**** BEGIN LOGGING AT Tue Apr 26 02:59:57 2011 Apr 26 06:47:21 morning Apr 26 06:47:39 * Jay7 is looking for better place for goocanvas recipe Apr 26 06:47:54 should I place it into recipes/gnome/? Apr 26 06:48:05 GooCanvas is a canvas widget for GTK+ that uses the cairo 2D library for drawing Apr 26 06:48:20 http://live.gnome.org/GooCanvas Apr 26 06:52:48 well.. I'll place it in own directory because it have no relation with gnome DE Apr 26 06:57:42 03Yuri Bushmelev  07master * r659a8f6754 10openembedded.git/recipes/gpredict/gpredict_1.3.bb: gpredict: new recipe: Gpredict is a satellite tracking and prediction application Apr 26 06:57:45 03Yuri Bushmelev  07master * rbb030383c1 10openembedded.git/recipes/aewan/aewan_1.0.01.bb: aewan: new recipe: Aewan is a ascii-art/animation editor Apr 26 06:57:48 03Yuri Bushmelev  07master * r8be5ba0467 10openembedded.git/recipes/goocanvas/goocanvas_1.0.0.bb: goocanvas: new recipe: GooCanvas is a canvas widget for GTK+ Apr 26 07:15:37 good morning Apr 26 07:17:51 mckoan: moin... Apr 26 07:17:59 mckoan: any eta on xinput-calibrator? :) Apr 26 07:23:33 JaMa|Wrk: hi, actually I no longer care about Xorg related stuff because I'm using Qt now Apr 26 07:25:43 ok, then please update MAINTAINERS file Apr 26 07:47:47 03Paul Menzel  07master * r559a5fb4be 10openembedded.git/recipes/libav/files/omapfbplay-errorhandling.patch: Apr 26 07:47:47 omapfbplay: fix typos in `omapfbplay-errorhandling.patch` Apr 26 07:47:47 Signed-off-by: Paul Menzel Apr 26 07:54:35 what is the most lightweight browser we got in oe ? minimo ? Apr 26 07:54:45 links Apr 26 07:54:48 w3m Apr 26 07:54:57 telnet? Apr 26 07:55:03 hehehe Apr 26 07:55:13 ok got ya Apr 26 07:55:14 netsurf Apr 26 07:55:38 whats netsurf? Apr 26 07:55:45 browser Apr 26 07:56:41 hm what graphic's backend it is using Apr 26 07:56:48 gtk or fb Apr 26 07:56:55 hm oh fb Apr 26 07:56:56 as I understood from recipe Apr 26 07:57:20 i will check netsurf out .. maybe its the one enough for me purpose Apr 26 07:57:31 its the one from RiscOS isnt it? Apr 26 07:58:04 NetSurf is a multi-platform web browser for RISC OS, UNIX-like platforms (including Linux), Mac OS X, and more Apr 26 07:58:09 http://www.netsurf-browser.org/ Apr 26 07:58:34 well its designed for some really really slow cpus so should fly on modern hardware :-D Apr 26 07:59:07 but gtk wasn't designed to run on cpus w/o math :) Apr 26 07:59:38 yes I found the website Apr 26 07:59:44 hm it mentioned mostly gtk Apr 26 08:00:01 those old RiscPCs were almost clockwork :-D Apr 26 08:00:11 There are actually two versions of NetSurf available for Unix-like systems. In addition to the GTK front end, which is for normal desktop usage, there is a framebuffer front end for NetSurf which requires no windowing environment or GUI toolkits. It currently supports the Linux framebuffer and ABLE's framebuffer. Apr 26 08:00:19 from http://www.netsurf-browser.org/downloads/gtk/ Apr 26 08:01:25 jlime is using netsurf fb version iirc Apr 26 08:01:29 hm update the recipe to 2.7? Apr 26 08:01:47 B_Lizzard: ping Apr 26 08:01:50 wonder if we could build RiscOS in OE :-) Apr 26 08:02:04 woglinde: ask B_Lizzard :) Apr 26 08:02:12 yes yes? Apr 26 08:02:28 he and anarsoul are touching netsurf :) Apr 26 08:02:37 Ah, it works pretty well Apr 26 08:02:52 But you'll need a mouse emulator for non-touchscreen targets Apr 26 08:02:57 B_Lizzard: there is 2.7 version Apr 26 08:03:05 Ah, they released? Apr 26 08:03:08 yes Apr 26 08:03:17 Well, the fb version is from SVN Apr 26 08:03:26 hm ah Apr 26 08:03:27 Old SVN, but still. Apr 26 08:03:49 I'll have to update some of the peripheral recipes though Apr 26 08:04:01 gogo Apr 26 08:04:09 B_Lizzard: btw, I've pushed aewan Apr 26 08:04:21 so it is in OE master now Apr 26 08:04:25 Great, Jay7 :) Apr 26 08:04:44 Netsurf is farking great, though Apr 26 08:05:01 Renders Engadget, a complex website perfectly within 5 seconds on the Nanonote Apr 26 08:05:19 Makes you wonder what the Dillo people have been doing Apr 26 08:08:28 morning all Apr 26 08:08:51 hi bluelightning Apr 26 08:09:40 B_Lizzard okay it has no html5 Apr 26 08:09:56 thats maybee the only downside Apr 26 08:10:00 and flash Apr 26 08:10:13 iPhone have no flash :) Apr 26 08:10:32 Well, it doesn't have javascript either so it's got some road ahead Apr 26 08:10:50 hm.. JS is must have now :( Apr 26 08:11:17 hm oh Apr 26 08:11:22 I thought it has Apr 26 08:11:23 hm Apr 26 08:11:47 morning woglinde Apr 26 08:11:56 he ka6sox-away Apr 26 08:12:08 ka6sox spend some easter holidays? Apr 26 08:12:59 woglinde, just 1 day... Apr 26 08:13:10 back to it and still working. Apr 26 08:13:23 hm Apr 26 08:13:32 good morning Apr 26 08:13:49 there is someone who is on almost as much as me. Apr 26 08:14:04 bluelightning: now that opie 1.2.5 is out can we remove old stuff? Apr 26 08:14:19 ant our janitor Apr 26 08:14:23 *g* Apr 26 08:14:30 ^_^ Apr 26 08:14:48 ant_work: I did remove 1.2.3... are you suggesting removing 1.2.4 as well? Apr 26 08:14:55 I was doing home-cleanings and just continued that work at keyboard :) Apr 26 08:15:24 +1 to 1.2.4 removal :) Apr 26 08:15:52 bluelightning: strangely atm OPIE won't even boot... OOM Apr 26 08:16:16 ant_work: hmm, not good... do you have any more info? Apr 26 08:19:23 well, I was testing yesterday, very shortly Apr 26 08:19:36 console-image boots, x11-image too (no touchscreen) Apr 26 08:19:53 iir cudev caused OOM Apr 26 08:20:00 hmm netsurf is crappy slow to start Apr 26 08:20:01 on OPIE Apr 26 08:20:32 bluelightning: we finally need to reorder initscripts/modutils.sh/udev creap...then I'll retest Apr 26 08:21:33 thema is under discussion in oe-core and oe-dev ML Apr 26 08:21:47 ant cudev? Apr 26 08:21:51 whats that? Apr 26 08:22:01 s/iir cudev/iirc udev/ Apr 26 08:22:04 hm Apr 26 08:22:04 :) Apr 26 08:22:19 ant yes Apr 26 08:22:25 discussion I mean Apr 26 08:22:44 I hope is just bad interaction with OPIE init scripts Apr 26 08:23:06 (was not so broken few weeks ago...) Apr 26 08:23:26 btw, I was using minimal distro, not A. Apr 26 08:24:24 A. did recently some more blacklisting so I'll have to retest it Apr 26 08:24:53 (i.e. util-linux-ng hotplug udev conflict) Apr 26 08:25:18 minimal is bitrotting Apr 26 08:25:30 even psplash is broken :) Apr 26 08:25:42 oh...about this... Apr 26 08:26:29 we have to ask A. folks : how to get rid of psplash-angstrom if one distro uses psplash-zzap? Apr 26 08:26:42 me and Tartarus we got two commits reverted Apr 26 08:26:46 :p Apr 26 08:31:38 hi mickeyl Apr 26 08:47:08 03Paul Menzel  07master * r9fb55ee617 10openembedded.git/recipes/python/python-scons_1.3.0.bb: (log message trimmed) Apr 26 08:47:08 python-scons_1.3.0: add `python-native` to `DEPENDS_virtclass-native` Apr 26 08:47:08 A clean build fails with the following error [1]. Apr 26 08:47:08 | ERROR: Function 'do_compile' failed (see /home/sgh/oe-new-oe/build-vmc16/tmp/work/x86_64-linux/python-scons-native-1.3.0-r1/temp/log.do_compile.26596 for further information) Apr 26 08:47:08 | + cd /home/sgh/oe-new-oe/build-vmc16/tmp/work/x86_64-linux/python-scons-native-1.3.0-r1/scons-1.3.0 Apr 26 08:47:08 | + do_compile Apr 26 08:47:09 | + distutils_do_compile Apr 26 08:47:50 *sigh* paul with his oversized commit-msg Apr 26 08:59:12 heh Apr 26 08:59:23 morning woglinde Apr 26 09:24:15 pingswept: did you get nginx going? Apr 26 10:35:54 help with the problem. there is a fee to the AT91SAM9263 processor. The Linux kernel with ALSA 02.06.30 module - snd_atmel_ac97c. If I try to play a mono file - there is a message "aplay: set_params: 959: Channels count non available ". A stereo played normally. How do I get around? Apr 26 10:36:47 spirin try the kernel updates from ulf's branch Apr 26 10:36:52 otherwise mail him Apr 26 10:37:29 http://cgit.openembedded.org/cgit.cgi/openembedded/log/?h=ulf/linux-2.6.30-2011-04-16 Apr 26 10:41:53 ok. thanks Apr 26 10:43:54 re hrw Apr 26 10:43:57 hi esben Apr 26 11:18:51 wow dietlibc reworked Apr 26 11:25:04 ant_work: yes; but I expect the only the -git version works on ARM Apr 26 11:25:15 cvs did some progress but misses still a lot Apr 26 11:30:15 ant_work what? Apr 26 11:32:11 ensc dont you have commit rights? Apr 26 11:33:20 woglinde: I have Apr 26 11:33:42 but changes seem to be too big to submit them without discussion Apr 26 11:34:17 commit tcp-wrapper and bietlibc Apr 26 11:34:23 ups dietlibc Apr 26 11:34:28 you have my ack Apr 26 11:34:44 ok Apr 26 11:36:04 lunch now Apr 26 11:36:36 enjoy Apr 26 11:37:00 03Enrico Scholz  07org.openembedded.dev * reb55e5e8bc 10openembedded.git/recipes/tcp-wrappers/tcp-wrappers-7.6/ldflags.patch: Apr 26 11:37:00 tcp-wrappers: place LIBS after object files Apr 26 11:37:00 This fixes builds with '-Wl,-as-needed'. Patch changes the previous Apr 26 11:37:00 'ldflags.patch' files instead of adding a new one. Apr 26 11:37:00 Signed-off-by: Enrico Scholz Apr 26 11:37:00 Cc: Jeremy Lainé Apr 26 11:37:50 03Enrico Scholz  07org.openembedded.dev * r6d7d5df79a 10openembedded.git/recipes/dietlibc/ (19 files in 3 dirs): Apr 26 11:37:50 dietlibc: rewrote recipes Apr 26 11:37:50 dietlibc was splitted into target and cross recipes. These recipes Apr 26 11:37:50 build now the latest stable version, from upstream CVS or from a github Apr 26 11:37:50 repository with lot of fixes. Apr 26 11:37:50 Signed-off-by: Enrico Scholz Apr 26 12:15:08 Err what happened to tinderbox Apr 26 12:16:45 dont know Apr 26 12:16:48 setup your own Apr 26 12:27:28 damn, firefox still does not build Apr 26 12:27:43 JavaRocky, the design did not scale Apr 26 12:28:20 Crofton|work, in what sense? builds too long? Apr 26 12:29:03 it overwhelmed the server collecting data Apr 26 12:30:31 crazy. thanks Apr 26 13:00:04 Tartarus: new bluez no longer installs /etc/init.d/bluetooth script, so using update-rc.d ends with "update-rc.d: /etc/init.d/bluetooth: file does not exist", could you please check what else could be used from http://git.openembedded.org/cgit.cgi/meta-openembedded/commit/?id=3a493483b59cfe700400737a38b8147583685de6 ? Apr 26 13:06:14 crofton whats the problem with firefox now? Apr 26 13:08:27 any python distutils wizz present ? Apr 26 13:08:50 mlip not really Apr 26 13:08:53 but whats the problem? Apr 26 13:09:59 I am still having fun with gpsd: autotools guesses python and friends (looks like this is corrected), but calls setup.py on its own (not through disutils) Apr 26 13:10:13 this results in /usr/lib being in -L Apr 26 13:10:23 ... warning -> error Apr 26 13:10:33 03Philip Balister  07org.openembedded.dev * rce56683db5 10openembedded.git/recipes/uhd/ (4 files): uhd : Update to latest release. Both source and fpga images. Apr 26 13:11:03 I don't know where setup.py gets its base LDFLAGS from (its not LDFLAGS or PYTHON_LDFLAGS) Apr 26 13:11:31 i can append further paths using --libdirs when calling build_ext, but I can't get rid of the /usr/lib ;) Apr 26 13:11:46 hm Apr 26 13:12:04 hm did you made a grep? Apr 26 13:12:09 maybee its hardcoded Apr 26 13:12:17 woglinde, I should check again, I may not have pulled ... Apr 26 13:12:19 but didnt it worked? Apr 26 13:12:28 crofton paul comitted the fix Apr 26 13:14:00 woglinde, I did grep, but couldn't find anything relevant; thx only thing that is still set to /usr/lib is configure's libdir Apr 26 13:51:33 JaMa|Wrk: k, thanks Apr 26 14:08:09 03Paul Menzel  07master * r9b214725ad 10openembedded.git/recipes/tcp-wrappers/tcp-wrappers_7.6.bb: Apr 26 14:08:09 tcp-wrappers: add variable `HOMEPAGE` Apr 26 14:08:09 Signed-off-by: Paul Menzel Apr 26 14:08:20 03Paul Menzel  07master * r0e00471307 10openembedded.git/recipes/tcp-wrappers/tcp-wrappers_7.6.bb: Apr 26 14:08:21 tcp-wrappers: reorder recipe according to style guide [1] Apr 26 14:08:21 [1] http://openembedded.org/index.php/Styleguide Apr 26 14:08:21 Signed-off-by: Paul Menzel Apr 26 14:08:24 03Paul Menzel  07master * rd7be16e86a 10openembedded.git/ (2 files in 2 dirs): Apr 26 14:08:24 doc: s:console/networking:console/network: Apr 26 14:08:24 All recipes use and the table in the manual [1] documents the correct value as »console/network«. Apr 26 14:08:24 [1] http://docs.openembedded.org/usermanual/usermanual.html#section_variable Apr 26 14:08:24 Signed-off-by: Paul Menzel Apr 26 14:16:42 hi CosmicPenguin Apr 26 14:20:17 03Tom Rini  07master * re557584a06 10openembedded.git/recipes/bluez/ (bluez4.inc bluez4_4.42.bb): Apr 26 14:20:17 bluez4: Move the update-rc.d bits to 4.42 only Apr 26 14:20:17 Noted by Martin Jansa, taken from what he did in oe-core/meta-oe. Apr 26 14:20:17 Signed-off-by: Tom Rini Apr 26 14:56:32 In local.conf am i supposed to specify the full pathname of the directory where my sources will be stored? Apr 26 14:56:43 as opposed to what? Apr 26 14:58:14 say DL_DIR = "sources" Apr 26 14:58:46 yeah, that's not going to work. the current directory changes at points during the build Apr 26 14:59:02 so the relative path would resolve to different absolute paths at different points Apr 26 14:59:12 ${TOPDIR}/sources is likely what you want Apr 26 14:59:33 okay Apr 26 16:02:00 Hi everyone. Is there a way to tell bitbake to always look in overo-oe/sources for source, instead of going out to the respository sources to get it? Apr 26 16:02:25 ? Apr 26 16:03:29 One of my recipes calls for a repository which seems to be down. but I already have the source in overo-oe/sources Apr 26 16:03:40 I want to force bitbake to "fetch" from there instead Apr 26 16:04:24 just copy it over? Apr 26 16:04:35 run md5sum foo > foo.md5? Apr 26 16:04:47 it's already in overo-oe/sources Apr 26 16:05:04 but when when I bake, the recipe goes out and tries to fetch it again Apr 26 16:05:11 and errors out with 400 Apr 26 16:05:20 is the problem that md5sum is wrong? Apr 26 16:05:48 excuse me, Error 404: not found Apr 26 16:06:00 I dont have such a crystall ball Apr 26 16:06:06 trying to ftech ti_cgt_c600_6.1.16_setup_linux_x86.bin which I already have Apr 26 16:06:17 where is the problem to copy over from overo-oe/sources to sources? Apr 26 16:06:31 and run md5sum Apr 26 16:07:49 Oh, you mean copy into tmp/deploy/glibc/sources? Apr 26 16:08:11 what could be easier? Apr 26 16:08:31 I thought all sources go in overo-oe/sources Apr 26 16:09:18 no crystallball for your local.conf Apr 26 16:21:36 thank-you guys again. generating the md5sum worked to get my build going. I'm seriously not worthy Apr 26 16:33:17 woglinde, you were telling me last week that gstreamer-ti was broken with gstreamer-0.10.32 - do you have any detail on that? Apr 26 16:33:28 no Apr 26 16:33:33 I only heard it second hand Apr 26 16:33:39 where at? Apr 26 16:33:41 didnt test it myself Apr 26 16:33:45 beagle channel Apr 26 16:33:54 k, thx Apr 26 16:33:57 I only tested an mp3 Apr 26 16:34:03 some years ago Apr 26 17:59:02 anybody here with scons experience ? Apr 26 18:08:47 nvm, the scons file i try to use is f**** ;) Apr 26 19:04:41 mlip: what do u need w.r.t. scons Apr 26 19:05:44 khem disko frameworks uses scons Apr 26 19:05:53 hi btw. Apr 26 19:06:49 hello Apr 26 19:07:10 he ant Apr 26 19:07:12 should I insist cleaning out 2.4 bits? Apr 26 19:07:18 thats early Apr 26 19:07:28 now comes problematic part: classes and tasks Apr 26 19:07:42 let see what oe-core does Apr 26 19:09:54 khem, I wanted to try the git version of gpsd, since the 2.96 release is totally useless regarding python compilation Apr 26 19:10:55 khem, 3.0~dev version from git doesn't handle PATH though, so pkg-config doesnt work Apr 26 19:11:11 (and one has to set CC/etc. including the toolchain paths) Apr 26 19:12:48 03Jeff Lance  07org.openembedded.dev * r77b78a65a0 10openembedded.git/recipes/ti/am-benchmarks_1.2.bb: Apr 26 19:12:48 am-benchmarks: add return values Apr 26 19:12:48 * add return values to dhrystone and linpack for automated test Apr 26 19:12:48 Signed-off-by: Jeff Lance Apr 26 19:12:48 Acked-by: Chase Maupin Apr 26 19:12:48 Signed-off-by: Koen Kooi Apr 26 19:16:36 03Andrea Adami  07org.openembedded.dev * r6346a2c6b9 10openembedded.git/recipes/ (7 files in 2 dirs): Apr 26 19:16:36 nylon: move more files to /obsolete Apr 26 19:16:36 * distro has not been updated and is still on 2.4 kernel Apr 26 19:16:36 Signed-off-by: Andrea Adami Apr 26 19:16:39 03Andrea Adami  07org.openembedded.dev * r482fb6fb36 10openembedded.git/classes/ (3 files): Apr 26 19:16:39 nylon: move specific classes to /obsolete Apr 26 19:16:39 Signed-off-by: Andrea Adami Apr 26 19:16:41 03Andrea Adami  07org.openembedded.dev * r174e523aa5 10openembedded.git/classes/wrt-image.bbclass: Apr 26 19:16:41 wrt-image: move class to obsolete Apr 26 19:16:41 Signed-off-by: Andrea Adami Apr 26 19:21:45 khem: about 2.4 memories, in bitbake.conf we have PCMCIA_MANAGER ?= "pcmcia-cs" Apr 26 19:21:45 Apr 26 19:22:17 hm its pcmciautils now Apr 26 19:22:23 khem made some patch to oe-core Apr 26 19:22:53 yes, that's why I'm pestering him Apr 26 19:23:16 same for util-linux Apr 26 19:24:10 vs. old linux-hotplug Apr 26 19:24:28 and semi-old hotplug-ng Apr 26 19:25:09 ANGSTROM_BLACKLIST_pn-linux-hotplug = "hotplug functionality is provided by udev" Apr 26 19:25:42 ANGSTROM_BLACKLIST_pn-hotplug-ng = "hotplug functionality is provided by udev" Apr 26 19:26:36 we have a strange mix of situatiuons: 2.4 kernels / pre-udev, 2.6.old kernels not udev friendly, recent kernels Apr 26 19:27:06 do we still support all three? Apr 26 19:27:15 ant did you move all old simpad kernels? Apr 26 19:27:21 ah, no Apr 26 19:27:31 only 2.4 Apr 26 19:28:11 we really need a layer asap Apr 26 19:28:18 for handhelds Apr 26 19:28:53 hi cbrake Apr 26 19:28:56 hm? Apr 26 19:54:11 03Andrea Adami  07org.openembedded.dev * r761421ee27 10openembedded.git/recipes/linux/linux-kexecboot.inc: Apr 26 19:54:11 linux-kexecboot: bump PR to follow kexecboot upgrade and force rebuild Apr 26 19:54:11 Signed-off-by: Andrea Adami Apr 26 19:54:15 03Andrea Adami  07org.openembedded.dev * r72b87556b4 10openembedded.git/recipes/kexecboot/ (kexecboot-klibc_git.bb kexecboot_git.bb): Apr 26 19:54:15 kexecboot: move git versions to 51940a6ce14e8cfc8d1068858075ee19bacd8493 Apr 26 19:54:15 * bump PR Apr 26 19:54:15 Signed-off-by: Andrea Adami Apr 26 20:23:34 he effem Apr 26 20:23:52 hi woglinde, all Apr 26 20:24:05 hello Apr 26 20:37:09 kergoth: around ? Apr 26 21:12:43 anyone here using dsplink/dspbios? Apr 26 21:12:54 sometimes Apr 26 21:13:28 have you rebuilt it recently? Apr 26 21:13:55 dsplink for beagle yes Apr 26 21:14:07 i'm having no luck Apr 26 21:14:31 are you on the dev branch? Apr 26 21:14:42 yes Apr 26 21:14:53 hrm Apr 26 21:15:00 whats your problem? Apr 26 21:15:18 do_install is failing because the bios_ directory is empty Apr 26 21:15:35 uh? Apr 26 21:15:58 maybee the bios_ file wasnt downloaded? Apr 26 21:16:42 well, there's a setuplinux_.bin file that's unpacked into the work directory Apr 26 21:17:04 so something was downloaded... Apr 26 21:18:21 can you do a `bitbake ti-dspbios -c compile` Apr 26 21:19:14 and ls tmp/work/beagleboard-angstrom-linux-gnueabi/ti-dspbios_ Apr 26 21:21:22 started now Apr 26 21:21:30 thanks Apr 26 21:21:36 it's pretty quick Apr 26 21:22:21 its ti-dspbios_5.41.07.24.bb here Apr 26 21:22:31 me too Apr 26 21:22:42 okay Apr 26 21:22:51 not sure why though, there's newer recipes at least for dspbios Apr 26 21:23:08 other platforms Apr 26 21:23:21 * Nedlinpopo hasn't figured out how to tell which version will be chosen Apr 26 21:23:56 ti-dspbios-324434 Apr 26 21:24:12 would be one variant to build it Apr 26 21:24:21 the other is look at the angstroem conf files Apr 26 21:26:10 no problems here Apr 26 21:26:28 do_compile worked Apr 26 21:26:54 do you have stuff in the bios_ directory? Apr 26 21:27:01 mine is empty after do_compile Apr 26 21:27:10 hm Apr 26 21:27:35 hm whats your shell? Apr 26 21:27:39 bash Apr 26 21:27:42 on ubuntu Apr 26 21:27:56 where is /bin/sh poiting to? Apr 26 21:28:06 bash Apr 26 21:28:17 okay so its not a dash problem Apr 26 21:28:20 /bin/sh: symbolic link to `bash' Apr 26 21:28:21 yeah Apr 26 21:28:30 hm try -c clean Apr 26 21:28:39 and remove the bios.tar Apr 26 21:28:44 okay Apr 26 21:28:46 and try to build it again Apr 26 21:29:44 interesting, it's not a tar, it's bios_setuplinux_version.bin Apr 26 21:29:52 yes Apr 26 21:30:07 rebuilding Apr 26 21:31:12 try do_install Apr 26 21:32:24 :/ same error Apr 26 21:32:53 I wonder if do_compile failed silently Apr 26 21:33:39 hm maybee you have an empty staging ipk? Apr 26 21:34:37 i think do_install predates the ipk doesn't it? Apr 26 21:35:05 okay make a clean Apr 26 21:35:12 let us try unpack first Apr 26 21:35:17 okie dokie Apr 26 21:35:41 cleaned Apr 26 21:35:44 unpacking... Apr 26 21:35:52 unpacked Apr 26 21:35:52 after unpack I have bios_5_41_07_24/ bios_setuplinux_5_41_07_24.bin* temp/ Apr 26 21:36:18 me too Apr 26 21:36:34 okay Apr 26 21:36:38 no do_compile Apr 26 21:36:46 which shouldnt change somehting Apr 26 21:36:51 do you also have all the temp/run.*? Apr 26 21:37:04 khem: was at the store, what's up? Apr 26 21:37:38 kergoth: I am trying to play with adding|removing gettext deps thru gettext class Apr 26 21:37:51 and this BBCLASSEXTEND is giving me pains Apr 26 21:37:52 woglinde_: do_compile adds a bios_5_41_07_24 Apr 26 21:38:27 no Apr 26 21:38:37 so now do_install Apr 26 21:38:43 khem: yeah, the way bbclassextend and overrides interact is a bit annoying. also, DEPENDS_virtclass-native is always defined, and due to evaluation order, DEPENDS_ doesn't affect the virtclass-native version, you have to append to both Apr 26 21:38:55 ups install Apr 26 21:39:00 do_install fails because bios_5_41_07_24 is empty Apr 26 21:39:11 kergoth: why is immediate evalutions done in those classes Apr 26 21:39:15 hm ah Apr 26 21:39:19 is it empty for you after do_compile? Apr 26 21:39:23 so do_compile makes something worng Apr 26 21:39:29 ORIG_DEPENDS := "${DEPENDS}" Apr 26 21:39:33 DEPENDS_virtclass-nativesdk ?= "${ORIG_DEPENDS}" Apr 26 21:39:41 let us return to unpack Apr 26 21:39:50 if I change ORIG_DEPENDS = "${DEPENDS}" everything works as expected Apr 26 21:40:00 khem: eh Apr 26 21:40:03 you really, really don't want that Apr 26 21:40:15 DEPENDS_virtclass-nativesdk gets mangled, DEPENDS doesn't Apr 26 21:40:23 setting one to the other is going to screw you Apr 26 21:40:48 ok so it branches out the evaluation order in that bbclass Apr 26 21:40:56 woglinde_: my log for do_compile says there was nothing to compile, becuase there was no makefile Apr 26 21:41:01 does yours? Apr 26 21:41:03 right Apr 26 21:41:14 so I have to check if its inheriting one of those classes then I should update the corresponding DEPENDS override Apr 26 21:41:26 Nedlinpopo please -c clean and -c unpack again Apr 26 21:41:44 or just do what most of the classes (e.g. base and autotools) do, append/prepend to DEPENDS, virtclass-native, virtclass-nativesdk. but obviously this isn't ideal, and it doesn't scale well Apr 26 21:41:52 done Apr 26 21:42:01 kergoth: ah blind appending hmm ? Apr 26 21:42:02 we need a better general solution, RP has mentioned the issue on the list a couple times Apr 26 21:42:02 is it empty now? Apr 26 21:42:26 kergoth: the problem is I have to remove some deps as well Apr 26 21:42:32 so I have to make sure that var exist Apr 26 21:42:41 ugh, that's going to hit issues Apr 26 21:42:58 ensuring you're the last thing to operate is going to be difficult Apr 26 21:42:59 i only have: bios_setuplinux_5_41_07_24.bin* temp Apr 26 21:43:04 e.g. anonymous functions run in the order they're defined in Apr 26 21:43:05 ah Apr 26 21:43:12 so unpack goes wrong Apr 26 21:43:14 depends = bb.data.getVar('DEPENDS_virtclass-nativesdk', d, True) or "" Apr 26 21:43:15 :) Apr 26 21:43:21 bios_ is created later Apr 26 21:43:49 examine log.do_unpack Apr 26 21:43:57 khem: heh, cheating so it doesn't matter, since you aren't relying on overrides? Apr 26 21:44:00 whatever works :) Apr 26 21:44:08 NOTE: Unpacking sources/bios_setuplinux_5_41_07_24.bin to tmp/work/beagleboard-angstrom-linux-gnueabi/ti-dspbios-5_41_07_24-r1/ Apr 26 21:44:13 is the entire log Apr 26 21:44:19 I really think we need to teach bitbake to support plugins Apr 26 21:44:25 but it looks like it was just a copy Apr 26 21:44:27 kergoth: yeah actually I have to make sure that gettext deps are added/removed from depends in the end Apr 26 21:44:36 whatever the recipe type is Apr 26 21:44:38 thats not right Apr 26 21:44:45 there should be more in Apr 26 21:45:18 Nedlinpopo try to execute it from your bash Apr 26 21:45:25 the shell script? Apr 26 21:45:29 no the bin Apr 26 21:45:31 file Apr 26 21:45:31 er, the bin Apr 26 21:45:35 yes Apr 26 21:45:54 khem: long term, we should probably try to avoid some of this "magic" behaviors. we jump through a *lot* of hoops to make it so the recipe writer doesn't have to care about certain things, but at a certain point I'd rather we just operated simply and cleanly instead.. Apr 26 21:45:56 nothing Apr 26 21:46:01 ? Apr 26 21:46:02 hey this is a 32bit binary Apr 26 21:46:02 heh Apr 26 21:46:10 are you on a 32bit machine? Apr 26 21:46:13 yes Apr 26 21:46:20 hm Apr 26 21:46:24 I wonder.... Apr 26 21:46:29 perhaps you need linux-ia32 libs Apr 26 21:46:39 it claims to be statically linked Apr 26 21:46:58 i ownder if i'm missing the 32bit compatibility module in kernel Apr 26 21:46:58 kergoth: I agree we are shooting outselves in the foot Apr 26 21:47:57 woglinde_: also, how do i tell that the binary was supposed to be run? I can't find the bitbake file which does it Apr 26 21:48:02 Nedlinpopo okay good we find out Apr 26 21:48:11 khem: I'd really like to see more recipes inherit their classes at the top of the file, and just use +=/=+/.=/=. more appropriately. I think one of the things we try to work around is the case where someone just does = and loses deps, but if they do that its their own fault :) Apr 26 21:49:47 kergoth: I agree we are shooting outselves in the foot Apr 26 21:49:48 woglinde_: well I'm installing ia32-libs just in case it does some more setup Apr 26 21:49:53 hehe Apr 26 21:50:03 course, we'd probably need better recipe sanity checking Apr 26 21:50:05 but that's doable Apr 26 21:50:06 Nedlinpopo it will be executed and the answers automagiclly answered with y Apr 26 21:50:27 are statically linked ELFs stilly dynamically linked with libc? Apr 26 21:50:43 no Apr 26 21:50:50 hmm Apr 26 21:50:54 but you could strace the run Apr 26 21:50:56 ;) Apr 26 21:51:01 not a bad plan Apr 26 21:51:13 kergoth: yeah I think one problem I still have understanding recipes is the evaluation order Apr 26 21:51:25 I wrote a bitbake patch somewhere that automatically makes 'inherit' declarative, takes those and loads them before loading the recipe itself Apr 26 21:51:35 i was planning on doing some comparisons of metadata before/after to catch any potential issues Apr 26 21:51:43 one of these days.. Apr 26 21:52:35 that would help Apr 26 21:53:29 it's a start, anyway, we have a lot of work to do to make things less confusing Apr 26 21:53:30 wow, I have no explanation Apr 26 21:53:36 we're probably always going to need to care about when things happen Apr 26 21:53:41 but it'd be nice to reduce it, at least Apr 26 21:53:57 Nedlinpopo whats strace saying? Apr 26 21:53:59 after installing ia32-libs it runs Apr 26 21:54:05 hehe Apr 26 21:54:43 i think maybe file lied Apr 26 21:54:49 open("/lib/ld-linux.so.2", O_RDONLY) Apr 26 21:55:16 open("/lib32/libc.so.6", O_RDONLY) Apr 26 21:55:20 et cetera Apr 26 21:55:46 but: $ file bios_setuplinux_5_41_07_24.bin Apr 26 21:55:46 bios_setuplinux_5_41_07_24.bin: ELF 32-bit LSB executable, Intel 80386, version 1 (GNU/Linux), statically linked, stripped Apr 26 21:56:15 weird Apr 26 21:56:19 very Apr 26 21:56:56 woglinde_: it worked now! thanks Apr 26 21:57:03 no prob Apr 26 21:57:20 hmm, maybe i should try that inherit change with oe-core, the metadata is quite a lot smaller, maybe it wouldn't be quite so painful to wrap my head around the issues that way Apr 26 21:57:21 heh Apr 26 21:57:35 kergoth: yeah oe-core would be fine Apr 26 21:58:29 sorting out the recipes will be relatively simple, once I implement a sanity checker for it, its the classes that won't be fun Apr 26 21:58:30 heh Apr 26 21:59:02 wonder if there's any cases where we could remove anonymous python functions entirely Apr 26 21:59:04 hmm Apr 26 22:00:00 khem: the thing to remember with the virtclass classes is they're parsed at the end, by bitbake, so their anonfunc is going to run after just about everything else Apr 26 22:01:03 yeah figured Apr 26 22:04:26 i wish the bitbake parser supported just handing it a string and getting back ast nodes Apr 26 22:04:28 need to add that Apr 26 22:04:58 okay good nite Apr 26 22:05:01 night Apr 26 22:06:06 https://gist.github.com/914343 Apr 26 22:59:44 03Paul Menzel  07master * rb5a7a87e73 10openembedded.git/recipes/orc/orc_0.4.11.bb: Apr 26 22:59:44 orc: add version 0.4.11 Apr 26 22:59:44 This version is build tested using `angstrom-2010.x` and `minimal{,-uclibc}` for `MACHINE = "beagleboard"`. `schroedinger_1.0.9.bb` also builds fine with this version. Apr 26 22:59:44 Signed-off-by: Paul Menzel Apr 26 22:59:44 Acked-by: Peter Gsellmann Apr 26 23:15:37 khem, it finished... Apr 26 23:17:02 1340m34.714s Apr 26 23:17:41 22hrs 22minutes Apr 26 23:17:51 rather 20mins Apr 26 23:18:11 thats for 4 arches Apr 26 23:18:18 5? Apr 26 23:18:18 doing a console-image Apr 26 23:18:23 oh. Apr 26 23:18:24 yes 5 Apr 26 23:18:30 kk... Apr 26 23:18:42 console-image has ~3200 tasks Apr 26 23:18:47 so "daily" is doable. Apr 26 23:18:59 yes Apr 26 23:19:20 I am sure it will be a bit better if we did it w/o rm_work Apr 26 23:19:32 okay we need to add yocto's buildstat.bbclass to our task lists. Apr 26 23:19:42 right now it has used 25G Apr 26 23:19:57 so every 2~3 days we clean Apr 26 23:20:11 or we run the "reformat routine" Apr 26 23:20:22 let me try just one build w/o rm_work Apr 26 23:20:24 hey guys, quick question. how do packages generally depend on a specific version of other packages? Apr 26 23:20:31 and see how much disk it eats Apr 26 23:20:45 khem okay lets clean first. Apr 26 23:20:49 all the ones i've seen so far just depend on the existence, but the latest version is not being built oddly Apr 26 23:20:54 then go ahead and start from zero. Apr 26 23:21:05 ka6sox-farfarawa: yes I will rm -rf /scratch/ Apr 26 23:21:20 khem, kk...then that will be a legit test. Apr 26 23:21:45 Nedlinpopo, sorry...I just work on infra...no help here... Apr 26 23:21:59 what's infra? Apr 26 23:22:10 Infrastructrue. Apr 26 23:22:14 ah Apr 26 23:22:17 servers and stuffa. Apr 26 23:22:27 well that's important Apr 26 23:22:36 good job Apr 26 23:22:49 it seems to be fully functionaly Apr 26 23:23:05 mostly..we are wroking on fixin' what "ain't" Apr 26 23:24:02 Nedlinpopo: build time deps are not version trackable Apr 26 23:24:18 what do you mean? Apr 26 23:24:20 Nedlinpopo: but you can set a PREFERRED_VERSION of a given recipe Apr 26 23:24:47 if you have a package depending on say glib 2.26 buildtime Apr 26 23:24:53 oh, I'm trying to find out what place inteh dependancy chain specified the version that's so old the source is no longer online Apr 26 23:25:09 and our preferred version is 2.30 then there is no way to express build time dependency on 2.26 Apr 26 23:25:14 i wonder if forcing it with PREFERRED_VERSION will work? Apr 26 23:25:29 there are few things Apr 26 23:25:48 first if you specified PREFERRED_VERSION then it will choose that Apr 26 23:26:00 say you mentioned it in local.conf or some other conf file Apr 26 23:26:09 makes sense Apr 26 23:26:32 secondly it will look for DEFAULT_PREFERENCE level of recipes Apr 26 23:26:42 one with highest number wins Apr 26 23:27:00 if none of the above is true then it will simply chose the newer version amongst all Apr 26 23:27:11 that's across all recipes that provide the same thing? Apr 26 23:27:16 say nano_1.2.bb and nano_1.2.1.bb Apr 26 23:27:21 it will chose nano_1.2.1.bb Apr 26 23:28:12 but if in nano_1.2.bb you said DEFAULT_PREFERENCE = "1" and in nano_1.2.1.bb you said DEFAULT_PREFERENCE = "-1" then it picks nano_1.2.bb Apr 26 23:28:24 yes across the recipes contention Apr 26 23:30:06 okay, so if there are no DEFAULT_PREFERENCES in a collection of recipes, then it's reasonable to conclude that somewhere else along the chain a PREFERRED_PROVIDR was set up Apr 26 23:30:58 if its not picking up the latest bb then yes Apr 26 23:31:17 you can do bitbake -e -b > /tmp/xx Apr 26 23:31:24 and inspect /tmp/xx Apr 26 23:31:55 search for PREFERRED_VERSION Apr 26 23:32:23 khem, its already cleaned...wow Apr 26 23:32:26 hmm, that produces a lot of "ERROR:" lines should I be concerned? Apr 26 23:32:40 no Apr 26 23:33:08 oh, MultipleMatches is an unhandled exception Apr 26 23:33:21 ka6sox-farfarawa: so I think rm_work wont have much impact Apr 26 23:34:02 agreed... Apr 26 23:34:19 you didn't do rm_work that last time right? or you did? Apr 26 23:34:26 I did Apr 26 23:34:33 the one now running is without rm_work Apr 26 23:34:50 I want to check home much memory it will eat Apr 26 23:35:14 ah ha! Apr 26 23:35:17 okay...btw, I killed hudson which was what was eating the RAM. Apr 26 23:35:33 sure ok Apr 26 23:35:41 indeed somewhere a PREFERRED VERSION is defined Apr 26 23:36:01 which recipe ? Apr 26 23:36:42 can't tell yet Apr 26 23:39:14 khem, you are doing the same test without rm_work console images... Apr 26 23:39:26 ? Apr 26 23:40:22 hmm, i don't see the part where a recipe is included Apr 26 23:40:44 ka6sox-farfarawa: yes but only for 1 arch Apr 26 23:40:56 oh, okay Apr 26 23:41:00 is that a FILE= line? Apr 26 23:41:05 you have stats for 1 arch then? Apr 26 23:41:09 I want to check how much disk it eats up Apr 26 23:41:10 as in timing? Apr 26 23:41:13 yes Apr 26 23:41:16 kk Apr 26 23:41:32 these are all build from scratch Apr 26 23:41:54 okay so nothing staged before... Apr 26 23:41:55 got it. Apr 26 23:42:10 39.03% max for /Scratch Apr 26 23:42:23 during the last buildcycle. Apr 26 23:42:38 ok Apr 26 23:42:44 this time it should be lot more Apr 26 23:42:59 ya, but that was 5 arches before. Apr 26 23:43:40 okay laters...bailing out to drive home. Apr 26 23:44:13 khem: how do I tell when a new file is parsed? Apr 26 23:45:11 i saw oe-core from yocto and angstrom, however openembedded's git stays the same, are those related work(oe-core) on a different branch? Apr 26 23:46:06 oe-core is core as it says Apr 26 23:46:18 its used by yocto and angstrom Apr 26 23:46:47 what about the original openembedded git tree? Apr 26 23:46:48 oe-core is a subset of classic oe Apr 26 23:47:05 is it the classic oe? does it also adopt oe-core somewhere Apr 26 23:47:14 or, is it _the_ oe-core? Apr 26 23:47:23 yes next release will be based on oe-core Apr 26 23:47:33 oe will have layers Apr 26 23:47:51 oe-core meta-oe and distro layers and machine layers Apr 26 23:48:48 maybe it's pulled out to yocto/angstrom then feedback to the original openembedded project? Apr 26 23:49:09 i thought it's the other way around, i.e. openembedded creates a oe-core, others build based on it Apr 26 23:49:21 yes thats how it is Apr 26 23:49:30 oe-core is maintained by oe folks Apr 26 23:49:44 other projects leverage it Apr 26 23:50:12 early adopters are yocto and angstrom Apr 26 23:50:19 which way is it? angstrom/yocto send back their oe-core to oe? or oe creates it ffor angstrom/yocto? Apr 26 23:50:24 hmm Apr 26 23:50:24 we expect more and more projects and distros to use it Apr 26 23:50:37 if it's the latter case, then where is oe-cre in oe's git? Apr 26 23:50:37 yocto is not really parallel to Angstrom Apr 26 23:51:22 i have three checkout, saw meta-core in yocto/angstrom, but openembedded has no such changes at all as far as the new design is concerned Apr 26 23:51:34 which is why i'm asking, where is the original oe-core residing now Apr 26 23:51:41 oe-core is hosted on oe git Apr 26 23:51:59 git.openemebedded.org/openembedded-core Apr 26 23:52:01 you mean there is a oe-core git separate from openembedded.git? Apr 26 23:52:05 yes Apr 26 23:52:10 ok...that answers my question then, let me check out Apr 26 23:52:36 and...openembedded will be in parallel with angstrom/yocto, all three use oe-core as the core then? Apr 26 23:52:45 if you are new to it. You can try something like angstrom which has some trivial layer tooling Apr 26 23:52:53 to checkout multiple layers Apr 26 23:52:56 one of them is oe-core Apr 26 23:53:17 yes i use angstrom at work, just curious on how openembedded is progressing these days Apr 26 23:54:40 http://cgit.openembedded.org/ that answers it. Apr 26 23:54:47 xxiao: http://sakrah.dontexist.org/blog/2011/03/using-openembedded-core-to-build-angstrom-for-qemu/ Apr 26 23:55:02 that should help you to setup angstrom with oe-core Apr 26 23:55:05 guess i should follow non-classical openembedded.git from now on Apr 26 23:55:10 khem: thanks Apr 26 23:55:53 yes, we strive to make it better Apr 26 23:56:53 oe is very cool, hope one day it has an android-meta, that has davlik to run all those apps on top of oe :) Apr 26 23:57:41 well android is getting more close sourced then it started with Apr 26 23:57:56 there is a davlik for meego in the making... Apr 26 23:57:56 honeycomb is only available to partners Apr 26 23:58:12 yeah i hate that, our company is switching from android to possibly oe Apr 26 23:58:19 yes but having dalvik would be a good thing Apr 26 23:58:41 so I guess Apr 26 23:58:45 thats good Apr 26 23:59:02 oe remains one option which is pure linux based stack Apr 26 23:59:04 as far as apps are concerned, davlik is very interesting Apr 26 23:59:26 the rest of android rootfs is not open source friendly at all Apr 26 23:59:28 I would agree Apr 26 23:59:47 yeah meta-android would be something to look out Apr 27 00:00:31 intel needs to pour some fund there :) Apr 27 00:00:39 anyway thanks Apr 27 00:02:05 Alien Dalvik hmmm Apr 27 00:02:44 but its not open source Apr 27 00:03:13 anyway if they would like to port it over onto oe well and good Apr 27 00:05:31 * xxiao wishes he could do it and open source Apr 27 01:13:08 hi , i've 2 operating systems , 1 is linux other is windows , i want to run some code before windows opens from another hdd and then redirect to windows boot back , is it possible ? Apr 27 01:29:57 JDuke128: wrong channel. Apr 27 02:07:11 if i define PREFERRED_VERSION twice, which one takes effect? Apr 27 02:08:44 seems like the last one Apr 27 02:08:46 Depends on when each is evaluated, which is dependant on how the assignments are done. Apr 27 02:09:15 If it is A=1, then A=2, then yes the last one. Apr 27 02:09:35 it's just whichever was evaluated last right? Apr 27 02:09:57 If they are just = and =. If it is = and ?= then the first one. Apr 27 02:10:45 =? is set if unset. Apr 27 02:10:57 ah, cool **** ENDING LOGGING AT Wed Apr 27 02:59:58 2011