**** BEGIN LOGGING AT Thu May 23 03:00:00 2013 May 23 04:30:23 sgw_: patches submitted May 23 04:43:47 I addtask'ed a custom task to my libgcc recipe, and have set [depends] on it...but bitbake-dumpsig claims my task doesn't depend on any other tasks...is there something special about libgcc? Same thing in other recipes and it _does_ show up in bitbake-dumpsig.... May 23 04:48:02 I'm trying to get a shared-work like effect on my custom task between libgcc and other recipes. If it weren't for this one bizarre difference, the task hashes would be equal and it would be working.... May 23 05:01:59 huh...the dependency does show up in task-depends.dot though...weird.... May 23 05:45:41 okay...what am I missing? I define do_foo, addtask foo..., do_foo[depends] = "thing-native:do_populate_sysroot", then bitbake-dumpsig the sigdata for foo, and it claims "Tasks this task depends on: []" for every recipe except the ones with -cross in their name. However, the task dependency shows up in task-depends.dot.... May 23 06:20:26 I'm trying to build core-image-minimal but the package prelink gives me an error: http://pastebin.com/aXWiJu99 May 23 06:20:33 What could be the problem? May 23 07:11:00 good morning May 23 08:51:21 morning all May 23 09:09:37 kergoth`: i don't suppose you have a fix for bb not working with bitbake master yet? :) May 23 10:19:21 kergoth`: very much a hack job on making bb work with new bitbake on my fork May 23 10:19:44 kergoth`: it does remove a lot of duplicate code from bbcmd which is good, but i'm mostly doing it without clue :) May 23 11:02:59 What recipes do I need let gnome3 work on core-image-minimal? May 23 11:03:59 packagegroup-core-x11-base and gdm? May 23 11:26:29 http://marcin.juszkiewicz.com.pl/2013/05/23/my-uk-trip-london/ May 23 11:26:32 ops May 23 12:01:20 hi May 23 12:01:47 how can i rebuild a single package? i've changed some source files and therefore a rebuild is needed May 23 12:02:22 derRichard: bitbake May 23 12:02:52 Zagor: this does nothing. bitbake thinks nothing has to be done May 23 12:03:00 bitbake -c clean May 23 12:03:01 i've replaced the tarball in downloads May 23 12:04:19 nope May 23 12:04:54 now run "bitbake " again May 23 12:04:58 i'll paste you the logs May 23 12:05:01 one moment May 23 12:05:16 possibly you'll even want -c cleansstate to avoid it pulling in the sstate cache of the old version May 23 12:06:21 see: http://paste.debian.net/plain/5973 May 23 12:06:30 will try cleanstate May 23 12:08:34 Zagor: thanks a lot! cleanstate did it! :-) May 23 12:08:43 excellent May 23 12:09:56 what is the perfered way to work with sources in yocto? i'm changing all the time some source files within packages. May 23 12:11:49 Is there a packagegroup available for yocto images to run gnome? I'm trying to create an image with it but I can't seem to succeed May 23 12:15:10 Something like angstrom-packagegroup-gnome May 23 12:25:30 derRichard: don't use bitbake for software development. rather build your test system in yocto and then copy your test binaries there with scp, nfs or some other way. May 23 12:25:54 rebuilding the whole system for each test during development is overkill May 23 12:27:07 look, currently i'm hunting down a bug in wvdial on linux/arm. so i have to patch various packages all the time (eglibc, wvstreams, wvdial). if i'd build all these all the time by hand. i'd not need something like yocto May 23 12:28:26 derRichard: ah, ok. May 23 12:30:19 currently i use bitbake -c cleanstate foo ; bitbake -c fetch foo ; patching the source taball... ; bitbake foo May 23 12:30:25 this kinda sucks :D May 23 12:35:22 derRichard: one option is to clone the repos of packages you modify and work against those (change SRC_URI and SRC_REV in the recipe) May 23 12:36:50 if you're wanting to compile after changing source code under the workdir, you should use bitbake -c compile -f foo May 23 12:37:22 or to do that and then continue on and do the rest of the tasks for the recipe: bitbake -C compile foo May 23 12:38:41 ah, -C is good to know May 23 12:38:56 bluelightning: what is -C? May 23 12:39:03 bitbake -h does not offer this switch May 23 12:39:26 derRichard: really? it does for me. which version are you on? May 23 12:39:28 derRichard: it means force the specified task and then run the default task for the specified recipe (as if you'd done bitbake foo) May 23 12:39:38 derRichard: you're probably using an old version of bitbake then May 23 12:42:49 bluelightning: BitBake Build Tool Core version 1.15.1, bitbake version 1.15.1 May 23 12:43:08 derRichard: yep that version does not have that feature May 23 12:43:18 it should support -c compile -f however May 23 12:44:05 ok May 23 12:52:21 bluelightning: hi bluelightning! May 23 12:52:28 hi Net147 May 23 12:53:08 bluelightning: how's it going? May 23 12:53:18 Net147: great thanks, and you? May 23 12:53:39 bluelightning: yeah, all okay May 23 12:55:33 bluelightning: Yocto is pretty large to understand, but i'm learning gradually May 23 12:56:47 Net147: thanks for your recent patches May 23 12:57:06 bluelightning: you're welcome May 23 13:11:10 is there a list of boards that Yocto is commonly tested/used with? May 23 13:14:27 Net147: the Yocto Project QA team tests with the four reference hardware machines (atom-pc, beagleboard, mpc8315e-rdb and routerstationpro) May 23 13:15:12 bluelightning: any particular brand/model of atom-pc? May 23 13:16:00 BSP authors will do their own image testing and we build images for some of these on a regular basis including BSPs from Intel, Freescale ARM and Freescale PPC May 23 13:16:34 Net147: if you have a look at the README.hardware I think we list a bunch of machines we have tested with although it should work with a wide range of machines May 23 13:17:49 bye! May 23 13:20:18 bluelightning: does that include Atom boards currently in production using PowerVR SGX graphics (e.g. on newer Acer Aspire One netbooks)? May 23 13:20:57 Net147: we wouldn't build the proprietary drivers for those no, but the machine should still boot I would think May 23 13:21:30 Net147: we do have some atom-based BSPs in meta-intel that do build the Intel EMGD drivers for similar platforms May 23 13:23:48 bluelightning: I think BSP was removed because the PowerVR graphics drivers aren't compatible with the 3.8 kernel May 23 13:25:40 bluelightning: Intel Cedartrail BSP May 23 13:26:47 bluelightning: http://git.yoctoproject.org/cgit/cgit.cgi/meta-intel/commit/?id=1d6b218a45e43441ca6299b899c4d84ec976082c May 23 13:28:26 Net147: that and that ISG were not support cedartrail anymore May 23 13:28:45 rburton: ISG? May 23 13:28:57 its mainly due to the kernel thing May 23 13:29:01 they weren't interested in porting May 23 13:29:30 however, patches welcome! the kernel bit is small and easy, so if someone wants to use cedartrail, they can update the kernel driver bit May 23 13:29:36 can EMGD drivers be used for Cedartrail platform? May 23 13:29:42 no May 23 13:29:50 aww May 23 13:30:34 I am using xf86-video-modesetting and kernel gma3600 support at the moment May 23 13:31:10 will probably need porting to newer X.Org as well May 23 13:31:28 Net147: meta-intel has its own X for the binary drivers May 23 13:31:55 rburton: I see. so perhaps just kernel stuff. May 23 13:32:04 Net147: the cedartrail bsp has the binary drivers May 23 13:32:25 rburton: okay May 23 13:32:50 maybe I will port it to newer kernel if I get bored May 23 13:33:25 one of the intel team does have a cedartrail netbook so he can test it in his spare time/gets bored, but he's also fully loaded with stuff that's a priority :) May 23 13:34:27 rburton: I got a whole shelf full of cedartrail boards May 23 13:35:03 Net147: patches very welcome then :) May 23 13:35:43 rburton: only at work. at home I got a cedartrail netbook though. May 23 13:36:40 isn't Ubuntu also doing some porting of the drivers to newer kernels? May 23 13:37:01 Net147: no idea if they're actually supporting cdt going forward May 23 13:37:30 rburton: well I imagine there would be many netbooks and Atom boards using it, until they produce a board that uses another chipset May 23 13:39:57 Net147: yes, you would May 23 15:12:05 hum... I'm trying to include grub_0.97 in my image, and I get 'checking whether i586-poky-linux-objcopy works for absolute addresses... no' and the configure fails. May 23 15:22:42 does it just not work, or do I need to do some magic? May 23 15:23:33 Garibaldi|work: why not use grub2? May 23 15:24:37 well, there are GPL-v3 concerns about us building grub2; the lawyers are extra-cautious May 23 15:25:16 and I'm trying to install it on a disk image (not an actual device) and I haven't figured out how to get anything grub2 gives me to do it. May 23 15:28:00 I can, for instance use old grub's 'grub' command to install grub on a disk image file. May 23 15:29:53 I actually tried grub2, but when I build it into my nativesdk, there were path issues (grub-mkimage: error: cannot open `/opt/poky/1.3/sysroots/x86_64-poky-linux/usr/lib/grub/i386-coreboot/moddep.lst); I didn't install the SDK in /opt/poky/1.3/ May 23 16:19:37 maybe I can find a different boot loader May 23 16:29:44 otavio: hi May 23 16:30:39 what do you mean with "author name not complete"? May 23 16:31:04 Garibaldi|work: syslinux? May 23 16:39:27 otavio: have to go now. Will reply you to your email, 'cause it's not clear to me what I'm missing May 23 16:41:39 panda84kde: Your .gitconfig is (were) not properly set May 23 16:43:18 ok, but were do you see that? What's missing? May 23 16:43:39 panda84kde: 'Diego <...>' not 'Diego Surname <...>' May 23 16:45:31 I'm searching in the mail, but can't find any occurrence of "Diego <...>" May 23 16:45:45 forgot: otavio May 23 16:47:08 otavio: this is the good one. Are you referring to this? https://lists.yoctoproject.org/pipermail/meta-freescale/2013-May/002816.html May 23 16:52:02 should go. bye everybody May 23 17:24:38 * mranostay yawns May 23 18:34:20 hi otavio, I'm Diego. I'm back if you want to clarify the problem with the wrong name May 23 21:29:26 any siggen.py gurus out there? I'm trying to do a shared-work sort of thing, but there's something I don't understand happening which is making my custom task have one hash in some recipes and another in the rest. If I bitbake-sigdump, the only difference is some say "Tasks this task depends on: []" while others say "Tasks this task depends on: ['customthing.bb.do_populate_sysroot']" May 23 21:30:26 thing is all the recipes in question inherit a bbclass with do_mycustomtask[depends] = "customthing-native:do_populate_sysroot" at the bottom May 23 21:31:19 so I'm assuming the siggen code sees the inter-task dependency in the one case and somehow doesn't see it in the other May 23 21:32:19 weird May 23 21:33:38 the really interesting part is that task-depends.dot shows a task dependency everywhere, even bitbake-sigdump claims there aren't any May 23 21:33:55 s/even/even though/ May 23 21:34:47 maybe richard has an idea, I can't think of anything that could cause such a weird behavior offhand May 23 21:35:01 could use getVarFlag to sanity check that the depends flag is what you think it is in all cases May 23 21:35:15 (i'm sure it is, but i tend to jump to bitbake -e equiavlent when things start beahving weird, to check assumptions) May 23 21:35:19 heh May 23 22:38:18 khem: did you want me to squash your uclibc patches into the systemd patch? May 23 22:48:21 * kergoth kicks the PR server repeatedly May 23 22:48:30 stupid going backwards errors when i didn't chnage anything May 23 22:57:01 evanp: The sstate code doesn't track over the native to non-native boundary May 23 22:57:42 evanp: So it will notice depending on and depending on but but not crosses of these May 23 22:58:10 evanp: I suspect in your case this is causing the problem since you mention both a -native and a non-native? May 23 22:58:25 kergoth: any objections to killing the global method space in bitbake? May 23 23:00:42 RP: I'm all for it May 23 23:00:44 RP: an example pair of affected recipes are my custom gcc-cross and custom libgcc, does that jive with what you are saying? May 23 23:00:48 just causes problems, really May 23 23:01:00 we started in that direction when we added the context stuff, so this is just the next logical step May 23 23:01:13 (saw the patches on the list this morning, hadnt had time to replyt hough) May 23 23:01:58 kergoth: I thought you'd agree, just thought I'd ask though. It continues down the road and is a good thing... May 23 23:02:04 * kergoth nods May 23 23:02:21 evanp: so that is mixing cross and target which will likely suffer as I mentioned May 23 23:02:23 we'll need this sort of thing to be able to have multiple builders in a single process space May 23 23:02:40 evanp: This is one of the reasons we stash libgcc into the sysroot and then libgcc simply packages it May 23 23:03:28 kergoth: I was a bit horrified when I realised resetting the cooker didn't touch the contexts :/ May 23 23:04:05 heh, indeed May 23 23:04:59 kergoth: the parallel parsing was also causing people interesting one build in twenty error messages from methodpool May 23 23:08:52 * RP -> Zzzz May 23 23:50:59 what's up with OECMAKE_BUILDPATH? why doesn't it just obey ${B}? May 23 23:57:18 * sgw_ reminds people that they should put in expense requests for Q3 May 23 23:57:31 * sgw_ sez oops wrong window! May 24 00:19:29 * kergoth wonders why ptest isn't implemented as PACKAGECONFIG, rather than DISTRO_FEATURES directly May 24 00:20:21 the EXCLUDE_FROM_WORLD Is extremely questionable too, not all ptests are separated into separate recipes May 24 00:39:16 hi kergoth May 24 00:41:43 sgw_: i know i know :) **** ENDING LOGGING AT Fri May 24 02:59:58 2013