**** BEGIN LOGGING AT Thu Aug 25 02:59:58 2011 Aug 25 04:28:49 can someone please help me get this agere hermes ii wifi card working on my c700 Aug 25 06:27:01 http://www.youtube.com/watch?v=vIJINiK9azc Aug 25 06:27:03 lol Aug 25 08:11:55 hi JaMa Aug 25 08:12:17 I have never created braches ..... so just need more information Aug 25 08:12:43 where should I create my own branch ...... in github and how can I do that Aug 25 08:14:06 the procedure would be i create a brach ...... clone it locally .... make chages and commit to local .... send them to my branch and commit there and then send pull request? Aug 25 08:15:25 ttp://www.openembedded.org/wiki/GitPhraseBook#Create_your_own_short_lived_feature_branch and answer for 2nd question was already in my e-mail meta-oe-contrib Aug 25 08:24:58 JaMa: back my system was hanged so did not get any message if you have send any Aug 25 08:25:43 http://www.openembedded.org/wiki/GitPhraseBook#Create_your_own_short_lived_feature_branch and answer for 2nd question was already in my e-mail meta-oe-contrib Aug 25 08:35:21 thanks JaMa Aug 25 09:01:43 hi all Aug 25 09:03:50 hi pb_, all Aug 25 09:05:17 hi bluelightning Aug 25 09:05:25 hi anarsoul Aug 25 09:07:39 JaMa: I was doing this to create the branch git checkout -b noor/benchmark git://git.openembedded.org/meta-openembedded-contrib .... getting an error fatal: Not a git repository (or any of the parent directories): .git Aug 25 09:08:06 should I first clone the meta-oe-contrib git repo? Aug 25 09:08:35 Noor: yes and some RTFM seems usefull too (sorry no time to guide you) Aug 25 09:09:02 JaMa: no problem Aug 25 09:32:12 Noor: you probably want to add meta-oe-contrib as another remote to your meta-oe checkout Aug 25 09:41:19 bluelightning: I have clone meta-oe repo .... can I do some thing with that repo or I have to start from scratch Aug 25 09:41:38 in clone meta-oe I have been making changes Aug 25 09:41:47 previously Aug 25 09:42:02 it's going to be easier to have it as a remote for your meta-oe checkout Aug 25 09:42:26 do you have push access to meta-oe contrib? Aug 25 09:42:44 er, I meant meta-oe-contrib Aug 25 09:45:28 bluelightning: I dont think so .... I am totally new commer in this world Aug 25 09:46:08 right, so are you adding it just to be able to pull someone else's branch? Aug 25 09:47:34 bluelightning: see on the ML koen has asked me to add my changes to some brach and send a pull request to bring the stuff Aug 25 09:48:10 bluelightning: previously I was sending patches to ML Aug 25 09:48:44 bluelightning: now have to create my own branch .... push my changes there and send pull request to mailing list Aug 25 09:49:21 so not I am in process of creating my own branch .... but not able to do so Aug 25 09:50:45 not -> now Aug 25 09:52:49 if you want to use meta-oe I think you'd need to get access for that first, but I don't imagine that would be too difficult Aug 25 09:52:59 er, I mean if you want to push to meta-oe-contrib Aug 25 09:53:30 otherwise you can just create a repo somewhere on github and use that for your pull requests Aug 25 09:53:40 bluelightning: I was told to make my own branch from meta-oe-contrib Aug 25 09:54:25 pb_: can you grant access to meta-oe-contrib for people? Aug 25 09:54:38 * bluelightning has no idea what the process is these days Aug 25 09:54:47 presumably it involves sending an ssh key to someone... Aug 25 10:02:31 has anyone an idea why the build for eglibc tries to overwrite/remove my host includes ? Aug 25 10:22:39 bluelightning: I have no idea either, sorry Aug 25 10:23:49 Noor: Martin just said "some repository"... it doesn't have to be meta-oe-contrib Aug 25 10:24:10 probably easiest to host a contrib tree on github or somewhere for the time being Aug 25 10:24:21 I don't think there's anything very magic about meta-oe-contrib Aug 25 10:24:25 is the org.openembedded.dev deprecated ? Aug 25 10:24:40 by some, yes Aug 25 10:25:01 I think it's probably fair to describe it as obsolescent although it does still have its adherents. Aug 25 10:25:11 pb_: .. and replaced by what ? Aug 25 10:26:14 bluelightning: some repository ie meta-openembedded-contrib branch Aug 25 10:26:38 pwgen: oe-core and the associated layers Aug 25 10:26:40 any way I just need to find a way to send pull request Aug 25 10:27:12 I never created my own repo so that is why finding it difficult Aug 25 10:27:36 my own repo/branch Aug 25 10:27:37 it's pretty trivial to create one at github, just a case of clicking one button Aug 25 10:27:47 pb_: is there any description how to use oe-core ? Aug 25 10:28:06 pb_: can I send pull request from there? Aug 25 10:28:06 I think there's a wiki page. Aug 25 10:28:17 yes, you can send a pull request from whereever you like. Aug 25 10:28:48 pwgen: http://www.openembedded.org/index.php/OpenEmbedded-Core Aug 25 10:29:58 we should really document how to submit patches as well if we haven't already Aug 25 10:30:45 bluelightning: There is a Wiki page »How to submit a patch« or so. Aug 25 10:31:12 PaulePanter: yes, it does not cover the new oe-core workflow however Aug 25 10:31:22 hi PaulePanter Aug 25 10:31:32 someone should put a note to the openembedded/Getting_Started page .. Aug 25 10:31:45 bluelightning: Well, sending patches to the list is still a valid way too, is not it? Aug 25 10:31:48 Noor: Hi. Aug 25 10:32:51 PaulePanter: it is; however we now encourage the use of pull requests; etc... multiple mailing lists; which layer should it go in, etc. Aug 25 10:33:51 bluelightning: I see. Thanks. Aug 25 10:36:16 I guess the policy for submitting patches is (at least potentially) different for each layer. oe-core itself doesn't seem terribly fussy about pull requests but apparently meta-oe is more enthusiastic about them. Aug 25 10:38:09 and, I suppose, for those layers which do want pull requests, there ought to be a wiki page explaining how to set up a repo to pull from. Aug 25 10:40:04 pb_: I am not sure about that. I guess there have to be plenty of resources explaining that, so a URL should be sufficent. Gitorious or GitHub have outstanding documentation if I remember correctly. Aug 25 10:41:11 if meta-oe want folks to use the meta-oe-contrib repo then well, yeah, I was thinking more of the meta-oe-contrib repo. if the meta-oe folks want people to use that then I think it probably ought to be incumbent on them to document it. Aug 25 10:41:31 er, sorry, that last sentence seems to have come out a bit jumbled Aug 25 10:41:38 I think you probably get the sense of what I was trying to say though Aug 25 10:47:36 I suppose the other thing that comes to mind on that topic is that I would rather hope that the larger corporate users of oe would provide a bit of internal support for their own developers rather than just chucking them into the pool and hoping that "the community" will sort them out. Aug 25 11:09:44 mickey|office: Gratulation fuer Lara !! Aug 25 11:14:16 is there somewhere a description how to add a layer to openembedded-core ? Aug 25 11:25:23 hello! Aug 25 11:26:00 have a bit problem, i want to build image and make some modifications in config files Aug 25 11:26:10 for example in wpa_supplicant.conf Aug 25 11:26:44 but build system tell's me error like: "This file provided by another package" Aug 25 11:28:05 for this i create new .bb with just this conf. i know it is wrong way, but ROOTFS_POSTPROCESS_COMMAND is too short to make all modifications Aug 25 11:29:04 I guess OE should have right nice way to done it.. Aug 25 11:29:17 ROOTFS_POSTPROCESS_COMMAND is also wrong way Aug 25 11:29:25 it has .bbappend in your layer Aug 25 11:29:38 and modify wpa-supplicant recipe Aug 25 11:31:16 i can't modify recipe,becouse i need to have few images with different configs Aug 25 11:33:55 JaMa, can u referer to good doc or usefull example? Aug 25 12:01:53 pwgen: what kind of layer? machine support, or for adding some set of general packages? Aug 25 12:01:59 or, a distro layer? Aug 25 12:03:45 : i had done a long time building angstrom images for c3200 and netwalker . now everything changed and i am confused. i think these strange setup-scripts are the right place to start ... Aug 25 12:04:18 well the setup scripts will help you set everything up for building Aug 25 12:04:30 but creating a layer of your own is something different Aug 25 12:04:49 what you want sounds like a BSP layer (machine support); if so, I would recommend checking out the yocto project BSP developer's guide Aug 25 12:05:01 http://www.yoctoproject.org/docs/bsp-guide/bsp-guide.html Aug 25 12:06:28 : i would be happy if i could buuild something that was former running. doing own layers is something i will do when i could build at least a working image ... Aug 25 12:08:29 pwgen: right... well either oe-core on its own (distro-less) or the angstrom setup scripts will get you a working image for some other machine (e.g. qemuarm, qemux86) Aug 25 12:09:49 FYI, machine support for zaurus is being worked on (partly by me) Aug 25 12:11:33 hi all. Aug 25 12:11:46 hi lumag Aug 25 12:12:37 I've stumbled upon a problem. opkg-make-index in do_rootfs fails if there is a do_write_ipk task running in parallel. Is there any way to interlock those two tasks? Or at least to lock do_rootfs not no be called in parallel with other tasks. Aug 25 12:24:34 lumag: hmm, I'm surprised that would be an issue... I've not heard of it happening Aug 25 12:24:55 is this something that's happening a lot in your setup or only rarely? Aug 25 12:25:17 it seems it happens quite a lot with initramfs image Aug 25 12:26:05 bluelightning, http://pastebin.de/18462 Aug 25 12:29:22 lumag: maybe the RDEPENDS for initramfs-image are not correct somehow? Aug 25 12:29:32 rdepends? Aug 25 12:30:14 yes, through recrdeptasks, rootfs_ipk.bbclass ensures that do_package_write_ipk has completed for everything in RDEPENDS Aug 25 12:30:17 AFAICT Aug 25 12:30:51 bitbake -e initramfs-image | grep RDEPENDS will let you examine what it's being set to Aug 25 12:35:20 bluelightning, I see the problem. initramfs-kexecboot-klibc-image is being built indirectly (it's not the target). So while that image has valid RDEPENDS (kexecboot), there are other packages being built. Aug 25 12:35:36 ... at the same time. And that confuses opkg-make-index Aug 25 12:42:01 Probably I'll hack a opkg-make-index to ignore an exception there Aug 25 13:29:19 lumag: it shouldn't need to be the explicit target to get the dependencies correct Aug 25 13:31:19 bluelightning, Basically, I build console-image. As one of it's indirect dependencies it has initramfs.....-image. So when initramfs do_rootfs is running, nothing stops bitbake from building other console-image in parallel. So it does and sometimes does do_package_write_ipk at the time of opkg-make-index running. Aug 25 13:33:08 lumag: ah right, I see now... hmm Aug 25 13:57:00 does anybody know what the "export OS" in module-base.bbclass is good for? Aug 25 13:57:11 seems a little bit random as far as I can tell. Aug 25 13:58:41 hmm, no clue. could try a blame, or better yet, a git log -S'export OS' --follow classes/module-base.bbclass, to try to find the commit that added it Aug 25 13:59:44 hmm, goes back to 2004 bk merge commit Aug 25 14:00:14 ouch Aug 25 14:01:29 well, I guess it hasn't been causing any grief at least :) Aug 25 14:01:45 but it would still be worth knowing what it's for Aug 25 14:01:50 yeah, it just happened to be causing me some grief because (combined with "make -e") it was stomping on one of my local makefile variables Aug 25 14:06:55 guh, that make -e thing should really go away. we wanted to make it trivial to add recipes, but it just encourages people to not read the makefiles, which bites them later on Aug 25 14:06:58 heh Aug 25 14:07:29 yeah Aug 25 14:08:01 dunno how we would really accomplish that now, though. add an oe_runmake_sanely() that doesn't use -e, and convert recipes to use it I suppose. Aug 25 14:08:38 obviously just dropping -e from oe_runmake without taking any other steps would cause widespread carnage, probably not an acceptable way to proceed. Aug 25 14:08:44 heh, yeah, would be a bit painful for the existing recipes. would have to go through all the non-{autotools,cmake,scons} stuff Aug 25 14:15:50 yeah Aug 25 14:16:15 that said, I dunno how many recipes there are (in oe-core at least) that don't use one of those frameworks. maybe not that many. Aug 25 14:16:51 pb_, kergoth maybe we should pull the switch in oe-core now, as there are still not a lot of packages? Aug 25 14:17:38 well, dunno, if it were up to me (which it isn't, obviously) I would not want to just pull the switch without checking to see what would actually break as a result Aug 25 14:18:57 and, of course, there are other layers like meta-oe which inherit oe_runmake from oe-core. Aug 25 14:34:22 yeah, it would have to be carefully done. I think ideally we'd go through all the recipes and find any whose EXTRA_OEMAKE contains -e in some fashion, and transition them to explicit variable passing Aug 25 14:34:37 then, at some hypothetical future time, consider changing the default Aug 25 14:54:18 * cbrake is working on a oe-core/meta-oe build and missing lots of packages that are available in OE classic ... Aug 25 14:54:29 so copying them over ... Aug 25 14:56:05 yeah, there is lots of stuff that still hasn't been migrated Aug 25 14:56:27 and, of course, the fragmentary nature of the oe-core layer universe makes it hard to tell exactly what has and hasn't Aug 25 15:04:55 pb_: we almost need some kind of auto-updated web-based search... Aug 25 15:05:20 yup, find is getting some extensive use here Aug 25 15:06:49 I say web-based because it would mean you wouldn't have to have every single layer checked out Aug 25 15:07:44 yeah, that would be handy Aug 25 15:07:51 sort of like the old ipkgfind thing did for feeds, I guess Aug 25 15:09:10 yes Aug 25 15:09:27 I don't have much in the way of webdev skills or I'd try to throw something together... Aug 25 15:21:10 I've occasionally thought about taking that a step further, and doing an apt-like tool that interacts with a server that parses recipes, which pulls down only what you need Aug 25 15:35:10 yeah, that's an interesting idea. might be a bit too scary though. Aug 25 15:36:07 I think what I would want is a tool to tell me what layer(s) contain a particular recipe, then I can take my own view on whether or not I think the maintainer of the layers in question is a lunatic before deciding to use the code Aug 25 15:41:37 pb_: well clearly it would need to tell you what layer it was in, otherwise you wouldn't know where to get it... Aug 25 15:42:04 ah, I thought kergoth's plan was that it would just download it automatically Aug 25 15:42:39 maybe I misunderstood what he meant, though if it wasn't that then I can't immediately think why it would be a step further than what you proposed. Aug 25 15:43:13 I don't know if it really could solve the problem that we have at the moment without being locally configured for every possible layer Aug 25 15:45:17 yeah, dunno. to a large extent, what I think we need at the moment is a more comprehensive set of "well known" and well maintained layers. anything that involves fishing around in random third-party repos is always going to be a little bit unsatisfactory. Aug 25 15:46:22 it's somewhat of a chicken-and-egg problem though... I suspect people look over the fence at oe-core, see that their favourite stuff is not there, then continue with oe-dev Aug 25 15:46:33 yeah Aug 25 15:47:15 besides, even a "poorly" maintained layer can be the source of a good recipe that could be pushed elsewhere Aug 25 15:48:45 yeah, indeed, I was thinking from the perspective of someone who just wants a recipe to use; ideally you want to feel that it is coming from a layer of generally good quality, rather than a layer which consists of junk maintained by crazed h4x0rs. Aug 25 15:48:59 true Aug 25 15:49:06 there is little documentation on oe-core that I've found Aug 25 15:49:19 haven't been following it closely, but only 2 wiki pages that describe it and then the meta layers Aug 25 17:08:49 toofar: well, I'd love to hear specific suggestions on what we need to document Aug 25 17:09:08 toofar: if you can give me some ideas I will try to write stuff down... Aug 25 17:12:22 * bluelightning -> home Aug 25 18:11:16 I've found a bug in the current oe-core binutils while building for PPC64.. anyone around have experience with binutils and section generaton.. Aug 25 18:11:26 I've been trying to figure out the cause and so far not had any luck.. Aug 25 18:11:38 (short story, the GNU_HASH section is not being written properly) Aug 25 18:57:15 Tartarus: have you looked at the differences between cpan*.bbclass between oe-core and oe, by chance? Aug 25 18:57:25 RP_: ping Aug 25 18:58:10 RP_: ah, nevermind Aug 25 19:10:02 kergoth__: not in a while Aug 25 19:10:14 iirc they were mostly re-synced Aug 25 19:12:17 yeah I think so, I was curious about the whole cpan unwrapping thing, is that a nonissue with the perl/nativeperl bits? Aug 25 19:12:28 I noticed that find/sed isn't in oe-core Aug 25 19:21:53 There was one, at one point for it Aug 25 19:21:58 but maybe it got re-re-done Aug 25 19:45:50 hmm. Aug 25 19:46:09 It'd be nice to have a convenient way to emit binary packages for all defined package groups/image features Aug 25 19:46:21 in case one wants them in a feed Aug 25 20:35:36 this no long seems to be working with oe-core Aug 25 20:35:38 SRCREV = "${AUTOREV}" Aug 25 20:35:39 PV = "1.0+svnr${SRCREV}" Aug 25 20:35:58 I end up with a AUTOINC instead of the rev # in the package name Aug 25 20:36:47 # AUTOREV=${@bb.fetch2.get_autorev(d)} Aug 25 20:36:47 AUTOREV="AUTOINC" Aug 25 20:36:56 hmmm Aug 25 20:37:17 * kergoth has no clue what's being done with fetch/fetch2 sadly Aug 25 20:40:49 OE classic has: Aug 25 20:40:51 SRCPV = "${@bb.fetch.get_srcrev(d)}" Aug 25 20:40:51 AUTOREV = "${SRCPV}" Aug 25 20:41:04 guess I need to ask on the maillist what is going on ... Aug 25 21:59:31 I am trying to build an image with OE and have included one of my own recipes in the image dependency Aug 25 21:59:38 but not all files show up in the filesystem Aug 25 21:59:59 the ones I install into etc/init.d/ from do_install show up Aug 25 22:00:22 but the ones I installed into boot/ in my recipe don't show up in the final filesystem image Aug 25 22:03:09 any ideas why? Aug 25 22:03:33 There are present in image/ in my recipe's temp dir Aug 25 22:19:30 joel_: see the FILES_ variables in bitbake.conf. only certain files go into certain packages. Aug 25 22:19:42 joel_: you'll lkely want FILES_${PN} += "/boot" Aug 25 22:23:46 joel_: It's important to look for installed but not shipped messages when building custom recipes Aug 25 22:41:25 bluelightning: in response to your question about oe-core, it has more to do with just learning about the oe roadmap Aug 25 23:25:40 kergoth, Tartarus thanks a lot :) Aug 25 23:33:49 toofar: so, as I said, let's have some specifics... ? Aug 25 23:38:42 When I send in patches, can I have a subject like: Aug 25 23:39:26 "[PATCH 1/10] [mplayer-dependency] libdvdread: Fixed foo bar baz" Aug 25 23:40:12 Like I want a way to group a set of patches into a common category and would like it to be mentioned in the Subject line Aug 25 23:44:54 But git am wont like it Aug 25 23:45:17 maybe [PATCH 1/10 - mplayer depedencies] ? Aug 25 23:47:49 So, are you using git send-email? Aug 25 23:47:55 Tartarus, yes Aug 25 23:48:29 OK, so it's a matter of fiddling with --subject or --subject-prefix Aug 25 23:48:41 for meta-oe+oe-core, the pull request scripts also have an option to set this for you Aug 25 23:48:57 Is it acceptable to put anything in [PATCH ] ? Aug 25 23:49:17 but usually it's best to just have your 0/10 be: [PATCH 0/10] Fixup mplayer dependencies and then just have everything else follow w/ the normal PATCH N/10 or PATCHv2 N/10, etc Aug 25 23:49:17 Just concerned about the policy , not the method ;) Aug 25 23:50:01 Tartarus, Ah, that sounds nice. So like 0/10 should be an empty commit? Aug 25 23:50:14 Nope, --compose Aug 25 23:51:09 Tartarus, so what goes into the body of the email? Aug 25 23:51:19 That files up $EDITOR Aug 25 23:51:28 Give it a whirl with a --to yourself:) Aug 25 23:51:43 That's something I do and kick myself when I don't, heh Aug 25 23:52:40 ah , gotcha :) Aug 25 23:52:50 thanks, I should rtfm more often Aug 25 23:54:11 heh **** ENDING LOGGING AT Fri Aug 26 02:59:57 2011