**** BEGIN LOGGING AT Mon Oct 14 02:59:58 2013 Oct 14 07:07:08 Crofton|work: no ;-) Oct 14 07:48:27 rburton: hi ! Oct 14 07:50:23 hi elbc Oct 14 07:51:04 rburton: I'm able to launch a core-image-weston with wayland/weston 1.3.0 but to do this, I had to remove weston-examples for core-image-weston.bb and to modify the init script of weston-init (because systemd was not able to execute the weston command but can only execute the weston-launch -u user command) . so my question is : what do I have to push exactly for patching ? Oct 14 08:00:21 elbc: the weston 1.3 release notes says that that there's a new option for installing examples Oct 14 08:00:36 rburton: just the commit relating to weston and wayland .bb files or do I have to push my changes for the init script and for the core-image-weston.bb file, I aske because it is a little confusing for me Oct 14 08:00:43 ok I'll take a look Oct 14 08:00:53 rburton: and for the init script ? Oct 14 08:01:10 feel free to send me everything you have as patches and i'll sort through it Oct 14 08:01:21 ok thanks ! Oct 14 08:02:44 rburton: what is the similar of ~/oe/org.openembedded.dev/recipes/ in Yocto? Oct 14 08:02:56 rburton: where did you found the 1.3.0 notes ? Oct 14 08:03:21 rburton: did you find sorry Oct 14 08:04:05 rburon: I was following this link(https://pixhawk.ethz.ch/tutorials/omap/dsplink) for dsp and it says to save the downloaded files in /recipes Oct 14 08:05:20 elbc: http://lists.freedesktop.org/archives/wayland-devel/2013-October/011419.html Oct 14 08:08:06 rburton: hi there Oct 14 08:09:03 rburton: do you remember we talked about qvga? I discovered I lacked a remote MACHINE_FEATURE qvga used by sato. Oct 14 08:10:18 rburton: oh, and about formfactor, do you remember where are/were used DISPLAY_DPI and DISPLAY_SUBPIXEL_ORDER ? Oct 14 08:15:22 good morning Oct 14 08:22:10 ant_work: my hunch is those were only used by kdrive Oct 14 08:23:05 as I suspected Oct 14 08:24:08 is there any possible future use wrt DPI / aliasing? Can we remove that vars and have the PPI calculated on the fly? Oct 14 08:24:49 the sibling question is: do we still need to provide an xorg.conf or is now automagically recreated? Oct 14 08:25:12 ant_work: xserver doesn't need xorg.conf Oct 14 08:25:21 (sorry, I'm not up to-date, I don't spend any time in userland nowadays...) Oct 14 08:25:34 aha Oct 14 08:25:47 ant_work: it can auto-detect all HW so you only need it if you have stupid drivers (EMGD...) or need to override specific options Oct 14 08:26:11 ok Oct 14 08:27:34 rburton: Hi, Oct 14 08:27:49 I was following this link(https://pixhawk.ethz.ch/tutorials/omap/dsplink) for dsp and it says to save the downloaded files in /recipes, what is the similar of ~/oe/org.openembedded.dev/recipes/ in Yocto? Oct 14 08:32:49 zaif: if it's saying to use recipes/ then it's quite likely the recipe will need to be ported anyway Oct 14 08:33:17 zaif: but oe-classic recipes/ is just anywhere in a recipes- directory in yocto systems Oct 14 08:34:18 rburton: I guess we needed xorg.conf for i.e. Option "Rotate" "CW" Oct 14 08:35:03 rburton: this should take DISPLAY_ORIENTATION Oct 14 08:35:42 ant_work: is it machine-specific? then a machine-specific config fragment in /etc/X11/xorg.conf.d/rotate-display.conf would make just as much sense. Oct 14 08:36:10 rburton: do I need to save the downloaded files in ~/yocto/build/downloads? and bitbake it? Oct 14 08:36:13 rburton: heh, I fear tehre are different layers of dust to clean here... Oct 14 08:36:50 zaif: if by downloaded files you mean sources, i guess so Oct 14 08:37:00 * rburton -> get dressed Oct 14 08:37:07 i only walked by to turn the computer on Oct 14 08:37:27 rburton: ok Oct 14 09:25:23 Hi all, After adding meta-ti to bblayers.conf, i get no recipes available error for these recipes: mesa_git.bbappend, mesa_9.1.6.bbappend, init-ifupdown_1.0.bbappend Oct 14 09:25:46 added this path: require yocto/poky/meta-ti/recipes-graphics/mesa-omap3-common.inc Oct 14 09:26:47 where am I doing wrong? Oct 14 09:27:17 those are in oe-core, so you must have accidently deleted meta/ from your bblayers.conf Oct 14 09:28:52 rburton: no meta\ exists in my bblayers.conf. Oct 14 09:29:34 that's the problem then Oct 14 09:29:49 are you using poky, or oe-core+bitbake, or what? Oct 14 09:32:39 i meant yes, the meta\ exists in bblayers.conf Oct 14 09:32:45 i am using poky Oct 14 09:32:54 sorry for the confusion Oct 14 09:39:54 so then I predict you are using an old release of poky, but meta-ti from git master. Oct 14 09:43:08 rbuton: I built poky from github(https://github.com/gumstix/Gumstix-YoctoProject-Repo), 1 week ago, deleted meta-ti and added meta-ti separately from http://git.yoctoproject.org/cgit/cgit.cgi/meta-ti/ Oct 14 09:44:00 zaif: that's a rather old poky release Oct 14 09:44:16 the release was called danny, so you want to use the danny branch of meta-ti Oct 14 09:44:41 though why did you delete and re-add meta-ti? Oct 14 09:45:38 before building meta-ti, i thought if i clone the latest then it will be updated Oct 14 09:47:08 rburton: gumstix-yoctoproject-repo is old? or meta-ti is old? Oct 14 09:47:44 gumstix-yoctoproject-repo is pulling an old release of poky Oct 14 09:47:50 and presumably a matching meta-ti Oct 14 09:48:02 if you want a newer meta-ti then you'll want to update poky to match. Oct 14 09:49:51 where can I pull the updated/latest version of poky? Oct 14 09:50:00 from yocto website? Oct 14 09:53:13 the poky directory that repo created is a clone from yocto. do a fetch in there Oct 14 09:53:41 then you can checkout either origin/dora (about to be released) or origin/dylan (latest release), checking out the same branch in meta-ti Oct 14 09:53:49 Morning all! Is there a quick and dirty way in yocto to enable sparse? Oct 14 09:54:14 And is anyone going to the Edinburgh do? I'm considering heading up! Oct 14 09:55:04 rbuton: ok. Thank You. no recipes available error was solved when i added meta-ti from danny(I am using poky-danny). I will do the fetch. Oct 14 09:55:20 zaif: if you have no reason to use master, then danny should work. Oct 14 09:55:30 and if that's what gumstix recommend, i'd stick with it Oct 14 09:55:49 rburton: ok Oct 14 09:55:50 pev: lots are heading to edinburgh, as there's a yocto developer day there Oct 14 09:56:57 Ace... hopefully will make it - will buy you a pint :-) Oct 14 10:03:37 I was looking at the schedule and thinking I ought to do the "Yocto for dummys" track :-D Oct 14 10:10:54 hello, is it possible to include file from another layer in a recipe in my custom layer? Oct 14 10:14:16 beuh: http://www.openembedded.org/Layers_FAQ#How_do_I_include_an_inc_file_from_another_layer.3F Oct 14 10:15:07 bluelightning, thanks Oct 14 10:53:50 So anyway... original question : Is sparse (or any other static analysis tool) already integrated into Yocto and easy to just enable or do i need to use them manually, I guess via devshell? Oct 14 11:13:07 Hm, I've got a bbappend that I'm using in my layer to override some kernel defconfig settings from the base layer but for some reason some of the CONFIG_ vars I want to unset don't get unset...? Any ideas or suggestions of directions to look for working examples to see if I've done something obviously silly? Oct 14 11:16:54 Hi, can somebody please tell why compiling a static binary with x86 toolchain (SDK created by -c populate_sdk) fails ? Oct 14 11:18:44 this is the failure log while compiling a simple hello world program as static: http://pastebin.com/2QKzRyQ7 Oct 14 12:29:15 maxin: do you have the staticdev packages installed? Oct 14 12:29:22 maxin: looks like its missing libc.a Oct 14 12:57:59 RP: sorry, I was out. Thanks! I will check it Oct 14 13:10:40 RP: that did the trick . Thank you very much ! Oct 14 13:14:05 maxin: great :) Oct 14 13:16:39 RP: most of the cross tool-chains support static linking as a default option. Can we do the same for Yocto toolchain as well ? Oct 14 13:18:34 considering static linking doesnt work at all on some of the toolchains... Oct 14 13:18:36 <_< Oct 14 13:18:42 (im looking at you uclibc toolchain) Oct 14 13:20:42 maxin: Its easy enough to include, it just makes the tarballs a lot bigger and in reality not many people use it Oct 14 13:21:37 Daemon404: RP: ok. In that case, can we document this behavior somewhere ? Oct 14 13:21:58 well... Oct 14 13:22:01 maxin: I'm sure Scott would love a patch to the manual :) Oct 14 13:22:01 at one point it worked Oct 14 13:22:06 then it didnt Oct 14 13:22:09 maybe it works now Oct 14 13:22:20 We Shall See Oct 14 13:22:48 any documentation would have to probably be a table of "which libcs play nice statically with yocto" Oct 14 13:22:51 i think. Oct 14 13:23:21 Daemon404: uclibc should work but it isn't used by as many people so it sees less exposure :/ Oct 14 13:24:02 i know Oct 14 13:24:20 i used to have a yocto uclibc buildbot for us (libav/ffmpeg) Oct 14 13:24:29 ill check if it works and/or fix it... bring it back Oct 14 13:24:42 s/buildbot/regression testing and buildbot/ Oct 14 13:25:10 RP: will send an update :) Daemon404: I generally don't use uclibc (though used it for a project in the past) Oct 14 13:25:42 ive used it to easy binary pushes in various situations in the past Oct 14 13:25:47 i know its evil... Oct 14 13:26:05 but glibc is more evil Oct 14 13:33:09 btw a request for faq entry: how on earth do you pronounce "poky" Oct 14 13:33:14 pocky or pokey? Oct 14 13:33:45 Daemon404: the former Oct 14 13:33:52 it's definitely a FAQ :) Oct 14 13:34:08 when i interned at WRS, nobody seemed to know. Oct 14 13:36:01 I'm pretty sure Mark, Jeff and Jason know how to pronounce it at least :) Oct 14 13:36:21 and Bruce as well Oct 14 13:41:52 In the Yocto kernel dev guide it mentions you can create a kernel "fragment" file which you can use to change a subset of the kernel configuration file. It seems to imply that you use this to "enable" certain options. What's not clear is if you can use the same mechanism to "disable" a kernel configuration item? Oct 14 13:42:34 Do the kernel "fragment" files allow you to turn-off an option that's enabled the the defconfig? Oct 14 13:42:49 gschmottlach: you can, it just needs to do: # CONFIG_SOME_VALUE is not set Oct 14 13:42:51 maxin: if it is for a few binaries consider klibc Oct 14 13:43:12 gschmottlach: (just as with a full config) Oct 14 13:44:04 having a quick browse through DISTRO/MACHINE_FEATURES is seems as though ext3 is specificed in a tune file, but is never used anywhere, ext2 exists and is used, but not ext3. Does anyone have any light to shed on this? Oct 14 13:44:47 ant_work: thanks. Haven't looked into klibc yet. here, the project requires static linking for c and c++ projects and the first toolchain version was based on eglibc.. Oct 14 13:45:14 Okay, so effectively the fragment file almost has the same syntax a a regular kernel config file . . . but just contains those options you want to tweak (either enable or disable), right? Oct 14 13:45:35 gschmottlach: correct Oct 14 13:46:00 Thanks for the help Oct 14 13:46:27 gschmottlach: the syntax is the same, not almost ;) Oct 14 13:47:00 perfect . . . thank you. Oct 14 13:49:47 dear all i'm facing an issue with mp3 decoder on imx6q board, i found this https://bugzilla.yoctoproject.org/show_bug.cgi?id=4449 Oct 14 13:49:48 Bug 4449: normal, Medium, 1.5, daiane.angolini, RESOLVED FIXED, GStreamer mp3 playback on Nitrogen6X - seg faults Oct 14 13:49:54 but i dont know what is MM Oct 14 14:05:03 does anyone have a uclibc machine with sdk they can run a five-line C program on? Oct 14 14:07:34 http://pastebin.com/ZzEHH18v <-- is this intentional? Oct 14 14:07:42 meta-toolchain has a hard dep on eglibc? Oct 14 14:12:30 Daemon404: hmm, how did you select uclibc? Oct 14 14:13:04 $ tail -2 conf/local.conf Oct 14 14:13:04 TARGET_OS="linux-uclibc" Oct 14 14:13:04 POKLIBC="uclibc" Oct 14 14:13:17 ... too bad i typo'd POKY Oct 14 14:13:20 * Daemon404 facedesk Oct 14 14:13:44 oh Oct 14 14:13:48 doesnt matter. still fails the same. Oct 14 14:16:16 looks like i needed TCLIBC Oct 14 14:16:19 didnt see it in the docs Oct 14 14:22:10 Daemon404: right, TCLIBC is the way to set it Oct 14 14:22:28 Daemon404: you don't need to set TARGET_OS, that will be set automatically from TCLIBC Oct 14 14:22:52 (TCLIBC replaced POKYLIBC about 3 or 4 releases ago I think) Oct 14 14:24:30 i simply followed the poky handbook (which is the only thing google ever returns for poky documentation Oct 14 14:24:39 which mentions TARGET_OS but not TCLIBC Oct 14 14:24:43 gah, we really must kill that document Oct 14 14:24:47 it is positively ancient Oct 14 14:25:01 perhaps make it more clear *where* the proper document is Oct 14 14:25:07 its very non-obviou Oct 14 14:25:07 s Oct 14 14:25:08 the current documentation is always available from here: http://www.yoctoproject.org/documentation Oct 14 14:25:36 i see Oct 14 14:25:46 handbook dominates search results Oct 14 14:27:00 .. i still cant find te poky documents off what you linked Oct 14 14:27:03 it's a maze of lines Oct 14 14:27:06 that leads nowhere Oct 14 14:27:10 s/lines/links/ Oct 14 14:27:51 there is no specific "Poky" manual anymore; the main manuals are the Yocto Project Reference Manual and the Yocto Project Development Manual Oct 14 14:28:10 there it is... Oct 14 14:28:11 TCLIBC Oct 14 14:28:11 Specifies which variant of the GNU standard C library (libc) to use during the build process. This variable replaces POKYLIBC, which is no longer supported. Oct 14 14:28:12 (which is why the old docs keep on appearing - the current documentation is for Yocto, not Poky) Oct 14 14:28:17 well this is just downright incorrect Oct 14 14:28:26 now all of the libc are gnu Oct 14 14:28:56 where the former is a reference for specific variables/classes/features and the latter is more of a task-oriented guide on how to accomplish things Oct 14 14:28:56 rburton, yes but yocto != poky Oct 14 14:28:59 which is my point Oct 14 14:29:05 yocto is a superset Oct 14 14:29:14 sure, that's my point too Oct 14 14:29:34 i wasn't are tclibc was a GNU project these days Oct 14 14:29:46 tclibc is a local.conf var Oct 14 14:29:53 set to eglibc or uclibc Oct 14 14:29:54 Poky is the reference build system for the Yocto Project; but information given in the Yocto Project manuals also applies to other systems that are based on the Yocto Project Oct 14 14:29:56 latter sure isnt gnu. Oct 14 14:30:07 then we should correct that, I'd have to agree Oct 14 14:30:55 wouldn't OE and bitbake be the reference Oct 14 14:31:05 poky is listed as a yocto project. Oct 14 14:31:10 not a "reference" Oct 14 14:31:20 it has it's own site too. with incorrect docs. Oct 14 14:31:23 its* Oct 14 14:31:30 yocto even links to it in some areas. Oct 14 14:31:45 it's all a bit schizophrenic Oct 14 14:31:50 Daemon404: please report any links from yocto to poky you find - they should have all been destroyed Oct 14 14:31:58 and the pokylinux site is up for historical purposes, really. Oct 14 14:32:19 some products have only just stopped using pre-yocto poky, believe it or not. Oct 14 14:32:36 https://www.yoctoproject.org/tools-resources/projects/poky Oct 14 14:32:48 it links directly to it Oct 14 14:33:07 and thats is linked directly from tools/resources Oct 14 14:34:13 Daemon404: right, but nothing likes from the front page of the poky site to the old poky docs Oct 14 14:34:33 ah, the quick start link goes to a really old version of the yocto docs Oct 14 14:35:05 yes Oct 14 14:35:50 FWIW, I am going to ensure that the old manual is replaced with a link to the new documentation Oct 14 14:36:59 bluelightning: hopefully that will help the current docs google-juice Oct 14 14:37:46 ;) Oct 14 15:02:29 RP, tested... static uclibc linking works again in poky git master Oct 14 15:02:33 cheers. Oct 14 15:08:26 Daemon404: good to know :) Oct 14 15:09:08 * Daemon404 ponders on ressurrecting his tilera toolchain or not... likely not due to lack of available hw or emulators Oct 14 15:19:52 i spoke too soon Oct 14 15:20:14 libc.a and libpthread.a have overlapping symbols Oct 14 15:20:20 they both define __errno_location Oct 14 15:22:38 http://patchwork.ozlabs.org/patch/280399/ Oct 14 15:22:40 speak of the devi. Oct 14 15:22:42 devil* Oct 14 15:36:03 elbc: somewhat confused how you made weston compile, unless you edited your sysroot manually Oct 14 15:37:16 halstead: ping Oct 14 15:37:34 Hi scottrif Oct 14 15:37:56 halstead: hi - issue with links... I will switch to private messaging Oct 14 17:19:38 RP: any idea about mulitlib regression reported here https://bugzilla.yoctoproject.org/show_bug.cgi?id=5333 Oct 14 17:19:39 Bug 5333: major, Medium, 1.5.1, laurentiu.palcu, NEW , meta-toolchain: no package provides libc6 >= 2.18 when multilib enabled Oct 14 17:19:57 RP: I am seeing the same on ppc64/x86_64 as well on 1.5 Oct 14 17:26:35 Hi all, Does capturing Images using opencv with GUI works on gumstix-console-image? or should I do capture without GUI? Oct 14 17:32:31 bluelightning: i updated the nginx recipe but i got lost yesterday building a gnat bootstrap for arm Oct 14 18:18:00 All of the documentation links on yoctoproject.org are broken. Oct 14 18:24:42 Most of the links at https://www.yoctoproject.org/documentation/current redirect to the same page. Oct 14 18:28:33 halstead: ^^^^ Oct 14 18:32:17 * sgw_ off to swim Oct 14 18:42:19 unable to get yocto documentation? all links ref to https://www.yoctoproject.org/documentation/current Oct 14 18:43:41 zaif: works for me? Oct 14 18:44:01 oh! Oct 14 18:44:03 it doesn't Oct 14 18:44:05 how odd Oct 14 18:44:59 rburton:ya. searching for getting started manual. Oct 14 18:45:30 zaif: you can checkout yocto-docs and build it yourself until its fixed Oct 14 18:45:52 just prodded scott, or maybe it's something halstead needs to fix Oct 14 18:46:18 rburton. now working Oct 14 18:46:25 halstead: ping Oct 14 18:52:18 halstead: The links on the http://www.yoctoproject.org/documentation page are not working correctly. Something we did to fix that google search stuff made them not work. Try going to the doc landing page, clicking on "view all" for current, and then clicking on any particular manual. You are taken back to the landing page rather than having the manual open up. Oct 14 19:17:21 RP: I have tried http://pastebin.com/3qPWqtWU for bug #5333 Oct 14 19:17:22 Bug https://bugzilla.yoctoproject.org/show_bug.cgi?id=5333 major, Medium, 1.5.1, laurentiu.palcu, NEW , meta-toolchain: no package provides libc6 >= 2.18 when multilib enabled Oct 14 19:17:29 but no use Oct 14 19:21:29 scottrif, sgw_ I'll get it fixed. Oct 14 19:22:19 scottrif, sgw_ Bad regex repaired. Oct 14 19:26:13 dbox, Doc redirects are fixed. Thanks! Oct 14 19:31:58 great, thank you Oct 14 19:47:41 halstead: thanks! Oct 14 19:47:55 Any time. Oct 14 20:16:23 seebs: around? There might be some minor issues with your binutils alternatives patches Oct 14 20:22:28 I am around. And I wouldn't be surprised if there were; they have so-far worked for us, but I could have missed stuff. Oct 14 20:24:59 seebs: ld.gold and embedspu did not exist, also your patch synopsis don't have the right format (binutils: ....) Oct 14 20:25:41 The ld.gold and embedspu thing is a Known Issue which I don't know how to correct: Both *could* exist, but may not *always* exist. Oct 14 20:26:06 The symlink thing can do this by just creating simlinks for whatever ends up existing, but I can't see a way to do conditional-alternatives. Oct 14 20:40:18 So, if there is a reasonable mechanism for making the alternatives stuff handle embedspu and ld.gold, I'd be interested in updating it. But I couldn't find one; I don't see any easy way to tell in advance whether the package will produce those. Oct 14 20:41:12 seebs: I will take a look also, I can't think of one off the top of my head, but we can't have the warnings. Oct 14 20:42:04 It might make sense to just not give WARNINGS for an absent alternative in some cases. Oct 14 20:42:21 you should be able to figure out if those two items are going to exist "early" Oct 14 20:42:21 the gold thing I thought was based on a configure argument.. Oct 14 20:42:33 and I suspect the embedspu is also based on an arch flag or something Oct 14 20:43:03 It might be. Part of the problem is, if there is a definite rule for when they get created, I don't know it... I've never actually seen embedspu, I'm not even sure it ever happens. Oct 14 20:44:54 just have to put into the python chunk if ld.gold .... if embedspu Oct 14 20:45:15 Yeah. But I don't have any clue what the conditions are. I've never seen either get generated... Oct 14 20:46:26 seebs, this was with a world build with qemumips Oct 14 20:47:29 Pretty sure embedspu is for Cell, so it's probably a rare subset of powerpc. We might well be able to just completely ignore it and assume no one will ever care. Oct 14 21:20:57 khem: not looked at that one yet... **** ENDING LOGGING AT Tue Oct 15 02:59:58 2013