**** BEGIN LOGGING AT Mon Dec 13 02:59:58 2010 Dec 13 03:03:44 kergoth, around? Dec 13 03:03:53 bb commit 53740977521bc81ffa37adfa7bbeb8f2a80ea165 breaks pcmciautils Dec 13 03:05:06 http://tinderbox.openembedded.net/packages/1246193/ Dec 13 04:09:26 grg: are you running a bitbake with 8b0753c090b0065a982198708c840f7c78dc0d23 applied? that fixes a major problem with it which was breaking gcc-cross-initial's patching. Dec 13 04:10:01 kergoth, i'm running master, i bisected back to taht commit. posted a fix to the bitbake-dev list too... Dec 13 04:10:10 k Dec 13 04:10:54 what was the issue? Dec 13 04:11:14 kergoth, wrong working directory when execing do_compile Dec 13 04:11:36 i figured that, i meant in the code :) Dec 13 04:12:01 pcmciautils, or bitbake? Dec 13 04:12:02 eh, that patch is completely unnecessary Dec 13 04:12:05 i dont follow Dec 13 04:12:16 cwd is passed to subprocess Dec 13 04:12:20 ok, well it fixes the problem for me Dec 13 04:12:22 it changes the current directory before running the script Dec 13 04:12:57 that bit you added back was removed intentionally. Dec 13 04:12:59 i'll look into it Dec 13 04:13:19 thanks Dec 13 04:13:24 thanks for the report. Dec 13 04:13:36 interesting that it would only break that recipe Dec 13 04:13:55 yes, i thought that odd also Dec 13 04:14:08 very reproducibly Dec 13 04:14:21 s/ly/e/ Dec 13 04:21:34 hey, I'm having some difficulty with building psplash. It appears that bitbake doesn't copy the subversion sources from tmp/downloads into the psplash dir in tmp/work Dec 13 04:22:08 and I'm not sure why it isn't... Dec 13 04:22:48 (there's a svn://... URI in SRC_URI, which is checks out into tmp/downloads OK, but it does't do anything with it later :( Dec 13 04:51:17 kergoth, i didn't notice it before but there appears to be a whole bunch of shit from various built packages in the directory that i invoked bitbake from Dec 13 04:51:38 e.g. mtdutils binaries Dec 13 04:52:42 and no matter what i do, i can't get pcmciautils to leave a log.do_compile Dec 13 05:26:54 kergoth, both mtdutils and pcmciautils use $PWD in their makefiles Dec 13 05:27:37 changing the cwd with subprocess.Popen uses os.chdir to change directory, which doesnt affect the PWD env var. cd foo in the shell does change it Dec 13 07:35:26 03Khem Raj  07master * r2c85705525 10openembedded.git/recipes/gcc/ (gcc-4.5.inc gcc-svn.inc): (log message trimmed) Dec 13 07:35:26 gcc-4.5.inc/gcc-svn.inc: Dont do the sourcedir move dance. Dec 13 07:35:26 In some cases the rename_srcdir ended up moving the sourcedir Dec 13 07:35:26 as a sub-directory under gcc-4.5 which was not intended actually Dec 13 07:35:26 it is intended to be renamed and not moved. Which was seen recently Dec 13 07:35:38 03Khem Raj  07master * r641a2a7e3d 10openembedded.git/recipes/binutils/ (3 files in 2 dirs): Dec 13 07:35:38 binutils_2.21.bb: Fix the build when using libtool 2.4 Dec 13 07:35:38 Signed-off-by: Khem Raj Dec 13 07:37:19 03Maksym Parkachov  07master * r4a8caae899 10openembedded.git/recipes/mcpp/ (mcpp-2.7.2/slice2ccp-no-crash.patch mcpp.inc mcpp_2.7.2.bb): Dec 13 07:37:19 mcpp: fixed crash issue with slice2cpp Dec 13 07:37:19 * added patch to fix crash issue with slice2cpp Dec 13 07:37:19 * fixed package name by removing lib_package Dec 13 07:37:19 Signed-off-by: Maksym Parkachov Dec 13 07:37:31 03Maksym Parkachov  07master * r2704fd5da3 10openembedded.git/recipes/zeroc/ (7 files in 3 dirs): (log message trimmed) Dec 13 07:37:31 icee: New program ZeroC IceE Dec 13 07:37:31 * added version of 1.3.0 of libicee Dec 13 07:37:31 * added slice2cppe with both native and target version Dec 13 07:37:31 * fixed multipe issues with makefiles Dec 13 07:50:17 good morning Dec 13 07:51:11 mckoan: gm Dec 13 07:54:22 gm Dec 13 07:56:39 Hey JaMa Dec 13 07:57:07 FYI I have a local patch I am testing where now cross localedef can be used for OE eglibc Dec 13 07:57:15 so no more qemu worries Dec 13 07:57:41 once it comes good I will send it out to ml and would need help in testing it out Dec 13 08:03:44 * JaMa will test it for sure Dec 13 08:04:37 khem: btw I have noticed segfault in qemu-0.13 also while building emacs-x11 (with 0.12 it was ok) Dec 13 08:05:39 gm Dec 13 08:06:17 khem, you saw my mail that mediatomb/js is still not working ? Dec 13 08:10:29 JaMa: yes I think we should pin to 12.5 for now and DP = -1 for .13 Dec 13 08:10:37 JaMa: care to post a patch ? Dec 13 08:10:46 eFfeM_work: yes I saw Dec 13 08:10:54 but I do not understand your error Dec 13 08:11:06 for me js build went fine Dec 13 08:11:17 after the one patch I had Dec 13 08:11:24 khem: I can post it.. but I haven't seen that error while building locales only in emacs-x11 Dec 13 08:11:39 khem, strange, I had the error both at home and on the autobuilder at home early dec it was still ok Dec 13 08:11:45 JaMa: there are more than 1 problem Dec 13 08:11:48 with .13 Dec 13 08:12:11 eFfeM_work: hmm whats the error Dec 13 08:12:16 khem, I have the whole setup still untouched, I can still reproduce and diagnose it Dec 13 08:12:24 khem, one sec Dec 13 08:13:58 khem: /home/hudson/jobs/FM_TEST/workspace/tmp/sysroots/x86_64-linux/usr/armv5te/lib/gcc/arm-oe-linux-gnueabi/4.5.2/../../../../arm-oe-linux-gnueabi/bin/ld: conftest: hidden symbol `__gnu_thumb1_case_uqi' in /home/hudson/jobs/FM_TEST/workspace/tmp/sysroots/x86_64-linux/usr/armv5te/lib/gcc/arm-oe-linux-gnueabi/4.5.2/libgcc.a(_thumb1_case_uqi.o) is referenced by DSO /home/hudson/jobs/FM_TEST/workspace/tmp/sysroots/x86_64-linux/usr/armv5te/lib/g Dec 13 08:14:06 ah, will make that three lines Dec 13 08:14:25 /home/hudson/jobs/FM_TEST/workspace/tmp/sysroots/x86_64-linux/usr/armv5te/lib/gcc/arm-oe-linux-gnueabi/4.5.2/../../../../arm-oe-linux-gnueabi/bin/ld: conftest: hidden symbol `__gnu_thumb1_case_uqi' in Dec 13 08:14:26 /home/hudson/jobs/FM_TEST/workspace/tmp/sysroots/x86_64-linux/usr/armv5te/lib/gcc/arm-oe-linux-gnueabi/4.5.2/libgcc.a(_thumb1_case_uqi.o) is referenced by DSO Dec 13 08:14:37 /home/hudson/jobs/FM_TEST/workspace/tmp/sysroots/x86_64-linux/usr/armv5te/lib/gcc/arm-oe-linux-gnueabi/4.5.2/../../../../arm-oe-linux-gnueabi/bin/ld: final link failed: Nonrepresentable section on output Dec 13 08:14:37 eFfeM_work: ok that error I see. but I thought you had issue with js too Dec 13 08:14:51 khem, this is still the js issue Dec 13 08:15:06 after your patch mediatomb still bails out for me Dec 13 08:15:07 eFfeM_work: not mediatomb ? Dec 13 08:15:22 khem, this is still from mediatomb configure Dec 13 08:15:33 o.0 Dec 13 08:15:34 git head from satudrday Dec 13 08:15:42 khem: sent to ml Dec 13 08:15:51 rebuild from scratch removed both TMPDIR and pstage dir Dec 13 08:16:04 let me ask once again Dec 13 08:16:17 is it js error or mediatomb error ? Dec 13 08:16:24 there are two different recipes Dec 13 08:17:07 khem it is an error in mediatomb do_configure where it tries to find js; the test in mediatomb configure fails because of the linking error above Dec 13 08:17:31 ok got it. so its mediatomb recipe Dec 13 08:17:55 well when I tried to build mediatomb it bombed for me too but elsewhere Dec 13 08:18:01 so I could not reproduce your problem Dec 13 08:19:21 khem, I isolated the code from do_configure, did a bitbake -c devshell mediatomb then ran the same command as configure runs on the same file and it still gives the error, the source file and cmd are in my last post Dec 13 08:20:07 eFfeM_work: hmm thats nice let me see Dec 13 08:20:19 khem: http://article.gmane.org/gmane.comp.handhelds.openembedded/40488/match=mediatomb Dec 13 08:23:13 morning Dec 13 08:24:05 hi hrw Dec 13 08:30:59 eFfeM_work: I have started to build js which will take some hrs here I will try to see this tomorrow Dec 13 08:31:06 good night all Dec 13 08:31:20 khem, thanks a lot and good nite Dec 13 08:32:02 khem will be online tonigt at around 18.30 GMT so we can compare results and e.g. peek into files with objdump Dec 13 08:32:28 hrw, what other tools do you need for the wiki? Dec 13 08:37:36 ka6sox: bb in few Dec 13 08:38:13 hrw, kk Dec 13 08:50:02 ka6sox: if you know something to mass remove users... Dec 13 08:50:19 I think we need to add some captcha or sth like that to register form Dec 13 08:50:45 hrw, okay you are the third person to ask for captcha...I'll get it on there tommorrow. Dec 13 08:50:59 re: mass removal...only the proceedure I showed you. Dec 13 08:51:11 I'll refind it and post the link again. Dec 13 08:51:16 ok Dec 13 08:51:30 basically combine folks from a list and then delete it as 1 user. Dec 13 08:55:51 k Dec 13 09:01:30 khem: 641a2a7e3dda3694fb2304f1a2ac4f7041ab3378 seems not enough for target binutils-2.21 (fails because of full tmpdir path: ERROR: QA Issue with staging: libopcodes.la failed sanity test (workdir) in path /OE/tmpdir-shr/sysroots/armv7a-oe-linux-gnueabi/usr/lib) Dec 13 09:17:45 morning Dec 13 09:19:09 morning all Dec 13 09:22:25 heh instead of captcha we should just print a git hash and user has to enter the commiter :-) Dec 13 09:27:10 hah Dec 13 09:27:13 :) Dec 13 09:28:16 morning bluelightning moving that stuff now Dec 13 09:35:05 ka6sox: morning, thanks for that Dec 13 10:14:38 XorA: too easy Dec 13 10:14:43 even for bot :) Dec 13 10:15:19 anarsoul: doubt bot writers will customize for one site though :-) Dec 13 10:17:12 reCaptcha is a PITA but it works. Dec 13 10:17:21 so thats what I'm putting on. Dec 13 10:43:26 JaMa: I'm ack'ing qemu 'downgrade' too Dec 13 10:43:37 not sure about weight of my ack :) Dec 13 10:47:08 hrw, recaptcha installed Dec 13 10:47:46 Jay7: please do it as reply, so I will get it included for pw-am :) Dec 13 11:01:56 JaMa: Acked-by: Yuri Bushmelev Dec 13 11:34:50 Jay7: hmm patchwork is not reading replies on IRC, but OK I'll add it manually Dec 13 11:35:09 JaMa: ah.. you mean reply in email Dec 13 11:36:50 03Martin Jansa  07master * rab163b72ae 10openembedded.git/recipes/qemu/qemu_0.13.0.bb: Dec 13 11:36:50 qemu-0.13.0: add D_P -1 for now, some people have segfault with it where 0.12.5 worked ok Dec 13 11:36:50 * I've seen it on one host while building emacs-x11 Dec 13 11:36:50 * locale generation: http://lists.linuxtogo.org/pipermail/openembedded-devel/2010-December/027723.html Dec 13 11:36:50 Acked-by: Eric Bénard Dec 13 11:41:45 it seems like patchwork breaks utf8 encoded patches: http://patches.openembedded.org/patch/35/ Dec 13 11:42:07 i can't apply this one, but it looks OK both in git and in my inbox Dec 13 11:42:42 because "Jan UrbaÅ„ski" contains an utf-8 character Dec 13 11:44:35 khem: maybe this patch would help: http://www.mail-archive.com/patchwork@lists.ozlabs.org/msg00096.html Dec 13 11:46:33 obi: http://patches.openembedded.org/patch/52/ I had also utf8 encoded and it worked ok (pw-am.sh) Dec 13 11:47:01 check source email Dec 13 11:47:10 may be it was broken there Dec 13 11:51:07 it is base64 encoded utf-8 and thunderbird displays it correctly Dec 13 11:51:33 strange... Dec 13 11:52:52 obi: compare to JaMa's one Dec 13 11:53:26 =?utf-8?q?=5BPATCH=5D_libdrm=3A_remove_old_not_pinned_versio?= =?utf-8?q?ns=2C_update_libdrm=2Einc?= Dec 13 11:53:29 hehehe Dec 13 11:53:35 that is from header Dec 13 11:53:55 anyway, patchwork have troubles with parsing mails Dec 13 11:56:23 obi: your link from PW ML should help I think Dec 13 11:56:34 ka6sox: can you try to apply it? Dec 13 12:04:08 03Andreas Oberritter  07org.openembedded.dev * r56f68d1554 10openembedded.git/recipes/gstreamer/ (7 files in 2 dirs): Dec 13 12:04:08 gst-plugins-base: removed unpinned versions Dec 13 12:04:08 Signed-off-by: Andreas Oberritter Dec 13 12:04:08 Acked-by: Martin Jansa Dec 13 12:04:10 03Andreas Oberritter  07org.openembedded.dev * r292df46569 10openembedded.git/recipes/gstreamer/ (6 files in 2 dirs): Dec 13 12:04:10 gst-plugins-ugly: remove unpinned versions Dec 13 12:04:10 Signed-off-by: Andreas Oberritter Dec 13 12:04:10 Acked-by: Martin Jansa Dec 13 12:06:50 CIA-101 doesn't like utf-8 either ;) Dec 13 12:07:47 err, wait... the patch got lost Dec 13 12:08:12 partially Dec 13 12:19:15 jo Dec 13 12:19:24 hello woglinde Dec 13 12:19:31 03Andreas Oberritter  07org.openembedded.dev * red61395003 10openembedded.git/recipes/gstreamer/ (6 files): Dec 13 12:19:31 gst-plugins-good: remove unpinned versions Dec 13 12:19:31 * fixup previuos commit, which broke due to utf-8 problems Dec 13 12:19:31 Signed-off-by: Andreas Oberritter Dec 13 12:19:31 Acked-by: Martin Jansa Dec 13 12:19:43 I took another step Dec 13 12:19:46 hi woglinde Dec 13 12:19:53 I held my thesis speech today Dec 13 12:19:58 wasnt perfect Dec 13 12:19:59 but okay Dec 13 12:20:12 :-) Dec 13 12:20:13 \°/ Dec 13 12:20:21 now only 3 oral exams left Dec 13 12:20:53 hm...isn't thesis the latest there? Dec 13 12:21:01 no Dec 13 12:21:12 you can make this indivudal here Dec 13 12:21:21 they changed the conditions some times agai Dec 13 12:21:24 ago Dec 13 12:26:17 now the mind is free for more oe development Dec 13 12:26:58 jej, so openjdk will build ok with libtool-2.4 soon? :) Dec 13 12:27:41 jama hm oh did I wrote my email only to jaap Dec 13 12:27:51 jama delete libz.la from sysroot Dec 13 12:27:57 I've read about that dirty work around too Dec 13 12:27:58 and it its building Dec 13 12:28:19 .la are uselses mostly now Dec 13 12:29:20 woglinde: I'll need your help about PKG_CONFIG issues for python-pygtk and co. Dec 13 12:29:44 ant just ask Dec 13 12:29:49 I've hacked around the issue but... Dec 13 12:29:54 Anydone tried compiling OpenJade (arm) or OpenJave (native) ? Dec 13 12:30:12 first: we have pygtk but upstream is called pygtk2 Dec 13 12:30:21 if I'm not too wrong Dec 13 12:30:30 but this is a secondary issue Dec 13 12:33:07 woglinde: http://tinderbox.openembedded.net/public/logs/task/12740495.txt Dec 13 12:34:07 I can hack it with a sed like -e s:'`$PKG_CONFIG --variable=codegendir pygobject-2.0`':\"${STAGING_DATADIR}/pygobject/2.0/codegen\":g \ Dec 13 12:34:17 (like koen did in pygtk) Dec 13 12:34:38 but why does PKG_CONFIG goes mad with both recipes? Dec 13 12:35:22 fwiw, upstream is now pygtksourceview_2.10.1 but the music doesn't change Dec 13 12:35:52 ah, yes, second question... should we keep around all thre versions (2.6, 2.10, 2.16) ? Dec 13 12:36:35 third thing is the codegen bits are now staged by pygobject and not by pygtk...(though pygtk depends on pygobject) Dec 13 12:37:06 hm Dec 13 12:37:11 what package is it? Dec 13 12:37:19 exactly Dec 13 12:37:21 python-pygtksourceview Dec 13 12:37:36 (vom sugar.bb ) Dec 13 12:37:53 * ant_work was only trying to build sugar... Dec 13 12:39:35 about pygtk vs. pygtk2 we have one *single* recipe which DEPENDS on "python-pygtk2" Dec 13 12:39:44 root/recipes/gnome/hippo-canvas_0.3.0.bb (plain) Dec 13 12:39:57 I suppose this could be python-pygtk Dec 13 12:40:14 being pygtk2 is provided by the same recipes... Dec 13 12:43:25 woglinde: he... reading configure: WARNING: unrecognized options: --with-python-includes I tought there is a missing acinclude patch... Dec 13 12:45:03 ant yes saw this too Dec 13 12:45:33 but didnt fint which other recipes provides pygtj Dec 13 12:45:36 ups pygtk Dec 13 12:49:41 woglinde, finally, hwn I build pygtksourceview_2.10.1 one file is unpackaged (gtksourceview.defs). I suppose this .defs file belongs to ${PN}, isn't? Dec 13 12:50:40 hm Dec 13 12:50:52 will try to build it here Dec 13 12:57:38 oh, thx Dec 13 12:57:56 hm seems libglade dont likes my libz.la removale Dec 13 13:22:44 03Koen Kooi  07org.openembedded.dev * rfa3bdf28f4 10openembedded.git/recipes/gstreamer/gst-plugins.inc: gst-plugins: make inherit gettext actually work Dec 13 13:27:34 hi rp Dec 13 13:32:22 anyone any idea on the gettext problem I mailed to the list? Dec 13 13:33:03 effem um which one? Dec 13 13:33:23 woglinde: let me find a link Dec 13 13:33:53 woglinde: http://article.gmane.org/gmane.comp.handhelds.openembedded/40487 Dec 13 13:34:26 failing do_install on arm (but not on ppc, although the ppc output also seems suspicious Dec 13 13:34:34 hi Dec 13 13:34:40 hi ericben Dec 13 13:35:04 hi eFfeM_work Dec 13 13:35:42 bitbake head + openembedded head + angstrom 2008.1 + armv5 target => perl-5.8.8-r36 fails with : ./miniperl : command not found Dec 13 13:36:20 is this a known problem ? Dec 13 13:36:30 hm, that is created somewhere in the build process (as part of perl or perl native) but I must say I use 5.10.1 already quite a while Dec 13 13:36:49 effem config.log is needed Dec 13 13:37:11 effem might be the same problem as with libz on openjdk build Dec 13 13:38:24 Could someone confirm this one: http://bugs.openembedded.org/show_bug.cgi?id=5494 Dec 13 13:40:33 woglinde will upload the config.log in a few mins, need to fetch it from the builder system Dec 13 13:40:57 woglinde: but as a starter here is the tinderbox log: http://tinderbox.openembedded.net/builds/110467/ Dec 13 13:45:22 hm, anyone knowing an alternative for filebin? it does not seem to respond Dec 13 13:47:19 why is autoconf on the target constantly breaking now? Dec 13 13:47:30 crofton hm? Dec 13 13:47:33 yeah Dec 13 13:47:39 ??? Dec 13 13:47:39 I thought I fixed it last week Dec 13 13:47:41 grrrrr Dec 13 13:48:10 interesting : cleaning perl-native and perl and relaunching the build "fixe" the problem Dec 13 13:48:39 with bitbake 1.10 Dec 13 13:50:43 Oh, RP__ is cloning himself :-) Dec 13 13:54:20 woglinde: http://www.easy-share.com/1913272622/config.log Dec 13 13:54:32 and if you want to: http://www.easy-share.com/1913272623/config.status Dec 13 13:54:38 and http://www.easy-share.com/1913272624/configure Dec 13 13:55:42 ok the problem with perl 5.8.8 seems to be in bitbake head : bitbake head fails, bitbake 1.10 works Dec 13 13:57:34 ericben: I think I did not see the problem with head and 5.10.1, then again I got all kind of other problems, might try 1.10 to see if that helps Dec 13 14:00:53 eFfeM_work: I'm trying to bisect this in bitbake which should help to undestand the problem's root Dec 13 14:01:08 ericben: cool! Dec 13 14:01:14 while doing nanox development ... this is not my lucky day :-D Dec 13 14:06:11 ericben had that yesterday, caught a cold, git.infradead.org was down so the fetch of mtd-utils for the testing branch test failed, gettext failed for arm mediatomb/js failed for arm, .... Figured it was not my day so got a few drinks and went to bed early :-) Dec 13 14:07:12 ok 9caf65e79f95fe0045e727391e974c4c1e7411ff (7/12/2010) works Dec 13 14:08:09 eFfeM_work: starting the week with drinks and sleeping seems not the right solution for me today :-D Dec 13 14:08:14 mrmoku, he seem away: fidencio[AWAY] Dec 13 14:08:20 hi ericben Dec 13 14:08:26 hi GNUtoo|laptop Dec 13 14:08:37 GNUtoo|laptop: yeah Dec 13 14:10:01 ericben, will you accept some intern this summer? Dec 13 14:11:10 GNUtoo|laptop: you are welcome here ;-) 1 hour by train from arcachon so you can mix holidays & internship ! Dec 13 14:12:14 if so how much time in advance will I have to reserve the internship? Dec 13 14:12:17 thanks a lot btw Dec 13 14:12:20 GNUtoo|laptop: we just have to find a good subject but I sure we can find one by the begining of 2011 Dec 13 14:13:44 ok thanks a lot!!!! Dec 13 14:14:37 GNUtoo|laptop: we stay in touch. I'll send you by email some ideas of projects, how long should your internship last ? Dec 13 14:16:02 I don't know, I should take some infos Dec 13 14:16:20 ericben, have you seen the CELF project proposal page? Dec 13 14:16:21 ok thanks a lot for the projects ideas Dec 13 14:16:37 Crofton|work: I don't really follow CELF Dec 13 14:17:03 http://elinux.org/CELF_Open_Project_Proposal_2011 Dec 13 14:17:23 possibly you could use them to fund internships :) Dec 13 14:19:40 Crofton|work: why not thanks I'll see if we can find something to fit into this projects. with 200k USD it's sure they can fund many internships :-D Dec 13 14:19:49 :) Dec 13 14:20:13 * GNUtoo|laptop doesn't know a lot about intersnip but knows about openembedded Dec 13 14:20:51 and some other embedded stuff(kernel(altough I'm a kernel newbie),debugging,android etc...) Dec 13 14:22:11 what should bainstorm for some things to propose as projects alsio Dec 13 14:22:50 isn't intership like normal work,but with more learning and some optional salary Dec 13 14:23:20 GNUtoo|laptop: here I don't "take" interns to do commercial work, only to work on internal projects/evaluations Dec 13 14:23:30 ah ok Dec 13 14:23:50 in other words, no interns in the critical path :) Dec 13 14:24:51 indeed maybe it's better like that Dec 13 14:25:12 no risk that way Dec 13 14:25:14 not really : what the intern develops can be critical, but for internal projects Dec 13 14:25:20 ok Dec 13 14:25:37 and here using the intern to work on commercial projects is on the border line of the law Dec 13 14:25:46 ah ok Dec 13 14:25:53 I didn't know that Dec 13 14:26:06 and not paying interns for intership of more than 2 month is out of the law (even if many companies are doing that) Dec 13 14:26:17 ok Dec 13 14:26:48 the onyl thing is to have a convention between your school/university and us Dec 13 14:27:00 ok Dec 13 14:27:05 I'll try to arrange that Dec 13 14:28:04 Crofton|work: concerning CELF projects, when I see projects for 2010 and their links, it's not clear wether they were supported or even approved Dec 13 14:28:39 okay time for christmas party at the kindergarten Dec 13 14:28:47 You can email them for more info ... Dec 13 14:28:53 now back to nanox, I wish I had an internship to do this crap :-) Dec 13 14:29:07 I saw the announcement and figured we should submit some ideas Dec 13 14:29:13 Crofton|work: will do if I find an idea Dec 13 14:29:33 one idea is an auomated test system of the software on the hardware Dec 13 14:30:16 I havent yet found something enough customizable for this Dec 13 14:38:00 Crofton|work: I found a "fix" for the coreutils issue Dec 13 14:38:19 ah Dec 13 14:38:23 seems to be related to the -e test used before running update-alternatives Dec 13 14:38:26 autoconf is screwing up again Dec 13 14:38:32 I swear it worked once last week Dec 13 14:39:19 Crofton|work: perhaps not enough coffee, but I can't grasp why it doesn't work as written :-) Dec 13 14:39:36 Crofton|work: http://www.sakoman.com/cgi-bin/gitweb.cgi?p=openembedded.git;a=commitdiff;h=d394170ca9733c5e89f68aff48f7726cbb5213c0 Dec 13 14:39:47 removing the -e test makes things work Dec 13 14:43:18 so that fails on first boot, but not afterwards Dec 13 14:43:37 what is base_bindir set to? Dec 13 14:44:09 er and bindir Dec 13 14:46:47 ericben: kexecboot was approved Dec 13 14:47:06 but I'm still not finished work proposed there Dec 13 14:47:10 Jay7: ah ok, and you really got paid for the work ? Dec 13 14:47:31 ericben: Tim Bird was asked me about week ago about checkout Dec 13 14:48:12 I've requested to wait one-two week to push some changes into git Dec 13 14:48:50 damn something else is messing up autotools Dec 13 15:12:23 kergoth_: hi kergoth_ it seems c63e55564a8840083dbd8634b10fe6f76d1f1354 in bitbake breaks perl-5.8.8 (I get ./miniperl command not found), 4d081a0ed759bd526ab01849d650bd9e8d80ddd1 works, 53740977521bc81ffa37adfa7bbeb8f2a80ea165 fails. Dec 13 15:13:36 ericben: it would be 53740977521bc81ffa37adfa7bbeb8f2a80ea165 that broke it, if anything. i'm already investigating a related issue with pcmciautils as reported by grg, there's a patch on the bitbake list that works around it, but i want to understand the root cause. Dec 13 15:13:57 kergoth: yes maybe 5374*, I'm testing c63e55564a8840083dbd8634b10fe6f76d1f1354 to confirm this Dec 13 15:15:11 its strange. cwd is passed to subprocess, to ensure it runs the run script in the right dir Dec 13 15:15:19 yet somehow adding 'cd %s' % cwd to the shell script fixes it Dec 13 15:15:23 which makes no sense Dec 13 15:15:26 kergoth: if that can help, I can send you the run.* scripts for both commits Dec 13 15:15:30 no need Dec 13 15:15:32 i know what it emits Dec 13 15:15:35 i wrote the commit Dec 13 15:15:52 sure ;) Dec 13 15:16:16 some log maybe, feel free to ask if you want me to try something Dec 13 15:17:13 i'm sure i'll be able to repro it locally, but i'll let you know if not, thanks Dec 13 15:17:20 will work on it today Dec 13 15:22:32 kergoth: "If cwd is not None, the child’s current directory will be changed to cwd before it is executed. Note that this directory is not considered when searching the executable, so you can’t specify the program’s path relative to cwd." Dec 13 15:22:44 from subprocess.Popen docs Dec 13 15:22:51 right.. Dec 13 15:22:55 i'm not seeing an issue with that Dec 13 15:23:04 cwd is set, its non-null, and its not used to find /bin/sh Dec 13 15:23:11 or the run script, actually Dec 13 15:23:38 * kergoth 'll investigate after he's done with this meeting he's in Dec 13 15:23:40 ah. That's right, bb is writing out a script first to run the command Dec 13 15:24:21 * foerster gets back to day-job stuff Dec 13 15:24:29 yeah, has to, since much of what it runs are shell functions right now, not external scripts Dec 13 15:25:00 of course, we could try passing it all to a shell via stdin, or use pysh, or something, but this works for now Dec 13 15:34:03 http://www.linuxfoundation.org/news-media/announcements/2010/12/linux-foundation-appoints-new-fellow Dec 13 15:34:29 RP goes to Linux Foundation Fellows - great Dec 13 15:35:24 RP, congratulations Dec 13 15:35:26 congratulations this should help yocto to become the future standard of embedded linux ! Dec 13 15:47:32 RP: congrats - excellent news ! Dec 13 15:53:46 cool! RP congratulations! Dec 13 15:56:59 Crofton: sorry went to breakfast Dec 13 15:57:24 I can't figure out why the tests fail ... Dec 13 15:57:27 Here is what the postinst on the target looks like: http://pastebin.com/ufxyaYSi Dec 13 15:57:44 Crofton|work: me either! that's why I just commented them out Dec 13 15:57:55 is -e a bashism? Dec 13 15:57:56 Crofton|work: FWIW, Poky doesnt have those tests Dec 13 15:58:05 Crofton|work: yes it is IIRC Dec 13 15:58:20 but my image is running bash and I use that in other scripts Dec 13 15:58:33 are you sure bash is sh when the scripts run? Dec 13 15:58:33 and I am sure that other postinst's must also Dec 13 15:58:42 maybe they screw up also? Dec 13 15:58:58 Crofton|work: I'm not really sure of much in life :-) Dec 13 15:59:03 yeah Dec 13 15:59:14 Crofton|work: -e isn't a bashism, assumign your'e talking about the test/[ argument Dec 13 15:59:23 yeah Dec 13 15:59:34 I think bash has test builtin or something Dec 13 16:00:07 yep Dec 13 16:00:53 I'm just tossing out things that could be different when it runs at first boot Dec 13 16:24:46 re Dec 13 16:25:14 * kergoth grumbles Dec 13 16:27:34 he kergoth whats up? Dec 13 16:28:03 1) boo meeting, 2) half asleep, 3) the bb.event rewrite i started on friday proved to be a slightly larger pain in my ass than i expected Dec 13 16:28:15 * kergoth may postpone that until a later date Dec 13 16:28:49 kergoth: https://github.com/foerster/bitbake/commit/0cde71444ff85cb1c28d888624767aea0d759ee6 Dec 13 16:29:09 otherwise, no log messages seem to go to the ui (only events make it there) Dec 13 16:29:29 ah, yes, the LogHandler has to be registered with the main logger, good call Dec 13 16:29:55 the event stuff.. we have events that get fired at higher levels in my event dispatcher tree, but currently events only flow up, not back down Dec 13 16:30:00 * kergoth may have to rethink the design Dec 13 16:30:41 why the need to flow down? Dec 13 16:30:55 boo meeting? Dec 13 16:32:36 foerster: runqueue fires StampUpdate, but the event handlers in the metadata are registered in the worker process, not the server process. if events only flow up, the StampUpdate will make it to the UI, but wont be handled by the handlers in the metadata Dec 13 16:32:41 as one example Dec 13 16:34:29 hrm Dec 13 16:34:32 wait, no, i guess we're fine Dec 13 16:34:40 as long as the metadata handlers are from the config metadata, not the recipe Dec 13 16:34:42 so nevermind that Dec 13 16:34:45 :) Dec 13 16:34:52 sweet, problem solved :) Dec 13 16:35:24 course, it still isn't working right, but that's likely just implementation bugs, not flaws in the overall design, at least i hope so Dec 13 16:35:39 check out the latest goggle when you get a chance, it's quite neat. Dec 13 16:35:43 will do Dec 13 16:35:57 * kergoth kicks one off Dec 13 16:36:13 ah, if the last commit i showed you is in there Dec 13 16:36:19 you'll get OE build info too Dec 13 16:36:28 then build started, build done, etc Dec 13 16:36:36 ooh, thats nifty Dec 13 16:36:48 i like how you see both overall messages and the recipe trees Dec 13 16:36:53 yep Dec 13 16:37:30 now, if we redo runqueue to use a pool of existing processes or something, that'll probably break (unless we add additional info to the logs and events) Dec 13 16:37:39 it's based on pid now Dec 13 16:37:43 well, sort of Dec 13 16:37:49 pid and task info Dec 13 16:38:16 i think we should add that extra info anyway, so that's fine with me Dec 13 16:38:21 better to make the associations explicit than implicit Dec 13 16:39:08 i like explicit. Some people don't - like those who don't want to put "static" in front of a module private function (in c). Or use const correctness.... Dec 13 16:39:12 but i rant..... :) Dec 13 16:39:16 hehe Dec 13 16:39:26 hey, its part of the zen of python, so.. ;) Dec 13 16:39:41 http://www.python.org/dev/peps/pep-0020/ Dec 13 16:39:59 sweet Dec 13 16:40:42 i work with some folks who don't care about software design/quality/architecture/etc. If it (somehow, half assedly, brokenly, maybe) seems to work, they're thrilled Dec 13 16:40:43 lol Dec 13 16:43:41 foerster: how is the google interface launched ? Dec 13 16:43:58 bitbake -u goggle mypackage Dec 13 16:44:06 so, just -u goggle Dec 13 16:44:21 foerster: Invalid user interface 'google' specified Dec 13 16:44:34 spelling Dec 13 16:44:35 goggle Dec 13 16:44:37 two g Dec 13 16:44:39 on o Dec 13 16:44:41 foerster: mm sorry Dec 13 16:44:41 one* Dec 13 16:45:03 not in master yet, on my branch Dec 13 16:45:11 yes I'm on your branch to try that ;-) Dec 13 16:45:15 ah, cool Dec 13 16:45:23 seems to work I have a progress bar Dec 13 16:45:59 ncurses, and depexp are also back to mostly functional Dec 13 16:47:05 foerster: the way log is rendered is quite interesting Dec 13 16:47:36 ericben: for the initial message, or just in general? Dec 13 16:47:53 when building a package now I see log I've never seen before and that I should have seen :-) Dec 13 16:48:08 lol Dec 13 16:48:09 but using several BBTHRAD makes the log difficult to see Dec 13 16:48:16 when using command line Dec 13 16:48:55 I like goggle, seems useful. It captures all logs and events, and puts them under their task if it can determine where it goes. Dec 13 16:49:34 when doing a big build, too many packages get started, and active ones are off the screen. I'd like to make the list "MRU", where active ones will automatically make sure they're near the top. Dec 13 16:50:02 would also be nice to just leave this running, and put a command line at the bottom Dec 13 16:50:45 I get a DeprecationWarning at line 178 of lib/bb/build.py Dec 13 16:51:00 that's due to using python 2.7, most likely Dec 13 16:51:04 (on a Fedora 14 with python 2.7) Dec 13 16:51:09 can you paste the exact warning? Dec 13 16:51:12 i noticed I get do_build doesn't exist a lot - poky eliminated that in base.bbclass by defining a stub. OE should do that if it doesn't break anything. Dec 13 16:51:13 you were faster than my explanation ;) Dec 13 16:51:17 :) Dec 13 16:51:24 foerster: agreed Dec 13 16:51:32 foerster: how can I copy/paste a log in the goggle interface ? Dec 13 16:51:48 don't think you can Dec 13 16:51:58 doubt you can, would be nice to be able to poke into the task output log somehow Dec 13 16:52:00 it seems I can't Dec 13 16:52:07 right clic -> psatebinit :-) Dec 13 16:52:16 would be great ! Dec 13 16:52:20 TaskStarted should really give you the log file, if it doesn't already Dec 13 16:52:21 ericben: this was existing software, I just resurrected it Dec 13 16:52:22 hmm Dec 13 16:52:38 but surely, we can add lots of nice features Dec 13 16:52:48 i want to make the log file clickable or something Dec 13 16:52:52 foerster: yes I'm joking, this is a great job Dec 13 16:53:09 ericben: no, we should add useful things to make it useful. Dec 13 16:54:10 Jay7, patches are applied by either khem or stefan_schmidt. Dec 13 16:54:33 foerster: yes, I wish I knew python to help you ... but that's not the case so appart from testing and filling the wishlist I'm not of big help here :-D Dec 13 16:54:47 ericben: I don't know python :) Dec 13 16:55:29 I've done python for a grand total of probably 3 weeks in my life. It's easy to pick up. Dec 13 16:55:35 but certainly, wish list is good Dec 13 16:57:42 kergoth: the warning is : build.py:178: DeprecationWarning:With-statements now directly support multiple context managers with nested(logfile, bb.utils.fileslocked(lockfiles)): Dec 13 16:58:55 ericben: try bin/bitbake, line 100. add a line: warnings.filterwarnings("ignore", message="With-statements now directly support multiple context managers") Dec 13 16:59:02 ericben: pleaes confirm that fixes it for you, if you would Dec 13 16:59:05 then i'll add it to master Dec 13 17:00:07 kergoth: line 100 of bin/bitbake is between 2 parse.add_option lines ? Dec 13 17:00:33 hmm didn't think so, must be a branch thing. put it after the other warnings.filterwarnings lines Dec 13 17:00:41 kergoth: ok Dec 13 17:02:33 ericben: if you pull, you can now copy text Dec 13 17:02:47 double click an entry and you'll get "edit mode", where you can copy the text Dec 13 17:03:03 not ideal solution, but works for now Dec 13 17:03:41 kergoth: no more warning with this line Dec 13 17:04:12 foerster: cool will try Dec 13 17:05:34 03Chris Larson  07master * r1c38f19fb8 10bitbake.git/bin/bitbake: Dec 13 17:05:34 Silence python 2.7 nested context manager warning Dec 13 17:05:34 Signed-off-by: Chris Larson Dec 13 17:05:40 foerster: works fine (the line gets unselected once a new one appear but copy/past works fine) Dec 13 17:06:36 foerster: ncurses seems to work also. One problem : resizing the terminal kills bitbake with IOError: [Errno 4] Interrupted system call Dec 13 17:07:19 foerster: kergoth : in fact resizing the teminal even without any ui kills bitbake Dec 13 17:08:19 with the following log : http://pastebin.com/bM9QYv6R Dec 13 17:08:31 JaMa: yes that binutils 2.21 issue is known to me. Dec 13 17:08:36 gm all Dec 13 17:08:56 hi khem Dec 13 17:09:01 hi Dec 13 17:09:12 ericben: thanks. we need to handle that, as resizing signals winch event Dec 13 17:09:20 khem: do you have any news concerning the gcc 4.5 + linaro bug ? Dec 13 17:11:05 foerster: in fact it seems it's only during parsing, not after when running tasks Dec 13 17:11:33 foerster: at least for the default ui Dec 13 17:12:46 ah, right, that's progressbar -- it needs to know about window resizes to handle the expansion to the width of the window Dec 13 17:12:53 iirc Dec 13 17:13:04 ericben: interesting. I would think it would be the other way, as the progress bar handles the window change signal. Dec 13 17:13:15 yeah it should be handling it.. strange Dec 13 17:13:18 hmm Dec 13 17:14:03 tested in a F14 with gnome terminal 2.32.0 Dec 13 17:14:10 ncurses hangs even after parse Dec 13 17:14:48 ericben: after a resize? or always? Dec 13 17:15:03 foerster: only after a resize Dec 13 17:15:10 don't worry :) Dec 13 17:15:43 what happens is we do a blocking call to .get(), which waits for events/logs from the server. Dec 13 17:15:55 SIGWINCH is thrown when you resize, which causes an io error Dec 13 17:16:01 we just need to catch it and continue Dec 13 17:16:45 for the knotty interface, it seems that after parsing I can't get it to hang (in fact after "INFO: oestats: build xyz" line) Dec 13 17:27:27 ericben: no not much. For now you can use the workaround Dec 13 17:27:41 disable 99379 patch Dec 13 17:29:12 ok, at 604fe17b5407465f381dd2d4c586e621c9684e17 this command: Dec 13 17:29:25 grep balister ../tmp/work/armv7a-angstrom-linux-gnueabi/autoconf-2.65-r10.0/image/usr/bin/* Dec 13 17:29:29 returns no lines Dec 13 17:32:12 ericben: pull again - i don't think it'll die from window change any longer Dec 13 17:32:37 * foerster has to dig out of the foot of snow, and go to the gym Dec 13 17:32:42 damn blizzard Dec 13 17:35:06 somehow, the path to m4 in staging creept back into autom4te since then Dec 13 17:35:26 hm hm Dec 13 17:41:13 Why the hell omap3-console-image would need OpenJade, sgml stuff (possible docbook) Dec 13 18:09:12 autoconf builds ok here : 6c0ce137ba0781bb7d3ddd6dfcf6d7d6af3f58df Dec 13 18:10:39 Whats wrong with autoconf? Dec 13 18:21:17 Crofton|work: Your target is same as mine (armv7a-angstrom-linux-gnueabi), could you possible try build OpenSP (it does not any dependency listed). I am looking for someone to confirm this one: http://bugs.openembedded.org/show_bug.cgi?id=5494 Dec 13 18:21:56 It broke me for last release, I want to know of it is only for me or not. Dec 13 18:25:38 hmm, we need to improve bitbake's handling of SIGTERM at some point. killing off the UI process with it doesn't bring down the server Dec 13 18:26:50 davidlt, I will after I find the commit that broke autoconf again Dec 13 18:26:59 I ned to be careful not to mess up staging Dec 13 18:27:25 Crofton|work: thats fine for me, just write what found out in bug report. Dec 13 18:29:33 Anyone comment on libgnutls-dev (labeled LGPL) but containg files from the extra package (GPLv3)? Dec 13 18:30:09 what recipe installs the kernel headers? Dec 13 18:30:19 martinambrose: Time for a -extra-dev package Dec 13 18:30:25 Crofton|work: for modules or userland? Dec 13 18:30:30 userland Dec 13 18:30:32 linux-libc-headers does userland Dec 13 18:30:37 hmmm Dec 13 18:30:44 are we still getting millions of unpackaged kernel file messages, or did we finally get those fixed? Dec 13 18:31:46 kegoth thats fixed Dec 13 18:31:50 only gcc is left Dec 13 18:31:50 ah, whew Dec 13 18:31:56 that was always so annoying Dec 13 18:32:23 So all, just kicked testing-next builds. Are there already known fixes pushed into master? Dec 13 18:32:25 kergoth yes I found the reason and fix was easy Dec 13 18:32:33 but for gcc it isnt Dec 13 18:38:56 GNUtoo|laptop: ping Dec 13 18:44:05 Can someone enlight me? I am confused about collections and layers Dec 13 18:44:28 otavio: in what way? Dec 13 18:44:43 layers are simply collections supported in a more convenient way by bitbake itself Dec 13 18:44:52 they're essentially identical, other than how they're configured Dec 13 18:45:00 conceptually each is just a pile of metadata Dec 13 18:46:53 #r Dec 13 18:46:57 ops Dec 13 18:47:38 collections.inc -> bblayers, amend.inc -> bbappend Dec 13 18:52:06 XorA: I missed the text of teh current resolution but I guess it's something along the lines of "The members support and endorse the efforts of the board to establuish a relationship"? Dec 13 18:52:52 broonie: yes Dec 13 18:53:48 "The board would like the eV membership to endorse the concept of Dec 13 18:53:49 developing a close collaborative relationship with the Yocto Project. " Dec 13 19:02:59 kergoth_: at conf/layer.conf it adds oe as a collection Dec 13 19:03:06 kergoth_: but LAYERSDIR is empty Dec 13 19:03:12 someone keeps breaking autoconf on the target, I am hunting you! Dec 13 19:03:14 XorA: Ah, that's not quite what I was thinking - I was thinking reword it into a vote of confidence in the eixsting efforts. Dec 13 19:03:14 kergoth_: how does it supposed to work? Dec 13 19:03:52 XorA: Would address koen's (reasonable & sane) issue - it's how most organisations I'm familiar with do this. Dec 13 19:04:08 Crofton|work: what exactly is the problem? Haven't run into problems so far Dec 13 19:04:08 otavio: LAYERDIR is *NOT* empty Dec 13 19:04:14 otavio: bitbake sets it as it parses the files Dec 13 19:04:19 Lopi, hi Dec 13 19:04:27 doesn't work Dec 13 19:04:34 XorA: If there's a problem then it changes into "the membership wants the board to stop doing this idiocti crap" or whatever Dec 13 19:04:38 the path to m4 in staging is used Dec 13 19:04:51 I think this is a very recent commit Dec 13 19:05:57 broonie: feel free to answer in the thread, this is good feedback Dec 13 19:06:01 Hmmm... How can I check this? Maybe this is could cause my problems with OpenSP ? Dec 13 19:06:17 XorA: I wanted to check what the actual resolution said first :) Dec 13 19:07:11 broonie: http://lists.linuxtogo.org/pipermail/openembedded-members/2010-December/000419.html Dec 13 19:07:13 full email Dec 13 19:07:37 otavio: the main difference in configuration between collections and bblayers is that with the latter, the layer is what adds itself to the vars to do the build. it results in unnecessary boilerplate most of the time, but i think having the knowledge of what to do with a layer in the layer isn't a bad concept in geenral Dec 13 19:07:48 XorA: Ah, that's pretty much what the resolution says already. Cool. Dec 13 19:13:32 * kergoth_ discovered git --author's "me" alias today. git log --author=me definitely saves on the typing Dec 13 19:15:46 kergoth_: right Dec 13 19:15:56 kergoth_: do you know any layer for me to use as reference? Dec 13 19:16:08 yes, OE Dec 13 19:16:26 * otavio confused Dec 13 19:16:27 if you you're wondering about bblayers.conf, not layer.conf, i have a simple repo on github Dec 13 19:16:32 look at openembedded/conf/layer.conf Dec 13 19:16:44 once again, a layer is a pile of metadata. OE is a layer, just the base one Dec 13 19:17:09 hum Dec 13 19:17:17 kergoth_: I see Dec 13 19:17:32 then in bblayers.conf, you set BBLAYERS to point at your layers Dec 13 19:17:43 and you set BBPATH there, just with your build directory only, since the layers will add themselves to it Dec 13 19:18:08 https://github.com/kergoth/OE-BBLayers/blob/master/conf/bblayers.conf Dec 13 19:18:18 so using that, there's no need to set any env vars for bitbake anymore if you don't want to. Dec 13 19:18:22 just need bitbake in your PAHT Dec 13 19:18:47 that bblayers.conf assumes that your bblayers.conf and local.conf live in the same place, and you cd into there to run your builds Dec 13 19:22:39 https://github.com/kergoth/bitbake/commit/c181d437c78951306505ccdffb6eb411ae5fd462 Dec 13 19:22:47 definite work in progress Dec 13 19:22:55 but for the curious, thats what i started on on friday Dec 13 19:26:44 damnit, StampUpdate is going to be a pain Dec 13 19:27:14 its fired in execute_runqueue, during the initialization phase, so the metadata handlers need to be registered at that point in time Dec 13 19:27:31 yet the config metadata handlers need to be unregistered when the recipe is parsed and its handlers registered, or we get duplicate handlers Dec 13 19:28:00 or i wipe __BBHANDLERS after registering the config metadata handlers, but that would break a handler that responds to RecipeParsed being defined in the config metadata Dec 13 19:28:50 foerster|away: bitbake ncurses doesn't hang with your updates (the ncurses interface doesn't resize) Dec 13 19:29:30 foerster|away: same with knotty interface : no crash even during terminal resize Dec 13 19:33:34 i think it might be better to postpone the bb.event rewrite until after the bb.runqueue one Dec 13 19:34:54 alternatively, runqueue's initialization could be shifted to construction time, and then the cooker could unregister its handler after the creation of the runqueue object, but before executing it Dec 13 19:34:58 hmm Dec 13 19:35:42 holy blizzard Dec 13 19:35:51 complete white out :) Dec 13 19:35:55 yikes Dec 13 19:36:15 foerster: https://github.com/kergoth/bitbake/commit/b00fa19b512046a652ea5a5c583314fe7961d163 seem like a good idea to you? Dec 13 19:36:31 it seems like in cases where there's both an event and a log message, this should be consolidated and left to the UI's discretion Dec 13 19:39:26 kergoth: that seems sane to me Dec 13 19:40:41 03Chris Larson  07master * rab831e867f 10bitbake.git/lib/bb/ (runqueue.py ui/knotty.py): Dec 13 19:40:41 Move the runq task start/failed messages to the UI Dec 13 19:40:41 Signed-off-by: Chris Larson Dec 13 19:40:43 03Richard Purdie  07master * re7c7e1f2e5 10bitbake.git/lib/bb/runqueue.py: Dec 13 19:40:43 bitbake/runqueue.py: Create RunQueueExecute and RunQueueExecuteTasks classes, further splitting up runqueue Dec 13 19:40:43 Signed-off-by: Richard Purdie Dec 13 19:40:44 03Richard Purdie  07master * rbf978ca017 10bitbake.git/lib/bb/ (cooker.py runqueue.py): Dec 13 19:41:11 yeah rp Dec 13 19:41:32 merging more poky bits Dec 13 19:41:40 up next is the scenequeue and checksum bits Dec 13 19:42:01 oh poky has its own bitbake? Dec 13 19:42:38 yep Dec 13 19:42:42 its been split off for a long time Dec 13 19:42:47 its stored inside of the poky repo Dec 13 19:43:40 hm oh didnt notice that Dec 13 19:43:57 gm Dec 13 19:44:10 merging the bitbakes is extremely important, and we want to get it done before the next bitbake release, which we wnat to get out in the next couple weeks Dec 13 19:44:59 cool Dec 13 19:45:36 kergoth_ nice! Dec 13 19:46:06 kergoth_: can you push those scmdata=keep patches from obi? Dec 13 19:46:36 khem, ping, any progress on the mediatomb/js thingie? I' available for the next 2 hrs to compare notes, objdump libs etc etc Dec 13 19:47:23 JaMa: i suppose, it just seems silly since the fetcher rewrite is on the yocto roadmap Dec 13 19:48:14 bother, can't reproduce my autoconf issue by stepping forward through revs Dec 13 19:48:22 well if you don't have objections to scmdata as name for param then it would be great to support it in metadata Dec 13 19:48:42 kergoth_: I have 2 patches for efl to keep those as efl fails to build with bitbake master Dec 13 19:49:13 ericben|away: thanks for testing. I'll have to see why ncurses doesn't resize - but at least no crashes now :) Dec 13 19:49:25 ah. okay, well, it certainly won't hurt to get it in, will do so Dec 13 19:50:56 kergoth_: thanks Dec 13 19:52:54 What might be a problem, with never rev my OpenSP fails and it does not have any dependencies and the recipe was not changed from last used rev. Dec 13 19:52:57 I can't get it. Dec 13 19:53:36 davidlt: whats the error Dec 13 19:53:59 eFfeM: my js is still building should take another 5 mins Dec 13 19:54:10 khem: http://bugs.openembedded.org/show_bug.cgi?id=5494 Dec 13 19:54:18 khem, ok thanks, np Dec 13 19:57:06 foerster: the ncurses.py looks like it needs a sigwinch handler... I don't believe curses.wrapper does it for you Dec 13 19:57:45 rphillips: that's possible. I figured it would do it automatically. Dec 13 19:58:29 davidlt: can you remove ../lib/libosp.la from the link line and redo the linking part Dec 13 19:59:01 give me several minutes Dec 13 20:04:12 khem: it gave huge list on undefined refs now from OpenSP:: Dec 13 20:04:41 khem, if svn protocol does not work, http is infinitely faster :) Dec 13 20:05:03 bah. ncurses ui will remain without resizing for a while, as it seems a PITA. Have to do it manually I believe. Dec 13 20:05:07 Crofton|work: for some svn protocol work Dec 13 20:05:20 I know Dec 13 20:06:04 eFfeM: I can reproduce the problem you are seeing Dec 13 20:06:37 kergoth, 61c8f6dca9422edcb477c81bbcfd1a3b1490eb59 breaks autoconf on the target Dec 13 20:06:41 khem, great, that means it is not some oddity in my system Dec 13 20:07:49 kergoth, you acked it :) Dec 13 20:08:37 mickey|bbiab, did you figure out anything about i2c-dev? Dec 13 20:10:33 khem, if there is something I can do to help debugging the issue, please let me know Dec 13 20:11:00 Crofton|work: its just that before we switch default it will be nice to know what we are losing we know what we are gainin Dec 13 20:15:13 khem: both my build environments also uses the save version libtool Dec 13 20:16:16 I actually put virtual machine with clean environment for building OpenSP Dec 13 20:17:46 eFfeM: the issue happens in js Dec 13 20:18:10 eFfeM: so when libjs.so is built it pulls in some undefined references to static libgcc which is like death Dec 13 20:18:45 khem, I agree, just poking fun at people who have to work behind restrictive firewalls Dec 13 20:18:54 davidlt: can u build this cleanly when u use say angstrom2008 Dec 13 20:19:23 Crofton|work: they should work with their IT and create local mirror Dec 13 20:19:26 and be done with this Dec 13 20:19:36 khem: not right now, but later yes Dec 13 20:19:44 or fix their fireqall Dec 13 20:19:54 they will hav eloads of other fetching issues Dec 13 20:20:01 its insane if your IT does not let you go out and you want to build something that sucks from all over www Dec 13 20:20:28 so I want to protect interest of those who still enjoy the freedom :) Dec 13 20:23:26 ohhh.. Dec 13 20:23:49 khem: rerun on my stable environment after cleaning opensp and it failed too Dec 13 20:24:08 It's getting strange... Dec 13 20:24:29 there was zero updates on recipes and the whole system. Dec 13 20:25:26 just few days ago I had trouble with angstrom-version recipe, which didn't compile for two days and after that like compiled just fine. Dec 13 20:25:49 Let's see what client environment on vm gonna say. Dec 13 20:26:23 eFfeM: seems something is fishy in libgcc how its built Dec 13 20:26:36 eFfeM: I can offer you a workaround for now Dec 13 20:27:34 khem, ok Dec 13 20:27:42 eFfeM: add ARM_INSTRUCTION_SET = "arm" to js_1.5.bb then bitbake js -c clean;bitbake mediatomb Dec 13 20:28:02 will try that Dec 13 20:28:30 khem do you want me to commit a patch for js if it works ? Dec 13 20:29:37 * eFfeM kicks of a new testing bulld now mtd-utils fetches again Dec 13 20:32:34 hm.. Dec 13 20:32:45 how to cleanup tmpdir except toolchain? Dec 13 20:32:51 is it possible? Dec 13 20:33:29 clean sysroot/host Dec 13 20:33:32 at other side I can build toolchain, save tmpdir, then restore it when needed Dec 13 20:33:35 and the stamps for it Dec 13 20:39:18 03Klaus Kurzmann  07master * r99a553d1eb 10openembedded.git/recipes/udev/ (files/nokia900/70-persistent-net.rules udev_162.bb): Dec 13 20:39:18 udev_162.bb: fix rule for persistent wlan0 and bump PR Dec 13 20:39:18 Signed-off-by: Klaus Kurzmann Dec 13 20:42:22 khem, with the workaournd mediatomb builds fine, do you want me to push the js patch Dec 13 20:42:28 (and thanks for your help) Dec 13 20:44:31 http://pastebin.com/sQawisgb Dec 13 20:44:38 anyone see any problems with this? Dec 13 20:46:51 eFfeM: no Dec 13 20:47:10 eFfeM: the problem seems to be in how libjs is linked Dec 13 20:48:23 eFfeM: as I see its calling ld directly Dec 13 20:48:38 which will break all sort of linking unless one knows what he/she is doing Dec 13 20:48:47 and js folks seems to have no clue Dec 13 20:51:16 eFfeM: try to set LD="${CC}" in js recipe and see if that helps Dec 13 20:51:22 and if it does commit that fix Dec 13 20:53:39 ok, will try Dec 13 20:54:26 khem, I feel the issue is not in js but in whoever created the recipe Dec 13 20:54:26 EXTRA_OEMAKE = "'CC=${CC}' 'LD=${LD}' 'XCFLAGS=${CFLAGS}' 'XLDFLAGS=-L${STAGING_LIBDIR} -soname=libjs Dec 13 20:55:22 khem I'll try first without the CC and LD assignments in the line above Dec 13 20:56:24 eFfeM: yeah seems so Dec 13 20:57:41 wt* Dec 13 21:04:19 khem, I have this friend back again, had that also last week but in a different file Dec 13 21:04:20 | jsinterp.c:1744: warning: ���cs��� may be used uninitialized in this function Dec 13 21:04:33 got 5 of these Dec 13 21:04:48 trying now if it is reproducable Dec 13 21:04:59 previous time it went away when I tried again Dec 13 21:06:34 it does not now :-( Dec 13 21:11:03 eFfeM: these randomly occurring problems are often around here? Dec 13 21:11:37 davidlt: I've only seen them with js recently (and had a full rebuild) Dec 13 21:12:03 I have two in last week Dec 13 21:12:26 One fixed after two days, just cleaning/building many times Dec 13 21:12:28 davidlt: which recipe ? Dec 13 21:12:41 First was angrstrom-version, which is very simple Dec 13 21:12:44 And no OpenSP Dec 13 21:12:49 *now Dec 13 21:13:15 I just build OpenSP on stable build host two days ago, cleaned, rebuild -> fail Dec 13 21:13:58 khem, if I just remove the CC and LD definitions I get | arm-oe-linux-gnueabi-ld: Linux_All_DBG.OBJ/jsmathtemp.o: Relocations in generic ELF (EM: 3) Dec 13 21:14:03 With angr...-version lsb_release file was never copied to working directly, that was very strange Dec 13 21:14:54 khem and if I add LD=${CC} I get: | /home/frans/oe/tmp_minimal/sysroots/i686-linux/usr/armv5te/lib/gcc/arm-oe-linux-gnueabi/4.5.2/../../../../arm-oe-linux-gnueabi/bin/ld: cannot find -lgcc_s Dec 13 21:15:18 and I keep getting those strainge errors with nonprintable Dec 13 21:15:20 char Dec 13 21:18:41 khem the weird warnings seem to go away if I add PARALLEL_MAKE = "" Dec 13 21:18:57 * kergoth_ sighs Dec 13 21:19:40 I retract that, there are still some left Dec 13 21:20:31 eFfeM: it does not go away probably it just hides away Dec 13 21:20:39 did u check the compile.log Dec 13 21:20:52 yes, thtat's why i retracted my remark Dec 13 21:20:59 eFfeM: yes that lgcc_s error is because its really wanting to use bare ld Dec 13 21:21:02 is this known ? Dec 13 21:21:23 its known to me since 5 mins Dec 13 21:21:25 the strange chars Dec 13 21:21:27 ah ok Dec 13 21:21:36 js is garbage build system Dec 13 21:21:58 http://www.linuxfordevices.com/c/a/News/LF-appoints-Richard-Purdie-as-Fellow/?kc=rss Dec 13 21:22:03 eFfeM: it seems openebedded ist broken .. http://pastebin.com/gukPSmD3 Dec 13 21:23:00 mybuild now fails after a git pull . vit bitbake errors, no build error Dec 13 21:24:06 pwgen: which bitbake version? Dec 13 21:24:45 "BitBake Build Tool Core version 1.8.19, bitbake version 1.8.19" and "BitBake Build Tool Core version 1.8.18, bitbake version 1.8.18" Dec 13 21:27:24 pwgen: fixed in 1.10, also in the 1.8 branch of git, but afaik not in a 1.8 release yet Dec 13 21:27:44 (fixed in the 1.8 branch of git as of about 30 seconds ago ;) Dec 13 21:28:33 03Denis 'GNUtoo' Carikli  07org.openembedded.dev * r2b1ee3146f 10openembedded.git/recipes/fmtools/fmtools_1.0.2.bb: (log message trimmed) Dec 13 21:28:33 fmtools: fix QA staging(GNU hash) Dec 13 21:28:33 Without that fix we have: Dec 13 21:28:33 ERROR: QA Issue with fmtools: No GNU_HASH in the elf binary: '.../work/armv7a-oe-linux-gnueabi/fmtools-1.0.2-r1/packages-split/fmtools/usr/bin/fm' Dec 13 21:28:33 ERROR: QA Issue with fmtools: No GNU_HASH in the elf binary: '.../work/armv7a-oe-linux-gnueabi/fmtools-1.0.2-r1/packages-split/fmtools/usr/bin/fmscan' Dec 13 21:28:43 03Denis Carikli  07org.openembedded.dev * rfbc3a2e8ad 10openembedded.git/recipes/netbase/ (netbase/iphone3g/interfaces netbase_4.21.bb): Dec 13 21:28:43 netbase: add iphone3g support Dec 13 21:28:43 Without that patch, the user can't ssh into his device, Dec 13 21:28:43 because usb0 is not up on the device side Dec 13 21:28:44 Signed-off-by: Chris 'Lopi' Spehn Dec 13 21:28:51 making the user mess with this 121 controls(if he does he would Dec 13 21:28:51 spend hours trying to figure out how to make it work) Dec 13 21:28:52 Signed-off-by: Chris 'Lopi' Spehn Dec 13 21:28:52 03Denis Carikli  07org.openembedded.dev * r1ac08d1b21 10openembedded.git/ (3 files in 3 dirs): (log message trimmed) Dec 13 21:28:53 : ....testing .... Dec 13 21:28:53 Add initial support for iphone3g(machine config and kernel) Dec 13 21:28:54 Android was ported to Apple's iDevices including the iPhone3G by the Dec 13 21:28:54 iDroid Project ( http://www.idroidproject.org/wiki/Main_Page ) Dec 13 21:31:56 : seems to work. now starting building for Angstrom for ion Dec 13 21:33:33 khem peeked at the offending command. see http://www.pastebin.ca/2018800 Dec 13 21:33:48 it does apparently have a symbol or function that it wants to fetch from gcc_s Dec 13 21:38:46 khem, made some progress, gcc generates a collect2 line with two times --as-needed -lgcc_s in it, if I cut and paste the collect2 command (which I got from gcc -v) and remove those two references to -lgcc_s things seem to work fine, Dec 13 21:38:57 khem no idea how I can tell gcc not to generate these Dec 13 21:39:56 eFfeM: OK I fixed js for you Dec 13 21:40:17 give me a min to commit it Dec 13 21:40:38 khem, cool (& curious how you fixed it) Dec 13 21:40:41 thanks alot! Dec 13 21:43:42 see the patch Dec 13 21:43:55 thanks, have it Dec 13 21:44:54 03Khem Raj  07master * ra8aa382702 10openembedded.git/recipes/js/ (js/ld.patch js_1.5.bb): Dec 13 21:44:54 js_1.5.bb: Use gcc to link shared library and not bare ld Dec 13 21:44:54 Get rid of do_stage while here Dec 13 21:44:54 Signed-off-by: Khem Raj Dec 13 21:45:32 03Philip Balister  07org.openembedded.dev * r2fb096e8be 10openembedded.git/recipes/autoconf/autoconf.inc: Dec 13 21:45:32 autoconf : Fix autoconf running on target. Dec 13 21:45:32 Acked-by: Koen Kooi Dec 13 21:47:16 03Khem Raj  07master * rd7e761cd3a 10openembedded.git/recipes/binutils/binutils_2.21.bb: Dec 13 21:47:16 binutils_2.21.bb: Dont apply libtool-update.patch at all Dec 13 21:47:16 Signed-off-by: Khem Raj Dec 13 21:48:07 khem, should saying LD=${CC} not have fixed it too ? Dec 13 21:51:57 khem, anyway, I can confirm that your patch works Dec 13 21:52:09 eFfeM: no it wont Dec 13 21:52:17 LD is used in varied forms in js Dec 13 21:52:38 ah ok, thought the LD var would always be overwritten by CC Dec 13 21:52:40 sometimes it uses it for partial linking sometimes to link shared objects and so on Dec 13 21:53:02 * khem sent the cross-localedef patch to ml Dec 13 21:53:16 ah forgot to pay the dues to poky Dec 13 21:55:17 khem btw my log file still has those odd chars, not peeking into that any more today, need to go Zzz, up early tomorrow Dec 13 21:55:30 khem & thanks alot for your help! Dec 13 22:01:22 Hmm... All my build environments/machines now fails Dec 13 22:04:10 calling it a day! Dec 13 22:04:18 have fun & see you tomorrow or so Dec 13 22:05:23 03Klaus Kurzmann  07master * r363f05f027 10openembedded.git/recipes/freesmartphone/cornucopia.inc: Dec 13 22:05:23 cornucopia.inc: bump FSO_CORNUCOPIA_SRCREV Dec 13 22:05:23 This brings fixed alsa statefiles for openmoko and some Dec 13 22:05:23 nice new plugins for nokia900. Dec 13 22:05:23 Signed-off-by: Klaus Kurzmann Dec 13 22:06:48 wiw Dec 13 22:06:53 s/i/o/ Dec 13 22:07:09 khem: I see patch about cross-localedef Dec 13 22:07:17 I'll run testbuild now Dec 13 22:11:04 khem: funny thing is that I can't even compile OpenSP on my Ubuntu or Mac Dec 13 22:11:15 But I get different errors Dec 13 22:12:44 khem: JFYI git am log: http://pastebin.ca/2018843 Dec 13 22:18:05 khem: testing build started for {angstrom-2008.1, minimal} {collie, akita, efikamx, ben-nanonote, qemux86} console-image Dec 13 22:23:55 Jay7: ok Dec 13 22:24:12 I have not tested glibc so your angstrom 2008.1 should catch that Dec 13 22:24:28 I am all fine to keep using qemu for glibc Dec 13 22:24:49 but for eglibc we should use cross-localedef now on Dec 13 22:24:49 angstrom is going first Dec 13 22:24:54 so we will know soon Dec 13 22:25:10 btw, is it enabled by default? Dec 13 22:25:20 I've not changed any variables Dec 13 22:25:48 Jay7: yes for eglibc it is Dec 13 22:26:05 is it for glibc? Dec 13 22:26:27 ah, I see +LOCALE_GENERATION_WITH_CROSS-LOCALEDEF = "1" Dec 13 22:27:18 +# set "0" for qemu emulation of native localedef for locale generation Dec 13 22:27:29 seems *instead* is missing :) Dec 13 22:34:14 03Klaus Kurzmann  07master * r283a6b3e1a 10openembedded.git/recipes/freesmartphone/fso-specs_git.bb: Dec 13 22:34:14 fso-specs_git.bb: bump SRCREV as needed by current cornucopia Dec 13 22:34:14 Signed-off-by: Klaus Kurzmann Dec 13 22:34:17 03Klaus Kurzmann  07master * rff2ef6e71c 10openembedded.git/recipes/freesmartphone/libfso-glib_git.bb: Dec 13 22:34:17 libfso-glib_git.bb: bump PR to pickup the fso-specs bump Dec 13 22:34:17 Signed-off-by: Klaus Kurzmann Dec 13 22:35:20 Jay7: thx I squelched those trailing whitespaces now locally Dec 13 22:35:48 cool Dec 13 22:44:01 khem: http://pastebin.ca/2018872 Dec 13 22:44:14 but it is still building Dec 13 22:44:22 not sure how critical is it Dec 13 22:44:58 hm.. Dec 13 22:45:28 may be it was something other.. Dec 13 22:46:23 NOTE: package glibc-2.9-r37.4: task do_configure: Failed Dec 13 22:46:24 ERROR: Function 'do_configure' failed Dec 13 22:46:28 no, it is Dec 13 22:47:23 khem: http://tinderbox.openembedded.net/packages/1262322/ Dec 13 22:47:40 ah.. no log there.. Dec 13 22:49:31 Jay7: ok thought of that let me fix it here Dec 13 22:49:50 I'll wait :) Dec 13 22:50:15 meh Dec 13 22:50:43 kergoth: btw, I'm running on bb master :) Dec 13 22:50:47 so we will test it too Dec 13 22:50:49 nice Dec 13 22:52:17 what to bitbake to build *libc only? Dec 13 22:52:22 virtual/libc? Dec 13 22:53:06 I mean libc version preffered by distro Dec 13 22:53:22 Jay7: ok sent an addon patch to ml Dec 13 22:53:25 apply that Dec 13 22:53:28 for glibc on top Dec 13 22:53:51 * khem think why we have so much undertaking on OE Dec 13 22:53:59 s/on/in Dec 13 22:55:46 hm.. patchwork is slow.. Dec 13 22:56:07 I've mail in my inbox, but nothing in PW still.. Dec 13 22:56:16 got it Dec 13 22:56:35 yes pw takes about 5mins Dec 13 22:56:59 you can not poll more often than that other wise Mail servers kick you out Dec 13 22:57:36 is it grabbing mail via pop/imap? Dec 13 22:57:55 I've thought it was working via smtp aliases Dec 13 22:58:04 well, delivery aliases Dec 13 22:58:10 hey all... I was trying to build "console-image" but I am getting this error: error: #error GNU libiconv not in use but included iconv.h is from libiconv -- has anyone seen this before? Dec 13 22:58:54 orbarron: what recipe is failing? Dec 13 22:59:19 Jay7: glib-2.24.1 Dec 13 22:59:40 orbarron: what is distro and machine? Dec 13 22:59:50 03Chris Larson  07master * refa5485c8f 10bitbake.git/lib/bb/process.py: Dec 13 22:59:50 build: set PWD in the subprocess preexec hook Dec 13 22:59:50 Signed-off-by: Chris Larson Dec 13 23:00:20 btw.. Dec 13 23:00:36 sakoman, should I push the coreutils patch upstream? Dec 13 23:00:46 Jay7: distro: angstrom -- machine: pandaboard Dec 13 23:01:00 kergoth: do we have bb option just to report build configuration? Dec 13 23:01:09 not sure what you mean. Dec 13 23:01:17 its not a function of bitbake Dec 13 23:01:22 bitbake has zero knowledge of those variables Dec 13 23:01:24 its all oe Dec 13 23:01:25 its printed first if you use bb master Dec 13 23:01:34 and its printed by OE, from an event handler Dec 13 23:01:40 Jay7: opps: I mean omap4430-panda Dec 13 23:01:41 ah, well Dec 13 23:01:57 I just mean command to show this configuration w/o doing build Dec 13 23:02:17 to do debugging such cases a bit easy :) Dec 13 23:02:34 may be -n Dec 13 23:02:37 OE can't add commandline arguments to bitbake Dec 13 23:02:38 instead of asking bunch of vars just ask to show command results Dec 13 23:02:48 yeah, -n would do, its printed when the build starts Dec 13 23:03:19 orbarron: yes this error is seen sometimes its kind of a catch22 condition sometimes Dec 13 23:03:27 orbarron: whats your build os Dec 13 23:04:31 host: ubuntu 10.04 -- target: omap4430-panda Dec 13 23:06:26 khem: nope.. host is actually 8.04 Dec 13 23:06:46 hmm Dec 13 23:09:50 ka6sox: I got internal mediawiki error while trying to change page Dec 13 23:10:22 ka6sox: I've tried to update info on this page: http://wiki.openembedded.net/index.php?title=Testing_with_QEMU Dec 13 23:10:47 Jay7, thanks Dec 13 23:10:51 let me see whats up with that Dec 13 23:11:32 n8 Dec 13 23:11:34 hi folks, have a couple newbie questions about sdk distribution/installation if anyone can help? Dec 13 23:13:03 hbeck: just ask Dec 13 23:14:29 Jay7, were you trying to edit it?\ Dec 13 23:14:32 it works for me Dec 13 23:15:10 ka6sox: I've changed link to linuxtogo (prepended www.), mark 'minor edit' and filled summary Dec 13 23:15:16 then click save page Dec 13 23:15:22 orbarron: ok I have seen such issue on old ubuntu may be you are hitting same Dec 13 23:15:36 kergoth: how can I get reverse diff of a commit Dec 13 23:15:49 git show gives the normal diff Dec 13 23:15:56 git diff commit..commit~1 Dec 13 23:16:09 Jay7, oh, its probably needing approval now as I've made it so that we can try to eliminate the wikispam Dec 13 23:16:47 Does the tree location for the sdk have to be fixed (root) or can it be installed under something like /home ? Dec 13 23:16:52 ka6sox: but it just show me Internal wiki error page :) Dec 13 23:17:10 kk...let me look at why the plugin is broken Dec 13 23:17:29 hbeck: I think it would want root Dec 13 23:18:32 ha git show -R :) cool Dec 13 23:18:36 * khem hugs git Dec 13 23:20:53 khem: Any other good suggestions? My all build environments doesn't work anymore.. Dec 13 23:21:02 khem: ok so no way to make the compilers relocatable? If so, is there an install method (other than just using tar) for a non-linux-savvy user base? Dec 13 23:22:27 I should say: a usual install method other than tar Dec 13 23:25:41 03Chris Larson  07master * r76aeea2c23 10bitbake.git/bin/bitbake: Dec 13 23:25:41 Ensure that the invalid UI error goes to stderr Dec 13 23:25:41 Signed-off-by: Chris Larson Dec 13 23:29:48 khem: ah, nice. i like that its the same argument as patch :) Dec 13 23:32:24 03Andrea Adami  07master * r4a7c6815cf 10openembedded.git/recipes/python/python-pygtksourceview_2.6.0.bb: (log message trimmed) Dec 13 23:32:24 python-pygtksourceview: hack around PKG_CONFIG issues and unbreak build Dec 13 23:32:24 * temporarly apply same fix (sed) used in python-pygtk Dec 13 23:32:24 * | checking for pygtk codegen... /usr/share/pygobject/2.0/codegen Dec 13 23:32:24 * | no Dec 13 23:47:21 hbeck|away: I dont think there is one atm Dec 13 23:51:50 khem: hey Dec 13 23:53:20 now samba is failing here vs. gcc-4.5 ...http://www.pastebin.ca/2018942 Dec 13 23:53:48 ant__: heh known problem Dec 13 23:53:56 :/ Dec 13 23:54:18 good news is that I have a fix Dec 13 23:54:25 I will just send it to ml Dec 13 23:54:30 and you can test it for me ?? Dec 13 23:54:34 great, thx Dec 13 23:54:38 ofc Dec 13 23:54:49 Jay7, try again Dec 13 23:56:44 hm.. btw I could only push to master and not to org.openembedded.dev anymore ?!? Dec 13 23:56:50 khem: locales was generated Dec 13 23:57:20 ka6sox: moment Dec 13 23:57:50 ka6sox: same sh@t :( Dec 13 23:58:02 ant__: hey Dec 13 23:58:04 arrgh...okay Dec 13 23:58:14 ant__: samba was always failing vs gcc 4.5 :( Dec 13 23:58:31 ah.. hope to fix :) Dec 13 23:58:35 Jay7: he..I try now 'sugar' with angstrom-2010 Dec 13 23:58:53 no probs with older gcc Dec 14 00:03:14 ant__: Jay7 just posted another patch for gcc please try it out and see if samba builds after applying this patch to gcc Dec 14 00:03:20 you have to rebuild gcc Dec 14 00:03:38 khem: tomorrow then.. Dec 14 00:03:43 bitbake -c clean gcc-cross gcc-cross-initial gcc-cross-intermediate eglibc eglibc-initial Dec 14 00:03:49 Jay7: ok Dec 14 00:03:58 current build is running for console-image Dec 14 00:04:02 there is no samba Dec 14 00:04:08 ant__: you can try it out and report back to ml if this works Dec 14 00:04:20 k Dec 14 00:04:30 khem: Did that patch to stop doing the source dir move dance fix a problem w/ gcc failing to patch? Dec 14 00:04:37 hm.. but I can start parallel build for samba only Dec 14 00:05:01 WARNING: Function do_build doesn't exist Dec 14 00:05:05 hehe.. Dec 14 00:05:15 Tartarus: yes it did Dec 14 00:05:18 k Dec 14 00:05:29 i bet that's most of my "broken since this build" failures :( Dec 14 00:05:29 I've seen fix for this warning in ML already Dec 14 00:05:45 Tartarus: in some case instead of renaming gcc-4_5-branch to gcc-4.5 it moved gcc-4_5-branch into gcc-4.5 Dec 14 00:06:12 and then failed to apply patches obviously Dec 14 00:06:28 somehow someone created gcc-4.5 under WORKDIR before the move Dec 14 00:06:53 it was not required anyway Dec 14 00:07:15 Tartarus: Try the cross-localedef stuff I posted patches Dec 14 00:08:12 there are two second one is needed for glibc, in the end I will squash them into one I think that would be better Dec 14 00:08:28 tad busy atm, not sure if i'll get to them before they hit the next testing branch Dec 14 00:08:57 ok for that it would have to be committed :) Dec 14 00:12:39 khem: fwiw NOTE: Gettext required but not in DEPENDS for file /oe/build/tmp/work/armv5te-angstrom-linux-gnueabi/gcc-cross-initial-4.5-r25.0+svnr167449/gcc-4_5-branch/intl/configure.ac. Dec 14 00:12:39 Missing inherit gettext? Dec 14 00:12:50 dont worry about that Dec 14 00:12:57 gcc has its own internal gettext Dec 14 00:13:13 I thought those warnings had been disabled Dec 14 00:13:14 khem: Yeah, so better get those bits in ;) Dec 14 00:13:31 ok we still have couple of days Dec 14 00:13:51 I will see if someone finds issues before putting them in Dec 14 00:56:46 khem: still generating locales (my bad I missed the patch ;) Dec 14 00:57:00 samba is WIP Dec 14 00:58:24 * Jay7 was hit into out of max No of open files :( Dec 14 00:58:25 (listening to a wonderful hammond song in the meanwhile) Dec 14 01:19:44 kergoth_: ping Dec 14 01:27:23 03Paul Eggleton  07master * r01da496399 10openembedded.git/recipes/opie-oxygen/opie-oxygen_1.0.0.bb: Dec 14 01:27:23 opie-oxygen: remove obsolete 1.0.0 version Dec 14 01:27:23 This has long been superseded; version 1.2.4 is the current version. Dec 14 01:27:23 Signed-off-by: Paul Eggleton Dec 14 01:27:43 03Paul Eggleton  07master * rcb2e8e8ed6 10openembedded.git/ (187 files in 176 dirs): Dec 14 01:27:43 Opie: remove version 1.2.3 recipes Dec 14 01:27:43 Opie 1.2.3 is no longer referred to by any distro and has been superseded Dec 14 01:27:43 some time ago by 1.2.4. Dec 14 01:27:44 Signed-off-by: Paul Eggleton Dec 14 01:28:26 hm.. looks like mtd-utils was compiled with objdir == $OE_DIR Dec 14 01:28:39 not sure when it was happens Dec 14 01:28:41 03Paul Eggleton  07master * r5414daaf30 10openembedded.git/ (6 files in 3 dirs): Dec 14 01:28:41 Opie: remove opie-cardmon & opie-pcmciaapplet Dec 14 01:28:41 These have been superseded by opie-cardapplet as of version 1.2.4. Dec 14 01:28:41 Signed-off-by: Paul Eggleton Dec 14 01:28:53 03Paul Eggleton  07master * rb025571a85 10openembedded.git/ (187 files in 174 dirs): Dec 14 01:28:53 Opie: add version 1.2.5 Dec 14 01:28:53 Signed-off-by: Paul Eggleton Dec 14 01:28:53 03Paul Eggleton  07master * r256938213f 10openembedded.git/ (188 files in 176 dirs): Dec 14 01:28:53 Opie: switch recipes away from CVS Dec 14 01:29:12 wow Dec 14 01:29:16 new opie Dec 14 01:32:38 Jay7: you getting mtd-utils build files in the build directory and not the mtd-utils work directory ? Dec 14 01:32:44 dfoley: yes Dec 14 01:32:56 I got it right in OE tree dir Dec 14 01:33:05 seems current working dir Dec 14 01:33:20 Jay7: you using the current bitbake master ? Dec 14 01:33:24 yes Dec 14 01:34:03 hm.. refa5485c8f looks like fix Dec 14 01:34:04 me too Dec 14 01:34:18 dfoley: try to git pull Dec 14 01:34:40 I've pulled before that commit Dec 14 01:35:13 I mean pull bitbake from git Dec 14 01:35:30 Jay7: I think I'm still getting the problem. I'll have to confirm 100% though Dec 14 01:38:41 khem: whitespace again: http://pastebin.ca/2019001 Dec 14 01:39:22 and looks like samba still failing.. Dec 14 01:41:22 khem: http://tinderbox.openembedded.net/packages/1270999/ Dec 14 01:41:35 but better to wait results from ant__ Dec 14 01:45:05 well.. restarting testbuilder again Dec 14 01:45:39 * Jay7 -> sleep Dec 14 01:48:22 Jay7: mtd-utils still puts the object files in the wrong place Dec 14 02:40:17 kergoth_, efa5485c8f9d488ba058c40734cc55296d3de6eb does not fix the PWD problem Dec 14 02:40:29 run.do_compile still exports the wrong PWD Dec 14 02:53:27 I'm trying to build some images for Chumby hardware, and I'm getting an error when I try to do "bitbake console-image": "UnboundLocalError: local variable 'cooker' referenced before assignment". Anyone else encountered this/have a fix? Dec 14 02:56:39 cjameshuff: are you using bitbake 1.10? I believe that was fixed by commit 1074f9a5c529e364404daf545b8465415d63a556 on Nov 19th. Dec 14 02:56:43 Ubuntu 10.4, BitBake 1.10.1, OE from Git... Dec 14 02:57:58 cjameshuff: are you using bitbake from git? Dec 14 02:58:25 Nope, the last release...which was in October, so lacks that fix. I'll grab it from Git. Dec 14 02:59:41 if you like bleeding edge, bitbake master has parallel recipe parsing, makes parsing _much_ faster now **** ENDING LOGGING AT Tue Dec 14 02:59:57 2010