**** BEGIN LOGGING AT Thu Mar 21 02:59:58 2013 Mar 21 08:26:11 I have cloned poky and want to add my own layer. What is the normal way to do this? Add it as a git submodule? Mar 21 08:28:48 good morning Mar 21 08:30:12 tasslehoff: you can create your layer in local just for starting Mar 21 08:36:20 mckoan: I already have my layer ready in a gitrepo on our server. I have based my work on the angstrom setup script until know. Mar 21 08:43:16 tasslehoff: an alternative is to have your layer use submodules to pull in poky Mar 21 08:43:26 (which is what meta-guacamayo does) Mar 21 09:08:12 hello Mar 21 09:17:45 hi Mar 21 09:18:16 hello mks, mattnie, all Mar 21 09:22:44 rburton: ah. that's a good idea. Mar 21 09:23:05 tasslehoff: see https://github.com/Guacamayo/meta-guacamayo Mar 21 09:23:41 tasslehoff: upstream/ is where submodules for poky, meta-oe, and various BSPs appear Mar 21 09:24:57 obviously its just one of many ways, but it does nicely split your upstream layers from your own Mar 21 09:25:19 the downside is that you can't just go and change stuff in the other layers, you need to bbappend Mar 21 09:32:18 rburton: yeah, but that is the practice I've been following until now anyway. I like that my other layers are pristine :) Mar 21 09:40:25 certainly makes upgrading them a lot easier Mar 21 09:44:57 morning all Mar 21 10:26:27 where are you bluelighting Mar 21 10:27:43 I will be off duty I'm in china Mar 21 10:33:49 lyang0: I'm in the UK Mar 21 10:33:56 London to be precise Mar 21 10:34:59 lyang0: on IRC people generally follow UGT though: http://www.total-knowledge.com/~ilya/mips/ugt.html :) Mar 21 10:43:38 Hi all! Anyone knows if there is a bsp/machine configuration for i.mx6SoloLite EVK? Thanks! Mar 21 10:46:13 soldoKyn: have you looked in meta-fsl-arm already? that has some i.mx6* machines Mar 21 10:53:43 rburton: in meta-fsl-arm there are only machines for i.MX6Q unfortunately... Mar 21 10:53:59 soldoKyn: i've no idea what the differences are. is one of those a good match? Mar 21 10:54:14 * rburton doesn't have any fsl hw, so can't really help more :/ Mar 21 10:57:46 rburton: thank you anyway! Mar 21 10:58:23 soldoKyn: AFAIK there is not bsp for i.mx6SoloLite EVK Mar 21 11:14:53 our image is !x11. Util now I have used angstroms console-base-image.bb as a base and added qt and other packages. which image would you recommend starting from in yocto? Mar 21 11:17:31 tasslehoff: depends on what you want.... honestly though image recipes are so trivial they can usually just be written from scratch Mar 21 11:30:07 bluelightning: yeah. perhaps that is the best thing to do to really understand what I'm including Mar 21 11:30:27 esp. since one of my goals are to make sure our image is as small as possible :) Mar 21 11:32:30 Hi, i wish to include a python 2.7 package in my image but the only package I can find in poky is the python-nativ package. Can I include this package in my image or is that just used by yocto to create the image? Mar 21 11:34:34 tasslehoff: if you want a tiny image then bitbake+oe-core and starting from one of the small images as an example is probably the right thing to do Mar 21 11:34:52 the bigger distros like angstrom pull more in as they're more featureful, which isn't always what you want Mar 21 11:46:09 rburton: true. Mar 21 11:48:43 bluelighting good morning Mar 21 11:48:54 I'm at home Mar 21 11:59:06 can I overlay .inc files by simply having an .inc file in a bblayer with higher priority? the recipe just to a require foo.inc and I want to do modifications to foo.inc but not the .bb Mar 21 12:01:19 erbo: unfortunately no; you can only overlay recipes Mar 21 12:01:28 or append them Mar 21 12:01:58 erbo: however, an append just overriding the variables you want to override may be able to do the job Mar 21 12:14:22 bluelightning: I can probably to everything I want that way, thanks for the hint Mar 21 13:20:40 hi, how to add a sanity test script to yocto? I'd like to add a test for git tags so that I don't have to fix them to hashes manually, https://wiki.yoctoproject.org/wiki/How_do_I#Q:_How_do_I_create_my_own_source_download_mirror_.3F Mar 21 16:48:37 how repeatable are yocto builds? Compilation order seems be pretty difficult to handle. Various packages succeed if build from scratch or if rebuild after a successful build. I'm afraid that there's plenty of things like headers that go unnoticed if build order is changed. Mar 21 16:48:58 /succeed/succeed or fail/ Mar 21 16:49:41 mcfrisk: that's why i submitted wipe-sysroot to oe-core, so when adding a new package you can wipe your sysroot and check it build cleanly Mar 21 16:49:55 any problems with repeatable builds is a genuine bug Mar 21 16:50:00 that's unlikely. very few build order issues are likely to remain after all the parallel builds we've done and wwith sstate reuse issues Mar 21 16:50:04 indeed Mar 21 16:51:38 yes, I see the genuine bugs and am fixing them. There just seems to be alot of them. And by fixing I mean cherry-picking fixes from git. Mar 21 16:51:58 ah, you're on danny or denzil i presume? Mar 21 16:52:05 denzil Mar 21 16:52:28 hopefully we can move to newer release or git master soon.. Mar 21 16:52:31 denzil is nearing retirement.. Mar 21 16:52:38 1.4 is about to release Mar 21 16:52:49 even danny/1.3 would be a big step forward Mar 21 16:52:50 I really like how oe-lite resolves build order problems -- per-recipe sysroots Mar 21 16:53:19 aha, that's how OBS works, per package build environments. Mar 21 16:53:31 kergoth: there are plans to add that to oe-core Mar 21 16:53:39 * kergoth nods Mar 21 16:53:50 as an option Mar 21 16:54:01 I had a oe based prototype at one point, but it was prior to shared state even existing, not much of use anymore Mar 21 16:54:15 i suspect you'll get quite a performance dive from it though Mar 21 16:54:27 + extra disk usage Mar 21 16:54:31 oh I'm sure. though we can probably cache based on the set of deps in use Mar 21 16:54:37 and do something vaguely rm_work like Mar 21 16:54:37 well, hardlinking might help there Mar 21 16:54:38 w/the shared state (ignoring disk usage) it shouldn't be hard to enable per-recipe sysroots Mar 21 16:54:57 bluelightning: i'd expect hard links to be involved, yes Mar 21 16:54:57 I'm also wondering how much the build host affects the end result. On my Debian wheezy I'm seeing many more failures which others are not seeing on Ubuntu. For example the broken . Mar 21 16:55:21 mcfrisk, again.. if you can identify a problem with that, it's a bug.. Mar 21 16:55:32 we're take a lot of effort to avoid host contamination issues.. but there were some in Denzil.. Mar 21 16:55:48 mcfrisk: those are probably only with native builds, the stuff we build on the host to run to build the target stuff.. e.g. glibc 2.16 required a lot of fixes / native recipe updates Mar 21 16:56:11 ya.. in 1.4, the native (and nativesdk) packages are significantly more robust Mar 21 16:57:09 maybe I need to refrase: denzil has many issues but most of the are fixed in poky-git danny/master. So thank you for that! Mar 21 16:57:09 mcfrisk: i've found the best results with building old yocto are on old build hosts, i do most of my old release builds in aa centos5 vm personally Mar 21 19:10:44 howdy! just poking yocto a bit, bitbake -k core-image-sato eeks out with "Failed to fetch test data from the network"... is this a false positive? Mar 21 19:12:57 LetoThe2nd: could indicate a problem with a proxy or firewall if you are in a corp environment Mar 21 19:13:15 sgw_: not at all, and i'm pretty sure connectivity is fine. Mar 21 19:13:50 LetoThe2nd: then it might be a false positive as you suggest Mar 21 19:14:10 sgw_: how to ignore it, then? Mar 21 19:16:06 LetoThe2nd: which part is failing can you tell? Mar 21 19:17:45 sgw_: i only get http://fpaste.org/ujoH/ Mar 21 19:18:26 LetoThe2nd: checking something back in a minute Mar 21 19:19:21 sgw_: no need to hurry. Mar 21 19:25:22 LetoThe2nd: Just to verify, can you try going to the 2 links in the CONNECTIVITY_CHECK_URIS Mar 21 19:25:41 CONNECTIVITY_CHECK_URIS ?= " \ Mar 21 19:25:42 https://eula-downloads.yoctoproject.org/index.php \ Mar 21 19:25:42 http://bugzilla.yoctoproject.org/report.cgi" Mar 21 19:27:26 sgw_: in the browser, both seem to work. want me to curl them? Mar 21 19:28:04 Try setting the CONNECTIVITY_CHECK_URIS = "" in your local.conf, what version of poky btw (git master or a version)? Mar 21 19:28:26 danny-8.0 as suggested in the quickstart Mar 21 19:29:48 ok, after unsetting that in local.conf, it proceeds. Mar 21 19:58:03 this is strange, over the weekend when i was working on this image, the modified passwd utility from shadow-password was working fine and the build would end up with the password from the variable ROOT_PASSWORD but today the build fails Mar 21 19:59:36 the command run is: passwd --root /opt/oe/builds/y1.4-M4_som3517/tmp/work/som3517-emac-linux-gnueabi/emac-image/1.0-r0/rootfs -p emac_inc root, the fail is 'passwd: unable to chroot to directory /opt/oe/builds/y1.4-M4_som3517/tmp/work/som3517-emac-linux-gnueabi/emac-image/1.0-r0/rootfs' Mar 21 20:00:50 i guess that makes it unreliable Mar 21 20:02:40 i'd like to know what made it work over the weekend but not today...i am running the build as the same user (i know i have never run it as root) Mar 21 20:06:12 you need to run that command w/in pseudo Mar 21 20:06:31 you can't just run it on the command line... since you won't have permission to chroot Mar 21 20:06:36 pseudo will "fake it" and make it work Mar 21 20:09:49 fray, is there an example of doing this somewhere? I am greping through poky/meta right now looking for one Mar 21 20:11:16 ah, just "psuedo [command]" Mar 21 20:11:25 you have to set the environment.. Mar 21 20:11:44 pseudo is already running if you are in the do_rootfs context Mar 21 20:12:45 ohh...hmm, maybe i am? i am running the generated temp/do_rootfs.#### command to test it Mar 21 20:13:23 bitbake -c devshell Mar 21 20:13:46 PSEUDO_DISABLE=0 bash Mar 21 20:13:49 "id" Mar 21 20:13:54 shoud return 0 Mar 21 20:15:12 no, returns my development user's UID and GIDs Mar 21 20:15:26 i mean, not saying "no it shouldn't" just "no it doesn't" Mar 21 20:16:00 just a sec.. I have to be in a context I an run the commands Mar 21 20:17:30 well in any case running the run.do_rootfs.# command does work from there Mar 21 20:18:11 ok.. then that is fixing it Mar 21 20:38:57 I have an .. unusual problem. Sometime when I build I get an error out of binutils-crosssdk that I need a different version of autoconf. configure.ac:33: error: Please use exactly Autoconf 2.64 instead of 2.69. Mar 21 20:39:18 it doesn't happen every time -- I'm not exactly sure what the trigger is Mar 21 20:39:29 any ideas? Mar 21 20:41:41 As far as I can see, 2.69 is the only version I have a recipe for Mar 21 20:45:32 thanks for the help fray Mar 21 21:00:57 I'm open to any ideas anyone might have Mar 21 21:02:13 this was working on a different machine Mar 21 21:02:22 I'm not sure why moving from machine to machine would make a difference Mar 21 21:06:14 Garibald1: what OS does each machine use? Mar 21 21:07:59 waynr: one was Ubuntu 12.10, the other is some Red Hat variant Mar 21 21:08:28 I think I may have tracked down at least part of the problem -- it didn't like my machine name Mar 21 21:08:44 you said it doesn't happen every time, does it happen every time on one machine but never on the other? Mar 21 21:08:45 err, not machine name Mar 21 21:09:13 it's possible it happens every time, that I just tried something this time that had the dep Mar 21 21:09:38 it tried to run /bin/sh .../confi.su x86_64-foo-pokysdk-linux and failed Mar 21 21:09:54 but I tried running that manually with x86_64-foo_pokysdk-linux and it worked Mar 21 21:10:17 seems it's sensitive to the number of hyphens and what the pieces mean Mar 21 21:10:34 I'm going to try modifying my distro.conf and see if that helps Mar 21 21:10:42 waynr: thanks Mar 21 22:48:17 bitbake -g + dot is one of the more useless visualizations I've run across Mar 21 22:48:26 indeed Mar 21 22:48:39 just massive Mar 21 22:49:02 kergoth: what do you use instead? text editor? Mar 21 22:49:46 yep. grep and vim Mar 21 22:49:52 or depexp Mar 21 22:49:57 bitbake -u depexp -g foo Mar 21 22:50:14 or the whatdepends subcommand of my work-in-progress 'bb' tool Mar 21 22:50:22 (though it's limited at the moment) Mar 21 22:50:37 ah, I should probably look at that Mar 21 22:50:41 where do I get bb? Mar 21 22:51:05 https://github.com/kergoth/sub-bb Mar 21 22:51:14 the readme has usage and examples for most of the subcommands Mar 21 22:51:16 thanks Mar 21 22:51:42 whatdepends -r is pretty awesome Mar 21 22:51:51 i need to teach it to handle runtime and task level dependencies though Mar 21 22:54:12 you'll love bb-log too, much thanks to rburton for that and bb-contents :) Mar 21 23:38:03 is there any way to override selected bbappends? I am using meta-selinux, but there's slight version drift between its .bbappend and what's in my tree; so I made a clone of the repo and patched it, but that's less ideal since i have a micro-fork **** ENDING LOGGING AT Fri Mar 22 02:59:58 2013