**** BEGIN LOGGING AT Tue Mar 19 02:59:58 2013 Mar 19 06:52:09 hi Mar 19 06:52:30 how can we use if condition in bb file Mar 19 07:00:53 does echo command work in .bb file Mar 19 07:01:02 "echo" Mar 19 07:01:38 because for me its giving unparsed line when i'm using echo Mar 19 07:02:49 and when i'm writing it in do_configure function its giving "failure expanding do_configure, shellsyntaxerror Mar 19 07:12:47 there is good documentation on bitbake, are you reading that? Mar 19 07:13:20 http://www.openembedded.org/wiki/Documentation Mar 19 07:24:47 thanks for the documentation link, but why is echo not working? Mar 19 07:24:52 i dont get it Mar 19 07:28:40 a guess: because that's not a shell script Mar 19 07:44:32 Anyone has compiled MESA 9.0.1 ? Mar 19 07:44:45 or upgraded mesa to 9.0.1 from lower version? Mar 19 07:46:21 i've used echo command in do_configure in glib, it worked, it doesnt give any error Mar 19 07:51:06 hi Mar 19 07:55:51 http://patches.openembedded.org/patch/33105/ is still a problem Mar 19 08:26:26 morning Mar 19 08:26:36 what does this mean ? "ERROR: QA Issue: non -dev/-dbg/-nativesdk package contains symlink .so" Mar 19 08:28:06 it means your recipe is putting a shared library into a package other than those that are supposed to hold such files Mar 19 08:28:12 that some package (probably PN) contains file like /usr/lib/libfoo.so -> /usr/lib/libfoo.1.0.so Mar 19 08:28:17 yes Mar 19 08:28:24 retrying with pkgconfig Mar 19 08:28:28 :/ Mar 19 08:28:32 .so belongs to PN-dev Mar 19 08:28:41 it's a shared lib Mar 19 08:28:52 shoud be libPN Mar 19 08:29:16 only in few exceptions where .so are not versioned (and not symlinks) belong to PN Mar 19 08:29:45 but the host does not need those symlinks ? Mar 19 08:31:49 no, apps should link to libfoo.1.so not libfoo.so Mar 19 08:32:03 my recipe is for a library :/ Mar 19 08:32:10 not an app only recipe Mar 19 08:33:54 and? Mar 19 08:34:30 and my FILES_${PN} is wrong Mar 19 08:49:07 hi. Is there any provision to enable building packages with debug symbols ? Mar 19 08:49:58 e.g. I'd like to use gdb to debug sshd, but by default openssh is compiled without debug symbols Mar 19 09:20:39 eh. --enable-obsolete-rpc still in use ;( Mar 19 09:34:21 hi, I have one big tarball that is used by several recipes. my concern is that if I set this tarball's location as the SRC_URI, that it will be unpacked for every individual recipe, even though I only use a small subset of the tarball's contents each time. Mar 19 09:34:48 if this is the case (bitbake would unpack it every time for each recipe), is there a way to "share" the unpacking result amongst recipes? Mar 19 09:34:59 morning all Mar 19 09:35:26 dv_: I don't think there is a way to do that, no Mar 19 09:35:32 good morning bluelightning Mar 19 09:35:35 not a standard way, anyway Mar 19 09:35:38 hi apelete Mar 19 09:36:03 hmm :/ the only other option I can think of is to write a custom do_unpack function that specifies which files to unpack Mar 19 09:36:17 no idea how to actually do that though Mar 19 09:36:32 I'll dig through bitbake's unpacking facilities Mar 19 09:50:06 dv_: take a look at gcc recipes Mar 19 09:50:15 dv_: they use shared source Mar 19 09:50:25 ah! good. thanks. Mar 19 09:54:31 bluelightning: I started creating a "meta-jlime" layer, using the instructions you wrote at http://www.openembedded.org/wiki/Migrating_metadata_to_OE-Core Mar 19 09:56:53 apelete: nice... how is that going? Mar 19 09:57:05 I spent a lot of time reading about how a layer is used by the oe-core build system (in order to understand whatI actually have to do to create a layer) Mar 19 09:57:17 but I discovered that there's an awful lot of documentation on the yocto project website, that's great Mar 19 10:02:29 so, the work is moving slowly (doing it only on my free time, got day time job to take care of too), but your instructions + documentation found at yocto's is proving very helpful Mar 19 10:03:29 apelete: great! Mar 19 10:11:13 bluelightning: I'll try to push the layer online as soon as possible, in order to get some reviews. will keep you posted. Mar 19 10:11:24 apelete: ok, thanks Mar 19 11:04:27 how to avoid runstrip in qt? Mar 19 11:04:52 i m getting error like runstrip command failed for qmake-palm.. Mar 19 11:05:06 i m building qt for arm? Mar 19 11:46:29 argh.. Mar 19 11:47:06 "if os.path.exists(libpath):" in OE code returns False but directory exists ;( Mar 19 11:48:42 hrw: nfs? directory perms? Mar 19 11:49:02 bluelightning: ext4 and I have access there Mar 19 11:49:30 hmm... well it's a standard python function Mar 19 11:49:39 bluelightning: trying to make linaro toolchains fully working (with hacked oe-core) Mar 19 11:49:55 bluelightning: and same snippet works in ipython ;( Mar 19 11:50:07 In [10]: print os.path.exists("/home/hrw/devel/canonical/aarch64/openembedded/build/linaro-tcwg/gcc-linaro-arm-linux-gnueabihf-4.7-2013.03-20130313_linux//arm-linux-gnueabihf/libc/lib/arm-linux-gnueabihf/") Mar 19 11:50:11 True Mar 19 11:50:14 it's possible pseudo might somehow be getting in the way Mar 19 11:50:27 wouldn't have thought it likely though Mar 19 11:51:01 strace in progress anyway Mar 19 11:51:19 eglibc 2.15 sounds better than eglibc UNKNOWN Mar 19 11:55:01 1.3GB log generated Mar 19 11:55:12 11 mln lines Mar 19 12:09:18 done! Mar 19 12:09:24 one bb.data.expand() more Mar 19 16:55:33 moin Mar 19 16:56:07 anybody got a clue why running opkg-cl in the image recipe would pooch pango modules? Mar 19 17:02:29 mr_science whatẞ Mar 19 17:49:53 woglinde: not sure what "whatẞ" means... Mar 19 17:51:10 essentially, our custom image preprocess stuff short-circuits the "testlab" bbclass function so it never creates the testlab dir eith the build info Mar 19 17:51:22 *with Mar 19 17:53:04 so I added the equivalent functions to my image recipe, and somehow that dorks up opkg configure so that pango gets dorked up which kills the display component that uses pango/clutter Mar 19 17:53:17 is there a way to add URIs to SRC_URI programmatically? I tried putting shell script code in do_fetch_prepend, but thats not even getting executed Mar 19 17:53:47 anonymous python Mar 19 17:54:12 python () { d.appendVar('SRC_URI', ' file://foo') } Mar 19 17:54:18 lots of examples in the metadata Mar 19 17:54:25 could use an event handler also. either way it'd be determined at parse time Mar 19 17:54:27 ah Mar 19 17:54:35 nice, thanks Mar 19 17:54:37 if that's too early, you could use a prefunc on do_fetch Mar 19 17:54:48 you want to avoid _append/_prepend, they cause problems since they blindly concatenate Mar 19 17:54:54 e.g. you can ccidentally append shell to python or vice versa Mar 19 17:54:57 yeah, I noticed already Mar 19 17:55:01 or end up with python indentation mismatch Mar 19 17:55:21 do_fetch[prefuncs] += "some_function" do_fetch[vardeps] += "some_function" Mar 19 17:55:25 then do what you want in some_function Mar 19 17:55:30 but again hopefully you dont need to do that Mar 19 17:55:31 I had to use it for a script that is used in both a yocto-based and an arago-based setup Mar 19 17:55:36 metadata changes are best done from anonymous python Mar 19 17:55:39 the former uses OE-core, the latter OE-classic Mar 19 17:55:48 it was ... ugly. Mar 19 17:56:12 * mr_science is still stuck with arago/classic Mar 19 17:59:18 at least we have a nice NAND-based installer image to load a new release on the SSD Mar 19 17:59:38 finally getting away from "burning" sdcards... Mar 19 18:00:16 I forgot what the second parameter of getVar is Mar 19 18:00:20 its usually set to True Mar 19 18:00:33 mr_science: oh I have the same problem Mar 19 18:00:59 it's expansion Mar 19 18:01:06 controls whether you get teh value expanded or unexpanded Mar 19 18:01:12 (wrt variable references) Mar 19 18:01:21 True = expanded Mar 19 18:01:32 ah right Mar 19 18:01:58 also , hrw: thanks for the gcc suggestion for shared work, works like a charm Mar 19 18:06:13 the arago problem or the testlab problem? Mar 19 18:14:56 mr_science: why don't you just disable testlab completely? or do you still need something from it? Mar 19 18:17:32 i kinda wanted the installed-packages and installed-files lists Mar 19 18:28:35 mr_science: then which part of testlab you don't want? :) Mar 19 18:38:04 not really using the dot files or package sizes Mar 19 19:12:52 if I have multiple SRC_URI entries (separated by whitespaces), how do I specify the checksums for these? can SRC_URI[md5sum] hold a whitespace-separated list? Mar 19 19:16:16 ah hold on. https://lists.yoctoproject.org/pipermail/poky/2010-December/001252.html answers it Mar 19 19:25:55 argh, damnit. this doesnt work either. the http server doesnt like the name= fragemtn Mar 19 19:25:57 *fragment Mar 19 19:26:32 is there a way to disable SRC_URI checksums entirely? Mar 19 19:37:15 dv_ cannt you pastebin your recipe and the exact errormsg? Mar 19 19:37:36 nevermind, BB_STRICT_CHECKSUM = "" seems to work Mar 19 20:54:29 I seem to remember someone talking about objective-c, gcj and fortran support (patches?) for the oe-core toolchain a few weeks ago.. Mar 19 20:54:47 does anyone remember where that email was sent to/ Mar 19 21:02:47 Hmm.. looking like basic gcj and fortran support is alreayd there.. maybe that is what I'm thinking of.. Mar 19 21:02:53 but I thought objective-c was somewhere as well Mar 19 21:16:33 * mr_science goes to the dentist Mar 19 21:39:18 hello **** ENDING LOGGING AT Wed Mar 20 02:59:58 2013