IRC Logs for #crux-devel Monday, 2011-12-26

Romsterjaeger, for some reason having libgcrypt installed on the system makes libxslt-32 not compile saying that gcrypt is not compatible. which of course it's 64bit vs 32bit, but why would it be looking at the 64bit than to skip that check?03:05
jaegerRomster: probably have to dig through config.log to find out. it might be something like a gcrypt-config executable or something09:39
*** mike_k has joined #crux-devel14:23
*** y3llow has quit IRC16:17
*** y3llow has joined #crux-devel16:18
*** y3llow has quit IRC16:20
*** y3llow has joined #crux-devel16:20
*** y3llow has quit IRC16:22
*** y3llow has joined #crux-devel16:23
*** y3llow has quit IRC16:24
*** y3llow has joined #crux-devel16:25
*** mike_k has quit IRC17:59
Romsterjaeger, it's trivial to fix finddeps two small changes /usr/lib32 added to LD_LIBRARY_PATH and lib32 added to line 1519:50
jaegerfigured it would be easy to fix19:54
Romsteri'll look into gcrypt sometime and see what's going on.19:55
Romsternoticed another possible issue preferhigher yes in prt-get.conf could be disastrous on multilib currently.19:55
jaegerMy guess is it's either a -config script it runs and gets native results or finding a native library in the wrong place19:56
jaegerI periodically check for incorrect libs in lib32 and usr/lib32 with file and grep19:56
jaegeroops, I forgot to update the md5sum for the new pkgmk.conf19:57
* ente tries to figure out autoconf19:57
Romsterhah have fun with autoconf it's a nightmare20:01
enteautoconf or the rest of the autotools?20:01
Romster/usr/bin/ld: skipping incompatible /usr/lib/gcc/x86_64-unknown-linux-gnu/4.5.3/../../../ when searching for -lgcrypt20:06
Romsterthe entire set of tools for autoconf20:06
Romsterwell autoreconf20:06
jaegerso it's using the wrong lib, yeah... have to find out where it's getting that info20:08
Romsteryeah looking for it now.20:08
Romsterthough if i got the 32bit lib of that installed it finds it ok.20:08
Romsteryou probably should not strip out the binaries on libxml2-3220:26
Romsteri found in libxslt-32 configure.in20:26
RomsterAC_PATH_PROG(XMLLINT, xmllint, /usr/bin/xmllint)20:26
RomsterAC_PATH_PROG(XSLTPROC, xsltproc, /usr/bin/xsltproc)20:26
Romsterthat should be point to the -32 binaries.20:26
Romsteror would that not matter that those are 64bit?20:27
RomsterAC_PATH_PROG(LIBGCRYPT_CONFIG, libgcrypt-config, no) <- there is my problem though i need to 's/libgcrypt-config/libgcrypt-config-32/' you would think that if i added --program-suffix=-32 that would transform them names for me.20:30
Romsterthat would be the most logical idea ever.20:30
jaegerCould be an oversight on the part of the authors20:33
jaegerOn the other hand, they have no way to know how every distro will do its multilib setup, I guess20:33
Romstersed -i -e 's/libgcrypt-config/libgcrypt-config-32/g' configure20:35
Romsterto libxslt-32 fixes this.20:35
Romsterthough i'm not sure if xmllint xsltproc is ok staying as 64bit it seems to not care.20:36
Romsterprobably only runs the commands to process stuff.20:36
Romstercan you add that fix to libxslt-32 please, i might attack the --program-suffix fix for it later on if i can mange to do that and submit a patch upstream20:37
Romsterhonestly though why are the not using a *.pc file instead.20:38
jaegerI'd still like ot know why it wants gcrypt at all20:38
jaegerI don't have that requirement on my systems, it's not even installed20:38
Romsterit can link to it for additional functionally it's optional20:39
Romsterlike i said if 64bit libgcrypt is installed it brakes libxslt-3220:40
jaegerthat's lame, but I suppose it's unavoidable20:40
Romsterlibgcrypt is used for a few programs see prt-get dependent --all libgcrypt20:40
Romsternotably cryptsetup chromium gnupg so it's going to be a common case.20:41
Romsteri agree it's very lame.20:41
Romsterthat's what trippe dme up on installing wine and made me package them other two ports, i'll keep those aside for later in case i need them. but currently i don't if you apply that sed fix.20:42
jaegerI had to do the same with freetype-config, though20:42
Romsteranything that has a foo-config binary needs to be sed to -32 on the end.20:43
Romsterheck i could of probably done sed -i -e 's/libgcrypt-config/&-32/g' configure20:43
jaegernot necessarily... only the ones that use that even if pkgconfig is available20:43
Romsteri'd be more inclined to rm them stupid files.20:43
Romsteri keep forgetting & in sed20:44
Romsterjaeger, found another bug you need to list libpng-32 on wine's Depends on: line it does not get installed in a prt-get depinst wine20:52
jaegerIt's not required20:53
Romsterfixme:ole:CoCreateInstance no instance created for interface {00000103-a8f2-4877-ba0a-fd2b6645fb94} of class {27949969-876a-41d7-9447-568f6a35a4dc}, hres is 0x8000400520:53
jaegerwine may be able to use it optionally but it isn't a hard requirement for building20:53
Romsterit isn't a hard requirment but try using stuff without png support.20:53
Romsterelse we need to list this as a optional port in readme or something20:53
jaegerstuff like what? I haven't run anything in wine yet that requires it20:54
jaegerI'm not against adding it if needed, just want to know what needs it20:54
jaegerI try to keep things minimal20:54
jaegerIs there something that doesn't render correctly?20:55
Romsterhmm not certain yet, just add a optional or nice to have line or something or a mention in README file for wine.20:55
Romsteri'll get back to you on that one.20:55
Romsteri'm wondering if gl-select has been modified for use with multilib?20:56
Romsterdoesn't look like it.20:56
jaegernot by me, no. hasn't seemed necessary so far20:56
jaegerlooking back at a wine build log I do see it check for libpng and throw a warning20:57
Romsterany program that uses png in wine will balk saying there is no png support, but that's not hard for the user to add then recompile wine.20:58
Romsteri just assumed all the major used deps would be pulled in for me20:58
jaegerhow common is that going to be?20:59
Romsterstarcraft1 is running really slow for me so i'm looking over the nvidia symlinks20:59
Romsternot certain.20:59
jaegerworld of warcraft and source engine stuff run full speed for me so far without any symlink changes20:59
jaegerfor what that's worth20:59
Romsterprobably the gecko html engine for explorer for one.21:00
Romsteri have to have soemthing broken on my system even the 64bit glxinfo is saying this.21:00
Romster$ glxinfo21:00
Romstername of display: :021:00
RomsterX Error of failed request:  BadWindow (invalid Window parameter)21:00
Romster  Major opcode of failed request:  137 (NV-GLX)21:00
Romster  Minor opcode of failed request:  4 ()21:00
Romster  Resource id in failed request:  0x680000321:00
Romster  Serial number of failed request:  3221:00
Romster  Current serial number in output stream:  3221:00
jaegeryeah, looks that way21:00
Romsterglxgears si not working either and it was before21:00
jaegerhave you already tried 'gl-select use xorg; gl-select use nvidia' ?21:01
jaegerto reset them21:01
Romsteri think i need to reboot i've been rebuilding tons of crap and i've been putting off a reboot.21:01
Romsterwhile x is running?21:01
Romsterwonder if that would cause x to segfault21:01
jaegerperhaps, haven't tried21:02
jaegerI'll add libpng-32 and libjpeg-32 to wine21:02
Romsternope no segfault but no change either i'll clean up what i got open and reboot.21:02
*** Romster has quit IRC21:05
*** Romster has joined #crux-devel21:09
Romsterhmm just noticed is dead21:09
Romsterglxgears now works lets see how sc1 goes now.21:10
Romstererr:wincodecs:PngEncoder_CreateInstance Trying to save PNG picture, but PNG support is not compiled in.21:11
Romsteryeah png is really needed21:11
Romstersc1 just crashed21:11
jaegerjust pushed the change21:11
Romsterhope i'm not bugging yo with all this stuff21:12
jaegernope, testing is good. I don't mean to sound combative, either, just like to keep things minimal like I said21:12
jaegerWas it sc1 that needed png support?21:13
Romsteri think so but i just remembered that i hadn't recompiled after i removed them two extra ports so i got a broken lib atm.21:14
Romsteroh what the hell now my pypanel crashed.21:16
Romsteri'm having a wonderful day :D21:18
jaegerheh :/21:18
Romsterrebuilding it all for good measure.21:19
Romstercoffee while i wait21:19
Romsteri di dhave some left over crud but i used pkg-not to clean up untracked files other than the font cache and some other cache files.21:25
Romsteri don't understand all this stuff jsut enough to be dangerous i guess.21:26
jaegerThe way everything interacts is a pain, complicated21:28
Romsteryeah and one little error and it's a pain to find.21:30
Romsteri'm still wondering when libarchive will get SMP for compression21:33
Romsterlike i was using pigz inplace of gunzip in my pkgmk before i reverted to crux.21:33
jaegerdecompresion too slow for you? :)21:34
jaegerwell, either21:34
Romstertry gunzip then try pigz in my repo big speed difference.21:34
jaegerWell, my point was more of a question of it actually bothering you how slowly or quickly compression/decompression goes21:35
Romsterconsidering i compile a ton of stuff yes it bothers me.21:36
jaegerfair enough21:36
jaegerconsidered submitting a patch to the libarchive dev(s)?21:36
Romsteryes but i haven't even begun work on that yet.21:37
Romsteri've got somewhat lazy since work drains most of my motivation, fixing windows all day every day kills most of my desire to improve my system.21:38
Romsterthough i got holidays i go more time to not think about windows.21:38
Romsterwould love a linux sysadmin job.21:39
jaegerI'm lucky to have one, I love it21:39
Romsteri bet you do. your lucky.21:39
Romster-- Packages where update failed21:41
Romsterhmm need to see why those failed.21:41
Romstergdbm-32 i think needs the --host= added.21:41
jaegerare you using the latest pkgmk.conf setup and ports?21:42
Romsteri just did a sysup jsut before but i haven't rejmerged any new pkgmk.conf21:43
jaegeradd -m32 to LDFLAGS in pkgmk.conf21:46
Romsteri got a unsupported locale error for pypanel21:47
Romsteri got in rc.conf LOCALE=en_AU.utf821:47
Romsterit's worked in the past.21:47
jaegerno idea on that, I never set LOCALE21:47
Romsterthough when iw as using hvl i had a ingenious way to only compile the locales i wanted...21:48
Romsteri'll look at that later.21:49
Romsterit's one issue after another with me.21:49
jaegeris that not as simple as only generating the locales you need?21:49
Romsterstartng to think i'm jinxed21:49
Romsteri haven't looked how crux is doing the locales.21:49
Romsteri'll look at that in a sec21:49
Romster/usr/bin/ld: i386 architecture of input file `gdbmopen.lo' is incompatible with i386:x86-64 output21:56
Romsteron gdbm-3221:56
Romsterhmp adding --host did not work and i adde dint ath -m32 to LDFLAGS21:57
Romsteradded in that*21:58
jaegerdo you have the libtool sed in the gdbm-32 Pkgfile?21:58
Romstersed -i -e 's,^CC="gcc",CC="gcc -m32",' libtool21:58
Romsteryeah i see it there21:58
jaegerline 27722:02
Romsterline 262 says nothing about 32bit though it points to lib3222:02
jaegerthat sed line won't work if "ccache gcc" is set in libtool22:02
jaegerboth lines reflect the same issue22:03
Romsterfigures :D22:03
Romsterccaache has been fine setting it this way too.22:03
jaegerit seems a problem specific to this particular autoconf/libtool setup, doesn't affect many ports so far22:04
Romstergdbm to me looks very old and not maintained.22:04
Romsterwhat sues it perl?22:04
Romsterprt-get dependent gdbm |xargs22:05
Romsterperl python ruby apr22:05
Romsterprt-get dependent gdbm-32 |xargs22:05
Romsteri don't understand why they keep using that database, sqlite3 is better?22:05
Romsteri can understand perl using it as perl has been around for ages but python.22:06
Romsteri see what's happening with that sed line.22:07
jaegerUnfortunately I don't think it'd be possible to account for every possibility there, but maybe a couple common ones22:07
Romsterthat's just ugly.22:07
Romstera better way would be to edit and autoreconf if it doesn't complain too much22:08
Romstersed -i -e 's,^CC=.*$,CC="gcc $CFLAGS",' libtool22:16
Romsterfixes gdbm-32 ok22:16
Romsterhacky but works for now.22:17
Romsterreadline-32 is ok talloc-32 is ok22:19
Romsterand python-32 is ok22:20
Romstercan you commit that fix to gdbm-32 :)22:20
jaeger-m32 is no longer in CFLAGS, it won't fix the problem :(22:20
jaegerer, wait22:20
jaegerI'm mixing it up, no longer in CC22:21
jaegerwill that not disable ccache temporarily?22:21
Romsterhmm then use sed -i -e 's,^CC=.*$,CC="gcc $CFLAGS $LDFLAGS",' libtool22:21
Romsteror judt LDFLAGS22:21
Romstermy options in 32bit in pkgmk.conf are too hacky.22:21
Romsteroh good call.22:22
Romsteruse $CC $CFLAGS22:22
jaegeryour pkgmk.conf is very complicated22:22
jaegerI don't think $CC is set at that point22:22
Romsteryeah i need to nuke all that crap i jsut been trying to geta round all these compile issues.22:22
jaegerIt would be for you if you use CC="ccache gcc" but I do not set CC in pkgmk.conf22:23
jaegerwhich change I made to work around your use of ccache, heh22:23
Romsterexport CC="ccache gcc" in my case.22:23
Romsteryou could just add CC=gcc and CXX=g++ in pkgmk.conf22:24
Romsteror in the Pkgfile to fall back to gcc if no CC is set.22:24
jaegerI don't like this, adding hack upon hack, honestly22:24
jaegerI'll play with it a bit22:24
Romsteri've usually jsut set a line in the Pkgfile to set CC=gcc if CC is not already set.22:25
Romstermind pastebinning your pkgmk.conf i'll clean mine up22:25
Romsterthat way we can be closer to the same issue, hopefully.22:27
jaegerpushed some program-suffix changes, though not related to gdbm stuff22:28
Romsteryou know instead of CFLAGS="${CFLAGS} -m32" you can do CFLAGS+=" -m32"22:28
Romsterthough it might not be portable other than bash22:29
jaegeryeah, wasn't sure about that. my preference was for the long version22:29
Romsterok got 32bit same as yours only adition there is the22:30
Romsterexport CC="ccache gcc"22:30
Romsterexport CXX="ccache g++"22:30
Romsterbefore the case ${PKGMK_ARCH} in22:30
Romsterwouldn't matter much even if that few lines missed ccache in gdbm-3222:31
jaegerI suppose not, it's just annoying22:31
Romstermore so i don't want it to fail compiling.22:31
Romsteron the few ports that are spastic on there configuration22:32
Romsterccache is a great help to what i do.22:32
Romsteri may even get distcc working again sometime in the future since i got 2 quad cores here now.22:32
Romsterbut distcc can be a nut case on errors.22:33
Romstererr:wgl:X11DRV_WineGL_InitOpenglInfo  couldn't initialize OpenGL, expect problems22:34
Romsterglxinfo is fine on 64bit22:35
jaegeryou have mesa3d-32 and nvidia-32 both installed?22:37
Romsterargh it would help if i installed nvidia-32 -_-22:37
Romstersorry one of them days22:37
jaegerThat isn't a hard dep since not everyone uses nvidia hardware, but I might add something to the README for it22:37
Romsteryeah i can understand that.22:37
Romster-- Packages installed22:38
Romsterlets see now.22:38
jaegermind trying that for gdbm-32? I removed the sed line and added a CC export22:38
Romsterno crash now or strange errors but no sound or screen22:39
jaegerit builds properly for me without ccache that way22:39
jaegertry recreating your .wine dir?22:39
jaegeror check its terminal output, see if there's anything obvious22:40
Romsteryeah that works wonderfully and is less hacky looking.22:43
Romstercommit that that's even better than my sed line change.22:43
jaegeralright, I'll settle on that... it's still something of a hack but I blame the software :P22:43
Romstersome how updates turned my volume down.22:44
Romsteryeah it's cleaner than the other sed line though22:45
Romsterwe can always come back to it later and fix it better.22:45
Romsteryes sweet sc1 is working smoothly now.22:46
Romsterand with pixel shader on.22:46
Romsterdespite all the annoyances linux has to offer it sure does work when it wants too.22:47
jaegerheh, yes22:47
Romsteri'd put up with the bugs in linux than windows anyday22:47
Romsterdarn the annoyance of heaving to insert disc into cdrom for red alert two though22:50
Romsteri thought i hacked that with a nocd fix.22:50
Romsteri was sure it was a game update and copy the mix file to the directory.22:50
Romsteroh i should look at this locale issue too.22:51
Romstermkdir -p $PKG/etc/ $PKG/usr/lib/locale and yet you did not copy any of the locales into it, no wonder22:55
Romsterhrmm then again it looks like crux does not either.22:56
Romstersee what i've doe for hvl ages ago
Romsteri don't quite understand how crux locales fork for glibc.22:58
jaegerAs I understand it something like gentoo's locale-gen simply does for you what the CRUX handbook suggests about generating locales23:08
jaegerI believe the way we do it in CRUX is as close to default as possible23:08
jaegerI don't know for sure, though, I'm no glibc expert23:11
Romsteri thoguth that LNAG= line in rc.conf is meant to run the locale generation automatically. but u remember in the past alt he locales were installed and that just made a default symlink to the correct one.23:16
jaegerno idea on that, either23:17
Romsterah there is a /usr/share/i18n/locales/en_AU but no UTF8 one.23:22
Romsteror perhaps UTF8 is the default in it now *shrugs* editing rc.conf23:22
*** Romster has quit IRC23:26
*** Romster has joined #crux-devel23:32
jaegerWhat did you edit in rc.conf? There's no LANG setting there...23:32
Romsterthere is too :D23:33
jaegernot in mine23:34
jaegernor referenced in the rc scripts23:34
Romsteroh wait a minute ther eis not that's a leftover for hvl23:34
Romsterarge i'm getting sick of the crap i got here that i had for hvl getting me all confused.23:35
jaegerMight be time to consider a fresh install23:35
Romsterbah now LANG is unset anyways23:35
Romsterso its' all POSIX now.23:36
Romsteri'm starting to miss stuff i had before.23:36
Romsterthough i also like multilib.23:37
Romsteri might put the locale generation in glibc to a feature request in flyspray.23:37
Romsteri'd still be in the dame boat if i reused all my configuration files.23:39
Romsteri guess crux does not support unicode.23:41
Romsterwithout some extra effort.23:42
Romsteri'm taking a break bbl23:43

Generated by 2.11.0 by Marius Gedminas - find it at!