**** BEGIN LOGGING AT Fri Oct 08 02:59:57 2010 Oct 08 03:24:38 tharvey: how did ramfs thing go Oct 08 04:42:33 03Martin Jansa  07master * r64538f2527 10openembedded.git/recipes/eglibc/ (eglibc-package.bbclass eglibc.inc): Oct 08 04:42:33 eglibc-package.bbclass: fix missing RPROVIDES for virtual-locale-* Oct 08 04:42:33 Signed-off-by: Martin Jansa Oct 08 05:46:37 03Khem Raj  07master * r79d813ec2e 10openembedded.git/recipes/libvisual/ (libvisual-0.4.0/no-libdir.patch libvisual_0.4.0.bb): Oct 08 05:46:37 libvisual_0.4.0.bb: passing usual ${libdir} in LIBS is not required Oct 08 05:46:37 Signed-off-by: Khem Raj Oct 08 06:30:11 gm Oct 08 07:52:40 03Martin Jansa  07master * r19a33cdd64 10openembedded.git/recipes/freesmartphone/cornucopia.inc: freesmartphone: bump cornucopia SRCREV to rebuild with newer vala with fixed dbus-glib servers Oct 08 07:52:51 03Martin Jansa  07master * r177477f842 10openembedded.git/recipes/openmoko-3rdparty/mokosuite2_git.bb: Oct 08 07:52:51 mokosuite2: bump SRCREV a bit for fixed memory leaks Oct 08 07:52:51 Signed-off-by: Martin Jansa Oct 08 07:52:52 03Martin Jansa  07master * r2bfa042c74 10openembedded.git/recipes/freesmartphone/libfreesmartphone-glib_git.bb: Oct 08 07:52:52 libfreesmartphone-glib: bump SRCREV for fixed memory leak Oct 08 07:52:52 Signed-off-by: Martin Jansa Oct 08 07:57:17 nirnung Oct 08 08:15:39 good morning Oct 08 08:40:29 JaMa: (or anyone else) can you say something about the RDEPENDS class patches? if ok, I would like to push them before the testing branch is updated Oct 08 08:44:59 root Oct 08 08:45:37 eFfeM_work: you have my ack Oct 08 08:45:53 ok, thanks, will finished it later today Oct 08 08:46:09 eFfeM_work: I'm rebuilding from scratch with those (to test them) but because of also testing new libtool I won't have an image soon Oct 08 08:46:55 eFfeM_work: if you replace also those new recipes not using _${PN} (except ttf.inc) then also Acked-by.. Oct 08 08:47:46 ok, will try to do that, not sure whether I will make it in time (not sure at what time cbrake will update the testing branch) Oct 08 09:09:40 03Mario Schuknecht  07org.openembedded.dev * r239730044b 10openembedded.git/recipes/linux/ (linux-2.6.24/hipox/siocoutqsnd.patch linux_2.6.24.bb): Oct 08 09:09:41 linux-2.6.24: add additional tcp ioctl (for hipox machine only) Oct 08 09:09:41 Signed-off-by: Mario Schuknecht Oct 08 09:09:41 Acked-by: Steffen Sledz Oct 08 09:33:59 Is there a standard way that after doing a full build you can know which recipes and files you aren't using such that you could remove them to speed up the build process? Oct 08 09:37:24 loadammo: no Oct 08 09:39:55 bited! Oct 08 09:39:58 er, noted! Oct 08 09:41:32 loadammo: i would love to see a tool which i could run with the packages I want to build and get a list of all files (including inc files, required bb files etc) that are needed, but it is not there Oct 08 09:41:37 don't have enough experience to build it Oct 08 09:46:30 03Frans Meulenbroeks  07org.openembedded.dev * rc1a16c92ab 10openembedded.git/classes/rubyextension.bbclass: Oct 08 09:46:30 rubyextension.bbclass: RDEPENDS -> RDEPENDS_${PN} Oct 08 09:46:30 Signed-off-by: Frans Meulenbroeks Oct 08 09:46:30 Acked-by: Chris Larson Oct 08 09:46:30 Acked-by: Martin Jansa Oct 08 09:48:36 03Frans Meulenbroeks  07org.openembedded.dev * r69d8a687b7 10openembedded.git/classes/distutils-base.bbclass: Oct 08 09:48:36 distutils-base.bbclass: RDEPENDS -> RDEPENDS_${PN} Oct 08 09:48:36 Signed-off-by: Frans Meulenbroeks Oct 08 09:48:36 Acked-by: Chris Larson Oct 08 09:48:36 Acked-by: Martin Jansa Oct 08 09:49:12 03Frans Meulenbroeks  07org.openembedded.dev * r43d59cb7fe 10openembedded.git/classes/image.bbclass: Oct 08 09:49:12 mage.bbclass: RDEPENDS -> RDEPENDS_${PN} Oct 08 09:49:12 Signed-off-by: Frans Meulenbroeks Oct 08 09:49:12 Acked-by: Chris Larson Oct 08 09:49:12 Acked-by: Martin Jansa Oct 08 10:00:23 JaMa: can you say somethign about this: Oct 08 10:00:24 nogravity/nogravity_2.0.bb:RRECOMMENDS+${PN} = "${PN}-data" Oct 08 10:10:43 03Frans Meulenbroeks  07org.openembedded.dev * r110907784f 10openembedded.git/recipes/wesnoth/wesnoth-wvga_1.8.4.bb: Oct 08 10:10:43 wesnoth-wvga_1.8.4.bb: RCONFLICTS -> RCONFLICTS_${PN} Oct 08 10:10:43 Signed-off-by: Frans Meulenbroeks Oct 08 10:10:43 Acked-by: Martin Jansa Oct 08 10:10:43 03Frans Meulenbroeks  07org.openembedded.dev * rb41f8c74df 10openembedded.git/recipes/wesnoth/wesnoth_1.8.4.bb: Oct 08 10:10:43 wesnoth_1.8.4.bb: RCONFLICTS -> RCONFLICTS_${PN} Oct 08 10:10:44 Signed-off-by: Frans Meulenbroeks Oct 08 10:10:44 Acked-by: Martin Jansa Oct 08 10:10:45 03Frans Meulenbroeks  07org.openembedded.dev * rd533262dfe 10openembedded.git/recipes/tzdata/tzdata.inc: Oct 08 10:10:46 tzdata.inc: RCONFLICTS -> RCONFLICTS_${PN} Oct 08 10:10:46 Signed-off-by: Frans Meulenbroeks Oct 08 10:10:47 Acked-by: Martin Jansa Oct 08 10:14:49 03Frans Meulenbroeks  07org.openembedded.dev * rf3c94c14dd 10openembedded.git/recipes/opie-mediaplayer2/opie-mediaplayer2-skin-default-landscape_1.2.3.bb: Oct 08 10:14:49 opie-mediaplayer2/opie-mediaplayer2-skin-default-landscape_1.2.3.bb: RPROVIDES -> RPROVIDES_${PN} Oct 08 10:14:49 Signed-off-by: Frans Meulenbroeks Oct 08 10:14:49 Acked-by: Martin Jansa Oct 08 10:14:50 03Frans Meulenbroeks  07org.openembedded.dev * r06109ef29e 10openembedded.git/recipes/bootstream/bootstream-chumby-falconwing_1.bb: Oct 08 10:14:50 bootstream/bootstream-chumby-falconwing_1.bb: RPROVIDES -> RPROVIDES_${PN} Oct 08 10:14:51 Signed-off-by: Frans Meulenbroeks Oct 08 10:14:51 Acked-by: Martin Jansa Oct 08 10:14:52 03Frans Meulenbroeks  07org.openembedded.dev * r48adadc806 10openembedded.git/recipes/opie-mediaplayer2/opie-mediaplayer2-skin-default-landscape_1.2.4.bb: Oct 08 10:14:52 opie-mediaplayer2/opie-mediaplayer2-skin-default-landscape_1.2.4.bb: RPROVIDES -> RPROVIDES_${PN} Oct 08 10:14:53 Signed-off-by: Frans Meulenbroeks Oct 08 10:14:53 Acked-by: Martin Jansa Oct 08 10:33:01 eFfeM_work: also replace with _ imho just typo Oct 08 10:33:11 yes, just decided to do that Oct 08 10:33:16 03Frans Meulenbroeks  07org.openembedded.dev * r4e99727d78 10openembedded.git/recipes/opie-reader/uqtreader_cvs.bb: Oct 08 10:33:16 opie-reader/uqtreader_cvs.bb: RRECOMMENDS -> RRECOMMENDS_${PN} Oct 08 10:33:16 Signed-off-by: Frans Meulenbroeks Oct 08 10:33:16 Acked-by: Martin Jansa Oct 08 10:33:16 03Frans Meulenbroeks  07org.openembedded.dev * r5c0d82a4f7 10openembedded.git/recipes/devilspie/devilspie_0.22.bb: Oct 08 10:33:17 devilspie/devilspie_0.22.bb: RDEPENDS -> RDEPENDS_${PN} Oct 08 10:33:18 Signed-off-by: Frans Meulenbroeks Oct 08 10:33:18 Acked-by: Martin Jansa Oct 08 10:33:19 03Frans Meulenbroeks  07org.openembedded.dev * rf4ef8e04ff 10openembedded.git/recipes/mokomaze/mokomaze_0.5.5.bb: Oct 08 10:33:19 mokomaze/mokomaze_0.5.5.bb: RDEPENDS -> RDEPENDS_${PN} Oct 08 10:33:20 Signed-off-by: Frans Meulenbroeks Oct 08 10:33:20 Acked-by: Martin Jansa Oct 08 10:33:21 03Frans Meulenbroeks  07org.openembedded.dev * r72302e5fea 10openembedded.git/recipes/xfce-base/xfce4-mixer_4.6.1.bb: Oct 08 10:33:21 xfce-base/xfce4-mixer_4.6.1.bb: RDEPENDS -> RDEPENDS_${PN} Oct 08 10:33:22 Signed-off-by: Frans Meulenbroeks Oct 08 10:33:22 Acked-by: Martin Jansa Oct 08 10:33:35 03Frans Meulenbroeks  07org.openembedded.dev * r14a114723a 10openembedded.git/recipes/pidgin/msn-pecan_git.bb: Oct 08 10:33:35 pidgin/msn-pecan_git.bb: RDEPENDS -> RDEPENDS_${PN} Oct 08 10:33:35 Signed-off-by: Frans Meulenbroeks Oct 08 10:33:35 Acked-by: Martin Jansa Oct 08 10:33:35 03Frans Meulenbroeks  07org.openembedded.dev * re2a80b0668 10openembedded.git/recipes/libaio/libaio-oracle_0.3.0.bb: Oct 08 10:33:36 libaio/libaio-oracle_0.3.0.bb: RDEPENDS -> RDEPENDS_${PN} Oct 08 10:33:36 Signed-off-by: Frans Meulenbroeks Oct 08 10:33:37 Acked-by: Martin Jansa Oct 08 10:33:37 03Frans Meulenbroeks  07org.openembedded.dev * r9d333119d4 10openembedded.git/recipes/fakeroot/ (fakeroot-native_1.9.6.bb fakeroot_1.9.6.bb): Oct 08 10:33:38 fakeroot-native: RDEPENDS -> RDEPENDS_${PN} Oct 08 10:33:38 Signed-off-by: Frans Meulenbroeks Oct 08 10:33:39 Acked-by: Martin Jansa Oct 08 10:34:31 lunch :-) Oct 08 10:34:54 03Frans Meulenbroeks  07org.openembedded.dev * rded7e11012 10openembedded.git/recipes/nogravity/nogravity_2.0.bb: Oct 08 10:34:54 nogravity/nogravity_2.0.bb: RRECOMMENDS+${PN} -> RRECOMMENDS_${PN} Oct 08 10:34:54 Signed-off-by: Frans Meulenbroeks Oct 08 10:34:54 Acked-by: Martin Jansa Oct 08 10:49:16 eFfeM_work: bootstream/bootstream-chumby-falconwing_1.bb: change looks wrong, PROVIDES without 'R' doesn't need _${PN} not sure if it can have _${PN} guess not Oct 08 11:06:41 sry thanks will fix Oct 08 11:08:51 03Frans Meulenbroeks  07org.openembedded.dev * r525e35645f 10openembedded.git/recipes/bootstream/bootstream-chumby-falconwing_1.bb: Oct 08 11:08:51 bootstream-chumby-falconwing_1.bb: no _${PN} after DEPENDS Oct 08 11:08:51 Signed-off-by: Frans Meulenbroeks Oct 08 11:09:04 JaMa: btw there is also this one: mokomaze/mokomaze_0.5.5.bb:RDEPENDS_shr += "fsoraw" Oct 08 11:12:30 eFfeM_work: ok I'll fix this one Oct 08 11:22:39 eFfeM_work: there was PR instead of PN in your mokomaze commit Oct 08 11:22:40 03Martin Jansa  07master * r1611e2faf1 10openembedded.git/recipes/mokomaze/mokomaze_0.5.5.bb: Oct 08 11:22:40 mokomaze: fix RDEPENDS Oct 08 11:22:40 Signed-off-by: Martin Jansa Oct 08 11:22:42 03Martin Jansa  07master * r00e8e81d7d 10openembedded.git/recipes/mokomaze/mokomaze_0.5.5.bb: Oct 08 11:22:42 mokomaze: oe-stylize Oct 08 11:22:42 Signed-off-by: Martin Jansa Oct 08 11:22:59 oh oops Oct 08 11:24:17 thanks for the fix Oct 08 12:05:17 hi gnutoo Oct 08 12:06:00 hi woglinde Oct 08 12:08:40 woglinde, btw do you have an idea on how to integrate an /etc/profile.d/gtk-immodule.sh? Oct 08 12:09:26 should it be done in gtk recipe? Oct 08 12:09:34 or separately Oct 08 12:12:52 hm Oct 08 12:12:54 sorry no Oct 08 12:13:04 ok Oct 08 12:13:33 I've tried to implement it in gtk recipe but failed Oct 08 12:14:08 basically I wanted to add it to the xim immodule in gtk Oct 08 12:14:20 but xim is a dynamic packag Oct 08 12:14:25 *package Oct 08 12:14:48 so now we live with a hack Oct 08 12:14:56 for the keyboard phones Oct 08 12:15:13 we have now an n900 branch on shr repo Oct 08 12:15:20 branch of oe Oct 08 12:15:23 which is bad Oct 08 12:15:54 but it was necessary because we were unsure abuot what kernel version to use Oct 08 12:17:31 btw you're still doing java stuff? Oct 08 12:19:29 how do I avoid build a kernel for any image?? Oct 08 12:20:17 why is it bad? Oct 08 12:21:06 I just want a minimal image, no kernel (I rather work with git version) Oct 08 12:21:47 make your own image Oct 08 12:21:58 look into the existing recipe to get a clue Oct 08 12:22:39 woglinde_: but... the ones that configures the preferred kernel are the machine.conf Oct 08 12:33:18 easy way, build standard kernel then in a postprocess command remove it from the image Oct 08 12:59:57 03Steffen Sledz  07org.openembedded.dev * r0486c1885b 10openembedded.git/recipes/syslog-ng/syslog-ng.inc: (log message trimmed) Oct 08 12:59:57 syslog-ng: increase u-a priority to 200 Oct 08 12:59:57 Priority for update-alternatives was 50 before which was the Oct 08 12:59:57 same like the one for busybox-syslog. Resulting order now is: Oct 08 12:59:57 * 300 rsyslogd Oct 08 12:59:57 * 200 syslog-ng Oct 08 12:59:57 * 100 sysklogd Oct 08 13:10:38 eFfeM_work: do you happen to see the notes about multiple providers for modules-init-tools? Oct 08 13:11:12 iirc the -cross recipes is conflicting Oct 08 13:47:01 ant_work: i think I did see it at some point in time (maybe with a build here at work), but not with my build at home Oct 08 13:47:42 I have also seen a message wrt depmod but guess this is still a conf problem (I'm targeting to a new board) Oct 08 13:47:44 ok, is a constant building angstrom console - opie - x11 images Oct 08 13:48:04 I'll check later Oct 08 13:48:17 btw, about kernel tasks / append Oct 08 13:48:42 * Tartarus pokes people about updating libtiff Oct 08 13:48:51 I think the clean/standard way would be the recipes include linux.inc Oct 08 13:49:00 and specifically why it'd be bad to use PE proactively Oct 08 13:49:10 and have a do_configure_append fre for mangling Oct 08 13:49:12 ah ok, haven't build angstrom for quite a while, moved to minimal as angstrom pinned mythtv to 0.22 and I want to use 0.23 (which ofc I could also force in local.conf) Oct 08 13:49:36 now as it is, there is a dumb fw task whic could probably moved Oct 08 13:49:53 eFfeM_work: yes, seems a distro-issue Oct 08 13:50:14 Tartarus: I've no idea on the PE thingie, but yeah, I do prefer to have rc candidates have the right number in the recipe; maybe something for OEDEM ? Oct 08 13:51:37 eFfeM_work: Yeah, but I really don't want to wait 3 weeks to fix it Oct 08 13:51:42 In some ways, I can't :) Oct 08 13:52:27 Tartarus: you can still name it 3.99+4.0beta6, if you don't like 3.9.2+ Oct 08 13:52:30 i can update libtiff to rc6 (or whatever it was), but don't feel very comfortable with chainging the numbers Oct 08 13:52:49 JaMa: yeah Oct 08 13:55:01 JaMa: Yeah, I guess that works too. How does debian or fedora deal with things like this? Oct 08 13:55:22 where the previous line isn't done Oct 08 13:55:54 Tartarus: ie in gentoo 3.9_beta2 is lower version than 3.9 Oct 08 13:56:06 i'm not sure debain will to beta things Oct 08 13:56:24 The other question is why are we using the beta one, yeah :) Oct 08 13:56:55 Anyhow, i'll do 3.9.9+4.0.0beta6 or so Monday'ish, thanks guys Oct 08 14:23:43 Tartarus: it might be that some package required libtiff >= 4.0 Oct 08 14:24:08 4.0 seems to support BigTIFF files Oct 08 14:41:17 heh, would be nice if the packaging code was smart enough to adapt prerelease version numbers automatically, somehow, since its the packaging and feeds that care about it, we can teach bitbake to understand the prerelease postfixes easily enough for its end of things Oct 08 14:45:33 maybe that should get more thought -- it makes sense that the version comparison schemes wont necessarily match up between bitbake and whatever package manager it uses for the packaging and filesystem Oct 08 14:46:53 yes Oct 08 14:47:00 weekend :-) Oct 08 14:49:18 good morning Oct 08 14:49:25 * kergoth ponders Oct 08 14:49:28 morning khem Oct 08 14:49:36 (and everyone else) Oct 08 14:49:42 the fire alarm batterry is low need to replace it kept beeping all night Oct 08 14:50:01 heh, annoying Oct 08 14:50:16 disturbed sleep Oct 08 14:50:32 when do we use PE Oct 08 14:51:00 when we have to, is the short answer Oct 08 14:51:35 e.g when a package has been made available in a feed with a busted, screwed up version (e.g. just 4.0.0beta6), then we'd have to bump the epoch to ensure the real 4.0.0 is seen as newer Oct 08 14:51:48 only ever use it when we absolutely have to Oct 08 14:52:00 again, its more a function of distribution and packaging than OE's end of things Oct 08 14:54:43 hmm interestingly weird Oct 08 14:55:39 can read more about it at the debian policy manual Oct 08 14:55:55 oh its same for us ? Oct 08 14:56:02 cool I know the debian one Oct 08 14:56:15 did not strike to me eh Oct 08 14:56:21 now i remember Oct 08 14:56:44 console-image with libtool 2.4 is building and booting Oct 08 14:56:56 but target gcc needs love Oct 08 14:57:33 kergoth: I was thinking of adding override for libtool 2.4 Oct 08 14:57:43 so I could conditionally include patches Oct 08 14:58:12 hehe "needs love" :) Oct 08 14:59:15 kergoth: btw textual is nice Oct 08 14:59:17 PE is the N: part of debian package versions, aka epoch, its directly moved over to the ipks.. no idea whats done with it for package_rpm Oct 08 14:59:20 I tried it Oct 08 15:00:47 override is an interesting idea.. one part of me hates touching that varaible at all due to the scope, but another part says thats a clean way to do the conditionals given the scope of libtool, and given the fact that adding the override could be opt-in Oct 08 15:02:47 kergoth: we might have to live with libtool 2.2.x for quite a bit Oct 08 15:02:52 :( Oct 08 15:03:00 unless all distro's move to 2.4 Oct 08 15:03:22 and angstrom-2008 is not going to be EOLed soon Oct 08 15:03:35 its tempting to create a 'next' branch where we just say fuck it and move forward regardless of compatibility, and let dev rot without it until people move forward Oct 08 15:03:39 ;) Oct 08 15:04:29 infact if we had time based releases this was a no brainer Oct 08 15:04:44 but we keep forging on master and never release Oct 08 15:05:01 and distros start using master as a release branch Oct 08 15:05:28 I'd love to start doing OE releases on a 6 month cycle Oct 08 15:05:30 or may be now is the time for a stable branch Oct 08 15:05:34 would encourage us to get shit together for it Oct 08 15:05:38 rather than just floatinga long Oct 08 15:05:49 yeah I am all for time based releases Oct 08 15:06:09 it gives better visibility and raises user confidence Oct 08 15:06:18 now that we have some sort of weekly testing Oct 08 15:06:21 it should be doable Oct 08 15:06:22 i'd like to see both OE and bitbake on that cycle, ideally with the same cycle, and have OE use one release behind of bitbake at all times (just to give people a little time to catch up) or the current Oct 08 15:06:50 actually bitbake could be released within OE :) Oct 08 15:07:46 so I have added the releases as an agenda item for OEDEM Oct 08 15:07:56 that's true. it could remain a separate project, with its release cycle, but just *package* it with OE for convenience Oct 08 15:07:57 hopefully people will discuss it Oct 08 15:08:00 nice Oct 08 15:08:28 * kergoth wishes he could make it :( already planned a trip to visit his parents for the first time in 4 years that weekend, and a couple days before that is his fiance's birthday Oct 08 15:08:58 all we need to say is here is release 2010.12 and its supported untill next release Oct 08 15:09:13 oh :) Oct 08 15:10:27 kergoth: we should organise one OEDEM US also sometime Oct 08 15:10:41 yeah, indeed. it'd make things easier for people that need stability, also, knowing that the release is more stable than the day to day tree, and they could always stay on a previous one until they've completed testing with the new one Oct 08 15:10:55 khem: you shoud remove one 'E' then Oct 08 15:11:00 hehe Oct 08 15:11:01 OEDUM, OEDAM Oct 08 15:11:11 zecke: heh right Oct 08 15:11:32 ok I'm definately sync'd up now and still failing gdb-cross-sdk-7.1-r8.0 from meta-toolchain for x86 - ERROR: Task 609 (/home/tharvey/oe/org.openembedded.dev/recipes/gdb/gdb-cross-sdk_7.1.bb, do_configure) failed with 256 Oct 08 15:11:52 tharvey: whats the error Oct 08 15:11:55 OEDUM sounds .. um.. i'm not sure i can it diplomatically. i'll just say "bad" Oct 08 15:12:02 haha Oct 08 15:12:56 | ERROR: Function do_configure failed Oct 08 15:13:03 I see no error in configure log Oct 08 15:13:15 last thing in it is | config.status: creating Makefile Oct 08 15:13:31 tharvey: check log.do_configure Oct 08 15:13:40 tharvey: btw did ramfs help Oct 08 15:14:09 I didn't try it - seemed like the discussion indicated it wouldn't work to muck with the structure of TMPDIR Oct 08 15:14:44 tharvey: ok SSDs are fast too :) Oct 08 15:14:47 kergoth: OEDEM is a... well not disease but something medical too Oct 08 15:14:58 hehe Oct 08 15:15:07 We shoud just call it OE dev Summit Oct 08 15:15:23 Europe and Americas Oct 08 15:15:29 i like that Oct 08 15:15:35 so people from Argentina are also invited Oct 08 15:15:42 khem, already using an SSD but it didn't make a big difference strangely - I think it was because I went from a 32bit system with python-psycho to a 64bit system with SSD and no psycho Oct 08 15:16:02 tharvey: ok Oct 08 15:16:25 not clear what is an error in long.do_configure - there is the following on the first line but then it seems to go on:configure.ac:32: error: Please use exactly Autoconf 2.64 instead of 2.65. Oct 08 15:16:33 heh Oct 08 15:16:41 tharvey: thats the error Oct 08 15:16:46 khem, still giong to try to use ramfs with pstage/deploy a bind to a disk based fs though Oct 08 15:17:04 so its trying to do reconfigure stuff Oct 08 15:17:18 which recipe is that once again ? Oct 08 15:17:27 gdb-cross-sdk Oct 08 15:17:59 ok I see Oct 08 15:18:12 its a dated build machine... not clear when it stopped being able to build angstrom - I'll update autoconf Oct 08 15:18:28 tharvey: recipes/gdb/gdb-cross-sdk.inc add oe_runconf at the end of do_configure_append Oct 08 15:18:32 that should fix it Oct 08 15:18:50 err change do_configure_append to do_configure Oct 08 15:18:53 as well Oct 08 15:19:22 really? ok let me try that - will have to do in a bbappend or something Oct 08 15:20:29 tharvey: heh just touch the openembedded layer its ok :) Oct 08 15:20:35 if I change do_configure_append to do_configure It will clobber orig do_configure though... what about adding a preappend? (is there such thing?) Oct 08 15:20:40 if it works we might need it upstream Oct 08 15:21:47 tharvey: just try what I said Oct 08 15:21:50 and lets se Oct 08 15:22:05 testing Oct 08 15:22:34 there's do_configure_prepend if we need to put it there Oct 08 15:23:55 so I'm curious... how did you know the Augoconf 2.64 vs 26.5 error is need for oe_runconf? Is it that autoconf is not being run just configure? Oct 08 15:24:17 khem, your orig suggestion works - its on do_compile now Oct 08 15:25:11 tharvey: ok happy hacking Oct 08 15:25:22 time to feed kids bbl Oct 08 15:25:34 can you push that upstream when you get a chance? Oct 08 16:08:22 re Oct 08 16:25:41 gm Oct 08 16:41:39 with in 2 days we added almost 60 parsable recipes :) Oct 08 16:41:59 not that its bad but the I am amazed at the rat Oct 08 16:42:02 rate Oct 08 16:42:31 khem`: xorg version upgrades are pretty easy, I guess it makes 30-40 of that Oct 08 16:42:48 hmm yeah Oct 08 16:42:53 khem`: please check ML about last gcc patch Oct 08 16:43:02 the thumb one ? Oct 08 16:43:11 yeah I think I have to revert that Oct 08 16:43:53 I got similar errors I guess Oct 08 16:43:53 I'm not sure that it's cause for my issue Oct 08 16:44:01 but the dbus change looks harmless Oct 08 16:44:30 did u try to revert it locally Oct 08 16:44:36 and see if its the cause Oct 08 16:44:58 actually the patch is still under test here and I am seeing some regressions in gcc testsuite Oct 08 16:45:11 so its likely that its gcc Oct 08 16:45:58 not yet.. my locall build is busy with those recipes failing with libtool-2.4 Oct 08 16:46:06 this is from "official" builder Oct 08 16:47:21 ok Oct 08 16:47:34 khem: the /obsolete population is growing at higher rate ;) Oct 08 16:47:39 no worry Oct 08 16:47:48 JaMa all patches that you need are posted to ml now Oct 08 16:48:27 JaMa target gcc is not building with libtool update yet Oct 08 16:48:50 there are some QA errors due to .la files containing paths into workdir Oct 08 16:49:02 those needs to be taken good care Oct 08 16:49:10 right now I seeing it in openjade Oct 08 16:49:33 khem`: you mean that those libpython from host issues are not caused by libtool-2.4? Oct 08 16:50:11 because with all patches from you applied I still have similar list as Graham Gower Oct 08 16:51:21 JaMa it could be Oct 08 16:51:30 have to be looked into Oct 08 16:51:47 mostly the issues were hidden because we munged .la files Oct 08 16:52:14 stupid mac terminal does not support 256 colors Oct 08 16:52:28 * JaMa leaving, enjoy weekend and thx Oct 08 16:52:29 * khem` needs to stick with iTerm Oct 08 16:56:54 03Khem Raj  07master * r327ba88ab7 10openembedded.git/recipes/gcc/ (gcc-4.5.inc gcc-4.5/thumb-asm-out-of-range-pr43961.patch): Oct 08 16:56:54 Revert "gcc-4.5: Fix asm out of range errors seen on thumb when using -O1" Oct 08 16:56:54 This reverts commit 8e1eef82a09c14668a125c720cfcd354bbeb42cf. Oct 08 16:56:54 Signed-off-by: Khem Raj Oct 08 17:02:07 ant_work: guess i am the biggest contributor to obsolete Oct 08 17:02:11 here comes todays set Oct 08 17:02:27 go Oct 08 17:02:33 03Frans Meulenbroeks  07org.openembedded.dev * r8a6dfba616 10openembedded.git/recipes/classpath/files/ (6 files): Oct 08 17:02:34 classpath : moved unused files to obsolete dir Oct 08 17:02:34 Signed-off-by: Frans Meulenbroeks Oct 08 17:02:34 03Frans Meulenbroeks  07org.openembedded.dev * r104b736e07 10openembedded.git/recipes/coreutils/ (3 files in 2 dirs): Oct 08 17:02:34 coreutils : moved unused files to obsolete dir Oct 08 17:02:34 Signed-off-by: Frans Meulenbroeks Oct 08 17:02:35 cia is slow Oct 08 17:02:35 03Frans Meulenbroeks  07org.openembedded.dev * r3f511d5b43 10openembedded.git/recipes/cairo/ (3 files in 2 dirs): Oct 08 17:02:35 cairo : moved unused files to obsolete dir Oct 08 17:02:36 Signed-off-by: Frans Meulenbroeks Oct 08 17:02:36 03Frans Meulenbroeks  07org.openembedded.dev * r12d705f554 10openembedded.git/recipes/curl/files/ (3 files): Oct 08 17:02:38 curl : moved unused files to obsolete dir Oct 08 17:02:38 Signed-off-by: Frans Meulenbroeks Oct 08 17:02:39 03Frans Meulenbroeks  07org.openembedded.dev * r04a4b8329e 10openembedded.git/recipes/clamav/files/cross-compile-fix.patch: Oct 08 17:02:39 clamav : moved unused files to obsolete dir Oct 08 17:02:40 Signed-off-by: Frans Meulenbroeks Oct 08 17:02:40 03Frans Meulenbroeks  07org.openembedded.dev * r54837a4d34 10openembedded.git/recipes/gpe-package/files/search.patch: Oct 08 17:02:41 gpe-package : moved unused files to obsolete dir Oct 08 17:02:53 ah.. gpe-package is broken atm Oct 08 17:02:54 Signed-off-by: Frans Meulenbroeks Oct 08 17:02:54 03Frans Meulenbroeks  07org.openembedded.dev * r60cea0764c 10openembedded.git/recipes/connman/files/connman-build-and-work.patch: Oct 08 17:02:54 connman : moved unused files to obsolete dir Oct 08 17:02:54 Signed-off-by: Frans Meulenbroeks Oct 08 17:02:54 03Frans Meulenbroeks  07org.openembedded.dev * r424b29819a 10openembedded.git/recipes/gpe-login/files/ (4 files): Oct 08 17:02:54 gpe-login : moved unused files to obsolete dir Oct 08 17:02:55 Signed-off-by: Frans Meulenbroeks Oct 08 17:03:41 ant not sure what is broken or not, i just run my script, check that everything still patches, then push Oct 08 17:03:59 still to do: recipes starting with g, o, x Oct 08 17:04:08 eFfeM: you may be the biggest but I'm the one who created /linux inside there ;) Oct 08 17:04:27 ant_work: no problems with that :-) Oct 08 17:04:38 I have a scary patch queue about this... Oct 08 17:04:45 let's wait some more days Oct 08 17:06:06 khem`, your patch works fine... want to make sure I understand it though Oct 08 17:06:43 the issue is that gdb-cross-sdk recipe is not running oerunconf, but configure instead, so its not fully rebuilding autoconf setup right? Oct 08 17:06:55 tharvey, if you call default do_configure from autotools then it runs autoreconf which means it will try to regenerate configure Oct 08 17:07:04 thus its somewhat dependent on the autotools version that was used to build the distribution upstream? Oct 08 17:07:20 right... so isn't that the right thing to do on pretty much all recipes? Oct 08 17:07:54 tharvey some packages like gcc, binutils, gdb are special as they have complex build mechanism so they stick to given version of autconf automake etc and they bar you from using a different version if you want to regenerate the scipts Oct 08 17:08:29 tharvey many packages use them and its fine to use a newer version of auto tools Oct 08 17:08:38 but not for some Oct 08 17:08:43 and gdb is one of them :) Oct 08 17:09:32 ideally we could avoid doing autoreconf but since we can not keep n number of versions of autoconf etc. its not possible Oct 08 17:09:49 so we autoreconf them using whatever version we use in OE Oct 08 17:10:33 ok, going to push my patch to patchwork - haven't done it before so hope I'm doing it right Oct 08 17:11:07 btw is there a good way to add patches to one's git without the risk of accidently pushing them (without having to go for two trees) Oct 08 17:11:54 tharvey: no explicit pushing to patchwork is needed, patchwork will pick up a patch mailed to the list if it has PATCH in the subject line Oct 08 17:12:51 tharvey if you git format-patch and git send-email to the ml that enough for patchwork it will catch it Oct 08 17:13:37 eFfeM create a local branch and apply them and I know many people love to do git push Oct 08 17:13:47 khem`, right - following instructions here http://wiki.openembedded.net/index.php/How_to_submit_a_patch_for_dummies Oct 08 17:14:00 but make sure which of your local refs you want to push Oct 08 17:14:16 git push origin : Oct 08 17:14:20 is the correct logic Oct 08 17:14:29 git: 'send-email' is not a git command. See 'git --help'. Oct 08 17:14:40 and you wont do mistakes Oct 08 17:14:44 khem, yeah, local branch is a solution Oct 08 17:15:04 you lost me there - if I cat the patch created by git format-patch it looks right Oct 08 17:15:04 issue that I often have is that some patches are still under review which I want to keep locally Oct 08 17:15:11 and I am not really a git wiz Oct 08 17:15:21 tharvey its separate package probably on your host Oct 08 17:15:42 ah... thx 'git-email' on ubuntu Oct 08 17:15:54 eFfeM, and you can use git rebase --onto to punt a patch out if you want Oct 08 17:16:17 patch I mean local patch (not yet pushed to public git repo) Oct 08 17:16:51 git rebase --onto ~1 Oct 08 17:16:54 will do it Oct 08 17:17:13 there are things like this which are very helpful Oct 08 17:17:32 tharvey whats your host OS Oct 08 17:17:34 khem, ah ok, thanks, guess the gitphrasebook could use an update Oct 08 17:17:51 khem`, ubuntu 10.04 Oct 08 17:17:54 eFfeM yeah I always thought to add some wisdom there :) Oct 08 17:18:19 not clear how to specify the subject line for patch - it looks like it defaults to my entire commit message w/o line breaks Oct 08 17:19:16 tharvey ok on that send-email is a separate package Oct 08 17:19:32 tharvey, sudo apt-get install git-email Oct 08 17:19:56 btw decided to do a test run for recipe_sanity, got about 1700 issues and it is at letter x Oct 08 17:20:26 tharvey: see if git send-email is in synaptic Oct 08 17:20:26 and you can also install git-completion and will complete the git commands for you :) and also show useful options for given git command Oct 08 17:20:43 and the subject line normally is the first line of the commit Oct 08 17:20:59 afk for a few hrs Oct 08 17:22:48 do I need to prefix patch with [oe]? such as --subject-prefix="[oe]" ? Oct 08 17:22:50 khem`: if all you want to do is drop the top patch, you can always reset --hard HEAD~1 Oct 08 17:23:01 if you want to not lose the changes from the patch, but want to re-commit it, you can just git reset HEAD~1 Oct 08 17:23:08 then you're back to where you can add -p and create the commit again Oct 08 17:23:17 yes the one I suggested can punt out 1 single patch from anywhere Oct 08 17:23:29 in a long list of patches Oct 08 17:23:43 cute. i'd probably just rebase -i Oct 08 17:24:04 sometimes when I apply patches from patchwork its handy Oct 08 17:24:28 yeah, makes sense, particularly when dealing with a series Oct 08 17:24:51 * khem` is looking for home theater system on amazon Oct 08 17:25:44 I wonder why bose is like 5 times expensive compare to others Oct 08 17:26:40 i don't understand people who feel the need to use stgit and the like Oct 08 17:26:45 with rebase -i its hardly necessary.. Oct 08 17:29:57 yeah with git only git is enough Oct 08 17:30:56 handy git alias: amend = commit --amend -C HEAD Oct 08 17:30:59 kergoth, I have a good one for you Oct 08 17:31:08 amend a commit with your changes in the index, but don't prompt for the commit mesage again Oct 08 17:31:08 I have a srctree recipe Oct 08 17:31:13 actually, I ahve 2 Oct 08 17:31:25 and they both build if you bb recipe ... Oct 08 17:31:38 but, if you clean them, and bitbake an image that includes them Oct 08 17:31:45 they fail at some point Oct 08 17:31:50 what's the failure? Oct 08 17:32:28 I'll patebin next time I get it Oct 08 17:32:36 I only just worked out the pattern .... Oct 08 17:34:24 k Oct 08 17:35:18 * kergoth adds to his OE todo Oct 08 17:38:02 * khem` goes biking Oct 08 17:43:20 Hi everyone... Can anyone tell me what could possibly be going wrong if I am not able to detect any usb device on usb host port on beagleboard. 'lsusb' shows no product id and 'dmesg' shows so usb device attached or registered.. I am using Angstrom 2.62 on beagleboard. Oct 08 17:49:04 a bit off topic but what are you all doing about subscribing to maillist in order to post? I read the maillist via gmane and rss - don't really want to burden my gmail account with maillist Oct 08 17:50:00 its not much of a burden when you leverage gmail's Mute function Oct 08 17:50:09 mute a thread, it never appears in your inbox again Oct 08 17:50:17 Or filter it into a folder. Oct 08 17:50:21 of course, you could always .. right, that :) Oct 08 17:50:25 archive it and label it Oct 08 17:50:42 muting is per thread though not source correct? Oct 08 17:50:43 you've got plenty of space, so why not, at least during the period that you're messing with oe Oct 08 17:50:57 right, but you can create filters to make it go directly to the label and bypass the inbox Oct 08 17:51:02 then you read it when you choose to look at it Oct 08 17:51:14 which is what broonie brought up Oct 08 17:52:11 yes, over the years I typically have to subscribe for a period of time then unsubscribe, just seems like a pointless waste of bandwidth Oct 08 17:53:09 I'd really like to get a better, cleaner workflow for monitoring projects i'm interested in, but whose day to day development i don't care as much about Oct 08 17:53:19 i want to know if a major direction change happens, or something, but thats all Oct 08 17:53:27 but its hard to filter to that Oct 08 17:53:36 other than just subscribing to announce lists, anyway Oct 08 17:53:39 thats not always enough Oct 08 17:53:56 agreed Oct 08 17:54:58 like, busybox -- if you're in embedded, and work on systems in particular, you obviously care about what they're doing, but don't need to see every little thing.. Oct 08 17:55:23 maybe i should find one of those online personal assistant people and pay them crap wages to sort through the crap for me Oct 08 17:55:24 hehe Oct 08 17:55:40 I have about 10 subscriptions to maillists which I don't have time to read - just so that I can post to them in a blue moon - seems a waste Oct 08 17:55:51 read via gmane Oct 08 17:55:54 maybe we can write a class or recipe for that :) Oct 08 17:55:56 gmane is a good idea Oct 08 17:56:01 ya, I read via gmane but can't post Oct 08 17:56:06 (aside: WHY oh WHY is gmane's search so fucking horrible) Oct 08 17:57:03 but really, how much bandwidth is a list's traffic likely to use, given its text, unless its LKML? Oct 08 17:57:04 :) Oct 08 17:57:34 * kergoth doesn't know, would be curious to find out Oct 08 17:59:22 Is there a dedicated channel for beagleboard? Oct 08 17:59:31 there is Oct 08 17:59:41 i don't recall what it is offhand, beagle or beagleboard or so :) Oct 08 17:59:51 #beagle Oct 08 18:00:05 :) thanks, my memory is weak Oct 08 18:00:35 tharvey: hmm, gmane does have a post function Oct 08 18:00:37 never tried it though Oct 08 18:00:43 ok, so my miniscule patch is posted to maillist - what now? Oct 08 18:00:44 and i dunno if you can reply to existing mails Oct 08 18:00:45 heh Oct 08 18:00:51 wait :) Oct 08 18:01:03 kergoth, ya I figured it would reject it though but I didn't try either Oct 08 18:01:30 well, gmane themselves are likely subscribed in order to archive the mail Oct 08 18:01:35 so if they send it on your behalf.. Oct 08 18:01:41 * kergoth shrugs Oct 08 18:01:57 thats a good point, but then you don't have the ability to mail from git directly Oct 08 18:02:05 can you explain this patch more? I'm not seeing how running oe_runconf and overriding do_configure runs autoreconf Oct 08 18:02:07 yeah, true Oct 08 18:02:18 @kergoth thanks .. its #beagle Oct 08 18:02:34 perhaps khem` can help explain - it was his idea ;P Oct 08 18:02:50 autoreconf is the default behavior for do_configure when inheriting autotools Oct 08 18:02:55 oe_runconf runs ./configure, not autoreconf Oct 08 18:03:00 so if anything its the other way around Oct 08 18:03:01 prior to that patch I was do_configure was complaining about my version of autotools Oct 08 18:03:08 unless i'm missing something.. Oct 08 18:03:33 its in a .inc, _append would apply regardless of what the recipe is doing Oct 08 18:03:47 whereas a definition of do_configure, if the recipe defines it, it would override it Oct 08 18:03:54 maybe thats whats happening there Oct 08 18:04:15 honestly I don't fully understand either.... I thought it should go in a do_prepend Oct 08 18:04:22 either way, that patch doesn't explain what its doing and why very well, i think it needs a bit more detail in the body of the commit message Oct 08 18:05:19 thought it was clear... re-runs autoconf instead of just configure Oct 08 18:06:19 the summary is clear, yes Oct 08 18:06:20 without it :configure.ac:32: error: Please use exactly Autoconf 2.64 instead of 2.65. Oct 08 18:06:33 that makes no sense. Oct 08 18:06:38 configure.ac would only error from *running* autoconf Oct 08 18:06:48 what you're doing is the opposite, you're running ./configure and bypassing autoconf execution Oct 08 18:06:52 most likely, anyway Oct 08 18:06:59 khem`: details, when you're around, please Oct 08 18:07:40 sorry, I'm not that up on autotools/autoconf - just frusterated by the fact that gdb-cross-sdk used to build and now doesn't on my build system Oct 08 18:08:10 I'm in the middle of a 'sync' our OE snapshot up with upstream and continually chasing new issues Oct 08 18:08:38 what i'm saying is, the description of that patch doesn't appear to jive with what its doing, unless there's some interaction that's unclear, in which case that should be mentioned in the message Oct 08 18:09:07 either way, i don't think it should be applied as is, unless khem disagrees, he's the toolchain expert :) Oct 08 18:09:21 agreed - I can't say I fully understand it (I thought I did until you pointed out its not doing what I thought) - all I'm trying to do is get a fix upstream Oct 08 18:09:57 which is good, we appreciate it, its just that commit messages need to comply with our policies Oct 08 18:10:00 speaking of which, that summary doesn't Oct 08 18:10:06 need to prefix with what it affects Oct 08 18:10:20 e.g. gdb-cross-sdk: fix on older build systems Oct 08 18:10:52 can always apply it and modify the message ourselves to comply, so i wouldn't worry about it Oct 08 18:11:04 i'll talk to khem about it Oct 08 18:11:11 because for the life of me I have not been able to get something like 'do_configure_prepend_pn-gdb-cross-sdk { oe_runconf }' in a local.conf to work - often find that I need to bandaid things like that Oct 08 18:11:28 well, that wouldn't stop it from doing what its doing Oct 08 18:11:40 that would only run ./configure *before* the things it was doing that cause errors.. Oct 08 18:11:47 prepend prepends, it doesn't replace Oct 08 18:12:18 prepend and append are special, you can't override one prepend with another Oct 08 18:12:24 they both get appended Oct 08 18:12:29 its like =+, only more lazy Oct 08 18:12:39 s/appended/prepended/ Oct 08 18:12:50 sorry what are you suggesting the summary say? the patch is to fix gdb-cross-sdk compilation - wouldn't that be the subject line summary? Oct 08 18:13:10 again, it has to be *prefixed* with what it changes Oct 08 18:13:13 you have it in the middle of the message Oct 08 18:13:23 gdb-cross-sdk: foo Oct 08 18:13:28 not "foo in gdk-cross-sdk" Oct 08 18:13:44 minor, but it is the policy :) Oct 08 18:13:50 ah... ok I understand thx Oct 08 18:13:54 np Oct 08 18:13:59 of course, there are exceptions Oct 08 18:14:09 if i have one commit that modifies 12 recipes, i'm sure as hell not listing them all :) Oct 08 18:14:10 good learning, I have several recipes I would like to push upstream just haven't had the time or knowledge on how to do so Oct 08 18:14:40 also not clear to me who (if anyone) commits them... I've seen things posted in the past that don't necessarily get nak'd yet never seem to be committed Oct 08 18:15:56 well, we have a specific policy, changes to core things, classes, toolchain won't be applied until they get two acks Oct 08 18:16:03 they could sit there forever and not be applied Oct 08 18:16:08 so sometimes those get resent, or whatever Oct 08 18:16:29 recipe changes, if there's no naks, they can be applied, but who does so is not defined, whomever decides to, or who goes through patchwork periodically Oct 08 18:17:19 ah I see (reading policy) Oct 08 18:17:30 personally, i usually ignore mails which are recipe changes unless its a recipe i maintain, but i tend to care more about infrastructure and work on bitbake and classes rather than little recipe things, just have different priorities than others do Oct 08 18:20:56 by the way, my discussion with khem about this was a few hours ago, its in the irc log Oct 08 18:21:12 although I can't say it was fully explained Oct 08 18:30:03 khem, you about Oct 08 18:36:40 kergoth, did you push image features? Oct 08 18:37:26 there's only one ack at the moment Oct 08 18:37:28 heh Oct 08 18:37:32 hmm Oct 08 18:37:46 I did a pull and it wnats me to merge something Oct 08 18:38:01 something else must have touched it Oct 08 18:40:53 i'm quite surprised at the lack of interest in IMAGE_FEATURES, i'd have thought the easy ability to add debugging files to an image would be more useful to folks Oct 08 18:41:22 maybe my summary wasn't interesting enough to get people to open the mail :) Oct 08 18:41:39 hmm, i know, i could break it up into two pieces. Oct 08 18:41:43 one adds package groups Oct 08 18:41:47 er Oct 08 18:41:49 package group support Oct 08 18:41:55 then a second adds the actual default groups, like debugging Oct 08 18:42:04 maybe that would get more attention Oct 08 18:42:36 does thi sline need to change to RDEPENDS_${PN} Oct 08 18:42:43 RDEPENDS += "${__PACKAGE_INSTALL}" Oct 08 18:45:40 kergoth: would my ack be worth anything, I like the idea Oct 08 19:24:24 I have a python assert failing, how do I get a backtrace of where it happens? Oct 08 19:51:29 hi Oct 08 19:52:41 AliAnwar: welcome Oct 08 19:53:01 kergoth_: never knew that its such a large community :) Oct 08 19:53:04 thanks Oct 08 19:53:09 :) Oct 08 19:53:10 np Oct 08 19:53:25 the links in the topic should keep you busy a while, i expect :) Oct 08 19:53:48 yes just going through wiki Oct 08 19:55:59 re Oct 08 20:00:25 re Oct 08 20:09:18 khem, ping Oct 08 20:11:27 gm all Oct 08 20:11:34 is this a recent regression? | checking lex output file root... configure: error: cannot find output from flex; giving up Oct 08 20:12:23 likewise: w|zzy haven't seen it Oct 08 20:12:51 aarg system was slow, typed ahead, no intention to mention w|zzy, sry Oct 08 20:13:38 and whoever asked earlier about multiple providers for module-init-tools-depmod:L I have that too Oct 08 20:13:39 I had a fixed build, threw away tmp/ and now the build fails. Hope my Ubuntu 10.04 package updates didn't regress me into a non-working setup. Oct 08 20:18:09 likewise: thats unlikely Oct 08 20:18:41 2 hours climb 20 miles not bad Oct 08 20:19:54 kergoth: you have my ack for image features Oct 08 20:20:00 khem: not 20 miles in 2 hours, right? Oct 08 20:21:10 khem: I hope the regression is so unlikely that it doesn't occur after a new "rm -rf tmp; bitbake console-image" :-) Oct 08 20:21:26 likewise: total 20miles Oct 08 20:21:33 16miles climb Oct 08 20:21:44 total time 1:59 Oct 08 20:21:46 :) Oct 08 20:22:01 likewise: whats the error Oct 08 20:22:05 khem: sounds like a good leg muscle trainer Oct 08 20:22:43 likewise: it was 16 miles climb and then 4 miles downhill so many times I looked back and though about U turn ;) Oct 08 20:22:50 | checking lex output file root... configure: error: cannot find output from flex; giving up Oct 08 20:22:52 NOTE: Task failed: /home/leon/sandbox/ccm/topha/openembedded/tmp/work/i686-linux/flex-native-2.5.35-r5.1/temp/log.do_configure.29323 Oct 08 20:22:53 NOTE: oestats: task failed, see http://tinderbox.openembedded.net/public/logs/task/8519755.txt Oct 08 20:24:48 likewise: whats in your ASSUME_PROVIDED Oct 08 20:24:59 khem: okay, thanks Oct 08 20:26:00 khem: didn't set it. The second attempt reaches much further already... Wonder if there is (yet) another race (this is heavily multithreaded). Oct 08 20:27:41 khem: I'll leave that error for now, have loads of work to do... Oct 08 20:30:26 likewise: ok np Oct 08 20:30:34 races are nasty Oct 08 20:30:41 probably missing depends or something Oct 08 20:31:14 kergoth: that gdb patch is correct Oct 08 20:31:43 you explained it yourself what do_configure is doing Oct 08 20:33:30 i wasn't arguing against the patch Oct 08 20:33:36 only that the commit message should reflect what its doing Oct 08 20:41:37 kergoth: yeah I explained it to tharvey earlier today we can explain more in commit message git commit --amend :) Oct 08 20:43:09 :) Oct 08 20:44:09 kergoth: whats your opinion on libtool 2.4 Oct 08 20:44:16 should I create a branch Oct 08 20:44:26 or merge it into master Oct 08 20:44:52 I am afraid if I create a branch the merge will destabilize master later Oct 08 20:45:03 1. not many devs will use the branch Oct 08 20:45:15 2. they may not care about 2.2 compat Oct 08 20:45:25 when fixing recipes for new libtool Oct 08 20:46:10 on master default will still be 2.2 so if anybody breaks a recipe with 2.4 that will get noticed pretty soon Oct 08 20:47:52 i like the idea of merging to master personally, not sure what others think Oct 08 20:48:22 khem, i'm afraid your analysis wrt the branch is correct, so probably better merge into master Oct 08 20:48:22 would be nice if you endorse my email then :) Oct 08 20:49:16 eFfeM: and I think since we are not changing defaults hell is not breaking on anybody :) Oct 08 20:51:29 khem wittten a small email and if it does not change default there should not be a problem; haven't yet tested it though so not too sure if I can sent an ack, but you have my blessing Oct 08 20:51:32 fwiw Oct 08 20:52:05 * eFfeM feels dev is for development :-) Oct 08 21:00:36 stupid Q: should RDEPENDS_append be RDEPENDS_${PN}_append ? (or but that one seems less likely: RDEPENDS_append_${PN}, or leave a sis?) Oct 08 21:01:36 _append_${PN} won't work. Oct 08 21:01:41 ${PN} isn't an override Oct 08 21:01:45 its just a variable named that way Oct 08 21:02:07 I'd do RDEPENDS_${PN} +=. if absolutely necessary RDEPENDS_${PN}_append, but its not likely necesary Oct 08 21:03:11 well was peeking at this: Oct 08 21:03:11 rootfs_ipk.bbclass:RDEPENDS_append = " ${@base_conditional("ONLINE_PACKAGE_MANAGEMENT", "none", "", "${IPKG_VARIANT} opkg-collateral", d)}" Oct 08 21:03:26 recipe_sanity seems to complain about this Oct 08 21:03:43 (and don't worry, won't commit a patch for it, this is way out of my league) Oct 08 21:04:50 oh, its fine ther. Oct 08 21:04:59 images don't generate packages Oct 08 21:05:04 so making it package specific wouldn't make much sense Oct 08 21:05:25 true :-) Oct 08 21:05:52 did a recipe_sanity run (see ML) was peeking at the result and trying to interpret them (and maybe fix a thing or 2) Oct 08 21:06:06 but not today any more 23.04 here Oct 08 21:10:12 calling it a day, cya tomorrow or so Oct 08 21:16:09 khem, you have the ability to commit the gdb-cross-sdk patch and amend the comments then? I'm still not sure how to explain why the fix works Oct 08 21:24:43 tharvey dont worry I will take care Oct 08 21:26:03 much appreciated Oct 08 21:26:17 np Oct 08 21:26:46 tharvey sometime soon i am buying a new machine with load of memory :) Oct 08 21:27:28 Can anyone else take a look/apply the patches I posted to: http://lists.linuxtogo.org/pipermail/openembedded-devel/2010-September/024657.html Oct 08 21:27:41 Or at least comment :P Oct 08 21:29:19 hi all , sorry this channel is not for that topic but i need to ask here because here C++ coders may help my question... Oct 08 21:29:23 how can i set default CRT version for my project & solution in visual studio 2010 ? someone know ? Oct 08 21:31:53 kelvie_ seems reasonable to me. let me see if I can queue it up in my patch queue Oct 08 21:32:52 JDuke128, M$ has nice knowledge base for such things search there Oct 08 21:33:11 sry :) Oct 08 21:43:06 khem`: Thank you, sir Oct 08 21:44:30 kelvie_, I do a weekly moppin of patchwork and if a patch is good and does not cause much pain I take it Oct 08 21:45:41 ok console-image boots good in qemu built using libtool 2.4 :) thats something Oct 08 21:48:16 hmmm the logic to detect image size is flawed when it comes to generating ext3 image Oct 08 21:49:08 my rfs it 285M untarred and its calling genext2fs -b 302064 ..... Oct 08 21:49:30 to generate the image and dies genext2fs: couldn't allocate a block (no free space) Oct 08 21:49:59 if I change the -b option so something above 400000 Oct 08 21:50:04 it works Oct 08 21:50:06 03Frans Meulenbroeks  07org.openembedded.dev * rf5080f0b25 10openembedded.git/recipes/shr/ologicd_git.bb: Oct 08 21:50:07 shr/ologicd_git.bb: added DESCRIPTION Oct 08 21:50:07 Signed-off-by: Frans Meulenbroeks Oct 08 21:55:46 03Frans Meulenbroeks  07org.openembedded.dev * r711204ea10 10openembedded.git/classes/module.bbclass: Oct 08 21:55:46 module.bbclass: RDEPENDS -> RDEPENDS_${PN} Oct 08 21:55:46 Signed-off-by: Frans Meulenbroeks Oct 08 22:04:36 heh reading this http://groups.google.com/group/android-x86/browse_thread/thread/44af30c18f04b828?pli=1 Oct 08 22:04:55 these folks seems to have copied this from OE Oct 08 22:05:39 all the commands and even extra sizes match :) Oct 08 22:07:44 heh Oct 08 22:12:15 wiki page tals abt bitbake 1.8 its not the latest I guess? Oct 08 22:12:31 wget http://download.berlios.de/bitbake/bitbake-1.8.18.tar.gz Oct 08 22:12:48 is it? Oct 08 22:13:09 1.10.0 is hte latest Oct 08 22:13:31 but you may find it nicer to just use the git repository Oct 08 22:13:37 easier to try other versions, apply fixes, etc Oct 08 22:14:16 Ok I have started checking out.. Oct 08 22:16:40 so we lack scipts to create build folder/configuration files etc.. Oct 08 22:16:59 according to wiki atleast Oct 08 22:17:01 there's a reason mentor includes the fact that we do as a value add in the marketing materials Oct 08 22:17:28 there are multiple tools floating around, none are particularly advanced, though the kaelios(sp?) one looks promising Oct 08 22:17:47 and of course poky has a setup script, other oe based projects do their own thing Oct 08 22:19:42 Crofton so you pinged me earlier Oct 08 22:19:45 whats up Oct 08 22:21:15 Crofton is your stuff moving along Oct 08 22:21:36 yikes, that recipe_sanity run is painful :) Oct 08 22:21:38 not surprising, but ouch Oct 08 22:23:13 kergoth: genext2fs comes from e2fsprogs I guess Oct 08 22:24:25 no Oct 08 22:24:32 genext2fs comes from genext2fs Oct 08 22:24:35 genext2fs != mkfs.ext2 Oct 08 22:24:39 two different tools Oct 08 22:24:59 right just found it Oct 08 22:25:04 rm work deletes all Oct 08 22:25:07 so its hard to find Oct 08 22:25:37 heh, yeah. should write a script that searches a dir of ipk files for a given pattern Oct 08 22:25:45 poke into the tar in the ar file, grep the file list Oct 08 22:25:50 would be easy to throw together Oct 08 22:25:56 could use it to search sysroot Oct 08 22:27:58 kergoth_ IMAGE_EXTRA_SPACE needs to increase Oct 08 22:28:31 my console image is not fitting into whatever size is computed Oct 08 22:28:38 it needs few kb extra Oct 08 22:29:06 I wonder how koen came to this number IMAGE_EXTRA_SPACE = 10240 Oct 08 22:29:07 the estimation is only that, an estimation, it can't perfectly determine the size without generating the filesystem twice Oct 08 22:29:11 dunno Oct 08 22:29:45 is it just 1024 and then a 0 Oct 08 22:29:52 may be Oct 08 22:30:11 these many blocks are added extra Oct 08 22:30:21 is block size 1k Oct 08 22:31:26 iirc block size is usually 512, but i don't know Oct 08 22:31:30 haven't messed with that stuff in ages Oct 08 22:34:38 it does du -ks Oct 08 22:34:47 must be 1k Oct 08 22:35:49 * khem` deserves a Trail mix bar today :) Oct 08 22:40:44 kergoth_: In textual how can I end the tab completion of nicks with say a , or : Oct 08 22:41:16 khem`: Preferences -> Interface -> Completion suffix Oct 08 22:41:52 kergoth_: heh I bet I did that and it did not work but now it does Oct 08 22:42:07 heh, dunno, worked fine here Oct 08 22:42:12 textmate needs to go Oct 08 22:42:22 is there a nice free editor for mac Oct 08 22:42:24 only thing i wish it did was a context sensitive completion, only end with ": " if at beginning of line Oct 08 22:42:29 some vi clon Oct 08 22:42:36 there's bbedit, textwrangler, etc Oct 08 22:42:37 oh Oct 08 22:42:39 macvim is *awesome* Oct 08 22:42:54 its just gvim, basically, but all mac-like Oct 08 22:42:57 :) Oct 08 22:44:31 * kergoth likes that his nice vim configuration works flawlessly on his macbook Oct 08 22:45:41 yeah macvim seems what I need :) Oct 08 22:46:22 i like textmate, but not enough to make me drop vim, i have too much invested in my setup Oct 08 22:46:23 hehe Oct 08 22:46:30 besides, this works portable Oct 08 22:46:32 portably Oct 08 23:17:13 hmm, i should set up a cron job to email me logs of the recipes / classes i have myself listed as maintainer of, on a regular basis Oct 08 23:49:03 heyo Oct 08 23:49:15 anyone around who uses or has used a quickcam pro 9000 with OE? Oct 09 00:00:39 aaaanyone ? Oct 09 00:19:28 03Chris Larson  07master * r516df11549 10openembedded.git/classes/image.bbclass: Oct 09 00:19:28 image.bbclass: revert the RDEPENDS change Oct 09 00:19:28 Image recipes don't emit packages, so it doesn't make any sense to make the Oct 09 00:19:28 RDEPENDS package specific. Oct 09 00:19:28 Signed-off-by: Chris Larson Oct 09 00:51:27 oh fuck Oct 09 00:51:28 damnit Oct 09 00:51:32 hmmm Oct 09 00:51:43 well, i can work around this.. Oct 09 00:51:47 not sure if its ideal, but.. Oct 09 00:51:48 hrm Oct 09 00:52:16 by default, pkg-config looks not *only* in PKG_CONFIG_SYSROOT Oct 09 00:52:20 it also looks in its default search path Oct 09 00:52:28 which allows pkg-config checks for -natives to be found Oct 09 00:52:46 which is likely a good thing, for tools like gnome-doc-utils, which is checked for with pkg-config, and then *run* Oct 09 00:53:04 but a prebuilt pkg-config isnt relocatable, it only ends up obeying the sysroot after that, so it only finds target stuff, not native stuff anyomre Oct 09 00:53:19 ideally pkg-config would be more cross aware, and configure scripts could do checks for build vs target Oct 09 00:53:31 but until then, guess i'll have to make it more relocatable or something Oct 09 00:53:43 or explicitly add the native paths via the PKG_CONFIG_PATHS Oct 09 00:54:22 but i need to determine the order of PKG_CONFIG_PATHS vs PKG_CONFIG_SYSROOT, since the target ones have to be preferred over the native ones Oct 09 00:54:29 * kergoth_ mutters Oct 09 01:18:41 kergoth: how about PKG_CONFIG_PATH or something Oct 09 01:19:06 <3 gmail's Insert feature of Canned Responses to add Acked-by to patches Oct 09 01:19:09 PKG_CONFIG_DIR Oct 09 01:19:28 yeah, thats what i was referring to, but would have to verify that those are checked after the sysroot Oct 09 01:19:29 would hope so Oct 09 01:19:45 me too Oct 09 01:19:55 and I think we have a hacked version of pkg-congih Oct 09 01:20:44 * kergoth_ ran a full relocation test script which builds each recipe in the task-depends.dot graph, in order, with all of each one's dependencies prebuilt in a different TMPDIR path Oct 09 01:20:47 takes a loooooong time Oct 09 01:23:43 * kergoth_ works on splitting up the image features patch into better chunks, with a different variable name for the package groups Oct 09 01:26:00 now its PACKAGE_GROUP_ and PACKAGE_GROUP_[optional] = 1 to make it optional Oct 09 01:26:07 so it can be used outside of images Oct 09 01:29:19 need to give some thought to how best to restructure rootfs_*.bbclass and image.bbclass Oct 09 01:29:37 much of what's in image.bbclass would be common to any root filesystem population, not solely "images" Oct 09 01:30:02 move that to rootfs.bbclass or something as a first pass, then make its functionality available as a function, populate a rootfs at this path Oct 09 02:58:17 kergoth: CC for nativesdk is ${SDK_ARCH}${SDK_VENDOR}-${SDK_OS} Oct 09 02:58:44 kergoth_: this translates to x86_64-linux-gcc on my build machine Oct 09 02:58:55 but there is no compiler called x86_64-linux-gcc installed Oct 09 02:59:05 it should have been simply gcc isnt it **** ENDING LOGGING AT Sat Oct 09 02:59:57 2010