**** BEGIN LOGGING AT Thu Oct 12 03:00:02 2017 Oct 12 07:44:08 is there some official meson bbclass and if so, since which yocto version would that be available? Oct 12 07:53:41 a) found the bbclass - not part of yocto as of 2.4, right? Oct 12 08:15:46 Hi, Im new to yocto (but have been creating embedded linux systems with buildroot before), and Im having difficulty porting an existing recipe from open embedded to yocto / Poky Oct 12 08:15:52 Im trying to get zbar included in my embedded build, and have succeeded getting it compile and work with imagemagic (--with-imagemagick). Now Im trying to add --with-python, but Im getting headers not found issues for Python.h: Oct 12 08:15:58 tmp/work/cortexa7hf-vfp-neon-poky-linux-gnueabi/zbar/0.10-r0/zbar-0.10/python/zbarmodule.c:24:0 tmp/work/cortexa7hf-vfp-neon-poky-linux-gnueabi/zbar/0.10-r0/zbar-0.10/python/zbarmodule.h:24:20: fatal error: Python.h: No such file or directory Oct 12 08:16:10 If I navigate to tmp/sysroots/pico-imx6ul-emmc/usr/include/, and search for Python.h, I can see it's under 'python2.7' folder. Now if I symlink Python.h in usr/include to usr/include/python2.7, I will get complaint of the next header (which is again in the same folder) Oct 12 08:16:24 What would be the correct way to let zbar know that all the python headers are under python2.7 folder ? Oct 12 08:16:47 I assume it would be something in the bb recipe file, but Im not really sure how to configure these Oct 12 08:23:45 jlumme, what do you have in your recipe DEPENDS and RDEPENDS? Oct 12 08:32:05 melonipoika: please find the recipe file here http://paste.ubuntu.com/25725000/ Oct 12 08:32:36 Looking more carefully at the compilation, I noticed this string: "| cc1: warning: include location "/usr/include/python2.7" is unsafe for cross-compilation [-Wpoison-system-directories]" Oct 12 08:33:38 I put the whole compilation log here: http://paste.ubuntu.com/25725014/ Oct 12 08:33:56 I noticed that there is a complaint "| cc1: warning: include location "/usr/include/python2.7" is unsafe for cross-compilation [-Wpoison-system-directories]" Oct 12 08:34:58 my company blocks that page :-( but good that you found some leads Oct 12 08:37:38 oh... crap Oct 12 08:39:03 melonipoika: in the recipe I have DEPENDS = "pkgconfig intltool-native libpng jpeg imagemagick python" Oct 12 08:39:20 no RDEPENDS Oct 12 08:41:11 melonipoika: The original recipe I took from here http://cgit.openembedded.org/meta-openembedded/tree/meta-oe/recipes-support/zbar/zbar_0.10.bb Oct 12 08:42:06 Added imagemagic and python there, and slight modification for PACKAGECONFIG also because I think the recipe was from newer version of OE Oct 12 08:42:45 (replaced @bb.utils.filter with @bb.utils.contains) Oct 12 08:48:03 hi Oct 12 09:03:23 jlumme: you need to fix the configure script then, its hard-coding a look at /usr Oct 12 09:06:07 rburton: I agree that it should not look at /usr/include, but I wonder why adding a Python.h symlink to tmp/sysroots/pico-imx6ul-emmc/usr/include it will find the header and move forward. It appears as if it looks also in the "proper usr/include" location Oct 12 09:06:42 I was thinking if I could make it look under python2.7 in that same folder chain, it would find all the needed headers Oct 12 09:07:16 depends=python should give you the py2 headers in the sysroot Oct 12 09:09:18 This is what the compilation command looks like Oct 12 09:09:19 arm-poky-linux-gnueabi-libtool: compile: arm-poky-linux-gnueabi-gcc -march=armv7-a -mfloat-abi=hard -mfpu=neon -mtune=cortex-a7 --sysroot=/home/jlumme/imx_build/build-x11-pico-imx6ul/tmp/sysroots/pico-imx6ul-emmc -DHAVE_CONFIG_H -I. -I/home/jlumme/imx_build/build-x11-pico-imx6ul/tmp/work/cortexa7hf-vfp-neon-poky-linux-gnueabi/zbar/0.10-r0/zbar-0.10 -I./include -fno-strict-aliasing -DNDEBUG -g -fwrapv -O2 -Wall -Wstrict-prototypes -D_ Oct 12 09:10:08 apologies, thats not correct - I was trying something else Oct 12 09:10:19 this is what it is: | arm-poky-linux-gnueabi-libtool: compile: arm-poky-linux-gnueabi-gcc -march=armv7-a -mfloat-abi=hard -mfpu=neon -mtune=cortex-a7 --sysroot=/home/jlumme/imx_build/build-x11-pico-imx6ul/tmp/sysroots/pico-imx6ul-emmc -DHAVE_CONFIG_H -I. -I/home/jlumme/imx_build/build-x11-pico-imx6ul/tmp/work/cortexa7hf-vfp-neon-poky-linux-gnueabi/zbar/0.10-r0/zbar-0.10 -I./include -fno-strict-aliasing -DNDEBUG -g -fwrapv -O2 -Wall - Oct 12 09:10:38 sigh, another copy paste fail Oct 12 09:10:48 | arm-poky-linux-gnueabi-libtool: compile: arm-poky-linux-gnueabi-gcc -march=armv7-a -mfloat-abi=hard -mfpu=neon -mtune=cortex-a7 --sysroot=/home/jlumme/imx_build/build-x11-pico-imx6ul/tmp/sysroots/pico-imx6ul-emmc -DHAVE_CONFIG_H -I. -I/home/jlumme/imx_build/build-x11-pico-imx6ul/tmp/work/cortexa7hf-vfp-neon-poky-linux-gnueabi/zbar/0.10-r0/zbar-0.10 -I./include -fno-strict-aliasing -DNDEBUG -g -fwrapv -O2 -Wall -Wstrict-prototypes - Oct 12 09:10:50 ; ) Oct 12 09:11:14 oh - appears like its too long Oct 12 09:12:00 maybe the pastebin location I posted above is easiest to look at Oct 12 09:17:33 hi, I'm a new user and I'm wondering if something can be done about the time to takes to compile one thing. Currently I'm always waiting about a minute while recipes are being parsed before compilation even starts. This makes my debug cycle sluggish to say the least. Oct 12 10:58:31 bertrik_work please take a look at http://www.yoctoproject.org/docs/2.3.2/mega-manual/mega-manual.html#using-devtool-in-your-workflow Oct 12 11:22:42 Hi all, does someone see a non wanted behavior by sharing a download_dir withmultiple users ? Oct 12 11:23:13 I was wondering about for example patch, that have both same name but inside isn't the same Oct 12 11:27:10 well you'll get that problem with a single user too Oct 12 11:27:19 if multiple recipes download a file with the same name Oct 12 11:39:09 rburton : so when you apply a patch, you apply a patch from the file in downloads ? Oct 12 11:39:30 ChrysDaze: if the patch is in the layer as a file: then it doesn't go to downloads at all Oct 12 11:39:41 rburton : ok Oct 12 11:40:04 if you're fetching a file from the internet using http or whatever it goes into downloads and then copied/unpacked into workdir Oct 12 11:40:50 Actually, if it's downloaded it should have a checksum, and that checksum is re-verified AFAIR. Oct 12 11:41:14 I think I contributed a patch a while back that puts the file's checksum in the .done stamp file and compares them with the checksum in the recipe Oct 12 11:41:39 afair there isn't a reverification at unpack time Oct 12 11:41:49 afair that's exactly what I added back then Oct 12 11:41:55 i'll check because i've been meaning to do that, so that's great news :) Oct 12 11:42:35 fbd4a0d422cf7f43db2f9eab2e47c41246a9031e in the bitbake repo Oct 12 11:42:49 mario-goulart: wow, thanks for the quick service ;-) Oct 12 11:42:56 yw :-) Oct 12 11:43:51 https://bugzilla.yoctoproject.org/show_bug.cgi?id=5571 Oct 12 11:43:52 Bug 5571: enhancement, Medium, 1.8, alejandro.hernandez, RESOLVED FIXED, Re-validate checksums in do_unpack Oct 12 11:44:09 \o/ Oct 12 11:44:46 .oO( back when we still had time for contributions ;-) ) Oct 12 11:45:14 ha, productising is a pain isn't it Oct 12 11:45:31 WARNING: m4-1.4.18-r0 do_fetch: Checksum mismatch for local file /home/ross/Yocto/downloads/m4-1.4.18.tar.gz Oct 12 11:45:37 verified it still works ;) Oct 12 11:45:42 neverpanic : so if the checksum failed , go back downloading? Oct 12 11:45:46 yes Oct 12 11:45:56 yes, if the checksum doesnt match it renames and refetches Oct 12 11:46:09 neverpanic : so we can share the same dl_dir according to this Oct 12 11:46:20 the situation you're describing is unrelated to sharing Oct 12 11:46:26 you'll hit it with a single user Oct 12 11:46:33 Well, that's not to say that there might not be other issues with sharing a DL_DIR... Oct 12 11:46:42 it's still sharing rburton Oct 12 11:46:46 between two project Oct 12 11:46:55 But I'm sharing a DL_DIR with three other users and it seems to work fine for this small setup Oct 12 11:46:56 you'll hit it with a single user and a single roject Oct 12 11:47:03 (not so much for sstate, that breaks from time to time) Oct 12 11:47:12 all you need is two recipes which download a file with the same name Oct 12 11:47:22 use downloadfilename in the SRC_URI to control the name so they don't collide Oct 12 11:47:34 downloadfilename? Oct 12 11:47:49 (see the SRC_URI docs) Oct 12 11:48:06 Don't we have per-recipe directories in $DL_DIR these days? Oct 12 11:48:13 i think not Oct 12 11:48:23 it download directly in the dl_dir the file Oct 12 11:48:39 yeah what ChrysDaze said Oct 12 11:48:48 Sorry for the lame questions. Still in jethro... Oct 12 11:48:59 mario-goulart: per-recipe would mean you can't share tarballs between recipes Oct 12 11:49:00 No lame question, only lame reply. Oct 12 11:49:40 rburton: true. My major concern is name clashes for patch files. Oct 12 11:50:02 mario-goulart : was exaclty my concern too, that's why i ask Oct 12 11:50:11 "fix.patch" :-) Oct 12 11:50:23 mario-goulart: only relevant if you fetch patches over http Oct 12 11:50:27 that's even not a good name for a patch ahahah Oct 12 11:51:14 rburton: if you have files in layer dirs, and referenced in SRC_URI, they go to $DL_DIR too, no? Oct 12 11:55:22 mario-goulart: the only things that go into DL_DIR are remote URLs in SRC_URI Oct 12 11:58:01 rburton : the only probelm si that two people are bitbaking in the same time the same recipe with the same patch name but with different content. Oct 12 11:58:10 rburton: I have a lot of .patch.done files in my $DL_DIR (no .patch, though). So maybe the one the .done for patches. Oct 12 13:46:31 Does anyone have already use ftrace? Oct 12 14:55:24 RP, around? Oct 12 15:30:31 Tartarus: yes Oct 12 15:30:50 RP, hey, so an odd bitbake signature issue / question / sanity check, if you please Oct 12 15:30:59 What's done in https://git.automotivelinux.org/AGL/meta-agl/tree/meta-agl/recipes-graphics/wayland/weston-ini-conf.bb#n26 results in WESTONSECTION (and in turn a number of other variable-with-flags) not ending up in the signature state. This is a 'feature' and the resolution is to use do_...[vardeps] to track them in turn, yes? Oct 12 15:31:09 Or is something else going on? Oct 12 15:31:29 (this is on pyro atm, but we see it on morty too, and I'm assuming master/rocko, barring some changes) Oct 12 15:32:37 Tartarus: I suspect we have no magic for looking for getVarFlags (in the same way we do have magic for getVar) Oct 12 15:33:05 So how is PACKAGECONFIG tracked? Oct 12 15:33:20 Are there just enough bb.note/etc calls somewhere that inform on it that it is tracked? Oct 12 15:37:04 even if it did pick up the getVarFlags(), you're doing a second getVarFlags() on a variable, which bitbake wouldnt'; be able to track anyway Oct 12 15:37:11 so you'll need to add extra explicit deps either way Oct 12 15:37:31 OK, thanks guys Oct 12 15:38:14 so you'll want some inline python in the vardeps flag to dump the list of all the deps Oct 12 15:38:21 we do that in package.bbclass for example Oct 12 19:17:07 hi all, I'm addin some python3 app and it has dependency on scheduler, but I cnanot find such package in manifest (should be part of standard python) Oct 12 19:17:09 any ideas? Oct 12 19:20:24 open-nandra: do you mean https://pypi.python.org/pypi/schedule or https://docs.python.org/3/library/sched.html ? Oct 12 19:21:12 bluelightning: I have seen just import scheduler and import sched Oct 12 19:22:10 open-nandra: looks like python3-misc is the package you want Oct 12 19:22:34 that conains the sched module, can't find a "scheduler" module in the standard python distribution Oct 12 19:23:26 bluelightning: ok it works python3-misc Oct 12 19:23:28 thanks alot Oct 12 19:30:46 open-nandra: no problem Oct 12 19:31:17 FYI I found that using oe-pkgdata-util find-path */sched.py Oct 12 19:31:31 (sometimes you need .* in case it's a C module) Oct 12 19:31:57 that will only work if the package has actually been built first Oct 12 19:41:45 bluelightning: thanks for valuable info -> I planned to ask but don't want to annoy so much :D Oct 12 20:32:31 hi all, had a question about kernel-fitimage.bbclass Oct 12 20:33:18 i enable the fitimage using KERNEL_CLASSES += "kernel-fitimage" and KERNEL_IMAGETYPE = "fitImage" Oct 12 20:34:02 but when i able initramfs along with these settings, using INITRAMFS_IMAGE = "core-image-minimal" and INITRAMFS_IMAGE_BUNDLE = "1" Oct 12 20:35:32 kernel build fails, its mostly from kernel.bbclass because it tries to deploy fitimage https://github.com/openembedded/openembedded-core/blob/master/meta/classes/kernel.bbclass#L639 Oct 12 20:35:56 am i using it correctly? anyone else facing same issue? Oct 12 21:28:25 Tartarus: FWIW we track contains() usage specifically and therefore DISTRO_FEATURES access uses contains are handled specially Oct 12 21:28:59 Ah, ok **** ENDING LOGGING AT Fri Oct 13 03:00:02 2017