**** BEGIN LOGGING AT Tue Jan 17 02:59:58 2012 Jan 17 09:13:49 morning all Jan 17 16:36:15 RP__, sgw: see the error in the logs. don't see it here, but it should be easy enough to fix. I'm on it. Jan 17 16:50:14 zeddii: cool, thanks Jan 17 16:51:19 trivial fix. I'm firing up an atom-pc build just to see it break and then test my fix. we can measure the breakage in hours now. Jan 17 17:22:47 * zeddii smashes his head into the desk. Jan 17 17:23:31 anyone seeing perl build errors in master ? I'm trying to test my kernel fix, but perl and python are failing. Jan 17 17:33:15 zeddii: failing how? Jan 17 17:33:23 * RP__ wonders if the bitbake.conf change did this somehow Jan 17 17:33:57 claimed that ./miniperl didn't exist. I moved my sstate aside and restarted, just so I could tunnel through to my kernel fix. Jan 17 17:43:41 what's with all the do_populate_setscene failures? Jan 17 17:43:53 erm, sysroot_setscene Jan 17 17:53:27 also, why does the cooker not name the -e output something other than cooker.log.? it's clearly not a log Jan 17 18:09:44 kergoth, which recipes are failing do_populate_sysroot_setscene? Jan 17 18:10:17 erm, don't remember offhand, but here on average 3 fail every build. i know shadow-sysroot does, but there's usually 2 others too Jan 17 18:10:22 will have to check the logs Jan 17 18:11:21 kergoth, probably other recipes which are using useradd. I'm looking into this issue today Jan 17 18:11:25 dbus/avahi/etc Jan 17 18:12:15 ah, cool Jan 17 18:13:00 the fact that they now show in the error summary combined with the fact that there's no output from those in the cooker log could cause a bit of user concern and support load Jan 17 18:13:02 heh Jan 17 18:14:57 kergoth, also want to confirm - were you doing a build with a populated sstate-cache? Jan 17 18:15:06 yeah Jan 17 18:15:10 90% at least Jan 17 18:15:16 makes sense Jan 17 18:17:17 sgw. great. perf builds on my updated master. so I can't see the breakage, which gives me less confidence when fixing it! Jan 17 18:17:37 zeddii: awesome! Jan 17 18:17:51 I have a reproducer with atom-pc on my machine Jan 17 18:18:50 that's what I just built. but without my fix, perf built. Jan 17 18:18:56 so I'm trying to see what I missed. Jan 17 18:20:15 what's the incantation if I just want to trigger the perf build ? I never do it individually. Jan 17 18:23:25 indeed. I have no error here. and hence nothing to fix. Jan 17 18:23:40 I just build with -c compile_perf Jan 17 18:24:00 zeddii: let me know what I can help with on that one. Jan 17 18:24:04 yep. just did that. builds like a champ. fully up to date master. sstate, downloads and tmp deleted this morning. Jan 17 18:24:40 well, let's see what happens with master and the AB then. I am rebuilding right now also. Jan 17 18:25:12 ok. I'll keep trying. I have a fix, but I'd have no confidence that it didn't do worse .. since I can't see the breakage. Jan 17 18:25:33 5724ee9c3a99519fea96446638910b727b89898d is my top commit. Mark's rpm space fix from today. Jan 17 18:27:13 If I get the breakage again with master, point me at your fix and I will test it. Jan 17 18:28:14 sounds good. Jan 17 18:49:49 zeddii: well, not sure what's going on but my build was also good here. Let's see what happens with the AB over the next couple of hours. Jan 17 18:52:24 sgw. ok. I'm back to debugging edison and workiong on the 3.2 tree now :) will wait to hear more on this one. Jan 17 18:52:36 Ok sounds like a plan Jan 17 19:04:42 zenlinux: ping Jan 17 19:04:53 hi msm Jan 17 19:05:12 zenlinux: just wanted to let you know, i'll be out for a while soon - but if you find something to fix the bug let me know and I can start some tests later tonight Jan 17 19:05:36 msm, definitely, I'll keep you posted and push something to a contrib tree if I think it's ready for testing Jan 17 19:05:40 zeddii: I wonder if there was an ordering issue, that once we built whatever was being depended on that issue went away, I know that's really vague, but that might be the issue. I will have to look at the file that was missing. Jan 17 19:06:22 could be. I don't like a mystery either. Jan 17 19:06:42 zeddii: I think I know the issue! Try setting your EGLIBC PREFERRED_VERSION to 2.15, that's what's different Jan 17 19:06:51 aha!! cool. doing that now. Jan 17 19:07:47 so when we do upgrade that we will be ready, you might have to rebuild python and tcl we had build issues from those, but I think I fixed the other ones in master. Jan 17 19:08:47 I'll dump my sstate and go completely clean on the rebuild. The patch in question is all about eglibc prep in the kernel, so it makes perfect sense. I'll make sure it builds for both cases. I actually hope this breaks :) Jan 17 19:10:56 you should not have to dump sstate for this. Jan 17 19:11:09 ok. so I'll be brave and leave it in place. Jan 17 19:12:43 zeddii: it's better that way, sstate is in much better shape now, we are using it on the AB regularly even flipping back and forth to 1.1.1 build with it in place. Jan 17 19:14:41 seems to be doing the right thing at the moment. watching the compiler and c library build. Jan 17 19:44:02 sgw, so my rebuild completed. I can still build with eglibc 2.15 and the latest kernel. let me know if it breaks for you. Jan 17 19:58:56 This is a place as good as any to ask about the gstreamer/video question - So, I have sato working on my target hardware, and I see a Video icon on the desktop and I can not play any video - MPEG 1, MPEG 2 and .mov files (H.264). I get the same error - Your GStreamer is missing a plug-in. Jan 17 19:59:24 How do I 1) include the plugin or 2) let GStreamer know where the plugin is Jan 17 19:59:38 My image is core-image-sato-sdk Jan 17 19:59:51 I am on Edison - 6.0 Jan 17 20:01:24 glxgears gives me 100 fps or so, so, the emgd driver is properly integrated with X, before that, I could only get 4-6fps Jan 17 20:01:50 MACHINE is crownbay Jan 17 20:02:31 Guest27563, is this the commercial license stuff autif is hitting? Jan 17 20:05:01 I have logs from video - -gst-debug-level=5 file:///myfile.mpg - if that will help Jan 17 20:05:47 Hi ztom :-) Jan 17 20:06:13 autif: hi autif Jan 17 20:07:34 autif: i've only tried using the command-line gst-launch, which should pick up the plugins, at least it did for me Jan 17 20:08:11 autif: e.g. gst-launch playbin2 uri="file:///sintel-1024-stereo.mp4" Jan 17 20:08:24 I tried that too - gstlaunch playbin2 usr="file://path/to/file.mpg" Jan 17 20:08:26 same error Jan 17 20:08:28 yup Jan 17 20:08:42 wow - look at my typos Jan 17 20:08:58 autif: or something like this manually: gst-launch-0.10 filesrc location=./big_buck_bunny_1080p_h264.mov ! qtdemux name=demux ! queue ! MixVideoDecoderH264 ! vaimagesink demux. Jan 17 20:09:38 are those ! or | Jan 17 20:10:07 ! Jan 17 20:10:12 autif: gstreamer uses ! as an analogue for | Jan 17 20:10:33 ok Jan 17 20:11:50 ztom: what would be the decoder for MPEG1 or MPEG2? I assume it will be the same Jan 17 20:11:53 autif: anyway, getting the right commandline can be tricky with gstreamer, i'm a novice Jan 17 20:13:08 autif: iirc the video acceleration is for h264 and mpeg4 and a couple others, see the EMGD doc for the list (and also commandline examples) Jan 17 20:13:13 ztom: "no element qtdemux" Jan 17 20:14:25 autif: do you see it in the gst-inspect output? Jan 17 20:14:43 ztom: can you please prompt me to the EMGD doc? Jan 17 20:15:17 I am not sure how to get to the gst-inspect output Jan 17 20:15:46 autif: somewhere on this page: http://www.intel.com/p/en_US/embedded/hwsw/software/emgd Jan 17 20:16:20 autif: the page has upgraded to emgd 1.10, but 1.8 which is what edison uses should be there too Jan 17 20:16:30 autif: just type 'gst-inspect' Jan 17 20:20:46 ztom - I can not understand the output of the inspect - http://pastebin.com/yKgUC9e3 Jan 17 20:21:33 what is expected for mpeg 1, mpeg 2 and H.264? Jan 17 20:24:49 autif: hmm, should show isomp4/qtdemux iirc Jan 17 20:26:09 autif: so this is plain old out-of-the-box edison crownbay? Jan 17 20:27:04 yes Jan 17 20:27:07 no tweaks Jan 17 20:27:09 yet! Jan 17 20:27:28 it is core-image-sat-sdk, not core-image-sato Jan 17 20:27:36 ztom Jan 17 20:27:45 autif: ok, i'll have to build and try here - it all worked fine last time i tried it Jan 17 20:30:58 OK, Will wait, meanwhile, I will send an email with all the details Jan 17 20:31:29 dvhart, incandescant, ztom - thanks a bunch Jan 17 20:31:58 autif: ok, will take a look and try things here Jan 17 21:07:46 zenlinux -- you here? Jan 17 22:43:45 RP__ or fray: Should Pseudo have a version 1.1.1 in bugzilla? I notice it has fewer versions than the other products. Jan 17 22:44:53 We try not to change it too much. :) Jan 17 22:45:33 I don't think that is needed.. Jan 17 22:45:41 That said, current version appears to be 1.2. Jan 17 22:45:56 And we do have 1.1.1 and 1.1.2 in our history. Jan 17 22:45:59 The current versions are 1.0, 1.1, and master. Jan 17 22:46:23 I didn't even know it was being tracked in bugzilla. Jan 17 22:46:40 *sniff* my little baby's all grown up. Jan 17 22:46:50 halstead: pseudo has its own versioning scheme Jan 17 22:47:11 Thank you all. I will leave it as is. Jan 17 22:47:48 seebs: its interesting seeing software out in the world :) Jan 17 22:47:51 It doesn't matter because pseudo doesn't have bugs, just user errors. Jan 17 22:47:53 yup.. Jan 17 22:48:01 seebs, I'd been monitoring htem for you. ;) Jan 17 22:48:13 fray: the users? :) Jan 17 22:48:22 the users and the bugzilla Jan 17 22:48:25 I think so far the WR branch has gotten one real bug report and one illucid bug report. Jan 17 22:48:33 * RP__ wonders how many errors he had Jan 17 22:48:39 er, has Jan 17 22:48:54 We got a bug report of the form "I made a large number of changes to the entire build system in the first public release of pseudo and I got error messages, FIX EVERYTHING". Jan 17 22:49:36 seebs: "Doctor, my arm hurts when I hit it" Jan 17 22:57:29 I still really want to know what he was actually doing, but I will probably never find out. Jan 17 23:02:43 back in about 30mins Jan 17 23:43:48 sgw: ping Jan 17 23:45:31 ant__: what's up? Jan 17 23:45:36 hi there Jan 17 23:45:59 I'd like to fix the 'touch symlink' issue which breaks each build here... Jan 17 23:46:23 I've verified I'm using touch binary provided by my host (Gentoo) Jan 17 23:46:48 so I'm not sure the '-h' option is supported by other distros Jan 17 23:48:03 ant__: your right about not support, I just checked ubuntu and it does not have the -h option. Jan 17 23:50:03 well, don't we have coreutils-native? Jan 17 23:55:33 ok, I'm looking at the sources in coreutils-8.14 and (-h) is supported Jan 18 00:24:22 gn Jan 18 01:40:58 and bitbake just segfaulted on me Jan 18 01:41:01 * kergoth sighs Jan 18 02:28:25 damnit, why the hell is do_compile_pref pulling in binutils explicitly, not the prefixed binutils virtual, resultin gin a build of binutils even when using an external toolchain? Jan 18 02:28:26 hrm Jan 18 02:28:50 mlprefix.. hmm Jan 18 02:29:01 * kergoth pokes around Jan 18 02:32:06 and binutils-native is being built now? jesus Jan 18 02:32:21 the amount of crap we need to compile a tiny image now is ridiculous **** ENDING LOGGING AT Wed Jan 18 02:59:56 2012