**** BEGIN LOGGING AT Wed Apr 15 02:59:57 2020 Apr 15 03:07:17 New news from stackoverflow: Need recipe for Realtek wifi dongle Apr 15 03:37:22 New news from stackoverflow: Change kernel config but defconfig already there Apr 15 07:48:05 has anyone added linux kernel checkpatch or sparse to bitbake builds? Apr 15 08:16:04 has anyone else seen the git fetcher locking up over NFS due to git alternates usage without our own lock files? Apr 15 08:24:44 RP: not locking up, but yesterday I've noticed some fetch failures with git alternates pointing to removed directory when DL_DIR was changed and the original location was removed Apr 15 08:38:42 JaMa: that is a different issue but also problematic :( Apr 15 09:55:59 Hello, is it correct place to ask about state of core-image-x11 on Raspberry Pi 4? Apr 15 10:02:17 tolszak, yes Apr 15 10:05:03 kroon: Hello, so does it actually work? I built it on zeus and master meta rpi 4 it booted but seems busybox is not installed Apr 15 10:05:59 kroon: should I look for some bug of mine or it is unknown land Apr 15 10:10:43 tolszak, if you say "it booted", I assume got was able to login to a shell ? Apr 15 10:33:27 kroon:Yes there is sh-5.0 prompt but no "cat", "ls" etc installed etc, anyways I noticed matrix group in meta-raspberrypi readme. Thanks! Apr 15 12:18:19 Hello all, after layer update I got many "Exception: ImportError: No module named 'oe.qa'" any idae what's append ? Apr 15 12:21:50 Ninic0c0: what and how did you update? Apr 15 12:23:07 qschulz openembedded / core /poky Apr 15 12:23:09 as usaul Apr 15 12:23:12 usual* Apr 15 12:26:24 Im having a build error with meta-qt on zeus does anyone know how i set the back end (im using a bbappend that worked for warrior) Apr 15 12:32:18 is there any way to keep the intermediate files created in a recipe's work directory? Apr 15 12:33:08 I'd like to inspect the files generated during the configure step, for example Apr 15 12:33:50 nameclash you can add a task to save what you want where you want Apr 15 12:36:10 ok, I thought maybe there's another way without modifying the recipe... maybe through some bitbake option Apr 15 12:37:03 nameclash: just remove rm_work from inherit, or use RM_WORK_EXCLUDE Apr 15 12:37:11 from INHERIT sorry Apr 15 12:37:40 thanks, that sounds promising Apr 15 12:38:11 rizzitello: if you don't give the error/logs and what you want to do, there's little chance someone will take the time to help you Apr 15 12:45:23 qschulz: good point Apr 15 12:49:45 It tells me no qpa platform is set and that there is no config for egl or opengl Apr 15 12:51:46 i have a single line in my bbappend of QT_CONFIG_FLAGS += " -qpa eglfs" . This was working for warrior i've updated to building with zeus Apr 15 12:53:16 i am attempting a rebuild now and when i get the full error i will provide a link to it Apr 15 13:01:32 the full error is here https://pastebin.com/KwNjhUkH Apr 15 13:03:42 So i guess my question is. What do i need todo to enable openGL es2 and eglfs for my build with zeus ? Apr 15 13:03:53 Ninic0c0: oe.qa is part of oe-core Apr 15 13:04:59 rburton Yes thx, some spaces/tab in TOPDIR andd BBPATH makes somes trouble. Fix now Thx Apr 15 13:09:37 rizzitello: have a look at the pACKAGECONFIG options in the original recipes. there is an eglfs one in qtbase for example. Also, you might need some DISTRO_FEATURES as matched in one of the PACKAGECONFIG_* variables in the same recipe Apr 15 13:09:58 ok thanks ill try that Apr 15 13:11:45 rizzitello: and you'll see for example that QT_CONFIG_FLAGS has PACKAGECONFIG_CONFARGS added to it, so I think it's a good lead :) Apr 15 13:12:04 rizzitello: if some variables are obscure, have a read in the mega-manual, this is well explained I think Apr 15 13:13:41 ok ill check there Apr 15 14:16:19 I have a Qt project that's built using CMake. The recipe inherits cmake and there are no issues building the project within the normal image build. However, we also build an SDK and in an ideal world, I would just source the SDK's environment setup script and run the cmake build outside of bitbake but that's not the case. Apr 15 14:17:23 Looking at what the cmake.bbclass does during configure, I think the key part is the toolchain.cmake file that's being generated there Apr 15 14:18:22 I made a dump of that file and now try to figure out, how to make that toolchain.cmake file available in the standalone build Apr 15 14:19:39 the hard part is the absence of a bunch of environment variables that exist in the context of the recipe Apr 15 14:20:11 you get some of them by sourcing the SDK's environment setup, but not all... Apr 15 14:21:59 now my question is, do I really need to go down that route to figure out the missing variables and write a setup script that generates me that file or is there a more elegant solution to this? Apr 15 14:22:18 I hope I was able to make my point :) Apr 15 14:23:13 nameclash: the cmake inthe SDK ships the pieces you need already Apr 15 14:23:24 export OE_CMAKE_TOOLCHAIN_FILE="$OECORE_NATIVE_SYSROOT/usr/share/cmake/OEToolchainConfig.cmake" Apr 15 14:25:34 I don't see that in my distro, what revision are you referring to? Apr 15 14:25:44 we're on rocko... Apr 15 14:26:46 its exported into the environment as part of the setup script Apr 15 14:27:47 if you're not putting nativesdk-cmake into your sdk, that's why Apr 15 14:29:49 brilliant, indeed, that's missing from the SDK Apr 15 14:30:15 rburton, thanks a lot man Apr 15 14:40:10 hello all! i'm having some trouble with systemd services not being enabled on a fresh boot of a yocto image. specifically, but not limited to systemd-networkd and systemd-resolved. i've added both to systemd's PACKAGECONFIG in a bbappend script. i've checked that SYSTEMD_AUTO_ENABLE is correctly set to 'enable' with bitbake -e. these services are Apr 15 14:40:10 listed as 'disabled' but vendor preset 'enabled'. i've also been able to manually link services with a 'do_install_append', which does make them enabled on boot, but is not ideal of course when i should just be able to inherit systemd... any tips for what I could check to discover why almost no systemd services are enabled on first boot? Apr 15 14:40:24 this is on the latest zeus btw Apr 15 15:36:26 beat_hitachi: stupid question, but you're setting SYSTEMD_AUTO_ENABLE per package right? (e.g. SYSTEMD_AUTO_ENABLE_systemd-networkd)? Apr 15 15:49:09 thank you for the response qschulz! at first, i was relying on the default enable from inherit, but I also tried per package in a bbappend file with no luck. right now i'm rebuilding from scratch after blowing away my build and sstate dirs, in case something is confused because of the move from warrior to zeus Apr 15 15:59:50 beat_hitachi: sstate isn't impacted by changes between major versions Apr 15 16:03:20 ah! i suspected as much, but felt it wouldn't hurt. good to know Apr 15 16:39:51 New news from stackoverflow: Yocto: Not able to building scikit-learn on yocto build system Apr 15 17:40:03 New news from stackoverflow: wifi driver periodically drops all connected hosts and data Apr 15 18:12:22 is there a release date for dunfell yet? :) Apr 15 18:20:31 nope. need to tidy up a few more things Apr 15 18:22:18 ok Apr 15 18:42:17 milloni: "soon", its been approved for release, we just need to sort the release notes Apr 15 19:12:10 zeddii: is warrior branch still maintained in meta-virtualization? - is it worth sending the hyperstart patch-fuzz QA issue fix there (the same one as master and zeus)? And any eta on branching dunfell in meta-virtualization (currently my last included layer still on master) Apr 15 19:21:06 I’ll definitely take the patch for warrior. I had been waiting for a few fixes to do the dunfell branch, but there’s no real reason to wait any longer. I can branch it tonight and just put fixes in master and then after I’ve done it. Apr 15 19:25:58 zeddii: thanks, warrior patch sent Apr 15 19:27:43 branch created and pushed: * [new branch] dunfell -> dunfell Apr 15 19:36:24 RP: nice! congrats Apr 15 21:00:17 gonna pull my hair out trying to get backports to behave in yocto Apr 15 21:29:53 Is there a quick way of checking what provides a certain virtual package? E.g. which libc am I using? Apr 15 21:31:33 I mean I can bitbake -f virtual/libc but that's a bit hacky heh Apr 15 21:34:57 zeddii: I just make master and dunfell track together for a while... Apr 15 21:55:57 dp: perhaps bitbake -e virtual/libc | grep ^PN= Apr 15 21:56:06 not very neat, but works Apr 15 21:56:25 or | grep ^FILE= will give you the recipe file Apr 15 22:00:06 Hmm ok thanks for the tip. I guess I can also use `oe-pkgdata find-path '/lib/libc.so.*'` Apr 15 22:00:37 bitbake -e is probably cleaner, under the assumption I don't know what files a virtual package places Apr 15 22:00:58 (err, oe-pkgdata-util, rather) Apr 15 22:05:43 right, oe-pkgdata-util will work if you want to start with the runtime context (and you've already built something to find) Apr 15 22:10:56 New news from stackoverflow: Yocto included cmake-project can't find stdio.h || How can I use yuicompressor with bitbake Apr 15 22:18:56 i can't seem to figure out how files in /etc/systemd/system/multi-user.target.wants/ are getting created :/ i can't use systemd.bbclass but thought i'd gotten the postinst all copied (and added a dependency on systemd-systemctl-native that doesn't even seem to run for things in there that also inherit the systemd bbclass) Apr 15 22:19:30 (this is specifically during do_rootfs, postinst in general seems ok, but my recipe has to defer while others do not?) Apr 15 23:17:45 read through sysemd.bbclass thats installing services either disabled/enabled based on SYSTEMD_SERVICE and SYSTEMD_AUTO_ENABLE Settings Apr 16 00:05:55 is it reasonable/sane to change how $WORKDIR is defined for a recipe? **** ENDING LOGGING AT Thu Apr 16 02:59:57 2020