**** BEGIN LOGGING AT Fri Mar 21 02:59:58 2014 Mar 21 06:10:43 Is there any problem with initramfs support in yocto for arm architecture? Mar 21 07:49:00 khem: otavio: hi. Final linking during build of the latest Chromium 35 from meta-browser master takes an awful lot of RAM. 4GB RAM + 2 GB swap aren't enough (headless server, so most of the memory is free) Mar 21 08:44:01 good morning Mar 21 08:55:14 hi. i have a recipe that inherits distutils as well as update-rc.d. i also have a custom do_install function in that recipe. Mar 21 08:55:35 for some reason do_install of distutils does not seem to be called Mar 21 08:56:08 is it legal to inherit multiple classes? and what happens if i define a do_install function - do i have to call the base classes do_install? Mar 21 09:00:59 ... i added "distutils_do_install" at the end of my do_install function, and it seems to work. still not sure if this is how it's supposed to work.. Mar 21 09:18:22 Hello, how do i make pkg_postinst_${PN} run after mariadb is installed and are running? Mar 21 09:27:40 how do i make pkg_postinst_${PN} run after mariadb is installed and running? Mar 21 09:39:18 hey, whats the SYSROOT_PREPROCESS_FUNCS is this the same as pkg_postinst? Mar 21 10:12:57 Krise1: pkg_postinst_${PN} is executed after the package manager installs ${PN} Mar 21 10:13:04 so that's what you want Mar 21 10:37:47 i want to be able to run a shell script after mariadb has installed the database Mar 21 10:53:15 morning all Mar 21 10:55:13 bluelightning: morning Mar 21 10:55:42 hi diego_r Mar 21 11:01:47 bluelightning: we also have http://autobuilder.yoctoproject.org/main/builders/nightly-arm-lsb/builds/109/steps/BuildImages/logs/stdio :( Mar 21 11:01:57 bluelightning: looks like its looking for x11 :/ Mar 21 11:02:12 " warning: library search path "/usr/X11R6/lib" is unsafe for cross-compilation" Mar 21 11:02:16 argh Mar 21 11:02:22 ok I will look into it Mar 21 11:03:14 * bluelightning looks forward to the day we stop supporting qt3 Mar 21 11:04:15 bluelightning: open an enhancement for 1.7 proposing we drop it regardless of lsb? :) Mar 21 11:04:15 is there a specific reason why python 2.x is at 2.7.3 and not at a more recent bugfix version? (2.7.5 or 2.7.6) or just because nobody stepped up for the upgrade? Mar 21 11:04:30 RP: I'm not sure we could really do that Mar 21 11:04:42 bluelightning: I'm beginning to think we could Mar 21 11:05:09 diego_r: coincidentally I looked at upgrading it the other day, it's a bit of a pain Mar 21 11:07:58 what is this qt3 thing you speak of? Mar 21 11:08:16 Crofton|work: an annoying piece of baggage we carry around solely for LSB purposes Mar 21 11:08:27 Hai, am cross compiling libnfc-1.7.0 getting link error lsusb-1.0.la . Here is log http://pastebin.com/YyMCMi63. Mar 21 11:08:30 (in a separate layer though, at least) Mar 21 11:08:41 you have just defined LSV :) Mar 21 11:08:44 LSB Mar 21 11:08:56 I almost pooped when my mage had rpm in it Mar 21 11:09:38 diego_r: pain because of cleaning up backported patches and making everything build with it? Mar 21 11:09:57 diego_r: that's the start of it yes Mar 21 11:10:23 diego_r: you can see my progress at http://git.yoctoproject.org/cgit/cgit.cgi/poky-contrib/log/?h=paule/python276-wip Mar 21 11:10:31 it currently fails at do_compile Mar 21 11:10:43 looks like building mesa is b0rk3d for arm, at least since dora :( Mar 21 11:11:21 bluelightning: thanks for the pointer Mar 21 11:11:54 LetoThe2nd: how do you mean? Mar 21 11:13:23 bluelightning: it compiles fine, but then breaks do_rootfs because the mesa package would be empty and therefore is not created Mar 21 11:13:53 LetoThe2nd: that is odd... I'd be surprised if we didn't build any images for arm that contain mesa... Mar 21 11:15:18 bluelightning: i'd think so too, but thats the way it is here right now. building a plain poky for qemuarm, and just adding mesa to IMAGE_INSTALL breaks. Mar 21 11:15:37 LetoThe2nd: the mesa package is meant to be empty Mar 21 11:15:41 that's why it breaks Mar 21 11:15:50 there *isn't* a mesa package Mar 21 11:16:02 oh, hang on, it's back Mar 21 11:16:03 ffs Mar 21 11:16:13 FastFourierSelection ;) Mar 21 11:16:29 rburton: plus, it builds fine for qemux86 Mar 21 11:16:30 ok my master build has a single directory in Mar 21 11:16:43 LetoThe2nd: what are you trying to achieve by installing "mesa? Mar 21 11:16:51 if you want the libraries, install the libraries Mar 21 11:17:07 there are no binaries. at one point that package contained a configuration file ,but that's been deleted. Mar 21 11:17:53 rburton: i'm fine if picking "mesa" just was the wrong thing to do (TM), but then i still wonder why its fine on x86 Mar 21 11:26:17 LetoThe2nd: its not meant to exist at all, its probably a bug on x86 Mar 21 11:28:41 YAY! a condition where a successful build is actually a bug Mar 21 11:29:30 but yes, changing the install to mesa-demos seems to build fine. Mar 21 11:36:30 LetoThe2nd: depends what you want to do, but that certainly pulls in some of the GL stack Mar 21 11:37:22 rburton: just some tinkering on getting GL (in software) onto ARM Mar 21 11:38:03 LetoThe2nd: oh god the pain. hopefully llvmpipe will work and be usable. you'll need to pull i meta-oe for that. Mar 21 11:38:42 rburton: ah yes? Mar 21 14:21:13 Hey, how do i make pkg_postinst_${PN} run after mariadb has initialized it's database? Mar 21 14:29:06 My guess is that you don't because that means it is already installed Mar 21 14:30:00 mariadb? Mar 21 14:30:41 i have another recipe that need to run a script after the db has been created. it's not beeing created before the first boot. then after that i need to run a sql script againts the database Mar 21 14:31:15 but how would i tell bitbake that the postinst should run after the mariadb installation script Mar 21 14:32:14 isn't making your packaage depends on mariadb enough ? Mar 21 14:32:27 i guess the installation ordre will respect that Mar 21 14:32:35 order Mar 21 14:33:44 but will it do that on the initial boot also? and dose rdepends do the same as depends Mar 21 14:33:46 ? Mar 21 14:34:46 this is the pkg_postinst of mariadb Mar 21 14:34:47 http://pastebin.com/Z5XjMDeF Mar 21 14:38:55 your postinst can indicate that it's unable to be run at image creation time, and so must run at first boot Mar 21 14:39:22 if $D is set to a non-empty string ([ -n "$D" ]) then the script is being run at image creation time, not first boot Mar 21 14:39:30 and if the postinst fails at image creation time, it'll be run at first boot Mar 21 14:39:39 so if [ -n "$D" ]; then exit 1; fi Mar 21 14:39:45 in the postinst should get it postponed Mar 21 14:40:06 and the first boot configuration should run in dependency order, so as long as you depend on mariadb, it should be run after mariadb's postinst Mar 21 14:40:09 kergoth: I would suggest stopping to propagate if [ -n "$D" ]; then exit 1; fi Mar 21 14:40:45 it makes it impossible to expand the recipe with a bbappend Mar 21 14:41:12 or to use update-rc.d in a bbappend Mar 21 14:41:20 and that is quite difficult to track Mar 21 14:41:40 oh, yes, you're right, the logic should be reverted Mar 21 14:41:50 * kergoth shrugs Mar 21 14:42:05 thanks, il try to change it up Mar 21 14:42:22 I'v been wanting to remove it from all the recipes but I didn't take the time yet Mar 21 14:45:05 so if i got this right. i should do something like this in my bb http://pastebin.com/kE6Te2e4 Mar 21 14:59:08 is there any point in using depends and rdepends? Mar 21 14:59:13 together Mar 21 15:04:58 it's still running postinit script before mariadb's post init. Mar 21 16:27:07 khem: otavio: hi. I'm having problems running chrome (actually Chromium, but the executable is still called chrome) as a normal user. While I do have the CHROME_DEVEL_SANDBOX=/usr/sbin/chrome-devel-sandbox exported I still get the "Running without the SUID sandbox" message. Mar 21 16:27:57 I mean... It was working with Chromium 29, but doesn't work anymore in Chromium 35 Mar 21 18:06:11 bluelightning: ping? Mar 21 18:08:11 zecke: pong Mar 21 18:10:50 bluelightning: I have something weird and wonder if you heard about it. I'm using dora without a PR server. I changed the git revision of OpenGGSN. This project provides a library libgtp (libgtp0 package name) and some apps linking to this library Mar 21 18:11:16 bluelightning: after the re-build due the changed SRCREV the packages of the binary dependend on the _old_ version of libgtp0. Mar 21 18:17:24 zecke: that suggests the apps didn't properly rebuild Mar 21 18:17:38 there is a declared dependency between the app and the lib recipes right? Mar 21 18:37:30 bbl Mar 21 19:23:19 fray: ping - could you look at my TMPDIR definition and see if that last paragraph I inserted is okay? http://www.yoctoproject.org/docs/1.6/ref-manual/ref-manual.html#var-TMPDIR. Mar 21 20:08:41 hi Jefro Mar 21 20:15:02 is there any workaround for https://bugzilla.yoctoproject.org/show_bug.cgi?id=4102 ? Mar 21 20:15:03 Bug 4102: normal, Medium+, 1.7, richard.purdie, REOPENED , Incremental builds do not work well when renaming recipes or changing architecture. Mar 21 20:36:32 Here is an land mine one of our developers (not me) stepped on today. Mar 21 20:36:48 You can not build in a directory containing the string HOST Mar 21 20:37:12 The fetcher will not fetch properly if you have a premirror. Mar 21 20:37:19 So, I just built what was described here: https://www.yoctoproject.org/docs/1.5.1/yocto-project-qs/yocto-project-qs.html Mar 21 20:37:35 and it finished... but where exactly did it put the finished product? Mar 21 20:38:12 Hodapp: Generally speaking in the images/deploy dir. Mar 21 20:39:05 Which may be in tmp-eglibc/deploy/images Mar 21 20:39:12 Depending if you just used the defaults. Mar 21 20:40:16 ahh, there it is, in tmp/deploy/image/qemumips Mar 21 20:40:25 images* Mar 21 20:40:27 thanks Mar 21 20:41:29 If the docs are not clear you can always mention it to scottrif. We want to make sure first timers don't have issues :-) Mar 21 20:44:44 https://www.yoctoproject.org/docs/1.5.1/yocto-project-qs/yocto-project-qs.html does not once mention 'tmp' or 'tmp-eglibc' so perhaps some explanation is in order on where to find the built image Mar 21 21:24:27 Hi folks - quick very basic question, but I can't quite figure it out from the documentation. Is the 'nativesdk' package(s) the compilers/libraries/tools needed to build applications from *within* the root file system? Mar 21 22:33:16 wotte: I was talking with someone and the answer to that question is "no". nativesdk is for SDKs that run on a "third" (possibly same as the build host i.e. the "first") system. Mar 21 22:44:50 Thanks scottrif Mar 21 22:45:35 wotte: Just passing info on :) Mar 21 22:46:09 So if I want to build a SDK bundle that I can distribute to someone else that has the rootfs, compilers, kernel, whatever Mar 21 22:46:17 is the correct command "bitbake image -c populate_sdk"? Mar 21 22:46:47 wotte: correct Mar 21 22:48:12 brilliant, thanks folks. **** ENDING LOGGING AT Sat Mar 22 02:59:59 2014