**** BEGIN LOGGING AT Thu Oct 31 02:59:58 2013 Oct 31 03:00:02 ah, so it builds a native (as in x86) version as well as a cross-compiled version Oct 31 03:00:26 ERROR: This autoconf log indicates errors, it looked at host include and/or library paths while determining system capabilities. Oct 31 03:00:26 Rerun configure task after fixing this. The path was '/home/sarnold/poky/build-rpihf/tmp/work/arm1176jzfshf-vfp-poky-linux-gnueabi/erlang/R16B02-r0/otp_src_R16B02/erts' Oct 31 03:00:29 I haven't tried running a native build outside bitbake Oct 31 03:01:17 yep, same error it gave me Oct 31 03:02:03 when I drilled down on this, it's the qa_configure greping the eras config.log file looking for problems like 'unsuitable for cross-compile' Oct 31 03:02:21 eras=erts Oct 31 03:06:01 that recipe is wrong... Oct 31 03:07:13 doh Oct 31 03:07:20 the native one? Oct 31 03:07:34 no, native builds fine Oct 31 03:09:06 so there's a problem with erlang_R16B02.bb Oct 31 03:09:16 looks that way to me Oct 31 03:09:36 is there a way to get bit bake to output exactly what it Oct 31 03:09:45 it's running Oct 31 03:10:37 when I look at the recipe it looks like it's just creating the comp conf file Oct 31 03:12:50 yes, but it needs to it right Oct 31 03:13:09 plus the "SKIP" thing doesn't work either Oct 31 03:17:05 so I do know that wx (haven't bothered to try and work out what that is) *doesn't* cross compile Oct 31 03:17:17 so perhaps if the skip thing doesn't work, that's causing the problem?? Oct 31 03:21:42 it looked like configure was okay with some of the missing deps Oct 31 03:22:08 best to disable them if the intent is a minimal erlamg install Oct 31 03:22:22 s/erlamg/erlamg/ Oct 31 03:22:33 I'm just a bit confused that when I run ./otp_build directly with the same conf file it all works fine Oct 31 03:23:21 when you say missing reps - do you mean the dlpi library? Oct 31 03:23:36 the headers are probably mostly the same , and configure finds the native lib path so it's happy Oct 31 03:23:49 hmm Oct 31 03:27:38 ok, thanks for your help Oct 31 03:27:51 will try to sort it out from here :-) Oct 31 03:34:58 still hacking the recipe Oct 31 03:35:19 configure stubbornly thinks it's not cross-compiling Oct 31 03:49:34 * nerdboy wonders if this recipe ever worked Oct 31 03:50:17 ah, one thing to be aware of - that otp_build configure step configures for *both* a native bootstrap version of erlang which is then used to build the cross-compiled version Oct 31 03:50:27 yup Oct 31 03:50:31 so configure is probably doing both native and cross-compiled Oct 31 03:50:48 i think that's why oe kills the configure Oct 31 03:51:40 which begs the question, why have a bootstrap build in the recipe when native is already built by the native recipe? Oct 31 03:51:42 yeah, although the warning about being unsuitable for cross compiling doesn't appear in the log when you run otp_build locally outside of oe Oct 31 03:52:16 or, why have native when the scripted build takes care of creating the bootstrap Oct 31 03:52:18 ? Oct 31 03:52:33 I remember reading somewhere that you need a native version to build the docs... Oct 31 04:27:34 this one is just weird... the whole bootstrap build thing needs an override or something... Oct 31 04:30:11 makes me feel better that I haven't worked it out Oct 31 04:32:06 well, mostly i have to work with oe-classic, which is why i made my own rpi layer Oct 31 04:33:43 so we pretty much need to disable the QA check that fails the bootstrap configure, or coerce erlang to build without building bootstrap first Oct 31 04:34:26 ie, make it think the native build *is* bootstrap Oct 31 04:47:58 yep, well I think that you can do that by just using configure directly as per the INSTALL-CROSS.md in the howto Oct 31 04:48:21 have tried that without success, but will try again later Oct 31 04:48:26 thanks for your help! Oct 31 04:48:41 have to go watch the kids play footy! Oct 31 06:36:18 i'm going back to the old recipe and update from there Oct 31 06:36:44 those new ones have only built in qemux86 Oct 31 07:12:25 After bitbake there is no UImage file and also no tar.gz, but there was only one warning... why? Oct 31 07:41:27 good morning Oct 31 09:40:52 Hey, I don't know how many people are online already, but I'm having a troublesome issue with my Perl @INC (include, library loading) paths. Oct 31 09:41:13 I have a recipe using the CPAN bbclass, and it lists it's @INC in such a way that it can't find Carp.pm Oct 31 09:41:19 It finds CGI/Carp.pm, but that's a different thing. Oct 31 09:41:33 Carp.pm, BTW, is a core module as per `corelist Carp` Oct 31 09:41:52 I am not exporting Perl, and I am using perl-module-carp in both DEPENDS and RDEPENDS Oct 31 09:42:02 bluelightning, Morning. Oct 31 09:42:07 Anyone know what could cause this? Oct 31 09:42:15 morning Stygia Oct 31 09:42:25 morning all Oct 31 09:42:37 bluelightning, You _just_ missed it, but I mentioned that I can't find Carp.pm (not CGI/Carp.pm) in the standard @INC (library include path) for my module, which is using cpan.bbclass and not exporting Perl Oct 31 09:42:55 stygia: where is Carp.pm installed, and what does @INC contain? Oct 31 09:43:27 zibri, http://pastebin.com/vHwjQv2T Oct 31 09:43:50 And, I find Carp.pm (Not CGI/Carp.pm) in /usr/lib/vendor_perl/5.14.3/ in the first dir of @INC Oct 31 09:44:40 try perl-module-carp-native? it's a bit ugly imho to explicitly depend on core modules though.. Oct 31 09:44:50 The module I'm creating a recipe for, BTW, is SQL::Abstract, i.e. libsql-abstract-perl Oct 31 09:44:53 ah Oct 31 09:45:08 zibri, Hmm? But AFAIK they aren't necessarily included unless you depend on them. Oct 31 09:45:11 i assume this is something that has to run on host? Oct 31 09:45:20 zibri, Yes this is compilation. Oct 31 09:45:29 stygia: yeah, but there's a perl-modules package iirc Oct 31 09:45:30 zibri, do_compile as generated by cpan.bbclass Oct 31 09:45:35 Sorry if that was unclear. Oct 31 09:45:45 zibri, Hmm, still. I'm not sure if I prefer being explicit to that. Oct 31 09:45:59 I could change my mind if I was given a good argument, but I'd lean towards slimness. Oct 31 09:46:46 sure, for embedded applications that may make sense. it's a balanace, but most upstreams will gladly add core modules without documenting the dependency Oct 31 09:47:21 zibri, Hmm, right. Oct 31 09:47:32 zibri, Well, thanks, I'll definitively consider that if I get trouble. Oct 31 09:47:53 zibri, Do you know how I could get Carp.pm in my .../usr/lib/perl/5.14.3/ though? Oct 31 09:47:55 And not native-perl. Oct 31 09:49:47 stygia: the issue you have is on host, not on target, right? Oct 31 09:50:10 zibri, Yes. Oct 31 09:50:30 Hmm it's actually failing on do_configure right now, it was do_compile, with the same error, at some point though. Oct 31 09:50:32 and Carp.pm only exists in /mnt/dev/fsl-community-bsp/build/tmp/sysroots/imx6qsabresd/usr/lib/perl/5.14.3/Carp.pm? Oct 31 09:51:29 No, only in tmp/sysroots/imx6qsabresd/usr/lib/perl/vendor_perl/5.14.3/Carp.pm, or of course subdirs to perl and company-image Oct 31 09:51:44 It _should_ be in .../usr/lib/perl/5.14.3/ Oct 31 09:51:46 ah... Oct 31 09:51:48 Seems to only be in vendor_perl Oct 31 09:52:54 I can, of course, sed on the script that gets called, FindBin.pm Oct 31 09:52:54 But that's hardly a solution Oct 31 09:53:24 How are other core modules installed? also in vendor_perl? Oct 31 09:54:39 Seems like strict.pm, for example, is in perl and native-perl. Oct 31 09:54:51 Not in vendor_perl... Oct 31 09:54:58 Hmm. Oct 31 09:55:14 Wait. How do I run things on native-perl, 'bitbake perl-vendor'? Oct 31 09:55:34 vendor-perl, sorry. Oct 31 09:56:41 what's vendor-perl? :) Oct 31 09:57:36 VENDORPREFIX in perl module toolchains is a place for a vendor (e.g. the distribution) to install perl modules. Oct 31 09:58:16 as opposed to perl itself (core modules, PERLPREFIX) or the system administrator (SITEPRFIX) Oct 31 09:59:57 Hmm, right. Oct 31 10:00:12 Then never mind, but I had a 'tainted' perl recipe for a while (long story, I conflicted with core modules), but it's cleaned up now. Oct 31 10:01:13 Interestingly the Carp.pm in the vendor_perl subdir remains after I cleanall perl and perl-native. Oct 31 10:13:48 Oh... Oct 31 10:14:00 I cleanall'd perl and native perl (again), recompiled, and not it exists. Oct 31 10:14:37 *now Oct 31 10:14:38 Weird. Oct 31 10:15:24 Hi all, How do I add a driver module to kernel in yocto? in debconfig? Oct 31 10:15:55 zaif, IIRC you can make a bbappend to the kernel config that adds a configuration file, like defconfig, where you can say MODULE=y Oct 31 10:17:35 zibri: hi, still awake? Oct 31 10:17:43 ant_work: sup? Oct 31 10:18:03 I'd have a naive kernel-question wrt sa1100 Oct 31 10:18:30 huh, are you thinking of somebody else? :) Oct 31 10:18:58 hm..donj' tell me !=ojn Oct 31 10:19:17 heh, not the same guy :) Oct 31 10:19:38 p) Oct 31 10:25:01 ant_work, What's the quesiton? Is it general to kernel stuff, or specific to yocto's build process? Oct 31 10:25:33 oh, someting very specific about ioremap and relocation Oct 31 10:25:41 sorry for the noise ;) Oct 31 10:34:50 ant_work, No worries, I'd help out if I could, but that does sound sorta beyond me, heh. Oct 31 10:37:50 yep, is for the other Olof ;) Oct 31 10:38:48 Multiple Olof's? Jævlå Svænskå. Oct 31 11:00:05 Stygia: Can I change CONFIG_VIDEO_VIVI=m in /yocto/poky/meta-gumstix/recipes-kernel/linux/linux-gumstix-3.5/overo/defconfig? Oct 31 11:00:22 for including vivi driver in my image? Oct 31 11:01:33 zaif, You can. Oct 31 11:01:49 Stygia: Done. Thanks. Oct 31 11:01:49 It's better if you make a BBAPPENd and have your own file, instead of messing with the recipes for the kernel itself. Oct 31 11:01:55 zaif, Wait. :P Oct 31 11:02:05 zaif, don't do that, you'll regret it. Oct 31 11:02:17 zaif, Make it easy on yourself, and mainable, by not tampering with the original recipe. Oct 31 11:02:26 zaif, Use a bbappend, and modify the recipe with your own recipe. Oct 31 11:03:02 zaif, Hang on, I can find you an example. Oct 31 11:03:06 zaif, This example is for buysbox. Oct 31 11:03:13 zaif, But it works the same way. Oct 31 11:03:22 Stygia: Where do I keep this bbappend file? in meta-gumstix/recipes-kernel/linux/linux-gumstix-3.5/overo? Oct 31 11:04:04 zaif, http://pastebin.com/ivvHx8R2 Oct 31 11:04:12 zaif, In your own directory. Oct 31 11:04:21 zaif, We have sources/meta-company/recipes-company/ Oct 31 11:04:30 zaif, To mimic the normal structure. Oct 31 11:04:46 zaif, I'd be happy to explain what the bbappend code itself does, it's fairly simple. Oct 31 11:06:44 zaif, I then keep the recipes I myself have made in the subdir recipes-company Oct 31 11:07:11 Stygia: I have a dir /poky/meta-bebot/conf & /recipes-bebot. So, this bbappend file goes in conf dir? is that right? Oct 31 11:07:46 zaif, So the busybox one has recipes-company/busybox/buysbox_1.20.2.bbappend and recipes-company/busybox/busybox/modules.cfg Oct 31 11:07:47 zaif, No. Oct 31 11:07:55 zaif, create a dircetory structure. Oct 31 11:07:58 Like you have sources/poky/meta/ Oct 31 11:08:07 Create sources/meta-company/recipes-company/ Oct 31 11:08:17 and/or things like sources/meta-company/recipes-kernel/ Oct 31 11:08:24 So you can easily organise your own recipes. Oct 31 11:08:36 zaif, I started up doing this a few months ago, trust me it's easier to just organise things well up front. :) Oct 31 11:11:52 meta-bebot is my own dir that I have created for my recipes. I kept my bebot recipes in recipes-bebot dir. So, it would be better to create a dir recipes-kernel in my dir and add the modules.cfg in that? Oct 31 11:14:42 Stygia: Understood the dir structure. Oct 31 11:16:42 Stygia: Now, my question is how do I add vivi kernel driver module for my image. by default it is set to #CONFIG_VIDEO_VIVI=is not set Oct 31 11:17:45 in defconfig Oct 31 11:21:50 zaif, Then, in the file you'd end up naming defconfig, you set CONFIG_VIDEO_VIVI=y Oct 31 11:22:30 zaif, For example, we have sources/meta-company/recipes-kernel/ Oct 31 11:22:45 Then we have linux-imx_3.0.35.bbappend Oct 31 11:23:29 zaif, That recipe looks like this: http://pastebin.com/Thvkw8m1 Oct 31 11:23:49 zaif, Then we have a subdir to that directory, i.e the dir 'linux' in recipes-kernel, files/ Oct 31 11:23:53 That has a file, defconfig. Oct 31 11:24:03 zaif, Which contains all the config options we want to set. Oct 31 11:24:13 zaif, I believe that would be .config after you run menuconfig, but I'm not 100% Oct 31 11:24:34 zaif, You would do that, and inthat config file, one of the options would be CONFIG_VIDEO_VIVI=y Oct 31 11:24:51 zaif, For example I have 'CONFIG_BLK_DEV_INITRD=y' in mine. Oct 31 11:25:09 zaif, Make sense? :) Oct 31 11:34:32 Stygia: Understood but a bit confused :) Will try that and get back. Oct 31 11:37:52 zaif, What're you confused about? Or, just overwhelmed? I'll be glad to explain something. Oct 31 11:38:33 I know it's sort of banned or unwelcome topic, but did TSC change opinion about whitespace in shell tasks? (would it make sense to send patches to use unified whitespace?) Oct 31 11:43:13 JaMa: surely the problem of backporting patches over such whitespace changes is still present? Oct 31 12:02:06 Stygia: sorry, I did not look at irc messages. I am a bit overwhelmed. I am making a reading and will get back to you for doubts. Oct 31 12:02:57 bluelightning: PR = "r0" drop also touched a lot of recipes Oct 31 12:03:20 bluelightning: I haven't seen single issue with meta-oe patches after changing it just before dylan release Oct 31 12:03:59 bluelightning: best time for it was before branching dora, but if needed we can backport such patch (so that only problem would be cherry-picking commit from master directly to dylan) Oct 31 12:08:13 hm, abou PR, is there any plan to remove the various *PR >0 ? Oct 31 12:09:25 JaMa: the PR = "r0" drop was a total of 137 lines removed; a rough count of tabs gives me about 9000 lines (in 550 bb/inc/bbclass files) that would be changed Oct 31 12:09:59 ant_work: you cannot drop them without version going backwards Oct 31 12:10:11 ant_work: and bumping PE is even worse than having few PRs around Oct 31 12:10:13 indeed Oct 31 12:10:21 they can just die a slow death as upgrades occur Oct 31 12:10:56 bluelightning: I'm not saying it's the same as PR drop, but I still don't see big problem with backports Oct 31 12:11:25 bluelightning: we're not going to mass-cherry-picking hundrets of commits do dylan, are we? Oct 31 12:12:02 bluelightning: and for few patches which are really needed in older then current dora release it's not hard to resolve conflict Oct 31 12:12:17 bluelightning: a lot easier than properly test each cherry-pick in runtime and world build :) Oct 31 12:12:31 not to dylan, but current count in dylan branch is 231 changes since dylan-9.0.0 Oct 31 12:12:52 bluelightning: and if someone accidentaly cherry picks few spaces instead of tabs to dylan, then nothing will break Oct 31 12:13:21 bluelightning: yes, that's why I said that dora should be changed too (because I expect many changes to dora, but not to dylan) Oct 31 12:13:48 in meta-oe I did it just before branching dylan Oct 31 12:13:49 JaMa: not long ago I've read that basically nobody cares about on-machine upgrade (maybe Angstrom they said)... Oct 31 12:14:07 so people were no longer cherry-picking to danny and people cherry-picking from master to dylan were OK Oct 31 12:14:13 ant_work: we have a bug open to add QA testing for that I think Oct 31 12:14:33 ant_work: I still care and I still report bugs (or reply to commits) when people break it Oct 31 12:14:35 JaMa: and afaik there is any recent public 'feed' Oct 31 12:14:42 ant_work: but it covers only my images Oct 31 12:14:56 ant_work: shr and webos still have reliable binary feeds Oct 31 12:15:08 great Oct 31 12:15:22 JaMa: I don't think we would make such a cleanup in dora, if it were to happen it would be before the next release Oct 31 12:15:26 well reliable but not very usable since PR server :) Oct 31 12:15:41 JaMa: but really RP is the one you have to convince, since he's the one who would have to deal with any fallout... Oct 31 12:15:43 bluelightning: OK, I'll ask about it before 1.6 Oct 31 12:16:17 JaMa: I was thinking about spaces/tabs while mangling machine.conf's Oct 31 12:16:19 ;) Oct 31 12:16:31 ant_work: since PR server most upgrades try to download almost all .ipk, so even with binary feed reliable you usually run out of space on smaller devices before opkg upgrade is finished Oct 31 12:16:36 ant_work: https://bugzilla.yoctoproject.org/show_bug.cgi?id=4304 Oct 31 12:16:37 Bug 4304: normal, Medium, 1.6, ionutx.chisanovici, NEW , Do automated upgrade path testing Oct 31 12:17:42 ant_work: so binary feed works fine when you're just changing small things without invalidating lot of signatures (like last "small" dora update) Oct 31 12:17:56 JaMa: I've made it clear I am not entertaining this tab/space change Oct 31 12:18:25 ant_work: or when you know how to "partition" the upgrade to smaller chunks which don't fill rootfs completely Oct 31 12:19:09 RP: I know, sorry about bringing it up again, but is the reasoning about doing it just before release really so bad? Oct 31 12:20:33 JaMa: last time I touched this, I had many people screaming at me about this. Unfortunately they did this privately, despite my suggestion they should do so in public. I am not going there again. Oct 31 12:20:59 There are plenty of more interesting changes I'd rather concentrate on Oct 31 12:21:22 right, but that wasn't done just before release and it was causing parse failures not estetic issues Oct 31 12:21:56 or couldn't be done just before release, because it needed a lot more testing and updates in other layers Oct 31 12:22:19 so this is completely different case in my eyes Oct 31 12:22:55 JaMa: For the python problem there was a real technical issue we were solving and I got shouted at badly over it. Doing it over something cosmetic, whether its just before release or not is not something I can justify Oct 31 12:27:09 RP: I feel uneasy too because typically a new recipe is done by copying an older one (and editing it) so things will only be worse with so many sparse examples Oct 31 12:28:41 obviously we can only hope the external layers will do cleanings as well but this is more a dream Oct 31 12:28:44 iirc I've also asked to just change styleguide on wiki and let tabs die a slow death just like PR variables.. Oct 31 12:29:19 JaMa: We have a style, I'd much prefer we just stick to the thing Oct 31 12:29:32 RP: one for all the indentation of SRC_URI ;) Oct 31 12:31:39 JaMa: I'm thinking about a skeleton for the machine.conf in meta-hh. Do you think ipaq's old examples could be a starting point? Oct 31 12:32:09 RP: the same style which was saying that tabs are bad for very long time on OE wiki, but I don't want to argue with you, you;re right that there are more interesting issues (I was just asking if people changed opinion at least a bit) Oct 31 12:32:15 unfortunately the stylize scripts cannot cope with strategically placed include/require :/ Oct 31 12:33:20 ant_work: the last thing I want is a ton of whitespace patches. They are an absolute nightmare to deal with, both to review and to deal with backports around Oct 31 12:33:40 I've heard those word before..where ? ah..Linus ;) Oct 31 12:34:03 ant_work: and what do Linus and I have in common? We have to deal with the patches... Oct 31 12:34:14 RP: you are fundamentally right Oct 31 12:35:56 let's janitor the patches on the ML and make our observations (PR not neede anymore, use tabs, ...) case by case Oct 31 12:40:00 JaMa: I've seen the failure for the initramfs image...I have a ton of thinmgs to test but I'll try to build qemux86 once Oct 31 12:40:58 JaMa: all normal for arm (maybe a bit of cruft more, empty dirs done by postinst/rc-update) Oct 31 12:44:21 ant_work: you mean in last world build? Oct 31 12:44:30 ant_work: you can ignore that one, it should be fixed already Oct 31 12:46:19 me..I was reading the old log then ;) Oct 31 12:48:23 ant_work: was it about that missing debug variable, right? Oct 31 12:48:43 that's what I've seen in log yesterday when generating report from world Oct 31 12:51:43 I don't remember exactly but the fix came just a couple of days later so I/we thought it was just bad luck Oct 31 12:52:26 there is no reason for do_rootfs to fail for qemux86 only... Oct 31 12:53:32 JaMa: thre is still the old issue with klcc-cross but you only see that when building multiple machines of the same arch Oct 31 12:55:34 RP: when you have a bit of time pls think to my problem about a shared place/sysroot for the headers arch-specific Oct 31 12:56:35 this path is hardcoded in klcc-cross Oct 31 13:01:03 ant_work: I don't think I know what "a bit of time" is these days but I will try and take a look Oct 31 13:02:13 RP: I've tried the quick hack to make it machine-specific but then it installs under /bin/ in sysroot and this is not in path Oct 31 13:03:44 ideally it lives side by side with the -gcc-cross Oct 31 13:08:09 JaMa, ant_work, RP: re. whitespace in new content, is there a tool similar to the kernel's check-patch that help ensure that future patches follow the current style guide? Oct 31 13:14:13 kmacleod: agreed Oct 31 13:49:50 zaif, No worries, I was AFK anyway. Oct 31 13:50:05 zaif, Say, BTW, 'zaif' is that the arabic word for sword or something else? Oct 31 14:04:45 for creating a SDK that i can deploy, whats the recommended way / documentation to read? do i understand it correctly that is sdk then contains all the dev-libraries that i decide to ship by default? Oct 31 14:16:07 LetoThe2nd: grep the docs for populate_sdk Oct 31 14:16:42 RP: thanks Oct 31 16:14:05 Hi everyone Oct 31 16:14:38 I've built an sdk for a freescale imx6 using yocto...the target is meta-toolchain-sdk Oct 31 16:14:52 unfortunately OPENGL headers are missing in the sdk Oct 31 16:15:09 how can I make them included automatically Oct 31 16:15:19 when building the sdk Oct 31 16:27:04 ok..I've just noticed that rpm dev packages are available Oct 31 16:27:15 and the files I needed are inside one of them Oct 31 16:27:33 has yocto been tested against ubuntu 13.10 (saucy) ? Oct 31 16:27:38 now how can I install those rpms so they are placed in the right place? Oct 31 16:27:46 (in the sdk) Oct 31 16:42:11 mrAlmond: there is an entry in the recipe where you can add packages Oct 31 17:07:32 Stygia: Hi Oct 31 17:07:55 Stygia: yes, my actual name is Zafrullah Syed. zaif id my nickname Oct 31 17:08:21 zaif, Ah alright. Zafrullah? Spelled out that'd be something like Zaif-Ul-Lah, right? Oct 31 17:08:53 Stygia: spelled as zaf-ru-llah Oct 31 17:11:01 zaif, Ah, alright. In which language? Oct 31 17:12:19 Stygia: It's in urdu Oct 31 17:12:30 I guess arabic also. Oct 31 17:13:07 zaif, Right urdu, in Arabic, purely, it would be Saif-Ul-Lah. Oct 31 17:13:18 zaif, Don't really have an Arabic keyboard here, but yea, heh. Oct 31 17:14:18 ya, but we don't speak arabic and my mother tongue is urdu(in India). :) Oct 31 17:18:02 Stygia: Is this correct way to write .bbappend ? Oct 31 17:18:02 http://pastebin.com/aGcQxgDA Oct 31 17:19:16 zaif, Ah yup, I'm not saying your name is "wrong" (which would be nonsense anyway), just chatting. :) Oct 31 17:19:26 zaif, It is an Urdu name, not Arabic. ^_^ Oct 31 17:19:37 zaif, I wouldn't put it as a subdir of poky. Oct 31 17:19:43 poky/meta/etc is it's own thing. Oct 31 17:19:51 I'd have it on the same level as poky Oct 31 17:19:52 cd dir/ Oct 31 17:19:57 peterwhitfield: i should have a working recipe later this evening Oct 31 17:20:07 ls: poky/ meta-bebot/ Oct 31 17:20:36 at least the 15B version or whatever, then bump to 16 Oct 31 17:20:42 zaif, Did you put CONFIG_VIDEO_VIVI =y in the bbappend? Oct 31 17:21:22 zaif, Plus, I don't think you used the pattern I recommended for the linux bbappend. :) Hang on. Oct 31 17:21:31 Stygia: ya, it's poky/meta-bebot/recipes-kernel/linux/defconfig Oct 31 17:22:00 and in /recipes-kernel dir i have put .bbappend Oct 31 17:23:10 Stygia: I didn't understand, so read this manual and followed it. http://www.yoctoproject.org/docs/current/kernel-dev/kernel-dev.html#changing-the-configuration Oct 31 17:23:30 zaif, http://pastebin.com/T6a1mEGh Oct 31 17:23:40 zaif, Yes, it is _almost_ right. :) Oct 31 17:24:16 zaif, please note that we keep our entire config file in this dir, under version control, and then copy it into place instead of .config, replacing default config or menuconfig changes Oct 31 17:24:45 zaif, But yes, your idea was right, but you placed the recipe directly into recipes-kernel/, I think. Oct 31 17:24:50 It needs to be in a subdir linux/ Oct 31 17:26:21 aha.. ok. and what about the version number? Is it just for my reference or does it actually have any importance in poky? Oct 31 17:26:22 zaif, In general, when you bbappend? It needs to be in recipe-name/ subdir, and in that dear you have recipe-name_recipe.version.number.bbappend Oct 31 17:26:33 zaif, You need the version of the kernel recipe. Oct 31 17:26:47 zaif, You are saying, by the naming, "Append to this recipe of this version, these changes" Oct 31 17:27:03 zaif, If the actual recipe is for 3.0.5 and you bbappend to 3.0.6, 3.0.5 won't change at all. Oct 31 17:28:22 Stygia: I am using linux-gumstix-3.5 version, so my .bbappend file will be linux-gumstix-3.5.bbappend ? Oct 31 17:28:31 zaif, Yes, exactly. Oct 31 17:28:50 zaif, And if linux-gumstix-3.5.bb is in a linux/ subdir, so must your bbappend be Oct 31 17:31:12 zaif, BRB Oct 31 17:32:21 Stygia: I copied the original defconfig file to my /meta-bebot/recipes-kernel/linux/files and changed this CONFIG_VIDEO_VIVI=y ?? Oct 31 17:35:51 Stygia: This is how my tree looks: http://pastebin.com/VEZSPEXK Oct 31 17:53:42 zaif, there. Note I was offline until now, so if you said something I didnt' get it. Oct 31 17:54:25 Stygia: I copied the original defconfig file to my /meta-bebot/recipes-kernel/linux/files and changed this CONFIG_VIDEO_VIVI=y ?? This is how my tree looks: http://pastebin.com/VEZSPEXK Oct 31 17:56:17 zaif, Yes, that looks right! Oct 31 17:56:27 zaif, Be aware that will be the final config, though, you will apply only these options. Oct 31 17:56:58 If you need to patch the config generated by do_config() in the original recipe, maybe write a loop and sed in the values you want (Read the manpage for sed, maybe) Oct 31 17:57:07 There could be a better way, there probably is, if you want to patch the config instead. Oct 31 17:57:09 Look in the manual. Oct 31 17:59:35 RP, is there anything written up on testing? Oct 31 18:00:12 Crofton: which part of testing? Oct 31 18:00:25 what we want to do about hardware testing Oct 31 18:00:25 Crofton: you mean what we have now, future plans or... ? Oct 31 18:00:30 yes Oct 31 18:00:31 :) Oct 31 18:00:46 I am sort of curious about automated testing via qemu Oct 31 18:01:13 Crofton: harware testing or testing via qemu? Oct 31 18:01:30 yes Oct 31 18:01:39 I have two thoughts Oct 31 18:01:46 Crofton: I have emailed a few people to try and herd cats. There isn't anything public afaik at this point but hopefully there will be some enhancement bugs soon Oct 31 18:02:01 one is testing some gnuradio stuff to make sure we do not regress some simd stuff Oct 31 18:02:17 I just want to make sure this doesn't get off list Oct 31 18:05:54 Crofton: I'll be ensuring it doesn't; FWIW I have been pinging people in various places (including the mailing list) but haven't always received a lot of response Oct 31 18:06:20 bbl Oct 31 18:11:44 Stygia: You will be working tomorrow? Oct 31 18:12:30 Stygia: I am exhausted for today. Will be here again tomorrow. Oct 31 18:13:48 Stygia: Thank You for your patience and helping me so much. :) Oct 31 18:19:40 zaif, No worries, somebody helped me when I didn't get things as well. Oct 31 18:19:46 zaif, Karma or something. Oct 31 18:19:57 :) Oct 31 18:59:09 mario-goulart: please paste the script you did for the customer somewhere so we can discuss it with tomz Oct 31 18:59:46 What a shame Oct 31 18:59:58 I hope tomz hasn't had any meal recently Oct 31 19:00:08 just a min Oct 31 19:00:29 mario-goulart: just e-mail it and i'll take a look - i'll need to digest it Oct 31 19:01:13 tomz: alright! Oct 31 19:01:29 tomz: heheh Oct 31 19:01:37 ;-) Oct 31 19:01:40 tomz: take a Scotch with you Oct 31 19:01:46 it'll be a ugly code Oct 31 19:01:47 heh Oct 31 19:22:18 should packagegroup.bbclass set INHIBIT_DEFAULT_DEPS? by default it gets that from allarch.bbclass but only when PACKAGE_ARCH == "all" Oct 31 19:58:06 @mr_science: that's awesome! I tried manually installing the erlang cross-compile last night but no luck running on the arm hardware Oct 31 19:58:49 7am here and off to work. will check in later Oct 31 23:28:00 hey guys :-) anyone know why network.service is masked by default, and how do I get networking to work nice with systemd Oct 31 23:28:45 I tried systemctr unmask network.service, that works, but the enable fails Oct 31 23:29:21 says no file found Oct 31 23:31:35 Cannot see /lib/systemd/system a network.service there .. Oct 31 23:32:29 I mean a networking.service in /lib/systemd/system Oct 31 23:54:45 cananyone help me with systemd stuff? Nov 01 01:11:20 how to avoid the split_and_strip_files for a certain package? Nov 01 01:11:42 I'm trying to add do INSANE_SKIP_${PN}-samples += "split-strip" but it's not working Nov 01 01:11:52 I'm using poky 10 btw Nov 01 01:14:51 INHIBIT_STRIP or something like that IIRC Nov 01 01:15:02 you still need split, right? Nov 01 01:35:35 JaMa: I'm not sure.. itÅ› a package that contains mp3 files only.. so probably not Nov 01 01:36:34 I can't find any reference to INHIBIT_STRIP in poky Nov 01 01:44:01 * nerdboy porting patches Nov 01 02:23:22 JaMa: ok, was a bug in package.bbclass, I'm already sending a patch **** ENDING LOGGING AT Fri Nov 01 02:59:58 2013