**** BEGIN LOGGING AT Fri Jul 31 02:59:59 2015 Jul 31 08:04:40 Hello Jul 31 08:05:50 Is this supposed to be a help channel or developers-only? I've got trouble with receipes trying compile image for my dev board (missing orc and bluez5-bluez4 conflict) Jul 31 08:51:06 Hi ! i solved my problem for a custom-xlnx kernel. There is no problem anymore on the recipe. Except that now the version taken is not the good one. It's back to the default (3.14) instead of mine (3.8) i settet LINUX_VERSION = "3.8" and KBRANCH ?= "xlnx_3.8" in my recipe and this in my local.conf : PREFERRED_PROVIDER_virtual/kernel = "linux-xlnx" and PREFERED_VERSION_linux-xlnx = "3.8". Somewere the version seems to be overrided. But i Jul 31 08:51:40 do you have any ideas ? Jul 31 08:52:01 hello, someone may have a sample makefile, which is running with a simple recipe? Jul 31 08:53:38 sorry i don't understand :/ you mean that the version might be hardcoded in a makefile or something ? Jul 31 08:54:29 morning all Jul 31 08:55:25 vtech: you meant PREFERRED not PREFERED right ? Jul 31 08:56:07 moin Jul 31 08:56:07 vtech: run bitbake -e linux-xlnx | less and search for LINUX_VERSION - see if it's being overridden Jul 31 08:56:43 vtech: you probably want 2 'R' at PREFERRED ;-) Jul 31 08:56:44 oh there is indeed PREFERED_VERSION_linux-xlnx = "3.8" in my local.conf, i test it with the correction right now >< Jul 31 08:57:57 karotin: the user manual has that. Jul 31 08:57:57 sometimes I think it would be nice if we could detect such typos, but I don't think it's practically possible with how our system works Jul 31 08:58:31 bluelightning: we could integrate aspell in bitbake ;-) Jul 31 08:58:41 Ok i have another output, there was indeed a problem here , sorry >< . It tries to take my version but : NOTE: preferred version 3.8 of linux-xlnx not available (for item kernel-image) seems to be the problem. Jul 31 08:59:03 i am using linux-yocto.inc but i don't thkink linux-yocto has a 3.8 kernel :/ Jul 31 08:59:43 vtech: that in and of itself shouldn't be a problem, as long as you are pointing at your own source Jul 31 09:00:38 vtech: if you made your own layer, did you add it to bblayers.conf? Jul 31 09:01:59 yes i did. And i am using the xlnx repo on the 3.8 BRANCH as specified by the KBRANCH variable. Jul 31 09:02:26 the stange things is that it seems to see the 3.8 as available and then it's not : NOTE: preferred version 3.8 of linux-xlnx not available (for item kernel-image) NOTE: versions of linux-xlnx available: 3.14-xilinx+gitAUTOINC+2b48a8aeea 3.8-xilinx+gitAUTOINC+297a37ee40 Jul 31 09:04:05 vtech: ah right, to match that you need to set PREFERRED_VERSION_linux-xlnx = "3.8%" Jul 31 09:04:12 (% is a wildcard in PREFERRED_VERSION) Jul 31 09:06:21 bluelightning ndec: That solved it :) Thanks again for you help :) Jul 31 09:06:28 np Jul 31 09:19:46 <[Sno]> bluelightning: wrt. two machines in one build-dir Jul 31 09:19:56 <[Sno]> $ MACHINE=bohr bitbake rdm-core-image Jul 31 09:19:56 <[Sno]> ERROR: Unable to determine endianness for architecture 'INVALID' | ETA: --:--:-- Jul 31 09:19:56 <[Sno]> ERROR: Please add your architecture to siteinfo.bbclass Jul 31 09:19:56 <[Sno]> ERROR: Failed to parse recipe: /home/sno/fsl-community-bsp/sources/meta-jens/recipes-extended/libstatgrab/libstatgrab_git.bb Jul 31 09:19:57 <[Sno]> ERROR: Unable to determine endianness for architecture 'INVALID' Jul 31 09:19:59 <[Sno]> ERROR: Please add your architecture to siteinfo.bbclass Jul 31 09:22:27 [Sno]: and that definitely does not happen if you set MACHINE = "bohr" in local.conf ? Jul 31 09:22:37 <[Sno]> bluelightning: yes Jul 31 09:22:56 [Sno]: is this with master? Jul 31 09:23:34 <[Sno]> bluelightning: I'm using that local.conf: https://github.com/rehsack/yocto-curie-workdir/blob/fido/conf/local.conf Jul 31 09:23:46 <[Sno]> and for the error I commented out the first line ;) Jul 31 09:23:53 <[Sno]> we have two machines, curie and bohr Jul 31 09:24:46 [Sno]: what does bitbake -e | less report about how MACHINE is being set? Jul 31 09:24:55 <[Sno]> and even in https://github.com/rehsack/meta-jens/blob/fido/recipes-extended/libstatgrab/libstatgrab_git.bb I don't see anything which should cause such error Jul 31 09:25:44 <[Sno]> bluelightning: PATTERN not found Jul 31 09:26:13 erm ... ? Jul 31 09:26:26 ndec: http://pastebin.com/gpKDtU3c still cant't find the error in the makefile..I'd like to compile a driver Jul 31 09:26:49 <[Sno]> I tried to add "MACHINE = ${@os.getenv('MACHINE', 'nope')}" to local.conf - and it reports "nope" Jul 31 09:27:20 karotin: for kernel module it's slightly different. one sec. Jul 31 09:27:32 <[Sno]> bluelightning: sno@waldorf:~/fsl-community-bsp/ornithologie$ MACHINE=bohr bitbake -e | grep MACHINE= Jul 31 09:27:32 <[Sno]> sno@waldorf:~/fsl-community-bsp/ornithologie$ Jul 31 09:28:34 karotin: http://git.yoctoproject.org/cgit/cgit.cgi/poky/tree/meta-skeleton/recipes-kernel/hello-mod/hello-mod_0.1.bb Jul 31 09:29:06 <[Sno]> same when grepping for "bohr" with above bitbake command Jul 31 09:29:23 ndec: the only thing i see, what is different, is the "inherit module" Jul 31 09:29:38 [Sno]: firstly, the environment is sanitised so that os.getenv isn't expected to work; secondly MACHINE is unexported so that grep isn't expected to work either Jul 31 09:30:06 [Sno]: you really do want to use bitbake -e | less and search, not grep Jul 31 09:30:56 <[Sno]> bluelightning: I used less and "/" - that results in "pattern not found" ;) Jul 31 09:31:54 [Sno]: MACHINE isn't in there at all? I find that hard to believe... Jul 31 09:32:11 <[Sno]> I can upload the output of bitbake -e Jul 31 09:33:18 <[Sno]> # Jul 31 09:33:18 <[Sno]> # $MACHINE [2 operations] Jul 31 09:33:18 <[Sno]> # set /home/sno/fsl-community-bsp/sources/poky/meta/conf/documentation.conf:271 Jul 31 09:33:18 <[Sno]> # [doc] "Specifies the target device for which the image is built. You define MACHINE in the conf/local.conf file in the Build Directory." Jul 31 09:33:18 <[Sno]> # set /home/sno/fsl-community-bsp/sources/poky/meta/conf/bitbake.conf:760 Jul 31 09:33:20 <[Sno]> # [unexport] "1" Jul 31 09:33:22 <[Sno]> # pre-expansion value: Jul 31 09:33:26 <[Sno]> # "None" Jul 31 09:33:28 <[Sno]> unset MACHINE Jul 31 09:33:47 ndec: looks good :) Jul 31 09:35:32 [Sno]: ah... in your external environment as set up by the environment setup script, does the value of BB_ENV_EXTRAWHITE include MACHINE? (it should by default) Jul 31 09:35:59 <[Sno]> bluelightning: I check that ... Jul 31 09:40:07 <[Sno]> bluelightning: it doesn't - we initially used such an environment created by fsl-release-bsp which broke more than it helped and we removed most of that without sane replacements Jul 31 09:40:33 <[Sno]> I dig deeper on that - maybe that solves the 2nd problem I had with different machines in one build-dir ;) Jul 31 09:40:39 [Sno]: ah, well, there's your problem then (well, at least an explanation of why it doesn't work) Jul 31 09:41:06 [Sno]: as I said earlier though there's nothing saying you can't just set the other machine in local.conf each time instead of passing it in from the environment, if you prefer Jul 31 09:41:36 <[Sno]> the BB_ENV_EXTRAWHITE explains a lot ;) Jul 31 09:43:32 <[Sno]> thanks a lot Jul 31 09:44:25 <[Sno]> rdmtob: I miss the PR for meta-cpan :) Jul 31 09:44:46 <[Sno]> I wanted this afternoon update some modules and add support for Tim's Slic3r wish Jul 31 09:45:50 Slic3r support? does that make even sense on embedded devices? Jul 31 09:46:58 <[Sno]> karotin: depends on what you embed where ;) Jul 31 09:47:32 <[Sno]> car infotainment is also embedded device and BMW uses Yocto for there infotainment platform Jul 31 09:47:57 telecommunications infrastructure is another oft-cited counterexample Jul 31 09:48:02 <[Sno]> my customer put's an xbmc on it's device paired with a bunch of Java services Jul 31 09:49:26 k :) Jul 31 12:24:37 Hi ! I am still trying to build my 3.8 kernel for xilinx , i have a problem with the meta. I have several .scc and .cfg in my layer and i am trying to use several .scc and .cfg from meta-xilinx layer. I think there is a conflict on that. as the error shown is : | ERROR. Could not locate meta series for zynq in the do_patch function. am i right on the cause of the problem ? Jul 31 12:30:18 joshuagl: ohai :) Jul 31 12:33:45 * joshuagl doffs cap Jul 31 12:41:51 If a .inc uses the variable ${THISDIR} , how can i use it from another folder without having conflict ? Jul 31 12:50:24 vtech: you may be getting into a situation where the .inc file isn't going to work for what you're trying to do Jul 31 12:55:06 how to write a do_compile for a simple c++ program with dependencies? i can compile the main.cpp now, but i faile to make a dependecie on clas.cpps.. Jul 31 12:55:12 ${CXX} ${CXXFLAGS} ${LDFLAGS} ${WORKDIR}/main.cpp -o ttkeys Jul 31 12:57:35 bluelightning: I think it is related to the usage of ${THISDIR} in the .inc which forbid it's use from anywhere else :/ overinding the THISDIR variable might do the job but this is the ugliest thing to do no ? :/ Jul 31 12:58:38 vtech: how is ${THISDIR} being used ? Jul 31 13:00:57 bluelightning: this way : FILESEXTRAPATHS_prepend := "${THISDIR}/linux-xlnx:" but the folder pointed is containing patches for 3.14 and 3.19 kernel which i don't use. I can't figure out why the meta xilinx cannot be found Jul 31 13:01:41 bluelightning: and the second one , which is , i think, the cause of the problem : FILESEXTRAPATHS_prepend := "${THISDIR}/config:" Jul 31 13:03:12 bluelightning: in the config folder (located inside the meta-xilinx layer) there are the base configs for the board. in a config folder located in my layer i have my .cfg and . scc. it seems that the meta paths are lost and do_patch fail :/ Jul 31 13:04:09 vtech: FILESEXTRAPATHS just specifies additional paths where do_fetch should look for for items specified in SRC_URI Jul 31 13:04:43 vtech: you can prepend your own paths there, or you can not have the items you don't want in SRC_URI, depending on what it is you need Jul 31 13:05:57 bluelightning: yes, that's why i started using this .inc but how can the base meta be lost. i am just appending my files with : FILESEXTRAPATHS_prepend := "${THISDIR}/config:" (this one is in MY layer). So the meta i added should be found no ? Jul 31 13:08:19 vtech: if by "meta" you mean local files specified with file:// in SRC_URI, yes Jul 31 13:11:40 bluelightning: In the .inc , the THISDIR variable is refering to my recipe location or the .inc location ? because they are in separated layers. It might be the problem . I know THISDIR is relative to the recipe currently being parsed. But what about a .inc ? Jul 31 13:16:41 vtech: I believe THISDIR will point to the path in which the current file is being parsed, and the current file at that point would be the inc file Jul 31 13:19:26 bluelightning: hmm seeing the error i'm getting i would say the opposite.. are you sure about that ? Jul 31 13:20:58 vtech: well, you can easily confirm what's really happening by using bitbake -e recipename | less and searching for FILESEXTRAPATHS Jul 31 13:23:10 bluelightning: indeed, you are right. it is relative to the .inc location. I have no clue about the error so.. thanks for you help once again.. ^ Jul 31 13:23:43 vtech: additionally you should be able to see in log.do_fetch where it is looking for each item Jul 31 13:29:24 bluelightning: every path is correct... i have no idea why the meta xilinx cannot be found.. Jul 31 13:29:58 vtech: what is the exact error you are receiving ? Jul 31 13:30:38 bluelightning: the error is : | ERROR. Could not locate meta series for zynq Jul 31 13:31:36 vtech: are you pointing at a repository that has a linux-yocto style meta branch + source branch, or only a source branch? Jul 31 13:33:23 Hi ! I would like to disable c++ languages during the build of gcc-runtime . So I override LANGUAGES by "c" and disable "--disable-libstdcxx-v3" on the EXTRA_OECONF. But it fail during gcc-runtime configuration . Somebody knows this error : "checking for exception model to use... configure: error: unable to detect exception model " Jul 31 13:33:46 bluelightning: i think it's a source only. But i'm using the the xlnx .inc and variables which are able to build a 3.14 and 3.19 kernels from a source only repository. (https://github.com/Xilinx/linux-xlnx/tree/xlnx_3.8) Jul 31 13:34:37 * zeddii reads Jul 31 13:35:48 * zeddii actually just improved the error messages in that area. Jul 31 13:35:55 vtech: what release are you using ? Jul 31 13:38:23 i'm using yocto fido Jul 31 13:38:38 zeddii: i'm using yocto fido Jul 31 13:39:31 ok. so my latest changes in master don't come into play .. that's good. but yes, chances are it isn't finding one of the references files and failing to generate a meta-series (which is the messages I've been working on). Jul 31 13:39:41 is any of this somewhere you can pastebin, and I can see it ? Jul 31 13:40:06 I'm heading out on vacation tomorrow, so have little time to get set up. but if I can jump start a debug of it, I can probably point it out quickly. Jul 31 13:40:22 zeddii: you want the error log ? Jul 31 13:40:40 it won't be helpful. but sure. it doesn't have much information, that's what I've been improviing. Jul 31 13:41:22 if you can find the log for the do_configme and do_patch phases, something may be lurking in there to point out the exact file it couldn't find. Jul 31 13:41:40 zeddii: ok, i'm posting that right now on a pastbin Jul 31 13:44:29 otavio: i'm using meta-java and trying to get a decent JAVA_HOME for my lib using openjdk7/8. Was hoping java.class would provide this, any tips? Jul 31 13:45:49 mario-goulart: also ^ Jul 31 13:46:33 zeddii: here is the do patch log : http://pastebin.com/8ufjqALZ but i have no log in the temp folder for do_kernel_configme Jul 31 13:47:08 zeddii: are you talking about the do_kernel_metadata instead ? Jul 31 13:47:21 yah. that's where I moved it. old habits :) Jul 31 13:47:38 ok i'm posting it too Jul 31 13:48:06 yah. as expected zero of value in http://pastebin.com/8ufjqALZ Jul 31 13:48:14 that's why I made some logging more verbose. Jul 31 13:48:27 zeddii: here is the metadata log : http://pastebin.com/0pHVyLHb Jul 31 13:49:41 ok. so that shows us something. Jul 31 13:49:58 bear with me if I ask you qusetinos that bluelightning already did. I didn't churn through the history. Jul 31 13:50:27 xilinx-drivers-linux-xlnx.scc, that is what you are trying to add ? Jul 31 13:50:41 * zeddii updates his meta-xilinx Jul 31 13:51:04 * zeddii sees that in the layer Jul 31 13:51:42 zeddii: no this one is in the xlnx layer Jul 31 13:52:45 i'm trying to add : framebuffer.scc and wifi.scc (they both have a corresponding .cfg and are located in meta-adeneo/recipes-kernel/linux/config/adeneo/features Jul 31 13:54:02 that .scc file is wrong in the layer. they don't actually reference it in any of the configs in meta-xilinx, and it is in fact referring to a file that doesn't exist. Jul 31 13:55:04 xilinx-ip-linux-xlnx.cfg isn't anywhere to be found ... are you saying that you are creating or importing that via another layer ? Jul 31 13:55:10 arfoll: I'd suggest to take a look at some lib recipes in meta-java Jul 31 13:55:11 it does exist, i added it in my layer. and used SRC_URI_append += " \ file://adeneo;type=kmeta;destsuffix=adeneo \ " to reference it. Jul 31 13:55:52 no i don't use xilinx-ip-linux-xlnx , it is not mine... Jul 31 13:55:57 is that the one causing the issue ? Oo Jul 31 13:56:08 yep. Jul 31 13:56:09 * ant_work thinks that layer is the only one using type=kmeta in SRC_URI Jul 31 13:56:19 ant_work, nope. all of master is now :) Jul 31 13:56:22 linux-yocto* Jul 31 13:56:55 how can the xilinx layer build a 3.14 kernel if there are missing files.. Oo Jul 31 13:57:06 yep but only in the master recipe, not in further appends/inc iirc Jul 31 13:57:14 vtech, because they don't actually reference that fragments. Jul 31 13:57:26 s/fragments/fragments/ Jul 31 13:58:19 when I grep the fido branch of meta-xilinx, I get zero hits on xilinx-drivers-linux-xlnx.scc Jul 31 13:58:42 ant_work, aha. I see. I do that locally in a meta-local, just to be sure it works. Jul 31 13:58:56 ok.. but neither i do.. well i thought so Jul 31 13:59:46 it is worth pinging the meta-xilinx list, it is probably just an orphan .. but not using it in your work is probably a good thing as well. Jul 31 13:59:52 zeddii: fwiw I have still to study the very last changes Jul 31 13:59:54 " After this change, linux-yocto can no longer process combined trees, Jul 31 13:59:54 and is simplified as a result." Jul 31 14:00:33 no more meta branch to cause confusion :) Jul 31 14:01:00 but i never reference it directly from my recipe Jul 31 14:03:51 looking at the fido branch of the xilinx repo i find both the .scc and the .cfg Jul 31 14:04:11 oh no my bad Jul 31 14:04:18 i made a mistake , there is nothing indeed. Jul 31 14:05:54 mario-goulart: do you have in particular? I see the inherit java-library seems to be the main thing but I dont see how that exports JAVA_HOME location which I need for my code to build Jul 31 14:09:24 vtech. something, somewhere must be referencing that .scc file if it is coming into the mix though. I'm not seeing it. hmm. Jul 31 14:12:57 zeddii: arf my last day of internship , so sad :p i'm looking for it too right now :) Jul 31 14:13:42 i have a reference to it in : meta-xilinx/recipes-kernel/linux/config/xilinx-base/bsp/xilinx/xilinx-drivers-linux-xlnx.scc Jul 31 14:14:03 zeddii: well it's referencing the cfg file Jul 31 14:14:15 zeddii: time permitting I'll try again to create some macro-groups of fragments. I.e. fs compiled in plus block devices to boot from anywere Jul 31 14:14:35 exactly. but unless a SRC_URI references that .scc, or another one includes it. it won't come into the build, and I'm not seeing that reference. Jul 31 14:15:30 yeah, me neither :/ Jul 31 14:16:00 if the layers or recipes were public, I could track it down. but looking at what I have to go on, I'm not seeing it. Jul 31 14:16:15 Hi All... I want to override a do_configure_prepend function in a .bbappend recipe. This does not seem to work. It will call both my do_configure_prepend and the one from original recipe. Jul 31 14:16:19 like anything there's no magic. a .scc file has to be referenced someewhere to be used. Jul 31 14:16:22 How do I do this ? Jul 31 14:18:54 zeddii: i'm currently searching my whole HDD for the file Jul 31 14:19:10 zeddii: well no match.. i'll give up maybe :/ Jul 31 14:20:13 * zeddii wants to know where that is coming from! Jul 31 14:20:29 zeddii: yeah me too >< i'm still searching Jul 31 14:20:33 if you do a bitbake -e and dump it to a file. Is it someone on the SRC_URI ? Jul 31 14:20:59 s/someone/somehow/ Jul 31 14:21:06 i tell you in a second Jul 31 14:23:16 not a single reference :/ Jul 31 14:23:37 how is that possible ? Jul 31 14:24:22 having written all the code that processes those files. there's nothing that would go out and reference it by itself. barring a horrible bug of course. Jul 31 14:24:39 vtech. can you go look at the meta-series itself ? it's in the kernel source directory Jul 31 14:25:03 tmp/worked-shared//kernel-source/.meta/meta-series Jul 31 14:27:08 i can't find any .meta folder.. i'm in /tmp/work/zc702_zynq7-poky-linux-gnueabi/linux-xlnx/3.8-xilinx+gitAUTOINC+297a37ee40-r0/linux-zc702_zynq7-standard-build Jul 31 14:27:26 that's the build Jul 31 14:27:37 the source is in split in fido to work-shared/ Jul 31 14:27:42 oh yes work-shared sorry << Jul 31 14:28:41 there is no file or folder names meta-series in .meta, i'm looking for it Jul 31 14:30:17 zeddii: there is nothing named like that in .meta Jul 31 14:30:46 in .meta, what do you see ? a file called top_tgt ? Jul 31 14:31:07 since that failure was in the generation of the meta series it won't be thre Jul 31 14:31:14 my bad for not being clear. Jul 31 14:31:54 yes there is top_tgt , no problem :) Jul 31 14:32:36 cat that file. where does it point ? Jul 31 14:34:17 i can't put it here >< don't know why here is the pastbin : http://pastebin.com/ZVmHSwJm Jul 31 14:36:36 vtech. ok. so that file is the start of everything. it includes other .scc files (and they can do the same), or fragments. if you open that file, and chase everything it includes, something must be yanking in that bad .scc file. if it isn't there, it is either via SRC_URI or KERNEL_FEATURES. exhaust those searches and it has to be found. Jul 31 14:37:17 o i'll look for that :) Jul 31 14:37:21 thaks :) Jul 31 14:37:24 thanks * Jul 31 14:46:44 arfoll: oh, sorry. So you _need_ JAVA_HOME set to build you library. Can't you just use the STAGING_... variables as defined in java.bbclass? Jul 31 14:51:28 morning Jul 31 14:54:59 vtech: found it? it seems config/xilinx-common/bsp/xilinx/soc/zynq.scc Jul 31 14:55:36 * ant_work lurking interested about new ways to (ab)use of fragments in the metadata Jul 31 15:17:25 mario-goulart: I could use those to translate I was just thinking there would be something pretty much good to go for people using JAVA_HOME Jul 31 15:26:06 RP: stupid question, are mirrors & sstate mirrors replacements applied recursively? Jul 31 15:26:27 kergoth: yes, I think so Jul 31 15:26:47 kergoth: I do have some sstate mirror pieces locally I keep meaning to submit... Jul 31 15:27:49 K, I figured that was probably the case. One related question: can it recurse infinitely? I'm assuming the same pattern won't be re-applied in the recursion, just all the others, but in theory two patterns could keep appending to one another recursively, forever Jul 31 15:27:57 thats just a thought, obviously haven't tested such a thing Jul 31 15:28:37 happened across your 'Add basic sstate dsitro version mapping for file urls' and was wondering if the distro fallbacks would recurse, so each new distro only has to fall back to the previous, not to all previous versions explicitly Jul 31 15:28:45 sounds like the answer is yes Jul 31 15:29:11 kergoth: yes, I did that deliberately. There are simplistic checks for recursion Jul 31 15:29:31 kergoth: I don't doubt someone could make it break Jul 31 15:29:47 k Jul 31 15:29:58 figured as much :) Jul 31 15:29:59 thanks Jul 31 15:30:15 hackers I have a question on how to structure a derivitative machine Jul 31 15:32:44 I have tweaked the beaglebone.conf file and added and include to the linux-yocto kernel recipe to add some features, and everything works well. Should I now create my own bsp layer using the yocto-layer create and move those changes in there, with my own machine.conf file, to keep them separate from meta-yocto-bsp layer? Jul 31 15:34:40 this is for internal use on a beablegone prototype, just wondering what the best way to package changes are for a machine that is almost the same as stock beagleboard but with a few minor additions Jul 31 15:59:02 peaceful here Jul 31 16:00:11 creating your own bsp layer sounds fine. if you wanted to, you could use a different machine name and have your own machine.conf require the other one, to avoid duplicating content there, and then use MACHINEOVERRIDES to that both your override and the original machine override apply Jul 31 16:00:27 we've done that before, e.g. mel-omap5-evm instead of omap5-evm Jul 31 16:01:12 that sounds cleaner, thanks Jul 31 16:02:23 e.g. require/conf/machine/.conf in the machine.conf, yous hould be able to grep to see examples of MACHINEOVERRIDES, but it's just a section of OVERRIDES, colon separated, with the last winning over the former. so e.g. MACHINEOVERRIDES = ":" Jul 31 16:02:41 that might not be needed, but it's a good idea just in case the original override is used anywhere, so you don't miss anything Jul 31 16:03:52 i think that sounds good. How would I handle the kernel fragment that I included in the kernel recipe for this machine? Jul 31 16:04:51 you could bbappend the kernel recipe to add your fragment, in your bsp layer Jul 31 16:05:06 or provide your own kernel recipe, really up to you, whatever works best Jul 31 16:05:25 I see. I will give this all a try. Thanks! Jul 31 16:07:53 np Jul 31 16:08:29 you'll discover there are lots of ways to just about everything, and only sometimes are there conventions over what's best, often it's case-by-case, and the project covers a lot of use cases, so you end up going with whatever is best for your situation Jul 31 16:10:15 Hi, I have a question about using qemu to generate some file in a recipe (like emacs recipe). I used the same template than the old emacs recipe and I get the same segfault error Jul 31 16:11:18 I investigated and it seems that qemu is loading the host libc and not the target libc, thus the segfault Jul 31 16:12:56 So my idea is that the emacs recipe is doing it the wrong way. Does someone now another recipe with the same mechanism that use qemu to run a program compiled for thetarget platform ? Jul 31 16:14:37 afaik any recipes needing to do that should be inheriting qemu, using qemu_run_binary(), and DEPENDS on qemu-native Jul 31 16:18:41 thanks, I think that would help me a lot Jul 31 16:19:30 RP: am i missing something, or is it impossible to add back a task which was removed with deltask? i think either it needs to be able to add it back (ideal) or the addtask command should result in some form of failure, rather than just doing nothing Jul 31 16:20:35 kergoth: I suspect its an oversight Jul 31 16:20:48 kergoth: I've never tried that Jul 31 16:21:56 I was wanting to experiment with appending a recipe and making it no longer build, but instead e.g. pull binaries. Since in the general case I don't know what tasks have been added between patch and install, i don't want to just override the configure/compile task content, it'd be ideal to wipe and reconstruct part of the task graph :) Jul 31 16:40:58 arfoll: what is the problem you are trying to fix? the javs should work out of box Jul 31 17:40:07 I'm attempting to create a package for use with meta-ros which encapsulates a catkin package. When I add my package to IMAGE_INSTALL in my image it install correctly; however when I try to install the generated RPMs directly it doesn't install the files I want Jul 31 17:40:34 Is there supposed to be a difference between using IMAGE_INSTALL and installing an rpm? Jul 31 17:41:54 skfax: erm, definitely not Jul 31 17:42:05 that sounds very strange... Jul 31 17:43:20 I tried to understand what goes into the install; is this determined by the FILES variable? The information put into it seems a bit weird: https://github.com/bmwcarit/meta-ros/blob/master/classes/ros.bbclass Jul 31 17:44:18 (I also tried to get one of their examples "chatter_talker" to install properly; however I have the same issue where using IMAGE_INSTALL works but the RPM does not) Jul 31 17:44:51 3 RPMs are generated; release, debug and dev. They are very small and can be installed producing a few files - but not the actual application Jul 31 17:47:53 :-( Jul 31 18:45:43 Has anyone tried to use lto (link time optimization) on a yocto build yet? (I'm considering it) Jul 31 19:48:21 * kergoth wonders whether its most intuitive for deltask to also clear the task deps.. if you deltask and then re-addtask, does it keep its old task deps? I'm inclined to say no, it should start fresh Aug 01 01:57:22 I'm trying to create an image based on master which includes a gtk+3 program, but whet i try to launch it I get "Couldn't open libGL.so.1: libGL.so.1: cannot open shared object file: No suche file or directory." I'm guessing I just need to add a package, but does anyone know which one? Aug 01 01:58:12 I'm targeting qemux86 right now, not sure if that matters… **** ENDING LOGGING AT Sat Aug 01 02:59:58 2015