**** BEGIN LOGGING AT Mon Sep 27 02:59:57 2010 Sep 27 03:24:22 kergoth_: bitbake world fetched stuff well with the SCM dir removal change Sep 27 03:24:58 I dont see any unpacking errors Sep 27 04:26:33 good time of the day all Sep 27 04:27:01 does anybody know what is "locale-base-en-us"? Sep 27 04:53:00 anybody around? Sep 27 05:05:28 IgorK, it is a package for translating the system into the en-us locale Sep 27 05:05:57 it depends upon other packages, which provide such things as a keyboard layouts and glibc error messages Sep 27 05:09:41 well for some reason this package was not present couple of days ago. can someone check on that please? Sep 27 05:32:36 so is this packet back in the tree? grg? anybody else? Sep 27 05:32:49 huh? Sep 27 05:44:30 grg: I cloned git on wednesday, and the build didn't find the package Sep 27 05:53:01 IgorK, i did a git pull this morning and my tmp dir contains locale-base-en-us along with all the other usual glibc locales Sep 27 05:53:14 (from scratch build) Sep 27 06:02:33 ok, thx, will try that in a sec Sep 27 06:02:40 gn Sep 27 06:12:49 gm everyone Sep 27 07:05:47 \/nick hrw Sep 27 07:05:49 morning Sep 27 07:29:37 good morning Sep 27 08:23:51 hello Sep 27 08:24:12 hi Sep 27 08:24:16 I'm trying to compile gcc-4.2.1 for SH4 target Sep 27 08:24:44 actually, I can do that when I drop zecke no host includes patch Sep 27 08:25:00 however I feel like it is not the right way Sep 27 08:25:30 (I'm worrying about side effects of using /usr/include path) Sep 27 08:25:36 yes Sep 27 08:25:44 why you need to drop the patch? Sep 27 08:25:52 btw. khem is working on sh4 Sep 27 08:25:55 how come gcc-4.2.4 can be compiled with zecke patch while gcc-4.2.1 cam Sep 27 08:25:59 can't Sep 27 08:26:13 the 4.2 gcc dont get much love these days Sep 27 08:26:18 may I paste few lines of logs? Sep 27 08:26:22 gcc-4.3 and gcc-4.5 is now used Sep 27 08:26:38 I need gcc-4.2.1 since I need glibc-2.6.1 Sep 27 08:26:53 sure you need glibc? Sep 27 08:26:59 target host won't start with glibc-2.6.1 compiled with gcc-4.4.4 Sep 27 08:27:39 try gcc-4.3 Sep 27 08:28:32 pthread_attr_setschedparam() returns errors on my SH4 box when I'm using glibc-2.10.1 (which can be compiled with gcc-4.4.4) Sep 27 08:28:59 also, I have kernel that must be signed by a binary tool made in 2005 Sep 27 08:29:07 gcc-4.4 ieehks Sep 27 08:29:09 ups Sep 27 08:29:11 ieehks Sep 27 08:29:12 signing Sep 27 08:29:31 it can sign kernels compiled with gcc-4.2.x Sep 27 08:29:35 hm wait for afternoon and ask khem about sh4 Sep 27 08:29:44 ok Sep 27 08:29:50 or on the oe-dev mailinglist Sep 27 08:38:22 hi guys Sep 27 08:38:40 i was going to crosscompile some java bytecode using gcj Sep 27 08:39:27 but it turned out that i manually need to export CLASSPATH pointing to the ecj.jar in my gcc-cross working directory before i can use gcj Sep 27 08:39:44 why you need gcj? Sep 27 08:40:01 i need to make target machine code out of the java stuff? Sep 27 08:40:34 i was wondering, where do i set the classpath so that it works automatically? Sep 27 08:40:36 javabytecode isnt enough? Sep 27 08:40:55 nope Sep 27 08:41:18 sorry that is not tested within oe Sep 27 08:41:23 it works though Sep 27 08:41:27 when manually doing it Sep 27 08:41:48 my question is just... do i have to set classpath in the gcc recipe anywhere Sep 27 08:42:00 or does it require modification to the recipes of the java tools? Sep 27 08:43:21 sorry dont know how gcj is build in oe and how it runs Sep 27 08:43:37 but dont you need gij? not gcj Sep 27 08:44:22 mipsel-oe-linux-gcj is what i run Sep 27 08:44:25 hm ah okay Sep 27 08:45:24 read on the gcj website it supports native machine code too Sep 27 08:50:39 hmm Sep 27 08:51:03 i just discovered that i actually have a /dream/oe1.6/dm8000/build/tmp/cross/mipsel/libexec/gcc/mipsel-oe-linux/4.4.4/ecj1 Sep 27 08:51:48 then i shouldn't be needing that ecj.jar anymore Sep 27 09:35:06 or how is that, woglinde? Sep 27 09:39:29 sorry dont know Sep 27 09:44:32 03Martin Jansa  07org.openembedded.dev * raefe32b4bd 10openembedded.git/recipes/kexecboot/kexecboot_git.bb: Sep 27 09:44:32 kexecboot: bump SRCREV for devicescan fix Sep 27 09:44:32 Signed-off-by: Martin Jansa Sep 27 09:44:32 03Martin Jansa  07org.openembedded.dev * r6be2a0d83f 10openembedded.git/recipes/kexecboot/linux-kexecboot-2.6.35/spitz/defconfig: Sep 27 09:44:33 linux-kexecboot-2.6.35: enable CONFIG_BLK_DEV_SD Sep 27 09:44:33 Signed-off-by: Martin Jansa Sep 27 09:44:34 03Martin Jansa  07org.openembedded.dev * r81daa524de 10openembedded.git/recipes/shr/shr-e-gadgets_git.bb: Sep 27 09:44:34 shr-e-gadgets: bump SRCREV Sep 27 09:44:35 Signed-off-by: Martin Jansa Sep 27 09:44:35 03Martin Jansa  07org.openembedded.dev * r9f979c42d1 10openembedded.git/recipes/linux/ (4 files in 2 dirs): Sep 27 09:45:01 jo jama Sep 27 09:45:30 linux-openmoko-2.6.34: add 3 patches from qtmoko Sep 27 09:45:30 Signed-off-by: Martin Jansa Sep 27 09:46:10 03Martin Jansa  07org.openembedded.dev * r2839bd9b81 10openembedded.git/recipes/xorg-lib/pixman_0.19.4.bb: Sep 27 09:46:10 pixman_0.19.4: increase D_P for SHR Sep 27 09:46:10 Signed-off-by: Martin Jansa Sep 27 09:48:22 woglinde: jo Sep 27 09:58:04 gm marex Sep 27 09:58:13 woglinde, hey Sep 27 10:01:45 woglinde... right now everything works when i put an export CLASSPATH = "/dream/oe1.6/dm8000/build/tmp/work/mipsel-oe-linux/gcc-cross-4.4.4-r0.1/gcc-4.4.4/ecj.jar:/dream/oe1.6/dm8000/build/tmp/work/mipsel-oe-linux/projectx-0.90.4.00.b32-r0/projectx-0.90.4.00.b32/src" into my projectx.bb Sep 27 10:02:00 the second part is clear, i can set that by using bb variables Sep 27 10:02:13 however, how do i correctly set the path to ecj.jar? Sep 27 10:07:42 fraxinath hm Sep 27 10:07:54 first think dont relay on work Sep 27 10:08:04 with inherit rm_work its gone Sep 27 10:15:02 hmm okay then i'll try to see if i can install the ecj.jar somewhere into build/tmp/cross at the end of gcc-cross Sep 27 10:16:32 hm we alreay have a ecj recipe Sep 27 10:16:42 that should work with gcj Sep 27 10:16:57 but I am not sure if gcj is installed under sysroot in a sane way Sep 27 10:17:22 hm and projectx should work openjdk Sep 27 10:21:55 hmm i don't have a specific recipe for ecj Sep 27 10:22:16 instead, i added a task to gcc-cross which invokes the contrib/download_ecj script from gcc Sep 27 10:24:42 fraxinath what do you think recipes/ecj/libecj-bootstrap_3.6.bb is for? Sep 27 10:25:50 hm you have a point Sep 27 10:26:14 hmm... qt is such a pain to build on my laptop Sep 27 10:26:18 hihi Sep 27 10:26:21 gm zecke Sep 27 10:26:26 bye a phenom Sep 27 10:26:29 ups buy Sep 27 10:26:34 or icore7 Sep 27 10:27:01 it is in the mail.. from Taiwan Sep 27 10:27:22 btw. pyqt compiles now Sep 27 10:27:30 will commit it this evening Sep 27 10:27:47 hi zecke Sep 27 10:27:54 qt 4.7 and declarative will come with the next pyqt release Sep 27 10:32:05 oah woglinde that pulls a lot of dependencies... Sep 27 10:32:08 classpath is biig Sep 27 10:32:24 sure Sep 27 10:32:29 jamvm looks big too Sep 27 10:32:38 o.O? Sep 27 10:32:48 icedtea/openjdk is big Sep 27 10:33:09 but than you have a java6 compatible compiler and vm Sep 27 10:33:20 lol and of course fails anyways Sep 27 10:33:29 hm might be Sep 27 10:33:30 i don't want a vm Sep 27 10:33:36 we didnt test it for mips Sep 27 10:33:36 woglinde: was just sending you an email Sep 27 10:33:41 lets see Sep 27 10:34:08 configure: error: cannot find libffi Sep 27 10:34:11 but it's installed Sep 27 10:35:22 zecke no email here Sep 27 10:35:31 woglinde: to heinold familie Sep 27 10:35:35 ha Sep 27 10:35:44 jamvm-initial-1.4.5-r1 configure fails to find libffi... it's part of gcc isn't it? Sep 27 10:35:56 however, i'm gonna get rid off the stuff again i think Sep 27 10:36:09 that brings too much overhead which i don't need Sep 27 10:36:43 zecke sure you send me the right email? Sep 27 10:37:05 no libffi is seperated Sep 27 10:37:14 was part of gcc some time ago Sep 27 10:37:25 hm I wonder why the depends dont work Sep 27 10:39:14 fraxinath than use find under sysroot to see if an ecj.jar is installed there Sep 27 10:39:43 i don't have master recipes though Sep 27 10:39:49 it's the opendreambox fork Sep 27 10:39:54 hm Sep 27 10:43:49 zecke: can I ask you about zecke-no-host-includes.patch and gcc-4.2.1 on SH4? Sep 27 10:53:40 pawelo_dcc: i am almost out of the door Sep 27 10:54:07 pawelo_dcc: lot of broken autoconf/Makefiles just put -I/usr/include in the compile line Sep 27 10:54:34 | fixproto: populating `include' Sep 27 10:54:34 | CROSS COMPILE Badness: /usr/include in INCLUDEPATH: /usr/include Sep 27 10:54:34 | fix-header: Internal error: abort in add_path, at c-incpath.c:371 Sep 27 10:54:39 pawelo_dcc: then weird stuff happens, e.g. when building on a glibc system for uclibc... it fails, or the wrong endianes is detected, or wrong freeytype etc get picked up Sep 27 10:55:31 zecke: I found that this problem returns from time to time, what is the typical solution? it is at gcc-cross-initial-4.2.1-r23.12.0 Sep 27 10:55:47 pawelo_dcc: so something is putting a -I/usr/include in the command line Sep 27 10:55:59 pawelo_dcc: check the config.log to see which autoconf macro it is... Sep 27 10:56:22 pawelo_dcc: the two options are a) someone finds a package that returns -I/usr/include b) some one just puts the includepath there Sep 27 10:59:40 zecke: I starded fesh build, I'll look for -I/usr/include at config.log for gcc when it fails Sep 27 11:06:46 Hi Sep 27 11:07:10 hi B_Lizzard_ Sep 27 11:07:27 I'm trying to create a recipe for dwm, which uses a simple makefile + a config.mk file which contains stuff like CC, CFLAGS etc. Sep 27 11:08:11 bitbake seems to ignore the CFLAGS and LDFLAGS definitions in config.mk and uses its own environment variables Sep 27 11:08:19 Is there any way of overriding that? Sep 27 11:08:38 B_Lizzard_ we have good reasons to ovveride the flags Sep 27 11:09:01 because stupid upstream authors dont care about cross-compiling Sep 27 11:09:09 I know Sep 27 11:09:14 03Martin Jansa  07org.openembedded.dev * r3f0fd6ab67 10openembedded.git/recipes/shr/shr-e-gadgets_git.bb: shr-e-gadgets: remove edje_convert calls (should be converted already in source) Sep 27 11:09:19 So I'd have to provide my own makefile? Sep 27 11:09:34 And just use different variable names where needed Sep 27 11:09:53 o.O? Sep 27 11:10:07 B_Lizzard_ why the dwm dont compiles with our flags? Sep 27 11:10:19 No, because it needs certain other things Sep 27 11:10:23 Example Sep 27 11:10:49 It has -DVERSION="5.8.2", which is passed in the CFLAGS Sep 27 11:10:58 Overriding it means it doesn't compile Sep 27 11:11:00 hm okay Sep 27 11:11:09 rewrite the makefile Sep 27 11:11:25 so its accepting the flafs from enviroment Sep 27 11:11:30 args flags Sep 27 11:11:44 I'll just define two LIBS and INCS variables in config.mk and add those in the makefile directly Sep 27 11:11:49 Thanks, woglinde Sep 27 11:13:52 hi all Sep 27 11:28:20 One more question Sep 27 11:28:38 I want to ship some machine-specific modifications to dwm Sep 27 11:28:54 machine or arch? Sep 27 11:28:59 Machine Sep 27 11:29:02 ben-nanonote Sep 27 11:29:22 whats the special? Sep 27 11:29:28 resolution? Sep 27 11:29:44 Nah, some keybindings and color changes and things like that Sep 27 11:29:51 dwm is configurable at build-time only Sep 27 11:30:28 If I put SRC_URI_append_ben-nanonote = file://config.h" for instance, the files will not replace the older ones Sep 27 11:30:49 I'd have to add code in do_configure to move them Sep 27 11:30:55 And that seems retarded Sep 27 11:30:59 Is there any better way? Sep 27 11:31:13 I suggest make it a patch Sep 27 11:31:20 I'd like to avoid patches to ensure buildability Sep 27 11:31:26 But if that's the only solution Sep 27 11:31:34 I'll do that Sep 27 11:31:37 o.O? Sep 27 11:32:05 Well, patches tend to break when versions are bumped Sep 27 11:32:19 so your config.h will do Sep 27 11:32:46 I've made modifications to both config.h and dwm.c, which is the entire dwm code Sep 27 11:33:04 So even if the version is bumped it will remain buildable Sep 27 11:33:19 make patch for config.h and a patch for dwm.c Sep 27 11:33:28 OK Sep 27 11:54:33 03Michael Smith  07master * r3619711f4c 10openembedded.git/recipes/openvpn/ (openvpn.inc openvpn_2.1.0.bb openvpn_2.1.3.bb): (log message trimmed) Sep 27 11:54:33 openvpn: change kernel-module-tun to RRECOMMEND Sep 27 11:54:33 RRECOMMEND, not RDEPEND, in case a particular platform doesn't have Sep 27 11:54:33 a separate kernel-module-tun package. Matches behaviour of other Sep 27 11:54:33 packages, like bridge-utils. Sep 27 11:54:33 Also switch to INC_PR. Sep 27 11:54:34 Signed-off-by: Michael Smith Sep 27 11:55:57 03Michael Smith  07master * rb64b1d26dc 10openembedded.git/recipes/gstreamer/gst-plugins-base_0.10.30.bb: Sep 27 11:55:57 gst-plugins-base 0.10.30: depend on gtk+ Sep 27 11:55:57 Some of the tests and examples require gtk. Sep 27 11:55:57 Signed-off-by: Michael Smith Sep 27 11:56:03 03Michael Smith  07master * r3d7f3a9e49 10openembedded.git/recipes/gstreamer/gst-plugins.inc: Sep 27 11:56:03 gst-plugins.inc: fix path to orcc when prefix != /usr Sep 27 11:56:03 Signed-off-by: Michael Smith Sep 27 11:56:03 03Michael Smith  07master * r96eb8296fe 10openembedded.git/recipes/webm/ (4 files in 2 dirs): Sep 27 11:56:03 libvpx: fix 0.9.1 build when prefix="" Sep 27 11:56:03 The configure script was changed in 0.9.1 and introduced some logic that Sep 27 11:56:04 breaks when prefix is an empty string. Sep 27 11:56:04 Also switch to INC_PR for 0.9.0 and 0.9.1. Sep 27 11:56:07 Signed-off-by: Michael Smith Sep 27 12:00:56 hi gnutoo Sep 27 12:24:47 Sorry for the ball-busting Sep 27 12:24:55 Yet another question Sep 27 12:25:40 zecke, are you still here? Sep 27 12:25:57 I'm making a recipe for dzen, and when wget downloads from http://gotmor.googlepages.com/dzen2-0.8.5.tar.gz I end up with a package name + a bunch of useless characters Sep 27 12:26:03 Is there anyway I can fix that? Sep 27 12:27:58 I'd have to override do_fetch, eh? Sep 27 12:31:03 B_Lizzard_ you are doing something wrong the url the packagename and the unpackedpackage is sane Sep 27 12:31:49 If you do "wget http://gotmor.googlepages.com/dzen2-0.8.5.tar.gz" on your machine you get a properly-packaged, um, package? Sep 27 12:31:54 I mean Sep 27 12:32:00 A properly named package Sep 27 12:37:08 ~curse bzr Sep 27 12:37:18 May the fleas of a thousand camels infest your most sensitive regions, bzr ! Sep 27 12:38:14 I used chromium Sep 27 12:38:22 and the package name was correct Sep 27 12:38:39 hrw *g* you choosed it Sep 27 12:39:25 I have no idea what's going on Sep 27 12:39:29 I'll try via svn Sep 27 12:39:30 woglinde: nope, I was chosen by others Sep 27 12:52:22 woglinde, hi Sep 27 12:52:49 lol about bzr Sep 27 12:52:56 I tried that with gnash head Sep 27 12:52:59 didn't work well Sep 27 12:53:05 btw I should fix gnash Sep 27 13:03:17 03Koen Kooi  07org.openembedded.dev * r1279c80d4f 10openembedded.git/recipes/usb-modeswitch/usb-modeswitch_1.1.4.bb: usb-modewitch: add 1.1.4 Sep 27 13:20:30 03Koen Kooi  07org.openembedded.dev * r030f3358ea 10openembedded.git/recipes/usb-modeswitch/ (usb-modeswitch-data_20100826.bb usb-modeswitch_1.1.4.bb): Sep 27 13:20:30 usb-modeswitch-data: add 20100826 Sep 27 13:20:30 usb-modeswitch: recommend the above Sep 27 13:23:41 hm, need to compile two kernels for the same machine; what would be the best way to do that (i started off creating two different recipes, but kernel recipes all create the same kernel ipk (kernel_2.6.34-r1_sygtd1.ipk) Sep 27 13:23:59 names seem to be fixed in kernel.bbclass Sep 27 13:24:35 eFfeM_work hm koen missused some targete for this look at the omap-psp recipe Sep 27 13:25:10 woglinde: thanks, will do (actually thought about creating a 2nd machine but that did not really seemed the proper way) Sep 27 13:25:55 hrms Sep 27 13:26:07 ~lart thread-handling in python Sep 27 13:26:08 * ibot drops a humongous exploding nuke on thread-handling in python Sep 27 13:27:09 woglinde: found multi-kernel.inc, that does produce multiple kernels, but no packages Sep 27 13:27:18 still need to see how to get it into my image then Sep 27 13:33:04 maybee I should use multiprocessing Sep 27 13:39:25 hi , i cant run libertas wifi driver , when i try to run , i got error "unknown symbol" wiphy_register , how can i fix this problem ? someone know ? Sep 27 13:40:11 when insmodding? if so insmod the module that provides that symbol Sep 27 13:41:12 yes , insmod Sep 27 13:41:31 it needs another module that you haven't loaded Sep 27 13:41:32 i edit defconfig file Sep 27 13:41:45 what kind of module i need for wifi ? Sep 27 13:41:49 libertas Sep 27 13:41:52 why can't you use modprobe? Sep 27 13:42:08 CONFIG_LIBERTAS=y Sep 27 13:42:40 ok i look modprobe too now Sep 27 13:43:04 if you used =y you do not get a module ! Sep 27 13:43:07 hm wow the Process from multiprocessing dont blocks keyboard Sep 27 13:43:36 how can i fix so ? Sep 27 13:44:15 no idea, insufficient info on hwat you did and insufficient knowledge on libertas Sep 27 13:44:25 http://pastebin.com/mXj4HTJL Sep 27 13:44:39 that is what i did Sep 27 13:46:31 and when do you get the message Sep 27 13:47:10 03Koen Kooi  07org.openembedded.dev * r0dc253d91e 10openembedded.git/recipes/usb-modeswitch/usb-modeswitch_1.1.4.bb: usb-modeswitch: fix recipe and packaging Sep 27 13:53:07 hmm how do i link a gcj-compiled java .o together with a c++ compiled.so? Sep 27 13:54:00 hmm Sep 27 13:54:20 fraxinath , use jc2go Sep 27 13:54:33 in the .java source there's a System.loadLibrary command Sep 27 13:54:55 fraxinath not that way Sep 27 13:54:55 and i compile all the source java files one-by-one into .objects Sep 27 13:55:00 args Sep 27 13:55:02 sorry Sep 27 13:55:08 that's how it works with projectx :) Sep 27 13:55:18 with this other program here the only difference is the external .so Sep 27 13:56:40 i forgot there is some compiler on net that makes java completely native C code Sep 27 13:56:58 ahh remember now Sep 27 13:57:03 http://www.ivmaisoft.com/jcgo/ <= use this Sep 27 13:57:19 fraxinath , its best Sep 27 13:57:26 or u can use PERC VM Sep 27 13:57:35 but i hate java has no framebuffer support on swing Sep 27 13:57:37 nah nah i don't want a vm Sep 27 13:57:45 i hate X Windowing system and slow X11 window Sep 27 13:57:49 i need to compile it to native mipsel code Sep 27 13:58:06 fraxinath , jcgo is now vm Sep 27 13:58:11 its native java compiler Sep 27 13:58:36 it took me a full week to get the whole gcj stuff to work :) now i'm gonna use it Sep 27 13:58:44 ugh Sep 27 13:58:54 thats uses an stonage classpath Sep 27 13:59:04 who ? Sep 27 13:59:13 who is using stonage classpath ? Sep 27 13:59:23 jcgo ? Sep 27 13:59:25 it's just a little linking problem here Sep 27 13:59:29 jcgo yes Sep 27 13:59:44 well , maybe but there is no %100 native compiler like Jcgo Sep 27 13:59:52 all others using Java on Layer Sep 27 14:00:03 JCGO directly makes ur java code into C Sep 27 14:00:19 i don't need java code into c Sep 27 14:00:35 bitbake -c menuconfig virtual/kernel <= when i do this , how can i recompile new configs and ignore defconfig ? Sep 27 14:01:14 exchange defconfig Sep 27 14:01:20 how ? Sep 27 14:01:39 bitbake -c menuconfig virtual/kernel <= when i do this , where the .config file stored so ? Sep 27 14:02:45 inside the kernel-dir Sep 27 14:03:24 /home/kadirbasol/OE/build/tmp-igep0020/work/igep0020-angstrom-linux-gnueabi/linux-igep2-2.6.33.7-0-r90/image/kernel/.config <= this ? Sep 27 14:04:26 yes Sep 27 14:07:30 ok Sep 27 14:08:30 then i do this : bitbake -c rebuild virtual/kernel Sep 27 14:08:39 hello, is there any OE developer who cross compiled nethack (I haven't found a recipe for it) Sep 27 14:08:44 btw , did someone tried java on framebuffer ? Sep 27 14:08:53 nethack ? Sep 27 14:08:56 whats nethack ? Sep 27 14:09:43 JDuke128: a game Sep 27 14:09:59 donno Sep 27 14:10:43 hm nethack should be there Sep 27 14:11:36 woglinde: are you a developer/maintainer? Sep 27 14:11:50 hm intressting Sep 27 14:12:09 than maybee some one want packaging and didnt it Sep 27 14:12:53 hm hm Sep 27 14:13:29 ok, if someone want to try to cross build it I hope he can tell me about to see if he can get an idea from a Pkgfile to build a recipe for it Sep 27 14:22:41 hi i've got a question. if i want to start build with bitbake i get thi error conf/bitbake.conf not found. what can i do? Sep 27 14:26:18 good morning Sep 27 14:27:01 adminuster: it seems you are missing local.conf file Sep 27 14:28:01 where should the local.conf be? Sep 27 14:28:02 khem: hi Sep 27 14:28:12 hrw howdy Sep 27 14:28:52 khem: want to check my linaro gcc 4.5-2010.09 recipes? Sep 27 14:29:14 adminuster: http://wiki.openembedded.net/index.php/Getting_started Sep 27 14:29:23 hrw: sure Sep 27 14:30:05 hi khem Sep 27 14:32:12 JaMa GNUtoo|laptop for the vectorization problem we could add -mvectorize-with-neon-quad to recipe's CFLAGS instead of diabling vectorizer Sep 27 14:32:26 hello woglinde Sep 27 14:33:32 khem: at Linaro we have a person to improve vectorization Sep 27 14:33:42 03Marcin Juszkiewicz  07org.openembedded.dev * r4292b4f8b7 10openembedded.git/recipes/gcc/ (gcc-4.4-2010.09.inc gcc-cross_4.4-2010.09.bb): gcc 4.4-2010.09: moved FILESDIR to inc file Sep 27 14:33:54 03Marcin Juszkiewicz  07org.openembedded.dev * r4b75473b41 10openembedded.git/recipes/gcc/ (6 files): (log message trimmed) Sep 27 14:33:54 gcc: added 2010.09 Linaro release of gcc-4.5 Sep 27 14:33:54 more about release: https://wiki.linaro.org/WorkingGroups/ToolChain Sep 27 14:33:54 This version does not have any OE patches applied (other then Sep 27 14:33:54 zecke-no-host-includes.patch in gcc-cross). Sep 27 14:33:54 There are build problems with this release in OE. But as it is not default Sep 27 14:33:55 version I am pushing it like it is. Sep 27 14:34:08 khem: check Sep 27 14:39:53 khem, ok Sep 27 14:40:53 hrw: I would have preferred the linaro patches on top of gcc 4.5 Sep 27 14:41:47 it becomes a maintainnance burden see the csl recipes Sep 27 14:42:25 any fundamental changes to gcc are very difficult so my goal is to consolidate the compiler Sep 27 14:44:44 khem: we do monthly releases so soon 4.5-2010.10 will be and I do not track bzr repository Sep 27 14:45:23 hrw: ok, I am tracking it Sep 27 14:45:28 ok then Sep 27 14:45:48 khem: but can you apply on top of gcc-4.5.1.tgz instead of fetching from svn? Sep 27 14:46:48 hrw: actually as gcc release its better to track the branch because only bugfixes go into it Sep 27 14:46:56 so minor relases dont matter Sep 27 14:47:01 ok Sep 27 14:47:02 thats why track svn Sep 27 14:47:52 feel free to drop gcc-4.5-2010.09 recipes then when you will merge your work Sep 27 14:47:56 olden days minor releases got features and ABI changes Sep 27 14:47:59 khem: you have my Ack Sep 27 14:48:05 hrw: thank you Sep 27 14:48:45 khem whats the status now? Sep 27 14:48:58 I need acks Sep 27 14:49:10 khem: no, I thank you - it is less work for me as you maintain OE part Sep 27 14:49:16 and testing if one can do Sep 27 14:49:24 hm right Sep 27 14:49:39 hrw sure Sep 27 14:50:18 hrw is Tejas part of Linaro ? Sep 27 14:50:31 does he work for Linaro I mean Sep 27 14:50:39 Tejas? Sep 27 14:50:51 he is one a compiler person at arm Sep 27 14:51:00 doing autovectorization Sep 27 14:51:28 hi Sep 27 14:51:35 khem: I lost track of linaro people Sep 27 14:52:17 k Sep 27 14:53:18 I have problem building esmart, it ends configure with "No package 'eina-0' found" error Sep 27 14:53:19 hi martix Sep 27 14:53:50 eina recipe is already builded Sep 27 14:54:22 how aboout eina-0 Sep 27 14:55:16 recipe eina-0 isnt exist Sep 27 14:55:30 Martix: change that to eina Sep 27 14:55:49 Martix: it was renamed in efl about 2 months ago (2 EFL bumps ago) Sep 27 14:56:33 JaMa: should I patch the esmart configure? Sep 27 14:57:01 Martix: yup.. I've patched everything built with SHR feeds (except zhone) Sep 27 14:57:10 but we probably don't build esmart Sep 27 14:59:13 ok, thanks Sep 27 15:23:33 hmm, i want a local git hook that prevents pushing any commits whose subjects are prefixed with WTF: to the 'origin' remote (or all but remote X) - would be a nice sanity check Sep 27 15:36:05 lol Sep 27 15:37:39 on x86_64 is there a way to get meta-toolchain etc to build for 32bit to publish an SDK? Sep 27 15:37:56 kergoth, there is only 1 commit with that word in it, but it date to 2006 Sep 27 15:38:29 tharvey, I think so but I don't know the exacts steps involved Sep 27 15:38:33 look at nativesdk Sep 27 15:38:36 or something like that Sep 27 15:38:55 well, that hook would only be useful if i remember to actually prefix the unfinished commits that way :) Sep 27 15:39:37 ok Sep 27 15:39:58 what about DO_NOT_PUSH Sep 27 15:40:02 hehe Sep 27 15:40:03 or something like that Sep 27 15:40:03 ? Sep 27 15:40:06 thatd work too Sep 27 15:40:11 it would be more polite and understandable Sep 27 15:40:37 * kergoth 'll set up a *local* hook for this for his own personal use, at least Sep 27 15:40:46 ok Sep 27 15:42:02 GNUtoo|laptop, perhaps canadian-sdk? Sep 27 15:42:27 I never looked into sdk but that is also a kind of sdk Sep 27 15:43:55 its a bit sad that canadian-sdk.bb and meta-toolchain.bb have the same DESCRIPTION - there is such a lack of documentation in the bb's Sep 27 15:46:28 new do_configure problem, with entrance-0.9.0.010-r3 - No package 'ecore-config' found No package 'ecore-data' found Sep 27 15:47:45 ecore is compiled with --enable-ecore-config and libecore_config*.so.* is in recipes/efl1/ecore.inc Sep 27 15:47:59 but not in deploy dir Sep 27 15:48:43 03Chase Maupin  07org.openembedded.dev * r0c1dfcbe94 10openembedded.git/recipes/qt4/ (files/qte.sh qt4-embedded.inc): (log message trimmed) Sep 27 15:48:43 qte: only export touchscreen setting if ts exists Sep 27 15:48:43 * Changed the qte.sh script to check for whether or not the Sep 27 15:48:43 touchscreen exists before exporting QWS_MOUSE_PROTO to use Sep 27 15:48:43 the touchscreen. This is so that devices that don't have Sep 27 15:48:43 a touchscreen can use the regular mouse support. Sep 27 15:48:44 * Bumped the PR Sep 27 15:50:06 i need defconfig for igepv2 that allows me to use SGX & WIFI , but i try to do my own kernel but losing my head here Sep 27 15:50:12 very hard and evil to make Sep 27 15:50:22 drivers/net/wireless/libertas_tf/main.c:340: error: ‘struct ieee80211_vif’ has no member named ‘addr’ Sep 27 15:50:35 when i add , wifi support , bitbake virtual/kernel Sep 27 15:50:38 i got this error Sep 27 15:50:43 someone have any idea? Sep 27 15:51:06 03Chase Maupin  07org.openembedded.dev * r278e683404 10openembedded.git/recipes/freetype/ (freetype_2.2.1.bb freetype_2.3.12.bb freetype_2.3.6.bb): (log message trimmed) Sep 27 15:51:06 freetype: update LICENSE field Sep 27 15:51:06 * The LICENSE.TXT file says that freetype may be licensed under Sep 27 15:51:06 either the FreeType license or the GPLv2 license. This patch Sep 27 15:51:06 updates the LICENSE field to reflect that this software can Sep 27 15:51:06 use either license. Sep 27 15:51:08 Signed-off-by: Chase Maupin Sep 27 15:53:15 03Chase Maupin  07org.openembedded.dev * r8a7be22800 10openembedded.git/recipes/fbset/fbset_2.1.bb: Sep 27 15:53:15 fbset_2.1: add dependency on bison-native Sep 27 15:53:15 * fbset requires bison in order to build. This patch adds Sep 27 15:53:15 bison-native to the DEPENDS. Sep 27 15:53:15 Signed-off-by: Chase Maupin Sep 27 15:53:15 Acked-by: Frans Meulenbroeks Sep 27 15:53:16 Signed-off-by: Denys Dmytriyenko Sep 27 15:57:30 03Jeff Lance  07org.openembedded.dev * r74e614260a 10openembedded.git/recipes/ti/ (3 files in 3 dirs): (log message trimmed) Sep 27 15:57:30 matrix-gui-e: updating to latest source revision Sep 27 15:57:30 * Updated recipe to use the latest source revision 131 Sep 27 15:57:30 * Included platform support Sep 27 15:57:30 * Removed init files from OE repository since they are now part Sep 27 15:57:31 of the matrix repository. Sep 27 15:57:31 * Bumped SRCREV to 136 in version 2 of this patch Sep 27 15:57:38 03Jeff Lance  07org.openembedded.dev * r8e17c351d7 10openembedded.git/recipes/ti/ (5 files in 3 dirs): (log message trimmed) Sep 27 15:57:38 matrix-gui: updating to use latest source revision Sep 27 15:57:38 * Use latest source revision 131 Sep 27 15:57:38 * Add platform support for multiple platforms Sep 27 15:57:39 * Remove un-used files from SRC_URI and repository. Sep 27 15:57:39 * init files are now part of the matrix repository Sep 27 15:57:40 * matrix-gui.desktop is now part of the matrix repository Sep 27 15:57:41 03Jeff Lance  07org.openembedded.dev * r2645779a26 10openembedded.git/recipes/ti/ (matrix-gui-common/browser matrix-gui-common_svn.bb): (log message trimmed) Sep 27 15:57:41 matrix-gui-common: update to latest sources Sep 27 15:57:41 * Updated to use repository revision 131 Sep 27 15:57:42 * This includes platform support Sep 27 15:57:42 * Removed the browser binary Sep 27 15:57:43 * Copy files based on PLATFORM Sep 27 15:57:43 * Bumped SRCREV to 136 in version 2 of this patch Sep 27 16:00:13 tzcode bumped their PV Sep 27 16:00:30 It is now 2010m Sep 27 16:12:06 entrance configure problem, nothing? Sep 27 16:25:15 ~lart ecore Sep 27 16:25:16 * ibot gives ecore a "free" copy of Windows and then charges double for "Upgrades" Sep 27 16:45:52 hi khem Sep 27 16:46:12 with last obtimisations flags I have that: X11 error: BadAtom (invalid Atom parameter) when using xv Sep 27 16:46:32 hm Sep 27 16:47:40 maybe it's not the flags? Sep 27 16:47:48 and maybe mplayer_git was bumped? Sep 27 17:04:03 03Chris Larson  07master * r25a0253a1f 10openembedded.git/classes/amend.bbclass: Sep 27 17:04:03 amend.bbclass: kill override specific Sep 27 17:04:03 Signed-off-by: Chris Larson Sep 27 17:18:29 GNUtoo|laptop: hmm what did you rebuild Sep 27 17:18:47 mplayer Sep 27 17:19:02 and what change did you make Sep 27 17:20:08 eFfeM: hello Sep 27 17:20:08 that one Sep 27 17:20:09 : Sep 27 17:20:17 o.0 Sep 27 17:20:52 -f-no-tree-vectorize -> -mvectorize-with-neon-quad Sep 27 17:21:11 not sure if it's the cause tough Sep 27 17:22:35 hi khem, all Sep 27 17:25:42 eFfeM: do u have armv7 hardware Sep 27 17:26:01 GNUtoo|laptop: ok Sep 27 17:26:14 GNUtoo|laptop: and you just recompiled mplayer nothing else ? Sep 27 17:26:24 I can try the old mplayer to be sure Sep 27 17:26:34 GNUtoo|laptop: try same sources Sep 27 17:26:44 re Sep 27 17:27:08 compile two different sets Sep 27 17:28:19 hmmm same issue with old one Sep 27 17:28:21 so... Sep 27 17:28:25 :) Sep 27 17:28:31 its not the option Sep 27 17:28:35 it worked before the full recompile Sep 27 17:28:45 indeed Sep 27 17:28:45 khem, yes, beagleboard Sep 27 17:29:12 eFfeM: if you have cycles try gcc 4.5 with the patches I posted Sep 27 17:29:50 GNUtoo|laptop: you mean to say you did a full recompile of system with gcc 4.5 with linaro patches and now you see this issue ? Sep 27 17:30:29 I mean to say: Sep 27 17:30:39 just before the full recompile with linaro it worked Sep 27 17:30:51 but I can't be sure it's that Sep 27 17:30:54 GNUtoo|laptop: hmmm ok Sep 27 17:30:59 because some things may have been updated Sep 27 17:31:02 in between Sep 27 17:31:05 GNUtoo|laptop: ah Sep 27 17:31:07 I'll git log mplayer Sep 27 17:31:28 GNUtoo|laptop: you should take the same snapshot that worked and recompile it with new compiler Sep 27 17:31:36 in OE there can be so many changes Sep 27 17:31:48 hmmm didn't change Sep 27 17:32:01 ok Sep 27 17:32:07 I've old images Sep 27 17:32:16 I'll check later(maybe after eating) Sep 27 17:32:21 GNUtoo|laptop: ok Sep 27 17:35:13 khem, will peek into it when I have time, but it might take a few days before i get to this Sep 27 17:35:27 unless I can kick off something on our autobuilder Sep 27 17:35:37 the patches are not on a branch, are they? Sep 27 17:36:50 they are Sep 27 17:37:00 in my user branch Sep 27 17:40:22 hmm anyone able to get networkmanager_0.8.1 to build yet? Sep 27 17:40:34 i know it's marked as DEFAULT_PREFERENCE="-1" Sep 27 17:41:22 having a problem with policykit dependency Sep 27 17:41:33 | checking for POLKIT... configure: error: Package requirements (polkit-gobject-1) were not met... Sep 27 17:45:18 khem, ah ok, that makes it simpler to run them on the autobuilder Sep 27 17:46:58 khem this one: khem/import-linaro-gcc-4.5 Sep 27 17:52:13 khem kicked off a build on our autobuilder, your branch, minimal, console-image Sep 27 17:55:53 and here's todays housekeeping Sep 27 17:56:03 03Frans Meulenbroeks  07org.openembedded.dev * r43a32bc5d7 10openembedded.git/recipes/python/ (3 files in 3 dirs): Sep 27 17:56:03 python: moved unused files to obsolete dir Sep 27 17:56:03 Signed-off-by: Frans Meulenbroeks Sep 27 17:56:05 03Frans Meulenbroeks  07org.openembedded.dev * rdc37764c5f 10openembedded.git/recipes/pimlico/files/tasks-single.diff: Sep 27 17:56:05 pimlico : moved unused files to obsolete dir Sep 27 17:56:05 Signed-off-by: Frans Meulenbroeks Sep 27 17:56:09 03Frans Meulenbroeks  07org.openembedded.dev * rc5a5631ae2 10openembedded.git/recipes/procps/procps-3.2.8/pagesz-not-constant.patch: Sep 27 17:56:09 procps : moved unused files to obsolete dir Sep 27 17:56:09 Signed-off-by: Frans Meulenbroeks Sep 27 17:56:09 03Frans Meulenbroeks  07org.openembedded.dev * r4e0548d675 10openembedded.git/recipes/pidgin/pidgin/status-icon-theme-crash.patch: Sep 27 17:56:09 pidgin : moved unused files to obsolete dir Sep 27 17:56:09 Signed-off-by: Frans Meulenbroeks Sep 27 17:56:13 03Frans Meulenbroeks  07org.openembedded.dev * r7173015a1b 10openembedded.git/recipes/pthsem/files/ldflags.patch: Sep 27 17:56:13 pthsem : moved unused files to obsolete dir Sep 27 17:56:13 Signed-off-by: Frans Meulenbroeks Sep 27 17:56:13 03Frans Meulenbroeks  07org.openembedded.dev * rbf0aff8bed 10openembedded.git/recipes/pulseaudio/files/ (3 files): Sep 27 17:56:13 pulseaudio : moved unused files to obsolete dir Sep 27 17:56:13 Signed-off-by: Frans Meulenbroeks Sep 27 18:10:49 howdy Sep 27 18:14:14 03Roman I Khimov  07org.openembedded.dev * rb52f8b25d9 10openembedded.git/recipes/iw/ (iw/kill-git-version-check.patch iw_0.9.20.bb): Sep 27 18:14:14 iw: fix build Sep 27 18:14:14 Signed-off-by: Roman I Khimov Sep 27 18:15:23 eFfeM: ok Sep 27 18:15:37 thx Sep 27 18:26:50 kergoth, eFfeM: can you please look at the patch I just sent to the maillist - it fixes the do_unpack()/SRC_URI problem... Sep 27 18:27:36 #join php Sep 27 18:27:36 will do Sep 27 18:28:31 kergoth_: thanks Sep 27 18:31:21 I have been trying to do a git with the following "git clone git://gitorious.org/gumstix-oe/mainline.git org.openembedded.dev" Sep 27 18:31:51 and I get the error fatal: The remote end hung up unexpectedly fatal: early EOF fatal: index-pack failed Sep 27 18:31:59 can someone help with why? Sep 27 18:36:11 nm found the error Sep 27 18:40:00 kergoth, khem: thanks for acks Sep 27 18:40:45 denix, will do, thanks for looking into this Sep 27 18:43:21 Islandmedic: go there http://gitorious.org/gumstix-oe/, pick http? Sep 27 18:43:43 Islandmedic: e.g. gitorious might have network issues, you might have network issues (being behind a firewall?) Sep 27 18:44:54 denix, reviewed the patch, let me do a quick test Sep 27 18:45:10 but looks good as far as I understand things Sep 27 18:45:40 eFfeM: ok, I'll wait for your test results Sep 27 18:46:33 btw, I only tested it with my failed case of amend.inc, hopefully it resolves yours with override too... :) Sep 27 18:48:08 denix, well from what i understood when I analysed this your patch seems to cure it Sep 27 18:48:36 guess it'll take a 5-10 mins before i have results Sep 27 18:54:14 re Sep 27 19:04:27 denix, it works here, ack emailed, thanks for the patch Sep 27 19:15:20 re Sep 27 19:26:15 03Vaudano Luca  07org.openembedded.dev * r31fbeac341 10openembedded.git/recipes/estardict/estardict_0.4.bb: Sep 27 19:26:15 estardict: add new version 0.4 Sep 27 19:26:15 Signed-off-by: Martin Jansa Sep 27 19:30:25 eFfeM: thanks for testing and verifying it! Sep 27 19:30:46 foo Sep 27 19:30:52 hi denix Sep 27 19:31:10 woglinde: hi Sep 27 19:33:14 i ve broken defconfig , how can i get its default settings ? Sep 27 19:33:40 git checkout? Sep 27 19:33:52 'broken defconfig'? Sep 27 19:33:57 defconfig for *what*? Sep 27 19:34:17 03Denys Dmytriyenko  07org.openembedded.dev * r661658f62b 10openembedded.git/classes/base.bbclass: (log message trimmed) Sep 27 19:34:17 base.bbclass: not use bb.fetch.urldata in do_unpack() for traversing SRC_URIs Sep 27 19:34:17 bb.fetch.urldata contains cached SRC_URIs, but in case SRC_URI is altered by Sep 27 19:34:17 amend.inc or machine override, the original SRC_URI is not removed from the Sep 27 19:34:17 cache, leading to calling do_unpack() multiple times on original and modified Sep 27 19:34:18 versions of the file. Don't rely on bb.fetch.urldata and traverse SRC_URIs Sep 27 19:34:19 directly. Sep 27 19:46:09 kergoth yeah bitbake is now fast Sep 27 19:55:08 khem: ping Sep 27 19:55:54 khem: ping too Sep 27 20:01:29 03Denys Dmytriyenko  07org.openembedded.dev * r3d1603cdcb 10openembedded.git/recipes/powervr-drivers/omap3-sgx-modules_1.4.14.2616.bb: Sep 27 20:01:29 omap3-sgx-modules: update SRC_URI with public link Sep 27 20:01:29 Signed-off-by: Denys Dmytriyenko Sep 27 20:05:06 denix: jo Sep 27 20:05:13 fraxinath: hello Sep 27 20:05:23 hiho khem Sep 27 20:05:59 do you know how to link mipsel-gcj-compiled java objects together with a c++ .so ? Sep 27 20:06:52 you need JNI Sep 27 20:07:02 * khem hates Java Sep 27 20:07:39 khem why? Sep 27 20:08:01 Its a un needed resource hog Sep 27 20:08:22 people should use other OO languages like even C++ Sep 27 20:09:06 *g* Sep 27 20:09:59 fraxinath: http://www.math.ucla.edu/~anderson/JAVAclass/JavaInterface/JavaInterface.html Sep 27 20:10:26 woglinde_: I have worked on dynamic compilers during my last years at Intel for all this byte code crap Sep 27 20:10:29 khem: in GCC there should be some basic compability between C++ and Java, e.g. a Java exception can be caught in C++ Sep 27 20:11:03 * woglinde_ waits for armqemu xorg-image Sep 27 20:11:05 khem: the sources are all done Sep 27 20:11:09 to test pyqt Sep 27 20:11:12 it's just a crosscompiling issue Sep 27 20:11:24 * zecke rebuilds qt4-embedded once again Sep 27 20:11:32 fraxinath: well, check if your library is loaded Sep 27 20:11:44 isn't Sep 27 20:11:46 qte now Sep 27 20:11:54 fraxinath: well, why is that the case? Sep 27 20:11:54 zecke: runtime systems are different Sep 27 20:12:04 I dont think they can interact at that level Sep 27 20:12:28 03Martin Jansa  07org.openembedded.dev * re105b26afd 10openembedded.git/recipes/linux/ (4 files in 2 dirs): Sep 27 20:12:28 linux-openmoko-2.6.32: upgrade to 2.6.32.23 and add 3 patches from qtmoko Sep 27 20:12:28 Signed-off-by: Martin Jansa Sep 27 20:12:33 i am not really sure how the linker command line has to look like *blush* Sep 27 20:12:53 fraxinath: well, how far are you? do you have a .so? Sep 27 20:12:53 does anybody see any issues with the mips64 patch series ? Sep 27 20:13:08 yes it's there Sep 27 20:13:08 here is last call before they go in Sep 27 20:13:11 khem: just wanted to ask why you didn't update omap-sgx-modules URL, I just pushed the fix... Sep 27 20:13:22 and when letting the java code run in the vm on the pc, it also imports it correctly Sep 27 20:13:30 khem: i have not seen everything, how did you generate the site-conf? by hand? Sep 27 20:13:38 fraxinath: is someone trying to dlopen this library? Sep 27 20:13:42 denix: ah forgot :) Sep 27 20:13:44 using System.loadLibrary Sep 27 20:14:09 khem: maybe it would be nice to fix the autoconf in sites to have all the tests there Sep 27 20:14:10 zecke: no its copy of mips 32 bit counterparts Sep 27 20:14:17 fraxinath: and what is the error? Sep 27 20:14:25 fraxinath: with strace, do you see a dlopen? Sep 27 20:14:30 zecke: you mean have all cached hmmmm Sep 27 20:14:43 zecke: you mean when i run the linker command with strace? Sep 27 20:14:56 fraxinath: no, the System.loadLibrary call Sep 27 20:14:57 zecke: actually we should cache all runtime tests yes Sep 27 20:15:14 zecke i dunno i am not at any point where i could run the code Sep 27 20:15:32 khem: well, I created an autoconf project with copies of m4 files at one point. Sep 27 20:15:50 fraxinath: you are confused. So you have a .so, it does not get loaded, but you do not have an app that executes it? Sep 27 20:15:53 the error is either "undefined reference to `hidden alias for java::lang::String* ...." Sep 27 20:16:20 zecke it's a java app which i want to crosscompile for mipsel Sep 27 20:16:32 when i run that java app on the pc then it imports the .so allright Sep 27 20:16:38 now i just need a static binary for mips Sep 27 20:17:01 fraxinath: well, have you built a static binary for PC? Sep 27 20:17:09 calling me confused! pfü! :) Sep 27 20:17:19 naah why would i do that Sep 27 20:17:24 fraxinath: there is a way to 'force' linking all the Java symbols Sep 27 20:17:39 fraxinath: to see that it is failing too and not mips specific? Sep 27 20:18:25 i am pretty sure it's not mips specifig Sep 27 20:18:28 -g +c Sep 27 20:18:44 fraxinath: http://gcc.gnu.org/wiki/Statically_linking_libgcj?action=show&redirect=Statically+linking+libgcj Sep 27 20:18:57 fraxinath: well, try it. :) Sep 27 20:19:12 zecke i am Sep 27 20:19:26 that's how i got projectx to work Sep 27 20:19:42 however, it fails in the second step pretty much Sep 27 20:19:52 fraxinath: is libgcj.so on target Sep 27 20:20:14 and see if its in DT_NEEDED when you do readelf -d app Sep 27 20:20:35 yeah libgcj is Sep 27 20:20:42 ok Sep 27 20:20:56 java::lang::String which .so has it Sep 27 20:21:03 but i don't know how to mix in the other .so Sep 27 20:21:38 fraxinath: you can ask static linker to emit them into DT_NEEDED Sep 27 20:21:38 if i put it as a -l then i get /dream/oe1.6/dm8000/build/tmp/cross/mipsel/lib/gcc/mipsel-oe-linux/4.4.4/../../../../mipsel-oe-linux/bin/ld: warning: libstdc++.so.6, needed by /dream/projects/bluray/aacskeys-0.4.0c-mipsel/lib/linux//libaacskeys.so, not found (try using -rpath or -rpath-link) Sep 27 20:21:49 or infact try o LD_PRELOAD them Sep 27 20:22:05 hmm LD_PRELOAD? Sep 27 20:22:07 never saw that Sep 27 20:22:50 libstdc++.so.6 why cant it be found Sep 27 20:22:53 is it not staged Sep 27 20:23:01 it's staged Sep 27 20:23:19 strange Sep 27 20:23:24 i probably have the syntax wrong Sep 27 20:23:31 since i can't use oe automake Sep 27 20:23:33 can you build a sumple hello.cpp Sep 27 20:23:50 then show the syntax here Sep 27 20:24:07 btw use g++ Sep 27 20:24:10 to link Sep 27 20:24:21 if you are using c++ libs/code Sep 27 20:24:28 ah okay good point Sep 27 20:26:39 khem: I get the same kind of warnings for libc.so Sep 27 20:27:05 zecke: hmmm 2.20.1 Sep 27 20:27:31 I did not apply patch to other binutils Sep 27 20:27:40 yeah, old angstrom binutils Sep 27 20:27:50 use 2.20.1 Sep 27 20:28:04 what does old angstrom use 2.18 something Sep 27 20:28:44 * khem this toothpick is bad it gets stuck in the teeth Sep 27 20:32:01 zecke, khem: http://pastebin.com/pEdyLm9K Sep 27 20:32:51 khem: yeah 2.18 Sep 27 20:33:17 zecke: sure you will see the problem Sep 27 20:34:12 fraxinath: can you link with -lstdc++ instead of -lsupc++? Sep 27 20:34:19 i shall try Sep 27 20:34:22 fraxinath: why not use g++ to link instead of gcj Sep 27 20:34:39 zecke: I think that may not work either Sep 27 20:34:45 khem: g++ doesn't accept the -static-libgcj but other than that it made no difference Sep 27 20:35:00 fraxinath: hmm Sep 27 20:35:00 fraxinath: there is the error that it does not find the lib (e.g. linking with gcc instead of g++), but your AACS code is using stl, so it should link to it Sep 27 20:35:29 yeah explicitly saying to link with libstdc++ is ok Sep 27 20:35:39 but I wonder why g++ can not find it Sep 27 20:35:52 it should find certain libs automagically Sep 27 20:36:13 specially the runtime Sep 27 20:36:39 http://pastebin.com/avaA49fS Sep 27 20:37:18 -Wl,--no-as-needed Sep 27 20:37:20 may be Sep 27 20:37:55 may be what? Sep 27 20:37:57 fraxinath: ok but it prolly did find libstdc++ Sep 27 20:38:50 fraxinath: so you want to do static linking with gcj Sep 27 20:39:02 aye Sep 27 20:39:12 then I think you need to specify -lstdc++ Sep 27 20:39:31 i put it in there already, cf pastebin Sep 27 20:39:34 just after -lrt Sep 27 20:39:54 ok Sep 27 20:40:04 and use gcj Sep 27 20:40:27 hmm no success Sep 27 20:42:30 http://pastebin.com/XKizQR3r Sep 27 20:43:14 fraxinath: that code looks like your app code is missing something Sep 27 20:43:15 os, sorry wrong pastebin Sep 27 20:43:42 here we go: http://pastebin.com/V9cn5rWs Sep 27 20:44:24 i don't think the app is missing anything... since it works all right in the VM (it's not my code either) Sep 27 20:46:22 fraxinath: at least some progress Sep 27 20:46:41 is AACSKeys.cpp public? Sep 27 20:46:45 khem how do I started qemuarm image with oe build? Sep 27 20:47:15 fraxinath: seems some classes are missing Sep 27 20:47:33 woglinde_: look into contrib/qemu/run-qemu.sh Sep 27 20:49:28 khem -> /scratch/oe/sysroots/i686-linux/usr/bin/qemu-system-arm Sep 27 20:49:34 hahahaha Sep 27 20:49:43 woglinde_: read the comments Sep 27 20:49:57 its not meant to be generic Sep 27 20:50:01 you need to adapt Sep 27 20:50:02 it Sep 27 20:50:43 zecke yeah it's open source Sep 27 20:51:09 http://rapidshare.com/files/187914704/aacskeys-0.4.0.tar.gz Sep 27 20:53:56 fraxinath: did you change the code in anyway? Sep 27 20:54:07 the aacskeys code, no Sep 27 20:54:15 the dumphd code i only too out the gui stuff Sep 27 20:54:21 fraxinath: move -Wl -static-libgcj to end Sep 27 20:54:28 aye Sep 27 20:54:45 oh i mean the -lgcj Sep 27 20:55:11 could you show me ./source/dumphd/aacs/AACSKeys.c*? Sep 27 20:56:05 aye Sep 27 20:56:09 http://pastebin.com/KRiqxH3C Sep 27 20:56:12 http://dumphd.sourcearchive.com/documentation/0.61-0.4ubuntu1/classdumphd_1_1aacs_1_1AACSKeys_d8137aee033c027a81d4ff7c75b8fad9.html#d8137aee033c027a81d4ff7c75b8fad9 Sep 27 20:56:20 why are these functions missing Sep 27 20:56:49 khem: didn't change anything with the new syntax Sep 27 20:57:45 I think something is missing in classdumphd Sep 27 20:57:50 do u have that linking in Sep 27 20:58:57 http://dumphd.sourcearchive.com/ Sep 27 20:59:09 khem is there a way to run qemuarm with xorg? Sep 27 20:59:19 woglinde_: it should run it Sep 27 20:59:23 x11-image Sep 27 20:59:36 couple of weeks ago it was running it Sep 27 21:00:06 khem hm vga stuff isnt found Sep 27 21:00:08 woglinde_: hmmm now I dont remember if I used the qemu shipped with my ubuntu box or the one we compile Sep 27 21:00:25 woglinde_: ok yeah SDL is not build into our qemu Sep 27 21:00:32 use the one thats on your build box Sep 27 21:00:37 args no-graphic Sep 27 21:00:43 remove that Sep 27 21:01:20 woglinde_: As a compiler guy I usually need a console and ssh Sep 27 21:01:22 thats it Sep 27 21:01:30 vnc seems to work Sep 27 21:01:35 woglinde_: cool Sep 27 21:01:42 are you able to connect Sep 27 21:01:50 I was not able to connect to vnc somehow Sep 27 21:01:57 hm only showd angstroem boot splash Sep 27 21:02:05 but now only cursor Sep 27 21:02:14 it takes a while Sep 27 21:02:25 to boot Sep 27 21:02:44 give it some minutes Sep 27 21:02:46 hm do you know what I need to boot into console for vnc? Sep 27 21:02:54 args Sep 27 21:02:59 put for console= Sep 27 21:03:23 usual console should be redirected Sep 27 21:03:30 isnt it happening Sep 27 21:03:38 no Sep 27 21:03:41 fraxinath: you must have two different AACCKeys.c* one from dumphd, one from acckeys Sep 27 21:03:46 I only see a blinking cursor now Sep 27 21:04:13 the dumphd one is not a c Sep 27 21:04:15 it's java Sep 27 21:04:47 fraxinath: yes then link this javaclass in Sep 27 21:04:51 http://pastebin.com/025Hsajh Sep 27 21:05:18 line 49 is where it tries to load the c++ so Sep 27 21:06:13 ah strg alt 3 Sep 27 21:06:25 woglinde_: yes Sep 27 21:06:33 thats also possible to use Sep 27 21:06:48 args Sep 27 21:06:49 xkb Sep 27 21:07:05 failed to compile keymap Sep 27 21:08:04 fraxinath: do you link aacckeys.o (result of aacckeys)? Sep 27 21:08:21 fraxinath: with objdump on that file, which symbols do you export? Sep 27 21:08:52 wait i'm confused this time really Sep 27 21:09:04 JNIEXPORT jstring JNICALL Java_dumphd_aacs_AACSKeys_getVersionString (JNIEnv *env, jobject jobj) Sep 27 21:09:06 the aacskeys.o comes from the java source Sep 27 21:09:16 that one doesn't export any symbols does it? Sep 27 21:09:21 fraxinath: you denied last time and zecke worked hard to prove it :) Sep 27 21:09:31 ublic native String getVersionString(); Sep 27 21:09:36 the libaacskeys.so i inspected with objdump earlier Sep 27 21:09:50 and that one does have a getVersionString function indeed Sep 27 21:09:53 where do i have it... Sep 27 21:09:58 lisppaste: is it exporting the Java_dumphd_aacs_AACSKeys_getVersionString? Sep 27 21:10:02 fraxinath: is it global Sep 27 21:10:10 00005390 : Sep 27 21:10:11 5390: 3c1c0003 lui gp,0x3 Sep 27 21:10:11 ...... Sep 27 21:10:37 fraxinath: use readelf -s Sep 27 21:10:44 fraxinath: use readelf -sW Sep 27 21:11:38 20: 00005390 32 FUNC GLOBAL DEFAULT 10 Java_dumphd_aacs_AACSKeys_getVersionString Sep 27 21:13:23 is that how it's supposed to look? Sep 27 21:15:04 seems right Sep 27 21:15:37 ./source/dumphd/aacs/AACSKeys.o can you dump it with readelf -sW Sep 27 21:16:58 http://pastebin.com/Mehfxvkt Sep 27 21:18:07 it's not the fault of "$(CROSS_COMPILE)objcopy -L '_ZGr8_$$_dummy' $@ " which i run in the %.o: %.java rule is it? Sep 27 21:18:12 hmmm thats a hidden alias for java::lang::String* dumphd::aacs::AACSKeys::getVersionString() Sep 27 21:18:23 i needed that with the new gcc for projectx Sep 27 21:18:29 so i put it in there aswell Sep 27 21:18:39 _ZGAN6dumphd4aacs8AACSKeys16getVersionStringEJPN4java4lang6StringEv Sep 27 21:18:58 zecke: yes exactly Sep 27 21:19:03 that is certainly a different signature Sep 27 21:19:10 JNI vs. CNI Sep 27 21:19:16 I guess so Sep 27 21:19:28 demangler is kicking in Sep 27 21:19:44 so I wonder if call site should say something about a JNI call Sep 27 21:19:46 i have no idea what you're talking about Sep 27 21:20:06 I should have said C++ mangler Sep 27 21:20:16 fraxinath: Java_dumphd_aacs_AACSKeys_getVersionString, is the classic signature for Java Native Interface (JNI) Sep 27 21:20:46 fraxinath: it is plain C abi, class name encoded in the symbol, params are included there as well, it is the result of javah (if one is lazy) Sep 27 21:20:55 fraxinath: can you post the code for AACSKeys.c or AACSKeys.cpp Sep 27 21:20:57 okay in other words... the label under which the linker is going to find the function? Sep 27 21:21:07 fraxinath: _ZGAN6dumphd4aacs8AACSKeys16getVersionStringEJPN4java4lang6StringEv, is a C++ symbol Sep 27 21:21:25 fraxinath: so when statically linking, GCJ appears to use CNI Sep 27 21:21:38 fraxinath: http://gcc.gnu.org/java/papers/cni/t1.html Sep 27 21:21:41 [22:55:48] http://pastebin.com/KRiqxH3C Sep 27 21:22:24 so you want -fjni Sep 27 21:23:04 -fjni Sep 27 21:23:05 With gcj there are two options for writing native methods: CNI and JNI . By default gcj assumes you are using CNI . If you are compiling a class with native methods, and these methods are implemented using JNI , then you must use "-fjni". This option causes gcj to generate stubs which will invoke the underlying JNI methods. Sep 27 21:23:33 # Sep 27 21:23:34 JNIEXPORT jstring JNICALL Java_dumphd_aacs_AACSKeys_getVersionString Sep 27 21:23:57 fraxinath: I need to see the call Sep 27 21:24:30 khem: the call is coming from Java code, all he did was declared the function as native Sep 27 21:24:36 -fjni is the answer. :) Sep 27 21:24:48 yeah right Sep 27 21:25:05 I would prefer him to use JNI instead of CNI Sep 27 21:25:27 hehe, he is unlikely interested in rewriting aackeys Sep 27 21:27:13 hmm -fjni will be slower as it will generate indirect calls but I think get it to compile and ship it Sep 27 21:29:02 damn xserver Sep 27 21:29:37 aye Sep 27 21:29:41 -fjni Sep 27 21:29:56 it's going with the compiler options in the gcj call? Sep 27 21:30:42 yeah Sep 27 21:31:16 no result :/ Sep 27 21:31:18 03Martin Jansa  07org.openembedded.dev * rfb3957c883 10openembedded.git/recipes/freesmartphone/zhone_git.bb: zhone: bump SRCREV Sep 27 21:31:57 http://pastebin.com/hRfbyfTU Sep 27 21:33:41 fraxinath: with the compiler Sep 27 21:33:50 fraxinath: you must have a header file where JNIEXPORT jstring JNICALL Java_dumphd_aacs_AACSKeys_getVersionString is declared Sep 27 21:34:03 fraxinath: remove AACCKeys.o, change the makefile Sep 27 21:34:08 khem: that is not true. :) Sep 27 21:34:27 zecke: I would generate a header file using javah Sep 27 21:34:34 and include that in AACCKeys.c Sep 27 21:34:51 so it gets right mangling when generating the .so Sep 27 21:34:53 ah sure, must recompile objects Sep 27 21:35:02 khem: AACCKeys.c (in that capitalisation) is actuall AACCKeys.java Sep 27 21:35:13 khem: there is a large amount of symbols though Sep 27 21:35:14 actually... and other typos.. Sep 27 21:35:28 hmm i c Sep 27 21:36:09 so, it has public native void getVersionString();... it is gcj that generates a CNI call for it, it should generate a jni one (-fjni) Sep 27 21:36:23 hrms how can start X without keyboard Sep 27 21:36:38 currently recompiling object files with -fjni Sep 27 21:37:30 WOOHOO Sep 27 21:40:34 n8 Sep 27 21:40:48 zecke: gn Sep 27 21:40:54 n8 zecke! Sep 27 21:41:02 http://pastebin.com/wpKHVp4a Sep 27 21:41:27 it's 6k+ lines of errors now Sep 27 21:41:48 fraxinath: make a clean build, you seem to mix things now Sep 27 21:42:01 i made clean Sep 27 21:42:25 fraxinath: well, then use -fcni for the ones that need it... Sep 27 21:42:38 e.g. what is bdvm? Sep 27 21:42:51 bluray disc virtal machine Sep 27 21:42:59 i think Sep 27 21:43:34 if you see stuff like... java::lang:: Sep 27 21:43:36 that is CNI Sep 27 21:43:47 if you see _Jv... that is JNI Sep 27 21:44:04 zecke i don't think it's gonna work that way... because AACSKeys.java is the one who imports the libaacskeys.so stuff and that one also gave errors when using -fjni Sep 27 21:44:48 hrms the ext2 fs isnt writeable Sep 27 21:45:02 there are mixed errors Sep 27 21:45:13 source/dumphd/aacs/AACSKeys.java:(.text+0x48): undefined reference to `java::lang::Object::Object()' Sep 27 21:45:17 source/dumphd/aacs/AACSKeys.java:(.text+0x130): undefined reference to `_Jv_AllocObjectNoFinalizer' Sep 27 21:45:22 well, remove -fjni from the link line? Sep 27 21:45:24 cni and jni Sep 27 21:45:31 ah good idea Sep 27 21:45:35 it's still there Sep 27 21:45:55 recompiling right now Sep 27 21:46:03 bbl Sep 27 21:46:15 woglinde_: add rw Sep 27 21:46:57 root=/dev/sda1 rw Sep 27 21:47:03 I have Sep 27 21:47:13 does it have udev working Sep 27 21:47:19 yes Sep 27 21:47:28 strange Sep 27 21:47:30 cat /proc/mounts Sep 27 21:47:33 'console=ttyAMA0 console=ttyS0 root=/dev/sda rw' -drive file=/devel/arm/oetmp-ang/deploy/uclibc/images/qemuarm/xorg-image-qemuarm.ext2 Sep 27 21:47:46 will try with ext3 now Sep 27 21:47:48 woglinde_: there may be no space left on it Sep 27 21:47:56 13 mb lef Sep 27 21:47:57 t Sep 27 21:48:01 woglinde_: k Sep 27 21:48:10 and says no space left Sep 27 21:48:22 woglinde_: you can try nfsboot Sep 27 21:48:31 *sigh* Sep 27 21:48:40 hm.. ext2 image is not drive image afaik Sep 27 21:48:48 this is filesystem image Sep 27 21:48:52 for qemu Sep 27 21:48:55 w/o fdisk partitioning Sep 27 21:51:43 khem: it didn't work either Sep 27 21:51:47 hehe.. DISTRO_PR.. Sep 27 21:52:04 woglinde_: use -hda /devel/arm/oetmp-ang/deploy/uclibc/images/qemuarm/xorg-image-qemuarm.ext2 Sep 27 21:52:07 does that help Sep 27 21:53:53 nope Sep 27 21:54:07 woglinde_: I run into same issue Sep 27 21:54:11 I started to use nfs Sep 27 21:54:17 hm qemu-convert Sep 27 21:54:31 yeah qcow2 would be an option Sep 27 21:59:04 khem we should build in the qemu-img command Sep 27 21:59:14 yeah thought so Sep 27 21:59:19 never got arounf doing it Sep 27 21:59:22 args Sep 27 21:59:24 if you do something Sep 27 21:59:25 same Sep 27 21:59:28 then please add Sep 27 21:59:33 maybee I am out if inodes? Sep 27 21:59:40 how big did u create Sep 27 21:59:56 oe standard Sep 27 22:05:45 khem, how would those suggested header files have to look like? Sep 27 22:07:12 03Khem Raj  07master * r21f9fffae6 10openembedded.git/ (5 files in 2 dirs): Sep 27 22:07:12 siteinfo.bbclass, site/: Add mips64/mips64el to siteinfo data Sep 27 22:07:12 Signed-off-by: Khem Raj Sep 27 22:07:12 03Khem Raj  07master * rd7482b9c43 10openembedded.git/conf/machine/ (include/tune-mips64.inc qemumips64.conf): Sep 27 22:07:12 conf/machine: Add qemumips64 machine configuration and tune file. Sep 27 22:07:13 Signed-off-by: Khem Raj Sep 27 22:07:13 03Khem Raj  07master * rb4ffa0f715 10openembedded.git/conf/distro/include/ (6 files): Sep 27 22:07:14 distros: Disable gnu-style hash for mips64/mips64el Sep 27 22:07:14 Signed-off-by: Khem Raj Sep 27 22:07:15 03Khem Raj  07master * r317011719c 10openembedded.git/ (3 files in 2 dirs): Sep 27 22:07:15 native.bbclass,package.bbclass,bitbake.conf: Make provisions for system libdir to be something else than '/lib' Sep 27 22:07:17 * Add base_libdir_native as target libdir could be different e.g. lib32 Sep 27 22:07:18 lib64 Sep 27 22:07:18 * Dont assume lib to be system libdir during packaging. Sep 27 22:07:19 Signed-off-by: Khem Raj Sep 27 22:07:19 Acked-by: Chris Larson Sep 27 22:07:20 03Khem Raj  07master * rc03a23a14c 10openembedded.git/classes/kernel-arch.bbclass: Sep 27 22:07:21 kernel-arch.bbclass: Add logic to account for mips64/mips64el in map_kernel_arch Sep 27 22:07:22 Signed-off-by: Khem Raj Sep 27 22:07:22 03Khem Raj  07master * rab0ea33492 10openembedded.git/recipes/netbase/netbase/ (qemumips64/interfaces qemumips64el/interfaces): Sep 27 22:08:12 03Khem Raj  07master * rfb354c78a4 10openembedded.git/classes/insane.bbclass: Sep 27 22:08:12 insane.bbclass: Recognise mips64 and mips64el Sep 27 22:08:12 Signed-off-by: Khem Raj Sep 27 22:08:12 (19 lines omitted) Sep 27 22:13:56 03Martin Jansa  07org.openembedded.dev * rdfbdd45c7e 10openembedded.git/conf/distro/shr.conf: Sep 27 22:13:56 SHR: drop preferred version for eglibc and gcc Sep 27 22:13:56 * same versions are now in sane-toolchain.inc Sep 27 22:13:56 Signed-off-by: Martin Jansa Sep 27 22:20:58 fraxinath: if you do javah .java Sep 27 22:21:05 it will generate the header file Sep 27 22:25:29 cool okay i'll try that tomorrow! Sep 27 22:25:31 n8 Sep 27 22:26:12 fraxinath: I think the problem is solvable Sep 27 22:26:15 n8 Sep 27 22:28:28 hey khem, do you have a moment for solving a doubt? Sep 27 22:29:13 sure Sep 27 22:29:26 this time is about klibc.so symlink Sep 27 22:29:30 ok Sep 27 22:29:39 well, klibc uses hash, no soname Sep 27 22:29:55 so how useful is the link? Sep 27 22:30:46 relative question: I suppose the symlink should be packaged with the klibc-hash.so, isn't? Sep 27 22:30:58 (if needed) Sep 27 22:31:28 now, I put the klibc.so in the -dev package because the QA was complaining Sep 27 22:32:13 good nite Sep 27 22:32:20 good night Sep 27 22:32:44 ant_: .so should not be in -dev Sep 27 22:32:48 as I understand it, the klibc utilities are linking with the 'true' klibc.so' so they don't complain about that Sep 27 22:33:03 (the shared utils) Sep 27 22:33:45 khem, ok, then should I skip the QA for the klibc package? Sep 27 22:35:31 INSANE_SKIP_${PN} = True Sep 27 22:35:42 no no Sep 27 22:37:19 hm.. I set again FILES_${PN} = "${base_libdir}/klibc-*.so ${base_libdir}/klibc.so" Sep 27 22:39:28 then I get Sep 27 22:39:30 ERROR: QA Issue with klibc: non -dev package contains symlink .so: klibc path '/work/c7x0-angstrom-linux-gnueabi/klibc-1.5.20-r1.0/packages-split/klibc/lib/klibc.so' Sep 27 22:45:50 khem, actually INSANE_SKIPS shuts down the QA, indeed "NOTE: package klibc skipped" Sep 27 22:47:14 ant_: whats the symlink about Sep 27 22:47:50 is created in do_install Sep 27 22:47:52 (cd ${D}${base_libdir}; ln -s klibc-*.so klibc.so) Sep 27 22:48:35 supposedly is not installed by klibc sources Sep 27 22:48:43 let me check Sep 27 22:49:05 ok you can package it into klibc package itself Sep 27 22:51:39 I see the similar QA about .so for other packages (libgtkstylus comes to mind) Sep 27 22:52:02 iirc in some rare case the QA is bogus Sep 27 23:07:47 ant_: yeah sometimes .so are linker stubs now a days Sep 27 23:08:06 so I wonder if we should check for .so or for .so.x Sep 27 23:08:16 because .so.x in a -dev package is a problem Sep 27 23:08:21 but .so could be anything Sep 27 23:08:32 like symlink Sep 27 23:08:41 or linker script Sep 27 23:11:31 64698 of 80462. Getting closer. Sep 27 23:18:16 03Tom Rini  07org.openembedded.dev * r5c3896a2a3 10openembedded.git/recipes/tzdata/tzdata_2010l.bb: Sep 27 23:18:16 tzdata: bump version to 2010m Sep 27 23:18:16 Signed-off-by: Tom Rini Sep 27 23:18:24 03Tom Rini  07org.openembedded.dev * r4838826097 10openembedded.git/recipes/tzcode/tzcode-native_2010l.bb: Sep 27 23:18:24 tzcode-native: bump version to 2010m Sep 27 23:18:24 Signed-off-by: Tom Rini Sep 27 23:29:26 thx khem. I'll fix it tomorrow Sep 27 23:29:31 good nite Sep 27 23:30:53 hi, all Sep 27 23:31:43 I am continuosly getting this: "unknown package: locale-base-en-use", while building x11-gpe-image. any resolution? Sep 27 23:42:39 anybody around? Sep 27 23:43:37 03Nobuhiro Iwamatsu  07master * r39cf877a8a 10openembedded.git/ (7 files in 5 dirs): Sep 27 23:43:37 machine/sh7785lcr: Add new machine for Renesas SH7785 reference board Sep 27 23:43:37 * Machine config for Renesas SH7785 reference board. Sep 27 23:43:37 * Add kernel config for 2.6.32 and 35. Sep 27 23:43:37 * Add tune file for sh4a. Sep 27 23:43:37 Signed-off-by: Nobuhiro Iwamatsu Sep 27 23:43:37 Signed-off-by: Khem Raj Sep 27 23:47:13 03Graham Gower  07master * r4e0a2ca681 10openembedded.git/recipes/mozilla/ (7 files in 3 dirs): (log message trimmed) Sep 27 23:47:13 firefox_3.6.8.bb: Add mipsel support. Sep 27 23:47:13 Patches taken from Gentoo and Debian. Sep 27 23:47:13 Runs almost acceptably on my 350mhz/128mb mipsel, after a lengthy Sep 27 23:47:13 startup delay. Should work for big endian mips without much more Sep 27 23:47:13 work, but I can't test. Sep 27 23:47:13 Signed-off-by: Graham Gower Sep 27 23:49:44 03Yann Dirson  07master * red5388f2fc 10openembedded.git/recipes/gnugo/ (5 files in 2 dirs): Sep 27 23:49:44 gnugo_3.8: Add new recipe. Sep 27 23:49:44 Cross-compiling patches included. Sep 27 23:49:44 Signed-off-by: Yann Dirson Sep 27 23:49:44 Signed-off-by: Khem Raj Sep 27 23:49:49 IgorK, it works for me. Can you paste your local.conf somewhere? Sep 27 23:50:24 grg: sure, hold on a sec.... Sep 27 23:50:55 03Enrico Scholz  07master * reb30ef6d8e 10openembedded.git/lib/oe/path.py: (log message trimmed) Sep 27 23:50:55 oe.path: added 'recurse' argument to remove() Sep 27 23:50:55 This makes it possible to specify whether the equivalent of 'rm -rf' or Sep 27 23:50:55 only this of 'rm -f' is wanted. Due to backward compatibility it Sep 27 23:50:55 defaults to the recursive variant. Sep 27 23:50:56 Signed-off-by: Enrico Scholz Sep 27 23:50:57 Acked-by: Chris Larson Sep 27 23:52:57 03Fahad Usman  07master * re701b369d2 10openembedded.git/recipes/python/ (python-pyrex-native_0.9.8.5.bb python-pyrex_0.9.8.5.bb): (log message trimmed) Sep 27 23:52:57 python-pyrex_0.9.8.5.bb: merged native and non-native recipes Sep 27 23:52:57 * removed the python-pyrex-native_0.9.8.5.bb file Sep 27 23:52:57 * added DEPENDS_pn-python-pyrex-native = "python-native" Sep 27 23:52:57 * added BBCLASSEXTEND = "native" Sep 27 23:52:58 * added NATIVE_INSTALL_WORKS = "1" Sep 27 23:52:59 * bunped PR Sep 27 23:53:40 03Fahad Usman  07master * r073f3eb34b 10openembedded.git/recipes/python/ (python-epydoc-native_3.0.1.bb python-epydoc_3.0.1.bb): (log message trimmed) Sep 27 23:53:40 python-epydoc_3.0.1.bb: merged native and non-native recipes Sep 27 23:53:40 * removed the python-epydoc-native_3.0.1.bb file Sep 27 23:53:40 * added DEPENDS_pn-python-epydoc-native = "python-native" and added DEPENDS_pn-python-epydoc = "python" Sep 27 23:53:40 * added BBCLASSEXTEND = "native" Sep 27 23:53:40 * added NATIVE_INSTALL_WORKS = "1" Sep 27 23:53:41 * bunped PR Sep 27 23:55:23 grg: http://pastebin.com/guGSGbPC Sep 27 23:56:54 IgorK, I'm just guessing, but i suspect this line is responsible: GLIBC_GENERATE_LOCALES = "en_GB.UTF-8" Sep 27 23:57:31 if you only build the en_GB locale and the image is looking for en_US, then it wont find it. Sep 27 23:57:39 really, the image should be fixed Sep 27 23:57:58 (or whatever is responsible for the en_US dependency) Sep 27 23:58:57 03Alex Bennee  07master * rca59f2509b 10openembedded.git/recipes/recode/recode_3.6.bb: Sep 27 23:58:57 recode: Fix configure step. Sep 27 23:58:57 Use += instead of _append to CC/CXX otherwise configure fails Sep 27 23:58:57 Signed-off-by: Alex Bennee Sep 27 23:58:57 Signed-off-by: Khem Raj Sep 27 23:58:57 xxxxxxxxxxxxx Sep 28 00:05:06 03Noor Ahsan  07master * r6e15655f69 10openembedded.git/recipes/sip/sip-native_4.7.9.bb: Sep 28 00:05:06 sip-native: Run oe-stylize.py script on the recipe and modified the the recipe accordingly Sep 28 00:05:06 * Run the oe-stylize.py script on the recipe and modified the recipe according to the output of the script. Sep 28 00:05:06 Signed-off-by: Noor Ahsan Sep 28 00:05:06 Signed-off-by: Khem Raj Sep 28 00:05:25 03Noor Ahsan  07master * r37bf830772 10openembedded.git/recipes/sip/sip-native_4.7.9.bb: (log message trimmed) Sep 28 00:05:25 sip-native: Convert to new style staging, remove do_stage() Sep 28 00:05:25 * Convert do_stage to do_install Sep 28 00:05:25 * Inherit recipe with python-dir for PYTHON_SITEPACKAGES_DIR variable Sep 28 00:05:26 * Replace ${STAGING_BINDIR_NATIVE} with ${D}${bindir} Sep 28 00:05:26 * Replace ${STAGING_PYDIR}/site-packages with ${D}${PYTHON_SITEPACKAGES_DIR} Sep 28 00:05:27 * Set NATIVE_INSTALL_WORKS = "1" Sep 28 00:05:28 03Noor Ahsan  07master * r1c8fbc364a 10openembedded.git/recipes/libsdl/libsdl-native.inc: Sep 28 00:05:28 libsdl-native: Convert to new style staging, remove do_stage Sep 28 00:05:29 * Remove do_stage () Sep 28 00:05:29 * Replace tabs with 8 spaces in do_configure Sep 28 00:05:30 * Bump INC_PR = "r3" Sep 28 00:05:30 Signed-off-by: Noor Ahsan Sep 28 00:05:31 Signed-off-by: Khem Raj Sep 28 00:06:46 grg: ok, so what should I do? comment out this line? Sep 28 00:19:18 IgorK, you could do that, or add an entry for en_US. Or if you're feeling adventurous, you could track down why there is a dependency upon the en_US locale. Sep 28 00:29:48 * kergoth mulls over how to revamp bb.fetch a bit Sep 28 00:34:21 hrm.. urldata needs to go, and we need a way to specify preference about which fetcher to use when multiple exist for a given scheme.. Sep 28 02:55:51 kergoth: you mean SRC_URI should also get some priority **** ENDING LOGGING AT Tue Sep 28 02:59:57 2010