**** BEGIN LOGGING AT Wed Jan 27 02:59:59 2016 Jan 27 08:27:50 Hi, how to map opkg binary package names to recipe or source package names after a build? Jan 27 08:36:53 hi, is there a way/command to show all licenses used for an image? Jan 27 08:41:36 ericbutters: I guess you could do "bitbake -e | grep LICENSES=" to get all the license names used in the image Jan 27 08:43:01 ericbutters: ah, the variable is actually SRC_DISTRIBUTE_LICENSES and I'm not sure what it contains Jan 27 08:44:27 ericbutters: hi, you can also find it in your build folder tmp/deploy/licenses. It contains the license for each package and for your image. Jan 27 08:44:47 ericbutters: ok, forget about it. This is not what you want. Jan 27 08:45:36 MyLene: yes i saw this, and that is good to have.. but I was wondering if there is a way to get simple output of license per package Jan 27 08:46:16 frsc: SRC_DISTRIBUTE_LICENSES => what is this list? Jan 27 08:47:10 ericbutters: I don't know. At first I thought it is what you are looking for, but it definitely is not. Jan 27 08:49:14 Can anyone help with an issue regarding the imx vivante galcore module not being loaded with linux-imx-3.14.38 (jethro)? Jan 27 08:50:10 It is installed, and autoloaded by default but not listed when I do lsmod Jan 27 08:53:31 i got it! it is in deploy/licenses/$IMAGE_BASENAME Jan 27 09:09:47 ok, I got my issue solved. My rootfs seemed to be corrupted... Jan 27 09:58:18 <_valle_> I'm struggeling with a u-boot recipe. Would be grateful if someone got any input on this issue. http://pastebin.com/hYPUwB9q Jan 27 10:07:13 _valle_: looks like it's missing and include path to sysroot Jan 27 10:07:25 s/and/an/ Jan 27 10:07:44 needs a ${STAGING_INCDIR} Jan 27 10:21:14 <_valle_> nerdboy: Should I set the STAGING_INCDIR or should i set something to ${STAGING_INCDIR}. Thanks for your help! Jan 27 10:25:20 your fw build needs to know $STAGING_INCDIR Jan 27 10:25:47 maybe add -I$STAGING_INCDIR to EXTRA_OEMAKE Jan 27 10:26:12 depends on what's in makefile Jan 27 10:26:33 should add tyhat flag to CPPFLAGS or CFLAGS probably Jan 27 10:27:16 * nerdboy has to go meet his pillow... Jan 27 10:51:45 <_valle_> nerdboy: Didn't manage to get it to work Jan 27 11:12:06 ryansturmer: Hello again; I managed to get it working by adding a bbappend and a modified file in the barebox/.../boot/. Jan 27 11:12:43 ryansturmer: Once the barebox is updated, the new settings (video in my case) are correctly setted Jan 27 11:12:50 ryansturmer: So thanks again Jan 27 11:19:28 Hello everyone. Jan 27 11:19:34 I have a question concerning postgresql, sdcard and lock file. Jan 27 11:19:37 My postgresql database is situated on my sdcard. Jan 27 11:19:41 I configured the fstab so that de sdcard is automaticallyt mounted in the correct folder. Jan 27 11:19:46 If the target is hard resetted, a lock file (postmaster.pid) is present and prevent my database server to launch correctly. Jan 27 11:19:50 I created a file in my init.d folder that removes such a lockfile if any. Jan 27 11:19:54 Unfortunately, the file isn't removed. Jan 27 11:20:00 If I execute my script once the target is started, the file is correctly removed. Jan 27 11:20:05 Does it ring a bell? Is the fstab not yet used when the init.d/ scripts are launched ? Jan 27 11:20:10 thanks Jan 27 11:26:16 Hay again Jan 27 11:26:42 I jut realised there was a typo in my bb file... Jan 27 11:26:56 once corrected, erverything goes smooth Jan 27 11:29:10 olivier___: haha, we all do it :) Jan 27 11:30:51 Hi everybody! Can I ask something about psplash in yocto after generate my rootfs? Jan 27 12:37:44 Hi. I'm executing a ./configure script in a recipe. How do I make yocto use native toolchain instead of yocto environment Jan 27 12:37:48 configure: inferring LLVM_CXX/CC from CXX/CC = arm-poky-linux-gnueabi-g++ -march=armv7-a -mfloat-abi=hard -mfpu=neon -mtune=cortex-a8 --sysroot=/home/ubuntu/data/poky/build/tmp/sysroots/beaglebone/arm-poky-linux-gnueabi-gcc -march=armv7-a -mfloat-abi=hard -mfpu=neon -mtune=cortex-a8 --sysroot=/home/ubuntu/data/poky/build/tmp/sysroots/beaglebone Jan 27 12:38:12 Its automatically inferreing the toolchain in yocto environment Jan 27 12:42:18 make the recipe inherit native and it will use the native compilers instead of the cross ones Jan 27 13:29:10 <_valle_> Hi, I think I have a problem with wrong linker is being used when running Make. Jan 27 13:29:33 <_valle_> Would be appreciated if somebody could have a look. http://pastebin.com/GaKCi7sd Jan 27 13:35:26 _valle_: the makefile should probably just use $(CC) instead of $(CROSS_COMPILE)gcc, and pass $(CFLAGS) too Jan 27 13:35:46 oh, uboot Jan 27 13:35:50 world of pain Jan 27 13:42:51 is fedora's mkdir failing if it tries to create a existing directory? Jan 27 13:44:22 fsdun: failure to create a directory is an error by default in mkdir Jan 27 13:45:00 $ mkdir conf ; echo $? Jan 27 13:45:00 mkdir: cannot create directory ‘conf’: File exists Jan 27 13:45:00 1 Jan 27 13:45:03 (debian) Jan 27 13:45:21 the magic you're looking for is mkdir —parents Jan 27 13:46:10 rburton, I'm on Ubuntu and it fails Jan 27 13:46:35 but AFAIK Intel has fedora as preferred OS Jan 27 13:46:46 and I don't have a fedora for testing Jan 27 13:47:57 as i said, mkdir is meant to fail if the target exists, unless you pass —parents Jan 27 13:49:47 ok. then I'm wondering how they built an image using provided layers ... Jan 27 13:56:01 I see this was merged http://git.openembedded.org/openembedded-core/commit/?id=9b3dc1bd4b8336423a3f8f7db0ab5fa6fa0e7257 Jan 27 13:56:11 I have some patches adding signing support to ipk's Jan 27 13:56:43 should I rewrite them to use this module? (they're standing on the mailing lists without a response for some months I think) Jan 27 13:56:52 yeah sorry about that, been meaning to review Jan 27 13:56:59 rewriting to use the module would be *awesome* Jan 27 13:57:13 ok Jan 27 13:57:21 thank you Jan 27 14:42:45 <_valle_> rburton: Thank you, solved the problem! Jan 27 14:52:47 hi Jan 27 14:53:20 is there any way to exclude some dependencies that rpm is finding in packaging stage?! Jan 27 14:54:14 when building some package i want to make use of the rpm feature: __requires_exclude Jan 27 15:14:14 Hi, is there a convenient method to disable a specific DISTRO_FEATURE for an image? Or do i always need to change the local.conf? Jan 27 15:15:17 ccdDISTRO_FEATUREs are as name suggests for DISTRO not for IMAGE to select Jan 27 15:17:56 thanks Jan 27 15:49:27 hi, is there any way to exclude a file from FILERDEPENDS_xxx chain in packagedata ?! Jan 27 15:50:56 cupidro: make a bbappend to remove it? Jan 27 15:53:15 it's a binary package and i can not do a bbapend .. i just want force the packager to exclude some dependecies Jan 27 15:53:20 rpm style Jan 27 15:55:34 so, there is no recipe that is bringing in the binary package? Jan 27 15:57:49 yes it is, and when the packages packas the data it creates a set of rdeps which some i want to exclude Jan 27 15:59:34 found an workarround Jan 27 15:59:34 :) Jan 27 16:10:11 you can stop a recipe doing that with an append :) Jan 27 16:21:52 Any qt5 devs in the house? I have questions about the SDK. Jan 27 16:27:55 wyrm: ask, don't ask to ask. Jan 27 16:48:13 Does any one have example on adding the SRC_URI in the class file and then all the recipes will inherit that class? For example, I have couple of tarballs in a git repo and I have recipes using different tarballs. I don't want to put the full SRC_URI in each recipe because I don't want to change each recipe when ever the loacation of the git repo c Jan 27 16:48:13 hanges. I will just change the class file. Jan 27 16:58:07 ssb321: gnomebase.bbclass, then each recipe which inherits it uses += to add recipe specific items to SRC_URI Jan 27 17:02:07 ssb321: or have a .inc file, like all the xorg recipes have Jan 27 17:03:00 rburton: I asked yesterday. Today, I'm asking who I should poke. Jan 27 17:03:30 wyrm: maybe try the mailing list instead Jan 27 17:04:54 I might. Jan 27 18:04:24 _valle_: did you try just CPPFLAGS="-I${STAGING_INCDIR}" ? Jan 27 18:05:26 otherwise get in there and hack the makefile/makefile.am Jan 27 18:05:33 <_valle_> nerdboy: Yes I did Jan 27 18:05:45 <_valle_> nerdboy: I have solved it, thanks for your help! Jan 27 18:06:00 <_valle_> nerdboy: I had to patch the Makefile Jan 27 18:06:40 <_valle_> nerdboy: With the -I${STAGING_INCDIR} Jan 27 18:08:30 a basic static makefile for cross-compile should look kinda like this: https://github.com/VCTLabs/redis-ipc/blob/master/Makefile Jan 27 18:09:50 and C/CXXFLAGS should really be MY_CFLAGS... Jan 27 18:11:31 * nerdboy has seen more crazy upstream build systems than sane ones... Jan 27 18:26:22 <_valle_> yes, it didn't look like that unfortunately Jan 27 18:26:40 welcome marka Jan 27 18:30:46 <_valle_> nerdboy: My next issue is that I need to convert a bmp image by using imagemagic covert command Jan 27 18:32:13 <_valle_> I have added DEPENDS = "imagemagick" to the recipe but I'm getting convert: Command not found. What am I missing here? Jan 27 18:37:54 <_valle_> After installing imagemagick with apt-get to the host pc It works. Jan 27 18:38:12 <_valle_> But that shouldn't be necessary? Jan 27 18:50:14 probably want "imagemagick-native" if it exists Jan 27 18:50:36 otherwise add the pkg-class-native stuff to the recipe Jan 27 18:59:34 <_valle_> nerdboy: Thanks once again, imagemagick-native solved it Jan 27 19:00:19 <_valle_> If I would have added the pkg-class-native to the recipe, would that break the parts that does gets cross-compiled ? Jan 27 19:10:35 you don't need it if you already have it Jan 27 19:11:07 * nerdboy needs a damn haircut... Jan 27 19:16:06 joshuagl : Thanks for the feedback. I will look at the gnomebase.bbclass Jan 27 19:35:30 JaMa, Hi. Di you verify that sstate is working correctly for qt? Jan 27 19:38:11 JaMa, I have some strings in qmake which point to t different (nonexistent) sysroot Jan 27 21:40:17 I'm want to measure the response time of a the PR server while other stuff (builds, stress tests) happen on my system. Jan 27 21:40:43 the PR server doesn't have to run on the same machine, but yes, that doesn't sound like a bad idea Jan 27 21:40:57 I could loop running bitbake -f -c package foo but that's restarting the PR server repeatedly. Jan 27 21:41:11 only if you leave it configured to autostart Jan 27 21:41:15 you dont have to do that Jan 27 21:41:23 you can run it yourself and point all your builds to it Jan 27 21:41:40 I thought I'd run: bitbake -c devshell foo and but now I don't know how to execute the run.do_package python/bitbake code - tips? Jan 27 21:42:20 maybe richard's python devshell would be of use? Jan 27 21:42:25 kergoth: I'll run a global PR server once my mega-core server is available. :) Jan 27 21:42:45 you could still run it global to all your builds, but keep it on the same machine Jan 27 21:42:53 at least you avoid the restarting situation Jan 27 21:42:54 Does prserv always generate a new PR for each request or does it bundle that, e.g. using the recipe's hash? Jan 27 21:43:11 it goes based on checksum/signature change, yes Jan 27 21:43:34 it is possible for it to go backwards if you e.g. make a change and then revert that change, since it'd pull the previous value for the old cache, iirc Jan 27 21:43:39 OK, we should definitely have one soonish then... the PR game is driving me nuts. Jan 27 21:43:43 the problem being that for < 1% of our builds, the PRServer is timing out. Jan 27 21:43:56 vmeson: ah Jan 27 21:44:01 must be high load, i've seen that occasionally too Jan 27 21:44:28 quite irritating, but anything will keel over given sufficient load, really :) if it's taking more than 30 seconds to communicate over IPC to a local process, that's cause for concern :) Jan 27 21:44:37 and our timeout is 30 seconds -- we do push our builders pretty hard but I'm tired of just bumping the PR (and UI) timeout and I want to see what the response time distribution is. Jan 27 21:44:44 * kergoth nods Jan 27 21:45:01 so what's the about richard's python devshell -- shall I just google it? Jan 27 21:45:42 it's -c devpyshell, gives you an interactive python session, afaik, where you could run e.g. bb.build.exec_func() Jan 27 21:45:54 depends on what you're trying to do Jan 27 21:46:49 kergoth: thx. I've instrumented the client call to the PR server to get the response time, I figured I'd just run package repeatedly while another build is going on. Jan 27 21:47:09 cool, should be interesting Jan 27 21:47:24 I suspect that the root cause is that the PR server is occasionally flushing it's DB to disk and this is getting backed up behind other IO. Jan 27 21:47:52 ok, more to come, thanks again kergoth. Jan 27 21:49:42 initial (not very interesting) results: http://pastebin.com/extLU21F Under heavier IO load, I've seen 2-3 seconds latency but not 30 seconds yet. Jan 27 21:50:02 * vmeson heads home Jan 27 21:51:34 oh, the other idea (which is mine!) is to lower the priority / niceness of most processes so that the UI and PRServ get to run. Of course we'll see if it's an IO blocker first. **** ENDING LOGGING AT Thu Jan 28 02:59:58 2016