**** BEGIN LOGGING AT Tue Oct 19 02:59:57 2010 Oct 19 03:06:42 here's something you guys can probably actually help with Oct 19 03:07:01 i'm on the last build step, which is populating the toolchain Oct 19 03:07:15 and it says that it can't satisfy dependencies for libsdl-mixer, which compiled and got packaged Oct 19 03:08:12 there's also a worrying number of "unsatisfied recommendation for" messages Oct 19 03:56:48 hello guys Oct 19 05:23:31 is there some bitbake syntax to perform multiple tasks? e.g. bitbake foo -c clean,do_compile Oct 19 06:11:28 grg: i've been trying to find that out myself to little success Oct 19 06:11:34 wish there was though Oct 19 06:11:52 vorporeal, a quick look at the bitbake code reveals it doesn't support such a thing Oct 19 06:11:57 :( Oct 19 06:12:57 i suspect it could be extended to support a comma separated task syntax, but i'm not sure how deep into bitbake it would need to be propagated... Oct 19 06:16:56 hi all i m getting invalid md5 error while building the overo image ..can anybody tell me how to resolve this issue Oct 19 06:30:04 gm Oct 19 06:30:09 holy shit angstrom's servers are sloooooooooow lately Oct 19 06:36:59 aarti: what exact error are u getting Oct 19 06:37:57 aditya_1010: because they hit 1TB traffic already Oct 19 06:38:05 ah Oct 19 06:38:15 well apparently the beagle-xm comes with a tiny partitioned card Oct 19 06:38:27 so I didn't have enough space for mono anyway :( Oct 19 06:38:37 need to do an OE build I guess Oct 19 06:58:25 khem, my error is : http://pastebin.com/SGLQQFbE Oct 19 07:01:44 aloha Oct 19 07:04:15 aarti: rm /home/sitpl/openembedded_sources/libsamplerate-0.1.7.tar.gz* Oct 19 07:04:26 so you'll download a new copy Oct 19 07:04:59 (and providing more info is more likely to generate an answer) Oct 19 07:06:05 i have tried this but no solution Oct 19 07:06:44 ah ok, then you have an old recipe but with the new checksups Oct 19 07:06:47 checksums Oct 19 07:06:54 do a pull of git head Oct 19 07:07:00 or modify the recipe locally Oct 19 07:07:05 ok Oct 19 07:07:10 let me chk Oct 19 07:08:45 you want this: Oct 19 07:08:46 SRC_URI[md5sum] = "6731a81cb0c622c483b28c0d7f90867d" Oct 19 07:08:46 SRC_URI[sha256sum] = "78ed5d9ff1bf162c4a078f6a3e7432a537dd2f22dc58872b081fb01156027fcc" Oct 19 07:13:02 anyone have any ideas as to why glibc would try to use arm-angstrom-xxxx instead of arm-none-xxxx for cross-compile stuff? Oct 19 07:13:10 good morning Oct 19 07:13:11 this is oe-dev Oct 19 07:13:16 morning Oct 19 07:13:37 whoa, its 3am.... crap Oct 19 07:16:11 no it's not; it is 9 am :-) Oct 19 07:22:32 european bastards Oct 19 07:24:30 ty :-( Oct 19 07:24:46 you and your fine wine... Oct 19 07:24:48 and women Oct 19 07:24:51 aditya_1010 is now in my ignore list Oct 19 07:25:06 pssh Oct 19 07:25:29 cleared out me tmp dir Oct 19 07:25:35 hopefully a rebuild will help Oct 19 07:25:50 guess I'll find out when I wake up in about 3.5 hours Oct 19 07:26:09 niiigh Oct 19 07:26:10 t Oct 19 07:35:04 aarti: ok the problem is that md5 we have dont match with the tar you download Oct 19 07:39:39 aarti: it seems you are missing 40e5976b29d5ae95a5c483e87c53ad23e518c7c5 Oct 19 07:39:48 cherry pick that commit and you will be fine Oct 19 07:40:06 nap time Oct 19 07:40:14 good night Oct 19 07:40:16 nite khem Oct 19 07:41:21 XorA|gone: http://www.youtube.com/watch?v=fOuP_w0FZi0 Oct 19 07:43:19 anyone wants to add this glibc patch to OE: http://sourceware.org/ml/libc-hacker/2010-10/msg00007.html issue: http://seclists.org/fulldisclosure/2010/Oct/257 Oct 19 07:47:36 03Koen Kooi  07org.openembedded.dev * rcc8bc6d248 10openembedded.git/classes/image.bbclass: image bbclass: fix install_linguas usage on shared buildservers Oct 19 09:04:17 JaMa|Wrk: good work on the openmoko stuff Oct 19 09:11:03 trying to resend patch 09 Oct 19 09:11:16 it's 3.8MB.. Oct 19 09:16:28 yuk (saw the gumstix kernel clenaup was also big, couldn't really do a reply on it) Oct 19 09:18:16 -rw-r--r-- 1 bitbake bitbake 1.7M Oct 19 10:17 logo_linux_clut224.ppm Oct 19 09:18:23 that's what makes it big :/ Oct 19 09:18:27 duh Oct 19 09:18:39 included twice in diff (delete and create) Oct 19 09:20:07 ofc Oct 19 09:49:08 hi folks Oct 19 09:50:53 hi blindman Oct 19 11:07:46 I suffer a collision when I try to compile qt4-embedded-gles. I get this error : ERROR: Multiple .bb files are due to be built which each provide qt4-embedded, and it seems to compile the non-gles variant Oct 19 11:30:33 florian, ping Oct 19 11:36:02 what does mean "maximum recursion depth exceeded while calling a Python object while parsing" ?? Oct 19 11:47:43 XorA|gone: http://www.youtube.com/watch?v=squxkHIaIdY Oct 19 11:51:55 :D Oct 19 12:00:25 Crofton|work: pong Oct 19 12:04:25 hi Oct 19 12:04:35 do you know how the data to ltg was split? Oct 19 12:24:16 I tried adding my own /etc/fstab to get data=ordered instead of data=writeback on ext3. That worked well, but someone/something makes my root still mount with writeback. Oct 19 12:27:04 Anyone know what that someone/something is? Oct 19 12:28:09 is the 1 TB limit that angstrom hit also affecting the wiki ? Oct 19 12:28:09 it is sloooow Oct 19 12:28:23 eFfeM_work: ltg is one machine... Oct 19 12:28:59 aaarg, (actually thought angstrom was on a different machine) Oct 19 12:30:08 i find it not really acceptable that a distro can greatly reduce core services (already waiting for 2 mins for a wiki page) Oct 19 12:30:40 hrw: hehe Oct 19 12:30:52 eFfeM_work, no Oct 19 12:31:23 eFfeM_work, the "problem" is being successful Oct 19 12:31:29 and the wiki is not on ltg Oct 19 12:31:47 so I have no idea what point you are trying to make Oct 19 12:32:09 the openembedded wiki is at OSUOSL Oct 19 12:32:27 and the Angstroms feeds and some other stuff are on the linuxtogo machine Oct 19 12:32:38 Crofton|work: I didn't say the wiki was on ltg; I asked '(02:24:13 PM) eFfeM_work: is the 1 TB limit that angstrom hit also affecting the wiki ?" Oct 19 12:32:51 then hrw said "(02:24:13 PM) eFfeM_work: is the 1 TB limit that angstrom hit also affecting the wiki ?" Oct 19 12:32:54 oops Oct 19 12:33:00 "(02:24:27 PM) hrw: eFfeM_work: ltg is one machine..." Oct 19 12:33:11 i find it not really acceptable that a distro can greatly reduce core services (already waiting for 2 mins for a wiki page) Oct 19 12:33:16 which is not happening Oct 19 12:33:18 i'm not trying to make a point, i noticed the wiki was slooow Oct 19 12:33:49 Crofton|work: then hrw misinformed me (or I misinterpreted his reply) Oct 19 12:34:03 well Oct 19 12:34:10 I think he said something vague :) Oct 19 12:34:24 :-) Oct 19 12:34:30 the sky is blue Oct 19 12:34:34 ltg is one machine Oct 19 12:34:38 :) Oct 19 12:35:09 Hopefully Florian has some idea what serveices are using the bw Oct 19 12:35:15 maybe it is the ltg pr0n collection Oct 19 12:35:21 true (but I considered that it was an answer to my question) Oct 19 12:35:25 Crofton|work: url ? Oct 19 12:35:31 :-) Oct 19 12:35:44 actually should have used :P Oct 19 12:36:22 its my lego pr0n collection :-D Oct 19 12:37:24 hm, they didn't have that kind of lego puppets when I was young :-( Oct 19 12:38:19 I suspect narcisus downloads Oct 19 12:38:35 Crofton|work: It was the webserver... the machine is back to 100MBit Oct 19 12:39:04 XorA|gone: not exactly, there seem to be some large downloads which are very interesting Oct 19 12:39:24 florian: someone warezed the server? Oct 19 12:39:40 heh Oct 19 12:39:40 florian: or bad mirroring attemopts Oct 19 12:39:59 well some people might want to treat it like this ;) Oct 19 12:40:24 the highlight seems to be a Ubuntu image for some modern HTC cellphone Oct 19 12:41:00 ~700mb Oct 19 12:41:02 ah Oct 19 12:41:10 ouch Oct 19 12:41:30 we seem to have a crazy host trying to download the file 100 times ;) Oct 19 12:41:42 thanks for figuring this out Oct 19 12:41:53 it seemed like an unusual spike Oct 19 12:42:21 ubutnu on a cell phone ewwwww Oct 19 12:42:29 heh Oct 19 12:42:36 ubuntu everywhere Oct 19 12:42:57 * XorA|gone has decided the best OS for cell phones currently available is Windows Mobile :-D Oct 19 12:43:06 * Crofton|work beats XorA|gone Oct 19 12:43:24 hey, it can reliable make phone calls unlike Android and IOS :-D Oct 19 12:44:14 florian: set QOS for files >10MB? Oct 19 12:44:25 * XorA|gone hopes WIndows Mobile 6.5 keeps in development in parrallel to 7 Oct 19 12:44:52 hrw: fail2ban if same IP downloads file >2 Oct 19 12:45:12 XorA|gone: no, as it will retry Oct 19 12:45:16 Crofton|work: yes indeed... the provider statistics nils checked toda say something about 6TB in June Oct 19 12:45:21 XorA|gone: give it 2bps speed instead Oct 19 12:46:27 hrw: thought about this already but it might take some time... my experience with qos is quite limited Oct 19 12:46:41 03Frans Meulenbroeks  07org.openembedded.dev * r40c4e7823d 10openembedded.git/conf/machine/include/nios2.inc: Oct 19 12:46:41 nios2.inc: added missing definitions for TARGET_CC_ARCH BASE_PACKAGE_ARCH FEED_ARCH Oct 19 12:46:41 Signed-off-by: Frans Meulenbroeks Oct 19 13:14:34 gm Oct 19 13:29:37 hi likewise, you're early (about 7 hrs I'd say) Oct 19 13:32:23 btw anyone an idea who should deliver os2.h for os2stubs.c in libX11) Oct 19 13:44:54 IBM? Oct 19 13:57:33 03Henning Heinold  07org.openembedded.dev * rbcd69fc319 10openembedded.git/recipes/dietlibc/dietlibc.inc: Oct 19 13:57:33 dietlibc: move do_stage to do_install Oct 19 13:57:33 * bump INC_PR Oct 19 13:59:46 now native compilation is broke Oct 19 13:59:46 hm what was the patchwork commad Oct 19 13:59:52 huhhhu? Oct 19 13:59:53 wtf Oct 19 14:00:09 I have never seen dev be this bad before Oct 19 14:00:10 introspections tuff? Oct 19 14:00:16 for such a length of time Oct 19 14:07:06 woglinde: hi Oct 19 14:07:12 03Koen Kooi  07org.openembedded.dev * ra3c2ecd5d8 10openembedded.git/classes/module.bbclass: module bbclass: add KERNEL_DIR for ubunutu style modules Oct 19 14:07:13 03Koen Kooi  07org.openembedded.dev * r91dfafe516 10openembedded.git/recipes/wl12xx/tiwlan-wl1271_0.24.9.bb: Oct 19 14:07:13 tiwlan-wl1271: add kernel drivers and loaders for the wifi portion of a wl1271 chipset Oct 19 14:07:13 * based on the ubuntu makefiles, so still a messy wrt OE style Oct 19 14:07:36 woglinde: you mean contrib/patchwork/pw-am.sh ? Oct 19 14:08:12 yeah Oct 19 14:08:21 found it already Oct 19 14:08:23 you effem Oct 19 14:08:29 args yo effem Oct 19 14:08:33 xmlrpc now works? Oct 19 14:09:04 yes with the version as committed Oct 19 14:09:10 cool Oct 19 14:12:33 eFfeM_work: yeah, too early, isn't it? :-) Oct 19 14:21:48 khem, ping Oct 19 14:21:56 * Crofton|work knows it is early .... Oct 19 14:22:34 03Andreas Oberritter  07org.openembedded.dev * r29ed2a74ff 10openembedded.git/recipes/libsdl/libsdl-directfb_1.2.14.bb: (log message trimmed) Oct 19 14:22:34 libsdl-directfb: update to 1.2.14 Oct 19 14:22:34 Changes since v1: Oct 19 14:22:34 * removed patches. Oct 19 14:22:34 * use autoreconf to update libtool etc. Oct 19 14:22:35 * inherit lib_package Oct 19 14:22:35 * dropped some unused configure options. Oct 19 14:22:37 03Andreas Oberritter  07org.openembedded.dev * r52fad6de31 10openembedded.git/recipes/directfb/directfb_1.4.6.bb: Oct 19 14:22:37 directfb-1.4.6: fix packaging Oct 19 14:22:37 * Files were searched in the wrong directory, because $RV was wrong. Oct 19 14:22:38 Signed-off-by: Andreas Oberritter Oct 19 14:22:38 Acked-by:: Henning Heinold Oct 19 14:24:15 http://pastebin.com/0SDDkgxz Oct 19 14:24:54 hm stdio.h comes from libc Oct 19 14:25:24 crofton sdk? Oct 19 14:25:35 native sdk Oct 19 14:25:42 woglinde: will try to do the recipe that uses xmlrpc tonight Oct 19 14:25:50 /usr/include/stdio.h exists Oct 19 14:26:02 /usr/lib/gcc/arm-angstrom-linux-gnueabi/4.5.2/include Oct 19 14:28:21 I fear khem may have been too agressive removing host includes :) Oct 19 15:03:59 03Chase Maupin  07org.openembedded.dev * r019cddea9e 10openembedded.git/recipes/qt4/ (qt4-embedded.inc qt4-x11-free.inc qt4.inc): (log message trimmed) Oct 19 15:03:59 qt4: replace underscore with dash in library name Oct 19 15:03:59 * For libraries that are added to the QT_LIB_NAMES or Oct 19 15:03:59 QT_EXTRA_LIBS variables replace any "_" characters in the Oct 19 15:04:00 library name with "-" for use in the package name. Oct 19 15:04:00 * The actual library file names are left alone. Oct 19 15:04:01 * Bump the PR for packages using qt4.inc Oct 19 15:04:07 03Koen Kooi  07org.openembedded.dev * r15147ebda1 10openembedded.git/recipes/gnome/gnome-games_2.30.0.bb: gnome-games: update 2.30.x to 2.30.2 and add 2.32.0 Oct 19 15:07:31 heyho Oct 19 15:24:41 urgs Oct 19 15:24:43 /devel/arm/oetmp-ang/work/armv7a-angstrom-linux-uclibceabi/jpeg-8b-r1/image/usr/lib/sedKBMV4P Oct 19 15:24:47 what the hell Oct 19 15:27:16 sedWTF Oct 19 15:27:45 yeah Oct 19 15:28:05 but I will later look into Oct 19 15:34:36 Crofton|work: whats up Oct 19 15:36:57 you see my email Oct 19 15:37:08 can't compile hello world on an omap3 atm Oct 19 15:37:14 can't find stdio.h .... Oct 19 15:50:36 woglinde: btw wrt xmlrpc-c: I have not found what gnu-configize does or where it is defined Oct 19 16:08:59 hmm task-proper-tools installs pump Oct 19 16:09:08 which seems to cause issues with avahi Oct 19 16:24:46 who is using pump? Oct 19 16:24:50 anymore Oct 19 16:25:33 dunno Oct 19 16:25:38 but it is messing with me Oct 19 16:25:40 testing now Oct 19 16:41:19 jo kergoth Oct 19 16:44:20 Hi. trying to add gst-plugin-mad, and am getting the error during the the opkg_install_cmd: ... etc/device_table is different from the conffile in the new package. The new conffile will be placed at... Any sage hints as to how to correct? Oct 19 16:45:38 error? Oct 19 16:45:41 its a hint Oct 19 16:45:52 you should diff the files Oct 19 16:46:12 and why the hell gst-plugin-mad is installing etc/device_table Oct 19 16:46:26 (that was sort of my question...) Oct 19 16:46:50 proabably you didnt seeh the real error Oct 19 16:48:30 | Collected errors: Oct 19 16:48:30 | * opkg_install_cmd: Cannot install package gst-plugin-mad. Oct 19 16:48:30 | * resolve_conffiles: Existing conffile /export/space/mikew/oe/build/tmp/rootfs/mityomap-base/etc/device_table is different from the conffile in the new package. The new conffile will be placed at /export/space/mikew/oe/build/tmp/rootfs/mityomap-base/etc/device_table-opkg. Oct 19 16:48:44 NOTE: Task failed: /export/space/mikew/oe/build/tmp/work/mityomapl138-angstrom-linux-gnueabi/mityomap-base-1.0-r0/temp/log.do_rootfs.10251 Oct 19 16:48:47 OK. I'll keep looking. Thanks. Oct 19 16:49:15 03Petr Štetiar  07org.openembedded.dev * rd875046057 10openembedded.git/classes/oestats-client.bbclass: (log message trimmed) Oct 19 16:49:15 oestats-client.bbclass: send USERDISTRO value Oct 19 16:49:15 With this patch it's now possible to see the correct Angstrom distro version Oct 19 16:49:15 in the Tinderbox. Now the distribution value is always 'angstrom', with this Oct 19 16:49:15 patch it would be more detailed version 'angstrom-2008.1' for example. Oct 19 16:49:15 Signed-off-by: Petr Štetiar Oct 19 16:49:16 Acked-by: Frans Meulenbroeks Oct 19 17:26:06 wb khem Oct 19 17:39:19 Crofton|work: hi Oct 19 17:40:27 PG&E are installing some new equipment so they needed to turn off the power for few minutes Oct 19 17:40:30 you see my note about native compile being broke? Oct 19 17:40:34 lovely Oct 19 17:40:45 hopefully they do not hit a gas line ... Oct 19 17:40:53 heh Oct 19 17:41:00 who knowns Oct 19 17:41:07 I dont smell rotten eggs Oct 19 17:42:16 Crofton|work: ok so whats the problem with native gcc Oct 19 17:42:32 basially Oct 19 17:42:41 if you try to build hello world Oct 19 17:42:48 it can't find stdio.h Oct 19 17:43:06 hmm Oct 19 17:43:07 unless you do gcc -o hello hello.c -I/usr/include Oct 19 17:43:16 thats bad Oct 19 17:43:40 ok paste me the output of gcc -v hello.c Oct 19 17:44:24 I think that is in my email :) Oct 19 17:44:43 ah email I havent checked that Oct 19 17:44:49 now email is like snail mail :) Oct 19 17:44:57 people are connected 24/7 Oct 19 17:45:06 heh Oct 19 17:45:12 working on gettign it for you Oct 19 17:45:16 thru facebook twitter irc etc. Oct 19 17:45:27 I am assuming another 5 yw\ears people will forget email Oct 19 17:45:50 brb Oct 19 17:46:00 http://pastebin.com/ABu00AQE Oct 19 17:54:39 03Philip Balister  07org.openembedded.dev * r4dbe8edc63 10openembedded.git/recipes/vim/ (files/vimrc vim_7.2.bb): vim: Add new package vim-vimrc that installs a vimrc that makes me happy. Oct 19 17:54:45 03Philip Balister  07org.openembedded.dev * r43116ef15a 10openembedded.git/recipes/gnuradio/gnuradio.inc: gnuradio: Drop DEPENDS on sdcc now that we use uhd for usrp interfaces/ Oct 19 17:54:45 03Philip Balister  07org.openembedded.dev * r89814f2a9f 10openembedded.git/recipes/tasks/task-sdk-gnuradio-native.bb: task-sdk-gnuradio-native: Stop build failures due to missing packages. Oct 19 17:55:31 Crofton|work: two suspects Oct 19 17:55:43 :) Oct 19 17:55:51 search for the guilty! Oct 19 17:55:52 with-sysroot without a value Oct 19 17:56:37 khem remember, this is the most basic c ompilation :) Oct 19 17:56:43 on the target hw Oct 19 17:56:48 and with-local-prefix Oct 19 17:57:27 how do we fix this? Oct 19 17:57:28 Crofton|work: ok add --sysroot=/ option and recompile the hello Oct 19 17:57:39 I think I know to fix it Oct 19 17:57:55 but let me make sure the problem symptom Oct 19 17:58:23 root@usrp-embedded:~# gcc --sysroot=/ -o hello hello.c Oct 19 17:58:23 root@usrp-embedded:~# ./hello Oct 19 17:58:23 hello, world Oct 19 17:58:30 :) Oct 19 17:59:47 ok I think I know the problem Oct 19 18:01:00 # Oct 19 18:01:00 ignoring nonexistent directory "/usr/arm-angstrom-linux-gnueabi/sys-root/usr/loc Oct 19 18:01:00 # Oct 19 18:01:03 al/include" Oct 19 18:01:19 that mean sysroot screwed it Oct 19 18:01:31 heh Oct 19 18:18:47 I can't seem to find documentation on PACKAGES_DYNAMIC nor really must about where its used in OE/bitbake - can anyone explain it? guessing it auto-generates packages based on pattern matching of things in images/ ? Oct 19 18:24:22 Do someone has samba 3.5.x packaged? Oct 19 18:26:45 tharvey: PACKAGES_DYNAMIC is something you dont want to know anyway :) Oct 19 18:28:23 re Oct 19 18:29:12 khem is it okay for our oe uclibc if I steal fts.h from glibc? Oct 19 18:29:35 or is the license than incompatible Oct 19 18:29:59 * kergoth_ hacks on autoconf/automake relocation Oct 19 18:30:21 tharvey: it's for situations where you don't know exactly what you're packaging until its built Oct 19 18:30:32 tharvey: for example, locale packages, whose names are determined by what locales actually exist Oct 19 18:30:47 bitbake has to be told about these things, so it knows that it can satisfy those runtime dependencies Oct 19 18:30:53 hence the glob in the variable Oct 19 18:31:01 e.g. PACKAGES_DYNAMIC = "foo-bar-*" Oct 19 18:31:22 woglinde: its ok. Oct 19 18:31:29 but justify it Oct 19 18:31:30 okay Oct 19 18:31:33 hm? Oct 19 18:31:48 * kergoth_ works on a new version of his autotools branch Oct 19 18:31:48 woglinde: btw when you are at uclibc can you update it to latest git Oct 19 18:31:53 ah... ok, that makes sense then why util-linux-ng.inc has PACKAGES_DYNAMIC has libuuid* - new versions of util-linux-ng provide it and older dont - thx Oct 19 18:32:06 khem tse Oct 19 18:32:10 thats your job Oct 19 18:32:17 woglinde: heh Oct 19 18:32:18 khem you could update chromium Oct 19 18:32:25 that would be cool Oct 19 18:32:55 what do you guys think about splitting autotools.bbclass, to separate autoreconf from gnu-config + oe_runconf? Oct 19 18:32:59 khem do I can safely use the latest git? Oct 19 18:33:03 i did a branch while at montavista which created autotools_base Oct 19 18:33:08 woglinde: yes Oct 19 18:33:16 if you inherit that instead of autotools, you get just gnu-configize + oe_runconf, no need to do it manually Oct 19 18:33:16 kergoth whats the reason? Oct 19 18:33:22 hm Oct 19 18:33:25 woglinde: if not then flag it and we need to fix it in uclibc upstream Oct 19 18:33:46 kergoth how many recipe suffer from this? Oct 19 18:33:51 ideally we'd never do this, but we don't live in an ideal world :) Oct 19 18:33:54 besides xmlrpc-c Oct 19 18:34:07 plus there are legit cases to do it, in the recipes which are built before our autotools have been built Oct 19 18:34:09 kergoth_: you want to separate out running the config and recreating and running right ? Oct 19 18:34:11 recursive dep issue Oct 19 18:34:22 khem: yeah, separate autoreconf from oe_runconf Oct 19 18:34:31 kergoth_: thats a wise thing Oct 19 18:34:38 kergoth btw. it's a shame I need to workaround autoheader call within autoreconf with AUTOHEADER = "" Oct 19 18:34:53 if I dont want it Oct 19 18:35:01 you don't have to, use --exclude Oct 19 18:35:11 we added the option so you could exclude specific steps from the autoreconf Oct 19 18:35:12 kergoth_: and when auteconfing we should provide options Oct 19 18:35:14 autoheader is one of those Oct 19 18:35:19 which tools to run Oct 19 18:35:27 you cannt Oct 19 18:35:32 some packages are lame about using autoconf Oct 19 18:35:33 autoreconf has no option Oct 19 18:35:36 woglinde: huh? Oct 19 18:35:42 of course it does. we patch it in. Oct 19 18:35:48 well autoreconf is a collection of steps Oct 19 18:35:50 hm Oct 19 18:36:00 yes, and we add an option to it to let you exclude any of those steps Oct 19 18:36:01 khem autoheader is optional Oct 19 18:36:11 kergoth any pointer Oct 19 18:36:19 git grep EXTRA_AUTORECONF Oct 19 18:36:25 woglinde: yes thats what I am saying if we provide an option to what to run then it can be left out elegantly Oct 19 18:36:27 multiple recipes, and a couple classes, use it Oct 19 18:36:39 i'm not seeing how what we have today doesn't provide that Oct 19 18:36:39 its needed for sdl too Oct 19 18:36:43 good to know Oct 19 18:37:05 recipes/apt/apt.inc:EXTRA_AUTORECONF = "--exclude=autopoint,autoheader" Oct 19 18:37:05 hm ah Oct 19 18:37:08 yes Oct 19 18:37:10 thanks Oct 19 18:37:14 kergoth_: I mean sometime autoreconf is not what you can use Oct 19 18:37:15 np Oct 19 18:37:20 kergoth_: where is gnu-configize defined, could not find it Oct 19 18:37:21 khem: what do you mean? Oct 19 18:37:30 eFfeM: it's a script installed by gnu-config-native Oct 19 18:37:33 it isn't "defined" Oct 19 18:37:40 kergoth was the patches not exepted upstream? Oct 19 18:37:45 ah ok Oct 19 18:37:46 or didnt you try it? Oct 19 18:37:50 Aside: it *has* to be a script, "-" isn't allowed in shell functions Oct 19 18:37:55 woglinde: don't think it was ever attempted Oct 19 18:37:57 kergoth_: autoreconf needs certain files eg. *.am files and some packages dont have that Oct 19 18:38:05 huh? Oct 19 18:38:08 no Oct 19 18:38:08 but they still use autconf and autotools Oct 19 18:38:20 autoreconf doesn't run automake if you dont' use automake. Oct 19 18:38:24 last i checked, anyway Oct 19 18:38:30 so not really seeing the dillemma Oct 19 18:38:41 kergoth_: ideally yes but I have run into packages where autoreconf failed Oct 19 18:38:47 so fix them. Oct 19 18:38:53 because it did not find makefile.am Oct 19 18:38:56 kergoth thanks will clean libsdl-directfb Oct 19 18:39:17 then their buildsystem is broken, or worst case you exclude automake with --exclude in EXTRA_AUTORECONF Oct 19 18:39:21 still a nonissue Oct 19 18:39:29 with libtool 2.4 we need to autoreconf all packages Oct 19 18:39:39 otherwise they dont build Oct 19 18:39:43 we've always needed to autoreconf all packages Oct 19 18:39:46 also, not all packages use libtool Oct 19 18:40:13 kergoth_: that true but some packages I have fixed lately use libtool and bypass autoreconf Oct 19 18:40:16 that said, you can work around it by forcing LIBTOOL on the make commandline to the one installed in sysroots, rather than the one generated at ./configure time Oct 19 18:40:37 khem: so again, fix them. there's nothing more than needs to be added to the classes for it Oct 19 18:40:49 there's no point reinventing autoreconf, it does the job Oct 19 18:40:55 kergoth_: yeah I guess the partition wont make it worse Oct 19 18:41:04 and does it intelligently, using autoconf tracing to figure out what needs to be run Oct 19 18:41:10 I was just thinking if we could somehow make it better :) Oct 19 18:41:37 i'm certainly willing to hear ways to improve it, but i haven't heard any yet -- but i do think conceptually splitting the two operations is a good thing Oct 19 18:41:47 kergoth yes Oct 19 18:41:49 that I agree Oct 19 18:41:59 i've also wondered if we should split the *task*, so people realize that configure isn't just ./configure Oct 19 18:42:01 since I saw xmlrpc-c Oct 19 18:42:04 but that could be done in a later step Oct 19 18:42:08 s/step/commit/ Oct 19 18:42:19 yeah pregen task Oct 19 18:42:23 yeah Oct 19 18:42:29 so whats your plan for now Oct 19 18:43:00 first is to clean up the class, split it into two pieces, autotools and autotools_base (unless you have a better name), where the latter just handles gnu-config and oe_runconf Oct 19 18:43:08 * khem converted his ext3 filesystem to ext4 today Oct 19 18:43:15 haha Oct 19 18:43:17 while i'm at it, i'm reworking some of the logic to use autoconf tracing to figure out what needs to be run, and to run things in AC_CONFIG_SUBDIRS Oct 19 18:43:21 tomorrow it will crash Oct 19 18:44:10 but i'm looking for opinions on this, so if you guys have a better idea, or see an issue or lack of benefit to this, i'd like to hear it Oct 19 18:44:41 kergoth_: sounds ok to me Oct 19 18:45:06 it will make it a bit consistent thats my hope Oct 19 18:45:11 also the handling of automake/autoconf/libtool specially in the class rather annoys me, ideally i'd like to ditch the _prepend/_append DEPENDS manipulation in favor of +=/=+, and let the early recipes override the deps to prevent it Oct 19 18:45:39 yeah, thats what i wnat too, it seems silly that recipes have to construct their own oe_runconf configure task, and most end up forgetting about gnu-configize, and end up with ancient config.sub/config.guess Oct 19 18:45:44 heh Oct 19 18:46:03 right Oct 19 18:46:30 you want to drop _prepend/_append from face of bitbake ? Oct 19 18:46:55 or just in DEPENDS Oct 19 18:47:01 eventually, but for now just the non-overrides ones that make sense Oct 19 18:47:07 ah thats gnu-configize does Oct 19 18:47:10 yeah Oct 19 18:47:12 search for it Oct 19 18:47:14 ed Oct 19 18:47:33 it creates config.status Oct 19 18:47:45 the issue is the inherit location and use of DEPENDS = vs DEPENDS +=, if hte class doesnt use append/prepend, the recipe may accidently overwrite the deps Oct 19 18:48:04 hmm I see Oct 19 18:48:18 if the class does +=, then the recipe does DEPENDS = after the inherit, boom Oct 19 18:48:28 course, most set depends before the inherit per the recipe style conventions Oct 19 18:48:40 but not all, would have to do some -e comparisons when making the change Oct 19 18:49:15 kergoth_: yeah Oct 19 18:50:31 these are the sort of things that make oe brittle -- the recipe owner needs to understand implementation details -- how do i override what the class does depends on the details of the class, no consistency Oct 19 18:51:09 no easy solution to that, though Oct 19 18:51:11 heh Oct 19 18:51:33 thats flexibility :) Oct 19 18:51:43 heh, ineed Oct 19 18:51:44 indeed Oct 19 18:52:23 i think it would help if all classes were parsed before the recipe, and the recipe could always override the class except in rare cases Oct 19 18:52:43 e.g. move 'inherit' lines to the top of the recipe to start moving that way Oct 19 18:53:11 hm Oct 19 18:53:20 thats a long transition Oct 19 18:53:21 i'm sure people would disagree from a style standpoint, but at least it would be moving toward more consistent behavior Oct 19 18:53:23 kergoth_: yeah but that could break sand walls Oct 19 18:53:24 * kergoth_ nods Oct 19 18:53:31 but its worth Oct 19 18:53:50 kergoth_: I expect quite a bit of fallout Oct 19 18:54:02 I have seen many recipes depend on where the inherit line is Oct 19 18:54:23 khem: i wrote a bitbake patch that made 'inherit' declarative, and parsed the classes before the recipes Oct 19 18:54:27 that was a fun fiery explosion :) Oct 19 18:56:46 evening. when I try to compile qt4-embedded-gles, I get an error saying that both that one and qt4-embedded provides "qt4-embedded". is this a bug, or something I should just ignore? Oct 19 18:56:59 03Koen Kooi  07org.openembedded.dev * ra4b599ed54 10openembedded.git/recipes/pango/ (pango-1.28.3/no-tests.patch pango.inc pango_1.28.3.bb): pango: add 1.28.3 Oct 19 18:57:15 tasslehoff: if you want to be using the gles one, you should probably set the preferred provider to match Oct 19 18:57:16 re Oct 19 18:57:21 hey florian Oct 19 18:57:39 hi florian Oct 19 18:58:16 kergoth_: where is that one set? I must do _something_ cause it doesn't compile what it needs to :) Oct 19 19:02:58 packaging/package-naming question: why does ode*.bb recipe for example, produce a package libode*.ipk but if you want to include the package in an image you need to include ode-dev and not libode-dev? Is it that debian.bbclass named the package libode-dev? Oct 19 19:09:18 kergoth_: if a commented line as a space at beginning of line then bitbake cries foul Oct 19 19:10:02 i like the idea of parsing the classes first, i'm all in favour for a standard and deterministic system Oct 19 19:10:57 that would also make things like oe-stylize less error-prone Oct 19 19:11:28 tharvey: debian naming Oct 19 19:12:21 ok thx Oct 19 19:16:34 Crofton|work: ok I sent two patches to ml Oct 19 19:16:53 which should fix the native gcc issue. Please try then out and ack/nack them Oct 19 19:17:07 Crofton|work: let me know if more it needed Oct 19 19:19:25 hm, added a patch to rtorrent, forgot a / (so said file:/ instead of file://) and when parsing I get the error ERROR: Unable to open /home/frans/oe/openembedded/recipes/libtorrent/rtorrent_0.8.6.bb and a stack backtrace Oct 19 19:19:41 that doesn't look good Oct 19 19:20:26 forgot an "? Oct 19 19:33:10 no, forgot a / Oct 19 19:33:45 hehe Oct 19 19:33:48 khem no pr bump needed for the gcc-configure-target.inc patch ? Oct 19 19:34:05 woglinde: but a missing / in file:// should not give such an error message Oct 19 19:34:27 hm effem there were some better Oct 19 19:34:32 ask kergoth about it Oct 19 19:34:36 where they have gone Oct 19 19:34:45 or our parser is that bad Oct 19 19:38:02 he kgilmer Oct 19 19:38:13 hi woglinde Oct 19 19:38:48 woglinde, i am still trying to get openjdk on bug13. i got it working once and am now unable to reproduce... :< Oct 19 19:39:15 kgilmer hm uh Oct 19 19:41:28 03Henning Heinold  07org.openembedded.dev * rbc893b21af 10openembedded.git/recipes/libsdl/libsdl-directfb_1.2.14.bb: Oct 19 19:41:28 libsdl-directfb: use the options of our patched autoreconf, to prevent autoheader Oct 19 19:41:28 * thank's kergoth for the hint Oct 19 19:41:29 * no functional changes so no bump of PR's Oct 19 19:41:44 kgilmer, hi Oct 19 19:41:55 hignutoo Oct 19 19:41:59 by bug13 you mean hardware or 1.4.3? Oct 19 19:42:02 hi woglinde Oct 19 19:42:17 both GNUtoo|laptop Oct 19 19:42:48 kgilmer, are you interested in porting bug 1.x forward? Oct 19 19:42:56 my hack was to import the machine config into stable, modify the openjdk-common.inc to remove shark and llvm, and build with MACHINE="bug". Oct 19 19:43:11 it produces all the packages but i get a runtime error Oct 19 19:43:24 when i run after installation on a bug13 Oct 19 19:43:25 ah ok Oct 19 19:45:42 this is strange http://pastebin.org/318400 Oct 19 19:46:46 Ubuntu 10.04.1 64-bit i5-760 Oct 19 19:47:49 BB_NUMBER_THREADS = "2" Oct 19 19:47:55 yeah weird Oct 19 19:49:30 ynezz@gaben:~/oe/test$ patch --version Oct 19 19:49:32 patch 2.6 Oct 19 19:54:43 khem: Hi, can I add your ack to openmoko cleanup? Oct 19 19:56:52 hi guys Oct 19 20:02:39 i have an OMAP L138 problem Oct 19 20:03:52 reboot, biab Oct 19 20:15:08 woglinde, any guess as to how to resolve this? http://pastebin.com/6viEXNzP Oct 19 20:16:43 yes Oct 19 20:16:55 you used a not patched gcc Oct 19 20:17:07 which dont have the atomic patches Oct 19 20:17:45 ok Oct 19 20:18:16 so I should add PREFERRED_VERSION_gcc in the recipe and rebuild? Oct 19 20:18:21 woglinde ^ Oct 19 20:18:44 make sure you use a gcc with atomic Oct 19 20:19:10 same problem here Oct 19 20:19:13 by boost Oct 19 20:19:15 http://answerpot.com/showthread.php?716060-__sync_val_compare_and_swap_4 Oct 19 20:20:31 ok this post says 4.4.4 Oct 19 20:21:01 i tried setting both PREFERRED_VERSION_gcc and PREFERRED_VERSION_gcc-cross but i cannot tell if it is making a difference Oct 19 20:21:17 sure Oct 19 20:21:29 remeber we packported the patches to gcc 4.3 Oct 19 20:21:36 args backported Oct 19 20:21:42 for bug20 Oct 19 20:22:11 ok Oct 19 20:22:22 i do remember that, yes Oct 19 20:23:14 i do not find any recipe in oe-stable however that sets PREFERRED_VERSION_gcc or -cross which makes me think i'm doing it wrong. Oct 19 20:23:26 recipes don't set PREFERRED_VERSION Oct 19 20:23:32 ok :) Oct 19 20:23:33 if they did, it would do absolutely nothing Oct 19 20:23:47 that would explain why it did absolutely nothing Oct 19 20:23:49 PREFERRED_* are by definition configuration metadata, to control bitbake Oct 19 20:23:55 where should i set that? Oct 19 20:23:57 local.conf? Oct 19 20:24:06 whenever its appropriate. Oct 19 20:24:11 local.conf, distro, whatever Oct 19 20:24:22 bitbake really doesn't care where something is set, only that it is. Oct 19 20:24:50 it actually has zero knowledge of it. it parses conf/bitbake.conf, which also happens to include other files, adn then it uses the config metadata it gets back Oct 19 20:25:09 khem, I am back from a meeting Oct 19 20:25:16 ok thanks kergoth_ Oct 19 20:25:17 kgilmer set it in local.conf Oct 19 20:25:27 kgilmer or grep it in config Oct 19 20:26:09 ok woglinde i will, thanks Oct 19 20:28:10 ok i set _gcc and _gcc-cross to "4.3" Oct 19 20:28:21 hm Oct 19 20:28:22 bumped the openjdk pr Oct 19 20:28:28 thats not onough Oct 19 20:28:33 4.3.3 or 4.3.2 Oct 19 20:28:50 or look th recipe where we all apply the atomic patches Oct 19 20:29:23 ok Oct 19 20:30:25 yes i can see gcc-4.3.3 is provided in oe/recipes/gcc Oct 19 20:30:29 http://pastebin.com/hMpqXTTn Oct 19 20:31:39 you need gcc-initial gcc-initermideate and gcc-cross kglimer Oct 19 20:31:49 look at the angstroem config Oct 19 20:31:56 in configu Oct 19 20:32:00 args config dir Oct 19 20:32:06 ok woglinde Oct 19 20:33:44 ok i see Oct 19 20:34:41 good luck Oct 19 20:34:47 thanks woglinde Oct 19 20:36:40 03Martin Jansa  07master * rf781461f15 10openembedded.git/recipes/images/fso-image.inc: Oct 19 20:36:40 fso-image: remove obsoleted openmoko stuff Oct 19 20:36:40 Acked-by: Michael 'Mickey' Lauer Oct 19 20:36:40 Acked-by: Frans Meulenbroeks Oct 19 20:36:40 Signed-off-by: Martin Jansa Oct 19 20:37:00 03Martin Jansa  07master * r83847370bd 10openembedded.git/ (38 files in 10 dirs): Oct 19 20:37:00 openmoko: obsolete tasks and images, openmoko as distribution was obsoleted on Aug 10, 2010. Oct 19 20:37:00 Acked-by: Michael 'Mickey' Lauer Oct 19 20:37:00 Acked-by: Frans Meulenbroeks Oct 19 20:37:01 Signed-off-by: Martin Jansa Oct 19 20:37:01 03Martin Jansa  07master * rb11e4c4add 10openembedded.git/conf/machine/om-3d7k.conf: Oct 19 20:37:02 om-3d7k: remove machine definition, because it was canceled long time ago Oct 19 20:37:02 Acked-by: Michael 'Mickey' Lauer Oct 19 20:37:03 Acked-by: Frans Meulenbroeks Oct 19 20:37:03 Signed-off-by: Martin Jansa Oct 19 20:37:04 03Martin Jansa  07master * ra8caf896c9 10openembedded.git/recipes/openmoko2/ (21 files in 6 dirs): Oct 19 20:37:04 libmokoui2, openmoko-apps: move to obsolete as libmokoui2 doesn't build for a while and nobody cares to fix it Oct 19 20:37:08 Acked-by: Michael 'Mickey' Lauer Oct 19 20:37:08 Acked-by: Frans Meulenbroeks Oct 19 20:37:09 Signed-off-by: Martin Jansa Oct 19 20:37:09 03Martin Jansa  07master * r3ca828e418 10openembedded.git/recipes/linux/ (13 files in 3 dirs): Oct 19 20:37:10 linux-openmoko: obsolete 2.6.24 and 2.6.28 Oct 19 20:37:10 Acked-by: Michael 'Mickey' Lauer Oct 19 20:37:11 Acked-by: Frans Meulenbroeks Oct 19 20:37:11 Signed-off-by: Martin Jansa Oct 19 20:37:24 03Martin Jansa  07master * r1ea912e301 10openembedded.git/recipes/ (12 files in 4 dirs): Oct 19 20:37:24 openmoko: move initscripts, psplash, alsa-scenarios also to obsolete Oct 19 20:37:24 Acked-by: Michael 'Mickey' Lauer Oct 19 20:37:24 Acked-by: Frans Meulenbroeks Oct 19 20:37:25 Signed-off-by: Martin Jansa Oct 19 20:37:25 03Martin Jansa  07master * r58c6b39473 10openembedded.git/recipes/ (15 files in 9 dirs): Oct 19 20:37:26 openmoko: remove openmoko directories used from FILESPATH as openmoko distribution was obsoleted on Aug 2010 Oct 19 20:37:27 Acked-by: Michael 'Mickey' Lauer Oct 19 20:37:27 Acked-by: Frans Meulenbroeks Oct 19 20:37:28 Signed-off-by: Martin Jansa Oct 19 20:37:28 03Martin Jansa  07master * r3b2dd54a4a 10openembedded.git/ (38 files in 9 dirs): Oct 19 20:37:29 openmoko: obsolete old openmoko classes and recipes using those Oct 19 20:37:29 Acked-by: Michael 'Mickey' Lauer Oct 19 20:37:30 Acked-by: Frans Meulenbroeks Oct 19 20:37:30 Signed-off-by: Martin Jansa Oct 19 20:37:34 03Martin Jansa  07master * r8a9fa4db8b 10openembedded.git/recipes/ (2 files in 2 dirs): Oct 19 20:37:34 settingsgui: move both versions to obsolete, nobody have it in task/images now Oct 19 20:37:34 Acked-by: Michael 'Mickey' Lauer Oct 19 20:37:34 Acked-by: Frans Meulenbroeks Oct 19 20:37:35 Signed-off-by: Martin Jansa Oct 19 20:37:36 03Martin Jansa  07master * r97e37d5275 10openembedded.git/removal.txt: Oct 19 20:37:36 removal: add entry for removing obsoleted openmoko recipes Oct 19 20:37:36 Acked-by: Michael 'Mickey' Lauer Oct 19 20:37:58 Signed-off-by: Martin Jansa Oct 19 20:37:58 03Martin Jansa  07master * r0b0094d131 10openembedded.git/contrib/angstrom/sort.sh: Oct 19 20:37:58 angstrom/sort.sh: remove om-3d7k as it was canceled long time ago Oct 19 20:37:58 Acked-by: Michael 'Mickey' Lauer Oct 19 20:37:58 Acked-by: Frans Meulenbroeks Oct 19 20:37:59 Acked-by: Koen Kooi Oct 19 20:37:59 Signed-off-by: Martin Jansa Oct 19 20:38:00 03Martin Jansa  07master * rc7574afb3f 10openembedded.git/contrib/angstrom/build-feeds.sh: Oct 19 20:38:01 angstrom/build-feeds: remove openmoko targets Oct 19 20:38:01 Acked-by: Michael 'Mickey' Lauer Oct 19 20:38:02 Acked-by: Frans Meulenbroeks Oct 19 20:38:02 Acked-by: Koen Kooi Oct 19 20:38:03 Signed-off-by: Martin Jansa Oct 19 20:38:04 wuhu Oct 19 20:38:14 03Martin Jansa  07master * r909dd51e5d 10openembedded.git/recipes/meta/meta-angstrom-2007.bb: Oct 19 20:38:14 meta-angstrom-2007: remove openmoko-browser2 which was moved to obsoleted Oct 19 20:38:15 Acked-by: Michael 'Mickey' Lauer Oct 19 20:38:15 Acked-by: Frans Meulenbroeks Oct 19 20:38:16 Acked-by: Koen Kooi Oct 19 20:38:16 Signed-off-by: Martin Jansa Oct 19 20:39:49 Anyone know if the darwin-branch is currently buildable on OS X? Oct 19 20:40:27 tasslehoff: try the regular 'darwin' branch, rather than koens Oct 19 20:40:33 its rebased on current master and has a few more fixes Oct 19 20:40:36 its a work in progress, though Oct 19 20:40:47 * kergoth_ was working on it some this weekend Oct 19 20:40:53 the current issue is cp and sed arguments Oct 19 20:40:53 03Simon Busch  07master * r08fc3adf98 10openembedded.git/conf/distro/include/preferred-shr-versions.inc: Oct 19 20:40:53 preferred-shr-versions: choose git version of tslib as default Oct 19 20:40:53 Signed-off-by: Simon Busch Oct 19 20:40:53 Signed-off-by: Martin Jansa Oct 19 20:40:59 sed -i behaves differently on osx than in linux Oct 19 20:41:16 i've tried generating wrapper scripts for them to handle argument compatibility, but thats a problem too Oct 19 20:41:26 because the wrappers used getopt, but getopt portability too is a problem :) Oct 19 20:41:30 kergoth_: thx for the info Oct 19 20:41:31 osx getopt can't handle spaces in arguments Oct 19 20:41:32 :) Oct 19 20:41:43 so i tried creating python wrapper scripts for them Oct 19 20:41:50 but do_configure in a recipe seems to hang if i do that Oct 19 20:41:54 haven't gotten past that yet Oct 19 20:42:08 of course, we can adjust all recipes to use correct cp/sed arguments, and don't use -i at all, etc Oct 19 20:42:12 but thats a pain in the ass Oct 19 20:42:22 particularly for sed, sinc eyou have to > $file.new && mv $file.new $file or whatever Oct 19 20:42:33 has to be done manually, not scripted Oct 19 20:42:36 the fixes that is Oct 19 20:43:03 http://gist.github.com/630556 is the kind of thing i was playing with, before i realized getopt compatibility was a problem Oct 19 20:45:03 well i got the gcc-* stuff set to 4.3.3 Oct 19 20:45:12 rebuiding and crossing fingers Oct 19 20:45:17 kergoth_: I opened Pandoras box :) Oct 19 20:45:24 should we use sed-native? :) Oct 19 20:45:31 tasslehoff: heh, it's been open a while, just.. meh Oct 19 20:45:38 Jay7: i vote hell no Oct 19 20:45:46 hehe :) Oct 19 20:46:16 re ka6sox Oct 19 20:46:24 anyway seems we should provide SED_INPLACE macro/function then.. Oct 19 20:46:31 I'll try the branch as soon as I have the time. Night is falling here. Oct 19 20:46:35 like FreeBSD ports does Oct 19 20:47:19 hiya woglinde Oct 19 20:48:14 Jay7: agreed, like i said in the above gist, might have to implement it ourselves Oct 19 20:48:27 as long as we don't do so in every recipe that wants -i, i'll be happy.. Oct 19 20:48:28 :) Oct 19 20:48:46 i see gcc icedtea patches getting downloaded, taking that as a good sign. Oct 19 20:53:09 03Frans Meulenbroeks  07org.openembedded.dev * re857d9a0ac 10openembedded.git/recipes/xmlrpc-c/xmlrpc-c_1.06.41.bb: Oct 19 20:53:09 xmlrpc-c: add inherit binconfig Oct 19 20:53:09 Signed-off-by: Frans Meulenbroeks Oct 19 20:53:19 03Frans Meulenbroeks  07org.openembedded.dev * r8b75cbf789 10openembedded.git/recipes/libtorrent/ (files/checks.patch rtorrent_0.8.6.bb): Oct 19 20:53:19 libtorrent: added xmlrpc support Oct 19 20:53:19 so web clients can interface with it Oct 19 20:53:19 Signed-off-by: Frans Meulenbroeks Oct 19 20:53:26 03Frans Meulenbroeks  07org.openembedded.dev * r2d578648ac 10openembedded.git/recipes/lighttpd/lighttpd_1.4.28.bb: Oct 19 20:53:26 lighttpd_1.4.28.bb: added checksums Oct 19 20:53:26 Signed-off-by: Frans Meulenbroeks Oct 19 20:58:56 khem, what is the minimum stuff I can get away with rebuilding? Oct 19 21:02:25 Crofton: just bitbake -c clean gcc;bitbake gcc Oct 19 21:03:42 k Oct 19 21:03:43 trying now Oct 19 21:11:28 ah ha Oct 19 21:12:00 getopts is in SuSv3, a shell builtin it seems, and doesn't seem to die on args with spaces, unless i screwed up my test Oct 19 21:12:11 can use that instead of getopt Oct 19 21:13:08 gn Oct 19 21:15:17 03Frans Meulenbroeks  07org.openembedded.dev * r626174dcf5 10openembedded.git/recipes/transmission/transmission_2.04.bb: Oct 19 21:15:17 transmission_2.04.bb: added checksums Oct 19 21:15:17 Signed-off-by: Frans Meulenbroeks Oct 19 21:15:43 nite everyone Oct 19 21:15:49 night Oct 19 21:30:31 03Koen Kooi  07org.openembedded.dev * r55f2ba0c38 10openembedded.git/recipes/qt4/ (qt-4.7.0.inc qt4-embedded.inc qt4.inc): Oct 19 21:30:31 qt4: move QtDeclarative to qt4.inc Oct 19 21:30:31 * This avoids pulling in Qt 4.7.0 into meta-toolchain-qte when 4.6.3 is specified Oct 19 21:30:31 * No further changes in behaviour or packaging Oct 19 21:42:24 these wrapper scripts are a giant pain in my ass Oct 19 21:43:58 I wonder how to find what have failed http://tinderbox.openembedded.net/builds/97162/ Oct 19 21:44:53 kergoth_, all wrapper scripts are. Oct 19 21:45:06 true Oct 19 21:47:37 * kergoth_ tests autoconf,automake,libtool relocation fixes/workarounds Oct 19 21:48:11 kergoth_: your suggestion did not qork Oct 19 21:48:14 work Oct 19 21:48:18 worked fine for me Oct 19 21:48:19 i tested it Oct 19 21:48:30 not in the recipe, but in a different one with a test var Oct 19 21:48:48 hmmm Oct 19 21:49:00 PG&E shutdown power once more Oct 19 21:49:11 and I am not at home to switch that box on Oct 19 21:49:27 will try again in the evening Oct 19 21:49:46 k Oct 19 21:49:51 ext4 seems zappier Oct 19 21:49:56 anything we can do to reduce the anonymous python is a good thing for performance Oct 19 21:50:09 * khem` nods Oct 19 21:56:06 20000 line of code in a static function :) Oct 19 21:56:12 yikes Oct 19 21:56:20 and all if it is if else goto Oct 19 21:56:35 shit load of labels Oct 19 21:56:46 sometimes I pity gcc Oct 19 21:56:52 what all people can abuse it for Oct 19 21:57:42 woo, coreutils-native and gnu-config-native built fine with the new wrappers for cp/sed -- sad that thats progress Oct 19 21:58:06 * kergoth_ 'll have to test performance before/after, might be a better approach, or maybe an opt-in to check sanity option Oct 19 21:58:23 yeah perf is important Oct 19 21:59:10 can always explicitly opt into the wrappers based on build distro, so it only affects those with non-gnu tools at least Oct 19 21:59:11 heh Oct 19 22:00:22 * kergoth_ fires off a task-boot on the osx box and crosses fingers Oct 19 22:02:33 osx ??? Oct 19 22:02:37 does it work at all Oct 19 22:02:37 kergoth, when that is setup and going I'll setup a tester. Oct 19 22:08:53 khem`: http://git.openembedded.org/cgit.cgi/openembedded/log/?h=darwin Oct 19 22:12:57 kergoth_, are you using a ppc or intel for it? Oct 19 22:13:56 I also wonder Oct 19 22:14:18 are you using macports? or it's "stock" osx? Oct 19 22:17:04 kergoth_: cool so you are reviving the old darwin-host-fixes branch thats fantastic Oct 19 22:17:14 ynezz: you need macports Oct 19 22:17:29 and ports in freebsd Oct 19 22:19:20 khem`: yeah, started by rebasing koen's branch on a current tree and started trying to build, working through the problems as they show up Oct 19 22:19:25 i'm sure there's still a lot to do Oct 19 22:19:26 heh Oct 19 22:20:45 kergoth_: nice. I can do a git pull on my MBP :) Oct 19 22:20:49 and join you Oct 19 22:20:57 my old tree is there Oct 19 22:21:07 I just ran out of interest Oct 19 22:21:13 last time Oct 19 22:21:48 at the moment i'm knee deep in dealing with sed -i Oct 19 22:22:31 it's all over :p Oct 19 22:22:34 heh sed was a PITA Oct 19 22:22:36 and also cp Oct 19 22:22:45 we assume gnu cp all the time Oct 19 22:23:00 good..I have a ppc mini that is bored... Oct 19 22:23:39 I was trying to do the same few days ago, but then realized I would need to spend month or so to just build minimal-image and forget about it :) Oct 19 22:24:03 khem`: i'm currently trying use of wrapper scripts which either adjust args or abort immediately when using incompatible bits, but its proving to be a huge pain in the ass Oct 19 22:24:11 which isnt too surprising Oct 19 22:24:17 and was thinking why it's not using python builtins instead of those awk/sed kungfu Oct 19 22:25:15 itd be nice to use more of that in our stuff Oct 19 22:25:57 damnit, using getopts helped the space in args issue, but now running into other issues.. maybe i should give my python wrappers another try Oct 19 22:26:19 * kergoth_ wishes dealing with strings with spaces was less painful in shell Oct 19 22:27:15 i have a set of arguments, and i want to replace one with something else, or skip just that one argument, and pass the rest onto the real binary without breaking on spaces -- nightmare Oct 19 22:27:27 heh Oct 19 22:27:33 if you shift the args off the stack, you can't "unshift" something back on Oct 19 22:27:41 if you add it to another variable, you lose the spaces unless you add extra quotes Oct 19 22:27:54 at which point you're hosed because not all arguments have extra quotes, so you cant use eval Oct 19 22:28:10 elm street Oct 19 22:28:15 if you try to add extra quotes to everything, and the argument itself has quotes, you're screwed Oct 19 22:28:23 is there something that triggers a rebuild of all sdk components if for example I've changed the SDK_PATH? Oct 19 22:31:36 tharvey: I dont think so Oct 19 22:31:55 its will probably trigger reparse only Oct 19 22:32:08 * khem` misses var tracking Oct 19 22:32:28 ya, var tracking would be nice, but probably add a lot of overhead Oct 19 22:33:37 khem`, ok Oct 19 22:33:44 finally tested, patches look better Oct 19 22:33:52 going to try something harder :) Oct 19 22:34:31 Crofton|work: unfortunately the second patch you have to try what kergoth suggested Oct 19 22:34:41 Crofton|work: unfortunately the second patch you have to try what kergoth suggested Oct 19 22:34:46 urg Oct 19 22:34:59 as that seems a better approach Oct 19 22:35:13 it did not work for me but I might have goofed up something Oct 19 22:35:22 hmm Oct 19 22:35:28 and now I dont have access to anymachine with OE src on it Oct 19 22:35:29 seems like yuo have the basic concept Oct 19 22:35:42 thats true Oct 19 22:37:04 tharvey: not as much as you'd think Oct 19 22:38:54 ya, I suppose you just need to cache them and look for changes, if change present then perhaps some per-variable facility to act on change Oct 19 22:39:03 heh Oct 19 22:39:18 I am going to try copiling something useful Oct 19 22:39:26 then I will eitehr have a beer Oct 19 22:39:32 or update the patch Oct 19 22:39:50 like the way you think Oct 19 22:50:54 khem`, your patch seems to not fix lib search paths though Oct 19 22:51:58 03Paul Eggleton  07org.openembedded.dev * r132d5ea735 10openembedded.git/recipes/opie-mediaplayer2/opie-mediaplayer2-skin-default_cvs.bb: Oct 19 22:51:58 opie-mediaplayer2: remove reference to OPIE_CVS_PV and default OPIE_GIT_PV if not set Oct 19 22:51:58 I'm a bit skeptical of having these defaults; anyone building Opie from git Oct 19 22:51:58 *should* be including preferred-opie-git-versions.inc and not trying to build Oct 19 22:51:58 without doing so. For consistency however I'm including the defaults here. Oct 19 22:52:03 hmm Oct 19 22:52:08 03Paul Eggleton  07org.openembedded.dev * ra1d962b3b9 10openembedded.git/conf/distro/include/preferred-opie-cvs-versions.inc: Oct 19 22:52:08 conf/distro/include: Remove preferred-opie-cvs-versions.inc. Oct 19 22:52:08 This include file is now obsolete; persons wishing to build the latest Oct 19 22:52:08 development version of Opie should now build from git by including Oct 19 22:52:08 preferred-opie-git-versions.inc instead. Oct 19 22:52:54 yeah, not linking againast installed libs Oct 19 22:54:09 http://pastebin.com/7HaeZsqC Oct 19 23:18:53 khem`, ping Oct 19 23:19:38 hmm, anyone have a copy of eautoreconf laying around, or know where i can find it on the gentoo site? Oct 19 23:29:28 Crofton|work: ok the pastbin link does not work Oct 19 23:31:46 kergoth_, does it matter if the box I have is a G4 or a G5? Oct 19 23:34:24 oh pastebin.com is kaput Oct 19 23:37:55 urg Oct 19 23:38:53 heh Oct 19 23:39:03 yeah I just figured that out myself Oct 19 23:39:40 ka6sox: doubt it, haven't even started looking at ppc yet Oct 19 23:41:25 /usr/lib/gcc/arm-angstrom-linux-gnueabi/4.5.2/../../../../arm-angstrom-linux-gnueabi/bin/ld: warning: libm.so.6, needed by /usr/lib/gcc/arm-angstrom-linux-gnueabi/4.5.2/../../../libfftw3f.so, not found (try using -rpath or -rpath-link) Oct 19 23:41:50 there are a few more lines Oct 19 23:41:54 similar Oct 19 23:42:33 /lib/libm.so.6 exists Oct 20 00:00:38 Crofton|work: can you readelf -d /usr/lib/gcc/arm-angstrom-linux-gnueabi/4.5.2/../../../libfftw3f.so Oct 20 00:03:11 hrm Oct 20 00:03:56 oh I should be on road Oct 20 00:03:58 ttyl Oct 20 00:06:03 damnit, this autoreconf thing is bugging me Oct 20 00:06:03 hrm Oct 20 00:06:07 * kergoth_ tries something Oct 20 00:26:41 khem yeas Oct 20 00:27:23 bother Oct 20 00:27:30 http://pastebin.com/JVhN6FNw Oct 20 00:38:38 denix, ping Oct 20 00:38:42 pong Oct 20 00:38:55 hi Oct 20 00:38:58 hi Oct 20 00:39:08 it looks like I will go to Cambridge on Monday Oct 20 00:39:33 when is your arrival at LHR? Oct 20 00:40:54 1005AM Oct 20 00:41:28 you? Oct 20 00:41:54 same. UA-924? Oct 20 00:42:28 no USAir 728 Oct 20 00:42:35 I will coem from PHL Oct 20 00:42:41 my route sux Oct 20 00:42:49 ah, mine is from IAD Oct 20 00:42:53 ROA->LGA->PHL->LHR Oct 20 00:43:03 whoa Oct 20 00:43:15 the problems living in a cow town Oct 20 00:43:26 not sure what is up with the ROA-PHL flight Oct 20 00:43:37 how far ROA is from IAD? Oct 20 00:43:45 4-5 hours Oct 20 00:43:56 US air does not use IAD Oct 20 00:44:07 my international gateways are PHL and CLT Oct 20 00:44:11 it is annoyong Oct 20 00:44:19 ah, thought it was slightly closer, maybe 3 hours... Oct 20 00:44:23 what's CLT? Oct 20 00:44:23 yeah Oct 20 00:44:35 3 hours if you are really fast Oct 20 00:44:40 Charlotte Oct 20 00:44:52 I see Oct 20 00:45:21 so, do you want to take the train to Cambridge? Oct 20 00:45:40 not sure Oct 20 00:45:46 I need to get a place to stay Oct 20 00:46:03 haven't you booked a hotel? Oct 20 00:46:06 and on the way back I need to get from Cambridge to WOking Oct 20 00:46:07 no Oct 20 00:46:14 my registration was up in the air Oct 20 00:46:31 yeah, I'll end up in someones basement Oct 20 00:46:54 I ended up booking at Travelodge - no rooms left at Univ Arms... about 1 mile away Oct 20 00:47:15 k Oct 20 00:47:19 they do not have parking Oct 20 00:47:29 well, they have expensive parking Oct 20 00:47:40 I need to play around with the europcar website Oct 20 00:48:07 I looked last week and the proces were not crazy Oct 20 00:48:19 you still want to drive? :) Oct 20 00:48:43 I like driving :) Oct 20 00:48:52 especially on the proper side of the road Oct 20 00:48:57 me too, but not in the UK! :) Oct 20 00:50:23 my current thinking is get a car to get back from Cambridge since I need to stop at my uncles Oct 20 00:50:26 I guess it pays to have British roots and past experience... :) Oct 20 00:50:31 yeah Oct 20 00:50:35 and no fear Oct 20 00:50:56 :) Oct 20 00:51:33 I need to write the postcodes for where I will be Oct 20 00:51:41 that should make immigration much easier Oct 20 00:52:05 what do you mean? Oct 20 00:52:26 well, last year, he really wanted the postcode for my uncle's house Oct 20 00:52:34 I knew about everything but the post code Oct 20 00:52:39 ah, interesting :) Oct 20 00:52:47 he suggested I renew my british passport Oct 20 00:52:54 he said it would make things easier Oct 20 00:53:00 I bet :) Oct 20 00:53:33 driving on the "wrong" side oyf the road...don't forget to look the other wa Oct 20 00:53:35 y Oct 20 00:55:53 Crofton: so let me know if you decide to ride the train to Cambridge and take the car on the way back - would be nice to have a company at least one way :) Oct 20 00:57:45 ok Oct 20 00:57:57 I will resolve all this tomorrow Oct 20 00:58:48 great! check the Travelodge - if they still have rooms left and if you don't need to park... Oct 20 00:59:38 they had non-refundable 55 pounds/night rate... Oct 20 01:13:24 damnit Oct 20 01:29:26 i'm building a toolchain, and it's on the last few steps, and it's failing on do_populate_sdk for the toolchain package, saying it can't satisfy the dependencies for the toolchain due to not finding libsdl-mixer Oct 20 01:29:31 which builds just fine Oct 20 01:30:13 any ideas? Oct 20 01:30:30 vorporeal: what are you building which image Oct 20 01:30:38 meta-toolchain-pandora Oct 20 01:30:50 provided by a pandora project overlay Oct 20 01:31:24 vorporeal: hmmm that could be specific to pandorabox Oct 20 01:31:43 why would it not be able to satisfy the dependency if it builds though? Oct 20 01:34:39 vorporeal: is it a run-time dependency? Oct 20 01:40:27 khem, ping Oct 20 01:41:31 i suppose it's possible that something broke when i fixed libsdl-mixer not using libtool2.4, but i don't think that's likely Oct 20 01:41:31 i basically just followed a commit you had made to fix libtool2.4 issues in something else Oct 20 01:41:31 libcurl, i think Oct 20 01:41:31 (sorry, by the way, for asking so many questions over the past few days - even though it may not seem that way, i do a lot of searching the internet and trying things myself before i ask here) Oct 20 01:41:49 vorporeal: it may not be providing that package Oct 20 01:42:30 Crofton|work: hi Oct 20 01:45:30 khem, did you need something from me? Oct 20 01:46:12 http://pastebin.com/SxrxYciN Oct 20 01:46:15 I think you asked for Oct 20 01:46:30 khem: ah, it may depend on it but nothing "provides" it? Oct 20 01:46:42 vorporeal: yes Oct 20 01:48:12 Crofton|work: ok, seems fine Oct 20 01:48:28 denix0: yes, it's an RDEPENDS Oct 20 01:48:48 libsdl-mixer-dev, to be exact Oct 20 01:48:52 Crofton|work: can you paste the error again Oct 20 01:49:05 Crofton|work: I mean the linking error Oct 20 01:51:31 vorporeal: do you have said package in deploy? Oct 20 01:53:44 gimme a minute to figure out what that means - i started using bitbake/OE about 3 days ago Oct 20 01:53:55 :) Oct 20 01:55:02 ah, i see what you mean - i'll check Oct 20 01:56:09 denix0: there's nothing in deploy except for eglibc... Oct 20 01:56:37 wait Oct 20 01:56:40 i'm an idiot Oct 20 01:56:41 hold on Oct 20 01:57:18 it is not in deploy Oct 20 01:57:25 if i find it, should i copy it in? Oct 20 01:57:32 or should i try to "properly" fix it? Oct 20 01:57:40 where is it? Oct 20 01:58:18 denix0: for starters, i'm going to try to rebuild it, see if that fixes anything Oct 20 01:58:59 aha, it copied in libsdl-mixer-dev and -dbg, but not libsdl-mixer Oct 20 02:02:47 which is because it believes the libsdl-mixer package to be empty Oct 20 02:05:34 my guess is i screwed something up when i tried to get it to work with libtool 2.4 Oct 20 02:05:45 http://pastebin.com/3ESatrEw Oct 20 02:07:09 where does libfftw3f.so come from? Oct 20 02:07:31 Crofton|work: ok now paste content of /usr/lib/libfftw3f.la Oct 20 02:07:37 on target Oct 20 02:10:54 http://pastebin.com/cKh3ZyR2 Oct 20 02:11:23 nothing crazyy looking Oct 20 02:12:15 Crofton|work: yes. Is libm.so symlinked correctly to /lib/libm-2.12.1.so Oct 20 02:13:35 Crofton|work: on command line can you try gcc -o conftest -g -O2 -xc /dev/null -lfftw3f Oct 20 02:13:46 and also add -v Oct 20 02:17:32 ok Oct 20 02:17:38 I am sort of not working ... Oct 20 02:21:04 http://pastebin.com/uMsmXsbv Oct 20 02:23:21 Crofton|work: hmmm I think I understand the issue Oct 20 02:23:32 Crofton|work: are you able to try a small patch Oct 20 02:23:36 noq Oct 20 02:23:38 bow Oct 20 02:23:45 now !!! finally Oct 20 02:29:04 crofton|work: try this patch http://pastebin.com/2ZXva8ar Oct 20 02:29:25 bitbake -c clean binutils;bitbake binutils Oct 20 02:29:40 and then repeat the above experiment and it should work I believe ? Oct 20 02:29:53 Crofton|work: I will wait for your test results Oct 20 02:33:00 k Oct 20 02:34:41 khem: any idea why the Makefile for libsdl-mixer would have -lmad but on the command-line output, it is replaced with /usr/lib/libmad.so? Oct 20 02:35:01 vorporeal: libtool Oct 20 02:35:06 would do that Oct 20 02:35:30 khem: any setting/value i should look for? Oct 20 02:35:39 i'm using 2.4 now, which fixed a LOT of those issues before Oct 20 02:35:47 except for this one Oct 20 02:36:07 vorporeal: its possible that this package needs to get new macros Oct 20 02:36:09 which recipe is it Oct 20 02:36:18 libsdl-mixer_1.2.11.bb Oct 20 02:36:43 the libtool.m4 file comments say it's for libtool2.2.6 Oct 20 02:37:06 but if i only run through the build step and check the libtool in the source dir, it's 2.4 Oct 20 02:37:21 and it seems like the m4 files in the acinclude directory never get used Oct 20 02:37:43 vorporeal: do_compile is what fails ? Oct 20 02:38:09 yes Oct 20 02:38:29 when linking libSDL_mixer-1.2.so.0.10.1 Oct 20 02:40:19 vorporeal: can you delete do_compile () in "libsdl-mixer_1.2.11.bb" Oct 20 02:40:23 and rebuild this recipe Oct 20 02:42:21 khem: no luck, fails in the same place Oct 20 02:43:11 vorporeal: ok. I wonder why it did not get new libtool macros Oct 20 02:43:12 the output includes "-Wl,-rpath -Wl,/usr/lib", which i remember you saying was not good Oct 20 02:43:34 vorporeal: thats fixed in latest libtool in OE Oct 20 02:43:40 I took care of this for cross building Oct 20 02:44:01 vorporeal: you might have to delete the .m4 fileds Oct 20 02:44:03 files Oct 20 02:44:14 before do_configure Oct 20 02:44:35 so -c patch, delete, continue? Oct 20 02:45:01 what all .m4 files are there in workdir of libsdl Oct 20 02:46:16 libtool.m4 ltdl.m4 lt~obsolete.m4 ltoptions.m4 ltsugar.m4 ltversion.m4 pkg.m4 sdl.m4 smpeg.m4 Oct 20 02:48:14 rm libtool.m4 ltdl.m4 lt~obsolete.m4 ltoptions.m4 ltsugar.m4 ltversion.m4 Oct 20 02:48:19 alright Oct 20 02:48:43 see do_configure_prepend in "gnutls_2.10.1.bb" Oct 20 02:48:48 thats what you need Oct 20 02:48:59 I mean you need to adapt that to this case Oct 20 02:57:20 if i delete those m4 files, it doesn't create the libsdl-mixer package, claiming that it's empty **** ENDING LOGGING AT Wed Oct 20 02:59:56 2010