**** BEGIN LOGGING AT Thu Nov 14 02:59:59 2013 Nov 14 06:51:56 hello Nov 14 06:52:53 im having a problem trying to build gstreamer1.0-plugins-base_1.2.0 with gobject introspection Nov 14 06:53:04 is there anyone here that could possibly help me? Nov 14 08:25:08 Morning Nov 14 08:25:12 Morning Nov 14 08:27:35 JaMa, I am not sure which branch of meta-qt5 I should be using... My environemnt users yocto 1.3 (Danny). I am at the moment cloning master and so far it works OK for us... Nov 14 08:27:52 but not sure if I might run into problems in the future because of this Nov 14 08:28:18 any recommendatiosn? Nov 14 08:34:22 is Khem Raj here? Nov 14 08:48:55 melonipoika: there isn't official branch for danny (only for dylan, dora), if you have adapted meta-qt5/master to work with danny (e.g. update cmake .bbappend) then you can use it (any "newer" branch), it's just not officially supported and some tweaks are expected Nov 14 08:49:43 ok, thanks! Nov 14 08:50:23 can I avoid in a .bbappend to apply a patch defined in the .bb? Nov 14 08:51:38 actually i didn't have to modify the cmake.bbappend, i had to hack slightly qtbase to get it working on my board Nov 14 08:56:35 n01: set SRC_URI in your bbappend Nov 14 08:56:45 leave out what you don't want Nov 14 08:58:03 nerdboy: right, tnx Nov 14 09:01:33 melonipoika: do you have BB_DANGLINGAPPENDS_WARNONLY set? Nov 14 09:01:59 i am not sure, at least I have't set it Nov 14 09:02:03 let me check Nov 14 09:02:25 melonipoika: without this meta-qt5/master and oe-core/danny won't parse, because .bb for meta-qt5/recipes-devtools/cmake/cmake_2.8.12.bbappend isn't in oe-core Nov 14 09:03:09 well you didn't say which revision you're using from meta-qt5/master, but unless it's something very old then you will have this ^ Nov 14 09:03:24 I am using the top of the repo Nov 14 09:04:21 then i must have BB_DANGLINGAPPENDS_WARNONLY set :-) Nov 14 09:05:13 how can i check where that variable is set? Nov 14 09:05:42 bitbake -e should show it Nov 14 09:06:42 yep: BB_DANGLINGAPPENDS_WARNONLY="true" Nov 14 09:09:38 I ahve been checking outmeta-qt5/master 84cd4be2491895b0c3a0af64e4c83d58910d1f6c, I am trying now 7327781083aea593b4911b6da7105e4984e28c68 Nov 14 09:10:36 damn, it's been so long i almost forgot how to use cvs... Nov 14 09:14:58 good for you, leaving bad habbits behind :) Nov 14 09:20:05 yeah, i also need to do another svn->git migration and push it to github Nov 14 09:20:15 the fun never stops Nov 14 09:21:11 but it's now 0120 here and i still have to work tomorrow... :/ Nov 14 09:22:42 How can I handle a source repos(svn) that requires authentication? Nov 14 09:23:35 hi JaMa, any change you can apply my qtwebkit-examples minor fix in dora branch in meta-qt5? you've merged it into master alreayd. Nov 14 09:24:06 tc_: you have to do the authentication upfront and store svn credentials in $HOME/.svn Nov 14 09:24:38 ndec: will do right now Nov 14 09:24:49 ndec: Thanks! I will try that! Nov 14 09:25:10 JaMa: thanks ;-) Nov 14 09:25:40 tc_: i use that everyday..so it should work. you basically need to do svn info for each you need. Nov 14 09:35:12 morning all Nov 14 09:37:05 hi bluelightning. morning. btw, we were able to sort out the gitsm situation last night. see https://bugzilla.yoctoproject.org/show_bug.cgi?id=5525 Nov 14 09:37:06 Bug 5525: normal, Undecided, ---, richard.purdie, NEW , gitsm.py fetcher requires git version > version 1.8.5 Nov 14 09:37:17 iirc, you left in the middle of the conversation ;-) Nov 14 09:37:29 ndec: ah right, yes, I had to leave the office Nov 14 09:37:30 ndec: thanks Nov 14 09:38:00 bluelightning: well, i think it's okay to leave the office. well, sometimes ;_) Nov 14 09:43:20 ndec: Works like a charm, thanks for the help! :) Nov 14 10:28:46 hi all, I want to use xz, but I want a GPLv3 free filesystem, can anyone tell me how that should be done? Nov 14 10:48:45 TuTizz: iirc, native tools are exempt from the no-gpl3-button Nov 14 10:55:36 any idea why this patch https://github.com/openembedded/oe-core/blob/master/meta/recipes-core/uclibc/uclibc-git/0001-add-posix_fadvise-for-arm.patch breaks my compilation? Nov 14 10:56:00 error: unknown type name 'off64_t' Nov 14 10:56:13 sorry rburton, I did not understand. Qt depends on tiff and tiff depends on xz (http://tukaani.org/xz/). Should I swap xz by iirc? Nov 14 10:57:13 oh wasn't aware that tiff depended on xz. either turn off tiff support in qt, or turn off xz support in tiff, i guess Nov 14 10:58:01 ok I will try that. Nov 14 12:37:54 hello everyone Nov 14 12:41:23 denix, any plans to make a mcsdk release based on yocto dylan? Nov 14 12:41:56 the latest mcsdk release (3.02.14 iirc) is based on danny Nov 14 13:24:41 When I bitbake foo, how can I know where is the recipe's location? Nov 14 13:28:08 one of easiest options is to use bitbake foo | tee and log will show you the path Nov 14 13:30:15 ok perfect ty Nov 14 13:54:16 Hi guys... i m creating a very simple recipe (poky 1.5) and i have a (stupid) problem regarding versioning... my recipe's name is "foo_2.11_rc1.bb", do_fetch gets a file (foo-2.11-rc1.tar.gz")... the problem is the destination direction ... do_unpack() unpacks in foo-2.11-rc1 BUT do_populate_lic() is done is foo-2.11 (which is empty) .... any suggestions ??? Nov 14 13:56:46 set S variable Nov 14 13:59:18 How is S used ? Nov 14 14:00:31 it points to source, grep existing recipes for examples Nov 14 14:01:02 ok... i ll check that ...thanks Nov 14 14:17:16 Hey, I'm confused about a minor issue I'm having. Nov 14 14:17:32 I removed MySQL from my image, but noted that libmysqlclient.so.16 was still found on the rootfs after flashing the new image. Nov 14 14:18:04 However, when I do find tmp/ -name libmysqlclient.so.16 in the build folder, the only place I see it is in subdirs to mysql5 and mysql5-native Nov 14 14:18:09 something probably links against it still Nov 14 14:18:28 Alright? Wouldn't I be able to see the lib in that package's packages-split/ dir, then, or what? Nov 14 14:19:17 something like buildstats or buildhistory should help Nov 14 14:20:01 Stygia: you'll only see that library in the mysql5 build directory because thats the package that builds it Nov 14 14:20:33 you can ask the package manager what depends on libmysqlclient Nov 14 14:20:50 Ah alright, well I'm not actually concerned then. :) Nov 14 14:21:22 I just thought, since I only saw it in packages-split, it was only used by that package. Nov 14 14:21:35 If it still gets in there if another package uses them, I'm completely okay with it, I don't mind, just wanted to be sure it wasn't a crap file. Nov 14 15:01:24 hi guys, im having troubles building gstreamer 1.2.0 from meta-gstreamer1.0, gstreamer1.0-plugins-bad is having problems with egl Nov 14 15:01:45 it requires x11 libs which i dont have (or want) any suggestions on ho to fix it? Nov 14 15:02:14 adding disable-egl in the reciepe did not help Nov 14 15:07:25 Dylan, Do you mean, it depends on libx11, which you don't have or want on your host system? Nov 14 15:08:04 yes, the gstreamer-egl plugin required X11 headers Nov 14 15:08:07 Dylan, In that case, you shouldn't be installing libs on your host system to fix this anyway, you should use DEPENDS if it's a build-time dependency and RDEPENDS_${PN} if it's a runtime dependency. Nov 14 15:08:27 Then, you must find the recipe that provides libx11, and add that as a DEPENDS for the recipe. Nov 14 15:09:13 that would put libx11 on my target system Nov 14 15:10:19 Dylan: without looking the recipe probably use PACKAGECONFIG to enable egl, so your disable-egl is being overriden Nov 14 15:10:34 If you put it in DEPENDS, it would not. If you put it in RDEPENDS_${PN}, then yes, it would. Nov 14 15:10:35 that it does sir Nov 14 15:10:57 ah ok i see stygia Nov 14 15:11:27 Dylan: just remove egl from PACKAGECONFIG Nov 14 15:12:12 Dylan, Although, if you link against it dynamically and have it only in DEPENDS, then the dynamically linked binary on the box won't be able to find it. So, if you _were_ wanting to include it (I know you don't), you'd need it in both DEPENDS and RDEPENDS_${PN} Nov 14 15:12:38 PACKAGECONFIG[eglgles] = "--enable-eglgles,--disable-eglgles,virtual/egl virtual/libgles2" Nov 14 15:12:54 i added PACKAGECONFIG[eglgles] = "--disable-eglgles,--disable-eglgles,virtual/egl virtual/libgles2" to a .bbappend Nov 14 15:12:59 no Nov 14 15:13:00 trying it now Nov 14 15:13:10 set PACKAGECONFIG to the value in the .bb, but without egl Nov 14 15:13:14 well, "eglgles" Nov 14 15:13:50 ah ok i see Nov 14 15:13:54 to override it Nov 14 15:14:04 Stygia: the dynamically linked library will link to the library, which will then get turned into a rdepends at package build time Nov 14 15:14:31 Stygia: (otherwise you'll have to add every library dependency to every package, which you don't need to do) Nov 14 15:14:36 morning walters Nov 14 15:14:37 rburton, Oh? It knows to do that dynamically? Fancy. Nov 14 15:14:41 rburton, Thanks, then! Nov 14 15:14:49 Dylan, Listen to rburton, he's right if he contradicts me. :) Nov 14 15:14:54 Stygia: every distro build system does that :) Nov 14 15:15:03 (doing it manually is crazy) Nov 14 15:15:03 hi rburton Nov 14 15:15:14 rburton, Heh, fair enough. Nov 14 15:15:21 Yocto is the first and only distro build system I've worked with. Nov 14 15:16:31 sorry stupid question but what does ??= do? Nov 14 15:17:03 Dylan: its a very lazy evaluation, set to this only if its not set anywhere else Nov 14 15:19:26 that means the packageconfig assignment in my ..bbappend cant use this? Nov 14 15:19:38 Dylan: copy the value, remove eglgles, use = Nov 14 15:19:44 10-4 Nov 14 15:24:04 unfortuantly that did not fix it Nov 14 15:24:41 while it stopped eeglgles from being build it did not stop gstreamer-egl from being built Nov 14 15:25:09 you probably need more lines like eglgles Nov 14 17:07:39 Hey, anyone know if I can specify a dependency to depend on a certain version of a package? Nov 14 17:08:07 I am trying to compile the 'squid' proxy, but it dies with a certain error that some searching has led to me believe is caused by an incompatible version of the Berkeley DB. Nov 14 17:08:30 So, I'd love to say something like DEPENDS = "virtual/db-4.6-or-lesser..." If that's possible? Nov 14 17:09:42 Stygia: no, can't do that Nov 14 17:10:03 Damn. Nov 14 17:10:17 Ah, thanks. I suppose I'll have to find some way to make a newer squid compile, then... Cheers anyway. Nov 14 17:11:33 oe-core only has a db 5.3 from what i can tell Nov 14 17:11:44 so you could just rename your db recipe to db-4.6 or whatever Nov 14 17:11:48 and then depend on db-4.6 Nov 14 17:12:03 (assuming db is parallel installable) Nov 14 17:12:21 Hmm, I could give that a try, yea. And yes, I did only find a 5.x Nov 14 17:12:48 Thanks. I'll try and see if I can find a more recent recipe somewhere, but it seems like every working recipe patches the source code rather heavily - so it isn't easy to cross-complie, it would seem. Nov 14 17:12:56 if you don't have a db <5.3 yet then it's all academic :) Nov 14 17:33:36 rburton, Yup, finally got it to compile with it, I'll see if my sources were right about db > 4.6 causing the bug I saw. :) Nov 14 17:54:44 Hello folks Nov 14 17:55:06 I got several reports of people complaining about xserver-common not being fetchable Nov 14 17:55:23 Can it be add to the Yocto mirror? Nov 14 18:06:36 Good morning everyone, does anyone have a tutorial on submitting a recipe patch? Nov 14 18:06:49 Also, on adding package to OE mirror? Nov 14 18:09:14 otavio: does your new lttng patch replace the two from earlier? Nov 14 18:09:57 otavio: I think I just received your email regarding libmpd Nov 14 18:10:00 b1gtuna: there is a patch guidelines page here: http://www.openembedded.org/wiki/Commit_Patch_Message_Guidelines Nov 14 18:10:10 sgw_: cool thank Nov 14 18:10:11 is that what your looking for Nov 14 18:10:12 thanks Nov 14 18:35:18 b1gtuna: i assume you searched for any existing libmpd bugs? Nov 14 18:37:18 mr_science: I asked in the mailinglist - this has been confirmed by Paul Eggleton, ANdreas Muller and Otavi Salvador. Nov 14 18:37:37 mr_science: Andreas is too busy at the moment to take care of this Nov 14 18:40:43 just needs a fetchable SRC_URI abut i'm still not sure where that version number came from Nov 14 18:49:59 mr_science: http://download.sarine.nl/Programs/gmpc/11.8/libmpd-11.8.17.tar.gz is one suggestion from Paul Nov 14 19:08:22 b1gtuna: how does that relate to upstream? why is it different? Nov 14 19:08:48 not asking to find out, just pointing out it doesn't seem consistent Nov 14 19:08:53 *you Nov 14 19:19:33 rp: hi, still nothing on adding debian-7.2 to sanity tested distros? Nov 14 19:24:07 zibri: sorry, keep meaning to merge that Nov 14 19:25:04 heh, don't worry. i guess you have a lot going on :) Nov 14 19:25:22 zibri: done Nov 14 19:25:46 thanks! Nov 14 19:27:58 w Nov 14 19:28:01 oops Nov 14 19:29:00 mr_science: I added MIRRORS += " http://download.sarine.nl/Programs/gmpc/11.8/ \n" to libmpd recipe. It's not getting recognized. Probably something stupid that i am doing. Nov 14 19:35:37 unless there's something weird i'm missing, i see no reasno not to point at upstream stable Nov 14 19:37:51 mr_science: ahh sorry, where is the upstream stable? Nov 14 19:38:05 is this a known thing when trying to eclipse-adt debug against qemuarm? "Could not determine GDB version after sending: gdb --version" Nov 14 19:38:10 should be musicpd.org Nov 14 19:38:39 i just checked, and the gentoo ebuild is the same wonky version and URI Nov 14 19:38:51 still no idea why... Nov 14 19:39:42 mr_science: i see what you are saying Nov 14 19:42:57 mr_science: sorry to keep on bothering - I just want to learn how to add a mirror. Do you see why the mirror wouldn't be working in this recipe? - http://paste.ubuntu.com/6417528/ Nov 14 19:44:30 mr_science: libmpd isn't official, it seems, and the wiki on musicpd says it's deprecated. the official website for gmpc/libmpd points to download.sarine.nl Nov 14 19:45:06 http://mpd.wikia.com/wiki/ClientLib:libmpd and http://gmpc.wikia.com/wiki/LIBMPD_0.17.0 Nov 14 19:45:26 then it seems odd that they point to 0.13 (or whatever) as stable release Nov 14 19:45:46 where? Nov 14 19:45:51 musicpd.org Nov 14 19:46:20 i don't see it Nov 14 19:46:44 i can't get to it right now (it's blocked) but that's what i saw two days ago Nov 14 19:47:11 it's in the downloads Nov 14 19:47:17 right. perhaps they did some clean up Nov 14 19:48:10 hum, interesting that musicpd.org hosts a 0.20.0, when gmpc says the latest is 0.19.0. :) Nov 14 19:48:31 like i said, doesn't make sense to me either... Nov 14 19:48:55 AFAIK musicpd.org is upstream Nov 14 19:49:25 what "officially" points to that oddball version/URI ? Nov 14 19:50:44 http://mpd.wikia.com/wiki/ClientLib:libmpd links to http://gmpc.wikia.com/wiki/LIBMPD_0.17.0 which links to sarine.nl Nov 14 19:51:05 http://repo.or.cz/w/libmpd.git lists qball sarine.nl as owner Nov 14 19:51:22 so probably not that oddbal, just the author's domain name Nov 14 19:51:40 and musicpd.org mirrors it. from what i can tell at least Nov 14 19:51:57 ah, that makes a little more sens then Nov 14 19:52:05 *sense even Nov 14 19:52:22 except that they should really be more clear about what's what Nov 14 19:53:07 maybe. but the mention of it on musicpd.org is in the wiki from what i can tell Nov 14 19:53:10 the only mention* Nov 14 19:53:42 well, if you go straight to musicpd.org all you know is -> downloads Nov 14 19:54:11 i didn't see any reference to the other site, certainly not an obvious one Nov 14 19:54:48 except for the directory listing, i can't find any references to it outside of the wiki Nov 14 19:57:20 my psychic powers aren't working today, so i don't have answer... Nov 14 20:22:33 is this a known thing when trying to eclipse-adt debug against qemuarm? "Could not determine GDB version after sending: gdb --version" Nov 14 21:03:18 sgw_: yes, it does. The new one has all in Nov 14 21:03:25 b1gtuna: good Nov 14 21:05:37 otavio: thanks Nov 14 21:08:25 where would you add usbutils? in the machine conf file? Nov 14 21:09:53 it seems to be hardware dependent, so AFAIK it would go into the BSP layer Nov 14 21:11:21 sgw_: yw Nov 14 21:14:39 Hi Guys, in the repo, there is postgresql server, but where is the pgsql client ? Nov 14 22:04:17 weebette: without looking, my guess would be a package created by the postgresql server recipe Nov 14 22:04:48 you could look at the server recipe and see what PACKAGES are defined Nov 15 01:20:45 vmeson: No comments on the email but I found out that the compilers are being set in `meta/conf/*conf` files. Now I need to figure out how to overwrite those variables, and use the new variables. Nov 15 01:23:41 If someone has already done something similar, that would have been very helpful. Nov 15 01:37:06 mtahmed: I don't know of anyone who has done this exactly. Certainly people override the open source gcc with a vendor supported one Nov 15 01:38:14 what are you trying to do exactly? Nov 15 01:40:54 kergoth: Trying to use llvm+clang for oe/poky. Nov 15 01:41:41 as a crosscompiler toolchain, or native? Nov 15 01:41:56 The first "goal" could be to try to build core-image-minimal for poky-tiny using llvm+clang. So far, I have been able to compile the clang cross-compiler. Nov 15 01:42:16 As a cross-compiler toolchain for building the images. Nov 15 01:42:18 afaik stock clang/llvm doesn't even support easy selection of target and corresponding sysroot yet, hence the existance of projects like ellcc Nov 15 01:42:24 could be mistaken, though Nov 15 01:42:58 I would definitely love to see this as a possibility though. There's definite value in it, both from a tooling perspective, and for folks who are gpl-averse Nov 15 01:43:52 * vmeson recalls seeing ellcc... a year or two ago but I thought that clang no longer needed that help. Nov 15 01:43:55 it's also worth noting that neither clang nor llvm replace binutils, there's no pure bsd binutils replacement afaik. there's elftoolchain, which is promising, but it's still missing the assembler and linker. not sure if you care about that or not Nov 15 01:44:19 clang has an argument to select the target, but i don't know whether or not its smart enough to associate that with the appropriate sysroot for multilib Nov 15 01:44:24 * kergoth shrugs Nov 15 01:45:06 kergoth: mtahmed (and I) are just trying to take the first few steps to understand what needs to be done so your comments are certainly helpful. Nov 15 01:45:26 mtahmed: http://ellcc.org/ (in case you can't use google. :) ) Nov 15 01:45:52 http://www.embtoolkit.org/ apparently has some llvm stuff too, never tried it though Nov 15 01:45:52 i've started trying to look at this in the past too, mostly was at the information gathering stage though Nov 15 01:45:52 also wanted to try using musl instead of eglibc, as an experiment/exercise :) Nov 15 01:45:59 Haha. I already saw that :D. "associate that with the appropriate sysroot for multilib"? I am kind of noob with this stuff right now. Nov 15 01:46:14 http://wiki.musl-libc.org/wiki/BuildingLLVM may be of interest for that Nov 15 01:46:20 we could hard-code some thing for one target arch as an initial proof of concept project since mtahmed has only 4 more weeks of co-op. Nov 15 01:46:35 yeah, that would work, just a separate toolchain for each target configuration, no multilib Nov 15 01:47:04 i honestly have no idea what clang's default search paths are for libs & headers, that'd be worth investigating so you know where to put things like crti.o Nov 15 01:47:15 sounds like a fun project though, good luck with it Nov 15 01:47:20 will definitely be curious to hear how it goes Nov 15 01:48:12 http://ellcc.org/viewvc/svn/ellcc/trunk/ is still active - that's good. If only they used git! Nov 15 01:48:19 git-svn is our friend ;) Nov 15 01:48:53 remember that you don't need to build llvm+clang in yocto necessarily, a first step could be to find an existing one and alter the external toolchain stuff to grab its C library & headers Nov 15 01:49:03 could then move on to the more complex bootstrapping process from there Nov 15 01:49:11 your call Nov 15 01:49:15 but something to keep in mind Nov 15 01:49:54 at least that would get you to the point where you'd be able to fix any build failures resulting from the use of clang rather than gcc :) Nov 15 01:49:57 (theoretically) Nov 15 01:56:14 Are there some small examples of using a non-host compiler for builds? Nov 15 01:57:33 not sure what you mean by that Nov 15 01:57:40 yocto always uses a non-host compiler for builds Nov 15 01:57:45 it builds its own crosscompiler Nov 15 01:58:08 unless you configure it for an external toolchain, e.g. the mentor graphics / codesourcery one (see https://github.com/MentorEmbedded/meta-sourcery) Nov 15 01:58:33 Well, I am quite noob. Then are there examples of using an external one? Nov 15 01:58:34 you should be able to use that as an example if you wanted to, say, use a prebuilt ellcc rather than trying to get yocto to build llvm+clang Nov 15 01:58:36 Okay I could look at that. Nov 15 01:58:45 poky includes an external toolchainc onfiguration right in it Nov 15 01:58:47 in oe-core Nov 15 01:59:05 but it's not well supported, and its a sourcery example, so you're better off using meta-sourcery as an example, i think Nov 15 01:59:39 Okay that would be (much?) easier? Just build clang separately, drop it in somewhere, and use to with the appropriate target arguments? Nov 15 02:00:09 And then see where/why I am getting errors (with include paths etc.) Nov 15 02:00:11 tough to say without digging in in more detail, but from a look at hte musl page, there are quite a few components you'll need to build, which means you'd need recipes for each Nov 15 02:00:16 * kergoth shrugs Nov 15 02:00:27 I've never built llvm+clang from scratch myself, so I'm no expert :) Nov 15 02:01:01 I have experience with llvm+clang ... I am maintaining (some) archlinux packages for it as well. Nov 15 02:01:22 I have little experience with bitbake/yocto/oe/poky stuff though. Nov 15 02:01:44 Which is what I am seeking help with. And thanks for the tips. I need to save them somewhere now :) Nov 15 02:02:36 no problem Nov 15 02:11:30 git svn clone -s http://ellcc.org/viewvc/svn/ellcc/ ellcc.git didn't work -- I have a custom git in /usr/local so I'll have to fix: Nov 15 02:11:32 Repository has been moved: Repository moved permanently to '/viewvc/svn/ellcc'; please relocate at /usr/local/share/perl/5.14.2/Git/SVN.pm line 310 Nov 15 02:11:37 tomorrowish... Nov 15 02:20:19 Okay. Nov 15 02:26:39 viewvc isn't an svn repository url, afaik, its likely the web browsing interface to it Nov 15 02:26:52 per http://ellcc.org/blog/?p=251: % svn checkout http://ellcc.org/svn/ellcc/trunk ellcc **** ENDING LOGGING AT Fri Nov 15 02:59:58 2013