**** BEGIN LOGGING AT Fri Oct 16 02:59:58 2015 Oct 16 07:26:45 morning all Oct 16 07:38:47 <_4urele_> Hi all Oct 16 07:40:55 <_4urele_> I have a team mate which is using the sdk, he is modifying a library which exists in the sdk, so to my mind he should modify his local sdk, or is there another option? (installing in another directory and tell the system to look into it) Oct 16 07:51:04 _4urele_: the buzzword answer would be: use the extensible sdk :) Oct 16 07:55:43 LetoThe2nd: is the extensible sdk added in yocto 1.8 or 1.9? Oct 16 07:56:22 fredcadete: added in 1.8 and usable from 2.0 on, if i understood it correctly. Oct 16 07:58:41 thanks Oct 16 07:59:38 fredcadete: maybe http://events.linuxfoundation.org/sites/events/files/slides/yocto_project_dev_workflow_elc_2015_0.pdf helps a bit Oct 16 08:05:51 <_4urele_> LetoThe2nd, thanks I will look this extensible sdk Oct 16 08:12:13 Hi. If I want to report a bug for a recipe, does it go under Product: Layers, Component: Layers? I did not see any components that would say "recipe". in bugzilla.yoctoproject.org Oct 16 08:20:25 gatisp: if it's a oe-core recipe then product "OE-Core", and pick a component that's a close match Oct 16 08:23:54 jku, ok, i will check. it is from "meta-openembedded" Oct 16 08:25:04 gatisp: most of the meta-oe layers are not handled in yocto bugzilla (because they're not really part of yocto). Check the layer readme Oct 16 08:27:58 right, we don't track bugs in meta-openembedded layers in the YP bugzilla Oct 16 08:28:16 unless they are a direct result of an issue in OE-Core or bitbake Oct 16 08:28:30 (in which case the bug would be about that underlying issue) Oct 16 08:29:32 interesting, the read me file says "Send patches / pull requests to openembedded-devel@lists.openembedded.org" Oct 16 08:29:43 so there is no way to report bugs, just send patches :) Oct 16 08:31:48 hello. i facing some problems using nativesdk. see my nativesdk-commonapi_2.6.1.bb recipe here http://paste.ubuntu.com/12796856/ the link also provides the directory structure in tmp-glibc/work/x86_64-nativesdk-oesdk-linux/nativesdk-commonapi/2.1.6-r0 -- the expected folders under sysroot-providers are not present.. i do not find the files in the sdk after running populate_sdk Oct 16 08:32:06 please, support me here :) Oct 16 08:33:04 gatisp: you can report issues via that mailing list as well Oct 16 08:33:23 line 18 is test/ not ttt/ Oct 16 08:34:16 bluelightning, I will send a mail to maintainers of that layer Oct 16 08:34:50 ericbutters: is your recipe called nativesdk-commonapi or commonapi? if the latter, the name is wrong - BBCLASSEXTEND will add that prefix for you when it creates the nativesdk variant of the recipe Oct 16 08:35:05 ericbutters: er, the former, I meant Oct 16 08:35:28 it is called nativesdk_commonapi Oct 16 08:35:43 _ or - ? Oct 16 08:35:45 nativesdk-commonapi Oct 16 08:35:46 sry Oct 16 08:35:53 that's not correct then Oct 16 08:36:10 should it be named commonapi? Oct 16 08:36:12 either rename it to commonapi, or instead inherit nativesdk rather than BBCLASSEXTENDing to nativesdk, then you need the native- prefix Oct 16 08:36:26 er, nativesdk- Oct 16 08:36:29 native- ? not nativesdk- ? Oct 16 08:36:32 ah Oct 16 08:36:40 no, I mistyped, sorry... lack of sleep :/ Oct 16 08:36:46 i tried that but that did not work also Oct 16 08:37:03 so my recipe seems to be okay? Oct 16 08:37:12 the content? Oct 16 08:37:32 i try to rename to commonapi and use BBCLASSEXTEND Oct 16 08:37:42 bluelightning: a developer suffering lack of sleep? how unbelievable! Oct 16 08:38:07 ericbutters: I can't immediately tell Oct 16 08:38:47 ericbutters: at this stage, what's going into image/ is of interest rather than sysroot-destdir Oct 16 08:48:45 bluelightning: the recipe is named commonapi-nativesdk_2.1.6.bb -- i changed to inherit nativesdk -- and i call "bitbake nativesdk-commonapi" -> in image/ i got the stuff i copied in do_install_append_class-nativesdk Oct 16 08:49:01 er Oct 16 08:49:26 making nativesdk a suffix isn't right either Oct 16 08:49:59 I'm sure it's in there, the question is is it under the correct path? Oct 16 08:52:40 therefore i copy the content twice, i see it in image like: http://paste.ubuntu.com/12797016/ Oct 16 08:58:20 Morning fols Oct 16 08:58:41 bluelightning: therefore i copy the content twice, i see it in image like: http://paste.ubuntu.com/12797016/ can you provide a simple example that just copies a text file into the nativesdk to usr/test? Oct 16 08:59:05 bluelightning: thanks for support Oct 16 08:59:08 ericbutters: I can't off the top of my head, but there must be such an example in the metadata Oct 16 08:59:19 in existing recipes, I mean Oct 16 09:03:27 ericbutters: in your recipe, I see you use ${D}/usr. I think it should be ${D}${prefix} Oct 16 09:14:59 bananadev: bluelightning: here is a more basic example: http://paste.ubuntu.com/12797209/ please find in that link my basic recipe and how i call -- and the result in the work directory Oct 16 09:30:17 bananadev: bluelightning: populate_sdk then ends with error: Unknown package 'nativesdk-ericbutters' -- i added to my image to TOOLCHAIN_HOST_TASK_append: nativesdk-ericbutters Oct 16 09:41:00 you mean you can bitbake nativesdk-ericbutters success but not meta-toolchain ? Oct 16 09:41:36 Did you try rename your recipe to nativesdk-erricbutters_1.0.bb Oct 16 09:42:55 I mean nativesdk-ericbutters_1.0.bb Oct 16 09:45:30 indeed, I mentioned moving nativesdk to a suffix wasn't right above... Oct 16 10:00:16 bluelightning: hey, around? Oct 16 10:00:33 bluelightning: I have recently run across something that may be worth noting in the "porting to Jethro" guide Oct 16 10:00:39 bluelightning: http://lists.freedesktop.org/archives/wayland-devel/2015-June/022548.html Oct 16 10:02:18 joshuagl: hmm, interesting - would you be able to word a small note in the form we might need it for the migration guide and send / paste it to me? Oct 16 10:02:35 bananadev: bluelightning: okay, i renamed to nativesdk-ericbutters_1.0.bb but populate_sdk still ends with error, pls see here: http://paste.ubuntu.com/12797465/ Oct 16 10:03:28 bluelightning: sure, I'm just about to step AFK so may forget. Please remind me if I do Oct 16 10:03:38 bluelightning: happy to supply something as a git patch if easier? Oct 16 10:03:52 but, i only see *-dbg and *-dev in deploy-ipks/x86_64-nativesdk so thats why right? Oct 16 10:04:26 joshuagl: Scott usually likes to do the final formatting, so I usually send him the raw material and let him take care of putting it into the doc Oct 16 10:04:37 bluelightning: ack Oct 16 10:10:40 ericbutters: right, whatever you're installing is not being packaged Oct 16 10:10:49 ericbutters: normally you would see a warning to that effect Oct 16 10:18:56 hello internet Oct 16 10:19:48 I am trying to install my scripts, but I don't see fetched 'System' directory :-( Here is the recipe of mine: http://dpaste.com/1NQD7MV Oct 16 10:26:48 bluelightning: got it! i only use {PN} for PACKAGES and and FILES.. thanks! Oct 16 10:43:30 wow it doesn't fetch the sripts o_O\ Oct 16 11:23:05 guys, could somebody help me with this one: http://dpaste.com/2Y5MFTV ? Oct 16 11:23:14 why the scripts were not be fetched? Oct 16 11:40:08 Ox4: which recipe? Oct 16 11:43:26 fl0v0: Oct 16 11:43:26 [13:19] Oct 16 11:43:26 *** IvanSB (~IvanSB@2a01:2000:2000:efcd:f279:59ff:fe64:3a8) has joined channel Oct 16 11:43:26 #yocto [13:23] Oct 16 11:43:29 bluelightning: got it! i only use {PN} for PACKAGES and and Oct 16 11:43:33 FILES.. thanks! [13:26] Oct 16 11:43:37 *** IvanSB (~IvanSB@2a01:2000:2000:efcd:f279:59ff:fe64:3a8) has quit: Ping Oct 16 11:43:40 timeout: 250 seconds [13:28] Oct 16 11:43:43 *** IvanSB (~IvanSB@2a01:2000:2000:efcd:f279:59ff:fe64:3a8) has joined channel Oct 16 11:43:46 #yocto [13:41] Oct 16 11:43:50 wow it doesn't fetch the sripts o_O\ [13:43] Oct 16 11:43:50 -YoctoAutoBuilder- build #519 of nightly-multilib is complete: Success [build Oct 16 11:43:53 successful] Build details are at Oct 16 11:43:56 http://autobuilder.yoctoproject.org/main/builders/nightly-multilib/builds/519 Oct 16 11:44:00 [13:44] Oct 16 11:44:03 *** IvanSB (~IvanSB@2a01:2000:2000:efcd:f279:59ff:fe64:3a8) has quit: Ping Oct 16 11:44:07 timeout: 240 seconds [13:46] Oct 16 11:44:10 *** rob_w (~bob@unaffiliated/rob-w/x-1112029) has quit: Quit: Leaving [13:48] Oct 16 11:44:13 *** psnsilva_ (~psnsilva@193-126-29-154.net.novis.pt) has joined channel Oct 16 11:44:16 #yocto Oct 16 11:44:20 *** ftonello (~quassel@81.145.202.106) has joined channel #yocto [13:51] Oct 16 11:44:23 *** IvanSB (~IvanSB@2a01:2000:2000:efcd:f279:59ff:fe64:3a8) has joined channel Oct 16 11:44:23 #yocto [13:59] Oct 16 11:44:26 *** IvanSB (~IvanSB@2a01:2000:2000:efcd:f279:59ff:fe64:3a8) has quit: Ping Oct 16 11:44:29 timeout: 250 seconds [14:07] Oct 16 11:44:32 *** nobu (~nobu@62.20.3.162) has joined channel #yocto [14:13] Oct 16 11:44:37 *** medicalwei (~medicalwe@122-116-81-121.HINET-IP.hinet.net) has quit: Ping Oct 16 11:44:40 timeout: 264 seconds [14:15] Oct 16 11:44:43 *** IvanSB (~IvanSB@2a01:2000:2000:efcd:f279:59ff:fe64:3a8) has joined channel Oct 16 11:44:44 the recipe looks like it ought to work... I'm assuming the repeated semicolons aren't causing a problem Oct 16 11:44:46 #yocto [14:19] Oct 16 11:44:50 *** bananadev (~dev@117.6.99.240) has quit: Quit: Leaving [14:21] Oct 16 11:44:53 *** Aethenelle (~Aethenell@107.138.98.226) has joined channel #yocto [14:22] Oct 16 11:44:53 *** medicalwei (~medicalwe@122-116-81-121.HINET-IP.hinet.net) has joined Oct 16 11:44:56 channel #yocto Oct 16 11:44:59 guys, could somebody help me with this one: http://dpaste.com/2Y5MFTV ? Oct 16 11:45:02 [14:23] Oct 16 11:45:07 why the scripts were not be fetched? Oct 16 11:45:10 *** IvanSB (~IvanSB@2a01:2000:2000:efcd:f279:59ff:fe64:3a8) has quit: Ping Oct 16 11:45:13 timeout: 250 seconds [14:26] Oct 16 11:45:16 *** sameo (~samuel@192.55.55.37) has quit: Remote host closed the connection Oct 16 11:45:20 [14:33] Oct 16 11:45:23 *** ericbutters (~daniel@217.110.68.82) has quit: Quit: leaving [14:35] Oct 16 11:45:23 *** IvanSB (~IvanSB@2a01:2000:2000:efcd:f279:59ff:fe64:3a8) has joined channel Oct 16 11:45:26 #yocto [14:39] Oct 16 11:45:29 Ox4: which recipe? [14:40] Oct 16 11:45:32 Oct 16 11:45:37 oops Oct 16 11:45:43 sorry guys, I miss button :-( Oct 16 11:46:10 it's alright, I once pasted a private discussion into a public IRC channel Oct 16 11:46:39 bluelightning: ouch Oct 16 11:46:51 hopefuly the channel wasn't logged... Oct 16 11:47:01 fl0v0: http://dpaste.com/1NQD7MV Oct 16 11:47:06 luckily there was nothing especially private in there, it was just a bit embarrassing Oct 16 11:47:42 fl0v0: but if I manually download scripts to the System directory the recipe is executed successfully Oct 16 11:55:56 Ox4: beats me. the scripts are normally checked out when you clone into the repo manually? Dont see why that shouldn't work Oct 16 11:59:57 fl0v0: yes Oct 16 12:02:35 Ox4: well, FWIW I just tested subpath= here and it did what I expected it to Oct 16 12:06:48 bluelightning: does it fetch all stuff? Oct 16 12:07:06 it fetched everything in the subpath I specified to a directory of the same name yes Oct 16 12:09:46 Ox4: does it fetch the right revision? I didnt know about AUTOREV but maybe that messes up Oct 16 12:25:40 In bbappend, can some packages/regexs override/conflict in FILES_${PN}? Trying move some bins into a test package but main package is holding on to them. Oct 16 12:31:21 mcfrisk: how are you adding the test package to PACKAGES? Oct 16 12:34:14 bluelightning: qtbase from meta-qt5, I'm bbappending some more examples and trying to push them to separate binary package. So FILES_${PN} from qt5.inc seems to override my bbappend with FILES_${PN}-tests Oct 16 12:35:14 mcfrisk: I guess what I was driving at is where exactly is the package in PACKAGES in relation to ${PN} Oct 16 12:36:02 if it comes after then it's not surprising the main package picks up the files still because its FILES value will match them first Oct 16 12:45:26 if I don't want to use the fetcher, and simply want to fetch code myself by my own means Oct 16 12:45:40 I just don't declare a SRC_URI and implement my own do_fetch Oct 16 12:45:44 right? Oct 16 12:49:25 if you feel you must yes... but it's very rare to need to do that Oct 16 13:15:21 is there a construct for adding lines to ld.so.conf ? Oct 16 13:20:20 cause files in ld.so.conf.d are not included unless ld.so.conf says so? Oct 16 13:27:03 include /etc/ld.so.conf.d/* usually Oct 16 13:27:15 [mhatle@msp-dhcp23 trunk]$ cat /etc/ld.so.conf Oct 16 13:27:15 include ld.so.conf.d/*.conf Oct 16 13:27:21 thats the line from a RHEL machinef or instance Oct 16 13:29:48 several people have asked why we don't do that out of the box Oct 16 13:29:56 I don't know that there has been a satisfactory answer Oct 16 13:30:49 answer is that many of our systems simply don't ever use ld.so.cache.. Oct 16 13:31:17 secondary reason -- so far it's VERY rare that you (as the -system- developer need to ever put an entry into ld.so.conf (let alone create ld.so.conf.d/*.conf entries) Oct 16 13:31:29 we want to actually discourage that behavior -- as it usually means broken system items.. Oct 16 13:32:10 (ld.so.cache is often used as a crutch for broken libraries that do not have proper filename/sonames... so the RTLD (ld.so) can't find them w/o the cache) Oct 16 13:32:39 the other piece is for libraries installed into non-standard locations.. but they shouldn't be cached, because the point of non-standard locations is to keep them outside of the regular loading namespace.. Oct 16 13:32:50 so if you suddenly cache them -- they are now availabel globally.. Oct 16 13:32:55 "just say no" Oct 16 13:33:58 (another way to look at it -- if the system requires ldconfig to have run -- you have a broken system.. if you are running it as a performance improvement.. GREAT! that is what it is there for) Oct 16 13:36:34 ok, perhaps I'd overestimated how often people would need to poke at that Oct 16 13:36:43 thanks for the detailed explanation :) Oct 16 13:37:05 fray/tasslehoff - I have the same issue, and will migrate to using a standard lib directory instead. We have deployed systems that do require poking that file. Oct 16 13:37:31 bluelightning: so in bbappend I need to append do_install to install files to different location, then at do package time pick them into correct package and move back to right path, or change the meta-qt5 side of recipe... Oct 16 13:39:01 mrk377 why are they using a non-standard lib dir? Is it to hide the libraries from the system.. or was it just a convienent place to store them? Oct 16 13:39:32 Usually you can use 'relative' library paths (See $ORIGIN in the linker docs) or an LD_LIBRARY_PATH= environment option in a shell wrapper to avoid this Oct 16 13:40:37 (if the purpose is to hide them, the relative/ld-library-path mechanism is the right one to use.. if it's just a convienence thing.. then I'd suggest moving them -- or at least symlinking fromt he regular libdir to where they were installed -- if you choose not to move them) Oct 16 13:40:44 mcfrisk: er, no... no moving around of files required Oct 16 13:40:46 Well, it was how it was done in the past. Mainly handling legacy code base, and most of us were learning. Yeah, we have used the shell wrapper too. I like your stance Oct 16 13:41:21 mcfrisk: you have two choices - rearrange PACKAGES such that your package comes first, or change FILES_${PN} so that it doesn't match the files you are trying to split out Oct 16 13:42:04 fray: I agree. I install an nvidia binary-driver recipe on my target (from meta-jetson-tk1). That one puts a file in /etc/ld.so.conf.d, but the contents of the file are wrong, and ld.so.conf is empty Oct 16 13:43:19 the nvidia case is interesting.. last time I looked they did it because they are intentionally polluting the SONAME namespace in order for their library to load.. the problem is that mechanism is somewhat fragile Oct 16 13:43:23 I've only got yacto compiling on RHEL7 this week (Great tool. love it for the read-only filesystem). Oct 16 13:43:48 Question FRAY: I have 4 source packages (autotools built) that need to incorporated into image. What is best strategy and howto? Oct 16 13:44:00 (and I don't have a good answer for the nvidia case.. since they did some very strange things) Oct 16 13:44:20 * fray looks over at bluelighting and suggest 'devtool' / recipetool Oct 16 13:44:54 fray: as long as I can get "the usual" ld.so.conf I should be fine. seems I must modify the libc-recipe for that. Oct 16 13:45:09 unless I can modify it in an image postprocess command or something Oct 16 13:45:36 FRAY: Is the mega-manual best reference for devtool/recipetool examples? Oct 16 13:46:17 There is a steep barrier to entry on this! Oct 16 13:46:49 I'm not sure if the manual is there yet for devtool/recipe tool.. (I've only had limited experience with it.. Bluelightning and others helped write these tools.. as they are fairly new) Oct 16 13:47:09 that is the recommended way of doing what you ask moving forward.. (at least setting up the recipe and such) Oct 16 13:47:09 devtool is covered in the development manual Oct 16 13:47:26 I still do it the "old fashion way" and write a new recipe from scratch.. but thats what I've been doing for about 5+ years now.. Oct 16 13:47:31 http://www.yoctoproject.org/docs/current/dev-manual/dev-manual.html#dev-modifying-source-code Oct 16 13:49:35 mrk377 here is something i found quickly on a google search.. Oct 16 13:49:36 https://enchildfone.wordpress.com/2010/03/23/a-description-of-rpath-origin-ld_library_path-and-portable-linux-binaries/ Oct 16 13:49:52 I've not read it all.. but the RPATH is what I was referring to with the $ORIGIN Oct 16 13:50:16 Fray - Thx mate Oct 16 13:50:17 you can embed that into the binaries when they are linked, and then they will search that path relative to the execution location.. completely avoiding the need for the ld.so.conf in most cases... Oct 16 13:50:47 (as I said, nvidia and other 'graphics override' type libraries are unique cases, one that I don't have a good answer for -- since each one seems to be unique in how it's written) Oct 16 13:51:00 (often though the LD_PRELOAD= is enought o make them work.. but that then requires a special wrapper) Oct 16 13:51:22 bluelightning: Thanks! PACKAGES_prepend did the trick. Oct 16 13:53:07 one thing not in the link I posted.. there are ways to set/change RPATH after a binary is linked as well.. a 'chrpath' command available on some distributions.. so if the install location of an executable can be changed.. what you end up doing is using a really long default path.. and then use the chrpath at install time to set the actual location.. that however is usually only necessary -if- an absolute path is required for some reason -- it rarely is Oct 16 13:53:18 the $ORIGIN path is the normal right way to handle it all.. Oct 16 13:53:25 * fray goes to get some breakfast Oct 16 14:44:43 Guys, I ran$ recipetool -d create -o /WHERE/zeromq-3.2.5.bb http://download.zeromq.org/zeromq-3.2.5.tar.gz Oct 16 14:45:34 Where should i put the zeromq-3.2.5.bb file to get included. I run bitbake -k core-image-minimal-dev Oct 16 14:45:39 hey. i built fsl-image-x11 and it created fsl-image-x11-t4240rdb-64b-20151016001455.rootfs.ext2.gz fsl-image-x11-t4240rdb-64b-20151016001455.rootfs.ext2.gz.u-boot uImage--3.12-r0-t4240rdb-64b-20151016001455.bin and other such files. how to setup this now? Oct 16 14:46:46 mrk377: look where the core-image-minimal-dev is defined and add it to the IMAGE_INSTALL += "zeromq" Oct 16 14:47:07 K. Oct 16 14:47:20 mrk377: it is in meta-something/images (at least for the freescale and raspbi stuff i've been playing with) Oct 16 14:51:52 anyway, is there a general guide how to put that freescale stuff on a sd-card? Oct 16 15:01:11 qknight: dd the .sdcard file to your sdcard Oct 16 15:01:54 joshuagl: problem: i don't have the sdcard image only the files listed above and i've been trying IMAGE_FSTYPES="tar.bz2 ext3 sdcard" but i'm not on ARM but on PPC and maybe the sdcard isn't built there... Oct 16 15:02:16 joshuagl: not sure how to format the sd-card and where to place all the files... yet Oct 16 15:02:40 https://community.freescale.com/thread/320655 <- here they state: sdcard is a class defined only under meta-fsl-arm (meta-fsl-arm - Layer containing Freescale ARM hardware support metadata) this is the one that creates the binary *.sdcard. Oct 16 15:05:31 oh, right - the sdcard stuff is from meta-fsl-arm Oct 16 15:05:38 I have no idea about the ppc parts, sorry Oct 16 15:45:35 <[Sno]> bluelightning: I'm a coder - if you'd love someone hacks code to support you guys, kick my ass (not this week nor month, but generally) Oct 16 15:45:46 <[Sno]> bluelightning: mind I hate python ;) Oct 16 15:52:54 * armpit nice to be home Oct 16 15:53:11 armpit, recovered from all of the guiness? Oct 16 15:54:48 QUESTION: If my final image will support a PC-104 PCI bridge, must I add a MACHINE_FEATURES += "pci" (In my build/local.conf, is that standard)?? Oct 16 15:54:50 not enough guinness Oct 16 15:55:11 fray, had a 3 day cust visit tacked on to elce Oct 16 15:55:17 whee Oct 16 15:55:36 hopefully not too painful Oct 16 15:56:16 not too bad. pain part was it was colder. they had to de-ice the plan when i left Oct 16 15:58:14 HA! Oct 16 17:23:28 Has anyone every forced a specific version of the Kernel? How far can you go back? Oct 16 17:32:57 mrk377 your kernel and glibc need to be compatible.. but other then that you can go back quite a ways.. Oct 16 17:33:14 there is a minimum kernel version set for glibc.. I don't remember the parameter offhand.. but that is the farthest point back.. Oct 16 17:34:00 fray: If I had to base my project on a 2.6 kernel can this be done? Oct 16 17:34:13 variable is 'OLDEST_KERNEL' Oct 16 17:34:16 ./bitbake.conf:OLDEST_KERNEL = "2.6.32" Oct 16 17:34:16 ./bitbake.conf:OLDEST_KERNEL_aarch64 = "3.14" Oct 16 17:34:29 that is simply for compatibility with glibc.. Oct 16 17:34:50 thats not to say other things won't break.. but it does give you a reasonable idea how far back you can go at least on the toolchain side Oct 16 17:37:27 Are you warning me that going to a 2.6.32 kernel will be difficult? I'm nervous now. Oct 16 17:38:11 it very well may be.. I don't know.. Oct 16 17:38:16 but 2.6.32 or newer will work with glibc Oct 16 17:38:30 just the fact you know the bounds for glibc will give you a higher chance of success Oct 16 17:39:04 kernels older then 2.6.32 are missing required APIs.. thus the 'oldest_kernel' setting Oct 16 17:42:04 fray: Thanks. How do I force a bitbake to use a Linux-yacto baseline kernel that is older? Oct 16 17:42:29 Since you make branch specific features. Oct 16 17:42:35 I think there is documentation there.. but basically you have to create an older linux-yocto recipe and then set preferred version Oct 16 17:42:47 but I don't do much on kernels.. I'm toolchain, userspace and build system Oct 16 17:43:20 I look into it. thx :) Oct 16 17:48:47 Anyone used to Xwayland? Oct 16 20:24:13 Men, in my poky/meta/recipes-kernel/linux folder (I have a linux-yocto_3.14.bb). How do I force this kernel? Oct 16 20:24:42 because I added PREFERRED_VERSION_linux-yocto_forcevariable = "3.14%" (The kernel did not change. Scratching head) Order of operation issue. Oct 16 20:24:51 https://bugzilla.yoctoproject.org/show_bug.cgi?id=4204 Oct 16 20:24:53 Bug 4204: normal, Undecided, ---, richard.purdie, RESOLVED NOTABUG, PREFERRED_VERSION_linux-yocto does not override machine specific settings in poky.conf Oct 16 21:04:29 Anyone knows how to remove dependencies check for a recipe? Oct 16 21:39:23 mtetreault_: just edit the file maybe? Oct 16 21:49:16 qknight: The issue is that I have a kind of hackish situation. I have Qt4.8.6Embedded compiled and deployed to the rootfs by yocto. But I also manually copy over Qt4.8.6 to the rootfs afterward (I had some issue with the version compiled by yocto). The recipe that fails is one that copies binaries needing qt4.8.6 that is not compiled/copied by yocto. That's why I'd like to remove the check for that recipe Oct 16 21:50:15 I know this is far from been the best solution. This is a really temporary solution. **** ENDING LOGGING AT Sat Oct 17 02:59:58 2015