**** BEGIN LOGGING AT Mon Nov 28 03:00:00 2016 Nov 28 08:47:31 Morning!. I want to port the opencv python bindings to python3 Nov 28 08:47:38 http://git.openembedded.org/meta-openembedded/tree/meta-oe/recipes-support/opencv/opencv_3.1.bb Nov 28 08:48:54 the issue is that the python2 bindings will be lost Nov 28 08:49:04 would that be a problem? Nov 28 08:53:45 can you make that a PACKAGECONFIG? Nov 28 08:56:40 the issue is that python3 : inherits distutils3-base and python2: inherits distutils-base Nov 28 08:57:30 can I select inerits based on PACKAGECONFIG? Nov 28 09:02:19 you can use variable in inherit ${FOO} and you can set this variable based on python3 in PACKAGECONFIG Nov 28 09:03:23 good hack :) Nov 28 09:03:25 thanks! Nov 28 09:03:35 I will give it a try Nov 28 09:20:12 morning Nov 28 14:55:08 * kergoth yawns Nov 28 16:06:39 khem: new webkit in mut causes gcc to ICE with musl Nov 28 17:03:24 | ERROR: oe_runmake failed Nov 28 17:03:25 | ../../orc-0.4.26/orc/orccompiler.c:13:31: fatal error: valgrind/valgrind.h: No such file or directory Nov 28 17:03:25 | #include Nov 28 17:05:21 hmm, clean sstate and rebuild makes it go away .... Nov 28 17:20:46 sounds like a floating dep on valgrind, need to make it explicit Nov 28 17:23:39 there is a confgiure check Nov 28 17:24:02 checking valgrind/valgrind.h usability... no Nov 28 17:24:02 checking valgrind/valgrind.h presence... no Nov 28 17:24:03 checking for valgrind/valgrind.h... no Nov 28 17:24:08 from the successful build Nov 28 17:24:24 some form of reace I guess Nov 28 17:30:54 ugh, i hate when i'm blocked on task progress by a slow build :| Nov 28 17:31:03 task 1800 of 8900 Nov 28 17:31:06 * kergoth taps foot Nov 28 18:57:36 kergoth: what's your builder? A 128 thread/64 core box builds world in 1.5 hrs... 128 core: 1:31:31, 4091% cpu. compared to 24 core: 3:23:52, 1740% cpu. -- where world is WR Linux world. Nov 28 19:15:34 vmeson: impressive. i do my builds on an 8 core VM :) Nov 28 19:15:49 i think our automated builds are better than that, though, just not the active development VMs Nov 28 19:17:47 lol Nov 28 19:27:17 kergoth: the focus should be on the poor human... bots can context switch more easily. we run 2-4 builds in parallel to keep the system through-put maxed. :) Nov 28 19:28:38 the delay in the development/test cycle is what i hate most about doing oe development, particularly since i don't like doing much multitasking anymore — too much context switching, i lose track of what i'm doing with >2-3 tasks Nov 28 19:47:46 Crofton|work: its a dangling dep on valgrind Nov 28 19:48:04 rburton:hmm link ? Nov 28 19:48:36 how do we fix? Nov 28 19:48:45 khem: Applying patch 0001-use-lnr-wrapper-instead-of-looking-for-relative-opti.patch Nov 28 19:48:46 patching file Makefile.am Nov 28 19:48:46 Hunk #1 succeeded at 321 (offset 1 line). Nov 28 19:48:46 patching file configure.ac Nov 28 19:48:46 Hunk #1 FAILED at 108. Nov 28 19:48:46 1 out of 1 hunk FAILED -- rejects in file configure.ac Nov 28 19:54:07 Crofton|work: you need to add a packageconfig may bw Nov 28 19:54:12 rburton: still ? Nov 28 19:54:29 rburton: what machine do I need to select Nov 28 19:54:36 qemux86-64 ? Nov 28 19:54:38 yeah hunting for orc homepage :) Nov 28 19:56:03 khem: intel-corei7-64 Nov 28 19:56:28 oh feck it Nov 28 19:56:30 we've a bbappend Nov 28 19:56:32 sorry Nov 28 19:57:07 hm no still breaks with that Nov 28 19:57:17 i wonder if the bsp bbappend is triggering for qemu builds too Nov 28 20:00:25 khem: no definitely breaks for qemux86-64 here with no other layers Nov 28 20:04:45 bah, all my natives are rebuilding due to the removal of caps from NATIVELSBSTRING :P Nov 28 20:04:49 * kergoth adds some symlinks Nov 28 20:04:57 cough uninative ;) Nov 28 20:06:05 doesn't help with the oe-core/nodistro testing Nov 28 20:07:21 The differences between the current build and any cached tasks start at the following tasks: Nov 28 20:07:21 /scratch/yocto/examples-sanity/oe-core/meta/recipes-devtools/quilt/quilt-native_0.64.bb:do_configure Nov 28 20:07:25 oh that bodes well Nov 28 20:07:30 hrmph Nov 28 20:08:22 hades Nov 28 20:08:33 * kergoth gives up and switches to uninative for his nodistro builds too Nov 28 20:09:09 can't disable valgrind in orc Nov 28 20:09:34 not even with autoconf cache seeing? Nov 28 20:09:46 ac_cv_HAS_VALGRIND=false or whatever the variable it uses is Nov 28 20:10:28 I am just reading configure script Nov 28 20:10:55 if it uses AC_CHECK_* and doesn't abort if it isn't found then you can pre-seed the check to force it Nov 28 20:12:07 http://pastebin.com/uzrF1k2f Nov 28 20:12:30 ah Nov 28 20:12:31 better Nov 28 20:12:43 AC_CHECK_HEADERS([valgrind/valgrind.h]) Nov 28 20:12:56 no abort Nov 28 20:13:24 there you go Nov 28 20:13:30 * Crofton|work makes a note to llok after finsdihing what he started to do :) Nov 28 20:13:38 ac_cv_header_valgrind_valgrind_h=no is the magic you want in your PACKAGECONFIG[valgrind] Nov 28 20:26:59 or patch it with an argument, could potentially go upstream, assuming it's maintained Nov 28 20:30:02 kergoth, also an idea Nov 28 21:27:33 rburton: OK I can reproduce it now. I think we have some patches that apply conditionally to systemd but not to systemd-boot which is causing the same patch to be not applicable in both places Nov 28 21:27:51 rburton: I have sent a v5 hopefully builders will like it this time Nov 28 21:35:38 thanks khem **** ENDING LOGGING AT Tue Nov 29 03:00:00 2016