**** BEGIN LOGGING AT Thu Sep 10 02:59:58 2015 Sep 10 07:27:20 morning folks Sep 10 07:35:23 good morning Sep 10 14:20:41 does anyone know of a good image viewing application for wayland/weston? I just want my image to display some pictures after booting, for a demo Sep 10 14:27:58 xerent: gst-player is very spartan but it's in oe-core... I don't remember testing on weston, but try "gtk-play [filenames]" after including gst-player in the image Sep 10 14:28:16 jku: thanks, I'll try it Sep 10 14:34:25 Nothing RPROVIDES gst-player Sep 10 14:36:15 I guess I don't have the recipe Sep 10 14:41:40 nah I'm too old version Sep 10 14:46:03 xerent: have you tried weston-image? Sep 10 14:47:28 it should be a part of weston-examples package Sep 10 14:48:27 yeah, seems like it's already part of my image then Sep 10 14:48:36 root@vit:~# weston-image Sep 10 14:48:36 error: XDG_RUNTIME_DIR not set in the environment. Sep 10 14:48:39 \o/ Sep 10 14:51:59 xerent: export XDG_RUNTIME_DIR=/tmp ? Sep 10 14:52:28 yep, I stole the path from the weston init script :) Sep 10 15:29:52 after letting few packages to use gcc only world build for OE-Core looks pretty clean for x86-64 and aarch64 http://errors.yoctoproject.org/Errors/Build/8628/ and http://errors.yoctoproject.org/Errors/Build/8630/ Sep 10 15:30:07 thats with clang Sep 10 15:39:22 khem`: gj! Sep 10 15:43:39 help Sep 10 15:48:57 ? Sep 10 15:49:01 HELP Sep 10 16:22:37 ~ask Sep 10 16:22:57 we would need Crofton|work bot in this channel too :-D Sep 10 16:23:40 hi guys, in my yocto build system, where can I see the procedures for creating my rootfs, for example core-image-xx.tar.bz2? Sep 10 16:24:01 This is because I need to add some binaries into the rootfs during build Sep 10 16:28:52 weston: meta/recipes-core/images/core-image-minimal.bb Sep 10 16:29:03 weston: find meta | grep image Sep 10 16:36:27 mckoan|away: thanks. But I dont see any procedures for compressing Sep 10 16:42:14 Hi all , any other inputs? Sep 10 16:50:33 JaMa: next I am going to go for QT5 Sep 10 16:50:46 so meta-qt5 will be interesting Sep 10 17:43:28 is there a way to tell wic to pad the image size out beyond the end of the rootfs? Sep 10 17:43:49 hmm Sep 10 17:46:00 Is there any way to set PREMIRRORS to a relative path? Sep 10 17:46:08 or use some variable (I was unsuccessful with ${TOP_DIR}) Sep 10 17:49:59 not enough information to say, would need to elaborate Sep 10 17:50:40 I'm trying in local.conf to set a pre mirror to file:// Sep 10 17:51:01 any fixed starting point really, I just don't want local.conf tied to any absolute path Sep 10 17:51:26 to be specific I'm using SOURCE_MIRROR_URL and inherit own-mirrors, but the principle is about PREMIRRORS Sep 10 17:55:05 I know it's odd :-) Most people have a fixed location for their mirror. In this case I'm working on a setup where one is provided with the project. Sep 10 17:56:11 Don't know the location this setup will be cloned into by the user. so bblayers are addressed relatively using ${TOP_DIR}/../meta-whatever, need similar for local.conf Sep 10 17:59:07 Is there any variable with a path in local.conf? Environment variable? Sep 10 18:00:04 A way to insert a python snippet to figure out an absolute path to... well any part of the project really? Just need any fixed point to add the relative path to. Sep 10 18:02:23 gunnarx: ${TOPDIR} sounds like a perfect place to go based on, if you want it reltaive to build dir, otherwise ${COREBASE} to be relative to oe-core/poky Sep 10 18:02:34 and it should work, right? Sep 10 18:03:18 wait... seems to be working, once I did it relative to DL_DIR Sep 10 18:03:21 if it doesn't, you have something else going on. file:// mirrors work fine (are known to work fine) Sep 10 18:04:16 hmm, no idea what mistake I did with TOP_DIR, I was sure the path was right. Sep 10 18:04:36 check the do_fetch log. there are debug messages showing the url replacements being applied from the mirror lines Sep 10 18:04:43 good idea Sep 10 18:04:44 so you can make sure its returning what you wanted it to return Sep 10 18:05:27 i have a standalone python script somewhere around here that just applies mirror lines to a url and gives you the results, so you can tweak PREMIRRORS/MIRRORS to make sure they're right Sep 10 18:05:35 not sure where i put it, but maybe should submit it to contrib Sep 10 18:06:53 yeah, checking now, i don't think it should need heavy debugging Sep 10 18:07:13 mostly needed to confirm my approach was ok Sep 10 18:18:20 kergoth, back on track with that thing, thanks. Sep 10 18:18:26 NP Sep 10 18:18:28 er, np Sep 10 18:18:48 I have another thing though but I haven't fully confirmed it either... Sep 10 18:19:20 so it seems a git repo is fetched over network (attempted, since I enabled BB_NO_NETWORK) even though it exists in the mirror Sep 10 18:19:50 I haven't tracked it down fully, but one theory is this seems to happen if SRCREV is not defined to absolute hash Sep 10 18:20:00 rather the git URL includes a tag=xxx Sep 10 18:20:34 Do you think that is something? Might bitbake be fetching the repo if nothing else to see that the tag did not move? :) Sep 10 18:21:36 sorry don't want this to be a wild goose chase, but as I said I'm trying to narrow down why this happens for some repos only, it seems Sep 10 19:03:40 khem: all your patches for musl "Unset need_charset_alias when building for musl", you say it should be fixed in gnulib. has that been fixed yet? Sep 10 19:03:59 rburton: trying to determine if you meant "it is fixed upstream" or "someone needs to fix it upstream" Sep 10 19:04:42 musl community has been discussing it I dont think a final resolution is reached yet Sep 10 19:05:10 Not Required isn't a valid upstream-status and they're the only ones firing a warning from my tool :) Sep 10 19:05:22 OK change it to Pending Sep 10 19:05:27 ok :) Sep 10 19:05:36 mostly it will be changed in gnulib Sep 10 19:06:03 yeah, and bubble down Sep 10 19:06:05 Total patches found: 1686 Sep 10 19:06:05 Patches missing Upstream-Status: 90 Sep 10 19:06:05 Patches with malformed Upstream-Status: 0 Sep 10 19:06:06 Patches in Pending state: 734 Sep 10 19:06:19 the 0 is an improvement at least Sep 10 19:07:57 (down from 94) Sep 10 19:13:40 I should just send them to components e.g. coreutils etc. itself Sep 10 19:13:49 in interim Sep 10 19:13:56 its still valid fix Sep 10 19:14:03 if gnulib is not fixed Sep 10 19:14:13 just at a different place Sep 10 19:14:17 gnulib route is long Sep 10 19:14:37 rburton: agreed Sep 10 19:14:58 may be we should start a Wiki page on pending patches Sep 10 19:15:09 and see if folks sign up for some of them Sep 10 19:15:16 its quite an effort to upstream Sep 10 19:15:29 700 odd parches Sep 10 19:41:38 my patch review tool is in rossburton/ross-tools on github, btw Sep 10 19:41:51 currently just does upstream-status verification Sep 10 20:51:36 Does anyone know if the devtool support SDK updating in master yet? Sep 10 22:13:45 kergoth: --extra-space is what you are looking for Sep 10 22:14:31 by default, it uses IMAGE_EXTRA_SPACE which is 10MiB Sep 10 22:14:34 heh, i knew we'd end up needing WIC_ARGS for IMAGE_CMD_wic Sep 10 22:15:11 no, --extra-space goes in the wks Sep 10 22:15:25 oh, right Sep 10 22:15:34 okay, thanks, will gave that a shot Sep 10 22:16:24 Hmm, should add a script to the rootfs to grow the ext4 rootfs to the size of the partition it lives in, i know the debian beaglebone rootfs provides one Sep 10 22:16:33 obviously, if you use it along with --align, you will probably end up with more extra space Sep 10 22:17:06 you should probably not need that Sep 10 22:17:22 wic creates the partition and then populates it with the rootfs Sep 10 22:17:43 so it has the correct size Sep 10 22:17:46 wic created me a 250 meg image for a rootfs about that size, and i'm writing it to a 8gb sd card Sep 10 22:17:51 clearly not ideal. Sep 10 22:19:34 fray: huh, so the new failing mirror recursion test, i'm guessing each mirror replacement ends up altering one after another, recursively, first one, then the other? so it's fine now when you have just the one entry, but when you have two, we can't catch that recursion bouncing back and forth? Sep 10 22:21:36 kergoth: why don't you use --size 7G for example ? :) Sep 10 22:22:02 yeah, probably will. was hoping to avoid hardcoding particular disk sizes in the wks, but guess thats probably the best way to go Sep 10 22:22:53 either that or do what i was thinking before, but would have to both extend the partition size and the fs size based on disk size Sep 10 22:22:53 yeah, I have different wks for different SD card sizes Sep 10 22:24:38 ah, cool Sep 10 22:25:02 working with the beaglebone black at the moment Sep 10 22:26:45 I love my BBB. Sep 10 22:26:49 hmm, wonder how best to integrate configuring configfs into the startup. maybe just a configfs.d which holds shell scripts Sep 10 22:26:56 better than any other dev. board Sep 10 22:26:58 would like to get the gadget stuff configured better Sep 10 22:27:23 yeah, it's pretty nice. i like how compact and light it is, and able to be usb powered, but with more onboard capes than the old beaglebone like hdmi Sep 10 22:27:38 * kergoth brings it and an lcd/touch panel, and cables up to the coffee shop to work Sep 10 22:27:50 I like the boot from 4Gig of onboard flash. Sep 10 22:28:05 beats the hell out of booting from SD card. Sep 10 22:29:43 oh yeah, sd boot is SO SLOW once you get used to emmc Sep 10 22:30:16 if they put 4gigs of flash (bootable) on any arduino mega2560 I'd be in heaven. Sep 10 22:31:30 these dev boards - they need to be bootable and run a real OS. Sep 10 22:31:50 and then, on top of that....do shields and sensors and motors and shit. Sep 10 22:32:49 o...and for christs sake...POE. Sep 10 22:33:01 down with wall warts. Sep 10 22:35:54 seriously, how much could a full blown dev board with everything + more goodies cost? Sep 10 22:36:25 the issue with POE is that it can be up to 60V Sep 10 22:36:31 802.11 and BT built in, of course Sep 10 22:36:39 so an LM7805 will not do Sep 10 22:37:06 others manage it. Sep 10 22:37:43 yeah, it is a matter of price Sep 10 22:37:43 but I suppose doing that prices it out of the starving college student kind of price range. Sep 10 22:37:51 I'm not saying it is not doable :) Sep 11 02:31:11 How do build just the kernel? Sep 11 02:32:34 I tried `bitbake -c compile linux-imx` but judging from the output it looks like it's building non-kernel stuff (like openssl). **** ENDING LOGGING AT Fri Sep 11 02:59:58 2015