**** BEGIN LOGGING AT Wed Nov 03 02:59:57 2010 Nov 03 03:00:26 cool mrj10..you use angstrom-setup-scripts too :) Nov 03 03:00:38 i had see here : "mrj10@mjlap:~/beagleboard/kernel/build-validation-image/angstrom-setup-scripts/sources/openembedded$" Nov 03 03:02:08 yeah, i practice what i preach :P Nov 03 03:02:39 ok thanks everybody Nov 03 03:02:41 i'll sleep Nov 03 03:02:41 i originally came to those setup scripts through some other scripts mentioned here http://code.google.com/p/beagleboard/wiki/BeagleBoardDiagnosticsNext Nov 03 03:08:09 * khem thinks of blogging on setting up OE Nov 03 03:09:18 khem, good plan. Nov 03 03:09:48 ka6sox: btw. should I change the owner permission think on melo for sources Nov 03 03:10:07 * kergoth sifts through the branches he was working on before the trip to minnesota Nov 03 03:10:24 kergoth: ah welcome back :) Nov 03 03:10:28 did u vote Nov 03 03:10:35 hadn't seen my dad or stepdad in 5 years Nov 03 03:10:43 khem, ya, I'm going to do that now...I think I understand what is wrong. Nov 03 03:11:02 ka6sox: yeah the dir is in var Nov 03 03:11:07 is there a reason the sources repo is *not* browsable? Nov 03 03:11:18 ka6sox: it doesnt have to be Nov 03 03:11:29 it will hit google hit needlessly Nov 03 03:11:44 kergoth: cool Nov 03 03:11:49 okay so we can't rely on robots.txt? Nov 03 03:12:13 ka6sox: actually there is no need to make it browsable I think Nov 03 03:12:31 khem, okay Nov 03 03:12:35 its a fallback mirror for OE srcs and we can keep it hidden from limelight Nov 03 03:13:15 okay where is the primary? Nov 03 03:13:19 ka6sox: can we use hudson now ? Nov 03 03:13:30 primary is the web :) Nov 03 03:13:39 ah, got it... Nov 03 03:13:40 IOW the original location of the tars Nov 03 03:14:19 okay, hudson just needs me to finish up a -minimal test on the autobuilder and then its good to go. Nov 03 03:14:33 fantabulous Nov 03 03:22:34 https://github.com/kergoth/openembedded/compare/srctree drops coreutils-native and stagemanager-native recipes - need to do more test builds with it before i send to the list though Nov 03 03:22:37 * kergoth kicks one off Nov 03 04:21:16 kergoth: that new install script is bash like Nov 03 04:21:29 may be you want #!/bin/bash Nov 03 04:21:32 it doesn't use any features that are bash specific. Nov 03 04:22:13 I thought local functions were bash Nov 03 04:22:17 getopts is actually much more portable than getopt Nov 03 04:22:20 nope Nov 03 04:22:26 defining them with 'function' is bash specific Nov 03 04:22:28 right I just check susv Nov 03 04:22:30 but foo() {} isn't Nov 03 04:23:13 may be instead of which you could use type Nov 03 04:23:32 sed is always not in /usr/bin Nov 03 04:23:44 freebsd comes to mind Nov 03 04:25:25 kergoth: the change looks sane to me Nov 03 04:26:09 kergoth: how does github workflow work Nov 03 04:28:52 ah, right, will drop the /usr/bin/sed bit, that was a copy/paste remnant from the sed wrapper from the darwin stuff Nov 03 04:28:54 heh Nov 03 04:33:58 type is more verbose than which Nov 03 04:35:12 some systems may not have which Nov 03 04:41:02 true Nov 03 04:41:06 will check into that Nov 03 04:41:56 the question is, is the format of type's output portable Nov 03 04:42:00 susv3 doesn't define the format Nov 03 04:42:04 and the output is more than just the path Nov 03 04:46:43 kergoth use type -ap Nov 03 04:50:26 khem: not supported by dash. Nov 03 04:50:31 nor is that part of susv3 Nov 03 04:51:19 kergoth: hmmm Nov 03 04:51:59 then pass to cut or some sed Nov 03 04:55:42 that's what i was thinking, remove the current install script dirname from PATH and exec the real version with that path Nov 03 04:57:44 kergoth: on a different note try to fetch opencvs_cvs.bb Nov 03 04:57:54 exposed a bug in cvs fetcher :) Nov 03 04:58:46 fetcher does a basename on SRC_URI Nov 03 04:59:11 which gives cvs and sees that cvs dir already exist and bails out Nov 03 04:59:35 but if I shuffle SRC_URI properties then it works ok Nov 03 04:59:57 http://pastebin.com/Sytdfknk works ok Nov 03 05:00:16 all I did is shuffle method bit Nov 03 05:02:03 kergoth: on yet anothe note whats your take on this issue http://patchwork.openembedded.org/patch/3460/ Nov 03 05:02:26 its being seen by many users of 1.8.x Nov 03 05:02:29 bitbake Nov 03 05:04:34 khem: that's already in Nov 03 05:04:37 see current base.bbclass Nov 03 05:07:27 03Graham Gower  07master * r6d71a1c8e6 10openembedded.git/site/ (mips-common mips64-common): Nov 03 05:07:27 site/mips{,64}-common: Add apache vars. Nov 03 05:07:27 Also, fix CORBA pointer alignment for mips64 that I missed previously. Nov 03 05:07:27 Signed-off-by: Graham Gower Nov 03 05:07:27 Signed-off-by: Khem Raj Nov 03 05:07:30 ok Nov 03 05:07:38 03Graham Gower  07master * r12795100f3 10openembedded.git/recipes/telepathy/telepathy-gabble_0.7.27.bb: Nov 03 05:07:38 telepathy-gabble_0.7.27.bb: Add libsoup-2.4 to DEPENDS. Nov 03 05:07:38 checking for SOUP... configure: error: Package requirements (libsoup-2.4) were not met: Nov 03 05:07:38 No package 'libsoup-2.4' found Nov 03 05:07:38 Signed-off-by: Graham Gower Nov 03 05:07:39 Acked-by: Frans Meulenbroeks Nov 03 05:07:39 Signed-off-by: Khem Raj Nov 03 05:07:40 03Graham Gower  07master * r8fc09c071a 10openembedded.git/recipes/python/python-cython_0.13.bb: Nov 03 05:07:40 python-cython_0.13.bb: Add python to DEPENDS. Nov 03 05:07:41 /usr/bin/cython is a python script. Fix the license while here. Nov 03 05:07:42 Signed-off-by: Graham Gower Nov 03 05:07:42 Signed-off-by: Khem Raj Nov 03 05:07:43 there we go, fixed the install wrapper - https://github.com/kergoth/openembedded/blob/b286f43/bin/install Nov 03 05:07:43 03Anders Darander  07master * re43aca568f 10openembedded.git/recipes/resolvconf/resolvconf_1.45.bb: Nov 03 05:07:56 will adjust the other wrapper scripts to match Nov 03 05:08:15 kergoth: how about this one http://patchwork.openembedded.org/patch/3455/ Nov 03 05:08:16 Resolvconf 1.45 is only available from the snapshot mirror. Nov 03 05:08:16 Remove the recipee for 1.45. Nov 03 05:08:16 Signed-off-by: Anders Darander Nov 03 05:08:16 Signed-off-by: Khem Raj Nov 03 05:08:41 khem: that does need to go in Nov 03 05:08:53 will look at it tomorrow Nov 03 05:09:40 kergoth: ok the patch is sort of broken by MUA but you can prolly take care of that :) Nov 03 05:09:45 * kergoth nods Nov 03 05:09:48 the principle is sound Nov 03 05:09:51 the install script now looks ok Nov 03 05:10:12 tried sed at first, but it can't replace with \n, tr did the job well enough :) Nov 03 05:10:23 * kergoth certainly is learning some shell tricks recently Nov 03 05:12:01 yeah tr is fancy Nov 03 05:12:57 kergoth: how should we work together with yocto, I want OE and yocto and poky to work in sync Nov 03 05:13:12 any ideas you have Nov 03 05:14:04 right now poky has done lot of cleanup of metadata and oe has prolly done its part and added more recipes and newer versions of some and also some features Nov 03 05:14:21 I think this all could merge in Nov 03 05:14:31 and community as a whole benefit Nov 03 05:16:17 yeah, i think ideally OE would get the metadata reorg, bsp layers, etc, as well as other cleanups from poky, but the resistance to change has been high in the past Nov 03 05:16:21 heh Nov 03 05:17:26 mrmoku: armv6j should work well with gcc 4.5 and eglibc yes I dont know of any issues Nov 03 05:18:04 I think these things are good for oe Nov 03 05:18:12 we should be agile Nov 03 05:18:35 and make it possible to make changes easily Nov 03 05:18:58 * khem should zzz too Nov 03 05:19:19 good night Nov 03 05:22:38 nn Nov 03 07:20:01 gm Nov 03 07:46:56 khem: ok, thanks Nov 03 08:27:29 re Nov 03 08:41:59 good morning Nov 03 09:45:44 good morning Nov 03 09:46:39 is there a way to have bitbake be a little bit more verbose for parsing errors ? Actually I get 4 x line xx, in function messages then "ParsonErrosFound" whihc makes it diffcult to locate the error Nov 03 09:46:43 ? Nov 03 09:48:17 add -D Nov 03 09:49:42 hrw: thanks but that doesn't change anything Nov 03 09:50:32 ericben, -v will also help and you can do it multiple times, if you want the max do soemthing like Nov 03 09:50:37 doing bitbake -b the recipe brings the log Nov 03 09:50:37 ericben: add few -D Nov 03 09:50:39 bitbake -D -D -D -v -v -v whatever Nov 03 09:50:39 -DDDDDDD Nov 03 09:50:43 ah ok Nov 03 09:51:00 hrw, that is quite a size :-) Nov 03 09:51:16 ? Nov 03 09:51:56 ok -DDD brings more log thanks Nov 03 09:52:06 I knew -v but not -D Nov 03 09:52:18 ericben: -h then should be next Nov 03 09:52:21 hrw: google images for double-D (or maybe DD), guess you do not want to do that at work :-) Nov 03 09:52:30 hrw: yes :) Nov 03 09:52:36 nor not when the Mrs is looking ;-) Nov 03 09:52:42 eFfeM_work: ah.. ok Nov 03 09:52:58 eFfeM_work: google for 38MM Nov 03 09:53:22 guess I better do that at home ;-) Nov 03 09:53:22 but increase debug level was not exactly what I would enable to find parsing error in a recipe as I would expect he option being useful for bitbake debug not for recipe debug Nov 03 09:54:05 eFfeM_work: better do not even attempt Nov 03 09:54:16 :-) Nov 03 09:54:21 eFfeM_work: DD are better ;D Nov 03 09:54:53 ~curse git bisect for being good only for bad commits Nov 03 09:54:54 May the fleas of a thousand camels infest your most sensitive regions, git bisect for being good only for bad commits ! Nov 03 09:55:08 * hrw manually bisects binutils to find that good commit Nov 03 10:18:22 03Koen Kooi  07org.openembedded.dev * r61a4d1805c 10openembedded.git/recipes/xorg-app/xterm_266.bb: Nov 03 10:18:22 xterm: add 266 Nov 03 10:18:22 Signed-off-by: Koen Kooi Nov 03 10:39:48 03Paul Menzel  07org.openembedded.dev * r93dca1c604 10openembedded.git/recipes/vlc/vlc_1.1.4.1.bb: (log message trimmed) Nov 03 10:39:48 vlc_1.1.4.1: add `lua5.1` to `DEPENDS` Nov 03 10:39:48 After VLC 1.0.6, being the latest version in OE until addition of 1.1.4.1 in commit 30e362 [2], some interfaces were implemented in Lua in VLC 1.1.0 [1]. Nov 03 10:39:48 Interfaces: Nov 03 10:39:48 * Renamed the legacy rc, telnet and http interfaces to oldrc, oldtelnet and oldhttp. Nov 03 10:39:48 * rc, telnet and http are now implemented using the lua interface system. Nov 03 10:39:49 Task `configure` fails with the following error message. Nov 03 10:52:17 03Martin Jansa  07master * rb2781103d5 10openembedded.git/recipes/linux/linux-2.6.34/use-noclone-attribute-for-naked.patch: Nov 03 10:52:17 linux-2.6.34: remove unused patch, removed from SRC_URI in 5786f08ba6066e9af60104c5d5606c106e8df7d2 Nov 03 10:52:17 Signed-off-by: Martin Jansa Nov 03 11:21:41 khem: (armv6j) it turned out to be a problem with arch linux as buildhost Nov 03 11:26:13 can't find where can I replace bisybox's /bin/ip with native ip utility. Can't find bisybox reciept even! Nov 03 11:26:52 dv: try in recipes/busybox/ Nov 03 11:27:30 ericben, I don't have it! in /openembedded/recipes/ Nov 03 11:27:51 ah Nov 03 11:28:07 try busybox with a u and not a i as the second letter ! Nov 03 11:28:47 sorry, see it... but can't see how can I replace /bin/ip with native utility Nov 03 11:30:11 seems oestats-client is very sensitive for network problems.. Nov 03 11:30:40 * Jay7 have a lot of builds in state 'Started' Nov 03 11:32:50 how can I replace /bin/ip with native utility instead of symlink to busybox ? Nov 03 11:34:30 dv: installing iproute2 may help Nov 03 11:38:06 ericben, I added iproute2 into PACKAGES without success Nov 03 11:39:16 dv: then you have a ip.iproute2 binary in your rootfs ? Nov 03 11:40:50 ericben, "find / -name "ip*" | grep rout" gives nothing Nov 03 11:44:18 Hi all Nov 03 11:48:53 ericben, thank you! seems, I know what's wrong Nov 03 13:40:18 i want to say sorry here,because i'm newbie on git-email,and i send 2 equal patches Nov 03 13:49:09 hi mrj10 Nov 03 13:49:30 hey Nov 03 13:50:12 i could send the patch by git Nov 03 13:50:14 but... Nov 03 13:50:18 per accident i send 2 Nov 03 13:50:20 :) Nov 03 13:50:28 not a big deal Nov 03 13:51:08 did you already do that? Nov 03 13:51:35 angelox_123: 2 pathes are better than no one :) Nov 03 13:51:39 it happens, i've sent a needless email before Nov 03 13:51:57 but i mean that i send 2 emails Nov 03 13:51:58 :) Nov 03 13:51:59 even if they are equal :) Nov 03 13:52:03 2 emails with 2 patches Nov 03 13:52:19 we all here humans Nov 03 13:52:23 :) Nov 03 13:52:24 I hope at least :) Nov 03 13:52:33 with exceptions of bots :) Nov 03 13:52:37 yes Nov 03 13:52:43 like ibot Nov 03 13:52:44 :) Nov 03 13:52:56 hm.. Nov 03 13:53:04 alwaysinnovating have smartbook Nov 03 13:53:11 very fun device Nov 03 13:53:19 is? Nov 03 13:53:27 http://www.alwaysinnovating.com/products/smartbook.htm Nov 03 13:53:27 can install linux into it? Nov 03 13:53:35 it is OE-based Nov 03 13:53:47 good Nov 03 13:53:48 very good Nov 03 13:55:10 well.. I should finish my testbuilder script today and continue to work on kexecboot Nov 03 14:33:27 what is current status of multimachine build? Nov 03 14:33:44 is bb 1.10 supported? Nov 03 14:35:19 those are unrelated questions, i hope? the latter has nothing to do with the former Nov 03 14:35:27 for multiple target right? Nov 03 14:35:35 JaMa: yes Nov 03 14:35:55 I mean something like MACHINE="tosa akita collie" bitbake console-image Nov 03 14:35:57 yes it works and as kergoth said, bitbake version is irelevant Nov 03 14:36:09 hmm not in this form :) Nov 03 14:36:20 what is right form then? Nov 03 14:36:21 to do that you have to set BB_ENV_EXTRAWHITE Nov 03 14:36:33 kergoth: I have :) Nov 03 14:36:38 for m in tosa akita collie; do MACHINE=$m bitbake console-image; done Nov 03 14:36:44 MACHINE is one machine, not a list Nov 03 14:36:49 yeah, JaMa's line is it Nov 03 14:36:54 hm.. Nov 03 14:37:18 seems I misunderstand this feature then :) Nov 03 14:37:33 i.e. one bitbake run is for one MACHINE only? Nov 03 14:37:36 it's multiple machines in one build dir Nov 03 14:37:46 well.. now I see Nov 03 14:38:22 so you don't have to build -native stuff more then once and also if machines have same arch, then it's also built just once Nov 03 14:39:26 you need two different toolchains for armv4 and armv5te though Nov 03 14:41:07 but toolchain build is automated so no worry Nov 03 14:41:38 thanks to all :) Nov 03 14:42:07 * Jay7 is implementing "groups" in testbuilder Nov 03 14:42:14 set of machines to build in one tmpdir Nov 03 14:42:34 is it useful, btw? :) Nov 03 14:43:23 I see only two points: 1) this can save some time; 2) this will test multi-machines build :) Nov 03 15:11:53 mickeyl: ping? Nov 03 15:15:07 hmm, I am very close now, everything seems to have built Nov 03 15:15:10 just one failure: Nov 03 15:15:11 | * satisfy_dependencies_for: Cannot satisfy the following dependencies for task-base-extended: Nov 03 15:15:15 | * angstrom-version * Nov 03 15:15:17 | * opkg_install_cmd: Cannot install package task-base-extended. Nov 03 15:15:21 why on earth would angstrom-version not build? Nov 03 15:18:44 oh..there was some talk about a 'midnight build' issue Nov 03 15:19:23 bitbake angstrom-version now Nov 03 15:19:57 ant_work: I'm using the tested_oct_29 tag so the midnight build stuff shouldn't matter Nov 03 15:20:11 there was a thread in the ML Nov 03 15:20:54 if your build has started right before midnight, angsrom-version could fail Nov 03 15:21:09 is a ghost-bug Nov 03 15:21:18 still unclear Nov 03 15:23:09 hmm, task-base-extended failed due to angstrom-version. I built angstrom-version and task-base-extended manually without any trouble Nov 03 15:23:14 trying opie-kdepim-image again Nov 03 15:23:50 iirc angstrom version fails due to date mismatch Nov 03 15:24:06 hmm, and that fails Nov 03 15:24:21 ant_work: oh I see. Nov 03 15:24:24 this isn't hte case here Nov 03 15:24:26 can fail under certain circumstances Nov 03 15:24:38 it happened to me long ago Nov 03 15:24:54 but recently the issue arose again Nov 03 15:25:54 http://pastebin.ca/1980714 Nov 03 15:26:00 that's the full error (9 lines) Nov 03 15:26:21 I'm sure it's making perfect sense to someone more familliar with oe, but to me it looks like it found a problem, corrected it and still failed Nov 03 15:46:10 gone for now, back tonight from home, cya & have fun! Nov 03 15:54:57 tzanger: do you have openembedded/recipes/angstrom/angstrom-version.bb? Nov 03 16:00:00 03Alex Ferguson  07org.openembedded.dev * rb6a3de656e 10openembedded.git/recipes/ttf-fonts/ (ttf-inconsolata/OFL.txt ttf-inconsolata_20100526.bb): Nov 03 16:00:00 ttf-inconsolata: Added recipe for the monospaced Inconsolata font. Nov 03 16:00:00 * Added OFL.txt, which contains the text of the SIL Open Font License Nov 03 16:00:00 as per the license terms. The text in ttf-gentium cannot be reused Nov 03 16:00:00 because it has more specific text. Nov 03 16:00:00 Signed-off-by: Alex Ferguson Nov 03 16:00:01 Signed-off-by: Kristoffer Ericson Nov 03 16:00:10 03Alex Ferguson  07org.openembedded.dev * r51a7e5bed3 10openembedded.git/recipes/echinus/ (6 files in 3 dirs): (log message trimmed) Nov 03 16:00:10 echinus: Add recipes for echinus (a wm) and ourico (a panel for echinus). Nov 03 16:00:10 * Added echinus-jlime.patch, which contains jlime-specific changes Nov 03 16:00:10 to the echinus window manager Nov 03 16:00:11 * Added ourico-jlime.patch, which contains jlime-specific changes Nov 03 16:00:11 to the ourico panel. Nov 03 16:00:12 * Added Makefiles for both echinus and ourico which add two additional Nov 03 16:00:12 03Alex Ferguson  07org.openembedded.dev * rceb9dc2430 10openembedded.git/recipes/linux/ (4 files in 2 dirs): (log message trimmed) Nov 03 16:00:13 linux-jlime-ben-nanonote: Remove linux-jlime-ben-nanonote-2.6.36, remove 2.6.32 Nov 03 16:00:13 * Included linux_logo_clut224.tar.gz which is a full-screen (320x240) logo Nov 03 16:00:14 image and which replaces drivers/video/logo/linux_logo_clut224.ppm Nov 03 16:00:14 (to avoid any source-code changes. Nov 03 16:00:15 * Included modifier-keys.patch, which adds proper support for the Ben Nanonote Nov 03 16:00:15 keyboard. Nov 03 16:00:16 03Alex Ferguson  07org.openembedded.dev * r0e16856e62 10openembedded.git/recipes/xdialog/ (files/fix-duplicates.patch xdialog_2.3.1.bb): (log message trimmed) Nov 03 16:00:17 xdialog: Add recipe for Xdialog, gtk-1.2 equivalent to dialog. Nov 03 16:00:28 dzen2: Add recipe for dzen2, a small messaging/menu program. Nov 03 16:00:28 * Added Makefile, which adds the INCS and LIBS variables due to the Nov 03 16:00:29 way OE handles CFLAGS and LDFLAGS. Nov 03 16:00:29 * Added dzen-extras.tar.gz which adds a launcher script containing Nov 03 16:00:30 several functions (including time display, memory usage, cpu usage Nov 03 16:00:30 and battery status) for use of dzen as a panel application, along Nov 03 16:00:32 03Alex Ferguson  07org.openembedded.dev * r6c8af6f29e 10openembedded.git/recipes/gmu/ (8 files in 4 dirs): (log message trimmed) Nov 03 16:00:32 gmu: Add recipe for gmu-0.7.1, a music player based on SDL. Nov 03 16:00:32 * Added Makefile, which adds an install hook and adds the INCS and Nov 03 16:00:33 LIBS variables, due to the way OE handles CFLAGS and LDFLAGS. Nov 03 16:00:33 * Added unknown.mk, the generic mk file, which replaces the version Nov 03 16:00:34 shipped with gmu and fixes compilation in OE. Nov 03 16:00:34 * Added gmu.desktop, which uses the generic audio-player icon. Nov 03 16:00:35 03Alex Ferguson  07org.openembedded.dev * rc63797cc58 10openembedded.git/recipes/psplash/ (5 files in 3 dirs): (log message trimmed) Nov 03 16:00:35 psplash: Added patches, splash and bar for jlime, modified recipe. Nov 03 16:00:36 * Modified configurability.patch so that it applies to the latest psplash Nov 03 16:00:36 SVN (revision 422). Nov 03 16:00:37 * Added jlime/psplash-poky-img.h and psplash-bar-img.h for jlime. Nov 03 16:00:38 * Added progress-color.patch, which sets the color of the progress bar Nov 03 16:00:38 to black, effectively making it invisible. Nov 03 16:00:50 * Added CCACHE = "", disabling ccache because compilation is broken Nov 03 16:00:50 with ccache installed. Nov 03 16:00:51 Signed-off-by: Alex Ferguson Nov 03 16:00:51 03Alex Ferguson  07org.openembedded.dev * rf8ff1a485e 10openembedded.git/recipes/base-files/base-files/ben-nanonote/ (issue issue.net): Nov 03 16:00:52 base-files: Add issue and issue.net for ben-nanonote. Nov 03 16:00:52 Signed-off-by: Alex Ferguson Nov 03 16:00:53 Signed-off-by: Kristoffer Ericson Nov 03 16:00:53 03Alex Ferguson  07org.openembedded.dev * r47bc7b6287 10openembedded.git/recipes/mplayer/ (files/mipsel-support.patch mplayer_0.0+1.0rc2.bb): Nov 03 16:00:54 mplayer: Added support for mipsel processors for version 1.0rc2. Nov 03 16:00:54 * Added mipsel-support.patch, which adds the correct hooks in configure Nov 03 16:00:55 for compiling mplayer for a mipsel processor. Nov 03 16:00:58 * Modified mplayer-0.0+1.0rc2, adding the above patch to SRC_URI. Nov 03 16:00:58 Signed-off-by: Alex Ferguson Nov 03 16:00:58 Signed-off-by: Kristoffer Ericson Nov 03 16:00:58 03Alex Ferguson  07org.openembedded.dev * r7fe160aa4d 10openembedded.git/conf/distro/jlime-2010.1.conf: (log message trimmed) Nov 03 16:00:58 jlime-2010.1.conf: Remove reduntant defs, add provider for psplash. Nov 03 16:00:58 I removed the PREFERRED_VERSION definitions for eglibc since they Nov 03 16:00:59 are useless (they are derived from sane-toolchain.inc) and added Nov 03 16:00:59 PREFERRED_PROVIDER_virtual/psplash = "psplash" since bitbake Nov 03 16:01:11 Signed-off-by: Kristoffer Ericson Nov 03 16:01:16 03Alex Ferguson  07org.openembedded.dev * r9573c1eecb 10openembedded.git/recipes/jlime/ (5 files in 2 dirs): (log message trimmed) Nov 03 16:01:16 jlime: Add several small jlime-specific recipes. Nov 03 16:01:16 * Added fileselector-1.0, a simple file-selector in C and SDL. Nov 03 16:01:16 * Added fnkey-1.0, which is used to emulate a Mode_Shift + Shift Nov 03 16:01:16 keypress using xdotool on pressing the Fn key. This is mainly for Nov 03 16:01:16 the ben-nanonote, and is hackish, but works. Nov 03 16:01:16 * Added gtk-engine-industrial for gtk-1.2, since we still use gtk-1.2 Nov 03 16:06:35 sorry for spam :) Nov 03 16:11:15 is the sdk generation desribed in the manual still the best way to create an sdk? Nov 03 16:12:52 bitbake meta-toolchain-foo? Nov 03 16:13:56 https://github.com/kergoth/openembedded/commit/72f3fd1 - any thoughts? this one could be slightly controversial Nov 03 16:15:23 Tartarus, I need to add my specific libraries and headers Nov 03 16:16:06 Well, most people set rm_work, but since the addition is destructive, and disk space cheap, I say keep the current behaviour Nov 03 16:16:23 I wouldn't mind but I'm on a laptop with a 160GB disk Nov 03 16:16:47 If I was on a desktop PC with more than 500GB of disk space, I wouldn't mind Nov 03 16:17:24 the destruction is of *intermediate* files Nov 03 16:17:30 I know. Nov 03 16:17:44 i don't really see the point in keeping around files which are only used to debug specific issues Nov 03 16:17:54 kergoth, so after a build, I could turn it off and collect work dirs to debug in? Nov 03 16:18:15 Well, thing is, you don't need you know you need the directories until after you do :) Nov 03 16:18:32 right, so set the var for what you need to debug and rebuild Nov 03 16:18:50 it'll just rebuild the one recipe and retain the dir that time around Nov 03 16:19:04 That might be unfeasible with some kernel stuff, where system.map needs to be retained and so on. Nov 03 16:19:15 hm? Nov 03 16:19:19 that doesn't make any sense. Nov 03 16:19:28 System.map is captured into the pstage package and is deployed into deploy Nov 03 16:19:33 removing workdir is not a problem Nov 03 16:19:50 Ah, I suspect it's an issue with my recipe then. Nov 03 16:19:53 "issue" Nov 03 16:21:44 Crofton|work: with rm_work inherited, or using the new commit, the workdirs are wiped when the recipe is built entirely, you can't magically restore them after the fact without rebuilding the recipe Nov 03 16:22:10 that's the downside, the need to rebuild the recipe with it disabled if you discover a need for the intermediate files of that recipe Nov 03 16:22:58 I like having everythign around Nov 03 16:23:36 most of the poeple who need rm_work, would be better served with better sdk's and otehr dev tools Nov 03 16:24:09 that said, I don't have a strong position on your approach since I can easily erstore the old behavior Nov 03 16:24:20 You will most likely get a few people saying WHERES MY SOURCES? Nov 03 16:24:57 Having rm_work as an option doesn't hurt anyone Nov 03 16:25:02 I personally disagree. That assumes that every user using OE is a developer. That's not the case. Even for developers, the common case is to never need or use those files Nov 03 16:25:28 see part 2 of my comments Nov 03 16:25:28 if you want to jump into the source tree, you should be using devshell :) Nov 03 16:26:02 if you want to assemble images you should be using narcissus :) Nov 03 16:26:13 well, i don't have a strong opinion on it either, it just seems pointless to optimize for rare cases Nov 03 16:27:59 * kergoth ponders Nov 03 16:30:31 can oe generate yaffs2 images? Nov 03 16:31:08 Maybe, if not it's easy to add Nov 03 16:31:15 see the existing examples for jffs2, ext2, etc Nov 03 16:31:32 * LexsZero trying to build custom rootfs for use with existing kernel on chinese android tablet Nov 03 16:31:44 LexsZero: yes, check conf/bitbake.conf Nov 03 16:32:29 LexsZero: and machine/neuros-osd2.conf + include/neuros.inc for usage Nov 03 16:33:11 oh, yep Nov 03 16:33:24 local.conf.sampe comments misinformed me Nov 03 16:33:34 thx Nov 03 16:33:48 Crofton|work: https://github.com/kergoth/openembedded/compare/master...srctree - seem sane to you? i'm playing around with a local collection that uses srctree for everything it builds, heh Nov 03 16:44:17 03Alex Ferguson  07org.openembedded.dev * r70b504ea5a 10openembedded.git/recipes/puzzles/ (10 files in 2 dirs): (log message trimmed) Nov 03 16:44:17 stppc2x: Added recipe for stppc2x, a port of puzzles to SDL. Nov 03 16:44:17 * Added dirs-fix.patch, which changes the default directories Nov 03 16:44:17 for the game data to a systemwide /usr/share/games/stppc2x Nov 03 16:44:17 setting and changes the default font to DejaVuSans-Condensed. Nov 03 16:44:17 * Added enable-fullscreen.patch, which, um, makes the game Nov 03 16:44:18 fullscreen. Nov 03 16:44:23 03Alex Ferguson  07org.openembedded.dev * r5764ede4ad 10openembedded.git/recipes/puzzles/stppc2x/stppc2x.png: Nov 03 16:44:23 stppc2x: Change icon for stppc2x Nov 03 16:44:23 I changed the icon for stppc2x to something more modern-looking Nov 03 16:44:23 from the Faenza iconset. Nov 03 16:44:23 Signed-off-by: Alex Ferguson Nov 03 16:44:24 Signed-off-by: Kristoffer Ericson Nov 03 16:47:12 kergoth: may be NO_RM_WORK="yes" Nov 03 16:47:18 instead of RM_WORK="" Nov 03 16:48:03 or maybe make it RM_WORK = 0/1, or yes/no, or whatever Nov 03 16:48:11 the empty or non-empty is pretty ugly Nov 03 16:48:18 but at all I'll vote for doing this default behaviour Nov 03 16:48:34 I have rm_work inherited always here.. Nov 03 16:48:37 there's a utility function in lib/oe/ for the boolean from string thing, which is used by the apply=yes bits, could use that for the variable Nov 03 16:49:08 kergoth: btw, should we create variable naming policy for such cases? Nov 03 16:49:18 probably Nov 03 16:49:21 e.g. freebsd rc.conf have good examples Nov 03 16:49:23 i did a prototype once of variable typing Nov 03 16:49:28 you'd set a flag Nov 03 16:49:33 FOO[type] = "boolean" Nov 03 16:49:40 and it'd automatically check the value for sanity for that type Nov 03 16:49:48 looking good Nov 03 16:49:55 another advantage of that is you could write a configuration frontend that's smarter Nov 03 16:50:00 it could show a checkbox, or whatever Nov 03 16:50:49 i'll resurrect that Nov 03 16:52:25 well.. my Frankenstein seems almost alive Nov 03 16:54:33 03Philip Balister  07org.openembedded.dev * r500e8f790a 10openembedded.git/recipes/uhd/uhd-firmware_0.1.bb: uhd-firmware : Add recipe to install USRP firmware into images. Nov 03 16:54:34 03Philip Balister  07org.openembedded.dev * r0307e67ec4 10openembedded.git/recipes/uhd/uhd.inc: Nov 03 16:54:34 uhd : Add dependency on uhd-firmware so the USRP1 and USRP2 firmware is Nov 03 16:54:34 installed in an image. Nov 03 16:56:54 while talking of rm_work, do you have any idea why I get this "ERROR: Task do_rm_work_all does not exist for target mytest-image" ? Nov 03 17:01:06 I keep running into some issues with openjade-native (dep of docbook-utils-native from iputils from task-proper-tools) - occasionally (often) it fails on a clean build, a followup build fails in docbook-utils-native. I can't see to figure out what to clean/rebuild to get past those. anyone else encounter this? Nov 03 17:01:28 if I understood docbook better it would probably help - I think something gets improperly staged when the first build fails Nov 03 17:01:49 ok I have my answer ... this is because of multiple inherit image due to require of an image in an image, stupid me Nov 03 17:03:50 Question. Ubuntu 10.04.1 (LTS) as build host, Bitbake 1.10.0 and latest OE recipies on Gumstix Overo. Nov 03 17:03:56 apache2, modphp and php packages do not compile out of the box. Nov 03 17:04:09 php and php-native has even wrong links to sources. Nov 03 17:04:38 Oh mistake, apache2 compiles, but building modphp requires some files from apache2 which are not found Nov 03 17:13:58 php packages even have wrong links to sources. Nov 03 17:15:39 | CCLD libcairo-trace.la Nov 03 17:15:39 | /bin/grep: /usr/lib/libz.la: No such file or directory Nov 03 17:15:48 this sounds familiar .... Nov 03 17:15:55 angstrom 2010 Nov 03 17:36:06 whoops...missed florian Nov 03 17:45:55 mrj10: sorry for the late response Nov 03 17:46:05 mrj10: yes, I definitely have an angstrom-version recipe Nov 03 17:46:11 I can make it with bitbake angstrom-version in fact Nov 03 18:48:22 gm Nov 03 19:06:58 I think something you did broke openembedded for MACHINEs with hyphens in their name Nov 03 19:07:38 I cloned a new openembedded repo, tried building for jlime-2010.1 and ben-nanonote Nov 03 19:07:53 It spits out EOL errors Nov 03 19:08:42 Ugh Nov 03 19:09:03 Not exactly, actually. Nov 03 19:12:19 how to ask oe to not build kernel? Nov 03 19:16:15 Not possible AFAIK -- there are recipes that require at least some of the products of a kernel build IIRC. Nov 03 19:17:10 is linux-headers not enough? Nov 03 19:23:17 hello webkit-efl_svn.bb does not compile any solutions available ? Nov 03 19:26:01 re Nov 03 19:29:10 Hi, I started from scratch and I'm getting errors like this: http://pastebin.com/hZazRh1r Nov 03 19:29:20 I'm building for ben-nanonote and jlime-2010.1 Nov 03 19:29:40 If I change either MACHINE or DISTRO, I stop getting that Nov 03 19:29:48 But both of them produce such issues Nov 03 19:30:27 Using the machine and distro files from an older (1 month) openembedded tree I have which works doesn't change a thing Nov 03 19:30:52 Newer bitbake does the same Nov 03 19:31:18 Has anyone seen anything like this and can help? Nov 03 19:35:22 Hello, i have a problem compiling webkit-efl http://pastebin.com/zAmegie5 Nov 03 19:40:53 B_Lizzard: that long string is looking as directories where to look for patches/files/etc Nov 03 19:41:31 may be something was changed there Nov 03 19:42:41 Hmmm Nov 03 19:42:52 Could it be some kind of clash? Nov 03 19:42:56 I can't see any other way Nov 03 19:43:24 Hmmm Nov 03 19:44:02 It happens with all mipsel targets Nov 03 19:44:23 check bitbake -e on angstrom e.g. and jlime Nov 03 19:44:34 may be you will find something Nov 03 19:44:56 or on minimal even Nov 03 19:46:32 ${FEED_ARCH} is only variable in that string Nov 03 19:48:27 You are right Nov 03 19:48:29 OK Nov 03 19:49:52 FEED_ARCH = "${TARGET_ARCH}" fixes it Nov 03 19:50:04 Not sure if that's correct, I'll see what Angstrom does Nov 03 19:52:45 Anyone, May help me, please? I tried to use "bitbake 'AnyRecipe' " , but I always get this error: "NOTE: :EOL while scanning string literal (FILESPATH, line 1)". I tried a lot things founded at internet, but nothing solved Nov 03 19:53:19 day of broken string literals Nov 03 19:53:50 roliveira, do you get that a lot? Nov 03 19:53:57 Are you building for mipsel Nov 03 19:53:58 > Nov 03 19:57:29 So adding FEED_ARCH = "mipsel" and BASE_PACKAGE_ARCH = "mipsel" in tune-mips32.inc as per all the other tune- files fixes the issue Nov 03 20:03:23 arg Nov 03 20:03:31 * Crofton|work needs a beating Nov 03 20:13:53 * eFfeM slaps Crofton|work with a large trout :P Nov 03 20:14:06 your wish is my command ;-) Nov 03 20:24:28 03Philip Balister  07org.openembedded.dev * rd66b527e43 10openembedded.git/recipes/uhd/uhd-firmware_0.1.bb: uhd-firmware: Add checksums. Nov 03 20:31:14 B-Lizzard: I get that error for all recipes and its dependencies Nov 03 20:31:36 B_Lizzard: I get that error for all recipes and its dependencies Nov 03 20:32:48 roliveira, which MACHINE are you building for? Nov 03 20:33:28 B_Lizzard: The MACHINE is x86 and DISTRO: minimal Nov 03 20:33:48 oestats is really strange.. Nov 03 20:34:19 most of builds are in state 'started' Nov 03 20:35:10 roliveira: Could you pastebin a whole error message? Nov 03 20:35:17 From NOTE: to the next NOTE: Nov 03 20:35:39 B_Lizzard: Of course, just a minute, thanks Nov 03 20:40:36 B_Lizzard: The Pastebin File: http://pastebin.com/UYjBXbS2 Nov 03 20:42:09 ${MACHINE_CLASS} ? ;) Nov 03 20:42:43 where is OEBASE set Nov 03 20:43:28 khem is right, unless that's a real directory it's your mistake Nov 03 20:43:59 I'll take a look, thanks Nov 03 20:54:32 JaMa: what BS mean :) Nov 03 20:55:50 first it should mean bisecting, but I guess that usuall asociation is bull**it so I removed it :) Nov 03 21:03:24 khem: wrt USE/DISTRO_FEATURES discussion, I would say that gentoo pretty much locks all features which are autodetected by autotools Nov 03 21:03:31 03Antonio Ospite  07org.openembedded.dev * r845c38177e 10openembedded.git/recipes/tzdata/tzdata_2010m.bb: Nov 03 21:03:31 tzdata: bump version to 2010o Nov 03 21:03:31 Signed-off-by: Antonio Ospite Nov 03 21:03:31 Acked-by: Khem Raj Nov 03 21:03:32 03Antonio Ospite  07org.openembedded.dev * ra8c1ef6eab 10openembedded.git/recipes/tzcode/tzcode-native_2010m.bb: Nov 03 21:03:32 tzcode-native: bump version to 2010n Nov 03 21:03:32 Signed-off-by: Antonio Ospite Nov 03 21:03:32 Acked-by: Khem Raj Nov 03 21:04:00 khem: that's mostly because for every sane option there is already USE-flag turning it on/off Nov 03 21:04:32 JaMa: can you say that with conviction Nov 03 21:04:38 I dont like *mostly* Nov 03 21:04:48 its relative we do that too in OE Nov 03 21:05:28 JaMa: btw. how is gcc 4.5 behaving now Nov 03 21:05:32 do u have more pain points Nov 03 21:05:58 I still am interested in measuring perf on some real world apps using gcc 4.5 from OE and some older gcc Nov 03 21:06:03 see how far we gain Nov 03 21:06:13 on ARMv7+NEON Nov 03 21:06:24 I'm trying to solve misterious stack corruption, but cannot say if it's some nasty gcc bug or something else Nov 03 21:06:37 hmm, wonder if we should patch autoreconf to do the glib-gettextize, intltoolize bits we do in autotools.bbclass Nov 03 21:06:55 but last state was that I've tried to downgrade all possibly related libs to versions where it was supposed to work and still it segfaulted Nov 03 21:07:02 though, i do dislike relying on patched versions of things. hrmph Nov 03 21:08:15 khem: I can try to check if gentoo has some QA check for autodetected options, but every ebuild I could think of has plenty of enable/disable with/without switches Nov 03 21:09:56 http://pastebin.ca/1980714 Nov 03 21:10:16 any idea why bitbake is erroring out with that? I'm trying to build opie-kdepim-image Nov 03 21:10:40 I can build task-base-extended and I can build angstrom-version separately Nov 03 21:10:49 but when trying to build opie-kdepim-image it errors out Nov 03 21:11:22 khem: if you have minute, can you check http://pastebin.ca/1981006 if you see something suspicious arround 0x40551880? there is just preprocessor macro (with few shifts and adds) and then PC jumps to strange address (outside /proc/pid/maps and stack) Nov 03 21:11:23 of course you can. the error is due to a file from one package being overwritten by a file from another package -- its by definition a problem with the rootfs population, and therefore the image, so yes, you'll be able to *build* the other things fine, its the *install* of them thats the problem Nov 03 21:12:06 anyone made apache2, modphp and php compile on Gumstix? (Angstrom) Nov 03 21:12:13 kergoth: hmm, it doesn't show that that's an error though, just a warning Nov 03 21:12:31 or rather it SOUNDS like a warning because it states what it did because of it Nov 03 21:12:33 yes, i see that, no idea why opkg would behave that way Nov 03 21:14:25 hmm, I'm going to try modifying one of those recipes to not create the problem Nov 03 21:17:04 Hmm... I updated apache2-native to copy 'build' directory into sysroot, in apache2 and apache2-native added --enable-so. Now apxs when compiling mod_php complains that can not find shared libexpat. There is only expat.bb, but no native version. Should I now provide expat-native.bb? Nov 03 21:24:01 btw, which version of bitbite is best to use? Nov 03 21:24:21 davidlt: if you are brave use git master Nov 03 21:24:33 What do you mean by brave? Nov 03 21:24:47 Right now I am using 1.10.0 is that ok? Nov 03 21:24:50 because sometimes infrequently there might be some breaks Nov 03 21:25:39 There is a 1.10.1, but yes, 1.10.0 is fine Nov 03 21:25:55 Is it only me or no one cares about apache2 and php? Nov 03 21:26:22 No, it's fair to say apache2/php aren't being used by many people right now Nov 03 21:26:45 JaMa: whats the value of r3 at that point ? Nov 03 21:26:50 davidlt: got it working 1.5 years ago on armv5, didn't try since Nov 03 21:26:54 JaMa: the code looks ok Nov 03 21:27:04 JaMa: somehow r3 does not carry right address Nov 03 21:27:11 is it on thumb ? Nov 03 21:27:32 I trying to make it work on 3 different setups (build host, bitbake version, recipies). It just doesn't work. No matter how much I try to make it. Nov 03 21:27:45 davidlt: what doesn't work ? Nov 03 21:28:12 a) php 5.2.13 has wrong source links, that easy fixed Nov 03 21:28:31 khem: it's on om-gta02 with thumb enabled (didn't check if it's currently in thumb mode and what is in r3), mmt will check later Nov 03 21:28:50 b) php 5.2.13 does not find libcrypto, but those are in armv7 and i686 lib dirs. Nov 03 21:29:04 JaMa: hmmm r3 is loaded couple of instrs above Nov 03 21:29:29 Removing all options that require SSL, you run into problems with MySQL, now it can't find MySQL libs and that continues... Nov 03 21:29:36 which is a ptr on stack Nov 03 21:30:03 JaMa: set watchpoints Nov 03 21:30:08 that will help you Nov 03 21:30:09 kergoth: shouldn't bitbake -c install task-base-extended then also fail? Nov 03 21:30:21 c) modphp does not compile, first apxs is looking for /usr/build folder which is not staged. (fixed). Now apxs can not find shared libexpat Nov 03 21:30:32 tzanger: there is some duplication of files provided Nov 03 21:30:43 and I think we dont tell opkg to force install Nov 03 21:30:48 thats why it fails Nov 03 21:31:09 install is task which is different here I think Nov 03 21:31:09 khem: I understand, I'm just trying to find the files it's complaining about Nov 03 21:31:15 hmm Nov 03 21:31:35 where would I begin looking to tell opkg to ignore errors for this particular recipe? Nov 03 21:31:35 davidlt: you can submit a bug report with details (and possibly send patches for what you already have fixed) Nov 03 21:31:38 /mnt/hd/overo-oe/tmp/rootfs/opie-kdepim-image/etc/device_table Nov 03 21:32:12 Okay, I am gonna do this. Nov 03 21:32:12 khem: ok I'll learn about watchpoints first :) Nov 03 21:32:34 JaMa: what is underlying arch for om-gta02 Nov 03 21:32:56 JaMa: I think it could be using thumb addresses and somewhere it forgot to flip the bit Nov 03 21:32:57 arm902t (armv4t) Nov 03 21:32:59 or something Nov 03 21:33:07 920t sorry Nov 03 21:33:10 tzanger: no. its installation of hte ipk into a filesystem that's the issue, not its do_install Nov 03 21:33:23 yeah in sane-toolchain for armv4t we ask for thumb (iow trouble) Nov 03 21:34:09 kergoth: if you have a moment can you look into the target override patch ? Nov 03 21:34:17 k Nov 03 21:34:29 kergoth / khem: yes I understand, that's why I'm trying to alter the package to not have that file in one of the packages Nov 03 21:34:36 I think whats happening is that some variables get evaluated prematurely Nov 03 21:34:57 and then later when native.bbclass contents kick in they dont change Nov 03 21:35:09 khem: that's the destination yes, but I need to edit the appropriate recipe or rather source to build a package without the file Nov 03 21:35:11 khem: do oyu have a specific example of a failure i can use to test? Nov 03 21:35:16 and CONFIGUREOPTS get --prefix etc set to /usr/ bla bla Nov 03 21:35:55 kergoth: apply the patch then do bitbake -e jamvm-native >& xx and then in xx look for CONFIGUREOPTS Nov 03 21:36:11 kergoth: try that with and without patch Nov 03 21:36:35 you will see that without patch --prefix will point into native sysroot which is correct Nov 03 21:36:51 but with patch it will point to /usr/include which is correct for target recipes Nov 03 21:36:58 but this one is a native recipe Nov 03 21:37:24 tzanger: you need to grep around Nov 03 21:37:35 tzanger: or dump the ipks and go from there Nov 03 21:38:00 hi guys Nov 03 21:38:06 there is anyone that can help me? Nov 03 21:38:22 lynx_deb: what happened parachute is not opening ? Nov 03 21:38:34 it looks like the device_table stuff is done in initscripts-1.0-r124 Nov 03 21:38:40 ahahha no, that problem can be more easy to solve Nov 03 21:38:50 i cannot compile with OE Nov 03 21:38:57 wxwidgets Nov 03 21:39:04 using gcc 4.2.1 and glibc 2.5 Nov 03 21:39:30 JaMa: hello, webkit-efl does not compile, http://pastebin.com/zAmegie5 is the problem known ? Nov 03 21:41:07 lynx_deb: ok whats your distro/machine Nov 03 21:41:13 and the error Nov 03 21:41:17 paste it somewhere Nov 03 21:41:18 i'm using debian lenny 5.0 Nov 03 21:41:29 lynx_deb: I meant OE distro/machine Nov 03 21:41:39 i.e. your target distro/machine Nov 03 21:41:47 angstrom.2008 sheevaplug Nov 03 21:42:10 nschle85: no, haven't seen this one on my host or shr buildhost.. Nov 03 21:42:41 http://pastebin.com/DafRtc6Y Nov 03 21:42:44 this is the error.. Nov 03 21:43:03 i've found this problem on google Nov 03 21:43:04 JaMa: i cannot read the logfile its very confuse... Nov 03 21:43:16 seems that on default libc-header 2.6.31 there is not procinfo.h file Nov 03 21:43:47 so i've told to bitbake to get libc-header 2.6.32 where the file is present Nov 03 21:43:58 it compile glibc.25 native, but stop on intermediate Nov 03 21:44:11 also seems that in tmp/work/sysroot/usr/include Nov 03 21:44:14 khem: yeah I'm find tmp -type f -name device_table and it's turned up only the one: initscripts-1.0-r124 Nov 03 21:44:29 there aren't all *.h file included in the header Nov 03 21:45:01 tzanger: does it make into more than one installs somehow Nov 03 21:45:18 lynx_deb: thats a known issue in glibc 2.5 any reason you cant use newer ones Nov 03 21:45:25 khem: that's what I am not sure how to figure out Nov 03 21:45:41 i need to compile a program and put on my nas Nov 03 21:45:45 that use glibc 2.5.. Nov 03 21:47:13 khem: is there any way to ask bitbake what recipes depend on other recipes? Nov 03 21:47:43 nschle85: try to read a bit higher if this was real cause for "collect2: ld returned 1 exit status" Nov 03 21:48:12 initscripts is not mentioned explicitly by any task-base-extended nor opie-kdepim-image, but there are some others Nov 03 21:49:48 bitbake -g Nov 03 21:49:52 tzanger: bitbake -g Nov 03 21:49:57 ah Nov 03 21:50:02 so khem, base on your response i cannot compile wxwidgets (and other deps) based on glibc 2.5? Nov 03 21:50:09 lynx_deb: got it Nov 03 21:50:17 lynx_deb: well you can Nov 03 21:50:39 lynx_deb: we need to fix the old recipe for glibc 2.5 Nov 03 21:51:36 what we need to do? Nov 03 21:51:41 lynx_deb: from this big patch here http://sourceware.org/ml/crossgcc/2006-12/msg00076.html Nov 03 21:51:55 you need to extract glibc-arm-no-procinfo_h.patch Nov 03 21:53:16 i've already use that patch and seems to work, but don't know if this create problem with other deps. Nov 03 21:53:25 because after glibc compile Nov 03 21:53:34 when bitbake try to compile zlib for arm Nov 03 21:53:47 give me error like: c compiler cannot create executable Nov 03 21:54:24 nice, what's the util that graphs the output of that? digraph sounds very familliar Nov 03 21:54:57 graphviz Nov 03 21:54:59 lynx_deb: that means some headers got unstaged Nov 03 21:55:49 lynx_deb: if you have that patch applied then the error you reported should not happen Nov 03 21:55:56 yes i know Nov 03 21:56:02 with that patch i can compile glibc Nov 03 21:56:17 but as i've told before, zlib doesn't compile Nov 03 21:56:25 may be was a problem Nov 03 21:56:29 with multiple thread.. Nov 03 21:56:33 i'll try again Nov 03 21:57:21 lynx_deb: ok do this Nov 03 21:57:58 bitbake -c clean glibc-initial glibc gcc-cross-initial gcc-cross-intermediate gcc-cross; bitbake zlib Nov 03 21:58:16 it will take time but should get it right in the end Nov 03 21:58:28 * khem has typed that command nth time into this room now Nov 03 21:58:38 perfect! Nov 03 21:58:51 i've done some mistake Nov 03 21:58:57 trying to copy header manualy Nov 03 21:59:06 can i clean only libc header and reinstall? Nov 03 21:59:17 so i apply the patch and try again Nov 03 22:05:51 lynx_deb: apply the patch and then do the above Nov 03 22:06:11 i need to make patch file Nov 03 22:06:15 and add to bb Nov 03 22:11:04 khem Nov 03 22:11:25 i need to apply patch only to glibc-2.5 or also on glib-intermediate? Nov 03 22:11:55 lynx_deb: just apply it to glibc should be enough Nov 03 22:11:59 ok Nov 03 22:12:13 it should get included automatically into initial recipe Nov 03 22:13:12 JaMa: "relocation truncated to fit: R_ARM_THM_CALL against symbol" i think -mthumb-interwork option may be the problem but iam not sure Nov 03 22:13:31 Jay7, its hard override in bitbake.conf Nov 03 22:13:47 oops sorry - old msg Nov 03 22:14:14 tharvey: this is answer but I've forgot question already :) Nov 03 22:14:23 seems about TMPDIR Nov 03 22:14:36 heh... ya, sorry didn't realize I was scrolled so far back Nov 03 22:14:44 it happens :) Nov 03 22:14:51 was just excited to be able to answer a question rather than ask one Nov 03 22:16:36 nschle85: neither am I.. sorry cannot help you with it right now (already doing 3 things in parallel..) Nov 03 22:16:55 well.. seems I'm ready to publish my testbuilder script :) Nov 03 22:16:56 JaMa: ok Nov 03 22:17:30 Jay7, did you figure out why bitbake returns non zero error code? Nov 03 22:17:43 tharvey: no, but I've some statistics Nov 03 22:18:03 I still maintain that it occurs when a non-fatal QA Issue is found Nov 03 22:18:06 it will return 0 if you will run it over already built tmpdir Nov 03 22:18:28 and seems oestats-client is somehow bound with this Nov 03 22:18:54 on clean build it always return 1 Nov 03 22:19:02 interesting - makes me want to change my script to look for deployed file and not look at retcodes Nov 03 22:19:20 I bet if you look at your clean build logs you will find some 'QA Issue's Nov 03 22:19:30 but if you will run this again on same tmpdir it will return 0 Nov 03 22:19:52 tharvey: It is very possible :) Nov 03 22:19:54 no QA Issues snd time around Nov 03 22:19:58 true Nov 03 22:20:30 http://tinderbox.openembedded.net/public/logs/qa/904411.txt for example Nov 03 22:20:37 so I'm just reporting this code but doing nothing with this Nov 03 22:20:55 and cups: http://tinderbox.openembedded.net/public/logs/qa/904923.txt Nov 03 22:21:35 put the following in your local.conf: INSANE_SKIP_gcc = True and INSANE_SKIP_cups = True Nov 03 22:21:42 bet you'll find retcode 0 on clean tmpdir Nov 03 22:24:35 03Eric Bénard  07org.openembedded.dev * r65e847e20d 10openembedded.git/recipes/meta/meta-toolchain-qte.bb: Nov 03 22:24:35 meta-toolchain-qte: fill OE_QMAKE_AR Nov 03 22:24:35 this allows the sdk to be used to compile libraries, else, we get : Nov 03 22:24:35 make[1]: cqs : command not found Nov 03 22:24:35 Signed-off-by: Eric Bénard Nov 03 22:24:36 Acked-by: Koen Kooi Nov 03 22:24:39 03Eric Bénard  07org.openembedded.dev * r4cf0f23354 10openembedded.git/recipes/ (4 files in 3 dirs): (log message trimmed) Nov 03 22:24:39 qt4e: add 2 images Nov 03 22:24:39 - qt4e-base-image : contains qt4 embedded libraries and can be used Nov 03 22:24:39 as a base rootfs when developping using the sdk Nov 03 22:24:39 - qt4e-demo-image : contains qt-base-image + qt-demo + a script to Nov 03 22:24:39 launch qt-demo at boot, this image can be used to quickly setup Nov 03 22:24:39 a demonstration of qt capabilities on a machine using qtdemo Nov 03 22:25:29 tharvey: hmmm package_qa_make_fatal_error returns non zero if error types are not 0, 5, 7 wich is negated again in package_qa_handle_error Nov 03 22:25:48 so I think there is a problem here Nov 03 22:26:06 didn't notice the double negation - wouldn't that make bb.fatal get called and halt building? Nov 03 22:26:30 yeah Nov 03 22:26:38 double negation is trouble in any case Nov 03 22:26:49 it winds my logical mind Nov 03 22:26:58 true... Nov 03 22:27:13 perhaps the intent was to be fatal on 0,5,7? Nov 03 22:27:25 tharvey: could be Nov 03 22:27:41 tharvey: so which error is the one being non fatal Nov 03 22:27:49 reporting non zero ret code Nov 03 22:28:01 for gcc 0 - non dev contains .so Nov 03 22:28:17 for cups - 7 - the desktop file is not valid Nov 03 22:28:24 ok thats error 0 Nov 03 22:28:26 khem Nov 03 22:28:34 can i use -j4 (i've a quad core) Nov 03 22:28:38 to compile gcc and glibc? Nov 03 22:28:42 right... woudln't think non dev contains .so should be 'fatal'? Nov 03 22:28:47 so I never questioned it Nov 03 22:29:19 lynx_deb: yes you can set PARALLEL_MAKE = "-j8" Nov 03 22:29:23 in your local.conf Nov 03 22:29:34 hope i doesn't make problems.. Nov 03 22:29:35 and it will heatup your room fast Nov 03 22:29:41 no it wont Nov 03 22:29:41 lynx_deb, if a recipe can't handle parallel make it will disable it Nov 03 22:29:41 openwrt Nov 03 22:29:48 ah ok Nov 03 22:29:50 perfect Nov 03 22:30:19 khem, curious why do you say 8 on quad core? wouldn't you say 4? Nov 03 22:30:44 usually I have seen n*2 loaded thre system optimally Nov 03 22:30:51 in my experience Nov 03 22:31:16 ok, started clean and cross fingers.. Nov 03 22:31:22 so for my Q9400 Core2 Quad CPU you would also use 8? Nov 03 22:31:26 guys, can you tell if this string is right? Nov 03 22:31:28 PREFERRED_VERSION_wxwidgets ?= "2.8.10" Nov 03 22:31:46 yes but its conditional Nov 03 22:31:55 so it will only kick in if someone else did not set it Nov 03 22:32:14 there is a documentation on the preferred_version variable? Nov 03 22:32:16 if you want to thump your authority you need PREFERRED_VERSION_wxwidgets_local = "2.8.10" in local.conf Nov 03 22:32:23 the only link i've found Nov 03 22:32:26 is on openmoko Nov 03 22:32:28 wiki Nov 03 22:32:42 well most of OE is tribal knowledge Nov 03 22:33:06 tharvey: hmmm so it says error is non fatal right Nov 03 22:33:16 but returns non zero ? Nov 03 22:33:32 can someone checkout sources.openembedded.org to see if they can get things out of the mirror? Nov 03 22:33:51 khem, right, if I bitbake meta-toolchain currently which shows that QA Issue 0 for gcc, bitbake returns a '1' Nov 03 22:33:57 but does not stop building... Nov 03 22:35:00 hmmm... how do I override a do_install_prepend thats coming from an included recipe? Nov 03 22:35:53 ok.. started again.. Nov 03 22:35:57 thank you guys for the help Nov 03 22:36:04 i hope to solve this issue.. Nov 03 22:36:10 or i come back here :D Nov 03 22:36:33 tharvey: git show f0ade73c Nov 03 22:37:00 I think its not correct return not package_qa_make_fatal_error(error_class, name, path, d) Nov 03 22:37:45 tharvey: ok try to change that line in insane.bbclass Nov 03 22:37:50 to Nov 03 22:37:57 return package_qa_make_fatal_error(error_class, name, path, d) Nov 03 22:44:32 http://wiki.openembedded.net/index.php/Testing:TestBuilder Nov 03 22:45:02 that is my suppa-screept Nov 03 22:47:36 wonderful - I can't recreate the non zero retcode issue now - Jay7 you must be right about tmpdir having something to do with it Nov 03 22:48:02 even if I clean and rebuild cups, QA Issue and all I get retcode 0, but if I clean tmpdir according to all my previous builds I will get retcode=1 Nov 03 22:48:16 yeah.. same thing here Nov 03 22:48:30 tharvey: try what I suggested Nov 03 22:48:56 do u still get the QA error ? Nov 03 22:48:58 still working on that... bitbake is VERY slow to recache when conf has changed Nov 03 22:48:59 but no message Nov 03 22:49:08 I mean ret = 0 Nov 03 22:49:23 tharvey: yeah true Nov 03 22:49:30 well even without your change like I said just a clean/build of cups produced QA issue but 0 retcode Nov 03 22:49:59 Jay7, this is meant for Tinderbox? Nov 03 22:50:29 ka6sox-work: did you mean that script? Nov 03 22:50:31 tharvey: Jay7 : is'nt that because of some parsing notes ? From memory when I had this problem with hudson this was because of some NOTE concernign preferred_provider or something like this Nov 03 22:50:50 I'm using it to do builds of testing-next Nov 03 22:51:10 but if you mean wrong bb retcode - this is bound somehow Nov 03 22:51:28 okay...got it. Nov 03 22:51:31 ericben, ah... yes you may be right Nov 03 22:52:17 are those part of insane.bbclass as well perhaps - I know that if I INSANE_SKIP_gcc I don't run into it Nov 03 22:52:39 INSANE_SKIP will avoid doing those checks Nov 03 22:52:43 for a given recipe Nov 03 22:52:50 so u curcumvent it Nov 03 22:53:08 never been clear when those 'multiple provider' checks are actually done Nov 03 22:53:19 err circumvent Nov 03 22:53:19 doesn't seem to happen on a clean/build of a package Nov 03 22:53:58 well multiple provider happens when a package is emitted by two different recipes Nov 03 22:54:16 sometimes they are alternatives to each other Nov 03 22:54:29 sometimes its just that one recipe is not in sync with the other Nov 03 22:54:47 bitbake -g helps in such cases Nov 03 22:59:32 I don't see 'multiple providers' coming from insane.bbclass - seems to come from tinderclient.bbclass which I would think would be from inherit oe-stats Nov 03 23:00:45 there is the 'ERROR: Multiple .bb files are due to be built which each provide' and the 'multiple providers are available for' - perhaps one of these is responsible for non fatal but retcode=1 Nov 03 23:01:28 khem, removing the 'not' - I didn't see any change in my clean/build of cups - I still see the QA Issue and still get the retcode=0 Nov 03 23:02:11 shoot... hold on - would help if I actually saved the file Nov 03 23:02:58 * tharvey is sick of Out of date cache found, rebuilding... Nov 03 23:04:48 tharvey: good that you dont do toochain stuff :) Nov 03 23:04:58 I have to face it all the time Nov 03 23:05:27 * Jay7 have looked over this a lot of time when writing testbuilder.. Nov 03 23:05:31 thats what I was thinking psycho sped up, but alas I'm on x64 and can't test that Nov 03 23:06:23 perhaps that push to get rid of obsolete/unmaintained recipes would make things faster :) Nov 03 23:07:02 hm.. can we mask obsolete dir (if not already)? Nov 03 23:08:04 its already masked Nov 03 23:08:08 ah, ok Nov 03 23:08:36 we should now look into dormantly unused recipes Nov 03 23:08:51 and move them to may be unused or something like that Nov 03 23:09:16 which then can be exluded as well Nov 03 23:11:33 khem, ok, removing that 'not' the QA issue with cups produces a fatal error Nov 03 23:12:00 so I think the 'not' logic is correct and should 'not' be changed ;P - I think the bitbake retcode comes from elswhere Nov 03 23:12:38 Jay7, so your still fighting this retcode issue right? I see your script reports it Nov 03 23:12:51 tharvey: I'm not fighting :) Nov 03 23:12:59 I'm just doing nothing with this Nov 03 23:13:14 well... its certainly not clear to me how to determine if a build was a 'success' Nov 03 23:13:19 * tharvey wonders what oe-stats considers Nov 03 23:13:24 look at oestats :) Nov 03 23:13:30 the only way Nov 03 23:14:12 khem: is there any good way to get bitbake output into file? Nov 03 23:14:21 I fear that scrolling slash :) Nov 03 23:14:34 I've tried to run bb under script and that was fun Nov 03 23:15:07 Jay7, bitbake foo 2>&1 | tee build.log Nov 03 23:15:39 tharvey: recipes parsing part is looking very fun :) Nov 03 23:16:21 no just >& bitbake.log Nov 03 23:16:23 oestats considers success = (e.getFailures() = 0) Nov 03 23:16:48 ah.. seems it disable output when running in pipe Nov 03 23:16:55 tharvey: well if you see images in deploy then its a success Nov 03 23:16:59 that was screen Nov 03 23:17:11 so prod into deploy Nov 03 23:17:14 thats what I do Nov 03 23:17:14 s/screen/script issue/ Nov 03 23:17:34 khem, ya - I'm thinking of changing my buildscript to look for deployed images, but then its more complex as I need to track deployed files along with machien/recipe combo that builds them Nov 03 23:17:36 sorry then :) Nov 03 23:17:51 I should release 0.2 ;) Nov 03 23:17:59 just seems silly... bitbake seems broken with regards to retcode Nov 03 23:18:42 tharvey: ok try to pin point the issue we might fix it Nov 03 23:18:55 heh... thats what were doing... Nov 03 23:19:43 would also like to get gcc/cups QA issues fixed, but the issues are beyond my skillset Nov 03 23:19:48 hmm those .dot files don't parse with any graphviz I have Nov 03 23:20:32 there are a lots of lines in them - I was able to render them but it was useless Nov 03 23:20:38 tzanger: I remember some commit about that Nov 03 23:20:47 tzanger: do you have latest oe maser? Nov 03 23:20:54 the deps for hello-world makes for an interesting example of how complex a build system really is Nov 03 23:21:04 Jay7: I'm at the oct29 tested tag Nov 03 23:21:51 Jay7: I don't really care too much, I'm ust trying to figure out why opie-kdepim-image fails and how to get around it. khem and kergoth have been helpful in pointing it out it's because two packages are installing etc/device_table but I can't figure out which two are doing it Nov 03 23:22:33 tzanger: yes you need to patch bitbake with http://patchwork.openembedded.org/patch/3455/ Nov 03 23:23:03 just sent updated patch for bitbake mater Nov 03 23:23:12 master Nov 03 23:23:20 ericben: there are so many patches from you lately on patchwork Nov 03 23:23:26 can you update the list Nov 03 23:23:35 khem: is bitbake $image | tee $log enough? Nov 03 23:23:38 and keep the ones that are really pending Nov 03 23:23:46 Jay7: should be Nov 03 23:23:50 ok Nov 03 23:24:03 ericben: so I should just pull from the bitbake repo (I'm on master) Nov 03 23:24:12 do bitbake output errors on stderr? Nov 03 23:24:23 wow rebuilding cache takes 5m15secs on my quad core w/ SSD Nov 03 23:24:40 tzanger: patch is on oe's ml Nov 03 23:26:53 Jay7, tee is enough Nov 03 23:27:08 I'm trying to save only stderr Nov 03 23:27:22 it should be easy but I can't remember.. Nov 03 23:27:37 then grep for 'Missing inherit gettext?' ;) Nov 03 23:27:43 yep :) Nov 03 23:27:57 and for 'cp: cannot stat *' Nov 03 23:28:00 :/ Nov 03 23:28:00 nah.. Nov 03 23:28:08 glibc compiled.. but same error on zlib Nov 03 23:28:33 khem.. http://pastebin.com/rtFGDteS Nov 03 23:29:41 ah, a new recipes in the legacy-staging list: stat_3.3.bb Nov 03 23:29:45 hm Nov 03 23:30:32 ok, needs native treatment too Nov 03 23:30:43 ericben: you forgot to send the patch to bitbake-devel list ? Nov 03 23:30:44 hmm that patch doesn't seem to apply Nov 03 23:31:16 khem: oops, I'm not subscibed to this one so I forgot yes Nov 03 23:32:00 tzanger: tzanger: it shoudl apply at least on top of 75ea005ac8fc05b2b3afca803d77a6b5f558efee in bitbake Nov 03 23:32:45 lynx_deb: ok can you paste config.log somewhere ? Nov 03 23:32:48 I'm only one commit ahead of that (64ade) Nov 03 23:33:04 what are the errors ? Nov 03 23:33:10 lynx_deb: thats the config.log of zlib Nov 03 23:33:19 unexpected end of file Nov 03 23:33:33 how did you extract the patch from the mail ? Nov 03 23:33:45 wget on the raw url Nov 03 23:33:54 then fixed the end of line breakage so it looks like a proper patch Nov 03 23:34:07 it's 15 lines total now Nov 03 23:34:35 khem: http://pastebin.com/3B4sNUXb config.log Nov 03 23:35:42 lynx_deb: ok got it Nov 03 23:36:12 ericben: http://pastebin.ca/1981115 Nov 03 23:36:24 lynx_deb: now edit zlib.inc Nov 03 23:36:27 looks like a good patch but without any trailing context, which is maybe what patch is complaining about Nov 03 23:36:28 in the recipes/zlib Nov 03 23:36:57 tzanger: http://pastebin.com/tzruCUfa , you are mising 1 line Nov 03 23:37:11 yes opened Nov 03 23:37:13 lynx_deb: see the line CFLAGS_EXTRA_append_arm = " -fforward-propagate" ? comment that out Nov 03 23:37:34 ahh, that wasn't in the email Nov 03 23:37:40 yes Nov 03 23:37:46 need to comment all CFLAGS? Nov 03 23:37:55 just that line Nov 03 23:37:58 tzanger: it was, that's the same file ;) Nov 03 23:38:07 change it to # CFLAGS_EXTRA_append_arm = " -fforward-propagate" Nov 03 23:38:18 ok done Nov 03 23:38:24 try again to compile.. Nov 03 23:38:32 now save it and bitbake -c clean zlib;bitbake zlib Nov 03 23:38:57 ericben: http://patchwork.openembedded.org/patch/3455/ Nov 03 23:38:59 I don't see it there Nov 03 23:39:10 in fact it's a different patch altogether Nov 03 23:39:21 line 435 not 387 Nov 03 23:40:09 tzanger: ok I understand : you took the patch khem posted the link Nov 03 23:40:13 yep Nov 03 23:40:20 I was talking of the one I sent which is ok for master Nov 03 23:40:45 original patch from Matt Johnson was from bitbake 1.10 I thjink Nov 03 23:40:54 think Nov 03 23:41:17 and it was broken Nov 03 23:41:31 it wont apply cleanly may be he uses a crappy MUA Nov 03 23:41:58 so I updated it to master and it fix the problem in the dotfile Nov 03 23:42:14 good night ! Nov 03 23:43:26 ericben|away: your patch is better Nov 03 23:43:32 it applies cleanly Nov 03 23:43:56 khem, your trick seems good, it past configure now Nov 03 23:45:10 ericben|away: thanks, giving it a shot now Nov 03 23:45:13 now i think i'm going to bed.. and leave to compile wxwidgets, hope it doesn't stop this time :D Nov 03 23:45:40 khem, thank you again! Nov 03 23:47:53 03Martin Jansa  07master * rb339eafdee 10openembedded.git/recipes/e17/e-wm_svn.bb: Nov 03 23:47:53 e-wm: temporary bump SRCREV here, fix big icons, labels without ellipsis Nov 03 23:47:53 Signed-off-by: Martin Jansa Nov 03 23:50:08 good night guys Nov 03 23:53:33 should machine override work with IMAGE_CMD variables? ie IMAGE_CMD_jffs2_om-gta01 ? Nov 03 23:54:16 03Andrea Adami  07org.openembedded.dev * rb3246d9606 10openembedded.git/recipes/stat/ (stat-native_3.3.bb stat_3.3.bb): stat: merge target and native recipe, remov elegacy staging. Bump PR. Nov 03 23:54:31 JaMa: should Nov 03 23:54:42 ah time to drive home Nov 03 23:54:44 ttyl Nov 03 23:55:42 JaMa, I rebuilt stat and stat-native here before committing, still pls give a check ;) Nov 03 23:55:44 there is some custom code handling OVERRIDES in image.bbclass IMAGE_CMD Nov 03 23:56:16 and overrides seems to be ignored (also in -e are shown as different variable) Nov 03 23:56:29 I can just push those to machine config but still curious :) Nov 03 23:57:34 JaMa: another call for help: remove legacy staging from xorg-proto/xextproto-70-includes_7.0.5.bb. You are our X guy now ;) Nov 03 23:58:04 ah sorry it probably does work Nov 03 23:58:38 ant_: I think I've already said that it's too old for my tests and interest :) Nov 03 23:58:56 I don't have kdrive on any machine :) Nov 03 23:58:59 ok, I have no idea about that headers Nov 03 23:59:08 I'll do a research sooner or later Nov 04 00:01:05 * JaMa going to bed, gnite Nov 04 00:01:14 gn Nov 04 00:03:25 03Martin Jansa  07master * r2f09d32fe2 10openembedded.git/recipes/images/shr-image.inc: Nov 04 00:03:25 shr-image: remove /boot from jffs2/ubi/ubifs images only on om-gta01 Nov 04 00:03:25 * om-gta02 has usually enough space on NAND Nov 04 00:03:25 * qi-bootmenu can use those kernels and supports only om-gta02 now Nov 04 00:03:25 * also make sure -boot is empty before moving /boot there Nov 04 00:03:25 * move /boot back even when last command mkfs.ubifs/ubinize/mkfs.jffs2 fails (no && here) Nov 04 00:03:26 Signed-off-by: Martin Jansa Nov 04 00:15:19 03Andrea Adami  07org.openembedded.dev * rb6756a1add 10openembedded.git/recipes/xorg-proto/xextproto-70-includes_7.0.5.bb: xextproto-70-includes: remove superfluous do_stage. Bump PR. Nov 04 00:24:29 good night Nov 04 00:39:53 03Andrea Adami  07org.openembedded.dev * ra33cf73683 10openembedded.git/recipes/xorg-proto/xextproto-70-includes_7.0.5.bb: xextproto-70-includes: remove obsolete commented do_stage code. Nov 04 00:40:04 03Andrea Adami  07org.openembedded.dev * rfebcae3b7b 10openembedded.git/recipes/libqtaux/ (4 files): libqtaux: clean do_compile task. Add INC_PR and bump it to "r1". **** ENDING LOGGING AT Thu Nov 04 02:59:57 2010