IRC Logs for #crux-devel Friday, 2007-04-06

*** jaeger has quit IRC00:24
*** clb has joined #crux-devel00:39
*** Romster has quit IRC01:28
*** Romster has joined #crux-devel01:29
*** Romster has quit IRC05:50
*** Romster has joined #crux-devel05:51
*** treach has joined #crux-devel06:06
*** treach has quit IRC06:49
*** treach has joined #crux-devel06:54
*** prologic has joined #crux-devel08:11
*** Viper_ has joined #crux-devel08:23
*** jaeger has joined #crux-devel09:03
*** Viper_ has quit IRC09:08
*** aon has quit IRC09:08
*** Viper_ has joined #crux-devel09:08
*** aon has joined #crux-devel09:08
*** aon has quit IRC09:42
*** cptn has joined #crux-devel10:22
*** ChanServ sets mode: +o cptn10:22
cptnhi there10:22
jaegerheyo, cptn10:23
*** jflash has joined #crux-devel10:40
*** jflash has quit IRC11:13
*** jflash has joined #crux-devel11:13
*** jflash has left #crux-devel11:36
tilmancptn: did you ever do a debug build of qt4, by chance?12:02
cptnhi tilman12:14
cptnno, sorry12:14
cptnI'll have a look though, wanted to build a new qt4 package anyway12:20
cptnthere's a "-debug" flag to configure though12:22
cptn(building now, will take a while)12:24
cptnmore entries in ports db = better :-)12:30
*** cptn has quit IRC13:03
*** cptn has joined #crux-devel13:04
*** ChanServ sets mode: +o cptn13:04
cptntilman: it seems it worked (the debug build)13:16
cptnnot sure if it's a correct qt4 build13:16
cptnpathes etc13:16
tilmani tried -debug, sane CFLAGS and CXXFLAGS, and i used -ns -kw13:17
cptnI used -debug and -ns13:17
tilmancould you paste/upload your pkgfile?13:17
tilmani'm using sepen's (i think) port13:17
tilmantwo weeks ago i tried 3 builds i think13:18
tilmaneach taking 70-80 minutes13:18
cptnmmh, I remember sepen's port can't be installed in parallel to qt313:18
tilmanafter that i kinda gave up :D13:18
cptnyeah, it's a pain :-/13:18
tilmani don't have qt3 anyway13:18
cptnI could also upload the package if you want to13:18
tilmanthat would be nice :)13:18
tilmanqt4 has issues with a xcb-based libx11, i wanted to try to fix that13:19
tilmanso eventually, i'll have to build it myself13:19
cptnyeah, I actually saw that at work last week13:19
cptnmmmh, the package I built has wrong pkg-config files13:19
cptndoes that matter?13:19
tilmandon't think so13:20
tilmandid you verify you actually got debug symbols?13:20
tilman(gdb qtdemo or somesuch ;)13:20
cptnyeah, I tried the designer13:20
cptnmmh, it'll take like 50 minutes to upload13:22
tilmanbtw, openbox has correct support for argb windows these days13:23
tilmanif you want it in blackbox, i could dig out a diff13:23
tilmani probably mentioned that some months ago though o_O13:25
cptnI think you did, yes :-)13:25
cptnwhere would I see an improvement with argb support?13:26
tilmanfull screen redraws are usually faster with xcompmgr running13:28
tilmanbut if you don't want translucent windows, it's not that useful13:28
cptnwell, I've stopped using xcompmgr because it was too slow13:28
cptnso I can't really tell ;-)13:28
cptn~ 20 minutes for the upload13:31
cptngot faster, apparently13:32
tilmani'm currently building it13:33
tilmani think ccache is hitting the cache \o/13:33
cptnunfortunately, my X stuff isn't debug enabled13:37
cptnso the backtrace isn't really meaningful13:37
tilmancan you pastebin it?13:37
tilmani'm not sure that having debug builds of libX* really matters13:38
tilmanthat message is wrong13:41
tilmanit should read:13:41
tilmanBacktrace stopped just when things got interesting.13:41
tilman(since the bug is inside qt)13:41
cptnmmmh, is this a problem of my qt build?13:43
tilmannope, i got the same two weeks ago13:44
tilmani didn't try your build yet13:44
tilmanmine just finished13:44
cptnalso note that I haven't updated to 2.3 yet13:46
cptnso my xorg stuff is somewhat outdated13:46
tilmani get exactly the same behaviour here13:46
tilmanbut at least with your pkgfile, i'm getting binaries with debugging symbols :)13:47
tilmanasking zack rusin for help now13:48
cptnmy libxi is at 1.0.1 only13:48
cptnZack should know :-)13:48
tilmanmeh, i think i looked at that thread two weeks ago, but overlooked that libxi part apparently (or the whole mail)13:49
tilmandang, i thought i could blame qt413:49
cptnmmmh, your tree already has 1.0.2 though13:50
cptnnow I get a core dump13:50
cptnoh wait, still the sam13:51
cptnso chances are I'll have to rebuild everything13:51
cptnat least to get a consistent X13:52
cptnah, I could just install everything from the 2.3 ISO13:53
cptnbut then I'd essentially be where you are now13:53
tilmanyes :)13:53
tilmandon't bother13:53
tilmanmaybe zackr has some clues13:54
cptnhave you seen this:13:57
tilmanno, but i don't think it's related13:59
tilmani mean, we had that bug in mesa3d, too, of course13:59
tilmanand i fixed it, too ;)13:59
cptnah :-(13:59
tilmanall i need is a usable backtrace >:14:00
tilmani should try to find out what gdb's error message means, or what it's caused by14:00
tilmanbreaking in XQueryExtension gives me a nice bt14:05
tilmantsk, now gdb can't resolve a symbol in libXi, although i have a perfectly debug build of it :x14:11
*** cptn has quit IRC14:17
tilmanreproduced the problem in a dumb test app14:17
tilmanwtf, i'm getting the same error from gdb for the test program14:24
tilmanis libXi evil?14:24
*** cptn has joined #crux-devel14:25
*** ChanServ sets mode: +o cptn14:25
cptnI'm sorry, I'm having some hardware issues here :-/14:25
tilman(21:17) <    tilman> reproduced the problem in a dumb test app14:27
tilman(21:24) <    tilman> wtf, i'm getting the same error from gdb for the test  program14:27
cptnyes (, libxi is evil)14:27
tilman(21:24) <    tilman> is libXi evil?14:27
tilmanlogs ftw etc14:27
cptnsounds strange though that no one else saw this so far14:28
cptnis this related?14:29
tilmanno, that bug was fixed in 1.0.214:29
tilman"it SHOULD be fixed" maybe :)14:30
tilmanthat's all Xi bugs (4)14:30
treachtinyurl..? :P14:30
tilman8663: "The current head version of XGetExtensionVersion (XGetVers.c) does not release14:31
tilmanthe display lock when the XInput extension is not enabled on the server."14:31
tilmanso it's not our bug14:31
tilmanugh, what the hell?14:32
tilman8663 is open in bugzilla14:32
tilmanbut the fix has landed in 1.0.214:32
cptnwell, the patch in 8663 does a lot more than the git commit you posted before14:33
tilmangood, i just asked why it isn't closed yet14:33
tilman(good as in d'oh)14:34
tilmanwhat the hell?? XGetExtensionVersion() is in libXi14:36
tilmanit sounds like it should be in libx11 or maybe libxext14:37
cptnpatch doesn't apply against 1.0.2 of course14:38
tilman"I've committed a much simpler patch that I believe addresses this bug."14:38
tilman(jamey in comment #6)14:38
tilmani have an idea anyway14:38
cptnat least your comment is fine then :-)14:38
tilmani wonder what libXi/libXext does to the stack14:42
tilmanjust as i thought:14:44
tilmanXGetExtensionVersion() locks the display, then calls _XiGetExtensionVersion()14:44
tilmanwhich happens to call XInput_find_display14:45
tilmanwhich locks the display14:45
tilman might be our bug14:45
tilmanresp #5424, cause it has a patch attached to it that's looking good14:46
cptnsounds good, yes14:47
tilmandoesn't help though14:47
tilmanjesus, the x libs are riddled with locking errors14:50
tilmanit's just not funny14:50
cptngentoo has libxi 1.1.014:54
tilmangentoo is on crack14:54
cptnand 1.0.1 has a fix-bad-unlock.patch patch14:54
tilmanoh? :)14:54
tilmangot the url handy?14:54
tilmani see14:55
cptnit seems it's in 1.0.2 already14:56
tilmannot in my copy of the tarball14:56
cptnoh, still had 1.1.0 in the version14:57
tilmanokay, it seems it's in 1.1.014:58
cptnwith the patch (ignoring all the rejects) designer works!14:59
tilmanokay, i think i made daniels release 1.0.315:01
cptnhow long might that take? :-)15:05
tilmani don't dare to ask15:05
cptnokay, want a diff against 1.0.2 then of the gentoo stuff?15:06
tilmanyeah, if you have it15:07
tilmanawesome, thanks15:11
cptnmaybe we should attribute the gentoo folks :-)15:12
tilmandunno, it's in git, too15:13
tilmannot sure what they did15:13
tilmani'm hoping daniels will release 1.0.3 really soonish anyway15:13
tilmanin that case i won't patch the port15:13
cptnit's a git diff they use15:14
cptnso it's probably straight from fd.o git15:14
cptnmeanwhile the two of use can exclusively enjoy qt4 on crux15:14
tilmantoo bad the guy in #crux who reported it left15:14
tilmanhe probably installed arch by now ;)15:15
cptnanyway, beer for me :-)15:22
cptngood night15:22
*** cptn has quit IRC15:23
tilmanalso, that qt4 designer looks like motif15:31
*** prologic has quit IRC16:12
*** prologic has joined #crux-devel16:13
*** aon has joined #crux-devel16:50
*** prologic has quit IRC17:13
*** prologic_ has joined #crux-devel17:13
*** prologic_ has quit IRC17:39
*** prologic__ has joined #crux-devel18:28
*** prologic_ has joined #crux-devel18:56
*** prologic__ has quit IRC18:57
*** ___mavrick61 has quit IRC19:03
*** ___mavrick61 has joined #crux-devel19:04
*** prologic_ has quit IRC19:19
*** Viper_ has quit IRC20:06
*** prologic_ has joined #crux-devel20:08
*** prologic_ is now known as pl20:48
*** pl is now known as prologic20:49
*** treach has quit IRC21:38
*** jaeger has quit IRC23:57

Generated by 2.11.0 by Marius Gedminas - find it at!