**** BEGIN LOGGING AT Fri Jul 26 02:59:59 2019 Jul 26 06:44:21 New news from stackoverflow: Yocto Linux installation failed Jul 26 06:49:12 problem building any nodejs module with warrior: https://pastebin.com/hpdTQbQC Jul 26 06:50:29 Dvorkin: this happens inside npm... Jul 26 06:51:40 LetoThe2nd: exactly. And I have a patch: https://pastebin.com/T1cj6PQK Jul 26 06:52:55 afaik, it is not fixed in the new nodejs-npm. Girls who is maintaining the npm package are slightly not clever too much to do it :) Jul 26 06:53:34 so it's easer to add the patch into oe/yocto then to push them to fix it Jul 26 06:54:38 well we carry a lot of patches for upstream. send a patch :) Jul 26 07:44:31 New news from stackoverflow: pathspec 'gnulib' did not match any file(s) known to git while building Linux image with Yocto Jul 26 08:07:10 hi, just to point out that bitbake build fails with gawk: ../../libgpg-error-1.35/src/mkerrcodes1.awk:84: warning: regexp escape sequence `\#' is not a known regexp operator for gawk 5.0 Jul 26 08:07:45 https://dev.gnupg.org/T4492 Jul 26 09:14:43 New news from stackoverflow: How to build yocto with complete gtk+ support || How to deal with symbolic links in Yocto recipe using cmake? || I can't get flatc installed into my yocto based SDK does anyone know when the sstate-cache takes the files from DEPLOYDIR of a recipe and put them in DEPLOY_DIR_IMAGE? Jul 26 10:10:36 for anyone poking at cmake recipes and packageconfig and magic dependencies: https://cmake.org/cmake/help/latest/variable/CMAKE_DISABLE_FIND_PACKAGE_PackageName.html Jul 26 10:30:23 hi. Is Qt5 EGLFS QPA backend supported for i.MX8M Mini platforms? I see there are some generic patches here: https://github.com/Freescale/meta-freescale/tree/master/dynamic-layers/qt5-layer/recipes-qt/qt5/qtbase Jul 26 11:06:08 JPEW: i think i got the reproducibile test down to just a build path in the kernel Jul 26 11:19:35 Is there a way to see in yocto shlibs if a library is registered there? Jul 26 11:20:33 because I add the RDEPENDS_${PN} = "recipe-name" of the lib and yocto says that "fp requires lib......so, but no providers found in" Jul 26 11:31:11 Hi guys. What is a right way to use a previous version of a recipe? F.e. I use 'thud' layer which provides the xserver v1.20.x, but this xserver gets crashed in my board.. so, I vant to try a previous xserver sersion (e.g. 1.19.6). But previous version provided by 'sumo' layer.. And, when I call bitbake, it say that a core layer 'sumo' is not compatible with my base 'thud' layer.. Jul 26 11:31:39 Hi Jul 26 11:31:43 How I can use the previous v1.19.6 xserver version for 'thud' layer? Jul 26 11:32:19 Can someone help me about yocto issue? Jul 26 11:32:41 So I'm trying to build yocto morty Jul 26 11:32:50 Can I just copy/paste its *.bb configuration from previous layer to the new layer? And to use a preferred option? Jul 26 11:32:53 I'm running arch linux rolling Jul 26 11:33:16 and I got error something like this "Please port gnulib fseeko.c to your platform! Look at the code in fseeko.c, then report this to bug-gnulib." Jul 26 11:33:47 I know yocto doesn't officially support arch. But can someone help me with this? Jul 26 11:34:08 kuzulis: AFAIK That should be the way to go. Jul 26 11:34:36 donkeysnore: https://yoctoproject.org/docs/current/ref-manual/ref-manual.html#detailed-supported-distros Jul 26 11:34:56 alessioigor: Ok, thx, I'll try Jul 26 11:36:22 jofr: I know yocto doesn't support linux but is there any way to run yocto properly? Jul 26 11:36:41 jofr: Maybe running official docker ? Jul 26 11:36:42 donkeysnore: I would suggest running bitbake in a docker image Jul 26 11:37:08 bjobjo: Any official support about docker image? Jul 26 11:37:14 https://hub.docker.com/u/crops You can have a look at the images here Jul 26 11:37:41 and you can check the manual for CROPS (although exactly which one describes this escapes me at the moment) Jul 26 11:39:01 https://www.yoctoproject.org/docs/2.7.1/dev-manual/dev-manual.html#setting-up-to-use-crops Jul 26 11:39:23 (although it won't tell you much beyond a pacman -S docker to be hones) Jul 26 11:39:26 *honest Jul 26 11:39:57 bjobjo: whoops. I didn't realize that it is specified clearly in the docs. Thank you so much Jul 26 11:41:16 NP, I hope it works out for you Jul 26 12:41:06 RP or rburton -- how would you recommend someone setup the autobuilder to support multiple yocto releases? I'm not seeing a way to override the layer branch/tag per-build with the -helper json. Jul 26 13:16:13 rburton: That's awesome! Is that with core-image-minimal? Jul 26 15:09:58 * kergoth yawns Jul 26 15:52:58 tgoodwin: we use multiple branches in -helper. So same autobuilder2 repo but different branches in the helper Jul 26 15:56:46 RP: Is that somewhere other than the default config.json? Jul 26 16:04:08 The only references I find are over in the release selector section of the autobuilder schedulers.py. Other than that I haven't seen the way to select the release in the build template or overrides. Jul 26 16:13:46 tgoodwin: Its done in the force scheduler in the autobuilder UI when you trigger a buil Jul 26 16:15:37 zeddii: there? Jul 26 16:16:05 RP: I see. So how would a tagged release work (e.g., 2.6.2)? Jul 26 16:21:31 What I mean to say is, how would you select a tagged release of Poky for a build? Jul 26 16:23:17 tgoodwin: put the tag name in the revision field? Jul 26 16:24:33 I don't have it up and running so I can't see the UI except for what's exposed on YP's main one. I've been focused more on trimming it down to only builds and workers that we'll need. The more I dig the less this seems like a good idea. Jul 26 16:24:44 (the trimming, I mean) Jul 26 16:56:11 tgoodwin: trimming should be reasonable. There is probably a case for some simpler default examples of it too :/ Jul 26 16:56:42 tgoodwin: Note that you can start a controller with "everything" quite safely which would let you experiment Jul 26 16:57:06 RP: I have it up right now and wandering around but I'm not entirely sure for what I'm looking. Jul 26 16:57:14 lol Jul 26 16:58:47 tgoodwin: you usually want the console view (or I do anyway) Jul 26 16:59:34 tgoodwin: to start a build, find a target like a-full, then the start button top right Jul 26 16:59:50 tgoodwin: our config requires you to be logged in to start builds Jul 26 16:59:54 Alright, I see what you're talking about in Builders -> pick one -> etc. The top panel seems to get a bit confused on the web ui. Jul 26 17:00:25 RP: right, ultimately I would like to do that too however we use LDAP and trying to get that working with buildbot was a bit of a challenge. Jul 26 17:00:30 (nginx frontend) Jul 26 17:00:44 tgoodwin: we just did something simple Jul 26 17:10:23 RP: Can't say I blame you after my experience. I've setup my worker's password to "pass" like it seems to be the default in the configuration and then added its name to the workers list. When I start the worker, I see unauthorized login, check name and password. But at the UI, I see the worker with a "healthy" state. Jul 26 17:11:19 Hmm... never mind, it's saying unknown user at the master. Jul 26 17:12:09 zeddii: would appreciate a glance at https://bugzilla.yoctoproject.org/show_bug.cgi?id=13458 if you're feeling kbuildy today Jul 26 17:12:09 Bug 13458: normal, Undecided, ---, bruce.ashfield, NEW , Kernel embeds build paths Jul 26 17:13:20 JPEW RP: I think ross/reprod is 100% reproducible for core-image-minimal now (albeit without the improved test) Jul 26 17:13:33 (the systemd and kernel changes are not mergable yet) Jul 26 17:30:35 RP: Are worker names cached somewhere? I updated my config.py and restarted but old worker names are still there. Jul 26 18:16:26 rburton_: Cool. I'll take a look. The 2 clean builds option is really to make builds more reproducible in the face of rebuilds, so if you're reproducible without it you should be reproducible with it Jul 26 19:16:23 New news from stackoverflow: How to use perf or any profiling tool for finding the cpu-cycle or amount of cpu utilized by the functions of my application or program Jul 26 19:22:57 RP: thanks for your help today. I'm still chipping at it and getting close. Jul 26 20:27:06 JPEW: 2019-07-26 21:24:59,883 - oe-selftest - INFO - RESULTS - reproducible.ReproducibleTests.test_reproducible_builds: PASSED (1948.85s) Jul 26 20:30:15 rburton_: Yay! Jul 26 20:34:11 just two hacks Jul 26 20:34:26 and that's with the qa test, so same-host Jul 26 20:38:12 Ya, I think that the QA test needs a little more work to be *really* reproducible.... it should use faketime-esq to set the clock ahead 1 year, 2 days, 1 hour, 1 minute and 1 second into the future, and also fakehost(?) name to change the hostname Jul 26 20:38:51 I tried faketime, but it didn't play well with pseudo Jul 26 20:47:39 i'd say its pretty damn close though Jul 26 20:47:49 hostname would be an interesting one to just add as a qa test Jul 26 20:48:00 like buildpaths but hostname Jul 26 21:28:05 rburton_: nice! :) Jul 26 21:28:19 tgoodwin: old workers will be in buildbots DB **** ENDING LOGGING AT Sat Jul 27 02:59:57 2019