**** BEGIN LOGGING AT Fri Nov 21 02:59:58 2014 Nov 21 08:39:56 hi, i got: subversion_1.8.9.bb, do_compile failed with exit code '1' -- how can i prevent bitbake to build subversion without rebuilding all? is there an option to just move on on error? Nov 21 09:02:51 hi all, Im getting a dependency issue when populating a qt5 based sdk : qtwebkit-mkspecs * qtwebkit-qmlplugins * qtwebkit (= 5.3.2-r0) * are missing Nov 21 09:03:05 how do I fixe that ? Nov 21 09:18:52 <_valle_> Anyone got qtmultimedia in dizzy working with gstreamer0.10? Nov 21 09:24:37 <[Sno]> RP: regarding my cpan-patch - how do I modify a git-commit-message when it's not the last commit (then commit --amend would do) Nov 21 09:33:55 rburton: Sorry for the late reply. Thanks for the update! Nov 21 09:35:19 ericbutters: bitbake won't rebuild anything unless it needs to, Nov 21 09:35:23 [Sno]: git rebase -i Nov 21 09:36:10 ericbutters: so i'm curious what you actually want to do? if you can fix the reason why subversion is not building, re-running bitbake won't build anything else - it will carry on from where it got to Nov 21 09:36:31 ericbutters: if you want it to ignore the subversion error and carry on building as much as it can now, instead of later, use bitbake -k Nov 21 09:36:48 ie people do bitbake -k world to build as much as possible in one go Nov 21 09:39:21 rburton: it is from subversion-native, which is needed on the host if i am right, but this issue is releated to ubuntu-12.04, i installed libserf-dev and bitbake now complains about error Please update your version of serf to at least 1.2.1 Nov 21 09:42:51 <[Sno]> rburton: will try and resend Nov 21 09:49:23 ericbutters: subversion-native should depend on serf-native Nov 21 09:50:29 which in dizzy, which i assume you're using, is at 1.3.6 Nov 21 10:10:50 morning all Nov 21 11:10:07 rburton: right, serf is installed inside the sysroot for i686, but why subversion-native does not find the headers? Nov 21 11:44:51 ericbutters: very good question :/ Nov 21 12:21:49 hello there Nov 21 12:22:24 I just realized there is a new thing called 'multilib' in Yocto Nov 21 12:22:43 how does that work? before that I would just pick C library, either uclibc or eglibc Nov 21 12:22:53 Xz: hi Nov 21 12:23:21 multilib has been around for a while, though using it to support a different libc flavour is a slightly more recent type of usage of it AFAIK Nov 21 12:24:08 in basic terms it creates a separate "namespace" for the build & packages to exist in Nov 21 12:25:20 if you want to see it working in that manner have a look in meta-intel-iot-devkit (in the devkit-daisy-1.6.1 branch) Nov 21 12:53:02 Hi, Nov 21 12:53:04 Hi, Nov 21 12:53:39 Hi, I have another problem regarding image building. on beaglebone am 335x we have 3 i2c, however in /dev i see only i2c-0, any ideas? Nov 21 12:54:16 JustasMika: maybe you have in your dts only 1 i2c entry? Nov 21 12:54:53 dts? (sorry, im pretty new in building) Nov 21 12:56:58 JustasMika: device tree structure in the linux kernel Nov 21 12:57:06 you may not be even using it, but that is not a Yocto question if so. Nov 21 13:06:10 how to change yocto default splash screen? Nov 21 13:07:00 ericbutters: use psplash Nov 21 13:07:40 ericbutters: you can convert your image to a .h file Nov 21 13:12:07 ericbutters: https://github.com/Guacamayo/meta-guacamayo/tree/master/meta-guacamayo/recipes-core/psplash Nov 21 13:15:26 Hi! I am trying to add https://github.com/meta-qt5 to a daisy based project, but the fetch task for qtsystem fails. The fix in dizzy and master is to use the '5.3' branch instead of the 'stable' branch, but would that work for daisy? Unfortunately, there is no 5.2.1 branch in upstream. Nov 21 13:32:02 lpapp: is there easy way to enable i2c-1 and i2c-2? Just I have deadline coming up for university, dont want to learn DTS as it seems complicated area Nov 21 13:32:23 JustasMika: pay someone for solve it for you, then. Nov 21 13:32:49 okay okay, got the idea Nov 21 13:32:50 I do not think that this kind of detail is a Yocto issue, but more like Linux kernel. Nov 21 13:32:56 for solving* Nov 21 13:33:20 and even I do not know DTS well, but you can ask on the linux kernel mailing list or #kernelnewbies (irc.oftc.net) Nov 21 13:48:10 JustasMika: FWIW it would probably be worth talking to the maintainer of the BSP layer you are using (or the specific mailing list / forum if there is one) Nov 21 13:50:02 Yocto bugzilla entries still confuse me. If I want to report a bug for a recipe which is in meta-openembedded where product should I pick? It's not oe-core, it's not meta-yocto, should I really select "other layers"? Nov 21 13:50:44 diego_r: meta-oe doesn't come under the yocto bugzilla... Nov 21 13:50:55 to file a bug, mail the oe-devel mailing list Nov 21 13:51:15 isn't it in the layer description? Nov 21 13:52:49 (apparently not, why?) Nov 21 13:54:16 lpapp: yeah, no indication in the READMEs Nov 21 13:54:29 rburton: ok, will do. Nov 21 13:54:31 IMHO, each nice layer should mention bugreports explicitly. Nov 21 13:54:50 hello, does someone know where i could find a freeimage recipe ? Nov 21 13:56:35 sce: https://gitorious.org/opencsbc/openembedded/source/bb74c9fb7ac59ad1883bf78803b4dabfb2589f97:recipes/ogre/freeimage_3.11.0.bb Nov 21 13:57:10 or http://cgit.openembedded.org/openembedded/tree/recipes/ogre/freeimage_3.11.0.bb?id=dd5c666d2182d6d839bde4ba4737bbea65680854 Nov 21 13:58:23 bit old though as I have 3.16 on my arch. Nov 21 13:59:05 lpapp, thx a lot Nov 21 14:02:40 sce: that's an oe-classic recipe so likely needs a bit of a clean to work in a oe-core/yocto environment Nov 21 14:03:03 yes i cant have the staging Nov 21 14:03:14 i'm facing a problem with rpm dependency Nov 21 14:18:07 rburton, when i want to install the rpm Nov 21 14:18:37 rburton, the freeimage rpm there is dependencies to libstdc++-dev-4.8.1 or libc6-dev-2.18-r0@cortexa9hf_vfp_neon Nov 21 14:19:05 sounds like the recipe is bust Nov 21 14:19:14 rburton, i dont want this dependency which are taking a lot of space for nothing, i just want the shared object file Nov 21 14:19:42 (or you're installing freeimage-dev without meaning to) Nov 21 14:20:33 freeimage is depending on freeimage-dev Nov 21 14:20:38 i dont know how to remove this Nov 21 14:21:04 bluelightning: for the multilib - I've never heard of having more than 1 C library on the target, I guess multilib does not change that? :) Nov 21 14:24:31 sce: you'll have to look at the contents of the packages to work that out Nov 21 14:24:49 without the recipe your building there's only so much i can help with Nov 21 14:25:13 can i change the package content? it seems that rpm is giving the dependencies ... Nov 21 14:25:25 rpm is creating it for a reason Nov 21 14:26:05 https://gist.github.com/dabrain34/21166220df9f7475a7fa Nov 21 14:26:14 and thats what i dont understand ;) Nov 21 14:26:32 on how i can change rpm behaviour ;) Nov 21 14:26:33 Xz: no, it does - it has to in order to work Nov 21 14:28:13 Xz: if you run an iotdk image on a galileo you'll see there is one C library in /lib and another in /lib32. That's how we're able to run uclibc sketches Nov 21 14:29:38 sce: bitbake warns me of this problem: Nov 21 14:29:43 ERROR: QA Issue: freeimage rdepends on freeimage-dev [dev-deps] Nov 21 14:30:03 yes Nov 21 14:30:08 and looking at the file listing shows why Nov 21 14:30:11 and i dont know how to remove this dependency Nov 21 14:30:30 you *need* that dependency with the current packaging Nov 21 14:30:38 as otherwise you don't get a library Nov 21 14:31:08 have a look at how the packages were split in the work directory Nov 21 14:32:20 basically, freeimage don't know who libraries are versioned on linux Nov 21 14:33:13 so you'll have to work around that by changing FILES Nov 21 14:33:44 yes Nov 21 14:33:58 that's what i'm trying libfreeimage-3.11.0.so is in -dev Nov 21 14:47:04 arfoll: so eglibc sits under /lib and uclibc under /lib32 on iotdk image? Nov 21 14:49:51 Xz: pretty much Nov 21 14:50:48 Xz: the trick is that there are two loaders in /lib, ld-linux.so.2 (eglibc) and ld-uclinux.so (uclibc). Depending on which is called by a binary the loader will go to the correct folder Nov 21 14:50:58 is it possible to have no dev or dbg package ? Nov 21 14:51:05 in recipe Nov 21 14:51:58 rburton, ERROR: QA Issue: File '/usr/lib/libfreeimage-3.11.0.so' from freeimage was already stripped, this will prevent future debugging! Nov 21 14:52:02 arfoll: ok, that's interesting. Does iotdk support default Galileo Arduino IDE? Nov 21 14:52:18 arfoll: I had a run through image/readmes etc. I tried I think XDK and Eclipse Nov 21 14:52:37 arfoll: so far Nov 21 14:52:43 sce: http://www.yoctoproject.org/docs/current/ref-manual/ref-manual.html#qa-issue-already-stripped Nov 21 14:53:28 bluelightning, thx Nov 21 14:53:47 Xz: yes, you can run arduino sketches compiled on the IDE Nov 21 15:23:36 rburton, https://gist.github.com/dabrain34/84393a6b1e86d65bd72b Nov 21 15:23:40 for your information :) Nov 21 15:25:43 lol at @ meta-embeddedgeeks Nov 21 15:42:33 sce: best practise there would be to patch out the strip command Nov 21 15:53:37 arfoll: if I just compile my simple hello-world.c app against uclibc toolchain - will it automatically pick up uclibc on target? Nov 21 15:53:50 arfoll: or I need to do some extra steps? Nov 21 15:54:24 arfoll: let's say I have eglibc and uclibc toolchains on host machine, I compile my hello-world.c twice - a) with eglibc toolchain, b) with uclibc toolchain Nov 21 15:54:34 arfoll: will both binaries work on target no problem? Nov 21 15:58:31 this is very strange, I am building stunnel in our layer and I only get -dev and -dbg packages. What could be the reason for that? Nov 21 16:01:46 Xz: should work Nov 21 16:02:06 here is the recipe: http://pastie.org/9734715 Nov 21 16:05:12 lpapp: perhaps nothing happening during do_install? the -dbg and -dev packages are marked ALLOW_EMPTY so you get those even if they have nothing in them Nov 21 16:06:14 lpapp: we do have a stunnel recipe in meta-networking btw, it inherits autotools Nov 21 16:07:32 bluelightning: I submitted that back then, yes. Nov 21 16:07:36 once it was working in our layer. Nov 21 16:07:48 I am not integrating a whole layer for one recipe, I guess. Nov 21 16:08:05 (not that the meta-networking recipe works, but that is a different issue) Nov 21 16:08:19 you did indeed, I didn't recall Nov 21 16:08:36 my only change that was accepted (after some nitpicking of course). Nov 21 16:08:42 ok but surely you'd at least start with that one... Nov 21 16:08:52 well, I do not know. Nov 21 16:11:02 bluelightning: same issue after using the same recipe. Nov 21 16:12:27 ls stunnel/4.56-r0/packages-split/ Nov 21 16:12:27 stunnel stunnel-dbg stunnel-dev stunnel-doc stunnel-locale stunnel-staticdev Nov 21 16:13:07 but those are empty directories only. Nov 21 16:15:18 bluelightning: http://paste.kde.org/p2cmm06b3 Nov 21 16:15:41 lpapp: I see, well there's your problem I guess Nov 21 16:16:03 bluelightning: where exactly? Nov 21 16:16:21 I'd assume it should compile something? Nov 21 16:16:43 if it doesn't compile anything there's nothing to install, and hence nothing to package... Nov 21 16:17:03 sure, but what I am trying to ask is why it does not compile? Nov 21 16:18:13 I think you'll just have to work through it, there aren't any clues in that log Nov 21 16:18:17 this is the do_configure: http://pastie.org/9734752 Nov 21 16:18:40 that looks suspiciously empty Nov 21 16:18:42 how to work through it? Nov 21 16:18:55 the log is kinda pointless to me. Nov 21 16:19:04 well, how would you do it if you were building it outside of the build system? Nov 21 16:19:20 you would run through the steps manually, I would assume Nov 21 16:19:24 so you could open up a devshell and try that... Nov 21 16:19:25 yes, I did that, that worked. Nov 21 16:19:45 it must be some Yocto thing. Nov 21 16:19:54 ok then look at the commands it is running in run.do_* and where they are being run Nov 21 16:20:03 The first thing I always suggest verifying when encountering a problem like that is to make sure S is set correctly, since a few of the default tasks will just do nothing if the expected makefile/configure doesn't exist, rather than erroring Nov 21 16:20:17 drop into a devshell and run the commands there Nov 21 16:20:33 yes that's a good idea, wrong S might cause this Nov 21 16:22:06 S is alright Nov 21 16:22:34 S="/home/lpapp/Projects/foo/Yocto/poky-dylan-9.0.1/build/tmp/work/armv5te-foo-linux-gnueabi/stunnel/4.56-r0/stunnel-4.56" Nov 21 16:22:49 based on bitbake -e stunnel | grep ^S= Nov 21 16:24:06 this is what log.task_order writes: http://pastie.org/9734763 Nov 21 16:24:09 and that path exists, and the appropriate build files are there? Nov 21 16:24:18 yes Nov 21 16:24:52 personally, i'd add set -x to do_configure and do_compile and do_install, or build with -D, and examine the more verbose logs so you can see exactly what shell commands are being run Nov 21 16:25:22 that's why I suggested looking at run.do_*, you'd see the commands there Nov 21 16:26:07 FWIW, I just tried the stunnel recipe in master, it builds here (for qemux86-64 that is) Nov 21 16:26:11 bitbake -e stunnel | grep ^S= | tr -d 'S="' | xargs ls -> shows stuff fine. Nov 21 16:27:18 yeah, you could look at the run scripts and run those commands yourself in a devshell too. you can also look one task at a time. e.g. run do_compile, check S/B to see if it was compiled or not, and if so, then check do_install and check ${D} to see if it installed anything, .. Nov 21 16:27:24 work through it one by one to see where it falls over Nov 21 16:28:48 * kergoth heads to yet another doc appt for his pregnant wife (only 2 weeks to go) Nov 21 16:29:39 kergoth: congrats! Nov 21 16:29:48 the stunnel binary is definitely not built Nov 21 16:29:55 kergoth: (+ in advance as well) Nov 21 16:30:21 not even with bitbake -f -c compile stunnel Nov 21 16:30:42 lpapp: no, that couldn't help if the underlying build system refuses to do anything as it apparently is Nov 21 16:31:09 I guess, but it is difficult for me to read those run files Nov 21 16:31:18 so I leave that as last resort. Nov 21 16:32:39 is there a force clean and rebuild without rebuilding the whole image? Nov 21 16:32:49 this is the base_do_compile fwiw: http://pastie.org/9734786 Nov 21 16:32:59 lpapp: that would be bitbake -c cleansstate stunnel then bitbake stunnel Nov 21 16:33:09 bluelightning: I tried -c cleanall already. Nov 21 16:33:19 which, afaik, is more bruteforcy Nov 21 16:33:34 lpapp: right, that additionally wipes out the downloaded source Nov 21 16:33:42 yeah. Nov 21 16:33:48 I have no any clue, I am afraid. Nov 21 16:33:52 lpapp: ok, have you tried disabling icecc for that particular recipe? e.g. by adding stunnel to ICECC_USER_PACKAGE_BL Nov 21 16:33:54 even the run file is greek or chinese to me. Nov 21 16:34:21 in local.conf? Nov 21 16:34:34 I would guess so yes Nov 21 16:34:39 I don't use that class myself Nov 21 16:34:45 I've no idea if this will help, but it's something to try Nov 21 16:36:30 bluelightning: that reenforces doing a lot of stuff Nov 21 16:36:32 500+ tasks. Nov 21 16:37:03 I would prefer something quicker than waiting a lot now. Nov 21 16:37:22 well, that's just poor behaviour of that class then... you could try setting it in the recipe Nov 21 16:38:22 this is the syntax that I used, just in case: ICECC_USER_PACKAGE_BL = "stunnel" Nov 21 16:39:04 and putting it into the recipe did not help fwiw. Nov 21 16:39:28 ok, well I really don't know anything about that class to be honest Nov 21 16:39:35 beyond just looking at it Nov 21 16:40:05 I just suggested disabling it for this recipe because it's one difference from here where it is building correctly Nov 21 16:40:44 does it make sense to try to fill the do_configure and do_compile myself with the dummy thing? Nov 21 16:41:46 bitbake -e stunnel | grep ^S= | tr -d 'S="' | xargs cd Nov 21 16:41:46 xargs: cd: No such file or directory Nov 21 16:41:54 but xargs ls shows the content, how weird. Nov 21 16:42:49 actually it looks as if you did set_icecc_env() { : } in the recipe it would stub out what icecc.bbclass is doing Nov 21 16:43:33 anyway, that is not the main point (xargs cd). Nov 21 16:44:24 recipes-support/stunnel/stunnel_4.56.bb:13: unparsed line: 'set_icecc_env() { : }' Nov 21 16:45:04 without the ':': stunnel/4.56-r0/temp/run.do_configure.12906: set_icecc_env: not found Nov 21 16:45:05 sorry, I wasn't clear - you'll need newlines between {, :, and } Nov 21 16:45:34 right. Nov 21 16:45:56 that did not make the stunnel executable generated in src/, sadly. Nov 21 16:46:11 hmm, wait a sec Nov 21 16:46:16 is this an S != B issue? Nov 21 16:46:22 no clue Nov 21 16:47:01 but they are the same Nov 21 16:47:07 well, no it wouldn't be Nov 21 16:47:12 based on bitbake -e stunnel | grep ^B= and bitbake -e stunnel | grep ^S= Nov 21 16:47:23 I am puzzled what is going on ... Nov 21 16:47:24 at least the version I built doesn't inherit autotools-brokensep Nov 21 16:47:59 I guess you are using daisy or earlier so that's probably not an issue Nov 21 16:48:37 hmm, I am not inheriting inherit autotools Nov 21 16:48:50 I am using dylan. Nov 21 16:48:58 ok, I did mention inheriting autotools earlier... Nov 21 16:49:08 the latest recipe does not do that in master. Nov 21 16:49:23 ah, the one I'm looking at in meta-networking master here does... Nov 21 16:49:56 I am confused, but why cannot Yocto detect this issue somehow? Nov 21 16:50:05 we could have spared some time. Nov 21 16:52:19 lpapp: well, to be fair you could also have started with the stunnel recipe you yourself submitted ;) Nov 21 16:52:42 hmm? Nov 21 16:53:04 it is possible as I was removing the earlier imported meta-networking, so basically copied the recipe at the time of submitting. Nov 21 16:53:07 here's the recipe as it was first added: http://cgit.openembedded.org/meta-openembedded/commit/meta-networking/recipes-support/stunnel/stunnel_4.56.bb?id=9f3cea60ed06730e6d26e5dae32ddf0312aee423 Nov 21 16:53:09 and I see some improvements upstream Nov 21 16:53:30 I have no idea how I got my recipe that was not working, then. Nov 21 16:53:43 I was missing the last three lines. Nov 21 16:54:05 thanks. Nov 21 16:54:17 no problem Nov 21 22:07:35 moo Nov 21 22:33:27 Hi Nov 21 22:33:44 just a brief question, is meta-toolchain-sdk still a valid target in Yocto >= 1.6 please ? Nov 21 22:34:01 or what is the difference between meta-toolchain and meta-toolchain-sdk please ? Nov 21 23:10:35 JaMa: ping? Nov 22 00:30:01 is openembedded.org down ? Nov 22 00:50:18 * armpit ok its back **** ENDING LOGGING AT Sat Nov 22 03:00:00 2014