**** BEGIN LOGGING AT Tue Oct 17 03:00:03 2017 Oct 17 07:10:55 Hi Oct 17 07:11:16 it was discussed here before, but I didn't get proper answer yet :) Oct 17 07:11:24 I'm bumping from jethro to rocko Oct 17 07:11:44 we use yocto to build our C++ application and create final image Oct 17 07:12:25 for development purposes (and to avoid long build times) we use manual corss compilation of our application by usgin yocto generated sysroot + toolchain (use toolchain file with correct paths) Oct 17 07:12:38 AFAIU it's not the case anymoe for rocko as every pckaga have separate sysroot Oct 17 07:12:48 is ther epossibility somehow to use our old approach>? Oct 17 07:13:03 or do we need to build SDK from yoct oand use that for cross compilation? Oct 17 07:14:33 open-nandra I think I did answer to you :-) Oct 17 07:15:05 Please check the developer manual, there is a tool called devtool that I think is a better workflow than using sysroot and external toolchain Oct 17 07:15:06 http://www.yoctoproject.org/docs/2.4/sdk-manual/sdk-manual.html#using-devtool-in-your-sdk-workflow Oct 17 07:15:10 melonipoika: but we cannot use yocto for our application development Oct 17 07:15:10 it took long time Oct 17 07:15:28 so basically you cross compila manually and then when we create final image we use yocto Oct 17 07:15:34 it's much faster Oct 17 07:16:09 compilation took some time so we choose this approach which is preferabe Oct 17 07:16:10 Do app developers have access to the yocto builds? Oct 17 07:16:11 compilation took some time so we choose this approach which is preferable Oct 17 07:16:27 open-nandra: for external application, i'd say sdk is the proper way, not really hard to do :) Oct 17 07:16:48 yes I know Oct 17 07:17:14 nayfe: but I hope that there exist also "old way" Oct 17 07:18:19 open-nandra: old way looks hackish to me Oct 17 07:18:26 compared to sdk Oct 17 07:19:09 nayfe: well but it works fine because we have sysroot + toolchain so :) Oct 17 07:20:07 nayfe: OK I'll generate SDK and use that one for cross compilation Oct 17 07:20:24 nayfe: it's just one step more -> took more time ;) Oct 17 07:21:44 open-nandra: indeed Oct 17 08:20:42 Hi, is this appropriate channel for technical questions? (to be more precise, I am having trouble getting git clone of private repo using ssh, and google has not been helpful) Oct 17 08:25:16 Well, I'll shoot anyway. So this command works, using ~/.ssh/id_rsa: git clone git@github.com:MyStuff/project.git Oct 17 08:26:27 but I have not been able to come up with SRC_URI, which would work. I always get error message: Oct 17 08:26:40 fatal: unable to connect to git@github.com: git@github.com: Name or service not known Oct 17 08:28:32 hyde: try git://github.com Oct 17 08:31:14 I have tried that., among other variants. but I'll try again just in case I did some mistake there Oct 17 08:32:28 hyde: can you share with us your SRC_URI? Oct 17 08:33:50 ah, got it! Oct 17 08:33:55 this worked: SRC_URI = "git://github.com/MyStuff/project.git;protocol=ssh;user=git" Oct 17 08:34:28 (rubber duck problem solving showing it's power once again...) Oct 17 08:34:36 thanks :) Oct 17 08:47:22 zero_note: Your solution yesterday, with the do_configure_append() worked beautifully. Now the only thing I need to patch is to put the dts in a u-boot Makefile so it gets compiled. I.e. I can change both the config and the dts without having to generate patches every time. Thanks! :D Oct 17 09:00:49 JoiF: glad I've helped you ;) Oct 17 11:32:46 rburton: how did you learn about the wpa replay problem? Oct 17 11:33:24 I mean, yes, it's all over my facebook feed, but do we have a formal channel for tracking the high profile security issues? Oct 17 11:34:07 yocto itself doesn't have a backchannel into the embargo security issues Oct 17 12:40:13 kanavin: it would be nice if we did have but its hard to establish and its based on personal trust... Oct 17 12:57:29 denix: there? Oct 17 12:57:47 denix: meta-ti is breaking other machines builds Oct 17 15:42:03 Hi, for imx6 users, someone is using imx_usb_loader with SDP protocol ? Oct 17 17:33:37 Hey, I get build errors when building meta-toolchain-qt5 for the raspberrypi. The are some moc files during qtxmlpatterns which the compiler don't find. Has anybody a suggestion what the problem can be? Here's the log. https://pastebin.com/E6Bt0Lt0 Oct 17 18:18:51 splashx11: seems a problem in qt5-ptest.inc for me ? what version do you use Oct 17 18:26:42 nefe: I am using poky 2.3. meta-qt: master:2c3ef00f53683fbdce5b390950b49b67da85d3a1 Oct 17 18:32:43 splashx11: did you try master-next just in case ? Oct 17 18:37:28 nefe: not yet, I will try Oct 17 19:26:21 lamego: looks like the scons upgrade means it uses py3 now, which means serf as the sconscrupt file or whatever its called won't parse anymore (see the errors in the mut run on the ab) Oct 17 19:27:52 rburton: will look into that. Thanks for the heads up. Oct 17 19:28:17 print 'Warning: Used unknown variables:', ', '.join(unknown.keys()) Oct 17 19:28:17 ^ Oct 17 19:28:19 SyntaxError: invalid syntax Oct 17 19:29:10 i can replicate with just bitbake serf, fwiw Oct 17 19:32:12 lamego: its because scons is still using py2 but it does from __future__ import print_function, which means you need to use py3-style print Oct 17 19:32:29 i hope this was release noted! Oct 17 19:32:46 should be trivial to fix serf and whatever else (nothing?) uses scons in oe-core Oct 17 19:34:17 rburton: that is of great help. Oct 17 19:35:24 erm no they didn't put it in the changelog Oct 17 19:36:15 ha my old boss is responsible Oct 17 19:36:17 small world Oct 17 19:38:11 import from future doent force you to use printf from py3 does it ? Oct 17 19:39:03 if nothing uses scons in oe-core then we should move a recipe to oe-core which uses scons Oct 17 19:39:41 alljoyn was based on scons and there might be other layers which expect scons support from oe-core Oct 17 19:41:43 khem: serf uses it Oct 17 19:42:09 ok then let serf survive there Oct 17 19:42:09 khem: the print_function import makes non-function print a syntax error Oct 17 19:42:17 we need serf to build svn Oct 17 19:42:30 trust me i've checked a number of times that we still need it Oct 17 19:42:37 I see, so it sort of pushes you to forward port your code while you still use py2 Oct 17 19:42:49 yeah they're finally sort of migrating Oct 17 19:43:01 i *think* scons 3 will let us use py3 Oct 17 19:43:15 it just would have been nice if this was in the release notes :) Oct 17 19:43:39 OK, I wonder by the time the world migrates away from py2, we will be ready for py4 Oct 17 19:44:03 py2 migration will never fully happen Oct 17 19:44:04 or dump python for something else like boa Oct 17 21:59:27 otavio: how do you figure? just tried poky/master with several BSPs and meta-ti built for different machines/archs - see no breakage Oct 18 01:34:45 just run 2to3 :) **** ENDING LOGGING AT Wed Oct 18 03:00:00 2017