11:26 <RP> khem: could you have a look at https://autobuilder.yoctoproject.org/typhoon/#/builders/156/builds/169/steps/12/logs/warnings please?

12:32 <dvergatal> RP: I can gladly confirm you that patches from here https://bugzilla.yoctoproject.org/show_bug.cgi?id=13419 fixes our issues but there is still an error occuring regarding do_package_setscene: package: useradd command did not succeed.

12:34 <dvergatal> RP: I was wondering does these fixes are touching do_package_setscene at all or is it rather related to the postinst-useradd-* ?

12:37 <JaMa> RP: khem: libcamera was blacklisted due to python-3.12 in https://git.openembedded.org/meta-openembedded/commit/?id=dd402176a6e28df76e4b737cec7d266b5bfbfa06 meta-raspberrypi has similar issue as dynamic-layers/multimedia-layer/recipes-multimedia/libcamera-apps/libcamera-apps_git.bb depends on libcamera as well

13:29 <RP> JaMa: we should probably add the SKIP_RECIPE there too?

13:30 <RP> dvergatal: thanks for the confirmation. Those fixes change ordering/dependencies of the tasks but not the postinsts themselves

14:39 <dvergatal> RP: n/p yeah I know that they do not touch postinst, we've been discussing it recently and agreed on that that the commit message is wrong:P I was just wondering - as mentioned earlier - if do_package_setscene is related to the postinst-useradd-* ?

14:42 <dvergatal> RP: because if it is, than this https://bugzilla.yoctoproject.org/attachment.cgi?id=4972 patch is not sufficient I think...

15:16 <RP> dvergatal: do_package_setscene can run thost postinst so yes it is "related" but I still don't know what you mean

15:17 <RP> Those patches in attachment 4972 have not merged yet as they're not ready and we don't have working test cases. Whether the patches are complete or not, I don't know

15:19 <Xogium> here's probably a dumb question, but I'm trying to picture the size of the yocto dev community in my head and quite unsure

15:20 <RP> Xogium: it depends which groups of people you class as "devs" :)

15:20 <Xogium> for reference buildroot has 3 regular maintainers, and 2 of them sometimes backports fixes to older releases. The rest of the community is mostly typical users submitting patches here and there

15:21 <RP> Xogium: If you look at the authors for git commits in any of our main repositories you'll see there are quite a few regular maintainers

15:22 <RP> Xogium: the number of layers (each with an author/maintainer) also gives some idea of scale

15:23 <Xogium> yeah commit graphs doesn't work too well with screen reader unfortunately. I wonder if there are other ways to access such stats

15:29 <RP> Xogium: https://git.yoctoproject.org/poky/stats/ might help but the forms aren't working for me to change the period/authors listing

15:30 <RP> Xogium: we should probably contact helpdesk@yoctoproject.org about that and see if we can fix it

15:32 <Xogium> huh, that's cool

15:32 <Xogium> might be good to fix it yeah

15:33 <dvergatal> RP: yeah because Jan was working on it and he has to fix

15:34 <RP> dvergatal: anyone can send or improve patches ;-)

15:34 <dvergatal> RP: and what I meant is that it is not sufficent because I'm still getting this error which I mentioned about from time to time

15:35 <dvergatal> RP: 16:34 < RP> dvergatal: anyone can send or improve patches ;-) < if someone will handle this earlier, than why not :D

15:36 <RP> dvergatal: ok, so what you're actually meaning is that you're still getting some kind of error on some set of backported patches to some previous LTS release ?

15:38 <RP> Xogium: I've asked helpdesk about it. With it being the holidays it may take a while

15:38 <Xogium> sure, no worries :)

15:38 <Xogium> I'm mostly being curious about yocto as a project and how the scale is compared to other things like buildroot or ptxdist

15:39 <Xogium> buildroot was fun, but sometimes... It just doesn't cut it, y'know. Plus I figure if I manage to put yocto knowledge on my resume it might look good

15:41 <Xogium> like for making an initramfs I had to make two totally different defconfig, one of which required to be built first, or the entire build would fail. Couldn't even build them in parallel

15:41 <dvergatal> RP: currently I'm on kikrstone and your patchset from 13419 fixes the issue for me for user add within do_populate_sysroot_setscene but we have in addition this patch from attachement 4972 which was suppoused to fix this useradd within ordering of postinst-useradd-* scripts but still from time to time an error `do_package_setscene: package: useradd command did not succeed` is occuring and it is random now

15:46 <dvergatal> RP: in addition I don't have to remove a package from sstate cache now as previously but just restart the build and that's it so that is why I'm thinking that the patch 4972 is not full fix of the issue...

17:03 <dvergatal> RP: btw. great news regarding ACLs, today has happend an update for tar 1.35 in source repositories of ubuntu so we can probably expect changes in the days to come I hope

19:12 <khem> RP: JaMa yeah sadly I should have held back on some of the python3.12 patches for things like libcamera and maybe even polkit, its trying to chew too much, anyhow I have fixed the libcamera issue and reverted the patch excluding it from parsing

19:12 <khem> I have a habit of unnessarily generating work for myself

19:18 <khem> RP: the issue with uid/gid mismatch is quite reproducible

19:18 <khem> using ipk backend and doing bitbake core-image-ptest-quilt will reproduce it as a ptest failure

19:19 <khem> the reason is that its expecting directories to have ptest gid as writable but its using a different gid in image and hence cant write to the ptest directory and test fails

19:19 <khem> this is recent regression AFAICT

19:27 <RP> khem: thanks for sorting libcamera. For the uid/gid issue we should probably open a bug. A reproducible issue would make a nice change!

19:28 <RP> dvergatal: ok, thanks, that explains things a bit more. I wonder if there is any more info in the logs anywhere about why the useradd is failing sometimes

19:28 <RP> I can believe there are more gremlins in there :/

19:29 <RP> Xogium: Yocto Project is used by quite a large variety of companies so it does work at a larger scale to some other projects. It has a lot of useful features for commercial usage but that also means it is a bit more complex too...

19:39 <dvergatal> RP: u're welcome, I dunno the only logs I've seen are in log.do_package_setscene and there is nothing only that the useradd command has failed due to wrong order

19:40 <dvergatal> RP: yeah i suppouse there are, but we are closer to fix it than before :)

19:41 <dvergatal> RP: btw. do you want me to write on bugzilla my conclusions?

20:40 <RP> dvergatal: adding the info shouldn't hurt...

20:45 <dvergatal> RP: :P

20:46 <dvergatal> RP: your comment made my laugh :D

21:29 <Xogium> RP: makes sense :)