**** BEGIN LOGGING AT Wed Oct 20 02:59:56 2010 Oct 20 03:00:33 khem: does the acinclude directory get processed by aclocal by default? i thought things normally used the m4 directory Oct 20 03:01:31 vorporeal: never heard of an acinclude directory. aclocal uses acinclude.m4 if it exists, and any paths specified with -I -- by default, autotools.bbclass will search the source tree and add any dirs which have m4 files to the search path Oct 20 03:01:58 so it doesn't matter what directory the m4 files are in, then Oct 20 03:02:07 unless the recipe overrides acpaths, yep Oct 20 03:02:34 * kergoth_ thinks this is one case where the class is trying to do too much for us Oct 20 03:03:53 kergoth_: well to a certain extent its ok. packages dont put it in standard dirs Oct 20 03:04:08 so in a way it provides some abstraction at some cost Oct 20 03:04:20 i don't think it is. any package providing local macros should be passing the -I's to aclocal with the variable in the toplevel Makefile.am Oct 20 03:04:29 so there should be no need for our search Oct 20 03:04:39 if it doesnt use automake, its not that hard to set acpaths Oct 20 03:05:40 kergoth_: there are packages with dont have makefile.am still use autotools Oct 20 03:05:44 yes, there are Oct 20 03:05:53 [20:01] kergoth_: if it doesnt use automake, its not that hard to set acpaths Oct 20 03:06:17 i think the "magic" in our classes is just encouraging people to add shit without becoming familiar with the buildsystem of what they're adding Oct 20 03:06:23 causing more trouble than its solving Oct 20 03:09:17 khem: still nothing - removing those files results in an empty libsdl-mixer package, and the compile step occurs almost instantly (so it seems as though nothing actually gets compiled), and it never creates a .libs dir with the compiled library files Oct 20 03:11:06 vorporeal: paste your diff Oct 20 03:12:47 i just manually removed them :) Oct 20 03:12:56 between patch and configure Oct 20 03:13:20 vorporeal: well I gave you a recipe to look at and add a do_configure_prepend Oct 20 03:13:35 do_configure_prepend() { Oct 20 03:13:35 cd ${S}/acinclude Oct 20 03:13:35 rm libtool.m4 ltdl.m4 lt~obsolete.m4 ltoptions.m4 ltsugar.m4 ltversion.m4 Oct 20 03:13:35 } Oct 20 03:13:38 err Oct 20 03:13:44 pastebin? Oct 20 03:14:25 i'm a computer science major, but i'm not too accustomed to linux utils yet, hence the manual change - i'll see what i can do Oct 20 03:14:40 is it possible to create mach-o files with oe? Oct 20 03:14:49 neekers: no Oct 20 03:14:57 darn Oct 20 03:15:21 khem: do you know of any toolchains that will work with scratchbox that can create them? Oct 20 03:15:50 khem: also, i looked at the gnutls recipe, and there was no do_configure_prepend there - there weren't any defined functions at all, actually Oct 20 03:17:11 vorporeal: you have to look at master Oct 20 03:18:02 khem: i'm checking master... Oct 20 03:18:06 ok, well thanks Oct 20 03:18:36 khem: strange, guess i wasn't Oct 20 03:18:36 got it Oct 20 03:19:27 neekers: no I dont know Oct 20 03:20:06 neekers: we have iphone-sdk Oct 20 03:20:22 may be that one is something you want Oct 20 03:21:29 khem: yes, i want to use apple's arm gcc but it's not easy getting scratchbox working on mac and i need it to handle dependencies Oct 20 03:24:33 khem: http://pastebin.com/LPxdwUVH (sorry for being so inept) Oct 20 03:25:04 neekers: it may not be functional right now there might be some rotten Oct 20 03:27:23 vorporeal: looks ok Oct 20 03:27:57 you can leave do_compile in there if this does not work Oct 20 03:28:04 alright Oct 20 03:31:27 khem: yeah, still doesn't work (with or without do_compile) Oct 20 03:43:46 khem: i got lazy and just removed the configure option that caused it to use libmad for decoding - that "fixed" it Oct 20 03:43:55 * kergoth_ grumbles Oct 20 03:45:53 khem: you know, if we changed autotools.bbclass to include the system dirs before the local dirs, it would avoid the need for the removal of all these things. the only problem would be recipes with local macro modifications or dependencies on newer macros, either of whicha re likely much more rare than the cases you're fixing, and would be easily dealt with Oct 20 03:46:07 heh Oct 20 03:46:30 hmm Oct 20 03:46:51 course, thats not the way aclocal normally functions, but.. Oct 20 03:47:43 given this is for us specifically, i think its worth considering Oct 20 03:47:57 although, that would cause problems with version upgrades Oct 20 03:48:03 building a new version of something with old version macros installed Oct 20 03:48:06 hrmph. Oct 20 03:48:08 nevermind Oct 20 03:48:38 although, that might not be a concern with packaged staging, due to uninstallation at setscene Oct 20 03:48:40 hmmm Oct 20 03:53:43 khem: alright, more or less back to the original problem - it can't satisfy the dependencies for the toolchain because of libsdl-mixer-dev, which is definitely in the deploy directory (i checked) Oct 20 03:54:22 is there a SRC_URI format for using a local git repo? Oct 20 03:54:48 I want to use my local clone of the kernel tree Oct 20 03:54:54 tharvey: git:///usr/local/foo;protocol=file Oct 20 03:55:06 but you should look into srctree and gitver Oct 20 03:57:11 thaytan: http://bec-systems.com/site/711/openembedded-srctree-and-gitver Oct 20 03:57:18 ahar - that option isn't mentioned in http://docs.openembedded.org/usermanual/usermanual.html#src_uri_variable :) Oct 20 03:57:23 ta Oct 20 03:58:11 git:// just tells bitbake to use its git fetcher, that doesn't control whcih protocol is passed to git on the cmdline (e.g. git:// vs ssh vs file) Oct 20 03:59:09 hey Oct 20 03:59:15 so got me a beagleboard-xm Oct 20 03:59:32 I read that since it doesn't have NAND I have to copy files other than just uImage to the FAT partition Oct 20 03:59:44 is this just u-boot.bin or are there even more files? Oct 20 04:12:16 03Khem Raj  07master * r48d7b34698 10openembedded.git/classes/autotools.bbclass: Oct 20 04:12:16 autotools.bbclass: Move --with-sysroot to CONFIGUREOPTS Oct 20 04:12:16 * This is added to CONFIGUREOPTS so it can be filtered out Oct 20 04:12:16 when needed. Recipes like gcc need this to be filtered Oct 20 04:12:16 Signed-off-by: Khem Raj Oct 20 04:12:35 03Khem Raj  07master * rf1148d2674 10openembedded.git/recipes/binutils/binutils.inc: Oct 20 04:12:35 binutils.inc: Filter out --with-sysroot for target binutils Oct 20 04:12:36 * This is not needed but libtool 2.4 will add it so we filter Oct 20 04:12:36 it out here. Oct 20 04:12:36 Signed-off-by: Khem Raj Oct 20 04:12:36 03Khem Raj  07master * r565b3f4965 10openembedded.git/recipes/gcc/gcc-configure-target.inc: Oct 20 04:12:37 gcc-configure-target.inc: Filter out --with-sysroot Oct 20 04:12:40 * For target gcc we dont need to specify --with-sysroot Oct 20 04:12:40 even with libtool 2.4 Oct 20 04:12:40 Signed-off-by: Khem Raj Oct 20 04:26:24 anyone? Oct 20 04:35:50 aditya_1010: also MLO for x-load, but #beagle would be more proper channel... Oct 20 04:37:31 oh err, was using OE Oct 20 04:37:55 guess I'll ask in beagle then Oct 20 05:02:30 03Denys Dmytriyenko  07org.openembedded.dev * r497f4e756b 10openembedded.git/recipes/gstreamer/gst-plugins-good_0.10.25.bb: Oct 20 05:02:31 gst-plugins-good: add 0.10.25 Oct 20 05:02:31 Signed-off-by: Denys Dmytriyenko Oct 20 05:02:31 03Denys Dmytriyenko  07org.openembedded.dev * r141c81dea1 10openembedded.git/recipes/gstreamer/gst-plugins-ugly_0.10.16.bb: Oct 20 05:02:31 gst-plugins-ugly: add 0.10.16 Oct 20 05:02:31 Signed-off-by: Denys Dmytriyenko Oct 20 05:15:59 is it allowed to set ${AUTOREV}" in an experimental recipe? Oct 20 05:21:00 playya__: yes Oct 20 05:23:31 ok Oct 20 06:05:11 playya__: in public one which is pushed to tree? Oct 20 06:08:00 you mean out tree on fso? Oct 20 06:09:44 morphis/work Oct 20 06:10:11 i just finished my image and I'll push my stuff after getting up Oct 20 06:10:14 good night Oct 20 06:10:27 playya__: I mean that all recipes (except 4 strange kernel recipes) in oe tree have sane SRCREV and AUTOREV set by distro/local config, and that's not bad rule at all Oct 20 06:11:02 ok Oct 20 06:11:17 just want to be shure to get the stuff upstream Oct 20 06:13:25 sure just put SRCREV known to build ok to recipe and AUTOREV to fso-autorev or where it belongs :) Oct 20 06:14:28 playya__: otherwise when you use SRCPV in PV, bitbake will connect to that git repo during every parse of that recipe Oct 20 06:15:08 which will be bad for people not even using this new recipe Oct 20 06:20:32 gm Oct 20 06:24:31 gm Oct 20 06:55:04 For a Core i7 with 4 cores and 8 threads, what should be good settings for PARALLEL_MAKE and BB_NUMBER_THREADS? Oct 20 06:57:37 03Roger Monk  07org.openembedded.dev * rdccf16d06c 10openembedded.git/recipes/egl/egl.inc: Oct 20 06:57:37 egl: Add tu816x as combatible SOC_FAMILY Oct 20 06:57:37 Signed-off-by: Roger Monk Oct 20 06:57:37 Signed-off-by: Koen Kooi Oct 20 07:07:39 tasslehoff: experiment; this would also depend on the amount of memory, disk speed etc, note also that larger values also have a somewhat higher probability to encounter issues wrt synchronisation Oct 20 07:08:16 I seem to recall a guy here runs with -j12 and BBT=4, not sure though Oct 20 07:08:53 my compu is less fast :-( Oct 20 07:09:22 hi all m getting this error: http://pastebin.com/3TbLuvJy while building omap3-console-image. Oct 20 07:09:33 eFfeM_work: noted. I have the i7 Q740@1.73GHz, 4GB RAM, and 7200RPM disk, and have used j4 and BBT4 until now Oct 20 07:10:17 do some benching & if you want to improve maybe add more mem (i think we use 6 or 8 for new systems) Oct 20 07:13:42 will do. Oct 20 07:16:08 another Q. I've been struggling to build qt4-embedded-gles to get the powervr ipk, and yesterday koen told me that he uses "bitbake -b" to compile it. For me, qt4-embedded and qt4-embedded-gles both provide qt4-embedded, and it seems that non-gles wins. How do I solve this? kergoth_ said something about setting PREFERRED_PROVIDER, but I'm not sure how, and the docs are withholding on the subject :) Oct 20 07:17:10 you can indeed use preferred provider, do a grep -r for in in the conf dir, plenty examples, you can just do that in your local.conf Oct 20 07:17:50 and if you do bitbake -b you can select the recipe to build (bitbake -b path/to/the/recipe.bb) Oct 20 07:19:03 eFfeM_work: I made bitbake -b work, but I need to use preferred provider to get it into the big picture Oct 20 07:19:15 see the conf dir Oct 20 07:22:14 03Koen Kooi  07org.openembedded.dev * r47df1036c1 10openembedded.git/conf/distro/include/angstrom-2010-preferred-versions.inc: angstrom next: switch to pango 1.28.3 Oct 20 07:22:29 eFfeM_work: I suspect something like PREFERRED_PROVIDER_qt4-embedded = "qt4-embedded-gles" Oct 20 07:22:38 yes Oct 20 07:22:47 i'm not sure if thre is docs on it Oct 20 07:23:03 * tasslehoff pats self on back and appreciates eFfeM_work helping Oct 20 07:23:09 yw Oct 20 07:24:45 good morning Oct 20 07:52:47 heyho Oct 20 08:21:01 03Simon Busch  07org.openembedded.dev * rdfe42d7630 10openembedded.git/recipes/palmpre/read-tokens_git.bb: Oct 20 08:21:01 read-tokens: add initial git version Oct 20 08:21:01 read-tokens is a little utility especially for the palmpre. It reads so alled 'tokens' Oct 20 08:21:01 from the internal flash and stores them in a key-value file so other system components can Oct 20 08:21:01 read them from there. Oct 20 08:21:01 Signed-off-by: Simon Busch Oct 20 08:21:02 03Simon Busch  07org.openembedded.dev * rdff92c6745 10openembedded.git/recipes/tslib/ (tslib.inc tslib_git.bb): Oct 20 08:21:02 tslib: adjust various recipe settings for the palmpre machine Oct 20 08:21:03 * set PACKAGE_ARCH = "${MACHINE_ARCH}" for palmpre machine Oct 20 08:21:03 * provide special EXTRA_CONF setting to build plugin for the cy8mrln touchscreen Oct 20 08:21:04 Signed-off-by: Simon Busch Oct 20 08:21:14 03Simon Busch  07org.openembedded.dev * r32adfe3f19 10openembedded.git/recipes/netbase/netbase/palmpre/interfaces: Oct 20 08:21:14 netbase: powerup usb0 interface automatically on palmpre machine Oct 20 08:21:14 Signed-off-by: Simon Busch Oct 20 08:21:14 03Simon Busch  07org.openembedded.dev * ra26bdb3e9d 10openembedded.git/recipes/shr/ (2 files in 2 dirs): Oct 20 08:21:14 initscripts-shr: remove unused usb-gadget script for palmpre machine Oct 20 08:21:15 This script was resonsible for activating networking over usb but is not needed anymore as Oct 20 08:21:15 our default kernel for the palmpre machine activates usb networking per default. Oct 20 08:21:16 Signed-off-by: Simon Busch Oct 20 08:21:16 Acked-by: Martin Jansa Oct 20 08:21:17 03Simon Busch  07org.openembedded.dev * r53a9c9a485 10openembedded.git/recipes/tslib/ (tslib/include-cy8mrln-header.patch tslib_git.bb): Oct 20 08:21:20 tslib: add patch for git version to compile cy8mrln plugin Oct 20 08:21:21 The cy8mrln plugin needs the cy8mrln.h which is shipped with the kernel but currently not Oct 20 08:21:21 deployed by any kernel-headers packages. So we include the header file for now directly Oct 20 08:21:21 within tslib. Oct 20 08:21:21 Signed-off-by: Simon Busch Oct 20 08:21:21 03Simon Busch  07org.openembedded.dev * rce435227ea 10openembedded.git/recipes/palmpre/tellbootie_git.bb: Oct 20 08:21:22 tellbootie: add initial git version Oct 20 08:21:23 tellbootie is a little utility for the palmpre machine which tells the bootloader over Oct 20 08:21:23 some spare space in the memory about what to do on next restart (e.g. fall in recovery Oct 20 08:21:24 mode). Oct 20 08:21:24 Signed-off-by: Simon Busch Oct 20 08:40:21 good morning. Oct 20 08:41:57 Sorry if off-topic, How can I "bitbake -c menuconfig virtual/kernel" without working in the X windows desktop? Oct 20 08:42:53 Because my link to the build host is slow, so I want to avoid X. Oct 20 08:43:28 hi mfhk Oct 20 08:43:50 hi florian Oct 20 08:44:30 mfhk: i don't know offhand but if it does not you can do it manually Oct 20 08:44:47 how? Oct 20 08:45:20 Even in devshell, I cannot do menuconfig without X. Oct 20 08:45:20 first -c configure, the run "make ARCH=arm menuconfig" in the working directory and then bitbake again Oct 20 08:46:01 Let me try... Oct 20 08:47:50 florian: hi, did you found the root of the issue with the download speed on angstrom-distro.org? Oct 20 08:47:56 menuconfig requires an xterm with curses in it (afaik); you can alwasy do make oldconfig or so Oct 20 08:48:20 you can set the TERM to screen Oct 20 08:48:32 there's some value for it, in bitbake.conf Oct 20 08:48:38 don't know offhand Oct 20 08:48:44 ah ok never used that Oct 20 08:49:56 # Set a default Oct 20 08:49:56 TERMCMD ?= "${GNOME_TERMCMD}" Oct 20 08:49:57 TERMCMDRUN ?= "${GNOME_TERMCMDRUN}" Oct 20 08:50:36 just change it to Oct 20 08:50:37 SCREEN_TERMCMD = 'screen -D -m -t "$TERMWINDOWTITLE"' Oct 20 08:50:38 SCREEN_TERMCMDRUN = '${SCREEN_TERMCMD} $SHELLCMDS' Oct 20 08:51:20 ynezz: yes should be fixed... ubuntu for some htc phone was guilty :) Oct 20 08:51:25 ynezz, Thanks! I will try. Oct 20 08:51:42 florian: sorry? :) Oct 20 08:51:56 I don't get it Oct 20 08:52:51 ynezz: massive downloads of a large file Oct 20 08:53:07 ah, so :) Oct 20 08:53:15 cool, that it's fixed, thanks Oct 20 09:06:49 Doing "bitbake -b qt4-embedded-gles_4.6.3.bb" worked, so I added PREFERRED_PROVIDER_qt4-embedded = "qt4-embedded-gles" to my image.bb. That makes it build the non-gles variant again though. How can I make my image use the gles-variant? Oct 20 09:08:25 tasslehoff: no idea, I recently also ran into such a case but didn't get to diving into it; actually thought it was my setup, i'm definitely interested in the solution Oct 20 09:09:32 eFfeM_work: as a hack, I guess I could overwrite non-gles with gles, but that wouldn't be a particularly good solution :) Oct 20 09:10:01 tasslehoff: i definitely would not do that, then better just rm the non-gles bb file Oct 20 09:10:05 still yukky though Oct 20 09:11:01 eFfeM_work: tried that, buth then it just picks the previous non-gles. 4.5.something Oct 20 09:11:41 tasslehoff: rm all of them, or add a default preference to the gles one (or a dp of -1 to the non gles one) Oct 20 09:11:52 note that these are still workarounds, not real solutions Oct 20 09:15:38 eFfeM_work: yeah. perhaps I should ask the mailinglist. Oct 20 09:16:23 it might be a bug somewhere (actually you might try with e.g. bitbake 1.8.18 to see if the problem remains) Oct 20 09:16:55 btw not sure if this will work, but you could also append _local. e.g. Oct 20 09:17:08 PREFERRED_PROVIDER_qt4-embedded_local = "qt4-embedded-gles" Oct 20 09:17:17 and stuff that in your loccal.conf Oct 20 09:21:08 eFfeM_work: could stuff that in my image.bb as well? Oct 20 09:21:24 nope Oct 20 09:22:18 the image.bb can as far as I know not specify variants (at least I haven't seen it and I don't know how to do it (which does not preclude that it is still possible :-) )) Oct 20 09:22:51 eFfeM_work: that's what I've done now, so perhaps that's why it didn't work. Oct 20 09:23:12 you should stuff the preferred provider line in your local.conf Oct 20 09:26:30 eFfeM_work: I think that actually did the trick :) Oct 20 09:26:37 great Oct 20 09:26:48 have fun with it :-) Oct 20 09:27:04 hah, I will :) Oct 20 09:56:12 03Martin Jansa  07master * r30a3ea891d 10openembedded.git/recipes/mesa/ (5 files in 2 dirs): Oct 20 09:56:12 mesa: add 7.9 release Oct 20 09:56:12 Signed-off-by: Martin Jansa Oct 20 09:56:12 03Martin Jansa  07master * rfad1dc47e1 10openembedded.git/recipes/mesa/mesa-7.8.2.inc: Oct 20 09:56:12 mesa-7.8.2: remove D_P_shr as we want 7.9 now Oct 20 09:56:12 Signed-off-by: Martin Jansa Oct 20 09:56:13 03Martin Jansa  07master * rafce52663d 10openembedded.git/recipes/talloc/talloc_2.0.1.bb: Oct 20 09:56:13 talloc: add recipe for talloc, needed by new mesa releases Oct 20 09:56:14 Signed-off-by: Martin Jansa Oct 20 09:56:14 03Martin Jansa  07master * r88a9cbf5b0 10openembedded.git/recipes/linux/ (3 files in 3 dirs): Oct 20 09:56:15 linux-openmoko-2.6.34: add patch from qtmoko for resume_reason Oct 20 09:56:16 Signed-off-by: Martin Jansa Oct 20 10:17:00 hi all I am not able to fetch u-boot-omap3 source pls help me Oct 20 10:43:22 anybody looked already on "failure building gcc-cross-initial" thing? Oct 20 10:44:42 morphis: cleaning and rebuilding whole toolchain usually helps in those cases Oct 20 10:44:51 already did that Oct 20 10:45:09 cleaned all gcc/binutils/eglibc? Oct 20 10:45:16 rm -rf tmp Oct 20 10:45:34 somebody on the ML has the same problem Oct 20 10:46:03 I had the same, but rebuild did work for me Oct 20 10:46:24 see http://tinderbox.openembedded.org/public/logs/task/8947680.txt Oct 20 10:46:25 hm Oct 20 10:46:33 I started a second rebuild Oct 20 10:46:36 maybe it works now Oct 20 10:47:38 morphis: config.log is more important Oct 20 10:48:08 conftest.c:11: fatal error: ac_nonexistent.h: No such file or director Oct 20 10:48:27 but ie shr buildhost also built gcc-cross for 2 archs in last 3 days too Oct 20 10:48:39 hm Oct 20 10:51:31 morphis: that's imho not real error, just what "checking how to run the C preprocessor" expects to see Oct 20 10:51:45 ok Oct 20 10:51:54 will wait until the build fails and then look again Oct 20 10:55:23 hi all i m getting error in u-boot-omap3 do_fetch error is :http://pastebin.com/N1dZGEF9 pls help meout Oct 20 10:55:58 florian: 2010-10-20 12:20:26 (0.01 B/s) - Read error at byte 18/46973967 (Connection timed out). Retrying. Oct 20 10:56:28 florian: --2010-10-20 12:38:50-- (try: 4) http://www.angstrom-distribution.org/unstable/sources/eglibc-2_11_svn.eglibc.org_.branches_10690_.tar.gz Oct 20 10:58:27 ynezz: strange... works for me Oct 20 11:02:55 03Martin Jansa  07master * rfabce288e4 10openembedded.git/recipes/mesa/mesa-7.9.inc: Oct 20 11:02:55 mesa-7.9: add talloc dependency Oct 20 11:02:55 Signed-off-by: Martin Jansa Oct 20 11:09:19 JaMa|W: minimal-image seems to build fine Oct 20 11:30:19 morphis: perhaps try a clean build (no pstage or tmp); i could build slugos for nslu2 (le and be) without problem (branch testing-next) Oct 20 11:34:13 eFfeM_work: already tried that Oct 20 11:34:16 two times Oct 20 11:34:19 did not wokred Oct 20 11:34:24 s/wokred/worked/ Oct 20 11:34:28 ah ok, what machine/distro ? Oct 20 11:34:40 i tried slugos, but that is on an older gcc i think Oct 20 11:35:10 and I did minimal/console-image for sheevaplug (which is armv5te) Oct 20 11:38:14 palmpre/shr Oct 20 11:41:38 do minimal and shr use a different gcc version or rev ? Oct 20 11:41:44 has the recipe been changed since ? Oct 20 11:41:55 that are the things you want to look at Oct 20 11:42:54 some infos from config.log: http://pastie.org/1235203 Oct 20 11:43:06 recipe did not changed Oct 20 11:43:16 shr choose new gcc version some days ago I think Oct 20 11:43:33 ah damn it Oct 20 11:43:38 I forgot some lines Oct 20 11:45:06 it tells me /home/morphis/oe/palmpre/tmp/work/armv7a-oe-linux-gnueabi/gcc-cross-initial-4.5-r12.0+svnr164562/gcc-4.5/build.x86_64-linux.arm-oe-linux-gnueabi/./gcc/as: line 83: exec: : not found Oct 20 11:46:28 morphis: yup that's what I had Oct 20 11:46:55 JaMa|W: hm ok Oct 20 11:47:04 should I try to rebuild a second time? Oct 20 11:47:14 äh a third time Oct 20 11:47:35 pastebin /home/morphis/oe/palmpre/tmp/work/armv7a-oe-linux-gnueabi/gcc-cross-initial-4.5-r12.0+svnr164562/gcc-4.5/build.x86_64-linux.arm-oe-linux-gnueabi/./gcc/as Oct 20 11:48:34 morphis@nirvana ...gcc-cross-initial-4.5-r12.0+svnr164562 % ls /home/morphis/oe/palmpre/tmp/work/armv7a-oe-linux-gnueabi/gcc-cross-initial-4.5-r12.0+svnr164562/gcc-4.5/build.x86_64-linux.arm-oe-linux-gnueabi/./gcc/as Oct 20 11:48:34 /home/morphis/oe/palmpre/tmp/work/armv7a-oe-linux-gnueabi/gcc-cross-initial-4.5-r12.0+svnr164562/gcc-4.5/build.x86_64-linux.arm-oe-linux-gnueabi/./gcc/as* Oct 20 11:48:41 it's there Oct 20 11:49:44 and what's on line 83? it want's to exec ${empty var} IIRC Oct 20 11:52:53 mom Oct 20 11:53:54 03Martin Jansa  07master * r85f8336c9b 10openembedded.git/recipes/libxml/libxml2.inc: (log message trimmed) Oct 20 11:53:54 libxml2: enable debug module (default is on) Oct 20 11:53:54 * it's not runtime debug (--with-run-debug or --with-mem-debug) Oct 20 11:53:54 * it's needed by python API (which is not limited to provide bindings Oct 20 11:53:54 based on enabled modules so ie python call param_node.lsCountNode() Oct 20 11:53:54 always calls libxml2mod.xmlLsCountNode(self._o), but this function is Oct 20 11:53:55 available only with debug module enabled. Oct 20 12:02:24 JaMa|W: http://pastie.org/1235220 Oct 20 12:06:38 ORIGINAL_AS_FOR_TARGET="" is that empty var :/ Oct 20 12:07:33 morphis: maybe it's related to http://git.openembedded.org/cgit.cgi/openembedded/commit/?id=f1148d26740e57cba4fe29fd78c234359868a50e Oct 20 12:07:47 morphis: I haven't build from scratch after this commit Oct 20 12:09:28 but that's included only in target binutils (as commit says) so it shouldn't be related Oct 20 12:10:06 hm, will remove this line and try again Oct 20 12:10:48 rather check how ORIGINAL_AS_FOR_TARGET is decided Oct 20 12:12:16 jepp Oct 20 12:14:40 hmm first I have seen it 20 days ago.. Oct 20 12:14:42 ./openmoko-cdevel/openmoko-cdevel-2010-10-01:13:13 < JaMa> /home/shr/shr-unstable/tmp/work/armv4t-oe-linux-gnueabi/gcc-cross-initial-4.5-r10.0+svnr164562/gcc-4.5/build.x86_64-linux.arm-oe-linux-gnueabi/./gcc/as: line 83: exec: : not found Oct 20 12:17:16 will try again at home ... Oct 20 12:17:23 cya Oct 20 12:29:03 Good evening everyone. Oct 20 12:29:53 My bitbake require X problem is solved. Oct 20 12:30:58 In bitbake.conf I set TERMCMD ?= "${SCREEN_TERMCMD}" Oct 20 12:31:09 TERMCMDRUN ?= "${SCREEN_TERMCMDRUN}" Oct 20 12:31:44 Thanks ynezz and others gave hints. Oct 20 12:41:24 mfhk: it would be nice to have that info in the wiki, could you write a short section on this? Oct 20 12:42:45 Sure, place it in the faq or what? Oct 20 12:59:45 mfhk: guess making a short page on it and put a link in the faq would be the best, there is not really a page on kernel building, you might want to start one; we could add some text on where to store the config file etc on that page too Oct 20 13:02:26 ok, I'll see what I can do. I have not worked with mediawiki before, so it may take some times. Oct 20 13:13:30 thanks alot! Oct 20 13:16:56 you're welcome, I am glad to contribe back to the community! Oct 20 13:47:59 03Koen Kooi  07org.openembedded.dev * re66dd27544 10openembedded.git/contrib/angstrom/sort.sh: angstrom feed sorter: optimize duplicate finder and add toggle for updater Oct 20 13:52:04 effem, page added at http://oe-wiki.leggewie.org/index.php/How_to_use_bitbake_without_X_Windows Oct 20 13:53:04 But, it seems I have no rights to edit the faq properly. Oct 20 13:54:03 mfhk: thanks, (acutally I always use this url; http://wiki.openembedded.net/index.php/Main_Page but it is the same place) Oct 20 13:54:42 I just copied the link from browser. Oct 20 13:54:51 mfhk: no need to edit the faq, add Oct 20 13:54:52 [[Category:FAQ]] Oct 20 13:54:56 to your page Oct 20 13:55:05 see eg http://wiki.openembedded.net/index.php?title=Interactive_debugging&action=edit Oct 20 13:55:12 tasslehoff: as i told you yesterday, preference variables will do absolutely nothing in a recipe. bitbake doesn't look there for them, at all, period. Oct 20 13:55:42 kergoth_: ok. missed/forgot that :) Oct 20 13:56:03 and thanks for the info (actually could have used that a week ago, when I had to ssh to a customer system :-) ) Oct 20 13:56:40 Very glad you find it useful :) Oct 20 14:17:40 JaMa|W, eFfeM_work: If I revert http://git.openembedded.net/cgit.cgi/openembedded/commit/?id=565b3f4965277f3a893b6417940ac59d78a97fb0 and http://git.openembedded.net/cgit.cgi/openembedded/commit/?id=f1148d26740e57cba4fe29fd78c234359868a50e Oct 20 14:17:51 I can compile gcc-cross-initial without problems Oct 20 14:21:24 best talk to khem about that, i would think Oct 20 14:25:51 morphis: please recheck if those .inc are included for gcc-cross-initial or binutils-cross (my quick check shown they isn't) Oct 20 14:26:47 ah sure the later is.. through ./binutils-cross_2.20.1.bb:require binutils_${PV}.bb Oct 20 14:27:05 ah ok Oct 20 14:27:09 then just binutils Oct 20 14:27:15 and as I did a clean rebuild Oct 20 14:27:20 binuitils now builds correctly Oct 20 14:27:40 than the revert of the first patch is not needed Oct 20 14:30:44 effem, kernel building page started here : http://oe-wiki.leggewie.org/index.php/Kernel_Building Oct 20 14:53:48 khem: ping Oct 20 14:54:26 * kergoth_ sighs at autotools Oct 20 15:07:11 hmmmmm - http://softlayer.dl.sourceforge.net/project/checkbaskisms/2.0.0.1/checkbashisms Oct 20 15:17:08 morphis: hi, I think the binutils patch is causing this. Oct 20 15:22:02 khem, your patch last night fixed the last problem Oct 20 15:22:26 Crofton|work: yeah but it caused binutils-cross to fail Oct 20 15:22:35 I am working on a fix Oct 20 15:22:39 k Oct 20 15:24:03 ah ha! I think http://www.seanius.net/blog/2009/03/saving-and-restoring-positional-params-redux/ may have the solution to my argument handling dilemma for my wrapper scripts Oct 20 15:27:23 03Khem Raj  07master * r096ab4990c 10openembedded.git/recipes/binutils/binutils.inc: Oct 20 15:27:23 binutils.inc: Dont filter out all --with-sysroot strings Oct 20 15:27:23 * Binutils.inc is included into cross recipes because the Oct 20 15:27:23 cross recipes include target recipe. We only need to remove Oct 20 15:27:23 --with-sysroot full word only and not --with-systoot=... Oct 20 15:27:23 Signed-off-by: Khem Raj Oct 20 15:27:27 morphis: try to update and see if it fixes your issue now Oct 20 15:47:24 * kergoth_ tests the above parameter saving/restoring for his wrappers Oct 20 15:51:52 gm Oct 20 15:52:30 hey eFfeM Oct 20 16:00:27 i think i might be almost done with my new autotools class cleanup branch, finally.. now to test the shit out of it Oct 20 16:06:22 khem: http://git.openembedded.org/cgit.cgi/openembedded/log/?h=kergoth/autotools - dropping acpaths and not using --force and oefatal on non-generated aclocal.m4 are all just experiments, may well drop those, but should give the idea of what i was trying to do Oct 20 16:06:43 http://git.openembedded.org/cgit.cgi/openembedded/tree/classes/autotools_base.bbclass?h=kergoth/autotools Oct 20 16:06:43 http://git.openembedded.org/cgit.cgi/openembedded/tree/classes/autotools.bbclass?h=kergoth/autotools Oct 20 16:07:31 planning on cleaning up the deps handling still Oct 20 16:07:43 wanted to sanity check this first Oct 20 16:13:44 hmmmm Oct 20 16:15:31 * kergoth_ has an idea -- the shell parsing code inside his signature code should be able to zip through, parsing the shell tasks in our recipes, and I could use its parse tree to find executions of cp or sed, and report back on problematic usages Oct 20 16:15:37 * kergoth_ ponders Oct 20 16:31:56 khem, should I test again? Oct 20 16:34:20 Crofton|work: yes Oct 20 16:35:54 ok Oct 20 16:36:08 03Martin Jansa  07master * r2de5bea0ef 10openembedded.git/ (2 files in 2 dirs): Oct 20 16:36:08 libice: add 1.0.7 Oct 20 16:36:08 Signed-off-by: Martin Jansa Oct 20 16:36:11 apparently you also know about my other problem Oct 20 16:36:20 03Martin Jansa  07master * rbd79cb8ef7 10openembedded.git/recipes/xorg-xserver/ (4 files in 2 dirs): Oct 20 16:36:20 xserver-xorg_git: bump SRCREV Oct 20 16:36:20 Signed-off-by: Martin Jansa Oct 20 16:36:28 ifndef __ARCH_FORCE_SHMLBA Oct 20 16:36:30 ... Oct 20 16:40:05 khem, did you push the -I fix also? Oct 20 16:40:15 also, I will need to clean binutilas? Oct 20 17:04:29 * kergoth_ kicks ncurses_5.7.bb Oct 20 17:06:01 03Chris Larson  07org.openembedded.dev * rac0b7d1cf7 10openembedded.git/recipes/coreutils/ (3 files in 2 dirs): Oct 20 17:06:01 coreutils-6.0: apply patch for USE_STATVFS Oct 20 17:06:01 Fixes compile error on platforms where statvfs does not have an f_type member Oct 20 17:06:01 (e.g. OSX). See Oct 20 17:06:01 http://www.mail-archive.com/bug-coreutils@gnu.org/msg06552.html for details. Oct 20 17:06:01 Signed-off-by: Chris Larson Oct 20 17:06:02 03Chris Larson  07org.openembedded.dev * r1c24a488b8 10openembedded.git/recipes/quilt/quilt-native.inc: Oct 20 17:06:02 quilt-native: remove '?' url in _append_build-darwin.. Oct 20 17:06:03 Signed-off-by: Chris Larson Oct 20 17:09:28 how do you disable PREMIRRORS for a recipe? Oct 20 17:09:47 PREMIRRORS_pn- = "" Oct 20 17:11:52 beautiful, thanks Oct 20 17:14:17 so much nicer to get eglibc direct from SVN instead of waiting 3-4 hours for it to download from angstrom unstable... Oct 20 17:15:11 hehe Oct 20 17:18:47 re Oct 20 17:26:05 uh oh, bitbake world -c clean is failing... Oct 20 17:26:40 that is how to "reset" your OE dir, right? Oct 20 17:27:16 Crofton|work: all fixed are in yes you need to clean binutils yes Oct 20 17:27:30 ok Oct 20 17:31:21 kergoth_: I am looking at the autotools branch Oct 20 17:31:54 kergoth_: pushd/popd is bashism may be you can get rid of it Oct 20 17:35:17 is there any way to force OE to clean everything? it's failing on "world -c clean" because some recipe is unbuildable Oct 20 17:36:23 vorporeal: clean everything could be done with removing the tmpdir Oct 20 17:36:25 define "everything" Oct 20 17:37:25 khem: i guess that works Oct 20 17:37:56 Crofton|work: i'm trying to rebuild this toolchain from scratch Oct 20 17:38:07 rm -rf tmp Oct 20 17:38:13 the oe tmp dir Oct 20 17:38:17 yep Oct 20 17:38:21 not /tmp :) Oct 20 17:38:23 vorporeal: clean the tmpdir thats ideally best Oct 20 17:39:27 thanks for the help, guys (i really appreciate it) Oct 20 17:58:41 khem: not easily. if you set e.g. olddir=$PWD and then cd $olddir, your'e hosed, because unles you use 'local' in the function, another bashism afaik, that olddir var is a global -- if the function is recursive, as that one is, to go into subdirs, it'll overwrite its own backup directory and never come back out Oct 20 17:58:49 khem: really requires a stack Oct 20 17:59:04 khem: we can always add a pure posix shell implementation of pushd/popd/dirs, if we can find one or write one Oct 20 17:59:14 i'm gonna double check susv3 about local variables though Oct 20 18:00:02 yeah, not portable, i figured as much Oct 20 18:00:19 (http://www.unix.org/single_unix_specification_v3/ - for reference) Oct 20 18:01:06 jo Oct 20 18:01:40 khem: http://docstore.mik.ua/orelly/unix3/korn/ch04_07.htm has a shell implementation of dirs/pushd/popd, but unfortunately they rely on advanced expansions like %% and # Oct 20 18:03:14 damn Oct 20 18:03:21 readline-native doesn't install in osx Oct 20 18:03:29 this is the farthest I've gotten with an osx build :) Oct 20 18:05:06 03Khem Raj  07master * r7cb89bb309 10openembedded.git/recipes/gcc/ (gcc-4.5.inc gcc-4.5/gcc-linaro-fix-lp-653316.patch): Oct 20 18:05:06 gcc-4.5.inc: Backport gcc trunk svn revision 165495 to fix ICE on udev Oct 20 18:05:06 Signed-off-by: Khem Raj Oct 20 18:05:21 uh Oct 20 18:05:23 udev fixed Oct 20 18:07:07 kergoth_: hmm I am ok if we demand bash too but if we could avoid that would be ideal Oct 20 18:07:42 we already do :) Oct 20 18:07:52 we have bigger problems to fix it than this, i think Oct 20 18:08:13 khem I will try now to update uclibc Oct 20 18:08:16 and put fts.h in Oct 20 18:08:21 khem: how do the changes look otherwise? seem sane? Oct 20 18:08:34 woglinde: yes thats good we should harden that patch in OE before pushing to uclibc Oct 20 18:08:45 kergoth_: yes it seems good otherwise Oct 20 18:09:20 hm so let see whats the latest revisiob Oct 20 18:09:21 is bash avaialable on all kind of hosts we intend to run OE builds on ? Oct 20 18:10:42 kergoth_: %% and # is not posix ? Oct 20 18:10:55 didn't think so, but if so that would be lovely Oct 20 18:11:06 khem: oe won't even run if /bin/sh is dash Oct 20 18:11:13 so i certainly hope so Oct 20 18:11:32 freebsd have bourne shell as /bin/sh Oct 20 18:11:33 not bash Oct 20 18:11:42 and not sure about OS X Oct 20 18:12:22 khem: i stand corrected Oct 20 18:12:53 :-, -, :=, =, :?, ?, :+, +, %, %%, #, ## are all in SuSv3 at least Oct 20 18:13:00 http://www.opengroup.org/onlinepubs/000095399/utilities/xcu_chap02.html#tag_02_06_02 Oct 20 18:13:11 yes I am seeing it Oct 20 18:13:26 so those implementations of the functions should work fine :) Oct 20 18:13:41 right. so you can add them Oct 20 18:13:42 http://www.freebsd.org/cgi/man.cgi?query=sh&manpath=FreeBSD+8.1-RELEASE Oct 20 18:13:49 may be they should be added to a general place Oct 20 18:13:50 man sh from freebsd Oct 20 18:13:51 i should add a class for these sorts of portability functions / scripts Oct 20 18:13:53 khem: :) Oct 20 18:13:53 JFYI Oct 20 18:13:56 where all shell can inherit Oct 20 18:14:28 Jay7: freebsd still uses ash Oct 20 18:15:13 I think its a good thing to be portable Oct 20 18:15:35 kergoth_: yeah posix-compat.bbclass Oct 20 18:15:54 or may be portable.bbclass Oct 20 18:15:56 and ugly-workarounds.bbclass :) Oct 20 18:15:57 khem: http://softlayer.dl.sourceforge.net/project/checkbaskisms/2.0.0.1/checkbashisms seems interesting Oct 20 18:16:27 yeah for sanity.bbclass Oct 20 18:16:33 we should run that script Oct 20 18:16:33 yeah thats what i was thinking Oct 20 18:16:38 scan the ${S} for #!/bin/sh Oct 20 18:16:41 cool Oct 20 18:16:50 might be better as an explicit operation like recipe_sanity, rather than automatic Oct 20 18:16:59 but either way would be good i think, to start moving in the right direction Oct 20 18:17:07 yeah recipe_sanity is more appropriate Oct 20 18:17:25 eventually such things should be mandatory Oct 20 18:17:54 we can enable this for all recipes but as warning, not error Oct 20 18:17:57 the only reason i did it the way it is is that some of those things require user thought - its not a hard guide, just recommendations Oct 20 18:17:58 for first time at least Oct 20 18:17:59 * kergoth_ nods Oct 20 18:18:28 kergoth_: if its not error devs dont care so much Oct 20 18:18:36 woglinde woglinde_ openjdk for bug13 works after your suggestions. thanks again! Oct 20 18:18:38 khem: we should maybe split it up Oct 20 18:18:41 gcc 4.3.3 Oct 20 18:18:43 somet hings optional, some not Oct 20 18:18:46 recipe_sanity i mean Oct 20 18:18:51 wtf Oct 20 18:18:57 opkg-native defines AC_CONFIG_MACRO_DIR twice Oct 20 18:19:02 one right after another Oct 20 18:19:06 what exactly is the point of that Oct 20 18:19:17 heh doube yanking Oct 20 18:19:19 double Oct 20 18:19:23 (broke my autoconf -t bits for gnu-config) Oct 20 18:19:25 heheh Oct 20 18:20:07 oh, hey i discovered an automake option i didn't know about Oct 20 18:20:09 not sure if others do Oct 20 18:20:13 but it has an option like -j for make Oct 20 18:20:17 thread the generation Oct 20 18:20:22 its just an env var Oct 20 18:20:31 kergoth_: yes I know that Oct 20 18:20:37 we should play with that Oct 20 18:20:55 although there were issues when I played with it Oct 20 18:21:00 ah Oct 20 18:21:02 not many people know Oct 20 18:21:05 and code according Oct 20 18:21:07 to it Oct 20 18:21:08 hehe Oct 20 18:21:12 insufficient testers Oct 20 18:21:16 coffee time Oct 20 18:21:19 would be a good reason for us to play with it though, report upstream Oct 20 18:21:42 fuck, my sed wrapper still isn't full proof Oct 20 18:21:43 argh Oct 20 18:21:47 khem you posted a couple of other patches on the ml, did they get commited? Oct 20 18:24:02 wtf Oct 20 18:24:12 "NOTE: package diffstat-1.47-r0: task None: Failed" Oct 20 18:24:15 i like that task. Oct 20 18:24:31 * kergoth_ grumbles Oct 20 18:26:53 kergoth lol Oct 20 18:27:00 about opkg-native Oct 20 18:28:25 i'm having issues with the stage-manager script in my newly generated sysroot Oct 20 18:29:05 it failed on "except OSError, e:", which I changed to "except OSError as e:" and now it's complaining about inconsistent tabbing/spacing Oct 20 18:29:14 khem I will take the poor man choice for fts.h for now and copy it over at do_install Oct 20 18:29:18 i'm assuming someone changed something and screwed it up? Oct 20 18:30:31 vorporeal: except OSError, e is just fine in any python version that isn't 3.x Oct 20 18:30:39 the 'as' version only works in 2.6+ Oct 20 18:30:41 oe doesn't require 2.6 Oct 20 18:31:14 kergoth_: ah, i guess this is a byproduct of arch having switched default to python3 Oct 20 18:31:19 yes Oct 20 18:31:19 *sigh* Oct 20 18:31:22 bitbake isn't 3.0 Oct 20 18:31:25 so its definitely on your end Oct 20 18:31:33 not everything is 3.x compatible yet Oct 20 18:31:39 i've been running it explicitly with 2.7, which has worked so far Oct 20 18:31:49 that should still work fine, as far as i'm aware Oct 20 18:31:54 i started doing the 2to3 conversions on master to prep for 3.x Oct 20 18:31:54 and i assumed it wouldn't run a separate instance of python for other scripts Oct 20 18:32:00 but we aren't breaking compatibility yet Oct 20 18:32:01 which i'm assuming it just did Oct 20 18:32:09 its not magic. Oct 20 18:32:17 i'll try changing the /usr/bin/python symlink and see if that fixes things Oct 20 18:32:18 if bitbake runs something, and that something has a #! line, its going to use that Oct 20 18:32:30 i'm not that linux illiterate :) Oct 20 18:32:34 its just a fork+exec, not in bitbake's control Oct 20 18:32:48 hmm, i suppose Oct 20 18:34:19 now, what you could do, if you really want to keep 3.x as default on your machine, is construct a chroot to use for bitbake executions, then its a stable environment Oct 20 18:34:24 thats what i've been doing recently Oct 20 18:34:48 i'm using bitbake to build this toolchain so that i didn't have to make a chroot to use a prepackaged one... Oct 20 18:34:58 you can only be so lazy Oct 20 18:35:38 well, switching it back to python2 fixed it Oct 20 18:35:39 thanks Oct 20 18:35:45 hopefully that won't break my arch install Oct 20 18:35:52 * kergoth_ thinks we should make available pre-constructed chroots to use for bitbake/oe builds, which have everything required to run it, to improve build reproducibility Oct 20 18:36:11 np Oct 20 18:36:20 damn, checking gcc out of source control takes a long ass time Oct 20 18:36:27 * kergoth_ hasn't done one from scratch in ages Oct 20 18:36:29 03Koen Kooi  07org.openembedded.dev * rc8b2396bc7 10openembedded.git/recipes/binutils/binutils.inc: binutils: bump INC_PR to force rebuild Oct 20 18:36:36 when I build an image should the postinst of a package be run or does this only happen on first boot ? Oct 20 18:36:47 eFfeM: all postinsts are run at image creation time Oct 20 18:36:50 if they return failure Oct 20 18:36:53 they get postponed to first boot Oct 20 18:36:56 kergoth_, on the Amazon cloud thing Oct 20 18:37:01 this is why many check against ${D} Oct 20 18:37:07 or rather, $D Oct 20 18:37:14 if D is set, its image creation time, if not, its first boot time Oct 20 18:37:32 so scripts taht require they run on the device, e.g. to run a target binary, will exit non-zero if D is set Oct 20 18:37:55 * kergoth_ hasn't messed with cloud stuff much, should do so Oct 20 18:38:21 * kergoth_ ponders Oct 20 18:38:23 kergoth_: that is what I expected will check what is wrong Oct 20 18:38:38 speaking of which, we should stop relying on opkg to do it Oct 20 18:38:43 and just have a general mechanism Oct 20 18:38:54 which gets installed in the rootfs regardless of ONLINE_PACKAGE_MANAGEMENT Oct 20 18:40:25 kergoth_: I'll create lxc container for OE soon Oct 20 18:40:35 I can package it Oct 20 18:40:53 it will be based on Debian 5 Oct 20 18:41:49 nice Oct 20 18:42:34 * kergoth_ 's been trying to build a wide set of chroots for build testing -- the centos and fedora ones are proving most annoying -- have fedora 11, 12, 13, 14 happy, but not 10 Oct 20 18:44:18 kergoth_: reason i was wondering about the postinst is because on first build i see it generate the dropbear key Oct 20 18:45:56 yes Oct 20 18:46:10 how would it do that when making the image, if the binaries are arm or something? :P Oct 20 18:48:44 so at least some of the stuff is done after image creation, haven't really figured out how Oct 20 18:52:39 i just explained it Oct 20 18:52:49 i don't know how to explain it any clearer than that Oct 20 18:53:01 many postinsts require that they be run on first boot, so they exit non-zero when D is set Oct 20 18:53:06 dropbear is likely one of them Oct 20 18:58:44 ah ok, now I understand Oct 20 18:59:26 still seems the postint I added to mythplugins does not work for me, strange Oct 20 19:01:00 D is the path to the rootfs, if you're doing postinst work at image creation time, remember to prefix your paths with it Oct 20 19:01:50 ah, ok checking Oct 20 19:02:51 03Martin Jansa  07master * rae3bc32a49 10openembedded.git/recipes/linux/ (2 files in 2 dirs): Oct 20 19:02:51 linux-openmoko-2.6.34: add patch for jitter less ts Oct 20 19:02:51 Signed-off-by: Martin Jansa Oct 20 19:04:40 kergoth_: ${D} was indeed missing Oct 20 19:04:58 thanks, fixed it, testing it now Oct 20 19:05:31 jo gnutoo Oct 20 19:08:10 eFfeM: no, not ${D}, $D Oct 20 19:08:22 iirc, anyway Oct 20 19:09:17 oh, ok, Oct 20 19:09:19 oops Oct 20 19:09:43 verify that though :) Oct 20 19:09:55 will do Oct 20 19:12:15 it is indeed $D Oct 20 19:16:54 * kergoth_ kicks readline Oct 20 19:18:03 Jay7: do u build OE on fbsd ? Oct 20 19:18:14 khem: I'll do Oct 20 19:18:27 just for testing it :) Oct 20 19:18:54 Crofton|work: all patches are in now Oct 20 19:18:58 I have HW for this already Oct 20 19:19:26 just need some time to build lxc containers and kvm virtual machine with freebsd Oct 20 19:19:28 Jay7: that will be cool Oct 20 19:19:52 khem: I hope this will be ready in 2-3 days Oct 20 19:19:58 my son likes the imp on freebsd box Oct 20 19:20:09 so I might install it on my old laptop Oct 20 19:20:12 and give it to him Oct 20 19:20:21 Jay7: cool Oct 20 19:20:50 woglinde: create a proper patch Oct 20 19:20:58 no fiddling with in do_install Oct 20 19:21:01 khem ts Oct 20 19:21:03 khem, cleaning out old stuff and getting into test mode Oct 20 19:21:04 no time Oct 20 19:21:48 woglinde: err then just commit the upgrade but not fts.h thing and send it to ml Oct 20 19:21:54 I can fix it and commit it Oct 20 19:25:35 hi stefan Oct 20 19:25:44 hi woglinde Oct 20 19:25:57 woglinde: Trying to backport your icedtea and openjdk patches to stable Oct 20 19:26:09 woglinde: failing so far. classpath build breaks Oct 20 19:26:41 Crofton|work: cool. I think it should all be settled in now Oct 20 19:26:43 hm? Oct 20 19:26:51 I build it last weak Oct 20 19:26:55 there it worked Oct 20 19:27:01 woglinde: on stable? Oct 20 19:27:05 woglinde: dev is fine Oct 20 19:27:07 bitbake world is much better now Oct 20 19:27:13 didnt test stable Oct 20 19:27:15 I have 50k tasks completed Oct 20 19:27:16 for long time Oct 20 19:27:19 woglinde: I was going to bring them into stable Oct 20 19:27:21 30k more to do Oct 20 19:27:36 khem hihi Oct 20 19:27:38 but most of them will be rm_work and qa Oct 20 19:27:40 woglinde: yeah, as I thought. was going to do this. Oct 20 19:28:00 woglinde: its a good excercise as we find so much cruft that needs fixing Oct 20 19:28:08 khem sure Oct 20 19:28:12 woglinde: copying 17.3 over to 1.7.5 and adjusting openjdk should be all I need? Oct 20 19:28:17 I think metadata should have all things lively and kicking Oct 20 19:28:23 no place for cruft and staleness Oct 20 19:28:36 java is one PITA Oct 20 19:28:42 one mans cruft is another mans release .... Oct 20 19:28:45 * stefan_schmidt gets remind to start the world build again as well Oct 20 19:28:55 to much trouble with my exams right now... Oct 20 19:29:13 Crofton|work: no no, I dont mean kill of stuff. I mean to make it build Oct 20 19:29:30 khem: yeah, even more if your customers relies on it working good in OE ;) Oct 20 19:29:44 But thatnks to tarent I did not have that much trouble yet :D Oct 20 19:30:02 hm Oct 20 19:30:03 * denix appends to the quote collection: one mans cruft is another mans release .... Oct 20 19:30:11 isnan patch dondt apllies Oct 20 19:30:28 right now I think we only build 30-40% recipes regularily Oct 20 19:30:34 woglinde: I did the copy in stable directly Oct 20 19:30:43 woglinde: so patch=1 stays there Oct 20 19:31:48 denix: heh, I wonder how can be a non-compiling recipe be anothers release Oct 20 19:32:09 denix: we have to contain things within out env Oct 20 19:32:27 if one pushed a recipe because it built with his external toolchain I think thats wrong Oct 20 19:32:34 it should as well build with internal toolchain Oct 20 19:33:32 lots of different OE use cases though Oct 20 19:34:01 I need to fix git-send-email Oct 20 19:34:04 Crofton|work: I think we have recipes for things from 2006 which build with gcc 3.3 Oct 20 19:34:19 I wonder if we can scope to support toolchains that wide range Oct 20 19:34:27 there are other incompatibilities Oct 20 19:34:33 like autotools Oct 20 19:34:36 damn, autotools branch breaks freetype-native Oct 20 19:34:50 kergoth_: in what way Oct 20 19:35:04 odd compile errors, checking configure log now Oct 20 19:36:04 my idea is that if a recipe is pushed upstream it needs to be maintained with global changes we do and evolve all along Oct 20 19:36:11 and bitbake world will help Oct 20 19:36:32 right now it takes a week to run bitbake world on my laptop :) Oct 20 19:36:59 woglinde: could you have a look at my patches if you see something obvious wrong in there? Oct 20 19:37:00 we must be a pack rats Oct 20 19:37:02 woglinde: http://www.datenfreihafen.org/~stefan/0002-openjdk-Backport-security-update-to-1.8.2.patch.txt Oct 20 19:37:09 woglinde: http://www.datenfreihafen.org/~stefan/0001-icedtea6-native-Backport-security-update-to-1.7.5.patch.txt Oct 20 19:38:43 hm seems okay Oct 20 19:39:35 maybe it is something else that was done in dev but missing in stable Oct 20 19:40:44 *sigh* seems I have to do more poking at it Oct 20 19:41:47 woo, task 128 of 480 in osx so far Oct 20 19:41:55 slowly getting there Oct 20 19:43:44 hmm gcc-cross-initial 4.3.3 is failing in do_compile Oct 20 19:43:48 configure-target-libgcc is failing Oct 20 19:44:29 configure:2605: error: cannot compute suffix of object files: cannot compile Oct 20 19:44:38 tmp/work/i486-oe-linux/gcc-cross-initial-4.3.3-r18.1/gcc-4.3.3/build.x86_64-linux.i486-oe-linux/./gcc/as: line 77: exec: : not found Oct 20 19:44:48 mzs_: do u pull Oct 20 19:44:52 some variable is not getting substituted i guess Oct 20 19:44:54 Yeah Oct 20 19:45:00 just pulled Oct 20 19:45:09 mzs_: do u have latest binutils commit from me Oct 20 19:45:40 096ab4990ce8aaa12196ee10d8ad23a86063a289 Oct 20 19:45:40 this one? binutils.inc: Dont filter out all --with-sysroot strings Oct 20 19:45:56 mzs_: yes Oct 20 19:45:57 Yeah got it Oct 20 19:46:47 damn, dies in module-init-tools-native now.. hmm Oct 20 19:46:58 mzs_: bitbake -c clean binutils-cross;bitbake binutils-cross Oct 20 19:47:14 Yeah I hit it when starting from a fresh build. Oct 20 19:47:40 my tmp/work/i486-oe-linux/gcc-cross-initial-4.3.3-r18.1/gcc-4.3.3/build.x86_64-linux.i486-oe-linux/gcc/as has: Oct 20 19:47:46 ORIGINAL_AS_FOR_TARGET="" Oct 20 19:47:56 seems wrong.. Oct 20 19:48:23 mzs_: ok paste run.do_configure of gcc-cross-initial and binutils-cross for me Oct 20 19:51:44 03Ben Gardiner  07org.openembedded.dev * rf3b29a0371 10openembedded.git/conf/distro/ (4 files in 2 dirs): (log message trimmed) Oct 20 19:51:44 binutils-cross: http://pastebin.ca/1968357 Oct 20 19:51:44 distro, toolchain-external: override TARGET_VENDOR with TOOLCHAIN_VENDOR Oct 20 19:51:44 Trying to use an external toolchain like the 2009q1 toolchain from codesourcery Oct 20 19:51:44 has some strange results since the toolchain has vendor 'none' but Oct 20 19:51:44 distributions like angstrom set TARGET_VENDOR unconditionally. Oct 20 19:51:45 This patch overrides the TARGET_VENDOR with the value of TOOLCHAIN_VENDOR, Oct 20 19:51:46 if set. TOOLCHAIN_VENDOR can be set in the local configuration alongside the Oct 20 19:53:05 mzs_: I need full long plz Oct 20 19:53:17 I mean full file run.do_configure.xxxx Oct 20 19:55:50 khem: binutils-cross http://pastebin.ca/1968360 ; gcc-cross-initial http://pastebin.ca/1968362 Oct 20 20:00:14 mzs: I dont see any problems Oct 20 20:00:28 mzs: whats your distro Oct 20 20:00:52 mzs: which libtool are you using btw. Oct 20 20:00:53 khem hm this is fixed? Oct 20 20:00:54 | checking for suffix of object files... configure: error: in `/devel/arm/oetmp-ang/work/armv7a-angstrom-linux-uclibceabi/gcc-cross-initial-4.5-r13.0+svnr164562/gcc-4.5/build.i686-linux.arm-angstrom-linux-uclibceabi/arm-angstrom-linux-uclibceabi/libgcc': Oct 20 20:00:58 | configure: error: cannot compute suffix of object files: cannot compil Oct 20 20:01:08 it's custom... and i'm still on libtool 2.2.6b iirc Oct 20 20:01:10 denix, ping Oct 20 20:01:12 woglinde: that's what i'm getting. Oct 20 20:01:18 woglinde: it should be Oct 20 20:01:24 * woglinde is on old libtool too Oct 20 20:01:31 woglinde: are you on latest git master Oct 20 20:01:44 checked out this evening Oct 20 20:01:59 woglinde: whats you top rev Oct 20 20:02:03 gcc-cross-initial-4.5-r13.0+svnr164562 Oct 20 20:02:11 oh thats new Oct 20 20:02:26 khem: does anything need to be added to distro config to make it keep working with old libtool? Oct 20 20:02:35 some variable with a default value that i missed? Oct 20 20:02:36 mzs: no nothing Oct 20 20:02:42 it should *just* wor Oct 20 20:02:43 k Oct 20 20:03:19 woglinde: post your run.do_configure of gcc-cross-initial Oct 20 20:04:38 Maybe I'll try switching to 2.4 Oct 20 20:04:47 just blew away my tmp so i might as well anyway Oct 20 20:05:26 hm why the hell its not finding exec in this context Oct 20 20:05:39 mzs: angstrom 2010 uses 2.4 Oct 20 20:05:51 woglinde: I dont understand either Oct 20 20:05:55 woglinde: configure is calling a wrapper script around "as". look at the top of the wrapper and i bet you'll see the variables are empty Oct 20 20:06:21 most probably binutils-cross problem Oct 20 20:06:28 I am not sure Oct 20 20:06:38 mzs? Oct 20 20:06:41 Crofton|work: pong Oct 20 20:06:51 woglinde: if you're seeing the same issue i am Oct 20 20:06:52 exec is shell builtin Oct 20 20:07:02 khem: morphis reported build ok after reverting last 2 patches (gcc-target and binutils) Oct 20 20:07:24 khem: so maybe $ in the end for that replace is not enough Oct 20 20:08:06 woglinde: it calls "exec ${empty_var}" that's what it says Oct 20 20:08:20 JaMa: could be Oct 20 20:08:30 thats why I am asking for run.do_configure Oct 20 20:08:38 and they contain the proper sysroot Oct 20 20:08:50 so it baffles me Oct 20 20:08:55 hm okay Oct 20 20:08:58 woglinde: post yours Oct 20 20:09:06 khem its same Oct 20 20:09:12 no need to post Oct 20 20:09:12 what the hell Oct 20 20:09:13 hrm Oct 20 20:09:17 run.do_configure for binutils-cross and gcc-cross-initial Oct 20 20:09:19 ORIGINAL_AS_FOR_TARGET="" Oct 20 20:09:30 original=$ORIGINAL_AS_FOR_TARGET Oct 20 20:09:37 exec "$original" ${1+"$@"} Oct 20 20:09:42 khem: just noticed my do_configure for gcc-cross-initial had a --with-sysroot, does that make sense with libtool 2.2.6b? Oct 20 20:10:25 mzs: yes with libtool 2.2.6 it should not be there without a '=' Oct 20 20:10:54 woglinde: those problem can happen if it does not find binutils placed right Oct 20 20:11:05 woglinde: post the files I am asking for Oct 20 20:11:07 hm, how can i see that my postinst has run when creating an image; apparently it isn't but no idea where to peek or when/how this is done; the postinst file is in usr/lib/opkg/info Oct 20 20:12:29 eFfeM: if it's like dpkg, i think the "status" file will show it as unpacked but not configured Oct 20 20:12:32 or try opkg -l Oct 20 20:13:26 http://pastebin.ca/1968375 Oct 20 20:13:30 binutils Oct 20 20:14:09 http://pastebin.ca/1968376 gcc Oct 20 20:18:50 khem ping Oct 20 20:20:31 woglinde: yes Oct 20 20:20:46 hmm Oct 20 20:20:48 I fail to see the problem :( Oct 20 20:20:57 khem, gcc-cross-initial is not building Oct 20 20:21:09 Crofton|work: ok Oct 20 20:21:15 same issue as woglinde Oct 20 20:21:19 ? Oct 20 20:21:19 woglinde: i just had the "suffix of object files" issue Oct 20 20:21:34 i think i fixed it with a rm -rf tmp and a rebuild Oct 20 20:21:36 ok guys let me see whats wrong Oct 20 20:21:41 thansk Oct 20 20:21:47 yep me too :) Oct 20 20:22:47 hehe Oct 20 20:22:58 is it working with libtool2.4? Oct 20 20:23:12 woglinde: i guess so Oct 20 20:23:17 woglinde: i'm trying that now... Oct 20 20:25:16 http://www.devere.co.uk/our-locations/university-arms Oct 20 20:26:50 no its not working with 2.4 either Oct 20 20:27:20 hehe Oct 20 20:27:24 ~blame khem Oct 20 20:27:24 * ibot blames khem (and Canada) for all the evil in the world Oct 20 20:29:37 woglinde: I am able to reproduce it on a clean build Oct 20 20:29:47 ok Oct 20 20:29:47 so should be fixable now for me Oct 20 20:29:54 quick update of the OEDEM page Oct 20 20:32:29 khem: i'm currently compiling gcc-cross-initial on an up-to-date (and previously clean) checkout Oct 20 20:32:38 something got it past that configure error Oct 20 20:35:22 vorporeal: what ? Oct 20 20:35:35 it works Oct 20 20:35:39 03Chris Larson  07master * r0c9b1efb3d 10openembedded.git/recipes/diffstat/diffstat_1.47.bb: Oct 20 20:35:39 diffstat: fetch from upstream Oct 20 20:35:39 Upstream has archived version tarballs, so it seems, so stop relying on debian Oct 20 20:35:39 for the fetching. Oct 20 20:35:39 Signed-off-by: Chris Larson Oct 20 20:35:47 i guess i'm confused? Oct 20 20:35:47 * khem is confused even more Oct 20 20:36:01 03Chris Larson  07master * r8708532de0 10openembedded.git/recipes/expat/ (expat.inc expat_1.95.6.bb expat_1.95.7.bb expat_2.0.1.bb): Oct 20 20:36:01 expat: use INC_PR, set acpaths Oct 20 20:36:01 Signed-off-by: Chris Larson Oct 20 20:36:27 wrt my postinst problems, found what was wrong, there was a fault in the postinst script but that does not trigger an error Oct 20 20:38:43 no idea how we could trap that reliably. Is there a way to tell bash to exit on failure ? Oct 20 20:38:58 khem, I saw it in an incremental and clean build Oct 20 20:39:04 eFfeM: set -e Oct 20 20:39:35 mzs, thanks! Oct 20 20:39:42 overllooked that one Oct 20 20:40:40 kergoth_: PR = "1", really? in diffstat recipe Oct 20 20:42:12 khem: nevermind, it failed :( Oct 20 20:42:56 * khem 's confusion point lowers Oct 20 20:43:05 sorry about that Oct 20 20:43:15 Crofton: what's up? Oct 20 20:43:28 booked the Travelodge Oct 20 20:43:55 khem hehe last tow binutils and gcc-commits Oct 20 20:44:16 hmm, you posted link to univ-arms - thought you got a room there... Oct 20 20:44:26 heh Oct 20 20:44:31 no, pastebining Oct 20 20:44:34 last time I checked 2 weeks ago they didn't have rooms Oct 20 20:44:41 updating oedem wiki Oct 20 20:44:46 ah Oct 20 20:45:36 Crofton: how long is your stay and when are you leaving? I'm leaving Sunday morning Oct 20 20:45:49 I'm leaving Saturday Oct 20 20:45:58 I want to go visit my family Sat night Oct 20 20:46:04 fly out Sunday AM Oct 20 20:46:27 what's the agenda for saturday? Oct 20 20:46:46 I mean OEDEM agenda :) Oct 20 20:47:01 make entries on the wiki Oct 20 20:47:23 the only problematic entry for the GA is the voting syste, Oct 20 20:47:40 I don't have an issue with the current system Oct 20 20:47:46 but some people do not like it Oct 20 20:48:13 I am interested in suggestions, but am afraid people will get to talking and inventing a new system Oct 20 20:48:24 that is no real improvement Oct 20 20:48:25 I just suggested to talk about the LICENSE field, if we have time... Oct 20 20:48:30 and doesn't actually work :) Oct 20 20:49:08 that would be good Oct 20 20:49:09 can we do license early so I can drop in later Oct 20 20:49:13 we need a resolution there Oct 20 20:49:18 * eFfeM finds license very boring ;-) Oct 20 20:49:34 lots of actual stuff is boring :) Oct 20 20:49:42 true Oct 20 20:49:46 :) Oct 20 20:52:34 but i hope there are some topics that are more interesting than license though :P Oct 20 20:53:06 * mwester tries to build SlugOS from scratch for the first time in two weeks, and finds that it fails miserably attempting to build the cross compiler :( Oct 20 20:53:19 :( Oct 20 20:53:20 if we attempt to sort topics by what people are interested in, madness will ensue Oct 20 20:53:22 mwester: yes I broke it Oct 20 20:53:25 Says something about being unable to figure out the suffix Oct 20 20:53:31 mwester: join #oe Oct 20 20:53:33 and it will be fixed soon Oct 20 20:53:37 oop Oct 20 20:53:37 s Oct 20 20:53:46 i was confused about which room i was looking at :) Oct 20 20:53:46 mwester, yes, we are all staring at khem Oct 20 20:53:54 O.O Oct 20 20:54:14 its not easy when you have a slow box Oct 20 20:54:22 Ah, ok. I was worried that it was something committed in the past weeks that I missed, and that it would be days of bisecting to sort out. :) A problem identified is almost fixed, so that's good. :) Oct 20 20:54:22 gcc takes forever to compile Oct 20 20:54:41 no its fairly fresh bleeding OE :) Oct 20 20:54:56 yeah, i went out to lunch to let it compile and it was still going when i got back... Oct 20 20:55:00 mwester: I'm surprised; I build slugos/slugos-image for both nslu2le and nslu2be last weekend using the testing-next branch iwthout a problem Oct 20 20:55:16 Cool. Oct 20 20:55:45 I usually try to blow away tmpdir weekly, but haven't got to it lately. Oct 20 20:55:57 So I will go get some lunch instead. Oct 20 20:56:44 * Crofton|work thinks we should collect some money to get khem a faster machine Oct 20 20:57:10 mwester, i have both on our autobuilder, try to build the testing-next every week Oct 20 20:57:17 enjoy lunch Oct 20 20:57:24 (almost bedtime here) Oct 20 20:57:29 ok I think I found the culprit Oct 20 20:57:33 now on to fix it Oct 20 20:58:11 hehe Oct 20 21:00:25 with this earch iteration is like 20mins :( Oct 20 21:07:19 hmmm := evaluate is early enough and then we inherit cross bummer Oct 20 21:07:48 this binutils needs restructuring and flattened out Oct 20 21:07:59 its like a italian spaghetti Oct 20 21:08:01 thats what i hate about .inc usage, ends up a twisty maze Oct 20 21:08:07 :\ Oct 20 21:08:36 guess its just a matter of stepping back and restructuring it, hopefully Oct 20 21:08:48 CONFIGUREOPTS = "${@oe_filter_out('--with-sysroot$', '${CONFIGUREOPTS}', d)}" Oct 20 21:08:50 guess you can always use the anonymous python snippet for now, since thats evaluated at the end Oct 20 21:08:57 kergoth_: do u think thats ok Oct 20 21:09:07 it will be evaluated at end I suppose Oct 20 21:09:07 yeah, can always revisit it later Oct 20 21:09:09 yep Oct 20 21:09:27 let me try to replace := with = Oct 20 21:09:36 doubt it'll work, recursion Oct 20 21:09:41 self reference Oct 20 21:10:28 hmm yeah Oct 20 21:10:45 let me create yes another inc file Oct 20 21:10:56 binutils-configure-target.inc Oct 20 21:11:09 that wont help either Oct 20 21:11:29 because binutils_{PV}.bb is includes in cross recipes Oct 20 21:13:51 i guess as the things stand I can not avoid anonymous python Oct 20 21:13:57 in this case Oct 20 21:14:27 so lets use that meanwhile think of restructuring binutils recipes Oct 20 21:14:32 what say Oct 20 21:14:37 yeah, good plan Oct 20 21:14:58 I agree :) Oct 20 21:16:32 khem: are you trying to override the --with-sysroot added in autotools.bbclass? Oct 20 21:16:53 it's a bit hacky but you could define something like DONT_SYSROOT = "yes" in binutils Oct 20 21:16:59 and add it to t he conditional in autotools Oct 20 21:17:15 we have more than enough variables already Oct 20 21:17:28 Yeah. Depends what you hate more, vars or anon python :) Oct 20 21:18:24 I hate both Oct 20 21:18:24 yes. Oct 20 21:18:26 :) Oct 20 21:18:53 anon python is like one night stand Oct 20 21:18:53 I hate everything and everyone Oct 20 21:18:58 so lets get away with it Oct 20 21:20:58 alright can you guys try this patch http://pastebin.com/6bbsscaU Oct 20 21:23:23 i only need to clean binutils-cross & gcc-cross-initial first right? Oct 20 21:23:52 mzs: also clean gcc-cross-initial Oct 20 21:25:07 calling it a day, cya Oct 20 21:28:43 celinux-test.conf and chinook-compat.conf and nylon.conf are maintained still ? in oe master Oct 20 21:29:15 is someone interested in getting these building with latest oe Oct 20 21:30:29 khem: so far so good, my "as" wrapper has values in the variables where before it had none Oct 20 21:30:30 khem: binutils-cross built, now onto gcc-cross-initial Oct 20 21:31:25 machine/mtx-3.conf Oct 20 21:31:29 what crap Oct 20 21:33:30 khem, I'll have to flush a distro and some related files. Straight remove or..? Oct 20 21:33:54 ant_: move them to obsolete Oct 20 21:33:54 we don't have obsolete for conf Oct 20 21:34:00 create one Oct 20 21:34:11 is above openembedded Oct 20 21:34:23 oh right Oct 20 21:34:29 well delete it Oct 20 21:34:40 was already acked on ML Oct 20 21:34:42 someone who needs it can recover it from SCM Oct 20 21:34:44 and irc Oct 20 21:34:46 go ahead Oct 20 21:34:49 ok Oct 20 21:35:22 We should call out for people using/maitaining the distro's we have Oct 20 21:35:41 and also some of questionable machines Oct 20 21:35:57 like one I saw mtx-3.conf Oct 20 21:36:27 khem: my gcc-cross-initial just got past populate_sysroot Oct 20 21:36:28 hmm Oct 20 21:36:30 with libtool 2.4 Oct 20 21:36:35 there's something off with overrides handling still Oct 20 21:36:37 hrm Oct 20 21:38:49 khem: yep, just finished for me too Oct 20 21:47:11 * denix thinks Crofton is a great source of quotes - I hate everything and everyone Oct 20 21:50:20 03Andrea Adami  07org.openembedded.dev * rbaca8ebb40 10openembedded.git/conf/distro/sharprom-compatible.conf: Oct 20 21:50:20 sharprom-compatible: remove unmaintained distro. Oct 20 21:50:20 * embedix-2.4 kernels have been moved to obsolete Oct 20 21:50:20 * external toolchain with gcc-2.95 was required Oct 20 21:50:20 * ack'ed by the former OpenZaurus maintainers Oct 20 21:50:22 03Andrea Adami  07org.openembedded.dev * r2ec4c55a4b 10openembedded.git/conf/distro/amsdelta-oe.conf: Oct 20 21:50:22 amsdelta-oe.conf: no need to specify preferred-gpe-versions-2.8 Oct 20 21:50:22 * default should work flawlessy (implicitly ack'ed on irc by GPE maintainer) Oct 20 21:50:33 03Andrea Adami  07org.openembedded.dev * r4477ea5d1a 10openembedded.git/conf/distro/include/ (preferred-gpe-versions-2.8.inc preferred-gpe-versions.inc): preferred-gpe-versions: remove now unused files. Oct 20 21:50:56 R.I.P Oct 20 21:51:54 ant_: so nothing needed gcc 2.95 now Oct 20 21:52:03 I guess it's time to sell my SL-5500 :) Oct 20 21:52:16 03Khem Raj  07master * r435d22298a 10openembedded.git/recipes/binutils/binutils.inc: (log message trimmed) Oct 20 21:52:16 binutils.inc: Use anon python to grok CONFIGUREOPTS Oct 20 21:52:16 * Immediate evaluation worked fine when the package was for target Oct 20 21:52:16 but failed for cross because cross bbclass is inherited after Oct 20 21:52:16 this evaluation happens and the configure parameters are pre Oct 20 21:52:16 evaluated which are incorrect for binutils-cross build, therefore Oct 20 21:52:17 we need to evaluate this expression at very end of parsing the recipe Oct 20 21:52:22 alright guys off my back Oct 20 21:52:34 denix, no no Oct 20 21:52:39 2.6.36 is coming Oct 20 21:52:55 khem: not that fast! your back is so comfortable... :) Oct 20 21:53:04 ant_: what's the status? Oct 20 21:53:19 denix: I've received my SL-5500 finally Oct 20 21:53:40 so.. at least kexecboot should support it :) Oct 20 21:53:58 Jay7: finally? how long did it take, a year? Oct 20 21:54:17 denix: almost :) Oct 20 21:54:49 yeah the cargo dropped it into black sea accidently and it had to be dried up before delivering to jay7 Oct 20 21:54:49 I hope lumag will receive his tosa too soon Oct 20 21:56:42 btw, jlime people are progressing Oct 20 21:57:00 I'm thinking about extending jlime and shr support to Z :) Oct 20 22:01:28 grrr Oct 20 22:02:32 keep running into the strangest errors in both the osx build and the autotools build Oct 20 22:04:30 about amsdelta distro, is it really necessary to pin qte and OPIE versions ? Oct 20 22:04:58 bluelightning, would OPIE_VERSION = "1.2.1" run to-day? Oct 20 22:05:48 ant_: it probably won't build, not only are the recipes gone but there would be a number of compile fixes missing Oct 20 22:06:06 ok, another case of distro bitrot Oct 20 22:06:32 there's really no reason not to be using 1.2.3+ Oct 20 22:07:17 kergoth_: I hope you are not doing autotools work on osx machine :) Oct 20 22:07:23 oh god no Oct 20 22:07:24 bluelightning, then we could delete the preferred-opie-versions Oct 20 22:07:27 i'm not that much of a glutton for punishment Oct 20 22:07:29 :) Oct 20 22:08:17 bluelightning: is it ok to delete the older opie recipes now that you have committed shiny new ones Oct 20 22:08:26 bluelightning, I'll send a note to eFfeM_work for punishment ;) Oct 20 22:08:28 may be you can obsolete them Oct 20 22:08:50 he'll hunt for those unpinned versions Oct 20 22:08:50 khem: I don't want to drop 1.2.3 until 1.2.5 is out Oct 20 22:09:03 but there could be some other cleanups possibly Oct 20 22:09:17 one of the preferred-* files is definitely obsolete Oct 20 22:09:58 bluelightning: ok fair enough Oct 20 22:10:02 chinook who maintains that Oct 20 22:10:11 ok, grep gives only a few distros Oct 20 22:10:29 angstrom jlime, kaeilos, minimal, shr Oct 20 22:10:42 and our amsdelta ;) Oct 20 22:11:22 and micro Oct 20 22:11:36 hm.. ? d/conf$ grep -R preferred-opie-versions . Oct 20 22:12:20 gn Oct 20 22:12:40 hey grg Oct 20 22:13:06 if you set DL_DIR_local in site.conf and don't set DL_DIR outside of bitbake.conf, DL_DIR ends up the local value in bitabke -e, but not in bitbake -e recipe Oct 20 22:13:11 very strange. Oct 20 22:13:24 bluelightning, it seems to me only preferred-opie-versions..2.4.inc is included Oct 20 22:13:40 i think the var must be used somehow before the override is applied Oct 20 22:13:42 hmm Oct 20 22:13:48 apart amsdelta-oe.conf ;)) Oct 20 22:14:46 morning khem Oct 20 22:15:02 ant_: preferred-opie-versions.inc is definitely out of date, I'll remove it Oct 20 22:15:09 pls do :) Oct 20 22:15:41 look at res of /oe/openembedded$ grep -R preferred-opie-versions . Oct 20 22:16:28 fyi, use git grep Oct 20 22:16:29 much faster Oct 20 22:16:55 right...old school here Oct 20 22:18:02 * khem loves git grep Oct 20 22:18:31 grg: bitbake world has improved Oct 20 22:18:50 khem, oh? what goodies have you committed? Oct 20 22:19:01 grg: not much Oct 20 22:19:09 its combined stuff over the period Oct 20 22:19:36 I just ran it recently and less breakages Oct 20 22:19:38 ok, it seems th eamsdelta thing has been touched by mickey|dinner and koen Oct 20 22:20:05 we still need to separate out some of ti stuff which get included for non ti cortex-a8 Oct 20 22:20:42 My plan is to run bitbake world twice each weekend, both on the same testing branch. Oct 20 22:21:04 good boy Oct 20 22:21:09 and try to sort out some races Oct 20 22:21:28 and/or missing deps Oct 20 22:21:38 kergoth_: autoreconf: `configure.ac' or `configure.in' is required it seems not its your turn :) Oct 20 22:21:52 kergoth_: this is expat-native Oct 20 22:21:53 ? Oct 20 22:21:56 grg, what's puzzling me, I have race issues with gtk+-native *only* on Gentoo with self-compiled kernel :/ Oct 20 22:22:13 at least until a couple of months ago Oct 20 22:22:26 on Ubuntu / 32b is ok Oct 20 22:22:27 hmm Oct 20 22:22:28 ant_, could be that it is easier to hit with that combo Oct 20 22:22:35 http://pastebin.com/g5tNMy9w Oct 20 22:22:36 bot 32 bits Oct 20 22:22:42 *both Oct 20 22:22:57 perhaps kernel scheduler? Oct 20 22:23:13 same filesystem (OE on another partition) Oct 20 22:23:35 03Chris Larson  07master * r4c0ffd9aaa 10openembedded.git/recipes/expat/expat.inc: Oct 20 22:23:35 expat: set acpaths=, not =+ Oct 20 22:23:35 Signed-off-by: Chris Larson Oct 20 22:23:48 grg: I'll do almost the same :) Oct 20 22:24:00 another good fellow Oct 20 22:24:19 khem: that should fix it, sorry about that Oct 20 22:24:47 kergoth_, may I ask you what is exactly DEPLOY_DIR_TOOLS for? Oct 20 22:25:04 no idea Oct 20 22:25:11 firmwares? Oct 20 22:25:13 I wish to do something like: for r in recipes/*/*.bb; do bitbake $r || echo $r failed > log; rm -rf $OETMP; done Oct 20 22:25:19 Hello Oct 20 22:26:17 Jay7: who will take care of build time deps ? Oct 20 22:26:39 Jay7: to what purpose? Oct 20 22:26:41 i made a xorg.conf for my device,that runs on Framebuffer 24BPP,but when i put "Depth 1" Oct 20 22:26:43 ops Oct 20 22:26:43 khem: this check is for lost depends :) Oct 20 22:26:44 sorry Oct 20 22:27:00 Jay7: that's not going to work. you likely want bitbake -s, not a glob Oct 20 22:27:05 i pressed by accident enter (lol) Oct 20 22:27:15 well.. that was concept :) Oct 20 22:27:30 Jay7: i have a similar script, but it's for checking for relocation bugs Oct 20 22:27:41 I hope there is possibility to script something around bitbake classes :) Oct 20 22:27:43 for each recipe, it prebuilds all of its deps and rebuilds it using the prebuilts that came from a different tmpdir Oct 20 22:27:47 slllloowwww Oct 20 22:27:51 but useful Oct 20 22:28:07 wow Oct 20 22:28:13 ant_: actually I think DEPLOY_DIR_TOOLS is redundant they should have just installed it into native sysroot Oct 20 22:28:18 i made a xorg.conf for my device,that runs on Framebuffer 24BPP,but when i put "Depth 24" the server don't start, it only start with parameter with 16..any help? i need pastebin my xorg.conf?? Oct 20 22:28:22 bunch of native recipes use it Oct 20 22:28:44 kergoth_: don't loose that script :) I'll ask for it soon :) Oct 20 22:28:47 * kergoth_ pushes some updates for darwin branch Oct 20 22:29:10 angelox_123, sounds like the driver doesn't support 24bpp, only 16 Oct 20 22:29:18 strange Oct 20 22:29:21 because if i do: Oct 20 22:29:27 Xorg -fbbpp 24 Oct 20 22:29:30 it opens Oct 20 22:29:52 ok, then you need to check the Xorg.log Oct 20 22:30:01 ok Oct 20 22:30:06 i made here work Oct 20 22:30:07 see: Oct 20 22:30:24 Xorg -weight 666 Oct 20 22:30:29 work with 18bpp selected Oct 20 22:30:48 but,have anyhow to don't need that -weight parameter?anything to place on xorg.conf Oct 20 22:32:00 gah, stupid wrappers Oct 20 22:32:03 angelox_123, sorry i'm not an xorg expert. Perhaps #xorg might help further? Oct 20 22:32:08 lately x fails to start on qemu Oct 20 22:32:16 says missing xmodmap Oct 20 22:32:19 i never hear about that channel Oct 20 22:32:22 i'll try thanks Oct 20 22:32:25 :) Oct 20 22:32:29 grg: do u boot with X on qemu Oct 20 22:32:35 i've never had xorg work for me. I always have to use kdrive Oct 20 22:32:42 khem, no. never tried X on qemu Oct 20 22:32:54 grg: hmmm Oct 20 22:32:59 me too,i ever used kdrive,but paulf(from openmoko) helped me :D Oct 20 22:33:11 we need some fancy gui booting for qemu as showcase Oct 20 22:33:25 may be latest QT Oct 20 22:33:26 and now flex-native dies in the autotools build Oct 20 22:33:28 * kergoth_ growls Oct 20 22:36:16 grg,worked here,without help of #xorg,i'll say here what i did to work(if anybody have the same problem,we'll can help) Oct 20 22:36:40 i put parameter: Weight 6 6 6 under Depth parameter on subsection of screen from xorg.conf Oct 20 22:36:43 and it worked fine Oct 20 22:37:41 ah Oct 20 22:38:00 before i forget,i have here a patch to apply on all ***-image from Angstrom Oct 20 22:38:51 it did work here on my phone shutdown and restart function Oct 20 22:45:25 is there a var to select which gst-plugins to use Oct 20 22:45:41 multiple providers are available for runtime gst-plugin-playbin (gst-plugins-bad, gst-plugins-good, gst-plugins-base, gst-plugins-ugly) Oct 20 22:47:16 ant_: do we have a sane image for initramfs which uses klibc Oct 20 22:47:39 you should modify the only one Oct 20 22:47:47 but basically, yes Oct 20 22:48:03 ant_: which image recipe it will be Oct 20 22:48:18 btw, should I rename initramfs-kexecboot-image -> initramfs-kexecboot-klibc-image ? Oct 20 22:48:24 ^^ Oct 20 22:48:42 one would be *libc agnostic Oct 20 22:50:08 if it uses klibc then its good to mention that in name Oct 20 22:50:22 ok, then wait a mom ;) Oct 20 22:51:08 I had this in queue Oct 20 22:51:32 * ant_ hopes nothing broke ;) Oct 20 22:52:17 03Andrea Adami  07org.openembedded.dev * r455a48017a 10openembedded.git/recipes/zaurus-utils/ (6 files): zaurus-utils: apply oe-stylize Oct 20 22:52:17 03Andrea Adami  07org.openembedded.dev * rcd9800e7f3 10openembedded.git/recipes/images/initramfs-kexecboot-klibc-image.bb: initramfs-kexecboot-klibc-image: klibc variant of initramfs-kexecboot-image Oct 20 22:52:17 03Andrea Adami  07org.openembedded.dev * rac6200745d 10openembedded.git/recipes/kexecboot/ (4 files): kexecboot: split kexecboot-klibc recipe and oe-stylize Oct 20 22:52:28 03Andrea Adami  07org.openembedded.dev * rd3a9a67eff 10openembedded.git/recipes/kexec-tools/kexec-tools_2.0.2.bb: kexec-tools: explicit packaging of kexec and kdump Oct 20 22:58:53 03Andrea Adami  07org.openembedded.dev * rf19ce6de6c 10openembedded.git/recipes/linux/linux-kexecboot.inc: Oct 20 22:58:53 linux-kexecboot: set INITRAMFS_IMAGE = "initramfs-kexecboot-klibc-image" Oct 20 22:58:53 * initramfs-kexecboot-image is now *libc agnostic Oct 20 22:59:05 khem,: done Oct 20 22:59:43 khem: I'd appreciate if you could boot some linux-kexecboot kernels on qemu Oct 20 23:00:05 ...or try to... Oct 20 23:01:54 ant_: how to do that Oct 20 23:02:03 kergoth_: http://patchwork.openembedded.org/patch/2601/ what do u think about this patch Oct 20 23:02:10 some qemu* machines are supported Oct 20 23:02:20 ant_: qemuarm ? Oct 20 23:02:20 just bitbake linux-kexecboot Oct 20 23:02:29 ok Oct 20 23:02:30 then Oct 20 23:02:34 I'm looking... Oct 20 23:02:44 can you add a wiki page plz Oct 20 23:02:56 how to work with kexecboot Oct 20 23:03:07 I see qemuarm and qemux86 ?? Oct 20 23:03:13 hem: doesn't seem unreasonable Oct 20 23:03:18 no idea about the test done... Oct 20 23:03:33 kergoth_: I am tempted to queue it in my staging Oct 20 23:03:44 khem, we have some infos Oct 20 23:04:32 http://wiki.openembedded.org/index.php/Zaurus Oct 20 23:04:40 expat-native dying on one machine Oct 20 23:05:02 oprofile on another Oct 20 23:05:20 Crofton|work: re expat, do you have 4c0ffd9? Oct 20 23:05:23 khem, http://elinux.org/CELF_Project_Proposal/Improve_kexecboot Oct 20 23:05:23 that commit should fix it Oct 20 23:06:02 ant_: I've written some info on Angstrom wiki about starting images under qemu-system-arm Oct 20 23:06:17 pls edit oe wiki Oct 20 23:06:37 I should start kexecboot.org one time.. Oct 20 23:10:35 i'm trying to compile blackbox Oct 20 23:10:40 but i get a error Oct 20 23:10:45 http://pastebin.com/1fizVFG3 Oct 20 23:11:51 kergoth_, checking Oct 20 23:12:52 kergoth_, just did a pull and restart Oct 20 23:14:25 angelox_123, looks like blacckbox does not build with newer g++ Oct 20 23:14:34 k Oct 20 23:14:37 usually these have been fixed upstream by now Oct 20 23:15:04 " usually these have been fixed upstream by now" you said that to me? Oct 20 23:15:36 ah ha, the sed wrapper script is losing empty arguments, that would exlpain this behavior.. Oct 20 23:15:38 hrm Oct 20 23:15:44 angelox_123, yes Oct 20 23:15:55 ok,i did now a "git pull" Oct 20 23:15:58 trying again Oct 20 23:16:00 khem, have you seen NOTE: multiple providers are available for runtime module-init-tools-depmod (module-init-tools, module-init-tools-cross) Oct 20 23:16:24 what is blackbox? Oct 20 23:16:34 a window manager Oct 20 23:16:49 lightweight Oct 20 23:17:44 we have the latest version Oct 20 23:17:50 but it is five years old Oct 20 23:18:13 if you want to build it, you will need to create a patch that lets it compile with newer g++ Oct 20 23:18:14 Do you know another light window manager? Oct 20 23:18:39 not really Oct 20 23:18:49 I do mostly console work :) Oct 20 23:18:53 me too Oct 20 23:18:59 but i'm building to final user Oct 20 23:19:00 :D Oct 20 23:19:25 i created a project to a operational system.... Oct 20 23:19:29 to mobile devices Oct 20 23:21:44 angelox_123: machbox is widely used on x11-based mobile env Oct 20 23:22:15 don't works fine here Oct 20 23:22:29 check fluxbox too then Oct 20 23:23:06 i'll check Oct 20 23:28:11 here is the oprofile failure Oct 20 23:28:12 http://pastebin.com/PnwK3Yrp Oct 20 23:28:20 smells like a libtool problem Oct 20 23:30:35 i did git pull but it'll rebuild all Oct 20 23:30:39 what is the problem? Oct 20 23:31:31 angelox_123, i've been playing with fvwm lately. Its good if you've got the time for reading the mountainous documentation Oct 20 23:31:47 * kergoth_ <3 the awesome wm on his linux laptop Oct 20 23:31:56 ok Oct 20 23:32:02 but about my problem Oct 20 23:32:39 i did git pull and after a bitbake fluxbox and it is building Running task 814 of 1823 Oct 20 23:32:44 yes, so? Oct 20 23:32:47 it are building all from zero Oct 20 23:32:50 no, it isnt Oct 20 23:33:00 its rebuilding things that had their versions/revisions bumped due to upstream changes Oct 20 23:33:14 hum Oct 20 23:33:14 ok Oct 20 23:33:17 sorry Oct 20 23:33:19 :) newbie Oct 20 23:34:04 i need read more about the fantastic OE Oct 20 23:37:04 03Paul Eggleton  07org.openembedded.dev * ra5c71447bf 10openembedded.git/conf/distro/include/preferred-opie-versions.inc: Oct 20 23:37:04 conf/distro/include: remove obsolete preferred-opie-versions.inc Oct 20 23:37:04 Specific version include files should now be used - this old one points Oct 20 23:37:04 to 1.2.2 which is no longer supported in the OE dev branch. Oct 20 23:41:43 damn, damn, damn Oct 20 23:42:10 ? Oct 20 23:42:44 03Andrea Adami  07org.openembedded.dev * r98643fe45a 10openembedded.git/recipes/sharp-binary-only/ (9 files in 4 dirs): sharp-binary-only: remove unused libs/utils for Zaurus Oct 20 23:42:55 03Andrea Adami  07org.openembedded.dev * r6ba1aada00 10openembedded.git/recipes/sharprom-toolchain/sharprom-toolchain-native_1.0.bb: sharprom-toolchain: remove now obsoleted/unbuildable toolchain Oct 20 23:43:22 kergoth_: http://patchwork.openembedded.org/patch/2883/ would you take it for bitbake Oct 20 23:45:37 khem, oprofile is being a pain Oct 20 23:45:45 http://pastebin.com/PnwK3Yrp Oct 20 23:45:54 I removed it from my image for the time being Oct 20 23:46:09 are you in the building off 237? Oct 20 23:47:25 khem: yeah, that looks like a pretty obvious one Oct 20 23:51:25 Crofton|work: yes Oct 20 23:52:37 heh Oct 20 23:52:48 my friend out there knows where you are :) Oct 20 23:53:41 where are the ac vars from the 'site' directory parsed? Oct 20 23:53:42 ok, I need to go afk Oct 20 23:53:50 I've had enough for today Oct 20 23:53:51 Crofton|work: you need libtool macros update Oct 20 23:53:56 Crofton|work: heh Oct 20 23:53:57 khem: what's the scoop on libtool 2.4 problems, hitting one where libtool says the LT_INIT wasn't from a matching version -- the odd thing is, an aclocal --verbose shows it using the libtool.m4 from sysroots, not the local one Oct 20 23:54:04 grg: see autotools.bbclass, CONFIG_SITE Oct 20 23:54:06 is there a sampel commit? Oct 20 23:54:15 kergoth_, thanks Oct 20 23:54:25 hrm Oct 20 23:54:28 Crofton|work: ummm look at gnutls_2.10.1.bb Oct 20 23:54:34 thanks Oct 20 23:54:41 I will look at that when I can Oct 20 23:54:44 kergoth_: yeah that means its not regnerating configure Oct 20 23:54:59 hmmm, odd Oct 20 23:55:18 * kergoth_ tests with --force back in the autoreconf exec Oct 20 23:55:38 kergoth_: or there are some .m4 in the tree which are used Oct 20 23:55:43 and not the ones from sysroot Oct 20 23:56:05 didnt see any in the --verbose, but will double check Oct 20 23:56:17 kergoth_: whic h recipe is it Oct 20 23:57:07 this is me knee deep in my autotools branch, trying to build a non-default version of gettext-native Oct 20 23:57:09 somebody shoot me Oct 20 23:58:04 kergoth_: heh Oct 20 23:58:25 inexplicably flex-native failed to build due to the gettext version / automake version MKINSTALLDIRS crap Oct 20 23:58:36 so i was building 0.17 to run gettextize to update the build files to a less shitty version Oct 20 23:58:40 heh Oct 20 23:58:42 gettext 0.18 has to be rammed down the throat Oct 20 23:58:57 i figured i'd use 0.17 since it updates the minimum version, and we still have 0.17 in the tree Oct 20 23:58:59 kergoth_: gettext is pretty much like libtool Oct 20 23:59:01 updating to 0.18 was pretty painless Oct 20 23:59:14 once you upgrade you cant go back easily Oct 20 23:59:23 we can drop .18 Oct 20 23:59:27 ups .17 Oct 20 23:59:29 its too bad so dman many gettext versions are buggy as hell Oct 20 23:59:38 yeah Oct 20 23:59:44 no one cares for cross compiling Oct 21 00:00:00 Crofton|work: oprofile 0.9.6 build all fine for me Oct 21 00:00:11 i was really hoping to avoid --force Oct 21 00:00:13 hrmph, oh well Oct 21 00:00:18 * kergoth_ hopes that fixes it Oct 21 00:00:30 clean build? Oct 21 00:00:45 i think we need a site/mips32-common and a site/mips64-common Oct 21 00:00:53 too much mips variations... Oct 21 00:01:10 khem, I am going to work around for now Oct 21 00:01:19 there is orbit2 alignment stuff in mipse-common which is wrong for 64 bit btw Oct 21 00:01:32 heh, yep, sounds like its time for the split Oct 21 00:02:19 grg: 64bit is a mere copy of 32bit you should fix it as you find issues Oct 21 00:02:59 khem, i'm not using 64 bit, so am unlikely to run into issues. I just saw that mips-common has a 4 byte alignment for long. Which is wrong for 64bit Oct 21 00:03:03 kergoth_: bitbake decides to remove stamps for recipes on rebuild even when there was no change in recipe Oct 21 00:03:12 I wonder why would it do that Oct 21 00:03:35 grg: morally you should fix it :) Oct 21 00:03:55 now that I know mips64 exist Oct 21 00:03:57 in OE Oct 21 00:04:04 khem, yes. i'll split orbit stuff out of mips-common into mips{32,64}-common Oct 21 00:04:46 confirmed the only LT macros it used were from the sysroot, must be the configure thing Oct 21 00:05:09 (note: add aclocal --verbose; exit 1 to the configure, then grep for 'found macro' in the log) Oct 21 00:05:14 kergoth --force is there for a reason I guess :) Oct 21 00:05:18 :) Oct 21 00:05:23 its usually the case with that sort of thing Oct 21 00:05:34 though a hammer Oct 21 00:05:34 but autoreconf has a lot of intelligence to avoid rebuilding things when it doesn't have to Oct 21 00:05:48 it doesn't blindly handle aclocal, it runs it and sees if what it produced changed Oct 21 00:05:55 always runs aclocal, no matter what Oct 21 00:06:03 so there's no risk of it not updating it, in theory Oct 21 00:06:05 it could spend less time trying to be smart and instead just regenerate the damn thing Oct 21 00:06:15 dont try to outsmart dumb humans Oct 21 00:06:17 what it does doesn't take that long, as far as i can tell Oct 21 00:06:31 there should be no need for us to forcibly regenerate the world :| Oct 21 00:07:14 grg: you could move that definition to mip{64el}-linux{-uclibc} files Oct 21 00:07:24 * kergoth_ glares at gnu Oct 21 00:07:26 and not create mips64-common Oct 21 00:07:31 03Denys Dmytriyenko  07org.openembedded.dev * r43cdcc1bf9 10openembedded.git/conf/distro/include/toolchain-external.inc: Oct 21 00:07:32 toolchain-external: fix circular recursion if TOOLCHAIN_VENDOR is not defined Oct 21 00:07:32 Signed-off-by: Denys Dmytriyenko Oct 21 00:07:46 khem, yeah i suppose. I want to add some alignment defs for postgres too Oct 21 00:07:52 right Oct 21 00:08:00 common means common :) Oct 21 00:08:10 khem, and is mip64el-linux supposed to be mips64el-linux? Oct 21 00:08:21 i wonder how safe itd be to just add LIBTOOL=...path..to..sysroot/libtool to EXTRA_OEMAKE and say fuck using the generated one Oct 21 00:08:22 hehe Oct 21 00:08:23 yes Oct 21 00:08:55 gotta run Oct 21 00:08:57 ttyl Oct 21 00:09:05 later Oct 21 00:09:26 kergoth_: I don't think it's going to work ;) Oct 21 00:09:52 dj-death: its done in some of the recipes, someone gave up on the regeneration Oct 21 00:10:01 git grep EXTRA_OEMAKE.\*LIBTOOL Oct 21 00:10:18 kergoth_: we're using it in buildroot Oct 21 00:10:54 it makes me sad that we have to regenerate so much -- its certainly necessary to ensure our fixes in macros make it into the configure scripts of the things we're building, but its still annoying as hell Oct 21 00:11:01 kergoth_: but some makefile check if there is a generated libtool script and prefer to use it over the one in the path Oct 21 00:11:27 that being said, we could still remove the generated script ;) Oct 21 00:13:17 i didn't say use PATH :) Oct 21 00:13:29 i said explicitly override LIBTOOL makefile variable on the make commandline :) Oct 21 00:13:39 yeah same thing Oct 21 00:14:41 basicly Makefiles contain that : Oct 21 00:14:43 LIBTOOL = $(SHELL) $(top_builddir)/libtool Oct 21 00:16:31 hello Oct 21 00:16:40 if you're lucky, the Makefiles are going to check whether $(top_builddir)/libtool exists, if yes use it, otherwise use $LIBTOOL indeed Oct 21 00:17:21 is it possible to override ANGSTROM_FEED_CONFIGS or DISTRO_FEED_CONFIGS from an image file? or whatever other file? Oct 21 00:21:27 I'm back Oct 21 00:21:32 i did: bitbake fluxbox Oct 21 00:21:36 but i get a error on final: Oct 21 00:21:41 http://pastebin.com/ZM3Yr6q1 Oct 21 00:22:54 --force doesn't solve it, and rm -f'ing configure doesn't either.. what the hell is going on here Oct 21 00:26:38 i guess not :) Oct 21 00:30:35 anyhelp Oct 21 00:30:36 ? Oct 21 00:43:02 cat: ../.git/HEAD: No such file or directory Oct 21 00:43:04 angelox_123, Oct 21 00:43:22 me Oct 21 00:43:22 there is the error Oct 21 00:43:38 yeah, in that pastebin Oct 21 00:44:23 any fix? Oct 21 00:44:41 well, i'm not a dev - i just thought you missed the actual error Oct 21 00:46:34 it seems that the makefile expects the .git folder to be there; i suppose the recipe was written for the git version of fluxbox and something (a makefile patch) slipped to the -1.1.1 version Oct 21 00:48:05 but again, i'm guessing Oct 21 00:57:47 khem: I told you it was too soon to get off your back! :) your anonymous function in binutils.inc breaks something else for me... Oct 21 01:15:42 goodbye all* Oct 21 01:15:46 Thanks by all Oct 21 01:34:57 03Khem Raj  07master * rca22079a28 10openembedded.git/recipes/qpe-gaim/qpe-libgaim_1.5.0.bb: Oct 21 01:34:57 qpe-libgaim_1.5.0.bb: Add new location of glib-2.0 headers to compiler include file search Oct 21 01:34:57 Signed-off-by: Khem Raj Oct 21 01:34:57 03Graham Gower  07master * rb72db8f97f 10openembedded.git/recipes/nail/ (files/makeconfig.FILES.patch nail_11.21.bb): (log message trimmed) Oct 21 01:34:58 nail_11.21.bb: Fix build failure due to dumbness in a configure-like script. Oct 21 01:34:58 Presumably this build failure is due to a race condition, whereby the LIBS Oct 21 01:34:59 file contains some garbage when it is autogenerated by the 'makeconfig' script. Oct 21 01:34:59 mipsel-oe-linux-gcc -march=mips32 -L/mnt/oe/tmp/sysroots/mipsel-oe-linux/usr/lib Oct 21 01:35:00 -Wl,-rpath-link,/mnt/oe/tmp/sysroots/mipsel-oe-linux/usr/lib -Wl,-O1 aux.o Oct 21 01:35:00 base64.o cache.o cmd1.o cmd2.o cmd3.o cmdtab.o collect.o dotlock.o edit.o fio.o Oct 21 01:35:38 03Graham Gower  07master * rf4bd3c4532 10openembedded.git/recipes/xoscope/xoscope_1.12.bb: (log message trimmed) Oct 21 01:35:38 xoscope_1.12.bb: Fix build failure when comedilib has already been staged. Oct 21 01:35:38 If comedilib is already staged, xoscope tries to use it and fails to build. Oct 21 01:35:39 mipsel-oe-linux-gcc -march=mips32 -DHAVE_CONFIG_H -I. -I.. Oct 21 01:35:39 -I/mnt/oe/tmp/sysroots/mipsel-oe-linux/usr/include/gtk-1.2 Oct 21 01:35:40 -I/mnt/oe/tmp/sysroots/mipsel-oe-linux/usr/include/glib-1.2 Oct 21 01:35:40 -I/mnt/oe/tmp/sysroots/mipsel-oe-linux/usr/lib/glib/include Oct 21 01:35:41 03Ash Charles  07master * rfb69129560 10openembedded.git/recipes/linux/ (48 files in 3 dirs): Oct 21 01:35:41 linux-gumstix-2.6.15 & gumstix_2.6.5-gnalm1: removed outdated recipes Oct 21 01:35:42 * removed two old recipes in the linux kernel with pre-2.6.20 kernel Oct 21 01:35:51 as suggested by Marcin's RFC of 20 Oct 2010. Oct 21 01:35:51 Signed-off-by: Ash Charles Oct 21 01:35:51 Signed-off-by: Khem Raj Oct 21 01:35:52 03Andreas Oberritter  07master * rf698175c29 10openembedded.git/recipes/libsigc++-1.2/libsigc++-1.2_1.2.5.bb: Oct 21 01:35:52 libsigc++-1.2.5: convert to new style staging Oct 21 01:35:53 Signed-off-by: Andreas Oberritter Oct 21 01:35:53 Signed-off-by: Khem Raj Oct 21 01:35:54 03Petr Štetiar  07master * rb78d5938aa 10openembedded.git/recipes/busybox/ (busybox.inc files/hwclock.sh files/ts72xx/hwclock-default): Oct 21 01:35:54 busybox: add ts72xx hwclock-default file and modify hwclock.sh Oct 21 01:35:55 This patch simplifies the hwclock.sh init script and adds a possibility to Oct 21 01:35:55 fallback to default RTC device even if the device specified in HWCLOCKDEVICE Oct 21 01:35:56 (128 lines omitted) Oct 21 01:40:12 denix: bummer what did it break now Oct 21 01:40:50 which package should have staged sysroots/x86_64-linux/usr/bin/java Oct 21 01:46:59 khem: it gives me a circular recursion evaluating variables in binutils-canadian-sdk recipes Oct 21 01:47:25 hmm Oct 21 01:47:28 which variables ? Oct 21 01:47:55 is it unearthing something latent or creating a new obstaclw Oct 21 01:47:55 khem: I'm setting SDK_PATH based on TOOLCHAIN_PATH Oct 21 01:48:28 hmm how does that trickle into CONFIGUREOPTS Oct 21 01:48:59 it worked fine before that change and if I revert it locally... Oct 21 01:49:17 not sure, I guess something in canadian/sdk class plus anonymous function... Oct 21 01:50:13 it gives this error - NOTE: :maximum recursion depth exceeded while calling a Python object while evaluating: Oct 21 01:50:39 and complains about following variables: ${SDK_PATH}, ${prefix}/bin, ${STAGING_DIR_HOST}${bindir}, ${STAGING_BINDIR} Oct 21 01:52:47 not sure about direct connection, but git-bisect shows that commit to blame Oct 21 01:52:51 i have to run some errands but you can fiddle around with where to place that inherit stuff Oct 21 01:53:04 I can see the connection Oct 21 01:53:16 how can I reproduce it Oct 21 01:53:26 bitbake binutils-cross-sdk ? Oct 21 01:53:30 or something Oct 21 01:53:43 will look into later when I return Oct 21 01:54:05 I'm not even building anythinbg cross Oct 21 01:54:22 it cycles while initially parsing all the recipes Oct 21 01:59:29 weird does not happen for me Oct 21 01:59:47 your local.conf might have something thats triggering it Oct 21 02:00:14 are you setting SDK_PATH on local.conf ? Oct 21 02:00:31 whee, darwin branch is coming along Oct 21 02:01:18 kergoth_: how far are you Oct 21 02:01:35 kergoth_: I might try it on MBP here Oct 21 02:01:45 not sure exactly how many tasks remain -- i have yet to complete a task-boot, even Oct 21 02:02:23 oh hell, openssl wants to use a version script for ld, osx ld oesn't know what to make of that Oct 21 02:03:00 hmmm Oct 21 02:03:11 kergoth_: macports might have newer binutils Oct 21 02:03:37 i'm sure it does, but i'd rather avoid relying too much on macports or homebrew or anything, if possible Oct 21 02:03:38 hrmph Oct 21 02:03:50 khem: yep, I'm setting SDK_PATH in local.conf. it's slightly complicated - I'll try to extract the relevant part... Oct 21 02:05:59 denix: hmmm **** ENDING LOGGING AT Thu Oct 21 02:59:57 2010