**** BEGIN LOGGING AT Wed Oct 01 03:00:00 2014 Oct 01 03:41:29 hi guys Oct 01 03:41:33 I get a strange error Oct 01 03:41:47 Recipe is trying to change PV from '1.0' to '10.1.3'. This will cause do_package_write_* failures since the incorrect data will be used and they will be unable to find the right workdir. Oct 01 03:41:54 but I have no such thing in my recipe Oct 01 04:07:03 gebreselaisi: you have it in your sysroot Oct 01 04:25:56 JaMa|Off, it was adding the same files twice in FILES_PN Oct 01 04:33:46 that's not what this error means Oct 01 05:21:36 well, that fixed it Oct 01 07:17:13 Morning all Oct 01 07:18:28 I see that $SDKTARGETSYSROOT in the sdk env. script got fixed recently. However, $PKG_CONFIG_PATH still looks incorrect Oct 01 07:30:13 morning all Oct 01 08:14:26 kroon: patch in master-next Oct 01 08:15:12 RP, that was fast! thanks Oct 01 08:15:23 kroon: was already know :) Oct 01 08:15:30 aha :) Oct 01 12:01:01 hello, is there a way to silence the "No generic license file exists for .." warning? Oct 01 12:02:05 i need it because of some lousy license that does not exists in std dir: http://dejavu-fonts.org/wiki/License Oct 01 12:02:19 "Arev Fonts Copyright " Oct 01 12:03:16 i tried to add "LICENSE_PATH += "${WORKDIR}/"" in my recipe, workdir contains ArevCopyright.txt, but warning is still there Oct 01 12:03:28 http://tavmjong.free.fr/FONTS/ArevCopyright.txt Oct 01 12:10:22 gatis: you need to put the license file in a directory off the root of the layer, add to LICENSE_PATH in conf/layer.conf within the layer, and the file must be named to match the entry in LICENSE within the recipe Oct 01 12:11:37 bluelightning, ah ok, i will try that Oct 01 12:19:56 bluelightning, and then from my recipe "fonts-dejavu.bb" i need to download the license into "${LAYERDIR}/my-layer/additional-licenses"? Oct 01 12:20:48 no, you just put the file there yourself Oct 01 12:22:02 i was thinking about that initially, but will then yocto track the changes to license file? Oct 01 12:23:27 if by track you mean detecting changes to the upstream license that's a separate mechanism (LIC_FILES_CHKSUM) Oct 01 12:24:27 yeah i know about LIC_FILES_CHKSUM, but you say to download and put that file there my self, so its pretty static then Oct 01 12:28:46 gatis: yes; I guess it's assumed that the license is static Oct 01 12:29:04 I had the problem, that my proxy settings where not properly used and wget timed out. Although http_proxy / https_proxy was defined properly. I could fix it by defining the proxies again in ~/.wgetrc. But that shouldn't be the case. Oct 01 13:43:13 Hello\ Oct 01 13:44:36 i need fslvpuwrap to build vpuapi, but i can't get where do i download it not using fsl Oct 01 13:44:46 i mean not using yocto build system Oct 01 14:03:49 Is there a way to have bitbake print the git versions of the layers it's using? (Just like it does when building, except I don't want to build anything) Oct 01 14:33:13 anyone had an issue with a module recipe hanging on do_make_scripts ? Oct 01 14:38:13 work folder contains no log of do_make_script either Oct 01 15:18:31 So, BB_ENV_EXTRAWHITE="$BB_ENV_EXTRAWHITE MYGITVERSION" bitbake myimage Oct 01 15:18:38 gets me past the sanity check Oct 01 15:18:48 (the git version sanity check) Oct 01 15:19:08 but then later, MYGITVERSION is cleared AGAIN from the environment Oct 01 15:19:16 I can see that with strace -v Oct 01 15:19:44 which makes my build pick up a old and broken git version again Oct 01 15:20:08 BB_ENV_EXTRAWHITE will allow the variable into the metadata (run bitbake -e and you should see it), but it won't export it into the tasks bitbake is running automatically Oct 01 15:20:17 add 'export MYGITVERSION' to local.conf if you want that exported everywhere Oct 01 15:20:29 trying right now... Oct 01 15:20:35 (will change all signatures for all tasks and thereby cause a rebuild of everything, unless you also add it to the hashbase whitelist) Oct 01 15:20:53 rebuilding everything is the least of my worry right now Oct 01 15:21:05 "first make it work; then make it work faster" Oct 01 15:21:26 that's expected, i just wanted to make sure it was on the radar Oct 01 15:21:32 so you know it will happen Oct 01 15:21:38 your concern is very much appreciated :-) Oct 01 15:21:49 (and your expertise much more) Oct 01 15:23:16 wow Oct 01 15:23:41 needs more testing but looking good so far Oct 01 15:23:58 what is the best place to read about the local.conf black magic? Oct 01 15:24:15 first, local.conf isn't special, it's just one of many config files that get parsed Oct 01 15:24:28 good to know Oct 01 15:27:23 hmm, I'm not seeing BB_ENV_EXTRAWHITE in the yocto project documentation, that's quite an oversight Oct 01 15:27:47 do I still need the BB_ENV_EXTRAWHITE or is the export enough? Oct 01 15:28:15 export will do nothing if there's nothing to export Oct 01 15:28:33 bitbake filters the environment, and BB_ENV_EXTRAWHITE controls what it lets into the metadata at all Oct 01 15:28:34 I get you Oct 01 15:28:40 so you'll want both Oct 01 15:30:31 it's in the bitbake manual... Oct 01 15:30:59 I've still yet to work out how much of the bitbake manual's variable glossary should be copied into the ref manual Oct 01 15:31:04 ah. i was expecting either step by step in the main manual, or the variable in the ref manual Oct 01 15:31:08 * kergoth nods Oct 01 15:35:55 So, I have this code lurking in a gist which will flatten a license string obeying distro or user preferences. E.g. `MIT | GPLv3 | Proprietary` -> `MIT` if LICENSE_PRIORITY = "MIT * Proprietary" (* is "everything not explicitly listed", the var is in order of highest priority to lowest). I'm thinking we could use this for incompatible license, so if an incompatible license has alternatives, those would be chosen in preference to it, as well as for Oct 01 15:35:56 packaging, so the binary package license fields express the distro choices about what licenses they want to use, when they have such an option. Does this seem reasonable? It's been on my personal todo for a while, thinking about integrating it for the 1.8 timeframe Oct 01 15:35:57 From a documentation perspective I think this needs to be backed by some kind of flow diagram Oct 01 15:36:51 I'm referring to: BB_ENV_EXTRAWHITE will allow the variable into the metadata (run bitbake -e and you should see it), but it won't export it into the tasks bitbake is running automatically Oct 01 15:38:30 kergoth: I've certainly thought about that capability in the past, seems reasonable to me Oct 01 15:41:07 k, just wanted to make sure i wasn't entirely crazy, i'll see about opening a feature request in bugzilla Oct 01 15:41:10 thanks Oct 01 15:43:15 the only thing you'd want to try to cover at the same time is the license manifest - some way to counter "OMG I disabled GPLv3 but it's still in there !!!1" Oct 01 15:43:23 bluelightning: excuse me for interrupting you, but have you ever heard of do_make_script hanging? Oct 01 15:43:59 xerent: I haven't... but it could be something waiting erroneously for user input; have you checked the log for the task? Oct 01 15:44:02 good point Oct 01 15:44:12 I have, there's no log at all Oct 01 15:44:18 which is... weird Oct 01 15:44:44 there's a symbolic link to the run.do_make_script but it points nowhere Oct 01 15:44:47 dvhart, thank you for yesterday's baytrail related answers -- corei7-64 BSP image silently hung on this tablet shortly after efi bootloader (just as android-ia did), I've tried this vector too: https://embedded.communities.intel.com/message/10270 Oct 01 15:45:19 xerent: did you check what processes are running at the time? Oct 01 15:45:26 is it just bitbake, or is it waiting on something it ran? Oct 01 15:45:28 it would be nice to know if there's some baytrail-specific place, seems it's still got enough unsolved common bugs so far Oct 01 15:46:02 xerent: if you look at meta/classes/module-base.bbclass, do_make_scripts doesn't actually do very much itself, it's pretty much just running "make scripts" Oct 01 15:47:01 kergoth: can I check that with ps or is there some bitbake-specific tool Oct 01 15:47:19 no need for anything bitbake specific, just look at the process list Oct 01 15:47:41 I'll check Oct 01 15:48:02 Here's the recipe, btw: http://pastebin.com/61hfjZ8c and the code is here https://github.com/groeck/nct7802/ Oct 01 15:48:05 It's a rather simple module Oct 01 15:52:54 kergoth: here's the output http://pastebin.com/P2KeNjUB , it doesn't tell me much Oct 01 15:56:10 I think I'm going to check in all my changes Oct 01 15:56:22 delete the entire yocto folder Oct 01 15:56:25 and start over Oct 01 15:56:32 with a clean checkout :) Oct 01 15:59:29 xerent: really, wiping everything out is unlikely to help Oct 01 15:59:38 it helped once :) Oct 01 16:03:26 xerent, but that flood was quite some time ago :) Oct 01 16:04:31 as long as I commit all my animals to the ark, I can build a new yocto world after the rain of rm -rf Oct 01 16:04:59 jokes aside, if you have any better idea bluelightning ... :) Oct 01 16:06:13 bluelightning, SDR devroom accepted at FOSDEm Oct 01 16:06:22 and I see belen1 has one also Oct 01 16:06:22 Crofton: \o/ Oct 01 16:06:42 have I asked you to send in the OE stand request? Oct 01 16:06:59 xerent: I don't on this specific issue Oct 01 16:07:04 Crofton: yes and I haven't done it yet Oct 01 16:10:02 rm -rf yocto/ Oct 01 16:10:05 \o/ Oct 01 16:10:27 thanks Oct 01 16:18:54 hi guys Oct 01 16:19:19 I have a variable TEST="something" Oct 01 16:19:28 in my recipe Oct 01 16:20:08 and I want to do something like include first-${TEST}.inc Oct 01 16:20:14 in my recipe Oct 01 16:20:42 so that I cinlude either a first-something.inc file or a first-nothing.inc file I have Oct 01 16:20:49 so that I include either a first-something.inc file or a first-nothing.inc file I have Oct 01 16:21:12 but it does not include my first-something.inc file Oct 01 16:22:37 first of all, change that from include to require, otherwise it'll silently ignore a missing file. second, TEST must be defined above the include/require Oct 01 16:23:32 yes it is Oct 01 16:24:04 but TEST is defined on if a var is in DISTRO_FEATURES Oct 01 16:24:11 it is parsed allright Oct 01 16:24:14 but not included Oct 01 16:24:32 again, use require, that way it'll fail if it's trying to include something other than what you expect Oct 01 16:24:37 k Oct 01 16:24:38 let me try Oct 01 16:24:44 then use bitbake -e to examine the value of TEST to make sure it's what you think it is Oct 01 16:25:37 yes, I dod just that Oct 01 16:25:45 I used require Oct 01 16:25:53 and still the same Oct 01 16:25:58 TEST is what I expect Oct 01 16:26:03 but does not ibclude my file Oct 01 16:26:07 don't know what to tell you, we do includes like that all the time, and it works just fine Oct 01 16:26:26 here's how I set TEST Oct 01 16:26:31 what makes you think it doesn't include your file? have you added a variable there and then checked its value with bitbake -e? or even add a line that's invalid syntax to make sure bitbake errors out? Oct 01 16:26:32 TEST = "${@base_contains('DISTRO_FEATURES', 'wayland', 'wayland', 'x11', d)}" Oct 01 16:26:35 i don't are how you set test Oct 01 16:26:44 try setting it just to wayland or x11, does that work? Oct 01 16:26:51 yes, that works Oct 01 16:26:52 as with any debugging, you need to isolate Oct 01 16:27:11 so TEST is set as wayland Oct 01 16:27:24 but the x11 files is included instead Oct 01 16:27:29 but the x11 file is included instead Oct 01 16:47:22 kergoth, so it will "require" or "include" exactly the other tested param which is not in DISTRO_FEATURES Oct 01 16:47:24 so strange Oct 01 17:27:01 It looks like meta-qt5 sdk install is also broken from the multilib changes, qmake points to the wrong location. Oct 01 17:27:09 in master that is. Oct 01 17:33:47 Hmm, meta-minnow readme should mention mkefidisk requires parted / dosfstools, or at least mkefidisk should be fixed to properly error out if you're missing tools it requires, rather than happily chugging along trying to continue Oct 01 17:34:16 yes, it really should check up front Oct 01 17:34:19 bbl Oct 01 18:19:57 hi all Oct 01 18:20:56 i'm trying to compile modphp with simplexml enabled, so i enabled libxml2 and also included the recipe for libxml2, but configure for modphp keeps on complaining that it can't find xml2-config Oct 01 18:21:28 i do see a couple of *-config binaries in ${STAGING_BINDIR_CROSS}, but xml2-config is indeed missing Oct 01 18:21:51 it's only in the build path for libxml2 itself, not being copied to the sysroot Oct 01 18:30:31 if i add libxml2-dev to IMAGE_INSTALL, it would get added to the image, but it still doesn't provide xml2-config when configuring modphp Oct 01 18:34:12 how can i point it to the build path of libxml2, instead of using '--with-libxml-dir=${STAGING_BINDIR_CROSS}' Oct 01 18:35:56 we don't support binconfig scripts anymore afaik, it should be fixed to use pkg-config the way it's supposed to Oct 01 18:45:21 okay, but how can i get it to compile until it's fixed? Oct 01 18:55:41 hello Oct 01 18:57:17 one question, is possibile to change" make" version to make-3.82 only for build of a package? Oct 01 20:38:55 otavio, Do you have a moment to look at https://autobuilder.yoctoproject.org/main/builders/nightly-fsl-arm/builds/52/steps/BuildImages/logs/stdio with me? Oct 01 20:42:07 * rburton looks too Oct 01 20:42:22 oh new mesa shipping VG headers Oct 01 20:42:42 otavio: maybe a "vg" packageconfig for mesa? Oct 01 20:43:42 hm, maybe not, our mesa isn't shipping those headers Oct 01 21:54:22 Hi just trying to integrate this recipe to my image: gtest from https://github.com/openwebos/meta-webos/tree/master/recipes-upstreamable/gtest Oct 01 21:54:30 So I run bitbake -b gtest_1.7.0.bb -c cleanall -f and bitbake -b gtest_1.7.0.bb without problem Oct 01 21:54:38 bitbake -e gtest | grep ^PACKAGES= give: PACKAGES="gtest-dbg gtest-staticdev gtest-dev gtest-doc gtest-locale gtest" Oct 01 21:54:44 Adding gtest to my image then bitbake image result in Error: gtest not found in the base feeds? Oct 01 21:54:50 Adding gtest-dev since recipe have FILES_${PN}-dev but it result in Computing transaction...error: Can't install gtest-dev-1.7.0-r1@armv7a_vfp_neon: no package provides gtest = 1.7.0-r1 Oct 01 21:54:57 and the package seem there tmp/work/armv7a-vfp-neon-poky-linux-gnueabi/gtest/1.7.0-r1/deploy-rpms/armv7a_vfp_neon/gtest-dev-1.7.0-r1.armv7a_vfp_neon.rpm Oct 01 21:55:04 Any clue what I'm missing? Oct 01 21:59:34 kapare: your bitbake line didn't tell you what packages were *output* so presumably it didn't generate a gtest package, but libgtestsomething Oct 01 22:00:11 have a look in tmp/work/[machine]/gtest/1.7.0/deploy_* to see what packages were generated Oct 01 22:02:26 it did I mean I'm having gtest-dbg-1.7.0-r1.armv7a_vfp_neon.rpm gtest-dev-1.7.0-r1.armv7a_vfp_neon.rpm in the tmp/work/armv7a-vfp-neon-poky-linux-gnueabi/gtest/1.7.0-r1/deploy-rpms/armv7a_vfp_neon/ Oct 01 22:44:36 how to I override a library? I need to replace libnl-3.2.8 with libnl-3.2.25. Oct 01 22:55:28 rbbratta: assuming libnl doesn't do anything special, high versions take priority Oct 01 22:56:00 rbbratta: unless something (distro, recipe, local.conf) sets the preference explicitly (PREFERRED_VERSION or DEFAULT_PREFERENCE) Oct 01 22:57:30 it builds both libnl and then messes up the symlink Oct 01 22:59:36 okay, I'll try PREFERRED_VERSION Oct 01 23:39:15 turns out it was a typo, libnl-3.2.25.bb instead of libnl_3.2.25.bb, it couldn't find the new version I guess **** ENDING LOGGING AT Thu Oct 02 03:00:00 2014