**** BEGIN LOGGING AT Mon Mar 05 02:59:58 2012 Mar 05 07:39:22 good morning! Mar 05 08:27:15 ~/IMBA Mar 05 08:27:25 good morning Mar 05 10:51:39 is there any way to get only the base of the name, so I can get libjpeg instead of libjpeg-native for the native build, which I get with ${PN}? Mar 05 10:53:37 ${BPN} (Base Package Name) ? Mar 05 13:05:48 hi Mar 05 13:06:25 i have some issues with the latest version of openembedded Mar 05 13:07:17 i am providing the hard code path in SRC_URI Mar 05 13:07:46 #SRC_URI = "http://www.kernel.org/pub/software/scm/git/git-${PV}.tar.bz2;name=src" Mar 05 13:07:47 SRC_URI = "file://../../../build/sources/git-${PV}.tar.bz2" Mar 05 13:08:51 instead of unpacking the source at /build/tmp/work/i586-linux/git-native-1.7.5/ Mar 05 13:09:15 it's getting unpacked at build/sources Mar 05 13:09:35 anyone have some idea regarding this Mar 05 13:17:31 anuj, maybe change S = Mar 05 13:17:44 ok Mar 05 13:17:46 let me try Mar 05 13:17:56 2) I guess you're using the obsolete oe classic Mar 05 13:22:19 denisATeukrea : now it shows to unpack in the right directory but its empty Mar 05 13:27:16 we have already provided the S=S = "${WORKDIR}/git-${PV}" Mar 05 13:27:46 SRC_URI = "file://../../../build/sources/git-${PV}.tar.bz2" S = "${WORKDIR}/git-${PV}" Mar 05 13:27:53 like this way Mar 05 13:30:19 hi there Mar 05 13:31:27 if I have a package in depends of package X... I am assuming that X do_configure task is dependent on populate_sysroot task of dependent package ... is this correct? Mar 05 13:38:31 anuj, can't you provide git-${PV}.tar.bz2 in the recipes directory? Mar 05 13:38:37 and you didn't comment on 2) Mar 05 14:05:09 hmmm Mar 05 14:05:19 how to get libz.la? Mar 05 14:05:27 I get only libz.so or libz.a Mar 05 14:05:31 for the native zlib Mar 05 14:16:29 hmmm zlib faq states: Mar 05 14:16:40 7.Why don't you use GNU autoconf or libtool or ...? Mar 05 14:16:45 Because we would like to keep zlib as a very small and simple package. zlib is rather portable and doesn't need much configuration. Mar 05 14:16:47 hmmm Mar 05 14:16:55 and gtk+-native looks for zlib.la Mar 05 14:16:57 oops Mar 05 14:16:59 libz.la Mar 05 14:36:41 khem, hi Mar 05 14:45:33 I'll try to disable the part that makes problem then Mar 05 14:45:43 because it's cups backend Mar 05 14:55:49 after updating OE, I get the following error when build my image: ln: failed to create symbolic link `/home/user/openembedded/oe/build/tmp-_2012_x-eglibc/deploy/images/panda/omapfb-image-panda.tar.bz2': File exists, any ideas on why and what I can do to fix it? Mar 05 14:58:03 is that file a link or a tarball? Mar 05 15:01:31 denisATeukrea: that file is a link, the existing one Mar 05 15:01:59 hmmm then maybe look at the permissions Mar 05 15:02:34 nothing wrong with the permissions. The problem is it tries to link to to the same file, with the same link twice Mar 05 15:02:49 and of course, the second time it complains about an already existing link Mar 05 15:03:25 ah ok no idea then Mar 05 15:03:42 maybe you it's fixed in the core based oe Mar 05 15:04:00 maybe, I'll try that Mar 05 15:04:01 thanks Mar 05 15:04:05 do you use oe-classic? Mar 05 15:04:20 if so nearly everybody switched to the core based oe Mar 05 15:08:24 no, I use the layered oe. Mar 05 15:55:19 Is (bitbake.)berlios.de accessible or is it just my network over here? Mar 05 15:56:23 PING bitbake.berlios.de (195.37.77.138) 56(84) bytes of data. Mar 05 15:56:23 ^C Mar 05 15:56:23 --- bitbake.berlios.de ping statistics --- Mar 05 15:56:23 9 packets transmitted, 0 received, 100% packet loss, time 8063ms Mar 05 15:56:33 awozniak: thanks! Mar 05 15:56:46 * kenws wonders what the differecence between ??= and ?= is Mar 05 15:57:55 ?= is evaluated at the time it's read.. if the item is NOT set, then it gets that value, otherwise it's ignored Mar 05 15:58:18 ??= is evaluated at the very end of the process.. if nothing has set the value during the var loading, then it's finally set to that value.. Mar 05 15:58:26 foo ??= "bar" Mar 05 15:58:32 foo ?= "bar2" Mar 05 15:58:37 yeah, the key differentiating factor is when you have multiple ??= defined, the last one wins, whereas with ?= the first one does Mar 05 15:58:38 foo will equal "bar2" Mar 05 15:58:45 * kergoth wanders off in search of coffee Mar 05 15:59:14 aah, that's why the user is supposed to specify the MACHINE using ??= Mar 05 15:59:30 to override whatever an external layer might say Mar 05 15:59:37 no.. the user is supposed to specify machine with "=" Mar 05 15:59:59 ??= is used to define a default value in case they didn't specify one.. Mar 05 16:00:01 the default local.conf seems to use ??= Mar 05 16:00:15 yes.. it specifies a default in case the user doesn't specify one.. Mar 05 16:00:34 you can also do MACHINE="qemux86-64" bitbake core-image-minimal and set it from the nevironment.. Mar 05 16:00:39 ??= will allow that to work Mar 05 16:00:52 (as will "?=" = simply sets it) Mar 05 16:01:34 yep, I think I got it - thank you! Mar 05 16:01:38 np Mar 05 16:03:14 I'm creating a layer that introduces a new MACHINE and I'd like to specify the default MACHINE in the conf/layer.conf Mar 05 16:07:11 hmmm the native toolchain is gone in core-based-oe it seems Mar 05 16:07:25 like HOST_CC etc... Mar 05 16:07:57 which pointed to i686-...-gcc Mar 05 16:09:12 kenws, it's not a good idea to specify the machine in a layer.. thats really up to the user's local.conf Mar 05 16:09:23 you can provide the configuration files, and everything else though.. Mar 05 16:09:35 I recommend simply placing in the README file (for the layer) what to set the machine to.. Mar 05 16:09:47 (this all assumes you will be publishing the layer) Mar 05 16:10:08 kenws, also make your bbappend machine arch or something like that Mar 05 16:10:11 if needed Mar 05 16:10:14 for something internal, i.e. a product layer.. then you can specify a new default local.conf file in your layer that has the machine already set.. Mar 05 16:10:20 *machine specific bbappend Mar 05 16:10:23 so when the user does the oe-init-build-env it'll use your local.conf Mar 05 16:10:39 (at least I'm pretty sure that works) Mar 05 16:11:40 Aah, so I'd rather create a local.conf for the layer. Interesting. Mar 05 16:12:12 I believe if you call it conf/local.conf.sample it'll use it.. Mar 05 16:12:30 but verify that, I may be wrong.. (I know we talked about making sure that would work a while back, and I think it got implemented) Mar 05 16:13:40 you should be able to do the same w/ bblayers.conf.sample as well.. Mar 05 16:13:46 ok.. I just looked.. it's not automatic.. Mar 05 16:13:56 you need to call oe-init-build-env w/ TEMPLATECONF set to the layer's conf dir Mar 05 16:14:12 so TEMPLATECONF=meta-my-layer/conf source oe-init-build-env will do it.. Mar 05 16:14:43 and this implies the user is sourcing the oe-init-build-env after the layer was added : ) Mar 05 16:14:44 looks like you'll will need both the bblayers.conf.sample and local.conf.sample to do that... Mar 05 16:14:54 no, thats why you need to specify it manually.. Mar 05 16:15:05 order is expected... user downloads oe-core, bitbake and your layer.. Mar 05 16:15:15 then runs TEMPLATECONFIG=yourlayer/conf source oe-init-build-env Mar 05 16:15:26 that will then use the samples from your layer, instead of the default "meta" layer Mar 05 16:16:03 What is there is to make it easier for distribution layers/creators to be able to setup default files for users.. Mar 05 16:16:09 alright, sounds convenient to me Mar 05 16:16:18 but since it's an added layer, something has to tell it where to look (instead of meta) Mar 05 16:16:55 (note, I've not used it directly in any layers I've generated, but the Yocto Project's Poky does..) Mar 05 16:17:32 ok Mar 05 16:18:34 * fray goes to get ready for his trip to Dallas... (speaking at RTECC tomorrow to promote the Yocto Project [and oe-core]) Mar 05 16:18:46 assuming my cold doesn't kill me first.. Mar 05 16:19:52 fray: Good luck! Mar 05 16:19:55 * fray wrote his slides heavily influences by cold medicine.. we'll see how bad they turn out to be when I give them.. :P Mar 05 16:20:03 thanx.. ;) Mar 05 16:20:04 heh :) Mar 05 19:24:35 kergoth: regarding ?= ??= dont the assignments get life when they are referenced so I suppose its the difference when bb is resolving the reference right ? Mar 05 19:26:18 no Mar 05 19:26:30 ?= is *immediate*. if unset at that point in the ast evaluation, it gets set right there Mar 05 19:27:18 hmm, can't remember on ??= if its on check or at RecipeParsed. I suspect the latter Mar 05 19:27:22 to avoid getVar overhead Mar 05 19:30:22 khem: imo ideally both ?=/??= (default value) and overrides, and even append/prepend would be done at getvar time with intelligent caching, thereby avoiding some issues relating to when such things occur Mar 05 19:30:28 but it's a bit late for that :) Mar 05 19:32:13 assignments are murky in bitbake Mar 05 19:33:27 it's tough to manage all the possibilities, from a design perspective. for example, var expansion in variable names.. to postpone that to getvar time would be nontrivial Mar 05 22:53:01 I'm playing with a gumstix overo airstorm. it seems to use some firmware for the wifi/bluetooth chip, and it works fine when I use the angstrom (oe-based) distro. I am toying with an archlinux dist and along the way I wanted to get the bluetooth chip working as it is in angstrom now, where are the gumstix-specific magic in the bitbake/oe build process that I could actually try and transfer over the the archlinuxarm side of the house? Mar 05 22:56:29 the bluetooth chip on there is a CSR BC04 **** ENDING LOGGING AT Tue Mar 06 02:59:58 2012