**** BEGIN LOGGING AT Wed Feb 11 02:59:58 2015 Feb 11 07:48:17 I am having troubles with my device tree file not being copied to my boot partition. Feb 11 07:49:22 When I enter the folder deploy/glibc/image/machine and run create_image.sdcard from there, the boot.img file produced contains the device tree. Feb 11 07:49:50 Where is create_image.sdcard ran from? Feb 11 07:50:55 Also, this only happens on my dev machine. My buildserver seems to produce a boot partition image that does contain the device tree Feb 11 08:58:07 If I add the following line to image.py:generate_image(arg) -> mypwd = subprocess.check_output('pwd', shell=True, stderr=subprocess.STDOUT) => the logs show that pwd was equal to the OE_BASE folder from where you execute bitbake. create_image.sdcard contains script that copies the device tree which does not contain absolute paths. Feb 11 08:58:13 I suspect that is a bug Feb 11 09:07:07 http://hastebin.com/igefazayik.bash Feb 11 09:22:12 patch submitted :) Feb 11 09:25:37 morning all Feb 11 09:33:13 hi bluelightning Feb 11 09:33:29 hi woglinde Feb 11 09:48:11 isn't ${AUTOREV} supposed to always fetch the latest version on a repository ? Feb 11 09:48:35 abelloni: if you also ensure the PV value contains SRCPV it will, yes Feb 11 09:48:53 e.g. PV = "1.2+git${SRCPV}" Feb 11 09:49:04 we really ought to document that... Feb 11 09:49:30 it is inheriting kernel so it odes Feb 11 09:49:33 does Feb 11 09:50:12 then assuming you have SRCREV = "${AUTOREV}" it should work... Feb 11 09:51:09 SRCREV = "${AUTOREV}" is what I have Feb 11 09:51:28 I just pushed a new commit and it didn't get fetched Feb 11 09:52:15 I don't even go through srcrev_internal_helper Feb 11 09:53:13 sigh Feb 11 09:53:20 can you please file a bug? Feb 11 09:56:38 it is indeed working for a clean build Feb 11 09:57:13 but I remember it being working even when the recipe has been build once Feb 11 10:01:24 after a cleansstate, it fetches the correct revision Feb 11 10:01:38 yeah, well that's not how it is supposed to work Feb 11 10:02:05 I wonder if you have to be using the PR service - do you have that enabled? Feb 11 10:08:20 don't Feb 11 10:08:23 I don't Feb 11 10:53:10 http://pastebin.com/PwAjquem - Support Engineer in Dusseldorf. OE/Yocto knowledge required Feb 11 10:53:33 if interested then ping me for details (I got recruiter mail) Feb 11 11:02:11 * Crofton|work needs to send in th epyzmq recipe today Feb 11 14:35:35 abelloni: are you using SRCPV in PV? if not, SRCREV won't result in signature changes, afaik. iirc some folks have considered adding SRCREV to the vardeps of do_fetch to fix that, but I don't think its been done yet Feb 11 14:37:32 kergoth: as said, I inherit kernel, so yes Feb 11 14:37:52 that doesnt' make any sense Feb 11 14:38:04 kernel doesn't magically change your PV to include ${SRCPV} Feb 11 14:39:35 oh crap Feb 11 14:39:45 you are absolutely right Feb 11 14:40:29 I was assuming it did Feb 11 14:40:39 not all kernels use SCM repositories Feb 11 14:40:41 plenty use tarballs Feb 11 14:40:45 so that wouldn't really be ideal Feb 11 14:42:54 yeah actually I have another project were it was done elsewhere, I got confused Feb 11 14:45:07 I'll try that, you can disregard my bug report... Feb 11 14:45:21 heh, mixing up projects and tasks, been there, done that :) Feb 11 14:45:46 i do think we should think about adding SRCREV to the do_fetch vardeps so tasks get rerun even if the version doesn't change as a result, but we'd probably only want to do so if there are scm urls in SRC_URI Feb 11 18:24:18 is the autotools class updating rdepends ? Feb 11 18:24:46 I have Feb 11 18:24:47 WARNING: QA Issue: obp-utils rdepends on libusb1, but it isn't a build dependency? [build-deps] Feb 11 18:25:03 but my RDEPENDS are "" Feb 11 18:25:44 the packaging process injects automatic dependencies based on the contents of the library (soname) usage (and provide) Feb 11 18:26:10 this is saying that odp-utils has something in it that is depending on something from 'libusb1'. But whatever provides libusb1 is NOT int he recipes 'DEPENDS' Feb 11 18:26:33 the problem with this is that if you build in a different order you could either get a failure -- or a different output.. causing non-deterministic builds Feb 11 18:27:54 yeah, I just wasn't sure were the rdep was coming from **** ENDING LOGGING AT Thu Feb 12 02:59:58 2015