**** BEGIN LOGGING AT Fri Feb 27 02:59:58 2015 Feb 27 08:29:50 JaMa: ping Feb 27 08:30:58 pong Feb 27 08:31:56 hello Feb 27 08:32:18 please push the patch removing orphaned kernels in meta-initramfs Feb 27 08:32:31 I don't see it in master-next yet Feb 27 08:33:01 I have the following meta-handheld fixes almost ready Feb 27 08:33:33 and I'll send a pull request later today Feb 27 08:33:44 thanks in advance Feb 27 08:42:13 ok, will do Feb 27 08:55:21 good morning Feb 27 09:32:05 morning all Feb 27 09:38:57 hi bluelightning Feb 27 09:39:04 hi mckoan Feb 27 09:46:33 morning Feb 27 10:34:23 hi all Feb 27 10:34:47 i've run into an issue with busybox Feb 27 10:35:50 although the reproduction rate is thin but the issue seems to affect the rootfs that is created in a way that it does not even boot Feb 27 10:36:08 my build ran out of space and was killed by bitbake Feb 27 10:36:23 so i cleared up some space and start the build again Feb 27 10:36:57 i use the .hddimg and dd it to a usb for using on the target which is an x86_64 Feb 27 10:37:28 now when I booted through the usb key kernel panicked Feb 27 10:38:52 complaining that it could not mount rootfs Feb 27 10:39:33 now when I looked through the boot log just before the panic printing there was an output from busybox "sh: applet not found" Feb 27 10:39:58 I flattened up the initrd from this .hddimg and mounted it locally Feb 27 10:40:34 a comparison with a sane initrd revealed that busybox symlinks (alternatives) were nowhere to be seen Feb 27 10:41:21 when I looked at the cooker log I saw that busybox's do_compile was killed when my system ran out of space Feb 27 10:42:38 I tried recreating the issue by doing a cleansstate and compile for busybox while throwing a ctrl-x twice at it on do_compile every now and then but only succeeded 1 Feb 27 10:42:42 however Feb 27 10:43:16 that revealed another issue where the do_package for busybox fails if do_compile is terminated in the way I was trying Feb 27 10:44:08 bluelightning: ping, can you please give it a thought? :) Feb 27 10:45:02 I see that the .config is updated in the do_compile process as well and it would be in half cooked state if do_compile is terminated Feb 27 10:45:36 wouldn't you just not interrupt the build during do_compile and avoid the issue entirely? Feb 27 10:46:10 and as the do_compile for busybox does not cleanup any leftovers from last terminated build this would cause issues Feb 27 10:46:27 bluelightning: is there not a problem? Feb 27 10:50:28 so, it's a bug to be sure, but I don't know that we've ever guaranteed to recover from tasks interrupted in this manner Feb 27 10:50:46 bluelightning: I rather not interrupt it myself but the same can happen I mean space running out and then the interruption Feb 27 10:50:49 if a file is half-written out, make will still think the file is there Feb 27 10:51:01 bluelightning: obviously the chances are very very thin Feb 27 10:51:09 yup Feb 27 10:51:41 I've a patch, should i submit to oe-core so i guess greater audience would comment Feb 27 10:57:10 sure, certainly Feb 27 10:57:39 hello all Feb 27 11:11:09 hi mago_ Feb 27 13:58:28 JaMa, otavio: so, is anything preventing you from applying the recursion removal patch in meta-qt5 ? Feb 27 14:00:21 abelloni: it was already applied couple hours ago Feb 27 14:00:30 oh ok :) Feb 27 14:00:38 I didn't check since yesterday Feb 27 14:00:48 better to check before asking :) Feb 27 14:01:04 yesterday I was still skiing in austria :) Feb 27 14:02:10 actually, I don't see it on github Feb 27 14:02:57 was the snow ok? it is knid of a crappy winter here Feb 27 14:03:16 abelloni github is only mirror Feb 27 14:03:23 look at the main repo Feb 27 14:05:02 it is probably worth updating http://layers.openembedded.org/layerindex/branch/master/layer/meta-qt5/ Feb 27 14:07:18 hm Feb 27 14:07:23 or my bad Feb 27 14:07:38 and mixed up meta-oe and meta-qt5 Feb 27 14:08:20 * Crofton|road is missing he good skiing on the East ocast :( Feb 27 14:13:06 I'm using a bitbake version 1.8.18. Can I add something like "addtask configure_with_test after do_configure before do_compile". Is addtask supported in this version? Feb 27 14:15:44 driftingblues: I think so, but you should be able to confirm that by using "git grep" to look for other examples Feb 27 14:16:03 bluelightning: Okay. let me check Feb 27 14:20:49 abelloni: I see it as 2nd commit in master: https://github.com/meta-qt5/meta-qt5/commit/d9c741cce23ed6ff741ed67b77f57e04e5b440b8 Feb 27 14:21:04 https://github.com/meta-qt5/meta-qt5/commits/master Feb 27 14:21:14 abelloni: snow was perfect Feb 27 14:21:23 bluelightning: Something like this there in one recipe "addtask deploy before do_install after do_compile" Feb 27 14:21:27 do_deploy () { Feb 27 14:21:29 } Feb 27 14:21:54 indeed Feb 27 14:22:04 do you plan to apply it on dizzy ? Feb 27 14:22:19 bluelightning: Can I add a statment in do_deploy 'echo "hello"' and check. I wanted to know if it really calling this do_deploy? Feb 27 14:23:01 driftingblues: you can do echo but that would only appear in the task log Feb 27 14:23:27 bluelightning: What is the easy way to check? Feb 27 14:23:53 well, it'll be reporting that it's running the task when you build the recipe Feb 27 14:24:00 if it's not reporting it then it's not being run... Feb 27 14:24:15 bluelightning: Okay Feb 27 14:24:49 bluelightning: Looks like if I add in bb file it is being run. When I add in amend.inc it is not being run Feb 27 14:24:54 bluelightning: Thanks :-) Feb 27 14:25:10 abelloni: I would prefer to keep in master for a bit longer until we consider merging it to dizzy Feb 27 14:25:17 just in case something creeps in Feb 27 14:25:24 ok Feb 27 16:25:14 What does this error mean? IndexError: list index out of range Feb 27 16:25:45 I added addtask in amend.inc and when I build I get this error, Any idea> Feb 27 16:28:25 you hit a bug Feb 27 16:29:09 but that's in an ancient version of bitbake... it may have been fixed in a later version, but then you would have to be careful to find one that's still compatible with OE-Classic Feb 27 16:29:44 I'm assuming you are indeed using OE-Classic if you are using bitbake 1.8.18 Feb 27 16:32:08 heh, use of amend.inc raises a red flag, that must be before bbappend existed :) Feb 27 16:34:15 driftingblues, We highly recommend moving off oe-classic. It is now very, very obsolete Feb 27 16:38:11 bluelightning: Crofton|road: Yes I'm using OE-classic. But we need to use this version for a customer project. Feb 27 16:38:25 I'm forced to work with this version :-) Feb 27 16:38:47 driftingblues, OK, but do not say you haven't been warned Feb 27 16:39:10 Crofton|road: :-) Feb 27 16:39:50 Crofton|road: hi - btw what does "Removing the oe-core and meta-oe repos. " mean on the OEDAM agenda? Feb 27 16:39:55 * Crofton|road adds an item to OEDAM aenda, what to do about OE-obsolete repo Feb 27 16:40:03 the renamed ones Feb 27 16:40:14 oh you mean the old named github mirrors? Feb 27 16:40:19 right Feb 27 16:40:35 ah ok - if you're editing that page can you make that clarification? Feb 27 16:41:00 we're also still short a place for presentations & promo material btw Feb 27 16:41:11 kergoth: I was actually backporting some recipes from latest version to old bitbake version. Since my bitbake version does not have bbappend supported I had to rename the file to amend.inc Feb 27 16:42:24 it has some addtask in it Feb 27 17:00:24 moin Feb 27 19:36:30 apparently one of my scale talks actually made it to video Feb 27 19:36:42 * nerdboy checks for camera/ccd breakage Feb 27 20:55:03 JaMa: more no opengl patches incoming ;) Feb 27 20:56:50 more no? Feb 27 20:57:25 * nerdboy was apparently moving around too much Feb 27 20:57:31 abelloni: please chech after rebasing on top of current master-next first Feb 27 20:57:36 camera and video survived Feb 27 20:58:57 abelloni: good, both shouldn't be needed in master-next :) Feb 27 20:59:55 * nerdboy lost his parser config Feb 27 21:08:17 JaMa: yeah, this is obviously not needed for 5.4.1 Feb 28 00:19:11 * nerdboy 's work channel filling up with "kinda bummed today" "he's dead, Jim :(" "i don't feel good..." Feb 28 00:22:19 it's sad to be sure... but he lived long and,,, well, you know the rest Feb 28 00:43:30 hmm Feb 28 00:43:41 it looks like "out of tree" builds are default now? Feb 28 00:43:50 is it possible to override that back to the old behaviour for one package somehow? Feb 28 00:49:49 for autotools, sure - just inherit autotools-brokensep Feb 28 00:49:56 (instead of autotools) Feb 28 00:50:18 its not autotools... its some hand crafted crap + a makefile Feb 28 00:50:32 which I really would not want to rewrite from scratch :) Feb 28 00:51:24 well, separate build dir is *not* the default other than for autotools and cmake Feb 28 00:51:43 you can see B is still set to "${S}" in meta/conf/bitbake.conf Feb 28 00:51:43 aha Feb 28 00:52:00 ooh, I'm inheriting autotools for some reason Feb 28 00:52:03 I guess I should not Feb 28 00:52:25 that might solve it Feb 28 00:52:56 yes! thank you :) Feb 28 00:55:58 np :) Feb 28 00:58:38 :) Feb 28 00:58:40 nite **** ENDING LOGGING AT Sat Feb 28 02:59:59 2015