**** BEGIN LOGGING AT Thu Apr 21 02:59:58 2016 Apr 21 05:55:58 anyone seen: #error "LONG_BIT definition appears wrong for platform (bad gcc/glibc config?)." before? This seems to happen when a project uses python pyport.h with a 64bit host and 32bit target Apr 21 07:01:34 Hi, is it possible to have something like do_populate_sdk_prepend? Apr 21 07:01:42 so _append / _prepend to tasks? Apr 21 07:17:25 addtask my_task before do_populate_sdk Apr 21 07:17:34 http://elinux.org/Bitbake_Cheat_Sheet Apr 21 07:17:59 yes, you can append/prepend to tasks — there are several do_configure_append and do_install_append in OE-Core Apr 21 07:18:57 thanks! Apr 21 08:07:08 dependency question: recipe A depends on B. B depends on B-native. Should A depend on B-native as well if it needs a tool during build or should the above just work? Apr 21 08:07:33 Practical example is matchbox-panel-2 that fails to build after wipe-sysroot because dbus-binding-tool is not there (matchbox-panel-2 depends on dbus-glib, which depends dbus-glib-native) Apr 21 08:39:21 I'd say that if your recipe requires a native tool at build time, you need to add it to DEPENDS Apr 21 09:16:53 I've asked about the ADT plugin before, but not gotten any response. Is it dead? There's a build for Eclipse Mars so I'd hardly think so. Apr 21 09:26:25 Ok, I _finally_ found this: http://comments.gmane.org/gmane.linux.embedded.yocto.general/27698 Apr 21 09:41:47 For the [meta-qt5] guys: https://bugreports.qt.io/browse/QTBUG-52739 Apr 21 09:44:51 In case you know the proper solution or can find it the Qt guys. Apr 21 09:44:58 *with Apr 21 10:19:46 I'm using Yocto 2.0 and it seems to be putting kernel sources to work-shared directory which is okay if you build just one platform. I have some platform depended patches that'll be applied to kernel so how can I remove/disable do_shared_workdir which I think causes that or is there otherway to disable that functionality? Apr 21 10:20:25 on Yocto 1.7 the kernel sources where in work - dir so there was no problem Apr 21 10:33:30 is there some handy way to display all the vardeps (explicitly declared and inferred) for a function? Apr 21 10:36:18 JaMa, j'ai tenté de clarifier un peu mieux le souci. J'ai ouvert des bugs report côté Qt. https://bugreports.qt.io/browse/QTBUG-52739 Apr 21 10:36:22 Oopppp Apr 21 10:36:48 JaMa, bug opened related to https://bugreports.qt.io/browse/QTBUG-52739 EGL opended on Qt side. Apr 21 10:37:24 JaMa, if you can have a look... I'm pretty sure it's a Qt mainstream bug now. Apr 21 11:08:44 buildhistory: are the generated RDEPENDS listed there too? Apr 21 11:16:13 ah buildhistory has both the source/recipe and binary 'latest' files, and generated RDEPENDS is only in the binary one. Apr 21 11:17:10 I'm seeing a race condition or sstate cache dependant automatic detection of libssp RDEPENDS, and libssp is provided by gcc runtime. weird. Apr 21 11:17:34 this on ~week old jethro based tree Apr 21 12:08:06 Hey where can I get a list of the packages I can add to my Yocto build ? Apr 21 12:08:21 EXtra-iamge- features Apr 21 12:42:06 hi there, is anyone already succed to build core-image-minimal on jetho with isafw? Bitbake claiming for python-lxml-native and I don't know what to do now. Apr 21 12:44:29 kapt: looks like you need to add meta-python Apr 21 12:45:50 rburton, ok I will retry but I think I will have the "unrecognized command line option '--should-not-have-used-/usr/bin/xml2-config'" error Apr 21 13:09:41 EXtra-iamge- features Apr 21 13:09:42 Hey where can I get a list of the packages I can add to my Yocto build ? Apr 21 13:12:51 find meta -name *.bb Apr 21 13:13:16 well, layers.openembedded.org will let you browse the available layers and recipes Apr 21 13:13:24 so you can see what is out there instead of what you have on disk now Apr 21 13:55:52 folks is there anyway to put the source of package on the target, to make debugging easier? Apr 21 13:56:48 the -dbg packages help a lot Apr 21 13:56:55 not sure if they have complete source Apr 21 13:57:03 Think that is symbols only Apr 21 13:57:16 Putting gdb on there as well Apr 21 13:57:22 tools-debug ... etc Apr 21 14:03:13 they have sources in for gdb to read Apr 21 14:07:09 for debugging -dbg has everything GDB needs.. but it's not a full copy of the source, only the sources that the dwarf symbols reference Apr 21 14:12:48 I have a really basic question. When doing GUI development, where you will be booting straight into your GUI with soemthing such as Qt, is there any reason for any terminal from window management systems such as matchbox? Has anyone seen any case where it might be necessary? Apr 21 14:23:54 riz__, lol we do the same thing :P Apr 21 14:24:05 riz__, No you don't want X system. Apr 21 14:24:46 present: So, X system is only required if the application depends on it, correct? Apr 21 14:24:58 Other than that it is just extra weight I am assuming Apr 21 14:25:14 Because Qt can write straight to the framebuffer from what I have read Apr 21 14:25:43 riz__, EGLFS Apr 21 14:26:29 present: That is where I get confused. Where does EGLFS come into play in regards toeverything? Apr 21 14:26:46 Does that talk to the fb? Apr 21 14:27:21 http://doc.qt.io/qt-5/embedded-linux.html#eglfs Apr 21 14:27:56 riz__, I'm not all the one who can answer this question. Apr 21 14:28:13 haha I was reading that as you sent it Apr 21 14:28:28 present: So you have completely left out x11 in your image? Apr 21 14:28:33 yep Apr 21 14:28:43 No wayland or anything? Apr 21 14:28:47 nope Apr 21 14:29:04 Have you found any hurdles or disadvantages in Qt by doing this? Apr 21 14:29:05 As long as I remember no wayland neither. Apr 21 14:29:18 Or do you have full functionality? Apr 21 14:29:32 full I'ld say Apr 21 14:30:04 OK. I guess using Qt5_CinematicExperience as a starting point hurt me then Apr 21 14:30:15 That depended on x11 and I had no choice but to add it Apr 21 14:30:22 By the way Qt5.7 is getting to be only commercial related to Qt Quick Compiler... Apr 21 14:30:42 Just read it Apr 21 14:30:54 http://blog.qt.io/blog/2016/04/21/qt-5-7-beta-released/ Apr 21 14:31:25 riz__, I'm out also for this question. Apr 21 14:31:54 I use to have some troubles with a virtualkeyboard but it works on EGLFS too in the end. Apr 21 14:32:45 Did I understand that correctly, they will provide some commercial features under the open licenses for 5.7? Apr 21 14:33:18 I just gave the link. I understand that QML will be commercial only for th 5.7. Apr 21 14:33:41 Oh. That isnt good Apr 21 14:33:51 May be that's not the good interpretation. Apr 21 14:34:49 Oh wait Apr 21 14:34:50 I don't get much how they want to close open source and commercial related to the compiler. Apr 21 14:34:53 But it is only for 5.7 Apr 21 14:35:05 present: 5.8 it will go back to opensource Apr 21 14:35:10 yes and then in 5.8 a kind of lighter version Apr 21 14:35:16 of the compiler Apr 21 14:35:33 I see Apr 21 14:36:10 At some point they need to make money. Apr 21 14:36:28 I don't know what is the price of the commercial version. Apr 21 14:36:59 That is true Apr 21 14:54:08 rburton, ok you were right, everything is fine with meta-python. Thx Apr 21 16:11:38 Does it make any sort of difference if I place my build directory within my poky source directory? Apr 21 16:16:53 not much Apr 21 16:17:10 just take care that you are in a git repository Apr 21 16:17:29 but build* is already ignored Apr 21 16:17:35 so no worries Apr 21 16:17:56 I'll say that the most important is to change that variable: Apr 21 16:18:12 DL_DIR ?= "${TOPDIR}/../downloads" Apr 21 16:18:38 You will keep everything you have downloaded while erasing the build directory. Apr 21 16:18:49 It will save you hours. Apr 21 16:19:45 Just read all the comments in ./conf/local.conf Apr 21 16:21:36 OK cool Apr 21 16:21:38 Thanks Apr 21 17:56:36 good job Atlassian, editing simple wiki page which shows some difference in version of installed packages (few hundred lines) is for my 8 core machine as difficult as building those images Apr 21 17:58:10 confluence? Apr 21 17:58:12 cpo: there are also post-functions for do_populate_sdk, see http://git.yoctoproject.org/cgit/cgit.cgi/poky/tree/meta/classes/populate_sdk_base.bbclass#n85 and following Apr 21 18:36:22 hmm, plymouth in meta-oe deps on dracut (though why i have no idea ,ocnsidering you don't need dracut to use plymouth) and dracut is in meta-initramfs, so that's an implict dep from meta-oe to meta-initramfs.. hmm Apr 21 18:53:52 fledermaus: yes confluence Apr 21 19:19:17 hello Apr 21 19:19:43 I've heard there is a webpage where you can find recipes for capabilities Apr 21 19:20:30 I need to add iptables or netfilter or whatever its called now so I can forward packets from one interface to another on my yocto board Apr 21 19:21:25 there is this, http://git.yoctoproject.org/cgit.cgi/poky/plain/meta/recipes-extended/iptables/?h=bernard Apr 21 19:21:48 and that name looks familiar, ie. bernard Apr 21 19:21:59 so perhaps that is a legit source Apr 21 19:23:03 davis: you wouldn't want to be looking at the bernard branch, that's ancient Apr 21 19:23:12 are you looking for http://layers.openembedded.org/layerindex/branch/master/recipes/?q=iptables Apr 21 19:23:55 heh, i thought bernard was someones name I met either here or at elc. Apr 21 19:24:16 that link neverpanic sent looks like what I heard people showing me once Apr 21 19:25:26 so once i save off my existing setup, I could add that recipes-extended/iptables/iptables_1.6.0.bb to my poky/jetrho build? Apr 21 19:25:46 jethro already comes with a copy of iptables Apr 21 19:26:09 hmm. Apr 21 19:26:40 (and the netfilter kernel module that you need for iptables to work) Apr 21 19:28:25 yeah, i can add kernel config mods ok now. Apr 21 19:29:00 i guess i need to add a similar layer to my kernel config layer and in it, add a recipe for iptables Apr 21 19:32:28 I am building my image without x11, but gstreamer-vaapi-1.0-dev because there is a dependency on x11 (libva-x11). Has anyone run into this? Apr 21 19:38:52 davis: you don't need to add the recipe, just add the package for it to your image Apr 21 19:38:59 the recipe is already there Apr 21 19:39:15 find meta -name iptables*.bb and you will see it Apr 21 19:47:02 riz__: seems like I saw a discussion on the mailing list about fixing (i.e., breaking) that dependency, but I could be wrong. Have to go back and look. Apr 21 19:47:56 neverpanic: thanks, I already ended up with a postfunction Apr 21 19:48:20 bluelightning: ok many thanks. I've got to do some ios stuff now, but i've added your help to my notes. I'll work on it tomorrow. my rsync to save off my yocto work is still going. Apr 21 19:50:05 billr: I saw one about libva-egl depending on libva-x11 for some reason, but did not see a fix Apr 21 19:50:18 I will keep searching the mailing list Apr 21 19:50:25 huh, with only minor tweaks bitbake runs fine under pypy Apr 21 19:50:26 riz__: maybe that's what I was remembering Apr 21 20:03:08 I've added a package to my core image's IMAGE_INSTALL. I've got a file from that package which is staged in my target MACHINE's sysroot. That file doesn't appear on my core image's rootfs. What's going on? Apr 21 20:05:05 zeddii: currently trying to figure out why a config fragment added to SRC_URI isn't being applied to the kernel config. it's definitely fetched and unpacked, but doesn't seem to be applied. Apr 21 20:05:52 zeddii_home: ^ I'm also seeing that the do_kernel_configcheck log is blank, is that expected? Apr 21 20:13:16 not expected,but I was just fixing some things in there. is this something I can try myself ? Apr 21 20:13:40 one sec, let me email you the fragment Apr 21 20:16:21 done Apr 21 20:17:00 karobar, use IMAGE_INSTALL_append and add it a space, such as XXX_append = " mypackage" Apr 21 20:17:27 bluelightning, I'll add that now and launch a config. Apr 21 20:17:33 zeddii: superb, thanks Apr 21 20:17:45 interesting. under pypy, there are a ton of taskhash mismatches that don't occur under cpython Apr 21 20:17:56 looks like they're all do_package Apr 21 20:18:00 hmm Apr 21 20:31:07 Hey, Apr 21 20:31:25 I am seeing that libva_1.5.0 depends on libva-x11 Apr 21 20:31:42 sorry, libva-egl depends on libva-x11 Apr 21 20:32:04 How does that make sense? Shouldnt you be able to use libva-egl without x11?? Apr 21 20:33:44 Hmm, does runqemu only support either initrd *or* a real rootfs, but not both? I'd think both would be useful to test out a real world initramfs Apr 21 20:34:21 guess you could manually set it up in qemuparams Apr 21 20:50:16 ulf`: thanks for the suggestion, unfortunately i have the same result Apr 21 21:04:36 * ulf` signs off for 1 hour to switch to home office Apr 21 21:35:11 Hi, there is a problem with the yocto git? I can not connect to fido repo Apr 21 21:36:58 viktorivan: works for me Apr 21 21:37:55 viktorivan: I just verified: I can connect to the fido repo. Apr 21 21:38:30 helps if you say what the problem is Apr 21 21:38:38 anyway, g'night Apr 21 21:41:05 oks, thanks... Apr 21 21:53:44 * kergoth chuckles at https://www.sunfrog.com/No-Category/Senior-Software-Developer-3226-RoyalBlue-Guys.html?9654 Apr 21 22:04:08 If I want to not install the following package, how do I do that: gstreamer-vaapi-1.0-dev Apr 21 22:04:51 IMAGE_INSTALL_remove? Apr 21 22:05:24 that won't help if it's pulled in via a dependency Apr 21 22:05:55 WEll I have a core issue where gstreamer-vaapi-1.0-dev is dependent on libva-x11, but I removed x11 Apr 21 22:06:18 kergoth: I dont think there should be that dependency but I dont know how to fix it Apr 21 22:08:14 when I use bitbake -u depexp it doesnt show that dependency wit gstreamer-vaapi-1.0-dev so I am not sure where the dependency comes from. What I did find is that libva-egl depends on libva-x11 and that could be pulling it in Apr 21 22:16:50 riz__: if you're using meta-intel, it looks like it goes into MACHINE_HWCODECS, which does into FEATURE_PACKAGES_hwcodecs, which is installed in core-image through IMAGE_FEATURES Apr 21 22:17:15 MACHINE_HWCODECS_remove might work but I haven't tried it Apr 21 22:18:06 So I just put "MACHINE_HWCODECS_remove = "libva-x11: in local.conf? Apr 21 22:18:27 clsulliv: and that will hopefully remove the depndency? Apr 21 22:18:47 riz__: gstreamer-vaapi-1.0 Apr 21 22:19:04 Oh ok Apr 21 22:20:22 clsulliv: What if I just remove hwcodecs completely from my IMAGE_FEATURES? Apr 21 22:20:42 I have a custom image and I put it there but I have no idea why I did to be honest Apr 21 22:20:48 riz__: that would probably work too Apr 21 22:20:54 OK. THanks Apr 21 22:20:56 Off to try Apr 21 22:21:51 you also lose va-intel. Not sure what that does but genericx86 works on intel boards so you should be OK Apr 21 22:45:55 zeddii, what do you mean by dropped? Are you saying all kernel options in mismatch.txt stating "Value requested for" need to be there for the fragment to be merged? **** ENDING LOGGING AT Fri Apr 22 02:59:58 2016