**** BEGIN LOGGING AT Tue Apr 15 02:59:59 2014 Apr 15 06:17:40 if I want to provide an alternative location for some (but not all) git repos used by my build, is a PREMIRROR pattern for each of those repos the best way to go? or should I create .bbappend-files for those recipes and override SRC_URI? Apr 15 06:18:11 (for other repos, i'd like to use the BB_GENERATE_MIRROR_TARBALLS function and have them be fetched as normal files from the PREMIRROR) Apr 15 06:26:12 mago_: I think premirror could do what you want Apr 15 06:27:55 koen: was it a weird question? im just wondering how i'm suppose to be able to develop efficiently, clearly if all repos are turned into tarballs, i cannot easily edit them. What is your typical development workflow in OE when working with various recipes? Apr 15 06:42:31 make a patch in ${S}, send a patch upstream and add the patch to the recipe as well Apr 15 06:42:43 but I only do integration Apr 15 06:42:48 I'm not a developer :) Apr 15 06:56:13 koen, but making a patch from ${S} when it's not a git repo would mean finding the original sources, unpacking it elsewhere & diffing? Apr 15 06:58:51 no Apr 15 06:59:05 git init . ; git add . ; git commit -a -m "foo" Apr 15 07:02:02 i see, that would work. and bitbake / OE does not care if the ${S} dir is dirty? Apr 15 07:02:57 not at all Apr 15 08:24:14 goody, more patches fixing symptoms instead of the actual problem Apr 15 08:25:30 Using meta-qt5, I'm trying to get qmake to play with QtCreator. I've added "nativesdk-packagegroup-qt5-toolchain-host" to TOOLCHAIN_HOST_TASK, but when I add the qmake binary in QtCreator, it complains about missing mkspecs.. Apr 15 08:29:56 * kroon tries adding qtbase-mkspecs to TOOLCHAIN_TARGET_TASK Apr 15 08:37:15 * kroon tries inherit populate_sdk_qt5 in image recipe Apr 15 08:59:22 morning all Apr 15 09:00:36 kroon: there's a clone of meta-qt5 that atleast have that working for qt 5.1.1, https://github.com/Gagi2k/meta-qt5 Apr 15 09:01:30 I'm sure sure exactly what the diff from upstream is Apr 15 09:02:27 I want to look at that some day though to try to make it work nicely upstream (if that is not the case already) Apr 15 11:26:35 erbo, inheriting populate_sdk_qt5 in the image recipe did the trick Apr 15 11:54:25 Project ERROR: Unknown module(s) in QT: opengl, for qt3d_git.bb, do_configure, using meta-qt5 master Apr 15 12:23:05 i know OE is OS-independent for developers, but is there a unbutu server version that is best suited for working OE ? Apr 15 12:34:46 afournier, i've used Ubuntu server 12.04 LTS to build OE stuff, worked fine. But any ubuntu variant should be OK Apr 15 12:35:12 mago__: thanks for the info i was going to install that version btw :) Apr 15 15:16:26 Good morning everyone. Apr 15 15:17:36 I got a question about bitbake recipe file syntax. In Apr 15 15:17:44 CONFFILES_${PN} = "${sysconfdir}/fstab ${@['', '${sysconfdir}/hostname'][(d.getVar('hostname', True) != '')]}" Apr 15 15:18:53 The second part of string,${@['', '${sysconfdir}/hostname'][(d.getVar('hostname', True) != '')], what does it mean? Apr 15 15:19:19 Is it bb syntax or Python syntax? Apr 15 15:20:28 DavieT: it's bitbake syntax which is used to run python Apr 15 15:20:56 i.e. ${@ ... } tells bitbake to consider ... as python to be executed when expanding the variable value Apr 15 15:22:14 So the remain part of line is executed as Python? ['', '${sysconfdir}/hostname'][(d.getVar('hostname', True) != '')] Apr 15 15:23:23 Thanks bluelightning! What is the keyword for this Python syntax? Apr 15 15:24:12 DavieT: yes Apr 15 15:24:17 DavieT: this is BitBake's in-line python syntax Apr 15 15:24:41 http://www.yoctoproject.org/docs/1.6/bitbake-user-manual/bitbake-user-manual.html#inline-python-variable-expansion Apr 15 15:24:44 Any ref links?Have to learn some bb' in-line Python finally Apr 15 15:24:54 Great! Thanks a lot! Apr 15 15:25:33 that's from the upcoming 1.6 release version of the docs, but is equally applicable to current stable / older releases Apr 15 16:29:16 moin Apr 15 16:30:22 was having trouble with the meta-ti/beagleboard stuff... Apr 15 16:32:00 so i ended up forking meta-beagleboard and made a bootable build with working hdmi Apr 15 16:32:50 switched u-boot-denx to static source release and made a new kernel recipe Apr 15 16:34:27 i think the other kernel recipes need a uEnv maybe? nothing booted out of the box, although it seems like the mainline 3.8 recipe *should* work Apr 15 16:35:25 anyway, not being a clone of denix i made it as simple as possible Apr 15 16:54:05 Should PACKAGE_GROUP basically just be replaced with FEATURE_PACKAGES nowadays, or does it have a different meaning now ? Apr 15 16:54:43 I'm asking because of this warning I get: "WARNING: PACKAGE_GROUP is deprecated, please use FEATURE_PACKAGES instead" Apr 15 16:55:28 yes, just rename Apr 15 16:55:40 was to avoid confusion with packagegroup- recipes Apr 15 16:56:49 kergoth, ok, thanks Apr 15 17:06:21 The old line in the recipe was actually PACKAGE_GROUP_qtcreator-debug, but I assume the same substitution applies.. Apr 15 17:06:41 I don't understand the question Apr 15 17:06:55 FEATURE_PACKAGES_qtcreator-debug ... Apr 15 17:06:58 there is no PACKAGE_GROUP variable, only PACKAGE_GROUP_something Apr 15 17:07:04 so this is exactly what should be changed, yes Apr 15 17:07:23 gotcha Apr 15 17:13:54 kroon: it's covered in the migration guide FYI: http://www.yoctoproject.org/docs/1.6/ref-manual/ref-manual.html#moving-to-the-yocto-project-1.6-release Apr 15 17:14:22 (not that I would automatically expect users to look there, but it's a good place to check) Apr 15 17:16:27 bluelightning, aha, good to know Apr 15 17:16:49 bluelightning: if you got a minute, could you please try 2 patches I sent to the list? thanks! Apr 15 17:17:00 denix: sure, will try now Apr 15 17:30:33 denix: hmm, that seemed to be the same; but I realised I was still using old gcc/binutils from a previous test, so I'm just trying with current ones now Apr 15 17:32:38 bluelightning: hmm, I was using the default gcc/binutils, not sure it matters. the way I tested is to make the TMPDIR very long and build a kernel in it... Apr 15 17:33:36 right, I am doing the same (effectively add your changes then "bitbake u-boot virtual/kernel") Apr 15 17:34:07 binutils/gcc ought not to make a difference but I'd rather be sure I'm testing the same thing Apr 15 17:39:26 bluelightning: ok, never mind, it still fails - I was fooled by sstate the first time :( Apr 15 17:39:45 denix: ah, ok Apr 15 17:39:59 at least that rules out the uimage completely Apr 15 17:40:23 indeed, so something is inside the kernel Apr 15 17:59:14 bbl **** ENDING LOGGING AT Wed Apr 16 02:59:58 2014