**** BEGIN LOGGING AT Wed Jan 04 02:59:58 2012 Jan 04 05:24:54 agreed. I remember why I added that, but I doubt there's much need for it anymore Jan 04 05:25:07 also, at the time there were no real users of on-target development :) Jan 04 05:33:43 hello Jan 04 06:36:28 any one got a mesa package error http://pastebin.com/9c6ARZLc ,i am compiling it for uclibc Jan 04 10:02:58 morning all **** BEGIN LOGGING AT Wed Jan 04 10:54:15 2012 Jan 04 14:59:47 hi. can i use yocto with a thirdparty toolchain for my machine? Jan 04 15:02:49 i only see armv5te from poky but need (and have) one for armv7a Jan 04 15:20:17 Hi there Jan 04 15:20:59 I'm just beginning with yocto, and got stuck on a simple issue Jan 04 15:21:07 basically I need my git binary to have SSL support Jan 04 15:21:13 since I clone stuff over https Jan 04 15:21:25 however, it looks like libcurl was build without ssl support Jan 04 15:21:49 /meta/recipes-support/curl/curl_7.21.7.bb does contain "--without-ssl" Jan 04 15:22:05 so I changed that to with-ssl, recompiled, but it recompiles for the target device, not for the SDK :) Jan 04 15:22:18 (when I run bb curl) Jan 04 15:22:46 does anyone know the way to recompile the curl package for the SDK? I tried to set MACHINE=native or MACHINE=i686 without success... Jan 04 15:42:15 abustany: try bitbake curl-native -c cleanall && bitbake curl-native Jan 04 15:43:52 might be cool to have say bitbake -C cleanall curl-native Jan 04 15:44:06 where -C means run this clean target first, then do the default Jan 04 15:47:24 does that work also with forced compile? eg. bitbake -C compile -f curl-native Jan 04 15:48:27 walters: I've often wondered about that... Jan 04 15:50:11 walters: re: the -dev dependencies, we should split those into a separate package. There is a case for both Jan 04 15:52:09 RP__, can you reply on list so i can follow up later? Jan 04 15:52:31 walters: ok, I didn't realise this had moved there Jan 04 15:57:36 re the -dev dependencies.. it would be -really- nice if we had a dependency scanner that would search headers for their dependencies and add them as appropriate. I'm not sure how easy/difficult that is to do. Jan 04 15:57:54 (especially for dependencies that may be ifdef'd and/or broken intentionally...) Jan 04 15:58:20 any .a dependencies can't be discovered -- I think.. and shared library dependencies should already be discovered and can be inherited easily enough Jan 04 15:59:02 this still leaves a hole for things that require m4, make, etc to run in the dev environment.. so it would likely introduce additional efects... Jan 04 15:59:41 fray: we have the DEPENDS variable... Jan 04 15:59:59 yes, which from my understanding is being inherited into the RDEPENDS for a -dev package right? Jan 04 16:00:09 and that is the complaint -- there is too much stuff there in the general case Jan 04 16:00:22 fray: amongst other things, yes Jan 04 16:00:57 (it's a "safe" approach, but the alternative is above.. additional automated deps & manual intervention when necessary) Jan 04 16:01:02 fray: The complain is there are two cases - you want to build something that uses X and you want to build X Jan 04 16:01:09 yup Jan 04 16:01:23 Unfortunately we default to the latter which doesn't make much sense now Jan 04 16:01:37 Ya, the former is the more typical case.... Jan 04 16:03:41 Its due to us matching the -dbg and -dev dependency handling Jan 04 16:03:56 with -dbg you'd typically want the latter Jan 04 16:04:14 Ahh yes.. I forgot about that Jan 04 16:04:36 there is also the new-ish staticdev package with covers the .a case.. I suspect they're not getting any dependencies right now.. (which is likely fine) Jan 04 16:05:13 fray: correct Jan 04 16:08:55 has this been entered as a feature enhancement -- revising the -dev package dependencies in a better way? I know a solution can be devised.. just not sure of the time table we could do it.. Jan 04 16:09:04 * fray notes he's willing to help someone else with some of the dependency issues Jan 04 16:19:43 hello ppl Jan 04 16:20:04 I'm working on a pyside layer, and right now, on adding it's examples as a separate package Jan 04 16:20:37 I'd like not to "pull" the whole qt stuff in, but I need to depend on qt's fonts for the exampels Jan 04 16:20:40 *examples Jan 04 16:21:17 although I've added the several "qt4-embedded-fonts*" to RDEPENDS_${PN} Jan 04 16:21:36 no font's apear on the /usr/lib/fonts where they should Jan 04 16:21:46 what could I be doin' wrong? Jan 04 16:23:14 s0undt3ch: that sounds like it should work Jan 04 16:23:37 bluelightning: I've now issued -c clean on my image recipe to see if that changes anything Jan 04 16:23:55 s0undt3ch: I don't think that will do anything Jan 04 16:24:10 hmm Jan 04 16:24:34 presumably you're sure the RDEPENDS value is being used? i.e. it shows up in bitbake -e recipename | grep RDEPENDS Jan 04 16:25:32 yep, image size remains the same Jan 04 16:25:36 fray, walters: we should open an ehancement bug in the bugzilla Jan 04 16:25:43 bluelightning: cheking Jan 04 16:25:50 walters: I did reply on list Jan 04 16:26:52 bluelightning: RDEPENDS_python-pyside-examples="python-math qt4-embedded-fonts qt4-embedded-fonts-pfa qt4-embedded-fonts-pfb qt4-embedded-fonts-qpf qt4-embedded-fonts-ttf-dejavu qt4-embedded-fonts-ttf-vera python-pyside-embedded" Jan 04 16:26:55 bluelightning: so yes Jan 04 16:27:39 bluelightning: my image recipe should add that python-pyside-examples on INSTALL or also on RDEPENDS? Jan 04 16:27:55 bluelightning: the examples are being installed though Jan 04 16:28:02 s0undt3ch: IMAGE_INSTALL# Jan 04 16:28:06 er no # Jan 04 16:28:16 hmm Jan 04 16:28:26 s0undt3ch: did you bump PR or use "-c cleansstate python-pyside" after changing it before building it or the image again? Jan 04 16:29:05 bluelightning: I havent' touched the python-pyside recipe, only python-pyside-examples, and I issued -c clean python-pyside-examples Jan 04 16:29:10 do I also need the cleanstate Jan 04 16:29:42 s0undt3ch: oh right python-pyside-examples is a separate recipe Jan 04 16:29:49 s0undt3ch: well, try it Jan 04 16:30:01 bluelightning: hi, have you built qt-4.8 on qemux86-64? Jan 04 16:30:14 JaMa: no, do you experience problems? Jan 04 16:30:42 bluelightning: ERROR: Task do_cleanstate does not exist for target python-pyside-examples Jan 04 16:30:42 I have build native on x86-64 machine although I appreciate that's not the same Jan 04 16:30:54 s0undt3ch: cleansstate not cleanstate Jan 04 16:30:59 heh Jan 04 16:31:21 bluelightning: builds fine on arms but on qemux86-64 gcc fails Jan 04 16:31:31 argh Jan 04 16:31:44 JaMa: can you file a bug please? Jan 04 16:31:45 bluelightning: so I was wondering if it fails only with gcc-4.6 from meta-oe or if it's the same with gcc-4.6 from oe-core Jan 04 16:32:00 JaMa: I'll fire off a build now Jan 04 16:32:04 * s0undt3ch is creating the image again Jan 04 16:32:18 JaMa: embedded or x11-free? Jan 04 16:32:22 bluelightning: and someone (morphis) send me just this http://pastie.org/3125695 Jan 04 16:32:26 bluelightning: x11-free Jan 04 16:33:04 bluelightning: will file bug as soon as we narrow gcc version (I can build it with oe-core version too if you want) Jan 04 16:33:11 JaMa: hrm that paste doesn't look good either Jan 04 16:33:22 JaMa: ok #good idea Jan 04 16:33:41 argh I should really stop switching to a US keyboard at home, it messes up my typing at work :( Jan 04 16:34:26 bluelightning: yes, but I have it built for armv4t and armv7a and previous qt-4.8 recipes from your branch also on another host for armv7a without this problem Jan 04 16:35:08 bluelightning: ah.. but in both cases I didn't have qt installed on host Jan 04 16:35:26 bluelightning: so that's maybe the problem for morphis's paste Jan 04 16:38:29 bluelightning: dam, still nothing on /usr/lib/fonts Jan 04 16:39:04 bluelightning: the examples recipe only installs docs, so I'm adding to IMAGE_INSTALL python-pyside-examples-doc Jan 04 16:39:09 bluelightning: would that matter? Jan 04 16:39:22 ie, not python-pyside-examples package is created Jan 04 16:39:28 ie, no python-pyside-examples package is created Jan 04 16:39:36 s0undt3ch: ah yes since you're actually adding it to RDEPENDS_python-pyside-examples Jan 04 16:39:53 s0undt3ch: add it to RDEPENDS_${PN}-doc instead Jan 04 16:40:06 k Jan 04 16:40:08 trying that Jan 04 16:40:59 bluelightning: just in case you will get ICE too, here is my output http://paste.pocoo.org/show/529894/ Jan 04 16:42:01 JaMa: oh, it's an ICE, I had missed that... Jan 04 16:42:23 anyway, build started, we'll see what happens Jan 04 16:43:55 bluelightning: now I get satisfy_dependencies_for: Cannot satisfy the following dependencies for python-pyside-examples-doc: python-pyside-embedded Jan 04 16:45:03 leaving RDEPENDS as it was and adding the fonts to -doc Jan 04 17:00:18 bluelightning: JaMa: may I ask you why do we have 2 gcc_4.6 ? Jan 04 17:00:50 RP__, thanks Jan 04 17:01:24 s0undt3ch: er... what's your RDEPENDS line look like now? Jan 04 17:04:51 run_do_rootfs.30810: line 205: rootfs_install_all_locales: command not found Jan 04 17:05:05 * kergoth tests a stock image instead of his own Jan 04 17:06:03 bluelightning: http://paste.pocoo.org/show/529903/ Jan 04 17:06:13 bluelightning: that's the recipe Jan 04 17:06:15 as it is Jan 04 17:07:16 hmm, core-image-minimal fails for me Jan 04 17:07:28 * kergoth bisects Jan 04 17:08:13 bluelightning: ok, like that I had no complaints about the fonts Jan 04 17:08:37 kergoth: is the last commit in oe-core I think Jan 04 17:08:40 now it's python's datetime :) but that's simple and I've been answered regarding those deps on the ml Jan 04 17:09:02 oh, right, my screwup. local iamge.bbclass ;) Jan 04 17:09:04 * kergoth mutters Jan 04 17:13:01 bluelightning: I didn't understand those splitted RDEPENDS though, ie, I know why the -doc RDEPENDS should be there, but I didn't got why python-pyside-embedded was needed on RDEPENDS instead of the -doc RDEPENDS Jan 04 17:13:11 ant_work: meta-oe version has more linaro patches Jan 04 17:13:24 s0undt3ch: is it needed at all? Jan 04 17:13:37 bluelightning: python-pyside-embeded? Jan 04 17:13:51 s0undt3ch: the RDEPENDS on it, yes, is that needed at all? Jan 04 17:14:28 it seems to me if you got an error before it's because it doesn't exist... it only doesn't cause a problem now because the ${PN} package never ends up getting installed Jan 04 17:14:46 yeah, for those examples, python-pyside-embedded is the python qt bindings, the fonts are needed by the UI I guess, or at least, an example should be requiring it Jan 04 17:15:03 ah Jan 04 17:15:07 hmm Jan 04 17:15:29 well, I added python-datetime to RDEPENDS, let's see if that gets pulled Jan 04 17:15:38 not the -doc Jan 04 17:16:30 s0undt3ch: I'm still a little confused as to why you'd need to set RDEPENDS_${PN} to anything if it's actually empty (as it appears to be looking at the recipe) Jan 04 17:16:48 ("it" being the ${PN} package) Jan 04 17:16:54 empty? Jan 04 17:17:11 well, it contains no files... since you only install stuff to docdir Jan 04 17:17:20 or have I missed something? Jan 04 17:17:23 ah Jan 04 17:17:46 maybe it's not failing because my image depends on python-pyside-embedded Jan 04 17:18:01 and because the datetime module did not get pulled Jan 04 17:18:32 bluelightning: is there a way to make this package a PN package and not just a PN-doc one? Jan 04 17:18:56 I've just started to play with the bitbake recipes, so I might be very lost here Jan 04 17:19:13 s0undt3ch: is python-pyside-* already in oe-core? Jan 04 17:19:36 s0undt3ch: or you are using classic-oe? Jan 04 17:19:36 morphis: don't think so, It's a custom layer on mine Jan 04 17:19:42 s0undt3ch: ah ok Jan 04 17:19:52 s0undt3ch: you imported the recipes from oe-classic to your layer? Jan 04 17:20:07 morphis: yes, and adapted Jan 04 17:20:12 morphis: or from angstrom Jan 04 17:20:17 s0undt3ch: and they are working fine? Jan 04 17:20:17 morphis: http://dev.ufsoft.org/projects/yocto Jan 04 17:20:19 s0undt3ch: if that's what you want for this, you can set FILES_${PN}-doc = "" and then add the docdir path to FILES_${PN}, that would do it Jan 04 17:20:29 morphis: that's what I'm trying to find out ;) Jan 04 17:20:33 s0undt3ch: it could make more sense for an examples type package Jan 04 17:20:50 s0undt3ch: ah ok as I wrote them for oe-classic some time ago Jan 04 17:20:57 morphis: you did!? Jan 04 17:21:08 morphis: want to help out with this port? Jan 04 17:21:15 morphis: I can grant write access to my git repo Jan 04 17:21:28 morphis: and I'm pretty new to all this Jan 04 17:21:55 s0undt3ch: I can try to compile them Jan 04 17:22:05 morphis: well, they did compile Jan 04 17:22:06 but I don't have enough time for it right now Jan 04 17:22:10 ah ok Jan 04 17:22:14 morphis: at least for qemuarm Jan 04 17:22:27 and pacakges are generated? Jan 04 17:22:31 and if I pull the whole qtdemo and it's deps, the examples do run, at least the ones I tried Jan 04 17:22:40 ok Jan 04 17:22:52 I'm just trying to make it as small as possibnle Jan 04 17:23:08 I don't want the qt demo stuff, just whats needed for pyside Jan 04 17:23:41 you mean from qt4-x11-free? Jan 04 17:24:51 morphis: !? Jan 04 17:25:05 right now I'm interested in the framebuffer Jan 04 17:25:08 -qws Jan 04 17:25:13 then qt4-embedded Jan 04 17:25:24 as that's what I'll need for a project that is comming my way Jan 04 17:25:28 morphis: yeah Jan 04 17:25:31 you're wanting only the packages from qt4-embedded need to run pyside? Jan 04 17:25:40 bluelightning: http://paste.pocoo.org/show/529914/ something like this? Jan 04 17:25:50 morphis: yes Jan 04 17:26:03 morphis: and the python modules pyside needs Jan 04 17:26:14 since python is split up into multiple packages Jan 04 17:27:12 ah ok Jan 04 17:27:32 so you want to adjust the packaging of pyside Jan 04 17:27:37 morphis: yeah Jan 04 17:28:10 morphis: and I'm currently building 1.0.2 which is now considered old, but since that's what was available on the old oe, first that one, then, the lastest pyside Jan 04 17:28:28 s0undt3ch: rather FILES_${PN} += "${D}${docdir}" Jan 04 17:28:43 s0undt3ch: ok Jan 04 17:28:53 bluelightning: JaMa told me you were working on qt 4.8 Jan 04 17:29:02 bluelightning: and leave de do_install as it was right? Jan 04 17:29:10 s0undt3ch: also you don't need the PROVIDES= line Jan 04 17:29:24 s0undt3ch: since every recipe PROVIDES ${PN} by default Jan 04 17:29:24 k Jan 04 17:29:30 morphis: that's right Jan 04 17:29:52 bluelightning: I builded it here and it failed cause webkit wants to link against libQtSensors Jan 04 17:30:11 morphis: JaMa sent me a paste with your error in it, he seems to think it's attempting to link to host libs, it kind of looks like that Jan 04 17:30:12 but afaik libQtSensors is part of QtMobility which is not part of the Qt Core right now Jan 04 17:30:49 morphis: yuck... I wonder why it's attempting to link to it then? Jan 04 17:31:04 but I am wondering how that should work to build webkit with QtMobility support when Qt needs to be build first? Jan 04 17:31:13 I don't know Jan 04 17:31:30 morphis: I have to admit I know little about webkit/qtwebkit Jan 04 17:31:32 there seems to be some autodetection inside the QtWebKit stuff for QtMobility Jan 04 17:31:54 bluelightning: me too Jan 04 17:32:28 bluelightning: I should ignore WARNING: For recipe python-pyside-examples, the following files were installed but not shipped in any package: Jan 04 17:32:30 morphis: right... probably left over from when qtwebkit was separate; I can't see how you could easily build qtmobility first if that depends on qt Jan 04 17:32:39 yes Jan 04 17:32:40 for the example files Jan 04 17:32:50 s0undt3ch: no, if you're getting that something is wrong Jan 04 17:32:58 hmm Jan 04 17:33:14 ah, yes, an error, bellow Jan 04 17:33:17 bluelightning: I will remove qtmobility from my host and try building Qt again Jan 04 17:33:31 * opkg_install_cmd: Cannot install package python-pyside-examples. Jan 04 17:33:57 s0undt3ch: ah, I made a mistake, should have been FILES_${PN} += "${docdir}" Jan 04 17:34:01 s0undt3ch: can you give me the pull url for your layer? I don't find it on the link you gave me Jan 04 17:34:09 s0undt3ch: sorry Jan 04 17:34:48 morphis: http://dev.ufsoft.org/projects/meta-qt is de dev page for it, the pull url git://dev.ufsoft.org/oe/yocto/meta-qt.git Jan 04 17:35:02 bluelightning: k, trying that Jan 04 17:36:48 s0undt3ch: there is no recipe in the layer; maybe you forgot to push it? Jan 04 17:37:02 morphis: I've also started a mini2440 layer, so the initial url I gave you was for both Jan 04 17:37:04 morphis: lemme see Jan 04 17:38:52 morphis: pull again Jan 04 17:39:56 s0undt3ch: ok, thanks Jan 04 17:39:59 anyway I have to leave Jan 04 17:40:00 bye Jan 04 17:40:04 morphis: np Jan 04 17:40:21 morphis: if you find anything wrong, please let me know or submit a bug report Jan 04 17:42:20 bluelightning: satisfy_dependencies_for: Cannot satisfy the following dependencies for python-pyside-examples: python-pyside-embedded * Jan 04 17:42:59 s0undt3ch: right, then that specific named package must not exist Jan 04 17:43:14 hmm Jan 04 17:43:52 s0undt3ch: maybe python-pyside-embedded ends up being empty? Jan 04 17:44:26 bluelightning: actually "ls -lah tmp/deploy/ipk/* | grep -i pyside" Jan 04 17:44:38 bluelightning: libpyside-python2.6 Jan 04 17:44:46 no python-pyside-embedded Jan 04 17:45:25 right, I guess that's the issue Jan 04 17:46:08 there's a python-pyside in PROVIDES from python-pyside.inc Jan 04 17:49:03 bluelightning: changing python-pyside-embedded to python-pyside make it fail Jan 04 17:49:11 NoProvider: python-pyside-examples Jan 04 17:49:19 Missing or unbuildable dependency chain was: ['python-pyside-examples', 'python-pyside'] Jan 04 17:50:05 but the generated image had PySide installed :\ !? Jan 04 17:55:39 eu-readelf: shdr->sh_size > 0 && shdr->sh_addr >= phdr->p_vaddr && (shdr->sh_addr + shdr->sh_size <= phdr->p_vaddr + phdr->p_memsz)) Jan 04 17:55:47 oops Jan 04 17:57:05 s0undt3ch: PROVIDES is build time, not runtime Jan 04 17:57:14 s0undt3ch: RDEPENDS is runtime Jan 04 17:57:49 s0undt3ch: you need to find the appropriate runtime package to depend on and use that instead of python-pyside(-embedded) in RDEPENDS_${PN} Jan 04 17:57:52 bluelightning: dam, how do I make a package depend on pyside!? Jan 04 17:58:02 well Jan 04 17:58:51 bluelightning: libpyside, libpyside-python, libpyside-python, libpyside-python2.6, none of these work Jan 04 17:59:49 s0undt3ch: libpyside-python2.6 should work, didn't you already verify that that exists? Jan 04 18:00:03 I did, still failed Jan 04 18:00:55 bluelightning: Missing or unbuildable dependency chain was: ['libpyside-python2.6'] Jan 04 18:01:34 which is the generated ipk name Jan 04 18:01:52 s0undt3ch: can I have a look at your python-pyside recipe? Jan 04 18:02:41 bluelightning: sure, http://dev.ufsoft.org/projects/meta-qt/repository/revisions/master/entry/recipes-qt/pyside/python-pyside-embedded_1.0.2.bb Jan 04 18:06:11 s0undt3ch: so looking at that, if it does not produce a python-pyside-embedded package it has to be broken AFAICT Jan 04 18:06:33 s0undt3ch: do you get any packaging warnings when you build python-pyside-embedded ? Jan 04 18:06:57 bluelightning: dam, if I add python-pyside-embedded to IMAGE_INSTALL, pyside exists on the generated image Jan 04 18:07:05 and it's usable Jan 04 18:07:22 at least import PySide which is the minimum, works Jan 04 18:07:42 hmm, lemme clean it and rebuild it Jan 04 18:08:00 then I have no idea why adding python-pyside-embedded to RDEPENDS_${PN} wouldn't work Jan 04 18:08:20 dam Jan 04 18:09:44 s0undt3ch: wait a sec Jan 04 18:10:00 k Jan 04 18:10:21 s0undt3ch: maybe an all-arch package depending on an arch-specific package won't work, that could be the source of your issues Jan 04 18:10:33 (your recipe does "inherit allarch") Jan 04 18:10:36 hmm, so remove that inherit? Jan 04 18:11:28 removed it, recreating image Jan 04 18:11:40 I think you have to in this instance if you really want that dependency relationship Jan 04 18:11:54 I could be mistaken but I can't think of another reason why this would not work Jan 04 18:12:15 we'll see if that's the issue in some mins :) Jan 04 18:13:04 no complains from building the image, let's run it in qemu Jan 04 18:13:56 image is quite small :| Jan 04 18:14:14 yep, no PySide in it :\ Jan 04 18:15:49 bluelightning: no more ideas right? Jan 04 18:16:28 s0undt3ch: can you pastebin the current version of your recipe? Jan 04 18:16:46 bluelightning: http://paste.pocoo.org/show/529956/ Jan 04 18:17:09 did you have it in IMAGE_INSTALL ? Jan 04 18:17:19 by some means Jan 04 18:17:21 khem: right now, no Jan 04 18:17:23 s0undt3ch: yeah what's your current IMAGE_INSTALL? Jan 04 18:17:31 then do it Jan 04 18:17:33 khem: I'm trying to make the -examples package pull it Jan 04 18:17:56 IMAGE_INSTALL += " \ Jan 04 18:17:56 busybox \ Jan 04 18:17:56 psplash \ Jan 04 18:17:56 python-math \ Jan 04 18:17:56 python-pyside-examples \ Jan 04 18:17:57 " Jan 04 18:18:26 bluelightning, khem: that's from the image recipe Jan 04 18:19:31 s0undt3ch: so did the examples package get installed in your image? Jan 04 18:19:40 bluelightning: yes, it did Jan 04 18:20:09 s0undt3ch: does your image have the packaging system installed? Jan 04 18:20:31 bluelightning: don't think so, those are the only things in install right now Jan 04 18:20:54 hmm, I was wondering about querying the packaging system Jan 04 18:20:56 it's requiring recipes-core/images/core-image-minimal.bb Jan 04 18:21:04 oh right, then the answer is no... Jan 04 18:21:36 I can add it though, opkg? Jan 04 18:23:04 yes, you'll need to comment out the line in core-image-minimal.bb that removes the packaging data files too Jan 04 18:23:32 bluelightning: what other, not that big, image can I inherit from then? Jan 04 18:23:45 so I don't need to comment out the line you're refering to Jan 04 18:23:57 I'm not sure there is one right now, it's something I want to tidy up Jan 04 18:24:15 the others end up building X which in your case is probably not desirable Jan 04 18:24:21 does python-pyside-examples RDEPEND on missing package ? Jan 04 18:24:53 khem: well, python-pyside-embedded only generates libpyside-python2.6 Jan 04 18:25:09 khem: no python-pyside-embedded package is created Jan 04 18:25:23 but I can't seem to depend on libpyside-python2.6 Jan 04 18:25:25 it fails Jan 04 18:25:38 s0undt3ch: I can't see how that can work, the recipe indicates it produces an ${PN} package Jan 04 18:25:48 ok Jan 04 18:25:53 I'll clean it and rebuild it Jan 04 18:26:07 let's see Jan 04 18:26:09 unless debian library naming is munging things Jan 04 18:26:18 quite some more mins though Jan 04 18:26:20 dunno Jan 04 18:26:23 what does bitbake -e your.bb | grep ^PACKAGES= says Jan 04 18:26:28 khem: Jan 04 18:26:31 looking Jan 04 18:27:04 khem: PACKAGES="python-pyside-examples-dbg python-pyside-examples python-pyside-examples-doc python-pyside-examples-dev python-pyside-examples-staticdev python-pyside-examples-locale" Jan 04 18:27:28 I mean for the missing one Jan 04 18:27:32 ah Jan 04 18:27:40 s0undt3ch: another way to introspect these packaging-end issues is to look in tmp/work/*/${PN}-${PV}/packages-split/ Jan 04 18:27:53 khem: PACKAGES="python-pyside-embedded-dbg python-pyside-embedded python-pyside-embedded-doc python-pyside-embedded-dev python-pyside-embedded-staticdev python-pyside-embedded-locale" Jan 04 18:28:05 s0undt3ch: might be worth looking at that dir for python-pyside-embedded Jan 04 18:28:50 so add python-pyside-embedded to either RDEPENDS or to IMAGE_INSTALL Jan 04 18:29:19 if you want the package to be names something else then munge the PACKAGES variable in recipe Jan 04 18:31:10 khem: which RDEPENDS? the examples recipe has it -> http://paste.pocoo.org/show/529956/ or you mean RDEPENDS instead of RDEPENDS_${PN}? Jan 04 18:32:01 bluelightning: http://paste.pocoo.org/show/529962/ Jan 04 18:32:33 bluelightning: broken build? Jan 04 18:32:45 s0undt3ch: well, that dir listing looks ok Jan 04 18:32:57 rebuilding pyside Jan 04 18:33:26 s0undt3ch: with cleansstate first? Jan 04 18:33:32 I'd recommend that Jan 04 18:33:39 bitbake -c clean -c cleansstate python-pyside-embedded Jan 04 18:33:46 good? Jan 04 18:33:51 omit -c clean Jan 04 18:33:57 k Jan 04 18:34:03 -c cleansstate performs the do_clean task Jan 04 18:34:08 ok Jan 04 18:35:26 hmm, it seems that rebuilding the image after the cleanstate is not making pyside also rebuild(since it was cleansstate'd) Jan 04 18:36:01 yep, python-pyside-embedded was not rebuilt Jan 04 18:36:36 forcing -> MACHINE=qemuarm bitbake -k python-pyside-embedded Jan 04 18:36:45 huh!? Jan 04 18:36:59 it didn't compile Jan 04 18:37:34 like it was already compiled Jan 04 18:37:42 cleansstate again Jan 04 18:37:50 recompiling Jan 04 18:37:52 er... ??? Jan 04 18:38:43 bluelightning: I think it was because of that -c clean before Jan 04 18:38:58 s0undt3ch: which one? Jan 04 18:39:14 bluelightning: I issued both -c clean and -c cleansstate Jan 04 18:39:27 I don't think I had re-run with just -c cleansstate Jan 04 18:40:11 I don't think that would make any difference, unless it just resulted in performing the do_clean task only and the sstate checksum did not change Jan 04 18:40:17 ok, now, making my image rebuild is also rebuilding python-pyside-embedded Jan 04 18:40:34 bluelightning: it probably just did the clean command Jan 04 18:40:40 ie, the first one Jan 04 18:40:50 don't know if bitbake acumulates commands Jan 04 18:41:04 hmm, we should probably fix that Jan 04 18:41:09 either it should run both or error out Jan 04 18:41:14 yep Jan 04 18:41:15 :) Jan 04 18:41:29 anyway, it's compiling now Jan 04 18:41:43 good, I was about to tear my hair out... Jan 04 18:41:58 I'll say something when it finishes Jan 04 18:42:22 bluelightning: well, don't pull your hair :) and Thanks for your help so far! Jan 04 18:42:34 no worries :) Jan 04 18:42:39 khem: you too. Thanks! Jan 04 18:43:35 if we still have trouble with sstate checksums not changing (i.e. -c clean and despite a variable value having changed, the sstate package being used instead) then we really need to fix that as well Jan 04 18:45:44 there are still cases where bitbake was unable to automatically determine the variable dependencies, but we haven't added explicit ones Jan 04 18:45:52 hopefully not too many, but.,. Jan 04 18:53:47 by the way, anyone who worked with a mini2440 around? And has anyone had issues flashing a unit using a "USB to Serial" cable? Jan 04 19:02:26 kergoth: it just seems all too often people do -c clean and then rebuild and it doesn't... I have nothing to quantify it though Jan 04 19:02:51 odd Jan 04 19:52:03 dam, bluelightning is gone Jan 04 19:52:09 khem: still around? Jan 04 19:52:42 after rebuilding python-pyside-embedded, there's still no python-pyside-embedded ipk Jan 04 20:11:46 s0undt3ch, morphis did pyside recipes but he's not online currently Jan 04 20:12:10 GNUtoo: yeah, I talked to him today and gave him the link to my git layer Jan 04 20:12:17 ok Jan 04 20:12:25 GNUtoo: hopefully he'll come around and help me solve these issues Jan 04 20:15:45 yes maybe tomorrow Jan 04 20:16:01 else you could write a mail Jan 04 20:21:56 GNUtoo: yeah, that's probably even better Jan 04 21:03:11 GNUtoo, khem: dam, removed any file which contained pyside-examples, then did a -c cleansstate, rebuilt the image, and now, python-pyside-embedded got pulled into the image Jan 04 21:13:39 ok Jan 04 23:28:38 kergoth: ping? Jan 05 00:08:59 RP__, I'll take a look, probably tomorrow Jan 05 00:09:11 thanks for the reproduction patch, that makes it a lot easier Jan 05 00:10:59 * kergoth is unfortunately pretty swamped at work lately, the majority of his team has been sucked away to other projects recently Jan 05 00:11:38 kergoth: np, I'm hoping you can reproduce and see whats wrong easily :) Jan 05 00:11:44 here's hoping Jan 05 00:12:09 kergoth: the key seems to be to trigger a subset of recipes failing to parse (the reproducer makes all the -natives fail) Jan 05 00:12:37 shutdown logic with threads can be a pain, and unfortunately python uses an internal thread to feed any queue object. if you're not careful you can deadlock due to those threads never returning Jan 05 00:12:46 ah, interesting Jan 05 00:14:11 kergoth: its something like that which is going wrong... **** ENDING LOGGING AT Thu Jan 05 02:59:57 2012