**** BEGIN LOGGING AT Tue May 17 02:59:58 2016 May 17 03:58:37 hi there, I'm pulling my hair with linux-yocto recipe May 17 03:59:20 I simply extended already existing .bbappend by adding a patch file. I know Yocto picks up the file because I tried to break it (both, file path and patch file contents). May 17 03:59:39 however after build is finished I searched whole Yocto dir for my .c file and there are only 2 occurences May 17 03:59:43 and both are unmodified May 17 03:59:48 so weird May 17 04:37:59 Xz: any way you can send this to me so I can try to reproduce it? May 17 04:44:58 Xz: yeah May 17 04:45:06 bluelightning: yeah, that's easy May 17 04:45:21 bluelightning: it's literally one patch and one layer May 17 07:22:31 good morning May 17 07:26:05 guys, how am I supposed to use 'yocto-kernel' script? it wants me to specify BSP May 17 07:26:30 I tried specyfing MACHINE and path to my bsp layer and many different configurations May 17 07:26:34 cannot get it to work May 17 07:48:21 Hi May 17 07:49:14 how do I tell toaster to accept repositories it apparently knows (e.g. meta-openembedded/*), but does not list them and denies to add them? May 17 07:49:43 I get "layer-exists-but-with-different-url" errors on the console log May 17 07:54:35 or to reword it: toaster refuses to add repositories, showing no error message in the web ui. It only logs to the console that the repository exists, but it is nowhere listed May 17 07:57:54 give the layer a different name? i.e. instead of meta-foo meta-foo1 ? May 17 08:06:22 I'm still fighting, looks like I have to use 'yocto-kernel' python script, except it breaks on my layer :o May 17 08:06:38 adding patches to kernel was easier back then May 17 08:06:48 SRC_URI was enough, right now it's not! May 17 08:07:10 would appreciate any yocto-kernel specialist May 17 08:08:36 there is a file in poky, ./scripts/lib/bsp/substrate/target/arch/x86_64/recipes-kernel/linux/files/machine-user-patches.scc May 17 08:08:50 that should contain kernel patches in right order May 17 08:23:43 looks like easier than adding a patch is to spin a new repo on github, patch it and feed that to Yocto... May 17 08:44:16 I guess the problem is that toaster's web-ui does not show 'remote' layers, if 'local poky' was selected at the beginning, but matches them when one tries to import layers May 17 08:56:09 that seems likely, I wonder if we're tracking that in bugzilla May 17 09:14:24 quite a few seeming related issues are filed; https://bugzilla.yoctoproject.org/buglist.cgi?quicksearch=toaster%20layer%20local&list_id=579624 May 17 09:30:01 Hey, I have a question to bitbake, I've built my first distro with yocto and I want to replace my actual build server with a yocto version but i forgot for whatever reasons to install all required python modules ... can I install them with easy_install or pip? May 17 09:31:10 is there a command which i can use to install all python modules required to run bitbake? May 17 09:32:07 for now I installed some manually but i still get an error: no module named future_builtins May 17 09:34:29 by the way i installed future already but i still get the error May 17 10:36:13 Hi guys! May 17 10:37:01 You gave me tips about the customization of the psplash recently. :) For the getenv(fbdev) and the colours with the cppflags. May 17 10:37:19 You told me that we can't use the FBENV variable... May 17 10:37:33 I'm on my way to add an option like the rotation one. May 17 10:37:53 Should I remove the getenv(FBENV) as you told me it was useless? May 17 10:40:03 rburton, you might know this one. :) May 17 10:40:23 Or Richard Leitner as he was the last commiter. May 17 10:41:47 present: its not useless, its just not trivial to configure May 17 10:46:40 rburton, the rotation way should work right? May 17 10:48:07 literally haven't looked at the psplash source for about a decade May 17 10:48:25 getenv(FBDEV) :) May 17 10:48:39 This is just that. May 17 10:50:46 Anyway this is just one line. I'll use is if the command line fbdev parameter is not given. May 17 10:50:53 *it May 17 10:53:19 rburton: is there a command which i can use to install all required python dependencies to run bitbake? May 17 10:55:46 cyball: bitbake just needs python May 17 10:56:36 rburton: well that was my thougts also but in fact id does need much more ... I figured out how to boot the image I've created yesterday, now i just installed it on my build server and tried to run bitbake on it May 17 10:57:12 rburton: with the result the i got a lot of named modules not found messages May 17 10:57:38 cyball: such as May 17 10:57:43 then i installed as much as i could but it does still show me the future_builtins import error May 17 10:58:48 I had to install multiprocessing, micropython_pty, tty, argparse, pysqlite, ... May 17 10:59:50 taking the first one, multiprocessing, that's part of the core python library May 17 11:00:06 so just install the whole python runtime May 17 11:00:15 http://pastebin.com/ZL6pVdqK May 17 11:00:47 rburton: python was already preinstalled version 2.7.11 May 17 11:00:58 clearly not if you had to install multiprocessing May 17 11:01:05 dpkg -S /usr/lib/python2.7/multiprocessing/__init__.py May 17 11:01:05 libpython2.7-stdlib:amd64: /usr/lib/python2.7/multiprocessing/__init__.py May 17 11:01:26 but the python command it self was availabe May 17 11:01:59 what distro? May 17 11:02:20 it was the core-image-minimal :) May 17 11:02:26 (debian and ubuntu split python to into a proper and minimal, maybe you've got the minimal one installed, which is only for the installer really) May 17 11:02:39 oh, you're trying to use bitbake inside an oe image May 17 11:02:43 install python-modules May 17 11:02:44 yes May 17 11:03:28 ok ... but can I can not for now because I installed it already on my hdd so i need to get it running for now :) May 17 11:03:46 because this is my build machine :) May 17 11:04:09 rebuild the image, adding python-modules to the image May 17 11:04:30 and all the other build dependencies for bitbake whilst you're at it, see the docs for the list May 17 11:04:54 is that in the bitbake docs? May 17 11:05:19 iirc packagegroup-self-hosted is the package group for all of the build dependencies of OE May 17 11:05:31 ok thx :) May 17 11:05:33 no, yocto docs, as i presume you want to do more than just run bitbake May 17 11:06:13 yes ... that's true ... ok i will give it a try thx for the help May 17 12:27:40 does anybody know if I want to use packagegroup-self-hosted if i can just skip the libgl stuff? or do i really need it? May 17 12:33:30 cyball: if you wont be doing anything with graphics, you can drop the whole packagegroup-self-hosted-graphics May 17 12:36:43 jku: how can i do that? May 17 12:40:24 cyball one way is that instead of adding packagegroup-self-hosted to your image you add the packages it depends on (except -graphics): "packagegroup-self-hosted-debug packagegroup-self-hosted-sdk packagegroup-self-hosted-extended packagegroup-self-hosted-host-tools" May 17 12:41:39 jku: ok, thx for that :-) May 17 12:48:12 jku: i still get the error of missing chain libgl even without packagegroup-self-hosted-graphics May 17 12:49:12 Hello everyone , i have built core-image-minimal successfully , i wanted to add an ssh-server-dropbear into the image using EXTRA_IMAGE_FEATURES =" ssh-server-dropbear" May 17 12:49:34 when i have added this feature "ssh-server-dropbear" , i can no longer login to the core-image-minimal using root and empty password May 17 12:49:48 any body knows what dropbear did to the root credentials ? May 17 12:52:03 cyball: and you are not adding packagegroup-self-hosted to the image anymore? May 17 12:52:46 jku: no but i have added at dev-pkgs in IMAGE_FEATURES does that makes sense? May 17 12:55:32 shouldn't hurt but shouldn't be needed for self-hosting yocto either May 17 12:55:53 jku: http://pastebin.com/pUh4tF5V May 17 12:56:37 jku: i guess it is easier if you can see the image content :) May 17 12:58:59 jku: http://pastebin.com/tAdDtSbC May 17 12:59:14 the last paste is the error message May 17 13:10:50 Hello everyone , when i start the core-image-minimal , at the login prompt , there is poky (Yocto project reference) 2.1 , how can i change these sentences May 17 13:11:02 cyball: it's saying packagegroup-self-hosted-debug depends on libgl. I have no idea how that could happen May 17 13:11:18 i want to make them something different May 17 13:12:17 jku: could you please help me ? May 17 13:12:24 snouto: you need a conf/distro/xxx.conf file in your layer May 17 13:12:37 what is xxx standing for ? May 17 13:13:00 snouto: you can choose what you want May 17 13:13:32 snouto: there are different DISTRO variables you can set there May 17 13:13:45 where can i find an example ? May 17 13:13:55 is there any example in poky for distro ? May 17 13:14:27 snouto: see meta-poky/conf/distro/poky.conf May 17 13:14:37 snouto: just have a look into meta-poky/conf/distro/poky.conf May 17 13:15:26 jku: i will try to just re initialize the build folder ... I'm curious if it could be the cache May 17 13:17:19 jku: ok i still get the error ... so i guess packagegroup-self-hosted-debug really needs libgl May 17 13:19:37 it's just a little hard to believe. You haven't modified packagegroup-self-hosted.bb right? May 17 13:19:43 jku: how can i make bitbake feels these changes , since when i run bitbake core-image-minimal , it did not compiled anything ? May 17 13:20:16 jku: no i didn't May 17 13:20:23 cyball: how can i make bitbake feels these changes , since when i run bitbake core-image-minimal , it did not compiled anything ? May 17 13:24:34 snouto: you can do that by cleaning the image cache like: bitbake -c cleansstate core-image-minimal but I'm not sure if you really need that or if it is enough to just clean the rootfs May 17 13:25:18 Hi. I hope I can find some advice here... I am trying to get started with Yocto and followed the setup instructions on a Ubuntu 14.04 host with poky krogoth (host gcc version is 4.8.4). But building core-image-minimal fails at the gcc-runtime_5.3 recipe in do_compile with the error "compatibility-c++0x.cc:215: undefined reference to `__dso_handle'". Does somebody know what is going wrong here? Unfortunately I have no clue at all h May 17 13:25:40 jku: if i remove the packagegroup-self-hosted-debug it still requires libgl and that is true for all packagegroup-self-hosted-* packagegroups May 17 13:25:55 jku: if i remove them all then it works but that is not what i wanted :) May 17 13:26:28 jku: so i guess the packagegroup-self-hosted bb file somehow has this requirement probably through the REQUIRED_DISTRO_FEATURES = "x11" May 17 13:28:12 hairyheron: have you run the cmd: sudo apt-get install gawk wget git-core diffstat unzip texinfo gcc-multilib build-essential chrpath socat libsdl1.2-dev xterm ??? May 17 13:28:31 cyball: that's not really how it works May 17 13:28:59 jku: ok, but it does not work for whatever reasons :) May 17 13:29:23 cyball: Yes, no problems there. When I run it again, it says everything is already the latest version. May 17 13:32:45 hairyheron: have you tried to do a apt-get update,upgrade,dist-upgrade ? May 17 13:35:08 cyball: Yes, and I did in fact receive an update to gcc-4.8 this morning, but it had no visible effect (I started a fresh compilation afterwards). So my system is up to date May 17 13:36:38 hairyheron: well this is probably a really dump question but have you tried to turn it off and on again? after the update? May 17 13:38:11 cyball: :D Hm, no. Okay... why not - let me do so now May 17 13:38:42 cyball: it didn't work May 17 13:39:06 jku,cyball: it didn't work after clearing cleansstate and core-image-minimal then bitbaking it again May 17 13:39:14 nothing happened , still , Poky ...etc May 17 13:39:27 i have added distro inside conf folder with a poky.conf inside May 17 13:39:46 snouto: well, then you can try to build it again ... I can remember that i had a similar issue but after a full rebuild it worked May 17 13:40:12 cyball: do i have to configure anything else before this step ? May 17 13:40:28 like adding a reference to the newly created folder "distro" anywhere May 17 13:40:37 or just removing tmp and sstate-cache May 17 13:40:42 then rebuilding everything again May 17 13:41:04 snouto: well, it is probably a good idea to just think whatever you want to do so if you plan to run network etc. it is probably not a bad idea to install resolveconf also May 17 13:41:23 what is resolveconf ? May 17 13:41:28 cyball: what is resolveconf ? May 17 13:41:42 snouto: ohh sorry i mean resolvconf May 17 13:41:58 cyball: yes , what is it and what does it do ? May 17 13:42:07 snouto: http://git.yoctoproject.org/cgit.cgi/poky/plain/meta/recipes-connectivity/resolvconf/resolvconf_1.48.bb?h=bernard May 17 13:42:43 cyball: Still the same, unfortunately May 17 13:44:25 hairyheron: that sounds not so good ... last time a red something about similar it was a gcc bug related to static linking problems i guess it would take a while to figure out why that happend to your system ... is the 10.04 your workstation or a build system only? May 17 13:45:15 hairyheron: you may need to upgrade to 16.04 or run a slower virtual instance as a quick fix May 17 13:46:52 It is my workstation. Alright... at least I apparently don't do anything completely wrong May 17 13:48:04 hairyheron: no i guess it is related to your installation May 17 13:53:19 cyball: Ok, no good news, but thank you very much for your support and your time - now at least I know what to do :) May 17 14:12:17 are there any (long term) plans to separate the concept of "image" from the concept of "rootfs" ? May 17 14:12:38 I find myself having to generate weird stuff for time to time where that concept would help May 17 14:12:56 like : a USB bootable image that contain the fs to install on the NAND May 17 14:18:34 Hi guys! May 17 14:19:22 I have a IMAGE_CMD_* function which is a sh function May 17 14:19:41 I want to run inside a function that is python and is provided by a bbclass. May 17 14:19:46 snouto, https://wiki.archlinux.fr/Resolv.conf May 17 14:19:54 Any idea how? May 17 14:26:36 alephan: I'm not sure (you should read the bitbake doc to be 100% sure) but ${@} might work May 17 14:26:53 though the python code will be run at parse time, not when the CMD is run May 17 14:27:02 Exactly May 17 14:27:07 That is only for variable assignment May 17 14:27:15 So it won't work for what I need May 17 14:28:57 alephan: can you do $(function_name)_append? Similar to do_install_append? May 17 14:29:44 I don't think I understand @tjamison May 17 14:29:49 how would that work? May 17 14:30:24 tjamison: no May 17 14:32:30 there's the do_install_append() function where you can add additional steps to the install process -- not sure if there are other analogous append() functions or which function you're looking to run inside of May 17 14:34:08 tjamison: Indeed. I am aware of that. What I'm asking is how to run in a sh task a python function? May 17 14:34:27 Or more generic, in a sh function a python function. May 17 14:35:37 ah, apologies, I think I missed part of the question May 17 14:36:15 It's fine. thanks. May 17 14:38:10 Hi all, I think I've found some weirdness with the gcc-cross built for wandboard using the meta-fsl-community layer. Maybe it's normal though? I was getting a failure to compile when pthreads.h couldn't be included. This was happening when trying to do the compile phase of a Cmake recipe in the case where the gcc-cross came from sstate populated by a build on another machine. May 17 14:39:32 I did -v on the compiler invocation to investigate and found that all the sysroot searches for the include were using the full path from the system that populated gcc-cross sstate, using its full paths not the path under workdir on this system. That's why pthread.h couldn't be found even though it was there. May 17 14:40:01 I am using a PRSERVER for all systems populating the sstate cache May 17 14:40:54 sounds like you forgot to pass --sysroot=, as we do in our CC variable May 17 14:43:10 also note that current oe sets the default sysroot to a path that doesn't exist, so you can't accidently break it like that now May 17 14:43:22 (at least i hope you're not using 2.1) May 17 14:43:48 yeah, i was wondering about that too, must be using an old release May 17 15:01:31 sorry, webchat. May 17 15:01:40 hi kergoth: and rburton: yeah I'm stuck on daisy in fact May 17 15:02:42 yeah that explains it May 17 15:02:47 kergoth: you say "forgot to pass --sysroot=" . You're thinking that the Cmake-based recipe could be updated? May 17 15:03:00 depends how you're passing CC May 17 15:03:08 :) yeah, I know... my hands are tied though. May 17 15:04:00 daisy's bitbake.conf has TOOLCHAIN_OPTIONS = " --sysroot=${STAGING_DIR_TARGET}" May 17 15:04:00 export CC = "${CCACHE}${HOST_PREFIX}gcc ${HOST_CC_ARCH}${TOOLCHAIN_OPTIONS}" May 17 15:04:42 thanks . I can ensure that CC is getting passed like that. May 17 15:04:43 but cmake is special/stupid and you have to pass flags separate from compiler May 17 15:04:58 but if your recipe inherit cmake then it should just work May 17 15:05:09 as that sets CMAKE_C_FLAGS correctly May 17 15:05:30 of course, the thing your build could promptly go and override that, had to fix a recipe that did that only last week May 17 15:05:40 I'll go check and confirm here May 17 15:05:41 so it sounds like it is a known daisy issue that the gcc-cross are built with the build system's full paths in its specs then? May 17 15:05:49 (ie the CMakeLists could set CMAKE_C_FLAGS directly) May 17 15:06:14 yeah post-daisy - can't recall when - we explicitly build the compiler with a sysroot like /this/path/does/not/exist so you *need* to pass the sysroot May 17 15:06:38 instead of it working assuming the path is the same as when it was built May 17 15:06:55 neat -- cool idea May 17 15:07:31 I tried a workarond here that was not such a great idea :) I added -I${TARGET_STAGING_DIR}/usr/include to TARGET_CFLAGS (except gcc-cross) May 17 15:07:55 the result was 100% sstate cache miss on any system with a different build absolute path :) May 17 15:08:30 I'm trying to understand how images are generated... I think I understand IMAGE_CLASSES IMAGE_TYPES and IMAGE_FSTYPES but I have one last question I can't find clearly in the documentation.. May 17 15:09:00 who is supposed to set IMAGE_FSTYPES ? is that a machine thing ? an image thing ? or is that something for me to choose ? May 17 15:09:12 rburton: actually, how does the later releases avoid that issue? if --sysroot= is passed and the dir is different on different machines doesn't that cause cache miss? May 17 15:09:55 because an image type contains stuff like "how to install u-boot" which should clearly be part of a machine description May 17 15:09:58 bengardiner_: no, the path to the sysroot is exempt from the hashes and hard-coded paths are generally worked around or editted out May 17 15:10:26 brilliant May 17 15:10:48 but a fstype is also decisions like "I want a /data partition that is 4Gb and is mounted read-write" which I could see going at the distro level or at the image level, depending on the use-case May 17 15:11:14 bengardiner_: you can verify/test that by just doing . oe-init-build-env foo ; bitbake core-image-sato ; . oe-init-build-env bar ; bitbake core-image-sato; and check that the second build is entirely sstate based May 17 15:11:32 wic is interesting in that in looks for a file named -.wks so it kinda agrees with my understanding May 17 15:12:43 rburton: definitely, and sstate hit rate to date proves that this is the case. Thanks for making that clearer for me May 17 15:18:10 cyball : do you know how can i package my custom core-image-minimal as .iso file to install it on vmware ? May 17 15:20:46 rburton: nice to see you again my friend May 17 15:21:34 snouto: have a look at: http://git.yoctoproject.org/cgit.cgi/poky/tree/meta/recipes-core/images/build-appliance-image_14.0.0.bb?h=krogoth May 17 15:21:39 rburton: core-image-minimal , turned out to include time limitation , the kernel will boot after 10 days , how can i get rid of this , when i go to yoctoproject.org/tlk , it tells me to read the documentation to know how to create get rid of this May 17 15:21:54 snouto: especially at the line MAGE_FSTYPES = "vmdk" May 17 15:22:09 snouto: you can generate a vmdk means vmware image :) May 17 15:22:17 yes May 17 15:22:25 cyball: thank you for this May 17 15:24:23 cyball: can you please tell me how can i get rid of this time limitation in core-image-minimal May 17 15:25:00 snouto: which time limitation do yo mean? May 17 15:25:32 cyball: core-image-minimal has time limitation , they say that the kernel will run for 10 days after that it will reboot May 17 15:25:40 cyball: http://www.yoctoproject.org/tlk May 17 15:28:38 snouto: i think that is ment to be valid for the intel specific embedded boards May 17 15:29:54 cyball: i am not using intel BSP layers altogether , but when i boot core-image-minimal in qemu , it shows me this banner "this image has time limitation , please refer to the website www.yoctoproject.org/tlk to get rid of this time limitation , never use this image in a product" May 17 15:31:36 if you'd actually read the page it links to, you'd realize it was the kernel that was time limited, not hte image, the fact that you're booting core-image-minimal is irrelevent May 17 15:32:42 kergoth: can you explain a little more May 17 15:32:57 kergoth: i am new comer to openembedded and yocto , what do you mean by this ? May 17 15:33:19 if you don't know what a kernel is, i'd advise reading some intro material on linux May 17 15:34:00 snouto: build your own image, that's a demo image. May 17 15:34:01 kergoth: i knew what is a kernel , of course , i knew , why are you so ironic May 17 15:34:58 snouto: if you did built your own image then remove the meta-tlk layer, and the question is where does it say to add that layer so we can fix whatever says to do that May 17 15:35:43 rburton : thank you , i will do that , i will remove meta-tlk layer from bblayers May 17 15:35:56 i just have added it there because i had to use psplash May 17 15:36:02 let me try modifying that and get back to you May 17 15:36:16 I get a taskhash mismatch -.- What's the reason and what's the solution? I tried to adapt https://www.mail-archive.com/yocto@yoctoproject.org/msg29001.html but that won't help. May 17 15:36:19 snouto: you don't need tlk for psplash, that's just an example of a layer that does fiddle psplash May 17 15:36:23 *while building an image May 17 15:36:33 snouto: the one thing tlk does is patch the kernel to be time locked :) May 17 15:37:15 rburton: man , you are very valuable member here , i would vote for you as the best helper at least for me :) May 17 15:37:30 rburton: take this digital kiss from me , Moooooah May 17 15:38:03 haha May 17 15:38:06 thanks :) May 17 15:38:51 kergoth: stay a side , let people who respect beginners like rburton to speak, one day , i will be better than you , believe me May 17 15:39:17 btw i don't think kergoth meant to be sarcastic May 17 15:39:40 that's true, yes, sorry if it came across that way May 17 15:40:08 rburton: btw. whats the point of tlk ? May 17 15:40:30 khem: so we can tell intel legal that we're not shipping binaries that can be used in production, they're for demo only May 17 15:40:41 heh thought so May 17 15:40:51 avoiding liability? May 17 15:41:36 yeah, forces people to build their own stuff instead of using the images as a starting point and pretending its redhat May 17 15:42:08 hehe May 17 15:43:08 not sure we should even bother shipping the binaries personally May 17 15:43:55 heh, its a different way of doing O_P_M May 17 15:44:44 i expect they're shipped to try to improve hte out of box experience, since a from scratch initial build is likely to take quite some time May 17 15:44:45 any input on that taskhas issue? :-/ May 17 15:44:49 though an sstate mirror can mitigate that May 17 15:45:36 a base android/brillo checkout is 34G May 17 15:45:39 wow May 17 15:46:00 yeah, next time someone moans about the size of yp tell them to do an android build May 17 15:46:16 heh :) May 17 15:46:49 I think OE is coming from other end May 17 15:47:13 where folks have one or two apps and base system more or less they dont modify it May 17 15:47:43 so they thing its waste of time for them to spend build cycles on the platform part May 17 15:48:07 but they need OE because size is important to them they dont want to build everything that debian would May 17 15:49:14 rburton: heh, to be fair, that checkout includes sources, whereas our builds fetch, so you'd ahve to include DL_DIR in the size :) May 17 15:49:35 i'll ignore that :) May 17 15:50:09 :) May 17 15:55:45 kergoth: but its serialized May 17 15:55:56 in OE case its parallel May 17 15:56:32 repo can do parallel clones, though May 17 15:56:37 not sure if they enable it in their manifest, though :) May 17 15:56:46 but that's just default,l you can enable it on the sync line May 17 15:57:42 thats right I think they have 4 threads May 17 15:59:03 kergoth: I did a du -sh on my download folder May 17 15:59:13 that sounds unpleasant May 17 15:59:17 i haven't done that in a while May 17 15:59:18 its 6G May 17 15:59:24 huh, not too bad May 17 15:59:36 for a QT5 image and a XFCE image May 17 15:59:44 20G /home/ross/Yocto/downloads/ May 17 15:59:58 how many machines do you build rburton May 17 16:00:07 with same downloads folder May 17 16:00:11 khem: 6g is only two copies of gcc, i don't believe you May 17 16:00:26 khem: mostly just x86 stuff and the odd qemuarm, but i haven't cleaned it for about two years May 17 16:00:35 the git2/ directory is 9g May 17 16:00:55 kraj@haswell ~OEROOT/downloads/git2/github.com.gcc-mirror.gcc May 17 16:00:57 % du -sh . May 17 16:00:59 2.4G . May 17 16:01:01 2.4G total May 17 16:01:30 rburton: this is clean build I did for angstrom yesterday May 17 16:01:41 its building two architectures May 17 16:01:48 arm7 and nios2 May 17 16:02:11 mine has been accruing downloads for a very long time, multuiple yocto releases, mel versions, etc, it's at 57G so far May 17 16:02:18 i expect there's some duplication between git2/ and git2_*.gz though May 17 16:02:23 thats more like it :) May 17 16:03:13 i hate to delete anything, never know when the upstream is going to vanish into the ether May 17 16:03:25 my core-image-minimal ( including build artifacts ) is less than 35G May 17 16:03:33 that's not bad May 17 16:04:16 android has these prebuilt SDKs May 17 16:04:33 and all of them are checked out in one go May 17 16:18:49 RP: I'm assuming shallow is waiting a while, which doesn't surprise me, but were there any objections to the other one-off bitbake patches? Specifically, http://patches.openembedded.org/patch/121277/, http://patches.openembedded.org/patch/121279/, http://patches.openembedded.org/patch/121283/, and http://patches.openembedded.org/patch/121285/ ? May 17 16:19:16 no rush, just wanted to see if there was anything I needed to do on those May 17 16:22:06 kergoth: those all seem reasonable to me offhand. The shallow one does worry me more, not found the time to think about it as much as I'd prefer to May 17 16:23:26 Yeah, I figured as much, they're invasive. Generally they won't affect anything unless the feature is enabled, but when it is enabled we lack the tests to cover all the possible states May 17 16:25:51 like i said, no rush, there's still plenty of time, just updating my document where i keep track of submission states :) May 17 18:38:24 Do the sharedstate and site.conf shared download directories handle concurrrent builds? Trying to optimize our build server times here. May 17 18:40:37 yes, it's completely safe to share DL_DIR and SSTATE_DIR between any of yoru builds, simultanious or otherwise, as lock files are used where needed May 17 18:40:57 Thank you, kergoth! May 17 18:58:52 QEGLPlatformContext: Failed to make temporary surface current, format not updated (3000) May 17 18:58:52 hmm May 17 19:39:46 what are packagegroups ? May 17 19:40:14 recipes that emit a package that has no content, only dependencies May 17 19:40:20 as the name would imply, it's a package grouping mechanism May 17 19:40:26 you can install the group to get everything the grou pdepends on May 17 19:40:28 i have read them through yocto development manual but still i couldn't understand it very well , is there any external tutorial or any resource to read about packagegroups in yocto ? May 17 19:40:35 it's not that complecated May 17 19:40:39 other distros call them metapackages or tasks May 17 19:42:21 kergoth: you knew , i had to modify distro folder inside meta-poky/conf/distro/poky.conf to change the distroName May 17 19:42:46 modifying an upstream layer isn't kosher, just make your own distro that includes poky.conf May 17 19:42:54 then changing the name is trivial May 17 19:42:56 kergoth: cuz cyball had told me to create a folder called distro inside my build conf directory May 17 19:43:19 when i bitbake core-image-minimal and run it onto qemu , it didn't take the change May 17 19:43:25 i had to change it in its original location May 17 19:43:40 lets backtrack: what do you want to do May 17 19:43:43 even bitbake -c cleansstate cor-image-minimal didn't do the trick May 17 19:44:09 rburton: i wanted to change the name "Poky (Yocto........)" May 17 19:44:20 header when the linux boots up to my product name May 17 19:44:44 for this , i had to ask people here in the room , cyball have told me to create a directory in my build conf called distro May 17 19:44:50 mybuild/conf/distro May 17 19:44:58 yeah that's wrong May 17 19:45:01 inside this , i have created a file called poky.conf May 17 19:45:12 if you want to make your own distro then make your own layer (yocto-layer will write the boiler plate), add it to bblayers.conf, and in there put a whatevermydistroiscalled.conf May 17 19:45:15 rburton: what is the proper way May 17 19:45:22 that file can be the contents of poky.conf if you want May 17 19:45:34 then in local.conf set DISTRO=whatevermydistroiscalled May 17 19:46:03 http://www.yoctoproject.org/docs/2.1/mega-manual/mega-manual.html#creating-your-own-distribution <— like this May 17 19:46:05 yocto-layer create local 1; bitbake-layers add-layer meta-local; mkdir -p meta-local/conf/distro; echo 'require conf/distro/poky.conf' >>meta-local/conf/distro/snouto.conf; echo 'DISTRO_NAME = "foo"' >>meta-local/conf/distro/snouto.conf; echo "LAYERDEPENDS_local = \"core poky\"" >>meta-local/conf/layer.conf; echo 'DISTRO = "snouto"' >>conf/local.conf May 17 19:46:16 ah, yes, referencing the docs is best, read that :) May 17 19:47:27 if you just copy poky.conf to your own conf then you don't need to have meta-poky around at all, and get to delete all the bits from poky.conf that you don't care for May 17 19:47:43 its your distro, poky.conf is just an example May 17 19:48:40 kergoth: i really thank you so much , i really appreciate your help and also rburton May 17 19:49:16 you know my friends , i am creating a new product May 17 19:49:26 a honeypot , where deception becomes defence May 17 19:50:07 i am a security big data architect , working with hadoop and apache spark , apache storm , for threat intelligence management May 17 19:50:28 i was always dreaming to understand building custom linux on embedded systems , that is my ultimate dream May 17 19:50:47 until recently , i got the courage to start this journey to study it May 17 19:51:35 i am coordinating with a chinese company to produce for me the network appliance May 17 19:52:14 and i should create the embedded linux distro customized to only this purpose , then i will have to harden it by selinux packages or recipes May 17 19:52:33 i think now , you understood what i am up to May 17 19:52:55 there's a selinux layer so that should be easy enough May 17 19:53:08 rburton: exactly May 17 19:53:22 time to eat May 17 19:54:04 kergoth: do you know how can i build an installer for my linux distro that allow me to install it directly on my appliance like ubuntu server , centos server....etc May 17 19:55:57 the 'live' image type provides that, either as a disk image fro a usb card or a .iso May 17 19:56:05 not very familiar with it personally May 17 19:56:35 live image ? May 17 19:56:45 do you mean bitbaking live instead of core-image-minimal ? May 17 19:58:02 kergoth: in your previous shell code May 17 19:58:07 you are saying May 17 19:58:19 yocto-layer create local May 17 19:58:21 and then May 17 19:58:30 bitbake-layers add meta-local May 17 19:58:35 are they the same thing ? May 17 20:09:32 rburton: kergoth: thanks for you tips this morning. The CMakelists wasn't using CC but was using CFLAGS; so I fixed the recipe with TARGET_CFLAGS_prepend = "${TOOLCHAIN_OPTIONS} " May 17 20:13:45 snouto: ? May 17 20:13:57 snouto: that question doesn't make any sense. they're two different commands May 17 20:14:16 snouto: and no, i mean the live image type. go se ethe yocto project documentation about IMAGE_FSTYPES May 17 20:18:38 kergoth: it gives me error , ERROR: Layer 'local' depends on layer 'poky', but this layer is not enabled in your configuration May 17 20:18:51 i have followed your steps one after the other May 17 20:19:21 likely need to either change that from poky to yocto by editing meta-local/conf/layer.conf, or remove the LAYERDEPENDS and copy poky.conf rather than including it, the way rburton advised May 17 20:19:48 forgot that while we renamed meta-yocto to meta-poky in the filesystem, the name in its layer.conf is still yocto May 17 20:21:23 i followed the first part of your solution by renaming poky into yocto , and it is working fine now May 17 20:21:39 LAYERDEPENDS_local= "core yocto" May 17 22:47:18 Hi May 17 22:47:29 Someone awake around? :) May 18 01:42:22 something I pulled in from yocto/oe late last week or early this week has triggered a spamfest of these msgs.... May 18 01:42:27 sh: warning: setlocale: LC_ALL: cannot change locale (en_US.UTF-8) May 18 01:42:34 anyone else seeing that? **** ENDING LOGGING AT Wed May 18 02:59:58 2016