**** BEGIN LOGGING AT Fri Dec 17 02:59:59 2010 Dec 17 03:17:20 kergoth, i get a segfault building kexec-tools-klibc, building on x86_64 for target qemumipsel distro minimal Dec 17 03:17:33 its the klcc process that dies Dec 17 03:18:04 but i didn't manage to reproduce outside of bitbaking the recipe Dec 17 03:18:19 havent looked into it... other stuff came up Dec 17 03:40:56 k, will try that machine Dec 17 06:06:20 03Khem Raj  07master * re0fc150cb1 10openembedded.git/classes/libc-package.bbclass: Dec 17 06:06:20 libc-package.bbclass: Add missing TARGET_ARCH to locale_arch_options Dec 17 06:06:20 Signed-off-by: Khem Raj Dec 17 06:18:55 khem, no problems with your updates for gcc and eglibc so far. tested minimal (x11-image, native-sdk-image, meta-toolchain) and micro (micro-image, native-sdk-image, meta-toolchain) Dec 17 06:19:30 grg: cool Dec 17 06:19:44 grg: just fixed an issue with locale generation Dec 17 06:19:48 do u use locales? Dec 17 06:20:02 hmm... not as such Dec 17 06:20:05 ok Dec 17 06:20:10 they still get built Dec 17 06:20:19 i just dont pay any attention to them Dec 17 06:20:20 few of them prolly Dec 17 06:20:25 ok Dec 17 06:32:11 kergoth_: same segfault with shr/spitz also on x86_64 Dec 17 06:34:28 gm Dec 17 06:39:04 ~gm Dec 17 06:39:04 gm is, like, Gambia Dec 17 06:41:51 khem, i still have issues with gettext, I'm suspecting libtool 2.4 I saw that these two lines were removed in 2.2.6 but not in 2.4 Dec 17 06:41:51 - # We cannot seem to hardcode it, guess we'll fake it. Dec 17 06:41:51 - add_dir="-L$libdir" Dec 17 06:41:51 I think these are causing that -L/usr/lib is used/injected/whatever libtool is run (btw thi is in libltdl/config/ltmain.m4sh) Dec 17 06:41:57 care to comment about this ? Dec 17 06:51:48 eFfeM_work: I dont see this issue Dec 17 06:51:57 eFfeM_work: I wonder why you see it Dec 17 06:52:03 tell me your distro/machine Dec 17 06:52:10 khem, you do not get the -L messages ? Dec 17 06:52:38 khem, one sec Dec 17 06:53:53 sheevaplug minimal james-image Dec 17 06:54:05 http://tinderbox.openembedded.net/builds/111300/ Dec 17 06:55:21 this is on our autobuilder (ubuntu 10.04 64 bit under vmsphere or so), sometimes @ home it will work ok Dec 17 06:55:36 and when libtool is processing gettext it adds a -L/usr/lib Dec 17 06:57:46 khem, here's the mailing list post with the log: http://thread.gmane.org/gmane.comp.handhelds.openembedded/40487/focus=40510, this one was for beagleboard minimal james-image Dec 17 06:58:18 let me pastebin for easier reference Dec 17 06:58:52 http://www.pastebin.ca/2021866 Dec 17 06:59:35 I think gettext need m4 macro updates Dec 17 07:00:01 Does it have its own libtool ? Dec 17 07:00:04 if yes then delete it Dec 17 07:00:14 lines are broken while pasting around, line 6 is the start of the libtool invocation, line 31 is generated by libtool, line 56 has -L/usr/lib , then line 63 and up complain Dec 17 07:00:17 khem, will check Dec 17 07:01:29 need a few mins for that Dec 17 07:05:15 khem, it indeed has ./m4/libtool.m4 Dec 17 07:05:15 ./gettext-tools/examples/hello-c++-kde/admin/libtool.m4.in Dec 17 07:06:38 should these be nuked? Dec 17 07:07:49 well you have to make sure you do the right thing Dec 17 07:08:14 what all is there under m4/ Dec 17 07:09:51 khem, frankly speaking i have hardly a clue when it comes to things like libtool and m4 Dec 17 07:10:09 ah misread, guess that was a Q Dec 17 07:10:24 fixautomake.m4 init-package-version.m4 libtool.m4 lt~obsolete.m4 ltoptions.m4 ltsugar.m4 ltversion.m4 woe32-dll.m4 Dec 17 07:14:40 good morning Dec 17 07:14:59 hi ericben Dec 17 07:15:07 hi eFfeM_work Dec 17 07:15:58 khem: build launched with your gcc updates. Dec 17 07:16:05 mailing lists seems not working Dec 17 07:36:01 ah I was surprised that now new e-mail arrived in ML :) Dec 17 07:37:09 * JaMa|Wrk has now qemu-arm segfault while building emacs-x11 even with 12.5 version :/ Dec 17 07:37:24 someone else seen this ^^ ? Dec 17 07:53:27 good morning Dec 17 07:56:01 eFfeM_work: delete lt~obsolete.m4 ltoptions.m4 ltsugar.m4 ltversion.m4 libtool.m4 Dec 17 07:56:23 and make sure that autoreconf is run Dec 17 07:56:31 during do_configure Dec 17 07:56:35 ericben: ok Dec 17 07:56:56 I dont know where ml is hosted Dec 17 07:57:06 we should prolly move it over ? Dec 17 07:57:12 khem, will do & test (i think I have seen other recipes doing this, will copy, try, then if ok submit patch for ack Dec 17 07:57:20 ML is on linuxtogo Dec 17 07:57:34 would it make sense to host it on openembedded.org ? Dec 17 07:58:01 it should be straight forward but we need archive and list info Dec 17 07:58:09 I am setup mailman Dec 17 07:58:11 no idea, guess you should ask ka6sox Dec 17 07:58:16 s/am/can/ Dec 17 07:58:37 who admins linuxtogo Dec 17 07:59:47 no idea, but I am pretty sure ka6sox will know that Dec 17 08:00:43 he does not admin ltg Dec 17 08:00:51 that much is sure Dec 17 08:01:05 no but i think he'll know the contact Dec 17 08:01:24 or contact the admin, address is at the subscription page Dec 17 08:02:12 iirc florian admins ltg Dec 17 08:02:48 and seems that also pipermail doesn't show new e-mails (I was hoping that only distribution to subscribers was off) Dec 17 08:03:01 flo_lap: ping Dec 17 08:03:36 hmm Dec 17 08:05:00 we should host the ml on say lists.openembedded.org Dec 17 08:10:55 khem: see http://lists.linuxtogo.org/pipermail/openembedded-devel/ Dec 17 08:11:09 ah I didn't read it right, sorry Dec 17 08:15:24 midnight Dec 17 09:05:25 hm, gettext keeps on amazing me; apparently it is not only libtool. if I build it from scratch for sheevaplug/minimal it builds fine and in temp I have only run.base_do_fetch.20348 run.do_build.5576 run.do_distribute_sources.20348 run.do_fetch.20348 run.do_package_stage_all.5560 run.SRC_DISTRIBUTECOMMAND.20348 Dec 17 09:05:25 (I do have rm_work enabled), I do a bitbake -cclean, then bake again and it fails in do_install, but the first time apparently there was no do_install Dec 17 09:08:51 fontconfig-2.8.0 fetch is broken : http://fontconfig.org/release/fontconfig-2.8.0.tar.gz Dec 17 09:08:58 http://tinderbox.openembedded.net/builders/eFfeM/, the build from 07.54.09 succeeds, does a do_build, do_package_stage_all and do_distribute_sources then is done, the build from 08.52.53 is the clean, and the build from 08.52.51 fails in do_install, it tries to do completely different tasks Dec 17 09:09:01 * eFfeM_work is flabbergasted Dec 17 09:29:44 morning all Dec 17 09:38:58 florian: hi, mailing list server seems to not relay the mails Dec 17 09:39:38 ericben: right, just noticed that its responsiveness seems to be limited too :-( Dec 17 10:12:48 can someone try to build emacs-x11? here it segfaults http://paste.pocoo.org/show/305952/ Dec 17 10:15:53 JaMa|Wrk: try with qemu build by OE Dec 17 10:16:04 seems you are calling your host's qemu Dec 17 10:16:29 no I'm not calling host's qemu Dec 17 10:23:40 and with OE from Oct24 it was building OK Dec 17 10:31:01 JaMa|Wrk: I can try to build for minimal/akita a bit later Dec 17 10:31:11 not it is building console-image Dec 17 10:38:12 thanks Dec 17 10:44:11 03Koen Kooi  07org.openembedded.dev * r995b148884 10openembedded.git/recipes/ruby/ (5 files in 2 dirs): Dec 17 10:44:11 ruby 1.8.7: fix socket extension Dec 17 10:44:11 The EXTRA_OECONF trick doesn't work for 1.8.7 Dec 17 10:44:11 Signed-off-by: Koen Kooi Dec 17 10:44:19 03Koen Kooi  07org.openembedded.dev * rd2e4a5141d 10openembedded.git/recipes/ruby/ruby-dbus_0.2.1.bb: Dec 17 10:44:19 ruby-dbus: fix libdir reference Dec 17 10:44:19 Signed-off-by: Koen Kooi Dec 17 10:45:20 build failed :( Dec 17 10:45:23 ERROR: fontconfig-2.8.0: http://fontconfig.org/release/fontconfig-2.8.0.tar.gz cannot check archive integrity Dec 17 10:45:31 http://pastebin.com/raw.php?i=5GbeRn9j Dec 17 10:53:58 03Koen Kooi  07org.openembedded.dev * r4d10e777f2 10openembedded.git/recipes/ruby/ruby/0001-socket-extconf-hardcode-wide-getaddr-info-test-outco.patch: Dec 17 10:53:58 ruby: add missing patch Dec 17 10:53:58 Signed-off-by: Koen Kooi Dec 17 10:59:45 hi woglinde Dec 17 11:01:18 he effem Dec 17 11:09:35 concerning fontconfig fetch problem : all versions seems affected, we can use http://cgit.freedesktop.org/fontconfig/snapshot/ to get the archives : is this an acceptable solution ? Dec 17 11:10:13 Hi all , Dec 17 11:10:49 I am getting this error while building the kernel for am3517-evm Dec 17 11:22:24 hi Dec 17 11:22:45 i want to use a specific kernel headers with a specific receipe Dec 17 11:22:53 can anyone tell me how to do that Dec 17 11:22:54 ? Dec 17 11:23:38 i tried to run "make KRNLSRC=/home/tapas/working_camera_copy/integretion-kernel-omap4" inside do_compile() Dec 17 11:23:45 but it didn't work for me Dec 17 11:28:14 ericben: what about http://fontconfig.sourcearchive.com/ ? Dec 17 11:28:47 hmm but there is with "orig" postfix Dec 17 11:29:36 hopefully they'll move it to http://xorg.freedesktop.org/releases/individual/ soon Dec 17 11:29:38 JaMa|Wrk: also possible but in that case we can also use GENTOO_MIRRORS or others Dec 17 11:30:15 JaMa|Wrk: my goal was to keep the archive download at the origin of the software Dec 17 11:32:30 ericben: fontconfig-ML seems pretty active about moving wiki page and stuff so maybe they'll update download location if someone reports it Dec 17 11:36:36 * JaMa|Wrk subscribing to that ML Dec 17 11:37:32 JaMa|Wrk: I sent an email to them this morning Dec 17 11:37:58 I don't see it on ml Dec 17 11:38:10 JaMa|Wrk: must be in moderation Dec 17 11:38:21 but OK, I wanted to reply on http://lists.freedesktop.org/archives/fontconfig/2010-December/003532.html Dec 17 11:39:06 but I'll do something else and let you to handle it ;) Dec 17 11:39:16 JaMa|Wrk: if you are subscribed to the list go ! Dec 17 11:39:26 re Dec 17 11:39:27 as I didn't subscribed to it Dec 17 11:40:23 one more fun thing : md5sum are different in our recipe vs gentoo archived files and vs the archive I get through their git Dec 17 11:40:32 ericben: ok Dec 17 11:40:58 I have original tarball if you want it for diff Dec 17 11:41:19 03Bernhard Reutner-Fischer  07master * r0ef8a9b378 10bitbake.git/lib/bb/fetch/bzr.py: Dec 17 11:41:19 bzr: use utils.remove instead of os.system Dec 17 11:41:19 Signed-off-by: Bernhard Reutner-Fischer Dec 17 11:41:25 03Bernhard Reutner-Fischer  07master * r6dc863f714 10bitbake.git/lib/bb/utils.py: Dec 17 11:41:25 utils: don't overwrite builtin dir() function Dec 17 11:41:25 Signed-off-by: Bernhard Reutner-Fischer Dec 17 11:41:32 03Bernhard Reutner-Fischer  07master * rc84a22b7a3 10bitbake.git/lib/bb/build.py: Dec 17 11:41:32 build: use utils.remove instead of os.system Dec 17 11:41:33 Signed-off-by: Bernhard Reutner-Fischer Dec 17 11:41:41 ericben: snapshot from git will be probably different as it probably doesn't create with make dist Dec 17 11:42:01 JaMa|Wrk: I also have one original file here Dec 17 11:42:53 and gentoo's md5sum is correct, that was my mistake Dec 17 11:42:53 JaMa|Wrk: no good, gettext is failing before emacs-x11 Dec 17 11:43:47 and kexec-tools-klibc is failing before gettext Dec 17 11:43:56 I'll try with -k Dec 17 11:45:00 RP, ping? Dec 17 11:46:01 Jay7: still that segfault from klcc right? Dec 17 11:46:03 JaMa|Wrk, how's the policy to fix commentary typos in oe? Do i need approval or can i just check them in? Dec 17 11:46:25 blindvt which region? Dec 17 11:46:31 JaMa|Wrk: yes Dec 17 11:46:43 blindvt`: I would say just check them in Dec 17 11:46:49 woglinde, help-text of the stage-manager, for example Dec 17 11:46:57 thats okay Dec 17 11:47:00 woglinde: hey Dec 17 11:47:00 check it in Dec 17 11:47:02 he zecke Dec 17 11:47:06 RP, never mind Dec 17 11:47:38 03Bernhard Reutner-Fischer  07master * r7d2bde76af 10openembedded.git/recipes/e2fsprogs/ (3 files in 2 dirs): Dec 17 11:47:39 e2fsprogs: add 1.41.12, remove old versions Dec 17 11:47:39 Signed-off-by: Bernhard Reutner-Fischer Dec 17 11:47:41 03Bernhard Reutner-Fischer  07master * r23e0cd2239 10openembedded.git/conf/ (bitbake.conf local.conf.sample): Dec 17 11:47:41 add ext4 image support Dec 17 11:47:41 Signed-off-by: Bernhard Reutner-Fischer Dec 17 11:47:47 Jay7: I've reproduced same qemu-arm segfault on another host Dec 17 11:47:52 03Bernhard Reutner-Fischer  07master * r2c0f30ae06 10openembedded.git/recipes/e2fsprogs/e2fsprogs.inc: Dec 17 11:47:52 e2fsprogs: fix alternatives of non-packaged native sysroot Dec 17 11:47:52 Signed-off-by: Bernhard Reutner-Fischer Dec 17 11:49:25 is something like "fsck=$(cd ${D}${base_sbindir}/" allowed in ie dash? Dec 17 11:49:37 03Bernhard Reutner-Fischer  07master * r0864e37e9b 10openembedded.git/recipes/stage-manager/files/stage-manager: Dec 17 11:49:37 stage-manager: fix typo in help-text Dec 17 11:49:37 Signed-off-by: Bernhard Reutner-Fischer Dec 17 11:49:44 JaMa|Wrk: why not Dec 17 11:50:10 Jay7: ok.. I wasn't sure if it isn't bash-only Dec 17 11:50:13 ah.. Dec 17 11:50:33 what is full construction? Dec 17 11:50:54 as I use only bash, so I'm not very good at spotting bashisms :) Dec 17 11:50:55 fsck=$(cd ${D}${base_sbindir}/ && ls -1 fsck.ext* | sed s/[^\.]*//) Dec 17 11:51:01 JaMa|Wrk, works fine with busybox's ash so it's ok as far as i'm concerned ;) Dec 17 11:51:40 JaMa|Wrk: it should work Dec 17 11:52:01 ok, fair enough, sorry for noise then Dec 17 11:52:21 jama is a dirty hack anyway Dec 17 11:52:42 ericben: sent http://lists.freedesktop.org/archives/fontconfig/2010-December/003535.html Dec 17 11:52:43 it is Dec 17 11:52:44 args Dec 17 11:52:48 jama is no hack Dec 17 11:52:56 why not? Dec 17 11:52:59 :) Dec 17 11:53:52 changing coffee to recipe metadata looks like hack to me Dec 17 11:54:05 kergoth_: can you ack gitpkgv change? Dec 17 11:54:13 kergoth_: I'd like to push it Dec 17 11:57:23 otavio: did you see my ack? sent it on ML but I don't see it on web archive Dec 17 11:57:38 so no idea if it arrived to todays ltg Dec 17 11:58:46 JaMa|Wrk: yes; i got it Dec 17 11:58:52 JaMa|Wrk: I ammended it already Dec 17 12:05:18 otavio: I mean todays reply http://paste.pocoo.org/show/305983/ Dec 17 12:21:24 hmm, no OE email in my inbox Dec 17 12:21:31 is the mail server dead? Dec 17 12:21:48 maybee Dec 17 12:22:12 cbrake: will there be a new testing-next branch? expected it yesterday ;-) Dec 17 12:44:56 | /usr/lib/libdl.so: file not recognized: File format not recognized Dec 17 12:45:07 ieeks Dec 17 12:45:16 hm.. something new.. before it was libgcc iirc Dec 17 12:45:32 Jay7: iirc I had this with mpd Dec 17 12:45:55 http://tinderbox.openembedded.net/public/logs/task/14190934.txt Dec 17 12:45:59 libungif Dec 17 12:46:08 Jay7: for long.. probably since switching to libtool-2.4 Dec 17 12:46:42 who the hell switched to 2.4 Dec 17 12:46:43 seems libungif have some lost dependency Dec 17 12:46:49 *g* Dec 17 12:47:10 I definitely should run lost dependency check Dec 17 12:47:19 for every package Dec 17 12:48:26 eFfeM_work: seems like mpd_0.15.12 still builds fine for you, which libtool version were you using for those builds on http://tinderbox.openembedded.org/packages/mpd/?page=1 Dec 17 12:48:41 Jay7: did you rebuild from scratch after switching libtool? Dec 17 12:48:48 JaMa|Wrk: this is clean build Dec 17 12:49:14 I'll rerun with bb 1.10 Dec 17 12:49:24 JaMa|Wrk: I think this was with libtool 2.4 Dec 17 12:49:43 it will allow to build kexec-tools-klibc at least Dec 17 12:50:06 eFfeM_work: and SYSROOT support enabled? Dec 17 12:50:27 omg.. bb w/o parallel parsing is sooooo slooooow..... Dec 17 12:50:39 jay7 yes Dec 17 12:50:39 not explicitly I think actually no idea how to do that, I think I just pinned it to 2.4 in my local.conf Dec 17 12:50:52 effem that was a joke Dec 17 12:51:08 could also be the cause of my gettext issue, Dec 17 12:51:13 koen already pinned it for angstroem Dec 17 12:51:20 woglinde: I'm using minimal Dec 17 12:51:33 eFfeM_work: you have to explicitily say LIBTOOL_HAS_SYSROOT = "yes" Dec 17 12:51:54 ah ok, will put that in my failing gettext build Dec 17 12:53:01 jama whats LIBTOOL_HAS_SYSROOT exactly doing? Dec 17 12:53:06 restarted on autobuilder, this will also build mpd Dec 17 12:53:19 eFfeM_work: http://gitorious.org/~jama/angstrom/jama-shr-experimental/commit/62f3abe8a5ff1294b181ae42b26ee67396abdef6 Dec 17 12:54:09 JaMa|Wrk: do this fixes your gettext build? Dec 17 12:54:18 or you was not affected? Dec 17 12:54:35 woglinde: allows libtool to use sysroot feature instead of autotools_prepackage_lamangler (grep bbclasses for details) Dec 17 12:54:41 Jay7: not affected Dec 17 12:54:47 hm.. Dec 17 12:55:55 seems only I and eFfeM_work was affected by gettext Dec 17 12:56:09 I see no other complains Dec 17 12:56:31 that could be because of ML is dead .. Dec 17 12:56:46 it happens about 2-3 days ago Dec 17 12:58:28 P_V for libtool was changed 2010-12-11 Dec 17 13:00:18 jay7 scroll back in the log, I had two successive builds, first one succeeds, 2nd one fails Dec 17 13:01:00 what I did (clean build, sheevaplug, minimal): bitbake gettext (works); bitbake -cclean gettext; bitbake gettext (fails) Dec 17 13:01:24 and the 2nd bitbake is doing different things first one does not run do_install, 2nd one does ! Dec 17 13:02:21 awesome Dec 17 13:09:45 florian, ping Dec 17 13:10:02 Crofton|work: i know Dec 17 13:10:06 ok Dec 17 13:10:25 I finished reading email way to soon this morning Dec 17 13:10:37 heh Dec 17 13:15:49 hi people i think this is not place to ask question like that but i think this channel people is pro coders so maybe someone have idea , please apoligise first for the question.... Dec 17 13:15:50 hi , i m having problem on win32 keyboard scan codes , i want to translate a scancode to WCHAR . I've tried to do it by utilizing function MapVirtualKeyExW but on some characters it gives error , returning 0 , some chars like 'i' 'ç' and 'ğ' , how can i solve pls help ? Dec 17 13:17:18 JDuke128: really wrong channel Dec 17 13:17:42 ya :) Dec 17 13:46:35 eFfeM_work: clean build of gettext (minimal/efikamx) failed with the same: /usr/lib/libc.a: could not read symbols: File format not recognized Dec 17 13:46:57 what machine/distro/host Dec 17 13:47:07 DISTRO=minimal MACHINE=efikamx bitbake gettext Dec 17 13:47:20 debian testing x86_64 Dec 17 13:47:57 strange, Dec 17 13:48:04 bb 1.10 Dec 17 13:48:13 what libtool Dec 17 13:53:06 2.4 Dec 17 14:09:42 here is some more info about qemu-arm segfault http://paste.pocoo.org/show/306007/ Dec 17 14:10:56 Jay7: do you use SYSROOT support as mentioned above by JaMa ? Dec 17 14:11:12 no, just building as is Dec 17 14:12:49 Jay7: JaMa|Wrk, khem I just added the LIBTOOL_HAS_SYSROOT = "yes" to my local.conf and now gettext seem to build fine for me Dec 17 14:12:56 interesting post : http://zecke.blogspot.com/2010/12/in-name-of-performance.html Dec 17 14:12:57 doing some more tests Dec 17 14:13:21 ericben: hehe. Dec 17 14:13:58 eFfeM_work: I'll try this too Dec 17 14:15:48 eFfeM_work: I think that almost nobody tried new libtool-2.4 actually without LIBTOOL_HAS_SYSROOT enabled so that could be the reason why sofar only you and Jay7 reported so much issues Dec 17 14:15:53 s/much/many Dec 17 14:16:24 JaMa|Wrk: we are just using distros as is :) Dec 17 14:16:38 zecke nice post, ericben, thanks for the link, I've optimised quite some stuff in the past and always made sure to bench things Dec 17 14:16:43 see e.g. http://git.kernel.org/?p=linux/kernel/git/next/linux-next.git;a=blob_plain;f=Documentation/mtd/nand_ecc.txt;hb=HEAD Dec 17 14:17:12 and actually on hindsight I'm not too happy with that bench as it was done on my dev sys and the results might not hold at all on e.g. arm or ppc Dec 17 14:17:13 Jay7: yup that patch with hard instead of weak assignment should be in much sooner ;/ Dec 17 14:17:15 eFfeM_work: I saw your nand_ecc.txt last week while searching for info on nand ecc Dec 17 14:18:09 JaMa|Wrk: could be that this is due the sysroot missing, I think I went to 2.4 quite early when there were still issues with sysroot support for some recipes Dec 17 14:19:12 eFfeM_work: with all shr dependencies including task-shr-feed I know currently about mpd and openjdk failing Dec 17 14:19:15 Jay7: I upgraded to testbuilder 0.3 : nice work the new config file is really far easier to understand Dec 17 14:19:36 JaMa|Wrk: guess that might be why mpd builds for me Dec 17 14:19:51 ericben: nice to hear :) Dec 17 14:19:52 eFfeM_work: there is mpd 0.16.0 release so maybe just upgrading recipe could be enough Dec 17 14:20:16 So, anyone around that can help w/ the ML? Dec 17 14:20:17 eFfeM_work: nice! Dec 17 14:20:24 eFfeM_work: and woglinde was already looking on issue with openjdk Dec 17 14:20:34 Tartarus: #linuxtogo :) Dec 17 14:20:43 but florian is already informed Dec 17 14:21:42 zecke: can't give the new mpd a stab now, but trying the current recipe on the autobuilder Dec 17 14:21:51 will try 0.16.0 sat or sun Dec 17 14:27:16 Jay7, thanks Dec 17 14:32:43 Anyone want to review https://github.com/trini/openembedded/tree/external-toolchain-updates while the ML is down? Dec 17 14:33:50 Hi Dec 17 14:33:58 when i building for am3517-evm kernel i am getting error like make: arm-angstrom-linux-gnueabi-gcc: Command not found Dec 17 14:34:21 Please help me in fixing this error Dec 17 14:44:04 morning Dec 17 14:45:31 morning Dec 17 14:45:39 gm Dec 17 14:45:48 kergoth_: will have UI stuff ready shortly Dec 17 14:45:51 nice Dec 17 14:57:44 03Eric Bénard  07org.openembedded.dev * re8853efcc8 10openembedded.git/recipes/fontconfig/ (fontconfig_2.6.0.bb fontconfig_2.8.0.bb): Dec 17 14:57:44 fontconfig: update SRC_URI Dec 17 14:57:44 fontconfig's website's configuration actually prevents to fetch the Dec 17 14:57:44 source archives so use gentoo's mirror until the website is fixed. Dec 17 14:57:44 Signed-off-by: Eric Bénard Dec 17 15:08:37 were there any objections to the variable typing stuff going in? given its entirely opt-in, the only downside is the extra flags placed in bitbake.conf and base.bbclass Dec 17 15:09:25 presumably no, but the ML has been down :( Dec 17 15:09:38 ah has it? lovely Dec 17 15:10:41 huh, yeah, guess my last mail from the list was 23 hours ago or so Dec 17 15:13:41 eFfeM_work, JaMa|Wrk: yes, that LIBTOOL_HAS_SYSROOT seems helped me Dec 17 15:19:14 cool digikey just shipped my pandaboard :) Dec 17 15:19:22 nice Dec 17 15:19:55 yes, yesterday it was planned for 06/01/2011, last week for 19/12/2010 and it ships today :-) Dec 17 15:28:17 meh Dec 17 15:29:33 * broonie just cloned OE in ~10 minutes. Dec 17 15:29:59 Rather than the 10+ hours it took to do the same thing with mtn :) Dec 17 15:30:30 * bluelightning remembers those days also (but not with fondness) Dec 17 15:30:39 kergoth_: https://github.com/foerster/bitbake/commit/8cb5900df6a2fbbdeab2e6bd96bb3aca4e1b2fc5 Dec 17 15:30:49 I had "mtn pull" in cron at 1:23 daily Dec 17 15:31:21 and has it been working?? :D Dec 17 15:31:37 mwester: it was Dec 17 15:32:35 Whenever I travelled and was away from the build machine for a week, I would pull down a new mtn snapshot rather than pull; it was faster... Dec 17 15:33:17 in other news sanity.conf really is shit Dec 17 15:33:55 Recent changes to it, you mean? I find the sanity checker quite convenient (perhaps because I am frequently insane, or rather do insane things) Dec 17 15:35:36 foerster: looks good, will push to upstream master now Dec 17 15:36:46 kergoth_: sweet. Still stuff can be fixed/added/updated/etc, but they're at least (mostly) functional now. Dec 17 15:36:53 * kergoth_ nods Dec 17 15:37:19 03Bob Foerster  07master * rb947e7aa40 10bitbake.git/ (7 files in 3 dirs): (log message trimmed) Dec 17 15:37:19 Resurrect alternative UIs Dec 17 15:37:19 The various alternative UIs have been updated to once again be functional Dec 17 15:37:19 with the latest bitbake internals. Each of the UIs still have much room for Dec 17 15:37:19 functional improvement. Dec 17 15:37:19 In particular, they have been updated to: Dec 17 15:37:19 - interact with the new process based server Dec 17 15:38:25 you snook in a feature with your fix commit :-p Dec 17 15:39:06 ? Dec 17 15:39:26 the pastebin feature is in that diff but not documented in the commit Dec 17 15:39:34 * kergoth_ points at foerster Dec 17 15:39:52 evil monkey style? Dec 17 15:39:57 exactly Dec 17 15:40:16 there's actually other improvements in that too, the coloring by task fail/success Dec 17 15:40:23 * kergoth_ should have thought to add that to the message Dec 17 15:40:44 * incandescant prefers more granular commits Dec 17 15:40:56 but not everyone agrees :-) Dec 17 15:41:29 after a git checkout -b release-2010.12 origin/release-2010.12 do I need to pull ? Dec 17 15:42:07 I like granular commits for functionality. the main reason i rebase -i before merging is to merge fixups back with the commit that implemented the feature and caused the bugs the fixups fix :) Dec 17 15:42:31 to be more clear, i think the best reason to use it is to avoid states in the history where unrelated breakage will hose someone's bisect Dec 17 15:42:56 * JaMa|Wrk nods Dec 17 15:43:05 need to use git test-sequence more Dec 17 15:44:30 do we still have origin/release-2010.12 branch? Dec 17 15:44:50 * JaMa|Wrk looking on gitweb and no Dec 17 15:44:59 with folks awake, anyone want to review... Dec 17 15:45:00 do I have to consider [new tag] tested_2010-12-10 == branch origin/release-2010.12 ? Dec 17 15:45:15 https://github.com/trini/openembedded/tree/external-toolchain-updates Dec 17 15:45:25 mckoan: git checkout -b release-2010.12 release-2010.12 Dec 17 15:45:26 or the milestone is simply the tag tested_2010-12-10 ? Dec 17 15:45:31 mckoan: http://www.openembedded.org/index.php/Release-2010.12 Dec 17 15:45:43 mckoan: release-2010.12 is the release tag Dec 17 15:45:55 Jay7: thx fine Dec 17 15:47:47 Jay7: is exactly what I did Dec 17 15:48:17 Jay7: my doubt is why I get a long msg if I do git pull after git checkout -b release-2010.12 release-2010.12 Dec 17 15:48:49 mckoan: git should complain about this Dec 17 15:48:53 http://pastebin.com/yubHkyCi Dec 17 15:49:24 I understand may be a problem with my git knowledge Dec 17 15:49:38 I am back Dec 17 15:49:39 mckoan: because git warns you that he won't pull the changes in this branch Dec 17 15:49:48 mckoan: git checkout master && git pull :) Dec 17 15:49:57 then git checkout release-xxxx Dec 17 15:50:00 Jay7: but pull is needed? Dec 17 15:50:10 this tag is not moving Dec 17 15:50:12 mckoan: if you checkout tag, then that local branch is not tracking any remote branch Dec 17 15:50:19 so you don't needed to fetch changes Dec 17 15:50:21 or release-2010.12 will be frozen forever? Dec 17 15:50:28 yes, afaik Dec 17 15:50:32 unless you add -t to git checkout Dec 17 15:56:29 maybe I didn't understand how the release method works Dec 17 15:57:16 release-2010.12 is not updated to master patches sent by 31/NOV Dec 17 15:57:25 * foerster should have split the UI commit up a bit more, but is time constrained with work that pays the bills :) Dec 17 15:57:44 where the release-2010.12 version comes from? Dec 17 15:58:52 mckoan: it was branch for stabilizing and after tagging it, the branch was removed Dec 17 15:58:53 for example, my last patch to master was on Mon Nov 29, so I expected to see it into release Dec 17 15:59:40 JaMa|Wrk: so I needed to patch into the branch and not into master? Dec 17 15:59:51 03Eric Bénard  07org.openembedded.dev * rb116445f60 10openembedded.git/recipes/busybox/ (2 files in 2 dirs): Dec 17 15:59:51 busybox 1.18.0 : add last fixes Dec 17 15:59:51 from : http://busybox.net/downloads/fixes-1.18.0/ Dec 17 15:59:51 Signed-off-by: Eric Bénard Dec 17 16:01:08 http://tinderbox.openembedded.net/packages/1326130/ Dec 17 16:01:19 e2fsprogs failed Dec 17 16:01:46 mckoan: if you wanted that patch in relase you should reply to http://lists.linuxtogo.org/pipermail/openembedded-devel/2010-November/027401.html or merge it yourself Dec 17 16:02:39 mckoan: and release branch was branched Nov 18 http://lists.linuxtogo.org/pipermail/openembedded-devel/2010-November/026975.html Dec 17 16:02:59 mckoan: since then only bugfixes were cherry-picked there on individual basis Dec 17 16:05:25 that is why I'll prefer having release tag inside release branch.. Dec 17 16:05:44 is ml working now ? Dec 17 16:07:00 khem: doesn't seems to work Dec 17 16:08:38 JaMa|Wrk: yes I read such msg from khem, but I thought that if I had pushed my patch to master it was included into release Dec 17 16:09:06 where do I find an explanation of relases generation policy? Dec 17 16:09:56 mckoan: Did you ask to include it in release ? if not then I probably missed it Dec 17 16:10:05 hey khem Dec 17 16:10:20 https://github.com/trini/openembedded/tree/external-toolchain-updates please :) Dec 17 16:10:20 mckoan: I send several messaged to ml askin for same Dec 17 16:13:07 bb.data.getVar('TOOLCHAIN_BRAND',d,1) in "csl": Dec 17 16:13:14 s/in/is/ ? Dec 17 16:14:25 khem: any hint for this qemu-arm segfault http://paste.pocoo.org/show/306007/ ? the same for 0.13.0 Dec 17 16:15:44 Tartarus: it should be ok to rename eglibc.inc to eglibc-internal.inc I think Dec 17 16:16:29 khem: lemme double check, I thought there were some direct includers of ${LIBC}.inc Dec 17 16:18:17 yeah, gumstix and a few others Dec 17 16:19:29 JaMa|Wrk: is qemu launching threads ? Dec 17 16:19:47 how can I check? Dec 17 16:19:55 ugh Dec 17 16:19:57 without reading whole strace :) Dec 17 16:20:06 khem, JaMa|Wrk : how couls we improve this policy? Maybe creating a rule and documenting it rather trusting that everyone has understood what is not so obvious. Dec 17 16:20:24 is not a criticism, is a proposal ;-) Dec 17 16:20:26 foerster: if the commandlineAction is empty, the UI exits without emptying the events from the queue, and we dont' see the log messages that indicate why the action was None Dec 17 16:20:29 * kergoth_ rolls eyes Dec 17 16:21:05 mckoan: pay attention to mailing lists regarding releases, I explicitly asked to mark messages [release xxxxx] in subject line Dec 17 16:21:09 etc. Dec 17 16:21:11 kergoth_: bah Dec 17 16:21:33 JaMa|Wrk: do .12.5 and .13 fail in same way Dec 17 16:21:34 khem: ok, I'll re-read them carefully Dec 17 16:22:00 khem: yes exactly the same Dec 17 16:22:13 * mckoan is still using master instead of switching to release-2010.12 :-( Dec 17 16:22:16 khem: and no pthread_create in strace log Dec 17 16:22:38 mckoan: you can switch and merge all your changes into your local branch Dec 17 16:22:41 * mckoan wonders what a mess will happen when yocto will be here ;-) Dec 17 16:22:55 foerster: as a workaround, i'm going to have shutdown() redirect any LogRecords left in the queue, but we'll have to teach the UIs to show the user everything that's queued up before they exit Dec 17 16:23:03 heh Dec 17 16:23:21 * JaMa|Wrk building pygobject to try goggle bitbake UI :) Dec 17 16:23:26 JaMa|Wrk: :D Dec 17 16:23:54 notify me when it will allow to connect to bitbake via tcp/ip :) Dec 17 16:25:13 Jay7: why not bitbake listening on irc channel :) Dec 17 16:25:34 khem: 'is' wouldn't be right in the above either, == is correct Dec 17 16:25:40 #oe-monster Dec 17 16:25:41 Jay7: we'll get there. Dec 17 16:26:12 Jay7: in the very near future we're creating a daemonize UI to spawn off a server with an xmlrpc socket to wait for connections, and displays the connection info, for use by non-python UIs Dec 17 16:26:14 kergoth_: yeah from type perspective correct Dec 17 16:26:23 if that listens on the ethernet interface.. Dec 17 16:26:38 khem: considering you can't really rely on 'is' for a number of things in python Dec 17 16:26:51 iirc 5 is 5 was false, or something, at least for really high numbers Dec 17 16:26:51 JaMa|Wrk: please commit that LIBTOOL fix Dec 17 16:26:59 ok Dec 17 16:27:00 libtool fix? Dec 17 16:27:22 poor eFfeM_work has been sufferring from it for so long :) Dec 17 16:27:36 khem: can I add your ack? Dec 17 16:27:40 jama found it? Dec 17 16:27:43 finally Dec 17 16:27:53 JaMa|Wrk: yes, pastebin it before you commit Dec 17 16:28:22 JaMa|Wrk: I was wondering if we could use cross-localedef for emacs too Dec 17 16:28:23 khem: this is it 13:52:19 < JaMa|Wrk> eFfeM_work: http://gitorious.org/~jama/angstrom/jama-shr-experimental/commit/62f3abe8a5ff1294b181ae42b26ee67396abdef6 Dec 17 16:28:28 khem: So otherwise happy then? Dec 17 16:28:35 Tartarus: yes Dec 17 16:28:48 acked-by? Dec 17 16:28:52 Sure. Dec 17 16:29:17 khem: I didn't look into emacs build but I was quite surprised to find whole sysroot copied to that qemu-treedir :) Dec 17 16:29:38 JaMa|Wrk: in emacs ? Dec 17 16:29:41 khem: yes Dec 17 16:29:45 hmm Dec 17 16:29:59 -L /OE/tmpdir-shr/work/armv4t-oe-linux-gnueabi/emacs-x11-23.1-r3/qemu-treedir is whole armv4t sysroot Dec 17 16:30:05 hm this Dec 17 16:30:07 okay Dec 17 16:30:07 hi all Dec 17 16:30:15 hi stefan_schmidt Dec 17 16:30:15 I thought jama found another one Dec 17 16:30:17 he stefan Dec 17 16:30:43 hi khem Dec 17 16:30:46 hi woglinde Dec 17 16:30:49 jama a link or real copy? Dec 17 16:30:59 real copy Dec 17 16:31:00 pb_: ping Dec 17 16:31:01 JaMa|Wrk: looks fine Dec 17 16:31:17 zecke, ericben: ping Dec 17 16:31:25 I guess it's considered too dangerous to run some build process loose on real sysroot with symlink Dec 17 16:31:33 stefan_schmidt: pong Dec 17 16:31:42 khem: as uclibc guru, whats your opinion on this one? http://patchwork.openembedded.org/patch/1008/ Dec 17 16:31:56 ericben: QT patch: http://patchwork.openembedded.org/patch/986/ Dec 17 16:32:05 ericben: Is something like this still needed? Dec 17 16:32:20 ericben: looks odd to me and it seems QT is in good shape... Dec 17 16:32:23 03Martin Jansa  07master * r7ea9be2b5d 10openembedded.git/conf/distro/ (include/sane-toolchain.inc shr.conf): Dec 17 16:32:23 sane-toolchain: use hard assignment instead of weak for libtool version and LIBTOOL_HAS_SYSROOT Dec 17 16:32:23 * bitbake.conf already has weak LIBTOOL_HAS_SYSROOT ?= "no" so setting to "yes" here was ignored Dec 17 16:32:23 Acked-by: Khem Raj Dec 17 16:32:23 Signed-off-by: Martin Jansa Dec 17 16:32:45 Tartarus: Any more comments on this one? http://patchwork.openembedded.org/patch/1067/ Dec 17 16:33:07 03Tom Rini  07org.openembedded.dev * r99498a6f35 10openembedded.git/recipes/libaio/libaio_0.3.107.bb: Dec 17 16:33:07 libaio: Add 0.3.107 Dec 17 16:33:07 Also remove legacy staging while in here. Dec 17 16:33:07 Signed-off-by: Irfan Sadiq Dec 17 16:33:07 Signed-off-by: Tom Rini Dec 17 16:33:15 khem: also the same emacs-x11 version was built with OE from Oct24 just fine, haven't found any change in emacs-x11 or qemu-0.12.5 since then Dec 17 16:33:36 stefan_schmidt: Did you fix those first comments? :) Dec 17 16:33:59 And can somebody enlight me about this patch? Does not seem to make sense for me. http://patchwork.openembedded.org/patch/1316/ Dec 17 16:34:07 03Tom Rini  07master * rfa6b202f9c 10openembedded.git/recipes/gettext/gettext_0.18.bb: Dec 17 16:34:07 gettext-native (0.18): Disable curses and enable relocatable Dec 17 16:34:07 Since we don't need the curses-based tools, just cli, we can disable that. Dec 17 16:34:07 We enable the relocation flag since if the tool wants to help us here, Dec 17 16:34:07 we should take it. Dec 17 16:34:08 Signed-off-by: Tom Rini Dec 17 16:34:13 Tartarus: I fixed nothing just reviewing patches in patchwork Dec 17 16:34:17 03Tom Rini  07master * rb7b5aa0a9b 10openembedded.git/recipes/meta/external-toolchain-csl.bb: Dec 17 16:34:17 external-toolchain-csl: Allow for sbin to be empty, update RPROVIDES Dec 17 16:34:17 Make the RPROVIDES list for glibc-dev match the real one. In some Dec 17 16:34:17 cases the sbin directory is empty, and that is OK. Dec 17 16:34:18 Signed-off-by: Tom Rini Dec 17 16:34:20 03Tom Rini  07master * r4ba8133097 10openembedded.git/conf/distro/include/ (4 files): Dec 17 16:34:20 sane-toolchain: Switch to TOOLCHAIN_TYPE Dec 17 16:34:20 Make use of TOOLCHAIN_TYPE and ${LIBC}-${TOOLCHAIN_TYPE} files Dec 17 16:34:20 like Angstrom/SHR/etc do. This lets external toolchains work Dec 17 16:34:20 with sane-toolchain.inc using distributions. Dec 17 16:34:21 Acked-by: Khem Raj Dec 17 16:34:21 Signed-off-by: Tom Rini Dec 17 16:34:22 03Tom Rini  07master * r307887a551 10openembedded.git/conf/distro/include/toolchain-external.inc: Dec 17 16:34:22 toolchain-external.inc: Uncomment linux-libc-headers Dec 17 16:34:25 ah stefan started with patchwork Dec 17 16:34:27 stefan_schmidt: 986 doesn't seems to be a patch for OE Dec 17 16:34:30 Tartarus: especialy old ones. :) No personal interest in binary packaging :) Dec 17 16:34:46 JaMa|Wrk: did you change anything on your build host ? Dec 17 16:34:48 Acked-by: Khem Raj Dec 17 16:34:48 Signed-off-by: Tom Rini Dec 17 16:34:53 ericben: no, its not, put maybe a issue when using OE Dec 17 16:35:15 Well, I still don't like the patch, no, it's not adding much time saving and has issues :) Dec 17 16:35:28 Tartarus: ah, and as you are doing external toolchain stuff right now :) Any comment on this one? http://patchwork.openembedded.org/patch/994/ Dec 17 16:35:56 khem: no afaik (the host where I built it fine before is just some VM where I don't have even root to say it for sure) Dec 17 16:36:05 Tartarus: ok, care to comment on it? Can go from patchwork then. Dec 17 16:36:35 * stefan_schmidt is doing a fast patchwork session right now. Need to prepare for the next x-mas party Dec 17 16:36:45 For 994, that doesn't look right at all and shouldn't be needed Dec 17 16:36:48 stefan_schmidt: I don't know if that's still needed. We don't seems to have problem to build qt4 applications so 986 may not be needed any more Dec 17 16:37:07 For 1067, isn't my original comment there enough? :) Dec 17 16:37:11 ericben: ok, will remove the patch from patchwork then Dec 17 16:37:29 Tartarus: fair enough :) Dec 17 16:37:42 Tartarus: you never got an reaction? On IRC or something? Dec 17 16:37:57 If not the patch can just go. Dec 17 16:38:48 No, I never got a follow-up from that guy, re 1067 Dec 17 16:38:54 ok, gone Dec 17 16:38:54 03Chris Larson  07master * r7c7648e62d 10bitbake.git/lib/bb/cooker.py: Dec 17 16:38:54 cooker: ensure parseCommandLine can get BBPKGS Dec 17 16:38:54 Without this, one can't use BBPKGS with bitbake -g, for example. Dec 17 16:38:54 Signed-off-by: Chris Larson Dec 17 16:39:01 03Chris Larson  07master * rdb56f34e22 10bitbake.git/ (bin/bitbake lib/bb/server/process.py): (log message trimmed) Dec 17 16:39:01 Ensure LogHandler is set up for the server construction Dec 17 16:39:01 If we add the LogHandler object in the server.main, rather than earlier, any Dec 17 16:39:01 log messages emitted during the construction of the server (and, therefore, Dec 17 16:39:02 during the construction of the cooker) will not be seen. This includes the Dec 17 16:39:02 error(s) which the cooker can emit when parsing the command line. With this, Dec 17 16:39:03 we again see the error if you try to use -g without specifying a recipe to Dec 17 16:39:23 03Chris Larson  07master * rb0cbc728e5 10bitbake.git/bin/bitbake: (log message trimmed) Dec 17 16:39:23 bin/bitbake: [HACK] Empty ui event queue on shutdown Dec 17 16:39:23 This is a workaround to ensure the user always sees any queued up log Dec 17 16:39:23 messages, even if the UI has shut down without showing them. This is far from Dec 17 16:39:24 ideal. The correct fix is to alter the UIs to shut down more cleanly, Dec 17 16:39:24 ensuring they display any remaining queued up events to the user before they Dec 17 16:39:25 exit. Dec 17 16:39:59 * JaMa|Wrk refused to build also pygtk in minimalistic chroot to try goggle UI :/ Dec 17 16:40:00 I will not have as much time this WE as I hoped. Will be online tomorrow for some hours and hopefully a bit more on sunday Dec 17 16:40:10 don't wait for me for patchwork cleanup :D Dec 17 16:40:38 * JaMa|Wrk will wait for separate client/server with server in minimalistic chroot and client outside in normal system Dec 17 16:42:41 stefan_schmidt: hi Dec 17 16:43:12 pb_: hi. How do you stand to your old patch? http://patchwork.openembedded.org/patch/691/ Dec 17 16:43:16 pb_: its the oldest one I patchwork :) Dec 17 16:43:40 ah, heh, that Dec 17 16:43:47 I thought zecke was going to check it in the other week, actually Dec 17 16:43:50 I guess he forgot :-} Dec 17 16:45:43 pb_: would you do it? Dec 17 16:46:10 so the patch collect which m4 functions is needed? Dec 17 16:46:14 are Dec 17 16:46:16 hrms Dec 17 16:46:28 ah, some more review, even better Dec 17 16:46:29 :) Dec 17 16:46:52 stefan_schmidt: yah, I'll try to do it this evening Dec 17 16:47:06 pb_: thanks Dec 17 16:47:38 whew, i think i managed to successfully kill off bitbake's hackish injection of os/bb into __builtins__ Dec 17 16:47:48 finally, good riddance Dec 17 16:47:55 oh, very good Dec 17 16:48:03 kergoth this helps us on which way? Dec 17 16:48:39 well, the way we were doing it was incompatible with python implementations other than cpython, for one Dec 17 16:49:55 ah Dec 17 16:50:00 * kergoth_ wonders how well bitbake would perform under pypy with its jit Dec 17 16:50:15 try it now? Dec 17 16:50:22 i plan on it :) Dec 17 16:51:46 at the moment i'm trying to reproduce Jay7's klcc segfault Dec 17 16:52:39 anyone happens to know who mainatins the metano distro in OE? Dec 17 16:53:22 stefan_schmidt: Angelo Dec 17 16:53:25 Ah, its actually angelo. Dec 17 16:53:36 kergoth isnt that a task for khem? Dec 17 16:53:42 ericben: And the patches for metano are also from him Dec 17 16:53:45 hmm Dec 17 16:53:57 :) Dec 17 16:54:23 ericben: do you know his nick here on IRC? Dec 17 16:54:35 woglinde: he reported that a bitbake change to stdout/stderr handling for task execution introduced it. the compiler still has a bug either way, but the change he reported caused it should only affect python tasks, not shell, and do_compile is shell, so I'm trying to see if bb.build is doing something wrong Dec 17 16:55:34 stefan_schmidt: can't remember, how can I get the list of connected peoples to the channel to check for it ? Dec 17 16:56:03 /names Dec 17 16:56:10 or /who, iirc Dec 17 16:56:18 ericben: oh, no big IRC skills over here. don't worry, will send him a mail tomorrow Dec 17 16:56:25 kergoth_: thanks Dec 17 16:57:03 * kergoth_ grabs vim 7.3 from natty for persistent undo Dec 17 16:57:46 have to hurry. will be back later tonight Dec 17 16:58:17 stefan_schmidt: it's angelox123 I think Dec 17 16:58:32 maybe with underscore before 123 Dec 17 16:59:33 thank you everybody, have a nice rest of the day Dec 17 17:06:19 gm Dec 17 17:09:06 hmmm Dec 17 17:16:34 kergoth oh thanks for explanation Dec 17 17:18:12 mickeyl: if you have some minutes this weekend review on http://patchwork.openembedded.org/patch/1631/ would be good Dec 17 17:43:40 urg, stupid email server Dec 17 17:45:43 stefan_schmidt, hi Dec 17 17:47:09 GNUtoo|laptop: gi ahead fast, on my way to an x-mas party Dec 17 17:47:21 s/gi/go/ Dec 17 17:47:27 just wanted to say hi Dec 17 17:47:32 ah, heh Dec 17 17:47:35 hi GNUtoo|laptop :) Dec 17 17:48:53 03Tom Rini  07org.openembedded.dev * r12275a1604 10openembedded.git/conf/bitbake.conf: Dec 17 17:48:53 bitbake.conf: Use normal bytes per inode param to genext2fs Dec 17 17:48:53 Without this we get an unusably small number of inodes in Dec 17 17:48:53 our filesystem images. Dec 17 17:48:53 Signed-off-by: Tom Rini Dec 17 17:50:01 Tartarus: commit fa6b202f9c954df43aa26ec8c9fa8b717d4074c1 does not build, you did not commit the patch itself Dec 17 17:50:21 isable-gettext-tools-tests.patch Dec 17 17:52:34 Er, I thought I skipped that hunk Dec 17 17:52:57 * Tartarus kicks git Dec 17 17:53:36 thanks Dec 17 17:53:55 Jay7: segfault occurs with or without stdout/stderr change. still looking Dec 17 17:54:04 03Tom Rini  07org.openembedded.dev * r1df959f029 10openembedded.git/recipes/gettext/gettext_0.18.bb: Dec 17 17:54:04 gettext: Drop hunk about disable-gettext-tools-test.patch Dec 17 17:54:04 I'd have sworn I told git not to add this hunk. Dec 17 17:54:04 Signed-off-by: Tom Rini Dec 17 17:54:24 Jay7: bingo, its stdin. Dec 17 17:54:31 Jay7: if stdin is /dev/null, klcc segfaults Dec 17 17:54:38 this amuses me Dec 17 18:03:15 Jay7: got it, fixed. Dec 17 18:03:25 03Chris Larson  07master * r566657c13d 10bitbake.git/lib/bb/ (build.py ui/uihelper.py): Dec 17 18:03:25 build: stop injecting os/bb into __builtins__ Dec 17 18:03:25 For one, we should be able to avoid this via _context in bb.utils. For two, Dec 17 18:03:25 __builtins__ only has to exist for CPython, so this causes problems running Dec 17 18:03:25 bitbake under other python implementations (though we could change to doing Dec 17 18:03:26 setattr()s against the __builtin__ module, let's avoid this instead). Dec 17 18:03:27 Signed-off-by: Chris Larson Dec 17 18:03:27 03Chris Larson  07master * r71f8b122dc 10bitbake.git/lib/bb/ (build.py utils.py): Dec 17 18:03:28 build: fix file mode for /dev/null Dec 17 18:03:28 Signed-off-by: Chris Larson Dec 17 18:03:29 03Richard Purdie  07master * r1537e94b9e 10bitbake.git/lib/bb/utils.py: Dec 17 18:03:29 bitbake/utils.py: Ensure the last lines of functions are printed in tracebacks Dec 17 18:03:30 Signed-off-by: Richard Purdie Dec 17 18:03:30 03Richard Purdie  07master * r1ad8b924c9 10bitbake.git/lib/bb/utils.py: Dec 17 18:03:31 bitbake/utils.py: Improve traceback to be more helpful/clear to users Dec 17 18:03:31 Signed-off-by: Richard Purdie Dec 17 18:05:07 03Denis 'GNUtoo' Carikli  07org.openembedded.dev * rf923ef91ba 10openembedded.git/recipes/freesmartphone/libisi_git.bb: (log message trimmed) Dec 17 18:05:07 Add new libisi recipe (libisi can talk to the nokia900 modem) Dec 17 18:05:07 libisi will be used in fsogsmd for the nokia900. Dec 17 18:05:07 As we can't use ofono directly we import some ofono parts in libisi, Dec 17 18:05:07 and use that library in the vala code of fsogsmd trough a vapi file. Dec 17 18:05:07 Thanks a lot mickeyl for help on #openmoko-cdevel on Freenode on irc: Dec 17 18:05:08 do you use =+ or += ? Dec 17 18:05:24 ??? Dec 17 18:05:42 the last not-me commit I had was from tom rini Dec 17 18:07:44 ha bitbake.git Dec 17 18:07:48 not org.openembedded.dev Dec 17 18:07:49 ok Dec 17 18:25:47 JaMa|Off: mpd 0.15.12 does not build here either, I am trying 0.16 now Dec 17 18:29:25 ah here neither Dec 17 18:29:29 mpd doesn't build Dec 17 18:29:33 it did long time ago Dec 17 18:36:18 actually it seems to be libtool related Dec 17 18:36:50 hmm, something's up with our log formatting Dec 17 18:36:59 guess i need to shift the level name changes into the formatter Dec 17 18:37:20 kergoth_: I think only place it's registered is in knotty Dec 17 18:38:06 http://cgit.openembedded.org/cgit.cgi/bitbake/tree/lib/bb/ui/knotty.py#n76 Dec 17 18:38:24 kergoth_, i've been adding shorthand accessors to datasmart for expandVar (but didn't yet test-convert oe). Do you think such shorthand accessors would be handy? Dec 17 18:39:38 possibly Dec 17 18:39:50 kergoth_, and i was thinking about ref-counting vars so i could warn about unused/used-once vars. Dec 17 18:40:03 foerster: yeah, i know. i guess the level no -> level name must happen when creating the log record, not on the display end, so when the processes were the name, it worked doing it fron knotty Dec 17 18:40:37 blindvt: i wouldn't bother, we already have reference tracking in the poky bitbake, it tracks exactly what variables refer to what other variables Dec 17 18:40:39 kergoth_, accessor would be something like http://paste.debian.net/102630/ Dec 17 18:41:05 blindvt: i've been experimenting with a datasmart wrapper in an oe.data module Dec 17 18:41:11 let me pastey what i have so far Dec 17 18:41:51 https://gist.github.com/745456 Dec 17 18:42:04 kergoth_, yes, i was about to add a wrapper too. the data references are too elaborate imho Dec 17 18:42:12 d.setVars(alpha=5, beta=4, theta=6) Dec 17 18:42:15 ;) Dec 17 18:42:59 kergoth_: whats underlying bug with compiler w.r.t klcc ? Dec 17 18:43:15 blindvt: hey Dec 17 18:43:38 kergoth_, i like the inherits(). It's a pita that inherits_class doesn't handle lists. Leads to really ugly and chatty code in oe Dec 17 18:43:56 khem, hi Dec 17 18:44:21 blindvt: I have seen a bug in uclibc building O=/x/y/z in install_headers Dec 17 18:44:32 khem, PR? Dec 17 18:44:34 blindvt: I will post the patch Dec 17 18:44:44 khem, oh, ok. TIA Dec 17 18:45:03 blindvt: problem is that headers exist in both srcdir and buildir which should be installed Dec 17 18:45:13 right now it only installs the one from srcdir Dec 17 18:45:22 which works well if O is not set Dec 17 18:45:27 or is same at topdir Dec 17 18:45:44 khem, oh darn. are there any generated headers we have to install? Dec 17 18:45:59 blindvt: inherits is in oe.utils now, btw Dec 17 18:46:04 khem, (apart from syscalls which supposedly works fine already) Dec 17 18:46:17 khem: it segfaults if stdin is a file descriptor that wasn't open for read ;) not that i blame it.. Dec 17 18:46:19 blindvt: yes uClibc_config.h Dec 17 18:46:22 sysnum.h Dec 17 18:46:28 and all include/bits Dec 17 18:46:41 kergoth: heh ok Dec 17 18:46:54 khem, mhm. good catch. Odd, i'm pretty sure that i once checked O= install Dec 17 18:46:57 I would pardon gcc for that Dec 17 18:47:48 blindvt: check the ml Dec 17 18:47:51 I just posted it Dec 17 18:48:19 add your build expertise to it if I missed something Dec 17 18:48:34 khem, btw.. i'm currently switching micro default package format from ipkg to opkg for the ipkg checkout doesn't seem to work for me. I intend to put that on some personal branch for review. Any branch-name rules i should obey to? Dec 17 18:49:49 blindvt: the main reason i did the wrapper was to make getVar default to expanded, so i can stop passing True every fucking time Dec 17 18:49:53 blindvt: under your name create a sensible name thats it Dec 17 18:50:01 blindvt: and also to make sure __getvar__ gets expanded also Dec 17 18:50:09 so d['CC'] gives you the expanded value Dec 17 18:50:25 kergoth: yes thats nice Dec 17 18:50:30 Lets see, the e2fsprogs failure has been reported now, yes? Dec 17 18:50:34 kergoth_, :) yea, there are only about 20 or so unexpanded users and perhaps even those would be able to work fine on the expanded vars (didn't really look yet) Dec 17 18:51:06 yeah, there are a few legit usages of unexpanded, but they're few and far between Dec 17 18:54:32 foerster: there, now all UIs will get the same level names -- of course they don't have to use them, but at least it isn't knotty specific now Dec 17 18:54:36 heh Dec 17 18:54:36 03Chris Larson  07master * rdc5a5c3929 10bitbake.git/lib/bb/ (msg.py ui/knotty.py): Dec 17 18:54:36 Fix logging level names for post-server-ui-split Dec 17 18:54:36 Signed-off-by: Chris Larson Dec 17 18:54:51 kergoth_: good stuff Dec 17 18:54:56 kergoth_, i like your oe.data. What about __setitem__ and __contains__ ? Dec 17 18:55:50 is ml up ? Dec 17 18:55:54 blindvt: DataSmart has all of that, at least in master Dec 17 18:56:08 it's a MutableMapping abc Dec 17 18:56:28 for portability, could shift some of that over Dec 17 19:00:27 damn Dec 17 19:00:30 foerster: bitbake -c foo busybox Dec 17 19:00:45 foerster: hangs after the runqueue raises SystemExit regarding the missing task Dec 17 19:00:46 :\ Dec 17 19:01:26 damn Dec 17 19:02:46 * foerster just thought of one thing that needs fixed Dec 17 19:03:04 ouch, worse than that, a keyboard interrupt won't exit it Dec 17 19:03:08 yea Dec 17 19:03:11 that's what i'm fixing Dec 17 19:03:13 cool Dec 17 19:03:19 every runCommand blocks forever for response Dec 17 19:03:23 if no server, no response :) Dec 17 19:03:40 need to time that out Dec 17 19:03:56 ^C is blocked while waiting for the response Dec 17 19:04:50 looks like exceptions from commands just get printed, but don't cause the server to exit. i wonder if its time to pass the exception back to the UI if a command fails Dec 17 19:10:16 kergoth_: https://github.com/foerster/bitbake/commit/3484e743ccd427672dba4cf95b3677823998264d Dec 17 19:10:28 now ui can still exit in this failing case we have Dec 17 19:13:29 jo re Dec 17 19:13:53 hm autoconf seems broken here Dec 17 19:14:00 after latest updates Dec 17 19:14:11 CRITICAL: global name 'bb' is not defined Dec 17 19:14:17 I see too Dec 17 19:14:35 hrm Dec 17 19:14:37 hold Dec 17 19:15:15 all do_patch Dec 17 19:15:25 CRITICAL: Function 'patch_do_patch' failed Dec 17 19:15:40 ah, i see what it is Dec 17 19:15:42 03Chris Larson  07master * rdc7893d0da 10bitbake.git/lib/bb/ (build.py ui/uihelper.py): Dec 17 19:15:42 Revert "build: stop injecting os/bb into __builtins__" Dec 17 19:15:42 Temporary revert, will resurrect. Dec 17 19:15:42 This reverts commit 566657c13de3fefb5043b71de953fb805799da95. Dec 17 19:15:44 well, here's a revert for now Dec 17 19:15:48 i'll bring it back after further testing Dec 17 19:15:56 okay Dec 17 19:16:00 kergoth: unit tests :) Dec 17 19:16:04 sorry for the bad news Dec 17 19:16:05 i had it fixed locally Dec 17 19:16:10 he khem Dec 17 19:16:15 it's not bad news Dec 17 19:16:17 i know what's causing it Dec 17 19:16:22 woglinde: hello Dec 17 19:16:24 its a python module in lib/oe/ without the necessary imports Dec 17 19:17:00 ah Dec 17 19:17:01 okay Dec 17 19:17:31 but i don't wnat master to stop working with old OE Dec 17 19:17:39 so i'll hold off on this bitbake commit until later on Dec 17 19:20:38 03Chris Larson  07master * r9723a1c474 10bitbake.git/lib/bb/fetch/__init__.py: Dec 17 19:20:38 Update persist_data usage to the new API Dec 17 19:20:38 Signed-off-by: Chris Larson Dec 17 19:20:40 03Chris Larson  07master * r2bc893dc40 10bitbake.git/lib/bb/parse/__init__.py: Dec 17 19:20:40 parse: use bb.utils.which Dec 17 19:20:40 Signed-off-by: Chris Larson Dec 17 19:20:48 03Chris Larson  07master * r40778a6e9e 10bitbake.git/ (TODO lib/bb/cooker.py lib/bb/data_smart.py): Dec 17 19:20:48 Move LAYERDIR expansion hack into DataSmart Dec 17 19:20:48 Signed-off-by: Chris Larson Dec 17 19:26:00 03Chris Larson  07master * r8caed59aac 10openembedded.git/lib/oe/patch.py: Dec 17 19:26:00 oe.patch: add missing bb imports Dec 17 19:26:00 Signed-off-by: Chris Larson Dec 17 19:28:07 in particular, i want to make sure the next bitbake release works with oe release-2010.12. might hold off on the __builtins__ commit until the next oe release Dec 17 19:29:38 03Bob Foerster  07master * r33e36916df 10bitbake.git/lib/bb/server/process.py: Dec 17 19:29:38 Timeout UI runCommand calls if server doesn't respond (or dies) Dec 17 19:29:38 Signed-off-by: Bob Foerster Dec 17 19:31:12 foerster: ran into an odd one. apparently the getMessage() method of the log record is what merges the args with the format string, so if you pass the wrong args to a logger.whatever call, it won't die until the uihelper tries to get the message Dec 17 19:31:16 hehe Dec 17 19:31:25 guess we'll have to check for that Dec 17 19:31:35 nice Dec 17 19:33:14 i know when I was doing event._msg % event.args manually, it threw an error in that case about not enough args. When I switched to getMessage, I didn't see the error any more Dec 17 19:34:52 i haven't seen it often, it must be this *one* particular message somewhere, not in a primary codepath Dec 17 19:54:27 kergoth_, could it be that you accidentally reverted RPs 1ad8b924c90bfb953c53db090cfe3abc324afc16 in your /dev/null fix (71f8b122dc8477b12911ff7b145a42c4c97a118f) ? Dec 17 20:00:05 kergoth_, re: bb.which: mind if i also change that in (the sample) conf/bitbake.conf for consistency? Dec 17 20:03:24 Hello Dec 17 20:03:45 Any one can recommend me a simpler app launcher that works under Fb? Dec 17 20:04:06 low memory resources is very important Dec 17 20:04:59 kristianpaul: what do you mean under app launcher? Dec 17 20:05:04 kind of WM? Dec 17 20:05:21 well not WM cause there is no windows (no X or related) Dec 17 20:05:29 kergoth_: thank you! Dec 17 20:05:35 just linux and framebuffer Dec 17 20:05:46 Tartarus: yes, I've reported here e2fsprogs error regarding uuid Dec 17 20:05:59 and some apps (SDL, GTK1) that ruin in fb Dec 17 20:06:07 kristianpaul: psplash, kexecboot :) Dec 17 20:06:09 but they need to be lanunched Dec 17 20:06:13 but Dec 17 20:06:18 i need a Menu Dec 17 20:06:23 a nasic one Dec 17 20:06:37 basic* Dec 17 20:06:58 kexecboot have (perhaps ugly) code for menu and can launch kexec tool :) Dec 17 20:07:12 and I'm developing it ;) Dec 17 20:07:16 he Dec 17 20:07:23 so feel free to ask Dec 17 20:07:23 good ! Dec 17 20:07:28 i can bug you Dec 17 20:07:41 look on kexecboot.org Dec 17 20:08:07 wait a min Dec 17 20:08:16 is that kexec linux feature? Dec 17 20:08:25 yes Dec 17 20:08:26 do you know wener? Dec 17 20:08:31 werner* Dec 17 20:08:38 mm... seems no Dec 17 20:09:17 please join #qi-hardware Dec 17 20:09:25 mom Dec 17 20:20:44 blindvt: apparently, yes, i don't know how i managed to do that Dec 17 20:20:57 blindvt: no, bb.which is portable, and oe doesn't require 1.10 yet, much less 1.11/1.12 Dec 17 20:21:04 blindvt: so the examples should remain compatible with 1.8 Dec 17 20:21:32 03Richard Purdie  07master * r70cba0e84d 10bitbake.git/lib/bb/utils.py: Dec 17 20:21:32 bitbake/utils.py: Improve traceback to be more helpful/clear to users Dec 17 20:21:32 Signed-off-by: Richard Purdie Dec 17 20:21:51 03Richard Purdie  07master * rb23356a96c 10bitbake.git/lib/bb/utils.py: Dec 17 20:21:51 bitbake/utils.py: Ensure the last lines of functions are printed in tracebacks Dec 17 20:21:51 Signed-off-by: Richard Purdie Dec 17 20:22:40 kergoth_, well conf/bitbake.conf from bitbake master is what i would expect to cover exactly this bitbake master version, but ok. Dec 17 20:22:54 oh, you didn't say *which* bitbake.conf you were talking about :P Dec 17 20:23:12 the bitbake one, sure Dec 17 20:24:20 03Bernhard Reutner-Fischer  07master * r447c87f006 10bitbake.git/conf/bitbake.conf: Dec 17 20:24:20 bitbake.conf: use bb.utils.which Dec 17 20:24:20 Signed-off-by: Bernhard Reutner-Fischer Dec 17 20:24:30 03Bernhard Reutner-Fischer  07master * r99cdb61b30 10bitbake.git/lib/bb/utils.py: Dec 17 20:24:30 utils: fix typo in error message Dec 17 20:24:30 Signed-off-by: Bernhard Reutner-Fischer Dec 17 20:25:44 kergoth_, erm, sorry. _of_course_ i ment bitbake's bitbake.conf :P Dec 17 20:26:03 :) Dec 17 20:45:18 I see some kind of patchwork weekend is started already :) Dec 17 20:45:26 I can start some testbuilds now Dec 17 20:45:36 every clean build is around 2 hrs Dec 17 20:45:39 so.. Dec 17 20:45:49 Jay7: cool Dec 17 20:45:55 but ml is still dead Dec 17 20:46:01 bad :( Dec 17 20:46:04 I will go ahead and set up mailman on openembedded.org Dec 17 20:46:10 tom or florian? Dec 17 20:46:21 ml is on ltg Dec 17 20:46:25 its florian Dec 17 20:46:28 florian is informed but seems still can't do anything Dec 17 20:46:28 which is florian Dec 17 20:46:50 Crofton|work: I was talking to florian other day and he thinks its a good thing to have it on oe.org Dec 17 20:47:14 ok Dec 17 20:47:21 I was going to ask that :) Dec 17 20:47:24 thanks for checking Dec 17 20:47:40 but he does not have time to do it :( Dec 17 20:47:43 thats bad thing Dec 17 20:47:45 yeah Dec 17 20:47:50 end of the year chaos Dec 17 20:47:58 I'll be afk most of the weekend also Dec 17 20:48:02 yeah mailman is also on vacation :) Dec 17 20:48:56 heh Dec 17 20:56:51 khem: I confirm you fixed samba's compile problem with the latest gcc update Dec 17 21:01:26 khem, not sure if you backread the channel log but Jay7 and my gettext problem was caused because we used libtool 2.4 but didn't have the SYSROOT support enabled Dec 17 21:01:41 yep Dec 17 21:02:32 eFfeM: yes I saw the message Dec 17 21:02:41 eFfeM: JaMa|Off should push a fix Dec 17 21:02:47 if he hasnt done that Dec 17 21:02:48 JaMa|Off: I can confirm mpd 0.16 works, will commit Dec 17 21:02:59 then you wont need to set that in local.conf Dec 17 21:03:17 what is the SYSROOT support ? Dec 17 21:03:47 ericben: libtool now knows about sysroot like cross gcc Dec 17 21:04:02 hm yeah lets see if it fixes the openjdk libz problem too Dec 17 21:04:02 so it generates the .la files properly sysrooted Dec 17 21:04:06 ericben: libtool 2.4 has sysroot support but for now you need to enable it in a conf file Dec 17 21:04:07 LIBTOOL_HAS_SYSROOT = "yes" Dec 17 21:04:36 khem: eFfeM: OK so that will become a default ? Dec 17 21:05:24 ah he has pushed it 7ea9be2b5d3ac4fb5da74506c2516b3412726495 Dec 17 21:05:24 cool Dec 17 21:05:37 so now it should work without issues Dec 17 21:05:42 ah nice Dec 17 21:05:47 eFfeM: you can punt it from local.cong Dec 17 21:05:56 is mailing list problem related to the exim remote root vulnerability ? Dec 17 21:06:12 ericben: no idea Dec 17 21:06:16 florian whould know Dec 17 21:06:19 will do, after the next pull Dec 17 21:09:00 khem: anything particular to test ? I'll be away most of the weekend so I can launch several builds if that helps Dec 17 21:10:53 ltg does seem a bit broken, not accepting shell logins. Dec 17 21:11:02 I think someone with hetzner access probably needs to reboot it. Dec 17 21:17:44 khem, are you interested in a qemux86_64 arch? Dec 17 21:20:00 hm Dec 17 21:20:05 that _ maay cause problems Dec 17 21:20:18 libz.la still has /usr/lib in staging for target Dec 17 21:20:32 Don't we support x8664 in some manner already? Dec 17 21:22:19 03Frans Meulenbroeks  07org.openembedded.dev * rb704b75747 10openembedded.git/recipes/musicpd/mpd_0.16.bb: Dec 17 21:22:19 mpd: upgraded to 0.16 Dec 17 21:22:19 This one works with libtool 2.4, the previous version did not. Dec 17 21:22:19 Signed-off-by: Frans Meulenbroeks Dec 17 21:24:08 damnit, got that getMessage error in runningbuild again Dec 17 21:24:09 hrm Dec 17 21:24:45 kergoth_: what's it throwing? can we just catch it? Dec 17 21:25:17 probably, it's just a TypeError coming from getMessage Dec 17 21:33:22 hmmm Dec 17 21:33:23 warning: Unable to find dynamic linker breakpoint function. Dec 17 21:33:33 I've that in gdb on my target Dec 17 21:33:37 I'll try gdb-cross Dec 17 21:35:45 GNUtoo|laptop: this usually mean the library is not loaded Dec 17 21:35:53 ok Dec 17 21:35:55 hi btw Dec 17 21:36:15 hi ;) Dec 17 21:37:41 I've strange things since some time with gdbserver+gdb: Dec 17 21:37:43 warning: no loadable sections found in added symbol-file /home/gnutoo/embedded/sshfs/usr/lib/.debug/libz.so.1.2.3 Dec 17 21:37:48 lots of stuff like that Dec 17 21:39:55 and indeed it's not loaded Dec 17 21:40:01 but very long time ago it worked Dec 17 21:40:32 I'll look with the binutils Dec 17 21:40:58 GNUtoo|laptop: doesn't that mean that the file don't have debug symbol inside ? (not sure at all) Dec 17 21:41:08 I'll look Dec 17 21:41:15 I don't remember well what it means Dec 17 21:42:50 oops no more battery, time to sleep ;) Dec 17 21:42:51 bye Dec 17 21:42:56 bye Dec 17 21:45:14 I bet it's some path errors Dec 17 21:45:16 kergoth_, is there any way to recover from "ERROR: running idle function" or should it be fatal? I'm seeing this while playing around in oe: http://paste.debian.net/102643/ Dec 17 21:45:28 some time ago I was told that it tried to load them as normal libs Dec 17 21:45:33 in #gdb Dec 17 21:46:52 blindvt: the fundamental problem is the runqueue failed to determine the cause of the circular dependency and died, and the server noticed it died (hence the error running idle function). there's no way that could be nonfatal, as it will fail to generate a runqueue to build. you have a circular dep that it can't figure out for some reason Dec 17 21:47:07 blindvt: the runqueue will need to be enhanced to handle whatever circumstance you're in Dec 17 21:49:13 03Chris Larson  07master * r2b07525d80 10bitbake.git/lib/bb/runqueue.py: Dec 17 21:49:13 runqueue: kill pre-server-ui-split remnant Dec 17 21:49:13 Signed-off-by: Chris Larson Dec 17 21:49:24 03Chris Larson  07master * r976e4f84a8 10bitbake.git/lib/bb/ (build.py runqueue.py ui/knotty.py): Dec 17 21:49:24 logger usage cleanup Dec 17 21:49:24 Signed-off-by: Chris Larson Dec 17 22:01:34 anyone an idea what package provides mkfs.jffs2 ? Dec 17 22:01:58 mkfs-jffs2 - 1.3.1-r0.6 - Tools for managing memory technology devices. Dec 17 22:02:06 maybe that package Dec 17 22:02:12 I didn't install it to check tough Dec 17 22:02:21 is debugging broken in oe? Dec 17 22:02:30 ? Dec 17 22:02:48 03Bernhard Reutner-Fischer  07master * r7498466f0e 10bitbake.git/lib/bb/runqueue.py: Dec 17 22:02:48 runqueue: use fewer newlines in error path Dec 17 22:02:48 Signed-off-by: Bernhard Reutner-Fischer Dec 17 22:02:57 kergoth_, I've that: Dec 17 22:03:01 I have mo recipe with jffs2 in the name, mkaing rootfs is failing for me becase mkfs.jffs2 is missing Dec 17 22:03:05 warning: no loadable sections found in added symbol-file /home/gnutoo/embedded/sshfs/usr/lib/.debug/libz.so.1.2.3 Dec 17 22:03:09 and a lot more like this one Dec 17 22:03:12 in gdb Dec 17 22:03:20 ah mtdutils Dec 17 22:03:21 I use gdb+gdbserver Dec 17 22:03:46 seems a dep for mtdutils-native is missing somewhere Dec 17 22:04:10 eFfeM, opkg infos can give clues on where the package comes from Dec 17 22:04:21 ah ok Dec 17 22:04:30 s/infos/info Dec 17 22:04:44 but the issue was that the package was not build, so guess opkg info would not have helped in that case Dec 17 22:04:49 Source: git://git.infradead.org/mtd-utils.git;protocol=git;tag=v1.3.1 file://add-exclusion-to-mkfs-jffs2-git-2.patch file://fix-ignoreerrors-git.patch file://lzo_1x-git.patch Dec 17 22:04:59 ok Dec 17 22:04:59 btw GNUtoo|laptop are you going to fosdem again ? Dec 17 22:05:02 yes Dec 17 22:05:07 nice Dec 17 22:05:10 I should normally go at fosdem Dec 17 22:05:26 hi khem Dec 17 22:07:39 now I've illegal instruction Dec 17 22:07:43 and still no debugger Dec 17 22:07:49 :( Dec 17 22:12:22 hmm Dec 17 22:14:52 blindvt: https://github.com/kergoth/bitbake/wiki/TODO - feel free to add anything else you know of Dec 17 22:15:33 calling it a day, nite all & have fun! Dec 17 22:15:52 night Dec 17 22:15:58 gn Dec 17 22:16:03 good night Dec 17 22:16:21 any pointers on gdb? Dec 17 22:25:42 good nite Dec 17 22:45:36 is there a subtle reason for the double mention of recrdeptask in package_ipk.bbclass or is it just a (harmless) bug? Dec 17 22:53:14 khem: how is mailman? Dec 17 22:58:51 does anyone else think we should stop doing the -e MAKEFLAGS="" madness? Dec 17 22:59:10 it makes some recipes "just work" with makefile based buildsystems, but it encourages people to not read the makefiles Dec 17 22:59:14 which i think is problematic Dec 17 22:59:19 better to be explicit about it Dec 17 23:00:33 Jay7: i thought the klcc segfault bug was pretty funny. i had accidentally opened stdin as write+append. no read access. it tries and segfaults because it isn't prepared for the error reading stdin, i assume :) Dec 17 23:00:47 why it should be reading stdin is beyond me, though Dec 17 23:01:29 kergoth_: someone should read klcc source ;) it's perl Dec 17 23:01:33 gah Dec 17 23:01:55 * kergoth_ 's never had a write only stdin before, first time for everything.. Dec 17 23:02:40 I'm thinking about doing run for lost depends Dec 17 23:03:01 how is better to get list of package names suitable to pass to bitbake? Dec 17 23:03:06 may be read from sqlite cache? Dec 17 23:03:31 I don't like idea of parsing bb-files with some shell script or vice versa Dec 17 23:06:04 kergoth_: any ideas about this? Dec 17 23:06:27 what do you mean by lost depends? Dec 17 23:06:45 something that should be in *DEPEND but not there Dec 17 23:06:59 I wish to do clean build of every recipe Dec 17 23:07:27 I'm sure there will be a lot of errors :) Dec 17 23:08:38 main idea is old: for r in `get_package_names_from_recipes`; do rm -rf $TMPDIR; bitbake $r; done Dec 17 23:09:06 ah Dec 17 23:09:27 i do have a script that builds every recipe using prebuilts for all of its deps, from a different build directory, to ensure the tmpdir path changes, to find path hardcoding Dec 17 23:09:37 that uses bitbake -g and then a small python script to parse the .dot file Dec 17 23:09:42 which was the quickest way to get to it Dec 17 23:10:12 kergoth_: hm.. can you share it? or even better - package list from current oe master :) Dec 17 23:11:48 may be good idea to place script into contrib Dec 17 23:12:01 i'll ask about it Dec 17 23:12:20 kergoth_: have sqlite cache list of packages? Dec 17 23:12:31 I can just read it from there Dec 17 23:12:38 have not seen yet inside Dec 17 23:13:19 the cache really has all that information, but you'd have to hardcode knowledge of a specific cache file version, or import bb.cache and have it load it Dec 17 23:15:22 second question is what machine/distro to choose to do this.. Dec 17 23:18:15 well.. having little python script that uses bb.cache to print out list of packages is good idea Dec 17 23:18:41 second usable script is mapper between bb-file and package name Dec 17 23:19:20 only second is enough in common case Dec 17 23:20:17 recipe<->package info is in pkgdata, after the recipes have been built anyway Dec 17 23:55:58 Jay7: reading the cache would be straightforward, but the cache data format is different depending on bitbake version :) at least between 1.8/1.10 and master Dec 17 23:56:06 would need to handle both Dec 17 23:56:06 hmm Dec 17 23:57:51 kergoth_: for me master is enough Dec 17 23:58:02 I don't need universal script Dec 18 00:00:06 c = bb.cache.Cache(d) Dec 18 00:00:17 but you'd need to give it at least basic metadata, it needs CACHE set Dec 18 00:00:30 d = bb.data.init(); and then set CACHE in it Dec 18 00:01:13 constructing the cache will load it from disk Dec 18 00:01:16 CACHE=$1 :) Dec 18 00:01:32 or $2 Dec 18 00:01:34 c.depends_cache is a dictionary holding key == filename and value = RecipeInfo object from bb.cache Dec 18 00:01:41 you can look at the class to see what info is in the object Dec 18 00:01:59 all th epackages, depends, rdepends, etc are there Dec 18 00:20:53 how do I debug QA issues with .debug directories when creating a new recipe? ERROR: QA Issue with pd: non debug package contains .debug directory: pd path /work/armv7a-angstrom-linux-gnueabi/pd-0.42-6-r0/packages-split/pd/usr/lib/pd/bin/.debug/pd-gui Dec 18 00:21:23 see conf/bitbake.conf, the PACKAGES and FILES variables Dec 18 00:21:30 also see the openembedded user manual about packaging Dec 18 00:22:06 03Richard Purdie  07master * rdb14a900e1 10bitbake.git/lib/bb/utils.py: Dec 18 00:22:06 utils: Allow copyfile to copy files which aren't readable Dec 18 00:22:06 Signed-off-by: Richard Purdie Dec 18 00:22:43 ok I'll re-read those now Dec 18 00:23:09 the error message tells you the problem, those .debug files need to go into the -dbg package, rather than the main one Dec 18 00:23:18 but the globs in the FILES variables are sucking those into the main one Dec 18 00:23:29 wildcards, if that's clearer Dec 18 00:23:38 * kergoth_ guesses not everyone uses the glob() function :) Dec 18 00:28:15 yes it looks like the FILES_${PN} is pulling in ${bindir}/* that includes the .debug dir created in the strip step. Dec 18 00:28:28 How do I get it to not create the .debug? Dec 18 00:29:08 you don't want to do that. Dec 18 00:29:34 the debugging info is stripped out of the binaries and placed in .debug dirs, which get split out into -dbg packages, so they can be installed if you need to debug the application with gdb or what have you Dec 18 00:31:20 so I need to be more specific in my recipes FILES_${PN} def so it doesn't wild card in the .debug stuff? Dec 18 00:31:47 exactly Dec 18 00:33:53 I think I got it. The following is wrong: FILES_${PN} = "${bindir} ${libdir} ${datadir}" Dec 18 00:34:31 I got it from a google search for someones attempt at the similar thing I'm doing. Dec 18 00:35:50 thanks! Dec 18 00:39:18 np Dec 18 01:01:03 hey, I saw new version of bitbake go up, seems much faster, good job to whoever was responsible Dec 18 01:03:18 no new versions have been released yet, just a lot of activity on master recently Dec 18 01:03:25 looking to get a new release out asap Dec 18 01:04:34 ah Dec 18 01:04:46 what version are you using? Dec 18 01:04:48 i checked the version on it, said 1.11.0 Dec 18 01:05:03 that would be from the master branch Dec 18 01:05:08 and yes, much faster :) Dec 18 01:08:44 if you're on current master, you should try out the other UIs Dec 18 01:08:48 bitbake -u goggle task-boot Dec 18 01:08:51 really nice Dec 18 01:10:13 cool, i'll give it a shot Dec 18 01:10:25 it's a gtk gui build monitor Dec 18 01:11:49 re Dec 18 01:11:59 hey florian Dec 18 01:12:05 kergoth_: sounds extremely promising :) Dec 18 01:12:17 very very nice, the ability to easily copy out of it is awesome Dec 18 01:12:43 yeah, if you right click on a log it'll pastebin it and give you the url Dec 18 01:12:50 foerster's been doing great work on it Dec 18 01:14:02 I see that search is a work in progress? Dec 18 01:15:11 i think pretty much every UI but the default is a work in progress ;) Dec 18 01:26:44 kergoth_: about bb.data.init() then setting CACHE Dec 18 01:27:25 should I use d.setVar("CACHE", "path_to_cache_dir", 0)? Dec 18 01:28:04 ah... no, last arg should be something other Dec 18 01:30:16 crofton you around? Dec 18 01:31:56 no need for a third arg at all Dec 18 01:34:27 oh.. python is not my language :\ Dec 18 01:34:55 i can take a look at it, but not just now, knee deep in the poky sync Dec 18 01:34:59 er.. closer to chest deep, really... Dec 18 01:35:17 I'm trying to write something Dec 18 01:35:42 kergoth_ as long as you don't go underwater. Dec 18 01:35:55 ka6sox: here's hoping.. :) Dec 18 01:39:14 no... I need learn python from scratch Dec 18 01:39:29 google :) Dec 18 01:39:35 kergoth_: if you will have time look at this please :) Dec 18 01:39:37 is your friend Dec 18 01:39:49 here is 4:38 Dec 18 01:39:54 there are python documentation resources on python.org, even links to free online books, though i don't know if any are any good Dec 18 01:39:57 bed is my friend now :) Dec 18 01:40:07 hehe Dec 18 01:40:13 will take a look after i get done with what i'm working on Dec 18 01:40:51 ok, I'll just run testbuilder now Dec 18 01:46:14 mickeyl: around ? Dec 18 02:06:36 OE mail begins to arrive... Dec 18 02:06:42 * kergoth_ gasps Dec 18 02:08:19 * ka6sox had nothing to do with it....either breaking or fixin' **** ENDING LOGGING AT Sat Dec 18 03:00:01 2010