**** BEGIN LOGGING AT Tue Jul 17 03:00:03 2018 Jul 17 09:58:26 rburton: killed the -next build as that sqlite upgrade is bust Jul 17 10:03:48 khem, hi, th egcc testcase https://pastebin.com/bhwA0n58 Jul 17 10:04:03 pls check it Jul 17 11:04:40 hmm, we never made that buildtoold release :( Jul 17 11:04:50 well, uninative Jul 17 11:06:01 does anybody know about python recipes? I try to build python3-awscli, but it fails on an RDEPENDS on python3-futures (with the "s"). Googling and grepping around I don't see where python3-futures would come from Jul 17 11:06:57 I am on Rocko. It seems like the code in question has been removed in Sumo and in older branches. But in Rocko it is still there Jul 17 11:09:19 https://github.com/intel-iot-devkit/meta-iot-cloud/blob/rocko/recipes-aws/python/python-s3transfer_0.1.10.bb Jul 17 12:15:20 RP: well, I sent out an email about backporting the multilib fixes Jul 17 13:51:09 Hi, Is it possible to setup a function to be executed at the end of each bitbake-session execution, just before calling sys.exit(ret) , but without patching the original bitbake script "poky/bitbake/bin/bitbake"? Is there a way to do this from a recipe (event handler, ....) ? Jul 17 13:57:41 I guess you could write a recipe my-final-fake-recipe.bb which DEPENDS on your image which you usually build. In that recipe you can execute whatever code you want. From that on you don't call bitbake whatever-image anymore, but bitbake my-final-fake-recipe. Obviously I haven't tried this, maybe there is an easier way Jul 17 14:04:07 shatrix: yes, an event handler on BuildComplete Jul 17 14:16:54 shatrix: https://github.com/rossburton/meta-ross/blob/master/classes/timer.bbclass Jul 17 14:51:10 hi Jul 17 14:51:49 I'm trying to debug our application by setting INHIBIT_PACKAGE_STRIP = "1" in the package recipe, however the resulting binary is stripped anyway Jul 17 14:51:52 any ideas? Jul 17 14:59:01 ente: the debug symbols are separated into the -dbg package, install that to access them Jul 17 15:00:33 I think qmake might be stripping the debug symbols before yocto gets a chance to Jul 17 15:01:02 you would get a warning if the symbols are stripped before being processed by the packaging system Jul 17 15:01:10 they are most likely in the dbg package.. Jul 17 15:02:32 Is there a YPTM today? Jul 17 15:03:49 JPEW: not aware of one Jul 17 15:04:26 JPEW_: thanks for the patches btw, need some time to digest them! Jul 17 15:05:52 RP: Sure. I think they still need some spit and polish, but I figured getting them on the list earlier rather than later would be good Jul 17 15:06:50 JPEW_: definitely Jul 17 15:07:17 thanks Jul 17 15:17:51 the package is not stripped, but the rootfs is. the dbg package is empty, the binary itself says "not stripped" when I extract the .deb Jul 17 15:31:44 Hmm, do_package of ltp spends 5 mins in split_and_strip_files and 4.5 mins in package_do_shlibs Jul 17 16:07:22 does anybody know what are the conventions in python packaging? I would have naively assumed that every recipe needs to have RDPENDS for all packages its code imports. However, that doesn't seem to be the case, I am adding RDEPENDS one by one after I get runtime errors Jul 17 16:22:46 al9pf and I just started a #yocto-training channel if anyone is specifically interested in YP training related discussion Jul 17 16:29:09 is that channel for trainers or trainees? Jul 17 16:39:58 I'm goin gto say he means trainers :) Jul 17 18:14:56 u1106_: that shouldnt be the case, all python packages are supposed to have RDEPENDS to what they require to import Jul 17 18:15:12 u1106_: what package are you having trouble with? Jul 17 20:56:08 rburton: sent a binutils update, its a one large patch, if it does not work you can cherry-pick from http://git.openembedded.org/openembedded-core-contrib/commit/?h=kraj/master&id=2f49964bffea1aa3e90122b01372cd4b784e1397 Jul 17 20:56:32 thanks :) Jul 17 20:56:54 rpi/risc-v world builds for musl/glibc showed now new fails for me Jul 17 20:57:04 so throw it on AB when you can Jul 17 20:58:38 fired Jul 17 21:07:35 link ? Jul 17 21:15:26 u1106_: For trainers. Jul 17 21:42:57 khem: https://autobuilder.yocto.io/tgrid top row Jul 17 22:04:47 aehs29: I was adding python-awscli @ rocko and needed 4 or 5 extra RDEPENDS to make it work at runtime. Only in the end I noted that not all (if any) were direct depedencies, also imported packages imported more, which they obviously did not RDEPEND on. Would need to do the whole exercise with 4-5 images again to be able to name the culprits :( Jul 17 22:25:53 Are we having yocto project mirror problems? Jul 17 22:47:40 u1106_: python packages do RDEPEND on what they need, indirect dependencies should be handled as well (since when you RDEPEND to A that RDEPENDS on b, you get b as well) Jul 17 22:48:01 u1106_: but if you are adding a new package you have to initially add the RDEPENDS it needs **** ENDING LOGGING AT Wed Jul 18 03:00:01 2018