**** BEGIN LOGGING AT Tue Nov 18 02:59:59 2014 Nov 18 08:28:16 good morning Nov 18 10:04:50 morning all Nov 18 10:04:58 hey bluelightning Nov 18 10:05:05 hi mago_ Nov 18 10:31:00 hi bluelightning, all Nov 18 10:31:06 hi mckoan Nov 18 10:31:34 hi Nov 18 10:33:31 hi splate Nov 18 10:43:02 is it possible to specify a given makefile in a recipe? Nov 18 10:46:48 splate: you could do EXTRA_OEMAKE += "-f yourmakefile" Nov 18 13:48:51 hi Nov 18 13:50:40 i use OE as part of the OE-alliance compilation (i know you cannot give support for OE-alliance's work) Nov 18 13:51:12 i was trying to add an own bitbake in the regarding "meta-local" layer Nov 18 13:52:14 while invokation of bitbake and parsing all the recipes, it gets parsed (i first got an error message, because i forgot to set LICENSE), but later it seems to be ignored, because no output files are created (i checked with "find") Nov 18 13:53:07 i also added echo commands, which write to custom files, to do_compile and do_install in the recipe, to check if this functions are executed in my recipe (which they aren't) Nov 18 13:54:23 so now i have no further clue what is missing there; i wonder that the parsing remarks the missing LICENSE, but later no executable is created etc Nov 18 13:55:53 this is my recipe: http://pastebin.com/SpgDh66c Nov 18 14:16:29 BasicMaster, what is the oe-alliance? Nov 18 14:17:01 lo Nov 18 14:17:26 does anyone care about sip and PyQt Nov 18 14:17:34 or know much the recipes Nov 18 14:19:05 I tried to use the meta-oe repository for yocto and I get an error "The URL: '${SAVANNAH_NONGNU_MIRROR}/hddtemp/hddtemp-0.3-beta15.tar.bz2' is invalid and cannot be interpreted"... does someone probably know what I did wrong? I don't find it Nov 18 14:19:06 * Crofton|work grumbles about people duping recipes in layers Nov 18 14:19:18 and updating in there layers and not sending updates to parent layers Nov 18 14:20:11 Crofton|work: it is a common project of some teams, which maintain images (e.g. openmips, openatv) for DVB set top boxes Nov 18 14:20:14 * Crofton|work is beginning to conclude the layer model is fundamentally broken, because will not work together unless forced to Nov 18 14:20:22 ah Nov 18 14:20:35 I wish they had a name that didn't use OpenEmbedded Nov 18 14:20:40 it is confusing Nov 18 14:21:05 i agree with you Nov 18 14:21:37 also the layer tool is great for exposing anti-social layers Nov 18 14:26:55 and Qt5 versus Qt4 makes the roblem harder Nov 18 14:27:04 RP, you going to keep riding the bike Nov 18 14:27:17 and the strava thing better not be you on a scooter :) Nov 18 14:29:13 Crofton|work: IMO we're a bit better off now, in the old days people just forked OE leading to bigger messes Nov 18 14:29:36 I'm just ranting Nov 18 14:29:45 meta-ivi-demo has some pyqt work in it Nov 18 14:29:53 to be fair it is for qt5 Nov 18 14:30:01 but still annoying Nov 18 14:30:16 I need to make pyqt actually work for gnuradio tools Nov 18 14:30:56 jaeckel: hi, perhaps you're not using the same branch between poky and meta-oe, you should make sure that you are Nov 18 14:31:21 bluelightning, it is probably a good topic for a tsc meeting Nov 18 14:31:31 Crofton|work: right Nov 18 14:31:38 hwo do we stop people from copying recipes into their own layers Nov 18 14:32:03 I do think we need to beat the drum about collaborating, instead of copying Nov 18 14:32:41 FWIW, I did add this page to make this sort of thing a bit easier to see: http://layers.openembedded.org/layerindex/branch/master/duplicates/ Nov 18 14:32:52 evil :) Nov 18 14:33:20 rofl 4 copies of alsa-plugins Nov 18 14:34:07 yeah, that recipe doesn't seem to be in a "central" place, perhaps that's the problem Nov 18 14:34:14 I think I will send that page to the list with the subjest "Wall of Shame" Nov 18 14:34:29 subjest is even a nice word :) Nov 18 14:34:53 in some cases movign recipes to a more central palce sould help Nov 18 14:35:13 you may wish to filter out meta-webos-ports, I think that's intentionally a copy of meta-webos although I could be wrong Nov 18 14:39:32 bluelightning: ty, that was it Nov 18 14:39:44 np Nov 18 14:40:09 why would a recipe build the file at first and not working when relaunching Nov 18 14:40:43 does it mess with unpacked files in S? Nov 18 14:40:47 splate: which task is failing, and what is the failure? Nov 18 14:42:05 well actually i had trouble with a lib, install failed but lib was compiled. now the lib is not even compiled Nov 18 14:43:57 is there a good cheat sheet for bitbake? Nov 18 14:46:16 jaeckel: well there is http://elinux.org/Bitbake_Cheat_Sheet Nov 18 14:46:31 well but that's far from complete :) Nov 18 14:47:12 sure, it's also quite old Nov 18 14:47:15 e.g. how do I build a specific version of a recipe? Nov 18 14:47:57 set PREFERRED_VERSION_recipename = "version" in your config file and then build it as normal is the best way Nov 18 14:51:41 is -c clean enough to clean the recipe? Nov 18 14:52:05 im having different behaviours Nov 18 15:00:35 splate: -c clean will delete just the work directory, if you want to also clean the results of completed output tasks you will need to do -c cleansstate instead Nov 18 15:03:22 bluelightning: thx, this doesnt solve my problem, im having two different results with the same pameters this is weird Nov 18 15:04:44 I'm still a bit in the dark as to what exactly is going on there, if you could explain further I might be able to help... Nov 18 15:06:19 bluelightning: first bake: makefile complains about unknown headers. after adding autotools and pkgconfig it does compile but fails install. after changing smth in the makefile it fails so undoing the changes and ... fails Nov 18 15:07:01 fails how? Nov 18 15:07:54 do_install but log.do_compile says : nothing to compile and nothing is compiled Nov 18 15:08:30 at first bake i had at least the lib compiled in workdir Nov 18 15:10:41 if log.do_compile is saying that then either it isn't building where it should (unlikely) or it isn't being cleaned properly Nov 18 15:12:49 "it isn't being cleaned properly": what can i do in this case? Nov 18 15:13:39 you're sure that what you are cleaning with -c clean is the same thing that's being built? Nov 18 15:15:47 yes Nov 18 15:15:59 ok, then I'm really confused as to what is going on Nov 18 15:16:47 logically if you have cleaned the recipe there can be no work files for it to claim that there is nothing to compile Nov 18 15:17:40 which version of the build system are you using, and are you prepared to share your recipe? Nov 18 15:19:02 bitbake version? Nov 18 15:19:20 and OE-Core version/branch Nov 18 15:19:41 bitbake is 1.18.0 Nov 18 15:20:03 Crofton|work: I just bought a new bicycle which goes a little better than my 40 year old one Nov 18 15:20:16 how can i get OE-CORE version Nov 18 15:20:31 where did you get the metadata from? Nov 18 15:20:47 i.e. the core recipes Nov 18 15:24:32 bluelightning: not sure its been a while now Nov 18 15:25:06 I see, ok... well I'm just going to assume the dylan branch since that would match up with 1.18 Nov 18 15:25:38 splate: are you prepared to share your recipe? Nov 18 15:27:08 the recipe is no problem, it just defines a SRC_URI="file://libtest.tar.gz" a workdir and inherit autotools pkgconfig Nov 18 15:27:15 not much yet Nov 18 15:27:58 btw removing autotools gets rid of the make install fail Nov 18 15:28:29 which wasnt a problem the first time .. Nov 18 15:28:45 is it actually an autotools recipe? Nov 18 15:28:58 it seems not :) Nov 18 15:29:21 either the library you are building uses autoconf/automake, or it doesn't Nov 18 15:29:29 if it doesn't, inheriting autotools would be incorrect Nov 18 15:31:04 it just has a makefile Nov 18 15:32:41 right, then you don't want inherit autotools Nov 18 15:33:24 you'll need to write your own do_install which does whatever is appropriate (either 'oe_runmake install ' or manual install commands to install into ${D}) Nov 18 15:34:32 this is clear to me but there is nothing compiled so .. Nov 18 15:35:30 "nothing to compile" Nov 18 15:42:53 RP, I notice. I'm going to invite you to the trolling group on Strava. Nov 18 15:42:58 do not worry about the name Nov 18 15:43:59 splate: you'll have to look at the context in which your makefile is being run and what it is doing, it's going to be specific to how the makefile is written# Nov 18 15:44:39 splate: you can look at run.do_compile under the temp directory in the workdir for the recipe as a reference for how the makefile being processed Nov 18 16:10:19 bluelightning: can the Makefile be named smt.mak or do i have to use only MAkefile as a name? Nov 18 16:10:54 splate: as I mentioned earlier, use something like EXTRA_OEMAKE += "-f smt.mak" Nov 18 16:11:06 tried this already Nov 18 16:11:11 and? Nov 18 16:11:41 really dont get what has changed since the first bake Nov 18 16:11:54 I've no idea either Nov 18 16:24:18 bluelightning: ok i have reproduced the previous behaviour by doing strange stuff Nov 18 16:25:40 dont know why but i renamed the makefile to mAkefile, then i got the complains from the beginning about the name that is not recognized, then putting it back to smt.mak and i got the lib compiled Nov 18 17:49:34 are the .sh files from the sdk generation compressed? Nov 18 17:51:03 yes Nov 18 17:51:08 with gzip **** ENDING LOGGING AT Wed Nov 19 02:59:59 2014