**** BEGIN LOGGING AT Mon Sep 21 02:59:59 2015 Sep 21 07:25:56 good morning Sep 21 07:59:53 mornin Sep 21 08:19:42 hi Jin^eLD Sep 21 09:30:54 morning all Sep 21 09:36:45 morning bluelightning Sep 21 09:37:21 morning nrossi Sep 21 09:41:34 hi bluelightning, nrossi, all Sep 21 10:15:41 is there a way to override meta/lib/oe/image.py in my own custom layer? Sep 21 10:25:22 Jin^eLD: OE extends the python path via BBPATH, which makes it relatively easy to add new modules, but overriding existing ones may not be easy thanks to the way python's import works, I'm not too sure of that though Sep 21 10:25:45 Jin^eLD: can you describe the issue you're trying to handle? Sep 21 11:29:17 gm Sep 21 11:29:27 What is #piggate? Sep 21 11:40:33 bluelightning: I need this https://bugzilla.yoctoproject.org/show_bug.cgi?id=8335 however I did not want to fix it in a commit in the poky submoduled but rather wanted to override it in my own layer to keep the originals clean Sep 21 11:41:25 Crofton|work: maybe this: http://www.theguardian.com/politics/2015/sep/21/piggate-hameron-twitter-react-david-cameron-pig-head-claims Sep 21 11:41:40 apologies for the almost literal spam Sep 21 11:41:52 :) Sep 21 11:42:43 Jin^eLD: hmm ok, in this case a genuine bug fix Sep 21 11:42:44 heh Sep 21 11:42:51 bluelightning: yep Sep 21 11:43:22 thanks for the info, my twitter feed had a burst of #piggate in it :) Sep 21 12:10:35 I don't even have a twitter account and somehow I still can't escape it... Sep 21 12:16:32 * Crofton|work blames pidge for my knowledge of #piggate Sep 21 12:59:47 bluelightning: so much for the right to be forgotten Sep 21 13:09:19 abelloni, the right not to know about crazy Internet memes Sep 21 13:12:19 I mean, I think everybody did something crazy at the university, maybe we can forget :) Sep 21 13:12:33 and not only on internet Sep 21 13:14:50 * Crofton|work is very very glad he was at Uni before we invented the Internet Sep 21 13:18:18 Crofton|work: LOL Sep 21 13:18:55 film was expensive, this limited the number of comical photos Sep 21 14:16:01 Crofton|work: ;D Sep 21 14:28:32 Jin^eLD: the fix attached to bug 8335 will fail if it's a dead link. it should be using os.path.lexists(), to check if a symlink exists, otherwise it's following the link and checking its destination. fyi. Sep 21 14:28:51 (this isn't the first time i've found a wrong use of exists instead of lexists in our code, actually, we need to watch that) Sep 21 14:29:31 also racy, better to just remove the thing and ignore the error if it was due to the file not existing Sep 21 14:29:33 * kergoth yawns Sep 21 14:31:29 * kergoth goes to get caffeine Sep 21 14:37:05 kergoth: hmm, yes, very good point Sep 21 15:27:10 any good examples of making init-ifupdown machine specific with custom interfaces file? Sep 21 15:27:41 Crofton|work: pretty sure all you have to do is put in in a machine-specific FILESPATH. iirc there's code that detects that and adjusts Sep 21 15:28:13 FILESEXTRAPATHS_prepend_ettus-e300 := "${THISDIR}/init-ifupdown:" Sep 21 15:28:13 PACKAGE_ARCH_ettus-e300 = "${MACHINE_ARCH}" Sep 21 15:28:16 is what I ahve Sep 21 15:28:57 i'd also put it in the machine specific filespath, to stick to existing conventions. that is, put the file physically in init-ifupdown/ettus-e300/ on disk Sep 21 15:29:03 but that should work fine Sep 21 15:29:16 yeah was thinking about that Sep 21 15:29:49 probably not *necessary* given you're manually adjusting PACKAGE_ARCH, but not a bad idea anyway Sep 21 15:30:15 stuff tends to get copied over and over Sep 21 15:30:41 I should just make the change in the base package Sep 21 15:30:55 moving from spa-driver wext to nl80211 Sep 21 15:31:02 wpa-driver Sep 21 15:31:40 Crofton|work: fyi, the bits that auto-set PACKAGE_ARCH to MACHINE_ARCH if the file:// is in a MACHINE-named path on disk is handled in base.bbclass, around line 576 Sep 21 15:31:53 so if you move it on disk, its up to you whether you want to keep the explicit PACKAGE_ARCH, it'll do it for you Sep 21 15:50:11 hmm, I'm updating from fido to master and my .sdcard image creation fails after the first time with mkfs.vfat: boot.img already exists - anyone seen this? Sep 21 16:27:56 Alan Cox has informed us as to the pig gate details via G+ Sep 21 16:46:50 kergoth, FILESEXTRAPATHS_prepend_ettus-e300 := "${THISDIR}/${MACHINE}:" Sep 21 16:46:55 is what I ended up with Sep 21 16:47:15 I assume this line is ignired if MACHINE isn't ettus-e300 Sep 21 16:47:21 that's not needed Sep 21 16:47:46 overrides are automatically searched Sep 21 16:47:55 hmm Sep 21 16:48:00 i.e. /${MACHINE} is redundant, it'll check under that path relative to the filespath dirs Sep 21 16:48:05 automaitcally Sep 21 16:48:14 but this is in a bbappend, so I need to modify the path bitbake searches Sep 21 16:48:18 just add to filespath the way you always do, but put the file one level deeper Sep 21 16:48:23 ah Sep 21 16:48:43 FILESEXTRAPATHS_prepend_ettus-e300 := "${THISDIR}/${BPN}:" or whatever, but put the file in ${BPN}/${MACHINE}/ on disk Sep 21 16:48:47 for example Sep 21 16:50:44 whould just THISDIR work I ownder Sep 21 16:53:40 * Crofton|work tries to recall the path from the times he messed it up :) Sep 21 16:54:54 should, yeah, just not the convention Sep 21 16:55:06 ok Sep 21 16:55:12 I can buy that Sep 21 16:55:21 afaik generally PN/BPN or files are the convention, but you can put it wherever you like :) Sep 21 16:55:22 so many ways, so few conventions Sep 21 16:55:30 this is the problem :) Sep 21 16:55:59 well for once there is a clear convention, so best to use it :) Sep 21 16:56:09 iirc 'files' is on the way out Sep 21 16:57:01 PN/BPN is a little redundant in some cases, but we do have cases where the recipes subdir has multiple recipes in it, so it's useful in that case, and we want to be consistent, so it makes sense Sep 21 17:08:13 thanks kergoth Sep 21 17:19:55 I've often wondered if we shouldn't just use some magic to auto-extend FILESEXTRAPATHS for bbappends automatically Sep 21 17:20:20 it could probably even be done without modifying bitbake Sep 21 17:24:04 touch foo_0.0.bbppends then? Sep 21 17:24:07 Hmm, possibly. we'd need the list of bbappends which were applied to the current recipe, not sure if that's available in non-internal variables at the moment. Sep 21 17:26:46 kergoth: it's in BBINCLUDED Sep 21 17:27:06 you'd have to filter it down to *.bbappend of course, but that's trivial Sep 21 17:27:27 ah, good point, forgot about BBINCLUDED Sep 21 17:28:15 we'd have to settle on an official default. probably /:/ or so Sep 21 17:28:30 good idea, though. should open an enhancement issue in the bts Sep 21 17:29:16 a good default would be what we apply for recipes Sep 21 17:29:27 yes, I'll do that Sep 21 17:30:01 I dunno about you, but I often come up with random ideas like that, and unless i prototype it then and there, i forget about it completely, sometimes for years :) Sep 21 17:30:41 actually, sometimes that's true even when it does get prototyped.. Sep 21 17:30:46 which is sad, but.. Sep 21 17:37:19 I tend to stick it in a todo file which just expands; occasionally I grab bits of it and file enhancements Sep 21 17:43:42 this is annoying, I need to override the ntp.conf file and everything become MACHINE specific :( Sep 21 17:43:55 I wish I'd had time to break it out Sep 21 17:47:05 OK Sep 21 17:47:21 Does anyone know if there's a good reason python-*-core includes /usr/include/pyconfig*.h ? Sep 21 17:47:42 rather than python-dev? Sep 21 17:56:16 good question **** BEGIN LOGGING AT Tue Sep 22 02:44:05 2015 **** ENDING LOGGING AT Tue Sep 22 02:59:59 2015