**** BEGIN LOGGING AT Mon Sep 21 02:59:59 2015 Sep 21 07:25:58 good morning Sep 21 07:34:05 good morning Sep 21 07:42:31 good night. :-) Sep 21 08:40:06 <[w00t]> Hi! Is it possible to add PREFERRED_VERSION_xyz = "1.0.0" to an image recipe, it does not seem to work in my tests. No matter what I put in the preferred_version, it seems to use the latest version. I'd like to avoid putting this in local.conf Sep 21 09:01:13 [w00t]: no, you have to put in local.conf or in the distribution config file, not in the image Sep 21 09:05:32 <[w00t]> mckoan: Ok, thnx. You know the reason for this? I'd like different images to contain different versions of my custom software. Sep 21 09:18:11 [w00t]: what do you want to do that ? Sep 21 09:18:21 oh Sep 21 09:18:54 [w00t]: your image might want to RDEPEND on particular version then (and probably collide with other) Sep 21 09:22:29 <[w00t]> Marex: Ok, and I can probably put RCONFLICTS to the "newer" versions to avoid collisions? Sep 21 09:28:47 anyone else having issue with systemd failing to start when compiled with GCC 5.2 from master branch? https://bugzilla.yoctoproject.org/show_bug.cgi?id=8291 Sep 21 09:28:48 Bug 8291: critical, Medium+, 2.0, raj.khem, NEW , systemd fails to start when compiled with GCC 5.2 Sep 21 09:29:00 wondering if it affects more than one platform Sep 21 09:30:52 morning all Sep 21 09:31:09 Morning - Sep 21 09:31:18 bluelightning: morning Sep 21 09:32:46 bluelightning: morning Sep 21 09:33:01 hi Net147, Marex, raykinsella78, mckoan Sep 21 09:33:04 [w00t]: I'd suspect so Sep 21 09:33:25 hm, 11.30 am, about time to hit the hay Sep 21 09:37:13 Marex: wow, long night... Sep 21 09:38:57 bluelightning: FPGAs and crystal meth have a lot in common, for one thing, you cannot sleep after doping them :-) Sep 21 09:39:37 bluelightning: I've been trying to port this ao486 design to my other FPGA kit here :-) Sep 21 09:40:36 hah Sep 21 09:41:03 one day I will have to play around a bit with FPGAs Sep 21 09:44:19 bluelightning: FPGAs are awesome, but once you get hooked, there's no way back Sep 21 09:48:31 bluelightning: let me know if you need some sort of gateway drug ... I mean FPGA introduction :) Sep 21 09:49:21 Marex: the first hit is free ;) Sep 21 09:52:23 bluelightning: well, given that intel bought altera this year, I guess you'll soon have the entire production line at your disposal :) Sep 21 09:54:15 that doesn't sound very profitable Sep 21 09:54:57 bluelightning: ++ Sep 21 10:31:39 bluelightning: it looks like there is a warning when installing base-files. opkg creates /var/run/opkg.lock but base-files has /var/run -> /run symbolic link and fails to unlink /var/run... Sep 21 10:32:43 bluelightning: should we change the lock path in opkg.conf or in the opkg defaults? Sep 21 10:34:10 Net147: I'm not sure, you'd need to ask the list Sep 21 10:35:28 hmm... well maybe better do it in opkg.conf since this is OE-specific. opkg is used in more places than OE... Sep 21 10:45:28 Net147, What was your issue? Sep 21 10:45:54 AndersD: hmm? which one? Sep 21 10:46:02 I was just about to ask about an issue I've finally succeeded in creating a minimal test case for.. Sep 21 10:46:09 Net147, /var/run . Sep 21 10:46:21 -/var/run and /run symlinks. Sep 21 10:46:26 AndersD: building any image results in a warning Sep 21 10:46:37 AndersD: when using systemd and opkg Sep 21 10:46:51 Ah, that's my issue likely. Sep 21 10:47:14 AndersD: I am writing a patch for it now Sep 21 10:47:33 Using opkg /var/run is creatad as a regurlar directory, while when using rpm it's a symlink to ../run Sep 21 10:47:55 Did that sound familiar? Sep 21 10:48:28 AndersD: opkg creates /var/run when it creates /var/run/opkg.lock Sep 21 10:48:44 Ah,I see. Sep 21 10:49:03 Then I'm looking forward to your patch! Sep 21 10:57:11 AndersD: so /var/run is a directory in the image? Sep 21 11:01:57 Yes, which leads to runtime issues as /run and /var/run is completely separate issues. Sep 21 11:02:20 AndersD: yes I an see that. I will update my commit to mention it. Sep 21 11:08:30 AndersD: patch posted to oe-core mailing list Sep 21 11:09:48 Net147, Cool! I'll grab it and test it. Sep 21 12:08:58 AndersD: how was the testing? Sep 21 12:13:59 Net147, Sorry for the delay. It was perfect! Now my images are built correclty again. (and in due time, I can continue forwardporting my own distro to the upcoming release) Sep 21 12:14:11 I'll send an ack on the mailinglist Sep 21 12:15:20 AndersD: thanks Sep 21 12:17:59 No problem, you saved me a lot of time debugging the difference between using ipk and rpm. It took a few hours to determine the minimal testcase; but the rest of the work would have been much harder... Sep 21 12:40:02 Hi looking for help in building git using a .bb file Sep 21 12:41:56 tarak: http://git.openembedded.org/openembedded-core/tree/meta/recipes-devtools/git Sep 21 12:44:19 thx. Sep 21 13:42:22 Hi how to tell bitbake to include git into the image file (eg., i'm trying to build core-image-sato-sdk) Sep 21 13:42:24 Hi how to tell bitbake to include git into the image file (eg., i'm trying to build core-image-sato-sdk) Sep 21 14:16:51 hi, who should I bother for a contribution to meta-raspberrypi? I updated the wiringpi library Sep 21 14:22:48 LocutusOfBorg1: http://layers.openembedded.org/layerindex/branch/master/layer/meta-raspberrypi/ Sep 21 14:23:04 the main maintainer is Andrei Gherzan Sep 21 14:25:56 thanks, I was hoping about some sort of github pull requests... Sep 21 15:16:39 Hello, I added this http://jpst.it/BVsb to my image recipe. But the symlink is not correct, it point on the full path. Any idea ? Sep 21 15:20:36 Piziwate: that's because your ln command isn't right Sep 21 15:21:26 Piziwate: I think you want: ln -s /lib/ld-linux-x86-64.so.2 ${IMAGE_ROOTFS}/lib64/ld-linux-x86-64.so.2 Sep 21 15:22:18 bluelightning : Does it works if source of the link doesn't exists ?? Sep 21 15:25:58 Piziwate: yes, ln doesn't care Sep 21 15:29:12 https://github.com/snoack/python-goto is twisted. Sep 21 15:29:35 Ok great ! Thank you again bluelightning ! Sep 21 15:46:32 kergoth: my god Sep 21 15:46:41 i know, right? Sep 21 15:47:00 getting around the syntax issue by using valid member/name lookup on label/goto objects is amusing and creative Sep 21 15:47:55 the whole thing is clearly evil, but amusing Sep 21 15:59:47 Can I add a Kernel Fragement through local.conf Sep 21 15:59:48 ? Sep 21 16:00:20 that wouldn't be ideal. possible, but not ideal, you'd have to hardcode a full absolute path to the file on disk Sep 21 16:00:59 can't you do SRC_URI_append_pn-linyx-yocto = "myfragment.cfg? Sep 21 16:01:03 pretty sure i did that before Sep 21 16:01:25 assuming you use a valid uri rather than a filename, yeah Sep 21 16:01:35 I'll give it a whirl, thanks Sep 21 16:01:37 but obviously he can't put it into filespath, so it'd have to be a full path Sep 21 16:01:47 it's really not that hard to do it in a layer, i'd recommend that Sep 21 16:02:33 yocto-layer create local 1; bitbake-layers add-layer meta-local; creates a new layer and adds it to your build. recipetool newappend meta-local virtual/kernel will create a .bbappend in that layer for your kernel, which you can edit to add the .cfg Sep 21 16:03:12 or you could add meta-mel from meta-mentor and use recipetool kernel_add_fragments meta-local /path/to/foo.cfg Sep 21 16:03:13 :) Sep 21 16:03:26 kergoth: fancy cli voodoo, I like it,. Sep 21 16:03:30 * kergoth should see if anyone is interested in those going upstream, or if they're too special purpose Sep 21 16:04:40 raykinsella78: :) i think a lot of folks don't know about yocto-layer, bitbake-layers, recipetool, and devtool, but they're all extremely handy Sep 21 16:09:02 kergoth: how do you delete a badly named layer? Sep 21 16:16:02 from disk, rm -rf, from the build, bitbake-layers remove-layer or edit conf/bblayers.conf and remov eit manually Sep 21 16:16:11 thx Sep 21 16:43:39 * kergoth wonders whatever happened with anibal's attempt to get my LICENSE_PRIORITY support merged Sep 21 16:44:01 our current INCOMPATIBLE_LICENSE handling for the manifest is much less capable and more prone to problems than that was Sep 21 16:45:59 hmm Sep 21 17:15:03 if i have recipe "foo" and "bar". if bar need to foo be in the target it needs the vairable DEPENDS ? Sep 21 17:15:50 what do you mean by 'in the target'? Sep 21 17:15:52 and when? Sep 21 17:16:08 DEPENDS is for build time dependencies -- bar needs foo to be able to configure/compile Sep 21 17:16:18 if it only needs it to run, on target, but not to build, then you want RDEPENDS_${PN} instead Sep 21 17:19:12 mmm ok but im seen same error that state ERROR: Nothing RPROVIDES 'foo-native' i have somenting missing in foo? Sep 21 17:22:04 aj_c: foo-native would be something on the build host, not the target Sep 21 17:23:56 ?? Sep 21 17:25:23 the -native suffix denotes things built for the build host Sep 21 17:26:25 anything build for the build host would not be appropriate in RDEPENDS Sep 21 17:27:13 RP: I had an idea a while back, but then forgot about it, and just found it buried in one of my todos. I was thinking about prototyping a bitbake change which would block access to variables which bitbake doesn't know about as vardeps in that context, most likely through a wrapper object around d Sep 21 17:29:12 * kergoth ponders Sep 21 17:56:31 perhaps i have a misunderstood does DEPENDS pulls up the dependencies to been build on the target? Sep 21 17:59:00 i just explained this. Sep 21 17:59:46 since what im looking is for my bar recipe have a DEPENDS to foo but what im looking is to foo be installed since bar needs it Sep 21 18:00:10 again, does bar need foo to compile, or only to run? Sep 21 18:00:19 only to run Sep 21 18:00:27 as i said earlier: "DEPENDS is for build time dependencies -- bar needs foo to be able to configure/compile " "if it only needs it to run, on target, but not to build, then you want RDEPENDS_${PN} instead " Sep 21 18:00:47 ok soy RDEPENDS_${PN} = "foo" it will installed foo? Sep 21 18:00:59 that's what rdepends means, yes Sep 21 18:01:09 see also the yocto project documentation Sep 21 18:41:02 kergoth: that is a good idea Sep 21 18:41:21 kergoth: initially might be a bit painful but would validate the metadata Sep 21 19:50:38 Creating a recipe that does not produce any packages but copies one file to sysroot for later pickup by another recipe. What's the right way to make this work properly with sstate cache? Neither custom do_install nor do_populate_sysroot seem to leave a ... Sep 21 19:50:42 ... record of the file in the cache. Thus, after new build from cache, the file could not be found by the second recipe. Sep 21 20:23:55 cesdv: you should install the file within do_install, but do not install directly into the sysroot - install it into a path under ${D} Sep 21 20:24:33 by default only a subset of the files installed by do_install get staged by do_populate_sysroot Sep 21 20:24:36 what kind of file is this? Sep 21 20:36:31 bluelightning: thanks! I'll try your suggestion. It's a part of bootcode image which will be assembled by another recipe. Sep 21 20:48:31 I'd like to build and generate ipk for two very similar machines (same processor, different board). I'm wondering how to best handle this while perserving PRSERV state, right now I build twice in the same build dir changing MACHINE= but I get errors about version going down for the odd packages that are machine specific. Sep 21 20:49:21 hey i want to build a kernel with networking enabled, what changes do i have to make to the config files? Sep 21 20:49:32 before i start bitbaking Sep 21 22:27:47 bluelightning: seems to work as expected. Had to include the temporary file into previously empty package to suppress the [installed-vs-shipped] warning. Sep 21 22:28:15 cesdv: right, you would need to do that yes Sep 22 02:14:42 Do I use flashcp with the .ubi or .ubifs RFS? **** BEGIN LOGGING AT Tue Sep 22 02:44:05 2015 **** ENDING LOGGING AT Tue Sep 22 02:59:59 2015