**** BEGIN LOGGING AT Thu Jan 12 02:59:57 2012 Jan 12 08:11:48 has anyone tried to bring X up in minimum configuration Jan 12 09:45:37 morning all Jan 12 09:58:02 hi bluelightning Jan 12 10:07:50 hi ant_work Jan 12 15:30:26 gm Jan 12 15:30:48 I'm attempting to build an image that does not use busybox Jan 12 15:31:58 I'm running into an issue where the image build fails because the python rpm package depends on /usr/bin/env Jan 12 15:32:41 coreutils in my image provides this functionality, but the link isn't created until the postinst is run at first boot Jan 12 15:33:16 what is the proper way to have the coreutils recipe indicate that it provides the /usr/bin/env file? Jan 12 15:34:45 update-alternatives is proper way Jan 12 15:35:04 heh, i know you can build a rootfs with busybox *and* the proper tools, but dont think i've ever seen anyone trying to do without it entirely Jan 12 15:35:05 just teach it to run in do_rootfs Jan 12 15:35:31 kergoth: there were some patches for OE-classic afaik Jan 12 15:36:09 that reminds me, wonder if we ever got the c implementation of update-alternatives into the build, might have value Jan 12 15:36:42 JaMa: is there an example somewhere of how to "teach it to run in do_rootfs" Jan 12 15:49:06 another area that would have a value in speeding up first boots on images with many gnome packages would be to give gtk-icon-cache.bbclass some love Jan 12 15:49:47 the postinst it installs burns a good 30 seconds for *every* gnome package! Jan 12 15:50:13 specifically the gtk-update-icon-cache call Jan 12 15:50:31 yow Jan 12 15:50:35 sakoman_: I've just checked /usr/bin/env and I have it pointing to env.coreutils in image (after do_rootfs) already Jan 12 15:50:44 it'd be nice to do something like what other desktop distros do with things like that Jan 12 15:50:52 where you have hooks run after all the packages are installed Jan 12 15:50:54 to update things like that Jan 12 15:50:57 rather than for each Jan 12 15:51:03 yes! Jan 12 15:51:37 JaMa: hmm . . . let me check Jan 12 15:52:15 perhaps it is an rpm thing Jan 12 15:52:42 and the coreutils rpm isn't advertising the availability of /usr/bin/env Jan 12 15:52:44 I'm using ipk Jan 12 15:52:55 JaMa: I figured :-) Jan 12 15:54:10 yeah, with rpm based image I don't see the links Jan 12 15:54:55 they don't get created till first boot Jan 12 15:55:34 so I wonder if that is a bug, or just an expected difference in behaviour for rpm Jan 12 15:56:55 sakoman_: maybe wrong VIRTUAL-RUNTIME_update-alternatives setting? Jan 12 15:57:59 or PREFERRED_PROVIDER_virtual/update-alternatives Jan 12 15:58:06 JaMa: using the poky default afaik, but will check Jan 12 15:59:14 default is update-alternatives-cworth afaik also for poky, but not sure if it's available during rootfs as update-alternatives-cworth is provided by opkg Jan 12 16:01:05 JaMa: thanks for the hint, will research that Jan 12 19:24:01 * zenlinux throws a dagger at zenlinux_ Jan 12 19:25:59 * incandescant takes not to be mindful around zenlinux and silverware Jan 12 20:28:09 damnit, getting a wrong exit code issue from bitbake again Jan 12 20:28:29 oh, wait, there was an error, but the error message summary/count didn't show it Jan 12 21:00:12 I'm thinking about doing some sort of build completion summary which shows info from buildstats/buildhistory and the like, particularly things like how many recipes were built from sstate vs not. does that seem like something that'd be useful? Jan 12 21:04:52 kergoth: how hard is it to get the buildstat/buildhistory setup and going? Jan 12 21:05:03 is it as simple as turning it on? Jan 12 21:05:52 buildstats is always running, emits info about task exec times, cpu usage, etc Jan 12 21:06:04 buildhistory is just a matter of inheriting it, adds info about package sizes, etc for each image you build Jan 12 21:14:04 do you have to say where the GIT repo is? Jan 12 21:14:16 (i should probably just read about it instead of asking questions) Jan 12 21:14:18 ;) Jan 12 21:16:01 i think it init's automatically Jan 12 21:45:24 hmm, anyoine hitting a samba do_qa_configure failure? Jan 12 22:34:31 * zenlinux loves build failures when upgrading a recipe - when I get to discover another patch of ours has been upstreamed. :D Jan 12 22:35:37 nice Jan 12 22:41:57 those are good pains Jan 12 22:42:17 consider chopping half a year of work in libc so find out a bug Jan 12 22:42:21 thats a bad pain **** ENDING LOGGING AT Fri Jan 13 02:59:57 2012