**** BEGIN LOGGING AT Thu Jul 09 02:59:58 2015 Jul 09 07:03:13 good morning Jul 09 08:24:54 morning all Jul 09 08:34:25 hi bluelightning Jul 09 08:40:37 hi mckoan Jul 09 10:01:07 RP: I have a fix for the alsa-plugins failure, but "error: Can't install pulseaudio-dev-6.0-r0@core2_32: no package provides libspeexdsp-dev" seems mysterious. Jul 09 10:02:06 When I wipe the tmp dir and run "bitbake speexdsp", libspeexdsp-dev gets generated just fine. Jul 09 10:02:24 hm, fun Jul 09 10:02:39 fwiw i did builds last night locally on six machines and never saw that either Jul 09 10:17:29 rburton, RP: I'll submit v2 of the speex patches with the alsa-plugins fix. If that libspeexdsp-dev continues to be a problem, I'll help if I can, but I don't have good debugging ideas of my own. Jul 09 11:17:49 tanuk: thanks, I had to admit I wondered why we only saw that one in a single build Jul 09 12:30:39 anyone know of an odroid xu3 bsp layer? Jul 09 12:41:48 oh lovely, valgrind check the kernel version of the build machine Jul 09 12:53:05 Crofton: xu3 -> https://github.com/akuster/meta-odroid or https://github.com/ARM-software/meta-mali ? Jul 09 12:53:33 thanks tlwoerner Jul 09 12:54:09 such duplication Jul 09 12:54:54 Crofton: i haven't tried either of those so ymmv Jul 09 12:55:52 Crofton: to be honest it's a bit of a mess out there; there are multiple BSPs floating everywhere in various states of completion Jul 09 12:56:01 yeah Jul 09 12:57:15 the layerindex certainly helps, but it would be great if the situation were better Jul 09 12:58:39 maybe i should put together a BoF for ELC-E? ;-) Jul 09 12:58:56 +1 Jul 09 12:59:26 or it's something we could discuss in the 2 days following the conference proper Jul 09 13:00:28 crfoton we should give badges with bronze silber and gold status for a layer and take money for it Jul 09 13:01:55 woglinde: i don't agree with the money part, but if the layerindex had some sort of feedback to indicate "does it build?" and "when was the last time someone ran this on actual hardware?" Jul 09 13:01:57 yeah Jul 09 13:02:18 interesting idea Jul 09 13:03:19 voting system maybe? Jul 09 13:04:11 i've found there are some layers (for example) which only have one branch (master) but haven't been worked on since Dora, so "master" should actually be "dora" but the README (etc) all assume it (still) works with oe's master Jul 09 13:04:16 more like we collect a list of things that annoy the crap out of us for bsp's and figure out how to test for them :) Jul 09 13:04:30 I need to go ride, finally stopped raining Jul 09 13:04:43 Crofton: ok, enjoy Jul 09 13:04:49 thanks for the pointer Jul 09 13:05:11 Crofton: I don't have an XU3 (yet?), i'd be interested in knowing how it goes Jul 09 13:06:44 yeah, the is a guy on the gnuradio list asking Jul 09 13:07:09 nrossi: a "does it build?" flag could be update from a CI loop, but it's understandable that The Yocto Project can only test the qemu MACHINEs and whatever is in the poky layer (beaglebone, edgerouter, etc) Jul 09 13:09:31 I am thinking like how easy to customize for your hardware (for bsps from people like Xilinx), does it mess with TUNE flags in ways that connat be overridden by distros Jul 09 13:09:49 does it make output that is easily installed into hw Jul 09 13:09:50 etc Jul 09 13:10:01 tlwoerner: i wonder if there is a way to validate that the recipes parse correctly and add some info to the layer index, could resolve some of the ambigious 'master' layers Jul 09 13:10:09 i think Koen submitted a proposal for a talk about OE BSPs Jul 09 13:10:33 koen: ^^^ Jul 09 13:12:54 nrossi: that's an interesting idea, i'm guessing there's some code somewhere (http://git.yoctoproject.org/cgit/cgit.cgi/layerindex-web) that parses the layers in the layerindex in order to populate the various database fields Jul 09 13:13:14 i wonder how often it runs, and whether other functions (such as this) could be added to it? Jul 09 13:29:17 tlwoerner: poke bluelightning, he's the layerindex guy Jul 09 13:38:14 koen: ok. earlier i suggested an ELC-E BoF about BSP layers which Crofton +1'd, but before getting too serious about it, i'd want to make sure it wouldn't be stepping on your toes (should your talk get accepted, which it should) Jul 09 13:38:55 (not that i have any say in what gets accepted, i'm just saying that a BSP talk from you should be accepted as a matter of course) Jul 09 13:40:35 I submitted 2 talks, the BSP one having a footnote saying "can be a bof if enough OE peeps are at elc" Jul 09 13:49:03 koen: great! Jul 09 13:51:38 is there a simple trick for getting, for example, the ext4 IMAGE_FSTYPES to include the kernel modules? Jul 09 14:00:13 tlwoerner: re the branch issue, there is in fact a bug open to try to address that: Jul 09 14:00:23 https://bugzilla.yoctoproject.org/show_bug.cgi?id=7792 Jul 09 14:00:36 it's a little bit tricky to solve but I do want to fix it Jul 09 14:01:16 I've also been thinking that maybe we should expose the ability for layer maintainers to map branches if they use a different naming scheme Jul 09 14:02:01 (the mapping capability already exists in the database and can be manipulated at the admin level) Jul 09 14:02:26 bluelightning: great! Jul 09 14:02:49 I'm open to suggestions about how to handle this btw Jul 09 14:03:22 are you planning to attend ELC-E? Jul 09 14:05:51 bluelightning: when or how often does the metadata in the layerindex get refreshed? at the time it is added, or does something get run periodically? Jul 09 14:21:08 tlwoerner: it's updated every three hours Jul 09 14:21:25 tlwoerner: I hope to be at ELC-E, yes, not confirmed yet though Jul 09 14:21:43 bluelightning: great, thanks Jul 09 14:30:06 I'm using "bitbake-whatchanged -v" to determine why do_package_write is being rerun for certain packages. I see several changes in BB_TASKDEPDATA from a very large map to None. What does this mean? I can't find a reference to this variable in the documentation. Jul 09 17:54:01 I have posted patch to runqemu again ( 4th time ) this time Jul 09 17:54:03 http://git.openembedded.org/openembedded-core-contrib/commit/?h=kraj/for-master&id=946bdac08d39152d5d25795d01bdf15b227378e2 Jul 09 17:54:17 its not applied and I dont have feedback Jul 09 17:54:30 why its not acceptable either Jul 09 17:57:49 We should be better at patch reviews and status for meta-openembedded one could check patchwork and still see the status of patch but for oe-core thats not possible so only way is mailing list and sometimes some patches from a pull request get applied some dont and there is no feedback on them as to why they are not pulled Jul 09 17:58:07 we should improve on this really Jul 09 17:59:19 providing a response on why a patch is not worth is more important so developers know what to do and they can react in time Jul 09 18:15:53 is master missing detecting when you add a patch to a kernl recipe? Jul 09 18:19:36 khem`: that is somethign itd' be nice to improve, there's no way to check status other than periodically checking the repo to see if it's on master-next or has been merged Jul 09 18:25:42 where is the kernel source in work now? Jul 09 18:26:59 hehe Jul 09 18:27:07 shared Jul 09 18:30:11 yeah I think I found it Jul 09 18:31:00 bb show -r virtual/kernel S or bitbake -e virtual/kernel | grep '^S=' Jul 09 18:31:01 :P **** ENDING LOGGING AT Fri Jul 10 02:59:58 2015