**** BEGIN LOGGING AT Fri May 10 02:59:58 2013 May 10 03:47:31 hei May 10 05:11:45 groovy May 10 05:15:01 If I remove my u-boot files under deploy/, what would be the generic way if installing the files again, "bitbake -c deploy virtual/bootloader" ? May 10 05:15:23 s/if installing/of installing May 10 05:47:10 maybe clean first and then rebuild it May 10 06:54:35 <_dv_> JaMa: works well here, there are still todos, but these are more about cleanups (like replacing the populate_package_prepend with PACKAGESPLITFUNCS May 10 06:57:37 JaMa: the _git versions are also used for a bluetooth project by a colleague (he needs the gstreamer 1.0 bluez plugins, which are currently only in the gstreamer git repo) May 10 08:00:30 good morning May 10 08:16:18 this is kinda cool... http://theonion.github.io/fartscroll.js/ May 10 08:26:50 morning all May 10 08:28:19 morning all May 10 08:33:23 night May 10 08:47:25 morning # May 10 09:40:36 hi all May 10 09:41:48 hi pb_ May 10 09:52:46 hi pb_, bluelightning May 10 09:53:12 pb_: may I fire ou a quick question? May 10 09:54:36 hi ant_work May 10 09:57:36 ant_work: sure May 10 09:57:59 we could finally compile kexec-tools-klibc for x86/x86-64 May 10 09:57:59 now a doubt about this arch only: a couple of x86 files are including May 10 09:57:59 arch/i386/pic.c needing outb() and arch/i386/vga.c using inb() outb() and outw() May 10 09:57:59 somehow kexec-tools provides its io.h May 10 09:58:15 afaik one should generally use the package-provided headers, isn't ? May 10 09:58:55 specifically: May 10 09:58:58 http://git.kernel.org/cgit/libs/klibc/klibc.git/tree/usr/include/arch/i386/sys/io.h May 10 09:58:58 http://git.kernel.org/cgit/utils/kernel/kexec/kexec-tools.git/tree/purgatory/arch/i386/include/arch/io.h May 10 10:01:19 I wonder why kexec-tools is providing its own. May 10 10:01:32 as far as I can tell from a quick glance at the files, they are equivalent and either one should work. May 10 10:02:19 oh, except the klibc one doesn't have read*()/write*() May 10 10:02:29 so yeah, I guess you should use the one from kexec-tools May 10 10:02:47 you know, I'd prefer the klibc headers...it's just happened to me with ubi headers outdated by kernel headers :/ May 10 10:03:23 but here it's just about a couple of funcs, really May 10 10:03:38 yeah, and the contents of this file is very unlikely to change. May 10 10:03:57 fine then, thx May 10 10:04:03 so I think it doesn't really matter all that much. you could use the klibc one if you wanted but you might have to patch in the bits it doesn't have. May 10 10:05:41 see, I have anyway to patch: with our flags (i.e. nostdinc) the is not found May 10 10:06:31 it's just easier to do May 10 10:06:33 -#include May 10 10:06:37 +#include "include/arch/io.h" May 10 10:06:59 and use the kexec-tools header, definitely May 10 10:07:39 heh, I wouldn't have suspected x86 has those "extras" ;) May 10 10:53:04 hi, could someone help me with some basic oe/bitbake questions May 10 10:55:45 just ask May 10 10:56:24 hi, i want to do bitbake openjdk-7jre May 10 10:56:35 lol May 10 10:56:38 yes May 10 10:56:41 but i want to make some changes to the package before it buids May 10 10:56:46 I am the maintainer of meta-java May 10 10:57:04 sorry for you May 10 10:57:16 if i do bitbake it trys to build cacao and jamvm May 10 10:57:17 oh lol May 10 10:57:18 its ed May 10 10:57:29 stupid me May 10 10:57:37 so i remove them from the common.inc May 10 10:57:48 and then redo the bitbake May 10 10:57:52 ednevill hm you do not want the libjvm from cacao and jamvm? May 10 10:58:24 but it still rebuilds jamvm/cacao May 10 10:58:52 so how can I pull the sources, make the changes before doing the bitbake May 10 10:59:26 obviously bitbake does not clean up properly when you redo the bitbake? May 10 11:01:01 is there a 'deep clean' option that doesn't remove the source May 10 11:01:01 ednevill again whats the goal you want to achieve? May 10 11:01:25 i am trying to build openjdk-7-jre for ARM AARCH64 May 10 11:01:32 ah okay May 10 11:01:52 bitbake -c cleansstate openjdk7-jre May 10 11:02:02 ok, thx will try that May 10 11:02:13 than change the configure option May 10 11:02:26 need to do a rebuild, might be some time:-) May 10 11:03:43 hi ednevill May 10 11:05:09 hi hrw May 10 11:07:07 ednevill ah I looked something uo May 10 11:07:21 you can set in local.conf WITH_ADDITIONAL_VMS="" May 10 11:07:42 and than you should be rid of cacao and jamvm too May 10 11:07:52 hi hrw May 10 11:08:17 so I make a wild guess linaro hired ed to make openjdk stuff working on aarch64 May 10 11:08:37 ednevill: meta-linaro/meta-aarch64/recipes-core/openjdk/openjdk-7_03b21-2.1.8.bbappend already has WITH_ADDITIONAL_VMS_aarch64 = "" set May 10 11:08:52 woglinde: or one of member companies May 10 11:09:02 hrw ah hehe May 10 11:09:27 good wild guess, watch this space! May 10 11:09:35 http://www.linaro.org/linux-on-arm/meet-the-team/edward-nevill/ - Linaro directly May 10 11:11:06 and good my hard unpaid work is usefull May 10 11:15:42 lunch now till later May 10 11:16:42 wogline: thx for your help, will keep ou update, ciao May 10 13:13:54 hello, I am having a issue with the recipes using autotools, and need to add an option to the LDFLAGS for all of them.. May 10 13:14:25 Does anyone know how to do that? May 10 13:40:32 I add a layer, bitbake see it, see .append, but not .bb ... I forget something? May 10 13:43:13 silvio_: what is the BBFILES value in the layer's conf/layer.conf? May 10 13:45:43 has anyone looked at building the gator module for the ARM DS5 debugger thing? May 10 13:51:12 bluelightning, BBFILES += "${LAYERDIR}/recipes*/*/*.bb ${LAYERDIR}/recipes*/*/*.bbappend" May 10 13:51:48 silvio_: looks correct to me... May 10 13:52:21 bluelightning, maybe I understand... wrong subdir May 10 14:01:52 bluelightning, ok discovered, maybe i found a bug, or issue on "networkmanager_0.9.8.0.bb" May 10 14:02:28 probably first I put the recipe in the incorrect directory May 10 14:02:53 after when try bitbake -p a receive this error: May 10 14:03:18 /home/oe-core/build/../stuff/meta-btw/recipes-connectivity/networkmanager/networkmanager_0.9.8.0.bb inherits systemd but doesn't set SYSTEMD_PACKAGES May 10 14:04:01 adding line " SYSTEMD_PACKAGES = "${PN}-systemd" " @ recipe I solved May 10 14:08:50 silvio_: you're using meta-systemd I take it? May 10 14:15:04 bluelightning, yes May 10 14:16:01 silvio_: right, that's probably what causes that problem May 10 14:16:17 silvio_: you should not need meta-systemd anymore (in most cases) May 10 14:16:30 assuming you are building from master or dylan that is May 10 14:19:17 bluelightning, I m building from the customized distro HW producer, based on Angstrom and OE commit "ae9dbd0e1e26ba2b35cbd08ec731aee62adedc23" May 10 14:43:02 for what it is worth, seems like libtool: libltdl/m4/libtool.m4 can do it.. May 10 14:43:24 and BUILD_CFLAGS with -Wl May 10 16:37:04 jeroen_: we have tried it on OSX May 10 16:37:22 so not a complete non starter May 10 16:39:23 debugging a pathological redis config... May 10 16:47:23 khem: what did you try? oe or trying to fiddle with the options for ld? May 10 16:48:23 khem: in my case ld need -zorigin for the $ORIGIN to be supported by ldrt May 10 18:12:06 I am still a bit in the dark how linux-libc-headers is supposed to work... May 10 18:13:02 It complains it cannot find the cross compiler (not sure it that is fatal) May 10 18:14:42 (the cross compiler is in armv7a-vfp-neon-bpp3-linux-gnueabi.gcc-cross-initial but not yet in the final directory), so that is not so suprising.. May 10 18:18:39 it fails at scripts/Makefile.build:44: SCCS: No such file or directory (which seems like a sed error).. May 10 18:19:01 so I tried to find the main repro of it, but that is unclear to me.. May 10 18:30:43 mmm seems like a complete linux tarball, mm May 10 18:49:50 can anyone confirm above, is a native linux make needed to install the cross compile headers? May 10 19:14:07 jeroen yes glibc needs the linux kernel headers May 10 19:17:12 somebody mentioned special EGL support in dylan, to avoid the issue with mesa and vendor-specific 3D drivers both providing virtual/egl May 10 19:17:25 which produces the error "ERROR: Multiple .bb files are due to be built which each provide virtual/egl" May 10 19:17:39 what special support is that, and how do I use it? May 10 20:04:06 jeroen_: you shoud install gnu version of tools May 10 20:04:32 jeroen_: freebsd utilites have subtle differences that will be a big headache May 10 20:05:05 basic core tools like awk/sed/ and friends May 10 20:08:13 khem: well, that is a matter of taste, where possible things should be POSIX complient... May 10 20:08:49 khem: I don't really like the GNU lock in.. May 10 20:09:12 hmm. to detect if a package is empty, I currently see if there is a FILES_${PN} variable, and whether or not it is empty. if both are true -> package is empty. May 10 20:09:16 subtle differences = 'we rely on gnu extensions' May 10 20:09:16 is there a better, cleaner way? May 10 20:09:47 (I mean, when I want to know if a package is empty in a python function) May 10 20:12:04 Daemon404: what gnu extensions do you rely on? May 10 20:13:14 *I* dont work on oe (much) May 10 20:13:24 just a general tl when people say bsd utils are different May 10 20:13:32 s/tl/translation/ May 10 20:15:20 Deamon404: mmm, difference are small, remember that linux tries to be a polite unix citizen May 10 20:16:06 jeroen_: with the possible exception of favoring /proc over calls like getrusage() May 10 20:16:21 dv_: haha May 10 20:16:39 jeroen_: if you wanna piss people off, ask about that in #posix :) May 10 20:17:21 dv_: correct, sudo mount -t linprocfs linproc /proc ;) May 10 20:19:47 dv_: I am not trying to piss off people, I just want things to work... May 10 20:20:15 jeroen_: sure, just saying that this has been the reason for some flamewars. :) May 10 20:20:17 <_rink> fwiw, freebsd has gnu tools with a g prefix if you have the correct port May 10 20:20:19 <_rink> gawk etc May 10 20:21:39 _rink: thanks, I know May 10 20:21:56 _rink: and autotools will find them etc May 10 20:22:52 _rink: so this is not a big change, it only takes a lot of time to figure out what needs to be changed... May 10 20:22:57 * _rink nods. May 10 20:23:12 <_rink> Well, I'm a longtime FreeBSD user and would appreciate the effort - just not use it :) May 10 20:30:41 I do cheat a bit btw, I use sed -i -e ... May 10 20:31:02 anyone aware of a platform not supporting the -i ? May 10 20:31:38 mac pleople, do you have sed -i? May 10 20:32:37 google says you have... May 10 21:20:40 hi there May 10 21:21:02 I'm trying to include tftp-hpa into omap3-console-image.bb for overo May 10 21:21:18 it's not picking up the tree of the image directory under the recipe build area May 10 21:21:21 am I doing something wrong? May 10 21:22:13 jakeman: the image directory is everything that has been installed within do_install May 10 21:22:34 jakeman: it then gets split into various packages, you can see the result of that under packages-split within the work directory May 10 21:22:43 thanks very much. ! May 10 21:22:49 jakeman: so the key is to install the packages containing the files you want May 10 21:23:24 aha....so how do I refer to the particular package from the omap3-console-image.bb image file? May 10 21:24:31 I SEE May 10 21:24:41 I need tftpd-hpa May 10 21:24:46 not tftp-hpa May 10 21:24:51 which is the package I'm looking for May 10 21:25:05 depends which tftp-hpa recipe you're building May 10 21:25:24 for the current one in meta-networking it's called tftp-hpa-server May 10 21:26:08 I'm building tftp-hpa_5.0.bb May 10 21:26:43 I suspect I want to call out tftpd-hpa in my image file May 10 21:27:21 thanks much for your help May 10 21:27:41 np May 10 21:35:41 can oe_runconf_prepend() be replaced by do_configure[prefuncs] ? May 10 21:36:19 dv_: why do you want to do that? May 10 21:36:54 I want to move away from _prepend functions May 10 21:37:16 oe_runconf_prepend() is being used here: https://github.com/openembedded/oe-core/blob/master/meta/recipes-multimedia/gstreamer/gst-plugins.inc May 10 21:38:14 dv_: honestly I think your push to do that isn't the right course May 10 21:38:36 hello there May 10 21:38:49 bluelightning: I thought prefuncs etc. are preferred over _prepend, because it is cleaner (it does not simply attach a piece of code) ? May 10 21:39:18 dv_: if they are that's news to me, we use a lot more prepends/appends than prefuncs/postfuncs May 10 21:39:23 hi eren May 10 21:39:34 bluelightning: how is layers.openembedded.org updated? May 10 21:39:39 is there a cron job? May 10 21:39:43 bluelightning: I guess it is because of the indentation issue May 10 21:39:50 eren: there is, it runs about every three hours May 10 21:39:52 bluelightning: also because of possible shellscript/python mixups May 10 21:40:04 dv_: just use separate branches for that May 10 21:40:18 bluelightning: okkie, it pulls git repository and processes them, then? May 10 21:41:33 eren: it does... you can see the code here if you're interested: https://git.yoctoproject.org/cgit/cgit.cgi/layerindex-web/ (the update script is layerindex/update.py) May 10 21:41:49 bluelightning: thanks May 11 01:37:09 * nerdboy waves bye-bye to the volkswagen for the last time **** ENDING LOGGING AT Sat May 11 02:59:58 2013