**** BEGIN LOGGING AT Thu Oct 28 02:59:57 2010 Oct 28 03:25:51 hmmmm Oct 28 03:26:21 * kergoth_ hacks on srctree.bbclass Oct 28 03:46:23 hrm, seems like something's still up with srctree's task merging Oct 28 03:46:25 hrmph Oct 28 04:28:55 vorporeal: whats the issue with libgcc Oct 28 05:16:26 khem: you there? Oct 28 05:16:55 i just submitted the beagleboard mac address patch. not sure what you mean by changing MACHINE_KERNEL_PR_append Oct 28 05:30:39 khem: three python packages are bringing in -L/usr/lib to the linker flags Oct 28 05:30:58 which, on arch, causes them to try to link against /usr/lib/libgcc_s.so Oct 28 05:31:08 python-numeric, python-fuse, and python-pygame Oct 28 05:31:16 03Graham Gower  07master * r74d7b8e7a0 10openembedded.git/recipes/gs/ (gs-tools-native_8.64.bb gs_8.64.bb): (log message trimmed) Oct 28 05:31:16 gs: Clean up and break into two recipes. Oct 28 05:31:16 * Break the native compilation steps into a separate native recipe. Oct 28 05:31:16 * Fix license Oct 28 05:31:16 * Convert to new style staging Oct 28 05:31:16 * Fix build on mipsel due to missing -fPIC for objects linked into so files. Oct 28 05:31:16 Signed-off-by: Graham Gower Oct 28 05:31:24 do the recipes inherit from autotools, by chance? Oct 28 05:31:31 sorry, that was for vorporeal Oct 28 05:31:32 nope, distutils Oct 28 05:31:37 ah Oct 28 05:31:54 bummer, would be cool if it was the same trivial problem as mine last night with jaaa/clxclient Oct 28 05:32:07 i had issues like that with autotools, so i switched to angstrom-2010.x to get libtool 2.4 which fixed that Oct 28 05:32:23 but this remains (same basic problem, i just don't know what to change to fix it) Oct 28 05:32:31 oh interesting. i am on 2010.x as well Oct 28 05:33:09 probably depends what images you build, i tried to build beagleboard-demo-image which included jaaa. i wouldn't imagine that package would be super popular, its like a spectrum analyzer for audio signals coming from JACK Oct 28 05:34:25 ah Oct 28 05:34:32 i had to install clxclient, but not jaaa Oct 28 05:34:52 the patch made to fix your issue helped me out, so thanks :) Oct 28 05:36:00 glad to be a squeaky wheel, khem deserves the real thanks Oct 28 05:36:38 absolutely - he's helped me many times over the past two weeks while i've struggled to get this working Oct 28 05:37:23 what's your end goal with your compiled distro? don't mean to be nosy, i just make a habit of asking what people are working on Oct 28 05:37:41 ah, i compiled a toolchain for the pandora handheld Oct 28 05:37:57 and have just these last few errors to wade through to get a root image Oct 28 05:38:34 doing this both for the purposes of being able to cross-compile code (my own and ports), as well as work on a more up-to-date base installation which may fix some of the issues that have been annoying me Oct 28 05:39:21 yeah, i'm trying to get the bleeding edge kernel working on the beagleboard to see if its performance or power management is any better than the stock 2.6.32 Oct 28 05:40:13 ah Oct 28 05:40:19 i'm not especially looking for bleeding edge Oct 28 05:40:41 but the official firmware has 2.6.27, and isn't using the most up-to-date versions of software Oct 28 05:40:53 yeah Oct 28 05:40:59 so i'm trying to get newer but hopefully still stable, more or less Oct 28 05:41:30 yeah. the kernel i'm using is rc7, so pretty close to the final 2.6.36 release Oct 28 05:43:00 well, night time for me Oct 28 05:43:03 class in the morning Oct 28 05:43:21 i wish i had the willpower to go to bed before 3 Oct 28 05:43:22 night man Oct 28 05:43:42 i went to sleep late last night and got up early to finish homework Oct 28 05:43:53 so i'm only going to bed now because i'm literally falling asleep in my chair Oct 28 05:44:02 after a week+ of this :( Oct 28 05:44:17 trying to turn it around, albeit unsuccessful so far Oct 28 05:45:11 night, gl with your image Oct 28 05:45:20 thx Oct 28 07:01:32 good morning Oct 28 07:02:07 morning Oct 28 07:02:14 you must be in europe :P Oct 28 08:43:57 hej, anybody else having problems with mesa-7.8.2 (or already working on it)? gsls-compile is not found Oct 28 08:57:02 what application has to be used to view .DSN files Oct 28 08:57:05 ??? Oct 28 08:57:12 ? Oct 28 08:57:22 google .DSN? Oct 28 08:57:54 woglinde: i did Oct 28 08:58:05 all the applications are proprietory Oct 28 08:58:11 i was looking for a free one Oct 28 08:58:24 aeh Oct 28 08:58:34 so why should we know? Oct 28 08:59:26 these files are used for showing schematics Oct 28 08:59:36 that is why i want a .DSN viewwer Oct 28 08:59:45 since this is OE Oct 28 08:59:55 ? Oct 28 08:59:59 i thought you guys might be using .DSN files extensively Oct 28 09:00:09 what the hell has openembedded to do with a DSN viewer? Oct 28 09:00:21 i am into embedded development Oct 28 09:00:23 hm.. Oct 28 09:00:24 http://lwn.net/Articles/412128/ Oct 28 09:00:33 i am trying to see beagleboard design files Oct 28 09:00:42 open embedded beagleboard Oct 28 09:00:46 we are making software Oct 28 09:00:51 not the hardware Oct 28 09:00:59 ask on #beagle for a viewer Oct 28 09:05:29 india Oct 28 09:05:35 I should have known Oct 28 09:05:53 http://www.yoctoproject.org/blogs/dvhart/2010/10/3/yocto-not-distribution Oct 28 09:06:01 am I slow on this? Oct 28 09:06:25 "dear mr. I am rikahsiana, please write me all steps with pictures for linux" Oct 28 09:06:55 * Jay7 wonders about yet another embedded linux workgroup Oct 28 09:07:17 jay7 yeah dontknow whats intel doing there Oct 28 09:07:30 intel is everywhere Oct 28 09:07:47 and nowhere at same time seems Oct 28 09:08:03 hm okay Oct 28 09:08:13 its linux foundation only Oct 28 09:08:22 maybee why the get kicked at moblin Oct 28 09:08:34 got Oct 28 09:09:05 hello all, in patch recipes/eglibc/eglibc-svn/eglibc-svn-arm-lowlevellock-include-tls.patch Oct 28 09:09:29 there is some make error message it should be removed Oct 28 09:09:35 hm? Oct 28 09:10:16 pastebin your error Oct 28 09:18:58 is not really error, i'm just wondering why are make error messages in this patch http://pastebin.com/FqCy1Nnp Oct 28 09:21:39 ? Oct 28 09:21:45 thats the error its fixing Oct 28 09:27:32 khem: the test.s with that gcc-cross you've requested is the same as before Oct 28 09:27:57 hi jama Oct 28 09:29:10 hi woglinde Oct 28 09:29:58 hm prelink-cross sounds nice Oct 28 09:30:10 jama have you tried prelinking for shr? Oct 28 09:30:18 does it gain some speed? Oct 28 09:32:04 no Oct 28 09:32:09 haven't tried Oct 28 09:32:23 http://www.yoctoproject.org/projects/cross-prelink Oct 28 09:34:18 * JaMa don't see so big gain from prelink on desktop, but benchmark from some small machine would be nice Oct 28 09:35:25 sure on desktop it wont much help Oct 28 09:35:31 but on our devices Oct 28 09:37:18 hmm only angstrom is using -Wl,--as-needed? that would help too Oct 28 09:38:13 hm? Oct 28 09:38:24 isnt micro using it too? Oct 28 09:39:16 git grep doesn't show it to me Oct 28 09:42:59 yeah, prelink-cross would be quite cool Oct 28 09:43:17 I spent a while trying to make that work a few years ago but I ran out of time and I don't think it ever did anything useful back then. Oct 28 09:46:09 uh pb stands up from the grave Oct 28 09:46:12 *g* Oct 28 09:46:14 hi pb Oct 28 09:46:32 hi woglinde Oct 28 09:48:14 pb not at elce? Oct 28 09:48:28 no, not cool enough to attend elc. Oct 28 09:48:35 I think one of my colleagues has gone though Oct 28 09:48:54 bit of a skeleton staff here in the office actually, two more of our guys are at droidcon. Oct 28 09:49:03 * pb_ not cool enough to go to droidcon either Oct 28 09:49:03 hehe Oct 28 09:49:04 okay Oct 28 09:49:27 pb your company is switching to android too? Oct 28 09:50:28 not switching, but we are starting to do a bit of android app development Oct 28 09:50:51 hm yeah lets see whats 3.0 will bring Oct 28 09:51:20 I hope google forces the manufactors for actual kernels Oct 28 10:43:04 hi Oct 28 10:53:05 jo hrw Oct 28 10:53:24 hrw are you now back? Oct 28 10:53:28 hi mickeyl Oct 28 10:57:42 woglinde: no, still usa Oct 28 10:58:09 will be @home on sunday evening Oct 28 10:58:13 hm than still jetlag Oct 28 10:58:42 phone had reminder set to 12:00 Oct 28 10:58:58 but 12:00 cest so woke me at 6:00 local ;( Oct 28 10:59:23 h the old automatic timezone problem Oct 28 10:59:58 yep Oct 28 11:00:06 hm the switch to est must be in next two weeks Oct 28 11:00:08 lets seet Oct 28 11:00:27 woglinde: not two weeks - this sat/sun night Oct 28 11:00:29 yeah Oct 28 11:00:32 this weekend Oct 28 11:00:42 during my flight Oct 28 11:17:38 hrw: did you have pandaboard, iirc? Oct 28 11:17:49 one guy is looking for shell on panda Oct 28 12:02:28 is it somehow possible to override a file from another package (i.e. overriding /etc/net/interfaces) ON image creation? (have an interface-opkg in my generated fs) Oct 28 12:04:39 and without changing the base recipe to use update alternatives Oct 28 12:18:46 Hi! I'm trying to understand a bitbake error "NameError: global name 'msg' is not defined Oct 28 12:28:56 Jay7: I own pandaboard but not have it yet Oct 28 12:29:25 ah, ok Oct 28 12:29:43 ideas? Oct 28 12:29:43 next week Oct 28 12:29:53 qdot: wrong BBPATH Oct 28 12:31:01 hrw: not really.. it seems correct - /vol/temp/qdot/src/newdeal/oe/build:/vol/temp/qdot/src/newdeal/oe/openembedded Oct 28 12:31:31 hrw: branch master Oct 28 12:32:02 ok, no idea at that time Oct 28 12:35:22 any special python versions? Oct 28 12:35:38 I used to be able to build openembedded just fine Oct 28 12:44:42 03Michael Smith  07master * r67017389a1 10openembedded.git/recipes/xdelta/ (xdelta-3.0z/xdelta-3.0z-use-oe-cflags.patch xdelta_3.0z.bb): Oct 28 12:44:42 xdelta 3.0z: fix GNU_HASH QA errors Oct 28 12:44:42 Signed-off-by: Michael Smith Oct 28 12:46:11 use 2.6 Oct 28 12:53:49 03Stefan Schmidt  07master * r32a308ab7b 10openembedded.git/recipes/images/illume-image.bb: illume-image: Remove obsoleted and commented out recipes Oct 28 12:53:51 03Stefan Schmidt  07master * r7ed0b3407b 10openembedded.git/recipes/e17/elementary-sms_svn.bb: Oct 28 12:53:51 elementary-sms: Move to obsolete. Oct 28 12:53:51 * Depends on libefso which moved to obsolete Oct 28 12:53:51 Signed-off-by: Stefan Schmidt Oct 28 12:53:52 03Stefan Schmidt  07master * r1c4b6a257e 10openembedded.git/recipes/tasks/task-shr-feed.bb: task-shr-feed: Remove obsoleted and commented out recipes Oct 28 12:56:32 hi Oct 28 12:56:37 hi ericben Oct 28 12:56:40 hi woglinde Oct 28 12:57:41 using mdev, there is a /etc/mdev/find-touchscreen.sh script which checks if the current devices probed by mdev is a touchscreen and creates a link to /dev/input/touchscreen0 Oct 28 12:57:57 this script assumes $MDEV is event0 or event1 or ... Oct 28 12:58:23 but , on a 2.6.36 kernel, $MDEV is input/event0 which makes the script failing Oct 28 12:58:33 I'm solving this by adding MDEV=`echo $MDEV | sed 's#input/##g'` Oct 28 12:58:39 at the top of the script Oct 28 12:59:04 is this an acceptable fix (it should'nt hurt even on older configuration) ? Oct 28 12:59:53 I should say on a 2.6.36 kernel + busybox 1.17.1 because this may be mdev which changed the way he reports the device Oct 28 13:00:57 ericben: send patch to ML for review? Oct 28 13:01:54 hrw: hrw|afk : I can do that but my last patch concerning mdev didn't get any feedback :-( Oct 28 13:02:17 ericben ping more Oct 28 13:02:26 and I don't want to push things like this without at least one review Oct 28 13:02:45 woglinde: will do now that I'm back to OE dev for a few days ;) Oct 28 13:03:00 we need to think about ways to get people to help with review of core things like this Oct 28 13:03:22 I'm sure both patches won't- break actual configuration Oct 28 13:03:31 because they handle special cases Oct 28 13:03:54 but ... just in case I missed something obvious this is better to get review Oct 28 13:05:58 cbrake: concerning qt 4.7: do you have any objection if I update the recipe to rename qtdemo to qtdemoE when compiling qt4-embedded ? Oct 28 13:06:08 in order to be consitent with 4.6 recipes Oct 28 13:08:43 hello Oct 28 13:09:02 (not using oe, just buildroot:) why on freshly built ARM926EJ-S MPU target executing make says "-sh: make: not found". rootfs was built with buildroot for arm926t. kernel came from earlier. same with other bigger executables, although ls, cat, .. work.. Oct 28 13:10:27 ? Oct 28 13:10:34 sledges: buildroot chanel is #uclibc , it seems you don't have make on your target and try to exectue it Oct 28 13:11:01 ericben hm I should example your polite answers Oct 28 13:12:00 ericben: thanks for the chan! make is there (i .config`ed it), /usr/bin/make ; so is fbdump, they are under path; but it simply does not get executed :( some .so, libc linking problems?.. Oct 28 13:12:25 try ldd Oct 28 13:12:59 woglinde: paradox: ldd is under /usr/bin/ldd but is under the same problem: -sh: /usr/bin/ldd: not found Oct 28 13:13:12 sledges: I don't know as it's quite a long time I've not used buildroot. Oct 28 13:13:31 sledges, ask in a buildroot channel Oct 28 13:13:32 sledges this means you have a fundamental problem Oct 28 13:13:42 * woglinde wonders why init is running Oct 28 13:14:07 ericben: i think the problem is not in the buildroot (that's why i'm stil here :D) I had a angstrom rootfs from a colleague (only binaries, dunno how he built it, om?) - was giving similas programs Oct 28 13:14:44 seldges than start using oe Oct 28 13:14:52 ericben: no, that is fine with me Oct 28 13:14:56 sledges: why do you need to run make on the target ? Oct 28 13:15:00 cbrake: ok thanks Oct 28 13:15:04 ericben as he said Oct 28 13:15:10 if nt ldd runs Oct 28 13:15:17 its like other stuff dont runs too Oct 28 13:15:33 oh yes didn't see this line Oct 28 13:15:53 sledges: ls -al /usr/bin/make : isn't that a symlink ? Oct 28 13:15:54 PREFERRED_PROVIDER_usbmodeswitch = "usb-modeswitch" ...whats wrong here ? bitbake still does complain about multiple providers Oct 28 13:17:12 ericben: i just took make as an instance. i don't have to run it; but it seems the majority from /usr/bin/* just cannot launch themselves (dfbscreen, df_fire, you name it) Oct 28 13:17:22 dc works though.. Oct 28 13:17:53 right, missed your lines there folks :) Oct 28 13:18:04 sledges: ls -al /usr/bin/dfbscreen ? Oct 28 13:18:23 no, ericben, no symlinks, -rwxr-xr-x 1 root root 10500 Oct 28 2010 /usr/bin/dfbscreen Oct 28 13:20:11 sledges: which dfbscreen Oct 28 13:21:13 ericben: /usr/bin/dfbscreen Oct 28 13:22:35 sledges: well here (arm926 + OE + angstrom) applications in /usr/bin are working fine so I think you're on your own here Oct 28 13:22:44 hi all Oct 28 13:22:59 ok thanks Oct 28 13:23:02 hi florian Oct 28 13:24:57 Hi, I'm building an image and I have a package which installs some extra stuff, scripts and files which are not useful elsewhere Oct 28 13:25:19 Some of the files in this package are designed to replace files shipped in other packages Oct 28 13:26:09 Is there any way I can tell opkg not to fail and overwrite the older files with the ones in the extras package I have during the image-building process? Oct 28 13:26:24 *and to overwrite Oct 28 13:26:42 hm, let me ask again - did anyone run into the "global name 'msg' not defined" error? Oct 28 13:26:59 with bitbake? Oct 28 13:27:21 B_Lizzard: maybe you can solve this by using CONFFILES_${PN} ? Oct 28 13:27:57 hi ericben Oct 28 13:28:04 have the same problem as B_Lizzard CONFFILES does not force an override but produce a -opkg Oct 28 13:29:06 can i somehow do a bitbake world, but just download the packages? not actually building them? Oct 28 13:29:11 just to get the sources Oct 28 13:29:24 Anything defined under CONFFILES will replace the older file and have it renamed as -opkg? Oct 28 13:29:29 That would work, I guess. Oct 28 13:29:30 mlip: but in the end, doesn't the new package's conffiles overwrite the old one (I didn't test so I may be wrong) ? Oct 28 13:29:45 * florian arrived in Cambridge this morning Oct 28 13:30:11 ericben would like to be in Cambridge but can't :-) Oct 28 13:30:25 florian fine Oct 28 13:30:26 ericben: if I define my own package with files as CONFFILES, the old files remain, and my own ones get renamed to opkg; this happens atleast at image creation Oct 28 13:30:33 dunno about invoking opkg on the board Oct 28 13:31:15 ericben: too bad... but quite some more suffer from this too Oct 28 13:31:52 woglinde: are you at elc? Oct 28 13:32:01 Wouldn't CONFFILES work only for the same package? Oct 28 13:32:07 Say, when installing a newer version? Oct 28 13:32:15 florian no Oct 28 13:32:21 only robert Oct 28 13:32:33 but I heared he will leave tomorrow Oct 28 13:32:34 Not if both xinit and jlime-extras have /usr/bin/startx Oct 28 13:32:53 B_Lizzard: I don't think so; I think the new files get renamed to -opkg to prevent overriding; this is the reason why conffiles doesnt work Oct 28 13:34:09 woglinde: yes i saw his note at the oedem wiki page Oct 28 13:34:26 mlip, I see. Oct 28 13:34:48 B_Lizzard: http://docs.openembedded.org/usermanual/usermanual.html#recipes_conffiles ; it is basically described here Oct 28 13:34:49 he crofton Oct 28 13:34:52 hi Crofton Oct 28 13:35:05 gm Oct 28 13:35:11 but there has to be some way ;) Oct 28 13:35:40 B_Lizzard: mlip: using --force-maintainer option of opkg ? Oct 28 13:35:57 How do you do that as part of a recipe? Oct 28 13:36:16 ericben: could work when invoking opkg, but how to do it in image creation ? Oct 28 13:36:41 B_Lizzard: I don't know hacking is necessary Oct 28 13:37:24 Maybe some of the OE developers know? Oct 28 13:37:36 I'm sure someone has asked before. Oct 28 13:38:38 B_Lizzard: mlip: in classes/rootfs_ipk.bbclass, there is IPKG_ARGS, maybe you can hack here ? Oct 28 13:39:14 Probably, but I'd like to have my stuff as compatible as possible. Oct 28 13:39:42 It would be easier just providing the files as part of each individual recipe but that's messy. Oct 28 13:39:47 Would never be accepted Oct 28 13:40:05 B_Lizzard: well you can have an overlay which contains these files Oct 28 13:40:47 Im' using this + amend.inc to create machine specific packages Oct 28 13:45:38 03Denis 'GNUtoo' Carikli  07master * ra3871c2416 10openembedded.git/recipes/xorg-xserver/ (2 files in 2 dirs): (log message trimmed) Oct 28 13:45:38 xserver-xorg-conf: nokia900's xorg.conf mark keyboard as CoreKeyboard Oct 28 13:45:38 * add CoreKeyboard option for nokia900 Oct 28 13:45:38 This seems to be needed to make it actually load it... Oct 28 13:45:38 Otherwise a pc105 model will be there instead of the Oct 28 13:45:39 needed nokiarx51 one... needs researching. Oct 28 13:45:40 Signed-off-by: Klaus Kurzmann Oct 28 13:45:49 03Martin Jansa  07master * r5fa427d069 10openembedded.git/recipes/efl1/edbus_svn.bb: Oct 28 13:45:49 edbus: fix upgradable path Oct 28 13:45:49 * FILES_edbus is empty now all libs packaged in lib*.ipk Oct 28 13:45:49 Signed-off-by: Martin Jansa Oct 28 13:46:00 03Martin Jansa  07master * r389115f89a 10openembedded.git/recipes/cairo/cairo_1.10.0.bb: cairo_1.10.0: bump PR to rebuild with gcc fix Oct 28 13:46:20 03Pau Espin Pedrol  07master * rdab2169ee8 10openembedded.git/recipes/e17/e-wm_svn.bb: Oct 28 13:46:20 e-wm: add subpackage with optional EFM desktop icon Oct 28 13:46:20 * you have to enable efm module in e-wm settings Oct 28 13:46:20 Signed-off-by: Martin Jansa Oct 28 13:46:36 bbl Oct 28 14:21:48 hi kgilmer Oct 28 14:29:08 hi woglinde Oct 28 14:29:23 are you in cambridge woglinde? Oct 28 14:29:37 kgilmer no Oct 28 14:30:08 kgilmer: hi Oct 28 14:30:13 hey hrw|uds Oct 28 14:30:21 how is sunny Florida treating you? Oct 28 14:30:26 kgilmer: it is strange feeling to talk with you in local time Oct 28 14:30:33 kgilmer: most of time in AC hotel ;d Oct 28 14:31:00 hrw|uds, i cannot get ubuntu to boot on my bug20. kernel boots, mounts rootfs and init just dies. recommended steps for debugging? Oct 28 14:31:24 hrw|uds, are you planning on visiting nasa or anything or just conference then return home? Oct 28 14:31:28 kgilmer hm enable userspace debug Oct 28 14:31:31 kgilmer: other then "send me one device"? Oct 28 14:31:37 kgilmer: conf and home Oct 28 14:31:59 at lave evening on sunday I will be home Oct 28 14:32:09 woglinde, how can i do that? Oct 28 14:32:12 mom Oct 28 14:32:22 hrw|uds, ok, i'll see if i can get one out to you. Oct 28 14:32:34 kgilmer: I was joking ;D Oct 28 14:32:39 damn Oct 28 14:32:41 i wasn't :) Oct 28 14:32:46 I should use a wiki for it Oct 28 14:33:00 kgilmer: cool ;) bug add video and debug stuff please Oct 28 14:33:16 yep Oct 28 14:33:44 woglinde, sorry don't understand. how can i enable userspace debug? Oct 28 14:34:52 kgilmer I am still searching Oct 28 14:35:00 there is one kernel option Oct 28 14:35:11 and boot cmdline activating Oct 28 14:35:27 kgilmer: can you share boot output? Oct 28 14:37:42 yeah sec Oct 28 14:37:48 oh..my bug20 was taken Oct 28 14:37:50 hmmpfh Oct 28 14:38:11 hrms Oct 28 14:38:17 khem knows it faster Oct 28 14:38:18 * kgilmer searches messy desk for tiny sd card Oct 28 14:39:14 * mwester has been there; searching floor around desk as well. Oct 28 14:41:45 i looked for my micro-sd to sd adapter for 3 days before i found it under some papers Oct 28 14:42:59 Of course, the fact that they are all flat black in color doesn't help any. :D Oct 28 14:43:35 I have one place where I keep all sd/cf/usb stuff Oct 28 14:45:18 ok i have booted something Oct 28 14:45:19 let me see if it's ubuntu Oct 28 14:47:07 damn Oct 28 14:47:10 where it is Oct 28 14:48:12 i'm usually better organized, this project was just 'i wonder how hard it would be to...' kind of thing. Oct 28 14:49:28 I added a base-conditional to rootfs_ipk.bbclass, something that may make upstream Oct 28 14:49:34 It works Oct 28 14:50:30 kgilmer sorry dont find it at the momen Oct 28 14:50:34 its ok woglinde Oct 28 14:50:39 i also cannot find my ubuntu sd card Oct 28 14:50:47 i will have to make another one and get the log output Oct 28 14:50:51 ask khem later Oct 28 14:50:56 wasnt that hard Oct 28 14:51:05 ok Oct 28 14:51:31 was user.debug=-99 or debug.user=-99 Oct 28 14:51:33 as cmdline Oct 28 14:51:41 and one kernel option activated Oct 28 14:51:43 kernel param woglinde? Oct 28 14:52:18 ah! i found it Oct 28 14:52:56 hrw|uds, http://pastebin.com/1Y0JNLAy Oct 28 14:53:07 so, i did the obvious googling on this situation Oct 28 14:53:14 x86 people get this with a bad fstab Oct 28 14:53:26 i made sure the fstab was not bad Oct 28 14:54:40 hm? Oct 28 14:54:43 this looks normal Oct 28 14:54:50 where is the error? Oct 28 14:55:46 kgilmer: hmm. it is normal boot Oct 28 14:55:51 kgilmer mom which image you are trying? Oct 28 14:56:10 recent ubuntu images was for 4 gig sd Oct 28 14:56:15 yours is 2 gig Oct 28 14:58:31 kgilmer: or does boot ends at that? Oct 28 15:02:01 is there a way to clean everything? Oct 28 15:02:26 bitbake -c clean XXXX would only clean that specific package Oct 28 15:02:50 rm -rf tmp? Oct 28 15:03:02 bbl Oct 28 15:13:03 woglinde, i am ust using rootstock with ubuntu-minimal Oct 28 15:13:13 so your own made Oct 28 15:13:33 yeah Oct 28 15:13:56 okay Oct 28 15:14:00 hrw|uds, that's the last message that comes off the serial line Oct 28 15:14:31 woglinde, i don't have a debian package for the bug20 kernel so i'm manually installing that and modules on the card after writing the rootstock images. Oct 28 15:15:44 kgilmer: did you gave few minutes for bug20 to continue boot? Oct 28 15:16:18 hrw you mean stupud udev starting up? Oct 28 15:16:51 hrw|uds, it's still running since i booted Oct 28 15:17:08 hm Oct 28 15:17:16 is ping working? Oct 28 15:17:27 let me check woglinde Oct 28 15:17:36 woglinde: ubuntu removed lot of boot messages Oct 28 15:17:36 but proably you didnt setup network yet Oct 28 15:18:04 moment Oct 28 15:18:32 kgilmer: you used 10.04 or 10.10 ubuntu? Oct 28 15:19:51 i used rootstock defaults hrw|uds. here is what i ran: http://pastebin.com/3gKtK1Ju Oct 28 15:19:56 maverik Oct 28 15:19:59 ok Oct 28 15:20:48 kgilmer: can you try lucid or even karmic? maybe udev require newer kernel then .31 Oct 28 15:20:54 lucid used .32 version Oct 28 15:21:31 sure Oct 28 15:21:35 i will try that today Oct 28 15:21:40 karmic is best? Oct 28 15:21:44 hrw|uds? Oct 28 15:21:55 udev works on .24 and above Oct 28 15:22:15 ka6sox: depends. newer ones wanted .27 Oct 28 15:22:29 oh right...forgot about that change. Oct 28 15:22:43 kgilmer: not best - I would just use it for test Oct 28 15:22:49 did you lock the ethernet to 1 mac address in teh rules? Oct 28 15:23:07 hi ka6sox Oct 28 15:23:13 morning woglinde Oct 28 15:23:24 ka6sox didnt need 154 a newer kernel? Oct 28 15:23:43 I'm using a .32 kernel. Oct 28 15:23:53 and which udev? Oct 28 15:23:58 from Lucid... Oct 28 15:24:05 ka6sox: and kgilmer uses .31 one Oct 28 15:24:19 hmmm...can't access it now as I"m out of town. Oct 28 15:24:23 kgilmer: test lucid. if it works then keep it for now. if not then try karmic Oct 28 15:24:39 ok hrw|uds i'll do that, thanks. Oct 28 15:24:49 lucid had udev 151 Oct 28 15:25:02 maverick has 162, karmic was at 147 Oct 28 15:29:53 hrw|uds, thats good to know... Oct 28 15:34:53 ka6sox its standing on the ubuntu packages website for udev *g* Oct 28 15:40:58 ya, I just didn't look...mostly I look for the deps on the Debian site (easier to read) Oct 28 15:46:25 re Oct 28 15:46:52 re florian Oct 28 15:46:54 florian, are you coming to OEDEM? Oct 28 15:48:23 Crofton: you wrote about agenda but can you remind a link to it? Oct 28 15:48:31 ah Oct 28 15:48:34 Crofton: I'm in Cambridge already :-) Oct 28 15:48:46 rofl Oct 28 15:48:50 * Crofton waves Oct 28 15:49:04 http://wiki.openembedded.org/index.php/Oedem/2010 Oct 28 15:49:08 gracias Oct 28 15:49:34 * florian waves up the road Oct 28 15:49:50 ok off for some coffee Oct 28 15:49:56 ELCE is winding up Oct 28 15:49:59 bbye Oct 28 15:50:03 Crofton: enjoy Oct 28 15:50:03 Crofton: will there be possibility to remote connect to gobby? Oct 28 15:50:25 hrw hehe Oct 28 15:50:30 thats what I thought too Oct 28 15:50:46 running the gobby server on the one of the oe Oct 28 15:50:51 hi tom Oct 28 15:51:01 hey Oct 28 15:51:16 that way I could atleast check what is going on at oedem Oct 28 15:52:04 as this is the first oedem which I not attend.. Oct 28 16:00:52 I'm occasionally running into autoconf issues with recipes that I'm mucking with such that do_configure fails with things like configure.ac:87: error: possibly undefined macro: PKG_CHECK_MODULES - I know if I wipe TMPDIR and rebuild it will be ok, but if I clean the package and rebuild its not - what regarding autoconf gets staged somewhere that do_clean doesn't take care of? Oct 28 16:04:56 I'm trying to add qtcreator (Nokia's QT IDE) to OE in order to integrate it to the SDK. The problem is that we don't have many dependencies ready to build for the sdk. Do you think there is an interest to integrate this to OE (require adding sdk / native to many recipes) ? Oct 28 16:05:45 if no interest, I'll do this out of OE. if you think this can be interesting, then I'll follow my tentative to get it into OE. Oct 28 16:06:00 ? Oct 28 16:06:09 just do it Oct 28 16:06:22 ericben: You think about building a native package as a part of the toolchain? Oct 28 16:06:29 florian: yes Oct 28 16:06:44 the idea is to produce a sdk which integrates the toolchain + qtcreator Oct 28 16:06:50 hm Oct 28 16:06:56 compiled static? Oct 28 16:06:58 like nokia's sdk Oct 28 16:07:00 otherwise Oct 28 16:07:05 that's the problem Oct 28 16:07:11 there is hundred of ways Oct 28 16:07:17 it will not run Oct 28 16:07:39 yes that's possible Oct 28 16:08:16 hm Oct 28 16:08:18 but on nokia's sdk it works fine Oct 28 16:08:21 its not the worst idea ever - you can keep it in sync with qt and don't have to wait for your distribution to update Oct 28 16:08:32 they simply hack the LD_LIBRARY_PATH and it works fine Oct 28 16:08:43 on the other hand it doesn't integrate into the desktop that well Oct 28 16:08:46 in fact I need a patch to get gdbserver interact fine with qtcreator Oct 28 16:08:49 hm hm maybee you need candian stuff for it Oct 28 16:09:08 hm hm Oct 28 16:09:16 but that would be same for toolchain Oct 28 16:09:23 woglinde: yes Oct 28 16:09:33 we had the problems with gdb and some libraries Oct 28 16:09:55 here the dependencies are huge (qt 4.7, python ....) Oct 28 16:10:43 I think I'm first going to get it compile using the host's libraries and then will try to integrate it better Oct 28 16:11:43 with this patch http://lists.trolltech.com/pipermail/qt-creator/2010-October/007901.html it's possible to use qtcreator simpy for remote debugging on the target Oct 28 16:12:12 so having this in the QT SDK seems great Oct 28 16:12:29 moreover this make it psosible to integrate plugins to qtcreator for some qt addons Oct 28 16:13:40 yes Oct 28 16:13:53 you are cloning madde a bit Oct 28 16:13:55 *g* Oct 28 16:14:05 what does that mean ? Oct 28 16:14:30 ok I just found madde Oct 28 16:14:34 didn't know that Oct 28 16:14:43 http://wiki.maemo.org/MADDE Oct 28 16:14:51 but thats okay Oct 28 16:14:53 *g* Oct 28 16:15:19 in fact they are using the binary sdk from nokia and just build a plugin Oct 28 16:15:53 oh intersting Oct 28 16:21:24 this is very similar but targeting normal devices (not meego / maemo) Oct 28 16:24:01 yes Oct 28 16:24:20 actually I have ti running but ... handbuilt ;-) Oct 28 16:24:35 sure Oct 28 16:25:09 the first problem is to build a native qt4-x11-free-sdk ... Oct 28 16:26:22 because qtcreator depends on qt4.7 and which is not standard on many distributions Oct 28 16:28:32 03Stefan Schmidt  07master * r3c0f83e86e 10openembedded.git/recipes/cacao/cacaoh-openjdk-native.inc: Oct 28 16:28:32 cacaoh-openjdk-native: Cleanup double inherit of autotools. Oct 28 16:28:32 Signed-off-by: Stefan Schmidt Oct 28 16:28:51 03Stefan Schmidt  07master * r44bc5a6380 10openembedded.git/recipes/cacao/ (4 files): (log message trimmed) Oct 28 16:28:51 cacaoh-*native: Try to be libtool version agnostic. Oct 28 16:28:51 Based on a suggestion from Enrico Scholz we should use libtool --mode=install Oct 28 16:28:51 for install libtool stuff. This fixes a regresion for me that came in with Oct 28 16:28:52 commit 93c4d9d84096c817e74abf3f4c02f82af4c95178 Oct 28 16:28:52 It is working fine again now with libtool 2.2.6b (angstrom-2008), testing with Oct 28 16:28:53 libtool 2.4 would be appreciated. Oct 28 16:29:06 Crofton|work: attempting to compile openMPI for gumstix as I type Oct 28 16:45:14 Crofton|work: mission UNaccomplished Oct 28 16:45:24 Crofton|work: receive the following error Oct 28 16:45:31 Crofton|work: configure: error: No atomic primitives available for armv7l-unknown-linux-gnueabi Oct 28 16:45:41 whats your gcc? Oct 28 16:45:56 Crofton|work: which is kind of what I expected from reading some others' accounts Oct 28 16:46:02 and where it comes from? Oct 28 16:46:31 oe based gcc >= 4.3 has atomic support Oct 28 16:46:37 gcc (Ubuntu 4.4.3-4ubuntu5) 4.4.3 Oct 28 16:46:54 hm uh Oct 28 16:47:00 hmm.... it's not from OE Oct 28 16:47:05 this should have atomic support Oct 28 16:47:15 I've installed ubuntu on the gumstix Oct 28 16:47:34 are you familiar with openMPI? Oct 28 16:47:42 no Oct 28 16:47:53 but I bit familiar with the atomic gcc stuff Oct 28 16:48:15 dpackard: this is #oe not #ubuntu... Oct 28 16:49:29 hrw|uds: right.. was just hoping to get a hold of crofton, we discussed openMPI for gumstix last week Oct 28 16:49:50 ah Oct 28 16:50:04 for Crofton it is 18:45 now as he is in UK Oct 28 16:50:13 I am think crofton is at beer now Oct 28 16:50:23 haha :) Oct 28 16:50:34 I wish I were there also Oct 28 16:52:07 anyway... i think the issue is that openMPI does not use the built-in GCC atomics Oct 28 16:52:32 a la - http://comments.gmane.org/gmane.comp.clustering.open-mpi.user/11461 Oct 28 16:52:44 why it than would error out at with configure: error: No atomic primitives available for armv7l-unknown-linux-gnueabi Oct 28 16:53:25 ah okay Oct 28 16:54:46 hm shouldnt be that hard to make some patches Oct 28 16:55:00 hehe =P hard for me Oct 28 16:58:12 dpackard you please grep in which m4 file atomic primitives is? Oct 28 16:58:48 woglinde: sure, give me a minute Oct 28 16:59:37 03Denis 'GNUtoo' Carikli  07org.openembedded.dev * r4534826895 10openembedded.git/recipes/efl1/eet_svn.bb: Oct 28 16:59:37 eet: depend on gnutls(else it fails to compile) Oct 28 16:59:37 If we don't depend on gnutls we have: Oct 28 16:59:37 | CC libeet_la-eet_connection.lo Oct 28 16:59:37 | eet_lib.c:50:28: fatal error: gnutls/gnutls.h: No such file or directory Oct 28 16:59:38 JaMa, ^^^ Oct 28 17:00:02 ~seen ynezz Oct 28 17:00:04 ynezz is currently on #oe (2d 21h 34m 6s) #uclibc (2d 21h 34m 6s), last said: 'mirrors.bbclass and angstrom-mirrors.bbclass'. Oct 28 17:00:23 hm right he is at elce too Oct 28 17:00:28 jo crofton Oct 28 17:00:33 where is the beer Oct 28 17:03:04 in the bar Oct 28 17:03:11 we are playing a game for prizes Oct 28 17:03:20 hehe Oct 28 17:03:31 prizes from TI? Oct 28 17:03:38 from CELF Oct 28 17:03:57 something special? Oct 28 17:04:20 a camera from sony that runs linux Oct 28 17:06:36 Crofton: heh sounds nice, so far I was aware of an old study at Ricoh only Oct 28 17:07:27 Crofton: If some OE people plan to meet for lunch later I might join Oct 28 17:07:52 ok Oct 28 17:07:56 where are you? Oct 28 17:08:37 hi Oct 28 17:09:39 woglinde: hey! sorry for the wait. This is what I found./opal/config/opal_config_asm.m4 Oct 28 17:11:15 khem, today i've pulled again after a long time, and now i get this funny error: http://paste.debian.net/98490/ Oct 28 17:12:35 does that sound familiar to someone? AS_NEEDED(-ltinfo) in a .so Oct 28 17:13:36 blindvt seems it copied the .la file instead of .so Oct 28 17:29:37 well no i'm using gold. let's use libtinfo.so which should work for both gold and bfd.. Oct 28 17:32:01 (i assume this was already fixed in gold, i'm using the rather old debian version ATM) Oct 28 17:32:15 re florian Oct 28 17:32:39 blindvt whats so much better on gold? Oct 28 17:35:34 woglinde, i personally like it's codebase Oct 28 17:48:35 woglinde: thanks for taking a look my open mpi troubles :) Oct 28 17:49:21 dpackard hm Oct 28 17:49:25 sorry Oct 28 17:49:30 didnt looked yet at it Oct 28 17:49:57 hey, i'm not in a hurry Oct 28 17:50:13 I'm just happy for the help Oct 28 17:50:25 ;) Oct 28 17:59:52 khem, in case anyone's interrested: http://busybox.net/~aldot/oe/0001-ncurses-fix-linker-script-to-also-work-with-gold.patch Oct 28 17:59:58 * blindvt` & Oct 28 18:02:29 khem: lemme know if/when you've got a minute to help me with this python/distutils/libgcc_s.so stuff Oct 28 18:15:53 khem, it would be nice if you could review http://www.mail-archive.com/openembedded-devel@lists.openembedded.org/msg08746.html (from august) Oct 28 18:16:47 blindvt`: hey dude how are you Oct 28 18:16:50 gm all Oct 28 18:17:14 blindvt`: will that ncurses fix work ok with ld too I doubt Oct 28 18:36:12 kergoth: around ? Oct 28 18:36:42 kergoth: I could use your braincells to get that recipe override thing going Oct 28 18:43:05 https://wiki.archlinux.org/index.php/QEMU#Starting_qemu_virtual_machines_on_boot Oct 28 18:43:31 JFYI: how to start qemu/kvm VMs on boot Oct 28 18:43:44 based on ArchLinux Oct 28 18:44:12 and most interesting - how to stop VMs on shutdown :) Oct 28 18:48:26 khem: with my patch i get CURL_FEATURES="zlib,cookies,crypto-auth,dict,file,ftp,http,telnet,tftp" Oct 28 18:48:31 in bitbake -e curl-native Oct 28 18:48:43 because only CURL_FEATURES is weak, not CURL_FEATURES_virtclass-native Oct 28 18:55:25 mzs: see how the anon python is evaled Oct 28 18:55:45 msz see the with-gnutls argument being passed to configure Oct 28 18:55:50 that should not happen Oct 28 18:56:08 out to lunch Oct 28 19:13:46 is there anyone that can advise on compiling kernel for ipaq h2200?? Oct 28 20:32:04 03Klaus Kurzmann  07org.openembedded.dev * rd2ad2e07bc 10openembedded.git/recipes/base-files/base-files_3.0.14.bb: Oct 28 20:32:04 base-files: bump PR to get /etc/profile.d built Oct 28 20:32:04 Signed-off-by: Klaus Kurzmann Oct 28 21:06:16 what regarding autoconf about a recipe would get staged somewhere that do_clean doesn't take care of? Oct 28 21:09:01 tharvey: reframe the question plz Oct 28 21:09:48 I'm occasionally running into autoconf issues with recipes that I'm mucking with such that do_configure fails with things like configure.ac:87: error: possibly undefined macro: PKG_CHECK_MODULES - I know if I wipe TMPDIR and rebuild it will be ok, but if I clean the package and rebuild its not Oct 28 21:10:49 not clear how I get into this state but if I understood how to recover it would save a lot of time as well as perhaps educate me on the ways of autotools Oct 28 21:11:11 I think something gets staged in your sysroots Oct 28 21:11:16 that interferes Oct 28 21:11:54 PKG_CHECK_MODULES is for pkg-config talking to autoconf Oct 28 21:14:30 tharvey: most probably its a bug in the package's m4 macros Oct 28 21:15:28 an example of this is here: http://pastie.org/1256514 Oct 28 21:16:39 so its possible I cntl-c'd a build in the middle of staging a package to sysrootfs and I would need to clean that particular pkg? Oct 28 21:17:10 if something is staged to sysroots, cleaning the package won't get it unstaged Oct 28 21:17:23 good point Oct 28 21:22:19 tharvey: is there pkg.m4 staged in your target sysrtoo Oct 28 21:26:32 yes there is one for that pkg/arch: tmp/work/armv7a-commonos-linux-gnueabi/gst-rtsp-0.10.5-r0/gst-rtsp-0.10.5/common/m4/pkg.m4 Oct 28 21:28:36 doh sorry... yes there are some in sysroot Oct 28 21:28:55 tmp/sysroots/x86_64-linux/usr/share/aclocal/pkg.m4 tmp/sysroots/armv7a-commonos-linux-gnueabi/usr/share/aclocal/pkg.m4 Oct 28 21:30:40 ok, my guess is something is wrong in that file Oct 28 21:30:52 pkg.m4 it might need some sort of update Oct 28 21:30:53 neither of those differ from what is in a 'good' tmpdir Oct 28 21:31:23 ok move this to some other place in staging Oct 28 21:31:36 then clean and rebuild your recipe in question Oct 28 21:32:09 * khem has a beer bash at work today in an hour or so Oct 28 21:32:23 * tharvey is jealous Oct 28 21:42:07 khem: Cheers mate! :-) Oct 28 21:43:53 likewise: hey Oct 28 21:44:29 rmoving pkg.m4's clean/rebuild pkg did re-generate tmp/sysroots/x86_64-linux-sdk/opt/commonos/commonos-devel/arm/share/aclocal/pkg.m4 but still fails in the same way :( Oct 28 21:45:02 tharvey: hmmm ok Oct 28 21:45:48 tharvey: you could try this little testcase Oct 28 21:46:43 tharvey: create a file called configure.ac Oct 28 21:46:45 and add Oct 28 21:46:46 AC_INIT Oct 28 21:46:46 PKG_CHECK_MODULES([FOO], [bar]) Oct 28 21:46:53 and save it Oct 28 21:46:57 then run aclocal Oct 28 21:47:01 khem: I'm stuck in Cambridge, UK for ELCE2010. Is there anything I should bring to the OEDEM attention tomorrow? Oct 28 21:47:04 and then run autoconf Oct 28 21:47:10 does that testcase work Oct 28 21:47:25 likewise: I have plenty Oct 28 21:47:43 khem, would need to run that testcase in a devshell correct? Oct 28 21:47:46 likewise: 1. Release. I would like to make one in Jan 2011 Oct 28 21:48:01 khem: as in stable? Oct 28 21:48:02 tharvey: yes Oct 28 21:48:23 likewise: yes having 2 releases in a year Oct 28 21:48:32 khem: I agree Oct 28 21:48:54 likewise: another thing is with continuous build Oct 28 21:49:10 khem: +1 Oct 28 21:49:10 khem: counterarguments are that the stable branch is not maintained. A counter-counter argument is that a unmaintained stable branch is better than non at all. Oct 28 21:49:14 I am working on setting hudson which will trigger a build of *common* combos Oct 28 21:49:18 on a commi Oct 28 21:49:20 commit Oct 28 21:49:42 likewise: I will take care of maintaining it if no one does Oct 28 21:50:12 likewise: maybe it's not necessary to have a stable branch, but at least a 1 month stabilization period before tagging a release Oct 28 21:50:15 likewise: but releases will bring us users Oct 28 21:50:32 khem: my reasoning is as follows: there is a stable branch. If you use it, maintain it. If you don't use it, don't bother. In other words, put the maintenance burden on the primary users. Oct 28 21:50:34 ericben: yep thats exaclty what I plan to do tag it Oct 28 21:50:38 and branch only if needed Oct 28 21:51:29 likewise: with weekly testing our master will be stabler Oct 28 21:51:38 khem: that would imply that during stailization period, major changes are postponed and only minor changes &fixes are acceptable Oct 28 21:51:44 and hopefully cutting out a release will be easier Oct 28 21:51:53 ericben: yes Oct 28 21:51:57 that all has to be planned Oct 28 21:52:00 that seems a good thing Oct 28 21:52:07 khem, that tests produces no stdout/stderr Oct 28 21:52:07 but for attracting more users we need to release Oct 28 21:52:14 yes Oct 28 21:52:27 even for actual user Oct 28 21:52:32 tharvey: hmm that means PKG_CHECK_MODULES is working in your sysrtoo Oct 28 21:52:52 tharvey: I think it will need debugging the confgure of the package itslef them Oct 28 21:53:24 likewise: another thing I want to bring to notice is obsoletions Oct 28 21:53:41 and resurrect bitbake world Oct 28 21:53:42 ya, probalby not worth it - just annoying that I've run into this several times int he past couple of weeks. sucks to ruin a workday by having to do a clean build Oct 28 21:53:48 thx for help Oct 28 21:54:07 tharvey: I can fix it surely but I will have to dig deep Oct 28 21:54:18 right now I only wait for beer bottles Oct 28 21:55:08 likewise: we need to drop certain old unmaintained versions of recipes Oct 28 21:55:16 and unused ones Oct 28 21:55:26 khem: there was a big announcement on http://yoctoproject.org/ yesterday on ELCE and last week on the web Oct 28 21:55:34 likewise: yes read that Oct 28 21:55:51 so its OE -> Poky -> yocto Oct 28 21:56:24 although intel is too involved in yocto that might have some questions on its non x86 support Oct 28 21:56:54 I know there are folks from WRS on it but then WRS is intel too now a days Oct 28 21:57:16 but I see yocto as a good sign for OE Oct 28 21:57:40 khem: ok, also agree on that; we should be wary but in general the backing is a good sign Oct 28 21:58:17 likewise: OE is a mothership of distros and pretty unbiased and broader in appeal Oct 28 21:59:31 likewise: that also brings a new question to table that we need to work more on getting OE and poky syncs going forward Oct 28 21:59:40 I'll see if I can address some issues during the meeting. Oct 28 21:59:45 and one reason why it does not happen is OE's intertia Oct 28 22:00:36 there are some good changes in poky that can be merged but OE support is supposedly large Oct 28 22:00:49 one idea is we move (most of our) metadata towards the yocto framework; it requires us to be selective and adapt to the cleaner framework Oct 28 22:00:52 so we need to keep what we can maintain in good shape Oct 28 22:01:06 another idea of course is to backport Oct 28 22:01:17 well you have it otherway around Oct 28 22:01:55 but we both mean same Oct 28 22:02:10 more seamless workflow between poky and OE Oct 28 22:03:44 likewise: We dont have clean profiles in OE Oct 28 22:03:48 splitting recipes in several layers like poky did could fit most needs : poeple who want to restrict to a fex catagories can do and people who want unmaintained recipeds (or not that much used recipes) can do it Oct 28 22:03:56 which means the recipes are loosely bound Oct 28 22:04:09 some stuff which is machine specific is appearing for all machines Oct 28 22:05:18 most of TI OMAP stuff e.g. should not be generic armv7 but right now it is Oct 28 22:05:38 the only problem with having several layers is when comes the time to choose where to put a recipe Oct 28 22:05:46 so givena machine it should not pick incompatible recipes Oct 28 22:06:17 ericben: well its not a hard and fast rule that you put it in one place and it cant be moved Oct 28 22:06:36 khem: yes, maybe it's even posible to have an incubator layer Oct 28 22:06:38 suppose a package that is now beagleboard specific might be ported to other machines Oct 28 22:06:46 and then can be moved to more general collection Oct 28 22:06:53 where a recipe starts and gets readyu to be integrated Oct 28 22:07:11 ericben: but out threshold is too low Oct 28 22:07:27 we just accept everything that comes our way :) Oct 28 22:08:14 or maybe layers for soc instead of machines. If a recipe is ok for beagle it should be ok for all other boards based on the same CPU as beagle ? Oct 28 22:08:37 ericben: its hard to say it really depends on nature of app Oct 28 22:08:53 sometimes they may be requiring stuff that only beagle has Oct 28 22:09:04 khem: yes of course Oct 28 22:09:07 Depends on why it's only for Beagle - if it's depending on some hardware feature that's not available on other boards (eg, form factor) Oct 28 22:09:08 but it can evolve Oct 28 22:09:28 and we can let it evolve without stepping on other's toe Oct 28 22:09:59 ericben: don't make too many (possibly) false assumptions on how to structure a project based on just a few machines/examples, that's what got us caught up in the current mess. Oct 28 22:10:02 yes but spliting with a too fine grain will make recipe duplicated and not in sync (ex : bootloader, kernel ...) Oct 28 22:11:20 likewise: you're right Oct 28 22:11:26 ericben: agreed, but we have to be careful about specialization as well as generalization Oct 28 22:12:35 OE has the smarter way to keep a recipe for multiple architectures and we should bank on it most of the time but sometimes we have to specialize Oct 28 22:12:53 khem: do you have a moment to help with the python/libgcc_s.so stuff? if not, no worries. Oct 28 22:13:16 and now how many levels of recipes we want to keep is something to think about Oct 28 22:13:27 vorporeal: sure Oct 28 22:13:37 in my projects here I'm having 4 collections : OE then company generic (base images, psplash) then machine specific (bootloader, kernel, cpu related libs/drivers), then project specific (applications, amend to oe recipes) Oct 28 22:14:33 khem: so three packages (python-fuse, python-numeric, and python-pygame (which depends on numeric)) are failing to build, citing /usr/lib/libgcc_s.so as an unrecognized file format Oct 28 22:14:41 and, of course, have -L/usr/lib in the linker flags Oct 28 22:15:03 vorporeal: are u using libtool 2.4 ? Oct 28 22:15:36 oh its python it might not be using libtool at all Oct 28 22:16:22 and with only 4 collections sometimes it's not easy to follow when 2 machines share the same CPU (but I'm quite new to OE so I don't have an enough global view of the beast so that's only my little experience ;) ) Oct 28 22:16:55 vorporeal: hmm once I was playing with idea of forcing linker to look into sysroot before it looks elsewhere but this may change the semantic in ways that make it more difficult Oct 28 22:17:46 ericben: there are several different ways the layers can be structured Oct 28 22:17:50 khem: yeah, i had similar issues with autotools packages, so i switched (at your suggestion) to angstrom-2010.x to get lt2.4 - these use distutils Oct 28 22:18:34 vorporeal: ok, do you have libgcc in /usr/lib Oct 28 22:19:20 khem: no, it's in /lib Oct 28 22:19:24 ericben: you could create based on machine subsystems then you can create based on device profile Oct 28 22:22:17 vorporeal: hmmm that seems oddball, could you move it to /usr/lib and see if linker finds it ? Oct 28 22:22:28 vorporeal: and paste me the failing linker commandline somewhere Oct 28 22:23:13 vorporeal: I specially added -L sysroot to distutils class Oct 28 22:23:28 just so that it finds these libs Oct 28 22:25:24 khem: would a symlink in /usr/lib suffice? Oct 28 22:25:40 vorporeal: should be ok Oct 28 22:26:02 likewise: so these are few points for you to represent me there at OEDEM :) Oct 28 22:26:05 khem: here's the link command: http://pastebin.com/Z9fjPuzW Oct 28 22:26:46 khem: what do you mean with "clean profiles" missing in oE? Oct 28 22:26:56 hi hrw Oct 28 22:27:15 likewise: if I choose a machine then it picks up incompatible recipes Oct 28 22:27:18 ho Oct 28 22:27:26 * hrw off Oct 28 22:27:29 likewise: so recipes which are not meant for other machines should be marked so Oct 28 22:28:10 likewise: its just that someone was lazy when committing those Oct 28 22:28:17 we need to get them cleaned up Oct 28 22:30:58 khem: looks like it worked perfectly Oct 28 22:31:04 * hrw|gone got efika mx smartbook today Oct 28 22:31:18 khem: ddi you reported that gcc-4.5/linaro cairo regression? Oct 28 22:32:38 hrw|gone: I did Oct 28 22:32:57 that patch I mentioned in the bug is dysfunctional in anycase Oct 28 22:33:39 vorporeal: ok thanks, if you can live with that workaround for a while then I can cook up something in gcc to install libgcc linker stub properly in usr/lib Oct 28 22:33:44 but it wont happen today Oct 28 22:34:47 khem: fine with me, thanks a ton Oct 28 22:50:58 hrm, my rootfs won't be built because there's no package satisfying kexec-tools, which builds -dev and -dbg packages as well as kexec and kdump packages. Oct 28 22:51:17 any way of specifying that kexec-tools doesn't create a kexec-tools package and that the rootfs can satisfy it with kexec and kdump? Oct 28 22:51:32 khem, can I get you to consider my gcc -ibad patch once again. I'm still running this patch and am seeing plenty of failures in do_qa_configure due to Badness with /usr/X11R6/include and /usr/local/include Oct 28 22:52:27 i'm not so attached to that particular patch or methodology, but the current zecke-no-host-includes makes it really difficult to add extra bad dirs Oct 28 22:53:31 vorporeal, what's the error? Oct 28 22:54:20 just RDEPENDS on kexec or klibc in your recipe Oct 28 22:54:52 ant_: cannot satisfy the dependencies for task_pandora_core: kexec-tools, in do_rootfs for pandora_core_image Oct 28 22:59:53 vorporeal, I found it in git.openpandora Oct 28 23:00:06 hm.. Oct 28 23:00:14 task-pandora-core.bb Oct 28 23:00:15 vorporeal: do you have openpandora? Oct 28 23:00:20 ant_: i changed kexec-tools to kexec and kdump Oct 28 23:00:22 106 kexec-tools \ Oct 28 23:00:37 Jay7: yep, working on a toolchain + newer rootfs for it Oct 28 23:00:39 remove and add 2 lines for kexec and kdump Oct 28 23:01:08 vorporeal: cool Oct 28 23:01:26 vorporeal: is it from latest part? i.e. not from part with broken joysticks Oct 28 23:01:53 Jay7: it doesn't have the _latest_ nubs (which are finishing testing/production), but better tested ones Oct 28 23:02:16 not to mention that they're going to do a small re-design on the nubs so they're socketed instead of soldered (and are backwards-compatible with current board designs) Oct 28 23:02:22 vorporeal: how is case quality now? Oct 28 23:02:26 so people with old nubs can remove them and plug in better ones Oct 28 23:03:07 Jay7: no issues with mine - it's sturdy and the paint is evenly distributed. still small issues with warping, but nowhere near deal-breaking Oct 28 23:03:14 * Jay7 wants HW of openpandora in zaurus c860 case :) Oct 28 23:04:29 Jay7: buy a pandora and send it to ben heck for a new case? Oct 28 23:07:08 if two packages provide a file, is there a way to tell OE that one package should overwrite the other package's file? Oct 28 23:07:53 you mean CONFLICTS? RCONFLICTS? Oct 28 23:08:07 i guess? Oct 28 23:08:27 package pandora-skel wants to install .bashrc, which is provided by base-files, hence failure Oct 28 23:08:55 hm.. this could be CONFIGFILES Oct 28 23:09:41 start here Oct 28 23:09:53 http://docs.openembedded.org/usermanual/usermanual.html and fwiw s/CONFIGFILES/CONFFILES/ Oct 28 23:10:47 couple of lines above Oct 28 23:11:49 probably not CONFLICTS/RCONFLICTS, as i assume pandora-skel doesn't provide all of base-files, just a couple of specific files Oct 28 23:11:55 i'll give CONFFILES a try Oct 28 23:13:44 ant_: ah, this package is providing skeleton files to be used for user creation - something must be getting put in the wrong place Oct 28 23:14:57 maybe i should specify a different FILES variable for the pandora system in base-files? Oct 28 23:15:56 oh, looks like Oct 28 23:16:38 hm, no, I don't see anything about bash.rc Oct 28 23:16:51 err .bashrc Oct 28 23:23:51 ant_: i don't know recipe syntax very well, but would it be possible to do a do_install_append_omap3-pandora() to prevent base-files from packaging skel/.bashrc and skel/.profile only on pandora builds? Oct 28 23:30:47 ant_: that fixed it :D Oct 28 23:34:16 I think you should merge pandora-skel here Oct 28 23:34:30 like other machines do Oct 28 23:34:54 ant_: i'm not a dev for the project, just a technically-savy user playing around ;) Oct 28 23:34:59 although i'll pass on the message Oct 28 23:35:06 sorry but I absolutely have no idea about openpandora Oct 28 23:35:38 imagine a totally self-contained beagleboard Oct 28 23:35:58 I mean relationship wrt git.openpandora and OE Oct 28 23:36:02 ah Oct 28 23:36:09 no idea about the overlay structure Oct 28 23:36:33 it's a separate set of packages which take precedence over the OE defaults, and they mirror the OE tree Oct 28 23:36:43 seems to me that having to change base-files and add a custom skel is a double work Oct 28 23:36:43 but it's quite out of date, hence me mucking around with stuff Oct 28 23:37:23 btw th ekexec-tools split is not old, just a couple of weeks ago Oct 28 23:37:45 breakage is recent :) Oct 28 23:39:38 nobody's touched the openpandora.git repo in at least a few weeks, though Oct 29 00:02:36 this is a weird question, but has anyone ever booted up a window manager (e.g. gnome, enlightenment) and found the icons to display fine, but all the text is little black rectangles? Oct 29 00:03:06 i imagine it's a problem with some text rendering or font library, but i'm not an expert Oct 29 00:05:55 mrj10, more likely you are missing fonts Oct 29 00:07:07 grg: any idea which recipe i might try installing via opkg from the device or distcleaning and rebuilding on my compile machine? Oct 29 00:07:28 mrj10, sorry i have no idea Oct 29 00:07:41 ok Oct 29 00:07:49 i have the same problem in gnome and enlightenment Oct 29 00:08:08 mrj10, what image did you build? Oct 29 00:08:32 i'm off the beaten path, it's a modified version of beagleboard-demo-image Oct 29 00:08:50 with angstrom 2010.x and linux-omap-2.6.36rc Oct 29 00:09:00 i suppose you look at the differences between your modified image and a stock image Oct 29 00:09:09 (if the stock image works) Oct 29 00:09:26 well, i tried linux-omap-2.6.36rc, and everything worked except dvi, so i downgraded back to linux-omap-psp-2.6.32 Oct 29 00:09:52 yeah, at this point it's likely easier to set up a new openembedded tree and port over the changes i know work Oct 29 00:09:54 grg: if you use 4.5 then we have a new warning called poison-system-directories Oct 29 00:10:08 grg: you can add that to CFLAGS Oct 29 00:10:09 i made a bunch of experimental changes that didn't end up working, and they probably put some other libraries in a bad state Oct 29 00:10:12 -Wpoison-system-directories Oct 29 00:10:23 that should help Oct 29 00:14:23 khem, ok i'll look ito that, thanks. What about recipes that override CFLAGS? Oct 29 00:21:06 heym anyone has any idea how I can find the position of ROM Table on Cortex A8 ? is there any way to detect it dynamically ? Oct 29 00:21:58 khem, I can't find the poison-system-directories option in the docs (certainly not here: http://gcc.gnu.org/onlinedocs/gcc-4.5.1/gcc/Option-Summary.html) any pointers? Oct 29 00:38:33 I see it now: gcc/gcc-4.5/linaro/gcc-4.5-linaro-r99297.patch Oct 29 00:39:55 But this does not address the issue that it is not a simple process to add more poisoned directories. Oct 29 00:50:21 * saa sees: .../beagleboard-angstrom-linux-gnueabi/linux-omap-psp-2.6.32-r90+gitra6bad4464f985fdd3bed72e1b82dcbfc004d7869/temp/run.do_compileconfigs.25634: line 679: [: missing `]' Oct 29 00:51:07 Looks like a missing space in "linux.inc: add knob to turn off cortex a8 fixups": http://cgit.openembedded.org/cgit.cgi/openembedded/diff/recipes/linux/linux.inc?id=88335ada413098838c3c8315b90183eef1ccd0d3 Oct 29 00:51:44 I think there needs to be a space after "yes" and before the closing square bracket. Oct 29 00:52:40 Should I just send email to koen@openembedded.org? Oct 29 01:02:24 saa, yeah and cc the list with the error you are seeing so that others know about it if/when they hit the same issue Oct 29 01:02:39 okie dokie Oct 29 01:02:43 Thanks. Oct 29 01:02:55 Still working on figuring out the protocol for such things... Oct 29 01:41:10 03Khem Raj  07master * re5bbac3a07 10openembedded.git/recipes/linux/linux.inc: Oct 29 01:41:11 linux.inc: Add space before ending ']' of shell conditional. Oct 29 01:41:11 * Thanks to Scott Anderson for spotting it. Oct 29 01:41:11 Signed-off-by: Khem Raj Oct 29 01:49:41 * saa waves thanks at Khem **** ENDING LOGGING AT Fri Oct 29 02:59:57 2010