**** BEGIN LOGGING AT Thu Apr 14 02:59:58 2016 Apr 14 10:01:05 * dv_ has gstreamer 1.8.0 recipe upgrades ready Apr 14 10:20:35 kergoth: how do you include layers 'on-demand' ? Apr 14 13:13:09 I implemented some suggestions from kergoth in yocto for modifying my kernel build so I can get a custom config Apr 14 13:13:32 what he suggested seems very on point and makes sense, but it fails to work Apr 14 13:13:47 I've got a rolling gist of this problem here Apr 14 13:13:51 https://gist.github.com/netskink/7700c036eec034281bab009a7f68bc33 Apr 14 13:14:23 any suggestions are much appreciated. This is a kernel.org kernel froom what I can tell and not a yocto kernel Apr 14 13:15:01 consquently my understanding is that kernel fragments will not work in this case. I tried that prevoiously and it also failed Apr 14 21:00:01 Crofton|work: am i reading your comments right and #9009 can be closed? Apr 14 21:00:16 hang on Apr 14 21:01:07 well, I suggest looking at that file on the device, lost of other host contaminatino Apr 14 21:01:18 but I think we fixed the problem I noticed Apr 14 21:03:20 I have a recipe for pyephem I need to submit :) Apr 14 21:03:35 :) Apr 14 21:05:02 Crofton|work: i'm about to go , can you close the bug? Apr 14 21:05:18 * rburton wrote a tool to mine git log looking for mentions of bugs that are not closed Apr 14 21:05:23 i am going to *own* the bug stats this week Apr 14 21:07:27 I did, with a reminder to watch the file on the machine Apr 14 21:09:49 Of course, I am working on a report for bluelightning now :) Apr 14 21:10:00 must practice conservation of open bugs Apr 14 21:33:23 hrm, is qemu support for an architecture mandatory to build core images ? Apr 14 21:33:37 I used to build nios2 images without qemu support for nios2, but now I no longer can do that Apr 14 21:33:45 maybe I should add nios2 support into qemu Apr 14 21:33:46 meh Apr 14 21:34:21 hmm, not sure. afaik it should be doable, though a lot would be postponed to first boot, but maybe not, possibly some recipe is unbuildable without it. what's failing, out of curiosity? Apr 14 21:35:02 kergoth: I will now do a fresh bitbake -k universe and will let you know Apr 14 21:35:19 give me a few hours , I am also rebootstrapping nios2 debian port now :) Apr 14 21:35:27 you shouldn't have to build the universe to build an image, though :) Apr 14 21:35:31 ah, cools Apr 14 21:35:33 er, cool Apr 14 21:35:41 kergoth: my computer is too slow :'-( Apr 14 21:36:32 i hear you, i'm doing oe-core world builds to test something, I swear nearly every pull touches *some* core piece and results in rebuilding the damn thing Apr 14 21:36:34 heh Apr 14 21:36:38 * kergoth taps foot Apr 14 21:38:07 * Marex should seek a position at altera (now part of intel) for unlimited build power ... they have a license for printing chips, afterall Apr 14 21:40:02 kergoth: I try to do a full rebuild to detect what broke at bulk on this nios2 arch, then I can start fixing it one after the other ... and repeat :) Apr 14 23:14:57 kergoth: gobject-introspection surely fails because of missing qemu-nios2 , but that's something I'll deal with once I get around to forward-porting the nios2 qemu patches Apr 14 23:16:48 Marex: add gobject-introspection-data to MACHINE_FEATURES_BACKFILL_CONSIDERED to get past it for now Apr 14 23:19:10 kergoth: it's not a blocker , so it's fine Apr 14 23:19:23 kergoth: but I didn't know you can do that, thanks :) Apr 14 23:20:44 oh lovely, libdrm triggers an ICE on nios2 , yay Apr 14 23:24:13 uh, why the heck would linux-yocto blow on taskhash mismatch on do_deploy ? This is becoming real weird Apr 14 23:26:13 ha, DATETIME in kernel-fitimage.bbclass Apr 14 23:26:22 usually there's a mismatch when the value of a variable changes globally vs in worker context, which.. yeah, it's usually a timestamp Apr 14 23:27:01 * kergoth thinks maybe we should think about passing DATE/TIME down to the worker as a special case Apr 14 23:29:56 * Marex wonders why does the fitImage filename even use DATETIME Apr 14 23:30:35 maybe the most stupid fix would be do_deploy[vardepsexclude] = "DATETIME" in kernel-fitimage.bbclass ? Apr 14 23:32:43 that'd do the job, yeah Apr 14 23:33:13 kergoth: does it sound correct to you ? Apr 14 23:35:42 looking around, yeah, it does make sense in my mind, so I will test it and ev. submit a patch tomorrow Apr 14 23:47:11 bluelightning, https://bugzilla.yoctoproject.org/show_bug.cgi?id=9469 Apr 14 23:48:39 Crofton|work: thx Apr 14 23:49:11 Ross made me close one bug, so I ahd to open anotehr Apr 14 23:49:24 hah Apr 14 23:49:28 sorry it took a while, had to visit my mother and was also filing toaster bugs Apr 14 23:49:32 np Apr 14 23:49:55 at least Fedora 23 alone isn't the issue, my machine here is running it Apr 14 23:50:08 someone needs to explain to peopl ethat many of us regard it as weird that you would checkout bitbake inside the oe-core directory Apr 14 23:50:15 weird Apr 14 23:50:26 the perl modules are not in the directopry it is using Apr 14 23:51:25 Crofton|work: is the location of bitbake material to this bug though? Apr 14 23:51:58 dunno Apr 14 23:52:14 running this from the collection of layers I am developing from Apr 14 23:52:49 it's unlikely to be related to those given that the check itself is in sanity.bbclass Apr 14 23:53:13 Hmm, was thinking about adding a sub-command to recipetool to gather packageconfigs from recipes ala list-packageconfig-flags.py, but i'm not sure it's in scope for the tool, thoughts? might be getting outside that scope Apr 14 23:54:16 kergoth: I'm also torn to be honest Apr 14 23:54:49 I did add a configure-help option to devtool but that's a little different since it needs the actual configure script whereas this would only need the metadata Apr 14 23:57:15 Crofton|work: here's a thought - if you set SDK_EXT_TYPE = "minimal" would you then be able to share the SDK installer with me to try here? (should be significantly smaller due to excluding sstate) Apr 14 23:57:38 would depend on whether it's got proprietary bits in it I guess Apr 14 23:59:16 I do think there's value in collecting useful tools that gather info from the available recipes, similar to some of what's in bb now, but i'm not sure recipetool is the right place for it. on the other hand, i also can't think of a better place for them, other than possibly an official bitbake command-based ui Apr 14 23:59:19 * kergoth ponders Apr 15 00:23:49 bluelightning, where do I et that? Apr 15 00:24:06 Crofton|work: local.conf would do Apr 15 02:32:52 * armpit that hurt my brain **** ENDING LOGGING AT Fri Apr 15 02:59:58 2016