**** BEGIN LOGGING AT Wed Aug 17 02:59:58 2016 Aug 17 08:17:10 compiling yocto with no gcc is fun :P Aug 17 08:37:49 khem: would something like this work from .conf context too?: do_compile[nostamp]_pn-recipename = "1" or do_compile_pn-recipename[nostamp] = "1" Aug 17 08:37:58 that could be useful.. Aug 17 09:03:33 hi all Aug 17 09:05:23 is it possible to built all python packages into an image except to put them all in "IMAGE_INSTALL_append"? Aug 17 09:05:58 hi guys, i was wondering how do I deploy u-boot on sdcard with bitbake ? I have an imx28evk board Aug 17 09:06:58 kbo: this is absolutely board specific, you need to look into the BSP/documentation for your hardware. Aug 17 09:10:24 LetoThe2nd: Im using meta-fsl-arm Aug 17 09:11:41 BlitzBlizz: well you can always create a packagegroup Aug 17 09:12:17 kbo: i geussed that from the imx28 mention. Aug 17 09:15:10 rburton: HEEELP! Aug 17 09:15:33 LetoThe2nd: what about including "python-modules"? Aug 17 09:15:35 rburton: I'm having a weird issue with the yocto base, wanted to run it past you Aug 17 09:15:44 (Or anyone else) Aug 17 09:16:07 For whatever reasons my current builds get: Aug 17 09:16:09 python -c "import sysconfig; print sysconfig.get_config_var('CC')" Aug 17 09:16:09 x86_64-unknown-linux-gcc -m64 -march=core2 -mtune=core2 -msse3 -mfpmath=sse --sysroot=/srv/sdkbuilder/work/build/x86_64/freedesktop-sdk-base-1-4/build/x86_64/tmp-glibc/sysroots/qemux86-64 Aug 17 09:16:46 BlitzBlizz: that might of course already be such a packagegroup Aug 17 09:16:57 And distutils uses basename(sysconfig.get_config_var('CC')) to figure out if the compiler is gcc Aug 17 09:17:07 and obviously basename picks up the / in --sysroot= Aug 17 09:17:20 But my old builds never got --sysroot in the CC var Aug 17 09:17:57 Has anyone seen something like this (jethro branch) Aug 17 09:18:01 LetoThe2nd: ok thanks. i will give it a try Aug 17 09:21:46 #join #imx6-dev Aug 17 09:25:21 alexlarsson: I have weird issues with jethro right now (though not exactly the same) because my host compiler migrated from gcc5 to gcc6, maybe it's related Aug 17 09:38:19 morning all Aug 17 09:40:10 khem: ping Aug 17 09:40:25 khem: are there any objections against https://patchwork.openembedded.org/patch/128887/ Aug 17 09:40:50 khem: who should I ping to get this merged? Aug 17 09:58:38 Hello, i have a question about pyqt, it is possible to use it with QWS and not with X11 ? Aug 17 10:08:29 Okay, last booting round I've got some errors with pythion subprocess missing, was easy to statisfy. But now I get "ImportError: No module named grp" Aug 17 10:08:49 Does someone know the recpie which is providing this python packages? Aug 17 10:08:54 python mdoule Aug 17 10:11:06 Hunk: I'm currently using pyqt with wayland Aug 17 10:12:34 How i can add wayland to my image? Aug 17 10:12:43 and thank your for your response Aug 17 10:14:27 CORE_IMAGE_EXTRA_INSTALL += "wayland weston" Aug 17 10:14:31 is that all? Aug 17 10:17:58 because i have still the same error Aug 17 10:18:07 ERROR: python-pyqt was skipped: missing required distro feature 'x11' (not in DISTRO_FEATURES) Aug 17 10:32:23 hi.. what is best way to update etc/passwd? I need to add more users and groups. I don't want to do this is individual recipes, but in one dedicated recipe/place.. any idea? Aug 17 10:53:58 btooth: http://www.yoctoproject.org/docs/2.1/ref-manual/ref-manual.html#ref-classes-extrausers Aug 17 10:53:59 btooth: look into 'EXTRA_USERS_PARAMS' Aug 17 10:55:58 does yocto do autoreconf for autotools based recipes ? Aug 17 11:08:09 boucman_work: yes Aug 17 11:08:52 Guest96062: hmm, ok... doesn't seem so (at least for the complicated case of the rpm-native recipe's recursive dependencies) Aug 17 11:29:31 Hi all. I am currently porting some of our projects to cmake and hence i included cmake recipe in our build to get OEToolchainConfig.cmake in our SDK. However I now noticed, that sysroot location is supplied as flags in CC, CXX, CPP, LD and KCFLAGS directly. However CMake has CMAKE_SYSROOT ( https://cmake.org/cmake/help/latest/variable/CMAKE_SYSROOT.html ) for this. Is there a reason it's not used? Aug 17 11:43:59 latest checkout of poky fails on sysvinit-inittab, cannot stat start_getty, its included next to the recipe, also in the recipe, but not showing up in WORKDIR, what could the problem be ? Aug 17 11:44:06 branch krogoth Aug 17 11:50:08 Has anybody tried building yocto in a build directory named "perl"? I have a reasonable suspicion that this might break perl-native: https://p.dnnr.de/kTLA01uEvnGlGESh Aug 17 11:52:45 Meh I forgot the package python-modules meh Aug 17 11:54:05 http://git.yoctoproject.org/cgit.cgi/poky/plain/meta/recipes-devtools/python/python-2.7-manifest.inc Aug 17 11:54:32 ok, that one baffles me... Aug 17 11:54:54 I am trying to compile master without gcc (i.e using BUILD_CC to provide the compiler) Aug 17 11:55:16 rpm-native:do_configure fails, complaining that my compiler can't produce correct binaries Aug 17 11:55:36 but when I run ./run.do_configure in the temp directory, it configures correctly Aug 17 11:56:03 or not. it did last time. I need to check what's going on Aug 17 11:59:42 joshuagl: thanks, I will try Aug 17 12:23:05 I think perl-native.bb should pass -Dinstallstyle="lib/perl5" -O to Configure Aug 17 12:56:55 I still having problems with mod_wsgi compiling. This package has to be processed with setuptools and autotools which is ok. But autotools is generating with his Makefile.in and configure.ac script wrong path. Aug 17 12:57:06 For exampple: LIBEXECDIR = /usr/lib/apache2/modules Aug 17 12:57:16 This is from LIBEXECDIR="`${APXS} -q LIBEXECDIR`" Aug 17 12:57:44 So the reason is APXS is used from my local computer instead from the apache2-native sysrootfs Aug 17 12:58:17 The reason for that is this part of the configure.ac script: AC_PATH_PROGS(APXS, apxs2 apxs, [apxs], [$PATH:/usr/local/apache/bin:/usr/sbin]) Aug 17 12:58:54 It searching for apx2 and apxs. Apxs2 is in my computer and apxs is in the sysrootfs of my yocto toolchain Aug 17 12:59:24 Now I have to options: deleting my local apxs2 (so autotools finds the yocto apxs) or better: Aug 17 13:00:03 I use this switch: for configure "--with-apxs=NAME" Aug 17 13:00:54 The meta-oe php_5.5.21 package is adding this configure switch this way: PACKAGECONFIG[apache2] = "--with-apxs2=${STAGING_BINDIR_CROSS}/apxs,,apache2-native apache2" Aug 17 13:01:01 But I don't understand the syntax Aug 17 13:01:06 Can someone help me? Aug 17 13:02:23 My apxs is there: oe-core/build/out-glibc/sysroots/colibri-vf/usr/bin/crossscripts/apxs Aug 17 13:16:30 Hello. What is the proper procedure in Yocto for changing the RGB TFT display used with an imx6ul development board? I am new to this. Any decent links to tutorials and/or documentation would be welcome. Aug 17 13:18:14 As far as I understand my mxs-lcdif driver is responsible for the display (info I get via fbset). I can enable and disable it via menuconfig. But how do I do proper configuration? Aug 17 13:19:12 I wish to switch from an 800x480 18 bit TFT to a 24 bit 480x272 TFT Aug 17 13:19:44 HyP3r: "if apache2 is in PACKAGECONFIG, then pass --with-apxs2=... to ./configure and also add apache2-native and apache2 as build-time (DEPENDS) dependencies." Aug 17 13:19:48 that's how you read it Aug 17 13:20:46 the empty second argument (,,) is what to pass to disable the feature in case "apache2" is not in PACKAGECONFIG. passing nothing does it in this case. Aug 17 13:21:05 see http://www.yoctoproject.org/docs/2.2/ref-manual/ref-manual.html#var-PACKAGECONFIG as well Aug 17 13:31:14 Ulfalizer: thanks. I solved that by adding EXTRA_OECONF = " --with-apxs=${STAGING_BINDIR_CROSS}/apxs " Aug 17 13:31:24 Ulfalizer: so the configure scripts finde the correct apxs Aug 17 13:32:27 Ulfalizer: but the next problem is even more trickier: while compilation make can't fine the Python.h file, and mod_wsgi is searching for this header file with a really akward way: Aug 17 13:33:10 CPPFLAGS1=`${PYTHON} -c 'from sys import stdout; from distutils import sysconfig; stdout.write("-I" + sysconfig.get_config_var("INCLUDEPY"))'` Aug 17 13:34:38 And there I get this error: PREFIX = os.path.normpath(sys.prefix).replace( os.getenv("BUILD_SYS"), os.getenv("HOST_SYS") ) Aug 17 13:34:43 TypeError: expected a character buffer object Aug 17 13:34:54 Seems like those to variables are not set Aug 17 13:35:04 Question: how to krogoth+meta-qt5+wayland (which flags are needed to enable) Aug 17 13:53:03 HyP3r: you shouldn't have to change EXTRA_OECONF. the proper way to enable apache2 is to include "apache2" in PACKAGECONFIG. the end of the glossary entry i linked has some examples showing how you can do that. Aug 17 13:55:18 HyP3r: i think i've seen that BUILD_SYS/HOST_SYS issue before. try googling e.g. yocto "export build_sys" Aug 17 13:55:57 Yeah, Python stuff requires those to be exported; the python bbclasses usually do this. Aug 17 13:58:01 MEH!!!! I'm slowly getting crazy, ok Ulfalizer I fixed this, is also working ok, neverpanic ok I fixed this and yes configure is working but I still cant compile. Aug 17 13:58:39 Seems like the CFLAGS still don't include the python header directory, but I don't know why Aug 17 13:59:09 HyP3r: only slowly? Aug 17 14:00:32 Seems like some apes on crack wrote this configure script Aug 17 14:03:12 stdout.write("-I" + sysconfig.get_config_var("INCLUDEPY"))' returns as include path for python header files in devshell (bitbake -c devshell mod-wsgi) "-I/python2.7" Aug 17 15:29:53 Hi guys ! I'm doing a bluetooth NAP with yocto on a small Hachiko board. I want to be able to make a PAN network. Aug 17 15:30:49 I saw meta-networking wich seem great but I'm stuck with a compilation issue. Aug 17 15:31:32 Has already somebody did this a bluetooth accespoint with PAN before ? Aug 17 15:32:41 I already added the BBLAYERS but it seems stuck in meta-oe Aug 17 15:33:46 *the BBLAYERS that are in dependencies : openembedded-core meta-python meta-oe Aug 17 15:34:46 ERROR: Failed to parse recipe: /home/architech/architech_sdk/architech/hachiko/yocto/poky/meta-openembedded/meta-oe/recipes-support/libsoc/libsoc_0.8.1.bb Aug 17 15:38:04 (in fact I just nett to use brctl to make my PAN bluetooth network work, is there a simple solution to do it ?) Aug 17 16:12:01 join #oe Aug 17 16:12:06 oops Aug 17 16:19:31 Hi! Can you make a yocto recipe default to the tip of a branch Aug 17 16:19:32 ? Aug 17 16:19:45 for example the tip of master for a particular project? Aug 17 16:21:00 nisha: RCREV = Aug 17 16:21:01 "${AUTOREV}" Aug 17 16:21:06 oops Aug 17 16:21:15 SRCREV = "${AUTOREV}" Aug 17 16:21:16 there Aug 17 16:21:51 ulf`, thanks! Aug 17 16:22:31 nisha: welcome Aug 17 16:24:05 ulf`, while your here, do you by chance know how to bypass the git branch --contains that bitbake does? Aug 17 16:24:51 If I pull multiple sources via git, how do I set the directory for doing build? Do I specify something like S=${WORKDIR}/git/therepotwhereIwanttobuild"? Aug 17 16:29:25 hmm. apparantly not. complains about SRCREV_FORMAT. Aug 17 16:30:03 davis: yes, set $S Aug 17 16:30:16 davis: to get aorund the branch --contains check, set ;nobranch=1 in the SRC_URI Aug 17 16:30:40 i tried that. its got this srcrevformat thing now, i'm trying to figure that out. Aug 17 16:30:46 When using multiple git checkouts you must set SRCREV_FORMAT Aug 17 16:30:58 SRCREV shows up in megamanual but that one does not. Aug 17 16:31:33 usually SRC_URI = "git://...;name=foo git://...;name=bar" SRCREV_FORMAT="foo_bar" or SRCREV_FORMAT="foo", then SRCREV_foo = "abc098" SRCREV_bar = "1234fda" Aug 17 16:32:13 I've recently read documentation for it, so there certainly is some Aug 17 16:32:17 I don't recall where though Aug 17 16:34:23 i'm pulling foo, goo, hoo. so I removed SRCREV=${AUTOREV} and replaced it with SRCREV_foo="${AUTOREV}", likewise for goo and hoo Aug 17 16:35:57 https://www.yoctoproject.org/docs/2.1/bitbake-user-manual/bitbake-user-manual.html#var-SRCREV_FORMAT Aug 17 16:36:12 here is one in the docs. section 5.21 Aug 17 16:37:06 joshuagl: thanks. i need to start search bitbake as well as mega Aug 17 16:37:20 np Aug 17 16:37:30 the fetcher is part of bitbake, so documented there Aug 17 16:38:09 gotcha Aug 17 16:38:47 hmm. Aug 17 16:39:01 might be useful to include the bitbake manual in the mega manual, if it isn't already Aug 17 16:40:40 this is what i'm trying Aug 17 16:40:43 https://gist.github.com/netskink/643bdfbc92017b265881e600670ca906 Aug 17 16:40:57 I am having trouble adding a recipe to Yocto. I have added a recipe to install the azure libraries. It creates 3 shared libraries. In the azure..rpm I see the libumqtt.so.1.1 libumqtt.so.1 libamqp.so.1.1 libamqp.so.1 libumock.so.1.1 libumock.so.1 files. In the the azure-dev...rpm file I set libmqtt.so libamqp.so and libumock.so along with header files etc. Aug 17 16:41:01 but that still gives SRCREV_FORMAT error Aug 17 16:41:09 davis: you need name=foo in the URIs Aug 17 16:41:16 A second recipe to test the installation links correctly against the libraries but when do_rootfs runs I get the message "Computing transaction...error: Can't install azure-test-1.0-r0@armv5e: no package provides libumqtt.so". Aug 17 16:41:23 Any idea what I could be doing wrong? Aug 17 16:41:54 comparing the log.do_package output with the libgcrypt output show the run.do_package script has nothing is 'sonames'. The libgcrypt output shows libgcrypt.so.20. I know this must be a clue but how do I use it? Aug 17 16:41:56 something like? git://ssh@devsupport:22/tfs/DefaultCollection/jfdtesty/_git/d3;protocol=ssh;name=d3 Aug 17 16:42:29 JohnniePeters: sounds like that library has an incorrect soname. The soname of the libumqtt library sohuld be libumqtt.so.1 Aug 17 16:42:53 JohnniePeters: you usually don't want -dev packages on target anyway, and the .so file is probably just a symlink Aug 17 16:43:09 davis: yes Aug 17 16:43:21 and then SRCREV_d3, not SRCREV_pn-d3 Aug 17 16:43:51 and of course SRCREV_FORMAT, which is missing from your paste Aug 17 16:44:46 The libumqtt.so.1 file is a symlink to libumqtt.so.1.1. The libumqtt.so file is not on the file system but is is in the azure-dev*.rpm file. The Makefile for the test recipe use -lmqtt. Aug 17 16:44:55 what should SRCREV_FORMAT look like? the bitbake entry has some text on it, but no example of what it looks like Aug 17 16:46:32 davis: there is an example in the text Aug 17 16:46:45 The test program compiles with no errors and the azure-test*.rpm file is created. Aug 17 16:47:08 JohnniePeters: Yeah, in that case check the soname of libumqtt. Also, if you have the choice, use a standard build system like CMake or autotools, they get this right for you. Aug 17 16:47:24 fwiw, i'm looking at this chapter in this book. they have one SRC_URI, but has a name=machine,meta entry at the end like you pointed out. Aug 17 16:47:52 then they also do a SRCREV_machine and meta like you pointed out as well. Aug 17 16:48:10 so it looks like they are pulling two version of the same repot Aug 17 16:48:27 yep. linux-yocto's fetch/checkout process is a little unusual Aug 17 16:48:42 similar but not the same as me. i'm pulling three repots but building only in one. Aug 17 16:50:25 I think the linux-yocto recipe isn't a very good learning resource Aug 17 16:51:34 try meta/recipes-core/glibc/cross-localedef-native_2.24.bb instead Aug 17 16:52:00 hmm. with additional tweaks its still giving me srcrev_format errors. the one in this book does not specify that variable. I'm using SRCREV_FORMAT ?= "d3_d3v16_pcmx" which correspond to the names for each of the three repots. Aug 17 16:52:37 That looks correct to me. Not sure why you're using a weak assignment, though. Aug 17 16:52:49 holy shit it looks like its building now Aug 17 16:53:18 nope Aug 17 16:53:33 let me adjust my gist Aug 17 16:54:58 https://gist.github.com/netskink/643bdfbc92017b265881e600670ca906 Aug 17 16:55:08 this has an error with do_unpack Aug 17 16:55:35 its saying fetcher fetch command failed, no such file or directory Aug 17 16:57:08 ive got a bunch of weirdness in my work/xxxx/pcmx dir Aug 17 16:57:35 Did someone manage to run/inherit autotools _and_ setuptools in one recpie? I don't get it :( Aug 17 16:57:52 hmm. one of them has just a readme in the git subdir Aug 17 16:57:56 odd Aug 17 16:59:37 neverpanic, thanks! nobranch=1 works Aug 17 17:02:22 it looks like of the three git repots i have, its only pulling one Aug 17 17:02:53 hi guys Aug 17 17:03:13 is ironport blocking your yocto emails? Aug 17 17:12:19 I changed the azure.bb file and attempted to rebuild by first doing a bitbake -c clean azure. But it does not recompile. It seems to be pulling in a stored of version of the package. How do I force it to recompile? Aug 17 17:16:36 I see cleansstate does what I want. Aug 17 17:17:24 i just did a rm -rf and restart because some of my mods are making dirs which I can not tell if they are legimate or not Aug 17 17:20:45 In the log.do_package file I see the line "DEBUG: LIBNAMES: pkg azure libs 1 bins 0 sonames []". I compared this to the same file for libgcrypt and I see "DEBUG: LIBNAMES: pkg libgcrypt libs 1 bins 1 sonames ['libgcrypt.so.20']" Aug 17 17:21:08 So the package script does seem to be seeing the libraries. Aug 17 17:22:25 In my azure.bb file I have file_${PN}-lib = "lib*.so*". What triggers the package to fine sonames? What could I be missing. Aug 17 17:42:59 After 3 days of trying things I finally figured it out. I had left the "-Wl,-soname,..." piece out of the link lines for the libraries. Aug 17 17:50:39 Failure :o Aug 17 17:55:32 Hi, I'm building multiple kernels in the same build folder and getting the error: The recipe ... is trying to install files into a shared area when those files already exist. Those files and their manifest location are: ... Aug 17 17:55:56 I'm already getting the output files that I need so I'd like to simply disable or supress this error Aug 17 17:56:29 I've tried deltask do_package and SSTATE_DUPWHITELIST = "${TMPDIR}/sysroots/imx6dlsabresd/pkgdata/runtime" in the recipe but neither seem to have any effect Aug 17 17:56:46 Does anyone here have experience with something like this? Aug 17 18:25:10 it looks like this is the same error as before Aug 17 18:25:45 https://gist.github.com/netskink/643bdfbc92017b265881e600670ca906 Aug 17 18:28:16 when i look in build/tmp-glibc/work/corei7-64-oe-linux/pcmx/1.0.0+gitAUTOINC+61d576b54e_61d576b54ev16_42d46f3e56-r0/temp/ Aug 17 18:28:45 i see the logs and tasks timestamp things? but I don't see the actual git pulls Aug 17 18:29:15 so its parsing the SRC uri's but not able to pull them, right? Aug 17 18:44:29 JohnniePeters: Yeah, as I said, that's why you use CMake or autotools, which don't forget -Wl,-soname for you. Aug 17 18:45:11 dmiess_: you shouldn't try to supress that error, rather fix the original problem and don't have two recipes install the same files? Aug 17 18:45:28 dmiess_: this can also happen if you move a file from one recipe to a different one, in which case you usually just delete tmp and try again Aug 17 18:45:35 I've some recpies with 'DEPENDS +=' and some with 'DEPENDS ='. Whats better or correct? Aug 17 18:46:21 davis: is seems the directory in which your fetch task runs is being deleted while it is trying to run there? Aug 17 18:46:57 neverpanic: got me Aug 17 18:47:01 HyP3r: There's nothing wrong with either of them; sometimes inherited bbclasses set DEPENDS to some values already, and for those it might make sense to use += to ensure you're not overwriting anything by accident Aug 17 18:47:28 But if you're sure DEPENDS is empty at the point where you set it (which is usually the case at the beginning of a recipe) DEPENDS = is fine Aug 17 18:48:11 ok, but for noobs is += then better.... Aug 17 18:48:34 davis: Or rather, the unpacking, which is also done by the fetcher Aug 17 18:48:49 neverpanic: any idea how to debug? does my gist look correct which shows the config? Aug 17 18:49:00 err, shows the recipe Aug 17 18:51:56 in that recipe, i took out the SRCREV line, its not used when use the SRCREV_FORMAT and individual SRCREV_xxx variables, right? Aug 17 19:04:39 davis: yes, didn't see any problems there. Try reading $WORKDIR/temp/run.do_unpack, although since that's Python code it's probably not very helpful Aug 17 19:05:13 Yes, although I think you can use SRCREV to set a default, but I've never tried that. Aug 17 19:07:46 neverpanic: fwiw, i've update the gist Aug 17 19:08:00 it shows a "working" version where it just pulls one git repot Aug 17 19:08:13 and a "non-working" version where it pulls multiple git repots Aug 17 19:08:15 https://gist.github.com/netskink/643bdfbc92017b265881e600670ca906 Aug 17 19:15:55 davis: instead of DEPENDS = "cmake", inherit cmake, which will add the appropriate dependencies Aug 17 19:17:19 No idea why the fetch or unpack would fail for you, I think you get to debug this on your own Aug 17 19:18:49 i actually have inherit cmake Aug 17 19:19:03 its just below in the recipe Aug 17 19:19:24 yeah, I was commenting before I had read the whole thing Aug 17 19:20:07 no worries. i appreciate your help. Aug 17 19:20:46 its odd the src gets pulled down into a git subdir when I do the single url repot Aug 17 19:21:19 when I do multiple either its left over dir but it is still in git, not git/pcmx git/d3 etc Aug 17 19:21:43 I think you can choose where you want the source with a fetcher option Aug 17 19:21:53 time to read the git fetcher docs and hope it's documented Aug 17 19:22:36 shudder, lol Aug 17 19:24:04 https://www.yoctoproject.org/docs/current/bitbake-user-manual/bitbake-user-manual.html#git-fetcher destsuffix= Aug 17 19:32:25 I'm trying to create a bitbake recipe for apitrace, which uses cmake. Aug 17 19:32:39 apitrace has 3 nested CMakeList.txt files, and bitbake doesn't seem to handle this Aug 17 19:33:03 it worked, except at the install phase. bitbake didn't know about the files installed by the subdirectory cmakes Aug 17 19:34:15 Any clue where the generic cmake recipe is? I'm new to yocto, but have a couple of years CMake and might be able to improve the recipe if it needs it Aug 17 19:34:32 bitbake is magic to me right now. Aug 17 19:41:10 I've been writing too many shell scripts. I just added #!include to a file Aug 17 19:43:46 hello! I want to debug why my sstate is not catching everything I think it should catch. I found some old discussion about adding more debugging information about this from 2013, what's the recommended way to debug this today? Aug 17 19:46:11 davis: meta/classes/cmake.bbclass Aug 17 19:46:53 daddio: ^ (not davis) Aug 17 19:49:09 blueCmd: if you see a rebuild you didn't expect, use bitbake-diffsigs Aug 17 19:49:44 neverpanic: I have a central sstate repository, so I don't really understand how I'm supposed to use that one Aug 17 19:50:05 the examples I got was more "use this to diff X and Y" Aug 17 19:50:07 I don't think it makes a difference for bitbake-diffsigs Aug 17 19:52:35 ok, I'll give it a try Aug 17 19:52:53 building right now so it will take a while until I can give it a try Aug 17 20:06:23 daddio: sounds like you should read cmake.bbclass Aug 17 20:06:33 and obviously make ure you inherit cmake Aug 17 20:07:33 kergoth: Thanks, I'll have a look. "inherit cmake python-native" is a s far as I got so far. Aug 17 20:07:53 compare what cmake.bbclass is doing to what the project recommends for build & installation. Aug 17 20:07:55 np Aug 17 20:09:11 neverpanic: so, this might be a weird question, but how do I know that it's building or using a sstate? right now I'm doing 'find ../build/tmp/stamps -type f | grep -v setscene' but I dunno if that's sane Aug 17 20:10:31 buildstats-summary.bbclass will display a summary at the end of the build with how many tasks were built from sstate and how many from scratch.. but examining stamps isn't an unreasonable way to go Aug 17 20:11:01 aha, i'll include that and have a look Aug 17 20:11:35 kergoth: my current goal is to understand why gcc is built every time. I have this stamp as the first one not called setscene: ../build/tmp/stamps/work-shared/gcc-5.3.0-r0.do_build.78d4a0ca4469e23e8c98a361a2141be3 Aug 17 20:11:51 do_build are irrelevent Aug 17 20:11:58 oh ok Aug 17 20:12:02 do_build is an empty task, a placeholder. all it does is depends on other tasks Aug 17 20:12:26 do_compile is maybe the one I want to focus on then? Aug 17 20:12:37 or maybe some do_package? Aug 17 20:12:44 when you 'bitbake foo' what bitbake acutally does is builds the ${BB_DEFAULT_TASK} task of recipe foo, and BB_DEFAULT_TASK is do_build. it doesn't do anything, it's just how we know what tasks are built by default Aug 17 20:12:52 really depends on what you're trying to do Aug 17 20:13:59 my super-end-goal is that I want to have a post-CI-analysis saying "This commit indirectly changed these packages" Aug 17 20:14:43 and I'm thinking I could use sstate for that - because in my world a perfectly tuned yocto would never re-build something from the same git commit of the build repo Aug 17 20:15:00 so my first step would be to arrive at this perfect world Aug 17 20:29:39 meta-qt5 maintainers here ? Aug 17 20:30:38 if i do cleansstate for a particular recipe, why is that I still have dirs in build/tmp-glibc/work/corei7-64-oe-linux/myfailingrecipe? Aug 17 20:30:57 is it ok, to simply rm -rf the dirs there? Aug 17 20:58:34 Is there a way to force-rerun a task like "do_populate_lic"? Aug 17 21:02:50 "bitbake -f" forces things, I feel dumb **** ENDING LOGGING AT Thu Aug 18 02:59:58 2016