**** BEGIN LOGGING AT Wed May 27 03:04:13 2020 May 27 06:26:30 smurray: ah thanks for the f32 heads up. i May 27 06:26:48 smurray: i'll have forgotten in 10 minutes, but hey, thanks for the effort :) May 27 07:52:30 RP: that was the idea, the amount of patches is so big that I don't want you to sort through the failures :) May 27 07:52:40 I messed up meson upgrade in particular somehow May 27 07:53:09 RP: but I will ask you to try again after fixing things, maybe we could make it possible for me to trigger those builds myself? May 27 07:53:57 kanavin_home: I'm fine with that, michael can set that up May 27 09:37:02 Hi May 27 09:44:16 I need some clarification on the proper way to organize layers. I f I properly understand, machine specific is isolated in it layer, reusable components (QT, applications, etc) are isolated in their own layers. Then I want ot build a product. What I would like to do is to create a dedicated layer for that product that create/customize a particular image, customize apps from other layers for my needs etc. So at the end, all layers are "generic" except my p May 27 09:44:16 roduct layer that will customize everything for my product needs. Is it the right way to go or did I misuse layers? May 27 09:45:45 ykrons: it depends a bit in how much customization your product brings/needs. May 27 09:45:46 ykrons: sounds right to me May 27 09:45:59 ykrons: but generally, it sounds pretty correct to me too. May 27 09:47:20 ykrons: things that affect the choice are for example if the hardware is highly custom, so not reusable for something else. also if you need runtime package management (avoid if possible at all!), if you have some binary compatibility restraints. May 27 09:48:18 ykrons: also, if and how you are going to automate building. May 27 09:48:26 Letothe2nd, RP: good news. There is no big customization. Mainy adding package, small patches to some applications .. May 27 09:49:19 ykrons: depending on if you have one specific thing or a "family" might also mandate a separate distro layer or not, for example May 27 09:49:32 How is it link tobuilding automation? May 27 09:50:41 ok, my understanding is that if it could be reused it must be generic and have its layer. So family probably comes in that category May 27 09:50:49 ykrons: kas for example supports "bundling" its own metadata into the final application layer. this can be very useful, or source of major annoyments. May 27 09:52:20 Letothe2nd: what did you mean by bundling? Inclusion of the build information (commit references for example) in the final image? May 27 09:53:14 ykrons: no. kas uses yaml files to define all layers and their revisions, as well as things that go into local.conf. thats what i mean. May 27 09:55:07 ykrons: there is usually one big rule though. The order of inclusion of layers shouldn't matter. And including a layer shouldn't break others (e.g. if you add a patch and it's specific to your machine or distro or whatever, make it so by using _ or instead of a simple bbappend May 27 09:57:50 Letothe2nd: I don't know kas. I'm using repo with some home made shel scripts at that time May 27 09:59:07 ykrons: :) May 27 10:03:48 Letothe2nd: go to discover it, that seems far less ugly that what I'm using! May 27 10:03:53 good May 27 10:04:50 ykrons: https://twitter.com/TheYoctoJester/status/1260293009909915653 May 27 10:05:30 the main problem is that i've got beer in the fridge but no demos prepared yet :( May 27 10:06:58 Ok that's auto promotion so! =) At least half the job is done if beers are cold May 27 10:07:53 ykrons: its promotion for yocto promotion, technically. i am not affiliated with kas :) May 27 10:10:55 Letothe2nd: hopefully I won't forget to tune in May 27 10:15:22 qschulz: as usual it will also be on youtube a little later. May 27 11:34:24 ykrons: at least i've finally fouind the motivation to actually set up the demo :) May 27 11:37:20 Letothe2nd: I'll definitely watch this one, I've been contributing a few improvements to kas lately so will be good to see your thoughts May 27 11:38:10 paulbarker: oh really? May 27 11:38:45 paulbarker: /me feels mightily proud having repeatedly poked mdp to actually try it at KG :) May 27 11:38:59 Letothe2nd: Patches currently sat on the next branch (https://github.com/siemens/kas/tree/next) but hoping they'll arrive in master soon May 27 11:39:28 paulbarker: actually i'll just show a generic directory setup, mostly generic yml for the bbb and a local overriding yml for dl/sstate May 27 11:39:46 paulbarker: any other pointers which are worth showing? May 27 11:39:48 Also got a couple of bugfixes to do and want to add new plugins `kas update` (fetch/checkout layers only, no build) and `kas foreach` (run a command in each git repo) May 27 11:40:24 Letothe2nd: A basic setup & introduction will be really helpful for people May 27 11:41:25 paulbarker: won't be much more i guess. "update" is certainly a nice feature too. will have to look into it more myself. May 27 11:41:53 paulbarker: the main objective of the session is to show people that it exists, to keep them away from wasting their time with custom scripts and/or repo :) May 27 11:41:56 Letothe2nd: I mean I've not implemented `update` yet but it's on my list for the next week or two May 27 11:42:04 Sounds good May 27 11:42:13 :) May 27 13:02:25 * ykrons not concerned by "keep them away from wasting their time with custom scripts and/or repo" May 27 13:04:42 ? May 27 13:08:59 Just because I exactly in the target I guess. I do ugly shell and repo usage because I'm not aware of the existence of kas! May 27 13:09:20 Won't be able to attend in live, but for sure I will have a look to the recording May 27 13:10:12 :) May 27 13:55:02 paulbarker: something like your description of "kas update" would make me more inclined to use it May 27 13:58:14 ah, is "kas shell" new, or did I miss it previously? May 27 13:58:44 smurray: I feel that kas as-is is great for downstream users only editing their own layer but a poor fit for us who work upstream and edit multiple layers May 27 13:58:52 `kas shell` has always been there May 27 13:59:38 if running it w/o -c just dropped you into the init-ed env, that'd be something I could see using, I'll have to try it May 27 13:59:58 paulbarker: and yes, that's kind of been my view so far May 27 14:45:02 Any way to do a shallow checkout with a yocto kernel? Notice it takes some time to do a full checkout. May 27 14:54:09 BB_GIT_SHALLOW = "1" ? May 27 16:13:36 has anyone encountered an error with the new npm class like "ailed: cache mode is 'only-if-cached' but no cached response available." ? May 27 16:14:56 I am experiencing it with a Node.js package that depends on https://www.npmjs.com/package/@nestjs/common May 27 16:16:00 Hi, now that i have systemd installed... Is there a C++ sdbus binding anyone could recommend? May 27 18:01:23 Guest5261:https://github.com/Kistler-Group/sdbus-cpp May 27 18:36:59 RP: I fixed the meson issue, so maybe a re-run? http://git.yoctoproject.org/cgit/cgit.cgi/poky-contrib/log/?h=akanavin/package-version-updates May 27 20:19:01 kanavin_home: will schedule it as soon as halstead gives me the ok... May 27 20:28:03 RP: I also got credentials from halstead, so thereafter I will try to do that myself May 27 20:28:30 kanavin_home: right, just sorted that. I ran the build as I wasn't sure if you were still around May 27 20:48:20 Exception: ModuleNotFoundError: No module named '_sysconfigdata' May 27 20:48:26 weird. anyone ever run into this? May 27 20:48:34 using python built by asdf/pyenv/python-build May 27 20:49:18 kergoth: I just added a "fix" to the head of warrior for that one in PR serv context May 27 20:49:41 what was the this? i'm hitting this trying to do a devtool modify. but regular builds are fine May 27 20:49:42 kergoth: same error, different context May 27 20:49:46 the fix, that is May 27 20:50:04 kergoth: http://git.yoctoproject.org/cgit.cgi/poky/commit/?h=warrior&id=ba1fb60b3980c86794f8a4c23e3a984f9e3590e1 May 27 20:50:18 kergoth: same patch is in master and zeus May 27 20:50:47 10 mionutes to happy hour https://zoom.us/j/99151508871 May 27 20:51:10 RP: thanks May 27 21:01:39 OE happy hour starting May 27 21:02:00 turned off the join/leave bell May 27 21:02:23 dl9pf: Is Crofton|cloud actually being happy? May 27 21:04:41 happy/sleepy hour for me - it's midnight local time :) May 27 21:13:31 * armpit passes on happy hour May 27 21:24:45 armpit, you'll regret it later when you are reviewing some craptastic patch. May 27 21:33:03 RP: already have that commit, still failing here :\ May 27 21:33:38 actually fails even without prserv enabled at all May 27 21:34:07 just import pydoc will call sysconfig.get_path() which enters that codepath May 27 21:34:22 so calling anything that imports bb from a task will explode May 27 21:34:33 cooker pulls in xmlrpc, which pulls in pydoc May 27 21:34:41 breaks devtool modify May 27 21:37:04 kergoth: right, it was more about it being a hint at the kinds of issues May 27 21:37:11 ah, fair enough May 27 21:37:26 i expect we may need to change how python3native is doing its thing May 27 21:42:55 kergoth: quite possibly. I never liked that fix and suspected there was more to it :/ May 27 22:21:39 RP: maybe something like https://github.com/openembedded/openembedded-core/compare/master...kergoth:python-sysconfig ? i'll mull it over. makes sense, though. we rename _sysconfig only for the python3-native recipe, so it'd make sense for us to only change that name for our python, not any python run from the task May 27 22:21:49 _sysconfigdata, that is May 27 22:46:36 kergoth: something like that would make sense, yes May 27 23:35:42 sakoman: I assume you watch bitbake mailing list as well, I just sent patch that should be in 3.1.1 (toaster-requirements.txt) May 27 23:36:57 * moto-timo should have CC'd you May 27 23:37:09 is it OK to add some small recipe to meta-selftest which reproduces some devtool issue even when it's not tested by oeqa (at least not yet)? May 27 23:53:12 anyone with experience adding new oeqa tests? When I add new to meta/lib/oeqa/selftest/cases/devtool.py, should it be immediately visible in "oe-selftest --list-tests" ? I don't see any registration for existing test cases, but don't see the new one **** ENDING LOGGING AT Thu May 28 03:00:05 2020