**** BEGIN LOGGING AT Mon Mar 23 02:59:57 2009 Mar 23 06:55:18 im trying to build a gpe-image, i get the error while fetching the package: rosetta_anoncvs.handhelds.org__20090320.tar.gz, it seems there is no package by that name Mar 23 08:44:31 03Koen Kooi  07org.openembedded.dev * r589db5a12d 10openembedded.git/ (2 files in 2 dirs): kirkwood: clean up kernel and machine file a bit Mar 23 08:44:42 03Koen Kooi  07org.openembedded.dev * re2b99b79f5 10openembedded.git/contrib/angstrom/sort.sh: Angstrom feed sorter: add sheevaplug Mar 23 08:47:08 good morning Mar 23 08:47:45 moring Mar 23 09:12:39 does someone know why are the linux-omap-pm_git, linux-omap_2.6.28 and ffmpeg_git recipes removed in the dev branch? Mar 23 09:15:16 * mpoullet|work is dumb:-(, forget the move packages->recipes, sorry for asking Mar 23 09:32:10 morning Mar 23 09:39:22 good morning Mar 23 09:48:19 why does not meta-e-x11 give me a file system image? Mar 23 09:48:35 arent all meta .bb files supposed to generate an image Mar 23 09:51:05 vsr: no Mar 23 09:51:20 vsr: images/*.bb are supposed to Mar 23 09:51:30 meta/* are just metapackages Mar 23 09:52:43 mornig all Mar 23 09:52:59 what is the difference? Mar 23 09:53:17 is meta package is a collection of packages Mar 23 09:54:10 yes Mar 23 09:54:17 but it generates packages Mar 23 09:54:29 empty packages with lot of dependencies on other packages Mar 23 09:55:13 hrw: can you give me an example, im finding it hard to understand, why would meta packages want to be like that Mar 23 09:56:54 vsr: for example I want to have image with 246 packages in it. there will be X11 system + native SDK + python environment Mar 23 09:57:24 vsr: so I create image which has task-native-sdk (metapackage), task-x11-system (metapackage), python-env(metapacakge) Mar 23 09:57:41 vsr: I do not have to check what those packages contains as they depend on all I need Mar 23 10:00:28 but task-x11-system, task-native-sdk arent in meta directory Mar 23 10:02:20 but big parts of meta/ recipes just groups tasks into metapackages Mar 23 10:06:48 hi gremlin[it] Mar 23 10:06:57 hrw: thanks Mar 23 10:07:24 hi mckoan ! Mar 23 10:12:15 anybody knows how to get a branch from OE repo? for example if I want origin/ulf/linux what do I have to do? Mar 23 10:17:26 git checkout origin/ulf/linux -b your-local-name-for-it Mar 23 10:18:00 thx Mar 23 10:19:44 * mckoan is waiting for O'Reilly book "Version Control with Git" Mar 23 10:20:23 will be published 27 May 2009 Mar 23 10:22:49 hrw: -b = Create a new branch named and start it at Mar 23 10:23:12 hrw: does it fetch it too? Mar 23 10:23:41 you can also 'git pull origin/ulf/linux' to merge it into your branch Mar 23 10:24:11 'git fetch' do fetching. 'git pull' do fetch + merge Mar 23 10:24:26 'git checkout' do checkout from already fetched stuff Mar 23 10:25:45 $ git pull origin/ulf/linux Mar 23 10:25:45 fatal: 'origin/ulf/linux': unable to chdir or not a git archive Mar 23 10:25:46 fatal: The remote end hung up unexpectedly Mar 23 10:27:02 good morning Mar 23 11:01:04 I seem to remember having asked this before, but I quite forgot the answer and I'm partioning disk space.. sooo. How much space is needed for OE? 20 GB ok ? Mar 23 11:08:49 schme: should be enough Mar 23 11:09:21 Great beans! thanks. Mar 23 11:11:10 schme, why are you setting up a different partition just for OE? Mar 23 11:11:12 schme, i saw some info about beagleboard target, to do full graphical image 45G is suggested Mar 23 11:13:03 kristoffer: I have no free space on my current partitions. Mar 23 11:13:22 gremlin[it]: yikes. Good thing I'm not doing graphical install then. Mar 23 11:13:23 Hmm.. Mar 23 11:13:28 I see you guys switched to git :) nice! Mar 23 11:14:51 schme, ah Mar 23 11:26:12 m Mar 23 11:26:53 ^^: My apologies for that. Typo made while joining a channel. Please ignore. Mar 23 11:30:31 Ok q#2, last time I was poking around with OE I gave up on composing mah own recipies, and instead I ended up with scratchbox. Reason was that some software needed the "remote execution" while building. Is there some easy way to do this using OE ? Mar 23 11:30:59 (ie. it needed to run on the actual device, and not in the emalutade environment) Mar 23 12:00:28 schme: what for they are need to be run? Mar 23 12:00:30 re Mar 23 12:01:57 hrw: Well I tried building some apps. Like clisp. It uses gcc to build, then it runs itself to build some more stuff. This has never worked for me without running the latter part on the actual device. arm emulation seems broken for this. Mar 23 12:02:59 hrw: but setting scratchbox up to run gcc locally, and run the clisp stuff on the actual device.. worked just fine. After I had installed me some ångström on it first, of course. Mar 23 12:03:08 (for the ssh and nfs and what not) Mar 23 12:04:47 how can i see what all components are there when one bb file is used Mar 23 12:04:56 according to clisp page 2.45 is cross-compilable Mar 23 12:05:16 hrw: According to real life it is not ;) Mar 23 12:05:27 Also I have the same issue with some other software. Mar 23 12:05:35 schme: too bad that I lack time to check ;D Mar 23 12:05:42 Indeed! Mar 23 12:05:53 I'm just curious about how to set up the remote execution thing with the oe Mar 23 12:05:54 03Robert Schuster  07org.openembedded.dev * r036b1362f3 10openembedded.git/recipes/llvm/ (llvm-2.5/fix-build.patch llvm_2.5.bb): Mar 23 12:05:54 llvm 2.5: Fixes Mar 23 12:05:54 - updated patch to include important preprocessor definitions Mar 23 12:05:54 - copy missing header into staging manually Mar 23 12:07:53 i want to know which packages are included when a particular bb file is used. how do i know Mar 23 12:09:13 sh1: you mean 'which pacakges will get generated as result of building of particaluar bb file'? Mar 23 12:10:30 i have chosen to do a bitbake x11-image. How do i know what constitutes a x11-image jffs2 file system Mar 23 12:12:34 ah. Mar 23 12:12:51 check what is in x11-image recipe, then check what those tasks contains Mar 23 12:13:42 that is a huge list. is there any easy way to do that. Mar 23 12:14:09 i need to open multiple .bb files and check dependencies and open some more files Mar 23 12:15:25 easier ways: 1. build it and check what is inside 2. fetch already built and check what is inside Mar 23 12:20:10 hrw: thanks, i will try them :-) Mar 23 12:52:33 is there a guide available explaining how to use a custom kernel? Mar 23 13:07:52 how can i select to compile a particular version of the kernel Mar 23 13:13:21 Gnutoo, i send you an email Mar 23 13:13:42 gremlin[it], ok thanks i'll look Mar 23 13:15:50 03Stefan Schmidt  07fso/milestone5.5 * r24914241bb 10openembedded.git/packages/connman/ (connman_0.10.bb connman_git.bb): Mar 23 13:15:50 connman: Moved from moblin.org to kernel.org Mar 23 13:15:50 Conflicts: Mar 23 13:15:50 packages/connman/connman-gnome_0.4.bb Mar 23 13:15:50 packages/connman/connman_0.10.bb Mar 23 13:15:52 packages/connman/connman_0.8.bb Mar 23 13:15:56 packages/connman/connman_0.9.bb Mar 23 13:15:58 03Stefan Schmidt  07fso/milestone5.5 * re73f13dd16 10openembedded.git/packages/connman/connman_0.10.bb: connman_0.10: Fix merge error. Mar 23 13:16:01 03Stefan Schmidt  07fso/milestone5.5 * rb0f835383f 10openembedded.git/conf/checksums.ini: Mar 23 13:16:04 checksums.ini: Add sums for connman packages. (Changed location) Mar 23 13:16:06 Conflicts: Mar 23 13:16:08 conf/checksums.ini Mar 23 13:16:10 03Stefan Schmidt  07fso/milestone5.5 * ra89a5bf239 10openembedded.git/conf/distro/include/preferred-om-2008-versions.inc: preferred-om-2008-versions: Bump connman to 0.13 Mar 23 13:16:15 03Stefan Schmidt  07fso/milestone5.5 * r80a1a8fba5 10openembedded.git/ (conf/checksums.ini packages/connman/connman_0.13.bb): connman: Add version 0.13 Mar 23 13:31:50 i have specified PREFERRED_PROVIDER_virtual/kernel = linux-omap-2.6.28. But it still says multiple providers are available for virtual/kernel Mar 23 13:37:37 sh1: the value should be in quotes Mar 23 13:39:36 yes it was specified in quotes Mar 23 13:40:54 but it wont work Mar 23 13:40:55 PREFERRED_PROVIDER_virtual/kernel = "2.6.28" I think Mar 23 13:41:09 wait Mar 23 13:41:16 PREFERRED_PROVIDER_virtual/kernel = "linux-omap" I think Mar 23 13:41:49 sh1: yeah, the version number is not part of a provider specifier Mar 23 13:42:07 sh1: unless the number is part of the actual recipe name (everything before the underscore) Mar 23 13:42:11 so it will take the higest version number kernel Mar 23 13:42:52 sh1: you can specify the version to use by putting Mar 23 13:42:52 PREFERRED_VERSION_linux-omap = "2.6.xy" Mar 23 13:42:52 into local.conf Mar 23 13:43:32 rschuster: cant i use both the version and linux-omap in virtual/kernel Mar 23 13:43:33 sh1: look at the angstroms distro configuration (below conf/distro) to see examples Mar 23 13:44:24 sh1: preferred version and virtual provider preference are two separate things. so no Mar 23 13:45:31 but having these two lines in your local.conf achieves what you want: Mar 23 13:45:31 PREFERRED_PROVIDER_virtual/kernel = "linux-omap" Mar 23 13:45:31 PREFERRED_VERSION_linux-omap = "2.6.xy" Mar 23 13:46:10 what exactly is meant by PREFERRED_PROVIDER_virtual/kernel Mar 23 13:46:31 should i specified the name of the directory where linux kernel where defconfig exists Mar 23 13:47:02 sh1: what would you prefer the package virtual/kernel to be built with ;-). Nope, it refers to a bb recipe. Mar 23 13:49:54 sh1: what 'defconf' have you edited? Mar 23 13:52:10 DJWills: I meant directory where defconfig resides, for example linux-omap-2.6.28 Mar 23 13:55:46 gremlin[it], ok i read the mail...but I've a question...could the sheevaplug be bought in a shop in milan or from somebody in milan...it would be easier for me than ordering from the internet Mar 23 13:59:18 03Angus Ainslie  07fso/milestone5.5 * r569bf44efa 10openembedded.git/packages/efl1/ecore-native_svn.bb: ecore-x needs to be enabled for edje-sdk to compile Mar 23 13:59:21 03Angus Ainslie  07fso/milestone5.5 * r633dc2c693 10openembedded.git/packages/xorg-lib/libxext-native_1.0.4.bb: New libxext-native recipe Mar 23 13:59:33 03Angus Ainslie  07fso/milestone5.5 * r43e379df35 10openembedded.git/: Merge branch 'fso/milestone5.5' of git@git.openembedded.net:openembedded into fso/milestone5.5 Mar 23 14:05:35 i have been going through the logs of oe. recently i have seen that some ppl say that "angstrom" is a ditro. A distro is a collection of version numbers of all the packages. Mar 23 14:06:20 going by this logic all angstrom boils down to is it specify what versions to use for which package. is my understanding correct? Mar 23 14:06:53 sh1: yep, plus some glue that makes it easy to switch libc implementation Mar 23 14:07:44 rschuster: i did not understand the glue part Mar 23 14:11:31 rschuster: are you saying apart from the package versions it also allows us to choose the libc/uclibc? Mar 23 14:11:42 sh1: I also do not know how its implemented but the result of is that you can decide against which c library you build by specifiny ANGSTROMLIBC Mar 23 14:12:05 sh1: sure Mar 23 14:12:19 great Mar 23 14:12:49 you can chose between uclibc, glibc and eglibc Mar 23 14:13:30 great Mar 23 14:13:35 however for certain architectures you do not have a choice, since a proper implementation is missing Mar 23 14:13:46 e.g. avr32 is only support by uclibc Mar 23 14:14:32 good thing. looks like glibc support for avr32 is underway: Mar 23 14:14:33 http://sourceware.org/ml/libc-ports/2008-11/msg00008.html Mar 23 14:15:01 cool Mar 23 14:15:15 * hrw waits for ngw100 Mar 23 14:15:16 so the difference between various distro in openembedded/conf/distro is just version numbers of packages and c libraries Mar 23 14:15:23 great Mar 23 14:16:28 sh1: yep. the definition of what is on your rootfs is made by an 'image' - see recipes/images/* Mar 23 14:19:56 http://www.youtube.com/watch?v=dq_SJ7CtnZI& Mar 23 14:24:48 Gnutoo: gremlin[it] : if you want we can buy a SheevaPlug together Mar 23 14:25:37 mckoan, do you live in milan? Mar 23 14:26:04 mckoan, mhhh iteresting more cause you are a company not a private ;) Mar 23 14:27:58 Gnutoo: Bergamo Mar 23 14:28:08 indeed ;-) Mar 23 14:28:34 I just asket to my Marvell representative Mar 23 14:28:51 mckoan: get one more so I will do not have to order too ;D Mar 23 14:29:21 ok stay tuned ;-) Mar 23 14:30:19 ok Mar 23 14:30:59 * hrw goes to hard task of recreating image recipe from /usr/lib/opkg/status file Mar 23 14:37:07 does OE have support for mp4 and mp3 codecs? Mar 23 14:37:27 there is no such thing as mp4 Mar 23 14:37:55 you want to play divx, h264 etc? mplayer Mar 23 14:38:03 or gst-ffmpeg Mar 23 14:38:07 or just ffmpeg Mar 23 14:58:19 I can't believe! every time I rebuild something with OE I face to a new error I never saw before Mar 23 14:58:45 some months ago I was able to build x11, now it fails Mar 23 14:59:02 some weeks ago I was able to build the toolchain alone, now it fails Mar 23 14:59:08 :-( Mar 23 14:59:36 we do need a stable snapshot ASAP Mar 23 15:00:24 I can't rely on this behaviour Mar 23 15:00:28 I recognize that, althoug I didn't build images a few months ago yet Mar 23 15:00:40 mckoan: true, hoerver after yesterday's fixe x-11 image built fine for me Mar 23 15:01:43 what's failing there? Mar 23 15:02:54 the only issues we see are with packaged-staging Mar 23 15:03:16 bitbake meta-toolchain worked last time I did it Mar 23 15:03:24 http://pastebin.com/m5dee16b8 Mar 23 15:04:29 hmm...our sdk guy here is Tartarus Mar 23 15:04:40 I think I've read something about ncurses Mar 23 15:07:54 03Robert Schuster  07org.openembedded.dev * r107b6a650d 10openembedded.git/recipes/classpath/ (2 files in 2 dirs): classpath-native 0.97.2: Retrieved back from 595952e9bbff5d25c370fd140d58c3aae9434690. Mar 23 15:07:57 03Robert Schuster  07org.openembedded.dev * r738f0e354b 10openembedded.git/recipes/classpath/classpath-native_0.98.bb: classpath-native 0.98: Set DEFAULT_PREFERENCE to -1. Mar 23 15:08:30 mckoan: woglinde: ncurses: set LEAD_SONAME to libncurses.so.5 and bump PR Mar 23 15:09:22 mckoan: I used to have the same experience with OE. So this made me use OE while developing, but for finalizing a project I used a specific, known version of OE and marked this in my main make file Mar 23 15:09:58 mckoan: for two projects I actually used only parts of OE, that is, I copied specific files from site, classes, conf, recipes and packages to have a know and repeatable state Mar 23 15:10:22 mckoan: however, using the "stable" branch of OE should solve your problem, too Mar 23 15:11:15 morning Mar 23 15:11:35 hi Chris Mar 23 15:11:50 the current stable branch is 'a little bit' outdated Mar 23 15:11:57 hey Mar 23 15:12:09 ah, the *stale* branch you mean ;-) Mar 23 15:12:12 kergoth: hi ! Mar 23 15:12:23 hey schurig, how's it going? Mar 23 15:13:07 tsjsieb: then someone should declare himself as stable maintainer and create a new stable one, e.g. like the Debian project: let some packages survive for some days, put them into testing, at some time finanalize testing and put it into stable Mar 23 15:13:15 schurig: thx, BTW stable should mean ord, isn't it? so other thinghs may be broken Mar 23 15:13:18 tsjsieb: but this is mailing list stuff :-) Mar 23 15:13:24 s/ord/old Mar 23 15:13:25 kergoth: fine! And you? Mar 23 15:13:32 pretty good, keeping busy Mar 23 15:13:35 tsjsieb, outdates is the definition of stable in one sense Mar 23 15:13:42 Someone volunteered to do a stable release in the last week or so. Mar 23 15:13:54 I did Mar 23 15:14:05 it is, but therefore it's not realy the solution for mckoan's problem Mar 23 15:14:24 mckoan: for me "old" means just ... old. Not the newest and best, but on my devices I seldom need that. They are not PDA-like and have a very defined set of software running on it, hardly with any updates. Mar 23 15:14:36 03Stefan Schmidt  07fso/milestone5.5 * rff5db81b50 10openembedded.git/packages/connman/connman_0.12.bb: Mar 23 15:14:36 connman_0.12: Add an older version. Mar 23 15:14:36 0.13 segfaults here when scanning for networks. Debugging still pending. 0.12 Mar 23 15:14:36 works fine though. Mar 23 15:14:37 hrw, we need to make some headway on this Mar 23 15:14:37 03Stefan Schmidt  07fso/milestone5.5 * r2072565a30 10openembedded.git/conf/distro/include/preferred-om-2008-versions.inc: Mar 23 15:14:41 preferred-om-2008-versions: Use an older connman version that does not fail Mar 23 15:14:43 while scanning. Mar 23 15:15:49 my problem now is to solve this ncurses bug Mar 23 15:16:21 hrw, ping Mar 23 15:16:27 ncurses bug? Mar 23 15:16:46 schurig: hrw is doing other stuff with me and all will be merged in the mainline, so I can't use stable Mar 23 15:17:06 mckoan: such is life :-) Mar 23 15:17:13 Crofton|work: we need Mar 23 15:17:23 yes Mar 23 15:17:24 schurig: ;-) Mar 23 15:17:30 Crofton|work: but I do not want to wait for next oedem with it Mar 23 15:17:36 me niether Mar 23 15:17:45 unless me can make that happen in a few weeks Mar 23 15:17:45 mckoan: you think wrong Mar 23 15:18:03 mckoan: you can use stable and improve it in same time. Mar 23 15:18:10 hrw, we should talk, but I must go eat lunch, back in a whiole Mar 23 15:18:21 mckoan: your improvements lands in .dev and then you ask for backport it to .stable Mar 23 15:18:28 Crofton|work: smacznego then Mar 23 15:18:29 the point is test before commit to stable Mar 23 15:18:41 ofcourse Mar 23 15:18:58 Crofton|work: obvious Mar 23 15:19:14 mckoan, not as obvious as I would like :) Mar 23 15:19:18 what "test" means is kind of fuzzy, i'm afraid Mar 23 15:19:21 LOL Mar 23 15:19:37 kergoth, we define it then Mar 23 15:19:44 hrw, thanks! Mar 23 15:20:06 now, if only google translate could teach pronunciation Mar 23 15:20:10 kergoth: for me it is: passing my 'build from scratch' buildbot with arm/x86 devices and few different images Mar 23 15:20:39 kergoth: and it change does not touch any of them then check does it builds for arm/x86 targets Mar 23 15:20:56 hrw: I woild add some other required recipe like meta-toolchain of course Mar 23 15:21:01 * kergoth wishes we could track which variables were actually -used- in the build of a given recipe, so we could dump a list of which of those variables are conditional, so we can test every one of those conditions Mar 23 15:21:08 s/woild/would Mar 23 15:21:12 kergoth: hehe, deprecate FILESDIR before getting rid of it :-) Mar 23 15:21:33 kergoth: every one? Wow, you must have a bunch of Intel Core7's :-) Mar 23 15:22:03 * kergoth rolls eyes, didn't foresee the unintended wildcard feature Mar 23 15:22:05 schurig: or farm Mar 23 15:22:36 kergoth: go ahead with house-cleaning ;-) Mar 23 15:22:37 kergoth: hey, I didn't want to stab on you, I made a :-) there for a purpose. Please unroll your eyes, will you? Mar 23 15:23:49 Oh, come on - thats the first mistake kergoth has made in what, 3 years? Thats a pretty good record... :) Mar 23 15:23:49 that wouldn't necessarily be needed, if you can track what variables go into a build, it would reduce the amount of package rebuilds necessary. not all of the packages in the dependency tree for the recipe being tested would be affected by the changes to the conditionals Mar 23 15:23:56 Mar 23 15:24:21 to be honest, i was going to send the patches to the list, but send-email decided to be difficult, and i gave up :P Mar 23 15:24:50 BTW, log-2009-03-02:19:26 Michael 'Mickey' Lauer did r43fa438235 openembedded.git/packages/ncurses/ (ncurses.inc ncurses_5.4.bb): ncurses: modernize and fix packaging problems Mar 23 15:25:07 kergoth: I'd need to look at the representation of variables now in Bitbake, it would slowdown BB a bit, but it would be possible. BB has changed a lot since I invented it (and later stopped working on it) Mar 23 15:25:07 after that I read people claiming fot that Mar 23 15:25:28 schurig: the problem is, we're basically using the same bb.data implementation that we put in the prototype. Mar 23 15:25:37 kergoth, the problem is that it is easier for everyone to commit to .dev, which leads to all of us having high blood pressure :) Mar 23 15:25:43 schurig: everything ends up in a big pile, variable refs untracked Mar 23 15:25:55 kergoth: I thought zecke made the thing suck much less than my implementation? Mar 23 15:26:03 kergoth: cow-aware, you know? Mar 23 15:26:18 also woglinde did something on ncurses on 2009-03-12 Mar 23 15:26:22 yeah, somewhat. it reduced the number of full deep copies of the metadata by tasks making their own local datastores Mar 23 15:27:59 mckoan, i was AFK ... i have collect request for about 7 sheeva in north italy ... if we can pass through your company we'll be happy ... Mar 23 15:28:36 gremlin[it]: I'll try as soon as I hear Marvell representative Mar 23 15:28:38 that reminds me, i've been thinking that tasks shouldn't be able to make changes to the datastore. those are serious side effects, and means behaviors could change based on which tasks i happen to be executing this build. some of those still exist, and should be audited. perhaps even hand every task their own DataSmart copy as 'd' instead of the original Mar 23 15:30:55 mckoan, let stay in touch :) Mar 23 15:34:33 anyone know about the GNU_HASH stuff in insane.bbclass? Mar 23 15:39:23 kergoth: you could start with something like http://pastebin.ca/1369191 Mar 23 15:40:27 what does "ERROR: QA Issue" mean ? Mar 23 15:41:02 mckoan: sanity checker found a problem. what the problem is varies Mar 23 15:42:41 kergoth: one recipe that just creates a new filesystem from existing IPKs here shows 275 unque variables. I wonder how much I'll get when I recompile GCC :-) Mar 23 15:42:50 yikes :) Mar 23 15:43:02 * kergoth was having fun a few weeks back toying with alternative bb.data apis - http://kergoth.pastey.net/108517 Mar 23 15:43:11 * kergoth didn't get anywhere near finishing such a thing though, was just a proof of concept Mar 23 15:44:58 schurig: I have recipe which creates package from rootfs created from packages. Mar 23 15:45:10 hrw: I have this for years, thanks! Mar 23 15:45:11 schurig: and resulting package is then used to create rootfs Mar 23 15:45:26 * kergoth chuckles Mar 23 15:53:53 hrw: would you give a look at my fixes from yesterday (qemu_svn, keymaps, opie-multikey) ? (FILESPATH / FILESDIR) Mar 23 15:55:01 could add new wildcard support that actually gathers together all the matching globbed items from everywhere in filespath, i wonder if that'd be worthwhile Mar 23 15:55:10 makes the builds slightly less deterministic Mar 23 16:01:43 despite such errors meta-toolchain builds Mar 23 16:01:45 NOTE: Tasks Summary: Attempted 990 tasks of which 0 didn't need to be rerun and 0 failed. Mar 23 16:02:37 kergoth: I don't know how many packages are using wildcards...not so many it seems Mar 23 16:02:50 only 6-7 were needing fixes Mar 23 16:03:04 or rather we found 6-7 being used Mar 23 16:03:10 but someone should build world Mar 23 16:03:15 he :-) Mar 23 16:03:52 i tried to build world a couple weeks ago Mar 23 16:03:56 after 48 hours it was almost half way through Mar 23 16:04:18 kergoth: and first it gave long list of errors before doing building? Mar 23 16:04:27 of course Mar 23 16:04:39 * kergoth doesn't recall what, should fire one off again Mar 23 16:05:04 fwiw I use 'buildall' and console-image, opie-image and x11-image are now building for arm. Now I'll reenable oestats (was too ashamed by the ton of errors) Mar 23 16:09:16 re Mar 23 16:22:40 ~curse DEPLOY_DIR change in minimal distro Mar 23 16:22:41 May you be reincarnated as a Windows XP administrator, DEPLOY_DIR change in minimal distro ! Mar 23 16:23:35 hey denix Mar 23 16:23:46 hey Mar 23 16:24:07 hrw: what's with the DEPLOY_DIR? Mar 23 16:24:29 1I have tmp/deploy/glibc/* and tmp/deploy/* now Mar 23 16:26:12 do you set it in your local.conf? Mar 23 16:26:24 and I wondered why I have wrong images Mar 23 16:26:51 denix: I do not use such vars in user configs Mar 23 16:27:18 my conf/* is 16 lines Mar 23 16:28:01 he, the second one is in bitbake.conf Mar 23 16:28:01 the thing is DEPLOY_DIR is now being weak-assigned in bitbake.conf and minimal.conf :( Mar 23 16:28:18 denix :-) Mar 23 16:28:19 03Koen Kooi  07org.openembedded.dev * r5261978d6f 10openembedded.git/recipes/busybox/ (busybox.inc busybox_1.13.2.bb): Mar 23 16:28:19 busybox: fix linuxrc link and put it in a seperate package Mar 23 16:28:19 * runtime tested on a beagleboard Mar 23 16:29:03 hrw: pls fix it once-for-all :-) Mar 23 16:29:49 Tom's idea in ML sounds good (USER_DEPLOY_DIR) Mar 23 16:30:13 ant_work: that's a workaround - needs to be put in every distro :( Mar 23 16:30:53 denix: no, only in ones which want to support it Mar 23 16:30:55 hi Liam Mar 23 16:31:16 well, having weak assign's in bitbake.conf and distro.conf obliges user to specify it in local.conf, otherwise Mar 23 16:31:30 no Mar 23 16:31:37 it does not oblige Mar 23 16:31:43 it allows Mar 23 16:32:23 hmm...then some distro (Angstrom) would not find pstage Mar 23 16:32:32 shit... Mar 23 16:32:58 qemu forgot how to cooperate with desktop when it comes to mouse<>touchscreen Mar 23 16:33:42 ant_work: why is that? Mar 23 16:34:20 I suppose because of the mising /*libc suffix Mar 23 16:35:20 >[17:24] 1I have tmp/deploy/glibc/* and tmp/deploy/* now Mar 23 16:35:34 (this was not Angstrom it seems) Mar 23 16:35:57 ant_work: I think hrw had 2 because he never cleans his deploy... Mar 23 16:36:48 that's minimal, but it is the same as Angstrom - $TMPDIR/deploy/$LIBC for both Mar 23 16:36:54 I have to retest, however I think rebuild from pstage is only working if DEPLOY_DIR is in TMPDIR Mar 23 16:37:02 while bitbake.conf sets it to just $TMPDIR/deploy Mar 23 16:37:59 ant_work: Hmm? why is that. I had my DEPLOY_DIR outside ot TMPDIR for half year (with the fix to base.bbclass) and it worked just fine Mar 23 16:46:41 its minimal distro Mar 23 16:46:57 and yes, builddir is more then 2 weeks old Mar 23 16:56:49 I'm experiencing a very strange bug with dbh_1.0-18.bb: http://pastebin.com/d24f001fe The same recipe was working before I upgraded yesterday my OE tree. So I don't think that it's the recipe. I debugged in python and strangely there is an empty string that appears when the SRC_URI is splited in do_unpack. oe_unpack is then passed with "", which is converted to ., hence trying to unpack the directory itself /OE/build/tmp/work/ Mar 23 17:01:47 that's .. strange Mar 23 17:05:15 it would be great if someone can do "bitbake dbh" on his system, so that at least I know if it's my system or if there is a bug in the git head Mar 23 17:07:38 building Mar 23 17:09:04 18:07 hrw@home:build$ bitbake -b /home/hrw/devel/kaeilos/openembedded/recipes/dbh/dbh_1.0-18.bb -e|grep ^SRC_ Mar 23 17:09:07 SRC_DISTRIBUTEDIR="/home/hrw/devel/kaeilos/build/tmp/deploy/glibc/sources" Mar 23 17:09:10 SRC_URI="http://downloads.sourceforge.net/dbh/dbh_1.0-18.tar.bz2 file://${FILESDIR}/configure.patch;patch=1" Mar 23 17:09:23 but do_distribute_sources failed Mar 23 17:09:57 ... yikes, why would someone make a file: url like that? Mar 23 17:10:02 file: already took care of that crap Mar 23 17:10:13 no idea Mar 23 17:10:24 * hrw -> off for some time Mar 23 17:10:51 gregoiregentil: if you remove the ${FILESDIR}/ part of the SRC_URI, does that help? Mar 23 17:12:01 03Marcin Juszkiewicz  07org.openembedded.dev * r74b803d675 10openembedded.git/recipes/netbase/ (3 files in 3 dirs): netbase: added eth0=192.168.7.2 for qemuarm/qemux86 (from Poky) Mar 23 17:12:26 kergoth: it does Mar 23 17:14:04 * kergoth grumbles and starts going through all remaining FILESDIR refs.. didn't intend to scrap its use entirely, you'll note it still exists in FILESPATH Mar 23 17:14:05 yes. if there is one line for SRC_URI with no \, it works indeed Mar 23 17:16:55 OK. it works. It was indeed the ${FILESDIR}. Arghhh! so it would be great to search for other similar occurences. Thanks for your help to fix this (at least it was a bug in the tree, not on my system) I have not wasted time of everybody Mar 23 17:18:15 yep, it wasn't your fault, it was mine, sorry about that Mar 23 17:19:14 if a recipe sets FILESDIR, it'll be used, and things won't break. its when a recipe tries to use it specifically for some reason, and doesn't set it, that there's a problem. i'll see about fixing those. Mar 23 17:23:35 looks like from recipes/linux/linux.inc the concept for machine specific defconfigs is to put them in $MACHINE/defconfig vs defconfig.$MACHINE - is the 2nd method considered deprecated? Mar 23 17:24:24 i don't remember the latter ever being the correct way of doing it Mar 23 17:25:37 thats what I figured... I just notice that there are a lot of $MACHINE.defconfig files out there under recipes/linux/... Mar 23 17:26:09 silly people Mar 23 17:32:20 anyone have an opinion on our gnu-configize stuff? seems a bit unnecessary given automake can already update one of the scripts, just not the other... guess it could be for autoconf but not automake packages Mar 23 17:32:23 * kergoth scratches head Mar 23 17:36:13 ah, there's an old comment on it Mar 23 17:43:08 hmm, doesn't look like intltoolize obeys AC_CONFIG_SUBDIRS. that's not good Mar 23 17:50:27 kergoth: gnu-configize recreates all necessary files if you want to regenerate configure from scrarch Mar 23 17:50:31 scratch Mar 23 17:50:46 khem: i know what it is, i created it. Mar 23 17:50:53 gnu-configize just copies config.sub, config.guess into place Mar 23 17:50:56 we patch autoreconf to call it Mar 23 17:53:36 yes autoreconf should be same Mar 23 17:53:48 you can punt it then if you want Mar 23 17:53:57 and replace it with autoreconf Mar 23 17:54:45 what? Mar 23 17:54:47 hmm autoconf prolly need it Mar 23 17:54:52 i think you're confused. Mar 23 17:55:04 yes I am Mar 23 17:55:38 * kergoth is just looking to simplify some things that seem oddly complex Mar 23 17:55:58 kergoth: why would you need to copy these files around ? Mar 23 17:56:12 configure scripts require config.sub and config.guess if they call the build/host/target macros Mar 23 17:56:35 automake and libtoolize will update them if needed, but not all packages use automake or libtool, some just use autoconf Mar 23 17:56:42 hence the separate script to do it Mar 23 17:58:25 autoreconf should call libtoolize Mar 23 17:58:40 if autoreconf is supported by package Mar 23 17:58:54 what? Mar 23 17:59:02 autoreconf does call libtoolize, if the package uses libtool. Mar 23 17:59:54 hmm I see so the packages you are talking are ones which just use autoconf but nothing else Mar 23 18:00:10 from autotools Mar 23 18:00:24 right. if a package uses just autoconf, and calls AC_CANONICAL_HOST, then its config.sub and config.guess wont' get updated by autoreconf normally Mar 23 18:00:38 but isnt that broken then Mar 23 18:00:54 i'd think so, yeah. probably something to talk about with upsream Mar 23 18:00:56 upstream, even Mar 23 18:01:08 do you know such a package I can poke into Mar 23 18:01:18 not offhand Mar 23 18:01:48 may be if I look for this gnu_configize thing Mar 23 18:01:58 i just made one.. just made a hello world .c, ran autoscan, moved that to configure.ac, added AC_CANONICAL_HOST, then ran autoreconf -f -i and it didn't install the files that configure needs to function Mar 23 19:11:10 there we are, got glib-gettextize and intltoolize running against all subdirs, not just ${S} Mar 23 19:11:31 03Khem Raj  07org.openembedded.dev * ra8f709faf5 10openembedded.git/: Merge branch 'org.openembedded.dev' of git@git.openembedded.net:openembedded into org.openembedded.dev Mar 23 19:11:58 03Khem Raj  07org.openembedded.dev * r45fb05982c 10openembedded.git/conf/distro/include/sane-srcrevs.inc: sane-srcrevs.inc: Bump the sane srcrev for uclibc. Mar 23 19:13:46 03Khem Raj  07org.openembedded.dev * rc1ab14e9ab 10openembedded.git/: Merge branch 'org.openembedded.dev' of git@git.openembedded.net:openembedded into org.openembedded.dev Mar 23 19:14:01 03Khem Raj  07org.openembedded.dev * rb45c90c449 10openembedded.git/: Merge branch 'org.openembedded.dev' of git@git.openembedded.net:openembedded into org.openembedded.dev Mar 23 19:14:05 03Khem Raj  07org.openembedded.dev * r8d0b0994a8 10openembedded.git/recipes/uclibc/ (9 files in 2 dirs): (log message trimmed) Mar 23 19:14:05 uclibc: Get uclibc working in thumb mode. Mar 23 19:14:05 * WARNING!! Do not use bintils 2.19 or 2.19.1 Mar 23 19:14:05 when compiling in thumb mode. The veneer generation Mar 23 19:14:07 does not work as expected. Use binutils 2.18 or binutils_cvs Mar 23 19:14:07 heh.. looks like I should rather avoid touching patchwork Mar 23 19:14:09 * Move LEAD_SONAME into uclibc.inc Mar 23 19:14:11 * Use no optimizations while compiling ldso.c Mar 23 19:39:49 hi to all ... Mar 23 19:40:15 hey Mar 23 19:41:48 hrw, what is the problem? Mar 23 19:42:16 Crofton|work: I changed mtd patches state to rfc and they are gone Mar 23 19:42:28 Crofton|work: and I cannot find any other of my patches Mar 23 19:43:14 upper left corrner Mar 23 19:43:19 adjust filter Mar 23 19:43:24 http://patchwork.openembedded.org/project/openembedded/list/?state=5 Mar 23 19:43:55 argh Mar 23 19:44:17 I would not call it userfriendly Mar 23 19:44:56 :) Mar 23 19:45:01 Crofton|work: and how to merge my email addresses with account? Mar 23 19:45:14 ask an admin Mar 23 19:45:15 :) Mar 23 19:45:25 Crofton|work: now I have todo(0) so looks like I have to select all patches and delegate them to me? Mar 23 19:45:36 not sure Mar 23 19:48:24 hmm Mar 23 19:48:50 well, since nobody argued against it, i guess we should start actually using MAINTAINER again Mar 23 19:49:03 Crofton|work: any way to save own queries? (other then bookmark all of them) Mar 23 19:49:08 * kergoth thinks Mar 23 19:49:33 bookmark the url maybe Mar 23 19:49:47 I am in usb cable hades Mar 23 19:49:56 need to attach a mini-A to regular B Mar 23 19:50:02 ;D Mar 23 19:54:56 Crofton|work, some kind of "enlarge your plug" pills ??? Mar 23 19:55:14 yeah Mar 23 19:55:23 I have the mini-a to mini-b Mar 23 20:05:45 Crofton|work: I think I have the right cables here to attach any usb to any usb :-) Mar 23 20:07:37 I lack usb host cable again. gave last one with sl-6000 Mar 23 20:09:02 I have mini-a mini-b and micro-b to normal A and an A->A adapter :-) Mar 23 20:10:07 anyone looking for a 12.3 inch widescreen monitor... awesom little piece of kit from meos Mar 23 20:10:17 monitoor dvd player tv etc. all in one Mar 23 20:10:32 Spyro: url and price? Mar 23 20:10:36 (Im usinig one for my linux dashboard project for my car :) Mar 23 20:10:50 Spyro: 1280x800 at least? Mar 23 20:12:27 http://www.wedodigital.co.uk/product/2068/12-lcd-tv/meos-dvd123b/ Mar 23 20:12:38 180 ish Mar 23 20:12:54 and yes Mar 23 20:12:57 1280x800 Mar 23 20:13:20 looks like the perfect fit for a golf Mk III dashboard module :) Mar 23 20:26:39 http://marcin.juszkiewicz.com.pl/2009/03/23/thinking-about-lcd-monitor-change/ Mar 23 20:31:35 * Jay7 remember some touchscreen monitors by Prology Mar 23 20:32:16 * hrw -> sleep Mar 23 20:52:38 I think I'll go through a hub Mar 23 21:02:47 Crofton|work I ordered such a cable from digitalstreet on ebay, haven't received it yet though, so no confirmation it works with my beagle Mar 23 21:03:18 let me know how it turns out Mar 23 21:03:31 actually it is digistreet, see http://www.auctiva.com/stores/viewstore2.aspx?id=158275&page=home&search=otg Mar 23 21:03:43 i hope to get it end of the week or so ordered it mid last week Mar 23 21:03:57 ping me if i forget to mention it Mar 23 21:04:38 will do Mar 23 21:04:47 I do not want to carry around an extra hub Mar 23 21:23:45 anybody seeing 'sh: rm: command not found' scrolling during build? it seems just before task do_setscene: completed Mar 23 21:25:15 ah..this is fresh :-) NOTE: the following files were installed but not shipped in any package: Mar 23 21:25:16 NOTE: linuxrc Mar 23 21:25:23 koen? Mar 23 21:26:39 got error building finale image "genext2fs: group descr 10. free blocks count == 0 (corrupted fs?)" ... mean have to increase image size or what ? Mar 23 21:34:56 koen is not here Mar 23 21:35:37 http://pastebin.com/m51dd7daf Mar 23 21:35:55 cannot copy a directory onto itself tupe error .... Mar 23 21:36:39 * kergoth looks Mar 23 21:37:06 ant__: I have seen Mar 23 21:37:29 updating also Mar 23 21:37:43 but seems it was harmless Mar 23 21:38:05 Jay7: http://rafb.net/p/5V3k4p88.html Mar 23 21:38:21 it seems some FILES_PN or something to check Mar 23 21:38:53 he..exactly like when we were deploying kexecboot Mar 23 21:39:19 Crofton|work: is that linux-omap_git.bb? Mar 23 21:39:25 yeah Mar 23 21:39:31 k Mar 23 21:39:43 I am addin ga ehci patch for overo Mar 23 21:40:13 * kergoth waits for the do_fetch Mar 23 21:40:37 crash! Mar 23 21:40:44 * mwester adds sound effects to the do_fetch ;) Mar 23 21:41:30 fail with current Mar 23 21:42:15 re Mar 23 21:42:37 * Jay7 is testing 2 SD cards for r and w speed Mar 23 21:44:17 sheesh, this is a long fetch :) Mar 23 21:44:44 kergoth: which package are you fetching Mar 23 21:45:00 one of the kernel git recipes. unsurprising Mar 23 21:45:14 oh Mar 23 21:45:26 git checkout then tarring Mar 23 21:45:32 aye Mar 23 21:45:59 for me when I check the mips-linux git or arm git it takes forever Mar 23 21:46:13 kernel.org is much faster Mar 23 21:46:32 seems that 0.6Mb/s is max write speed to SD for zaurus hardware Mar 23 21:46:39 ah Mar 23 21:46:43 * kergoth thinks someone should really rethink the whole git fetch/unpack methodology Mar 23 21:46:49 reading speed is ~1.9Mb/s Mar 23 21:46:53 the patch I addded is not in the linux-omap directory Mar 23 21:47:10 Jay7: so slow? Mar 23 21:47:11 so, SD 60x card is enough for zauruses :) Mar 23 21:47:30 ant__: I wondering that too.. Mar 23 21:47:35 Crofton: figured it was something like that, the change i made to oe_unpack_file needs to abort if it fails to locate it, it doesn't seem to be doing that at the moment.. Mar 23 21:47:47 deceptive error messages < * Mar 23 21:48:11 ant__: but this is FAT16.. ext2 should be faster Mar 23 21:48:20 right Mar 23 21:48:36 well, checking now Mar 23 21:48:50 otavio: ping Mar 23 21:49:11 otavio: you forgot to fix the psplash thing Mar 23 21:49:12 I wish we could get someone to work through bitbake and fix all the "difficult" error messages Mar 23 21:49:19 ant__: I see the 'sh: rm: command not found' too Mar 23 21:49:35 but it has never prevented me from completing a build Mar 23 21:49:41 only when rebuilding from pstage? Mar 23 21:49:59 Crofton|work: i've been thinking we should audit all the messages that OpenEmbedded and bitbake emit in general. lots of the classes are -way- more verbose than they should be, giving the user info they're not likely to ever need or want Mar 23 21:50:37 ant__: that could be, can not try right now Mar 23 21:50:44 here it happens just on 3 packages atm Mar 23 21:51:08 ant__: rebuild from pstage works well for me Mar 23 21:51:10 one of them 'package-index' :-) Mar 23 21:51:17 Crofton|work: with recent bitbake, there's no reason to dump the task started/completed messages anymore, since it says what task its executing.. Mar 23 21:51:21 the other 2 zaurus-specific Mar 23 21:51:33 ant__: my only issue is still the opkg-native Mar 23 21:51:37 kergoth: please do... Mar 23 21:51:50 kergoth: it's really too much log Mar 23 21:52:13 I always must remove it before starting a rebuild from pstage Mar 23 21:52:31 mpoullet: interestibgly I did'nt yet stumble in it... Mar 23 21:52:48 not my day Mar 23 21:52:53 mpoullet: pls try package-index Mar 23 21:52:54 Configuring staging-package-index-armv5te-angstrom-linux-gnueabi Mar 23 21:52:55 sh: rm: command not found Mar 23 21:54:33 ant__: must switch machine to do this, wait a minute Mar 23 21:57:57 kergoth: this is bitbake virtual/kernel from pstage Mar 23 21:58:03 http://rafb.net/p/kwiaRM88.html Mar 23 22:00:00 mpoullet|home: wb Mar 23 22:00:17 ant__: http://pastebin.com/d14a71753 Mar 23 22:00:22 mpoullet|home: now I see the infamous do_rootfs) failed Mar 23 22:00:25 :/ Mar 23 22:00:37 I don't see the issue with package-index Mar 23 22:00:55 but I'm sur that I've already seen this! Mar 23 22:00:55 I have it with virtual/kernel too... Mar 23 22:01:25 ant__: it was a rebuild from pstage though Mar 23 22:01:43 you only left pstage dir? isn't? Mar 23 22:02:53 I save the DEPLOY_DIR as explained by Koen on the m-l Mar 23 22:03:29 try to delete all but /pstage Mar 23 22:03:36 looks like stage-manager is confused here Mar 23 22:04:10 here is my setup http://pastebin.com/d4e23d3c3 Mar 23 22:04:56 mpoullet|home: for the last build I left pstage in TMPDIR Mar 23 22:05:41 ant__: I've pstage outside of TMPDIR for ~3 days, seems to work Mar 23 22:05:47 btw INHERIT += "packaged-staging" seems superfluous Mar 23 22:06:37 yes, it works...but weak-assigning DEPLOY_DIR in angstrom.inc has the only result of defaulting back to bitbake.conf value Mar 23 22:06:53 so you need to specify it in local.conf Mar 23 22:07:00 ..and that's bad Mar 23 22:07:15 Tartarus proposed a new variable Mar 23 22:07:24 I'd like to read Koen's opinion Mar 23 22:07:52 ant__: why is that bad? Mar 23 22:07:52 * mpoullet|home don't know oe good enough to know what it's bad Mar 23 22:07:52 yes Mar 23 22:07:57 Because, imho, it's a distro problem Mar 23 22:08:28 I mean, the patch should not oblige people to set it in locval.conf Mar 23 22:10:01 ant__: rebuild of package-index after having removed tmp dir: http://pastebin.com/d5b6a6938 Mar 23 22:10:11 mpoullet|home: anyway, now, on second rebuild from pstage, I see your prob Mar 23 22:10:14 touch: cannot touch `/oe/build/tmp/angstrom/deploy/ipk/Packages': No such file or directory Mar 23 22:11:09 * mpoullet|home feels not alone :-D Mar 23 22:11:24 btw the file is there Mar 23 22:14:15 I notice that busybox 1.13.2 (current angstrom preferred version) bails out with 'SIOCSIFFLAGS: Cannot assign requested address' when you add something like 'iface eth0:0 inet dhcp' to /etc/network/interfaces - refuses to run udhcpc on it b/c of the failure Mar 23 22:14:54 tharvey: you tried 13.3 and see if it is fixed? Mar 23 22:15:33 not yet... was wondering if anyone else had run into this - it seems that 'ifconfig eth0:0 up' always produces an error, but older busbyox didn't bail out and ran udhcpc anyway Mar 23 22:16:16 ant__: rebuild fron package-index, no tmp, deleted the ipk too: http://pastebin.com/d40df9d0 Mar 23 22:16:21 still no error Mar 23 22:16:31 mpoullet|home: I inherit 'rm_work' Mar 23 22:17:52 ant__: what do you mean? should I try this too? Mar 23 22:18:13 I'll try without :-) Mar 23 22:18:41 it's necessary to fit /tmp in RAM Mar 23 22:18:42 ok, got it, why do you inherit this? Mar 23 22:18:57 ant__: ok, thanks for the info Mar 23 22:19:38 generally for disk-space. Don't use it when patching :-) Mar 23 22:20:35 I've a git question regarding the move packages->recipes Mar 23 22:21:29 I've commited some changes in recipes in packages and now when merging I've a conflict of the form you've modified/they've deleted Mar 23 22:22:43 because e.g. packages/xyz/xyz.bb is now removed and I've commited changes there Mar 23 22:23:27 so what is the cleanest merge strategy in such cases? so that my changes get merge in the new location? Mar 23 22:23:33 if that branch isn't public, you could always rebase your changes instead of merging, then you wont push so many merge commits, and i think rebase is a bit smarter about the rename Mar 23 22:23:36 * kergoth shrugs, not sure Mar 23 22:28:06 btw: any ideas (people to contact / links to read) regarding my opkg-native issue? Mar 23 22:29:22 nobody have commented on it on the m-l, but I won't give up :) Mar 23 22:31:06 mpoullet|home: there are other problems... Mar 23 22:31:36 see what happens if you -c clean and rebuild virtual/kernel Mar 23 22:31:38 http://rafb.net/p/sj0ySP72.html Mar 23 22:31:53 after having wiped /tmp Mar 23 22:33:43 ant__: it might help: http://lists.linuxtogo.org/pipermail/openembedded-devel/2009-March/009005.html Mar 23 22:35:46 ant__: I've seen this too before Mar 23 22:36:29 mpoullet|home: I'm testing angstrom with DEPLOY_DIR ?= "${TMPDIR}/deploy" Mar 23 22:36:43 no joy Mar 23 22:36:59 now trying USER_DEPLOY_DIR Mar 23 22:37:01 ant__: here is my DEPLOY_DIR structure: http://pastebin.com/d7d2b7db3 Mar 23 22:37:15 it's called archives Mar 23 22:37:25 and I save pstage + ipk Mar 23 22:37:37 images doesn't matter Mar 23 22:37:49 can be wiped out like tmp Mar 23 22:38:41 * mpoullet|home is doing a rebuild of virtual/kernel for ant__ Mar 23 22:38:42 ipk too Mar 23 22:38:51 can be deleted Mar 23 22:39:09 ant__: I don't think so Mar 23 22:39:18 try :-) Mar 23 22:40:34 ant__: virtual/kernel works Mar 23 22:40:49 * khem never used staging Mar 23 22:41:03 ant__: you need the ipks to populate the staging Mar 23 22:41:07 from pstage Mar 23 22:43:14 ant__: as long as I remember it, I had the same issues with you e.g. with the kernels as long as I hadn't saved the ipk dir Mar 23 22:44:25 I mean after havin done what Koen said it worked far better Mar 23 22:44:27 ? Mar 23 22:44:42 the .ipk are in /pstage/* Mar 23 22:44:54 the staging.ipk Mar 23 22:45:34 I've rebuilt in that way one hour ago Mar 23 22:45:50 bur deleting kernel brought new problems Mar 23 22:46:25 I notice there are a bunch of .mtn2git_empty files still in oe-git Mar 23 22:46:34 hmm if they are there then it should not require Mar 23 22:46:47 anything else Mar 23 22:48:08 ant__: rebuild from virtual/kernel after removing tmp: http://pastebin.com/d3439f0c5 Mar 23 22:48:36 but you kept deploy/ipk Mar 23 22:48:42 I suppose Mar 23 22:48:55 ant__: yep, is that wrong? Mar 23 22:49:07 03Chris Larson  07org.openembedded.dev * r850ea31d59 10openembedded.git/classes/base.bbclass: Mar 23 22:49:07 base.bbclass: Abort early if localpath() was unable to find a local file for the url. Mar 23 22:49:07 Signed-off-by: Chris Larson Mar 23 22:49:17 03Chris Larson  07org.openembedded.dev * r6ddd16d877 10openembedded.git/classes/base.bbclass: Mar 23 22:49:17 base.bbclass: rstrip the git branch/revision. Mar 23 22:49:17 Signed-off-by: Chris Larson Mar 23 22:49:33 Crofton|work: that error shouldn't happen now, it'll abort with a proper error Mar 23 22:49:42 hmm looking at the staging packages the *-initial-* and *-intermediate-* are not required I think Mar 23 22:49:47 thanks Mar 23 22:50:05 my actual problem now is the patch doesn't apply :) Mar 23 22:50:09 but not your fault Mar 23 22:50:18 kergoth: what's the bare minimum to keep for packaged-staging builds? Mar 23 22:50:23 hehe Mar 23 22:50:38 ant__: not sure. i've done builds in the past only keeping the pstage/pkgs dir around Mar 23 22:50:39 stupid usb code in omap kernel Mar 23 22:52:03 kergoth: so I suppose it looks if pstage dir is there if not then only build otherwise just install the one from pstage ? Mar 23 22:52:35 no, if the pstage package for that recipe doesn't exist, then it does a normal build Mar 23 22:52:38 its not conditional on the dir itself Mar 23 22:52:44 see the packaged-staging.bbclass Mar 23 22:53:27 * mpoullet|home rebuilds virtual/kernel with pstage only Mar 23 22:54:00 * kergoth goes to put FILESDIR back for the time being, since he didn't intend that breakage Mar 23 22:56:08 so many recipes using FILESDIR? Mar 23 22:57:06 not a huge amount, but its the principle of the thing, it should've been marked deprecated before complete removal Mar 23 22:57:48 ant__: I must sleep now, but that would be great if we could put all the packaged staging infos somewhere Mar 23 22:58:08 ant__: like a patch for the docs or something like that Mar 23 22:58:49 ant__: I've removed tmp and archives/ipk + archives/images Mar 23 22:59:12 should not be too long Mar 23 22:59:16 ant__: I've redone a build a virtual/kernel from pstage only Mar 23 22:59:25 ant it has worked Mar 23 22:59:33 s/ant/and Mar 23 22:59:39 good Mar 23 22:59:50 03Chris Larson  07org.openembedded.dev * r9bb0079fdb 10openembedded.git/conf/bitbake.conf: Mar 23 22:59:50 bitbake.conf: resurrect FILESDIR, didn't intend its complete removal just yet. Mar 23 22:59:50 Signed-off-by: Chris Larson Mar 23 22:59:58 if you feel brave, bitbake -c clean virtual/kernel Mar 23 23:00:05 and rebuild it Mar 23 23:00:13 ant__: but I'm sure that I add your issue with the kernel before Mar 23 23:00:20 ant__: can't reproduce it now Mar 23 23:00:23 I'm not Mar 23 23:00:40 anyway thx Mar 23 23:01:06 ant__: I'm trying that tomorrow Mar 23 23:01:10 gb folks Mar 23 23:01:28 * mpoullet|home the quest for packaged-staging continues! Mar 23 23:01:34 'nacht Mar 23 23:01:55 neverending story :) Mar 23 23:18:22 Jay7: btw mpoullet had 'sh: rm: command not found' too in his logs on ML :-) Mar 23 23:42:19 Hmm, I see 2.6.29 now Mar 23 23:42:37 I should really make an mx31pdk or mx313stack machine conf and 2.6.29 recipe Mar 24 00:02:55 are classes searched for in BBFILES? I'm trying to use a collection to provide an override for a class and it seems to always want to pull the class from my upstream collection Mar 24 00:03:39 sorry, I mean are classes searched for in BBPATH Mar 24 00:38:22 ah I see my error... my BBPATH has the wrong path order Mar 24 01:29:22 03Angus Ainslie  07fso/milestone5.5 * r59fd9ef421 10openembedded.git/packages/efl1/ecore.inc: ship ecore-input for python ecore bindings Mar 24 01:29:32 03Angus Ainslie  07fso/milestone5.5 * rbff3555824 10openembedded.git/: Merge branch 'fso/milestone5.5' of git@git.openembedded.net:openembedded into fso/milestone5.5 Mar 24 01:29:33 03Angus Ainslie  07fso/milestone5.5 * r4ab6d86b56 10openembedded.git/conf/distro/include/sane-srcrevs.inc: Bump EFL version to latest for new python bindings **** ENDING LOGGING AT Tue Mar 24 02:59:57 2009