to find them, that will not be a real problem.11:38
Romsterjue, none at all, everything works as expected.11:56
juegreat, thanks for testing :)11:57
Romsterplus the extra stuff for moving the lib to /lib than /usr/lib can be removed out of libffi.11:57
Romstertested on mutlilib as well same result.11:58
Romsterno problem someone has to be a guinea pig.12:02
acruxjue: agree, i think it's an elegant solution12:09
*** mike_k has quit IRC14:35
*** clb has joined #crux-devel15:13
*** hawksbill has quit IRC15:38
*** hawksbill has joined #crux-devel15:43
frinnstjue: wrt glib: sounds like the way to go18:00
*** mavrick61 has quit IRC21:42
*** mavrick61 has joined #crux-devel21:43

Generated by 2.11.0 by Marius Gedminas - find it at!