**** BEGIN LOGGING AT Wed Nov 09 02:59:57 2011 Nov 09 05:16:58 Hi, fellows! I have a question about the subject: so, if get the base image and then build nano, for example, how can I put nano into my image? Nov 09 06:51:21 anyone else having trouble compiling udev? http://pastebin.com/8phizCrQ Nov 09 08:25:13 good morning Nov 09 08:28:26 mckoan: morning Nov 09 08:34:01 mckoan: can you answer my question above? :) Nov 09 08:36:02 what about reading the docs? :p Nov 09 08:36:36 or take a look for an example in images already available Nov 09 08:38:24 okay, ynezz, will read the docs, but I guess, that should be very simple move Nov 09 08:38:29 it is Nov 09 08:38:36 then find it yourself :p Nov 09 08:38:55 you'll of course learn something new also Nov 09 08:39:02 as a bonus Nov 09 08:39:14 thank you very much :) Nov 09 08:48:51 reisei: try to pastebin your imagefile.bb Nov 09 08:48:59 ~pastebin Nov 09 08:48:59 [~pastebin] A "pastebin" is a web-based service where you should paste anything over 3 lines so you don't flood the channel. Here are links to a few : http://www.pastebin.com , http://pastebin.ca , http://channels.debian.net/paste , http://paste.lisp.org , http://bin.cakephp.org/ , http://asterisk.pastey.net/ , or install pastebinit with yum or aptitude. Nov 09 08:57:32 should PREFERRED_VERSION_udev = "173" in local.conf work? it still tried 175 (which fails here). Nov 09 09:07:55 03Koen Kooi  07org.openembedded.dev * r8e69cfcd24 10openembedded.git/conf/machine/include/omap3.inc: Nov 09 09:07:55 Revert "omap3.inc: Remove hardcoded x-loader dependency breaking builds for every board without a matching x-loader configuration." Nov 09 09:07:55 Don't break existing boards without either a proper fix or consulting the maintainer Nov 09 09:07:55 This reverts commit 413ba9a4b137386a224d7063bb88a7449f3c269e. Nov 09 09:38:13 mornin Nov 09 10:11:29 mickey|babybusy: hey! Nov 09 10:48:10 in sysroot_stage_dirs, it is doing sysroot_stage_dir $from/usr/include $to/usr/include, where is the "/usr/include" coming from? my problem is: I have a package where it is appending the full path there, i.e. /myhome/myoedir/workdir/recipe/bla/usr/include Nov 09 10:48:44 and somehow I could not yet figure out why this is happening or where it is taking those paths Nov 09 10:49:04 my $from and $to paths seem to be correct though Nov 09 10:52:47 jineld which recipe and which soft? Nov 09 10:56:11 woglinde: the recipe is self-made, so that's surely the suspicion number one :) I am trying to build rubygems-native Nov 09 10:56:46 basically it only does a custom do_install and populates ${D} Nov 09 10:57:16 http://pastebin.mozilla.org/1377992 Nov 09 10:58:36 good morning Nov 09 10:58:46 hi florian Nov 09 10:59:37 export gem_dir=/usr/lib/ruby/gems Nov 09 11:03:32 woglinde: that is my seccond attempt, I had it with ${D} already Nov 09 11:04:08 but that does not change anything, do_install runs throuhg, but then in the staging step it - no idea fom where appends to $from and $to the full paths isntead of just the relative image paths Nov 09 11:04:22 btw that was the reason why I tried dropping the ${D} here, at this point am alreaedy experimenting Nov 09 11:07:18 with -e, this is how it should be: http://pastebin.mozilla.org/1378013 Nov 09 11:07:37 instead I have $to/home/bla/path/to/sysroot Nov 09 11:07:45 and $from/home/path/to/image Nov 09 11:08:03 where is it taking those paths from, that it is appending to $from and $to ? Nov 09 11:08:22 Jin^eLD what is your overall goal Nov 09 11:08:26 ruby support in oe? Nov 09 11:09:11 woglinde: I have a piece of software that needs "sprocketize" to be built, its a javascript UI; so for building that I need ruby, rubygems and the sprocketize gem, all native Nov 09 11:09:20 I had t working with ruby 1.8.5 in oe-classic Nov 09 11:09:45 sprocketize is a tool that processes a bunch of .js files that have special headers and spits out one .js file Nov 09 11:10:09 and since it's not available on all repos etc I'd prefer to build it within OE Nov 09 11:11:33 the same recipe that did work in classic did not work in core anymore Nov 09 11:11:55 and I can't figue out why because it's failing at an "autoamted" stage which i am not touching at all Nov 09 11:12:29 the classic version though, installed the stuff to the staging dir and then copied it over to ${D} Nov 09 11:12:34 the recipe in classic Nov 09 11:12:58 I got rid of that part because it did not work in core Nov 09 11:19:31 hmm ok from what I see the bbclass is doing: sysroot_stage_dir $from${includedir} $to${includedir} Nov 09 11:19:40 so that means that my includedir is completely messed up for some reason? Nov 09 11:48:55 it boils down to this: http://pastebin.mozilla.org/1378053 Nov 09 11:49:04 the dir that tar is getting is just wrong Nov 09 12:03:41 guys, please have a look: http://pastebin.mozilla.org/1378090 Nov 09 12:03:48 this simple recipe triggers the error that I am fighting with Nov 09 12:04:53 this is the error output that I get: http://pastebin.mozilla.org/1378091 Nov 09 12:04:59 is it a bitbake bug? Nov 09 12:05:05 or is such a recipe invalid? Nov 09 12:05:05 dont know Nov 09 12:05:16 raise it to the mailinglist Nov 09 12:05:25 intel guys repsonsing there more often Nov 09 12:05:52 which list exactly, is there an own bitbake ml? Nov 09 12:06:07 oe-core Nov 09 12:06:18 k, will do, thx Nov 09 12:31:42 what decides which kernel is added to my image? the symlink in deploy/board points to a kernel called 3.0 (that I've built), but it seems another one called 3.0+3.1rc is the one chosen. Nov 09 12:33:13 I've changed PREFERRED_PROVIDER_virtual/kernel in my machine conf, rebuilt the kernel and built my image again Nov 09 12:36:07 bitbake -DDD Nov 09 12:36:15 search the output Nov 09 12:37:15 tasslehoff: is the PREFERRED_VERSION set correctly? Nov 09 12:37:22 for the kernel? Nov 09 12:39:01 Jin^eLD: I've set the provider to linux-omap-sakoman, and there's only one recipe providing that - my testrecipe :) Nov 09 12:39:21 * tasslehoff *has* to start remembering -DDD Nov 09 13:01:41 hm. it talks about the right linux-recipe when I run bitbake -DDD myimage, but the image contains another kernel. Nov 09 13:02:16 hm Nov 09 14:02:17 hm - indeed. tried cleaning kernel/image and bumping MACHINE_KERNEL_PR, but it still ends up packing the wrong kernel Nov 09 14:04:19 oh well. kindergarten pickup. I'll dig more into this later. Nov 09 14:05:41 hehe Nov 09 16:16:09 is ther a way to somehow override this LIC_FILES_CHKSUM thing? Nov 09 16:17:22 you can set LICENSE = "CLOSED" which indicates its closed source, ala firmware, but it's not recommended, and using it for anything else as a hack Nov 09 16:17:35 better to just set LICENSE and LIC_FILES_CHKSUM correctly Nov 09 16:18:01 kergoth: well, I have a weirdo setup where I need to install a ruby .gem file as a native recipe because it provides some crap that I need to build another tool Nov 09 16:18:13 the .gem is not being unpacked by OE, hence I can not point to any of its contents Nov 09 16:18:38 a rubygems-native recipe is providing the gem utility that then installs the actual gem contents Nov 09 16:19:46 but hmm, with OE core it does not work out so well anyway, because at most of the times I get hit by the error that I asked about on the ML :P Nov 09 16:20:33 Jin^eLD: you can point to one of the common license files Nov 09 16:21:47 how? I grepped around but everyone seems to be poinitng to file://COPYING or similar Nov 09 16:22:05 which I assume is part of the package Nov 09 16:22:36 e.g LIC_FILES_CHKSUM = "file://${COMMON_LICENSE_DIR}/GPL-2.0;md5=801f80980d171dd6425610833a22dbe6" Nov 09 16:22:46 it's somewhat unorthodox however Nov 09 16:22:54 ah I see, thanks Nov 09 16:27:29 and yes, when no (or a relative) path is specified it refers to somewhere under ${S} Nov 09 17:17:50 hrms Nov 09 17:17:52 ERROR: Could not include required file recipes-graphics/xorg-xserver/xserver-xf86-dri-lite.inc Nov 09 17:18:16 very old oe-core? Nov 09 17:18:19 or meta-oe? Nov 09 17:18:29 no I pulled all stuff Nov 09 17:18:43 or some other old layer Nov 09 17:18:50 which recipe caused that? Nov 09 17:18:50 should I start building from clean Nov 09 17:18:52 hm Nov 09 17:18:57 old layer Nov 09 17:19:31 grep all layers for xserver-xf86-dri-lite.inc, it should be long gone Nov 09 17:22:54 intellayer it was Nov 09 17:23:32 but that was also fixed a while ago, wasn't it? Nov 09 17:23:47 sure Nov 09 17:24:04 hrms Nov 09 17:24:06 RROR: Nothing RPROVIDES 'initramfs-live-install' Nov 09 17:24:59 hi crofton Nov 09 18:14:58 jama is PACKAGECONFIG ??= right? Nov 09 18:15:14 woglinde_: where? Nov 09 18:15:32 libsdl? yes, that's intentional Nov 09 18:16:04 yes libsdl Nov 09 18:31:45 woglinde_, i have fleshed out your uclibc sched_* patch and pushed it as 19dd090a0f68765db87990ef8eda9bf77bb29581 . Thanks! Nov 09 18:33:36 woglinde_, i have also resurrected ppc e500 fenv support (i think) so if you could bump to current HEAd (and drop the patches that are now obsoleted) i'd be glad 8) Nov 09 18:36:14 darn. I must somehow find machines that have enough disk-space to actually build oe-core :( *wince* Nov 09 18:40:28 i have 8.8 GB of my 80GB left on my whole disk here nowadays so i'm essentially back to building a really small subset and that quasi manually. Sucks big time Nov 09 18:41:20 blindvt: are you using rm_work? Nov 09 18:43:33 to test some pm I decided to make my own kernel recipe. I called it linux-omap-sakoman_3.0.bb, and added PREFERRED_PROVIDER_virtual/kernel = "linux-omap-sakoman" to my machine.conf. When I bake my image it builds the kernel I want, but in /boot after deployment I find the meta-ti kernel thats default for my angstrom build. What am I missing? Nov 09 18:44:57 JaMa|Off, no? sounds like the appropriate thing! Since HP turned off their PARISC boxes some time ago, i had to fall back to my old setup with a ridiculous small set of stuff. Furthermore bitbake seems to do odd things dependency wise since the move away from "classic" oe so i was concentrating into looking into that madness, but would have much preferred not to, TBH Nov 09 18:49:56 manually building systems with just a hand-chosen, dedicated, really small set of stuff is just a blind flight imho. but we all know that, that's because we're #here anyway so sorry for the inappropriate rant. Let me see if rm_work gets me there to fit into my 8.8GB .. Nov 09 19:27:16 doh, what license does one pick for ca certificates? there's nothing in the tarball that would indicate the license Nov 09 19:35:14 re Nov 09 19:37:26 wb Nov 09 19:50:08 blindvt hm 8,8 could be okay Nov 09 19:50:23 blindvt try bitbake gcc-cross Nov 09 19:50:34 which builds you the toolchain Nov 09 21:03:04 I've tried clean and cleansstate for virtual/kernel and my image, but still the old kernel lies in work/beaglish-angstrom-linux-gnueabi/myimage-1.0-r0/rootfs/boot Nov 09 21:03:26 how do I make it put my new kernel there instead? Nov 09 21:06:14 did your PV/PR of the kernel go backwards? could end up with the old ipk lurking around in deploy/ipk Nov 09 21:08:58 kergoth_: indeed it did. I'm trying out a 3.0 kernel because of some power management issues in 3.1 Nov 09 21:09:32 might have to manually kill the kernel* ipk files and rebuild the image then Nov 09 21:10:06 kergoth_: that's fine, as long as I solve this :) Nov 09 21:10:10 If there is a particular kernel recipe you'd like to build your fs against Nov 09 21:10:16 you can force it by adding Nov 09 21:10:20 DP=1 Nov 09 21:10:26 to the kernel recipe Nov 09 21:11:42 (bitbake will use the recipe with the highest DP, or default priority) Nov 09 21:12:08 the issue isn't with kernel selection, it's that he changed it and his rootfs still has the old one Nov 09 21:12:08 pack3754: it used the recipe, but didn't pack right uImage into the rootfs Nov 09 21:15:20 ahh, ok Nov 09 21:16:15 so the kernel modules and everything were compiled against the "new" kernel Nov 09 21:17:11 but the new kernel wasn't installed to the boot dir Nov 09 21:22:51 think so. removed *kernel*.ipk from deploy, cleaned and am rebuilding now Nov 09 21:31:20 mm, does eglibc provide clock_settime/clock_gettime? I am adding -lrt, but still get unresolved symbols when linking Nov 09 21:31:23 same recipe worked fine with glibc Nov 09 21:38:10 never mind, patched it with a syscall Nov 09 23:10:51 good evening. Is there some documentation around which describes making recipes for virtual packages? I'd like to create a virtual package that includes packages for the binaries and configuration files for an application, so that I can create an alternate configuration set which uninstalls the default configuration Nov 09 23:21:28 "virtual" packages are just a convention. they aren't special. its just a package. when multiple packages provide the same functionality, then the user has to choose between them Nov 10 00:16:22 kergoth_: ok, I'll see what I can reference to create one Nov 10 00:16:51 is there anything special in saying "package B provides the same functionality as package A, so installing B will automatically uninstall A" ? Nov 10 00:21:35 in debian's universe, you do that by saying you conflict + replace + provide. see the debian packaging manual. I'm not sure if opkg supports that, though Nov 10 00:21:40 (afaik, anyway) Nov 10 01:16:51 kergoth_: thank you, I'll take a look. Half the battle's learning the nomenclature so I know what to look for **** ENDING LOGGING AT Thu Nov 10 02:59:58 2011