**** BEGIN LOGGING AT Thu Mar 27 02:59:58 2014 Mar 27 05:53:38 JaMa: where is the result of bitbake -S dumped to? Mar 27 07:50:31 good morning Mar 27 07:59:19 hello using FILESEXTRAPATHS_prepend seems to have no efect on an inc file the location being changed from bb to bb. Mar 27 07:59:55 does someone has a suggestion? Mar 27 08:04:07 yes, the inc in oe-core wins. Create and refer to a new one in your layer (it can just include/require the .inc in oe-core) Mar 27 10:15:04 morning all Mar 27 10:40:51 I would like to understand variable SELECTED_OPTIMIZATION a little bit more Mar 27 10:41:05 if I add some flag to SELECTED_OPTIMIZATION Mar 27 10:41:16 is it going to be passed to -native tools as well? Mar 27 10:42:45 RP: ping? Mar 27 10:43:50 zecke: pong Mar 27 10:44:11 Xz: I think BUILD_OPTIMIZATION is used for that Mar 27 10:44:24 Xz: no, the value of SELECTED_OPTIMIZATION goes into TARGET_CFLAGS i.e. for the target only Mar 27 10:45:48 RP: so BUILD_OPTIMIZATION will affect only my target binaries? Mar 27 10:46:10 bluelightning: RP: I'm puzzled now Mar 27 10:46:16 Xz: SELECTED_OPTIMIZATION only changes target binaries, BUILD_OPTIMIZATION only affects native ones Mar 27 10:46:55 Xz: BUILD meaning the build host Mar 27 10:47:49 RP: bluelightning now if I build SDK and added a flag into SELECTED_OPTIMIZATION - will SDK add that flag everytime smb uses it? Mar 27 10:48:22 RP: bluelightning I mean after someone installs SDK and starts using - will my flag be by default appended? Mar 27 10:50:14 Xz: the toolchain env setup scripts set CFLAGS from the TARGET_CFLAGS value set when building the SDK, yes Mar 27 10:51:40 RP: okay. so CONFFILES missing makes sense. What about do_install? Mar 27 10:51:57 RP: these were the two items I changed and that didn't reflect in a .X AUTOPR change Mar 27 10:52:11 sorry. my wife is calling for lunch. Mar 27 10:52:13 bbl Mar 27 10:52:32 zecke: do_install should trigger install to rerun for sure Mar 27 10:52:55 bluelightning: that's pretty cool, I like thah :) Mar 27 10:53:30 our docs are a bit thin on this area, I've added a note to improve them Mar 27 12:56:13 what's the best way to clean out old files from the downloads directory? Mar 27 13:00:30 at the moment I have a cache folder that I move downloads to, then I set the SOURCE_MIRROR_URL, INHERIT += "own-mirrors" and run bitbake -c fetchall someimage Mar 27 13:00:37 hi all Mar 27 13:01:37 Anyone here can help out a quick question? Attempting to build a Yocto I.MX6 image, works nicely - but libSDL2 is missing from my target image (it is in the SDK) Mar 27 13:01:55 Do I need to do more than 'DEPENDS += 'virtual/libsdl2' and 'IMAGE_INSTALL_append += libsdl2 ? Mar 27 13:07:58 rink_, where are you doing that DEPENDS? in local.conf? don't do that please Mar 27 13:08:07 and use IMAGE_INSTALL_append = " libsdl2" Mar 27 13:08:09 notice the sapce Mar 27 13:08:18 or IMAGE_INSTALL += "libsdl2" Mar 27 13:08:19 don Mar 27 13:08:31 don't mix += with append, I have no idea what happens... Mar 27 13:09:54 i've made a custom layer Mar 27 13:10:23 it ends with 'inherit core-image' Mar 27 13:10:59 I don't use IMAGE_INSTALL +=, that caused problems Mar 27 13:11:50 you can use IMAGE_INSTALL += *after* the inherit Mar 27 13:11:57 (in an image recipe) Mar 27 13:12:14 oh, i see! Mar 27 13:12:59 typically though assuming your application is packaged by the build system, you wouldn't explicitly add a library to be installed into the image - libraries should be brought in as dependencies of the application's package (automatically) Mar 27 13:13:37 well, the application is under development Mar 27 13:13:42 which is why I'm using the SDK to build it Mar 27 13:14:07 that works fine, but when I try to run it, it complains about missing shared libs (which makes sense, as I do not see them in the resulting image) Mar 27 13:14:33 so I'm guessing bitbake's dependency thing decides that it isn't needed by the runtime on the image and doesn't install them Mar 27 13:16:26 hmmm Mar 27 13:16:34 moving the 'inherit' to the top seemed to help things... Mar 27 13:17:50 disclaimer: I've started using Yocto yesterday; it's been a nice experience so far. Mar 27 13:36:18 alright... now to try to boot this kernel on a WR703N with UBoot, while not knowing much of anything about UBoot Mar 27 13:36:41 https://www.yoctoproject.org/download/ubiquiti-networks-routerstation-pro-routerstationpro-3 is very similar hardware (same chip, I think) but has instructions for RedBoot Mar 27 14:11:12 this chip isn't identical (AR7161 vs. AR7240) but it looks like it should be close enough; I wonder if this is just a matter of passing UBoot the right parameters Mar 27 14:29:55 bluelightning, You were saying yesterday to rink_ [13:11] that we should use IMAGE_INSTALL += after inherit. First, the doc does not say anything about that? Second does the doc enforce the fact that it's better to simply always use IMAGE_INSTALL_append? Mar 27 14:30:05 bluelightning, third will recommend changing this IMAGE_INSTALL += for IMAGE_INSTALL_append? https://github.com/gumstix/meta-gumstix-extras/blob/dora/recipes-images/gumstix/gumstix-console-image.bb Mar 27 14:32:06 kapare: the doc should probably be fixed, but to be honest I think += is better for the image recipe as long as it appears in the right place Mar 27 14:32:24 kapare: there's no need to change all recipes though if they already work Mar 27 14:35:39 bluelightning, the right place is after inherit? or after also other vars? or only this one? Is there other ordering that we must take care of in a recipe-image or local.conf? Mar 27 14:58:01 kapare: ordering around inherit is almost always a concern in every recipe, not just images. inherit is an immediate operation, so depending on the behavior of the class, some things nee dto be done after the inherit rather than before. there's a reason functions like do_configure always go after the inherit autotools :) Mar 27 15:00:06 kergoth, thx for the clarification still new to a lot of that process ;) Mar 27 15:02:06 Admittedly, it'd be nice to be more consistent in this regard. In my personal opinion, I'd like to see all inherits move to the top of recipes. It's almost never right for a class to override the recipe inheriting it, but if it does need to do that, that's what overrides are forl Mar 27 15:02:16 s/l$/ Mar 27 15:02:20 heh Mar 27 15:02:23 anyway.. Mar 27 15:04:32 kergoth, I was surprise to not see them there since I thought of them as import or include ;) Mar 27 15:50:06 I have a binutils bbappend that adds a patch. How can I make it not appending to -natve? Binutils recipe in poky does: BBCLASSEXTEND = "native" Mar 27 15:51:03 SRC_URI_append_class-target = " foo" will append just for the target recipe Mar 27 15:51:46 kergoth: good trick Mar 27 15:51:59 kergoth: will it work for Yocto 1.4.2? Mar 27 15:52:02 in the past we had to use an intermediate variable with native/nativesdk overrides to empty that variable Mar 27 15:52:08 not sure if class-target was available in 1.4 Mar 27 15:52:11 if not, use such an intermediate var Mar 27 15:52:33 MY_SRC_URUI = "foo", MY_SRC_URI_class-native or MY_SRC_URI_virtclass-native depending on version Mar 27 16:07:36 hrmmm, my build finished, and I have the squashfs image I asked for, but apparently no kernel? Mar 27 16:08:46 all I did that I remember that should have influenced that was KERNEL_IMAGETYPE = "uImage" rather than KERNEL_IMAGETYPE = "vmlinux" Mar 27 17:06:28 I build core-image-minimal. then I added IMAGE_INSTALL_append=" shadow", so my rootfs got that package. Then I removed the shadow IMAGE_INSTALL. Now I still get the shadow package automatically installed when I re-create the rootfs with "bitbake core-image-minimal -c rootfs -f". Mar 27 17:06:52 How do I reconfigure the rootfs configuration so it does not install no longer configured packages? Mar 27 17:16:07 RP: re. :} Mar 27 17:17:19 volker-: use bitbake -e core-image-minimal and see what is being in IMAGE_INSTALL. Does it still show your append? If not you can consule the image creation log. It might show you for which depedency shadow was installed? Mar 27 17:21:26 zecke: core-image-minimal doesn't has it as dependency and no other packages are installed. Mar 27 17:22:37 zecke: the bitbake -e command only shows: IMAGE_INSTALL="packagegroup-core-boot run-postinsts " Mar 27 17:24:47 maybe you grabbed the wrong image file, and the new one really doesn't have shadow? ;) Mar 27 17:25:19 kergoth: no, I check the qemux86_64-poky-linux/core-image-minimal/1.0-r0/rootfs/ folder which definitly gets rebuild Mar 27 17:44:48 zecke: welcome back :) Mar 27 17:45:01 zecke: I'm here for about 15mins Mar 27 17:45:27 zecke: when you run bitbake -S it places sigdata files into the STAMPS directory Mar 27 17:45:50 zecke: the first key thing to figure out is whether the sigdata files are different before and after your change Mar 27 17:46:24 zecke: is they're the same, the change wasn't detected. If it changes, that should trigger the PR server to increment its PR value Mar 27 17:46:35 s/is/if/ Mar 27 17:47:49 RP: ah thanks the "STAMPS" directory thing was missing. strace bitbake -S created too many lines of output to be just be lucky. :) Mar 27 17:48:13 RP: I will have another look tomorrow and report back. So far it looks like I was just unlucky with CONFFILES :) Mar 27 17:50:13 zecke: yes, we should have that one in the magic list. I wish there was a way to auto-generate it but out python ast parsing isn't good enough for that :) Mar 27 17:55:02 RP: "shadow" variables? A special cow dictionary and see which variables were used for do_package? It probably will not scale. :) Mar 27 17:55:32 zecke: the trouble is we need to compute the hash before the function ever runs Mar 27 17:55:55 zecke: i.e. can we fetch the output of this task from an sstate cache when we know the checksum Mar 27 17:56:11 RP: hehe, predicting the future. How can that be difficult? :} Mar 27 17:56:33 I do think we probably should sanity check we did "guess" the right things though after the task does execute Mar 27 17:58:34 RP: yes, this sounds like a good idea. I think it is just about missing variables so even a good error message naming the variable and where to insert it and where to report the issue would rock. (I use ZFS on my FreeBSD servers and the error messages are a different league) Mar 27 21:30:50 RP: come the end of the next AB run I need a quick downtime (updates). **** ENDING LOGGING AT Fri Mar 28 02:59:59 2014