**** BEGIN LOGGING AT Wed Jul 24 02:59:59 2013 Jul 24 03:46:11 okay is there a way to get a local (recipe) env variable into a python function? Jul 24 03:46:51 apparently it's always empty... **** BEGIN LOGGING AT Wed Jul 24 05:27:42 2013 Jul 24 07:07:37 good morning Jul 24 08:08:58 morning all Jul 24 08:18:42 hi bluelightning Jul 24 08:29:16 morning all, hi bluelightning hi mckoan Jul 24 09:13:37 hi there. i'd like to be able to set BBLAYERS in a 'dynamic' way. e.g. in our project the set of layers we use is 'dynamic'. there are some optional layers, and i'd like BBALYERS in bblayers.conf to adapt using some sort of os.isfile(xxx/layer.conf). but i can't get that to work. is there a known mechanism to do that. Jul 24 09:29:15 ndec: other than modifying bblayers.conf externally, I don't think there's a standard way to do that no Jul 24 09:30:17 bluelightning: it seems i can't use @base_ifelse in the .conf file, indeed. Jul 24 09:30:47 i can get some simple python evaluation in the variable, but i need if..else. Jul 24 10:05:10 in a layer under mine there is a "do_compile_append()" that append "things" at my "do_compile()" it is possible dont run it without modify the under layer? Jul 24 10:05:54 hi all Jul 24 10:06:09 hi pb Jul 24 10:07:01 silvio: not easily. there are hacks you could do to work around it, but there is no good way to inhibit an append that's happening in some other layer. Jul 24 10:08:03 o/ morning all Jul 24 10:10:21 silvio_: sometimes it's a case where you should work with the maintainer of that layer to be able to accommodate your requirements Jul 24 10:10:25 hi eren Jul 24 10:10:27 hi pb_ Jul 24 10:10:34 hi bluelightning Jul 24 10:17:58 I missed yesterday's meeting :( Jul 24 10:19:20 bluelightning, pb_ , I m not sure that is good, but I append a "angstrom-feed-configs.bbappend" to a previous layer "angstrom-feed-configs.bbappend" and I force/overwrite their "do_compile_append()" whit my "do_compile()" Jul 24 10:19:52 in practice I have "do_compile_append()" equal to "do_compile()" in my bbappend Jul 24 10:20:40 silvio_: do_compile_append will always be added on top of do_compile, it won't override it Jul 24 10:20:52 what does angstrom offer, btw? Jul 24 10:21:22 bluelightning, I ovveride the "do_compile_append()" of the inferior layer Jul 24 10:21:38 that was appended to my do_compile() Jul 24 10:21:42 silvio_: as I said, you cannot Jul 24 10:22:53 bluelightning, It seam works, but I dont like this hack Jul 24 10:23:03 seams Jul 24 10:23:49 bluelightning, It seems works, but I dont like this hack Jul 24 10:26:29 silvio_: I'm inclined to say, if you're customising angstrom-feed-configs what you're building is no longer angstrom and you'd be better off having your own distro config / feed recipe / etc. Jul 24 10:32:55 bluelightning, sure I think is best solution, but I have to disable the "angstrom-feed-configs" used by the layer between mine and angstrom Jul 24 10:33:44 silvio_: which layer is modifying angstrom-feed-configs? Jul 24 10:33:47 silvio_: it sounds a bit as though this intermediate layer is deranged and you should stop using it. :-} Jul 24 10:35:38 I'd have to agree Jul 24 10:42:05 pb_, bluelightning : you right unfortunately the layer is put by the hardware producer, and I cannot easily remove it Jul 24 10:42:32 that is unfortunate indeed. which vendor is this? Jul 24 10:42:39 toradex Jul 24 10:42:58 silvio_: you should talk to the vendor; touching anything distro-related in a BSP layer is wrong Jul 24 10:44:04 ok, I contact them to solve the issue Jul 24 10:44:25 good plan Jul 24 10:50:51 Still a question, is not so clear to me how to works ipks feeds across layers, in the dev-man talk only about "/etc/opkg/base-feeds.conf" but how is intendet manage it if different layers put different feeds, howto disable one and enable last? Jul 24 10:51:57 well, feeds are fundamentally a distro problem. can you be more specific about what exactly you are ttrying to do? Jul 24 10:58:40 remove the "angstrom-feed-configs" at all and put "silvio-feed-configs" Jul 24 11:00:47 can't you just modify whatever recipe is currently depending on angstrom-feed-configs to make it depend on silvio-feed-configs instead? Jul 24 11:03:19 pb_, i think no, but now I m investigate Jul 24 13:17:36 If I canghe some "files://" in a recipe, it is possible rebuild it without a clean? Like -c rebuilt? Jul 24 13:23:40 silvio_: should be done automatically because hashes (based on mtime) change Jul 24 13:25:21 ensc|w_, only if u change the .bb or .bbappend, but if you change content of files pointed by "file://" it does not happens Jul 24 13:27:19 silvio_: which version of the build system? Jul 24 13:27:57 silvio_: that is supposed to work, it generates a checksum of all files referred to in SRC_URI and forces do_fetch if any change Jul 24 13:28:06 (all local files that is) Jul 24 13:28:38 bluelightning, BB_VERSION = "1.16.0" Jul 24 13:28:54 based on yocto 1.3 Jul 24 13:29:03 silvio_: it should support that then Jul 24 13:33:03 bluelightning, cheching in the /builb/out-eglibc/work/arm...blabla/packagename/folder/namefile.txt seams that if I dont clean it remain the same Jul 24 13:33:15 and build ystem doesnt rebuild recipe Jul 24 13:34:02 and build system doesnt rebuild recipe when i do "bitbake packagename" whitout a -c clean/cleanall before :( Jul 24 13:36:28 silvio_: which 1.3 release are you based on? Jul 24 13:51:44 bluelightning, OPENEMBEDDED_CORE=ae9dbd0e1e26ba2b35cbd08ec731aee62adedc23 Jul 24 13:53:27 silvio_: it's the bitbake revision I'm most interested in Jul 24 13:58:06 bluelightning, bitbake --version Jul 24 13:58:07 BitBake Build Tool Core version 1.16.0, bitbake version 1.16.0, now I ask to supplier more info about Jul 24 14:30:24 Hello, I would like to specify a "-dbg" package in IMAGE_INSTALL, but I do not want to install all the other debug dependencies. The extra dependency "-dbg" packages won't fit into image (and I'm not interested in them) anyway. Is there a way to do that easily? Jul 24 17:06:46 moin Jul 24 17:37:29 in OE, what decides whether or not a package is being installed? in particular, I have a kernel-headers package that gets created during kernel build which I need to build some other stuff, but it is not being installed. If I put a DEPENDS="kernel-headers" into the package that build-depends on it, OE complains that nothing PROVIDES this. i need this for some user-space (test) apps, that rely on (extended) versions of e.g V4L2 headers. Adding a DEPENDS on the Jul 24 17:37:29 kernel.bb (recipe) doesn't help, my kernel-headers package is not installed into sysroot. Adding RDEPENDS doesn't help either, kernel-headers still isn't being installed (and RDEPENDS is for runtime, not build-time anyway, as far as I understand). What is the correct solution? split out my kernel-headers package into a separate recipe? Why is OE not installing this package in the first place? Jul 24 17:38:39 andre_d: if kernel-headers is a package name as opposed to a recipe, then it would belong in RDEPENDS (i.e. a runtime dependency) Jul 24 17:39:45 but RDEPENDS is for runtime? I've tried adding it to RDEPENDS nevertheless, OE doesn't install it for some reason when building the recipe that needs it Jul 24 17:40:48 I may have misunderstood Jul 24 17:42:22 andre_d: if we're talking about a build-time dependency, i.e. your other recipe needs the headers in the sysroot so it can build, then it's DEPENDS and it's DEPENDS on the recipe that provides the headers and not any package that that recipe might create Jul 24 17:46:02 bluelightning: I've tried that, but OE doesn't seem to install all packages that are created by the recipe - it only installs some of the packages from my kernel recipe (kernel-headers _not_ being one of them) Jul 24 17:49:22 andre_d: it installs to the sysroot only a limited selection of what is installed by do_install, FYI packages aren't involved Jul 24 17:50:00 andre_d: the implementation is in sysroot_stage_dirs() in meta/classes/staging.bbclass Jul 24 17:50:18 andre_d: if a recipe needs th ereal kernel headers, e.g. an out of tree module build, then it should inherit an appropriate class (e.g. module.bbclass) and/or DEPENDS onv ritual/ekrenl, and should look in STAGING_KERNEL_DIR for the files Jul 24 17:50:21 as you can see from there it's based on path Jul 24 17:50:35 the main sysroot has the sanitized kernel headers from glibc, we wouldn't want the real kernel headers there Jul 24 17:50:42 right definitely what kergoth said Jul 24 17:50:45 kergoth: thanks :) Jul 24 17:51:50 but this is for user-space, it needs the sanitised headers from my kernel-headers package, I don't think I can or should use STAGING_DIR_KERNEL. I will have a look at staging.bbclass. Jul 24 17:52:18 maybe the correct solution is to provide my own version of linux-libc-headers? Jul 24 17:55:14 andre_d: not sure, but I think that's only indended for use by the libc... Jul 24 17:55:18 * bluelightning -> home, bbl Jul 24 17:56:26 andre_d: userland applications should not be using kernel headers directly except in particular, rare, cases Jul 24 18:04:48 gah, modified one thing, read_only_rootfs_hook, though in global context via an _append, and it's rebuilding systemd and stuff Jul 24 18:04:56 i really hate how brittle our checksumming is at times Jul 24 21:31:50 Does lpapp ever do anything but complain? he's as bad on the mailing list as IRC Jul 24 21:32:08 <3 gmail's Mute. /ignore for email Jul 24 21:37:48 * kergoth fights read-only-rootfs some more Jul 24 22:57:30 * kergoth ponders the r/o stuff discussed on the list Jul 24 23:38:06 kergoth: you've seen the ro-rootfs patches in contrib? Jul 24 23:44:07 mr_science: actually I haven't, i checked the lists but didn't see many pending patches. will check contrib Jul 24 23:54:43 i applied what i could to the version we're using but it wasn't the best match Jul 24 23:55:06 the part that i could apply has been helpful Jul 25 00:13:07 night... **** ENDING LOGGING AT Thu Jul 25 02:59:58 2013