**** BEGIN LOGGING AT Thu Oct 25 03:00:01 2012 Oct 25 06:10:46 JaMa|Zzzz: Thanks for the help yesterday. The error was that I had a FILES_${PN} line there. After removing it it started to work. Oct 25 06:12:10 The google font directory has fonts with different licenses (in different directories) how should I handle that in bitbake recipe? Oct 25 06:13:01 Or is is better to write two receipes that share the same SRC_URI ? Oct 25 06:48:26 anybody aware of xinput layer beeing able to filter debounces ? Oct 25 07:09:00 inch: hehe, pity you haven't shown whole recipe, I was thinking about FILES_PN but then didn't mention it as fonts is not included there by default Oct 25 07:14:33 gm Oct 25 07:14:41 jama still in bed? Oct 25 07:22:31 woglinde: just wake up Oct 25 07:28:13 hi woglinde, all Oct 25 07:28:47 is it possible to specify a different user agent in a SRC URI ? Oct 25 07:29:02 in oe classic wget does not like to build libedit, fetches with error 406 Oct 25 07:29:17 adding --user-agent=Konqueror to wget manually works Oct 25 07:33:16 for now faked things with my own do_fetch Oct 25 07:33:17 do_fetch () { Oct 25 07:33:17 wget --user-agent=Konqueror ${SRC_URI} Oct 25 07:33:17 } Oct 25 07:33:42 effem no idea Oct 25 07:43:49 woglinde: the question was not specific to you, the hi was a reply to your gm from 15 minutes before Oct 25 07:44:36 not sure whether the meta-oe recipe builds or has the same error Oct 25 07:44:38 hi ant_work Oct 25 07:46:19 'morning eFfeM_work Oct 25 08:10:27 he stefan Oct 25 08:10:36 moin woglinde Oct 25 08:10:50 woglinde: how is life? Oct 25 08:12:30 depends Oct 25 08:12:32 *g* Oct 25 08:14:56 gm bluelightning Oct 25 08:15:10 right now I am trying to build qt5 with wayland support Oct 25 08:15:29 morning woglinde, all Oct 25 08:15:51 bl are there plans at intel side for wayland in oe? Oct 25 08:15:51 woglinde: ah, interesting... have you been talking to otavio? Oct 25 08:16:05 bl umh without oe sofar Oct 25 08:16:11 ah right Oct 25 08:16:15 just for trying on my desktop Oct 25 08:16:31 ah, boring :) Oct 25 08:16:54 woglinde: I think there might be some work on wayland in the 1.4 cycle yes Oct 25 08:17:09 one problem is that under ubuntu 12.10 egl and gles needs libwayland-client at linking time Oct 25 08:17:16 but thats easy to solve Oct 25 08:17:53 but nice is that you can build qt5 completely without xcb support Oct 25 08:18:05 args Oct 25 08:18:08 damn tabs Oct 25 08:18:14 I killed the conmpile tab Oct 25 08:19:17 ynezz btw. you are boring too ,) Oct 25 08:20:39 I know :) But well, I thought that you're doing it for beagleboard or so :) Oct 25 08:21:03 btw, hi Oct 25 08:22:31 haha Oct 25 08:22:37 not yet Oct 25 08:22:45 but should not be hard Oct 25 09:32:24 How do I make packages created by "PACKAGES_DYNAMIC" and "do_split_packages" not depend on the package whose name is the name of the .bb file? Oct 25 09:34:53 inch: this may be helpful as a general reference: https://wiki.yoctoproject.org/wiki/Handling_optional_module_packaging Oct 25 09:35:29 inch: specifically you can set extra_depends='' to not add that dependency Oct 25 09:36:09 Thanks. Oct 25 09:36:09 is there a way to easily remove distro features? For example I would like to keep the Poky distro (for now anyway) but remove the x11 distro feature. Would the best way to be just removing it from the oe-core default-distro-features? Oct 25 09:46:53 jackmitchell: you can set DISTRO_FEATURES from local.conf Oct 25 09:47:18 jackmitchell: ultimately the best solution is to take a copy of the poky distro config and set DISTRO_FEATURES there though Oct 25 09:47:40 (but renaming the conf to something different) Oct 25 09:48:29 bluelightning: yes, that was my itentention eventually Oct 25 09:49:33 bluelightning: does the distro conf settings change regularly or will I be constantly trying to keep a cusom conf in line? Oct 25 09:55:58 jackmitchell: occasionally they do yes, but not often enough to be painful in that regard I would think Oct 25 10:38:31 bluelightning: what is LOCAL_CONF_VERSION? Oct 25 10:39:14 no mention of it in the mega-manual, is it something poky specific that I can strip out for my own distro config? Oct 25 10:40:02 jackmitchell: it gets compared to LCONF_VERSION in local.conf to force the user to update it if the default local configuration has changed Oct 25 10:40:24 bluelightning: ok, thanks! Oct 25 10:40:36 jackmitchell: we don't change the value often, but if the local.conf template changes significantly then we bump it Oct 25 10:40:51 I suspect if you don't set it you'll have trouble with the sanity check Oct 25 10:41:00 bluelightning: np, I'll leave it in Oct 25 10:41:25 I think it's going to be a bumpy ride anyway with the disto-images i've stripped out Oct 25 10:41:40 s/distro_images/distro_features Oct 25 11:03:41 So with my new distro config I am getting the error: Failed to fetch test data from the network. Please ensure your network is configured correctly. Oct 25 11:03:41 ERROR: Execution of event handler 'check_sanity_eventhandler' failed Oct 25 11:04:04 however I have tried both of the connectivity check urls and can access both of them Oct 25 11:05:00 jackmitchell: sanity check has shorter timeout Oct 25 11:05:18 jackmitchell: fails here too (while wget used outside bitbake works) Oct 25 11:05:25 JaMa: ok, cheers Oct 25 11:05:40 JaMa: I have just removed the check, I'm competent enough to know if I'm online or not ;) Oct 25 11:37:27 I finally managed to package the fonts. I ended up creating 3 bb-recipes, one for each license: http://pastebin.com/ShiR9Gt4 Oct 25 11:37:57 morning all Oct 25 11:46:04 "File log.do_compile is large (118MB), really open?" Oct 25 11:46:06 * pb_ stabs webkit Oct 25 12:07:10 pb_: have fun with that! Is webkit getting slowly better or worse - I understand it's already pretty bloated, or is it all necessary? Oct 25 12:08:14 I don't know specifically but unfortunately the web is actually quite complicated these days :( Oct 25 12:09:15 There's certainly a lot of it, though quite a lot of the source code is platform specific and doesn't actually get built on any given system. Oct 25 12:09:44 It's hard to say that much of it is actually unnecessary, though there are a few features (like webgl which seems to bring the whole of ANGLE in) which I don't especially need/want. Oct 25 12:09:46 bluelightning: yes, I believe javascript engines take part for a lot of the blame - I think something really needs to superceed it Oct 25 12:10:37 there are dozens of languages popping up which compile down into javascript but I think someone needs to take the bull by the horns and get the ball rolling something new Oct 25 12:11:17 google are playing with a language called DART but the support is currently poor, and I would be suprised if anyone else is interested in supporting it Oct 25 12:11:39 fwiw, the JavaScriptCore directory is "only" 26M, which makes it the smallest of the three big subfolders within webkit. Oct 25 12:11:54 WebCore is the big one, that's 131M, and WebKit is 39M. Oct 25 12:15:10 pb_: that's a surprise! So what is (at high level) WebCores main function? Oct 25 12:16:26 http://en.wikipedia.org/wiki/WebKit#WebCore Oct 25 12:16:35 "WebCore is a layout, rendering, and Document Object Model (DOM) library for HTML and SVG" Oct 25 12:17:10 That's where all the heavy lifting happens. The stuff in WebKit/ is, confusingly, mostly just API glue and chrome. Oct 25 12:20:06 pb_: ah ok, very interesting Oct 25 12:29:14 I'm trying to fix up recipes that needlessly pull in X dependencies when x11 distro_features is not enabled Oct 25 12:29:22 how would I go about doing the following Oct 25 12:29:23 EXTRA_OECONF_X = "${@base_contains('DISTRO_FEATURES', 'x11', '', '#undef HAS_GTK_GUI', d)}" Oct 25 12:29:42 can I put bash into that function to undefine HAS_GTK_GUI> Oct 25 12:46:06 http://www.linaro.org/engineering/armv8/ if someone want to play with aarch64 and OE Oct 25 12:58:19 jackmitchell: I can't quite make out what that's supposed to do, but for most cases of simple configure script arguments you should use PACKAGECONFIG Oct 25 12:59:36 jackmitchell: ah I see your patch, for that case I guess PACKAGECONFIG isn't really applicable Oct 25 13:08:09 bluelightning: yes I switched it round, instead of disabling GTK I now enable it if required - was much easier ;) Oct 25 14:46:12 sigh... Oct 25 14:46:46 * bluelightning ponders whether to mangle the quagga recipe to put the binaries in /usr/lib as was supposedly intended in the OE-Classic version (to match debian) Oct 25 14:47:12 binaries in /usr/lib seems wrong to me somehow... Oct 25 14:47:18 well, executables Oct 25 14:47:49 ick Oct 25 14:48:55 what's the proper way of bbappending patches to the gcc-cross package? Oct 25 14:49:07 i have a gcc-cross_4.7.bbappend Oct 25 14:49:11 and it contains: Oct 25 14:49:14 FILESEXTRAPATHS := "${THISDIR}/gcc-4.7" Oct 25 14:49:14 SRC_URI += "file://iwmmxt_fix.patch" Oct 25 14:49:43 this doesn't seem to work as i don't see the patch in work-shared/gcc directory Oct 25 14:50:00 dromede: that should probably be FILESEXTRAPATHS_prepend := "${THISDIR}/gcc-${PV}:" Oct 25 14:50:17 although that's unlikely to be the cause of the issue Oct 25 14:50:31 i wonder if work-shared is a factor, as that change only affects one of th recipes sharing htat workdir. do all need to be adjusted? I never read the shared workdir code Oct 25 14:50:38 heh Oct 25 14:50:47 * kergoth gets food Oct 25 14:51:01 yes, I'm unclear on how shared workdir behaviour as well Oct 25 14:51:02 strangely enough if i do the same thing for gcc-cross-initial, the patch gets applied Oct 25 14:51:30 i'll try the "_prepend" Oct 25 14:52:30 oh and by the way, are the ":=" semantics in bitbake the same as the ones in Make? Oct 25 14:52:46 simply expanded variables? Oct 25 14:54:08 immediate expansion, ,yeah Oct 25 14:54:20 we based the expansion on make Oct 25 14:54:50 that's good to know, thanks Oct 25 14:56:25 np Oct 25 16:12:32 bluelightning: adding "_prepend" did the trick Oct 25 16:14:51 dromede: great, I guess we are setting FILESEXTRAPATHS elsewhere and that was stomping on the value you were setting; prepend allows the addition to happen later Oct 25 16:15:19 ah, that makes sense Oct 25 16:15:57 but still, pixman fails to build with iwmmxt code Oct 25 16:16:08 although this time it's not the compiler's fault Oct 25 16:16:12 it's a missing file Oct 25 16:24:21 hmm, this chromium recipe in meta-browser is a bit hokey Oct 25 16:24:27 dies if the build machine doesn't have ld.gold Oct 25 16:24:30 no gtk+ dep.. Oct 25 16:24:32 * kergoth rolls eyes Oct 25 16:28:49 * mwester is unfamiliar with ld.gold Oct 25 16:29:11 general replacement for ld in OE, or only selected recipes? Oct 25 16:29:31 gold is a replacement linker you can opt-into in oe using a distro feature Oct 25 16:29:41 but this recipe is requiring it as *native*, which is just bad Oct 25 16:29:47 thanks. will have to take a look at that. Oct 25 16:30:18 iirc the feature is 'ld-as-gold' Oct 25 16:30:22 haven't tried it myself Oct 25 18:12:21 kergoth: chromium needs udev ?! Oct 25 18:12:30 apparently it wants to link against libudev Oct 25 18:12:32 no clue why Oct 25 18:12:39 haven't bothered to dig further Oct 25 18:13:06 sounds a bit unusual for a web browser to need that... Oct 25 18:13:11 indeed Oct 25 18:25:15 kergoth: we've been using firefox here at ossystems. I'm quite sure otavio hasn't built chromium, but I may be wrong. Oct 25 18:25:41 okay, thanks. just wanted to check on it, before we mess with it Oct 25 18:26:11 Alright. :-) It would be nice to have a working chromium recipe. Oct 25 18:27:37 * kergoth nods Oct 25 18:55:12 qtwebkit needs udev Oct 25 18:57:56 woglinde: where do you have qtwebkit recipe? meta-webos is building sort of qtwebkit with webkit-webos :) Oct 25 18:58:17 or you mean webkit enabled in qt4 recipes? Oct 25 19:00:08 in qt5 and yes qt4 Oct 25 19:00:23 or it was something else in qt Oct 25 19:00:24 sorry Oct 25 19:44:26 where are cross toolchain installed? STAGING_DIR_NATIVE or STAGING_DIR_TARGET? Oct 25 19:53:59 native of course Oct 25 20:08:26 BTW if anyone is interested in RPM package issues, I recently sent an RFC about moving from Zypper to SMART to the oe-core list. Oct 25 20:08:34 I'm looking for feedback, if anyone has any **** ENDING LOGGING AT Fri Oct 26 03:00:01 2012