**** BEGIN LOGGING AT Mon Apr 24 03:00:02 2017 Apr 24 13:24:36 NOTE: I'll be working to move the openembedded.org wiki to a new server today, so it will be read only for a bit. Apr 24 13:26:28 wmat: it will have read only bits?!? ;-) Apr 24 13:26:45 heh Apr 24 14:25:03 did bitbake syntax change very recently? Apr 24 14:25:04 if d.getVar('MACHINE', False) != 'vexpress-qemu' and not bb.utils.contains('DISTRO_FEATURES', 'systemd', True, False, d): Apr 24 14:28:48 Crofton|work: whats the error? Apr 24 14:29:50 http://paste.debian.net/929117/ Apr 24 14:30:04 feels like weird interatcin with mender layers Apr 24 14:30:37 Crofton|work: yer, someone had the same issue. DISTRO_FEATURES and MACHINE_FEATURES no longer apply to native recipes Apr 24 14:30:46 ah Apr 24 14:30:48 Crofton|work: looking for the irc logs of it now Apr 24 14:30:56 that sounds really familiar Apr 24 14:33:32 so mender-systemd needs to stop chacking distro features for native recipes Apr 24 14:37:21 Crofton|work: something along those lines, cant find the discussion unfornately Apr 24 14:37:34 yerah, hunting git log now :) Apr 24 14:38:02 it was a patch by jussi in the last fortnight Apr 24 14:38:23 but yes, distro features is pruned so only reflects a very limited range of your actual selection for native Apr 24 14:39:01 how can I tell is I am building native in python? Apr 24 14:39:47 731744d5538e315702be828e6f2bd556309dee07 Apr 24 14:42:17 https://www.yoctoproject.org/irc/%23yocto.2017-04-18.log.html#t2017-04-18T15:02:36 <<< found the irc log i mentioned Apr 24 14:43:12 thanks Apr 24 14:43:14 Crofton|work: a better question is why the hell does a native recipe need to know the distro features of your target Apr 24 14:43:22 sure Apr 24 14:43:38 in this case mender is constantly checking for systemd Apr 24 14:43:42 and not filtering check Apr 24 14:48:08 so question is what do I add to pythong test to stop it checking Apr 24 14:48:32 unrelated, yet anoterh email asking if I got anywhere with pyqt5 :) Apr 24 16:04:08 nrossi, so looking at the commit suggest I need to add the new vars to local.conf (or distro) Apr 24 16:43:40 Crofton|work: not sure tbh, but meta-mender shouldn't expect systemd on native recipes. Best solution would be to fix that if statement so that it checks if the current PN is native before testing for systemd Apr 24 16:44:32 Crofton|work: something like "if not d.getVar("PN").endswith("-native"):" might be the easiest way Apr 24 16:50:42 Crofton|work: or you could test for class-target, e.g. "if d.getVar("CLASSOVERRIDE") == "class-target":". Not sure what the goal of that mender bbclass is Apr 24 16:55:35 also Apr 24 16:55:55 + if bb.data.inherits_class('nativesdk', d) or bb.data.inherits_class('native', d): Apr 24 16:55:55 + return Apr 24 16:56:06 seems to work Apr 24 16:56:16 although I admit not knowing what I am doing Apr 24 18:14:05 khem: nativesdk-glibc breaks *everywhere* with the latest glibc patch sorry :( Apr 24 19:12:57 rburton: ok no worries Apr 24 19:13:27 rburton: we can try our gcc alone meanwhile Apr 24 19:14:34 yeah i'll drop the glibc patch and see what happens Apr 24 19:15:10 rburton: right. gcc and glibc both is a bit too much to bite :) Apr 24 19:15:31 :) Apr 24 19:18:39 rburton: you pointed to an error related to cfi frames is that the same across all failures ? Apr 24 19:18:49 secondly, whats gcc version on your build host Apr 24 19:19:03 if bb.data.inherits_class('nativesdk', d) Apr 24 19:19:22 rburton, does that really detect if I am doing a nativesdk recipe Apr 24 19:19:27 same for native Apr 24 19:19:48 Crofton|work: yes it does Apr 24 19:19:52 thanks Apr 24 19:20:17 I don't want to show alt fixes Apr 24 19:22:06 Crofton|work: yes thats correct Apr 24 19:22:25 need to send it in to the mender guys Apr 24 19:22:33 khem: watch http://errors.yoctoproject.org/Errors/Latest/Autobuilder/?filter=e8832205641db1249876752c30d86575f9c372b9&type=commit for gcc7 errors (no glibc) Apr 24 19:24:37 khem: yes, errors are the same across a sample of instances i picked. interestingly i think the original glibc recipe worked better Apr 24 19:32:10 rburton: yeah, there is only version update which could have impacted it so some patch upstream is causing it Apr 24 19:32:21 rburton: I am bitbaking nativesdk-glibc lets see what I get Apr 24 19:33:02 cool, thanks Apr 24 19:33:13 best walk the dogs before charlie eats my knee or something Apr 24 19:43:56 rburton: ok fixed Apr 24 19:44:22 rburton: it was my attempt to enable unwinding during configure so I reverted that Apr 24 19:44:26 now it should work fine Apr 24 19:44:36 take kraj/master from contrib once again Apr 24 20:53:49 urg, now vim fails to build Apr 24 21:19:43 oh new glibc built here Apr 24 21:22:01 anything change recently how packagegroups process depends? **** ENDING LOGGING AT Tue Apr 25 03:00:02 2017