**** BEGIN LOGGING AT Thu Nov 07 02:59:58 2013 Nov 07 09:36:20 morning all Nov 07 09:40:36 mourning , too Nov 07 10:28:08 good morning Nov 07 10:37:12 hi all i have used poky with meta atmel layer,i have added a bluez-tools package externally,and it compiled properly bitbake bluez-tools,but when i create a rootfs it is not detected in target device like bt-obex... my bb file is http://pastebin.com/CeNtqPBm can you tell me the issues Nov 07 10:40:20 linu1: you probably need to add those tools into your image Nov 07 10:42:21 * bluelightning mutters to himself about layers not in the layer index... Nov 07 10:42:31 RP of course i have added that package into filesystem image as bitbake root-image,but those binaries are not in the target system Nov 07 10:44:04 linu1: can you be more specific about how you did that since something clearly isn't working Nov 07 10:44:53 You'd have thought autotools would be consistent about its cache file format wouldn't you :( Nov 07 10:53:21 RP,i just added 4 external packages for my at91sam9x5ek target, bluez-tools,fuse,obexd, obexfs obexftp, i have compiled each package separately like bitbake fuse;bitbake obexftp....., and finaly i created rootfs as bitbake 72xx-sdk,when i try to search bt-adapterbt-device, there are no files in target my bb files http://pastebin.com/mVi5a0xJ Nov 07 10:55:48 linu1: did you add these recipes to 72xx-sdk though ? Nov 07 10:56:18 linu1: you shouldn't have to build them indvidually, if the image depends on them, it will trigger the build of them Nov 07 11:01:16 RP,ya i forgot to add into 72xx-sdk,i got it..actually i am fresher to poky..thanks RP Nov 07 11:07:24 is sharing downloads and sstate over all of the system a good or bad idea? even when building for different aches? Nov 07 11:10:57 a very good idea Nov 07 11:13:04 rburton: so x86 vs. arm or such is not going to get me into troubles? Nov 07 11:14:07 LetoThe2nd: downloads are obviously architecture independant, and the shared state handles (and numerous other inputs that change the binaries) Nov 07 11:14:51 rburton: would've guessed it for the downloads, but the sstate is not completely in my head by now. thx. Nov 07 11:15:30 LetoThe2nd: in fact sstate is *designed* to be shared between builds/machines/etc. Nov 07 11:15:37 rburton: hm ok Nov 07 11:16:43 even over machines, e.g. several developers, that is? Nov 07 11:16:45 yes Nov 07 11:16:53 hmhmhm Nov 07 11:17:00 host-specific bits are split into per-distro directories Nov 07 11:17:48 RP i have added those packages to target rootfs it shows http://pastebin.com/TVsB79jD can you help me Nov 07 11:18:04 rburton: ok, i think i get it. Nov 07 11:25:39 Hi guys, I have a sysvinit script which does 'while true do ; mybinary ; done' in runlevel 5. I would like to create a stopping behaviour as well, any ideas? Nov 07 11:29:05 killall? Nov 07 11:56:02 hi all i have used poky with meta-atmel,i have tried to add obexfs package in my project externaly it shows the error http://pastebin.com/Epg7cYDR my bb file is http://pastebin.com/9HuUDy9J can you help me Nov 07 11:57:18 well find out why the configure failed. Nov 07 11:58:14 it even tell you which log to check. Nov 07 11:58:19 tells, that is. Nov 07 12:01:11 LetoThe2nd, you mean this file http://pastebin.com/zghLfN9S Nov 07 12:01:47 linu1: sorry, but why do you always only post error messages, but not read them Nov 07 12:02:20 linu1: line 64 is very clear about what fails. go check your dependencies. Nov 07 12:23:48 Hi guys... one question regarding dash/bash ... i m getting a "error: Can't install pm-utils-1.4.1-r1@x86_64: no package provides /bin/bash" on do_rootfs.... should bash be added to pm-utils recipe ??? Nov 07 12:40:22 linu1: what are the names of the packages you're trying to install? Do packages of those names exist in tmp/deploy/{ipk|rpm} ? Nov 07 12:45:06 Hello. Nov 07 12:45:24 Does anyone know how I can see which package depends on X package? I get ofonod as part of my image, and I'd rather take it out. Nov 07 12:46:09 Does anyone know how I can search my image to see which recipe ends up depending on it? Nov 07 12:46:13 RP i have added total four 5 packages 3 packages added successfully and rootfs also created successfully but the obexfs and obexftp even does not compile,i have some issues in bb file Nov 07 12:46:31 RP i have added total 5 packages 3 packages added successfully and rootfs also created successfully but the obexfs and obexftp even does not compile,i have some issues in bb file Nov 07 12:46:35 like i said - check the inter-recipe dependencies. Nov 07 12:47:25 LetoThe2nd, yes i am working on it Nov 07 12:51:15 RP yes ussp-push and bluez-tools have ipk in tmp/deploy Nov 07 12:55:48 stygia: try 'bitbake -g ' and look at the generated dot file "package-depends.dot" Nov 07 12:56:19 linu1: If the ipk files are there it should be able to install them... Nov 07 12:56:24 zibri, Ah, cheers. Nov 07 12:56:48 zibri, I looked at that and figured I could use it to see what package X depended on, but I can always grep it for the image itself. Nov 07 12:58:53 RP: from the configure log in linu1's case, it seems that package c does not even get built, because it does not find package a or b. i don't think that the packages directory is the place to search, then. Nov 07 13:01:19 LetoThe2nd: If there is a package missing then that is a different issue, yes. That wasn't clear from the log fragments that I saw :/ Nov 07 13:02:18 13:01 < linu1> LetoThe2nd, you mean this file http://pastebin.com/zghLfN9S Nov 07 13:02:24 RP ^^^^ Nov 07 13:42:51 rburton: the core-image-minimal-directfb build we talked about yesterday was successful after removing the check. i have no means of checking the result, but shall i file that as a bug then? Nov 07 14:09:54 LetoThe2nd: please do Nov 07 14:16:39 Is there any recipe for Python 3? Nov 07 14:16:59 kbart: there have been patches sent to oe-core previously, yes Nov 07 14:17:08 by otavio, iirc Nov 07 14:17:38 by khem Nov 07 14:19:12 http://cgit.openembedded.org/openembedded-core-contrib/log/?h=kraj/python3 Nov 07 14:19:17 kbart: ^ Nov 07 14:21:03 rburton, bluelightning thank you both Nov 07 14:21:34 Do you know by chance if it's ready for use or still under development/experimental? Nov 07 14:22:14 last I heard it still had some build-time issues Nov 07 14:22:32 (otherwise I think it would have been merged by now) Nov 07 14:23:56 I see.. Nov 07 14:29:07 rburton: #5504 Nov 07 14:35:06 * kergoth yawns and stretches Nov 07 15:01:22 Hi all, I got gumstix-console-image working very well. Any hints on how to setup wifi? and what recipes should I include for wifi? Nov 07 15:01:43 parekh: connman is one option Nov 07 15:01:57 or just do the old fashioned way in /etc/network/interfaces Nov 07 15:04:30 rburton: Ok. Thank You. Nov 07 15:06:11 rburton: I am unable to see what is inside of a any recipe at layers.openembedded.org. When i search for a recipe and when i open it it shows just blank Nov 07 15:06:34 parekh: tell bluelightning Nov 07 15:07:23 parekh: can you be more specific? Nov 07 15:07:36 parekh: i.e. can you paste the URL that's showing up blank for you? Nov 07 15:11:31 bluelightning: When I open any recipe it shows like this http://imgur.com/rwjXqDP Nov 07 15:14:03 Hi all. If I want to temporarely change u-boot files to try something - where do I make the changes? In STAGE_DIR? In /tmp/work dir? Nov 07 15:14:18 I.e. which sources are used for compilation? Nov 07 15:15:37 Found it in /tmp/work... Nov 07 15:20:54 parekh: hmm, seems like it's a problem with cgit rather than the layer index, if I browse from the root of cgit.openembedded.org I see the same problem Nov 07 15:23:24 bluelightning: ok. Nov 07 19:09:03 it's pretty quiet in here. i guess that means everything is working the way it should :) Nov 07 19:10:44 i've got a bitbake environment question, if anyone has a minute to assist Nov 07 19:11:12 don't ask to ask, just ask :) Nov 07 19:13:00 :) i'm trying to get bitbake to not use any data from a users ~/.bashrc file to keep any of their customizations out of the build... i'm checking it with bitbake -c devshell Nov 07 19:13:15 IRC is usually pretty asynchronous, often it's most effective to post a question and check back a day later. :) Nov 07 19:13:52 most notably the PATH variable which seems to allow the user's .bashrc file to override any of the bitbake environment Nov 07 19:14:22 Huh, that's a fascinating question. Nov 07 19:14:57 well, one thing: .bashrc is run only for interactive shells, in principle. Nov 07 19:15:19 And it may be sort of intentional that a devshell would pick up your normal environment, because otherwise it'll feel broken or unfamiliar. Nov 07 19:15:40 so you are saying that the normal bitbake environment would not have included the .bashrc file? Nov 07 19:15:45 rumdisk: bitbake already filters the environment heavily. any additions it does to the PATH are prepended, so override what the user had Nov 07 19:15:45 but the devshell would Nov 07 19:16:00 so i think this theoretical concern really isn't an issue Nov 07 19:16:35 hmm, the adt code needs someone to replace all mentions of poky with oe-core Nov 07 19:17:45 i had assumed that the devshell would operate in the exact environment that bitbake would run in when processing a target (i think i read that in the doc somewhere) Nov 07 19:18:25 upon noticing that the PATH had a prepended adjustment to the path that my .bashrc contains, I became concerned Nov 07 19:18:43 I don't really see why that's a concern Nov 07 19:19:30 for example, if I have ~/bin prepended to the path that contains something like install or make Nov 07 19:20:23 perhaps the full paths are given during the builds and it doesn't rely on path lookups Nov 07 19:21:14 imo the answer is, if you want sane builds, don't run builds on users that do that, or just plain dont do that Nov 07 19:21:37 but you could always override the PATH in your distro or local.conf to a minimized set, if you'd rather do that Nov 07 19:22:46 i would agree that overrides in the PATH from .bashrc is probably not best practice... the modification to local.conf hadn't occurred to me. i might do that Nov 07 19:23:27 boy is it hard to convince some people that windows is a crap platform for judging compliance with external standards Nov 07 19:23:44 i'm working on scripting builds in a way that the random user can run them without breaking them because they changed something that seemed like a good idea to them :) Nov 07 19:24:08 NTP being the current issue... Nov 07 19:25:03 can't i even get a "kergoth rolls eyes" for that one? Nov 07 19:31:30 rumdisk: If you really want deterministic, reproducible, isolated builds for all engineers, you might consider distributing VMs, or a chroot environment Nov 07 19:31:36 rumdisk: something to think about, anyway Nov 07 19:32:06 thanks, i'll give it some thought and see where i end up Nov 07 19:52:31 Hello all. Is KERNEL_DEFCONFIG only for poky kernels? Nov 07 20:18:44 jwhitmore. no, it isn't for them at all :) Nov 07 21:13:02 zeddii: sorry went for food. I'm a bit confused by your answer, sorry. I asked was the variable only for poky to which you replied "no". So I assume it's for them all. But you added that "it isn't for them all" Apologies but can you clarify? Nov 07 21:13:39 "poky" is the reference distroy for Yocto. Nov 07 21:13:52 that variable is used by the meta-raspberrypi kernel recipes, and no one else. Nov 07 21:14:14 so that's why the answer was no. it'll only work for that one board layer. Nov 07 21:14:55 Thanks. I'm having problems with it on the raspberrypi kernel. It basically does.. well it don't seem to do what I tell it. Nov 07 21:15:55 there have been some other threads on the mailing list(s) just recently on just that subject. Nov 07 21:16:16 what is it not doing ? not using your config ? Nov 07 21:16:17 well the most recent is me ;-) Nov 07 21:16:24 thought so :) Nov 07 21:16:53 did I follow up to yours ? I know I replied to one or two .. but I don't have my sent box handy. Nov 07 21:16:53 I just thought I'd check here before I made a fool of myself on the mailing list with a variable that I should not be using Nov 07 21:17:42 if the question is about SRC_URI order/finalization or typical variable overrides .. it applies on the lists. but if it is about the specific handling of it by the recipe, then the raspberrypi lists would be appropriate. Nov 07 21:20:52 Hmm I would have thought that the poky build system enables a user to configure the build that they're trying to perform and every layer should be consistent with that. If Raspberrypi layer starts doing it's own stuff then should I be reading the yocto documentation or forgetting that as it isn't relevant at all. Maybe the simplest thing would be to install gcc on the raspberrypi and just do the kernel work there. OK so that's 8 hours of a build probab Nov 07 21:20:53 ly ;-) Nov 07 21:22:00 well that's the thing. if any layer has recipes with routines and special processing of any given part of the recipe .. the core supports that, but if the processing in the recipe doesn't work .. that's the recipe author's that know best. Nov 07 21:22:50 Agreed but the raspberrypi kernel recipe is written in a way that precludes changing the config file. What's the sense in that? Nov 07 21:23:24 you'd have to ask the authors :) I'm only familiar with what is happening in that recipe based on what I've seen asked on the lists. Nov 07 21:24:11 I could email the maintainers of that layer. I was just trying to rewrite that recipe to do it the more established way Nov 07 21:24:15 cross post ;-) Nov 07 21:24:47 the thing about it is .. is that clearly they want you to change the config (hence the variable), but what I'm trying to glean from the list posts is .. has something changed in the base that changed the compute order ? Nov 07 21:24:48 I'll email the yocto list and CC them. Nov 07 21:26:18 their recipe has a do config_prepend() which hard codes the config file that they're using. I was trying to change it to use SRC_URI Which is what the hard code. But poky ignored my KERNEL_DEFCONFIG SRC_URI attemt Nov 07 21:26:25 zeddii: yes, there are recent quirks with overrides and in some cases it can be painful Nov 07 21:28:22 My thinking was that if their recipe used KERNEL_DEFCONFIG=${SRC_URI}/..... Then my bbappend could override that KERNEL_DEFCONFIG. That sounds like it's how it should work. It's not how it seems to Nov 07 21:29:11 I'll post on the list and CC the authors, but thanks you've been very helpful Nov 07 21:32:21 jwhitmore, you can cc' me as well if you like, I can poke at things to see if the base changed unexpectedly Nov 07 21:38:16 Zeddii: I'm only new on the mailing list so not sure your name to find your email address? Nov 07 21:38:27 bruce.ashfield@windriver.com Nov 07 21:38:37 thanks a million! Nov 07 22:07:38 damn do i feel dirty... Nov 07 22:08:17 * mr_science just had to install NTP on a windows box Nov 07 22:14:45 * kergoth fixes a little update-rc.d bug Nov 07 23:13:31 ant_home: good news, I understand your overrides issue Nov 07 23:13:41 rburton: same bug as you found Nov 07 23:42:11 RP: great, then fix is coming ;) Nov 08 00:53:52 RP: why nodistro and not oe-core? We already have DISTRO_VERSION = "oe-core.0" Nov 08 00:54:26 RP: the whole point is that THIS ought to be the default distro, not Poky ;) Nov 08 00:58:33 anyway, thx, I'll test tomorrow Nov 08 00:58:34 gn **** ENDING LOGGING AT Fri Nov 08 02:59:58 2013