**** BEGIN LOGGING AT Thu Dec 12 02:59:59 2013 Dec 12 06:47:49 afternoon all Dec 12 09:50:57 morning all Dec 12 10:07:34 good morning Dec 12 11:33:47 hii all Dec 12 11:34:44 how to WHITELISt the blacklisted module with out changing the actual file? i have used LICENSE_FLAGS_WHITELIST += "commercial_libxdmcp" like this Dec 12 11:34:58 don;t know how exactly i should do it!! :-( Dec 12 11:39:22 Sj___: libxdmcp doesn't have LICENSE_FLAGS set though does it? Dec 12 11:42:08 hm actually in one of our meta-layer they have used PNBLACKLIST[libxdmcp] = ".." but now i need to enable it with out modifying that file.. Dec 12 11:42:35 is there any way like PNWHITELIST[] ?? to unblacklist that? Dec 12 11:43:46 ah right Dec 12 11:43:54 PNBLACKLIST and LICENSE_FLAGS are completely unrelated Dec 12 11:44:01 ok Dec 12 11:44:09 PNWHITELIST can i use?? Dec 12 11:44:52 well, the question is why are you using something that blacklists the recipe if you don't want it blacklisted? are you re-using someone else's distro config? Dec 12 11:45:30 yes .. it is some one else distro Dec 12 11:46:03 hmm, well when you start wanting to override other people's distro policy then it's time to start thinking about whether using that distro is the right thing vs. creating your own Dec 12 11:46:09 but anyway, there is no PNWHITELIST Dec 12 11:46:23 you could try PNBLACKLIST[libxdmcp] = "" Dec 12 11:46:29 in their layer they have done it. but i'm using there layer on top of it i have created one meta-layer (need to use same distro) with some changes Dec 12 11:46:49 well this is the thing, when you change the distro policy it's no longer the same distro really Dec 12 11:47:49 ohh ok and one more doubt i have. we have bbappend files for bb files.. what about .inc files? is ther any way we can re config the variables used in .inc files? Dec 12 11:48:48 Sj___: no, but you can bbappend each of the .bb files that the .inc file is included in Dec 12 11:49:07 ohh ! ok thank you Dec 12 11:49:20 so only way fr me to use another distro config Dec 12 11:49:26 for my layer!! Dec 12 11:51:54 as libexecdir issues were not solved for dora; are there plans to fix them in the next release? Dec 12 11:53:52 ensc|w_: which ones still remain? a patch did go in for the sftp + dropbear issue you highlighted Dec 12 11:54:59 bluelightning: this does not solve the underlying problem that libexecdir must not contain $PN Dec 12 11:55:17 and this patch breaks my system because it has such a good libexecdir Dec 12 11:55:34 ensc|w_: must not contain - why? it's an FHS standard... Dec 12 11:56:05 ensc|w_: if it causes specific problems, we should definitely fix those problems Dec 12 11:56:40 the specific problem is that it must not contain $PN but it does ;) Dec 12 11:57:10 ensc|w_: can you explain the problem further? Dec 12 11:57:26 the dropbear patch is broken because it assumes a fixed path of sftp-server, but this program is installed in ${libexecdir} in the openssh recipe Dec 12 11:57:50 ensc|w_: i actually just dug out my patch to "fix" it Dec 12 11:58:04 bluelightning: libexecdir must not contain $PN but must be a path which is common across the whole system Dec 12 11:58:26 bluelightning: our libexecdir definition is contrary to every other distro. Dec 12 11:58:45 debian made the mistake to append the package name in some ancient release and it seems oe adapted it Dec 12 11:58:46 rburton: every? then how did it make it into the FHS? Dec 12 11:59:02 libexexdir is not fhs Dec 12 11:59:19 that was the justification for the change IIRC Dec 12 11:59:32 but even debian doesn't do that anymore :) Dec 12 11:59:38 argh Dec 12 11:59:46 exactly Dec 12 11:59:51 ok I'm not going to argue this because I'm clearly not in posession of the full facts Dec 12 12:00:04 talk to Saul, he made the change... Dec 12 12:00:27 i have a branch that cleans up any crazy libexecdir usage (thus the sato-icon-theme patch) and switches it Dec 12 12:00:31 obviously its a non-trivial change Dec 12 12:00:40 well, the change is trivial. the impact isn't. Dec 12 12:05:19 bluelightning, when has that ever stopped anyone from arguing? Dec 12 12:05:46 sounds like the ideal reason for arguing! Dec 12 12:05:52 heh Dec 12 12:05:52 :) Dec 12 12:06:27 stupid denian :) Dec 12 13:42:16 evening all Dec 12 13:42:38 hi pb_ Dec 12 13:44:21 JaMa: btw, are you still able to reproduce https://bugzilla.yoctoproject.org/show_bug.cgi?id=4790 ? Dec 12 13:46:29 hi bluelightning Dec 12 13:48:14 I'm trying to add mpd (or mplayer2) to an image but I get an error about ffmpeg license LICENSE_FLAGS_WHITELIST, how could I enable it? Dec 12 13:48:56 yay, no more external toolchain! Dec 12 13:50:21 LICENSE_FLAGS_WHITELIST += "commercial" seems useless Dec 12 13:52:08 bluelightning: I'll have to re-try Dec 12 13:52:11 damn! nevermind, it must go in local.conf not in image file Dec 12 13:53:35 FYI: meta-qt5 was just updated for 5.2.0 release, but not even build tested yet (it's in jansa/qt5-5.2.0 branch if someone wants to test) Dec 12 13:53:42 Crofton|work: well, we still have it, it's just in a separate layer Dec 12 13:53:53 I know Dec 12 13:54:09 but now people will use the layer, which people care about :) Dec 12 13:54:14 JaMa: do you use it with dora or master? Dec 12 13:57:39 mckoan: both Dec 12 13:58:02 but first I'll test it with master Dec 12 14:51:13 is anybody builing oe on XFS filesystems (spinning disk)? how is the performance compared to ext4? I have to revert the horribly slow btrfs setup and me wonders which fs is the best for building oe Dec 12 14:52:01 ext2 Dec 12 14:52:09 or ext4 without journal Dec 12 14:53:44 mmh... fsck on ext2 is a pain; I will go to ext4 then Dec 12 14:53:58 perhaps with external journal on a small ssd Dec 12 15:23:02 ensc|w_: way back when I set up my build machine someone recommended I use mkfs.ext4 -O ^has_journal,extents,huge_file,flex_bg,uninit_bg,dir_nlink,extra_isize Dec 12 15:23:13 I've been fairly happy with the performance overall Dec 12 15:24:24 I also have TMPDIR on a different drive to the rest of the system Dec 12 15:24:42 (which is formatted as above; I didn't do anything special with the root partition) Dec 12 15:36:49 I use ext3 or ext4 for all of my builds.. Dec 12 15:48:41 ensc|w_: ext4 on an ssd is a good idea anyway because of TRIM Dec 12 15:49:11 maybe also f2fs Dec 12 16:31:02 dv_: plain SSD is too expensive; I have around 10 reference platforms I build oe for. And this across 3 versions (dylan, dora, master). Then customer projects. Per build I calculate 30 GiB space in worst-case (compilation error during qt-embedded ;) ) Dec 12 16:31:48 ensc|w_: rm_work is useful there Dec 12 16:33:05 rburton: yes; I use this (and rm_work_all). But when bitbake aborts in the middle, there are still a lot of unpacked files Dec 12 16:33:42 true Dec 12 16:34:24 when build is completed, the tmp needs 5-10 GiB and this is not a problem. But when doing a batch build across the reference platforms I have to calculate 200-300 GiB Dec 12 16:35:19 I switched to btrfs due to its internal volume manager, but performance is horrible Dec 12 16:36:45 but thanks for the hints; I will move to ext4 and change my environment to allow tmp to be on an extra partition Dec 12 17:24:06 rburton: even with rm_work I need >80GiB per release Dec 12 17:25:00 dv_: be carefull with trim, it can slow down removals a lot when it kicks in Dec 12 17:26:33 koen: with how many additional layers ? Dec 12 17:27:46 26 git repos with layers Dec 12 17:28:00 7 architectures Dec 12 17:28:14 and that 80GiB is not counting sstate Dec 12 20:36:35 I am trying to understand why we are using ocf-linux and not cryptodev? Dec 12 22:50:21 Does Jansa hang around in here ? Dec 12 22:52:48 otavio, I couldn't resist trying out the new "jansa/qt5-5.2.0" branch in meta-qt5, but it seems the resulting Qt5 pkg-config files have there include and library paths somewhat screwed up I think Dec 12 22:53:34 jkroon_, he does Dec 12 22:53:42 wait martin jamsa? Dec 12 22:54:13 Crofton, Martin Jansa, one of the meta-qt5 layer maintainers Dec 12 22:54:29 i think he was poking fun at the typo :) Dec 12 22:54:41 oh, no, i just can't read Dec 12 22:54:43 * kergoth rolls eyes Dec 12 22:55:06 aha, "JaMa" was his nick, now I remember Dec 12 22:55:14 that reminds me, i need to send out the patch to make th consolekit pulseaudio dep conditional, for him Dec 12 22:55:17 indeed Dec 12 23:08:12 I wasn't sure about the typo :) **** ENDING LOGGING AT Fri Dec 13 03:00:00 2013