**** BEGIN LOGGING AT Tue Aug 31 02:59:57 2010 Aug 31 05:35:19 Morning. I'm compiling my sdk after an update. The last one I made didn't work because it couldn't find libmpfr.so.4 and a couple of other libs on my host. I'm on Ubuntu 10.04, and the versions of the libraries missing aren't in any repos. Does this mean the OE or tasslehoff has done something wrong? Aug 31 05:37:33 It accidentaly worked on my pc, since the library rpath of the executable pointed to the sysroots/i686-linux/usr/lib-folder of the folder where I compiled my sdk. On other peoples machines it failed. Aug 31 06:03:37 tasslehoff: having a rpath to something in your build directory is certainly wrong Aug 31 06:07:34 zecke: I suspected so, but have no idea where it went wrong :) Aug 31 06:08:46 zecke: the rpath is wrong, and libmpfr.so.4 isn't in my toolchain. Aug 31 06:09:04 tasslehoff: well, me neither. you need to trace back the binary using that lib, to the OE package Aug 31 06:11:46 zecke: it's gcc that fails loading shared libraries, more specifically /libexec/gcc/arm-angstrom-linux-gnueabi/4.3.3/cc1 Aug 31 06:17:00 anyone got some info on this error? Aug 31 06:17:01 | * opkg_install_cmd: Cannot install package apache2-native. Aug 31 06:17:02 | * resolve_conffiles: Existing conffile /home/OE/angstrom-dev/rootfs/ct-gcs-image/etc/device_table is different from the conffile in the new package. The new conffile will be placed at /home/OE/angstrom-dev/rootfs/ct-gcs-image/etc/device_table-opkg. Aug 31 06:17:11 I keep getting it in various places Aug 31 06:17:15 tasslehoff: then figure out which recipe that is, then find the build of that in the workdir Aug 31 06:17:39 but I'm not too sure why or how to fix it.. Aug 31 06:19:18 gm Aug 31 06:21:07 porcoesphino: it is very weird that there is a /etc/device_table in the apache-native recipe Aug 31 06:21:26 haha Aug 31 06:21:30 someone had opkg issues too yesterday or so, no idea if they are related Aug 31 06:21:35 that doesn't help :P Aug 31 06:21:56 I had issues with this sort of thing getting gcc-native and another toolchain app installed Aug 31 06:22:01 porcoesphino: you're restoring from pacakged staging ? Aug 31 06:22:15 the task-sdk-native wasn't compiling at that stage Aug 31 06:22:33 there is an apache-native file in your pstage dir, examine that one (hint: .ipk files are ar files) Aug 31 06:22:38 zecke: it'll take bit more time to compile, but I'll hunt it down afterwards Aug 31 06:22:44 I got around it by moving the work, tmp, deploy extra to another place and building from scratch Aug 31 06:22:51 if it does not have this file then something wrt the opkg metadata is broken Aug 31 06:23:02 hmm.. Aug 31 06:23:15 oh, it's for an image, if that changes anything Aug 31 06:23:20 and I'll look now Aug 31 06:27:02 eFfeM_work: I'm not trying to restore one from staging. I'm calling bitbake -c clean before running the bitbake... Aug 31 06:27:13 what should I be looking for in this .ipk file? Aug 31 06:28:41 tasslehoff: I also started a toolchain build Aug 31 06:29:47 zecke: great. Aug 31 06:35:06 tasslehoff: okay... i fail early with Aug 31 06:35:09 cp: `/data/openembedded/openembedded/recipes/tasks/task-sdk-host.bb' and `/data/openembedded/openembedded/recipes/tasks/task-sdk-host.bb' are the same file Aug 31 06:36:05 zecke: hm. didn't I have something like that when my build tried to deal with symlinks? Aug 31 06:36:22 no symlinks. need to figure out who wants to copy that Aug 31 06:57:17 Hi all Aug 31 07:03:32 porcoesphino, eFfeM_work i've looked into the device_table error previously. Aug 31 07:03:58 the "Cannot install package apache2-native" message appears to be separate to the device_table error. Aug 31 07:04:14 its just all the deferred errors are printed out at the end Aug 31 07:04:30 the device_table is added by initscripts Aug 31 07:04:50 and also in image.bbclass Aug 31 07:09:04 well, i might go home early... the bureau's radar images show rain and i'm riding today Aug 31 07:09:09 bye Aug 31 07:23:38 morning Aug 31 07:25:34 hi hrw Aug 31 08:33:21 If I compile an SDK on my 64-bit Ubuntu, will people using 32-bit be able to use it? Aug 31 08:39:41 no Aug 31 08:40:07 you want a canadian cross sdk (or such) and the support is not really working (AFAIK) Aug 31 08:43:04 tasslehoff: RPATH...my toolchain has this as well.. and it is a bug (maybe even a security issue... if people can create your path) Aug 31 08:43:58 zecke: hm. what about the other way around? will I be able to use a 32-bit sdk on my 64-bit install? Aug 31 08:44:03 zecke: ouch Aug 31 08:44:55 tasslehoff: the other way around is more likely to work (people can install a 32bit compat) Aug 31 08:49:36 okay, we are using BUILD_LDFLAGS instead of SDK_LDFLAGS... but this means I will need to ask as well Aug 31 08:50:01 back to the original problem. the sdk-gcc wants libmpfr.so.4, which is neither on my host or in the compiled sdk. is this right? should the library have been in the sdk? Aug 31 08:50:16 zecke: ok. could that cause the above problem as well? Aug 31 08:54:47 tasslehoff: well, it does not explain the lack of the mpfr libs and such Aug 31 08:57:29 zecke: they should have been part of the sdk? Aug 31 08:57:45 tasslehoff: no idea yet Aug 31 08:57:56 I will probably send an email and see if khem or RP can help. Aug 31 09:00:10 03Koen Kooi  07org.openembedded.dev * r4e3cee4562 10openembedded.git/recipes/abiword/abiword_2.8.6.bb: abiword: add 2.8.6 Aug 31 09:01:34 zecke: cool. thanks. Aug 31 09:05:22 huhu Aug 31 09:05:40 good morning Aug 31 09:17:18 how to revert only one commit and keep intack all the other commits Aug 31 09:17:35 Noor: git revert SHA1? Aug 31 09:17:51 when I use git reset --hard it reverts all the commits Aug 31 09:19:08 git revert SHA Aug 31 09:19:17 where SHA is the SHA1 of the commit Aug 31 09:19:28 or in your terms git revert Aug 31 09:19:40 OK Aug 31 09:21:25 all I have in RDEPENDS in task-tassle-toolchain-host.bb is qt4-tools-sdk. don't know if something more there could have given me the libs I'm missing. Aug 31 09:26:43 die apache runtime, ist das ein krampf zum cross compilen Aug 31 09:26:45 oops wrong window Aug 31 09:26:47 :) Aug 31 09:32:41 eFfeM_work: it says error: Entry 'recipes/tcltk/tcl_8.5.8.bb' not uptodate. Cannot merge. Aug 31 09:32:56 although i have commited it Aug 31 09:33:04 and i can see it in log command Aug 31 09:40:19 Noor i think this is because your tcl patch is not added upstream Aug 31 09:40:48 but cant I move to the previous file Aug 31 09:40:59 that I committed Aug 31 09:41:09 I mean to the original file Aug 31 09:41:09 you can always use git format-patch to get a series of patches, then hard go back then patch again Aug 31 09:41:51 but I am not really a git wiz might be your tcl recipe in recipes differed from what you committed Aug 31 09:41:52 the problem is I commited a file locally Aug 31 09:42:07 then i realzed that there is a problem in the commit Aug 31 09:42:34 OK Aug 31 09:42:55 is there any way to get the file that is present in original repo Aug 31 09:44:03 git checkout Aug 31 09:51:03 git checkout -b temp-branch Aug 31 09:51:26 then you can play with rebase Aug 31 09:54:48 eFfeM_work: please ignore my second main on tk Aug 31 09:54:54 i send it accidently Aug 31 09:54:56 ok Aug 31 10:29:16 florian: is it useful to get all those gtk+ versions back or only specific ones Aug 31 10:42:04 eFfeM_work: there are 3 - 4 pathces which are not merged yet Aug 31 10:42:14 and i didnt get any feedback on it as well Aug 31 10:42:27 hello, I would like build meta-toolchain-qte there is this error: ERROR: Nothing provides runtime dependency binutils-cross-sdk Aug 31 10:42:27 [oe] uicmoc3-native: convert to new style staging, remove 'do_stage()' Aug 31 10:42:46 http://pastebin.com/eTPhFfE9 Aug 31 10:42:50 [oe] tzcode-native.inc: Converted new style staging, remove do_datge() Aug 31 10:44:20 any idea? Aug 31 10:56:35 03Michael 'Mickey' Lauer  07org.openembedded.dev * r2a38acbdbd 10openembedded.git/recipes/vala-dbus-binding-tool/vala-dbus-binding-tool_git.bb: vala-dbus-binding-tool: catch up with vala 0.9.6.1 Aug 31 11:12:53 morning GNUtoo|htcdream Aug 31 11:14:48 noor, might have missed these thought i once peeked into uicmoc3 though Aug 31 11:15:47 time permitting i might look into it later this week, probably not tonight as school starts again, so from now on my tue evenings are spent on lecturing Aug 31 11:16:11 but there are also lots of other people on the list who can merge them (including tom) Aug 31 11:27:30 eFfeM_work: u teaches as well? Aug 31 11:27:47 yes, part time, network management Aug 31 11:28:36 but also the low level stuff Aug 31 11:32:32 eFfeM_work: hi i would like build meta-toolchain-qte but there is this problem http://pastebin.com/eTPhFfE9 Aug 31 11:32:42 can you help me Aug 31 11:32:51 ERROR: Nothing provides runtime dependency binutils-cross-sdk Aug 31 11:34:44 guess your tree is broken or your conf is wrong, binutlis-cross-sdk lives in recipes/binutils in your tree Aug 31 11:37:23 i think conf Aug 31 11:37:40 how can i modify my local.conf Aug 31 11:38:08 i must add ASSUME_PROVIDED ? Aug 31 11:38:50 no idea, i am not an sdk expert Aug 31 11:39:11 ok Aug 31 11:42:33 eFfeM_work: since went you started woking in OE? Aug 31 11:42:55 *when Aug 31 11:44:00 don't remember exactly, 2005 or 2006 I think, my roots in oe are nslu2 based http://www.nslu2-linux.org/wiki/Profiles/EFfeM not sure when I moved to oe Aug 31 12:01:08 eFfeM_work: have you any idea how can i build qte my distro is opendreambox-openpli machine is dm600pvr Aug 31 12:01:33 nope, suggest you talk to the distro maintainer Aug 31 12:01:59 (btw this is not a distro support channel) Aug 31 12:04:08 anyone have any thoughts on which install directories in work would be best for testing out of Aug 31 12:04:26 as in move it to the target and set up ld.so.conf etc Aug 31 12:06:43 althought the reailty is, I will be better off building an image with the package installed Aug 31 12:06:52 to amke sure all the depends make it Aug 31 12:18:36 there is channel of openpli? Aug 31 12:23:14 kamel: no idea, guess you got some info on it which got you started in the first place, would suggest to look there Aug 31 13:15:50 I'm trying to build a meta-toolchain for an avr32 system, but I get an error: "*** BFD does not support target avr32-oe-linux-uclibc." while building gdb-cross-sdk-7.1-r6.0 - Does anyone know how to fix or add support for this target? Aug 31 13:20:04 I have an SDK where GCC won't launch because it's missing a native libmpfr.so.4. Anyone got a clue if it should find this on my host or in the sdk? Aug 31 13:23:17 guys, is there somebody expert of gsoap? Aug 31 13:23:23 I cannot build a program Aug 31 13:25:15 it complains that it cannot find stdsoap2.h, which is stagend in the directory of the host and not the target Aug 31 13:26:33 ottavio: we do build an app that uses soap, and I had to modify the gsoap package a little to make gsoap as such build, but I do not really remember having a problem with stdsoap2.h Aug 31 13:27:14 which recipe are you using? Aug 31 13:28:10 custom: http://gitorious.digitalstrom.org/dss-oe/dss-oe/trees/master/targets/dss11/recipes/gsoap Aug 31 13:30:57 Jin^eLD, I'll try it and let you know in a few minutes Aug 31 13:34:32 byt the way, Jin^eLD, what do you specify as DEPENDS? just gsoap or something more? Aug 31 13:52:13 Jin^eLD, it works with you recipe! Thank you :-) Aug 31 13:54:15 I have a last question. By default, oe_runconfigure installs files in /usr, while I'd like to change it and use /usr/local Aug 31 13:54:20 how can I overwrite it? Aug 31 14:04:15 re Aug 31 14:04:18 sorry was on the phone Aug 31 14:04:23 * Jin^eLD reading backlog Aug 31 14:05:25 ottavio: I have just gsoap in DEPENDS Aug 31 14:05:38 I guess I should finally take some time and submit the patches then Aug 31 14:05:53 for the gsoap recipe at least Aug 31 14:06:11 not sure about runconf, don't know if you can modify the prefix Aug 31 14:10:56 03Koen Kooi  07org.openembedded.dev * r0ba9d6f94c 10openembedded.git/recipes/qmake/qmake2.inc: qmake2: use the new qt4-tools instead of the unmaintained uicmoc recipes Aug 31 14:24:49 03Steffen Sledz  07org.openembedded.dev * rec73d417a0 10openembedded.git/recipes/ (3 files in 3 dirs): Aug 31 14:24:49 various syslogs: keep syslog running as long as possible on shutdown Aug 31 14:24:49 Signed-off-by: Steffen Sledz Aug 31 14:24:49 Acked-by: Graeme Gregory Aug 31 14:29:36 hello! if I build a "helloworld-image" I get a error: Aug 31 14:29:38 cp: `/oe/openembedded/recipes/helloworld/helloworld_1.0.0.bb' and `/oe/openembedded/recipes/helloworld/helloworld_1.0.0.bb' are the same file Aug 31 14:29:40 distro is "minimal" and nothing special is configured. Aug 31 14:29:42 any ideas? Aug 31 14:34:00 dnjl: fix incoming. Aug 31 14:34:01 03Chris Larson  07org.openembedded.dev * r7b4cb64edc 10openembedded.git/classes/base.bbclass: Aug 31 14:34:02 Revert "base.bbclass: fix bug with absolute file:// uris" Aug 31 14:34:02 Drop this for now, as it breaks other things. Of course, without this, you Aug 31 14:34:02 once again can no longer have absolute file:// paths the way we used to be Aug 31 14:34:02 able to. Will look into it further. Aug 31 14:34:02 This reverts commit 599b03fe999d41114d124ce663b6c59fa57256d8. Aug 31 14:34:08 dnjl: pull and try again Aug 31 14:34:37 kergoth: k, tnx - i will try... Aug 31 14:40:01 kergoth: tnx, this works! :D Aug 31 14:40:10 np, sorry about the trouble Aug 31 14:41:36 * kergoth mutters Aug 31 15:48:50 http://www.redstar.co.jp/shogo.htm Aug 31 15:49:11 OS: Angstrom Linux Aug 31 15:49:49 Jay7: redstar reminds me North Korea Linux distro Aug 31 15:50:01 :) Aug 31 15:50:06 :) Aug 31 15:50:32 it reminds me my remote past :) Aug 31 15:51:27 I thought the Chinese had a Distro like that Aug 31 15:51:52 there was RedFlag :) Aug 31 15:52:20 oh thats it Aug 31 15:52:23 Japanese -> Polish google translation is so awful then I prefer to self translate it ;d Aug 31 15:52:48 hrw: japanese -> english is good enough Aug 31 15:52:54 anybody remember Altavista? Aug 31 15:52:59 I'm Aug 31 15:53:39 altavista was awesome back before google, was my engine of choice Aug 31 15:53:44 * kergoth_ mutters Aug 31 15:53:52 yep Aug 31 15:54:28 mine too Aug 31 15:56:50 * kergoth_ kicks off a helloworld-image meta-toolchain bulid to sanity check the fix for the absolute uri fix Aug 31 16:07:08 ericben, hi I'll PM you Aug 31 16:42:16 can anyone answer a question about a bitbake error? Aug 31 16:42:42 trying to bitbake an image for omap3 overo board Aug 31 16:45:03 several files OE doesnt download automatically so i did that manually and did a checksum to point to it Aug 31 16:45:27 this is the error :| cp: cannot stat `/home/amir/oe/tmp/work/overo-angstrom-linux-gnueabi/ti-cgt6x-1_6_1_9-r4/cgt6x_6_1_9/*': No such file or directory Aug 31 16:45:47 i checked and the directory exists Aug 31 16:50:20 03Chris Larson  07org.openembedded.dev * rdd478c59e5 10openembedded.git/classes/base.bbclass: (log message trimmed) Aug 31 16:50:21 base.bbclass: fix bug with absolute file:// uris Aug 31 16:50:21 The code which used host/path from the url to construct the destination of the Aug 31 16:50:21 copy was using os.path.join. os.path.join treats a "/" in an element as an Aug 31 16:50:21 indication to drop all previous elements, so the full path resulted in a Aug 31 16:50:21 destination exactly the same as the source, resulting in a no-op unpack. Aug 31 16:50:21 Fixed by using oe.path.join instead. Aug 31 16:50:24 that's not a fatal error, skedar Aug 31 16:50:40 you can ignore it, its not relevent Aug 31 16:52:17 ok..i can ignore these types and just look for FATAL OE_error..looks like this i think? Aug 31 16:52:29 kergoth, does that fix the problem I was having? Aug 31 16:53:06 Crofton_|work, yep Aug 31 16:53:14 awesome Aug 31 16:53:25 I was just thinking having a fix would be a big help Aug 31 16:53:37 means I can get the modules installed when I build an image Aug 31 16:53:42 sithout extra screwing around Aug 31 16:54:00 it would do os.path.join(workdir, /some/absolute/path/) and end up with a destionation of just /some/absolute/path/ Aug 31 16:54:02 hehe Aug 31 16:54:08 sometime we should make srctree work for kernels that use linux.inc also Aug 31 16:54:12 heh Aug 31 16:54:19 yeah that would be good Aug 31 16:54:33 file solves the problem for now Aug 31 16:54:43 but I can see a use for the srctree aproach also Aug 31 16:55:34 i could see srctree being useful while working on drivers, or fixing kernel bugs, but most kernel devs would likely work outside of oe entirely, i expect, to reduce overhead Aug 31 16:56:01 yes Aug 31 16:56:10 right now, I am in driver bug fix mode Aug 31 16:56:22 but the driver is a modulwe and I just copy it in Aug 31 16:56:33 I certainly do all my work outside the tree, I just totally ignore the kernels OE builds. Aug 31 16:56:36 but when I build an image, it is noce to get all the correct modules Aug 31 16:56:39 to start with Aug 31 16:57:13 srctree lets you have the best of both Aug 31 16:57:54 it looks like I can edit and recompile after OE confuigures and builds the tree once Aug 31 16:58:24 alot of overhead for a cross environment, but if you have a lot of depends, a big help Aug 31 17:06:20 hmm what am I doing wrong? I have FILES_${PN} =+ " ${datadir}/mediatomb " Aug 31 17:06:29 but still get NOTE: /usr/share/mediatomb/mysql.sql Aug 31 17:06:36 CMoH-notebook: the following files were installed but not shipped in any package: Aug 31 17:06:40 and list of the files like above Aug 31 17:06:46 what am I missing? Aug 31 17:07:02 huh? Aug 31 17:07:37 i think you mistyped the nick you addressed, Jin^eLD Aug 31 17:07:49 aem sorry that was tab completion in the irc client Aug 31 17:07:53 I guess I had a tab when pasting Aug 31 17:07:54 :) Aug 31 17:08:02 question was addressed to everyone :) Aug 31 17:09:30 Jin^eLD, add a * Aug 31 17:09:44 ...mediatomb/* Aug 31 17:09:53 and possibly ...*/* Aug 31 17:10:22 tried mediatomb/* as well already Aug 31 17:10:29 did not make any difference Aug 31 17:10:44 I'll try */* Aug 31 17:10:45 well I think the * form is correct Aug 31 17:13:38 kergoth, why do I need to update my repo before your fix works? Aug 31 17:16:13 FILES_${PN} =+ " ${datadir}/mediatomb/*/* " that did not help, FILES_${PN} =+ " ${datadir}/mediatomb/* " did not help either Aug 31 17:16:36 brb, gotta buy some food as long as its open Aug 31 17:18:29 weird, check the definition of datadair Aug 31 17:20:20 Crofton_|work, what do you mean? Aug 31 17:22:10 * Crofton_|work means he ran the bb file and it did not work Aug 31 17:22:17 forgetting I needed to pull :) Aug 31 17:23:23 hehe Aug 31 17:27:13 ok compiling now Aug 31 17:27:25 why won't this computer do what I think it should! Aug 31 17:29:06 Crofton_|work: re-adjust your thoughts or use gravity. Aug 31 17:29:28 gm all :-) Aug 31 17:30:03 * likewise should archive all the first sentences he sees when joining #oe - it would be a funny list of quotes Aug 31 17:35:56 * kergoth_ works on making do_unpack suck less Aug 31 17:42:01 So Aug 31 17:42:12 Are there any should build external kernel modules for say qemux86? Aug 31 17:50:42 re Aug 31 17:54:07 ok I now tried avoiding variables, setting explicitly: FILES_mediatomb-tiny =+ " /usr/share/mediatomb/* " Aug 31 17:54:14 and still same problem Aug 31 17:54:17 it won't package the files Aug 31 17:56:24 I'm running out of ideas... Aug 31 17:57:08 any hints on how to debug this problem? no combination will make the files go into the ipk Aug 31 17:57:27 setting explicitly will be a no-op Aug 31 17:57:35 hm Aug 31 17:57:44 if FILES_foo is set and FILES_${PN} is set, and PN is foo, the latter will *overwrite* the former Aug 31 17:57:50 and you'll end up losing the changes Aug 31 17:58:03 they're two different variables in the metadata Aug 31 17:58:13 aha... Aug 31 17:58:17 because PN isn't necessarily defined at the point where the FILES var is, its postponed until the finalize Aug 31 17:58:22 recipe_sanity will warn you about those Aug 31 17:58:28 well originally I was using FILES_PN =+ but that did not help Aug 31 17:58:43 do a bitbake -e, make sure PACKAGES and FILES are what you think they are Aug 31 17:58:45 check sanity Aug 31 17:59:04 ok, lets see.. Aug 31 18:05:16 hi kergoth, hi Jin^eLD Aug 31 18:05:22 hey likewise ! Aug 31 18:05:27 Jin^eLD: long time no see Aug 31 18:05:39 indeed! Aug 31 18:05:47 Jin^eLD: good to see you hit #oe once every now and then. Packaging up tiny goodies? Aug 31 18:06:06 Is tiny somewhat bigger than minimal? Aug 31 18:06:39 kergoth: I think the problem might be that the name of the package and the name of the installation directory are not the same, FILES_xxx only has this stuff: http://pastebin.mozilla.org/779208, I did not find my addition yet... Aug 31 18:07:01 likewise: actually I am doing more stuff with OE again, currently maintaining a rootfs Aug 31 18:08:19 Jin^eLD, PN is mediatomb-tiny? Aug 31 18:08:47 I guess, the recipe name is mediatomb-tiny_0.12.1.bb Aug 31 18:08:58 named that way because of special compile configuration Aug 31 18:09:10 FILES_${PN} += "foo" Aug 31 18:09:20 bitbake -e mediatomb-tiny|grep -E 'FILES.*foo' Aug 31 18:09:20 FILES_${PN} += " ${datadir}mediatomb/* " Aug 31 18:10:40 that spit out quite a few things... looking Aug 31 18:11:21 well FILES_mediatomb-tiny is definitely missing /usr/share/mediatomb, it only has /usr/share/mediatomb-tiny Aug 31 18:11:49 as if my setting was not there at all Aug 31 18:12:04 are you sure you're editing the right version? ;) Aug 31 18:12:09 bitbake -l Parsing -l Parsing -l Parsing Aug 31 18:12:12 is that file the one it's parsing? Aug 31 18:12:17 I'm an idiot :) Aug 31 18:12:34 I forgot a } after a postinst hook Aug 31 18:12:44 but somehow that did not trigger a parse error Aug 31 18:12:49 interesting Aug 31 18:12:50 the FILES_ statement was last line in the recipe Aug 31 18:12:56 and seems it got swallowed Aug 31 18:12:58 it should have caught that Aug 31 18:13:05 can you open a bug against bitbake about that? Aug 31 18:13:13 for tracking purposes Aug 31 18:13:20 * kergoth_ gets food Aug 31 18:13:42 I'll see if it works if I fix my recipe, then I'll "break" it again, if I can reproduce it I'll open a bug Aug 31 18:14:02 but I am pretty sure that was it Aug 31 18:15:36 lol Aug 31 18:15:38 still same problem Aug 31 18:16:24 mhm Aug 31 18:21:19 ok thats interesting Aug 31 18:22:07 I moved the FILES_ thing to the beginning ot the recipe and before the ipk hooks Aug 31 18:22:09 and now it works Aug 31 18:22:38 Hmmm Aug 31 18:22:58 I wonder if everything that needs kernel sources for stuff is OK with switching to the output from modules_prepare Aug 31 18:23:15 * Tartarus will rfc that once this first pass at making the kernel recipes not so loud is done Aug 31 18:23:39 Or are 2.4 era machines still supported? Aug 31 18:24:38 but I can confirm the missing paren bug... Aug 31 18:27:49 hrm Aug 31 18:30:55 submitted as bug 5469 Aug 31 18:42:17 * kergoth_ kicks do_patch Aug 31 18:48:17 ah, there we are Aug 31 19:09:38 i've got a error: Aug 31 19:09:40 arch/arm/mach-pxa/built-in.o: In function `h4000_set_led': Aug 31 19:09:42 | dma_needs_bounce.c:(.text+0x1bc0): undefined reference to `ipaq_asic3_set_led' Aug 31 19:09:44 ... Aug 31 19:09:46 at compiling linux-handhelds-2.6_2.6.16-hh9.bb. nothing special setup, distro is minimal, bb job was virtual/kernel Aug 31 19:09:48 any hints? Aug 31 19:10:33 (machine is looxc550, but no special mods) Aug 31 19:10:41 dnjl: grep the patches of that kernel, for ipaq_asic3_set_led Aug 31 19:13:07 likewise: but where? and why this happens - there is no modification done... Aug 31 19:38:50 dnjl: have you tried the solution that are mentioned in Google's findings when searching for ipaq_asic3_set_led, such as http://www.mail-archive.com/oe@handhelds.org/msg00498.html Aug 31 19:49:31 likewise: i dont use google :) but i will look at this - tnx! Aug 31 19:59:27 I'm lost in OE namespace. What is that class/whatever that copes with sane toolchains? Aug 31 20:00:48 ah, found it in conf/distro/include/sane-toolchain.inc Aug 31 20:00:53 hi florian, eFfeM Aug 31 20:01:11 re Aug 31 20:04:01 likewise, florian, hi, was about to revert the gtk+ commit, but git revert says dirty index, do you by any chance have an idea about that? Aug 31 20:04:41 and do you want to keep all of 14.1, 16.4, 16.6, 18.0 and 18.3? or should we e.g. only keep 14.1, 16.6 and 18.3 Aug 31 20:05:52 found the dirty index solution, apparently this is because I have uncommitted changes Aug 31 20:05:54 effem: the index is what I think is what you already commit, but didn't push yet Aug 31 20:05:55 need to park them Aug 31 20:06:06 eFfeM: yup, stash the index :-) Aug 31 20:06:28 GIT namespace is not well known to me either Aug 31 20:07:47 there you go Aug 31 20:08:13 03Frans Meulenbroeks  07org.openembedded.dev * r184be7e035 10openembedded.git/recipes/gtk+/ (20 files in 6 dirs): Aug 31 20:08:13 Revert "gtk+: remove old versions" Aug 31 20:08:13 This reverts commit 96503f06a6f1f0b7f9df44d97af90252717d8897. Aug 31 20:08:13 As requested by Florian and Leon because the footprint of Aug 31 20:08:13 these versions is smaller than the newer ones. Aug 31 20:08:13 Signed-off-by: Frans Meulenbroeks Aug 31 20:08:16 sorry it took some time, but school has started so need to lecture again on tuesday evenings Aug 31 20:08:38 eFfeM: no idea, i didn't touch it so far Aug 31 20:09:31 eFfeM: in fat i would like to keep several at least... i do not remember offhand all details which one was useful for which purpose Aug 31 20:09:39 s/fat/fact Aug 31 20:10:32 ok np, though e.g. 16.4 and 16.6 were quite close to each other, but no problem to keep them if it is deemed useful Aug 31 20:13:07 btw the low version numbers were not all gone, we still had 2.18.6, 2.6.10, 2.10.14 and a gtk_1.2 version in total there are now 11 gtk+ recipes Aug 31 20:31:10 anyway, calling it a day, stay well everyone! Aug 31 20:35:51 okay, http://github.com/kergoth/openembedded/commit/b1a799d081fde0f94d4238a3a5e51212acbd7752 Aug 31 20:36:28 I think it's cleaner and more explicit, and doesn't break globs in file: uris Aug 31 20:43:37 kergoth: I like this patch Aug 31 20:43:54 short question: Aug 31 20:43:56 I created a new machine and placed its config to build/conf/machine/looxn520.conf. this will use linux-handhelds-2.6 (arm). Aug 31 20:43:58 where do I place its kernel config? the original was openembedded/recipes/linux/linux-handhelds-2.6-2.6.16/defconfig. if I place it to build/recipes/linux/linux-handhelds-2.6-2.6.16/looxn520/defconfig it will not work. I think placing in dir openembedded is wrong... Aug 31 20:46:44 dnjl: build/recipes/linux/linux-handhelds-2.6-2.6.16/looxn520/defconfig should have worked Aug 31 20:46:58 do you know if you chose the right kernel Aug 31 20:48:10 khem: mmh, is uses openembedded/recipes/linux/linux-handhelds-2.6_2.6.16-hh9.bb Aug 31 20:49:47 dnjl: ok edit linux-handhelds-2.6.inc Aug 31 20:50:01 and add looxn520 to COMPATIBLE_MACHINE Aug 31 20:52:30 khem: k tnx - does it make sense to place the new linux-handhelds-2.6.inc to build/recipes/linux? I want to keep openembedded/ clean Aug 31 20:54:01 I try it.. Aug 31 21:01:53 dnjl: ? its already there Aug 31 21:02:01 what? Aug 31 21:02:07 recipes/linux/linux-handhelds-2.6.inc Aug 31 21:02:38 yes - but i doesnt work if its in build/ - i have to placed it in openembedded/ Aug 31 21:03:02 it has to be in metadat Aug 31 21:03:08 metadata Aug 31 21:03:19 why in build ? I dont get it Aug 31 21:05:11 I wanted to keep the directory openembedded/ clean - I will use it for other projects and keep it in sync with upstream Aug 31 21:05:57 I thought build/ works like a override-dir Aug 31 21:06:17 You can do stuff like this, yes Aug 31 21:06:27 But it's easiest to hack on openembedded, if you plan on pushing it upstream Aug 31 21:07:39 ok, I will do so - I am planning to push up the machine loox n520 if I get it work Aug 31 21:19:19 wtX? arch/arm/boot/compressed/head-h2200.o:(.start+0x0): undefined reference to `no symbol' Aug 31 21:29:45 dnjl: or you can use bblayer Aug 31 21:30:06 dnjl: http://sakrah.dontexist.org/node/2 read this Aug 31 21:31:23 khem: thanks - i will look for this.. Aug 31 21:32:29 khem: cow-mounts whould also do this job I think Aug 31 21:33:34 ahh - its already in oe - nice.. Aug 31 22:11:19 builder@jlime:/home/builds$ cat fresh/jornada7xx/tmp/work/x86_64-linux/gconf-native-2.6.1-r0/temp/log.do_unpack.13895 Aug 31 22:11:22 UH!? Aug 31 22:11:24 cp: `/home/users/builder/openembedded/recipes/gnome/gconf-native_2.6.1.bb' and `/home/users/builder/openembedded/recipes/gnome/gconf-native_2.6.1.bb' are the same file Aug 31 22:11:27 builder@jlime:/home/builds$ Aug 31 22:12:01 flip: update your oe dir Aug 31 22:12:53 dnjl: ah ok, somebody's finger has slipped ;)? Aug 31 22:13:09 i think so :) Sep 01 00:12:36 www.pynell.com Sep 01 00:12:53 Brazilian developer of embedded software! Sep 01 01:30:33 I Want to say thank's for all Openembedded developers. I already tried some compiles but all with error,now i'm compiling normal with all working ! Thanks ! Very much Sep 01 01:38:33 angelox_123: cool. Enjoy Sep 01 01:39:33 base image here - Running task 3613 of 4713 yeah! is going far! Sep 01 01:41:39 If i build a base-image later i can build a gpe-image yes? or i need re-start from scratch ? Sep 01 01:52:51 If i build a base-image later i can build a gpe-image using the same compiled packages?or it will re-build all from scratch ? **** ENDING LOGGING AT Wed Sep 01 02:59:57 2010