**** BEGIN LOGGING AT Wed Oct 23 02:59:58 2019 Oct 23 03:04:55 hey, i'm absolutely clueless about yocto and related tools, and i'm trying to push through to get something working, which isn't a great combo. Oct 23 03:05:39 I'm trying to package up python3-pyqtwebengine, looking at a recipe from meta-qt5 for python3-pyqt5 Oct 23 03:06:03 it builds just fine, finally.. but no files end up in the packages. the base package isn't created, and the -dev and -dbg packages are empty Oct 23 03:06:59 my RDEPENDS_${PN} is the same as the pyqt5 reference, and those pathspecs are places the install target seems to be installing to Oct 23 03:07:29 i'm kinda running out of places to look, since i don't really understand this well at all.. any guidance would be really appreciated Oct 23 03:11:39 (it's also saying 1 warning message was shown, but i don't see any warning messages :P Oct 23 04:13:27 icee: could you paste the recipe files somewhere? Oct 23 04:15:24 Also, after building, do you have any files installed in tmp/work//python3-pyqtwebengine//image/ ? Oct 23 04:30:59 ok, image is empty, that's a hint Oct 23 04:31:02 one sec for paste Oct 23 04:32:28 erbo: https://gist.github.com/mlyle/c890f2405eaefb63c6b9051c5b995106 Oct 23 04:32:42 sorry it's a bit of a mess, i've been flailing Oct 23 04:32:58 doesn't help that the qt stuff takes hours to build for each real iteration Oct 23 04:35:09 so, the files are in the -sysimage- Oct 23 04:35:22 is maybe the really evil build script for the pyqtwebengine awry? are they supposed to show up in image? Oct 23 04:39:24 icee: usually a recipe installs everything into image/, and that's where the packaging parts looks for files to put into the different packages generated by a recipe Oct 23 04:42:06 omg i have some stuff in the package Oct 23 04:42:09 not enough stuff Oct 23 04:42:35 thank you, erbo. now just to figure out how the heck this python+custom crap+qmake abomination chooses where to put stuff Oct 23 04:43:12 Good luck! I don't have much experience with PyQt so I won't be of any helt there :) Oct 23 04:46:18 it's evil :( but thanks, that was a very good hint to peek in image Oct 23 04:46:44 they have bifuricated where dependency sip / python / library files vs. where they go in the configuration script, but there's no consistency to the naming is the real problem Oct 23 04:46:58 so it'll be pretty much brute force to figure it out Oct 23 05:15:54 New news from stackoverflow: How do I write a yocto/bitbake recipe to copy a directory to the targe root file system Oct 23 06:46:07 New news from stackoverflow: yocto crops container's PATH variable cannot be modified Oct 23 06:47:36 good morning Oct 23 06:48:47 good morning Oct 23 06:49:07 depends :P Oct 23 07:00:46 LetoThe2nd: By your dishwasher? :-) Oct 23 07:02:53 alessioigor: at least that ones fine again. Oct 23 09:16:30 New news from stackoverflow: in Qt Webgl mode camera is not streaming correctly Oct 23 09:46:34 New news from stackoverflow: Where are bitbake python functions documented Oct 23 09:58:43 hello Oct 23 10:16:38 New news from stackoverflow: Yocto: _append python code to a shell task Oct 23 10:32:49 how can we remove an inherited class in *.bbappend? Oct 23 10:34:10 xtron: you cannot. Oct 23 10:34:48 (at least i'm pretty sure you can't :D ) Oct 23 10:36:30 LetoThe2nd, so if we want to modify a function/method (defined in a bbclass) we can re-define it in another *.bbclasss then inherit it in a *.bbappend in a layer with higher priority, right? Oct 23 10:36:54 xtron, overlay it in a higher priority layer or selectively _remove things which that layer added in bbappend Oct 23 10:37:10 or mark functions which that layer added as [noexec]. **** BEGIN LOGGING AT Wed Oct 23 12:06:06 2019 Oct 23 12:07:14 JaMa: let me explain. base-files is a DEPENDS in useradd.bbclass. So any recipe which inherits useradd will be "poisoned" by base-files machine-specific recipe, retriggering a full build of said package (e.g. dbus) when machine is changed. Though said package workdir is not by default in a machine-specific workdir Oct 23 12:07:47 JaMa: send it where? Oct 23 12:07:48 the issue is that you then have every package depending on e.g. dbus (e.g. qt) being rebuilt for each machine Oct 23 12:10:44 sven^: where README* says Oct 23 12:12:24 JaMa, "Fatal error: can't write 49 bytes to section .text of .obj/release/qopenglversionfunctionsfactory.o: 'file too big'" oO Oct 23 12:12:43 qschulz: I'm not seeing this behavior and sstate-diff-machines --analyze also doesn't report this Oct 23 12:13:07 "as: .obj/release/qopenglversionfunctionsfactory.o: too many sections (60193)". too bad :( probably has -ffunction-sections or something activated Oct 23 12:13:14 qschulz: because base-files is in SIGGEN_EXCLUDERECIPES_ABISAFE Oct 23 12:13:31 JaMa, hope you don't mind my issue reports on github Oct 23 12:14:33 litb: I definitely don't mind, but I never used meta-mingw, so patches are welcome, but I won't be able to spend any time on that any time soon Oct 23 12:14:52 sure, thanks Oct 23 12:14:57 JaMa: uh, I guess I won't do that right now. I don't want to subscribe to any mailing lists. Sorry Oct 23 12:15:27 sven^: no need to subscribe. just send the mail, you'll get a notice that a mod will look at it, and thats it. Oct 23 12:15:52 http://www.openembedded.org/wiki/How_to_submit_a_patch_to_OpenEmbedded says I need to subscribe to be able to send a patch Oct 23 12:17:26 sven^, without subscribing to amiling lists, how do you track the direction of the project? Oct 23 12:17:44 luz Oct 23 12:17:47 sven^: surprising. if you don't mind just give sending a try. the worst that can happen is that you get a reject message. and in that case, well, so be it. Oct 23 12:18:02 Crofton|work: how does that matter when I want to help by improving the project's code? Oct 23 12:18:10 Crofton|work: JaMa: don't care folks away :P Oct 23 12:18:13 LetoThe2nd: will do. Oct 23 12:18:17 sven^: thanks. Oct 23 12:18:30 *scare, even Oct 23 12:18:59 rburton: btrfs tools require a mounted device and cannot manipulate disk image files directly. the only way i found for now is to mount the image as loop device, but that requires root permissions Oct 23 12:22:50 JaMa: where is that SIGGEN_EXCLUDERECIPES_ABISAFE set for base-files? Oct 23 12:23:14 JaMa: also, previouslyk running krogoth, now thud so the "bug report" I got might not be based on thud yet Oct 23 12:24:20 JaMa: answering my question: in conf/layer.conf Oct 23 12:24:46 ugh. The git version on my development vm is too old to have git send-email -.- Oct 23 12:26:50 sven^: You may just be missing a dependency. `git send-email` has existed pretty much as long as git itself has Oct 23 12:27:38 Look up your distro on https://git-send-email.io and see what it says Oct 23 12:28:16 paulbarker: yes, thanks, got it Oct 23 12:28:45 sven^: lets make a deal. you dump the patch into a pastebin this one time, and we promise to look at it. on the other hand, if you want to submit your second patch, you promise to jump through the hoops of getting your email and ML setup sorted out, mkay? Oct 23 12:29:20 I only need to set up my smtp server in git and it should work Oct 23 12:29:45 was just an offer :) Oct 23 12:33:45 I got the cc. So I guess it went through Oct 23 12:57:59 hey guys, is there any way to create empty directories on yocto recipes? I'm trying to mkdir "${D}/mydir" on do_install(), but it seems like those don't make it to the final image Oct 23 12:59:49 Klanticus: did you add it in some package with FILES variable? Oct 23 13:01:11 JaMa: yes, i did. I have "install -d "${D}/run/mydir" on do_install(), and then "/run/mydir" on FILES_${PN} Oct 23 13:02:12 sorry, mkdir instead of install -d Oct 23 13:02:22 actually I tried both Oct 23 13:03:40 maybe if the package is otherwise empty, something might go wrong Oct 23 13:04:04 I guess the issue is you've got `/run` in that path. That's not persistent Oct 23 13:04:19 paulbarker: good spot Oct 23 13:05:10 hmm.. that may be a problem Oct 23 13:05:53 any way to workaround that? Init script to create the directories on every boot? Oct 23 13:05:58 `/run` is populated at boot time. There is support for adding things but I can't remember the details Oct 23 13:06:13 Search for something like `populate-volatiles` Oct 23 13:06:23 paulbarker: will do. Thanks Oct 23 13:06:48 I think you can add entries to the list of volatile directories to create at boot Oct 23 14:17:20 New news from stackoverflow: qt webview shows no https site Oct 23 14:44:59 rburton, milestone reached: qt5 compiles and installs! I think adding a mingw target is way less work than people think, it seems! Oct 23 14:48:09 litb: I'm very interested in what you did, we might need something similar (but less complex) in here. Is there any link you could share? Oct 23 14:49:45 qschulz, i need to ask my employer, but I think they will agree, so I can upload it soon Oct 23 14:51:33 qschulz, essentially, I just inherited from generic-x86, and adjusted TARGET_OS to be "mingw32", and TARGET_VENDOR to "-w64". SOLIBS should be .dll and SOLIBSDEV should be .dll.a. the remaining parts are just fiddling with compiler flags and individual recipes that need fixes Oct 23 15:04:22 litb: that might not be what I'm after afterall :/ My brain was in short-circuit I guess when I asked you about it. Oct 23 15:05:38 too bad. finally someone thinks i'm doing something useful, I thought :p Oct 23 15:07:31 Klanticus, also, I suspect that at boot time, someone mounts the tmpfs over /run. your file will still be there, I suppose, but it's "overlayed" by the mount in /run Oct 23 15:08:32 In very short, I'm looking for a way to have a few recipes with different CFLAGS/LDFLAGS, not the whole build. And those recipes will exist twice in the rootfs, once "normal" and once with modified CFLAGS/LDFLAGS. I was thinking about multilib and thought that's what you were doing. Definitely misunderstood :/ Oct 23 15:10:44 hm, I see. mabye BBCLASSEXTEND would be useful here? Oct 23 15:15:26 yeah, that's what I;m thinking, I just couldn't find a very easy example on how to do this. (also the modified CFLAGS/LDFLAGS have to apply also on top of multilib lib32 :/) Oct 23 15:16:12 if I have multiple .bbappend files that all match a given recipe Oct 23 15:16:17 in which order are they appended to the recipe? Oct 23 15:16:32 is it unspecified? Oct 23 15:17:03 hm, I see. i've not yet looked into how multilibs work Oct 23 15:18:25 something I've not yet understood is that there's a /usr/lib, but also a /usr/lib/ , the latter of which also implies the (for example i368-linux- implies =32 Oct 23 15:18:37 and both folders contain files.. Oct 23 15:19:45 litb: layer priority for bbappends. You can find out in which order empirically with bitbake -e, it tells you in which order your bbappend are applied Oct 23 15:21:26 for overrides, I expect there must be something like a "more specific" and "less specific" classification. so if i say "_append_" and "_append__" and both match, bitbake would take the second Oct 23 15:55:03 * zeddii just wondered something. Can I get into OEDAM without a yocto summit pass ? Oct 23 15:55:33 zeddii, Ask Lieu or Andreea Oct 23 15:56:01 for some reason, I thought they were related, but unrelated. Oct 23 15:56:10 well they sort of are Oct 23 15:56:13 I guess I can buy a pass and expense it, but last I heard the link wasn't working. Oct 23 15:56:22 (for registration) Oct 23 15:56:25 if it isn't we want to fix it Oct 23 15:56:48 I wasn't planning to go to anything Thursday, but if anyone asks, I did :P Oct 23 15:56:56 ok. I'll google up the link and try it. Oct 23 15:57:03 * zeddii verbed google Oct 23 15:58:37 according to: https://events19.linuxfoundation.org/events/open-source-summit-europe-2019/program/co-located-events/#yocto-project-summit, day 2 is just a reception and drinks. whoo. Oct 23 16:06:01 https://www.openembedded.org/wiki/OEDEM_2019 Oct 23 16:06:07 Can you check the lionks from here Oct 23 16:06:23 and if possible, can you point out your issues to Lieu/Andreea? Oct 23 16:15:11 seems to work. Oct 23 16:15:35 was holding off to see if there were any already bought passes that others don't need, etc. call me lazy :D Oct 23 16:17:04 zeddii, has issues ?? Oct 23 16:17:22 that's not up for debate!! Oct 23 16:35:45 zeddii: no debate needed? Oct 23 16:36:16 yup. guilty as charged :P Oct 23 18:18:03 New news from stackoverflow: Setting up Yocto on my Ubuntu (Ubuntu 18.04.3 LTS/bionic) with error "ImportError: No module named '_sqlite3'" Oct 23 18:33:13 Klanticus: note that volatiles is only used for sysvinit or busybox init setups, systemd uses tmpfiles, so recipes often need to set up both Oct 23 18:33:22 (very late response) Oct 23 19:18:11 New news from stackoverflow: Yocto Conflict between attempted installs Oct 23 19:23:51 LetoThe2nd: This mailinglist requires people to subscribe before posting. Here is the patch: http://sprunge.us/sT5062 Oct 23 20:43:21 Hi, I am building u-boot with yocto for ppc64 and when tring to debug it with TRACE32 Lauterbach the steps are pretty wire. I am susupecting that is because u-boot was build with optimization -O2. I would like to build it with optimization -O0. How to do that? **** BEGIN LOGGING AT Thu Oct 24 01:31:54 2019 Oct 24 02:19:36 if I'm building from x86_64 to rapsberry pi 4, and i'm bitbaking meta-toolchain-qt5 how do I get meta-qt5 to use the qmake.conf file found in https://code.qt.io/cgit/qt/qtbase.git/tree/mkspecs/devices/linux-rasp-pi4-v3d-g++/qmake.conf Oct 24 02:31:14 or if there is a particular forum to post this on **** ENDING LOGGING AT Thu Oct 24 02:59:57 2019