**** BEGIN LOGGING AT Mon Nov 19 02:59:59 2012 Nov 19 08:41:12 morning all ! Nov 19 08:47:41 hi silvio Nov 19 08:47:55 hi afournier , hi all Nov 19 09:29:28 morning all Nov 19 09:30:36 bluelightning: morning Nov 19 09:33:00 hi bluelightning Nov 19 10:27:45 uf. Nov 19 10:27:52 mysql fix sent to ML Nov 19 10:40:58 good morning Nov 19 11:11:12 know about trouble with netbase? my interfaces dont start at boot with angstrom Nov 19 13:38:48 why does bitbake -g image give me -dev packages and even some packages which aren't installed in the target rootfs? Nov 19 13:39:21 I though the only things in package-depends would be what was packaged and installed in the target root? Nov 19 13:40:13 e.g. I have a package libwebsockets, which is a library, and is installed just as a library, but package-depends only lists libwebsockets-dev Nov 19 13:41:42 sorry, I like it lists ALL the libwebsocket packages m -dev -dbg -locale, Nov 19 13:41:49 s/like/lie Nov 19 13:56:51 jackmitchell: the actual packages included in the root filesystem are computed by the package manager during do_rootfs Nov 19 13:58:20 so we don't have all of that information when -g is executed Nov 19 13:58:59 thus the -g output just includes all packages the build system knows about (which may not actually include all packages, e.g. in the case of optional modules) Nov 19 14:00:57 bluelightning: ah, ok Nov 19 14:02:10 if you would like information about packages installed into the image (including graphs) you can enable buildhistory; these will only be produced after the image has been built though Nov 19 14:03:05 JaMa: hi - any known issues with Qt 4.8.3 or is it working smoothly in your testing? Nov 19 14:34:20 bluelightning: no issue afaik Nov 19 14:34:59 JaMa: ok, great... ross just sent some patches to make 4.8.3 the default and remove 4.8.1 Nov 19 14:36:04 good Nov 19 14:42:17 is libtool meant to be included in meta-toolchain Nov 19 14:42:22 as I get the error: error: possibly undefined macro: AM_INIT_AUTOMAKE Nov 19 14:42:33 which after a quick google points to libtool not being present Nov 19 14:43:02 backinfo: I am trying ot convert a project to autotools and this error pops up when running autoconf -i Nov 19 17:14:10 JaMa: I got further with qt5 and found some serious issues Nov 19 17:14:33 JaMa: I guess you're building qtbase for same arch you're using in host, right? Nov 19 17:31:43 JaMa: The build system is trying to use run the local moc/uic/... binaries (which are ARM binaries in my case) so it fails Nov 19 17:42:52 levonmaa: ^ Nov 19 17:43:03 levonmaa: did you get this failure in qt5 too? Nov 19 17:53:43 bluelightning: referring to your patch on [meta-networking], why do you create an *.inc file and just write {md5,sha256}sum in an actual bb file? Nov 19 17:55:12 eren: the checksums are specific to the version since the tarball would be different; the rest are common parts that go into the inc Nov 19 17:55:34 (just allows additional versions to be added without having to duplicate the common parts) Nov 19 17:55:56 bluelightning: you didn't do it in proftpd package Nov 19 17:56:18 no, I basically kept to however it was set up in OE-Classic Nov 19 17:58:03 bluelightning: so the convention is to keep common code in *.inc, (summary, description, building, installation, user addition, etc), and put changin parts in bb file? In this way, updating the package is simply renaming the bb file and updating sha, md5 sums Nov 19 17:59:34 I also came across PR = "r0". Some said that it's not required, and should be removed because the default is "r0" while others keep PR variable specifically so that it's not forgotten to update the revision. Are there different approaches for oe-core, and for other repositories? Nov 19 18:02:05 eren: certain people insist PR = "r0" should be removed; I prefer to keep it in because it only has to be re-added if it is incremented and then it often gets added in a different place in the file than it was previously Nov 19 18:03:30 bluelightning: yeah, that also seems right to me. What about the convention on *.inc and bb? Nov 19 18:04:06 there's no hard requirement either way Nov 19 18:04:32 if there are multiple versions of the package with known compatibility issues it may be a good idea to have a common .inc though Nov 19 18:07:12 got it, thanks Nov 19 18:10:34 I've completed the half of the document. It includes building bitbake, explaining base class, adding tasks, overwriting the tasks (with autotools example), showing how the tasks are overwritten using run_TASK.PID files in temp. I will be releasing soon on mailing list when I finish explaining OpenEmbedded code :) Nov 19 18:10:59 cool Nov 19 18:13:39 otavio: right, have not seen it as I have been working on x86 so far Nov 19 18:14:04 levonmaa: yes but this doesn't show the native tools issues Nov 19 18:14:18 otavio: not sure I follow... Nov 19 18:14:49 levonmaa: currently qtbase is trying to run QT_SOURCE_DIR/bin/moc for example Nov 19 18:14:59 levonmaa: and this is an arm binary Nov 19 18:15:03 ah.. thats wrong Nov 19 18:15:09 levonmaa: yes; I know Nov 19 18:15:12 ;) Nov 19 18:15:21 levonmaa: and this is what needs fixing now Nov 19 18:15:27 yes Nov 19 18:16:34 JaMa, otavio: just to make sure we are not duplicating work, who should pick it up? Nov 19 18:17:34 levonmaa: I am working on this at this moment Nov 19 18:17:40 levonmaa: and what you're working on? Nov 19 18:17:52 otavio: ok, I Nov 19 18:17:55 moops Nov 19 18:17:57 oops Nov 19 18:18:37 Im looking into the installations and then I have prelim recipes for qtjsbackend and qtdeclarative Nov 19 18:19:02 levonmaa: I think we ought to sort out qtbase and qt-tools before Nov 19 18:19:03 but they are still work in progress Nov 19 18:19:41 otavio: agreed Nov 19 18:22:13 otavio: with the installations above I was referring to that fact that qtbase still installs wrongly Nov 19 18:26:31 woglinde: Are java building for danny? Nov 19 18:26:42 woglinde: mario-goulart had problem building it here Nov 19 18:27:00 otavio hm I am only testing against master Nov 19 18:27:09 and I have no time at the moment Nov 19 18:27:18 mario-goulart: do you remember the failure? Nov 19 18:27:53 no time for meta-jaba of course Nov 19 18:27:58 ups java Nov 19 18:28:56 otavio: I don't remember the exact error. I think I posted a log. I'm trying to find it. Nov 19 18:29:23 I'm afraid I set the expire time for something short, like 3 days or something... Nov 19 18:29:58 mario-goulart or did we talk about it already? Nov 19 18:32:00 woglinde: No, we don't. Nov 19 18:32:16 As I expected my error log posted to paste.debian.org has expired. Nov 19 18:32:40 I remember I got errors similar to the ones showed here: http://lists.linuxtogo.org/pipermail/openembedded-devel/2012-August/041120.html Nov 19 18:32:59 There's a patch at that thread, but I couldn't apply it, IIRC. Nov 19 18:33:24 than run it again Nov 19 18:33:29 and show me the error Nov 19 18:33:46 Ok Nov 19 18:38:06 hi Nov 19 18:38:20 woglinde: http://paste.debian.net/210771/ Nov 19 18:39:43 woglinde: I run "bitbake openjdk-6-jre" Nov 19 18:45:28 mario-goulart is that the native or target llvm package? Nov 19 18:45:37 I never build llvm for x86 targets Nov 19 18:52:06 woglinde: I don't know. How can I check that? Nov 19 18:54:03 I'm bitbaking llvm2.8 to try to find some info. Nov 19 19:01:34 otavio: seems like the native recipe is missing a patch that allows the configration of the qt.conf location Nov 19 19:01:43 levonmaa: yes Nov 19 19:01:57 levonmaa: I fixed it locally and I am finishing to test it for uploading Nov 19 19:02:06 otavio: groovy Nov 19 19:02:42 otavio: there are two patches, imo the 0001-Allowing-the-customization-of-the-paths.patch it the right one Nov 19 19:02:56 s/it/is Nov 19 19:04:56 levonmaa: Yes; I am running devshell to confirm it worked Nov 19 19:05:03 ok Nov 19 19:07:36 otavio: have you seen this before http://paste.debian.net/210781/ Nov 19 19:07:53 just popped up on my other machine... Nov 19 19:08:29 levonmaa: no; never Nov 19 19:14:33 woglinde: it is for target Nov 19 19:21:42 I guess it would be useful to have branch information on oe-commits list Nov 19 19:21:46 i.e Nov 19 19:22:02 [oe-commits] [master-next] Richard Purdie : scripts/pybootchart: Allow minimum task length to be configured from the commandline Nov 19 19:22:39 now, when skimming through commits, opening the mail explicitly is needed to just get which branch the commit has been made into Nov 19 19:23:26 it looks like a duplicate commit, and make me think that "hey, I remember it!" :) Nov 19 19:28:14 levonmaa: pushed Nov 19 19:38:17 eren: I'm using filter to move emails from oe-commits with "Branch: master$" to separate folder Nov 19 19:41:08 JaMa: hm, a possible solution Nov 19 19:41:52 JaMa: what about other branches, like master-next? Nov 19 19:42:40 + denny and denzil Nov 19 19:44:40 heh, I will just create folders for each branches for now :) Nov 19 19:57:19 otavio: why are you changing author of 0001-qlibraryinfo-allow-to-set-qt.conf-from-the-outside-u.patch ? please rebase it in qtcore/meta-qt5/oe-core-5 Nov 19 20:49:48 are SECTION and PRIORITY deperectaed now? Nov 19 21:36:10 guys, can we build stuff for android using OE? I saw some fork attempts but I wonder if anything is in "officially", would be quite handy for cross compiling various linux command line tools and so on Nov 19 21:48:07 Jin^eLD, there was an oe classic fork Nov 19 21:49:11 and with core, noone tried? Nov 19 21:49:33 it's old Nov 19 21:49:51 try to look in the openembedded mailing list Nov 19 21:50:26 I still do not quite understand what android actually is :) got my first tablet today and it looks like a crippled linux Nov 19 21:50:52 but I am not yet clear about the big differences Nov 19 22:08:35 jineld hm Nov 19 22:09:12 jineld android is linux-kernel+bluez+dbus+dalvik Nov 19 22:09:38 then why the hell is everything there so crippled? :) Nov 19 22:09:55 will google up on dalvik Nov 19 22:10:04 hm what you mean with crippled? Nov 19 22:10:15 you can install a bash if you want Nov 19 22:10:26 that I did by the way, but it's all very strange there Nov 19 22:10:31 path layout is completely messed up Nov 19 22:10:32 the bioniclib dont support all glibc features Nov 19 22:10:42 ah, so they use "bioniclib" as the c library? Nov 19 22:10:51 sure Nov 19 22:11:04 because its bsd licensends Nov 19 22:11:13 ah, I see Nov 19 22:11:27 ah I frogot webkit engine Nov 19 22:11:52 any special gcc tweaks or just the regular stuf? Nov 19 22:12:05 just regular stuff Nov 19 22:12:21 but since 4.1 or 4.0 the gcc is 64bit only Nov 19 22:12:25 because from what you told me it sounds like we should be able to build stuff in OE for it, provided I manage to replace the libc with this bionic stuff Nov 19 22:13:48 good nite Nov 19 22:14:05 nite, thanks for the hints Nov 20 01:02:38 JaMa: I did some changes and prepared it for sending to upstream; just that Nov 20 01:07:36 ok Nov 20 01:33:45 JaMa: but I didn't get what you meant with rebase? **** ENDING LOGGING AT Tue Nov 20 02:59:57 2012