**** BEGIN LOGGING AT Mon Sep 07 02:59:57 2009 Sep 07 03:05:10 i rm all trace of it Sep 07 03:05:24 it is not even in checksums.ini anymore Sep 07 04:56:22 anyone know of a local mta/mda which is good for local delivery (cron outputs and the like) ? Sep 07 04:56:32 ssmtp and esmtp don't seem to fit the bill Sep 07 04:56:43 i might try the development version of esmtp, which supposedly does local delivery Sep 07 04:56:52 was wondering if there were any other recipes that can do that Sep 07 05:00:37 maildrop? Sep 07 05:47:13 hmm . . . gpe recipes are failing in fetch because http://gpe.linuxtogo.org/download/source/ has permission issues :-( Sep 07 06:25:24 Goodmorning Sep 07 06:29:08 In the manual is said that the ROOTFS_POSTPROCESS_COMMAND can be used to change e.g. the roots password Sep 07 06:30:29 But what would be the appropriate way to do that? Just by changing / patching the default '/etc/passwd' or is there a better way? Sep 07 06:43:18 good morning Sep 07 06:53:02 tsjsieb, image.bbclass contains a zap_root_password function which is used in many of the image recipes' as a ROOTFS_POSTPROCESS_COMMAND Sep 07 06:54:51 I've seen that one indeed. But what If I want e.g. to add another user to the image. Sep 07 06:56:26 Could / should I do that to, using this command? (it's for a custom image, not to add to oe) Sep 07 06:57:34 Ass I want to run an app on quite a lot of machines wich will be installed with the same image Sep 07 06:57:50 (and I prefer not to run the app under the root user) Sep 07 06:58:14 gpe-image-blank-user.bb:ROOTFS_POSTPROCESS_COMMAND += "adduser -D user" Sep 07 06:58:25 * grg shrugs Sep 07 06:59:04 o.k. thanks I will see if that works Sep 07 06:59:21 there are a few examples in the various image recipes Sep 07 06:59:44 fso-image.inc has a function with bunch of random shit in it Sep 07 07:00:32 Thanks I grepped for the command, but I missed both images your pointing out (shame on me) thnx! Sep 07 07:27:43 grg: the adduser command doesn't seem to work in ROOTFS_POSTPROCES_COMMAND (and I think the format '-D user' is the one of the busybox adduser tool, so it seems logical that that one can not be used during image creation). I think I will test with patching the default 'passwd' file, although it doesn't look to me as the most beautifull way to do it. Sep 07 08:06:51 re Sep 07 09:18:02 morning Sep 07 09:20:40 hey hrw Sep 07 09:28:19 morning all Sep 07 09:28:27 broonie: I've not fogotten, don't worry! Sep 07 09:28:37 Good, good :) Sep 07 09:31:28 I'd really like to get them in for 2.6.32 if possible. Sep 07 10:48:48 hey, could you please review my imx patches ? http://github.com/tardyp/oe-imx/commits/ Sep 07 10:49:44 I'm particularly interrested in how to include recipes of binary proprietary software :-/, and wether they will be accepted in mainline. Sep 07 10:50:38 I think there is some precedent for wrapping proprietary binaries in .bbs, see the adobe flash bits and some of the TI code. Sep 07 10:50:45 tardyp: we already have some propertiary ones - ti ones for example Sep 07 10:50:45 So, that shouldn't be an issue in itself. Sep 07 10:51:41 tardyp: I do not like kernel recipe Sep 07 10:51:49 tardyp: we prefer to keep kernel configs in metadata Sep 07 10:52:20 tardyp: and instead of DEF_PREF=-1 use COMPATIBLE_MACHINE var Sep 07 10:54:29 hrw, ok, I took DEF_PREF=-1 from another kernel reciepe, but made the correct thing for gpu drivers Sep 07 10:56:32 tardyp: add URL to machine homepage in configs Sep 07 10:57:25 tardyp: imx35 is 1136? not 1176? Sep 07 10:58:38 hrw, I did not verify that. probably. Sep 07 10:58:44 python do_package_qa () {} is what for? Sep 07 10:59:09 QA is complaining that there is no GNU_HASH into libs. Sep 07 10:59:21 add note about it to recipe? Sep 07 10:59:32 good poing Sep 07 10:59:34 point Sep 07 10:59:36 otherwise you will have to answer it again Sep 07 10:59:57 tardyp: basically if you do something not standard then add info to recipe Sep 07 11:00:20 we need to standarize SRC_URI for such tarballs btw Sep 07 11:02:35 ti's codec bb ask user to put it on files/ dir. Sep 07 11:11:23 perl needs another set of changes ;( Sep 07 11:16:58 hrw: perl will allows need kludged I think Sep 07 11:17:53 perl-doc contains important files.. Sep 07 11:19:04 for the GNU_HASH thing, you might consider adding a switch to turn off the gnu hash checking specifically, rather than inhibiting the whole qa pass. Sep 07 11:19:19 irssi-perl on sheevaplug even works ;D Sep 07 11:19:24 and, if you must inhibit the whole qa, it might be neater to set INSANE_SKIP_${PN} rather than redefining do_package_qa(). Sep 07 11:22:04 a couple of other things about the amd-gpu recipe: (i) the definitions of DEPENDS and S look redundant; (ii) the definition of PROVIDES is probably wrong, I suspect it should be PROVIDES_${PN}; (iii) putting the whole of ${libdir} into FILES_${PN} looks a bit odd, though maybe this is correct; (iv) setting OE_ALLOW_INSECURE_DOWNLOADS here is a bit unwholesome: is the code so proprietary that even its md5sum must be kept secret? Sep 07 11:22:48 I doubt any of freescale's competitors will try to reverse-engineer the tarball by making up random bitpatterns until they hit on the matching md5sum :-} Sep 07 11:23:25 ;D Sep 07 11:24:26 oh, and (v) your hard-coded references to "usr/lib" and suchlike should probably be ${libdir} and so on. Sep 07 11:24:39 otherwise, you will lose with distros where prefix != /usr Sep 07 11:37:26 silly question, are the recipes tested automatically? Sep 07 11:37:33 (ones in master say) Sep 07 11:44:51 I don't thinks so (and I think it would practically be impossible to test every recipe for every different machine and architecture) Sep 07 11:46:18 well, I didn't mean all possible combinations Sep 07 11:46:41 and obviously build-test wouldn't verify that the build actually runs. Sep 07 11:46:53 just thinking aloud. Sep 07 11:50:11 Some people are running buildbot to test some basic images from scratch. Sep 07 11:56:33 but nothing coordinated? Sep 07 11:57:01 pb_, I'm back from lunch Sep 07 11:57:41 about OE_ALLOW_INSECURE_DOWNLOADS, I wasn't sure what is the procedure to add things in checksum.ini Sep 07 11:59:11 about PROVIDE, probably there is a need for something like virtual/libgles? Sep 07 11:59:47 yes, probably Sep 07 12:01:13 for checksums.ini, you basically just patch it like any other file Sep 07 12:01:47 pb_, is it sorted? Sep 07 12:02:03 yes, there's a script somewhere to do that (or you can just insert the entry in the right place by hand) Sep 07 12:02:28 pb_, I suspected that. Sep 07 12:04:32 tardyp: contrib/source-checker/ is dir to check Sep 07 12:04:40 I just saw it. Sep 07 12:05:07 Hi all, I've built a rootfs for mini2440 and my mplayer crashes with http://pastebin.ca/1557075 , is there a solution of that? Sep 07 12:05:12 czr: I think it is coordinated, but you should ask core-devs Sep 07 12:07:16 about usr/lib: that is the path in the fsl tarball. Sep 07 12:23:37 pb_, I dont understant your remark about PROVIDES_${PN} Sep 07 12:24:19 tardyp: yourrecipe-doc will also have PROVIDE... Sep 07 12:24:35 ok I see Sep 07 12:25:04 I will also need advice on how to provide libraries links correctly. Sep 07 12:25:14 I get a bunch of: Sep 07 12:25:16 NOTE: Couldn't find shared library provider for libgsl.so Sep 07 12:29:12 tardyp: oh right, I see. that's fine then. Sep 07 12:29:44 anyone sucessfully runs awt/swing on beagleboard? I'm able to run headless java applications but no awt/Swing apps. It complains about a gtkpeer lib is missing. Is there any extra recipe I need to build to make awt/swing working? Sep 07 12:31:07 pb_, what about my "Couldn't find shared library provider" messages? Sep 07 12:34:54 dunno, what package is libgsl.so in? Sep 07 12:34:59 is that part of this same thing? Sep 07 12:35:28 yes, there is a bunch of libs in the driver Sep 07 12:36:11 it sounds like package.bbclass isn't managing to extract the shlibs data from that file for some reason. does it have a correct SONAME? Sep 07 12:37:29 I can't see there is any Gtk-peer recipe.. Sep 07 12:38:11 bbiab, lunchtime for me now Sep 07 12:40:42 how to compile a gtk code using OE ? Sep 07 12:52:21 sx you have to build a recipe, you can base yours on another program using gtk. what program do you want to compile? Sep 07 13:24:07 re Sep 07 13:29:50 gm Sep 07 13:30:01 How can I found out *why* a recipe is skipped during scanning? Sep 07 13:32:18 Ah yes, COMPATIBLE_MACHINE needs a regex rather than space seperated list. --- but my question still stands for future issues. Sep 07 13:32:22 hi vivijim Sep 07 13:42:38 03Koen Kooi  07org.openembedded.dev * r4f2d73c53a 10openembedded.git/recipes/angstrom/ (angstrom-uboot-scripts.bb angstrom-uboot-scripts/pico.cmd): Sep 07 13:42:38 angstrom-uboot-scripts: add recipe to generate .scr files easily Sep 07 13:42:38 * the recipe puts them both in deploy and in ${datadir}/u-boot-scripts Sep 07 13:43:14 hi likewise Sep 07 13:49:56 likewise: -DD? Sep 07 13:50:00 likewise: for bitbake Sep 07 13:51:21 hrw: doesn't mention *why* it is skipping the recipe Sep 07 13:51:29 ok Sep 07 13:51:51 for newbies, it's hard to go from there./ Sep 07 13:52:11 most people do not even see *that* the recipe was skipped Sep 07 14:00:01 pb_: any new findings about endianess / BASE_PACKAGE_ARCH / HOST_ARCH (was armv5te is now armv5teb) ? Sep 07 14:00:31 ant_work: no, I haven't looked at it. as I said the other day, I think you need to take this up with khem. Sep 07 14:00:44 afaict, his patch broke it and he should be the one to fix it. Sep 07 14:01:05 ok, I was afk for a while, will try to catch hime :) Sep 07 14:02:02 hrw, pb_ PROVIDES += "virtual/vg" work, but with PROVIDES_${PN} += "virtual/vg", bitbake says ERROR: Nothing PROVIDES 'virtual/vg' Sep 07 14:04:11 tardyp: oh, sorry, forget what I said about PROVIDES Sep 07 14:04:23 I obviously wasn't thinking straight then. PROVIDES = ".." is correct Sep 07 14:09:48 mickeyl: good morning Sep 07 14:13:09 morning pb__ Sep 07 14:17:08 03Ghislain MARY  07shr/import * r6f51eb318e 10openembedded.git/recipes/freesmartphone/misc-vapi_git.bb: Sep 07 14:17:08 misc-vapi depends on alsa-lib Sep 07 14:17:08 recipes/freesmartphone/misc-vapi_git.bb | 1 + Sep 07 14:17:08 1 files changed, 1 insertions(+), 0 deletions(-) Sep 07 14:17:10 03Thomas Zimmermann  07shr/import * r0a8e30894c 10openembedded.git/shr-unstable/openembedded/recipes/brickshooter/ (brickshooter_0.04.bb brickshooter.desktop readyforfr.patch): Move Brickshooter reciepe to right location Sep 07 14:17:14 03Julien 'Ainulindale' Cassignol  07shr/import * r56a4da6dc3 10openembedded.git/recipes/freesmartphone/misc-vapi_git.bb: Sep 07 14:17:17 misc-vapi: Bump PR. Sep 07 14:17:19 Signed-off-by: Julien 'Ainulindale' Cassignol Sep 07 14:17:22 03Ghislain MARY  07shr/import * rcc78e7587b 10openembedded.git/recipes/shr/shr-theme-niebiee.bb: Allow shr-theme-niebiee to be empty so that the metapackage is created. Sep 07 14:17:26 03Sebastian Krzyszkowiak  07shr/import * r4031840719 10openembedded.git/recipes/openmoko-projects/paroli_git.bb: paroli: add tichy stuff to package Sep 07 14:18:23 03Thomas Zimmermann  07shr/import * r0b1120a80e 10openembedded.git/recipes/ (4 files in 2 dirs): Cherry pick python-manifest-2.6.inc and python.inc from .dev (fixes #549) Bump PV of python-bluez and add to feed (fixes #621) Sep 07 14:25:19 morning Sep 07 15:03:40 tardyp, ping Sep 07 15:04:31 sx, pong Sep 07 15:05:06 tardyp, it is enough to see how to compile a gtk hello word program using OE. I've looked some recipes but they are so complicated but I read some documents for that Sep 07 15:05:36 tardyp, above, I'd asked a question for compiling gtk apps with OE Sep 07 15:06:12 sx please tell what is your program, is it autoconf or plain makefiles, etc. Sep 07 15:07:16 tardyp, my program is only a .c file which contains gtk buttons and uses mplayer externally Sep 07 15:09:10 actually when I compile it using armv4t's arm-angstrom-gnueabi-gcc compiler I don't get errors but on the mini2440 board says "illegal instruction" and doesn't work Sep 07 15:09:20 sx Then you might want to read this: Sep 07 15:09:23 http://thread.gmane.org/gmane.comp.handhelds.openembedded/26156 Sep 07 15:10:35 because of that I think to compile it using OE, because the gtk applications in the x11 image work.. Sep 07 15:11:23 tardyp, no, I know this. I'm searching something for gtk exactly. for example I've done this for QT, Sep 07 15:11:39 the .bb file contains something like "inherit qt4" Sep 07 15:11:49 and everything is ok Sep 07 15:12:49 sx: used -mcpu=arm920t? Sep 07 15:13:11 hrw, armv4t, s3c2440 Sep 07 15:13:21 sx you just need a basic makefile like the one kergoth proposed for helloworld Sep 07 15:13:35 the update the makefile to use gtk+ via pkg-config Sep 07 15:14:47 tardyp, I've tried some things but I couldn't... is not there a simple example? Sep 07 15:15:23 the simple example is here: http://thread.gmane.org/gmane.comp.handhelds.openembedded/26156 Sep 07 15:16:28 tardyp, but there is no something with gtk, right? Sep 07 15:17:00 you just have to add "`pkg-config --cflags --libs gtk+`" at the end of the CC cmd line in the makefile Sep 07 15:17:59 tardyp, ok, I'll try this. Sep 07 15:18:21 hrw, /home/eray/OE/angstrom-dev/cross/armv4t/bin/arm-angstrom-linux-gnueabi-gcc src/gtk.c -o gtk `cat /home/eray/OE/libsv4.txt` Sep 07 15:18:37 hrw, I compile it such... Sep 07 15:19:11 then add -mcpu=arm920t and then try on board Sep 07 15:19:21 ok Sep 07 15:20:50 Alignment trap: gtk (1776) PC=0x408133c4 Instr=0x280069c0 Address=0xffffffff FSR 0x813 Sep 07 15:20:50 Segmentation fault Sep 07 15:20:57 I got this Sep 07 15:21:17 I have to leave now the office, tomorrow I'll write here again for this subject Sep 07 15:21:21 not illegal instruction this time Sep 07 15:21:28 thanks for everything Sep 07 15:21:31 hrw, yes Sep 07 15:21:38 I'll care about this Sep 07 15:21:40 thanks Sep 07 15:47:24 ant_work: the armv5te and armv5te recursion problem was solved by passing d,1 to the function Sep 07 15:47:31 ant_work: it should be ok now. Sep 07 15:47:37 what issue are you seeing Sep 07 15:50:07 uf. Sep 07 15:50:16 so finally sheevaplug became my irc machine Sep 07 15:50:31 hmm nice hrw Sep 07 15:51:05 but I had to build irssi by hand on it - perl in OE suxx Sep 07 15:53:17 hi Sep 07 15:56:32 khem: great Sep 07 15:56:44 hi woglinde Sep 07 15:57:55 hrw: then fix perl in OE ;-) Sep 07 15:58:25 hm I finally managed to get kernel-package working with ccache and distcc Sep 07 16:01:38 khem: I built 2 days ago...sure the issue isn't actual? Sep 07 16:03:32 RP: I do not think that it is worth time Sep 07 16:04:02 RP: it looks like perl authors do not even consider cross compilation of libperl using apps Sep 07 16:17:02 auch.. my build took >70GB Sep 07 16:18:29 disk space is cheap Sep 07 16:18:45 Discount for bulk purchase! Sep 07 16:18:55 hrw: this sucks Sep 07 16:19:06 that's too much imho Sep 07 16:19:26 for openssl with ansgtroem you need around 1.6 gig Sep 07 16:19:34 temp sapce Sep 07 16:20:19 florian: thats one month of building for few archs without rm_work Sep 07 16:20:22 affordable Sep 07 16:20:31 80GB now and still counting Sep 07 16:20:58 and like woglinde said - disk space is cheap Sep 07 16:21:01 hrw: ah ok, that's somthing different... Sep 07 16:22:42 wow.. 500GB hdd is now ~5€ Sep 07 16:22:49 shit. broken encoding Sep 07 16:22:52 50eur Sep 07 16:22:56 ą Sep 07 16:23:06 need to fix sheeva environment... Sep 07 16:23:33 hello i am trying to start opie on a fbcon only angstrom rootfs. does opie need a xserver to run ? when i start opie with this command here "/etc/init.d/opie start" i am getting as a result a black screen with this output here Sep 07 16:23:35 http://pastebin.com/m51b24794 Sep 07 16:23:58 whats wrong ? Sep 07 16:24:51 opie dont need a xserver Sep 07 16:25:02 its running on framebuffer directly Sep 07 16:25:40 looks like normal startup Sep 07 16:25:49 xperia: login remotely and run "qpe -nodaemon" Sep 07 16:26:18 hrw hm wasn there the -qws option too? Sep 07 16:26:27 woglinde: thanks a lot for the clarification. in this case do i need some special configsettings for the linux kernel or do i miss some libs. Sep 07 16:26:38 as hrw said Sep 07 16:27:05 hrw: i will test in a minute. thanls for the answer Sep 07 16:28:04 hm someone knows a tool which looks up your current system and makes a minimal kernel config? Sep 07 16:31:30 hrw: thats the output of "qpe -nodaemon" http://pastebin.com/m3b1daef3 Sep 07 16:31:36 ant_work: Did you see any problems when you built Sep 07 16:31:59 yes, all was for armv5teb Sep 07 16:32:05 I am trying to build native-sdk-console-image for micro but it always builds console-image Sep 07 16:32:14 xperia: line 31,32 says the problem Sep 07 16:32:17 ant_work: which machine Sep 07 16:32:21 c7x0 Sep 07 16:32:24 le Sep 07 16:33:32 hrw: hmm what does this mean. that the tmp/folder is not writable ? Sep 07 16:33:37 khem: bb.data.getVar magic ? ^_^ Sep 07 16:33:56 uh Sep 07 16:33:58 or is a folder missing in /tmp Sep 07 16:34:06 what system dont have semaphores? Sep 07 16:36:04 hrw: the folder exist in /tmp so it isnt a permission problem. what could be this for a problem ? Sep 07 16:36:06 QWSApplication::qt_init() - current terminal specified. Sep 07 16:36:07 Cannot create semaphore /tmp/qtembedded-0/QtEmbedded-0 'd' Sep 07 16:36:09 Error 38 Function not implemented Sep 07 16:36:47 no idea Sep 07 16:36:58 I touched opie last time over 2 years ago Sep 07 16:37:30 khem: btw TARGET_OS triple... Sep 07 16:38:10 khem: can you also solve all-angstrom-linux all-angstrom-linux-gnueabi tmp/work dirs? they should be same iirc Sep 07 16:39:32 hrw: http://lists.trolltech.com/qt-interest/2002-09/thread00968-0.html Sep 07 16:40:00 khem: btw I was building angstrom/glibc Sep 07 16:40:29 xperia: second answer is for you Sep 07 16:41:36 hrw: thanks a lot . now how to enabling semaphores in the kernel ! isnt this by default enbaled ? hmmm Sep 07 16:41:51 xperia: I do not know your kernel Sep 07 16:41:57 morning Sep 07 16:42:06 and opie on xperia x1? waste of device Sep 07 16:43:18 hrw: yes I know that problem Sep 07 16:43:36 ant_work: hmmm c7x0 doesnt seem to include a tune file Sep 07 16:43:45 hrw: just want to try it. what for a gui are you running if i may ask ? Sep 07 16:43:59 khem: c7x0 include includes/zaurus.inc which should take tune-* Sep 07 16:44:13 xperia: usually something x11 based: gpe or sato Sep 07 16:45:54 hrw: hmm seems not Sep 07 16:46:13 o.. Sep 07 16:46:20 it includes zaurus-2.6.inc Sep 07 16:46:43 which includes zaurus-kernel.inc Sep 07 16:46:55 and initramfs-kexecboot.inc Sep 07 16:47:03 khem: indeed.. bug Sep 07 16:47:28 xperia qt has no phone functions Sep 07 16:47:46 hrw: gpe would be the next thing for trying. i didnt find a recent rootfs with gpe. if i chose a x11 build from here do i get then the gpe gui ? Sep 07 16:47:59 commit e8a60c1e518ddfeb5b5b913c3631b5daa56f6dad Sep 07 16:47:59 Author: Rolf Leggewie Sep 07 16:48:03 that dropped it Sep 07 16:49:08 ant_work: so you have to fix zaurus configs (again) Sep 07 16:49:20 hrw: may be it should include zaurus-clamshell.inc Sep 07 16:49:23 I dont know Sep 07 16:49:41 khem: the only change I see is Sep 07 16:49:45 -BASE_PACKAGE_ARCH = "${@['armv5teb', 'armv5te'][bb.data.getVar('SITEINFO_ENDIANESS', d, 1) == 'le']}" Sep 07 16:49:46 +BASE_PACKAGE_ARCH := "${@['armv5teb', 'armv5te'][bb.data.getVar('SITEINFO_ENDIANESS', d, 1) == 'le']}" Sep 07 16:50:10 ok, I revert it Sep 07 16:50:18 but grepping around I discovered the value are changed by cross.bbclass Sep 07 16:50:28 khem: c7x0 included zaurus-2.6.inc and then zaurus-clamshell Sep 07 16:50:37 perhaps the := evaluates in awrong moment... Sep 07 16:50:56 ah...wait Sep 07 16:51:21 hrw: aren't the values inverted? Sep 07 16:51:36 armv5te -> le Sep 07 16:51:40 armv5teb -> be Sep 07 16:51:56 maybe are Sep 07 16:52:01 if that endianess check fails... Sep 07 16:52:10 those values are correct. they're not inverted. Sep 07 16:52:26 this would be a surprise :) Sep 07 16:52:41 does it actually help if you revert that = -> := change? Sep 07 16:53:03 I suspect the real bug is probably elsewhere but it would be useful to know for sure. Sep 07 16:53:07 I just set a plain BASE_PACKAGE_ARCH = "armv5te" Sep 07 16:53:27 and the build was fine Sep 07 16:53:43 well, sure, but that doesn't help with understanding why it failed. Sep 07 16:53:52 yep, some oefatal Sep 07 16:54:03 I'll add som debug Sep 07 16:54:11 and let you know Sep 07 16:54:35 florian: what is ssh fingerprint of new LTG? Sep 07 16:54:41 hrw: without := it does not work Sep 07 16:54:49 I'll have a quick play with it myself. I think I have an 4ng5tr0m build tree lying around somewhere. Sep 07 16:55:07 hi ph5 Sep 07 16:55:24 hej woglinde Sep 07 16:57:14 hrw: without := python went into infinite recursion Sep 07 16:57:20 ok Sep 07 16:58:46 we need to look more into it Sep 07 17:00:34 ah, yes, the tune-xscale.inc thing loses because the siteinfo vars aren't defined at that point. Sep 07 17:00:50 assigning it with := might avoid recursion, but only because it will always give you None :-} Sep 07 17:01:26 so, that change definitely seems bogus. Sep 07 17:01:30 auch Sep 07 17:01:45 khem: if it can help, a failure log http://tinderbox.openembedded.net/public/logs/task/2814057.txt Sep 07 17:03:19 of course, the original code was a bit fragile too: it isn't terribly clever to accept any string other than "le" as meaning big endian. but, at least, it did previously work. Sep 07 17:07:37 pb__: quick catch ! Sep 07 17:11:44 bbl, bye Sep 07 17:22:26 opkg is nightmare Sep 07 17:22:36 you have to be root to be able to use it for anything... Sep 07 17:22:51 'opkg files PACKAGE' req root... Sep 07 17:28:42 why does http://amethyst.openembedded.net/~koen/narcissus/ dont work ? if i try to build a new x11 rootfs 70% of the packages are missed ! Sep 07 17:28:44 hrw, in gentoo you can do that sort of things if you are part of the 'portage' group Sep 07 17:28:49 is there no group associated with opkg? Sep 07 17:30:30 there is opkg-devel on googlegroups Sep 07 17:34:52 hrw: that should be easy enough to fix. I don't think there's any good reason why it should need root for that operation. Sep 07 17:34:58 how does it actually fail if you aren't root? Sep 07 17:35:08 bye till later Sep 07 17:36:55 pb___: unable to obtain lock Sep 07 17:37:02 I saw where it is in code Sep 07 17:37:42 anyway need to go Sep 07 18:00:32 http://www.angstrom-distribution.org/ down.. Sep 07 18:13:59 hrw|gone, i meant to say unix group, as in `groups; id -Gn` Sep 07 18:24:37 pb__: hmmm Would it make sense to change the evaluation order Sep 07 18:27:39 pb__: most processors these days support LE and BE so IMO endianless could be made as DISTRO feature Sep 07 18:28:06 s/endianless/endianness/ Sep 07 18:29:21 Some distros support both, even. Sep 07 18:30:13 mwester I meant MACHINE_FEATURE Sep 07 18:30:52 That seems to make sense to me. Sep 07 18:30:55 mwester: or we could define separate tune files for BE and LE machines Sep 07 18:31:21 That would get rid of some hackery I put in somewhere, IIRC. Sep 07 18:32:05 I seem to recall something went fishy with the SDK at some point, but I don't recall how (or if) I worked around it. Sep 07 18:33:10 I also think it will simplify understanding of how tune file is included Sep 07 18:33:23 Yep Sep 07 18:34:44 mwester: I am trying to generate native-sdk-console-image it compiles everything pulls in all deps correctly but the image file is still console-image Sep 07 18:35:06 it does not include the stuff that native sdk should have Sep 07 18:35:25 any idea how to debug it Sep 07 18:35:42 I can fire of some tests if it worked Sep 07 18:36:38 :( I'm afraid not; it's mostly all black magic to me. I struggled through getting a slugos sdk working with a lot of help from folks here on IRC. Sep 07 18:37:06 actually it worked well few months back Sep 07 18:37:25 may be something special in micro distro is preventing it Sep 07 18:45:22 hmmm consulting siteinfo to determine endianness in a machine file does not seem right to me Sep 07 18:57:27 khem: I don't think it's just a problem of evaluation order, it goes deeper than that. Sep 07 18:57:57 The reason that you were getting infinite recursion originally is that, the way things currently stand, TARGET_ARCH is defined in terms of endianness, but endianness is determined by looking at TARGET_ARCH. Sep 07 18:58:22 Fundamentally, there needs to be something to say what the endianness currently is, and we don't have that at present. Sep 07 18:59:15 Making it be a DISTRO choice, like the eabi/oabi/half-eabi thing, seems reasonable. Sep 07 19:14:22 khem: you export IMAGE_BASENAME in recipe? Sep 07 20:01:14 re Sep 07 20:44:25 pb___: I posted a patch to avoid depending on siteinfo in machine files. I feel its better to not use siteinfo Sep 07 20:44:43 hrw|gone: no i dont Sep 07 20:45:15 hrw|gone: that said, previously the name was console-image only but it was populated with a lot more stuff Sep 07 20:47:57 hrw|gone: I have this MAGE_INSTALL += "task-native-sdk" Sep 07 20:48:06 in the .bb file for this new image Sep 07 21:27:48 re Sep 07 21:28:27 woglinde: hey Sep 07 21:29:47 jo khem Sep 07 21:37:25 woglinde: do you know if there are size limits on deploy image that can be imposed Sep 07 21:38:24 hm wasnt there a sie option? Sep 07 21:38:28 size Sep 07 21:38:38 can remeber for simpad Sep 07 21:38:46 grep conf/machine Sep 07 21:39:43 hey, has anyone had the need for sgi's libattr? Sep 07 21:39:53 the latest libcap seems to need it Sep 07 21:40:04 and i *think* vsftpd 2.2.0 needs a newer libcap Sep 07 21:40:24 m4t provide patches Sep 07 21:40:42 woglinde: IMAGE_INSTALL says "task-base libstdc++ task-base-extended task-sdk-native" Sep 07 21:40:55 sdk? Sep 07 21:40:58 in image Sep 07 21:40:58 but it seems to have packaged only task=base Sep 07 21:41:02 woglinde heh im trying Sep 07 21:41:07 thats sounds wrong Sep 07 21:41:12 woglinde: yes its sdk on the target Sep 07 21:41:23 native-sdk not cross-sdk Sep 07 21:41:50 hm Sep 07 21:42:40 hello to all. i have downloaded a qemuarm rootfs from this link here http://amethyst.openembedded.net/~koen/narcissus/ Sep 07 21:42:42 the strange problem is that somehow i miss some apllication in the folder could this be ? Sep 07 21:42:44 as a example this apllication is not it the folder Sep 07 21:42:45 $OPIEDIR/bin/opie-login the same thing for this essential application here Sep 07 21:42:47 /usr/bin/ssh-agent Sep 07 21:42:48 how can this be ? Sep 07 21:42:56 i got a patch for mdadm-3.0 / uclibc Sep 07 21:43:04 uclibc_realpath.patch Sep 07 21:43:12 verified working Sep 07 21:44:15 oh samba 3.3.7 should be good to go as well Sep 07 21:44:34 i almost want to weed out some of the avahi stuff Sep 07 21:44:55 avahi/dbus etc. Sep 07 21:46:10 m4t: good. send it to ml Sep 07 21:46:59 m4t dont forget git format-patch Sep 07 21:49:28 i actually wrote a few lines of c to get mdadm-3.0 compiling Sep 07 21:49:36 hms Sep 07 21:49:43 website not up again? Sep 07 21:49:43 it was using some canonicalize_path thing that uclibc doesnt have Sep 07 21:50:04 and realpath(var, NULL) didnt work because -Werror was on Sep 07 21:50:28 so is just a simple char buf[sizeof(var)]; realpath(var, buf); Sep 07 21:50:42 sizeof(var) + 1 Sep 07 21:52:18 how do i '-lgettext'? Sep 07 21:52:33 ie. what is the name ld takes for that Sep 07 21:52:41 m4t: realpath' accepts (char*, char*) Sep 07 21:52:47 What was warning you got Sep 07 21:54:37 i forget Sep 07 21:55:04 but it warned when i did #define canonicalize_blah(x) realpath(x, NULL) Sep 07 21:55:10 so it erroed because of -Werror Sep 07 21:55:28 so i had it append source_uri for uclibc with a quick patch Sep 07 21:55:54 its in platform-intel.c Sep 07 21:58:40 okay so -lintl fixed te binddomain etc. errors Sep 07 21:58:55 but i get undefined reference to '_' with fprintf(stderr, _(...)); Sep 07 22:05:28 jo ant Sep 07 22:06:31 03Florian Boor  07org.openembedded.dev * r9906662f3d 10openembedded.git/recipes/linux/linux-2.6.28/tx27/defconfig: linux: Update TX27 configuration for 2.6.28 Sep 07 22:06:42 03Florian Boor  07org.openembedded.dev * r5d8aa57291 10openembedded.git/recipes/linux/linux_2.6.28.bb: linux: bump 2.6.28 PR Sep 07 22:06:42 03Florian Boor  07org.openembedded.dev * r5fa71fccd6 10openembedded.git/recipes/canutils/canutils_3.0.2.bb: canutils: add 3.0.2 - needs recent linux-libc-headers Sep 07 22:09:20 hey woglinde Sep 07 22:18:29 jo mickeyl Sep 07 22:20:31 servus woglinde Sep 07 22:24:56 mickeyl: why did you chose to chop /usr stuff for minimal ? Sep 07 22:25:20 its a pain to teach every package to look into /lib instead of /usr/lib Sep 07 22:28:03 i didn't Sep 07 22:28:06 someone did that for micro Sep 07 22:28:11 but minimal should be same Sep 07 22:28:16 err Sep 07 22:28:20 s/same/standard/ Sep 07 22:47:44 is there a way to get oe_runconf to not discard the contents of include/config.h.in? there is a macro there which should keep the gettext undefined reference errors from popping up Sep 07 22:49:47 # define _(x)gettext(x) and the gettext-less alternatives are right there, but they disappear Sep 07 22:52:39 ???? Sep 07 22:52:54 config.h.in is generated Sep 07 22:53:02 bye autools Sep 07 22:53:15 woglinde: not in buildsystems that used old autoconf versions. someone will likely need to fix it up Sep 07 22:53:19 configure creates config.h Sep 07 22:53:21 m4t: oe_runconf doesn't care about it. autotools_do_configure does, as it runs autoreconf Sep 07 22:53:39 kergoth haha than the software is broken Sep 07 22:53:39 autoheader creates config.h.in, but old buildsystems didn't use autoheader. not sure when it was introduced Sep 07 22:53:49 but there are a number of recipes that patch old buildsystems to fix exactly this problem Sep 07 22:54:22 it isn't broken, it's only broken for people wanting to re-run autotools against it, which is.. us. just needs updating to current autotools Sep 07 22:55:10 heh Sep 07 22:55:20 kergoth for me I define its broken, when they dont update the autostuff Sep 07 22:55:47 why? users don't re-run them, by design. autotools are considered maintainer tools, hence --enable-maintainer-mode Sep 07 22:55:51 it's sgi's attr-2.4.43 Sep 07 22:56:56 which the latest libcap seems to require Sep 07 22:57:05 which the latest vsftpd seems to require Sep 07 23:11:06 i just patched all the affected c files Sep 07 23:11:23 config.in and the related might as well be brainfuck to me Sep 07 23:13:35 compiles at least, now to fix the install Sep 07 23:17:19 g'night Sep 07 23:36:00 03Mike Westerhof  07org.openembedded.dev * rb7e0565481 10openembedded.git/recipes/opkg/files/opkg_wget_nogpg_02_use_vfork_system.patch: opkg_wget_nogpg: adjust vfork patchset to accomodate general opkg patching. Sep 07 23:48:56 anyone no any good free dns sevice? Sep 07 23:53:32 damnit Sep 07 23:53:43 why my netbook has no a/b/g wifi Sep 07 23:55:46 hms Sep 07 23:57:46 hms only 1000 series has it Sep 08 00:03:00 any idea why bitbake would do a 'make DESTDIR=...' even if do_install is defined? Sep 08 00:03:47 look in the run files Sep 08 00:03:55 well i built it with -D Sep 08 00:03:56 | + do_install Sep 08 00:03:56 | + autotools_do_install Sep 08 00:08:17 http://pastebin.com/deb4c820 is the bb Sep 08 00:12:26 is autotools overriding it? Sep 08 00:19:32 http://pastebin.com/d326bd659 is the do_install log Sep 08 00:19:48 i mean, it clearly specifies a do_install() {} in the .bb Sep 08 00:20:12 is anyone aware of a condition under which something would be ignored? Sep 08 00:25:35 m4t, is something being included which overwrites your do_install ? Sep 08 00:26:06 no zero includes Sep 08 00:26:15 'inherit autotools' is what i'm using Sep 08 00:26:17 If you define your own tasks for configure, compile, stage or install (via do_) then they will override the methods generated by the autotools class. Sep 08 00:26:34 yes Sep 08 00:26:38 that was my understanding Sep 08 00:26:54 provided you inherit autotools before you define your task Sep 08 00:27:01 i added several 'warn' 'fatal' etc. into the various subroutines and i see none of them Sep 08 00:27:02 oh grg Sep 08 00:27:06 its at the bottom of th efile Sep 08 00:27:32 then the autotools will just declare its own do_install, overwriting your version Sep 08 00:27:41 cool Sep 08 00:28:16 | NOTE: this is broken Sep 08 00:28:23 :] Sep 08 00:29:26 there were expletives in other places, i wont paste those Sep 08 00:30:58 I'm running a an image based off gpe-image. When I hold the stylus on the touchscreen for a bit, a right click event is generated. Sep 08 00:31:09 Anyone know where the code responsible for that lives? Sep 08 00:32:34 I'm trying to debug an application bug, but whenever i run it from the serial console i can't seem to send right click events to reproduce the problem! Sep 08 00:37:25 grg may help he wrotes some parts of tslib Sep 08 00:37:30 args Sep 08 00:37:37 I meant grg kergoth may Sep 08 00:37:49 I have to go to bed now Sep 08 00:37:52 good nite Sep 08 00:37:57 cya Sep 08 00:37:58 -rw-r--r-- 1 matt matt 85104 2009-09-07 20:37 libattr1_2.4.43-r0.3_ppc405.ipk Sep 08 00:38:02 thanks, all Sep 08 00:38:04 woglinde, its not in tslib, i know that much :) Sep 08 00:38:06 grg or you are going to look into tslib Sep 08 00:38:09 hm Sep 08 00:38:11 okay Sep 08 00:38:31 sorry dont know much about gpe internals Sep 08 00:38:35 but ill check the patches anyway... Sep 08 00:38:38 grg maybee the x tslib driver Sep 08 00:38:55 i dont see any right click events when i run 'xev'... Sep 08 00:39:02 lol Sep 08 00:39:09 when where do it come from Sep 08 00:39:16 higher up Sep 08 00:39:17 but I reaaly have to go Sep 08 00:39:21 maybe gtk? Sep 08 00:39:24 :) Sep 08 01:00:11 its part of libgtkstylus Sep 08 01:00:43 i needed to export GTK_MODULES=libgtkstylus.so before running the app Sep 08 01:00:54 weird crazy shit Sep 08 01:01:51 /etc/X11/Xsession.d/45gtkstylus does this for X Sep 08 01:11:56 oh god Sep 08 01:12:21 after all the trouble of getting libcap 2.17 and lsattr installed Sep 08 01:12:26 vsftpd still craps out Sep 08 01:15:02 lsof shows its loading libattr and the libcap2 though **** ENDING LOGGING AT Tue Sep 08 02:59:57 2009