**** BEGIN LOGGING AT Mon Aug 24 02:59:59 2015 Aug 24 06:58:56 woglinde: http://hastebin.com/fugiyubeji.coffee Aug 24 07:04:59 I was wondering if you could give me a quick tip on how to proceed? Update, dig into recipes. Report the problem? Aug 24 07:32:41 pompomJuice: what does your PREFERRED_PROVIDERS look like Aug 24 07:34:19 especially PREFERRED_PROVIDER_virtual/java-native Aug 24 07:36:40 it should point to some native recipe like "jamvm-native" Aug 24 08:23:51 PREFERRED_VERSION_openjdk-7-jre = "75b13-2.5.4" Aug 24 08:23:51 PREFERRED_VERSION_icedtea7-native = "2.1.3" Aug 24 08:24:00 I dont have a virtual/hava-native Aug 24 08:24:03 let me put that in Aug 24 08:24:09 @khem Aug 24 08:25:03 aah I missed those from the README Aug 24 08:29:04 morning all Aug 24 08:44:03 khem, I am getting the same issue still with this config: http://hastebin.com/xuyehedisa.md Aug 24 09:12:52 externalsrc question: we use externalsrc for out project. in our project, the main software stack is in the same tree as our oe recipes, so the src_uri is simply a local path. we picked externalsrc for the recipe for our software because we dont know how to otherwise make sure that the software is rebuilt when it is modified. Aug 24 09:13:19 but the result is that it is recompiled *every* time. as if bitbake was running with the -f flag. Aug 24 09:14:09 are there any more refined solutions for this problem? I thought of perhaps somehow hook into git to check if something changed, but I dont know how this would help with staged and unstaged files. Aug 24 09:18:53 dv_: right, it does behave that way Aug 24 09:19:15 dv_: in theory you could influence the task signature based on md5sums or timestamps of all of the files in the source tree or perhaps some report from git status --porcelain; I just wasn't sure if that would scale so I never investigated it Aug 24 09:20:08 I know that waf (the build system) checks for changes in the files, not just the modification timestamps, and it does so incredibly quickly. and it is python based, like bitbake. Aug 24 09:20:17 perhaps its scanner approach could be adopted. Aug 24 09:20:23 another reason I didn't get into that was I had considered externalsrc to be something that people use temporarily while they are doing development rather than on an ongoing basis; but I do know there are people using it in the latter way Aug 24 09:21:07 I think in-tree sources are not that esoteric Aug 24 09:21:35 it gives us one platform which we can use both for development and for production builds Aug 24 09:22:10 OHAI! Aug 24 09:22:40 It appears that there two perl versions given for perl in in yocto 1.7 (perl 5.20.0 and 5.14.2) Aug 24 09:22:54 Which one is correct? Aug 24 09:23:10 Or am I hallucinating somehow (not enough coffe?) Aug 24 09:23:21 jeremiah: where are you seeing these two versions? Aug 24 09:23:30 When I do perl -V Aug 24 09:23:54 hmm, well, supposedly we are building 5.20 Aug 24 09:24:19 That's what it looks like when I do perl -v Aug 24 09:24:24 But when I do perl -V; Aug 24 09:24:28 Summary of my perl5 (revision 5 version 14 subversion 2) configuration: Aug 24 09:25:49 Its not the end of the world, I'm just trying to understand the delta between default perl version and oe version Aug 24 09:25:56 Since I'm looking for various modules Aug 24 09:30:42 ah, I think I see how that is happening Aug 24 09:32:41 :) Aug 24 09:32:52 we supply a config.sh file into perl when we build it, that has some pre-generated values in it that clearly haven't been updated Aug 24 09:33:05 I thought it might be something like that. Aug 24 09:33:16 Since it looks like you really do have a perl 5.20 Aug 24 09:33:41 it got fixed in master apparently, but the same issue is present in fido (1.8) Aug 24 09:34:12 Okay. Then I guess all is well and I shall go my merry way, no need for a patch. Aug 24 09:34:29 No point in patching an older 1.7 I suppose? Aug 24 09:35:06 I think we might fix this in 1.7 and 1.8 because it is a little ugly Aug 24 09:35:26 We're meant to be moving to 1.8 soon so that would be mighty helpful if it were updated. Aug 24 09:36:00 I'm writing some tests based on TAP and perl has some modules I need so an accurate version helps for my particular use case. Aug 24 09:36:02 I suspect what would also help would be a test which ensures the version has been properly updated on upgrade Aug 24 09:36:22 Oh, that would be very useful yes. Aug 24 16:55:17 has anyone had an issue with opkg not running postinst scripts correctly? I've been getting inconsistent results where applying an upgrade will upgrade a package and run the preinst script which removes symlinks to an initscript but never runs the postinst script to create the symlinks to the initscript. Aug 24 17:12:46 JaMa: yt ? Aug 24 17:13:00 qtbase is taking forever to fetch Aug 24 17:13:32 Can we use github mirrors instead of code.qt.io Aug 24 17:15:17 is it enough to re-define: conf/layer.conf:QT_GIT ?= "git://code.qt.io" Aug 24 17:15:36 or do you need to add some params to SRC_URI? Aug 24 17:16:17 I don't see long fetch time, but I also update them quite often so my clone was really close Aug 24 17:16:24 https://github.com/qtproject Aug 24 17:16:46 I think in general it will be good to use github Aug 24 17:17:04 just for its infra support worldwide Aug 24 17:18:04 I don't mind changing it to github.com, please send patch Aug 24 17:18:45 Let me see if it helps me out Aug 24 17:18:52 and I will send it along with my other rebased patchset Aug 24 17:20:46 agreed re teh github infrastructure, good performance Aug 24 17:45:03 github is far better all went quickly after I switched to it Aug 24 17:45:11 sent out my v2 series Aug 24 17:46:06 btw. if any clang enthusiasts here .. I have it going well now, I could build a core-image-minimal for arm and x86 and boot it Aug 24 17:46:17 ofcourse not all packages were compiled with clang Aug 24 17:46:28 but approximately 75% were Aug 24 17:46:38 I have got qt images going too Aug 24 17:46:45 but they dont yet boot so well Aug 24 17:47:02 I guess I am pushing the envelop too far by using thumb2 ISA Aug 24 17:47:33 but if anyone is interested to help out then clone meta-clang and it should default to clang now automatically Aug 24 17:47:46 so you can start sending patches to fix various recipes Aug 24 17:48:07 see how good the diagnostics are :) Aug 24 17:48:35 Errors/warnings/notes/ and colors with underlines where the issue really is Aug 24 17:48:47 its a good experience Aug 24 17:49:00 I must do a detailed write up Aug 24 17:49:15 you can also create SDKs with both gcc and clang in it Aug 24 17:49:16 now Aug 24 17:49:38 also compiler-rt and libc++ are working now for arm and x86 Aug 24 17:50:32 one thing I noticed that the compile time is significantly better I could see 30% or more reduction in qrwebkit compile Aug 24 17:58:18 khem` I'll have to find some time to try it **** ENDING LOGGING AT Tue Aug 25 02:59:58 2015