**** BEGIN LOGGING AT Wed Dec 27 03:00:01 2017 Dec 27 10:14:14 How can I remove some PACKAGECONFIG from a .bbappend? Dec 27 10:14:24 should this work? PACKAGECONFIG_remove_class-target = "threaded-resolver" Dec 27 10:17:55 hey guys! Is there a neat workflow to create .bbappends for existing recipes? As if, I don't want to touch the sources, I just need to add and verify a few lines to original .bb script? Dec 27 10:18:06 can't figure out if devtool works with it :-( Dec 27 10:25:39 luneff, for bbappend files, you can do it by hand easily, devtool modify should do it otherwise http://www.yoctoproject.org/docs/current/mega-manual/mega-manual.html#using-devtool-in-your-sdk-workflow Dec 27 10:28:58 yes, did it by hand as 'devtool modify' seems to be about editing source code, not the recipe itself; devtool edit-recipe doesn't work after a devtool modify has started Dec 27 11:18:47 Hey guys! I have a weird question. I use boost header-only and don't want to install it on target image. But I need some headers from it during the build stage. Is there an option somehow to put into the ${WORKDIR} those headers without compiling (rather huge) boost target? Dec 27 11:50:05 wanderer_> maybe create your own recipe without any do_compile(), etc and DEPENDS on it ? Dec 27 11:58:24 nayfe>hm, sounds promising. Dec 27 11:59:10 Anyways you mean overriding the system boost recipe, right? Dec 27 12:02:23 wanderer_: create a new recipe for instance boost-headers-only_1.64.bb, maybe there are other ways but it's how i would do that with my actual "knowledge" Dec 27 12:05:10 but i would probably use boost from oe-core, compile time with yocto is only one shot with cache system Dec 27 12:12:00 should I specify MACHINE variable to devtool or it's pointless ? Dec 27 12:15:47 dev1990: not sure i got you, you have to source your yocto env to work with devtool, so it will be added automatically, right ? Dec 27 12:22:11 nayfe: I'm switching between machine on the same enviroment Dec 27 12:24:19 in short default machine from my local.conf is i7 Dec 27 12:24:47 and I'm building sometimes for rpi3 without changing local.conf (for example MACHINE="raspberrypi3" bitbake retropie-roms-image) Dec 27 12:25:45 and when I'm using devtool I'm often specify MACHINE="raspberrypi3" explicit and I'm wondering if it is pointless Dec 27 12:28:22 dev1990: I think you need it, as devtool is a wrapper for bitbake Dec 27 13:19:11 Hi guys, is there a way to build **all** recipes from a meta-XXX layers? Dec 27 13:20:39 Hello, I'm running in a configure error with libgfortran on pyro (with GCCVERSION=5.4% and PREFERRED_VERSION_libgfortran="5.4%") Dec 27 13:20:52 The config.log has many lines like this: work-shared/gcc-5.4.0-r0/gcc-5.4.0/libgfortran/configure: line 3868: aarch64-poky-linux-gcc: command not found Dec 27 13:21:00 so for some reason it's unable to reach gcc? Dec 27 13:22:28 Unless there are other prerequisites than setting the "FORTAN" variable in my local conf. Bit at a loss here Dec 27 13:26:53 mjourdan: did you try without specifying preferred versions gcc/fortran? Dec 27 13:28:04 Nope I didn't try with 6.3. But I'll launch a build without the version overrides, see what happens. Dec 27 13:29:00 hello guys, how can I install mariadb with debugging symbols to the image? Dec 27 13:32:52 Ox4: Something like IMAGE_INSTALL += "mariadb-dbg" ? If this is more for a one-time thing, maybe setup a package server instead and install the debug packages as needed. Dec 27 13:33:51 mjourdan: should I set mariadb-dbg with mariadb package? Dec 27 13:34:13 ERROR: Nothing PROVIDES 'mariadb-dbg'. Close matches: Dec 27 13:34:13 mariadb Dec 27 13:34:13 mariadb RPROVIDES mariadb-dbg Dec 27 13:37:44 What change did you do exactly? Dec 27 13:41:00 mjourdan: replaced mariadb with mariadb-dbg in IMAGE_INSTALL_append variable Dec 27 13:49:09 and with "mariadb mariadb-dbg" ? Dec 27 13:49:52 nayfe: no luck, same error. The only thing I did was add "FORTRAN_forcevariable = ",fortran"" in my conf Dec 27 13:50:34 mjourdan> can you pastebin your full log error ? Dec 27 13:53:05 config.log: https://pastebin.com/s69Y1qex Dec 27 13:53:05 log.do_configure: https://pastebin.com/nKV7pqXu Dec 27 13:59:27 Guys, can I build OpenSwitch using Yocto? Dec 27 14:07:38 kostaz: it seems there are some layers here https://github.com/open-switch/ops-build/tree/master/yocto/openswitch Dec 27 14:24:09 mjourdan> not sure yocto supports fortran, and older irc logs show it seems broken "carlesfernandez oh, I’ve just realize that libgfortran is working in krogoth (but not in jetrho or morty)", don't know if it has been fixed in rocko Dec 27 14:24:31 ouch Dec 27 14:25:00 I'm bumping form krogoth to pyro, so that explains it Dec 27 14:25:42 welp, thanks Dec 27 14:26:30 major stuff between those is the recipe-sysroot split by recipe and maybe work-shared stuff, so I suppose configure script fails to find crosscompiler for this reason Dec 27 14:28:44 mjourdan: you can diff between https://git.yoctoproject.org/cgit.cgi/poky/plain/meta/recipes-devtools/gcc/libgfortran.inc?h=pyro and https://git.yoctoproject.org/cgit.cgi/poky/plain/meta/recipes-devtools/gcc/libgfortran.inc?h=rocko they added gcc-cross-${TARGET_ARCH} to DEPENDS for instance Dec 27 14:30:20 good idea, thanks Dec 27 14:34:21 the "recipe-sysroot split by recipe" Dec 27 14:34:32 in short, a custom sysroot is generated based on the DEPENDS of each recipe ? Dec 27 14:35:01 mjourdan: yes, I left mariadb-dbg only Dec 27 14:35:53 mjourdan> exactly Dec 27 14:36:18 Ox4: you need both, -dbg just add symbols Dec 27 14:36:41 nayfe: thank you, I will try Dec 27 14:37:17 I see. So indeed with the cross-${TARGET_ARCH} dep, configure does work. Compile fails. So I just copied over the whole .inc from master and it works.. Should have started by doing that >_<. Dec 27 14:40:29 mjourdan: good to know, maybe you can ask to fix pyro branch, but everybody are in vacation those days :) Dec 27 14:46:03 Sure, I'll send an email at the poky mailing list. Dec 27 14:46:06 hello, what is the best way to notify a Qt GUI application that a USB key was inserted on an embedded Linux systemd? Dec 27 14:47:39 systemd-udev? Dec 27 14:47:57 Is there a way to cause an initramfs image to be packaged in an RPM? Dec 27 14:48:35 nayfe: yes, but I mean, how do I actually interface with that? Dec 27 14:49:30 as far as I am aware systemd-udev will not do a d-bus broadcast uppon USB attachment on its own.. or am I wrong? Dec 27 14:51:33 UDisks2 does a d-bus broadcast on Kubuntu 16.04 for example uppon USB device attachment Dec 27 14:52:25 but having udisks2 as a dependency for my image seems a bit of an overkill Dec 27 14:54:03 it has a lot of gnome-related dependencies I would not need otherwise Dec 27 14:54:21 eduardas_m: afaik you can use libudev to listen to events Dec 27 14:55:35 eduardas_m: that solution seems a bit ugly but should work https://www.toradex.com/community/questions/10602/update-qt-application-from-usb-stick.html Dec 27 15:01:25 nayfe: thank you for the link. Actually my usecase is similar as mentioned in the thread, i.e. I need this for updates. Still, it is unclear to me what is considered "good practice" in this case. Dec 27 15:05:32 eduardas_m: in my case, i'd probably use SWupdate to update system, but i don't think there is one-fits-all solution :) Dec 27 15:06:18 Can someone actually say what they use to detect USB mass storage device attachment detection in production? Dec 27 15:08:30 nayfe: True. There are openswitch layers on github. But, then there is OPX build system, that does not seem related to Yocto... (https://github.com/open-switch/opx-build). Dec 27 15:09:05 nayfe: And also https://git.openswitch.net/cgit.... It all seems strange. Dec 27 15:10:03 nayfe: I and my colleagues have decided to use SWUpdate too, actually. Do you do full system update only or package updates via swupdate too? Dec 27 15:10:41 I'm asking because allowing both full-system and package updates seems tricky to do correctly. Dec 27 15:11:16 Because not all single package versions will be appropriate for all base systems. Dec 27 15:14:41 eduardas_m: the two snippets here https://github.com/gavv/snippets/tree/master/udev are a good start imho. Dec 27 15:17:12 mjourdan: thank you a lot... that actually looks really useful Dec 27 15:21:00 kostaz: seems that previous openswitch versions (ops) where based on yocto and they switched to debian for opx ? Dec 27 15:22:20 eduardas_m: for now, only full system update, but maybe some minor parts will be updated with package management Dec 27 15:23:55 nayfe: are you using initrd, dual-image or hybrid approach? Dec 27 15:25:31 2nd Dec 27 15:25:41 we are trying to implement the initrd approach with a small recovery/update image and a bigger "user" image Dec 27 15:26:09 as that way it looks like we have an immutable reliable fallback Dec 27 15:26:50 problem with this approach to me is the down-time, if memory space is not an issue Dec 27 15:27:58 nayfe: yes, implementing package updates is motivated mainly by this... Dec 27 15:29:07 makes sense ;) Dec 27 15:29:42 eduardas_m: btw we're still in POC phase Dec 27 15:31:11 i find ostree very innovative system too Dec 27 15:32:51 nayfe: Qt Company actually has an ostree-based update system as part of their Qt for device creation product... not sure how good or reliable it is though... or how often it is being used in the field Dec 27 15:33:49 http://doc.qt.io/QtOTA/ Dec 27 15:35:52 it is also used by gnome if i remember correctly, problems comes if there is a partition corruption as there is only one Dec 27 15:53:56 New news from stackoverflow: How to install dlib and update gcc on Yocto Dec 27 16:00:50 Is there a way to try to apply a patch that is failing to apply and get a conflict resolution with <<<< ||| >>>> as if I were using git cherry-pick ? Dec 27 16:05:47 ttllkk: something like "git am < patchfile.patch" ? Dec 27 16:06:12 or git apply patchfile.patch Dec 27 16:12:41 neither git apply or git am give conflict markers. iirc you can pass —reject or somesuch to leave a standard patch .rej file Dec 27 16:15:10 gnu patch itself does provide a --merge option that may do what you want, however Dec 27 16:15:43 note that it'll show limited information, not the full info available to cherry-pick, as it doesn't have the full original file available as is the case when working within git Dec 27 16:16:48 alternatively, there's the "wiggle" tool, but it doesn't just merge, it tries harder to apply the patch, which can be good and bad, it isn't perfect, but can save time, as long as you manually review what it does Dec 27 16:23:58 thanks Dec 27 17:24:25 Hey folks. I've notice something super weird when moving from Dizzy to Morty/Rocko. Bluetooth/Bluez and dbus seem to have some bad errors and I get bad memory leaks. None of this is present building the same source for Dizzy. Dec 27 17:24:50 Any tips on isolating the cause of these failures? Is there a way I can incrementally upgrade? Dec 27 20:53:28 Hi, I am trying to build homebridge source code for raspberrypi3. (See https://github.com/nfarina/homebridge). I've created a recipe using devtool and npm registry but homebridge requires libavahi-compat-libdnssd-dev installed on Linux. So, I got a build error. Here is the output https://pastebin.com/xgyUq2Nd I think I should add an RDEPENDS which provides necessary files. Should I edit the avahi recipe to add an extra package? Or do **** ENDING LOGGING AT Thu Dec 28 03:00:02 2017