**** BEGIN LOGGING AT Wed Jan 26 02:59:57 2011 **** ENDING LOGGING AT Wed Jan 26 04:23:39 2011 **** BEGIN LOGGING AT Wed Jan 26 06:04:15 2011 Jan 26 06:23:40 03Khem Raj  07master * r0af5de8b75 10openembedded.git/recipes/uclibc/ (uclibc-git/powerpc_copysignl.patch uclibc_git.bb): Jan 26 06:23:40 uclibc: Add copysignl implementation for powerpc to uclibc Jan 26 06:23:40 * On some ppc architectures like ppce500 gcc generates calls to Jan 26 06:23:40 copysignl but this function is not provided by uclibc or gcc itself Jan 26 06:23:40 we add it to uclibc. Jan 26 06:23:41 Signed-off-by: Khem Raj Jan 26 06:48:46 How do I build a uclibc version of the minimal-image? I did set TARGET_OS="linux-uclibc" but the resulting image was built using eglibc Jan 26 06:56:44 you'll need to set LIBC or so Jan 26 07:00:27 so LIBC="uclibc"? :) Jan 26 07:01:45 better use DISTRO=minimal-uclibc Jan 26 07:02:17 ok =) Jan 26 07:03:28 thans Jan 26 07:03:31 thanks Jan 26 07:05:40 sedrik: something like DISTRO=minimal-uclibc MAC Jan 26 07:06:04 DISTRO=minimal-uclibc MACHINE=qemuarm bitbake minimal-image Jan 26 07:06:15 will generate minimal-image using uclibc Jan 26 07:07:13 I have machine set to beagleboard Jan 26 07:07:20 will that make any differense? Jan 26 07:11:21 you can use use MACHINE=beagleboard Jan 26 07:11:32 I just gave an example Jan 26 07:14:07 ok. thanks :) Jan 26 07:22:54 03Víctor Manuel Jáquez Leal  07master * r820c31e493 10openembedded.git/recipes/python/python-scons_1.3.0.bb: Jan 26 07:22:54 python-scons_1.3.0: fix licence type Jan 26 07:22:54 According to the wiki [1], the licence of the project [2] is MIT, not GPL. Jan 26 07:22:54 1. http://www.scons.org/wiki/FrequentlyAskedQuestions#head-4fe23271137d5e5e95370a1360909ca303157c0d Jan 26 07:22:54 2. http://scons.tigris.org/source/browse/scons/trunk/LICENSE?revision=557&view=markup Jan 26 07:22:54 Signed-off-by: Víctor Manuel Jáquez Leal Jan 26 07:22:55 Signed-off-by: Khem Raj Jan 26 07:23:04 03Víctor Manuel Jáquez Leal  07master * r16d02558af 10openembedded.git/recipes/python/python-scons_1.3.0.bb: (log message trimmed) Jan 26 07:23:04 python-scons_1.3.0: add missing dependencies Jan 26 07:23:04 In order to launch scons in a board it depends on a couple python packages Jan 26 07:23:04 which were not managed by the recipe. Jan 26 07:23:05 This patch adds them. Jan 26 07:23:05 Signed-off-by: Víctor Manuel Jáquez Leal Jan 26 07:23:06 Acked-by: Koen Kooi Jan 26 07:25:00 03Andreas Oberritter  07master * r972d4902b6 10openembedded.git/recipes/zope/python-zopeinterface_3.5.1.bb: (log message trimmed) Jan 26 07:25:00 python-zopeinterface-3.5.1: fix packaging Jan 26 07:25:00 * Inherit setuptools instead of distutils. Jan 26 07:25:00 Install paths with distutils (built on x86_64 for mipsel): Jan 26 07:25:00 usr/lib/python2.6/site-packages/zope.interface-3.5.1-py2.6-linux-x86_64.egg/zope/interface/*.py Jan 26 07:25:00 Install paths with setuptools: Jan 26 07:25:01 usr/lib/python2.6/site-packages/zope/interface/*.py Jan 26 07:57:16 good morning Jan 26 08:32:08 is there someone attracted to sthe sgx video acceleration on omap3 platforms ..seems we dont have recipes in oe yet Jan 26 09:36:10 * stefan_schmidt is thinking about if we will have another release on march 1th as originally planned or if the oe-core stuff will void this Jan 26 09:43:57 stefan_schmidt: ask the ML (and maybe we could do the release just before going to oe-core). Jan 26 09:44:08 I am getting several Jan 26 09:44:19 eFfeM_work: will send a mail later today Jan 26 09:44:23 ERROR: Task 7835 (/home/hudson/jobs/FM_TEST/workspace/openembedded/recipes/libgsm/libgsm_1.0.13.bb, do_package_stage) failed with exit code '1' Jan 26 09:44:31 errors, anyone an idea Jan 26 09:44:43 no idea Jan 26 09:53:17 how can i find the package which delivers /etc/version ? Jan 26 09:55:06 rob_w: did you grep for it in the tree? Jan 26 09:55:22 sure Jan 26 09:55:36 rob_w: hmm, I would think that it maybe gets assembled in some class. Based on the distro you used, etc.. Jan 26 09:55:37 but either i grep wrong or it is generated .. Jan 26 09:55:49 right .. the bb classes .. Jan 26 09:55:57 maybe in do_rootfs or so .. thx Jan 26 09:57:14 classes/rootfs_ipk.bbclass: echo ${BUILDNAME} > ${IMAGE_ROOTFS}/${sysconfdir}/version Jan 26 09:57:21 wow thx Jan 26 09:57:48 there might be more stuff that get added there Jan 26 09:58:03 at least sysconfdir and version should be a start to grep Jan 26 09:58:28 thats ok .. i just wanna alter the string a bit in that file Jan 26 09:58:44 have fun Jan 26 09:59:07 rob_w: if you are using angstrom look also at: ./recipes/angstrom/angstrom-version.bb Jan 26 09:59:21 or ./recipes/distro-version/distro-version.bb Jan 26 09:59:28 i did this prior .. Jan 26 09:59:32 grep -r version . | grep sysconfdir Jan 26 09:59:40 showing some files that may fit you Jan 26 10:00:38 $BUILDNAME is now my suspect to alter .. Jan 26 10:01:33 k got it .. Jan 26 10:01:44 thx stefan_schmidt quick help is best help Jan 26 10:01:52 rob_w: no problem Jan 26 10:29:44 03Koen Kooi  07org.openembedded.dev * r8fae434ef8 10openembedded.git/recipes/xorg-driver/xf86-video-omapfb_git.bb: Jan 26 10:29:44 xf86-video-omapfb: bump SRCREV a bit Jan 26 10:29:44 Signed-off-by: Koen Kooi Jan 26 13:23:41 03Koen Kooi  07org.openembedded.dev * r78ffd08aa6 10openembedded.git/recipes/xorg-driver/ (4 files in 2 dirs): Jan 26 13:23:41 xf86-video-omapfb: bump SRCREV to latest, apply patches to keep XV working Jan 26 13:23:42 Removing patch #3 will get you working xrandr, but broken Xv Jan 26 13:23:42 Signed-off-by: Koen Kooi Jan 26 13:46:49 can I enable the output form git in someway when bitbaking and fetching repos, it seems to have gotten stuck on linux-omap-psp-2.6.32 Jan 26 13:47:11 I trid enabling BBDEBUG but it gave no extra info during from the fetch command Jan 26 14:01:47 I tried passing -v to bitbake but it seems to only affect bitbake related output Jan 26 14:06:45 * hrw -> lunch Jan 26 14:16:52 ERROR: Logfile of failure stored in: /dev/null Jan 26 14:16:52 Log data follows: Jan 26 14:17:01 I've that for all failures Jan 26 14:17:16 I bet staging issues Jan 26 14:17:51 I'll try -DDD Jan 26 14:17:58 I bet my staging area is in a bad state Jan 26 14:18:42 task staging_packager: Failed Jan 26 14:23:01 and worse, with -k it's able to make an image Jan 26 14:23:05 even if all is failed Jan 26 14:23:07 ah stage Jan 26 14:23:09 sorry Jan 26 14:28:31 http://pastebin.com/ziFsVU0 Jan 26 14:28:36 anyone that can spot the error? Jan 26 14:32:33 sedrik: - Unknown paste ID, it may have expired or been deleted! is the error :-) Jan 26 14:32:57 hehe Jan 26 14:33:01 lets me check that again Jan 26 14:33:14 zi3.. Jan 26 14:33:32 but I dont think it will say that much, git just fails to get the source from arago Jan 26 14:34:25 ask TI dudes on #beagle :-) Jan 26 14:36:39 XorA: will try, thanks :) Jan 26 14:41:13 khem, yay for the opkg-utils update and fixing gcc-cross-sdk on e500v2+uclibc :) Jan 26 15:51:05 sedrik: git server died on arago, should work now Jan 26 15:52:20 btw, directing everyone to #beagle is a bad idea - there's "evil" koen, who poops on anything not angstrom... :) :) :) Jan 26 15:54:01 ~hail angstrom Jan 26 15:54:01 * ibot bows down to angstrom and chants, "I'M NOT WORTHY!!" Jan 26 16:12:47 hi Jan 26 16:13:10 * ji is few weeks into OE Jan 26 16:13:34 * ji is struggling to override kernel_do_install Jan 26 16:13:40 is it even possible ? Jan 26 16:14:06 ji: what are you trying to accomplish? Jan 26 16:15:18 ji: see openembedded/recipes/linux/mnci-ramses_2.4.21-rmk2-pxa1.bb Jan 26 16:16:07 I am trying to do another kernel recipe, which installs this second kernel at a different location than boot Jan 26 16:16:31 but without overriding it ends up overwriting kernel in /boot Jan 26 16:16:44 like multi-linux.bb ? Jan 26 16:16:58 oops, multi-kernel.inc Jan 26 16:17:49 no the .bb Jan 26 16:18:04 * ji didn't think about multi-kernel.inc Jan 26 16:18:53 You can change KERNEL_IMAGEDEST Jan 26 16:19:54 right now KERNEL_IMAGEDEST = "boot" Jan 26 16:20:14 maybe KERNEL_IMAGEDEST_ji_machine = "newplace" Jan 26 16:20:23 yes I did, and that did work but in addition to install -d /${KERNEL_IMAGEDEST} there is Jan 26 16:20:37 install -d /boot also in kernel_do_install Jan 26 16:21:02 your right Jan 26 16:21:06 so it writes the image at boot also along with second location Jan 26 16:21:17 maybe kernel_do_install_append () Jan 26 16:22:00 doesn't that just create the /boot directory? Jan 26 16:22:11 It doesn't actually install anything Jan 26 16:22:41 yes but there are commands to write the image below install so those does it actually Jan 26 16:23:44 Can you manipulate ${D} ? Jan 26 16:23:50 never done that Jan 26 16:24:25 I feel I have to stop kernel_do_install from executing Jan 26 16:24:39 will look at ${D} Jan 26 16:25:06 Example is in mnci-ramses_2.4.21-rmk2-pxa1.bb Jan 26 16:25:14 example of override that is Jan 26 16:30:47 atl1971, thanks, I was looking for something exactly like this, even though I have tried it but prolly I messed up somewhere Jan 26 16:30:55 will try again, thanks! Jan 26 16:41:09 Anyone using an external toolchain (like Angstrom SDK) to produce a root fs? Jan 26 16:41:18 I need an example to follow. Jan 26 16:42:14 Can I just set TOOLCHAIN_TYPE in my machine.conf and it will work? Jan 26 17:05:01 03Koen Kooi  07org.openembedded.dev * r98cf2eed5d 10openembedded.git/recipes/opencv/ (opencv-samples_2.2.bb opencv_2.2.bb): Jan 26 17:05:01 opencv,opencv-samples: add 2.2 Jan 26 17:05:01 Signed-off-by: Koen Kooi Jan 26 17:06:33 hi, question in relation to (http://www.mail-archive.com/openembedded-devel@lists.openembedded.org/msg01875.html) about updating the packages list - is it required to manually run the package_update_index_ipk task to update the list? Jan 26 17:24:03 gm Jan 26 17:28:48 hi eFfeM Jan 26 17:29:16 Feel like you're up for reviewing / ack'ing kergoth's changes to move stagemanager-native / provide install ourselves and drop coreutils-native ? Jan 26 17:29:29 * khem yawns Jan 26 17:29:59 Tartarus: kergoth's changes look good, they should be pushed Jan 26 17:30:13 only one ack ;) Jan 26 17:30:26 (gotta love our process) Jan 26 17:31:03 Tartarus: was that reviewing Q to me? Actually I did post a stagemanager issue that I did get on the autobuilder today Jan 26 17:31:29 eFfeM: those issues could be fixed with kergoth's patches Jan 26 17:31:36 thats why you should review them :) Jan 26 17:31:58 ah ok, I'll have a peek at them, probably slide them in and the QA one from Koen and kick off a build later tonight Jan 26 17:32:11 first need to verify that tzdata also builds, want to push that first Jan 26 17:32:26 denix: aa ok, thanks :) Jan 26 17:36:19 hm, something strange is happening, I kicked off bb 5 mins or so ago but it is doing nothing (after printing the config), top tells it is not doing anything. Is this a known issue? ideas to fix ?? Jan 26 17:37:08 did you try to bitbake somehting which is in assume provided? Jan 26 17:37:11 iirc thats the behavior if you do that Jan 26 17:37:21 (explicitly bitbake it, on the commandline) Jan 26 17:38:53 kergoth: I tried bitbake tzdata don't think it is provided, btw I am using the cmd line Jan 26 17:40:05 kergoth, get khem to stage the patches in his git and accidentally push them :) Jan 26 17:41:05 i will try the staging patches later tonight Jan 26 17:41:13 did a btibake -cclean tzdata and that works, now retrying Jan 26 17:41:16 kergoth: build fired using your 4 patches, result in 1 hours or 2 Jan 26 17:41:21 kergoth: now it works Jan 26 17:45:46 03Frans Meulenbroeks  07org.openembedded.dev * rcebce0e782 10openembedded.git/recipes/tzdata/tzdata_2010o.bb: Jan 26 17:45:46 tzdata: updated to 2011a Jan 26 17:45:46 Signed-off-by: Frans Meulenbroeks Jan 26 17:49:48 Crofton: they are already staged btw. :) if need be I can push Jan 26 17:50:34 eFfeM: weird. Jan 26 17:50:41 no idea what would cause taht behavior Jan 26 17:51:16 i should have emailed those coreutils/stagemanager patches ages ago, they'd been sitting on a branch for quite some time, just forgot to send to the list Jan 26 17:51:19 heh Jan 26 17:51:25 * kergoth has too many branches in too many repositories Jan 26 17:52:52 yeah it would be nices if once could just push one ref Jan 26 17:52:59 khem: one can Jan 26 17:53:01 or git push Jan 26 17:53:10 'git push :' Jan 26 17:53:25 or if it's already there then just 'git push ' Jan 26 17:53:31 andyjpb: that pushes the whole set of commits Jan 26 17:53:39 hi Jan 26 17:53:54 I want to push only say commit foo but keep commit bar Jan 26 17:53:55 khem: ah... you're confused about the way git works then Jan 26 17:54:01 I know Jan 26 17:54:04 I am not confused Jan 26 17:54:06 if the commits are independent then you can Jan 26 17:54:08 khem: rebase -i origin/master, rearrange them so foo is applied before bar Jan 26 17:54:12 otherwise, semantically, you can't Jan 26 17:54:13 then git push origin :master Jan 26 17:54:17 i do it all the time Jan 26 17:54:21 go you can cherrypick the commit and then push it Jan 26 17:54:37 yeah of course you can use a staging branch for it Jan 26 17:54:51 kergoth: I use staging branches Jan 26 17:54:53 mostly Jan 26 17:55:02 but I have more than 1 Jan 26 17:55:11 and sometimes I confuse myself Jan 26 17:55:48 if it's independent then you can have a staging branch... if you need to share that bit of dev in isolation then you must have a staging branch Jan 26 17:55:56 kergoth: git push origin :master will that push commits after foo excluding foo ? Jan 26 17:58:31 that pushes the commits between foo and master Jan 26 17:58:37 so all commits *from master to foo* Jan 26 17:58:56 which is why i said to use rebase -i to rearrange them so that master to foo is just the bits you want to push, and teh bits after foo you don't want to push yet Jan 26 17:59:01 but maybe a staging branch is m ore convenient :) Jan 26 17:59:06 * kergoth does both, depending on the case Jan 26 18:00:05 ah master .... foo ... bar so it will push master...foo including foo Jan 26 18:00:37 hmmmI should try that Jan 26 18:06:30 git log foo..master will tell you what git push origin foo:master would push Jan 26 18:06:31 :) Jan 26 18:06:41 erm, master..foo i mean Jan 26 18:06:52 gotta love git Jan 26 18:08:46 kergoth:your patches do not apply, see http://www.pastie.org/1499891 Jan 26 18:08:54 this is on git head Jan 26 18:09:56 will update Jan 26 18:10:03 think i rebased locally, just need to put them somewhere Jan 26 18:11:55 ok, great Jan 26 18:12:45 eFfeM: https://github.com/kergoth/openembedded/commits/tasks-reduction Jan 26 18:12:49 btw who is making a mess of our patchwork, several unapplied patches seem to have moved to accepted or is this default in the new patchwork if a patch gets an ack; eg was peeking at this http://patches.openembedded.org/patch/323/ Jan 26 18:12:53 git pull git://github.com/kergoth/openembedded tasks-reduction Jan 26 18:13:02 will grab those commits adn merge with your current branch Jan 26 18:13:38 kergoth: great, but I see only 3 patches and you posted 4 is there a merge ? Jan 26 18:13:41 or is one missing Jan 26 18:14:11 i merged the add of the -D arg to install in with the commit that created the install script Jan 26 18:14:16 figured it wasn't worth keeping separate Jan 26 18:14:28 hey easy question I think...is it necessary to run do_package_update_index_ipk manually to get a package into the Packages list or should this be happening automatically? Jan 26 18:14:28 ah ok Jan 26 18:15:28 if I want to use my own toolchain to build qemuppc target, is it enough to overwrite TOOLCHAIN_PATH, HOST_PREFIX and TARGET_PREFIX? Jan 26 18:16:00 no comments on that 'rant about SECTION clutter' ? Jan 26 18:17:06 khem, i've just sent some additions to distro features. I think we should document them somewhere Jan 26 18:20:47 kergoth: applied, test has been started Jan 26 18:24:31 03Koen Kooi  07org.openembedded.dev * r9f1ba8973f 10openembedded.git/recipes/gstreamer/ (2 files in 2 dirs): Jan 26 18:24:31 gst-plugins-bad: add 0.10.21 Jan 26 18:24:31 Signed-off-by: Koen Kooi Jan 26 18:24:43 03Koen Kooi  07org.openembedded.dev * rc8c2f688a1 10openembedded.git/recipes/gstreamer/gst-plugins-base_0.10.32.bb: Jan 26 18:24:43 gst-plugins-base: add 0.10.32 Jan 26 18:24:43 Signed-off-by: Koen Kooi Jan 26 18:24:48 03Koen Kooi  07org.openembedded.dev * r9b3509fa36 10openembedded.git/recipes/gstreamer/gstreamer_0.10.32.bb: Jan 26 18:24:48 gstreamer: add 0.10.32 Jan 26 18:24:48 Signed-off-by: Koen Kooi Jan 26 18:26:35 eFfeM: thanks Jan 26 18:29:33 http://busybox.net/~aldot/oe/0002-RFC-initramfs-qemu-image-new-smallish-image.patch Jan 26 18:29:43 misham: try SDK_PATH=path TOOLCHAIN_PATH=path TARGET_VENDOR="-none" TOOLCHAIN_TYPE="external" TOOLCHAIN_BRAND="generic" (for an angstromtoolchain) I am having some success there. Jan 26 18:30:14 misham: which toolchain are you using? Jan 26 18:30:23 misham: you should look into external-toolchain Jan 26 18:30:30 (doesn't apply since it depends on some of the other patches just sent to the ML). Herein i'm introducing a private oe_contains(). See explanation in the hunk for b/recipes/images/initramfs-qemu-image.bb Jan 26 18:31:00 blindvt: there is a mips64 patch needed for getting mips64/nptl uclibc compiled even though it does not boot Jan 26 18:31:16 blindvt: I think we should apply that before .32 release Jan 26 18:31:20 misham: external-toolchain-generic.bb in recipes is a good starting place Jan 26 18:31:38 (that comment reads -- yes, sorry for the spam but it's just 3 lines) Jan 26 18:31:39 +# base_contains requires the full range of checkvalues to be _found_ so adding Jan 26 18:31:39 +# the compressed variants here doesn't work out of the box Jan 26 18:32:09 blindvt: http://git.openembedded.org/cgit.cgi/openembedded/tree/recipes/uclibc/uclibc-git/mips64-no-waitpid.patch Jan 26 18:32:49 when i wrote this, not a single user in OE used lists for base_contains. So my question is if we could use that (or something like) oe_contains instead of the current base_contains() ? Jan 26 18:33:04 atl1971: if you succeed in making external toolchain work can you do a favor by documenting all the steps you needed to get there ? may be in oe wiki or even a simple email to mailing list ? Jan 26 18:35:50 all the base_ functions have already been moved to oe.utils, the base functions are just wrappers now Jan 26 18:36:24 kergoth: Did you think more about target override option ? Jan 26 18:36:34 khem, does OE by now handle the different MIPS ABIs correctly? Back when i suggested the uclibc configury there was no machine/distro feature where i could properly set the abi since i only saw triplets and they didn't contain all required info. Anyway. isn't o32 dead and is there no way to impl waitpid on o32? Jan 26 18:36:50 khem: it proved to be a royal pain, and i haven't gone back to work on it since :P Jan 26 18:36:56 kergoth, yes, i'm talking about the ones in oe.utils Jan 26 18:37:17 * kergoth nods Jan 26 18:37:18 blindvt: for mips64 oe just supports n64 Jan 26 18:37:50 blindvt: we dont support multi-abis in single build atm Jan 26 18:38:10 kergoth, i can't say Jan 26 18:38:13 IMAGE_INSTALL += "${@oe_contains('IMAGE_FSTYPES', ['ext3', 'ext3.gz', 'ext3.bz2', 'ext3.xz'], 'initramfs-module-ext3', '', d)}" Jan 26 18:38:30 kergoth, with the current impl of whatever_contains() in oe.utils Jan 26 18:38:54 you want it to see the list and treat it as any() not all()? Jan 26 18:39:01 yes Jan 26 18:39:07 makes sense Jan 26 18:39:17 i can't imagine the other would be very useful at all Jan 26 18:39:24 ideally we should have options to select abi's and OE would magically spin the versions of libraries for you in /lib or lib64 or lib32 Jan 26 18:39:26 all() doesn't make sense Jan 26 18:39:31 but that easier said than done Jan 26 18:40:07 blindvt: what do u think of this patch for mips64 ? Jan 26 18:41:54 khem, the common/Makefile.in should use CSRC-filterout-y or something such since it's getting convoluted, but other than that that o32 ABI patch sounds fine, without having looked how to provide waitpid() in another way, sure. Please install. Jan 26 18:42:37 03Khem Raj  07master * r1a9a5b145c 10openembedded.git/recipes/samba/samba_3.2.15.bb: (log message trimmed) Jan 26 18:42:37 samba: Fix build issue on uclibc/arm/thumb Jan 26 18:42:38 We get following errors as of now. This patch fixes those Jan 26 18:42:38 /scratch/oe/sysroots/x86_64-linux/usr/armv5te/lib/gcc/arm-oe-linux-uclibceabi/4.5.3/../../../../arm-oe-linux-uclibceabi/bin/ld: Jan 26 18:42:38 /scratch/oe/sysroots/armv5te-oe-linux-uclibceabi/usr/lib//uclibc_nonshared.a(atexit.os)(.text+0x12): Jan 26 18:42:38 unresolvable R_ARM_THM_CALL relocation against symbol `__cxa_atexit' Jan 26 18:42:39 Signed-off-by: Khem Raj Jan 26 18:44:51 khem, if we only support n64 then it's perfectly fine with me (for OE), sure. But we should reflect that fact somewhere, either in uclibc-config.inc or in some docs Jan 26 18:46:06 blindvt: yes indeed thus far its tribal knowledge Jan 26 18:46:30 khem, speaking of docs. Where can we put a list of DISTRO_FEATURES that are commonly used? As you can see on the ML, i've added^Wsuggested to add xattr for the posix extended attributes. We should _document_ these somewhere so the same stuff is used throughout libcs Jan 26 18:46:32 blindvt: re. fakeroot please port upgrades for those remaining distros too Jan 26 18:47:05 blindvt: I think OE manual is the best bet Jan 26 18:47:11 khem, am i blind? which upgrades for which distros? Jan 26 18:48:54 khem, oh, you mean oplinux and chinook-compat, maemo5-compat Jan 26 18:49:18 * khem sees blind bt Jan 26 18:50:57 2/9 os.path I think there was oe.path too can we use that instead ? Jan 26 18:51:35 khem, to be honest, i've never heard about any of those. If they for whatever obscure reasons pin non-existing or ancient stuff then they just clutter up. I can just remove the pins but i honestly don't have time to listen to anybody yelling at me because i remove pins so i better leave my hands off any pin-removal. Not worth my time Jan 26 18:52:34 blindvt: yes they are unmaintained Jan 26 18:52:41 but we can keep them uptodate Jan 26 18:53:03 khem, if we're about to play the game of removing unwanted ancient stuff then i'm certainly one of the first to fuel up my chainsaw |) Jan 26 18:53:16 blindvt: 3/9 why do u use rm -f ; rmdir why not rm -rf Jan 26 18:53:40 with oe-core we will try to keep stuff maintained Jan 26 18:54:37 blindvt: 4/9 looks ok you could also unpin glibc versions Jan 26 18:55:09 5/9 looks ok Jan 26 18:55:14 khem, re 3/9: The whole IMAGE generation is a pile of crap. We genext2fs more often than i can count without stuttering Jan 26 18:55:45 6/9 looks ok Jan 26 18:57:48 khem, there should be an .ext2 task and we can make ext{3,4}{,.{Z,gz,bz2,xz}} out of that basic .ext2 Jan 26 18:58:44 7/9 looks ok Jan 26 18:59:03 khem, re 2/9 oe.path has no basename, so no Jan 26 18:59:15 ok Jan 26 18:59:18 8/9 is ok Jan 26 19:00:05 9/9 is ok too Jan 26 19:01:04 provided distros use the DISTRO_FEATURES Jan 26 19:02:11 khem, re 4/9: i'll not unpin glibc versions. I'm inheriting this from micro-uclibc so that pin is not my problem, really. I'm sure they know what they're doing Jan 26 19:02:30 ~glibc Jan 26 19:02:31 i guess glibc is libdrepper is a bloated pile you wouldn't want to use since there is uClibc which in turn is fabulous Jan 26 19:02:57 :P Jan 26 19:05:21 hmmm Jan 26 19:05:30 is there a quip for eglibc too Jan 26 19:06:15 anyone know if there's a document outlining the oe testing/release process? Jan 26 19:06:24 URL would be ideal Jan 26 19:07:11 ibot, eglibc is an attempt to de-drep libdrepper a.k.a glibc Jan 26 19:07:11 okay, blindvt Jan 26 19:07:30 khem, now there is Jan 26 19:09:40 khem, anyway, let's try to get at least binutils from sane-toolchain for a start Jan 26 19:10:01 k Jan 26 19:10:41 jconnolly: http://www.openembedded.org/index.php/Release-2010.12 lists the test combos Jan 26 19:10:51 khem: thanks a lot Jan 26 19:11:33 it does not have process per say but it does have info on how it was done Jan 26 19:13:04 evening Jan 26 19:13:37 khem, re 9/9 Our current util-linux handling is really suboptimal. We have that ancient package that doesn't even compile, let alone install (at least on micro). And there is of course busybox (adding PROVIDES is in my queue and works fine for me since a couple of months) and util-linux-ng. I for one would just blow away the old util-linux, not sure how controversial that is though so i'll leave my hands off it and instead make the bb pa Jan 26 19:13:37 ckage reflect reality Jan 26 19:16:57 atl1971: thanks, I'll take a look at those Jan 26 19:19:04 yeah util-linux should be removed Jan 26 19:21:57 khem, can you please elaborate what the potential problem with that qemux86_64 machine was/would be? You mentioned concerns about overlays because of the '_' or something like that? Jan 26 19:22:30 khem, ( for reference i asked you why we didn't have something like http://busybox.net/~aldot/oe/qemux86_64-nocona.tar.bz2 ) Jan 26 19:26:05 blindvt: there was issue calling the machine x86_64 Jan 26 19:26:14 since _ is out override separator Jan 26 19:26:28 if we called it amd64 it would be better Jan 26 19:26:42 but then we have to map the autconf triplets Jan 26 19:27:05 khem, short of using another char as separator we could always name it amd64, good old debian way :) Jan 26 19:27:24 yeah Jan 26 19:27:27 indeed. hmz Jan 26 19:28:12 it would be easier to use another overlay separator i guess (wasn't it ':' anyway or am i mixing things up) Jan 26 19:31:19 its for overrides not overlay Jan 26 19:31:31 CFLAGS_arm CFLAGS_x86 thing Jan 26 19:31:46 look at poky how did they manage it Jan 26 19:31:49 I have nt looked Jan 26 19:31:56 khem, ah! your previous suggestion was here: http://patchwork.openembedded.org/patch/1402/ Jan 26 19:32:16 So Jan 26 19:32:28 Why can't we just call it qemux8664 ? or x86-64 Jan 26 19:32:32 since _ is special Jan 26 19:35:14 Tartarus: read that thread Jan 26 19:35:24 Tartarus: I think we should go with amd64 Jan 26 19:36:02 x86-64 is fine too but why have - when we dont need it Jan 26 19:36:17 and amd64 is a familiar term Jan 26 19:36:48 Tartarus, the concern is the name of the machine $(think uname -m) it seems. CFLAGS_append_x86_64=-mtune=nocona\nCFLAGS_append_x86=-mtune=i386. bitbake tokenizes on _ so x86-64 or amd64 would work but x86_64 would not without tricking a bit Jan 26 19:37:04 right Jan 26 19:37:27 so if we cant use x86_64 then x86-64 or amd64 are same Jan 26 19:37:35 and amd64 is more familiar Jan 26 19:37:41 I guess the bit there used to be about well if we have amd64 what do we call the intel ones isn't a political thing anymore? Jan 26 19:38:00 heh Jan 26 19:38:00 unpleasant. I don't really have time to look into it Jan 26 19:38:29 why didnt intel call it x64 to start with Jan 26 19:39:09 x8664 will sounds like an office phone extention Jan 26 19:39:41 Tartarus, debian uses (or at least used to use) amd64 for x86_64 packages, so i guess it shouldn't matter much (except to intel employees and company marketing, perhaps) Jan 26 19:39:43 bbl Jan 26 19:39:45 is it possible to eg set a DP for a distro Jan 26 19:40:02 khem since we have perl 5.10.1 in minimal, I get Jan 26 19:40:03 ERROR: Multiple .bb files are due to be built which each provide libmodule-build-perl-native (/home/hudson/jobs/FM_TEST/workspace/openembedded/recipes/perl/perl-native_5.10.1.bb /home/hudson/jobs/FM_TEST/workspace/openembedded/recipes/perl/libmodule-build-perl-native_0.2805.bb). Jan 26 19:40:30 if I recall correctly the module is not needed or named differnetly in 5.10.1 Jan 26 19:41:13 actually this is the first line: NOTE: multiple providers are available for libmodule-build-perl-native (libmodule-build-perl-native, perl-native) Jan 26 19:41:14 perl-native 5.10.1 needs a provides of some sort added? Jan 26 19:41:40 Tartarus: the provides is there guess we need to set a preference or so Jan 26 19:42:24 That is funny, heh Jan 26 19:42:54 maybe a PREF_PROVIDER_libmodule... = "perl-native" in minmal.conf and probably some other distros you may or may not feel comfortable submitting a patch to Jan 26 19:43:30 Tartarus: will give it a try Jan 26 19:44:06 after the current build finishes Jan 26 19:48:30 Tartarus, i don't understand your question Re: 'simplify "if image" condition'. If nothing sets INITRAMFS_TASK then ?="" makes bitbake set it empty? Jan 26 19:49:36 Tartarus, $ python2.6 -c "print any([u'','',None,False,[],{},()])" Jan 26 19:49:36 False Jan 26 19:49:51 ok Jan 26 19:50:12 So you're just saying that given normal python mechanics we don't need all of those other tests Jan 26 19:50:25 exactly Jan 26 19:50:42 ok Jan 26 19:50:46 i'll follow up then, thanks Jan 26 19:51:47 same for that opkg-utils yocto patch that excepts several types in a row instead of just 'except (whateverException, anotherExceptionClass):' that khem sent to the list recently. Jan 26 20:04:15 btw.. is there some magical way to add ACKs when pushing instead of respinning every single patch with the respective ACKs added manually? Jan 26 20:08:04 Good question, git rebase -i ... and setting them each to edit or whatever it's called is slightly better Jan 26 20:08:07 or were you doing that already? Jan 26 20:10:06 Tartarus, no i wasn't. But that's a good idea, thanks! Jan 26 20:12:30 I want to set PR using a shell command, is that allowed in a bb file, like this PR = `pwd` Jan 26 20:47:49 blindvt: yeah if someone adds acks Jan 26 20:47:56 then you can use patchwork Jan 26 20:48:02 on a staging branch Jan 26 20:48:20 if not then you have to do interactive rebasing Jan 26 20:48:28 and ammend the commits as you go Jan 26 20:48:37 atl1971, i think that doesn't work. You could try = "{@os.getcwd()}" or something like that but i think you really shouldn't Jan 26 20:49:14 and PR should be sensible thing Jan 26 20:49:35 preferably rX Jan 26 20:55:41 I see examples of using the gitrev{SRCREV} and would like to do that, but use svn and I couldn't get the svn fetcher to work with proto=ssh Jan 26 20:56:56 so I really want to do " svn info svn+ssh://server/trunk | grep -i revision | cut -d" " -f2" which I am ashamed to even type in your presence Jan 26 20:57:07 khem, do you plan to ACK the 9 series on the list? AFAICS you acked all of them as-is #here already, yes? Jan 26 20:58:16 re Jan 26 20:58:38 atl1971, sounds like you want AUTOREV Jan 26 20:58:59 woglinde, hi Jan 26 20:59:26 jo blindvt Jan 26 21:00:20 khem, docs/usermanual/chapters/features.xml sounds like a natural place to document distro feature tokens, no? Jan 26 21:00:50 * Jay7 is looking into oestats code.. Jan 26 21:01:04 blindvt: I acked few of them Jan 26 21:01:15 hmz. embarassing, but how do i build the docs Jan 26 21:01:23 it shouldn't be hard to add all fields we are needed to do auto-report for testing branch Jan 26 21:01:27 I asked to restain one of IRml Jan 26 21:01:31 blindvt: yes that is what I want, but I couldn't get the SRC_URI to work Jan 26 21:01:48 but I need kergoth's advice Jan 26 21:02:25 or may be RP__'s :) Jan 26 21:02:46 RP__: do we have bitbake version info in bb class? Jan 26 21:03:05 atl1971, i'm not sure about svn+ssh nor git+ssh. We don't have your keys and stuff lives supposedly in open repos, sorry Jan 26 21:04:05 may be bb.data.getVar("something") Jan 26 21:07:06 Jay7, there's bb.__version__ but that looks like it's private. If it is, then ask him for a http://paste.debian.net/105748/ Jan 26 21:07:07 anyone know offhand if sbindir is actually /sbin or /usr/sbin ? Jan 26 21:07:15 i have a dhcp-client postinst script that does Jan 26 21:07:16 ln -s ${sbindir}/dhclient ${sbindir}/dhclient3 Jan 26 21:07:33 but: Jan 26 21:07:34 root@bug20:~# cat /usr/lib/ipkg/info/dhcp-client.postinst ln -s /usr/sbin/dhclient /usr/sbin/dhclient3 Jan 26 21:07:41 jconnolly, depends on the distro; micro collapses usr/ into /, the rest uses distinct dirs Jan 26 21:07:53 Jay7: bb.__version__ sounds like the right move Jan 26 21:08:13 blindvt: hmm thanks. I'll do the wrong thing and specify /sbin anyway Jan 26 21:08:17 ;D Jan 26 21:08:26 jconnolly: you might want base_sbindir Jan 26 21:08:51 RP__: much obliged Jan 26 21:09:18 base_sbin = /sbin sbin = /usr/sbin or in the micro case base_sbin == sbin Jan 26 21:09:48 RP__: 10x Jan 26 21:10:12 03Koen Kooi  07org.openembedded.dev * r60911f259a 10openembedded.git/recipes/ffmpeg/ (4 files in 2 dirs): Jan 26 21:10:12 omapfbplay: import 3 patches from git to fix playing weird sized videos Jan 26 21:10:12 Signed-off-by: Koen Kooi Jan 26 21:10:15 blindvt: 10x as well :) Jan 26 21:11:01 Jay7, at your service Jan 26 21:11:46 well.. next thing is user distro.. Jan 26 21:15:03 Jay7, ${DISTRO} ? Jan 26 21:15:18 blindvt: I mean distro on build host Jan 26 21:15:27 something like /etc/*-release Jan 26 21:21:16 well.. better to leave it as env/bb variable Jan 26 21:21:41 like OESTATS_BUILD_DISTRO Jan 26 21:30:03 jconnolly: base_sbindir is /sbin and sbindir is /usr/sbin iirc Jan 26 21:30:40 jconnolly: you can alsways bitbake -e helloworld | grep "^sbindir=" Jan 26 21:31:04 khem: ah yes thanks. sometimes I use #oe as my bitbake -e =/ Jan 26 21:31:09 03Chris Larson  07master * raf5f4debf1 10openembedded.git/ (7 files in 4 dirs): Jan 26 21:31:09 Add install wrapper rather than coreutils-native Jan 26 21:31:09 Signed-off-by: Chris Larson Jan 26 21:31:09 Acked-by: Khem Raj Jan 26 21:31:09 Acked-by: Eric Bénard Jan 26 21:31:10 Acked-by: Bernhard Reutner-Fischer Jan 26 21:31:24 03Chris Larson  07master * r0bf1d00fb4 10openembedded.git/ (6 files in 3 dirs): Jan 26 21:31:24 Move stage-manager-* into bin/ rather than a recipe Jan 26 21:31:24 Signed-off-by: Chris Larson Jan 26 21:31:24 Acked-by: Khem Raj Jan 26 21:31:24 Acked-by: Eric Bénard Jan 26 21:31:29 03Chris Larson  07master * rc1c058fc29 10openembedded.git/ (5 files in 3 dirs): Jan 26 21:31:29 bitbake.conf: include bin dirs from BBPATH in PATH Jan 26 21:31:29 Signed-off-by: Chris Larson Jan 26 21:31:29 Acked-by: Khem Raj Jan 26 21:31:29 Acked-by: Eric Bénard Jan 26 21:43:38 well.. next question Jan 26 21:44:07 kergoth_, RP__: from where can I get name of image that should be built by bb? Jan 26 21:44:25 hm.. wrong question Jan 26 21:48:08 kergoth_: actually just wanted to mention that I am at task 5345 of 9396 and did not encounter issues, but see the patch is just pushed. Jan 26 21:48:54 so no need for an extra ack Jan 26 21:51:00 ERROR: Only one target can be used with the --environment option. Jan 26 21:51:02 heh.. Jan 26 21:52:19 any idea why do_package_update_index_ipk would not be triggered for a recipe? (my package is not being added to the list unless I call this task manually) Jan 26 21:53:28 hbeck: I've noticed here that I have to "bitbake package-index" if I want something in my package list Jan 26 21:54:22 foerster: yeah I saw that recipe, but http://www.mail-archive.com/openembedded-devel@lists.openembedded.org/msg01875.html seems to indicate that it's the "old way" of doing it Jan 26 21:54:49 hbeck: ah. Seems with my current setup, that's the only way it works Jan 26 22:02:54 anyone seen an error when using bitbake that says "Error: No provider" and then lists your .bb file Jan 26 22:04:44 is there a needed ordering for FILES_* ? I have Jan 26 22:04:46 FILES_${PN}-dev += "${libdir}/liblirc_client.so" Jan 26 22:04:46 FILES_${PN} = "${bindir} ${sbindir} ${libdir} ${sysconfdir} ${exec_prefix}/var" Jan 26 22:04:53 but still I get a QA issue on the .so file Jan 26 22:04:55 how to fix ? Jan 26 22:05:25 do I need to reorder things in PACKAGES or do I need to make FILES_${PN} more explicit ? Jan 26 22:10:53 effem hm why you need to alter it at all? Jan 26 22:17:47 03Koen Kooi  07org.openembedded.dev * rd305d98172 10openembedded.git/recipes/ffmpeg/omapfbplay/ (3 files): Jan 26 22:17:47 omapfbplay: fix patch dir Jan 26 22:17:47 Signed-off-by: Koen Kooi Jan 26 22:26:03 woglinde_: because I get a QA error: ERROR: QA Issue with lirc: non -dev package contains symlink .so: lirc path '/work/calamari-oe-linux-gnuspe/lirc-0.8.7-r0/packages-split/lirc/usr/lib/liblirc_client.so' Jan 26 22:36:34 khem: ping Jan 26 22:39:07 nite all Jan 26 22:42:43 Jay7, how do I limit the HD bandwidth that a build takes? I"m getting killed. Jan 26 22:43:12 ka6sox: HDD is trashed? Jan 26 22:44:02 so much so that things have ground to a halt Jan 26 22:44:02 I'm building inside lxc with lowered priority Jan 26 22:44:09 this help sometimes.. Jan 26 22:44:14 okay Jan 26 22:44:30 but sometimes it is trashed too Jan 26 22:44:45 I have 2 VMS that consume 100% of the HD I/O 24/7 when they run. Jan 26 22:44:52 so the others are DEAD. Jan 26 22:44:55 for days Jan 26 22:44:57 there are some settings to look but I haven't looked yet Jan 26 22:45:17 what VM hypervisor is used? Jan 26 22:45:24 xen Jan 26 22:45:53 I"d like to lower their I/O bw for a bit till I can make better arraingements Jan 26 22:47:11 http://xen.1045712.n5.nabble.com/How-to-Allocate-Disk-Bandwidth-among-VMs-td2538448.html Jan 26 22:47:13 check this Jan 26 22:47:31 dm-ioband Jan 26 22:48:43 ya, looks like that might do it...I"ve been like a fireman recently puttin out fires. Jan 26 22:51:37 or just insert 2 more drives and dedicate them for that VM's :) Jan 26 22:57:11 Jay7, I'd love to...can somebody please donate 2ea 256GB SSDs? Jan 26 22:57:43 ka6sox: yocto? ;) Jan 26 22:58:02 Jay7, sure...that would be most helpful. Jan 26 22:58:20 or may be TI :) Jan 26 23:23:59 so im building angstrom_2010.x but in its config file i select prefered kernel to be 2.6.34 in order to get the 2.6.34 kernel right Jan 26 23:26:02 or do you override the config using local.conf Jan 27 01:49:06 what's the proper way to link two recipes ("A" depends on "B") so that installing "A" will automatically cause "B" to be installed? Jan 27 01:49:28 I *think* it is RDEPENDS_A = "B", but this doesn't seem to be working for some odd reason. Jan 27 01:50:06 "B" isn't showing up in the IPKG Depends line, so doesn't get installed. **** ENDING LOGGING AT Thu Jan 27 02:59:57 2011