**** BEGIN LOGGING AT Fri Apr 22 02:59:57 2011 Apr 22 10:41:09 hello Apr 22 10:42:51 I had build a filesystem "x11-image" for mini2440.... Now I had to install qt4 using command " bitbake -b qt4-embedded_4.7.1.bb -c install" but I m getting an errors.... Apr 22 10:43:22 can anyone help me how do I compile qt4 with x11-image for mini2440 in oe... Apr 22 10:44:26 sanket: you would want qt4-x11-free for x11 not qt4-embedded Apr 22 10:44:32 this are the errors.... Apr 22 10:44:34 http://pastebin.mozilla.org/1209814 Apr 22 10:44:36 qt4-embedded does not use X Apr 22 10:45:12 ok... I will try... btw what is the difference between both? Apr 22 10:45:17 sanket: also, do not use -b, it's for debugging only Apr 22 10:45:41 qt4-x11-free uses X for display, qt4-embedded uses the framebuffer directly Apr 22 10:57:19 but still I m getting an errors.... Apr 22 10:58:14 are you still using -b? Apr 22 11:01:28 yes Apr 22 11:01:42 without -b I m getting parsing error Apr 22 11:02:30 http://pastebin.mozilla.org/1209823 Apr 22 11:04:32 you should run bitbake qt4-x11-free Apr 22 11:04:42 no version, no .bb Apr 22 11:05:33 if it's not picking up the version you want you need to set PREFERRED_VERSION_qt4-x11-free in your local.conf Apr 22 11:06:52 but I will once again create image for qt4-x11-free na?... I want compiled package of qt4 in /tmp/work/armv4t-angstrom-linux-gnueabi/...... Apr 22 11:09:44 sanket: 13:02:20 < bluelightning> you should run bitbake qt4-x11-free Apr 22 11:10:23 that's not an image and it will stay in tmp/work unless you're using rm_work Apr 22 11:10:31 hey its working ... let c if it give any further errors Apr 22 11:10:35 or you can add -c build to prevent rm_work running Apr 22 12:15:38 03Michael 'Mickey' Lauer  07org.openembedded.dev * rc7c2ebead1 10openembedded.git/recipes/qt4/qt4-embedded.inc: qt4-embedded: enable support for the built-in LinuxInput keyboard driver Apr 22 15:16:32 lsb_release': No such file or directory Apr 22 15:16:42 how do i get out of this again? Apr 22 15:17:46 maybe touch local.conf Apr 22 15:27:25 hi all Apr 22 15:30:35 Crofton|work: install the lsb tools? Apr 22 15:30:59 no, some day rollover issue in angstrom-version Apr 22 15:31:05 need to force a re-parse Apr 22 15:31:15 and that causes the above error how exactly? Apr 22 15:31:25 i'm not seeing how a day rollover makes lsb_release vanish Apr 22 15:31:25 not sure Apr 22 15:31:33 maybe a cargo cult solution :) Apr 22 15:32:01 Tartarus, understands, i think Apr 22 15:32:11 messages are gone now Apr 22 15:32:19 Yeah Apr 22 15:32:22 bitbake -c clean angstrom-version Apr 22 15:32:40 * Crofton|work is thinking that did not work for him Apr 22 15:32:40 The problem is that the date stamp is in $S Apr 22 15:33:02 I had to also touch local.conf Apr 22 17:06:19 gm all Apr 22 17:07:30 #PV_U = `echo ${PV} | tr . _` Apr 22 17:07:49 Crofton|work: hmmm Apr 22 17:07:53 I need to convert a string with .'s into a string with _'s for a SRC_URI Apr 22 17:08:11 sed ? Apr 22 17:08:54 tr is slightly faster Apr 22 17:09:04 echo "khem.raj"|sed 's/\./_/' Apr 22 17:09:06 ;) Apr 22 17:09:10 but sed is called more frequently so it should be in cache anyway :) Apr 22 17:09:52 oh to replace all one would need to say g Apr 22 17:09:58 well, what I have with tr is n't working Apr 22 17:10:01 parse error Apr 22 17:10:09 echo "khem.raj"|sed 's/\./_/g' Apr 22 17:11:17 Crofton|work: try PV = `echo ${PV} | sed 's/\./_/g` Apr 22 17:13:06 for src_uri? i don't think `` is going to help him in a recipe :) Apr 22 17:13:21 PV_U = `echo ${PV} | sed 's/\./_/g'` Apr 22 17:13:22 yeah Apr 22 17:13:36 hints Apr 22 17:13:42 ${@'${PV}'.replace('.', '_')} Apr 22 17:13:57 urg Apr 22 17:14:08 I need to have PV and PV_U at the same time Apr 22 17:14:17 okay... Apr 22 17:14:24 I was beginning to wnder about what is python and what is sh Apr 22 17:14:48 *bitbake* needs src_uri, not just shell tasks, so you can't use shell things there Apr 22 17:16:10 that works Apr 22 17:16:18 SRC_URI = "http://www.ettus.com/downloads/uhd_releases/${@'${PV}'.replace('.', '_')}/images-only/UHD-images-${PV}.tar.gz" Apr 22 17:16:28 it is greek to me though Apr 22 17:17:34 I will be happier after Josh (the uhd guy) stops mucking with the fpga image paths Apr 22 17:19:16 thanks kergoth Apr 22 17:21:40 np Apr 22 17:22:07 it's just python. '${PV}' is a string, string objects have a replace method that replaces one thing with another Apr 22 17:23:42 kergoth: this fine line between python and shell confuses folks a bit truly Apr 22 17:24:04 kergoth: some tasks can be written in shell but some not Apr 22 17:24:50 khem: that's just due to how append/prepend behave. Apr 22 17:24:58 we're going to change it so a task is a list of functions Apr 22 17:25:04 run one by one Apr 22 17:25:07 I dont know if there is something we can do to improve this Apr 22 17:25:09 so it won't matter what you're appending to Apr 22 17:25:40 kergoth: these functions can be written in any script then ? Apr 22 17:25:44 it will also avoid the python indentation issues with append/prepending to python functions -- you have to know whether the function you're appending to uses tabs or spaces today Apr 22 17:25:45 they python perl shell Apr 22 17:25:50 we don't support perl. Apr 22 17:25:54 but otherwise yes Apr 22 17:26:06 hmm Apr 22 17:26:24 today if I look at recipes they look half shell half pythonic Apr 22 17:26:44 eh, not that many recipes have to resort to python Apr 22 17:26:53 it'd be nice to have a declarative way to specify things like do_split_packages Apr 22 17:27:11 yes Apr 22 17:27:16 I'd like to avoid the need for adding tasks from recipes. one common case is to adjust something in the source tree Apr 22 17:27:20 what we could do is provide hooks Apr 22 17:27:27 so if you define this function, it gets run after unpacking Apr 22 17:27:32 or aftering patching Apr 22 17:27:33 or whatever Apr 22 17:27:36 yes hooks would be better Apr 22 17:27:49 since the build process it same for every recipe Apr 22 17:27:54 there can be set of tasks Apr 22 17:28:01 which are to be done Apr 22 17:28:13 yeah. tasks should be defined outside the recipe, recipes should be as declarative as we can possibly make them Apr 22 17:28:16 and we can provide hooks into these tasks Apr 22 17:28:20 the recipe maintainer shouldn't have to worry about *how* bitbake works Apr 22 17:28:22 they just define things Apr 22 17:28:30 exactly Apr 22 17:28:40 i think we can gradually work toward that in the metadata Apr 22 17:28:47 yes Apr 22 17:28:57 that coupled with the forthcoming change to fix prepend/append of functions should get us most of the way toward a better situation, i think Apr 22 17:29:37 personally, I'd rather the python / task stuff not be in the metadata format at all. instead we could have python plugins which bitbake loads and uses hooks at various points Apr 22 17:29:43 but i doubt everyone would agree with that one :) Apr 22 17:29:58 but it would help to alleviate the problem you mention Apr 22 17:30:10 since python would be in python scripts, shell in metadata, other than inline python snippets Apr 22 17:30:40 it will be better but slightly complex to use for recipe writers Apr 22 17:30:48 how so? Apr 22 17:31:06 using plugins would just shift bits from classes into python modules, it wouldn't affect anything else Apr 22 17:31:19 one would have to maintain different files Apr 22 17:31:48 or do you mean the classes functionality Apr 22 17:32:31 inline python is ${@ .... ? Apr 22 17:32:41 or more ? Apr 22 17:32:48 yeah, ${@} Apr 22 17:33:20 I mean the non-recipe bound class work. assembling our tasks, QA checking, etc. all the global stuff could be done entirely in python Apr 22 17:33:39 so the classes would be more about defining things as well, just the things common to multiple recipes Apr 22 17:34:22 I think it's just the next logical step in making the metadata more declarative Apr 22 17:34:32 theoretically, it would also mean we could change file formats more easily Apr 22 17:34:57 today, how bitbake does its job is so bound to the metadata that switching file formats would be very difficult, since not all file formats might want to use key/value pairs + flags, for example Apr 22 17:35:25 anyway, just an idea :) Apr 22 17:36:07 the next steps for this area I think are adding hooks at commonly used points to remove addtasks from recipes, and adding a way to specify do_split_packages in a more declarative way Apr 22 17:37:00 and the function appending/prepending thing, of course Apr 22 17:43:24 kergoth: hi Apr 22 17:43:52 kergoth: yes that makes lot of sense Apr 22 17:44:20 kergoth: I have another weird problem :/ with older OE branch and newest OE bitbake/master I got a lot of unpackaged files (like when the file is not listed in any FILES_) Apr 22 17:44:52 kergoth: now it seems working after cherry-picking all "types" commits Apr 22 17:45:22 kergoth: but if it's known issue that newer bitbake won't play nice with older OE repos it would be worth documenting IMHO Apr 22 17:45:34 where? :) Apr 22 17:45:44 as far as i know it was only the FAKEROOT thing. nothing else should be broken Apr 22 17:46:05 kergoth: worst thing was that I've noticed in the end of image build when few recipes failed because file listed in CONFFILES was not included in any package Apr 22 17:46:16 kergoth: FAKEROOT is set in that branch Apr 22 17:46:18 that makes no sense Apr 22 17:46:28 hmm Apr 22 17:48:04 I did that before starting build.. and If I add more debug to package.bbclass it looks like if (not os.path.islink(file)) and (not os.path.exists(file)): is resolved false even when the file exists in workdir/package Apr 22 17:48:35 ie ./etc/frameworkd.conf in frameworkd-config-shr Apr 22 17:49:35 so I started to compare lib/ bin/ and classes/ with OE master and now I have ~20 commits and after those it works ok again Apr 22 17:50:17 kergoth: are you interested in that debug or should I push those 20 commits and forget about it? Apr 22 17:57:48 well, nothing bitbake could do could possibly affect os.path Apr 22 17:58:02 I think the root cause needs to be identified Apr 22 17:58:09 what branch are you using? Apr 22 17:58:55 shr/testing2011.1 Apr 22 17:59:12 distro/machine? Apr 22 17:59:13 try without todays patches (those after FAKEROOT) Apr 22 17:59:18 shr/om-gta02 Apr 22 17:59:18 k Apr 22 17:59:29 and fastest test case should be base-files Apr 22 17:59:43 where it fails with CONFFILES (fstab missing) Apr 22 18:00:00 okay, i'll try to repro Apr 22 18:00:00 thanks Apr 22 18:00:17 * JaMa restarting build from scratch with those patches on top Apr 22 18:00:19 i'm sure it doesn't hurt to have those commits, but its still strange that it would be needed Apr 22 18:00:47 kergoth: and btw I was able to reproduce it with current HEAD when I just checkouted that repo and rebuilt one single recipe with -b Apr 22 18:01:03 k Apr 22 18:01:17 I mean with tmpdir created with current HEAD.. Apr 22 18:02:41 and one more.. maybe it's some sort of race condition, because after switching to bitbake-1.12 I was able to build some recipe which was failing before Apr 22 18:03:18 and then few minutes later something failed the same as before, so not sure if 1.12/master diff is real cause Apr 22 18:03:53 but maybe it was just missing -c clean after swith for that failing one so discard last 3 messages Apr 22 19:52:48 damnit. gst-plugins-base depends on x264, which fails in configure Apr 22 19:53:10 http://pastebin.com/j2imAZwF Apr 22 19:53:16 is anyone able to compile x64? Apr 22 19:53:23 x264* Apr 22 20:07:58 hmm. Seems gcc hasn't yet been built. Odd. Apr 22 20:50:44 @ka6sox - you still around? Apr 22 20:54:03 darknighte, yup Apr 22 20:54:26 I'm surprised. Apr 22 21:08:33 anyone know off the top of their heads how to lower the verbosity of do_fetch ? Apr 22 21:26:08 kergoth: around ? Apr 22 21:26:26 kergoth: I need to find if the recipe I am compiling is a native or target Apr 22 21:26:34 in python whats the best way Apr 22 21:26:52 bb.data.getVar('PN', d, True) compared to something "-native" Apr 22 21:26:55 imo '${BPN}' == '${PN}' Apr 22 21:26:59 if that's true, you're in target Apr 22 21:27:11 alternatively, you can use oe.utils.inherits() or whatever it is Apr 22 21:27:14 hmm I am interested to know if I am native Apr 22 21:27:21 ah Apr 22 21:27:22 if oe.utils.inherits(d, 'native', 'nativesdk', ...): Apr 22 21:27:25 or whatever Apr 22 21:40:27 03Andrea Adami  07org.openembedded.dev * r7c99e43029 10openembedded.git/recipes/modutils/ (5 files in 3 dirs): Apr 22 21:40:28 modutils-collateral: remove stale overrides for poodle and tosa Apr 22 21:40:28 * recipe is obsolete und is not built anymore, bump PR though Apr 22 21:40:28 Signed-off-by: Andrea Adami Apr 22 22:31:33 using angstrom maintenance release No such file `fakeroot_1.14.5.orig.tar.bz2' is that supposed to be 1.15_1 ? Apr 22 22:32:22 or fakeroot_1.14.4.orig.tar.bz2 for example from here ftp://ftp.hr.debian.org/debian/pool/main/f/fakeroot Apr 22 22:33:18 Its eventually found here http://www.angstrom-distribution.org/unstable/sources/fakeroot_1.14.5.orig.tar.bz2 Apr 22 22:33:30 but it does say "unstable" Apr 22 22:37:23 debian bumps versions and removes old archives all the time Apr 22 22:38:56 kergoth: fair enough, once its fetched I suppose I'll never go looking again Apr 22 22:44:22 03Andrea Adami  07org.openembedded.dev * r60c7c7eeca 10openembedded.git/recipes/linux/ (linux-2.6.30/tosa/defconfig linux_2.6.30.bb): Apr 22 22:44:22 linux-2.6.30: remove stale/obsoleted references to tosa Apr 22 22:44:22 Signed-off-by: Andrea Adami Apr 22 22:44:24 03Andrea Adami  07org.openembedded.dev * r26d66cb098 10openembedded.git/recipes/linux/ (18 files in 2 dirs): Apr 22 22:44:24 linux-2.6.31: remove obsoleted collie references and files Apr 22 22:44:24 Signed-off-by: Andrea Adami Apr 22 22:44:35 03Andrea Adami  07org.openembedded.dev * ra56c19aabf 10openembedded.git/recipes/linux/ (234 files in 6 dirs): Apr 22 22:44:35 linux-rp: move to obsolete Apr 22 22:44:35 * qemu and zaurus were the typical targets Apr 22 22:44:35 * both did move to plain linux recipe Apr 22 22:44:35 * thanks RP for the outstanding work ! Apr 22 22:44:36 Signed-off-by: Andrea Adami Apr 22 22:44:43 03Andrea Adami  07org.openembedded.dev * r802a5a4f29 10openembedded.git/recipes/linux/ (318 files in 38 dirs): Apr 22 22:44:43 linux-kexecboot: move 2.6.2x and development 2.6.37 to obsoleted. Apr 22 22:44:43 Signed-off-by: Andrea Adami Apr 22 22:44:44 03Andrea Adami  07org.openembedded.dev * r067b569c24 10openembedded.git/recipes/linux/ (3 files): Apr 22 22:44:44 linux_2.6.29+2.6.30-rc4: remove stale tosa files Apr 22 22:44:44 Signed-off-by: Andrea Adami Apr 22 22:46:19 bluelightning: ping Apr 22 22:46:28 hi ant__ Apr 22 22:46:32 hey Apr 22 22:46:52 I catched a single opie override for Z Apr 22 22:47:01 ./opie-taskbar/opie-taskbar-images.inc:PIXMAP_SIZE_c7x0 = "-640x480" Apr 22 22:48:00 can we solve it using other vars e.g. MACHINE_DISPLAY_WIDTH_PIXELS = "640" Apr 22 22:48:00 MACHINE_DISPLAY_HEIGHT_PIXELS = "480" Apr 22 22:48:05 ? Apr 22 22:48:43 ant__: it would probably be neater, assuming all of those machines set those vars correctly Apr 22 22:49:15 ok, there is qte-common too... Apr 22 22:49:38 and task-opie, I see now Apr 22 22:50:46 ant__: what's in there? or are you just talking about machine-specific overrides? Apr 22 22:51:33 I'm cleaning the stale overrides, just grepping Apr 22 23:41:56 hmmm INHIBIT_DEFAULT_DEPS should remove deps added by any prevuous classes etc right ? Apr 23 00:00:55 03Andrea Adami  07org.openembedded.dev * r4dc6cc5675 10openembedded.git/conf/machine/hx2000.conf: Apr 23 00:00:55 hx2000: move to machine/nonworking Apr 23 00:00:55 * unfortunately support for the machine seems dead Apr 23 00:00:55 * linux-rp kernels have been obsoleted Apr 23 00:00:55 * machine needs fresh patches against recent kernels Apr 23 00:00:56 Signed-off-by: Andrea Adami Apr 23 00:26:09 khem: no. INHIBIT_DEFAULT_DEPS drops the default deps -- that is, the ones that come from base.bbclass. assuming the other classes use +=/=+ rather than _append/_prepend, you could just set DEPENDS = "" after the inherits, in addition to INHIBIT_DEFAULT_DEPS, to get it truly empty Apr 23 00:26:53 khem: the idea was we wanted removal of the base deps to be extremely explicit, and to make it hard to screw up (e.g. DEPENDS = vs += after the inherits) Apr 23 00:27:06 looking back, I rather dislike the approach, but it does get the job done Apr 23 00:32:52 kergoth: ok in gettext I have to take care of INHIBIT_DEFAULT_DEPS Apr 23 00:33:11 and check if INHIBIT_DEFAULT_DEPS is set then dont much with DEPENDS Apr 23 00:33:30 I realized that already when you were doing errands ? Apr 23 00:34:03 my wife sends me out for errands and 95% of the time I have to go second time and return that item I got in first place Apr 23 00:34:28 and funny part is that I always run into guys who are in same situation as me :) Apr 23 00:35:10 well time to drive home hopefull errandless today **** ENDING LOGGING AT Sat Apr 23 02:59:57 2011