**** BEGIN LOGGING AT Thu Nov 28 02:59:58 2013 Nov 28 09:19:34 Hello everybody. Is it possible to avoid inserting FEED_DEPLOYDIR_BASE_URI in local.conf ? because I want to be able to generate two images, the first one is configured with opkg and ipkg repository and the second is minimal without pkg repository and opkg and opkg info, etc.. Nov 28 09:44:01 hannken: yes. you can set it in the image .bb file instead. Nov 28 09:44:35 that said, you should be able to generate already 2 images like you said i believe. Nov 28 09:45:08 you just need to create 2 .bb files, one with package-management 'enabled', one 'disabled' Nov 28 09:51:26 ndec: I just did that but there is an error because FEED_DEPLOYDIR_BASE_URI is set in local.conf, should I move FEED_DEPLOYDIR_BASE_URI to .bb file ? Nov 28 09:52:39 hannken: what errot? Nov 28 09:52:41 error? Nov 28 09:53:07 i am doing something similar and it is working. which version are you using? Nov 28 09:53:57 ndec: work/raspberrypi-oe-linux-gnueabi/minimal-innerjbox-distro/1.0-r0/rootfs/etc/opkg/local-all-feed.conf: Directory nonexistent Nov 28 09:54:52 ndec: let me check other things then I will ask again.. maybe I messed up with something Nov 28 09:55:18 hannken: in which task are you getting this error? Nov 28 09:55:18 morning all Nov 28 09:55:23 bluelightning: morning! Nov 28 09:55:38 hi ndec Nov 28 09:55:48 ndec: task? do you mean do_rootfs ? Nov 28 09:55:52 yes. Nov 28 09:56:18 so it is in do_rootfs Nov 28 09:56:21 hannken: and which OE version are you using? Nov 28 09:57:16 mmm.. I think I am getting OE from git but I didn't update the source since two months Nov 28 09:59:03 hannken: ok. but that's probably not a too good idea ;-) you should either use a 'stable' branch if you aren't really 'developing' on OE, and just want to use it. and if you want to develop on OE, then you should use master, but track it ;-) Nov 28 10:04:26 ndec: I think that it's enough to set FEED_DEPLOYDIR_BASE_URI in .bb with package-management management activated.. No need to set it in local.conf Nov 28 10:05:17 well,yes. however it should work even in the other case... so you might have another problem in fact. Nov 28 10:06:03 ndec: thanks for your help ^^ Nov 28 10:06:35 hannken: do you set BUILD_IMAGES_FROM_FEEDS to 1? Nov 28 10:07:27 ndec: no Nov 28 10:54:18 Hi Nov 28 10:58:26 RP: I have a sysroot cleanup question Nov 28 10:59:01 RP: for perl I need to sed out path to sysroot for the files that go into the resulting packages, but I do need them in place for the files going into the sysroot Nov 28 10:59:05 RP: https://github.com/Angstrom-distribution/oe-core/commit/eda248d42ccba3e39acba88661e247a32bd49330 Nov 28 10:59:26 RP: that patch fixes the packages, but breaks when trying to build libxml-parser-perl Nov 28 10:59:46 so it there a method to only fixup the files for the packages? Nov 28 11:01:35 i can not apply a custom kernel defconfig to a kernel recipe, its allways overwritten from stuff by the do_configure section Nov 28 11:03:19 koen: Use PACKAGE_PREPROCESS_FUNCS Nov 28 11:04:30 thanks! Nov 28 11:04:40 * koen stabs perl just for good measure Nov 28 11:05:11 oh dear Nov 28 11:05:15 perl already has that Nov 28 11:05:24 * koen puts on dunce cap Nov 28 11:10:22 another undocumented variable... Nov 28 11:10:54 * bluelightning wonders if he and Scott will ever get to the end of the list Nov 28 11:11:09 ttp://hastebin.com/hafoxijiko.bash Nov 28 11:11:14 looks a lot better already Nov 28 11:11:17 +h Nov 28 11:12:57 bluelightning: These are some of the more obscure ones. I'd not advise this one be used under lots of circumstances Nov 28 11:13:13 bluelightning: perl is one of the crazy cases we need it :( Nov 28 11:13:28 RP: true, but I see we use this in a few other places too Nov 28 11:15:25 it's all about how we describe it in the entry Nov 28 11:16:02 bluelightning: right Nov 28 11:16:17 we have other variables higher on the list to document though of course Nov 28 11:45:38 how can i create a kernel package with a custom kernel .config ? is there a sample ? Nov 28 11:48:17 pwgen: just supply your config as "defconfig" and point to it in SRC_URI Nov 28 11:50:10 this does not help, the run.do_configue will erase and recreate my .config with a lot of variables Nov 28 11:51:59 is there a variable or switch i have to set to prevent it from automagically create the .config file ? Nov 28 11:52:29 in my freshly built dora-sdk, arm-angstrom-linux-gnueabi-gcc segfaults. is that a known issue? Nov 28 11:54:20 pwgen: which recipe are you basing your kernel recipe on? Nov 28 11:56:50 http://pastebin.com/BL0FbMZf i think i derived it from a raspberry recipe Nov 28 12:02:07 oh, linux.inc Nov 28 12:04:08 someone managed to get the kexec hardboot working on the ouya(:-)) Nov 28 12:04:32 time to give it a usefil bootloader Nov 28 12:05:55 tasslehoff, are you building on an Ububtu machine? Nov 28 12:09:39 Crofton|work: yep. 13.04, 64 bit. Nov 28 12:09:52 hmm Nov 28 12:10:48 there seems to be an issue with angstrom toolchains and Ubuntu that we have never been able to track down Nov 28 12:24:42 Crofton|work: no voodoo to get around it? clean and rebuild, or something similar. Nov 28 12:25:22 only voodoo I know is not use Ubuntu Nov 28 12:25:24 sorry Nov 28 12:25:46 koen, we are getting bit by ubuntu + angstrom again Nov 28 12:26:21 tasslehoff, be prepared to be told Ubuntu isn't a real Linux :) Nov 28 12:26:33 I use Fedora, so I've never seen the problem in person Nov 28 12:27:10 Crofton|work: yup, I've been in #beagle enough years to know that :) Nov 28 12:28:47 I blame some Linaro crap :) Nov 28 12:28:55 Been wanting to try something else, but everyone else here is running Ubuntu Nov 28 12:29:04 I think that is the difference in toolchains, but I've never looked very hard Nov 28 12:29:20 Perhaps I need linaro-crap? I don't have meta-linaro. Nov 28 12:31:01 hmmm Nov 28 12:32:07 I started with the angstrom-setup-scripts, but deleted layers. Now I have bitbake, meta-angstrom, meta-openembedded, meta-beagleboard, meta-ti, openembedded-core and meta-qt5. Nov 28 12:32:39 one of the workarounds was to delete the host libc from the sdk Nov 28 12:34:36 ok. so python-native goes bonkers if you have python stuff installed at $HOME/.local Nov 28 12:35:38 ...time well spent finding out that Nov 28 12:37:12 koen: that did the trick here as well. Nov 28 12:38:55 hmm. perhaps I spoke too early.. Nov 28 12:45:48 nope, but I spoke *again* too early. it works. thanks koen Nov 28 12:48:46 Crofton|work: so what is a good choice of distro for oe-work. anything !ubuntu? Nov 28 12:56:13 I use Fedora Nov 28 12:56:28 I am sure there are other good choices Nov 28 13:12:41 FWIW I use ubuntu with poky / plain OE-Core and have not seen this problem, or any other related to the host distro Nov 28 13:25:45 bluelightning, that does not make the problem any less annoying Nov 28 13:28:06 Crofton|work: well, it shouldn't be impossible to find if it's definitely in something custom that angstrom is setting over the OE-Core config Nov 28 13:28:14 and it seems that that is the case Nov 28 13:32:36 morning all Nov 28 13:37:06 hi pb_ Nov 28 18:07:42 hello ppl ;) Nov 28 18:08:21 * Jay7 is preparing to make talk about OE/Yocto on local mini-conference Nov 28 18:08:44 collecting good resources about Nov 28 18:42:53 hi Jay7 Nov 28 18:43:00 Jay7: great! Nov 28 18:57:23 Jay7: let us know if you need anything Nov 28 18:57:24 bbl **** ENDING LOGGING AT Fri Nov 29 02:59:58 2013