**** BEGIN LOGGING AT Wed Dec 09 02:59:57 2009 Dec 09 03:03:35 Man, you are the man! Let me try Dec 09 07:28:43 good morning Dec 09 07:37:01 good morning Dec 09 08:37:50 03Mike Rapoport  07org.openembedded.dev * r4c82c46829 10openembedded.git/recipes/linux/ (8 files in 2 dirs): Dec 09 08:37:50 linux-omap 2.6.32: add cm-t35 support Dec 09 08:37:50 Signed-off-by: Mike Rapoport Dec 09 08:37:51 03Mike Rapoport  07org.openembedded.dev * r2513dec674 10openembedded.git/conf/machine/cm-t35.conf: Dec 09 08:37:51 cm-t35: add machine configuration Dec 09 08:37:55 Signed-off-by: Mike Rapoport Dec 09 08:37:57 03Mike Rapoport  07org.openembedded.dev * r394a64335d 10openembedded.git/recipes/angstrom/ (angstrom-led-config.bb angstrom-led-config/cm-t35/leds): Dec 09 08:38:00 angstrom-led-config: Add LED setup for the cm-t35 and bump PR. Dec 09 08:38:02 Signed-off-by: Mike Rapoport Dec 09 08:38:04 03Mike Rapoport  07org.openembedded.dev * raf43bbdcad 10openembedded.git/recipes/u-boot/ (u-boot-git/cm-t35/cm-t35.patch u-boot_git.bb): Dec 09 08:38:07 u-boot git: add cm-t35 support Dec 09 08:38:09 Signed-off-by: Mike Rapoport Dec 09 08:38:11 03Mike Rapoport  07org.openembedded.dev * r31447b2f27 10openembedded.git/recipes/xorg-xserver/ (xserver-xorg-conf/cm-t35/xorg.conf xserver-xorg-conf_0.1.bb): Dec 09 08:38:14 xserver-xorg-conf: add cm-t35 support Dec 09 08:38:16 Signed-off-by: Mike Rapoport Dec 09 09:34:20 morning Dec 09 09:34:47 hi hrw Dec 09 09:36:56 good morning Dec 09 09:37:43 hi all Dec 09 09:42:28 morning all Dec 09 10:20:06 do we have mk68 coldfire in some machine.conf ? Dec 09 10:21:42 03Benjamin Schieder  07org.openembedded.dev * r696d681007 10openembedded.git/recipes/obexpush/obexpush_1.0.0.bb: Dec 09 10:21:42 obexpush: create /var/obexpush on installation Dec 09 10:21:42 * it is the default path in the ARGS, so it should better exist Dec 09 10:21:42 Signed-off-by: Sebastian Spaeth Dec 09 10:25:19 no Dec 09 10:26:57 how can i force bitbake to re-download a recipe's files? even deleting the files from oesources doesn't work - it just complains they don't exist Dec 09 10:27:42 mike_cw_ than the dont exist Dec 09 10:27:54 and you have to find another mirror for it Dec 09 10:28:23 woglinde: it isn't attempting the download as far as I can tell. it's on a local server Dec 09 10:31:05 i put the file in oesources manually using wget and it works Dec 09 10:32:18 mike_cw_: -cfetch -f Dec 09 10:32:30 hrw, tried that - didn't make any difference Dec 09 10:32:52 you removed file and file.md5? Dec 09 10:33:11 just the file actually Dec 09 10:33:39 so drop also file.md5 Dec 09 10:33:46 thats why bitbake complain Dec 09 10:34:04 ah ok Dec 09 10:34:46 thx. i need to fix the root of the problem really, which is that the buildserver doesn't add a version number to the archive it generates Dec 09 10:38:13 i have a recipe for building libxmlrpc-c - can I just submit the whole recipe as a patch for inclusion into OE? Dec 09 10:40:08 git format-patch Dec 09 10:40:11 hi cbrake Dec 09 12:06:40 NOTE: Multiple libraries (libexslt.so.0, libxslt.so.1) found and LEAD_SONAME not defined Dec 09 12:06:52 Heinervdm yo Dec 09 12:07:01 either split the package or define one Dec 09 12:07:21 LEAD_SONAME = "libxslt.so.1" in the recipe Dec 09 12:07:24 should be fine Dec 09 12:08:31 i just wanted to mention this, because some weeks ago someone said that those packages should ship only one .so Dec 09 12:08:44 i can't commit changes :) Dec 09 12:08:51 woglinde: patches for the gcc documentation are welcome :-) Dec 09 12:09:39 pb hm its anyway a strange option Dec 09 12:10:19 which option is this? Dec 09 12:12:37 -pthread Dec 09 12:12:44 which sets -lpthread Dec 09 12:14:32 in manual its only for ia64 Dec 09 12:14:48 but with specdump I can see it on x32 too Dec 09 12:16:39 ah, yeah, that is just historical cruft Dec 09 12:16:44 I don't think you want to be using that option. Dec 09 12:20:05 there is a damn .m4 macro which uses it Dec 09 12:23:05 ah, in which package? Dec 09 12:23:10 it should probably be patched to not do that Dec 09 12:23:18 sure Dec 09 12:23:35 I think found the solution Dec 09 12:23:51 the flags needs pthread as lib flag Dec 09 12:24:00 only pthreads was there Dec 09 12:32:57 Debian moved to gcc 4.4 now Dec 09 12:33:38 hrw jupp Dec 09 12:33:49 hmrs Dec 09 12:33:57 why -lphtread dont works Dec 09 12:38:35 hrms Dec 09 12:38:39 thats crazy Dec 09 12:41:38 pb hm the problem is much subtile Dec 09 12:41:54 when linke exectuable you should use -pthread Dec 09 12:42:04 when library -lpthread Dec 09 12:44:17 on what platform is that true? Dec 09 12:44:39 for most/all linux systems, -lpthread is correct for both executables and libraries. Dec 09 12:45:19 crap.. I lost my favorite multislot card reader Dec 09 12:45:29 pb seeems for 4.4 it isnt Dec 09 12:45:56 what the hell Dec 09 12:46:08 why it works now Dec 09 12:46:57 afaik, gcc only supports -pthread at all for compatibility with vendor compilers on platforms like solaris. Dec 09 12:47:19 pb I am puzzled Dec 09 12:48:16 hms maybee I had type Dec 09 12:48:19 args typo Dec 09 12:56:07 03Thomas Zimmermann  07org.openembedded.dev * r381196eba2 10openembedded.git/recipes/gtk-webcore/ (4 files in 3 dirs): Dec 09 12:56:07 midori_0.2.1: update shr specific parts of the recipe Dec 09 12:56:07 * Update default config Dec 09 12:56:07 * Add patch (from upstream) to retain the visibility of the navigationbar in fullscreen mode. Dec 09 12:56:07 It can be hidden manually, the way it used to be in 0.2.0 Dec 09 12:56:11 * This closes SHR bug #768 Dec 09 12:56:13 Signed-off-by: Sebastian Spaeth Dec 09 12:56:22 For libraries you can *get away* without linking to dependencies if the application links to them directly. Dec 09 12:56:35 But you always *should* link to them. Dec 09 12:58:34 03Sebastian Spaeth  07org.openembedded.dev * ra79466c88d 10openembedded.git/recipes/gtk-webcore/midori-0.2.1/a08caa02b95db6c9251a5342de0fb985fae8399b.patch: Dec 09 12:58:34 midori: delete extranous patch file from previous commit again. Dec 09 12:58:34 Signed-off-by: Sebastian Spaeth Dec 09 13:02:12 RP: ping about http://www.rpsys.net/openzaurus/patches/archive/pxa-linking-bug.patch Dec 09 13:02:27 ant_work: What about it? Dec 09 13:02:31 the patch needs minor edit for 2.6.32 Dec 09 13:02:44 but is *insanely* necessary to boot from nand... Dec 09 13:02:50 after 5 years Dec 09 13:03:03 ant_work: the bootloader on the device is broken Dec 09 13:03:13 so it's again Sharp? Dec 09 13:03:39 ant_work: yes, its Sharp's fault Dec 09 13:03:57 ok, second question is about tghe necessity of lzo_jffs2_sysfs patch Dec 09 13:04:14 ant_work: Its a nice to have feature, not essential Dec 09 13:04:24 2.6.32 has a lot of new options in .config Dec 09 13:04:29 I'm a bit lost... Dec 09 13:04:34 Never got that one upstream iirc :( Dec 09 13:04:46 should we stay favourlzo? Dec 09 13:05:05 ant_work: lzo does work better imo Dec 09 13:05:52 do you think we could use lzma to further compression of kernel in mtd1? Dec 09 13:07:02 ant_work: yes, probably Dec 09 13:11:02 RP: should I refer to the original pxa-linking patch and patch it in the recipe to adapt line #? Dec 09 13:12:33 hm ah Dec 09 13:39:23 ant_work: probably, or I can add a new version to the site Dec 09 13:39:53 I would be very thenakful ;) Dec 09 13:40:05 ant_work: you have an updated patch? Dec 09 13:40:07 oh..emotyon lead to mistyping Dec 09 13:40:16 well, not offhand Dec 09 13:40:23 just redid diff Dec 09 13:40:36 there is a bit more code on the top of head.S Dec 09 13:41:01 iirc patch still applies Dec 09 13:41:50 ant_work: if it still applies, whats the problem? :) Dec 09 13:42:06 the entry point are obviously broken :) Dec 09 13:42:22 but yes...no complaints atm ;) Dec 09 13:42:40 (it was 1.20 AM, though, better to recheck later on) Dec 09 13:43:39 btw linux-kexecboot overrides do_stage inherited from kernel. It's not using legacy staging :) Dec 09 13:43:58 it just overrides with an empty do_stage Dec 09 13:44:28 can this be optimized? Dec 09 13:44:58 ant_work: lzma compression of kernel can be big saver for zaurus Dec 09 13:45:28 I hope..I'm over the limits with 2.6.32 right now (+EXT4 + UBIFS) Dec 09 13:45:54 I falled back to using old/smaller 'kexec' binary Dec 09 13:46:19 I'll see how klcc cohoperates with -Os Dec 09 13:47:07 hrw: with kexec-tools we inherited a ton of unwanted/unnecessary code for our purpose Dec 09 13:47:44 ~hail sizecheck task then Dec 09 13:47:45 * ibot bows down to sizecheck task then and chants, "I'M NOT WORTHY!!" Dec 09 13:47:52 possibly next move will be linking the kexec code directly in kexecboot Dec 09 13:48:04 without external binary Dec 09 13:48:36 hrw: I'll try lzma first ;) Dec 09 13:49:43 http://www.flickr.com/photos/hrwandil/4171053791/ Dec 09 13:50:36 wuahah Dec 09 13:50:41 that was a hard fight Dec 09 13:50:44 but now its over Dec 09 13:51:56 ant_work: pxa-linking-bug-r1.patch Dec 09 13:52:07 A 5 year old patch :/ Dec 09 13:52:14 many many thx Dec 09 13:52:34 hm pushing it upstream? Dec 09 13:53:47 RP: just collie doesn't need it? Dec 09 13:55:37 RP: OT Dec 09 13:55:49 bitbake 1.10 parsing speed is ..wow..fast Dec 09 13:56:07 mostly when recalculating deps / runqueue Dec 09 13:56:53 big improvements over 1.8.18 Dec 09 14:00:03 hm Dec 09 14:00:07 I should update Dec 09 14:00:15 do Dec 09 14:00:31 hm git head? Dec 09 14:00:51 I checkout origin/1.10 Dec 09 14:01:48 03Sebastian Spaeth  07org.openembedded.dev * ra26054e8bd 10openembedded.git/conf/distro/include/sane-srcrevs.inc: Dec 09 14:01:48 sane-srcrevs: bump shr-settings Dec 09 14:01:48 Signed-off-by: Sebastian Spaeth Dec 09 14:02:10 woglinde: there is that same warning about 'set' module deprecated iirc Dec 09 14:04:53 yeah I know Dec 09 14:08:35 are we supposed to do apt-get install git or git-core? Dec 09 14:08:57 depends Dec 09 14:09:16 if you untar oe-meta stuff than not Dec 09 14:09:26 git-core Dec 09 14:09:33 git is another package Dec 09 14:09:36 same name Dec 09 14:09:40 I think Dec 09 14:10:41 would it matter for building? Dec 09 14:10:49 no Dec 09 14:10:54 ok cool Dec 09 14:11:00 oe builds git-native if its needed Dec 09 14:11:41 just trying to reduce things that can go wrong :P Dec 09 14:11:54 hopefully get atleast one build working Dec 09 14:12:20 hm? Dec 09 14:12:52 trying OE build again Dec 09 14:13:10 remember I had random problems with OE and was asking about distros and such? couple of days ago Dec 09 14:14:11 hm Dec 09 14:14:21 dont use fedora Dec 09 14:14:28 even when other claims it works Dec 09 14:14:28 using ubuntu 9.04 Dec 09 14:14:32 *duck and hide* Dec 09 14:14:40 hm Dec 09 14:14:46 I have some problems on it Dec 09 14:14:48 at work Dec 09 14:14:51 building glibc Dec 09 14:15:02 what's the safest lol Dec 09 14:15:21 cause my luck seems to be shitty enough that even the builds from koen dont work Dec 09 14:15:48 hm uh? Dec 09 14:16:03 actually maybe you can shed some light on that Dec 09 14:16:11 If prebuilt images from other people aren't working that suggsts something non-software - eg, dodgy RAM. Dec 09 14:16:28 well Dec 09 14:16:38 first I used the 2gb Sd card image version Dec 09 14:16:42 broonie *g* Dec 09 14:16:43 where you just dd it on there Dec 09 14:16:52 aditya which hardware again? Dec 09 14:17:04 hm I always tar Dec 09 14:17:14 when not using jffs2 Dec 09 14:17:15 I have bb revC Dec 09 14:17:21 ah Dec 09 14:17:29 I am unsing ext2 on my sdcard Dec 09 14:17:30 http://www.angstrom-distribution.org/demo/beagleboard/sd-images/2gb/... stuff from there Dec 09 14:17:31 thats okay Dec 09 14:17:44 it loads the modules fine Dec 09 14:17:50 but I cant install anything through opkg Dec 09 14:17:56 hm why? Dec 09 14:18:09 kept giving me errors about not able to get control file or something Dec 09 14:18:22 maybee the conf is wrong Dec 09 14:18:27 so its working Dec 09 14:18:33 so djlewis suggested I use one of the tar versions Dec 09 14:18:35 *sigh* Dec 09 14:18:40 *sigh* Dec 09 14:18:54 you should have tried to fix opkg conf Dec 09 14:18:54 when I tried those, even the modules wouldn't load Dec 09 14:19:08 well its not terribly hard to put the other one back on Dec 09 14:19:17 so something in opkg conf Dec 09 14:19:21 you should do Dec 09 14:19:22 * notes down * Dec 09 14:19:27 and tell the opkg error Dec 09 14:20:06 I'll have to do it in the evening though... need to get to work lol Dec 09 14:20:21 thanks for the help Dec 09 14:20:28 haha Dec 09 14:20:31 okay Dec 09 14:26:07 ping RP Dec 09 14:26:14 re effem Dec 09 14:26:17 ~ping RP Dec 09 14:26:18 pong RP Dec 09 14:28:04 morning Dec 09 14:28:31 jo kergoth Dec 09 14:29:31 hey woglinde, how's it going? Dec 09 14:29:55 hm Dec 09 14:30:00 fighting with autotools Dec 09 14:30:10 but becomming a guru more and more Dec 09 14:30:34 I'm not sure whether to congratulate you or say I'm sorry Dec 09 14:30:37 maybe both :) Dec 09 14:30:44 congrats Dec 09 14:30:50 I have no fear Dec 09 14:30:58 hehe Dec 09 14:30:58 with cmake I have to write more stuff Dec 09 14:31:11 need some food now Dec 09 14:31:33 autotools isn't nearly as bad as people make it out to be. it has its quirks and annoyances, but more often than not the problems people hit are due to the people who made the configure.ac, not the tools themselves Dec 09 14:33:43 hm I want to really get rid of cp in do_stage Dec 09 14:34:02 kergoth right Dec 09 14:34:16 but let people write there own makefiles Dec 09 14:34:27 is often much more hell Dec 09 14:34:47 with autotools I know in most cases I can crosscompile safely Dec 09 14:35:03 so food now Dec 09 14:35:10 till later Dec 09 14:43:49 http://marcin.juszkiewicz.com.pl/2009/12/09/n900-arrived/ Dec 09 14:49:56 re Dec 09 14:50:23 hi thebohemian Dec 09 14:51:58 thebohemian: did you saw my java handling patch on ML? Dec 09 14:52:19 hrw ah hm want to talk with robert about it Dec 09 14:52:27 but we are really busy at work Dec 09 14:52:46 hrw hm you dont often read planet.ltgo? Dec 09 14:52:50 ok, I am fine with delay Dec 09 14:53:11 otherwise you would have known about tinymail and pvanhoff Dec 09 14:53:16 on n900 Dec 09 14:53:22 woglinde: I do Dec 09 14:53:33 and I tend to ignore pvanhoff posts Dec 09 14:53:40 hm okay Dec 09 14:54:30 I did read them over year ago and then it was less and less interesting for me to I skip his posts Dec 09 14:56:59 woglinde: and recently his posts are rather about tracker then tinymail Dec 09 14:57:09 hrw sure Dec 09 14:57:12 hmm, nobody's populated a ix86-common site equivalent for x86_64 eh.. might have to add that to the todo Dec 09 15:13:34 03Jeremy Puhlman  07org.openembedded.dev * rfea5aa9ae8 10openembedded.git/site/x86_64-linux: Dec 09 15:13:34 site/x86_64-linux: Add in some test results for sudo, ORBit and IDL. Dec 09 15:13:34 Signed-off-by: Jeremy Puhlman Dec 09 15:13:35 kergoth: Did you ever get a reply from the gitosis author? Dec 09 15:14:12 huhu rp Dec 09 15:14:25 damnit, i forgot to fix jeremy's email addy in the author/committer.. fixed the signoff Dec 09 15:14:35 RP: never emailed them, thought you did :) Dec 09 15:14:38 rp can we get rid of the cp in teh base.bbclass for staging and jusr using the python stuff for copy files? Dec 09 15:14:51 kergoth: I did but I thought you'd also sent a patch or something Dec 09 15:14:51 hrm Dec 09 15:14:59 * RP 's memory is probably playing up :/ Dec 09 15:15:03 * RP never got a reply Dec 09 15:15:19 woglinde: probably Dec 09 15:15:44 there's a new thing called gitolite which seems to be much more actively developed these days. Dec 09 15:15:49 not 100% sure of the story why. Dec 09 15:16:18 RP: i have an issue with recipe compatibility. parse errors in oe vs mvl6 depending on which repository is used first in bbpath. thoughts? I can't think of an ideal solution short of versioning everything, other than what we're already heading toward — reducing recipe responsibility, making them more portable Dec 09 15:16:20 brolin: aha, need to look at that then - a similar idea to gitosis I presure? Dec 09 15:16:25 gitolite? hmmm Dec 09 15:16:39 ooh Dec 09 15:16:54 "Description: standalone, souped-up (started with per-branch permissions as the only extra feature, now lots more)" Dec 09 15:16:54 kergoth: Compatibility being the classes I presume? Dec 09 15:17:08 RP: yeah, and configuration metadata. infinite loops due to the layout_* changes, etc Dec 09 15:17:11 "per-branch permissions" sounds lovely Dec 09 15:17:13 it's unpleasant Dec 09 15:17:14 indeed Dec 09 15:17:24 gitolite may be exactly what we need Dec 09 15:18:46 * RP agrees Dec 09 15:21:59 "config syntax gets checked up front" Dec 09 15:22:04 finally :) Dec 09 15:24:08 03Chris Larson  07org.openembedded.dev * r163c8367bd 10openembedded.git/site/x86_64-linux: Dec 09 15:24:08 site/x86_64-linux: Add a few missing test results. Dec 09 15:24:08 Signed-off-by: Chris Larson Dec 09 15:30:17 03Brian Pomerantz  07org.openembedded.dev * rb56e04329d 10openembedded.git/site/ (arm-common mips-linux mipsel-linux powerpc-common): Dec 09 15:30:17 site/: Change the db_cv_mutex site defaults for db Dec 09 15:30:17 The rpm build is failing on the internal db3 build because it thinks Dec 09 15:30:17 db_cv_mutex=no is telling it that "no" is the mutext type. Setting Dec 09 15:30:18 this to proper values for the platforms that didn't have them set Dec 09 15:30:22 already should fix the rpm builds. Dec 09 15:30:24 Signed-off-by: Brian Pomerantz Dec 09 15:46:13 bitbake udev is fetching libsdl-native... what? Dec 09 15:46:17 * kergoth shakes head Dec 09 15:47:45 03Koen Kooi  07org.openembedded.dev * rbef71b66dc 10openembedded.git/recipes/linux/ (3 files in 3 dirs): Dec 09 15:47:45 linux-omap 2.6.32: bump SRCREV Dec 09 15:47:45 * Special thanks to Mike R. for updating the cm-t35 patches so quickly! Dec 09 15:53:14 I get the "can't open '/dev/misc/rtc': No such file or directory" error when running "hwclock". I googled and found that I'm not alone. In one of the threads there was a suggestion to create a symlink that points at /dev/rtc which also is missing Dec 09 16:06:37 03Martin Jansa  07org.openembedded.dev * rd79c373371 10openembedded.git/recipes/efl1/elementary_svn.bb: Dec 09 16:06:37 elementary: readd RDEPENDS on elementary-themes, accidentally removed in 51d8c69272483d1749587ba589f038e81718afc3 Dec 09 16:06:37 Signed-off-by: Martin Jansa Dec 09 16:32:59 morning Dec 09 16:35:42 morning XorA Dec 09 16:37:00 yo XorA Dec 09 16:38:45 bye all Dec 09 16:39:03 bye Dec 09 16:40:34 hms Dec 09 16:40:46 how can I prevent aclocal from running Dec 09 16:41:06 after my configure aclocal wnats to run Dec 09 16:41:26 pretty sure autotools.bbclass provided a variable which was a list of steps to skip, and we patched that support into autoreconf, or something.. check the class Dec 09 16:41:58 but the better solution would be to fix it to not break with aclocal, move their local macros (not pulled from other .m4 files) from aclocal.m4 into acinclude.m4 Dec 09 16:45:06 kergoth hm its not oe Dec 09 16:45:39 hm seems after patching Dec 09 16:45:43 its runs always Dec 09 16:46:19 03Martin Jansa  07org.openembedded.dev * r959fa7ec51 10openembedded.git/recipes/obexpush/obexpush_1.0.0.bb: (log message trimmed) Dec 09 16:46:19 obexpush: update INITSCRIPT_PARAMS to use only one digit param Dec 09 16:46:19 * update-rc.d_0.7 calls "printf %02d 09" inside and 09 is considered as Dec 09 16:46:19 octal number and it fails, then stop parameter is used as 00 and "sh: Dec 09 16:46:19 09: invalid number" is shown in postinst. Dec 09 16:46:23 * Maybe this should be fixed in update-rc.d as it suggests to use both Dec 09 16:46:25 digits "update-rc.d [-n] [-r ] [-s] defaults [NN | sNN Dec 09 16:46:49 remember autoconf can make those annoying rules to automatically re-run the autotools if timestamps changes on their files, to regenerate things.. we should think about trying to disable those for our stuff, since we force the re-execution anyway Dec 09 16:46:51 heh Dec 09 16:46:56 s/rules/makefile rules/ Dec 09 16:47:24 kergoth yes its not an oe problem Dec 09 16:47:27 as I said Dec 09 16:47:34 * kergoth nods Dec 09 16:47:52 hm will a touch all work Dec 09 16:47:56 trying Dec 09 16:52:51 ah okay Dec 09 16:52:58 after running configure Dec 09 16:53:05 find | xargs touch solves it Dec 09 16:54:45 03Martin Jansa  07org.openembedded.dev * rc5619454ae 10openembedded.git/recipes/ (poptop/poptop_1.3.4.bb rp-pppoe/rp-pppoe_3.8.bb): poptop, rp-pppoe: INITSCRIPT_PARAMS 08 is not octal number, lower values are considered octal too, but its OK for < 08 Dec 09 16:55:10 poptop? Dec 09 17:06:38 anyone looked at mips64 buidls with oe? seems there's no siteinfo/site files yet Dec 09 17:07:57 kergoth: once I had tried but then I did not push it Dec 09 17:07:57 kergoth: multilib needs some thought otherwise its ok Dec 09 17:07:57 kergoth we only have 32bit machines so far Dec 09 17:07:57 he khem Dec 09 17:08:00 woglinde: hello Dec 09 17:09:40 I did 22 years ago a working multilib build for x86_64 Dec 09 17:09:40 khem: multilib is even more complex since there's more abis than just 32/64 Dec 09 17:09:42 its a cold day here 1C or 33F Dec 09 17:09:46 kergoth: mips has 3 Dec 09 17:09:49 khem: its -10C here, your in tropical paradise Dec 09 17:09:49 or even 4 if you count O64 Dec 09 17:09:54 gmp has an abi list of n32, o32, 64 Dec 09 17:09:58 fun fun Dec 09 17:10:01 yeah those are commonly used Dec 09 17:10:29 n32 would be most commonly used on newer devices Dec 09 17:10:33 or n64 if full 64 bit power is to be used :) Dec 09 17:10:33 guh :) Dec 09 17:10:36 we really need to get automated site file generation going someday Dec 09 17:10:38 but O32 is has the widest support Dec 09 17:10:58 kergoth: I will try to fathom my memory and put something together for mips64/n64 Dec 09 17:10:58 no multilib for now Dec 09 17:11:06 my head hurts :) Dec 09 17:12:06 may be octeon or xlr would be a good target Dec 09 17:12:07 but I have none Dec 09 17:12:08 does qemu support mips64 well Dec 09 17:12:08 * kergoth still wants to leverage scratchbox w/ qemu for platforms that are supported to do site generation Dec 09 17:12:08 * kergoth shrugs Dec 09 17:13:09 I got mips32 going well on qemu which I have committed Dec 09 17:13:10 ah, nice Dec 09 17:13:10 let me experiment with mips64 for a change Dec 09 17:13:24 * kergoth thinks site/ is still quite a mess.. hard to keep straight which results belong where Dec 09 17:13:25 yeah Dec 09 17:13:43 site files should be restructured Dec 09 17:13:58 XorA: You are in Utah right ? Dec 09 17:14:29 I'd almost rather see a database of target information / test results Dec 09 17:14:59 hi esben Dec 09 17:53:39 khem: s/would/does/ Dec 09 17:54:22 does DEPEND=blah automatically become DEPEND=blah-native ? Dec 09 17:56:25 rp is BBCLASSEXTEND .18? Dec 09 17:56:31 woglinde: yes Dec 09 17:56:36 XorA: yes Dec 09 17:56:43 ah cool Dec 09 17:56:49 XorA: See the additions to native.bbclass Dec 09 17:56:50 RP: man your too awesome Dec 09 17:56:53 so I could convert some java recipes Dec 09 17:57:19 woglinde: yes Dec 09 17:57:24 coolie Dec 09 17:57:34 XorA: it also changes PROVIDES too Dec 09 17:57:43 RP: so if we add it to bitbake.conf would that mean for every recipe it will generate native packages too ? Dec 09 17:57:49 * XorA would have RPs kids if that were biologically possible Dec 09 17:57:59 khem: yes Dec 09 17:58:05 * khem would prefer clones Dec 09 17:58:09 * RP is scared Dec 09 17:58:14 XorA: ;-) Dec 09 17:58:20 * broonie watches the RP sekrit labs kick into action :) Dec 09 17:58:38 khem: can the world handle more than one RP? Dec 09 17:58:42 RP: but we do not need native packages for every recipe do we ? Dec 09 17:59:08 khem: no, but its just the principle that is nice Dec 09 17:59:15 aren't -natives still excluded from world? Dec 09 17:59:18 if so i don't see an issue there Dec 09 17:59:34 khem: there is a whole lot of hacks in OE that are there just because making native packages was a pain Dec 09 17:59:43 kergoth: yes, they are Dec 09 17:59:53 kergoth: assuming thats in native.bbclass Dec 09 17:59:58 which iirc it is Dec 09 18:00:03 * kergoth nods Dec 09 18:00:11 khem: so the theoretical possibility for any package to auto native is quite nice Dec 09 18:00:26 And it doesn't have a massive parsing overhead Dec 09 18:00:42 Converting a package to BBCLASSEXTEND reduces parsing time Dec 09 18:00:51 thats nice Dec 09 18:01:16 less recipes for me to keep track of wrt mvl6 vs upstream would make me less likely to go completely fucking insane Dec 09 18:02:02 XorA: right now I only have few native packages built but then if say a console-image includes 100 recipes wud it generate 100 native packages too or is there a way to stop that Dec 09 18:02:18 I dont wish to generate native packages for toolchain e.g. Dec 09 18:02:20 khem: console-image doesn't depend on all the native providers Dec 09 18:02:24 so I'm not sure why you think they would be built Dec 09 18:02:43 oh right someone has to pull in dep Dec 09 18:02:48 what exactly does the uriscr append_fail-fast mean? Is it suppose to add the patch if it notices an early failure? Dec 09 18:02:52 the runqueue only has tasks that need building, bbclassextend doesn't add them to that, just makes more providers available Dec 09 18:02:53 yeah Dec 09 18:02:58 khem: yes, it just makes the target available Dec 09 18:03:08 hm now reading python docu 11.10. shutil — High-level file operations Dec 09 18:03:38 one asks for foo it looks into foo.bb and also for foo-native Dec 09 18:03:41 got it Dec 09 18:04:23 RP: we should start converting the recipes Dec 09 18:04:26 :) Dec 09 18:04:31 khem: right :) Dec 09 18:04:36 and punting natives out Dec 09 18:04:38 khem: I already did in Poky ;-) Dec 09 18:04:44 hmm cool Dec 09 18:04:51 hm yes Dec 09 18:04:53 shutil.copytree Dec 09 18:04:57 lets try Dec 09 18:05:41 RP: are do_stage removal stuff also in ? Dec 09 18:06:05 khem: The core changes are in OE Dec 09 18:06:14 khem: Poky has been converted completely Dec 09 18:06:49 rp hm oh Dec 09 18:07:01 not that now some one doubles the work Dec 09 18:07:16 woglinde: ? Dec 09 18:07:57 RP: so lets start with BBCLASSEXTEND Dec 09 18:08:50 care still has to be taken to watch out for recipes that use update-alternatives Dec 09 18:09:20 rp for BBCLASSEXTEND in recipes Dec 09 18:10:18 woglinde: Well, the two things are independent in a way Dec 09 18:10:43 XorA: yes, that is true. We need a better way of handling that Dec 09 18:10:59 RP: I got bitten by coreutils-native Dec 09 18:11:28 RP: I checked a hack into OE as I wasnt sure how to do that with BBCLASSEXTEND Dec 09 18:11:41 hm whats packagedstaging_fastpath Dec 09 18:12:14 woglinde: A function populate_staging calls if packaged_staging is installed and we're not legacy Dec 09 18:12:48 XorA: coreutils should be in Poky, I wonder what I did Dec 09 18:13:33 XorA: Hmm, looks broken in poky :/ Dec 09 18:14:09 hms Dec 09 18:14:18 its a shell function again Dec 09 18:15:38 RP hm last question Dec 09 18:15:55 XorA: Probably a if BUILD_ARCH = HOST_ARCH test in do_install... Dec 09 18:16:11 whats the best way to get ${SYSROOT_DESTDIR}${TMPDIR}/ into python variable Dec 09 18:16:48 woglinde: bb.data.expand("${SYSROOT_DESTDIR}${TMPDIR}/",d, True) ? Dec 09 18:16:55 okay Dec 09 18:16:55 * RP < food Dec 09 18:17:06 wastn sure whats expand is doing Dec 09 18:17:11 back in ~ an hour Dec 09 18:17:14 okay Dec 09 18:42:38 khem, Im having same issue as http://bugs.openembedded.org/show_bug.cgi?id=5251, trying to find the patch you applied, you remember the name of it? Dec 09 18:52:48 03Maxime Petazzoni  07org.openembedded.dev * r9de20344a0 10openembedded.git/recipes/udev/ (8 files in 5 dirs): Dec 09 18:52:48 udev: exclude mtd* from persistent storage rules Dec 09 18:52:48 Exclude mtd* devices from the persistent storage rules to avoid invalid Dec 09 18:52:48 vol_id invocations on MTD partitions that may not be accessible. Dec 09 18:52:48 Signed-off-by: Maxime Petazzoni Dec 09 19:00:00 03Jeremy Puhlman  07org.openembedded.dev * r9bd199c537 10openembedded.git/recipes/u-boot/u-boot-mkimage-native_1.3.2.bb: Dec 09 19:00:00 u-boot-mkimage-native: Fix problem building on a system without strip. Dec 09 19:00:00 Signed-off-by: Jeremy Puhlman Dec 09 19:00:00 Signed-off-by: Chris Larson Dec 09 19:02:24 anyone recall where that wiki page was with the recipe style / variable order bits? Dec 09 19:02:31 s/\?$/was ?/ Dec 09 19:04:41 03Jeremy Puhlman  07org.openembedded.dev * r30dbf4deb5 10openembedded.git/recipes/tcltk/tcl_8.4.19.bb: Dec 09 19:04:41 tcl: Always include /usr/lib/tcl* scripts. Dec 09 19:04:41 Make sure that the tcl scripts are added in altabi builds. Dec 09 19:04:41 Signed-off-by: Jeremy Puhlman Dec 09 19:04:41 Signed-off-by: Chris larson Dec 09 19:08:54 http://wiki.openembedded.net/index.php/Styleguide was? Dec 09 19:09:33 ah, right, thanks Dec 09 19:09:36 * kergoth bookmarks Dec 09 19:12:13 sure gets annoying trying to apply fixes to every single version of a recipe we have Dec 09 19:13:06 kristoffer: are you using gcc 4.4.2 ? Dec 09 19:20:03 khem, yeah Dec 09 19:21:51 kristoffer: hmm I have not built for SH in long time Dec 09 19:22:01 but does your local.conf look like ? Dec 09 19:25:28 khem, aside from parallell make theres nothing weird about it, compiling for jornada6xx (machine) Dec 09 19:25:47 khem, this only happens on sh targets also. Done both mips/arm without any issues. Dec 09 19:27:10 it makes me sad that i have to wait for binary locale generation just to build a single recipe for the target. it doesn't need that to build & package itself.. Dec 09 19:30:05 kergoth: hmm, what point do we do the binary local generation? do_install? Dec 09 19:30:49 good question :) probably Dec 09 19:35:34 * khem should freshen up the patch to use cross-localedef in eglibc Dec 09 19:48:05 * * OE Bug 5345 has been RESOLVED (FIXED) by kristoffer.ericson(AT)gmail.com Dec 09 19:48:07 * * cdparanoia fails to run through configure on mipsel target Dec 09 19:48:09 * * http://bugs.openembedded.net/show_bug.cgi?id=5345 Dec 09 19:49:39 03Khem Raj  07org.openembedded.dev * r08c443b60d 10openembedded.git/recipes/moblin/librest_git.bb: Dec 09 19:49:39 librest: Update SRCREV. Dec 09 19:49:39 * Move to latest tip. Dec 09 19:49:39 Signed-off-by: Khem Raj Dec 09 20:02:59 re **** ENDING LOGGING AT Mon Dec 14 23:56:40 2009