**** BEGIN LOGGING AT Mon Mar 25 02:59:59 2013 Mar 25 08:09:29 good morning Mar 25 08:18:29 morning all Mar 25 08:35:23 bluelightning: hi Mar 25 08:54:32 morning all Mar 25 08:55:53 hrw, I try to build my own sdk, but I get a series of ERRORS on QA in the RDEPEND, like this: Mar 25 08:55:56 ERROR: QA Issue: packagegroup-btw-standalone-sdk-target rdepends on binutils-dev Mar 25 08:56:01 hi Mar 25 08:57:00 silvio_: those qa checks were added when I did not used OE so did not yet adapted to them Mar 25 08:57:29 hrw, thanks Mar 25 09:17:44 hi hrw Mar 25 09:17:49 hi silvio Mar 25 09:17:53 hi woglinde_ Mar 25 09:22:43 hi woglinde Mar 25 09:26:14 I am sending bunch of addons for OE today Mar 25 09:26:22 s/addons/new recipes/ Mar 25 09:26:48 hrw: nice Mar 25 09:27:03 some are cleaned resends Mar 25 09:28:19 hi bluelightning Mar 25 09:28:39 hi silvio_ Mar 25 09:30:51 good morning Mar 25 09:31:17 hi apelete Mar 25 09:31:59 bluelightning: hello Mar 25 09:32:38 I found some time to work on the jlime migration this week-end oe-core Mar 25 09:33:16 apelete: great, how is it going? Mar 25 09:34:11 bluelightning: nice spot with powertop Mar 25 09:35:01 hrw: I have an index here which lets me search most layers Mar 25 09:35:34 hopefully we should have a public metadata index some time this week... Mar 25 09:46:47 bluelightning: I pushed evrything I did on a personal repository: Mar 25 09:46:47 here is the layer (distro conf + image recipe) -> http://seketeli.fr/git/~apelete/meta-jlime.git/ Mar 25 09:46:47 machine conf is in a "add-nanonote" branch in meta-handheld like we discussed -> http://seketeli.fr/git/~apelete/meta-handheld.git/log/?h=add-nanonote Mar 25 09:46:47 and just FYI, I backed-up my builddir conf in a "builddir-backup" branch in oe-core -> http://seketeli.fr/git/~apelete/oe-core.git/commit/?h=builddir-backup&id=4e3c6c1a64b47fefd62fbd7c58787128144bf7fd Mar 25 09:46:51 Mar 25 09:47:52 bluelightning: can you remind me how to build uclibc with OE? Mar 25 09:48:47 bluelightning: it's just a start, I tried to build muffinman-image to test it and it failed (as expected :) ). bitbake output is here: http://paste.debian.net/244439/ Mar 25 09:49:07 hrw: TCLIBC = "uclibc" Mar 25 09:49:12 thanks Mar 25 09:50:08 building powertop/uclibc Mar 25 09:50:09 bluelightning: it seems I'll have to backport some recipes from oe-classic to make it work (stating with "echinus" recipe, according to the bitbake ouput) Mar 25 09:50:25 apelete: right, according to my lists there is no echinus recipe in OE-Core-based layers yet Mar 25 09:52:23 bluelightning: yeah, I couldn't finnd any echinus recipe, and there may be a few other recipes to port too Mar 25 09:53:41 hi bluelightning Mar 25 09:55:36 hi ericben Mar 25 09:56:28 bluelightning: if you got some time, could you review the code in http://seketeli.fr/git/~apelete/meta-jlime.git/, http://seketeli.fr/git/~apelete/meta-handheld.git/log/?h=add-nanonote and http://seketeli.fr/git/~apelete/oe-core.git/commit/?h=builddir-backup&id=4e3c6c1a64b47fefd62fbd7c58787128144bf7fd please ? Mar 25 09:56:29 I've been reported an issue when building qt-x11 on a host which has qtmobility installed : the qtmobility's conf file gets detected and thus qt gets -I/usr/include -I/usr/include/QtMobility which leads to a filaing compilation (that's using danny). Do you have an idea how we could force qt to not use the host's files ? From a quick analysis it seems in src/3rdparty/webkit/Source/WebCore/features.pri there is load(mobilityconfig, true) wit Mar 25 09:58:34 bluelightning: there is not much in there yet, but I may have done something wrong already. Mar 25 09:59:02 did someone built uclibc recently? Mar 25 09:59:08 | ./include/bits/arm_asm.h:6:1: error: expected identifier or '(' before '.' token Mar 25 10:00:18 ericben: I don't know of a specific fix, but clearly the way to fix it will be to find out where it is looking on the host and stop it from doing that... Mar 25 10:02:06 bluelightning: I see to way to solve that : either comenting out the faulty line in the .pri file or looking at the search path of qmake Mar 25 10:02:26 the second seems cleaner of course Mar 25 10:03:28 apelete: looks reasonable except for the distro inc files; I'm pretty sure you won't need either of them Mar 25 10:03:43 apelete: we have thumb handling in core AFAIK Mar 25 10:05:40 apelete: also in machine config I think you may need to be selecting DEFAULTTUNE instead of setting TARGET_ARCH Mar 25 10:08:35 apelete: looks reasonable except for the distro inc files; I'm pretty sure you won't need either of them Mar 25 10:08:35 you're talking about conf/distro/include/arm-thumb.inc and conf/distro/include/sane-toolchain.inc, right ? Mar 25 10:10:14 hrw hm which version? Mar 25 10:10:22 so far uclibc build here Mar 25 10:14:25 apelete: yes Mar 25 10:14:58 woglinde: 0.9.33+gitAUTOINC+9591091f150e8a86c47efbea5def21fa67fffb07-r8.4 Mar 25 10:16:05 woglinde: for armv7a Mar 25 10:16:31 will try for qemuarm Mar 25 10:16:51 bluelightning: ok, thanks. will take every you said into account, and start porting the needed recipes into the layer as soon as possible Mar 25 10:25:14 woglinde: built for qemuarm Mar 25 10:27:31 hrw yes Mar 25 10:32:15 hi likewise Mar 25 10:43:07 jo jama Mar 25 10:43:51 I'm getting "perl-dev: unsatisfied recommendation for perl-lib-dev" while trying to build an SDK, how to cope with this issue ? Mar 25 10:44:10 bluelightning: powertop/uclibc done, now powertop/ncurses instead of powertop/ncursesw Mar 25 10:44:47 hrw? Mar 25 10:44:52 done Mar 25 10:45:17 silvio_ latest master? Mar 25 10:45:19 woglinde: powertop 1.13 in OE was forced to link with ncurses not ncursesw and was uclibc buildable. Mar 25 10:45:31 woglinde: so I reimplemented those for 2.3 Mar 25 10:46:27 woglinde, nope. we freezed git commits to recover a stable build enviroment. I suppose I'm missing some packages in DEPENDS / RDEPENDS or what else ? Mar 25 10:47:45 or you missing the fixes rp made Mar 25 10:48:40 woglinde, I copied meta-toolchain-gmae and been adding packages to it, the original works fine, mine not yet. Mar 25 10:48:47 hrw: thanks for testing these cases Mar 25 10:48:56 bluelightning: thanks for reminding Mar 25 10:49:01 patch sent Mar 25 10:56:19 bluelightning: can you point me to recipe which can be used as alternative to powertop? Mar 25 10:56:35 hrw: koen says busybox... Mar 25 10:56:36 bluelightning: cause I do not see sense in u-a fields Mar 25 10:56:48 bluelightning: koen and his jokes ;d Mar 25 10:57:01 have you looked at git blame? Mar 25 10:57:11 not yet Mar 25 10:57:54 wow. busybox really got it. Mar 25 10:57:56 yes Mar 25 10:57:58 insane ;d Mar 25 10:58:06 pretty crazy yes Mar 25 10:58:10 added Mar 25 10:58:35 sent new ver Mar 25 10:58:48 ops, wrongly attached in thread Mar 25 11:00:44 hope that this ver is fine Mar 25 11:09:16 hrw: looks OK to me Mar 25 11:09:18 thanks Mar 25 14:50:21 is there an easy way to make an image with snae tools, not the busybox ones? Mar 25 14:52:55 Crofton|work: snae? Mar 25 14:53:16 sane :) Mar 25 14:53:21 normal Mar 25 14:54:54 Crofton|work: I would hardly see an embedded system without busybox ;-) Mar 25 14:55:08 heh Mar 25 14:55:28 but with multi G flash roots, we no longer need to save sapce Mar 25 15:00:41 Wish I could stop supporting legacy embedded systems with under 128M of root storage. So much easier now with so much space. So tired of shoe-horning stuff into NOR flash. Mar 25 15:06:23 georgem, sure, I am just looking to see if we have a fairly standard image for "large" embedded systems Mar 25 15:08:35 you can add bash, but afaik there is no package group to pull in all the other little tools you need Mar 25 15:14:56 Crofton|work: core-image-base is probably the closest thing Mar 25 15:17:57 yeah, I need to read the source Mar 25 15:18:09 after I learn more about multicast Mar 25 16:44:02 moin Mar 25 18:03:53 quit Mar 25 22:07:40 bluelightning: there ? Mar 25 22:07:51 onoffon: yep Mar 25 22:07:54 http://pastebin.com/j3kX3fYt Mar 25 22:08:19 bluelightning: have you seen this error Mar 25 22:08:46 khem: I haven't no Mar 25 22:09:15 whichever recipe is inheriting opie_git seems to have it Mar 25 22:10:43 is this new? Mar 25 22:10:51 relatively yes Mar 25 22:11:04 I am using angstrom-next Mar 25 22:11:12 which uses master of every layer Mar 25 22:11:38 unlike default angstrom which stays on stable snapshots or branches Mar 25 22:15:55 bluelightning: here is complete log http://paste.ubuntu.com/5647812/ Mar 25 22:17:50 bluelightning: is is some opie gitorious server issue Mar 25 22:19:02 khem: if the SRCREV values are set, and they should be since the distro config includes the opie srcrevs inc file, it shouldn't even be contacting the git server Mar 25 22:20:30 yes angstrom has include conf/distro/include/preferred-opie-git-versions.inc Mar 25 22:22:57 5 Mar 25 22:23:16 woglinde: hey Mar 25 22:23:23 * mr_science is finding several "fixed" bugs that aren't... Mar 25 22:23:36 our bugs that is... Mar 25 22:23:38 woglinde: I have a pending pull request for meta-java Mar 25 22:23:52 have you had time to take a look nto it Mar 25 22:27:41 uhm oh Mar 25 22:27:42 yes Mar 25 22:30:13 khem done Mar 25 22:30:23 thanks Mar 25 22:30:31 woglinde: I might have some more coming Mar 25 22:36:49 sure Mar 25 22:42:53 khem: fixed Mar 25 22:43:12 khem: layer.conf was setting OPIE_SRCREV and that meant the inc file's ?= did nothing Mar 25 22:43:34 bluelightning: ah I see Mar 25 22:43:41 bluelightning: thanks for quick fix Mar 25 22:43:50 khem: np Mar 25 22:44:08 I should really have some kind of automated build set up Mar 25 22:44:30 bluelightning: use angstrom-next if you dont hate it so much Mar 25 22:44:40 gn Mar 25 22:44:44 bluelightning: its a pig when it comes to layers Mar 25 22:44:45 khem: I don't hate it at all :) Mar 25 22:45:13 bluelightning: ok rebuild started Mar 25 22:45:14 bluelightning jenkins for the win Mar 25 22:45:31 woglinde: do you have jenkins scripts to build oe Mar 25 22:45:35 inclusive firering qemuimages up Mar 25 22:45:42 khem not yet Mar 25 22:45:49 zeckes server screwed uo Mar 25 22:45:52 up Mar 25 22:46:06 khem, JaMa has some Mar 25 22:46:06 so I could not get further with setting it up Mar 25 22:46:12 yes Mar 25 22:46:13 ka6sox: yes Mar 25 22:46:18 but he is already full Mar 25 22:46:20 I suspect pidge would not be happy if I didn't use autobuilder ;) Mar 25 22:46:28 :) Mar 25 22:46:31 but intel/linuxfoundation should have some Mar 25 22:46:35 they do Mar 25 22:46:39 machines to spare Mar 25 22:46:48 woglinde, what do you need? Mar 25 22:46:56 I have a few spare cycles. Mar 25 22:47:10 bluelightning: for me jenkins is better since internal github has plugins for jenkins Mar 25 22:47:16 but not for builbot Mar 25 22:47:22 khem: ah ok Mar 25 22:47:24 which is what autobuilder is based on Mar 25 22:47:30 indeed Mar 25 22:47:50 ka6sox hm 2 cpus and enough space for building meta-java Mar 25 22:47:56 and at least 2 gig Mar 25 22:47:58 ram Mar 25 22:48:00 time to get some sleep, gn all Mar 25 22:48:03 4 would be better Mar 25 22:48:06 bluelightning: I am setting up continuous integration where a commit will trigger a build and report back into the pull request Mar 25 22:48:08 woglinde, pm me. Mar 25 22:48:19 khem: sounds great! Mar 25 22:48:39 yeah gitlab is the tool for repo management Mar 25 22:48:51 its nice version of private github **** ENDING LOGGING AT Tue Mar 26 02:59:58 2013