**** BEGIN LOGGING AT Sat Feb 02 03:00:00 2013 Feb 02 06:01:10 is it possible to use the BBCLASSEXTEND functionality with images? Feb 02 06:28:45 yes, you could, but i doubt it'd be of much use, as images require packages. native/cross don't produce packages. Feb 02 06:28:53 so in reality depends on the class Feb 02 07:02:46 urg Feb 02 07:02:52 to early Feb 02 07:03:03 0800 at breakfast tie in Brussels Feb 02 07:22:04 kergoth: just made devimage.bbclass and qaimage.bbclass, and testing now. The idea is to allow extending a minimal device image to a bigger image with the parts necessary to assist in debugging failures, or to do more in depth testing of a product. Feb 02 07:26:24 basically, instead of having to have three images, can have one and don't have to use EXTRA_IMAGE_INCLUDEs from local.conf to get builds with debug symbols/etc. Feb 02 07:27:12 seems earlier here Crofton Feb 02 07:27:17 (1 AM) Feb 02 09:38:15 hello guys Feb 02 21:06:45 I have an unexpected parameter included in the yocto-linux kernel produced. Is there an easy way to find a) the list of fragments used. 2) which fragment a specific parameter came from? Feb 02 22:36:51 Does anyone know why binutils.inc removes ar and strings when installing? The last change was kergoth making it an 'rm -f' to avoid a failure mode with the Code Sourcery external toolchain, and the line otherwise dates back to the intiial checkin from RP in 2005, which appears to be a hand-off from SVN to git. Feb 02 22:37:30 This isn't breaking anything interesting that I know of, but I assume it must be solving some problem unknown to me. Feb 02 23:08:49 seebs: that sounds like very old code, I don't remember anything... Feb 02 23:12:01 seebs: ah, http://git.yoctoproject.org/cgit.cgi/poky/commit/?id=4d4a37609166adb9a86423f346e7a6bd6649da3a Feb 02 23:12:21 seebs: looks like its the symlinks its putting under update-alternatives control Feb 02 23:23:49 seebs: removal like that is probably to minimize number of packages providing a given entity. But since I'm not sure of any other package providing ar, not sure this case. Do know gcc for instance has items pulled out so it doesn't collide with binutils. Feb 02 23:28:40 Ahh, okay. Feb 02 23:29:02 Thanks, now I feel like I can just ignore that and not worry. :) Feb 02 23:29:39 a quick search for another package providing ar would let you know with a good level of confidence. Feb 02 23:29:47 (gcc maybe) Feb 02 23:34:12 I'm having trouble with dependencies, probably just my understanding. Looking at a package-depends.dot shows packages depending on like autoconf-native. Is thiis meaning that autoconf-native must be installed before that package can be installed? Feb 02 23:35:00 I am not totally sure, but I *think* "DEPENDS" is "must be in sysroot before I can be built" and RDEPENDS is "must be installed before I can be installed". Feb 02 23:42:07 My problem I'm trying to figure what is going wrong for is a package with RDEPENDS+="xserver-xorg-module-exa xserver-xorg-module-xaa". System ends up unbootable because the video driver is found but that driver can't load libxaa.so (from xserver-xorg-module-xaa). So how to find why an RDEPENDS is not bringing the dynamically loaded library in. Feb 02 23:48:02 woot! Found it (sort of). I have a RDEPENDS+= but bitbake -e shows RDEPENDS="" Feb 03 00:07:38 another case of bitbake syntax is too close to a number of other languages. RDEPENDS="xyz" and RDEPENDS = "xyz" are not synonomous. :( RDEPENDS+="xyz" and RDEPENDS += "xzy" really aren't. Feb 03 00:13:16 oh? which one is correct? Feb 03 00:31:40 the spaces are required Feb 03 00:33:13 none gave any errors, just RDEPENDS kept coming up completely empty. Feb 03 00:51:38 bitbake doens't care about spaces around operators Feb 03 00:51:41 never has, never will Feb 03 00:51:49 well, imo anyway Feb 03 00:51:55 if it's misbehaving, open a bug Feb 03 02:59:16 thanks kergoth, will file a bug. **** ENDING LOGGING AT Sun Feb 03 02:59:59 2013