**** BEGIN LOGGING AT Wed Sep 27 03:00:01 2017 Sep 27 03:06:45 behanw: heh, interesting experiment :) Sep 27 03:54:48 bluelightning: The really interesting thing to me is that Windows users would be able to trivially work on or use OE and/or YP Sep 27 06:44:09 How can I get C.UTF-8 enabled as system default? IMAGE_LINGUAS and GLIBC_GENERATE_LOCALES are unset. Sep 27 06:44:45 Which recipe provides locale data? tmp-glibc/work/corei7-64-refkit-linux/glibc/2.26-r0/package/ has no /usr/lib/locale at all. Sep 27 06:56:06 rburton: ^^^ Sep 27 07:05:13 I've set IMAGE_LINGUAS = "en-US C" but nothing gets installed into the rootfs. Nor do I get an error message. Sep 27 07:08:48 Is there some other variable which controls which locales are packaged? Sep 27 07:09:27 I checked a few examples, but none of them had locales. For example, packages-split/bash-locale is empty for bash. Sep 27 07:16:32 Hrrrm, refkit-image.bbclass overrides IMAGE_LINGUAS. Now that I have en-US added properly, I get: Sep 27 07:16:32 opkg_prepare_url_for_install: Couldn't find anything to satisfy 'locale-base-en-US'. Sep 27 07:16:40 Where should that package come from? Sep 27 07:18:50 Hi, is there a way to monkey patch bitbake during runtime like you can do it with rails? https://stackoverflow.com/questions/41705391/where-to-put-code-when-monkey-patching Sep 27 07:19:48 I know I can open up python classes but I'm not sure where to put my code so it gets evaluated before it gets run Sep 27 07:20:27 Jeena: out of pure interest, whats the usecase? Sep 27 07:26:35 we have a patch for one move function which we already sent upstream and we assume that it will be merged, but untill then we need that patch applyed locally and in our CI Sep 27 07:27:34 we can run the patch and patch the files before we run bitbake, but it's always one additional manual step which many people forget to do Sep 27 07:29:13 thanks. was just curious, and despite having no idea if it works or even if its a good idea, it sounds kinda interesting :) Sep 27 07:42:58 # localectl set-locale de_DE Sep 27 07:42:58 Failed to issue method call: Invalid Locale data. Sep 27 07:43:04 Grrrr. Sep 27 07:44:03 en_US is the way to go when you're a programmer ;) Sep 27 07:44:46 FWIW, IMAGE_LINGUAS = "en-us de-de" has an effect (localectl list-locales shows en_EN and de_DE), but they seem to be unusable on the device (qemu). Sep 27 07:45:06 Same error for "localectl set-locale en_US". Sep 27 08:07:27 Hmm, looks like "LANG=en_US.UTF-8 " does have an effect. My initial test was with fwupdmgr, which did not show localized messages because (for some reasons) it was looking for .mo files under /home/root/share. Sep 27 08:07:48 Localized messages worked in util-linux commands like "kill". Sep 27 08:08:26 Is there already a way to set up /etc/locale.def in an image? Or do I need a custom ROOTFS_POSTPROCESS command? Sep 27 08:13:25 Hi all, i tried to read the mega manual about kernel-modules but i didn't understand the purpose of it. What does the line " kernel-modules" means when putting into IMAGE_INSTALL? Thanks. Sep 27 08:14:05 Entries in IMAGE_INSTALL are names of packages that are going to be added to the image Sep 27 08:14:22 neverpanic : Yeah Sep 27 08:14:33 In the mega manual Sep 27 08:14:35 it says : Sep 27 08:14:36 Tarballs that contain all the modules built for the kernel. Kernel module tarballs exist for legacy purposes and can be suppressed by setting the MODULE_TARBALL_DEPLOY variable to "0". The deploy/images/machine directory can contain multiple kernel module tarballs for the machine. Sep 27 08:14:38 kernel-modules hence is a package generated by your build somewhere, presumably in a similarly-named recipe (or in the case of kernel modules, your kernel recipe) Sep 27 08:15:01 ChrysD: It essentially installs the modules from your kernel into the image Sep 27 08:15:08 Otherwise they aren't added to the image. Sep 27 08:15:39 behanw : the modules that was choosen by the menuconfig? Sep 27 08:15:53 Yes. Sep 27 08:16:01 behanw : i understand now thanks Sep 27 08:16:03 Or otherwise in you config file Sep 27 08:16:09 ChrysD: The section you quoted is unfortunately about a different 'kernel-modules', it's not a package, but a folder in deploy Sep 27 08:16:25 neverpanic: True. He's mixing the 2 things up. Sep 27 08:16:33 But that doesn't matter. Sep 27 08:16:43 ok thanks Sep 27 08:16:48 In this case it just installs the modules for the kernel which is what he cares about. Sep 27 08:17:40 Ok so it's more about the kernel.bb class Sep 27 08:17:50 kernel.bbclass* Sep 27 08:17:57 ChrysD: Basically the old way and the new better way Sep 27 08:18:07 Concentrate on the new way. Don't worry about the old way Sep 27 08:19:08 behanw : yeah thanks btw. i was in the wrong section of the mega manual. Sep 27 08:19:26 ChrysD: I figured. ;) Sep 27 08:19:48 That manual is good. But bloody huge... Sep 27 08:20:57 behanw : yeah so i've done a simple "CTRL-F" and was unlucky to find the same name for other thing even if it was for the same application ahahah. Sep 27 08:58:57 pohly: there's an outstanding bug to make all this a bit better to use Sep 27 09:00:17 rburton: that probably would be very useful. Do you have a bug number? Sep 27 09:09:21 hi.. anyone knows how to get docker with yocto? Sep 27 09:15:10 btooth: a docker container to do your builds? read up on https://crops.github.io/crops/update/2016/04/19/what-is-crops.html Sep 27 09:17:19 hi guys... I have base-files_%.bbappend to change my /etc/fstab.... but when looking at the rootfs my custom fstab is not there... I checked with bitbake-layers show-appends | grep base-files and it is there Sep 27 09:17:34 however another layer provides the same bbappend Sep 27 09:17:39 how to set priority? Sep 27 09:20:34 LetoThe2nd: no, I want to have docker.io in my image Sep 27 09:20:48 to be able to run docker images on my target Sep 27 09:21:05 btooth: https://layers.openembedded.org/layerindex/branch/master/recipes/?q=docker Sep 27 09:21:42 LetoThe2nd: thanks! do you know this works on armv8 (64bit)? Sep 27 09:22:05 btooth: i would expect, but not guarantee it. Sep 27 09:22:19 okay i will try Sep 27 09:22:24 thanks again Sep 27 09:24:51 have fun Sep 27 09:31:58 wooosaiiii: did you put FILESEXTRAPATHS_prepend := "${THISDIR}/files:" in bbappend ? Sep 27 09:33:05 wooosaiiii: layer priority are in meta-x/conf/layer.conf BBFILE_PRIORITY_X Sep 27 09:33:15 nayfe: my bbappend looks like this... Sep 27 09:33:16 FILESEXTRAPATHS_prepend := "${THISDIR}/${PN}:" SRC_URI += "file://fstab" Sep 27 09:34:46 not sure SRC_URI is needed as file already exists in .bb Sep 27 09:36:03 ok I changed my layer priority to higher number... Sep 27 09:36:26 I believe it will work now... Sep 27 09:49:43 Do you know if there is some recipe or a package to add the unit_tests for I.MX6 ? Sep 27 09:49:46 thanks by advance. Sep 27 09:51:32 oh god, so stupid, wrong manual again. I found. For people who may be interested in, add the package : imx-test. I will try Sep 27 10:15:20 I am writing a recipe for console/utility and after successful compilation during packaging process I am getting the error, "do_package_qa: QA Issue: -dev package contains non-symlink .so" Sep 27 10:17:11 I am not sure, which part of recipe should I be looking at to debug this? Sep 27 10:50:02 baali: the -dev package has a file /usr/lib/lib.so which isn't a symlink. either the library isn't versioned at all (bad library) and you need to fiddle FILES_ to put stuff in the right place, or the install is broken. Sep 27 10:52:50 rburton: Thank you for the insight. So I am not trying to install a library with this recipe, I just want to compile a github repository... https://pastebin.com/U6tAY8Gq Sep 27 10:53:06 baali: well something is installing a library Sep 27 10:53:32 baali: no need to write your own do_install as the cmake class does that Sep 27 10:53:44 okay.... Sep 27 11:32:10 Hi Sep 27 11:32:43 in GDP, we have PV = "1.3+snapshot-${DATE}" for naming the image Sep 27 11:33:44 however, because of the date changing (like everyday), we get task hash mismatch errors on subsequent rebuilds Sep 27 11:34:26 does anyone know a solution to this issue? Sep 27 11:44:04 remove DATE from the task hash with vardepsexclude Sep 27 11:44:44 ah by putting it in PV you're skirting around the existing exclusion i guess Sep 27 11:45:03 PV[vardepsexclude] = "DATE" should work Sep 27 11:46:45 presumably you're putting PV back into IMAGE_NAME as by default it isn't in there Sep 27 11:46:49 so this isn't a problem Sep 27 11:46:53 meta/conf/bitbake.conf:IMAGE_NAME = "${IMAGE_BASENAME}-${MACHINE}-${DATETIME}" Sep 27 12:28:49 I am trying to use OpenGL/GLX but "glxinfo" from recipe "mesa-demos" reports an error "Couldnt find RGB GLX visual or fbconfig", does anyone know what I could be doing wrong? Sep 27 12:40:20 LetoThe2nd: i get abort: 'https://code.google.com/archive/p/gosqlite' does not appear to be an hg repository Sep 27 12:40:22 any idea? Sep 27 12:44:16 wouterstreamit: missing drivers? what does the xorg log say Sep 27 12:45:55 rburton: I am getting a bit further, apparently my GPU (rk3288) does not support glx. But I am trying to use chromium and even with the --use-gl=egl flag, it uses the ANGLE library which still tries to get a glx backend for some reason Sep 27 12:46:17 btooth: well what recipe does throw this error? maybe it is outdated? and, are you sure you are not mixing branches? Sep 27 12:48:02 LetoThe2nd: meta-virtualization/recipes-devtools/go/go-sqlite_hg.bb on fido branch Sep 27 12:48:49 if i do: hg clone https://code.google.com/p/gosqlite -- i get the same error Sep 27 12:49:57 btooth: fido? are you serious? Sep 27 12:53:23 LetoThe2nd: yes :/ i am forced to use fido Sep 27 12:53:38 btooth: i guess then you are in for a lot of manual backporting Sep 27 12:54:10 okay.. i will forward this to my manager :) Sep 27 12:55:44 btooth: given the fact that meta-virtualization and docker are quite active (see http://git.yoctoproject.org/cgit/cgit.cgi/meta-virtualization/log/) and the last commit on fido was 2015-05-13, i suspect a lot of things might be broken there already. Sep 27 13:06:55 hi all Sep 27 13:07:17 there was a command to find the location of a recipe Sep 27 13:07:25 bitbake something Sep 27 13:07:31 for a specific target Sep 27 13:07:38 eg. bitbake u-boot Sep 27 13:07:46 where is u-boot.bb? Sep 27 13:07:55 * rburton uses find :) Sep 27 13:08:31 :) Sep 27 13:08:40 Costin1: bitbake-layers show-recipes -f u-boot Sep 27 13:09:23 thanks ross Sep 27 13:09:27 Ross sorry Sep 27 14:33:43 so I'm doing some ugly stuff, CONFIGURE_machine := "${@d.getVar("FOO").replace("--bla=${TARGET_SYS}", "--bla=other") Sep 27 14:34:07 it looks like even though I pass False to getVar, and this is an immediate assignment, my replace is not doing anything Sep 27 14:34:31 because ${TARGET_SYS} has already been expanded in the setting of CONFIGUREOPTS Sep 27 15:20:57 ramcq: first, depending on bitbake version, that getVar() without passing ,True will still default to True. second, yes, that ${TARGET_SYS} is expanded before that inline python is run. i'd suggest using getVar() without expansion on that if you want both getvars to be on the unexpanded value, otherwise expand both Sep 27 15:28:20 right, expanding all the way :) Sep 27 15:28:27 also realised I changed some stuff like Sep 27 15:28:36 FOO := ... getVar("FOO").mangle() Sep 27 15:28:45 to FOO_machine := getVar("FOO").mangle() Sep 27 15:28:49 except, there were two Sep 27 15:28:53 spot the error... :P Sep 27 16:12:03 khem: Morning, got a MUSL / Go issue possibly Sep 27 16:16:34 khem: when TCLIBC = "musl" and MACHINE = "qemux86-64" then we get some undefined references when building go-helloworld and go-dep, in linux_amd64_dynlink//libstd.so to __libc_{malloc, realloc, stack_end, free} Sep 27 18:42:22 kergoth: the inline python doesn't seem to be getting it's ${VARS} interpolated before running Sep 27 18:42:28 at least my recipe is pretty hosed still :/ Sep 27 18:48:32 * ramcq just getVar... looks more hopeful Sep 27 19:38:11 ramcq: inline python gets ${} expanded, but only if the variable is defined, anonymous python does not. it'll choke if you try to use a ${} that isn't set, so if necessary define a fallback default Sep 27 19:43:18 ramcq: use d.getVar("FOO") in python, ${FOO} doesn't get expanded in py Sep 27 19:48:23 that's not actually the case for inline python Sep 27 19:48:46 should we change that? :) Sep 27 19:56:28 yeah, if only because we have no way to escape it and prevent the expansion, and to improve consistency Sep 27 19:56:35 we should really add a way to escape expansion though regardless Sep 27 19:57:47 kergoth: yes, I know the lack of that has bothered a few people Sep 27 21:23:06 can openembedded generate windows SDKs? (with crosscompiler & binutils that run on windows) Sep 27 21:27:55 sobczyk: i don't think so Sep 27 21:29:27 sobczyk: check http://git.yoctoproject.org/cgit.cgi/meta-mingw Sep 27 21:29:40 but I've never used it Sep 27 21:30:35 isn't mingw supposed to compile for windows? Sep 27 21:31:05 isn't that what you wanted? Sep 27 21:31:31 I want to compile on windows for linux Sep 27 21:32:20 so, a cross canadian build of gcc or clang Sep 27 21:32:29 and binutils Sep 27 21:32:54 I'll just have to try it I guess Sep 27 21:44:37 would be easier to just use VirtualBox Sep 27 23:03:24 (52 **** ENDING LOGGING AT Thu Sep 28 03:00:01 2017