**** BEGIN LOGGING AT Tue Feb 19 02:59:57 2019 Feb 19 05:18:06 New news from stackoverflow: internal error in override_version, at ../../gold/resolve.cc:61? Feb 19 08:10:45 kanavin: https://autobuilder.yoctoproject.org/typhoon/#/builders/85/builds/233 :( Feb 19 08:48:47 New news from stackoverflow: How to declare custom variables in bitbake recipes having the same visibility like for instance ${bindir}? Feb 19 09:40:16 Maybe a bit "odd" question Feb 19 09:40:30 is there a way to extend standard machine configuration file? Feb 19 09:40:41 like I do have qemux86-64.conf Feb 19 09:41:02 and I would like to have foox86-64.conf with only my one, two adjustments Feb 19 09:41:50 I could add those adjustments to local.conf or distro.conf, but I'm wondering if I can reuse the qemux86-64.conf (something similar to bbappend) Feb 19 09:42:03 lukma: create qemux86-64-lukma.conf that as first line includes qemu86-64.conf Feb 19 09:42:03 lukma: you can require qemux86-64.conf, set MACHINEOVERRIDES and then customise Feb 19 09:42:28 lukma: poky-blleding orwhatsitcalled does about exactly that Feb 19 09:42:29 LetoThe2nd: snap :) Feb 19 09:42:54 RP: flatpak :) Feb 19 09:43:00 Feb 19 09:43:01 :) Feb 19 09:43:13 lukma: we even do this in some of the QA tests iirc Feb 19 09:43:37 RP: btw, a penny for your thoughts: what do think of some yocto-getting started stream like on twitch.tv? Feb 19 09:44:20 RP: in contrast to a pre-boiled video, it could be quite interactive in terms of reacting to questions Feb 19 09:44:40 LetoThe2nd: would love to see it Feb 19 09:45:07 RP: MACHINEOVERRIDES is not not necessary Feb 19 09:45:22 RP: i'm just working on finding a good timeslot and rest of infrastructure... would love to give it a try Feb 19 09:45:26 As I do only need to tune PREFERRED_Xxxx Feb 19 09:45:36 lukma: it is, since you change MACHINE name and you want to use the original override Feb 19 09:45:48 LetoThe2nd: cool! :) Feb 19 09:45:58 RP: but the point is that we kind of need to advertise it a little beforehand, because if 0 people watch and ask, be end up in a pre-boiled video again. Feb 19 09:46:08 RP: yes correct Feb 19 09:46:33 RP: probably something EU late afternoon so we catch the people on the other side of the pond too? Feb 19 09:46:44 LetoThe2nd: we have the mailing lists, that sounds reasonable Feb 19 09:46:56 RP: :) Feb 19 09:47:01 ndec: any thoughts? ^^^ Feb 19 10:18:46 RP: LetoThe2nd : hi! that could be a great idea... i have attended a couple of similar 'webinar' in the past. it is not straight forward to make them 'smooth' to follow i think. we would have to think a bit hard about what we want to do. Feb 19 10:19:32 isn't twitch.tv some gaming thing? e.g. would that be appropriate for some engineering thingy like YP? maybe there are alternate places? Feb 19 10:21:19 ndec: its a gaming thingy by main audience, thats right. but it seems their technology seems to be pretty streamline, and in my experience the streamers there encourage a very, very interactive style. which i do highly favor over classic webinars. Feb 19 10:22:34 LetoThe2nd: i agree that interactivity is important (and that it was really bad in the classic webimar I attended!) Feb 19 10:23:22 ndec: if you're concerned about the YP brand being tainted, i have no problem in doing it as a standalone person, so no official endorsement has to be given Feb 19 10:24:25 well, that was indeed my concern. but the idea is good. we should evaluate about doing it officially/sponsored by YP first. Feb 19 10:24:56 i think it's worth having a discussion on yocto-advocacy@yoctoproject.org mailing list. would you start a thread there? Feb 19 10:25:13 ndec: classic webinars tend to suffer badly from "a presenter basically does a presentation". whereas the style i think of is more like, "livecoding to show" Feb 19 10:25:25 like i would do for a new coworker that looks over my shoulder Feb 19 10:28:02 RP: thanks, nativesdk packaging seems to be the major issue Feb 19 10:28:08 LetoThe2nd: anyways, i think it's a great idea. thanks for suggesting this. it needs some thinking / discussion / planning, and i think the advocacy mailing list would be the right next step. Feb 19 10:28:12 or the common one rather Feb 19 10:28:22 ndec: just subscribing Feb 19 10:28:58 LetoThe2nd: you can send an email directly, i will make sure it goes through. Feb 19 10:29:58 ndec: oh no problem. i kinda need to come up with a proper wording anyways, i usually try to make sure my emails have higher standards than my irc conversational style Feb 19 10:30:14 hehe ;) Feb 19 10:31:15 kanavin: yes, I'm having trouble figuring out everything that failed there. Am rerunning with meson changes dropped to isolate Feb 19 10:33:31 ndec, LetoThe2nd: I don't think we need to overcomplicate this. As long as its clear its about the Yocto Project, not *by* the Yocto Project I think we should be ok Feb 19 10:33:56 RP: i generallythink the same, but additional input usually doesn't hurt Feb 19 10:36:24 RP: yep, clearly this should be handled in smaller chunks. But great that this time there was a response to auh :) Feb 19 10:36:33 LetoThe2nd: I just know the advocacy people get nervous about the project branding (understandably) Feb 19 10:36:41 kanavin: yes! :) Feb 19 10:37:02 kanavin: I pulled a few of Ross' patches together since I know he's away atm Feb 19 10:42:08 RP: we definitely don't need to overcomplicate.. but we need to plan a bit at least, no? Feb 19 10:44:08 i think the main thing to grasp is that this would not be a traditional teaching/presenting format Feb 19 10:44:31 it is even something that would probably have to be repeated regularly to have the best effect Feb 19 10:48:26 ndec: right Feb 19 10:56:19 ndec: oh, and for the "twitch vs. serious" thing: its mainly us old grumpy embedded hags not being up to date with the cool kids. i thought the same, and quickly realized that its often used for such: https://github.com/bnb/awesome-developer-streams Feb 19 10:56:49 * LetoThe2nd launches lunch, bbiab Feb 19 10:57:35 LetoThe2nd: right.. speaking about cool kids.. the 96boards team is moving to discord.. and i am resisting.. ;0 Feb 19 10:58:17 ndec: discord is often used as a secondary channel for twitch indeed. i'm kinda like, whatever works. Feb 19 10:58:34 oO(yet, 96borads & cool?) Feb 19 10:58:38 ndec: I saw talk about discord somewhere else, seems to be the trendy thing Feb 19 10:58:40 * LetoThe2nd ducks and runs quickly Feb 19 10:59:36 RP: the react js folks are on discord also. like i said, whatever. if it works for a given problem, then please use it. Feb 19 10:59:41 RP: right.. and a sign that we are becoming less trendy ;) Feb 19 11:02:08 ndec: one of several :/ Feb 19 11:02:19 heh Feb 19 11:03:18 what? runit is not in yocto?? this seem wrong Feb 19 11:04:41 according to https://github.com/madscientist42/meta-runit-init Feb 19 11:05:59 Hi all. I'm trying to remove packagegroup from another packagegroup in image recipe. I'm trying to do this: RDEPENDS_mainpackagegroup-full_remove = "mainpackagegroup-application". But it does not work. mainpackagegroup-application and full are declared in mainpackagegroup recipe. PACKAGES = "${PN}-full \ ${PN}-application" and RDEPENDS_${PN}-full = "${PN}-application" and RDEPENDS_${PN}-application = "some packages" Feb 19 11:06:07 thanks for your help Feb 19 11:06:37 i don't want mainpackagegroup-application in my SDK Feb 19 11:07:22 so i want to find a way to do not build and install package from mainpackagegroup-application in SDK image recipe Feb 19 12:13:24 RP: virgl tests seem to be persistently failing on ubuntu 16.04 (but not 18.04), I am looking into it, as I have a 16.04 laptop here Feb 19 12:15:18 kanavin: was this the missing host library problem? Feb 19 12:15:37 RP: the failure happening just now printed no useful info :( Feb 19 12:15:43 kanavin: still not gotten around to auditing the workers for that :( Feb 19 12:16:20 RP: I'm just running the same oe-selftest on a laptop, we can always skip it per-distro, although I'd rather make it work Feb 19 12:16:22 kanavin: the fact this build is getting this far suggests some more subtle locale problems in the meson changes for glib Feb 19 12:17:04 (on a 16.04 laptop I mean) Feb 19 12:17:24 kanavin: right, I'd rather understand why its failing Feb 19 12:17:37 RP: yep, I fixed the obvious things, but not sure whose fault the locale problem is :-/ Feb 19 12:18:26 kanavin: right now its looking like a patch I dropped from -next which were several of the meson ones Feb 19 12:18:48 (trying to bisect to get some things merged and narrow it down) Feb 19 12:18:57 RP: as you know, we use mesa-native for builds, but fall through to host mesa at runtime via libepoxy dlopen(). I'm not super happy about this, but we can't afford to build a full range of mesa drivers :( Feb 19 12:19:20 kanavin: right, its a tricky situation Feb 19 12:20:46 RP: the default is still sdl2 without opengl, so it's not going to suddenly break Feb 19 12:21:00 you need to opt-in by adding 'gl' to runqemu Feb 19 12:25:12 hello! Feb 19 12:36:36 I have packages, and then the package that RDEPENDS on them. those packages bring some settings files, ... is there a way to build install time deps file that includes path to those setting files and installs this file as part of a package Feb 19 13:04:34 RP: ubuntu 16.04 passed the virgl tests here, so it's probably something missing on the host site in the AB Feb 19 13:51:09 kanavin: also wondering why the poor log output :( Feb 19 13:56:10 RP: I also fixed the locale issue, I erroneously dropped a patch from glib thinking it's autotools-specific Feb 19 13:58:35 RP: just sent a new patchset Feb 19 13:59:20 kanavin: ah, great to have an explanation for that :) Feb 19 14:33:39 kanavin: in your new series did you update any of the patches I still have in -next ? Feb 19 14:34:33 RP: no, I don't think so Feb 19 14:34:44 it's been fixes to glib/gdk-pixbuf/pango Feb 19 14:37:15 kanavin: thats fine, just checking Feb 19 15:42:43 what exactly is "opkg --dest install " for? when add "--add-dest foo:/foo", if installs into /foo/foo Feb 19 15:43:01 instead contrast to --offline-root Feb 19 15:43:10 can someone explain please? Feb 19 15:43:43 i see "dest" can be set in the config file, but yet i fail to see the point Feb 19 16:00:41 okay changing defaulttune didn't do shit Feb 19 16:01:26 still got the same error: https://pastebin.com/899drCMD Feb 19 16:05:21 Example resulttool reports https://pastebin.com/PC0wwQp9 and https://pastebin.com/9PgFG4pb Feb 19 16:19:40 how do I make sure recipes are excluded from the sdk? like custom built ones. Feb 19 16:30:52 kanavin: I really need you/Khem to reach an agreement on the llvm patch Feb 19 16:31:35 hello. is there a keyword I can use in a recipe to get the version of a dependency from DEPENDS ? Feb 19 16:33:20 khem, ^^^ do you have a strong objection to my llvm patch? we can also switch to a 8.0 tarball once it is out. Feb 19 16:33:40 RP: the way I read his last message on the subject, there is no strong objection, but let's confirm Feb 19 16:36:06 kanavin: I think khem would like to have the tarball as the default and a git version we can keep testing with Feb 19 16:36:27 * RP would love to see us get back to poky-bleeding working Feb 19 16:36:59 RP: I'm fine with that, I just don't want to use any kind of 'unofficial mirror' that also doesn't have version tags Feb 19 16:37:10 kanavin: yes, that is worrying Feb 19 16:38:14 the official git is larger, but not dramatically - 'unofficial mirror' has 175k commits, official git has 310k, because they decided to use a single repo for all of the llvm projects Feb 19 17:20:30 New news from stackoverflow: Disable systemD services at compile time Feb 19 17:32:00 ndec: just wanted to give a short ping. i subscribed to advocacy and opened the confirmation link, but subscription doesn't seem to be finished. Feb 19 18:04:20 LetoThe2nd: if it doesn't work ask halstead :/ Feb 19 18:04:39 halstead: mailing list woes continue? :( Feb 19 18:05:09 RP, Yes. It doesn't look especially bad right now though. Feb 19 18:06:02 RP, LetoThe2nd in this case even after confirming someone from the advocacy team needs to approve. There are 4 outstanding requests. Feb 19 18:06:35 halstead: is ndec a moderator for that list? Feb 19 18:07:06 RP, Yes ndec and Jefro. I asked Andreea if she would be willing as well. Feb 19 18:08:57 halstead: ok, good. As long as we have some active people on it Feb 19 18:10:19 halstead: after our discussion I just scheduled two builds at once :) Feb 19 18:10:37 I can agree that needs work, received a question from Rich P about that also Feb 19 18:11:04 Crofton: moderators or lists in general? Feb 19 18:11:21 approvals for advocacy list Feb 19 19:11:46 5i hereby advocate advocaat. Feb 19 19:15:08 LetoThe2nd, stop drinking Feb 19 19:16:31 Crofton: that would seriously cut down booze supplies for OEDEM and YPDD Feb 19 19:16:40 so i better keep on drinking. Feb 19 20:02:51 Hi all -- newcomer bitbake question: Is it possible to "extend" a bitbake ".conf" file in a similar way to ".bbappend" files for recipes? I'd like to define some additional configuration variables that seem like they should belong in a machine.conf (rather than my local.conf), but the machine.conf is controlled in an external repository and I'd like to just "append" to what's already there. Feb 19 20:03:43 Is there a way to append a ".conf" file or do I need to create a new machine "variant" (e.g. my_machine-extra.conf), which contains my additional variables and then includes the upstream "my_machine.conf"? Feb 19 20:17:33 jtrimbl3: there's no .conf append functionality - I think you'd have to either use your variant approach or add the customisations to your distro config (with machine overrides if necessary) Feb 19 20:18:40 jtrimbl3: if you do use the "machine variant" approach then you will probably want to set MACHINEOVERRIDES =. "mymachine" within it so that overrides for the underlying machine still apply Feb 19 20:19:01 argh sorry that should be MACHINEOVERRIDES =. "mymachine:" Feb 19 20:29:50 @bluelightning: Ok, thanks for confirming. Also thanks for the note about MACHINEOVERRIDES -- that's complicated but I think I get why it's needed. Feb 19 21:45:18 kanavin: two builds both had gtk-native issues, I suspect the meson changes so have dropped them again Feb 19 22:04:45 JPEW: fwiw I think there is going to need to be a trivial fix to meta-mingw for the master-next patchset in testing Feb 19 22:05:00 RP: Ok, thanks Feb 19 22:05:16 JPEW: https://autobuilder.yoctoproject.org/typhoon/#/builders/89/builds/306 Feb 19 22:07:27 Thats a particularly ugly bbappend... I'll have to see if we can just drop it. Feb 19 22:10:43 JPEW: that may be less trivial, not sure :) Feb 19 22:26:08 kanavin: I dont have much of an objection, I was just worries about others since in past we have seem people not like huge repo fetches Feb 19 22:26:23 as I said it helps meta-clang so I am ok Feb 19 22:38:41 Hello. What happened to the tags at: https://hub.docker.com/r/crops/poky/tags ? Feb 19 22:39:00 seems like someone pressed the wrong button :) Feb 19 22:58:56 khem: thanks for clarifying. I think if we can switch to a tarball at release it would be ok Feb 19 23:21:39 New news from stackoverflow: Yocto: : does bitbake cleanall ,cleans dependencies as well Feb 20 00:46:39 kids Feb 20 00:55:31 RP: yes although its not clear how the sources will look like on release, this will be first release based on monorepo **** ENDING LOGGING AT Wed Feb 20 02:59:57 2019