**** BEGIN LOGGING AT Fri Apr 09 03:00:09 2021 Apr 09 04:25:37 in case this helps anyone else, i figured out how to get at the build dependencies: add -g to the bitbake command, which will result in a .dot file in the build directory, then using graphviz's "dot" command, convert to something viewable "dot .dot -Tpdf -o .pdf" Apr 09 04:25:59 this generates a humongous, levely dependency graph. Apr 09 04:26:06 danka yocto! Apr 09 04:26:14 lovely Apr 09 07:30:20 yo dudX Apr 09 09:30:24 o/ Apr 09 09:35:21 * LetoThe2nd feels very split. Apr 09 11:24:22 #zeromq Apr 09 13:57:23 yates: nice to know that option works... last time I tried it, 'dot' program never completed (left it running for hours...) Apr 09 14:10:35 qschulz: I finally seem to have solve my issue with gitsm fetcher... Apr 09 14:11:14 your tip of forcing the do_fetch was the key. Apr 09 14:11:22 thanks! Apr 09 14:13:23 rfs613: care to explain what was the issue so that it is part of this IRC channel archives? Apr 09 14:14:46 qschulz: the problem basically came down to the fact that the URL of a package using gitsm: changed (because sources got moved to another git server) Apr 09 14:15:17 at the same time, one of those submodules also gained a new submodule of its own Apr 09 14:17:41 evidently this confused yocto's tracking logic, so it never called the do_download, which then resulted in errors during do_unpack. Apr 09 14:18:38 your tip of doing "bitbake -c do_fetch -f recipe.bb" fixed the problem locally Apr 09 14:19:10 it was a bit more of a slog to do the same in CI but I managed eventually ;-) Apr 09 14:21:39 rfs613: mmmm, feels like a proper bug report should be opened for that? forcing a fetch shouldn't be needed Apr 09 14:25:25 qschulz: then I'd need to be able to reproduce it on demand... and having already banged my head on this for a week, I'm really not feeling up to any more... Apr 09 14:59:51 "CI"? Apr 09 15:00:23 rfs613: buy you a beer if you do! Apr 09 15:01:20 yates: Jenkins in this case... you know, a build system located on somebody else's servers, which you can't debug easily without involving a dozen 3rd parties, etc... Apr 09 15:01:33 but all the cool kids are using it so it must be good :-) Apr 09 15:01:49 yah, must be... Apr 09 15:02:38 i've [not] had the good fortune to use such things yet Apr 09 15:02:42 yates: in the end I think the error message yocto gave me was correct... it was just not making sense to me. As I had just changed the gitsm URL, I assumed the problem was with the syntax of my URL. So I started experimenting with different URLs, at one point switching back to git instead of gitsm, etc. Apr 09 15:03:08 so possibly I made the problem worse by those efforts. Apr 09 15:04:00 i find many of my problems in my technical work come down to semantic issues and/or definitions of words/phrases. Apr 09 15:04:11 the key realization was that it failed to download the repos (do_fetch) and yet tried to clone them (do_unpack). I only figured this out by adding logger.debug() prints inside the python code. Which of course was a whole level I wasn't planning on going into... Apr 09 15:04:51 correct, it didn't fail to download, rather, it did not even attempt to download. Apr 09 15:05:06 s/correct/correction/ Apr 09 15:05:25 sounds like a possible bug. would be good to bug-report it Apr 09 15:05:56 but i hear you - everything wants a piece of your time.. Apr 09 15:06:27 I hear you... but having been on the other end of such reports... I know where it will go... can I reproduce it using latest version? Apr 09 15:07:37 ooo, yeah, i hate that. do you know how many fedora bug reports i filed only to get a string of "EOL - won't fix" resolutions over the next few years..? Apr 09 15:08:26 (my home systems all use fedora - f32 now. started way back at fc3) Apr 09 15:09:19 what, not f33... ? :P Apr 09 15:09:29 talk about CI.. Apr 09 15:10:08 I stopped reporting fedora bugzilla on account of every single bug landing in that EOL pile... Apr 09 15:10:09 they need to start coming out with fractional release numbers... f32.0, f32.1, ... Apr 09 15:10:23 you too? hrm.. Apr 09 15:14:16 rfs613: are you using something older than gatesgarth on you CI system(s)? Apr 09 15:14:22 your Apr 09 15:14:32 yates: dunfell currently... Apr 09 15:31:38 how do i grab hardknott? Apr 09 15:32:49 ..or just browse it on line? Apr 09 15:33:09 i want to check of gcc has been updated to 10.3 instead of 10.2 in gatesgarth Apr 09 15:34:39 hi, is qtwebengine for mips64el supported by yocto? Apr 09 15:36:58 yates: https://git.yoctoproject.org/cgit/cgit.cgi/poky/tree/meta/recipes-devtools/gcc?h=hardknott , but I don't see 10.3 there Apr 09 15:37:57 i don't either. thanks theo Apr 09 15:38:00 i don't either. thanks theodiem Apr 09 15:43:07 is this just a matter of "haven't gotten around to it" or is there some reason 10.3 is not in the latest yocto? Apr 09 15:45:56 yates: gcc is never changed once a release has been made IIRC. You get bumps with new releases. For the release after hardknott, it'll be gcc 11 Apr 09 15:46:25 yates: c.f. https://git.yoctoproject.org/cgit/cgit.cgi/poky/commit/?h=master-next&id=8405c2ec333e15577725792f1f1ee0456d0cda9a Apr 09 16:11:04 qschulz: but gcc-11.0 isn't even released yet.. https://ftp.gnu.org/gnu/gcc Apr 09 16:26:08 yates hardknott (the next release) is still on gcc 10.2 Apr 09 16:26:21 the version after will be gcc 11 (most likely) based on current in-progress work Apr 09 16:30:42 qschulz: not true, we'd consider gcc bugfixes Apr 09 16:32:00 RP: would you consider an upgrade from 10.2 to 10.3 for example? Apr 09 16:32:26 we have in the past, but the upgrades were small, targeted and auditable.. Apr 09 16:32:36 qschulz: we have in the past, depending on what was in 10.3 Apr 09 16:33:05 qschulz: gcc point releases tend to be bugfix/stable Apr 09 16:34:13 yates: the earlier YP get started on using newer GCC, the better it is, because we can give feedback to GCC developers about potential bugs or issues before they even release. And also, it helps fixing warnings/errors in other SW and send patches to those projects to support newer GCC too. Hopefully it's not too far from reality what I'm saying but RP will correct me for sure :) Apr 09 16:34:28 RP: thx for correcting me Apr 09 16:35:07 qschulz: the general policy is to avoid upgrades, unless it is an upstream stable series with a policy we can align with, gcc tends to be relatively good at point releases Apr 09 16:36:08 gcc 11 is testing in -next so we have a chance to correct the final release if we did find issues and reported them Apr 09 16:44:44 👍 Apr 09 16:58:52 can anyone recall what fixes missing python3 raise exceptions logs from bitbake output when comparing sumo and master? spent ages looking at why some fetcher code was failing only to find out that raise FetchError("The SRCREV_FORMAT...") just isn't visible in failing task output. on master it is. Apr 09 18:05:39 note to self, old bitbake's really need 2f17b87f9712d0e218d656dae0037d15a019f9f6 to log exceptions properly Apr 09 20:05:52 where are global cert and key file usually found? in a "files/" directory at the root of the layer? Apr 09 20:18:15 fray, RP, qschulz: The reason I'm trying to do this is that I am attempting to build the toolchain for a new architecture: csky. I getting an error in the do_configure for libgcc and I suspect it may be solved with the latest, released version of gcc. Some changes for the csky architecture have been made in the past few months. Apr 09 20:18:34 i am trying to force an update to 10.3, that is Apr 09 20:18:55 I tried copying all the *10.2.* files in poky/meta/recipes-devtools/gcc to *10.3.* but it is still building for 10.2. How do i get the populate_sysroot to build for 10.3 instead? Apr 09 20:19:11 (that is, from my gatesgarth checkout) Apr 09 20:35:40 yates: have you got a poky/meta/recipes-devtools/gcc-*.inc file? For me (on dunfell) this one has PV = "9.3.0" hardcoded into it, and this int urn determine name of gcc tarball used. Apr 09 20:35:53 has anyone seen meson refusing to use the host getent ? or declaring it as non-executable ? Apr 09 20:36:47 yates: also there is a long list of patches, which probalby won't apply cleanly on a different gcc version... Apr 09 20:59:29 rfs613: good catch, but that wasn't it Apr 09 21:00:30 aha Apr 09 21:00:52 khem: thanks! just got your email Apr 09 21:01:12 yates: could be coming from another layer maybe? Apr 09 21:01:19 bump GCCVERSION as well in meta/conf/distro/include/tcmode-default.inc Apr 09 21:01:29 (from khem ^^^^) Apr 09 21:01:44 yates: sent a patch for 10.3 update as well Apr 09 21:01:50 so if you want to try it Apr 09 21:02:02 https://git.openembedded.org/openembedded-core-contrib/commit/?h=kaj/gcc11&id=f3f98b936ed9abb509d76d6f3c02f2d17fa468a7 Apr 09 21:02:24 khem: thank you! i will try it immediately Apr 09 21:03:12 k Apr 09 21:03:27 you can also try gcc-11 if you want which is in same branch :) Apr 09 21:03:53 if you are feeling adventurous Apr 09 21:04:11 baby steps... Apr 09 21:09:06 khem: are you from India? Apr 09 21:09:15 my wife is from Bangalore. Apr 09 21:10:01 she speaks Tamil, kannadu, and hindi Apr 09 21:32:41 Hi all! I would like to install a Desktop Environment so my device has a desktop with icons and a panel with clock and wifi widget. Which project is the lightest and easiest to configure? Does the need for touchscreen input reduce the set of available options? Apr 09 21:33:15 I'm talking about a Yocto-created distribution of course. Which packages shall I put inside? Apr 09 21:39:51 manuel__: XFCE would be something to try Apr 09 21:40:38 i was thinking matchbox-sato is the ticket. Apr 09 21:45:12 manuel__: see here: https://docs.yoctoproject.org/ref-manual/images.html?highlight=sato Apr 09 21:45:25 core-image-sato Apr 09 23:58:08 khem: https://autobuilder.yoctoproject.org/typhoon/#/builders/69/builds/3276/steps/18/logs/stdio - segfault in gcc11 Apr 10 02:13:30 RP: interesting, is it with the laltest I pointed you to or the one you picked from the mailing list that I sent RFT Apr 10 02:15:55 so I guess all I need is to build nativesdk-gnutls Apr 10 02:31:41 RP: I can reproduce it, seems to be a recent enough change Apr 10 02:31:52 will look into it, Apr 10 02:32:09 btw. I reshaped the go 1.16.3 patch a bit and sent v2 Apr 10 02:32:20 should be good now **** ENDING LOGGING AT Sat Apr 10 02:59:57 2021