**** BEGIN LOGGING AT Fri Jan 04 02:59:57 2019 Jan 04 03:35:17 use _append Jan 04 04:45:50 New news from stackoverflow: Failed do_rootfs for agl-demo-platform Jan 04 04:57:39 is there a recent image (or build point) for the later Sharp Zaurus like the C-3200? Jan 04 05:56:07 ds2: http://www.penguin.cz/~utx/zaurus/feed/images/spitz/ Jan 04 05:58:23 and there is http://cgit.openembedded.org/meta-handheld/tree/conf/machine/spitz.conf?h=master Jan 04 05:58:37 but dont know if it builds may be it does Jan 04 05:58:59 all you need is to add meta-handheld to your bblayers.conf Jan 04 06:00:13 hmm last date is 2013 Jan 04 06:00:25 not a quick 2hr fun time Jan 04 09:29:42 anyone uses dbus-broker ? Jan 04 09:30:28 isn't this super bleeding edge? Jan 04 10:05:58 derRichard: that's why i'm asking :) Jan 04 10:07:30 ak77: i'd ask bus1 folks. i don't know what the current state is. Jan 04 10:07:40 AFAIK they are on irc Jan 04 10:43:07 RP: http://git.yoctoproject.org/cgit.cgi/poky-contrib/log/?h=akanavin/perl-sanity is again ready for AB after another round of fixing :) Jan 04 10:43:14 rburton: ^ Jan 04 11:25:06 thanks kanavin Jan 04 12:01:34 kanavin: why the need to remove perl-native from RDEPENDS for native builds when there is a perl-native (bbclassextended, but it still exists) Jan 04 12:02:57 kanavin: quick build fired on the ab Jan 04 12:08:02 RP: fixed the epoxy thing by picking a meson patch from git Jan 04 12:20:03 rburton: http://git.yoctoproject.org/cgit/cgit.cgi/poky/tree/meta/recipes-devtools/meson/nativesdk-meson_0.49.0.bb?id=eb4898839fbf6060c440052cf40e6ce1d4e35ea7 implies a dependency on gcc, g++, etc. via the `meson.cross` yet does not RDEPEND on it. I that correct? Jan 04 12:21:10 it would actually imply a native+cross dependency. Meson needs both to do cross compilation. Jan 04 12:24:59 either that or there should be like a set of meson classes to inherit from meson-{bare,c,cpp,java,d,...} to automatically add the correct compilers to the sdk? Jan 04 12:39:26 or maybe each recipe that does `inherit meson` should explicitly depend on its compiler too so that the build system can infer the correct compiler tuple for the sdk? Jan 04 13:01:58 hi all Jan 04 13:03:20 can anybody tell me what happened to scripts/contrib/mkefidistk.sh? I'm following the quickstart tutorial and i can't flash my minnowboard Jan 04 13:06:01 wic has an efi command to do that Jan 04 13:06:26 iirc, if you build a wic image then you'll get something you can just dd Jan 04 13:23:08 thanks @rburton Jan 04 14:30:03 hey Jan 04 14:30:50 I tried wic, but it gave me the following error "ERROR: ("The artifact that couldn't be found was %s:\n %s", 'rootfs-dir', '/home/holoplot/asalle/poky/build/tmp/work/intel_corei7_64-poky-linux/core-image-sato/1.0-r0/rootfs')" Jan 04 15:11:53 which approach do you guys suggest to override a function in meta/classes/image-live.bbclass? i'd like to change do_bootimg() Jan 04 15:12:29 write a new class that provides the new implementation Jan 04 15:12:34 or better, send the patch Jan 04 15:13:46 rburton: well, when i create a new class which inherits image-live.bbclass, image.bbclass will still inherit the original image-live.bbclass Jan 04 15:14:14 but if your image inherits your class atfter image.bbclass, it will override the function Jan 04 15:15:40 ok, i think too much in c++/java i fear. bitbake "inherit" is different Jan 04 15:23:51 rburton, any chance you could run perl-sanity branch on the AB? Jan 04 15:24:35 kanavin: already fired, nearly done Jan 04 15:24:43 oh Jan 04 15:24:46 can I see? Jan 04 15:24:48 kanavin: i think the "perl-native doesn't exist' path can be dropped Jan 04 15:24:52 https://autobuilder.yoctoproject.org/typhoon/#/console Jan 04 15:25:10 the intltool change is still needed, but i'm not sure we want to get rid of the nativeperl wrapper Jan 04 15:25:36 ah target perl failed on my other machine, lets get the log off that Jan 04 15:30:35 rburton, not bad, there is one multilib failure (not a difficult one, conflicting -doc packages), and then one oe-selftest failed because it couldn't start qemu (unrelated to perl I guess?) Jan 04 15:30:52 most likely Jan 04 15:32:21 hi, i get following warning: WARNING: networkmanager-1.14.4-r0 do_configure: QA Issue: networkmanager: invalid PACKAGECONFIG: bluez5glib [invalid-packageconfig] when building an image for my target, i think i spotted the problem had have a valid patch. It is for recipes-connectivity/networkmanager/networkmanager_%.bbappend in the meta-gnome layer. Where should i send my patch? Jan 04 15:34:56 read the readme in the layer Jan 04 15:37:22 rburton: works, thanks lot for the point! Jan 04 15:37:23 thanks and sorry missed it Jan 04 15:37:25 *pointer Jan 04 15:46:56 kanavin: perl-sanity fails on my other machine with 'Can't locate DynaLoader.pm in @INC' when building perl itself Jan 04 15:47:41 this is an old ubuntu though, should get it upgraded Jan 04 15:54:30 rburton: hm, I haven't seen that :( and neither has the AB Jan 04 15:56:59 kanavin: nativeperl is useful because it lets you run the perl we built without needing to inherit perlnative. why remove it? Jan 04 15:59:03 rburton: complexity reduction Jan 04 16:00:28 needs to be spelt out clearly that perl-native has different behaviour then Jan 04 16:00:38 also different behaviour to py-native Jan 04 16:02:26 there's an argument that perl/nativeperl is redundant with RSS, but that's orthogonal to the perl upgrade. i'd prefer it the perl upgrade kept nativeperl, and perl in a subdir, then the native* debate can happen Jan 04 16:04:47 rburton: I basically wrote the perl recipe from scratch, adding pieces where necessary. Where the choice was between making the perl recipe more complex, and fixing other recipes, I chose fixing the other recipes (particularly not having perl-native as a separate recipe required some work). Jan 04 16:04:59 so it was not some evil plan :) Jan 04 16:05:24 having a single recipe is definitely good, the split in py hurts as we keep patching one but not the other Jan 04 16:05:41 but just mving bindir/perl to bindir/perl-native/perl for class-native isn't hard Jan 04 16:06:25 but what happens if some scripts want perl? Jan 04 16:08:56 the point is that the move to perl-sanity shouldn't change behaviour Jan 04 16:09:21 removing nativeperl wrapper, the bindir/perl-native folder, etc, can happen as a separate debate Jan 04 16:15:06 kanavin: local testing says you can drop the RDEPENDS_class-native = "" chunks from the recipes patch, and just rename it to intltool: don't use nativeperl Jan 04 16:17:27 right, I can bring those back Jan 04 16:21:04 and then drop that as intltool would use nativeperl again? :) Jan 04 16:22:08 yeah :) Jan 04 16:22:22 had a look at what still needs intltool, annoyingly quite a few recipes left Jan 04 17:25:10 RP, rburton, armpit, AB updates and reboots complete. Jan 04 17:25:24 thanks Jan 04 17:55:04 does anybody know off your head which kernel version the xilinx-linx releases are based on? Jan 04 18:44:51 Piraty: look into meta-xilinx it seem 4.14 see https://github.com/Xilinx/meta-xilinx/blob/master/meta-xilinx-bsp/recipes-kernel/linux/linux-xlnx_2018.3.bb Jan 04 18:45:11 good hint, thanks Jan 04 20:05:37 khem, ++ Jan 04 20:05:44 * Crofton curses XIlinx Jan 04 20:09:39 is there a way to enable systemd "@.service" in yocto? For instance I would like to enable wpa-supplicant@wlan0.service Jan 04 20:12:45 Crofton: :D yeah Jan 04 20:13:01 dkc: ship the symlink along :D Jan 04 20:15:24 :( Jan 04 20:16:43 why not. systemd is broken by design, why not play according to that philosophy? Jan 04 20:19:22 troll asides, Yocto has a way to enable systemd services so it would be nice to be able to cover all use cases Jan 04 20:25:47 it actually works, not making typo would help me to not ask stupid questions Jan 04 20:35:01 we need the fix-typo.bbclass that would greatly improve usability Jan 04 21:02:30 dkc: usually u create needed symlink via a bbappend Jan 04 21:03:40 Crofton: https://ownyourbits.com/2018/06/13/transparently-running-binaries-from-any-architecture-in-linux-with-qemu-and-binfmt_misc/ Jan 04 21:03:54 Crofton: do we need cross compilation frameworks anymore :) Jan 04 21:04:11 kergoth: ^ Jan 04 21:04:42 for some reasons it works to have SYSTEMD_SERVICE_${PN} = "wpa_supplicant@wlan0.service" but it creates wpa_supplicant@%i.service Jan 04 21:05:40 yeah %i should be replaced with interface name Jan 04 21:06:01 khem: heh, same sort of thing scratchbox was doing yeras ago no? Jan 04 21:06:38 yes indeed, but I think qemu-usermode is interesting here Jan 04 21:06:54 this is going to be near native performance Jan 04 21:13:45 it's weird, it's really systemctl that substitutes wlan0 for %i Jan 04 21:16:55 oh wait, of course systemctl is a custom Yocto shell script and not the tool Jan 04 21:18:54 right remember we are doing offline presets Jan 04 21:19:11 so its a cross tool Jan 04 21:25:48 okay, the cross tool is confused by these lines: Jan 04 21:25:50 [Install] Jan 04 21:25:52 Alias=multi-user.target.wants/wpa_supplicant@%i.service Jan 04 21:29:13 what do you think? Should we try to make systemd-systemctl handle these cases? Or forget it and let people symlink? Jan 04 21:49:26 Is 'runqemu' supposed to work automatically when using INITRAMFS_IMAGE_BUNDLE=1 ? It is complaining about failing to find rootfs. Jan 04 21:52:53 kroon: no, you have to call it with kernel parameter alone Jan 04 21:55:13 why is image-live.bbclass using grub-efi for efi and syslinux for pcbios? syslinux supports efi too Jan 04 21:56:51 khem, ok Jan 04 23:19:24 New news from stackoverflow: Yocto Project Bitbake Unexpected Termination Jan 05 00:09:03 Hi, Im seeing an issue with using esdk when built with autorev. I do see a bug that was filed in 2017 (https://bugzilla.yoctoproject.org/show_bug.cgi?id=11350), but we were not seeing this issue before moving to thud release. The specific issue I am seeing is : fatal unable to acccess https://github repo : error setting certificate verify locations. If i rebuild esdk with pinned commit ids and reinstall esdk it works. any ideas here Jan 05 00:09:03 ? Jan 05 00:09:04 Bug 11350: enhancement, Medium+, 2.99, paul.eggleton, NEW , eSDK should handle SRCREV = "${AUTOREV}" **** ENDING LOGGING AT Sat Jan 05 02:59:56 2019