**** BEGIN LOGGING AT Tue Nov 01 02:59:57 2011 Nov 01 04:00:51 Greetings! I wrote yesterday about a problem with libtool-native. And to me have answered : "[16:35] ivan____: try bitbake -c clean libtool-native ; bitbake -c clean gnu-config-native; bitbake gnu-config-native then retry your build" I tried, but after bitbake gnu-config-native deduces http://paste.org/pastebin/view/40393 And at attempt again to collect http://paste.org/pastebin/view/40394. There are still ideas? Nov 01 06:10:42 Greetings! I wrote yesterday about a problem with libtool-native. And to me have answered : "[16:35] ivan____: try bitbake -c clean libtool-native ; bitbake -c clean gnu-config-native; bitbake gnu-config-native then retry your build" I tried, but after bitbake gnu-config-native deduces http://paste.org/pastebin/view/40393 And at attempt again to collect http://paste.org/pastebin/view/40394. There are still ideas? Nov 01 08:50:47 following style from some other image recipes I've added http://pastebin.com/p95wdnsX to my image.bb. Should this be enough to override the usage of systemd in angstrom/oe-core? Nov 01 08:51:34 I have some issues w/systemd, and thought I'd try to make the other stuff I'm porting work before trying to sort out those issues Nov 01 09:02:37 mornin Nov 01 09:15:28 hmm, I am reading through the setup scripts and I still do not get the idea of this layer.conf, what is the difference to local.conf and package collections? Nov 01 09:16:03 and does one have to use the layer configuration now or is the classic way still supported (or is the classic configuration deprecated?) Nov 01 09:21:21 hmm, from what I see in the layer config, its just a normal collection definiton... Nov 01 10:36:28 hi, all! Nov 01 10:36:32 privet Nov 01 10:37:13 Jin^eLD: how is your flash memory? Nov 01 10:37:38 don't ask :) I almost booted from NFS yesterday, but something is wrong with my init script so I did not get a shell yet Nov 01 10:38:08 Jin^eLD: init=/bin/sh? Nov 01 10:38:37 that will do of course, I first wanted to see why the default init script is not working for me Nov 01 10:38:47 because I will probably have the same problem on the flashed image Nov 01 10:46:15 if I plan to use the 2011.03 release, it's the 2011.03-maintenance branch? Nov 01 10:52:03 Jin^eLD: dunno Nov 01 11:39:45 hi, does anyone know if there has been any work done on distributing oe builds across several machines at the task level (I know icecc can be used at a lower level). logically it should be very easy in homogenous environments with a shared filesystem (as is my case) and possible otherwise Nov 01 11:40:26 after changing IMAGE_INIT_MANAGER, what must I clean/rebuild? 'bitbake -c myimage' doesn't do the trick. Nov 01 11:53:33 tasslehoff: maybe add -f? Nov 01 12:02:41 sassan: I'll try, but I suspect maybe there's another recipe that must be cleaned. the one that makes the symlink to /sbin/init maybe. Nov 01 12:03:24 tasslehoff: for me the symlink was broken and was always pointing to busybox Nov 01 12:03:39 I fixed it manually, I did not dig into why the symlink was incorrect Nov 01 12:05:27 Jin^eLD: for me /sbin/init points to /bin/systemd, but I want it to point to /sbin/init.sysvinit. are we talking about the same symlink? Nov 01 12:06:12 hmm probably not, I was referring to /linuxrc :P Nov 01 12:06:41 :) Nov 01 12:07:13 Jin^eLD: I'm trying to boot with systemvinit instead of systemd, since something is broken on my system when using systemd Nov 01 12:08:09 thought to finish the porting from oe-classic to oe-core still using sysvinit, and then revisit systemd later Nov 01 12:08:24 I know what you are trying to do; however I never used systemd so far and nothing tried to pull it in; so I did not run into your problem yet Nov 01 12:08:41 what is oe-classic and oe-core? Nov 01 12:09:13 is this http://wiki.openembedded.org/index.php/Release-2011.03 not the "real" thing anymore? Nov 01 12:11:52 would be nice if someone could explain that to me :) I hear the "classic" term popping up now ant then but don't really get the idea Nov 01 12:15:09 Jin^eLD: http://www.openembedded.org/wiki/OpenEmbedded-Core Nov 01 12:15:25 people: why is this not on the frontpage of the wiki? Nov 01 12:15:35 thanks, I totally missed that Nov 01 12:15:56 tasslehoff: no wonder :) Nov 01 12:16:01 Jin^eLD: ^ Nov 01 12:16:53 doh :P Nov 01 12:16:57 Jin^eLD: I believe 2011.03 is the final release in classic, which will soon be made read only Nov 01 12:18:16 that means there is no point to update my current 2008 stuff to 2011.03 Nov 01 12:18:29 but I should rather migrate to oe-core, doh :( Nov 01 12:20:43 Jin^eLD: I think so. "Wasted" much time? Nov 01 12:21:43 not yet, actually updating to a newer classic would be very easy for me Nov 01 12:22:08 so actually changing my stuff to "core" will cost me some time... Nov 01 12:24:54 core makes a more complicated impression on me Nov 01 12:25:47 Jin^eLD: I started out using the angstrom setup-scripts, which sets up most of what you need. Nov 01 12:26:24 I do not like too much automagic Nov 01 12:28:22 Jin^eLD: not much automagic to it (for me at least). I've added my own meta-layer and setup everything the way I want there Nov 01 12:30:17 hm, so I guess I am next in the line for the fight against systemd then...? Nov 01 12:30:19 :P Nov 01 12:30:29 :) Nov 01 12:33:16 and angstrom lives in an own git now as a layer, do I understand it correctly? Nov 01 12:33:26 yes Nov 01 12:33:57 difficutl to look through what went where.. is there some migration guide? Nov 01 12:36:05 not really, this is something that needs improving Nov 01 12:37:09 Crofton: any hint or suggestion? or is the only thing I can do - go through each single package I am using to figure out in which repo it went and try to put it all together again? Nov 01 12:37:33 find is your friend :( Nov 01 12:37:37 doh... Nov 01 12:37:48 use the oebb.sh script to get close to a working build Nov 01 12:40:36 hm. IMAGE_BOOT doesn't seem to be used anymore, actually Nov 01 12:51:19 Jin^eLD, http://www.openembedded.org/wiki/LayerIndex Nov 01 13:05:28 yeah, thanks, I was already there looking at what is where Nov 01 13:34:17 I use oe-core for some new project, and for some requirement I need to use toolchain from layer (gcc-3.4-based with lots of patches), so I need to override oe-core toolchain and use one in layer for some images. Is it possible? Nov 01 13:35:07 that is for mips without unaligned-access instructions Nov 01 13:39:45 I'm not sure you can use gcc-3.4 with contemporary oe-core. Does that support sysroots? Nov 01 13:40:28 Using a toolchain from another layer is no problem in principle though. I think meta-oe has its own gcc for example. Nov 01 13:52:44 pb_: looks like it supports sysroot Nov 01 13:53:31 at least it support --with-sysroot configure option Nov 01 13:53:39 bah. how do I change from systemd to sysvinit? currently I'm trying a postprocess command to remove /sbin/init and create a new symlink.. Nov 01 13:56:09 tasslehoff: you could run bitbake with the -g option and analyse the dependency graph to figure out what is puling in systemd Nov 01 13:59:20 Jin^eLD: that I've never tried before. thanks :) Nov 01 14:00:08 tasslehoff: it will spit out .dot files that are more or less human-readable Nov 01 14:08:09 hm. rsyslog seems to be the only one depending on it Nov 01 14:10:45 only one is already more than you want :) Nov 01 14:30:07 Jin^eLD: yeah. but I guess that both systemd and sysvinit can be installed at the same time. it's just the darn symlink I want to change now :) Nov 01 14:30:56 well, does it install systemd in your rootfs? if yes, and if you do not want to use it - I'd make sure it does not go there at all... Nov 01 14:31:13 otherwise I'd try to rigure out who creates that link Nov 01 14:31:30 we'll see how fast I will get there ;) rigt now I am still messing around with migration to oe-core :P Nov 01 14:47:59 how exactly does this bbappend thing works? is there any documentation for it? Nov 01 14:48:57 add a glob for the bbappends to BBFILES, name the bbappend the same as the .bb Nov 01 14:49:46 so *.bbappend is loaded on top of the *.bb and overwrites and/or extens whatever was defined in the .bb? Nov 01 14:49:46 that's it Nov 01 14:49:46 its content gets appended to the recipe metadata Nov 01 14:49:46 goes entirely by filename to associate bbappend to bb Nov 01 14:50:03 again, it gets appended Nov 01 14:50:12 its like concatenating it to the end of the recipe file Nov 01 14:50:14 so if in the .bbappend I do a DEPENDS = instead of DEPENDS += ? Nov 01 14:50:26 then you'd replace the main depends, yes Nov 01 14:50:32 its no different from echo foo >> blah.bb Nov 01 14:50:40 ok thanks, I get the idea Nov 01 14:50:53 np Nov 01 14:51:06 is there a way to specify layer priorities? Nov 01 14:51:31 yes, that hasn't changed, same method we've used for like 5 years or something Nov 01 14:51:43 read the layer.conf in oe-core and meta-oe to see the methdo Nov 01 14:52:00 ah ok, so the regular collection method still works Nov 01 14:52:12 BBFILE_PATTERN blahblah Nov 01 14:52:25 I see, thanks Nov 01 14:52:38 same stuff as usual, just how the layers are specified and where they're configured that changed Nov 01 14:52:44 bblayers -> layer.conf, but their content is the same Nov 01 14:52:47 bbpath, bbfiles, bbfile_* Nov 01 14:55:16 slapin: that's not enough. does it support --sysroot at run time? Nov 01 14:55:42 morning kergoth Nov 01 16:31:15 bluelightning: Hello there ... Nov 01 16:31:28 hi otavio Nov 01 16:31:32 bluelightning: it seems I got an issue with combo-layer script once again heh Nov 01 16:31:56 otavio: hmm ok, what happens? Nov 01 16:33:04 bluelightning: it fails to apply a bitbake patch Nov 01 16:33:17 commit 8e4e75383e43d6da2c16ec5286186a0d0569b0f8 Nov 01 16:33:18 Author: Christopher Larson Nov 01 16:33:28 otavio: are you using it yourself for your own setup, or for poky, out of curiosity? Nov 01 16:34:25 kergoth_: my own Nov 01 16:34:30 ah, cool Nov 01 16:34:42 kergoth_: i am using it to maintain our internal tree Nov 01 16:34:48 how is that going for you? beyond this particular problem i mean :P Nov 01 16:34:57 * kergoth_ is experimenting with using android's repo tool, himself Nov 01 16:35:06 kergoth_: so far it is working well, I miss some useful features but it works quite well Nov 01 16:35:48 cool Nov 01 16:35:55 kergoth_: it is easy to use and people that work on our tree doesn't even know it exist most of time as they mostly work in our internal layer Nov 01 16:36:18 kergoth_: so I am the one responsable to update it against current trees and then I do it once a week or so Nov 01 16:36:24 is it a pain to extract changes back out? thats why I didn't go with one like that.. i like to cd into the individual repositories and hack on things and push Nov 01 16:36:32 ahh, i see. since its just you responsiblel for that... Nov 01 16:36:38 less of an issue Nov 01 16:36:39 cool Nov 01 16:37:03 kergoth_: what I do is to clone them and handle it by hand Nov 01 16:37:16 kergoth_: so i commit the changes on the local repos on same tree Nov 01 16:37:38 kergoth_: and then I do checkout -f on the root repo and use combo-update to sync what has been merged Nov 01 16:37:45 Hi, anyone built a oe-core + meta-micro image lately? Nov 01 16:38:06 martinambrose: i tried it yesterday Nov 01 16:38:12 failed for micro-uclibc, but the eglibc one worked Nov 01 16:39:14 otavio: well, it seems like the hook script is not the issue this time, I just checked and it's not doing anything nasty with the patch itself Nov 01 16:39:28 bluelightning: any idea? Nov 01 16:39:45 bluelightning: what worried me is that it seems to have applied the patch and fails after it Nov 01 16:39:51 bluelightning: but it said it fails on it Nov 01 16:39:53 thx. both failed for me. trying uclibc with ERROR_QA overridden. Which machine work for your eglibc build? Nov 01 16:40:01 otavio: I guess I will have to set up a similar configuration to what you have Nov 01 16:40:50 by moving dev-so to WARN_QA my micro-base-image built for qemuarm Nov 01 16:40:55 martinambrose: I was testing qemux86. Nov 01 16:41:04 kk. thx Nov 01 16:41:20 ah, cool. I'll have to try that qa workaround, thanks for the tip Nov 01 16:41:21 :) Nov 01 16:41:29 I like how small the task # is when using micro :P Nov 01 16:50:18 bluelightning: what I can do to help you with it? Nov 01 16:50:36 otavio: I'm just setting up a test now, will let you know what I find Nov 01 16:50:47 bluelightning: good Nov 01 16:50:48 which was the error you were getting with dev-so? Nov 01 16:50:58 I know libcidn is broken for eglibc but (afaik) uclibc should be ok Nov 01 16:55:13 it was a packaging error. the symlinks were placed in non -dev packages. Nov 01 16:55:49 with all the split packages it doesn't look like auto packaging is currently feasible Nov 01 16:56:36 well, yes, I guessed that it was a packaging error from the fact that it was dev-so. Nov 01 16:56:47 I meant, which package and which file had the problem? Nov 01 16:57:21 ah, ok. it was uclibc-thread-db Nov 01 16:58:32 righto, thanks Nov 01 17:03:38 pb_, do you think denix passes the Turing test? Nov 01 17:04:58 I guess he ought to, though I can't say I've ever tried it. Nov 01 17:05:03 why do you ask? Nov 01 17:06:55 write permission on the wiki. Nov 01 17:07:23 aha Nov 01 17:07:39 well, I've met him in person and he seemed convincing then. Nov 01 17:13:08 otavio: I think the problem is there was a merge commit; it doesn't handle those... Nov 01 17:13:35 although I did not get an error I do see it's using an incorrect revision in the commit you mentioned Nov 01 17:19:26 bluelightning: it might be Nov 01 17:19:32 bluelightning: but it ought to handle those, no? Nov 01 17:19:46 otavio: it ought to yes Nov 01 17:20:08 otavio: I just never expected to see one in bitbake/oe-core... Nov 01 17:20:27 bluelightning: I did the update and it did have no diff in the end Nov 01 17:20:35 bluelightning: but it didn't work automatically Nov 01 17:20:45 bluelightning: so it indeed ought to be realted to the merge Nov 01 17:20:51 bluelightning: well, merge commits are common Nov 01 17:20:55 bluelightning: imo it is critical Nov 01 17:21:05 otavio: yes I know, I will fix it Nov 01 17:23:55 bluelightning: when you're free, let's talk about the missing features of combo-update Nov 01 17:24:01 bluelightning: so you can work on those too Nov 01 17:24:23 otavio: ok sure, it'll probably be tomorrow if that's ok... Nov 01 17:24:32 bluelightning: ok Nov 01 17:24:34 bluelightning: good Nov 01 17:31:33 Is someone having trouble to package things that installs into lib64 as native? Nov 01 17:50:10 otavio: do you know what the original revision your bitbake repo was at when you ran the update that failed? Nov 01 17:53:08 bluelightning: 6cae2d11f8e254a1d01bf294eef916ac43a1881b Nov 01 17:54:12 martinambrose: mm, yes, it certainly is broken. weirdly it looks as though it has been broken for ages, and I can't see any way in which it would work on other distros either. Nov 01 17:54:46 the definition of FILES_uclibc-thread-db is just plain wrong, it needs to be more specific. Nov 01 17:55:04 It seems dbus has a build-failure today Nov 01 17:55:05 http://paste.debian.net/141845/ Nov 01 17:55:13 Our autobuilder has just catched it Nov 01 18:32:37 ka6sox, pb_: that's funny, you made my day! :) Nov 01 18:45:21 denix, :D Nov 01 21:01:18 how does OE handle firmware blobs? Nov 01 21:04:53 msm: a recipe pulls in the blob and places it at the needed folder for driver consumption. We have some magic for linux-firmware as well as recipes for marvel libertas firmware as examples Nov 01 21:06:22 next question: how does distro_tracking stuff come in to play? Nov 01 21:06:25 i see it's in the file Nov 01 21:06:28 is there some reporting tool? Nov 01 21:06:47 msm: no idea actually. Never had to play with it :) Nov 01 21:07:59 found 'bitbake world -f -c distro_check' **** ENDING LOGGING AT Wed Nov 02 02:59:57 2011