**** BEGIN LOGGING AT Thu Apr 16 02:59:57 2020 Apr 16 06:12:21 New news from stackoverflow: How to compile Makefile based recipe in bitbake? Apr 16 07:31:50 mischief: what exactly do you want to do? Apr 16 07:42:48 qschulz: i have a bug in a program that is somewhat out of my control. if the WORKDIR/S contain regexp metachars, it silently fails. Apr 16 07:43:39 qschulz: my solution right now was to assign WORKDIR/S to get rid of the git revs inserted by PV and it seems to work... Apr 16 07:45:03 and for anyone curious: the program with the bug is spatch from coccinelle :) Apr 16 07:56:18 mischief: I guess it could be worth reporting to coccinelle maintainters? surely something to fix during compilation right? Apr 16 08:13:54 qschulz: yes, perhaps Apr 16 08:14:37 its not easy because we use spatch as a host tool right now. building ocaml and then coccinelle in yocto is a huge PITA and we gave up so we are stuck with the host version for now. Apr 16 08:15:28 its simpler to try to make sure the workdir doesn't have regexp metacharacters Apr 16 11:47:09 Hi all, is there a standard setup-environment script ? looks like everybody is doing this kind of script in there repo Apr 16 11:54:55 clementp[m]: are you sure everyone isn't just wrapping the one in oe-core? Apr 16 11:56:23 rburton: I'm new to Yocto and I use mostly NXP/iMX and Mender meta-repo for now and both use different setup script Apr 16 11:57:10 https://github.com/mendersoftware/meta-mender-community/blob/sumo/scripts/setup-environment for Mender Apr 16 11:57:51 https://github.com/Freescale/fsl-community-bsp-base/blob/master/setup-environment Apr 16 11:59:13 the important bit is . ${script_dir}/sources/poky/oe-init-build-env ${build_dir} Apr 16 11:59:47 they're just repo-specific 'value-add' over the oe-core setup Apr 16 12:00:17 Yes I'm looking to have a standard / proper way to specify my BBLAYERS Apr 16 12:00:49 and local.conf Apr 16 12:01:56 your distro layer can ship templates for those Apr 16 12:02:39 https://github.com/rossburton/customdistro Apr 16 12:04:15 Haaaaa nice Apr 16 12:04:19 thank you :) Apr 16 12:10:39 RP: something for the usability discussion, add 'how to make a custom distro' to the docs ^^^ Apr 16 12:10:46 clean and efficient exactly what i was lookin for ! Apr 16 12:11:25 uses submodules obviously, but if you want to use repo or whatever it shows the structure of what you want to do Apr 16 12:22:18 Yes i'm using repo Apr 16 12:32:48 I switched to submodules from repo and am so much happier Apr 16 12:41:41 * alessioigor waves all Apr 16 12:41:43 Have someone incurred into "AttributeError: module 'platform' has no attribute 'linux_distribution'" building a Python3 package with dunfell on Ubuntu 18.04? Apr 16 14:05:32 hello! following up from my troubleshooting yesterday... after starting with a fresh build dir, i still cannot get systemd-networkd nor systemd-resolved enabled as default in a new image for core-image-base. reviewing bitbake -e, SYSTEMD_AUTO_ENABLE is still enabled by default. In a systemd bbappend recipe i've specified both PACKAGECONFIG to Apr 16 14:05:33 include systemd + resolved and SYSTEMD_AUTO_ENABLE_systemd-networkd specifically to enabled. I'm suspicious that a systemctl status on these service show them disabled, but vendor preset is enabled... but i'm at a loss as to what to look at next for debugging. could anyone a next step for me? Apr 16 14:07:22 khem: i did read through the systemd.bbclass, but didn't see in log.do_rootfs where systemctl enable was called to create say the symlinks under /etc/systemd/system/multi-user.target.wants/ (seems like the case *@*) fails when building the image so it's for ontarget only? but some packages (e.g. nginx) get symlinks created Apr 16 14:14:49 rburton: bugzilla entry? Apr 16 14:14:58 RP: the fix for perl worries me a lot on the license+multilib. Because there are many PACKAGES_DYNAMIC which aren't using either MLPREFIX or ${PN} which would mean (IIUC) that all of those will also have issues wrt licensing and mulitlib Apr 16 14:15:27 license+multilib +bug Apr 16 14:19:48 qschulz: How many are set in anonymous python though? Apr 16 14:20:56 qschulz: this is the consqeuence of that event handler change, it means code in anon python handlers may not be quite correct :/ Apr 16 14:22:56 RP: You're right, it seems only perl is doing it in the python anon function (git grep "d\..etVar.*PACKAGES_DYNAMIC") Apr 16 14:23:42 RP: unfortunately I have never even looked at how bitbake is working. So all I can do is test the suggested patch :/ I do not understand it unfortunately :( Apr 16 14:23:58 qschulz: its all basically an ordering problem Apr 16 14:24:30 qschulz: multilib does things before key expansion and things after but there was no dedicated event so ordering raced against base.bbclass and others Apr 16 14:24:43 you can apply various bandaids but the real fix is the one I've shared Apr 16 14:24:57 we may have to undo or fix other bits to match Apr 16 14:25:47 I'm building right now but I can already say there is something else to fix: Apr 16 14:26:06 WARNING: lib32-glibc-2.31+gitAUTOINC+71f2b249a2-r0 do_package: QA Issue: lib32-glibc package lib32-glibc - suspicious values 'ldconfig' in RRECOMMENDS [multilib] Apr 16 14:32:34 steve@sakoman.com Apr 16 14:35:11 khem: seems like it's do_image calling systemctl (systemd_preset_all()), ty Apr 16 14:48:34 Hello all, little question today: I have 2 image recipes. img2 require img1. What is the best way to set 2 different login/password for each image? Apr 16 14:53:29 Ninic0c0: https://www.yoctoproject.org/docs/latest/mega-manual/mega-manual.html#ref-classes-extrausers ? Apr 16 14:59:02 qschulz definitely a french reply :P Apr 16 15:03:22 Ninic0c0: was not meant to be an RTFM :) it's just that I've never done it and found this one in the documentation which explicits it can be used directly in the image recipe :) Apr 16 15:03:27 ah, he's gone :/ Apr 16 15:03:33 beat_hitachi: they are already in the packageconfig of systemd_243.bb, you may not to add them in a bbappend. But it seems that it's the systemd build that is not enabling resolved/networkd by default. Look at $(SYSTEMDSRC)/units/meson.build Apr 16 15:10:01 beat_hitachi: adding the .wants path in the resolved entry in in_units makes it auto enable for me. You could patch it to do that, or enable your systemd services manually in a bbappend. Apr 16 15:16:39 champagneg thank you for the response! i thought i was enabling them manually in a bbappend with the explicit SYSTEMD_AUTO_ENABLE_systemd-resolved. i will chase this down a little more with your info. i was able to link these explicitly in a bbappend prior. i've been mostly interested in the 'correct' way to enable the services, because i haven't Apr 16 15:16:39 understood why documented way to enable services was not working. Apr 16 15:53:39 beat_hitachi: yeah, you're probably right that SYSTEMD_AUTO_ENABLE would workd ideally. But it's handled by systemd.bbclass which mainly works with packages that provide additional systemd services. systemd-resolved is provided by the systemd package itself, which is not included in SYSTEMD_PACKAGES in systemd.bb. Apr 16 15:56:18 systemd.bbclass _is_ inherited by systemd because it has subpackages that do provide additional services, but the core systemd services are not affected by SYSTEMD_AUTO_ENABLE from what I understood from the recipe. Maybe you could also make it work by tweaking the variables that systemd.bbclass checks (SYSTEMD_PACKAGES, SYSTEMD_SERVICE_). Apr 16 16:44:28 champagneg: strangely, for me it looks like resolved is enabled and has .wants path in the in_units for ./units/meson.build. perhaps i'm misreading this. ['systemd-resolved.service', 'ENABLE_RESOLVE', join_paths(pkgsysconfdir, 'system/dbus-org.freedesktop.resolve1.service') + ' ' + join_paths(pkgsysconfdir, Apr 16 16:44:29 'system/multi-user.target.wants/')], In any case, i'm attempting to add the SYSTEMD_PACKAGES and SYSTEMD_SERVICE_systemd-resolved in the bbappend to see if it works Apr 16 16:58:00 well, unfortunately adding the various SYSTEMD variables to the systemd bbappend did not work either. i'm going to have to step away for a little bit. this is baffling Apr 16 16:59:30 i feel like i'm missing something fundamental because enabling the basic systemd-networkd and systemd-resolved shouldn't be this hard Apr 16 21:15:21 New news from stackoverflow: wifi driver periodically drops all connected hosts and data [closed] Apr 16 21:45:27 New news from stackoverflow: Need recipe for Realtek wifi dongle [closed] **** ENDING LOGGING AT Fri Apr 17 02:59:57 2020