**** BEGIN LOGGING AT Sat May 02 02:59:57 2009 May 02 05:23:37 03Roman I Khimov  07org.openembedded.dev * r7eef95e87b 10openembedded.git/recipes/bridge-utils/ (3 files in 2 dirs): May 02 05:23:37 bridge-utils: remove old versions May 02 05:23:37 Version 1.0.4 FTBFS and both with 0.9.7 are very old. No distro sets any May 02 05:23:37 version as preferred, so it's safe to assume that everyone is on 1.2 now. May 02 05:23:37 Acked-by: Otavio Salvador May 02 05:23:47 03Roman I Khimov  07org.openembedded.dev * red293046eb 10openembedded.git/ (4 files in 2 dirs): May 02 05:23:47 bridge-utils: add version 1.4 May 02 05:23:47 Acked-by: Otavio Salvador May 02 07:10:26 03Mike Westerhof  07org.openembedded.dev * rf8f9b7e4b4 10openembedded.git/conf/machine/ (nslu2be.conf nslu2le.conf): May 02 07:10:26 nslu2 machine configs: do not inherit machine-specific image class May 02 07:10:26 The creation of the NSLU2 full flash image (.bin) file should be May 02 07:10:26 image-specific, since not all images for machines similar to the May 02 07:10:26 NSLU2 have the same flash constraints. May 02 07:23:34 good morning May 02 07:24:29 morning mckoan May 02 08:33:46 giorno valhalla ! May 02 08:34:01 hi gremlin[it] May 02 08:36:52 good morning Italy :) May 02 08:45:18 morninf Jay7 ! May 02 08:48:49 morning JaMa May 02 10:35:48 03Graeme Gregory  07org.openembedded.dev * r91c920197a 10openembedded.git/recipes/uclibc/ (files/unifdef-fix.patch uclibc_0.9.30.bb): May 02 10:35:48 uclibc_0.9.30.bb : add fix for unifdef for modern gcc/glibc combos. May 02 10:35:48 This is probably suitable for all uclibc when someone tests them. May 02 12:15:03 pxa2xx-udc pxa2xx-udc: UDC connecting May 02 12:15:31 this is last message before akita's x11-image in qemu hangs May 02 14:51:30 nicklist fifo May 02 14:52:02 quit May 02 16:31:20 <_cpo_> hmm anyone knows why meta-toolchain does not build an installable toolchain? May 02 16:31:36 <_cpo_> it bitbakes a lot of packages but i dont get any tar.bz2 with the toolchain in the end May 02 16:32:57 It doesn't? May 02 16:33:02 tmpdir/deploy/sdk/ May 02 16:33:11 Or tmpdir/deploy/glibc/sdk, if using angstrom May 02 16:34:29 <_cpo_> Tartarus: no, thought so, too. May 02 16:34:37 <_cpo_> im using the minmal distro ... May 02 16:34:49 <_cpo_> but i dont think thats the reason May 02 16:36:46 And meta-toolchain completed? May 02 16:37:09 The last part in do_populate_sdk is to make a tarball and place it into ${SDK_DEPLOY}/ May 02 16:37:20 which is ${DEPLOY_DIR}/sdk/ May 02 16:37:32 distro-independant and all May 02 17:16:19 <_cpo_> seems that the curl-sdk package has an error May 02 17:18:32 really? May 02 17:18:36 dev or stable? May 02 17:19:07 I've had both build meta-toolchain/canadian-sdk in the last 24h or so May 02 17:19:35 <_cpo_> stable May 02 17:20:26 Sigh, I've still got updates that need ack'ing for stable May 02 17:20:29 but it should be OK May 02 17:20:55 <_cpo_> i just cleaned a lot of packages and trying to rebuild May 02 17:23:57 03Mike Westerhof  07org.openembedded.dev * r3965cd0064 10openembedded.git/ (2 files in 2 dirs): SlugOS: images - create nas100d and dsmg600 .bin firmware files with slugos-extd image. May 02 17:25:03 Tartarus: I'd ack them, but they are far, far beyond my ability to understand, so the ack would be meaningless :( May 02 17:25:09 <_cpo_> Tartarus: m128b843b May 02 17:25:12 <_cpo_> still failing May 02 17:25:56 <_cpo_> i mean http://pastebin.com/m128b843b May 02 17:28:43 Hmm May 02 17:28:58 _cpo_, is your stable up to date? I thought the fix for that made it into stable already May 02 17:29:10 <_cpo_> Tartarus: pulled 10 minutes ago May 02 17:29:25 mwester, well, khem ack'd them, and I just need a 2nd one from someone May 02 17:29:29 _cpo_, very odd.. May 02 17:29:34 <_cpo_> ok than i have to backmerge from dev May 02 17:29:38 <_cpo_> to my overlay May 02 17:30:05 _cpo_, well, in the oe git dir try git cherry-pick 5c69bdef4ca0d8413477db32cc82b82395806504 May 02 17:30:15 http://cgit.openembedded.net/cgit.cgi?url=openembedded/commit/&id=5c69bdef4ca0d8413477db32cc82b82395806504 is the fix for that problem, in dev May 02 17:30:25 <_cpo_> thx May 02 17:40:22 <_cpo_> Tartarus: that did it. would be nice to see this one in stable May 02 17:43:09 <_cpo_> also nice May 02 17:43:10 <_cpo_> | * ERROR: Cannot satisfy the following dependencies for task-sdk-bare: May 02 17:43:10 <_cpo_> | * locale-base-en-gb * May 02 17:46:53 03Mike Westerhof  07org.openembedded.dev * rf567230526 10openembedded.git/classes/nas100d-image.bbclass: SlugOS: nas100d-image.bbclass - use nslu2 kernel; we no longer generate machine-specific kernels. May 02 17:49:30 That's odd May 02 17:49:36 Did you disable all locale stuff? May 02 17:50:01 <_cpo_> just enable binary_locale_generation and retesting May 02 17:50:27 Nope, that's not it May 02 17:50:35 my buildbot has that off May 02 17:51:01 <_cpo_> uhm ... other ideas? May 02 17:51:09 <_cpo_> or just remove the locale dependency :) May 02 17:53:04 What machine are you using? May 02 17:53:11 and libc? May 02 17:53:36 It's been a while since I did eglibc build, but that should be fine, I think I fixed the rprovides errors there May 02 17:53:56 <_cpo_> minimal distro, qemuarm, glibc May 02 17:54:16 very odd May 02 17:54:20 lemme go kick something off then May 02 17:57:41 ok, pulled, using normal stable/2009, ie w/o my updates May 02 17:57:51 kicking a build now of minimal/qemuarm May 02 17:57:55 <_cpo_> kthx May 02 17:58:03 got about an hour before my next buildbot kicks off May 02 17:58:12 but this should get a good bit done in that time at least May 02 17:58:33 <_cpo_> just removed the locale now and thats ok May 02 17:59:23 This should all be fine out of the box tho :) May 02 17:59:31 mwester, ping May 02 17:59:46 <_cpo_> Tartarus: "should" :) May 02 18:00:01 <_cpo_> upcomming week will be fun by changing from old stable to stable2009 :) May 02 18:00:06 _cpo_, well, and it does for me on some combinations of machines :) May 02 18:00:21 <_cpo_> Tartarus: could be an issue with the qemuarm machine May 02 18:00:41 that shouldn't matter for meta-toolchain May 02 18:00:49 at91sam9263ek builds fine May 02 18:00:55 and both are armv5te May 02 18:01:30 <_cpo_> strange May 02 18:01:32 http://tucson.is.crashing.org:8010/builders/angstrom_glibc_full_stable2009_build/builds/0/steps/shell_2/logs/stdio May 02 18:14:34 03Mike Westerhof  07org.openembedded.dev * r6b0e2b924e 10openembedded.git/recipes/images/ (slugos-extd-image.bb slugos-image.inc): SlugOS: make sym-links to the images image-specific. May 02 18:16:45 mwester, wearing your slugos hat can you ack the LIBC change for dev? May 02 18:19:44 Ok, I'll apply that and test. May 02 18:25:40 thanks May 02 18:25:58 nothing popped out in a bitbake -g at least, so it shouldn't be a change really May 02 18:40:50 I see a lot of stuff in patchwork; which one specifically is that? May 02 18:49:55 trini/add-libc-overrides-2 almost applies; unslung.conf conflicts. May 02 18:50:15 I'll ignore that conflict, and see what happens with the result... May 02 18:50:56 Also in a branch May 02 18:52:35 _cpo_, stable/2009 built for me May 02 18:52:51 Ok, starting a fresh slugosle image build... May 02 18:56:58 _cpo_, sample local.conf w/ BB_NUMBER_THREADS=4/PARALLE_MAKE="-j6" May 02 18:57:05 and angstrom mirrors stuff thrown in to combat 404s May 02 19:44:35 <_cpo_> Tartarus with minimal distro and qemuarm image? May 02 19:45:02 yeap May 02 19:45:46 -rw-r--r-- 1 trini trini 40354135 May 2 11:50 tmp/deploy/sdk/minimal-.dev-snapshot-20090502-armv5te-linux-gnueabi-toolchain.tar.bz2 May 02 19:46:02 openembedded$ git branch | grep ^* May 02 19:46:02 * stable/2009 May 02 19:46:23 $ git log | head -n 1 May 02 19:46:23 commit e264996efb684cfad7c196ea19be1b1dc060dccc May 02 20:46:17 Bah, that's right, I need to move the uclibc stuff to stable May 02 20:46:23 after I get that 2nd ack on the first round of changes May 02 20:46:24 anyone? May 02 20:48:24 can somebody have a quick look at http://elinux.org/BeagleBoardAndOpenEmbeddedGit at the git checkout origin/stable/2009 -b stable/2009 May 02 20:48:39 part if that make sence and if so how to update for example May 02 20:49:34 as I understand that create a new local branch May 02 20:49:39 creates May 02 21:17:20 keesj, reading May 02 21:18:23 keesj, git pull then checkout May 02 21:18:27 But yes, that's fine May 02 21:18:34 stable/2009 will track origin/stable/2009 May 02 21:29:42 mwester, thanks May 02 21:30:12 Nice cleanup. :) May 02 21:31:14 more to come, hopefully May 02 21:31:40 I kinda want to make it easier to get 'build gcc/etc', 'external toolchain', or 'something else' May 02 21:32:26 I think that's the right direction for OE May 02 21:34:46 Ultimately, it would be excellent if we had not only a standard minimal image, but a standard minimal toolchain. Then we could get closer to where I think many OE users (hobby and commercial) would like: a standard minimal distro to start with, known to build with a standard external toolchain, on some "reference" host distro/release where we can minimize the number of -native recipes built. May 02 21:35:19 That would be an ideal starting point, and would help people understand why OE must build so much other stuff for their toolchain/host/distro. May 02 21:36:39 It would also help reduce the "gentoo"-ism of the solution; many commercial users would rather not have each developer build their own toolchain. May 02 22:04:38 Yeah, that too May 02 22:04:52 another thing on my todo list, after that, is to make some standard host conf files May 02 22:05:13 so you can do say, inherit += "ubuntu_jaunty.conf" and that will ASSUME_PROVIDED stuff for a normal install May 02 22:05:20 Cut way the hell down on the -native junk getting built May 02 22:05:41 That's where a lot of the "big"ness comes from, in building a "toolchain" May 02 22:05:46 it wouild be tremendously sexy for someone to have eclipse (on windows) build packages for an OE built machine May 02 22:05:51 Just doing gcc+binutils isn't hard May 02 22:06:18 mind you, I wouldn't be a user of such a beast :) May 02 22:07:29 I really should try a build of the micro-image with Angstrom in uclibc mode also :) May 02 22:08:26 The new micro thing? May 02 22:08:31 yeah May 02 22:08:39 should be interesting, yeah May 02 22:08:45 I ran micro-image with Angstrom May 02 22:08:49 but no uclibc May 02 22:08:59 The feed stuff / on-target pkg management is where some bloat comes in May 02 22:09:03 but, it is good the image built May 02 22:09:07 yeah May 02 22:09:14 Crofton|work, btw, can you review/ack the LIBC patchset? :) May 02 22:09:22 and there are loads of people that do not need that May 02 22:09:35 this is not my strength May 02 22:09:40 heh May 02 22:09:42 a working example is a very very good thing May 02 22:09:46 yeah May 02 22:09:53 I like micro-image for that reason May 02 22:09:59 yep May 02 22:10:05 basically, a taskless image May 02 22:10:35 it should also help machine maintenance; i.e. the micro image should be buildable, and create a usable image for every machine... otherwise your machine.conf is wrong. May 02 22:10:48 yeah May 02 22:12:45 I would push the image, but I ahve not been able to boot test it May 02 22:12:55 but, I may just push it tomorrow May 02 22:13:10 since it at least builds and seems to be a unique starting point May 02 22:13:27 it isn't going to break anything else May 02 22:13:36 micro versus minimal distros need to e combined in my mind May 02 22:13:38 yeah May 02 22:13:51 well, one needs to go May 02 22:14:10 to many distro files, it was good to see people willing to delete a distro :) May 02 22:14:31 ok, must get ready for dinner May 02 22:15:09 and derby? :) May 02 22:23:30 heh May 02 22:23:39 Melbourne cup? May 02 22:29:22 hello May 02 22:30:22 I'm trying to built/install python a target using oe.... but it seems python is built without a lot of really common stuff that i need, like array, struct, fcntl, etc May 02 22:30:26 what's the best way to include this? May 02 22:33:18 checked the other packages that are generated? May 02 22:38:59 which ones? May 02 22:39:13 Well, tmp/deploy/ipk//python* May 02 22:39:22 or deploy/glibc/ipk, if using angstrom or similar May 02 22:41:36 yeah, i see a python-io nd python-fcntl there... how can I include those in the rootfs? May 02 22:42:11 Are you adding python to an existing rootfs? May 02 22:42:17 er image May 02 22:42:24 image May 02 22:42:35 yeah May 02 22:42:39 Same way :) May 02 22:42:44 list python python-io ... May 02 22:43:06 hmmmm, but i can't do a bitbake python-io, it says package not found May 02 22:43:14 right May 02 22:43:19 'bitbake python' builds it May 02 22:43:31 Then same way you installed python into the rootfs May 02 22:43:36 add python-io May 02 22:43:38 so then if I put python-io in my image recipe, will it include it? May 02 22:43:57 `yeah May 02 22:44:02 okay May 02 22:44:25 sorry i'm not entirely up to speed with oe yet... how come I can include a package that I can't build directly? May 02 22:46:27 Roughly... May 02 22:46:42 'bitbake recipe' means you need to have recipe_someversion.bb May 02 22:46:59 And recipe_...bb can make many packages May 02 22:47:19 i see May 02 22:47:28 You can RDEPENDS (runtime depends) on something that's made from a recipe May 02 22:47:38 But not DEPENDS May 02 22:48:47 okay, thanks May 02 22:49:26 while I'm at it, do you how I can configure .conf files that get baked into my images so it's not necessary to change them each time? May 02 22:51:24 Hmm May 02 22:51:33 What distro? May 02 22:51:42 There's a few override location May 02 22:51:44 s May 02 22:51:45 gumstix May 02 22:52:20 You can put the modified one into openembedded/recipes//package-version/gumstix/ May 02 22:52:29 Then bitbake -c clean package && bitbake package May 02 22:52:32 then rebuild your image May 02 22:53:21 okay May 02 22:53:25 hey thanks again May 02 22:56:40 np **** ENDING LOGGING AT Sun May 03 02:59:58 2009