**** BEGIN LOGGING AT Tue Apr 05 02:59:58 2011 Apr 05 06:19:52 hello, when I compile 2.6.38.2 kernel against S3C64XX ARM board, it throws : drivers/input/touchscreen/eeti_ts.c:65:2: error: implicit declaration of function ‘irq_to_gpio’, does anyone know this ? Apr 05 07:17:47 how can I add some action in the receipt to make something after do_rootfs? Apr 05 07:18:18 bitbake says, "ERROR: No such task 'do_rootfs'" while executing receipt Apr 05 07:19:06 grep ROOTFS recipe/images Apr 05 07:19:06 isn't postinst good enough? Apr 05 07:32:18 postinst, postinst seems too receipt-specific. /etc/iittab is created at other receipt, I need to add some string to inittab from my own receipt without touching main image receipt Apr 05 07:33:19 bad idea, because next sysvinit upgrade will overwrite your change Apr 05 07:40:21 good morning Apr 05 07:55:36 JaMa, can I use ROOTFS_POSTPROCESS_COMMAND += "..." in my programm receipt? Apr 05 07:56:05 same problem Apr 05 07:56:30 better to provide/alter inittab in sysvinit recipe Apr 05 07:56:38 s/provide/provide own/ Apr 05 08:00:48 JaMa, but I need to provide my own inittab in my global image receipt? seems to me, I found a weak point of OE. :( Apr 05 08:01:17 sorry I don't see it Apr 05 08:02:19 JaMa, I'd like to provide my own inittab in my application building receipt, not in global image receipt Apr 05 08:02:33 if you want package management working as expected, then you have to customize packages, not to alter them after install while doing do_rootfs Apr 05 08:02:35 s/receipt/recipe ;-) Apr 05 08:03:35 mckoan, :) sorry Apr 05 08:03:56 then I don't understand why what you mean by recipe-specific and now application, 09:32:13 < dv> postinst, postinst seems too receipt-specific. Apr 05 08:04:30 if you want your application to mess with system inittab, then put it in application postinst (with check that inittab is already messed or not) Apr 05 08:05:23 and user has to understand that inittab was messed intentionally and carefully merge inittab changes during next sysvinit upgrade Apr 05 08:05:52 lucky you, inittab is in CONFFILES now Apr 05 08:06:17 JaMa, thank you for answering and your patience! :) Apr 05 08:06:43 "in CONFFILES"? what it means to me? Apr 05 08:07:01 that you should read manual or ML :) Apr 05 08:07:25 in short, that it won't be overwritten without user interaction during sysvinit upgrade Apr 05 08:07:41 something like this will be shown * resolve_conffiles: Existing conffile /usr/share/navit/navit.xml is different from the conffile in the new package. The new conffile will be placed at /usr/share/navit/navit.xml-opkg. Apr 05 09:19:08 03Koen Kooi  07org.openembedded.dev * r2ea4d3792f 10openembedded.git/recipes/networkmanager/networkmanager_git.bb: Apr 05 09:19:08 networkmanager git: bump to 0.8.3.999 Apr 05 09:19:08 Signed-off-by: Koen Kooi Apr 05 09:19:18 03Koen Kooi  07org.openembedded.dev * r84f0436d63 10openembedded.git/conf/distro/include/angstrom-2010-preferred-versions.inc: Apr 05 09:19:18 angstrom: remove networkmanager pinnings Apr 05 09:19:18 Signed-off-by: Koen Kooi Apr 05 09:19:23 03Koen Kooi  07org.openembedded.dev * r1ba5e0122d 10openembedded.git/recipes/networkmanager/network-manager-applet_git.bb: Apr 05 09:19:23 network-manager-applet git: move to 0.8.3.999 Apr 05 09:19:23 Signed-off-by: Koen Kooi Apr 05 09:33:17 good morning mickey|office Apr 05 10:56:31 why are the *_CC_ARCH flags not set in the CPP variable by bitbake.conf? E.g. there is a difference between 'arm-linux-uclibceabi-gcc -E --sysroot=...' and 'arm-linux-uclibceabi-gcc -E -mthumb -mcpu=cortex-m3 -mfpu=vfp -mfloat-abi=softfp --sysroot=...' Apr 05 11:16:17 sounds like just an oversight; I guess nobody has previously encountered a situation where it matters. Apr 05 12:35:35 hi, i am having a problem building the firefox recipe from 2011.3 maintance branch. I swore it was building in the past, but when i updated to the branch last night firefox fails to build complaining it cannot find glib.h Apr 05 12:43:57 firefox, IIRC, has been having troubles lately. Apr 05 12:47:18 ah Apr 05 12:47:56 was it a change in the bitbake? i did not think i saw any recent changes that would have impacted it. Apr 05 12:50:00 I suspect if it was easy to identify what caused it, it probably would have been fixed by now. :) Apr 05 12:50:26 Search the logs for this channel to get some clues, perhaps? Apr 05 12:51:04 it seems like this same error popped up in 2008 last... but i cant find a recent post about it Apr 05 12:56:22 geiseri: latest change was this http://git.openembedded.org/cgit.cgi/openembedded/commit/?id=3aaa064fbbed5809aeac9373eeaab235712b6ae7 Apr 05 12:57:54 was that in the dev branch? or the maintance branch? Apr 05 12:58:24 dunno, check git log Apr 05 12:58:53 * geiseri is waiting for firefox to get arround to opening the url :) Apr 05 13:04:14 okay, i have that commit... Apr 05 13:06:32 really it looks like the only change that would impact me is the mozilla.bbclass Apr 05 13:07:48 i need to see what the impact of that patch is Apr 05 13:10:59 03Koen Kooi  07org.openembedded.dev * rc3265b14b2 10openembedded.git/recipes/xorg-lib/ (40 files in 2 dirs): Apr 05 13:10:59 pixman: add 0.21.6 + fixes Apr 05 13:10:59 Signed-off-by: Koen Kooi Apr 05 13:10:59 Acked-by: Martin Jansa Apr 05 13:21:52 khem: regarding fennec: I've been wondering about adding flash support. is that possible via current org.oe.dev metadata ? Apr 05 13:22:09 koen had better not be commiting from his exams! Apr 05 13:22:36 khem: also I have been wondering if you picked a certain revision or just the latest (maybe we can pick a stable version) Apr 05 13:22:56 is khem awake? Apr 05 14:03:24 morning Apr 05 14:03:41 gm Apr 05 14:03:41 JaMa: i reverted that commit to the mozilla.bbclass and it compiles again. i have to test the image though to see if it runs Apr 05 14:04:17 JaMa: should i report this as a bug against openembedded? Apr 05 14:05:27 geiseri: firefox ? Apr 05 14:05:36 dcordes, yes Apr 05 14:05:52 it looks like the last commit to mozilla.bbclass breaks it Apr 05 14:06:46 geiseri: let's see if I can compile fennec Apr 05 14:07:16 i get an error about not being able to find glib.h Apr 05 14:11:52 geiseri: on firefox do_configure ? Apr 05 14:12:03 yes Apr 05 14:12:30 g'day kergoth Apr 05 14:27:01 geiseri: fennec do_compile worked for me in org.oe.dev it's compiling now Apr 05 14:27:10 s/do_compile/do_configure/ Apr 05 14:27:37 im in the maintnence branch Apr 05 14:27:51 so fennec compiles, but mozilla doesn't Apr 05 14:28:32 geiseri: remotes/origin/2011.03-maintenance ? Apr 05 14:28:39 yes Apr 05 14:39:19 khem: in? Apr 05 14:40:46 khem: Ever seen an autotools recipe seem to force aclocal.m4 be from libtool 2.2.6b not 2.4? Apr 05 14:41:11 There's _not_ the magic to say don't regenerate aclocal.m4 being set and I even tried rm -f 'ing the file at the start of do_configure, same deal Apr 05 14:44:22 autotools.bbclass removes aclocal.m4 Apr 05 14:44:29 check for a local libtool.m4 its pulling Apr 05 14:47:09 Nope, no libtool.m4 Apr 05 14:47:34 libtool macros in acinclude.m4? Apr 05 14:47:42 or any other .m4, for that matter Apr 05 14:53:35 dcordes it seems my fennec wont build anyway, it cannot build autoconf213. Either way im confused as to why that commit broke the build of firefox. i actually think the prior state of the mozilla.bbclass would create a broken binary Apr 05 14:54:05 autoconf213 problem fixed by khem in org.oe.dev maybe you should switch branches ? Apr 05 14:54:39 i wanted to try to stay in a stable build for a while since i want to get to a shipping state Apr 05 14:55:08 geiseri: Then you may need to cherry pick fixes from dev Apr 05 14:55:10 how do other guys do it for production? just sit at a revision and cherry pick until you want to rebase? Apr 05 14:55:19 And send a pull request to the ML for the maint branch Apr 05 14:55:40 Tartarus, ok Apr 05 14:56:04 * Tartarus is happy to take well formatted pull requests from people other than koen and denix, btw ;) Apr 05 14:56:39 * dcordes is curious Apr 05 14:56:45 geiseri: may I ask what your target platform is ? Apr 05 14:56:57 dcordes the dm8148 Apr 05 15:04:29 geiseri: nice strong dsp Apr 05 15:05:49 geiseri: NOTE: package fennec-1_2.0+4.0b5-r0: task do_compile: Succeeded Apr 05 15:32:22 kergoth: nope, no libtool bits in the other m4 files Apr 05 15:39:45 is the 2011.03-maintenance the current stable branch? Apr 05 15:43:32 It is a long term non-development branch Apr 05 15:43:50 With a defined set of DISTRO/MACHINE/image targets that are expected to build Apr 05 16:31:39 * Tartarus digs more Apr 05 16:33:50 hi all, Apr 05 16:34:14 I'm working with a fresh OE installation on 32-bit lucid lynx Apr 05 16:34:45 and my first build "bitbake omap3-console-image" Apr 05 16:35:40 was trying to download files that no longer exist? Apr 05 16:39:17 daniel__: yes, quite possible. What did it not find? Apr 05 16:44:10 well, among others: ncurses-5.7-20100424-patch.sh.bz2 Apr 05 16:44:25 Tartarus: ^ Apr 05 16:44:42 k Apr 05 16:44:48 I think it's time for us to update ncurses again ;( Apr 05 16:44:50 :( Apr 05 16:52:40 daniel__: this ncurses version should not be required anymore since 0f2a949439089e8793ee17c677cbe3b8bc01aae5 (Jan 12) Apr 05 16:54:14 03Simon Busch  07master * r33107e85a5 10openembedded.git/recipes/freesmartphone/fsogsmd_git.bb: Apr 05 16:54:14 fsogsmd: add connman as dependency Apr 05 16:54:14 Signed-off-by: Simon Busch Apr 05 16:54:25 03Simon Busch  07master * rfe4156fc56 10openembedded.git/recipes/freesmartphone/cornucopia.inc: Apr 05 16:54:25 cornucopia: bump SRCREV to latest Apr 05 16:54:25 Signed-off-by: Simon Busch Apr 05 16:54:32 03Simon Busch  07master * re75fc3073e 10openembedded.git/recipes/connman/connman_0.72.bb: Apr 05 16:54:32 connman: add recipe for version 0.72 Apr 05 16:54:32 Signed-off-by: Simon Busch Apr 05 16:54:39 03Simon Busch  07master * rb6fc3c342c 10openembedded.git/recipes/ (3 files in 2 dirs): Apr 05 16:54:39 meta-toolchain-shr: add initial recipe Apr 05 16:54:39 This is the start of a own toolchain for the SHR distribution. Later it Apr 05 16:54:40 will include several utilities to develop applications for SHR. Apr 05 16:54:40 Signed-off-by: Simon Busch Apr 05 16:54:41 03Simon Busch  07master * rb1c6600c94 10openembedded.git/recipes/tasks/task-fso2-compliance.bb: Apr 05 16:54:41 task-fso2-compliance: uncomment connman runtime dependencies and bump PR Apr 05 16:54:42 Signed-off-by: Simon Busch Apr 05 16:54:45 03Simon Busch  07master * rc83d603d4a 10openembedded.git/recipes/mplayer/mplayer_0.0+1.0rc2.bb: Apr 05 16:54:45 mplayer: prefer version 1.0rc2 for palmpre machine Apr 05 16:54:45 Signed-off-by: Simon Busch Apr 05 16:54:45 03Simon Busch  07master * rfc93121244 10openembedded.git/recipes/freesmartphone/msmcomm.inc: Apr 05 16:54:45 msmcomm: bump SRCREV to latest version Apr 05 16:54:45 Signed-off-by: Simon Busch Apr 05 16:55:05 ensc|w : thanks, Apr 05 16:55:35 ensc|w : so now I wonder why my .bb recipes are out of date, and how I can get the latest changees? Apr 05 16:56:30 man i love schroot Apr 05 16:56:34 alias bitbake="schroot -c maverick -p -- bitbake" Apr 05 16:57:14 no -dev packages installed on the desktop itself, only in the chroot, which only has what is required for the build, no gui Apr 05 16:57:54 i wish i'd stop trying to open oe-core/classes, the tab complete always grabs something useless, instead of oe-core/meta/classes Apr 05 16:58:02 maybe i should make the completion a bit more picky in my zsh setup Apr 05 17:08:39 Hi all , When i try to build gstreamer (branch 0.10.32) i am getting the following " error: gettext infrastructure mismatch: using a Makefile.in.in from gettext version 0.17 but the autoconf macros are from gettext version 0.18" I googled and tried this solution (http://comments.gmane.org/gmane.comp.handhelds.openembedded/42069) , but hangs after this message http://pastebin.com/5Tn9R6uj Apr 05 17:09:15 does anyone had similar error Apr 05 17:20:38 kergoth_: I have lxc container with similar purposes.. Apr 05 17:20:45 but this is slightly different Apr 05 17:20:50 i need to play with lxc some more Apr 05 17:21:00 i like the idea of being able to run init, etc without having to use qemu Apr 05 17:21:11 I'll describe my setup on wiki one time :) Apr 05 17:21:14 :) Apr 05 17:21:23 rob landley had a couple nice blog posts about it Apr 05 17:21:31 and may be even upload container somewhere Apr 05 17:21:44 good idea near containers is cgroups :) Apr 05 17:21:57 you may renice priorities there Apr 05 17:21:59 * kergoth_ nods Apr 05 17:22:03 cgroups are very cool Apr 05 17:22:27 * kergoth_ wonders if python's urlparse module works well on our uris Apr 05 17:22:31 kergoth_, thanks for the tips! Apr 05 17:22:32 and now I have RAID0 under OE lxc container Apr 05 17:22:43 Jay7: ooh, neat Apr 05 17:23:04 very-very neat I'll say :) Apr 05 17:23:29 damn, not quite, urlparse doesn't get our params, i guess we don't comply with the uri rfc.. we should look into this Apr 05 17:23:58 iirc our params are delimited by ';' Apr 05 17:24:10 but we can't change to '&' Apr 05 17:24:42 because '&patch' may be part of url Apr 05 17:25:12 hm.. but we can add '&oe_very_own_parameters=..' Apr 05 17:33:23 Hi, is there a way to guarantee my bibtbake recipes are all up to date? Apr 05 17:37:14 daniel__: git pull Apr 05 17:47:51 foerster: thanks! It looks like my recipes are up to date Apr 05 17:48:16 daniel__: no problem. Apr 05 17:49:06 daniel__: what branch are you using? Apr 05 17:49:08 git branch Apr 05 17:52:20 Tartarus: git clone git://git.openembedded.net/bitbake bitbake Apr 05 17:52:36 Tartarus: I used this command to checkout my copy Apr 05 17:52:42 That just gets you bitbake Apr 05 17:52:44 not openembedded Apr 05 17:53:08 Tartarus: sorry, Apr 05 17:53:30 Tartarus: I followed these instructions : http://www.gumstix.org/software-development/open-embedded/61-using-the-open-embedded-build-system.html Apr 05 17:53:40 git clone git://gitorious.org/gumstix-oe/mainline.git org.openembedded.de Apr 05 17:54:04 OK, so you're using overo Apr 05 17:54:13 I don't know if that's tracking mainline close enough then :) Apr 05 17:54:34 The site says it''s about 90% mainline Apr 05 17:54:56 check: 99.9% Apr 05 17:55:37 interesting why other 10% are not in OE? Apr 05 17:56:00 it's 99.9%, not 90 ;) my mistake Apr 05 17:57:00 daniel__: Well, we fixed that ncurses problem, I think, back in Jan Apr 05 17:57:07 It could be we need to update ncurses junk again Apr 05 17:57:19 Since I think hollisb reported that problem to me last night Apr 05 17:57:27 I just wondering why overo is not in mainline OE Apr 05 17:59:19 :Jay7: I'm not sure Apr 05 17:59:23 Jay7: because some people is lazy to send stuff upstream :) Apr 05 17:59:56 Tartarus: hey, my .bb recipe for ncurses is looking for this file: Apr 05 17:59:58 ftp://invisible-island.net/ncurses/5.7/ncurses-5.7-20110108-patch.sh.bz2 Apr 05 18:00:16 i.e. openmoko devs did nice s3c24xx support for qemu, but they were lazy enough to push it upstream. Apr 05 18:00:48 Jay7: maybe w/ oe-core overo will finally become just another layer :) Apr 05 18:01:31 well.. Apr 05 18:01:58 * Jay7 need some new ideas what to improve in kexecboot Apr 05 18:02:20 Tartarus: which tries to fetch it from here, I think Apr 05 18:02:21 http://www.angstrom-distribution.org/unstable/sources/ Apr 05 18:02:34 Well, first it tries the link you pasted :) Apr 05 18:02:38 and this is where I get missing file/ permissions problems Apr 05 18:02:40 Then it may try other mirrors Apr 05 18:02:49 ok Apr 05 18:02:49 I don't know if overo uses the angstrom mirror Apr 05 18:03:33 So is this a case of Overo not keeping up with changes in mainline? Apr 05 18:03:40 Don't know. Apr 05 18:03:52 the angstrom mirror is for the angstrom distribution Apr 05 18:03:54 but others use it Apr 05 18:04:16 It looks like the upstream FTP site might be down Apr 05 18:04:50 * Tartarus digs Apr 05 18:05:04 Or they re-orged Apr 05 18:05:08 :Tartarus thanks! Apr 05 18:05:19 I'll wait a bit, and hope it works out Apr 05 18:06:21 khem: ping Apr 05 18:07:34 khem: nm Apr 05 18:08:55 ensc|w: ping Apr 05 18:32:06 * Tartarus tries ncurses 5.9 on uclibc next Apr 05 18:33:10 In trying to build OE, I get a series of errors like this: undefined reference to `pthread_getspecific' Apr 05 18:33:28 pingswept uclibc? Apr 05 18:33:30 I suspect that means I need to tell bitbake where to find pthread, no? Apr 05 18:33:36 no Apr 05 18:33:38 uclibc? Apr 05 18:33:45 tell the damn package to link right Apr 05 18:34:15 Meaning that gcc is correct, but the package is not? Apr 05 18:34:24 The package is perl-native. Apr 05 18:36:56 woglinde_: I'm not sure how to tell the package to link right. (The package in question is perl-native.) Apr 05 18:37:52 whats your host system? Apr 05 18:38:11 Ubuntu 11.04 Apr 05 18:38:45 hm then look at the packages which are needed Apr 05 18:38:50 its on the wiki Apr 05 18:39:19 You mean the packages needed to compile perl-native, or for OE generally? Apr 05 18:40:17 the packages which are recommended for building openemebdded Apr 05 18:41:54 You suspect that the situation is that perl-native is correct, but Ubuntu is missing a library? That would make sense-- I'm 75% sure this problem appeared when I upgraded to Ubuntu 11.04. Apr 05 18:54:38 I am trying to build Gstreamer from GIT ( using branch 0.10.23 ) .But I am not able to run autogen.sh due to missing .git directory . Can any one let me know why the . git is getting erased and what I should do to stop getting that erased. (I am verymuch new to Git (though I have build Gstreamer using svn)o build Gstreamer using GIt Pointers for building Gstreamer using GIT would be helpful. Here is my recipe http://pastebin.com/tz Apr 05 18:55:12 sreddy Your paste was cut off. Apr 05 18:55:33 OOps Sorry I ll repaste that thanks for informing Apr 05 18:57:35 Tartarus: pong Apr 05 18:57:52 sreddy look at the serval git recipes in oe Apr 05 18:58:31 ensc_: I'm going to rev ncurses to 5.9, now that they're doing updates / releases again Apr 05 18:58:44 anything I should be sure works when build testing stuff? Apr 05 19:00:39 Tartarus: how different is 5.9 from 5.7? Apr 05 19:00:58 Not overly, at first glance Apr 05 19:01:07 Took the 5.7 recipe and just changed the hashes Apr 05 19:01:21 built for qemuarm+angstrom+eglibc Apr 05 19:01:27 doing qemuarm+angstrom+uclibc now Apr 05 19:03:06 Tartarus: there is some non-trivial stuff which is common in all ncurses recipes; perhaps it should be refactored into a .inc Apr 05 19:03:43 but when 5.9 is only 5.7+last-patchset plus some minor patches, 5.7 can be perhaps moved to 5.9 Apr 05 19:05:20 Yeah, 5.8 and then 5.9 are probably pretty close to what we had been doing Apr 05 19:05:24 since 5.8 came out in feb Apr 05 19:05:29 and 5.9 in march to fix some bugs Apr 05 19:12:10 Getting this error when I build Gstreamer from GIT http://pastebin.com/ryDLhL19 Apr 05 19:12:58 woglinde_ : I will look at the recipes .. but I am not able to run autogen.sh Apr 05 19:13:07 ? Apr 05 19:13:16 we have our own stuff Apr 05 19:13:34 inherit autotools Apr 05 19:13:40 ok Apr 05 19:14:10 and start from gstreamet recipes which we have Apr 05 19:15:20 woglinde_ : I am using the already existing Gstreamer recipes and auto tools are inherited Apr 05 19:16:20 woglinde_ :This is my actual problem http://pastebin.com/ryDLhL19 Apr 05 19:21:16 why youe need 10.30? Apr 05 19:22:14 and you are using arago which not always keeps up with oe Apr 05 19:28:49 hm, looks like tight deadline Apr 05 19:28:58 irc, oe-devel, angstrom-devel... Apr 05 19:39:48 * kergoth_ sighs Apr 05 19:40:01 this code worked fine last week, no longer does Apr 05 19:43:42 what the hell is going on here, this makes absolute no sense Apr 05 19:58:39 hello. any pointers on how to work around this issue? http://www.mail-archive.com/openembedded-devel@lists.openembedded.org/msg15734.html Apr 05 20:03:14 * kergoth_ knows nothing about qt4 Apr 05 20:05:57 General OE question: is it true that when, on an x86 host, I build an Angstrom rootfs for an ARM processor, bitbake first builds an ARM toolchain with the x86 tools, and then uses that to build the ARM binaries? Apr 05 20:06:49 pingswept: yes Apr 05 20:07:15 its called cross toolchain runs on x86 produced code for arm/mips whatever you configure for Apr 05 20:07:18 khem: Thanks. Very interesting. Kind of wonder how I missed that until now. Apr 05 20:07:49 CMoH: did you try with master Apr 05 20:08:19 I had the CodeSourcery tools installed from building the Linux kernel outside of OE, and I just realized that I never told OE about them, yet the build seems to be working. Apr 05 20:08:27 Anyway, thanks. Apr 05 20:08:34 khem, no Apr 05 20:10:53 CMoH: thats would be first thing you should try Apr 05 20:11:59 khem, what about the 2011.03-maintenance branch? Apr 05 20:12:29 CMoH: yes that would be ok too Apr 05 20:13:15 pingswept: you *can* make it use your external cs toolchain if you want to Apr 05 20:13:22 it just defaults to building its own Apr 05 20:15:28 speaking of which; are these plethora of oe branches documented on some web page? :D Apr 05 20:17:39 hmm, i need to figure out how to tell OE that its running in a given chroot Apr 05 20:17:46 could use that fake hostname app i guess Apr 05 20:17:50 but that was ugly and iirc required root Apr 05 20:17:51 hrm Apr 05 20:18:24 kergoth_: why OE should know this? Apr 05 20:18:36 i.e. for what purpose? Apr 05 20:19:48 i use site.conf for machine specific things, like ASSUME_PROVIDED Apr 05 20:19:58 and my site.conf loads site/${HOSTNAME}.conf for that Apr 05 20:20:07 but of course teh chroto has a slightly different environment Apr 05 20:20:26 1 Apr 05 20:20:29 ups Apr 05 20:20:37 hehe Apr 05 20:21:19 in this particular case, my $SHELL is set to zsh, yet my chroto doesn't have zsh Apr 05 20:21:19 chromium died just when I was pressing Apr 05 20:21:23 which screws up my devshell Apr 05 20:21:43 (the new devshell code that isn't in oe yet, which uses $SHELL) Apr 05 20:22:16 hehe.. Apr 05 20:22:50 will fix it to just automatically fall back to bash at some point here, but still :) Apr 05 20:23:08 and of course nowadays my ASSUME_PROVIDED is really mostly set by what's available in the system, so that's less of a concern than it used to be Apr 05 20:23:14 export SHELL=/bin/bash inside chroots .profile :) Apr 05 20:23:26 using schroot, which bind mounts home, etc Apr 05 20:23:31 same shell setups :) Apr 05 20:23:34 ah Apr 05 20:24:01 * Jay7 prefer to have full setup and different home :) Apr 05 20:24:07 * JaMa using normal chroot and binding with mount --bind Apr 05 20:24:41 | install: cannot change ownership of `/var/tmp/oe/minimal/arm/work/armv7a-oe-linux-gnueabi/portmap-6.0-r5/image/sbin/portmap': Operation not permitted Apr 05 20:24:43 o_O Apr 05 20:24:53 bitbake master, i assume Apr 05 20:25:12 yep Apr 05 20:25:21 see ee39de6856b628df792c4a56966528796fbe2c8e. right now we aren't running the fakeroot tasks under fakeroot, due to an implementation change in how the stuff is run Apr 05 20:25:28 need to work out how best to do it Apr 05 20:25:32 you can revert the commit for now Apr 05 20:25:37 ok Apr 05 20:25:39 (of course, its required for oe-core) Apr 05 20:25:55 i'll get to it in the next few days, most likely Apr 05 20:26:16 np, I can fallback to 1.12 Apr 05 20:26:30 kergoth_: Thanks-- might use external CS toolchain some day. More likely to go the other way, and use the OE toolchain for everything. Apr 05 20:27:04 Jay7: or, you could test something. FAKEROOTENV = "LD_PRELOAD=libfakeroot-sysv.so LD_LIBRARY_PATH=/usr/lib/libfakeroot:${STAGING_LIBDIR}/libfakeroot" Apr 05 20:27:12 Jay7: if you set that in local.conf, master *should* work for you Apr 05 20:27:19 Jay7: if you would test it and let me know if it works, that'd be cool :) Apr 05 20:27:26 kergoth_: I'll test a bit later then :) Apr 05 20:27:32 :) Apr 05 20:27:36 after reading man schroot and cup of tea :) Apr 05 20:27:51 the new way is entirely environment driven, as with oe-core/poky, the *entire* bitbake run is underneith pseudo, and its controlled by env vars Apr 05 20:28:13 oe's fakeroot usage is really not very useful, all it does is avoids errors like the above, it doesn't actually get us correct perms in the packages Apr 05 20:28:33 but we can punt on that and know that the situation is less crappy when we move to oe-core Apr 05 20:28:44 in the meantime though, do need something that doesn't break us :) Apr 05 20:30:30 imho, we should speedup moving to oe-core and meta-oe Apr 05 20:30:48 because slow move is very expensive Apr 05 20:31:57 kergoth_: well, you may replace hostname with own wrapper that will return something like scrhoot.`hostname` Apr 05 20:32:47 03Bob Foerster  07master * r1ee82aa258 10openembedded.git/recipes/tzdata/tzdata_2011d.bb: Apr 05 20:32:47 tzdata: update to 2011e Apr 05 20:32:47 Signed-off-by: Bob Foerster Apr 05 20:32:47 Signed-off-by: Antonio Ospite Apr 05 20:32:52 03Bob Foerster  07master * rf9d86ff119 10openembedded.git/recipes/tzcode/tzcode-native_2011d.bb: Apr 05 20:32:52 tzcode-native: update to 2011e Apr 05 20:32:52 Signed-off-by: Bob Foerster Apr 05 20:32:52 Signed-off-by: Antonio Ospite Apr 05 20:35:09 Jay7: agreed, the problem is the amount of work to transition the metadata. we aren't even done with oe-core, much less meta-oe and whatever other layers Apr 05 20:35:12 heh Apr 05 20:35:19 i think we should get the core bits over as soon as we can Apr 05 20:35:48 (also, i'm doing a lot of my day to day work in oe-core, and then backport to mentor's product and oe, but haven't pushed it all to oe itself yet) Apr 05 20:38:04 :( Apr 05 20:38:44 I'm only good for testing tasks here, not for real work :( Apr 05 20:40:29 btw, command-prefix=command,option1,option2,... Apr 05 20:40:35 from schroot.conf Apr 05 20:44:13 ah, interesting, just need a shell script to mangle HOSTNAME. won't help the things that run 'hostname' directly, but will work for this Apr 05 20:44:17 thanks for the tip Apr 05 20:44:28 imho, this is better way here.. Apr 05 20:44:42 because there is no reliable way to detect that you are in chroot Apr 05 20:44:55 alias bitbake='schroot -u maverick -p -- bitbake' <3 Apr 05 20:45:00 so you should say it from outter side :) Apr 05 20:45:29 alias bitbake='HOSTNAME=schroot .... :) Apr 05 20:45:43 hehe, good point Apr 05 20:46:04 or introduce CHROOTED variable Apr 05 20:46:14 IS_CHROOTED even :) Apr 05 20:46:33 well.. now cup of tea :) Apr 05 20:47:34 woglinde_ : we have our plugins compiled against new gstreamer (i.e, for branch 0.10.30).Yes we are using Aragto distro Apr 05 20:47:44 schroot already sets SCHROOT_SESSION_ID and debian_chroot, can just make a wrapper script t hat sets HOSTNAME based on that Apr 05 20:47:53 actually.. i could add an schroot hook script to do it automatically Apr 05 20:47:55 * kergoth_ does that Apr 05 20:48:14 hrm, *think* it'll let me do that, anyway Apr 05 20:53:32 kergoth_: testing your suggestion about fakeroot Apr 05 20:56:35 hrm, -p makes it lose the HOSTNAME change i added to .profile Apr 05 20:56:35 grr Apr 05 20:56:37 * kergoth_ digs Apr 05 20:57:35 kergoth_: FAKEROOTENV works for me Apr 05 20:57:49 nice. Apr 05 20:58:03 now i need to verify that it works both with fakeroot-native and fakeroot on the build machine, if it does i'll see about pushing to bitbake.conf Apr 05 20:58:07 * kergoth_ adds to todo Apr 05 20:58:33 may I help you somehow? Apr 05 20:58:47 I can build something ;) Apr 05 20:59:00 well, fakeroot-native is optional, i think, so could just kill it on the build machine and bitbake fakeroot-native, then clean and rebuild the test recipe (portmap?) Apr 05 20:59:17 yes, portmap Apr 05 21:01:35 I've deinstalled fakeroot here and will do clean build Apr 05 21:02:54 * kergoth_ wasn't sure if oe automatically builds fakeroot-native as a base dep or not, might want to make sure its built before rebuilding portmap :) Apr 05 21:03:10 that was point of clean build :) Apr 05 21:03:21 we will know soon :) Apr 05 21:03:39 khem, what do you mean by master btw? it's not the "master" branch, is it? Apr 05 21:04:27 :) Apr 05 21:04:46 afaik it's the org.openembedded.dev branch Apr 05 21:05:08 CMoH: that was renamed to master months ago :) Apr 05 21:05:14 the old name is old around for compatibility Apr 05 21:05:23 s/old ar/only ar/ Apr 05 21:06:07 hmm... i get 403 forbidden for ncurses on the "master" branch. dunno of the org.openembedded.dev but... Apr 05 21:06:28 org.openembedded.dev just points to master. Apr 05 21:06:30 they're always identical Apr 05 21:06:38 okay Apr 05 21:06:39 you should update to master though Apr 05 21:06:49 what about the 2011.03 maintainance? Apr 05 21:06:53 (just to avoid relyance on the old name) Apr 05 21:06:54 what's its story? :D Apr 05 21:07:00 that's a stable branch based on the most recent release Apr 05 21:07:03 only bugfixes since Apr 05 21:08:25 will new hardware targets be added to 2011.03 maintainance? Apr 05 21:08:32 highly unlikely Apr 05 21:08:36 NOTE: Running task 299 of 813 (ID: 119, virtual:native:/home/oe/openembedded/recipes/fakeroot/fakeroot_1.14.5.bb, do_setscene) Apr 05 21:08:36 but ask tom, he maintains it Apr 05 21:08:57 Jay7: ah, good news. now we just need to see if the FAKEROOTENV works fine with it and portmap installs Apr 05 21:09:07 okay, but if i was going to base a product on something in there that currently had board support it would be the best bet? Apr 05 21:09:07 waiting for Apr 05 21:09:57 geiseri_: imho, yes Apr 05 21:10:05 but you should test it for your env Apr 05 21:10:44 then you may just use it as origin for your local branch Apr 05 21:10:46 i am now... i was using org.oe.dev for some time, but wanted to make a line in the sand. Apr 05 21:11:37 kergoth_, the only results for the new master branch is the failure to retrieve ncurses: http://pastebin.com/1qYjqaXU Apr 05 21:11:40 well, fakeroot-native is compiled Apr 05 21:11:50 about 200 tasks left Apr 05 21:14:03 CMoH: yes the location has changed Apr 05 21:14:35 mkay; i suppose the recipe drifted behind in the meantime :) Apr 05 21:23:05 VIA Eden is sooo sloow on compression.. Apr 05 21:23:46 * Jay7 is merging netflow data on customers router Apr 05 21:36:27 | ERROR: ld.so: object 'libfakeroot-sysv.so' from LD_PRELOAD cannot be preloaded: ignored. Apr 05 21:36:29 kergoth_: ^ Apr 05 21:36:51 portmap is failed as a result Apr 05 21:48:01 Jay7 I got some portmap install failures that looked like fakeroot wasn't working and then i switch from bitbake git tips to bitbake 1.10 branch and things worked better for me. Apr 05 21:48:33 mgross: yes, I know this already :) Apr 05 21:48:39 we are trying to fix this :) Apr 05 21:48:47 Jay7: bah, okay, i was afraid of that. different soname for fakeroot-native than the buidl machine's fakeroot, will have to make bitbake obey FAKEROOT as well as FAKEROOTENV Apr 05 21:48:49 * kergoth_ adds to todo Apr 05 21:48:53 Jay7: thanks for testing Apr 05 21:49:13 well, perhaps not, could you do find tmp -name libfakeroot\*.so? Apr 05 21:49:20 moment Apr 05 21:49:25 at least make sure the path added to LD_LIBRARY_PATH is right :) Apr 05 21:49:30 oh Apr 05 21:49:34 it should be STAGING_LIBDIR_NATIVE Apr 05 21:49:39 do that first :) Apr 05 21:49:54 /var/tmp/oe/minimal/arm/sysroots/x86_64-linux/usr/lib/libfakeroot.so Apr 05 21:49:56 /var/tmp/oe/minimal/arm/sysroots/x86_64-linux/usr/lib/libfakeroot-0.so Apr 05 21:50:07 TMPDIR=/var/tmp/oe/minimal/arm/ Apr 05 21:50:14 ah, okay, so the soname is indeed different Apr 05 21:50:17 okay, thanks Apr 05 21:51:50 kergoth_: ping me when you will have something to test again :) Apr 05 21:57:24 hello Apr 05 21:58:20 i can't get past the ncurses-native recipe; where did it move to fetch it by hand until someone gets to fixing the recipe? Apr 05 21:58:39 ok, nothing builds for uclibc atm Apr 05 21:59:28 if i am looking for how a specific recipe gets included in a build (like which other recipes reference it), what would be the best way to find it? Apr 05 22:00:48 i have been grepping the recipes directories for it but so far have only found one other recipe that references the one for which i am searching Apr 05 22:01:15 (and i know that other recipe is not built) Apr 05 22:07:30 in the meantime i've copied them from a lucky computer that got the files before being relocated Apr 05 22:07:31 03Tom Rini  07master * r29d49f74c2 10openembedded.git/recipes/ncurses/ (3 files in 2 dirs): Apr 05 22:07:31 ncurses: Add version 5.9 Apr 05 22:07:31 This is relatively close to our 5.7 version in that we had been using Apr 05 22:07:31 5.7 + a snapshot of trunk which became 5.8 and then 5.9. Apr 05 22:07:31 Signed-off-by: Tom Rini Apr 05 22:08:12 which branch is recommended for commercial products? Apr 05 22:09:36 CMoH: Well, many of us would recommend the vendor we work for, so we're a bit biased Apr 05 22:10:23 CMoH: and re ncurses-native, git pull, unless you're on 2011.03-maintenance which doesn't have what i just pushed Apr 05 22:11:12 ty, Tartarus Apr 05 22:14:06 however on git branch -r i still can't distinguish which vendors are which :D Apr 05 22:15:12 thanks for the reply; we'll see i guess Apr 05 22:16:51 Tartarus: Does this apply to the Gumstix branch? Apr 05 22:18:26 CMoH, what board are you using? Most boards have their own favorite oe tree Apr 05 22:20:07 we're developing a new one; for now the beagleboard Apr 05 22:20:10 http://pastebin.com/bCBHRSra Apr 05 22:21:00 daniel__: probably, yes Apr 05 22:21:13 CMoH: Vendors don't have their branches upstream, you buy their product :) Apr 05 22:21:19 CMoH, i started using Arago's flavor Apr 05 22:21:50 CMoH, since that seemed to be what TI was pushing when we started using their dm8148 Apr 05 22:22:31 CMoH, i think they follow oe pretty close though Apr 05 22:22:49 Nevermind, missing dep on host OS. Is the wiki gobally writable? Apr 05 22:25:08 thanks for your comments. to sum up, I guess we'll have to branch from master or 2011.03-maintenance Apr 05 22:25:58 CMoH, if you are new to linux distros though i would strongly reccomend talking to one of the vendors though... you will save a hell of a lot of money and time. Apr 05 22:26:39 i am pretty satisfied with OE and angstrom for our purpose :) Apr 05 22:27:13 CMoh: any details on what you're building? I'm in a similar situation. Apr 05 22:27:33 (I also have concluded that 2011.03 maintenance is a good choice.) Apr 05 22:27:56 CMoH okay, just throwing that out there... note i do not work for a vendor :) Apr 05 22:28:06 hmm... it's what everybody is building i guess :) a tablet Apr 05 22:28:06 Tartarus: We're still not having luck with the overo ncurses... could definitely be a misunderstanding on our side though. Apr 05 22:28:20 lol Apr 05 22:28:23 Any wiki editors around. In the ArchLinux section docbook-xml should be docbook-xsl Apr 05 22:28:25 yay another linux tablet :) Apr 05 22:28:29 CMoH: cool. Apr 05 22:28:52 CMoH will it have a UI that wont suck, or require a 1Ghz cpu to render the wallpaper? Apr 05 22:28:56 (Of OEandYourDistro) Apr 05 22:28:59 Tartarus: New to OE, and need to learn the ins and outs before I start complaining too loudly ;-D Apr 05 22:29:05 well with some integrated proprietary software Apr 05 22:29:31 eh, it's targeted for a specific field, not a general-purpose tablet Apr 05 22:29:36 good call Apr 05 22:29:49 android has all the bottom feeders in the consumer market Apr 05 22:30:29 anyway, after the master branch i was following split into an unstable master and a 2011.3 stable branch, i also have to make a choice :D Apr 05 22:30:38 ty for your advice as well, pingswept Apr 05 22:31:01 CMoH, FYI i am using 2011.3 branch based off of the minimal distro Apr 05 22:31:03 No problem. Take it with a grain of salt, as I'm pretty new to OE. Apr 06 00:26:54 03Chris Larson  07master * rd6fba8bad6 10bitbake.git/lib/bb/persist_data.py: Apr 06 00:26:54 persist_data: make SQLTable a context manager Apr 06 00:26:54 This can be used for more control over the underlying transactions. Unlike Apr 06 00:26:54 the context manager of, say, a file object, we can still use the object even Apr 06 00:26:54 after the end of a given with block, as the context manager exit only ensures Apr 06 00:26:54 we've committed to the database, not that we have closed the database. Apr 06 00:26:55 Signed-off-by: Chris Larson Apr 06 00:26:59 03Chris Larson  07master * r7942833ca0 10bitbake.git/lib/bb/persist_data.py: Apr 06 00:26:59 persist_data: add has_key Apr 06 00:26:59 Signed-off-by: Chris Larson Apr 06 00:27:00 03Chris Larson  07master * r1190406c52 10bitbake.git/lib/bb/ (compat.py persist_data.py): Apr 06 00:27:00 persist_data: implement comparison, same as dict Apr 06 00:27:00 Signed-off-by: Chris Larson Apr 06 00:27:01 03Chris Larson  07master * ra4f6243384 10bitbake.git/lib/bb/ (4 files in 3 dirs): Apr 06 00:27:01 persist_data: raise KeyError on missing elements Apr 06 00:27:01 Signed-off-by: Chris Larson Apr 06 00:27:02 03Chris Larson  07master * r28958cd55e 10bitbake.git/lib/bb/persist_data.py: Apr 06 00:27:02 persist_data: don't allow non-string keys/values Apr 06 00:27:21 there, now it passes python's mapping unit tests Apr 06 00:38:33 * Tartarus posts pull request for oe-core for ncurses **** ENDING LOGGING AT Wed Apr 06 02:59:58 2011