**** BEGIN LOGGING AT Mon Sep 24 02:59:59 2018 Sep 24 06:46:46 New news from stackoverflow: aptina camera interface with gstreamer Sep 24 06:48:13 khem: well yes. but this doesn't change the fact that i have to use either local.conf or a distro config, right? Sep 24 06:51:27 yes Sep 24 09:17:16 New news from stackoverflow: Will install_requires in setuptools work without explicit yocto-recipe for certain packages Sep 24 10:00:55 rburton: should we wait for the meta-yocto non-IA bsp bumps or cut M3? Sep 24 10:01:49 any idea when the non-ia bumps will turn up? Sep 24 10:01:55 rburton: no Sep 24 10:02:14 lets do m3rc1 today then Sep 24 10:02:43 rburton: agreed. Its gone on long enough. I'd take those other bumps in M4 Sep 24 10:02:58 RP: musl bump on the list which looks kind of useful to have Sep 24 10:03:45 rburton: risk just merging and build? Sep 24 10:05:02 just fired a quick build with musl to see if it at least builds something Sep 24 10:07:49 rburton: I've set the AB at those... Sep 24 10:42:52 RP: the fribidi thing that khem/jama reported is real and might break a build on centos7. working on a fix now. Sep 24 11:18:13 rburton: ok, thanks! Sep 24 11:18:37 ok just did a test and it appears to work on our centos box so might not break anything Sep 24 11:22:29 rburton: does raise some questions about what the differences are... Sep 24 11:25:23 yeah Sep 24 11:32:13 I've also just remembered to change LAYERSERIES Sep 24 11:54:53 rburton: rc1 building... Sep 24 11:58:58 godspeed Sep 24 12:26:59 I want to use SRC_URI file:///foo in a postprocess hook in an image recipe, is that possible? Sep 24 12:27:37 I guess I need to enable a fetch step then... seems very hard :-) Sep 24 12:31:40 rburton, so the multithreaded xz can break packaging for rpm and opkg (oom) Sep 24 12:33:21 normally for small files <16MiB xz compression level 2 or 3 is more than enough, I bet we still have default 6 Sep 24 12:34:28 I think the multithreading code in xz is not mature enough (it wasn't last I've looked at the manual) Sep 24 14:00:37 What will the result of VAR ?= "something" and VAR_append = " else" be? Sep 24 14:03:02 "something else" or " else"? Sep 24 14:03:52 sveinse: I'm pretty sure it will be "something else" because ?= is evaluated immediately and _append is deferred until the end of parsing Sep 24 14:05:57 My reason is that conf/bblayers.conf is by default setup with BBLAYERS ?= and I need a programmically way to add new layers. Concatenating lines with BBLAYERS_append= is the simplest naïve approach Sep 24 14:06:47 sveinse: ?= means set if not already set so its immediate in that it will always have a value after that Sep 24 14:07:04 sveinse: ??= is the deferred version Sep 24 14:08:07 RP: thanks. Sep 24 14:09:43 RP: I found the (undocumented) PACKAGE_DEBUG_SPLIT_STYLE = "debug-with-srcpkg". You think that should be the default? Sep 24 14:09:50 BTW that is the use case for BBLAYERS ?= in conf/bblayers.conf? I.e. when is it expected to be overriden by anything? This being the top-level config and all. Sep 24 14:10:07 s/that/what/ Sep 24 14:13:46 JPEW: er, it was probably introduced with a plan in the back of my mind to switch :) Sep 24 14:15:25 RP: Ya makes sense... I *think* it might still have a few holes w.r.t. IMAGE_GEN_DEBUGFS. I haven't seen anything that would lead me to believe the -src packages would get added to the debugefs image. A change for 2.7 perhaps? Sep 24 14:17:26 JPEW: definitely 2.7 and yes, may need work but I'd be happy to see it Sep 24 14:17:45 JPEW: maybe an ehnancement bug for 2.7? Sep 24 14:19:11 Ok. Do you think this would be two different tasks potentially? The first would be the (possible, still verifiying) bug where the -src packages aren Sep 24 14:19:44 oops.... aren't in the debugfs (which might need backported) and the second being to change the default? Sep 24 15:00:51 RP the 'enabled warnings' in the AB is showing up on the wiki. is that something we can turn off for older stable branches ? Sep 24 15:04:43 armpit: hmm, I think we need to tweak the wiki code to separate warnings from errors Sep 24 15:04:47 at a minimum we need to warn swat Sep 24 15:05:01 armpit: yes, I'd not realised warnings would show there as errors Sep 24 15:05:28 armpit: need code changes to the wiki plugin code Sep 24 15:06:16 RP btw stable/sumo-next is good to go Sep 24 15:07:22 * armpit rocko has build failures needing correction Sep 24 15:27:21 RP: obviously pulling on the fribidi thread has revealed RAZOR WIRE Sep 24 15:27:28 Sep 24 15:28:02 rburton: :( Sep 24 15:28:16 rburton: sadly I'm not surprised, it looked at risk of that :) Sep 24 15:28:19 er, :( Sep 24 15:28:35 got a few fixes for our class and upstream are engaging with fixing assumptions in their code Sep 24 15:48:28 New news from stackoverflow: Use python3-pytz for an Application in Krogoth branch in yocto Sep 24 15:57:24 rburton: glad they're being responsive :) Sep 24 16:11:15 hi o/ - I am working on a recipe for electron that depends on another recipe for libchromiumcontent that is built separately. The thing is libchromiumcontent is usually built as part of electron itself, but electron also allows using an external build by passing the dir where to find the libchromiumcontent static lib (2GB) and sources/headers to it. Given the libchromiumcontent build artifacts are only needed to build electron, is Sep 24 16:11:16 there a way to use the build artifacts from it from the electron recipe? I thought of using something like the kernel shared_workdir for that but wonder if there is a more "proper" way to do it? Sep 24 16:11:21 any tip here? Sep 24 17:15:47 Has anyone seen wic generate a fat16 partition when fstype is set to vfat? Sep 24 20:07:18 I’ve created a git repository for my recipes (meta-sockeye-gm) and cloned it in the poky directory alongside the other meta-* directories. When I try to execute ‘devtool update-recipe sockeye-gm-process’, I get the following error: https://www.pastiebin.com/5ba943bd8cb34 Sep 24 20:07:28 What am I doing wrong? Sep 24 20:51:53 on sumo, what could cause the fetcher not to honor downloadfilename and subdir ? (here waf download in meta-oe's mpv) Sep 24 20:53:06 even -vDDD does not seem to give much insight of what the fetcher is doing Sep 24 21:39:42 I'm not sure what the hell is going on - if I append to do_unpack to copy the download waf to the correct place, the build proceeds, and my copy is replaced by a symlink to the waf-1.8.12 I had originally copied, but if I do nothing it just fails complaining it cannot find waf Sep 24 21:43:36 is it possible to inherit a class in a bbappend? Sep 24 21:44:05 jdel: yes Sep 24 21:44:59 yann: There were some changes to waf.bbclass to try and detect the version, it is possible they broke the recipe? Sep 24 21:46:27 possible - looks like do_configure creates the symlink, but waf_preconfigure runs first and expects that symlink to be present Sep 24 21:46:37 yann: That would do it Sep 24 21:47:27 OTOH, the SRC_URI seems to request waf-1.8.12 to be renamed to git/waf, which it is not (fortunately for do_configure, it would seem) Sep 24 21:48:00 yann: It is *really* weird that recipe downloads waf independently Sep 24 21:48:43 yann: It's supposed to be included in the source (although a lot of distros balk at that.... long argument that I'm not a part of :) Sep 24 21:48:47 JPEW: ok, digging that Sep 24 21:49:55 I must say I was surprised, but then when I saw that waf.bbclass does not do anything to get a waf into place... Sep 24 21:50:31 in fact, mpv ships a bootstrap.py which is tasked with downloading waf... Sep 24 21:50:46 yann: Hah. I guess thats one way around it. Sep 24 21:50:58 * yann hates those build systems with silly ideas Sep 24 21:51:21 yann: Ya waf is.... different... Sep 24 21:51:42 their bootstrap.py downloads 1.9.8 - guess the packager found that to be too old Sep 24 21:52:51 and well, doing a download separately from the fetch step is evil anyway Sep 24 21:52:59 yann: I give waf points for trying new things. It does solve some of the problems in a unique way, but it makes some more in the process :) Sep 24 21:53:37 yann: Anyway, hopefully that gets you headed in the right direction, I have to go. Sep 24 21:53:45 "oh, those n build systems all have problems, let's make a new one!" ? Sep 24 21:53:53 ok thx! Sep 24 22:14:27 JPEW: damned, I had missed that meta-rockchip was overloading the mpv recipe... Sep 24 22:25:34 Does anyone know what I need to do about this cmake related warning? SkippingbecauseOE_QMAKE_PATH_EXTERNAL_HOST_BINSisnotdefined Sep 24 22:26:01 I'm getting a build error because Qt include paths aren't on the command line and I'm wondering if this is related. Sep 24 23:15:42 Hi, what is the best way to rerun configure if a local file has been changed? I have tried manually adding to src_uri all the files in this directory, and tried setting do_configure[file-checksums] to the list of files in the directory. Both of these ways works if I ADD a file but not delete. any ideas? Sep 24 23:37:47 jae1: if you're trying to build from a local source tree that's not really the right way to go about it. I would recommend using the externalsrc class, or if this is just for local development use "devtool modify" Sep 24 23:49:38 bluelightning: I need some other files in addition to a source tree(which i am defining just like other recipes) Sep 24 23:51:26 ( it is files that need to be provided by the user so will be provided locally) **** ENDING LOGGING AT Tue Sep 25 02:59:59 2018