**** BEGIN LOGGING AT Sun Dec 12 02:59:58 2010 Dec 12 08:13:22 khem js/mediatomb problem is unsolved, still get it on a clean build Dec 12 08:15:17 i also seem to have an issue with gettext but only on the autobuilder (which is ubuntu 10.04/64 bit and not on my home system (opensuse 11.2/32bit) Dec 12 09:29:42 Anyone having problems with OpenSP ? Dec 12 09:29:43 http://build.sakoman.com/public/logs/task/112012.txt Dec 12 09:30:11 .../armv7a-angstrom-linux-gnueabi/opensp-1.5-r2/OpenSP-1.5/sx/XmlOutputEventHandler.cxx:789: undefined reference to `OpenSP::InputSource::get(OpenSP::Messenger&)' Dec 12 09:41:49 It looks like something is wrong with include paths Dec 12 09:47:44 eFfeM: morning Dec 12 09:55:09 hi davidlt Dec 12 09:57:36 eFfeM: had any problems with OpenSP lately? Dec 12 09:58:15 http://lists.linuxtogo.org/pipermail/openembedded-devel/2010-October/025761.html Dec 12 10:00:05 I was using November release of Gumstix, moved to December 3rd and OpenSP doesn't compile even thought there is no changes done. Dec 12 10:14:44 davidlt: saw your log, I have never ever build opensp, not even an idea what it is Dec 12 10:15:17 ask paul or sakoman Dec 12 10:15:20 For other problems (like http://bugs.gentoo.org/show_bug.cgi?id=100814#c23), they probably are related to having Dec 12 10:15:20 compiled opensp with crazy CFLAGS, because opensp wasn't using the same filters Dec 12 10:15:20 used in openjade. For those who still has those problems, please re-sync your Dec 12 10:15:20 portage tree and re-emerge opensp and openjade. Alternatively, please consider Dec 12 10:15:21 using sane CFLAGS only to compile your packages :). Dec 12 10:15:31 I found this one, // oh a bit too many lines Dec 12 10:15:55 And openjade was just updated Dec 12 10:16:30 What`s the best way to check out earlier version of openjade folder? Dec 12 10:16:42 Tried git checkout openjade, still gave me HEAD Dec 12 10:17:08 it should give you that srcrev Dec 12 10:17:30 or you might go for the latest stable Dec 12 10:18:41 does such exist in OE world :) Dec 12 10:18:44 does it work with openjade 1.3.2 ? Dec 12 10:19:01 i meant latest stable openjade Dec 12 10:19:26 It doesn't with latest changes in OE.devel Dec 12 10:19:48 in openjade or opensp Dec 12 10:19:51 I am trying to 2010-10-17 version of openjade to test if openjade changes breaks everything Dec 12 10:19:57 openjade Dec 12 10:20:04 opensp wasn't changed for long time. Dec 12 10:20:26 But looking from comments openjade/opensp compilcation might be tied together Dec 12 10:21:32 you might want to try a version of the openjade recipe from a month ago Dec 12 10:21:43 or chat with grg who did the last 2 commits Dec 12 10:22:01 I am already going back with in time and testing versions Dec 12 10:23:35 in oe you can go back with git checkout that will give you the situation at that point in time Dec 12 10:23:51 but you might also manually try to revert only the openjade changes Dec 12 10:45:37 eFfeM: I took the old files (openjade) from last working files I had, but still the same problem. Dec 12 10:46:39 hm, strange ask grg or sakoman, I have no idea and cannot even try it right now as I have a large build running Dec 12 10:46:54 and some very odd other problems Dec 12 11:45:36 which command will help me download files from linux using serial port Dec 12 11:49:40 rs1: for example, lrz/lsz from lrzsz package Dec 12 11:50:38 thank you. Dec 12 11:50:46 i had forgotten the command name Dec 12 14:06:26 * XorA is too hung over for reading oe-members Dec 12 15:06:00 calling it a day Dec 12 15:29:08 No success in OpenSP problem Dec 12 15:29:20 Even getting older OpenJade files this didn't solve problem. Dec 12 16:41:27 * khem yawns Dec 12 17:02:40 I have some problems with wvdial. According to several internet sources, getcontext() is broken on ARM, and thus wvdial asserts when I use it Dec 12 17:04:04 However, there are patches to get wvdial to work without getcontext Dec 12 17:04:21 What is the best approach if I want to use these patches? Dec 12 17:05:08 what are the patches targeted at? Dec 12 17:05:21 recipe? source? Dec 12 17:05:33 Ah, sorry Dec 12 17:05:40 WvStream I guess Dec 12 17:06:07 hm... Dec 12 17:06:47 There is a git repository available for wvstreams Dec 12 17:06:56 A branch, that is Dec 12 17:07:39 I guess if I could use that git branch instead of the ordinary wvstreams (already present in OE) I would be alright Dec 12 17:08:43 I also see people mention that earlier versions of wvstreams did not have this problem. Dec 12 17:10:37 sorry, I dont know Dec 12 17:11:41 you can try to modify the source between the unpack / patch and the configure phase of bitbake Dec 12 17:12:01 or modify the recipe to use the branch that has the patch Dec 12 17:12:16 If I want to compile a specific version of a recipe when I run bitbake (e.g. 'bitbake m2mecanum-image') should I have 'PREFERRED_VERSION_wvstreams = "4.4.1"' in my local.conf Dec 12 17:12:17 ? Dec 12 17:12:42 I'm probably noobier than you Dec 12 17:12:56 I don't know how to get the syntax to work right Dec 12 17:13:35 So, basically I have 2 possible solutions: 1. Use older wvstreams recipe. 2. Copy wvstreams recipe and change SRC (or similar) to use git repo insted. Dec 12 17:14:36 Sorry if I'm not making not much sense, the lack of sleep this weekend is getting to me Dec 12 17:15:12 those are the options I can think of Dec 12 17:19:26 hm, strange... /usr/share/zoneinfo/zone.tab isn't being installed :/ Dec 12 17:28:06 bluelightning: http://www.sakoman.com/cgi-bin/gitweb.cgi?p=openembedded.git;a=commitdiff;h=6377855244f8426ad10f90bee1bdee0ce73b07f8 Dec 12 17:28:24 I ran into the same issue Dec 12 17:29:08 anyone ever gotten: *** glibc detected *** mkfs.ubifs: double free or corruption (!prev): 0x0000000001c51100 *** Dec 12 17:29:28 during task do_rootfs for an image? Dec 12 17:43:29 risca: getcontext/setcontext are removed from POSIX 2008 so new implementations will never happen and you should either rewrite the code which uses them differently may be using pthreads and locks or mark the package not supported on ARM Dec 12 17:44:30 03Klaus Kurzmann  07master * r0f63dd1021 10openembedded.git/recipes/udev/udev_162.bb: Dec 12 17:44:30 udev_162.bb: actually install rules for nokia900 and bump PR Dec 12 17:44:30 Signed-off-by: Klaus Kurzmann Dec 12 17:44:31 risca: if wvstreams has started to use get/setcontext recently then they are going backward Dec 12 17:44:38 and should be informed Dec 12 17:47:33 bsl: JaMa saw it http://permalink.gmane.org/gmane.linux.drivers.mtd/29178 Dec 12 17:49:26 khem: shit, this step takes like an hour Dec 12 17:49:38 i rebooted my computer so hopefully it wont happen again Dec 12 17:54:44 sakoman: that patch should not be needed Dec 12 17:55:04 sakoman: zone.tab should have been installed via glibc ipks Dec 12 17:57:21 sakoman, any luck with coreutils yesterday? Dec 12 17:57:23 or pulse? Dec 12 17:58:10 Crofton|work: coreutils ? Dec 12 17:58:15 whats the issue Dec 12 18:02:06 god dammit, it happened again Dec 12 18:31:33 khem, That was my impression as well. I hope a more recent version of wvstreams removes the need for getcontext Dec 12 18:36:12 post inst script not running Dec 12 19:52:19 Crofton: i don't think i like the idea of switching to i2c-dev-user.h Dec 12 19:52:29 is this something blessed or an oe-thing? Dec 12 19:58:18 well, over writing the existing header file is a no no Dec 12 19:58:29 that is what another guy on the list did Dec 12 19:59:01 I'mm all for different solutions, that do not overwrite the kernel header Dec 12 20:08:51 ok, but now everything is broken Dec 12 20:09:05 everything that uses libi2c-dev is expecting i2c-dev.h Dec 12 20:09:15 khem: glibc & eglibc have been patched to remove installation of timezone files in favour of tzdata Dec 12 20:09:19 so all userland programs have to look for two headers Dec 12 20:09:29 i don't consider this a better situation than overwriting the header file Dec 12 20:09:49 khem: but tzdata only installs those files for eglibc Dec 12 20:11:16 hmmm Dec 12 20:11:22 what about different -I Dec 12 20:11:23 ? Dec 12 20:11:32 we workarround in the recipe Dec 12 20:11:41 and stage the header in different places Dec 12 20:12:57 can autotools differenciate between the 2 headers? Dec 12 20:13:03 how do other distros handle it? Dec 12 20:13:52 ubuntu uses some redirection mechanism Dec 12 20:14:01 such as? Dec 12 20:14:13 /usr/include/linux/i2c-dev.h Dec 12 20:14:13 package diverts others to: /usr/include/linux/i2c-dev.h.kernel Dec 12 20:14:20 dunno what that means :) Dec 12 20:14:30 but in the end it overwrites i2c-dev Dec 12 20:15:06 the thing is... we should rather check why it hurts Dec 12 20:15:10 it would be great if we could autodetect what header it needs and act accordinly Dec 12 20:15:15 it shouldn't hurt, since it just adds macros Dec 12 20:15:19 such as: Dec 12 20:15:21 err, inline statics Dec 12 20:15:22 inherit i2c Dec 12 20:16:06 but I've no idea at all on how to do the autodetection Dec 12 20:16:50 or maybe Dec 12 20:16:54 inherit i2c-user Dec 12 20:16:58 and inherit i2c-kernel Dec 12 20:17:40 maybe....maybe...ubuntu has a wrapper header? Dec 12 20:17:47 for instance Dec 12 20:18:01 #ifdef KERNEL Dec 12 20:18:22 #include Dec 12 20:18:26 #else Dec 12 20:18:37 #include Dec 12 20:18:40 I'll look Dec 12 20:19:46 I've only that in trisquel, which is based on the previous ubuntu: i2c-dev.h i2c.h Dec 12 20:19:48 hmmm Dec 12 20:20:13 fedora has the kernel one Dec 12 20:20:18 not sure about the other Dec 12 20:20:22 the wrapper header would be the best interface Dec 12 20:20:31 but I'm not sure how to make it work Dec 12 20:20:33 ok Dec 12 20:20:35 the i2c-tools guys should be beaten Dec 12 20:20:39 indeed Dec 12 20:20:48 should we go on their mailing list? Dec 12 20:22:42 Adding `diversion of /usr/include/linux/i2c-dev.h to /usr/include/linux/i2c-dev.h.kernel by libi2c-dev' is a postinstall message Dec 12 20:25:59 I'm confused why the code I have need the kernel one Dec 12 20:26:12 and why i2c-tools would over write it Dec 12 20:26:20 http://www.mjmwired.net/kernel/Documentation/i2c/dev-interface Dec 12 20:30:55 dpkg-divert --package libi2c-dev --divert /usr/include/linux/i2c-dev.h.kernel --rename /usr/include/linux/i2c-dev.h Dec 12 20:31:13 that's the postinstall code from ubuntu and derivated distro Dec 12 20:32:41 http://www.debian-administration.org/articles/118 Dec 12 20:32:54 basically it replaces the files Dec 12 20:34:45 ah ok it's a sort of update-alternatives Dec 12 20:34:52 hmmm Dec 12 20:51:13 the problem is what gets staged Dec 12 21:21:55 hi, does bitbake still requires python-psyco? Dec 12 21:28:51 anarsoul: I don't think it's ever required it has it? AFAIK it is only a recommendation Dec 12 21:30:33 oh, ok Dec 12 21:31:49 bluelightning: for eglibc thats intended change Dec 12 21:32:23 bluelightning: and slowly we should stop using glibc Dec 12 21:32:26 khem: ok, but for glibc zone.tab isn't being installed it would seem Dec 12 21:32:39 unless I'm mistaken, angstrom still uses it Dec 12 21:32:55 2008 yes but 2010 switched to eglibc Dec 12 21:33:26 hmm, so 2008 shouldn't really be used in the dev branch I guess Dec 12 21:34:00 2010 probably is better choice Dec 12 21:34:05 for master Dec 12 21:36:26 master, yes... ok I will change my environment here Dec 12 21:37:32 ok Dec 12 21:51:06 anyone around to help with some OE build issues? Dec 12 21:51:49 trying to build base-image from oe-dev Dec 12 21:51:52 getting: http://pastebin.com/tcdXhRrj Dec 12 21:52:05 does the build system not give you a log anymore? Dec 12 21:52:24 I tried non dev, but it does the same thing Dec 12 22:01:03 lostincake1: check ML for solution Dec 12 22:01:18 iirc, qemu-0.13 does something wrong there Dec 12 22:15:16 ML? Dec 12 22:15:54 oh Dec 12 22:15:55 nm Dec 12 22:16:00 mailing list Dec 12 22:44:04 Has anyone had problems with wvdial not recognizing wvdial.conf? Not even when created with wvdialconf? Dec 13 01:25:07 03Graham Gower  07master * rb0190e9cee 10openembedded.git/recipes/tasks/task-proper-tools.bb: Dec 13 01:25:07 task-proper-tools.bb: Exclude some packages if DISTRO_FEATURES lacks ipv6. Dec 13 01:25:07 Signed-off-by: Graham Gower Dec 13 01:25:07 Acked-by: Khem Raj Dec 13 01:56:08 03Graham Gower  07master * r61c8f6dca9 10openembedded.git/recipes/autoconf/ (2 files in 2 dirs): (log message trimmed) Dec 13 01:56:08 autoconf_2.65.bb: Fix build for DISTRO=micro. Dec 13 01:56:08 Update autoconf-dont-execute-perl.patch for autoconf-2.65. Dec 13 01:56:08 See aa0d1e870c41d3375866c8180c720cd0fc27ffcd. Dec 13 01:56:08 checking for perl... (cached) /bin/perl **** ENDING LOGGING AT Mon Dec 13 02:59:58 2010