**** BEGIN LOGGING AT Fri Feb 10 02:59:57 2012 **** BEGIN LOGGING AT Fri Feb 10 07:28:11 2012 Feb 10 08:17:55 good morning Feb 10 09:24:45 morning Feb 10 10:51:01 morning all Feb 10 11:33:40 Is there a recipe to add users (w/passwords) to my rootfs? Feb 10 13:47:21 tasslehoffwrk: there is a whole class to do that, see useradd.bbclass Feb 10 14:03:42 naja, ich habe gedacht es waer evtl sinnvoller das doch unter den feed dirs zu platzieren Feb 10 14:03:44 oops wrong window Feb 10 14:35:01 GNUtoo: hi Feb 10 14:35:17 GNUtoo: i noticed you're working in chromium? Feb 10 14:48:08 yes Feb 10 14:48:36 otavio, hi Feb 10 14:48:44 denisATeukrea: Hi Feb 10 14:48:49 otavio, are you sure of mozilla license in meta-mozilla? Feb 10 14:48:53 because it sounds strange Feb 10 14:49:02 LICENSE = "MPL-1.0 & GPL-2.0 & LGPL-2.1" Feb 10 14:49:07 shouldn't it be: Feb 10 14:49:09 denisATeukrea: are you willing to share work in Chromium? Feb 10 14:49:25 LICENSE = "MPL-1.0 | GPLv2+ | LGPL-2.1+" Feb 10 14:49:28 or something like that Feb 10 14:49:30 yes of course Feb 10 14:49:32 denisATeukrea: we'll need to support it too Feb 10 14:49:43 the only problem is that It needs some deps Feb 10 14:49:49 and I'm doing the nss dep right now Feb 10 14:50:00 I didn't even have chromium working on target Feb 10 14:50:01 yet Feb 10 14:50:22 denisATeukrea: If you can, please send a pull request fixing the license Feb 10 14:50:29 denisATeukrea: it is wrong, indeed. Feb 10 14:50:32 ok Feb 10 14:50:36 I'll do it now Feb 10 14:50:43 what's the place to send the patch? Feb 10 14:51:12 github Feb 10 14:51:21 url = git://github.com/OSSystems/meta-mozilla.git Feb 10 14:51:21 you can do a pull request Feb 10 14:51:22 ah oik Feb 10 14:51:23 yep Feb 10 14:51:24 ok Feb 10 14:51:30 easy and easy to manage Feb 10 14:51:44 I prefer gitorious personally Feb 10 14:51:47 denisATeukrea: I also want to sit and update firefox to 10 once i have time Feb 10 14:51:51 but anyway I'll send a pull request Feb 10 14:51:57 ok Feb 10 14:51:58 denisATeukrea: gitorious lack issues, that's why we use github Feb 10 14:52:03 ok Feb 10 15:12:24 done Feb 10 15:12:34 back on working on nss Feb 10 15:12:48 I've to find where the license is in the source Feb 10 15:14:11 btw, for LICENSE it's only | if there is a choice for the entire codebase... if it's a mix it's definitely & Feb 10 15:14:43 it's a choice I think Feb 10 15:16:04 ok, just checking :) Feb 10 15:18:26 Alternatively, the contents of this file may be used under the terms of - either the GNU General Public License Version 2 or later (the "GPL"), or - the GNU Lesser General Public License Version 2.1 or later (the "LGPL"), - in which case the provisions of the GPL or the LGPL are applicable instead - of those above. Feb 10 15:18:45 so I think it's | Feb 10 15:20:48 denisATeukrea: presumably that's the same header for all files? Feb 10 15:22:05 yes Feb 10 15:22:22 in mozilla stuff they apply the same header to all files trough configure or make Feb 10 15:22:32 at least that's what I saw in nspr and nss Feb 10 15:22:41 so I suppose it's not that different with firefox Feb 10 17:21:05 Do I need to clean any packages after switching IMAGE_INIT_MANAGER to "sysvinit"? Bitbake just did a short do_rootfs job and systemd is still running on the target. Feb 10 17:21:54 icanicant, I think it's a distro policy Feb 10 17:21:58 right? Feb 10 17:22:08 so I think a lot has to be rebuilt Feb 10 17:22:13 but I'm not sure Feb 10 17:22:52 hmmm forget what I told Feb 10 17:23:14 I remember beeing able to bitbake different images with systemd or without on the same distro Feb 10 17:23:33 so it might be an image issue Feb 10 17:24:08 GNUtoo: i based the image on angstrom's console-image and just changed the IMAGE_INIT_MANAGER field. bitbake -e shows sysvinit but i have a feeling i need to rebuild something (don't know what!) Feb 10 17:24:28 maybe do an image which doesn't depend on systemd Feb 10 17:24:49 because maybe there are stuff which deend on systemd in that image Feb 10 17:26:03 i read that there was a compatibility package so connman-systemd and friends will work. and there's no mention of systemd in bitbake -e output. Feb 10 17:26:33 hmmm Feb 10 17:26:39 maybe you need to rebuild a task then Feb 10 17:26:54 look for tasks that may include IMAGE_INIT_MANAGER Feb 10 17:27:09 GNUtoo: thanks Feb 10 18:38:22 is there a tool that selects a default toolchain? which would move /usr/bin/{gcc,as,ranlib,etc} to point at the correct implementation? Feb 10 18:56:34 msm, you mean like a devshell in oe-classic? Feb 10 18:56:37 yes there is Feb 10 18:56:44 it's on github or something like that Feb 10 18:57:51 msm: I'd think update-alternatives use in the toolchain packages would do the job Feb 10 19:02:09 ahh on target Feb 10 19:05:32 Anyone aware of a tutorial for creating a layer? Feb 10 19:08:53 create a directory. create conf in that directory. copy an existing layer.conf into there. edit and search/replace the layer name. now you have a layer. put whatever you want in it, config files, classes, recipes, and bbappend files. Feb 10 19:10:18 kergoth, Great! Thanks for the tutorial! Feb 10 20:01:21 kergoth: GNUtoo: yea, update alternatives Feb 10 20:01:33 ok Feb 10 20:03:24 it seems like that's just symlinking Feb 10 20:03:32 there is not shortcut to replace all the binutils at once? Feb 10 20:05:03 does this even work for rpm? Feb 10 20:05:20 update-alternatives is a standalone script. package manager is irrelevent Feb 10 20:05:24 irrelevant, even Feb 10 20:13:24 kergoth: doesn't seem like it does anything except symlink? Feb 10 20:13:32 I don't see what the problem with that is Feb 10 20:16:23 well, yeah, that's what update-alternatives does. what were you hoping for? Feb 10 20:16:40 could make a version that hardlinked, I guess, that'd be easy enough Feb 10 20:17:10 pb_: Feb 10 20:17:19 pb_: kergoth: update-alernatives binutils Feb 10 20:17:33 which would update all the binaries at once Feb 10 20:17:51 the arguments to update-alternatives seem to be the same arguments passed to symlink Feb 10 20:17:54 update them from what? Feb 10 20:18:15 update-alternatives is to choose between multiple alternatives (which, implicitly, need to all be installed already) Feb 10 20:18:32 if you're talking about installing newer versions atomically, that's a package manager problem Feb 10 20:18:41 pb_: no they are installed Feb 10 20:18:52 i need to switch these: /usr/bin/powerpc-fsl-linux-as Feb 10 20:18:52 /usr/bin/as Feb 10 20:18:52 /usr/bin/powerpc64-fslmllib64-linux-as Feb 10 20:19:12 i *can* change the symlink on my own Feb 10 20:19:16 not sure how update-alternatives is helping me Feb 10 20:19:23 im talking about doing this on a live system too btw Feb 10 20:20:54 am I missing something? Feb 10 20:20:55 update-alternatives is used for one link at a time (e.g. /usr/bin/vi being vim vs nvi), not an entire set of the things Feb 10 20:21:23 as far as I know, nothing like what you want exists, and I really doubt it should in any generic way Feb 10 20:22:26 those toolchain symlinks are taken care of during install anyway Feb 10 20:22:47 yes, except i want to switch between them Feb 10 20:23:17 hmm so want to have multiple toolchains ? Feb 10 20:23:21 yes Feb 10 20:23:35 since we don't support multilib yet, it's a stop gap solution Feb 10 20:24:00 actually update-alternatives in debian/ubuntu has the concept of "link groups" Feb 10 20:24:00 ok then change target binutils recipes to use update-alternatives for each binary you want that way Feb 10 20:24:15 actually Feb 10 20:24:18 I think we are using the cworth version of update-alternatives by default though Feb 10 20:24:23 i can just install "lib64-gcc-symlinks" Feb 10 20:24:48 and i will have 64 stuff, but i need to be able to switch between then on a live system Feb 10 20:25:49 bluelightning: ah, does it? interesting Feb 10 20:26:53 it all works quite well believe it or not… using multilib to install multi toolchains Feb 10 20:29:43 I have a source tarball that is no longer really found in any good place online (dead link now), but we still want to use that specific source. Is it possible to get at that in the SRC_URI using the local file: fetcher? Feb 10 20:30:41 possible, yes, advised, no Feb 10 20:31:24 not advised sure :) We are probably going to completely remove the package from our install in the near future so an upgrade or replacement isn't worth the time Feb 10 20:31:37 hbeck: you probably just need to set up a local mirror containing that one file Feb 10 20:31:40 hbeck, if you have one that has a correct MD5sum then you con point me to it. Feb 10 20:31:46 and i will put it in our source mirror Feb 10 20:35:42 ka6sox appreciated but probably not worth it :) I could email/send the file for grins if you want it around just in case Feb 10 20:36:14 bluelightning can a local mirror work without setting up for http type access? That is kind of a pain Feb 10 20:36:57 hbeck: yes; all the mirroring code does is do a replace on the path, so if you point it to a file:// location then that should work just fine Feb 10 20:37:25 interesting. That's not something I've played with yet. Where should I peek for an example? Feb 10 20:37:36 hbeck: have a look at classes/own-mirrors.bbclass (which you can adapt or use directly, whatever suits) Feb 10 20:38:47 although that uses PREMIRRORS (i.e. it will look there first rather than if fetching from the original URI fails) Feb 10 20:42:15 bluelightning I assume that can be inherited at the recipe level (and SOURCE_MIRROR_URL also) Feb 10 20:42:41 hbeck: yes, INHERIT += "own-mirrors" Feb 10 20:43:24 no, if he wants to inherit at the reciep level he'd use 'inherit' :) Feb 10 20:43:27 but should work fine yes Feb 10 20:43:32 kergoth: was about to ask that Feb 10 20:44:37 kergoth: ah right yes... although it's somewhat unorthodox Feb 10 20:45:03 yeah, not really intended to be used like that, but *shrug* Feb 10 20:45:26 well a bit unorthodox to have a local hosted relic package, too :) Feb 10 20:46:15 If I'm trying to specify it as file:// wouldn't I need to also muck with the FILESPATH or FILESDIR stuff? Or can I give it a fully qualified path somehow Feb 10 20:46:40 hbeck: not if you specify the full path Feb 10 20:47:15 I'm guessing it would be file:///full/path/to/mirror/dir/ Feb 10 20:50:37 ok yeah the file:/// with full path makes it happy. Of course with that I don't need the own-mirrors Feb 10 20:51:23 thanks for the help with the total hackery of the normal system :P Feb 10 20:51:39 hbeck: no problem :) Feb 10 22:29:49 any idea why a fakeroot do_install would give "operation not permitted" ? http://paste2.org/p/1900352 Feb 10 22:36:28 hmmm, bitbake issue??: http://comments.gmane.org/gmane.comp.handhelds.openembedded/44436 Feb 10 22:53:50 has someone found the breakage of gtk+ with current zlib 1.2.6/Z? Feb 10 22:53:55 has someone found the breakage of gtk+ with current zlib 1.2.6? Feb 11 02:16:21 otavio: what problem do you see. Feb 11 02:16:31 otavio: always post errors **** ENDING LOGGING AT Sat Feb 11 02:59:57 2012