**** BEGIN LOGGING AT Wed Oct 28 02:59:58 2009 Oct 28 06:46:06 anyone already had a stab on adding mysql5 ? Oct 28 08:07:10 Hi all, "eject" fails to compile as it tries to download the sources from geocities which doesn't exist anymore since 2 days... Oct 28 08:11:35 spaetz: any other source for it? Oct 28 08:15:09 don't don't know. I found a copy here: http://www.ibiblio.org/pub/Linux/utils/disk-management/eject* Oct 28 08:15:18 but not 100% sure if its the same source Oct 28 08:16:32 the ibiblio site has a somewhat older version than what dev.oe.org has Oct 28 08:17:50 freshmeat.net entry points to a website that doesn't exist anymore as well... Oct 28 08:19:53 good morning Oct 28 08:20:22 author of eject, a Jeff Tranter, points to ibiblio, but that only has 2.1.0 as the latest version. not 2.1.5 as dev.oe.org wants Oct 28 08:20:31 fuck Oct 28 08:20:34 sorry Oct 28 08:21:43 denix: about your email, I'm staying at http://www.aandbguesthouse.co.uk/ Oct 28 08:28:56 zecke: would it pose huge problems if I commited a change to revert "eject" to the latest available version on ibiblio? (on org.oe.dev that is) Oct 28 08:34:53 spaetz: yeah, it is a problem :) Oct 28 08:35:09 spaetz: could you not get the code from a Debian mirror? http://ftp.de.debian.org/debian/pool/main/e/eject/eject_2.1.5+deb1+cvs20081104.orig.tar.gz Oct 28 08:36:00 DJWillis: sorry, I don't know OE policies, can we just use a different upstream source which probably has distro specific stuff applied? Oct 28 08:36:09 I don't want to be the one making that change :) Oct 28 08:37:09 spaetz: no, Debian have the source 'pre mods' on there mirrors as do distros like Gentoo, nothing wrong with grabbing the 'pre mods' version from there. Should just be a chance to a SRC_URI then. Oct 28 08:37:25 morning Oct 28 08:37:44 morning hrw Oct 28 08:38:51 DJWillis: ok, will just change the SRC_URI then to a debian ftp site Oct 28 08:39:30 spaetz: well check it matches 1st but you get the idea, the code should be easy to find on a mirror somewhere. Oct 28 08:40:19 spaetz: as long as the checksums are fine changing the SRC_URI is an option Oct 28 08:47:47 zecke: good question, need to check new src checksum (it's a simple md5, right?) Oct 28 08:48:25 spaetz: checksums.ini, md5sum and sha some Oct 28 08:52:05 fun, the md5 s different between what we have recorded and the debian one. I give up. Oct 28 09:02:21 03Koen Kooi  07org.openembedded.dev * rb4c532efca 10openembedded.git/ (5 files in 3 dirs): Oct 28 09:02:21 gnome-python: fix 2.26.1 build, add 2.28.0 Oct 28 09:02:21 python-pygobject*: add 2.20.0 Oct 28 09:02:31 03Koen Kooi  07org.openembedded.dev * rde1d0546fd 10openembedded.git/conf/distro/include/angstrom-2008-preferred-versions.inc: angstrom: bump samba and python Oct 28 09:03:32 spaetzBBL: just google for eject-2.1.5.tar.gz, it finds the gentoo distfiles Oct 28 09:09:24 DJWillis: based on the uri that spaetz posted, I think the debian "orig" source in this case probably is already patched. Oct 28 09:09:35 if it was pristine then I wouldn't expect it to have things like "deb1" in the filename Oct 28 09:13:10 pb___: I realise now, it was the 1st google result and meant to be more of an illustration ;-) Oct 28 09:13:31 heh, fair enough Oct 28 09:16:13 "configure: error: C preprocessor "arm-angstrom-linux-gnueabi-gcc -E" fails sanity check Oct 28 09:16:13 " << I think this day is going to test my sanity ;-) Oct 28 09:17:54 03David-John Willis  07org.openembedded.dev * r25e876d130 10openembedded.git/recipes/xorg-driver/ (3 files in 2 dirs): Oct 28 09:17:54 xf86-input-evtouch: Use latest Ubuntu package diff and tweak configure. Oct 28 09:17:54 * Add configure support to generate calibration app and udev rules for touchscreens. Oct 28 09:17:55 03David-John Willis  07org.openembedded.dev * rb288fc18a0 10openembedded.git/recipes/pointercal/ (files/omap3-pandora/pointercal pointercal_0.0.bb): pointercal: Add calibration for the omap3-pandora touchscreen. Oct 28 09:17:58 03David-John Willis  07org.openembedded.dev * reb738d7c77 10openembedded.git/recipes/keymaps/ (files/omap3-pandora/keymap-extension-2.6.map keymaps_1.0.bb): keymaps: Add keymap extension for the omap3-pandora. Oct 28 09:18:02 03David-John Willis  07org.openembedded.dev * re65da86dc1 10openembedded.git/recipes/ (3 files in 2 dirs): u-boot/x-load: Add support for the omap3-pandora building from trees as git.openpandora.org. Oct 28 09:18:06 03David-John Willis  07org.openembedded.dev * r6d81ef0477 10openembedded.git/recipes/udev/ (udev_124.bb udev_141.bb): udev: Update recipes to package up local.rules. Oct 28 09:18:09 03David-John Willis  07org.openembedded.dev * r9c5d1cb40b 10openembedded.git/recipes/devicekit/ (devicekit-power_009.bb devicekit_003.bb): Oct 28 09:18:12 devicekit: Update to remove host dependency on Docbook XSLT files. Oct 28 09:18:18 * Bump PR Oct 28 09:18:22 * Update recipe to sed out the -nonet in the documentation XSLT conversion removing the reliance of the Docbook XSLT templates being installed on the host system (it will grab them from the net however). Oct 28 09:18:23 * Note: Another (maybe better) fix would be to package up docbook-xsl but that is overkill for these two recipes. Oct 28 09:18:25 03David-John Willis  07org.openembedded.dev * r4460e876e2 10openembedded.git/recipes/fbset/ (fbset-modes.bb fbset-modes/omap3-pandora/fb.modes): fbset-modes: Add fb.modes for the omap3-pandora LCD. Oct 28 09:18:29 03David-John Willis  07org.openembedded.dev * ra71f287b6f 10openembedded.git/recipes/tango/tango-icon-theme-enable.bb: tango-icon-theme-enable: Add recipe for enable the icon theme once installed (hackish). Oct 28 09:18:33 03David-John Willis  07org.openembedded.dev * r12eb9b8d48 10openembedded.git/recipes/x-load/x-load.inc: x-load.inc: Add support to build a version of x-load that can be loaded over OTG USB. Oct 28 09:18:40 03David-John Willis  07org.openembedded.dev * r7a2724f282 10openembedded.git/recipes/udev/ (3 files in 3 dirs): udev: Add local.rules for the omap3-pandora to open up /dev/event* to all users. Oct 28 09:18:43 03David-John Willis  07org.openembedded.dev * re286301652 10openembedded.git/recipes/libsdl/libsdl-gfx_2.0.11.bb: libsdl-gfx_2.0.11: Fix build by exposing LDFLAGS. Oct 28 09:18:46 03David-John Willis  07org.openembedded.dev * rc145b7f1d7 10openembedded.git/recipes/netbase/netbase/omap3-pandora/hosts: netbase: Add another host to the omap3-pandora hosts. Oct 28 09:18:49 03David-John Willis  07org.openembedded.dev * rbee83ea1d9 10openembedded.git/recipes/psplash/ (10 files in 2 dirs): psplash-omap3pandora: Add psplash splashscreen for the omap3-pandora. Oct 28 09:18:52 03David-John Willis  07org.openembedded.dev * r06b7f2e81a 10openembedded.git/recipes/base-files/ (3 files in 2 dirs): base-files: Add base files for the omap3-pandora. Oct 28 09:19:34 03David-John Willis  07org.openembedded.dev * r8d4ec184e9 10openembedded.git/recipes/aufs/aufs_cvs.bb: aufs_cvs: Fix LDFLAGS build error. Oct 28 09:19:34 03David-John Willis  07org.openembedded.dev * rb1b23b3efd 10openembedded.git/recipes/angstrom/ (2 files in 2 dirs): angstrom-led-config: Add LED setup for the omap3-pandora and bump PR. Oct 28 09:19:42 pigs flies... Oct 28 09:22:47 Whooo, urggg, I did not expect Koen to mirror that lot into oe.dev :-o Oct 28 09:38:05 good morning Oct 28 09:38:29 Morning florian Oct 28 09:39:23 hi florian Oct 28 09:39:33 florian: planet.ltg do not refresh again Oct 28 09:42:32 hrw: yes... some cronjob seems to mess up the permissions of the cache directory Oct 28 09:47:50 hrw: uh no, I take that back... its a new error and you seem to be guilty ;) Oct 28 09:49:51 o.. how? Oct 28 09:50:53 hrw: it actually gets updated but not your feed. "ERROR:planet.runner:Update of failed" Oct 28 09:51:10 ok, will check Oct 28 09:51:59 The feed is actually there... after a long traceback I see this at the end: "UnicodeDecodeError: 'utf8' codec can't decode bytes in position 32-33: invalid data" Oct 28 09:52:18 This looks like broken utf8... Oct 28 09:53:25 http://feedvalidator.org/check.cgi?url=http%3a%2f%2ffeeds.feedburner.com%2fHrwWebsite Oct 28 09:54:05 florian: can you tell me what is "in position 32-33"? Oct 28 09:54:15 Stupid piece of software ;) Oct 28 09:54:24 hrw: not yet :-} Oct 28 09:54:51 ok Oct 28 09:54:58 I go for coffee Oct 28 09:55:47 re Oct 28 09:58:52 hrw: hmm... un the header this would be a part of the encoding setting. In the data it would be somewhere in your last name in the feed title. Both do not look suspicious here. Oct 28 10:15:56 good morning Oct 28 10:16:50 TI PSP and OE on DM365 are going to open DaVinci !! Very nice! Oct 28 10:21:00 03Michael Smith  07stable/2009 * ra96afb1363 10openembedded.git/recipes/ (9 files in 2 dirs): (log message trimmed) Oct 28 10:21:00 glibc/eglibc: remove calls to obsolete base_set_filespath() Oct 28 10:21:00 Some recipes can use the default FILESPATHPKG, others need to override. Oct 28 10:21:00 Signed-off-by: Michael Smith Oct 28 10:21:00 Acked-by: Tom Rini Oct 28 10:21:02 Signed-off-by: Marcin Juszkiewicz Oct 28 10:21:06 Acked-by: Koen Kooi Oct 28 10:21:08 03Michael Smith  07stable/2009 * re68a11f9bb 10openembedded.git/recipes/glibc/glibc-initial.inc: (log message trimmed) Oct 28 10:21:11 glibc-initial.inc: remove FILESPATHPKG setting to fix glibc-initial_2.3.2 Oct 28 10:21:13 All v2.5+ recipes add glibc-2.4 as a FILESDIR or FILESPATHPKG if they Oct 28 10:21:15 need it, so it's not required in the inc anymore. Oct 28 10:21:17 Ref: Oct 28 10:21:19 http://projects.linuxtogo.org/pipermail/openembedded-devel/2009-June/011692.html Oct 28 10:21:21 Signed-off-by: Michael Smith Oct 28 10:23:55 recalcati__: oh... sounds interesting. link? Oct 28 10:26:09 http://wiki.davincidsp.com/index.php/DaVinci_GIT_Linux_Kernel_Releases Oct 28 10:26:36 time to prepare next set of patches for stable/2009 Oct 28 10:28:56 I'm using http://wiki.davincidsp.com/index.php/DaVinci_PSP_03.01_Linux_Installation_User_Guide . But now I have evmdm365 running from oe dev directly. So, TI PSP tells to use cs toolchain, instead oe use its one, and works. But I'm trying to enable analog video output, by now. Oct 28 10:31:13 very nice Oct 28 10:34:23 I've compiled the kernel, but , after compilation, the dir is deleted automatically from angstrom-dev/work/dm365-evm-angstrom-linux-gnueabi/linux-davinci-2.6.30+2.6.31-rc7-r10+gitr835d1ac43b1c0428cb0f7f91fbaf708ba8e7e504-r10/ . Any possibility to inspect kernel source after the compilation ? Oct 28 10:39:20 03Jeremy Lainé  07org.openembedded.dev * r73c37436b0 10openembedded.git/recipes/linux/ (linux-2.6.31/qemux86/defconfig linux_2.6.31.bb): linux-2.6.31: add defconfig for qemux86 Oct 28 10:39:31 03Jeremy Lainé  07org.openembedded.dev * r31f6e5ff32 10openembedded.git/: Merge branch 'org.openembedded.dev' of ssh://git@git.openembedded.net/openembedded into org.openembedded.dev Oct 28 10:41:15 I was wondering if kexecboot does support multiple kernels per partion. If yes, how to setup boot.cfg? Oct 28 10:43:54 hrw: Hi ! Oct 28 10:44:12 noglitch: hi Atmel man! Oct 28 10:44:42 hrw: on stable/2009 at91sam9263 we have a pink background... an issue with matchbox ? Oct 28 10:45:00 rather rgba order thing Oct 28 10:45:24 well, my X has good options: Oct 28 10:45:33 -rgba bgr Oct 28 10:45:36 noglitch: "bitbake xserver-common-1.30" and retest with it Oct 28 10:45:51 noglitch: drop that switch and test again? Oct 28 10:50:02 failed ! Oct 28 10:52:53 hrw: xserver-common/xserver-common-1.30/89xdgautostart.sh': No such file or directory Oct 28 10:54:59 hrw: Hi ! Oct 28 10:56:55 hrw: about noglitch issue only backgound and icons have wrong color Oct 28 11:03:09 Do we want to apply for a booth at Cebit? Oct 28 11:03:46 hrw: seems to be matchbox issue no ? Oct 28 11:04:20 anyone already tried to move mysql to v 5 ? Oct 28 11:07:11 florian: Cebit is no longer so interesting as it used to be Oct 28 11:07:38 mckoan: Yes that's true Oct 28 11:08:12 I guess OE is not the project that will raise interest there. Oct 28 11:08:34 EmbeddedWorld would be cool. Oct 28 11:09:09 florian: the only exhibition that is worth doing is EmbeddedWorld Oct 28 11:09:22 But the background for the question was that Cebit just published a call fro projects. Oct 28 11:09:38 florian: FLOSS projects? Oct 28 11:10:32 hi pvz Oct 28 11:10:47 I'm using pavel's kernel for spitz 2.6.32rc4. It seems soc modules are not working. Is that confirmed? Or do I miss something? Oct 28 11:10:49 Hi, I am using Pavels's kernel for spitz and soc modules fail to load. Is that confirmed or do I miss something ? thx Oct 28 11:10:49 test Oct 28 11:10:58 mckoan: yep... Oct 28 11:11:25 daal: wait for ant to be online Oct 28 11:11:37 pvz: so matchbox issue it is Oct 28 11:11:41 hrw: thx Oct 28 11:11:59 daal: he is zaurus maintainer now Oct 28 11:12:11 noglitch: hmm. strange a bi Oct 28 11:12:12 t Oct 28 11:12:26 hrw: good to know Oct 28 11:12:36 florian: link plz? Oct 28 11:12:58 does kexecboot support multiple kernels per partition? Oct 28 11:13:10 mckoan: let me see if I find something with less German text Oct 28 11:13:18 daal: no idea, never used Oct 28 11:13:55 florian: wunderbar danke! Oct 28 11:14:03 hrw: thx, will ask pavel Oct 28 11:14:49 noglitch: one thing for consider for atmel HW team: add mounting holes in next devboards Oct 28 11:15:03 noglitch: ngw100 have them, at91sam9263ek lacks them Oct 28 11:16:31 hrw: ok thanks I forward the information... Oct 28 11:17:17 noglitch: sam9g2 lacks too ;-) Oct 28 11:19:44 mckoan: aber sicher doch :-) http://www.cebit.de/opensource_e Oct 28 11:21:01 florian: definitely great! we can't loose this opportunity Oct 28 11:21:07 noglitch: thanks! :) Oct 28 11:21:56 when i insert my kernel module i get unknown symbol spi_register_board_info Oct 28 11:22:46 mckoan: on the one hand its a nice opportunity, on the other it is quite some effort and average Cebit vistors are not really interested in what we do. Oct 28 11:23:14 should i just edit spi.c and include the line EXPORT_SYMBOL(spi_register_board_info) then recompile the kernel or am i compiling my module wrong? Oct 28 11:23:17 MonkeySilent: insmod or modprobe? Oct 28 11:23:20 insmod Oct 28 11:23:30 MonkeySilent: then stop using insmod and start using modprobe? Oct 28 11:23:31 any chance you will include linux-fusion (DirectFB multi-app) as kernel module? and build DirectFB with fusion support? Oct 28 11:23:49 daal: patches welcome Oct 28 11:25:17 hrw: git linux-fusion from directfb should work out of the box TM :) Oct 28 11:26:24 that does not change situation Oct 28 11:27:30 hrw: well, I give it a try when I get pavel's kernel working and then ... Oct 28 11:28:45 submit a patch or something like that Oct 28 11:29:20 modprobe doesnt know where to look it for it yet, insmod should work as its just a very basic init module for spi Oct 28 11:33:07 can I have your opinion on DirectFB as a base for embedded applications on embedded systems? Oct 28 11:33:44 I am guessing i cant call the function from a external module as they dont want anyone calling it from kernel sapce? Oct 28 11:37:03 MonkeySilent: ok, I thought that spi_register_board_info was in other module Oct 28 11:43:27 do the architecture priority features of opkg actually work? Oct 28 11:44:08 yes Oct 28 11:44:23 at least for my arm targets Oct 28 11:44:45 so a machine architecture package should trump an armv5te one of the same name no matter what the version, or does it not work like that? Oct 28 11:45:03 arch.conf has the machine with a higher number than armv5te Oct 28 11:45:40 I'd not expect that, I'd expect the version number to have greater priority. Oct 28 11:46:49 version is more important Oct 28 11:46:55 see the problem is this. I have a special build of ffmpeg that includes extensions for the DSP on the SoC. The recipe gives it MACHINE_ARCH, but then once the target is running opkg always wants to use the version of FFMPEG from angstrom Oct 28 11:47:08 arch priority matter if you have same version for few archs Oct 28 11:47:21 mike_cw: "opkg hold ffmpeg" Oct 28 11:47:34 The Angstrom build has a colon in the front of the version number, which makes it always win Oct 28 11:48:00 mike_cw: fix your version then Oct 28 11:48:30 XorA|gone: I tried that, but bitbake chokes on the colon Oct 28 11:48:40 mike_cw: look for PE= Oct 28 11:52:27 XorA|gone: is that the variable that ends up before the colon in FFMPEG's version number? The only docs I can find for it say "fixme" :) Oct 28 11:57:43 anyone an idea what I did to deserve this when compiling a .c file: Oct 28 11:57:44 {standard input}:15243: Error: bad instruction `cmpl r6,r3' Oct 28 11:57:44 {standard input}:15244: Error: bad instruction `cmovl r3,r6' Oct 28 11:57:51 and that for many more lines Oct 28 11:58:13 inline assembly that isn't for your cpu Oct 28 11:58:35 will check Oct 28 11:59:53 no inline asm in the .c file Oct 28 12:00:22 btw this is for armv5te (sheevaplug) will try for beagleboard later Oct 28 12:01:22 mike_cw: yes thats the Package Epoch Oct 28 12:03:17 XorA|gone: and that has top priority presumably, being at the front? Oct 28 12:03:37 mike_cw: yes, its the same as dpkg Oct 28 12:03:49 Hi Oct 28 12:04:13 mike_cw: used when someone fucks up the PV and the only fix is to bump the epoch Oct 28 12:05:17 If I build a jffs2 image for my device, how can I add/delete folders, change permissions or modify a few files in etc? (_before_ the image is flashed on the device) Oct 28 12:05:36 XorA|gone: Thanks, I think this is going to be preferable to just holding FFMPEG. No other packages seem to be causing a problem at the moment thankfully. Oct 28 12:06:44 03Koen Kooi  07org.openembedded.dev * rebe5a33204 10openembedded.git/recipes/u-boot/u-boot-omap3pandora_git.bb: u-boot-omap3pandora: fix typo in recipe Oct 28 12:16:55 recalcati__, lots of hard work on TI over the past 18 months :) Oct 28 12:17:11 Crofton|work: you are at TI? Oct 28 12:17:16 no Oct 28 12:17:20 just a user Oct 28 12:17:39 if I was at TI I would not be cursing GPMC setup atm Oct 28 12:17:57 where is CE? Oct 28 12:18:01 er Cebit Oct 28 12:18:31 Hannover Oct 28 12:19:08 How far from Karlsruhe to Hannover? Oct 28 12:22:11 found it Oct 28 12:22:24 hannover is just 4.5h from me Oct 28 12:22:41 by train Oct 28 12:53:12 I followed http://elinux.org/BeagleBoardOpenEmbeddedDevelopment to make OE compile a recipe using my local sources. This recipe uses scons. When cleaning I want to run "scons -c". I added a do_clean_prepend function to my recipe, but it fails because CC, LDFLAGS and others are not set during clean. Is there a way to go around this to get a decent scons clean? Oct 28 12:54:36 I could make it work by just deleting the object directory, but that's not as nice :) Oct 28 13:04:40 Hi. I've got a flex (updated on sunday according to git log) build error at do_configure: onfigure: error: C preprocessor "arm-angstrom-linux-gnueabi-gcc -E" fails sanity check Oct 28 13:06:00 other things builds? Oct 28 13:06:33 Crofton|work: yes, I've seen. it's all very interesting. by now I'm trying to guess bootargs for fb component output. Oct 28 13:11:21 hrw: hum no ...looks like they all fail sanity check. Do I need to wipe erverything and build from scratch ? Oct 28 13:11:33 thats what I did Oct 28 13:13:06 hum ... ok, let's do it then. thx Oct 28 13:21:12 mickey|office: good morning Oct 28 13:35:18 good morning pb__ Oct 28 13:35:28 building an image with your patch attached Oct 28 13:35:33 flashing it after lunch Oct 28 13:37:40 mickey|office: how goes palmpre hacking? Oct 28 13:38:43 mickey|office: very good Oct 28 13:42:18 btw, I stopped fixing "eject-2.1.5". The gentoo tar.gz is much smaller than what is on a slightly older ibiblio version. And the 2.1.0 gives me errors during configuration... Oct 28 13:42:33 sucks to put sources on geocities.com.... Oct 28 13:43:01 and it sucks to have a more current version than what upstream advertises as the latest release... Oct 28 13:43:18 hrw: we're making good progress, phone call should be possible soon Oct 28 13:44:28 mickey|office: any plans to add n900 support? Oct 28 13:44:58 b96a6d4263122f1711db12701d79f738 eject-2.1.5.tar.gz Oct 28 13:45:03 spaetz: this file you need? Oct 28 13:45:16 hrw: yeah, why not. I'm not much excited about the n900, but if someone donates one to us or sends patches, I'm all ears Oct 28 13:45:25 yep, dev.oe.org uses it and the SRC_URI refers to a geocities.com address Oct 28 13:45:49 spaetz: well, did you check the md5sum of the gentoo sources? Oct 28 13:45:57 zecke: yep, different. Oct 28 13:46:01 spaetz: http://marcin.juszkiewicz.com.pl/download/oe/ Oct 28 13:46:39 hrw: so, should openembedded.org mirror that file, or should we put your URL into the SRC_URI? Oct 28 13:46:39 mickey|office: I am curious how they did things. Oct 28 13:46:51 mickey|office: and 300€ price is nice Oct 28 13:47:01 spaetz: my url is temporary Oct 28 13:47:08 yeah Oct 28 13:47:16 i do prefer the Palm Pre's formfactor though Oct 28 13:47:28 (even if the "keyboard" is like zaurus sl5500d) Oct 28 13:47:49 hrw: thanks, good to solve my immediate compile problems but we need to do something about that recipe and its obsolete SRC_URI as well... Oct 28 13:47:59 spaetz: sure Oct 28 13:48:14 pb__: do we have mirrors in oedem agenda? Oct 28 13:48:55 not that I know of. the agenda is on the wiki, feel free to add an item if you want. Oct 28 13:49:24 spaetz: you could copy eject-2.1.5 from the angstrom source mirror and host it someplace elese Oct 28 13:49:30 s/elese/else/ Oct 28 13:49:52 * hrw -> food Oct 28 13:50:08 ~bon appetit Oct 28 13:50:09 [bon appetit] smacznego. Guten Appetit. Eet Smakelijk. God Appetitt. Buon Appetito. Buen apetito Bom Apetite. buen apetito Smaklig måltid!. Hyvää ruokahalua. Bo Proveito Oct 28 13:54:40 pb__: yay, it works! Oct 28 13:54:41 khem: PING Oct 28 13:54:48 pb__: thanks for investigating Oct 28 14:00:32 mickey|office, khem is on the US left coast Oct 28 14:00:46 03Sebastian Spaeth  07shr/import * ra044bf682d 10openembedded.git/recipes/eject/eject_2.1.5.bb: eject: temporarily use angstrom as SRC_URI as geocities.com whent down. Not for upstreaming. Oct 28 14:01:12 Crofton|work: good for him :D Oct 28 14:01:29 i guess this indicates something wrt. timezones? ;) Oct 28 14:01:33 yeah Oct 28 14:01:40 ~lart timezones Oct 28 14:01:40 * ibot hooks into a hydrant and hoses timezones down Oct 28 14:01:41 0700 his time :) Oct 28 14:01:46 well Oct 28 14:01:52 he _could_ be awaken now ;) Oct 28 14:02:26 true, but a bit early for him I think Oct 28 14:02:41 * Crofton|work has a mental map of when people seem active ... Oct 28 14:04:48 03Koen Kooi  07org.openembedded.dev * r9c3d769ebf 10openembedded.git/ (conf/checksums.ini recipes/gtk-webcore/midori_0.2.0.bb): midori: add 0.2.0 Oct 28 14:04:49 03Koen Kooi  07org.openembedded.dev * rb34f0987f7 10openembedded.git/ (4 files in 3 dirs): webkit-gtk: bump to 1.1.16 Oct 28 14:04:50 03Koen Kooi  07org.openembedded.dev * r35e29f8924 10openembedded.git/conf/machine/omap3-pandora.conf: omap3-pandora: fix uboot provider Oct 28 14:14:31 mickey|office: good-o Oct 28 14:20:14 mickey|office: I guess someone should inspect the other glibc 2.6.1 patches and see whether there are any more that should be applied to newer versions. Oct 28 14:20:39 I'm not quite sure why they were suddenly dropped at 2.7, that seems a bit weird. Oct 28 14:21:45 pb__: sorry, I'm curious which patches got dropped? Oct 28 14:22:49 zecke: armv4t-interworking.patch is the one whose absence was vexing mickey Oct 28 14:23:15 ah thanks Oct 28 14:23:30 mickeyl has verified experimentally that interworking does not work on v4t without it :-} Oct 28 14:24:28 looking at eglibc_2.10.bb, it applies only one patch, whereas glibc_2.6.1.bb applies something like seventeen. Oct 28 14:24:53 * zecke wonders if the 40 hours travel time or the one flight with KLM is the crappy thing Oct 28 14:25:19 newer versions of glibc are also missing armv4t-interworking.patch; I haven't checked whether they are missing anything else as well Oct 28 14:25:29 zecke: heh, KLM are not that bad Oct 28 14:25:43 They were fine when I flew to Portland with them. Oct 28 14:26:22 pb__: really? They still have CRT monitors and no entertainment system at all Oct 28 14:26:28 pb__: did you every fly an asian airline? Oct 28 14:27:12 zecke: yeah, I have flown with cathay pacific several time, singapore airlines at least once or twice, and probably a couple of others. Oct 28 14:27:34 but, personally, I never use the entertainment system on planes anyway so its absence would not bother me very much Oct 28 14:27:56 pb__: ah, sorry then. I had one KLM flight to Taipei (actually roundtrip) and it was the last time... I got stuck in Hong Kong and Amsterdam for hours :) Oct 28 14:28:14 They had LCDs on the flights I was on. Oct 28 14:28:44 but I can't say I've ever had a burning desire to use an entertainment system on a flight. Oct 28 14:29:17 zecke: ah. yeah, I have had that problem as well; last time I went to taipei I got stuck in HK because CKS airport was closed due to a typhoon. Oct 28 14:29:31 didn't really seem fair to blame the airline for that though Oct 28 14:29:56 If you can blame the airline then angering them by choosing another carrier is asking for trouble :) Oct 28 14:30:43 pb__: oh, they just felt like putting me into another connection flight from amsterdam to berlin Oct 28 14:30:48 I don't think I have ever been becalmed at Schiphol although I did once get stuck in Zurich for 24 hours due to swissair Oct 28 14:31:03 pb__: before boarding the flight from hong kong to amsterdam Oct 28 14:39:07 zecke it greatly depends on what plane you have with KLM, whether they have crt or lcd Oct 28 14:39:21 they have both flavours Oct 28 14:40:49 eFfeM: right, I prefer companies where this is power outlets on every seat or once per row :) Oct 28 14:43:29 morning kergoth Oct 28 14:45:14 hey Oct 28 14:57:28 zecke: i prefer power too, klm has it in some planes in the front half (which ofc they only give to frequent fliers) Oct 28 15:14:14 eFfeM: :) Oct 28 15:20:53 03Jeremy Lainé  07org.openembedded.dev * rb3a385a228 10openembedded.git/recipes/linux/ (6 files in 2 dirs): linux-2.6.31: use standard GPIO drivers for boc01 Oct 28 15:40:56 03Frans Meulenbroeks  07org.openembedded.dev * rc5704e55e6 10openembedded.git/ (conf/checksums.ini recipes/mythtv/mythtv_0.22rc1.bb): Oct 28 15:40:56 mythtv: created recipe for 0.22rc1 (based upon svn version) Oct 28 15:40:56 DEFAULT_PREFERENCE=-1, still need to look at the patches from 0.21 Oct 28 15:40:56 builds for beagleboard, not for sheevaplug (armv5te) Oct 28 16:17:08 03Koen Kooi  07org.openembedded.dev * r8e785a41b3 10openembedded.git/recipes/mythtv/mythtv_0.22rc1.bb: mythtv 0.22rc1: fix PV Oct 28 16:45:40 bye Oct 28 16:51:13 pb__: hmm, python SIGSEGV w/ alignment traps in dmesg Oct 28 16:51:27 i'm afraid there's more than just this one patch necessary to bring eglibc up to par Oct 28 16:51:30 khem: ping? Oct 28 17:04:14 mickey|zzZZzz: doh Oct 28 17:31:28 03Koen Kooi  07org.openembedded.dev * rc8c654d5a6 10openembedded.git/ (6 files in 3 dirs): gtk+: add 2.18.3 Oct 28 17:31:35 03Koen Kooi  07org.openembedded.dev * rbea7f9fd85 10openembedded.git/conf/distro/include/angstrom-2008-preferred-versions.inc: angstrom: bump gtk+ Oct 28 18:31:02 mickeyl: can you capture a register dump for the crashing python? Oct 28 18:31:20 just the output from catchsegv might be enough, or a kernel trace with CONFIG_DEBUG_USER on, or from gdb. Oct 28 18:32:03 but, you might find it is easier just to go back to glibc for the time being. I don't think you will be getting much advantage from eglibc. Oct 28 18:34:21 * pb__ go home now Oct 28 18:34:23 bbiab Oct 28 18:42:48 Hi. I'm using Angstron (with 2.6.29-omap1) on a BeagleBoard and a TV (Plasma 52") and the graphical didn't start (previously on my 23" worked fine) Oct 28 18:43:29 I changed some kernel parameters, such as "console=ttyS2,115200n8 console=tty0 root=/dev/mmcblk0p2 rw rootfstype=ext3 rootwait video=omapfb:mode:720p60" Oct 28 18:43:54 maluta, check with #beagle Oct 28 18:44:09 but I still got "incompatible mode" (obs.: I can see the beagle logo from u-boot) Oct 28 18:44:14 Crofton|work: ok Oct 28 18:44:25 * Crofton|work thinks we need a contest for ratio of TV screen area to board area .... Oct 28 18:50:18 vai falar com vc todo dia Oct 28 19:04:38 re Oct 28 19:35:57 mickeyl: hello ? Oct 28 19:38:21 re Oct 28 19:39:46 florian: good evening Oct 28 19:43:13 * florian searches usb cables again... Oct 28 19:43:47 I can imagine that the kids eat cables. ;) Oct 28 19:44:02 pb_: armv4t-interworking.patch seems to be needed for eglibc I think we should also propose it upstream to glibc ports Oct 28 19:44:37 pb_: These routines were developed on armv5t so didnt catch the armv4t cases Oct 28 19:44:48 florian: very likely. my daughter certainly seems keen to eat anything that falls within her reach. Oct 28 19:45:12 khem: yes, it seems to be needed for both glibc and eglibc Oct 28 19:45:59 pb_: yeah... this sounds very familiar. Oct 28 19:46:31 khem: however, it seems that eglibc does not work on armv4t even with that patch, so maybe something else is needed. Oct 28 19:47:15 yeah I am sure there are some issues in lowlevel lock routines Oct 28 19:47:37 I will have a look. Oct 28 19:47:43 pb_: hmm... when did this eglibc on armv4t problem show up? Oct 28 19:47:47 What is the error Oct 28 19:48:17 florian: I think it has probably never worked and mickeyl was the first to try it. Oct 28 19:48:27 another problem could be that compiler is now defaulting to generate armv5t code for eabi unless told to do for other arms Oct 28 19:48:57 pb_: yeah I do not have armv4t hardware Oct 28 19:49:03 it can't be that, if your compiler was generating code for armv5 then nothing would ever work at all Oct 28 19:49:51 but if we pass with-arch correctly during configuring cross gcc then it should not be a problem Oct 28 19:49:56 the binaries that mickey showed me were mostly armv4t; the only obvious example of v5 code was memmove() though I didn't check it all that carefully Oct 28 19:50:42 and, in any case, v4t did use to work fine with glibc, so this is either an eglibc specific failure or a gcc regression Oct 28 19:51:21 pb_: I'm pretty sure I had it working some time ago on the micro2440. Oct 28 19:51:36 or binutils, I guess, though failures there tend to be more of an all-or-nothing thing Oct 28 19:52:09 florian: yeah, it definitely did work at one point. I had the whole fso stack running as thumb on my gta01 some months ago. Oct 28 19:52:28 actually, probably about a year ago since I think that was when I first wrote the armv4t-interworking patch. Oct 28 19:52:28 thumb should be better with 2.20 now Oct 28 19:53:10 so, clearly it has regressed somewhat in the last 12 months, it's just a case of figuring out why. Oct 28 19:53:12 pb_: oh yes right, I might had turned off thumb binaries for this... Oct 28 19:53:20 florian: ah, right Oct 28 19:53:48 pb_: so its thumb userspace ? Oct 28 19:53:55 khem: yes Oct 28 19:53:59 on armv4t ? Oct 28 19:54:02 yes Oct 28 19:54:19 I see and arm userspace works ok ? Oct 28 19:54:27 I think so. Oct 28 19:54:38 I remember a lot of failing binaries with thumb turned on. Oct 28 19:55:43 I can try thumb userspace on armv5t just in case if its general interworking failure I will catch that on osk too Oct 28 19:56:25 btw. eglibc can be compiled pure thumb2 code ( for cortex guys it might be interesting) Oct 28 20:03:08 heya Oct 28 20:03:32 how do i do a recipe to get the do_install to just copy the binary to /usr/bin? Oct 28 20:03:51 do_install() { install -m 0644 binary ${D}${bindir}/ } Oct 28 20:03:54 can i just say cp binaryname /usr/bin since it is chrooted? Oct 28 20:04:08 no; it isn't chrooted Oct 28 20:04:12 oh it isnt? Oct 28 20:04:13 k Oct 28 20:04:19 so what is $D then Oct 28 20:04:30 the "root" directory for that purpose Oct 28 20:04:38 oh ok Oct 28 20:04:49 and bindir is defined as /usr/bin somewhere? Oct 28 20:04:55 right, in bitbake.conf Oct 28 20:05:07 pb_: thanks a bunch dude have a great day Oct 28 20:05:19 np Oct 28 20:05:43 03Koen Kooi  07org.openembedded.dev * r7767c7c6d2 10openembedded.git/ (11 files in 5 dirs): Oct 28 20:05:43 gnumeric: add 1.9.14 Oct 28 20:05:43 libgsf: add 1.14.16 Oct 28 20:05:43 goffice: add 0.7.14 Oct 28 20:05:53 03Steve Sakoman  07org.openembedded.dev * r2715d18f86 10openembedded.git/ (conf/checksums.ini recipes/gnome/gedit-plugins_2.28.0.bb): gedit-plugins: add 2.28.0 Oct 28 20:15:53 03Khem Raj  07org.openembedded.dev * rbd449d860f 10openembedded.git/recipes/eglibc/ (4 files in 2 dirs): Oct 28 20:15:53 eglibc: Forward port armv4t-interworking patch Oct 28 20:15:53 * This patch is needed to support armv4t targets. Oct 28 20:15:53 Signed-off-by: Khem Raj Oct 28 20:15:53 what do you add to bb to keep it from cleaning up the working directory if you already have inherit autotools? Oct 28 20:17:53 03Frans Meulenbroeks  07org.openembedded.dev * r5a66b3c314 10openembedded.git/recipes/mythtv/mythtv_0.22rc1.bb: mythtv: add S, REALPV, PR Oct 28 20:29:17 my conference is fine :) Oct 28 21:02:06 RP, runqueue.py: I'm curious why you spell out "> 1" ? faeb86a6 (Richard Purdie 2007-08-25 09:20:32 +0000 663) if len(prov_list[prov]) > 1 and prov not Oct 28 21:03:22 blindvt: how else would you write it? Oct 28 21:03:37 RP, if len() and whatnot Oct 28 21:03:53 surely that's equivalent to "len() > 0", not > 1 Oct 28 21:03:55 there's no need to be that explicit to check if a list is empty in python, better to use __nonzero__ or whatever. just if prov_list[prov] ... Oct 28 21:04:03 oh, > 1, right, i'm blind.. Oct 28 21:04:05 :) Oct 28 21:04:33 http://i.imgur.com/EMKdt.jpg is amusing, on another note Oct 28 21:05:39 hehe Oct 28 21:05:52 pb_, i mean "if len(self.runq_depends[task]) == 0:" sounds odd to my untrained ears.. "if not len(self.runq_depends[task]):" ok Oct 28 21:05:57 * blindvt ducks Oct 28 21:06:15 pb_: I wonder if mickeyl is using thumb for first time Oct 28 21:07:00 blindvt: yes, the latter is better than the former in that one Oct 28 21:07:52 blindvt: yes, in that case there are various other ways of writing the same expression and most of them would probably be better :-} Oct 28 21:08:41 I'm not sure that dealing with stylistic issues in the bitbake code is the most productive use of time though. there are probably more important issues to address in that code. :-} Oct 28 21:09:00 how can i set a default PREFERRED_PROVIDER that can be overridden by a distro? E.g. conf/distro/micro.conf: PREFERRED_PROVIDER_virtual/coreutils-native = "busybox-native" would override a default of coreutils/native that i'd like to set somewhere Oct 28 21:09:21 * blindvt nods pb_ Oct 28 21:09:39 blindvt: if you want the local.conf to be able to override the distro, the distro should use ?=, generally Oct 28 21:09:56 blindvt: you could set it with "?=" somewhere else, though I'm not quite sure where you had in mind to be the default. Oct 28 21:10:21 wont _local override work ? Oct 28 21:10:41 for what? you wouldn't want the DISTRO defining a _local override, that would be backwards. Oct 28 21:10:52 no in local.conf Oct 28 21:11:13 kergoth, yes, that would work, but i want the distro to override it, eventually. Dunno where i could stick the default of PREFERRED_PROVIDER_virtual/coreutils-native = "coreutils-native". Perhaps into bitbake.conf? I don't think so Oct 28 21:11:59 it sounds like you need to rethink your approach Oct 28 21:12:08 out of interest, why are you tryng to make micro use busybox-native instead of coreutils? Oct 28 21:12:09 the use-case is that i want e.g. distro/micro.conf to use busybox but not foobar anybody else Oct 28 21:12:41 I'm not entirely convinced that this is a good idea, busybox is not nearly as well tested on the host systems as coreutils is. Oct 28 21:12:56 i don't really see the gain Oct 28 21:13:02 some tiny amount of disk space saved for a buidl? Oct 28 21:13:04 pb_, compat check and i just don't want to download coreutils Oct 28 21:13:10 blindvt: is busybox having merits over coreutils on build box Oct 28 21:13:32 if your disk space or bandwidth is that precious, you have other problems than just coreutils Oct 28 21:13:37 some time saved in building coreutils and space prolly yes Oct 28 21:14:30 khem, i need a testbed and simply do not want coreutils, not even for building stuff (i need a testbed ;) Oct 28 21:14:35 well, anyway, leaving aside for a moment the question of whether this is actually a good idea or not, oe doesn't really have any way to specify a "default preferred provider" in the sense that you want. the general expectation is that each DISTRO will call out explicitly the providers that it wants, if it cares. Oct 28 21:15:04 and there's always DEFAULT_PREFERENCE Oct 28 21:15:05 you could set DEFAULT_PREFERENCE = -1 in busybox-native, which would have roughly the same effect: DISTROs that don't specify a preference would not get it. Oct 28 21:15:41 pb_, k, i'll try that, thanks! Oct 28 21:16:42 how do you bitbake clean on a bb file you created yourself? Oct 28 21:16:50 ? Oct 28 21:16:51 I did a bitbake -b somefile.bb Oct 28 21:16:56 or, as a last resort, given that this dependency is injected by base.bbclass, you could attack it there. Oct 28 21:16:56 the same way you bitbake clean anything else. Oct 28 21:17:03 i can not do a bitbake -c clean file.bb Oct 28 21:17:03 bitbake -c clean -b foo.bb or add foo.bb to bbfiles and bitbake -c clean foo Oct 28 21:17:07 of course not Oct 28 21:17:14 did you even bother to read bitbake --help? Oct 28 21:17:27 set COREUTILS_NATIVE ?= "coreutils-native" in bitbake.conf, override it in your distro.conf, and make base.bbclass use ${COREUTILS_NATIVE} instead of coreutils-native. Oct 28 21:17:39 -b doesn't change how the following arguments are interpreted, it accepts one following argument which is a filename Oct 28 21:17:40 yuck, but all these answers are yuck to some extent Oct 28 21:17:45 kergoth: that worked thanks Oct 28 21:19:12 me estrepei Oct 28 21:19:22 yo: np Oct 28 21:20:54 in a .bb file, for the do_install, I have install -m 0644 ${S}/binary ${D}${bindir}/ but it is trying to put it into the workingdir... Oct 28 21:20:57 grrr Oct 28 21:22:02 oh woops Oct 28 21:22:09 i think i forgot an install -d Oct 28 21:27:37 ERROR: QA Issue: Architecture did not match (40 to 3) on /work/armv5te-angstrom-linux-gnueabi/aes-1.0.7-r0/install/aes/usr/bin/aes ??? Oct 28 21:28:36 sounds like its build didn't obey the bitbake variables (CC, etc) Oct 28 21:29:20 kergoth: but it got past the compile stage Oct 28 21:29:26 and? Oct 28 21:31:38 kergoth: shouldnt it be install -m 0755? Oct 28 21:31:46 for the binary? Oct 28 21:31:52 if you want to execute it, yes Oct 28 21:32:03 kergoth: then possibly thats the problem... :) Oct 28 21:32:07 i doubt that will magically fix the fact that the binary is compiled for the wrong architeture Oct 28 21:32:10 but hey, have fun Oct 28 21:39:57 kergoth: ok so how do i fix the arch then I have the Makefile it lists CC, CFLAGS, LDFLAGS, LDLIBS, AR, RANLIB Oct 28 21:40:49 i don't have time to teach you how make works and the ways by which you pass variables into a build from outside. if it's autoconf+automake, its mostly automatic, but custom buildsystems need individual attention Oct 28 21:41:17 im doing inherit autotools Oct 28 21:42:01 that tells me nothing. i asked about the buildsystem of the project you're building, not your recipe Oct 28 21:43:08 run 'file' on the binary out in the workdir of the recipe, see what it is. Oct 28 21:43:11 and look at the do_compile log Oct 28 21:45:53 kergoth: holy cow man I just realized the problem I was logged in as a different user and environment was all f&*( up Oct 28 21:46:20 ahhh oops :) permissions will hose you there, can't share tmp between users at the moment, not safely Oct 28 22:00:47 kergoth: what would you recommend for aes 256 bit encryption in a mounted loop? Oct 28 22:01:32 I wouldn't be the best person to ask, very limited experience with encryption in general, including disk/volume level Oct 28 22:03:33 I'm trying to build the PSP environment for my TI DaVinci processor with: MACHINE=dm355-evm bitbake board-set arago-demo-image but it fails with compiler errors: http://pastebin.org/49101 What do I do? Oct 28 22:06:34 Rovanion, try #arago Oct 28 22:15:31 "The external toolchain could not be found in /home/rov/CodeSourcery" Oct 28 22:15:34 sounds fairly explicit Oct 28 22:15:45 I guess you have to debug the external toolchain installation. Oct 28 22:17:17 has anyone gotten truecrypt working on any OE device? Oct 28 22:18:10 pb_, that to :) Oct 28 22:18:38 hmm, can anyone think of any types for bitbake variable type checking other than list and boolean? Oct 28 22:18:43 pb_: Ouch, had tons of problems with that yesterday. Guess I hav eto redo it. Oct 28 22:19:05 How can I prevent udev in openembedded (very minimal own distro) from autoloading modules? Oct 28 22:20:17 kergoth: well, string, obviously, but I guess you could consider that as a degenerate list. python-code, shell-code? Oct 28 22:20:37 hmm, perhaps Oct 28 22:21:03 depends a bit on what problem you're trying to solve Oct 28 22:21:04 03Frans Meulenbroeks  07org.openembedded.dev * re7c38fbecd 10openembedded.git/recipes/mythtv/mythtv_0.22rc1.bb: mythtv: made it compile for armv5te too Oct 28 22:21:35 i have this checker in the metadata, but mulling over how to integrate into bitbake, cause it's probably best done there, that way += could obey the separator for the list in question.. Oct 28 22:21:35 heh Oct 28 22:42:20 hmm Oct 28 22:55:01 when OE runs configure, how does it know where the system root is Oct 28 22:55:40 what do you mean? Oct 28 22:55:55 the configure it runs uses /usr/bin etc as paths Oct 28 22:56:08 so something has to point it at staging ... Oct 28 22:57:49 yes, cflags, ldflags, etc. Oct 28 22:57:57 configure picks them up from the environment Oct 28 22:58:07 yeah Oct 28 23:38:25 Is it normal, that the first mount of a fresh jffs2 image on my arm920t takes several minutes? the image is just 4 MB big and the partition is 32MB big Oct 28 23:38:37 anything I can do about that? Oct 28 23:40:03 reading from the nor flash with dd is quite fast (1-2MByte/s) Oct 28 23:40:17 well ... fast enough ;-) **** ENDING LOGGING AT Thu Oct 29 02:59:57 2009