**** BEGIN LOGGING AT Thu Apr 20 03:00:03 2017 Apr 20 06:30:39 I updated from jethro to morty, and it seems like I lost a bunch of OpenSSL symbol versions: https://github.com/flatpak/freedesktop-sdk-images/issues/25 Apr 20 06:30:45 Is this a known issue? Apr 20 06:33:01 I am creating an image intel 64 bit then I will have a .wic file. How can I play my new linux 64 bit intel in virtual box or vm player ? Apr 20 06:51:31 hamdy: what machine and image did you build? Apr 20 06:54:04 Core image minimal, x86-64_linux Apr 20 06:54:43 hamdy: and you did get *only* a wic image? Apr 20 06:56:32 The image is not finished now but the result will be a .wic image like they say in the documentation Apr 20 06:56:55 hamdy: which documentation, exactly? Apr 20 06:58:07 i mean, a cmdk or such is never generated by default. but you should be able to use runqemu to play with your image just nicely Apr 20 06:59:24 I did an image for qemu and it works. This is the yocto documentation. I want to open it in virtualbox Apr 20 07:01:09 hamdy: i personally would give tlwoerners old documentation a try: http://twoerner.blogspot.de/2013/12/using-oeyocto-to-create-vms.html Apr 20 07:01:44 ok I will look Apr 20 07:12:11 I added IMAGE_FSTYPES += vmdk to the conf/local.conf and now I have a parser error Apr 20 07:14:17 hamdy: its just missing the quotes. here it is a little more exhaustively explained, jsut ignore the genivi bits: http://gmacario.github.io/howto/gdp/genivi/yocto/virtualbox/embedded/ivi/2015/11/14/running-yocto-image-inside-virtualbox.html Apr 20 07:17:37 ok I do it Apr 20 07:18:40 armpit: Are you akuster808@gmail.com ? Apr 20 07:20:36 mivond: I'm interested in your build but I'd like to proceed one step at a time. What layers are you using ? meta-raspberrypi and meta-mender ? I guess you're using uboot as you told me before, isn't it ? How did you set MACHINE var and where you've set the uboot feature ? Apr 20 07:21:19 mivond: what RPi boards did you tested with it ? Apr 20 07:21:31 mivond: BTW, excellent work ! Apr 20 07:53:59 armpit: I'm looking at the openssl version script change you did in the 1.0.2g import. Apr 20 07:54:11 armpit: and it seems to have issues: https://github.com/flatpak/freedesktop-sdk-images/issues/25#issuecomment-295616999 Apr 20 07:57:18 * RP fires the M4 rc1 build Apr 20 08:05:35 RP: normal fire or total fire? ;-) Apr 20 08:09:38 LetoThe2nd: firing builds is so much more satisfying that starting them ;-) Apr 20 08:10:42 RP: exactly. thats why everybody knows demolition man, and nobody a construction man ;-) Apr 20 08:17:07 hi! Apr 20 08:17:19 the current poky "morty" branch doesn't work Apr 20 08:17:24 I get an error at compilation Apr 20 08:17:31 "ERROR: logrotate-3.9.1-r0 do_fetch: Checksum failure fetching https://fedorahosted.org/releases/l/o/logrotate/logrotate-3.9.1.tar.gz" Apr 20 08:17:39 "ERROR: logrotate-3.9.1-r0 do_fetch: Fetcher failure for URL: 'https://fedorahosted.org/releases/l/o/logrotate/logrotate-3.9.1.tar.gz'. Checksum mismatch!" Apr 20 08:17:49 "ERROR: Task (poky/meta/recipes-extended/logrotate/logrotate_3.9.1.bb:do_fetch) failed with exit code '1'" Apr 20 08:29:21 It looks like that website was retired Apr 20 08:29:22 https://fedoraproject.org/wiki/Infrastructure/Fedorahosted-retirement Apr 20 08:30:55 I wonder if this is the same https://github.com/logrotate/logrotate/archive/r3-9-1.tar.gz Apr 20 08:34:34 I guess so Apr 20 08:36:10 Hey all. Are this channel still actiev? Apr 20 08:36:40 MichaelLarsen: yes Apr 20 08:36:54 only if you bring cookies and coffee! Apr 20 08:37:04 oh yes, coffee Apr 20 08:37:08 mines a flat white Apr 20 08:37:57 :-( Apr 20 08:39:42 rburton: sounds great. Apr 20 08:40:04 are there any of you guys that have any knowledge on ROS Apr 20 08:40:05 ? Apr 20 08:40:18 combined with embedded linux Apr 20 08:40:26 didil: armpit mentioned earlier this patch: https://patchwork.openembedded.org/patch/139233/ Apr 20 08:41:10 MichaelLarsen: depsite the question being extremely meta, there is https://github.com/bmwcarit/meta-ros Apr 20 08:41:59 LetoThe2nd: I actually stumbled upon that repository in my search. Is this the "most" active embedded projecT? Apr 20 08:43:04 MichaelLarsen: we have little to no repots concerning ROS here. Apr 20 08:43:27 radanter: thanks :) Apr 20 08:49:33 LetoThe2nd: ok. Apr 20 09:26:39 Hi, I am trying to build some image and the build of mtd-utils (from poky) fails - I get the following note in the beginning: mtd-utils-1.5.2-r0 do_compile: NOTE: mtd-utils: compiling from external source tree /tmp/devtoolqasb5c9lst/source - could someone explain to me why this happens? Where is that defined? I did not change anything in the recipe Apr 20 10:03:07 There is no EXTERNALSRC set on purpose, how can I figure out where this is set? I am still unsure whether this is even the reason Apr 20 10:58:44 m4ho: from the path name I would guess you've used devtool to work on that at some point and it's still in the workspace ... "devtool reset " should remove it from the devtool workspace (you probably lose the changes you may have made) Apr 20 11:07:53 jku: thanks a lot, devtool status shows exactly that, I am sure never to have used it. However, somehow it must have happened. Now I know what that looks like, I never actually used devtool until now Apr 20 11:10:45 well that's strange Apr 20 11:11:25 'strange' describes my experience with yocto quite well ;) Apr 20 11:44:27 I created an intel minimal image then I opened the .iso in virtualbox. Now it's read only. What can i do ? Apr 20 11:47:00 Hi everyone, there is a way to find out which recipe in which directory provide the recipe which install a package ? I would like to find out where does the qt plugins are installed. It's not in the build/conf neither in the recipe of the image.... Thanks. Apr 20 11:52:28 Hi ChrysD. You can find that in buildhistory. Apr 20 11:53:04 mario-goulart : I'm not familiar with it. Can you explain more or it's information easy to get ? Apr 20 11:53:14 mario-goulart : Thanks by the way. Apr 20 11:53:31 ChrysD: sure. Are you familiar with buildhistory? Apr 20 11:53:59 mario-goulart : No i'm not Apr 20 11:54:35 ChrysD: ok. Here's some documentation about it: http://www.yoctoproject.org/docs/current/ref-manual/ref-manual.html#enabling-and-disabling-build-history Apr 20 11:54:54 ChrysD: what yocto version are you using? Apr 20 11:54:56 mario-goulart : thank you ! Apr 20 11:55:01 mario-goulart : Krogoth Apr 20 11:57:34 I created an intel minimal image then I opened the .iso in virtualbox. Now it's read only. What can i do ? Apr 20 11:57:48 Ok. Unfortunately, krogoth doesn't have the patch that will add information that maps recipes to layers in buildhistory. Apr 20 11:57:54 ChrysD: ^ Apr 20 11:58:25 mario-goulart : Thank you some much ! It's better to know that something doesn't exist instead of looking at it for a while ahahah Apr 20 11:58:36 ChrysD: d8e59d1f840e4282859ad14397d1c06516b8eb11 could probably be cherry picked, though. Apr 20 11:59:06 mario-goulart : So i still need to dig into files to know where the (IMAGE_INSTALL=" qtplugins" ) is written x) Apr 20 11:59:19 ChrysD: but keep in mind that .bbappend files may influence on packaging (and other things). So, only knowing the base recipe sometimes is not enough. Apr 20 12:00:10 ChrysD: you want to know what recipe drags a recipe into an image? Apr 20 12:00:22 s/recipe/package/ Apr 20 12:01:03 ChrysD: rephrasing to avoid confusion: you want to know what recipe drags a package into an image? Apr 20 12:03:46 mario-goulart : I want to know in which recipe is written that "those" packages will be installed in the image Apr 20 12:04:15 mario-goulart : for exampel i would like to add at qt plugins, and i would like to add this qt plugins by a patch and a .bbappend. But i would like to add it in the list of the others. Apr 20 12:04:36 mario-goulart : Have checked my build/conf and the recipe-images/ already Apr 20 12:05:31 mario-goulart : add a qt plugin * Apr 20 12:08:48 ChrysD: not sure if I understand your problem, but here's some info. Recipes specify what they depend on and what packages they generate. Images are described by recipes. Image recipes usually only specify the packages that they ultimately want (not the whole chain of dependencies). So, to get what [I think] you want, you have to take a look at the dependency graph and find out which package in the image depend on the qt plugins. Apr 20 12:08:48 You can generate it with bitbake -g. Apr 20 12:09:45 rburton: Did you see this: https://bugzilla.yoctoproject.org/show_bug.cgi?id=11396 its somewhat painful for the flatpak runtime... Apr 20 12:09:46 Bug 11396: normal, Undecided, ---, richard.purdie, NEW , OpenSSL soname "wrong" Apr 20 12:09:58 mario-goulart : For example, the simple fact of putting a "RDEPENDS " can install a package in the iamge ? Apr 20 12:10:57 mario-goulart : And if I would like to add a package, i should do a new image recipe or i should patch the image recipe which have been given ? Apr 20 12:11:36 ChrysD: it depends on where you put RDEPENDS. :-) It must be specified for a package that is installed in the image. Apr 20 12:12:05 It all depends on how your project is structured. Bot alternatives should work. Apr 20 12:12:45 Both* Apr 20 12:12:55 mario-goulart : So in my case, if i don't find where does the qtplugins are installed, could be possible that the fact of putting a RDEPENDS packagroup of qt plugins make the qt plugins installed right ? Apr 20 12:12:59 I created an intel minimal image then I opened the .iso in virtualbox. Now it's read only. What can i do ? Can someone help me please ? Apr 20 12:16:18 alexlarsson: eww. Somehow this always happens with openssl at release time Apr 20 12:16:28 <_arthur_> Are there know issues with IMAGE_DEPENDS_foo ? I'm trying to add a new image type, and I need a custom tool to be installed (presumably in build/tmp/sysroots/x86_64-linux), but the dependency doesn't seem to make it somehow. Apr 20 12:16:33 something like this anyway Apr 20 12:17:18 jku: i remember openssl ABI being a shitshow 10 years ago Apr 20 12:17:31 how come its still a problem Apr 20 12:17:36 amazing Apr 20 12:18:36 alexlarsson: urgh flaming death to openssl Apr 20 12:18:40 smcv has some critique for us in the end of that good post that probably does apply: I'm not sure we have anyone who pays _as much_ attention as one apparently would need to keep openssl running well Apr 20 12:18:49 https://github.com/flatpak/freedesktop-sdk-images/issues/25 Apr 20 12:19:42 agreed Apr 20 12:20:21 "copy exactly what debian does" is a sane policy if you manage to keep the 'exactly' bit Apr 20 12:22:24 grabbing the patch from debian to bump soname seems like the best solution here Apr 20 12:22:25 <_arthur_> Let me rephrase the question: How do I ensure tool is available while an image is generated (e.g. creating a ubifs image requires mkfs.ubifs). Apr 20 12:22:34 especially as we just rolled 2.3 RC1 *this morning* Apr 20 12:23:16 rburton: its what i did: https://github.com/flatpak/freedesktop-sdk-base/commit/cff4acfd64da088bfe3dbc0469a7fdb3d9432c17 Apr 20 12:23:19 jku: got time to take this? i should eat something shortly. Apr 20 12:23:37 rburton: seems to work Apr 20 12:23:57 alexlarsson: thanks. cursed openssl. Apr 20 12:24:28 time to write a class to run libabigail over binary output for ABI review Apr 20 12:24:30 rburton: I can send a patch Apr 20 12:24:51 thanks jku Apr 20 12:26:46 rburton: maybe things will get better with 1.1 :) Apr 20 12:26:59 supposedly yes Apr 20 12:27:10 kanavin: not sure if thats a serious expectation or irony Apr 20 12:27:14 :) Apr 20 12:27:57 rburton: for a start, I've thrown away the old recipe, and all of its patches Apr 20 12:28:11 rburton: I'm not sure myself :) Apr 20 12:37:39 rburton, what exactly does "copy what debian does" sound sane for? Apr 20 12:39:26 kanavin: the only good news with openssl 1.1 is that for the first time, everyone agrees what the soname is Apr 20 12:39:57 I had some help to convince Fedora to switch to using the Debian convention, even though it's the wrong convention for openssl anyway Apr 20 12:40:08 because I figured no one would switch to the Fedora convention anyway Apr 20 13:19:02 <_arthur_> I'm trying to understand how build/tmp/sysroots/x86_64-linux/usr/sbin/mkfs.ubifs came to be installed there. I (think I) need to accomplish the same for my tool, but simply adding it to IMAGE_DEPENDS_foo doesn't seem to be working. Apr 20 13:19:35 <_arthur_> I see IMAGE_DEPENDS_ubifs = "mtd-utils-native", but doing the same for my tool doesn't work. What could I be missing? Apr 20 13:26:57 _arthur_: have you checked if the tool you need ends up installed properly in native sysroot ? first somewhere like tmp/sysroots-component/x86-64/ and then I guess in the image $WORKDIR/recipe-sysroot-native Apr 20 13:28:15 <_arthur_> It's not installed, I believe. It is listed in tmp/sysroots/x86_64-linux/sysroot-providers/ along with mtd-utils-native Apr 20 13:29:40 _arthur_: oh sorry, which yocto version? Apr 20 13:32:50 <_arthur_> 1.8.2 Apr 20 13:33:13 <_arthur_> fido, iirc. Apr 20 13:33:49 ah yes, my advice was for much newer sysroot locations. Sorry Apr 20 13:34:23 <_arthur_> No worries. I'm aware that we're on older code, but ... embedded, you know how it is .... Apr 20 13:55:17 chbae: hi Apr 20 13:55:27 err.. wrong channel :) Apr 20 14:19:17 <_arthur_> Ah, I think the problem is not with the dependency, but with the package. sstate-cache/Debian-8.5/7e/sstate:mtd-utils-native:x86_64-linux:1.5.1+git... contains the binaries, but my package does not. Looks like some confusion between ${D} and ${sbindir}. Apr 20 14:39:13 I have added IMAGE_FSTYPES += "vmdk" to my local.conf but I still don't have a .vmdk image for virtualbox. I Followed this tutorial : http://gmacario.github.io/howto/gdp/genivi/yocto/virtualbox/embedded/ivi/2015/11/14/running-yocto-image-inside-virtualbox.html Can someone help me please ? Apr 20 14:57:59 now my vmdk image works. How can i resize a yocto partition ? Apr 20 15:19:00 Someone can tell me how to resize my yocto image ? Apr 20 18:17:44 hello there. i have problems to get ffmpeg with ffplay (needs sdl) compiled. i bitbaked libsdl and ffmpeg can't find the headers and libs. both are recipes of Apr 20 18:17:46 meta Apr 20 18:18:34 how can i direct the ffmpeg recipe in the right direction? Apr 20 18:38:07 * tx0h sighs Apr 20 18:42:17 to be precise: it didn't find the sdl stuff during the do_configure Apr 20 18:43:36 <_arthur_> Does your ffmpeg recipe depend on libsdl? Apr 20 18:43:44 yes Apr 20 18:43:50 it does Apr 20 18:44:51 DEPENDS = "alsa-lib zlib libogg yasm-native libsdl" Apr 20 18:45:15 and sdl files are in the sysroots Apr 20 18:46:02 <_arthur_> Okay, so presumably the configure just needs to be pointed at it somehow. Is this an autoconf recipe? Apr 20 18:47:20 it's the recipe which is in poky/meta/recipes-multimedia/ffmpeg Apr 20 18:47:42 <_arthur_> Yeah I don't have that one around ;) Apr 20 18:49:09 <_arthur_> It might be worth taking a look at how the ffmpeg configure looks for libsdl. Apr 20 18:49:12 oh, ok. Apr 20 18:50:52 thats the verdict: check_pkg_config sdl SDL_events.h SDL_PollEvent Apr 20 18:51:18 where check_pkg_config is a function in the ffmpeg configure script Apr 20 18:53:04 <_arthur_> Did your libsdl recipe install the .pc file? Apr 20 18:53:19 also in the sysroots? Apr 20 18:53:39 <_arthur_> I think so, yes. Apr 20 18:56:02 yes, there is a file 'sdl.pc' in the pkgconfigs Apr 20 18:57:26 <_arthur_> Do you get error output from the configure step? Apr 20 18:57:36 yes Apr 20 18:57:40 <_arthur_> I'm looking at the latest version of ffmpeg/configure and it looks for sdl2, not sdl. Apr 20 18:57:46 <_arthur_> What does it say? Apr 20 18:57:50 a ffmpeg config.log and the bitbake errors Apr 20 18:58:25 aha?! i have both. let me change it to libsdl2 Apr 20 18:59:21 <_arthur_> Well hang on, that might not be the problem ;) Apr 20 18:59:37 <_arthur_> Look at the bitbake errors and the config.log to figure out what's going on first. Apr 20 19:01:57 SDL_events.h is definitively in both packages. and thats what it looks for (at first) and i think, this can't be found Apr 20 19:04:00 <_arthur_> It might be worth sticking in a debug patch to log what the configure script sees. Check the pkg-config output, and check_func_headers output. It seems to compile a test app so that might fail. Apr 20 19:07:33 oh, what does this mean? arm-rk-linux-gnueabi-gcc: error: /tmp/ffconf.loKPYZ9o.m: Objective-C compiler not installed on this system Apr 20 19:08:02 thats for the QTKit test in ffmpeg Apr 20 19:08:14 maybe thats the issue?! Apr 20 19:08:45 <_arthur_> Yeah that looks likelu. Apr 20 19:08:58 how do i enable objective-c so? Apr 20 19:11:09 <_arthur_> Good question. I don't know. Apr 20 19:11:15 i thought this should be standard in the gcc.. Apr 20 19:16:12 * tx0h looks puzzled Apr 20 19:31:44 source is in build/tmp/work-shared/gcc-6.2.0-r0/gcc-6.2.0/libobjc but i can't find libobc* Apr 20 19:32:21 it seems it's not made Apr 20 19:43:34 Is there a way to use do_split_packages to select directories and their contents to become a new package? With 'allow_dirs=True', it appears that only the directories (but not the files in them) are included in the packages. Apr 20 19:44:34 Or do I need to just do what do_split_packages does internally myself? Apr 20 19:51:45 jmesmon: i'd be surprised if it can't do what you want :/ Apr 20 19:52:09 rburton: I'm trying out using aux_files_pattern. Might work Apr 20 19:58:03 added objc to LANGUAGES in gcc-configure-common.inc. lets see how it works.. Apr 20 20:05:17 ah, turns out I was just missing 'prepend=True'. No need for messing with aux_file_pattern, just need to have the dynamic packages be packaged before PN so it doesn't hoover up the dyn-package files. Apr 20 20:21:32 * tx0h recompiles the whole setup :-( Apr 20 22:54:33 tlwoerner: I seem to recall someone mentioning not that long ago that devtool had stopped picking up the name from a source tree correctly, but I can't find the message in the mailing list archives - was that you by any chance? Apr 20 22:55:08 * tlwoerner ponders... Apr 20 22:56:44 bluelightning: i did notice this, i think it was around the time of the OEDAM, it used to be enough to say "devtool add " but now it needs "devtool add "? (or something like that) Apr 20 22:57:19 tlwoerner: right yeah that was what I recall reading - that's certainly not the result of an intentional change, have you got an example reproducer? Apr 20 22:57:47 (sorry it's taken me so long to follow up...0 Apr 20 22:58:45 bluelightning: you don't owe anyone any apologies! Apr 20 22:59:07 to be honest, it's not a really bad thing Apr 20 22:59:20 i don't have a reproducer on-hand, but could go looking for one Apr 20 22:59:40 I thought there was an example in the original message but I can't remember for sure Apr 20 22:59:49 was it in the minutes perhaps? Apr 20 22:59:50 * bluelightning checks Apr 20 23:00:40 nope, doesn't seem to be there at least Apr 20 23:00:46 (i.e. no mention at all) Apr 20 23:05:13 bluelightning: haha, funny, i can't find it either Apr 20 23:05:35 *mass hallucinations* Apr 20 23:07:59 tlwoerner: ok, no worries... either that or some kind of telepathy :D Apr 20 23:08:11 in any case if you do find a reproducer please do let me know Apr 20 23:08:20 bluelightning: ok, np Apr 20 23:08:25 :-) Apr 20 23:08:25 thanks Apr 21 00:42:57 bluelightning: ahh!! not me Robert Day Apr 21 00:43:17 (right country, even right province, wrong person) haha Apr 21 00:44:17 bluelightning: http://lists.openembedded.org/pipermail/openembedded-core/2017-March/133848.html Apr 21 00:47:42 funny how easy it was to "remember" it was me Apr 21 00:48:43 in my defense, bluelightning is *very* persuasive, he's using and old Yocto mind-trick on me! Apr 21 00:55:33 bluelightning: ...and i have been able to reproduce it Apr 21 00:55:36 brb Apr 21 00:56:13 if you checkout the code that was "current" at that time, you can reproduce it Apr 21 00:56:34 take bitbake and openembedded-core Apr 21 00:56:38 then, in each do Apr 21 00:56:41 git checkout $(git rev-list -n 1 --before="2017-03-10" master) Apr 21 00:57:07 then go through the "source"'ing step, then the "devtool add" step Apr 21 00:59:33 oh... seems to happen with current master too Apr 21 01:23:37 tlwoerner: oops - sorry for the mixup Apr 21 01:24:24 damn it I even googled for "site:lists.openembedded.org devtool name" and didn't find it :/ Apr 21 01:24:39 thanks for digging it up Apr 21 01:29:52 I can indeed confirm the issue here... now to find out how it's happening Apr 21 01:36:47 so it's picking this up from the spec file... it contains Name: %{name} Apr 21 01:37:26 I wonder if there's a default for %{name} that says pick it up from the spec file name? Apr 21 01:37:46 can't seem to find any documentation that indicates how this would be interpreted **** ENDING LOGGING AT Fri Apr 21 03:00:02 2017