**** BEGIN LOGGING AT Tue Oct 05 02:59:57 2010 Oct 05 05:32:32 Iam facing problem in do_rootfs task of omap3-desktop-image.bb Oct 05 05:32:37 resolve_conffiles: Existing conffile /media/500_p4/open_embedded/overo-oe/tmp/rootfs/omap3-desktop-image/etc/device_table is different from the conffile in the new package. The new conffile will be placed at /media/500_p4/open_embedded/overo-oe/tmp/rootfs/omap3-desktop-image/etc/device_table-opkg. Oct 05 05:32:43 opkg_install_cmd: Cannot install package swfdec-mozilla. Oct 05 05:32:47 opkg_install_cmd: Cannot install package cheese. Oct 05 05:32:52 opkg_install_cmd: Cannot install package task-proper-tools. Oct 05 05:32:57 opkg_install_cmd: Cannot install package task-base-extended. Oct 05 05:33:12 Any clue ! what i was missing ? Oct 05 05:49:47 gm Oct 05 05:51:26 Hi eFfeM_work Oct 05 05:51:40 you seem to be early riser :) Oct 05 05:51:42 how are you doing? Oct 05 05:51:49 yup, 7.50 am here Oct 05 05:51:52 good. Oct 05 05:52:03 and you are already at work ? Oct 05 05:52:11 yes Oct 05 05:52:23 reminds me of my working days in .de Oct 05 05:52:24 wife & one kid had to get up early so I joined them Oct 05 05:52:34 I use to come in last Oct 05 05:52:41 :-) Oct 05 05:52:49 that was like 8:45 Oct 05 05:52:51 some people here start at 9-9.30 Oct 05 05:53:04 here if I get up at 8:45 thats early ;) Oct 05 05:53:44 btw. I have libtool 2.4 upgrade patches Oct 05 05:53:51 yeah, I know the times in the valley Oct 05 05:53:56 ah cool, Oct 05 05:54:06 I need a lot of support to test them Oct 05 05:54:20 libtool is very very core Oct 05 05:54:35 and I would like to make sure that it works in all cases Oct 05 05:54:45 i can give it a try building things, but can't test today, have to teach on tue evening Oct 05 05:54:58 So I am going to post a version today to ml Oct 05 05:55:05 cool Oct 05 05:55:13 seems I also have some issues with the testing branch Oct 05 05:55:16 and hope that people will feel enthused to test it more Oct 05 05:55:38 hmm I have no issues with the branch Oct 05 05:56:38 i got my kernel note things back (at least some of them) and the patch for redefining SRC_URI also does not seem to work Oct 05 05:56:52 I have only built minimal-image with them so I am expecting more helping hands to do larger builds thats why the early post of patches. I know they may not be final yet Oct 05 05:57:04 eFfeM_work: hmm Oct 05 05:57:04 okie Oct 05 05:57:16 well found the problem, my kernel.bbclass is definitely not current Oct 05 05:57:24 :) Oct 05 05:57:43 dont you do git pull --rebase on the testing-next Oct 05 05:58:06 only a git pull, no --rebase Oct 05 05:58:27 yeah pull is file as long as you dont have local changes Oct 05 05:58:33 s/file/fine Oct 05 05:59:12 i have no local changes, this is on our autobuilder Oct 05 05:59:16 k Oct 05 05:59:24 where and what do u teach Oct 05 06:01:22 local university, network management Oct 05 06:01:31 part time job (tue evening) Oct 05 06:01:53 can you peek at this: http://pastebin.ca/1954548 Oct 05 06:02:15 maybe I introduced a problem when moving to the linaro branch Oct 05 06:03:22 actually the script I run is: http://pastebin.ca/1954549 Oct 05 06:03:33 MACHINE, DISTRO and TARGET_RECIPE are set from outside Oct 05 06:04:22 hi i am truing to compiling c++ file .. but its giving one warning Oct 05 06:04:26 not 100% about the checkout Oct 05 06:04:39 warning: libc.so, needed by /home/abhilash/overo-oe/tmp/sysroots/i686-linux/usr/armv7a/lib/gcc/arm-angstrom-linux-gnueabi/4.3.3/../../../../arm-angstrom-linux-gnueabi/lib/libgcc_s.so, not found (try using -rpath or -rpath-link) Oct 05 06:04:47 any idea?? Oct 05 06:05:49 ?? Oct 05 06:05:58 anyone alive?? Oct 05 06:28:57 khem didn't see why i still got old files, decided to nuke the complete dir and fetch git from scratch; will keep an eye on it next week when pulling and will rerun all tests Oct 05 06:48:33 eFfeM_work: hmmm how do u checkout the branch Oct 05 06:48:41 are you tracking remote testing-next Oct 05 06:48:49 if not then it wont et yupdate Oct 05 06:49:01 no tracking Oct 05 06:49:04 r0ck_: which distro/image Oct 05 06:49:07 see the pastebin above Oct 05 06:49:17 thought git pull would update Oct 05 06:49:37 git checkout -b testing-next origin/testing-next Oct 05 06:49:40 thats good Oct 05 06:50:06 followed by a git pull (near the end, after the here document to generate local.conf) Oct 05 06:50:25 should I add --track to the checkout ? Oct 05 06:53:24 anyway a fresh restart gave me a good git Oct 05 06:54:15 yes it should Oct 05 07:06:57 moin Oct 05 07:08:16 good morning Oct 05 07:34:39 03Koen Kooi  07org.openembedded.dev * r8e35a2ac8f 10openembedded.git/recipes/mplayer/ (mplayer/vofw-swscale.diff mplayer_svn.bb): mplayer svn: make swscale accept larger input images regardless of arch Oct 05 07:35:49 khem: still there? Oct 05 07:38:24 ant_work: yes Oct 05 07:38:39 hi, I've a question about Oct 05 07:38:40 http://cgit.openembedded.org/cgit.cgi/openembedded/commit/?id=5ac52371b09a25518b60e23e80c8c812cdea6e36 Oct 05 07:38:57 I had Oct 05 07:38:59 SRC_URI = "file://${LOGO_SIZE}/logo_linux_clut224.ppm.bz2" Oct 05 07:39:26 morning Oct 05 07:39:50 now my logo ends-up in ${WORKDIR}/${LOGO_SIZE} Oct 05 07:40:04 but linux.inc searches in ${WORKDIR} Oct 05 07:40:10 ant_work: what is LOGO_SIZE defined to Oct 05 07:40:14 so, apart of copying it, Oct 05 07:40:25 how can I fetch it straight to WORKDIR? Oct 05 07:40:35 ah, is 'vga or qvga' Oct 05 07:40:43 is an 'override' Oct 05 07:42:42 on the rush, I copied the check in linux.inc and added it, modified, to do_configure_append Oct 05 07:43:07 but seems not elegant Oct 05 07:45:07 I briefly peeked in base.bbclass but could not find a way to have it ignore the subdir Oct 05 07:49:40 ant_work: file:// Oct 05 07:50:01 khem: base.bbclass Oct 05 07:50:04 if subdirs: Oct 05 07:50:06 destdir = oe.path.join(workdir, *subdirs) Oct 05 07:51:02 does it matter where it is ? Oct 05 07:51:35 somehow... Oct 05 07:51:46 linux.inc expect it in WORKDIR Oct 05 07:53:13 not a big issue, just to know if there are other similar cases Oct 05 07:58:51 I think we should extend the check in linux.inc Oct 05 09:13:14 jo Oct 05 09:13:48 good morning Oct 05 09:16:32 hm intressting thinks happen in autotoolsland Oct 05 09:16:42 seems aclocal will soon vanish Oct 05 09:23:47 03Martin Jansa  07master * r4f1139422c 10openembedded.git/recipes/navit/navit_svn.bb: navit: bump SRCREV for speedwarner feature Oct 05 09:27:29 jama hehe Oct 05 09:30:16 woglinde: well I was only waiting for chollya's fix Oct 05 09:30:41 and today someone even asked on ML to upgrade it :) Oct 05 09:37:44 I will integrate some patches too today Oct 05 09:42:07 woglinde: oh Oct 05 09:42:22 zecke? Oct 05 09:42:33 navit Oct 05 09:43:42 woglinde: why is aclocal going away? Oct 05 09:44:52 because its only a hack Oct 05 09:45:08 to determine additional directories for m4 Oct 05 09:45:22 but m4-2.0 will have this feature self Oct 05 09:45:45 * woglinde read this on the libtool-ml Oct 05 09:47:19 all started when the libiconv/gettext maintainer asked, why he get the "please use the M4_ADDITIONAL_DIR in configure.ac" Oct 05 09:47:29 messages Oct 05 09:48:35 woglinde: http://www.gnu.org/software/hello/manual/automake/Future-of-aclocal.html.. okay Oct 05 10:11:01 Is it correct that you don't have to download the OMAP Graphics SDK manually to compile libgles-omap3 anymore? I didn't at least, and I was just wondering if the comments in the recipes should be changed to reflect that. Oct 05 10:11:24 Unless some build _do_ require a manual download of course Oct 05 10:12:48 yes its correct Oct 05 10:13:05 since 4.0 or late 3. the url is open for everyone except iran Oct 05 10:15:32 d Oct 05 10:16:11 hey , i tried to use clutter compiling for my card , but i got error "unable to get gles headers" Oct 05 10:17:11 you used clutter to compile? Oct 05 10:17:28 I always thought one needs a compiler like gcc Oct 05 10:17:57 hmm Oct 05 10:18:17 i use arm-angstrom-linux-gnueabi-gcc Oct 05 10:18:19 maybee you meant Oct 05 10:18:34 you try to compile clutter or an application which uses clutter Oct 05 10:19:04 and when i try to use , i got error : configure error: Unable to locate reqiired GLES headers Oct 05 10:19:27 which machine? Oct 05 10:19:38 igepv2 Oct 05 10:20:03 how can i solve this problem ? i did bitbake omap3-libgles Oct 05 10:20:15 it successfully did all tasks Oct 05 10:24:56 please help Oct 05 10:29:15 hm, testing-next builds seem all to fail on building sqlite3-native: http://tinderbox.openembedded.net/public/logs/task/8402014.txt Oct 05 10:50:08 03Holger Hans Peter Freyther  07master * r3f3128c291 10openembedded.git/recipes/ncurses/ncurses_5.7.bb: Oct 05 10:50:08 ncurses_5.7.bb: Disable the rpath hack for the ncurses binaries Oct 05 10:50:08 Do not have a RPATH in the binaries to the sysroot. Oct 05 12:09:53 jo mickeyl Oct 05 12:15:42 hi kgilmer Oct 05 12:15:46 thats early Oct 05 12:20:14 hi woglinde Oct 05 12:20:18 yeah i'm at home Oct 05 12:20:54 awhile ago i asked you about openjdk recipes that didn't build shark, you asked why Oct 05 12:21:08 the reason is for hw platforms that don't support lvm but zero will work Oct 05 12:21:16 for example bug10 Oct 05 12:21:22 hm Oct 05 12:21:42 hm google tv Oct 05 12:21:49 :) Oct 05 12:22:37 kgilmer hm dont know offhand how to determine a plattform can not build lvm Oct 05 12:22:45 why ther bug10 dont compiles lvm Oct 05 12:22:50 arhgs Oct 05 12:23:01 llvm Oct 05 12:23:34 yeah i don't think the recipe should determine that, but if the recipes are more fine-grained to a specific jvm backend then the user can decide. Oct 05 12:24:14 llvm: some freakish error, i didn't get too deep into it. i can file a bugzilla if you want. Oct 05 12:25:23 i would like to (unofficially) support older bug hw with newer software releases if it's fairly easy to do so. Oct 05 12:28:37 hm sure mail or paste the error Oct 05 12:28:44 maybee I or xrandby can help out Oct 05 12:41:20 ok woglinde, will do when i'm in the office. Oct 05 12:41:20 thx Oct 05 12:50:47 hi jconnolly Oct 05 12:50:52 eFfeM_work: I quickly checked the bb file for sqlite3 native and that has a DEPENDS="tcl-native". When is tclsh needed exactly? Oct 05 12:51:23 janp thats fixed Oct 05 12:51:28 only native needs tcl Oct 05 12:51:59 woglinde: it is, but unfortunately an hour or so after the testing branch was tagged Oct 05 12:52:23 so testing fails, suggested already on ML to retag testing Oct 05 12:53:37 woglinde: so DEPENDS is not taken into account for native packages? Oct 05 12:53:47 *sigh* Oct 05 12:53:53 hi woglinde Oct 05 12:54:18 DEPENDS_append_virtclass-native = " tcl-native" Oct 05 12:55:53 woglinde: the remark from janp made me look at the commit with the fix: http://cgit.openembedded.org/cgit.cgi/openembedded/commit/?id=9f503dbde0f2f6e9d4a94f6ab4be1b04dfc99950 Oct 05 12:56:08 but the old text read -DEPENDS = "tcl-native readline ncurses" Oct 05 12:56:52 so i'm a little bit puzzled why sqlite3-native fails for me Oct 05 12:57:06 staging? Oct 05 12:57:33 http://cgit.openembedded.org/cgit.cgi/openembedded/commit/?id=9f503dbde0f2f6e9d4a94f6ab4be1b04dfc99950 was not optimal anyway Oct 05 12:57:53 http://cgit.openembedded.org/cgit.cgi/openembedded/commit/?id=9189f87363be2865404f7de70bcda10f8087053e Oct 05 12:59:21 staging? don't really understand what you mean with that Oct 05 12:59:41 * eFfeM_work is checking his logs Oct 05 12:59:43 eFfeM_work: sqlite3-native fails on customer site too, I'm checking on my side Oct 05 13:00:31 Running task 3453 of 5429, I'm reaching the error Oct 05 13:02:21 well, i was building the testing branch which had the DEPENDS above, but tcl-native is not build; this is a clean build Oct 05 13:03:14 here's my full log: http://tinderbox.openembedded.net/builds/92319/ Oct 05 13:04:36 or does the code from yesterday adds -native to it and do I for testing now get a depends on tcl-native-native ? (whihc ofc does not exist) Oct 05 13:08:50 eFfeM_work: mine looks dofferent http://tinderbox.openembedded.net/public/logs/task/8392573.txt Oct 05 13:09:21 mckoan: this is dev head? Oct 05 13:09:34 eFfeM_work: yes Oct 05 13:09:54 ok so it has the patches given by woglinde Oct 05 13:10:08 now I am testing myself Oct 05 13:11:56 woglinde is my assumption correct that the original code (in the testing branch) would lead to a dep on tcl-native-native (for sqlite3-native)? or do I completely misunderstand things ? Oct 05 13:12:15 * eFfeM_work then had expected an error that the dep could not be met Oct 05 13:12:47 * eFfeM_work is puzzled Oct 05 13:13:20 hm seems the append dotn works Oct 05 13:14:52 hey , when i try to do "bitbake clutter" oe says , nothing provides clutter but when i look recipes , there is clutter folder inside ? why it says nothing provides ? Oct 05 13:15:13 hm we need -i again Oct 05 13:15:14 JDuke128: the recipe is not for your machine Oct 05 13:16:04 hmm , i tried to compile myself also , configure works fine but when i tried to make , i see some /usr/lib files used on my host machine and gives errors Oct 05 13:17:30 wait i post them on pastebin please if someone help Oct 05 13:18:37 I am trying += now Oct 05 13:18:51 kergoth suggested this too yesterday Oct 05 13:19:19 args Oct 05 13:19:26 sqilte-native vs. sqlite3-natviwe Oct 05 13:20:33 tcl is in the depends Oct 05 13:20:38 as expected Oct 05 13:24:51 when i try to configure clutter it works fine as url shown : http://pastebin.com/X5m40G2d , but when i try to make it i got error : http://pastebin.com/PrEXZZ8M , someone know how to fix ? or whats wrong ? Oct 05 13:25:49 i suspect on "/usr/include" is used on make , CROSS COMPILE Badness: /usr/include in INCLUDEPATH: /usr/include/cairo , how to fix that problem ? Oct 05 13:26:35 JDuke127: how surprising, you run configure on the target and expect to get proper results for a cross build ???????? Oct 05 13:27:45 sorry , so how to fix ? Oct 05 13:29:15 use bitbake :-) Oct 05 13:29:19 and build the recipe Oct 05 13:29:42 bitbake -cclean clutter; bitbake clutter Oct 05 13:29:52 i did , but "bitbake clutter" , but system says nothing provides clutter Oct 05 13:30:20 (03:16:53 PM) eFfeM_work: JDuke128: the recipe is not for your machine Oct 05 13:30:23 i m using igepv2 , more advanced then beagleboard Oct 05 13:30:25 hm sqlite3-native build fine Oct 05 13:30:40 ERROR: Nothing PROVIDES 'clutter' Oct 05 13:30:55 woglinde: same here sqlite3-native build fine Oct 05 13:30:58 jduke i think the recipe depends on omap3 and igepv2 is not in that group Oct 05 13:31:13 no Oct 05 13:31:18 igepv2 is omap3 Oct 05 13:31:24 clutter.inc has COMPATIBLE_MACHINE = "(zylonite|mx31litekit|omap-3430ldp|omap-3430sdp|mx31ads|qemuarm|qemux86|ipodtouch|am3517-evm|dm37x-evm|am37x-evm|beagleboard|overo|omap3evm|omap5912osk)" Oct 05 13:31:24 woglinde: so the patch is fine Oct 05 13:31:25 woglinde: maybe customer has a corrupted fs Oct 05 13:31:33 i m %100 sure , igepv2 is omap3 Oct 05 13:31:40 add igep there Oct 05 13:31:40 JDuke127, i know, but it might not be defined as such Oct 05 13:31:44 *sigh* Oct 05 13:31:59 this search cost me 10 seconds Oct 05 13:32:08 and one beer for you Oct 05 13:32:40 so , i must compile ? Oct 05 13:32:55 configure works fine for igepv2 Oct 05 13:33:04 * eFfeM_work suggests starting to read Oct 05 13:33:41 JDuke127: how surprising, you run configure on the target and expect to get proper results for a cross build ???????? Oct 05 13:33:57 so , how can i fix that ? Oct 05 13:34:13 JDuke127 learn the difference between host and target Oct 05 13:34:43 take a reading course, we've already told you a few times what to do Oct 05 13:34:46 *sigh* Oct 05 13:35:39 --host=arm-angstrom-linux-gnueabi Oct 05 13:35:40 clutter.inc COMPATIBLE_MACHINE = "(zylonite|mx31litekit|omap-3430ldp|omap-3430sdp|mx31ads|qemuarm|qemux86|ipodtouch|am3517-evm|dm37x-evm|am37x-evm|beagleboard|overo|omap3evm|omap5912osk|igep0020) Oct 05 13:35:42 this ? Oct 05 13:35:48 now its 2 beers Oct 05 13:36:11 woglinde: you Oct 05 13:36:17 woglinde , so why not clutter is not working but other bitbake cmds working fine ? Oct 05 13:36:21 woglinde, you're not by car, are you ? Oct 05 13:36:34 dont know Oct 05 13:36:40 this was the last hint Oct 05 13:36:44 now mental ignore Oct 05 13:36:59 * eFfeM_work joins woglinde Oct 05 13:37:56 d Oct 05 13:38:34 good morning for those in the western hemi Oct 05 13:39:11 whats the proper way to migrate to a new version of gcc to my oe toolkit Oct 05 13:40:05 i tried moving from 4.4.2 to 4.4.4 ... it was kind of an epic failure Oct 05 13:40:07 we already have the newest version of gcc in Oct 05 13:40:36 right ... i am just asking how to clean or ... force it to replace the other version Oct 05 13:40:44 03Dmitry Eremin-Solenikov  07master * rd523823c53 10openembedded.git/recipes/libopie/ (libopie2/libopie2-tosa.patch libopie2_1.2.4.bb): Oct 05 13:40:44 libopie2: fix path to backlight instance on tosa Oct 05 13:40:44 Signed-off-by: Dmitry Eremin-Solenikov Oct 05 13:41:29 like on gentoo i just run the crossdev command and it emerges the packages in the right order .... and am getting configure: error: Link tests are not allowed after GCC_NO_EXECUTABLES. when building the final stage of the gcc Oct 05 13:41:53 whether i try 4.4.4 or 4.5 Oct 05 13:41:56 03Dmitry Eremin-Solenikov  07master * r78069a1982 10openembedded.git/recipes/qt4/ (qt4-tools-sdk.inc qt4-tools-sdk_4.6.3.bb): Oct 05 13:41:56 qt4-tools-sdk: fix building when no native freetype/X11 are installed Oct 05 13:41:56 Signed-off-by: Dmitry Eremin-Solenikov Oct 05 13:42:31 cargoudel make a clean build Oct 05 13:42:40 all other is nonsense Oct 05 13:43:13 really ... so i lose everything .. it took a while to build on my slow as machine Oct 05 13:43:37 well if thats the way to do it ... i can handle that Oct 05 13:44:01 cargoudel how do get all packages rebuild with gcc? Oct 05 13:44:59 oh i thought you mean clean ... everything Oct 05 13:45:55 thats just it ... i started by cleaning glibc and gcc-cross-initial gcc-cross-immediate and gcc-cross Oct 05 13:46:19 then i built and everything built expect for that last stage Oct 05 13:46:29 except* Oct 05 13:46:52 i search for this error configure: error: Link tests are not allowed after GCC_NO_EXECUTABLES. Oct 05 13:47:08 but in the cross compile word its like search for a seg fault error Oct 05 13:47:32 i get so many results ... and it means basically you complier is not working Oct 05 13:47:47 sure remove the oe-tmp dir Oct 05 13:47:51 und build all new Oct 05 13:48:15 good overnight usage of your system :-) Oct 05 13:49:26 so ... for recipes that all already packages removing tmp doesn't erase those? Oct 05 13:51:40 so this is interesting ... when i built it the first time it built gcc-cross-4.3.2+csl-arm-2008q3-39-r1 .... as part of the bootstrap process ... even though that wasn't my final version Oct 05 13:52:03 gcc 4.4.2 is required because i am eabi on armv4 Oct 05 13:53:06 did it build gcc-cross-4.4.2 with gcc-cross-4.3.2 ... i guess i can read the dep tree Oct 05 13:53:24 woglinde, http://pastebin.com/TP50yan2 Oct 05 13:53:48 dmitry? Oct 05 13:54:54 kgilmer bug1 was armv5? Oct 05 13:56:09 kgilmer one google search Oct 05 13:56:15 http://markmail.org/message/zm2dtvrzcdnugxv2#query:__clear_cache%20llvm+page:1+mid:cp23ozrqq4avt5pl+state:results Oct 05 13:57:19 so you're saying woglinde that changing the gcc version that bug machine uses should *just work*? Oct 05 13:57:21 woglinde: what is our revert strategy? the above Qt4 patch is rubbish... and it applies a patch that does not exist Oct 05 13:57:46 zecke pull a koen Oct 05 13:59:20 kgilmer when xrandby tell this sure Oct 05 14:01:15 i will try that, thanks woglinde Oct 05 14:01:42 kgilmer try a least one with the atomic patches Oct 05 14:02:09 thanks guys for the help ... i am still a oe-newb .... but i am learning fast Oct 05 14:03:33 woglinde: sent an email, but the build is now broken. :) Oct 05 14:03:42 not sure what you mean specifically woglinde Oct 05 14:07:17 kgilmer try latest gcc-4.3 Oct 05 14:07:22 or 4.4 Oct 05 14:07:25 or 4.5 Oct 05 14:07:28 they sould work Oct 05 14:07:46 zecke find the author and ask for a revert or find two tsc members for an ack for a revert Oct 05 14:08:13 eFfeM_work: googled that... i wonder how much google knows me that revert policy has OE at first hit Oct 05 14:08:29 :-) Oct 05 14:09:46 or as woglinde sayd "pull a koen" ;-) (if my interpretation of woglindes words is ok) Oct 05 14:19:56 mickey|office: live? Oct 05 14:30:01 mickey|office, hi, we need to talk, it has been several days that I'm trying to talk to you, so ping me when you'll have some time Oct 05 14:30:28 i'll have some time today in the evening, will ping you Oct 05 14:30:35 greetings! anyone know how to change udhcpc's default arguments? I'm grep'ing my entire filesystem and I can't find it Oct 05 14:30:55 ok thanks a lot Oct 05 14:31:09 hi kergoth Oct 05 14:31:34 I see /usr/share/udhcpc/default.script, which runs what's under /etc/udhcpc.d/, but no udhcpc arguments are there Oct 05 14:34:28 hey Oct 05 14:46:11 03Steffen Sledz  07org.openembedded.dev * re65bf047fe 10openembedded.git/recipes/linux/linux-2.6.24/hipox/defconfig: Oct 05 14:46:11 linux-2.6.24: build bitbash i2c and related modules for hipox machine Oct 05 14:46:11 Signed-off-by: Steffen Sledz Oct 05 14:46:13 03Steffen Sledz  07org.openembedded.dev * r7e8cfb36e7 10openembedded.git/recipes/linux/ (3 files in 2 dirs): Oct 05 14:46:13 linux-2.6.24: enable PoE board switching using gpo for hipox machine Oct 05 14:46:13 Signed-off-by: Steffen Sledz Oct 05 14:54:49 kergoth_: is there any way to fetch ignoring the subdir? e.g. SRC_URI = "file://${LOGO_SIZE}/logo_linux_clut224.ppm.bz2" in ${WORKDIR} and not in ${WORKDIR}/${LOGO_SIZE} Oct 05 14:55:48 PoE nice Oct 05 14:56:32 khem: ping Oct 05 15:00:23 ant_work: s/fetch/unpack/; don't think there's an option for that at the moment Oct 05 15:00:53 oops, right Oct 05 15:03:10 http://docs.openembedded.org/usermanual/html/src_uri_variable.html <-- I'm using SVN for my SRC_URI. My repository has a password. Is there a way to store a password that would allow bitbake to get the files or do I have to make an account that allows anonymous access? Oct 05 15:10:14 foobaz: you should, as far as i know, be able to specify it in the url Oct 05 15:12:06 kergoth_: about that subdir, it comes from LOGO_SIZE = '${@base_conditional("GUI_MACHINE_CLASS", "bigscreen", "vga", "qvga", d)}' Oct 05 15:12:17 we need to extend it for more sizes! Oct 05 15:12:46 okay, go ahead :) Oct 05 15:13:10 how many sizes should we support? is there an 'official' list? Oct 05 15:13:24 no idea Oct 05 15:13:31 i expect git grep is your friend here Oct 05 15:13:43 I mean outside ;) Oct 05 15:14:03 ok http://en.wikipedia.org/wiki/List_of_common_resolutions Oct 05 15:14:47 too many! Oct 05 15:14:50 well, i expect GUI_MACHINE_CLASS is intentionally less granular and more about categorization .. "small" vs "large" Oct 05 15:14:55 LOGO_SIZE on the other hand is more specific Oct 05 15:15:25 where's LOGO_SIZE defined, in that recipe? Oct 05 15:15:37 now is only in linux-kexecboot.inc Oct 05 15:15:46 but logo processing is done in linux.inc Oct 05 15:16:17 if [ -e ${WORKDIR}/logo_linux_clut224.ppm ]; then Oct 05 15:16:24 why not make it a ?=, so the distro can specify a larger one directly if a logo is in the repo for that size, rather than messing with the class Oct 05 15:16:30 ah Oct 05 15:16:40 er, s/distro/machine/ Oct 05 15:16:45 we play using the overrides Oct 05 15:16:59 i.e. angstrom has own vga and qga Oct 05 15:17:05 logos Oct 05 15:18:31 other distros pick the default OE logo Oct 05 15:20:11 kergoth_ I'm tempted setting LOGO_SIZE ?= "" in linux.inc and modify the test Oct 05 15:21:06 in if [ -e ${WORKDIR}/${LOGO_SIZE}/logo_linux_clut224.ppm ]; then Oct 05 15:24:15 yeah, that seems reasonable to me Oct 05 15:24:19 or do ?= "." in linux.inc Oct 05 15:24:26 "//" isn't supported everywhere Oct 05 15:24:42 heh Oct 05 15:24:55 iirc osx might be the one that chokes on it Oct 05 15:24:57 can't recall Oct 05 15:25:30 kergoth_: I don't believe SVN allows usernames and passwords in the URL unless I'm mistaken. Oct 05 15:25:55 foobaz: but the url is given to the fetcher, and the fetcher passes it on to svn. Oct 05 15:26:01 it's not unreasonable to think it could translate it. Oct 05 15:26:27 i don't know offhand if it does or not, but arguably not doing so is a bug in the fetcher, since authentication is an important requirement Oct 05 15:26:38 kergoth_: thx Oct 05 15:26:40 bbl Oct 05 15:26:48 i'd suggest taking a look at bitbake/lib/bb/fetch/svn.py Oct 05 15:26:50 (iirc) Oct 05 15:27:33 re Oct 05 15:27:40 hey woglinde Oct 05 15:29:37 Are you like a developer? I see where it builds the SVN command. hmm now just need to follow where it's getting everything :P Oct 05 15:34:08 foobaz grep in recipe directory is too hard? Oct 05 15:37:09 foobaz: yes, you could say i'm a developer Oct 05 15:39:52 03Dmitry Eremin-Solenikov  07master * r4847c2dd25 10openembedded.git/: Merge branch 'master' of git://git.openembedded.org/openembedded Oct 05 15:40:18 03Dmitry Eremin-Solenikov  07master * rf109fd1be1 10openembedded.git/recipes/qt4/ (3 files in 2 dirs): Oct 05 15:40:18 qt4-tools-sdk: correct my last commit: add a patch I've forgot Oct 05 15:40:18 Signed-off-by: Dmitry Eremin-Solenikov Oct 05 15:40:18 03Dmitry Eremin-Solenikov  07master * r4931b37c8f 10openembedded.git/recipes/qt4/qt4-tools-sdk.inc: Oct 05 15:40:18 qt4-tools-sdk.inc : drop freetype include as we build with -no-freetype Oct 05 15:40:18 Signed-off-by: Dmitry Eremin-Solenikov Oct 05 15:40:48 foobaz: Oct 05 15:40:48 if ud.user: Oct 05 15:40:48 options.append("--username %s" % ud.user) Oct 05 15:40:48 if ud.pswd: Oct 05 15:40:49 options.append("--password %s" % ud.pswd) Oct 05 15:40:51 woglinde, I'm not sure what you mean. I found ud.pswd in the _buildsvncommand method which is called by go, but that isn't where ud is set. I assume there's a parser. Oct 05 15:41:03 foobaz: ud.user and ud.pswd are the username and password components of the url, as i had pointed out Oct 05 15:41:06 kergoth_: my thoughts exactly Oct 05 15:41:11 so again, put it in the url :) Oct 05 15:41:57 how? What's the format? That just says how the SVN command is built. It doesn't say what format ud.pswd was in Oct 05 15:42:46 ;user=username;password=pass? at the end of the URL? Oct 05 15:43:03 zecke: o_O How did you find that information? Oct 05 15:43:28 That's not in the documentation. Oct 05 15:45:07 foobaz: it is a guess. Oct 05 15:45:17 foobaz: feel encouraged to fix the bitbake manual Oct 05 15:45:22 foobaz: uh Oct 05 15:45:24 its a url. Oct 05 15:45:31 how about using a username and password the standard way you do for a url? Oct 05 15:45:38 svn://foo:bar@baz/ Oct 05 15:45:58 I thought that didn't work for SVN? Oct 05 15:46:01 looks good too Oct 05 15:46:01 * foobaz tries Oct 05 15:46:08 how many times do i need to explain this? Oct 05 15:46:17 that url isn't passed *as is* to svn Oct 05 15:46:21 the url goes into the fetcher Oct 05 15:46:26 the fetcher decides what to pass to svn Oct 05 15:46:38 not that complicated Oct 05 15:46:45 so i don't really give a shit what svn supports Oct 05 15:46:50 the fetcher translates to svn arguments Oct 05 15:46:54 as i pasted above Oct 05 15:47:00 it converts to --username and --password Oct 05 15:47:22 kergoth isnt it a URI with username and password? Oct 05 15:47:33 08:39 < kergoth_> foobaz: ud.user and ud.pswd are the username and password components of the url, as i had pointed out Oct 05 15:47:42 makes sense. I was looking for that in the source. :P Never used python before though. Oct 05 15:47:50 components of the url, not parameters of the url Oct 05 15:48:01 I assumed there was a parser that would tell me the correct format Oct 05 15:48:04 that has nothing to do with python really Oct 05 15:48:08 this isn't about python, it splits urls as per the RFCs Oct 05 15:48:13 (or should be, anyway) Oct 05 15:48:22 Silly question. FILES globs are shell globs not regex so ${base_libdir}/libgcc*.so should catch /lib/libgcc_s.so, yes? Oct 05 15:48:28 Tartarus: yep Oct 05 15:48:35 And libgcc*.so.* /lib/libgcc_s.so.1 Oct 05 15:48:40 So why isn't it, dang it.. Oct 05 15:48:59 http://en.wikipedia.org/wiki/Uniform_Resource_Locator Oct 05 15:49:04 package ordering issue? not sure it allows you to include things in multiple packages today Oct 05 15:49:11 it being picked up by an earlier package? Oct 05 15:49:19 worse, nothing catches it Oct 05 15:49:23 weird Oct 05 15:49:47 verify your FILES var is being set the way you think it is, and isn't being overwritten by a variable name expansion or override? Oct 05 15:49:52 bitbake -e it Oct 05 15:51:05 # FILES_libgcc=${base_libdir}/libgcc*.so.* Oct 05 15:51:06 FILES_libgcc="/lib/libgcc*.so.*" Oct 05 15:51:21 strange, looks just fine Oct 05 15:51:22 # FILES_libgcc-dev=${base_libdir}/libgcc*.so Oct 05 15:51:22 FILES_libgcc-dev="/lib/libgcc*.so" Oct 05 15:51:24 yeah Oct 05 15:51:35 * Tartarus waits for one more build to finish Oct 05 15:51:54 * kergoth_ hopes next week is a bit better than this one Oct 05 15:54:54 if someone interested in migrating clutter to javafx , let me know. Oct 05 16:20:03 Is this a bug: http://pastebin.com/efd4Aw8M I have my bb file included and the output when I run bitbake on it. I'm cross compiling for angstrom, but why is it trying to use my username with angstrom to look for my source files? Oct 05 16:21:13 you dont need the second install -d ${D}${base_bindir} Oct 05 16:21:16 yeah, that's a bug in the mirror handling Oct 05 16:21:29 it shouldn't be bringing username/pass across to the mirrors Oct 05 16:21:45 if you would, email the bitbake/oe lists about it so i dont' forget to look into it Oct 05 16:24:18 hm mesa 7.9 is out Oct 05 16:27:11 * kergoth_ thinks about adding 'target' to overrides when its a target recipe.. Oct 05 16:27:14 foobaz: oh, by default we hit a TARBALL_STASH. E.g. to put load from servers to fetch certain revs... we might have them as a tarball on a server Oct 05 16:27:19 * kergoth_ .. in the default overrides, he means Oct 05 16:27:27 03Chris Larson  07master * r17579c48cd 10openembedded.git/ (3 files in 2 dirs): Oct 05 16:27:27 Adjust rdep/rrec for gtk+* and gtk-icon-cache.bbclass Oct 05 16:27:27 Drop the runtime dependencies / recommendations unless the recipe is for the Oct 05 16:27:27 target. This is necessary because right now bitbake obeys *_${PN}, even if Oct 05 16:27:27 ${PN} is not in PACKAGES. This should kill the unnecessary build of the Oct 05 16:27:27 crosscompiler when doing 'bitbake gtk+-native'. Oct 05 16:27:28 Signed-off-by: Chris Larson Oct 05 16:27:34 foobaz: it tries to download them... I see how that is bad for your user/password.. :} Oct 05 16:27:35 zecke: yes, but the problem is its using teh username/password from the original url in it Oct 05 16:27:41 * kergoth_ nods Oct 05 16:28:06 either a bug in the patterns used in our MIRRORS/PREMIRRORS, or the mirrors handling in bitbake needs to know to strip those automatically Oct 05 16:28:41 foobaz: the easiest way to disable it, could be to set CVS_TARBALL_STASH="" (not sure if it works) Oct 05 16:29:14 kergoth_: something like ;private_credentials... could make sense Oct 05 16:30:08 it doesn't make much sense to bring them across in general, the likelihood of the mirror sharing the same authentication scheme is low i would think -- maybe a parameter to flip it the other way around, ;mirror_auth=1 Oct 05 16:30:12 hrm Oct 05 16:31:08 true Oct 05 16:31:16 * zecke hunts some food Oct 05 16:32:13 is it safe to delete stuff in build/tmp/work .... or will that cause packages to be rebuilt ? Oct 05 16:32:20 INHERIT += "rm_work" Oct 05 16:32:25 will do it for you when the recipe is complete Oct 05 16:32:34 and yes, its safe as long as those recipes are done Oct 05 16:32:41 k thanks Oct 05 16:33:57 oh cool. Thanks for that hint about the tarball stash. Did you mean SVN_TARBALL_STASH? Oct 05 16:34:28 Also I don't really know how to use mailing lists. I have in the past with great difficulty. Might want to write it down. :\ Oct 05 16:34:39 iirc there's an SCM_TARBALL_STASH or something, but falls bcak to CVS_TARBALL_STASH even for non-cvs for compatibility Oct 05 16:37:38 kergoth_: is there a documentation page for local.conf with this information? Oct 05 16:38:05 it isn't specific to local.conf Oct 05 16:38:22 should be in the OE manual, not sure if it is, but.. Oct 05 16:39:22 oh I see it Oct 05 16:42:58 kergoth_: wait the CVS_TARBALL_STASH would go into the local.conf or is this also in the URI. I notice the documentation refers to them as variables and not options. Oct 05 16:43:09 how can i add new include h files ? "bitbake clutter" automatically adds clutter libraries to my arm gcc compiler or i need to add it manually ? if yes , how from ipk files on host machine? Oct 05 16:43:31 foobaz: its a variable, so it goes in a .conf Oct 05 16:43:38 JDuke127: ? Oct 05 16:43:45 JDuke127: if a recipe requires clutter, it should depend on clutter Oct 05 16:43:52 and the files will be made available for it Oct 05 16:44:20 no i ll write C++ code Oct 05 16:44:52 http://pastebin.com/MdXCRrqV Oct 05 16:45:06 but i dont have include libraries when i look include folder Oct 05 16:45:12 "/home/kadirbasol/OE/build/tmp-igep0020/sysroots/armv7a-angstrom-linux-gnueabi/usr/include" Oct 05 16:45:37 this doesnt have any clutter.h but i did "bitbake clutter" Oct 05 16:45:50 bitbake doesnt automatically add header files on bitbake ? Oct 05 16:45:58 how does oe determine what compiler to compile gcc-cross with Oct 05 16:46:05 you shouldn't be using the sysroot outside of OE. Oct 05 16:46:09 cargoudel: ? Oct 05 16:46:17 cargoudel: it doesn't. gcc's configure script does Oct 05 16:46:30 JDuke127: if you want a toolchain to use outside of oe, you should be using meta-toolchain Oct 05 16:47:09 i try helloworld code now Oct 05 16:47:58 hmmmm ... wild ... i am used to setting a gcc .. i'll have to watch the configure carefully Oct 05 16:49:17 hello.c:7: error: ‘ClutterColor’ undeclared (first use in this function ) Oct 05 16:49:30 arm-angstrom-linux-gnueabi-gcc hello.c Oct 05 16:51:13 JDuke127: any earlier warning? missing header files? Oct 05 16:51:27 hello.c:1:29: error: clutter/clutter.h: No such file or directory Oct 05 16:51:30 yes sorry Oct 05 16:51:44 how can i add these includes ? Oct 05 16:51:53 and static libraries Oct 05 16:53:16 JDuke127: bitbake devshell, then source the resulting file, then use pkg-config --cflags NameOfTheClutterModule (e.g. clutter-1.0... but i amnot sure) Oct 05 16:53:59 arm-angstrom-linux-gnueabi-gcc hello.c `pkg-config clutter-1.0 --cflags --libs` Oct 05 16:54:09 ok i try now i didnt know what devshell is Oct 05 16:58:09 i did Oct 05 17:00:50 but i m having same error Oct 05 17:01:07 arm-angstrom-linux-gnueabi-gcc hello.c `pkg-config clutter-1.0 --cflags --libs` Package clutter-1.0 was not found in the pkg-config search path. Oct 05 17:25:17 zecke: About that tarball. I added the CVS_TARBALL_STASH="" and reran bitbake. http://pastebin.com/AMxxWF3j Line 32 is of interest. Does that mean it's still looking for only a tarball and not trying to do a checkout for the source files? I don't have direct access to the SVN server so I don't think I can configure it to make tarballs unless it's a per repository thing I can quickly setup after every commit or something. Oct 05 17:28:31 foobaz: well, read the output. Oct 05 17:31:07 Tartarus: hello Oct 05 17:31:10 gm to all Oct 05 17:31:36 i get a internal gcc error if compiling mplayer for the palmpre (shr): http://pastebin.com/2NSNyq37 Oct 05 17:32:28 khem, did you push the linaro patches over the weekend? Oct 05 17:35:00 JDuke127: if that's not a recipe, its not expected to work. like i told you before, using sysroot outside of oe is a bad idea, and you're on your own with that Oct 05 17:35:29 so , whats outside of oe mean ? Oct 05 17:35:47 how can i go inside oe Oct 05 17:36:25 docbook-sgml .. native is failing for me Oct 05 17:38:14 inside oe means , its something like my x86 opens an arm based virtual console and write everything there ? Oct 05 17:38:25 like an emulator ? Oct 05 17:38:27 hi Oct 05 17:38:30 hi Oct 05 17:38:45 kergoth_ , ? Oct 05 17:39:55 I have problem with XDG menus, e-wm doesnt install /etc/xdg/menus/applications.menu Oct 05 17:40:07 | install-catalog: "/home/balister/oe/tmp/sysroots/x86_64-linux/etc/sgml/sgml-docbook.cat" does not exist or is empty. Oct 05 17:40:13 this ring a bell with anyone Oct 05 17:40:26 so I dont have any applications in Enlightenment menu Oct 05 17:41:03 and I have xdg-utils already installed Oct 05 17:42:04 JDuke127: no, it means creating a recipe rather than trying to use its files externally Oct 05 17:42:15 zecke: can I not fetch when a certificate is expired? Oct 05 17:42:19 * kergoth_ grumbles Oct 05 17:42:35 ahh for C++ projects also i need to make recipes ? Oct 05 17:42:56 again, if you want to use oe's toolchain outside of OE, you should use meta-toolchain to construct an sdk for you to use Oct 05 17:43:22 meta-toolchain Oct 05 17:43:33 yes, that's what i said Oct 05 17:43:37 which one you recommend most ? Oct 05 17:43:42 what? Oct 05 17:43:46 meta-toolchain is a recipe in oe Oct 05 17:43:52 bitbake meta-toolchain Oct 05 17:44:05 there's a variable you can use to add your own things to it, which you'll want to use for clutter Oct 05 17:44:51 how can i go inside OE ? by devshell ? Oct 05 17:45:01 bitbake -c devshell clutter Oct 05 17:45:01 no, you're misunderstanding Oct 05 17:45:03 like this ? Oct 05 17:45:09 by "inside of OE" I mean building with oe, using a recipe Oct 05 17:45:19 * kergoth_ gives up Oct 05 17:45:35 hmm Oct 05 17:46:01 btw , i successfully compiled with "bitbake -c devshell clutter" it opened me new console Oct 05 17:46:13 arm-angstrom-linux-gnueabi-gcc-4.3.3 hello.c `pkg-config clutter --cflags --libs` -o test Oct 05 17:46:19 and it gave me test file Oct 05 17:46:28 successfully Oct 05 17:46:37 Crofton: yes Oct 05 17:47:43 yeah Oct 05 17:47:52 I just saw linaro in patches Oct 05 17:47:59 playya__: how old is your OE snapshot Oct 05 17:48:31 hrm Oct 05 17:49:41 zecke: yeah I have no idea. I'm going to go have them refresh their certificate and get back to this problem. Oct 05 17:49:52 Thanks for your help today kergoth_ and zecke :) Oct 05 17:49:56 np Oct 05 17:55:33 hey khem Oct 05 17:55:47 Tartarus:hello Oct 05 17:56:14 khem, do you know why we make 'gcc' have a symlink to libgcc_s.so ? Oct 05 17:56:23 And also why we don't ship it? Oct 05 17:56:45 (4.4.1 at least, qemuppc) Oct 05 17:57:11 Tartarus: target gcc Oct 05 17:57:38 Yes Oct 05 17:57:48 Triggers a QA error we don't skip (easy) Oct 05 17:57:54 and none of the target packages at least ship it Oct 05 17:58:01 is it supposed to be coming in from elsewhere? Oct 05 17:58:36 Tartarus: yeah gcc-cross Oct 05 17:59:02 khem, yesterday Oct 05 17:59:06 I think should be in -dev Oct 05 17:59:09 playya__: ok Oct 05 17:59:34 playya__: can you preprocess this one file Oct 05 17:59:39 and send over to me Oct 05 18:00:18 i just clean gcc and glibc and pulled HEAD to check if there were any updates Oct 05 18:01:02 khem, if you tell me how I'll send you the file Oct 05 18:01:19 playya__: add -E option to the failing commandline Oct 05 18:01:33 ok Oct 05 18:01:33 that should generated the intermediate file I need Oct 05 18:01:53 if you have -o then this file will be dumped into that file Oct 05 18:01:59 otherwise to stdio Oct 05 18:03:38 NOTE: package gtk+-native-2.20.1-r10.3: task do_configure: Failed Oct 05 18:03:40 hmm Oct 05 18:03:43 what is up Oct 05 18:04:28 hmmm modules are missing from my rfs Oct 05 18:04:32 what changed ? Oct 05 18:05:00 I have the .tgz but usually it was also part of the rfs image Oct 05 18:06:01 03Michael 'Mickey' Lauer  07org.openembedded.dev * rdb91b3f33f 10openembedded.git/classes/vala.bbclass: vala.bbclass: export XDG_DATA_DIRS to point upstream vala to where the vapis are Oct 05 18:06:12 03Michael 'Mickey' Lauer  07org.openembedded.dev * r64a53b0ab6 10openembedded.git/recipes/libgee/libgee_0.6.0.bb: libgee 0.6.0: depends on gobject-introspection Oct 05 18:06:18 03Michael 'Mickey' Lauer  07org.openembedded.dev * r9b47372447 10openembedded.git/recipes/vala/vala_0.11.0.bb: Oct 05 18:06:18 vala: add 0.11.0 Oct 05 18:06:18 NOTE: Finally, we can now use upstream vala Oct 05 18:06:26 03Michael 'Mickey' Lauer  07org.openembedded.dev * r93169ecb3c 10openembedded.git/recipes/libgee/ (libgee_0.5.2.bb libgee_0.6.0.bb): libgee: 0.5.2 -> 0.6.0 Oct 05 18:06:35 03Michael 'Mickey' Lauer  07org.openembedded.dev * r14318872c0 10openembedded.git/recipes/gobject-introspection/gobject-introspection_0.9.10.bb: gobject-introspection: new recipe; introspection-library for gobject Oct 05 18:11:07 oh. nice. upstream vala Oct 05 18:11:59 kergoth, | configure: error: *** libX11 not found. Check 'config.log' for more details. Oct 05 18:11:59 NOTE: package gtk+-native-2.20.1-r10.3: task do_configure: Failed Oct 05 18:14:28 no idea Oct 05 18:14:40 is libx11-native in its DEPENDS? Oct 05 18:14:42 you pushed something today .. Oct 05 18:14:52 checking Oct 05 18:15:06 i pushed something which fixes it to stop building hte entire crosscompiler to package gtk+-native Oct 05 18:15:24 if it doesn't work now, the recipe was always broken, and we just never realized it Oct 05 18:15:27 does not look like Oct 05 18:15:38 * kergoth looks into it Oct 05 18:15:51 it uses BBCLASSEXTEND Oct 05 18:15:54 i know Oct 05 18:16:03 so add libX11 to depends? Oct 05 18:16:29 this machine uses lots of BBTHREADS Oct 05 18:16:37 khem, still compiling. should i recompile gcc without the linaro patch and check if it works? Oct 05 18:16:37 yeah, if it needs it, should go there Oct 05 18:16:41 it is pretty good at finding missing depnds :) Oct 05 18:17:05 let me try bitbake libX11-ntaice Oct 05 18:17:08 -native Oct 05 18:17:14 then build it again Oct 05 18:17:26 if that fixes it, I'll add libX11 to the DEPDNS Oct 05 18:18:45 hmm Oct 05 18:18:50 no libX11-native Oct 05 18:20:11 Crofton: libx11 vs. libX11? Oct 05 18:20:33 er Oct 05 18:20:36 which is correct Oct 05 18:20:56 checking the lower case option now Oct 05 18:21:26 ah Oct 05 18:21:50 libx11-native exists Oct 05 18:24:07 still broken Oct 05 18:24:59 hmm Oct 05 18:25:12 it exists in the x86 sysroot Oct 05 18:26:28 playya__: no I think I know the problem just provide the preprocessed file Oct 05 18:27:33 Crofton|work: it's really sad how much stuff just happens to work by accident today.. Oct 05 18:27:39 yeah Oct 05 18:27:44 trying a clean and rebuild Oct 05 18:27:57 * kergoth kicked off a fresh rebuild in a minimal chroot too Oct 05 18:27:58 same thing Oct 05 18:28:01 crap Oct 05 18:28:12 I can find the lib in the sysroot Oct 05 18:28:14 that doesn't make any sense :( Oct 05 18:29:14 i need to figure out how to fix the runtime dependency handling for natives Oct 05 18:29:20 khem, do you mean -e ? Oct 05 18:30:48 http://pastebin.com/3fYPuUx8 Oct 05 18:30:56 there is the config.log of the failure Oct 05 18:31:19 I am tempted to nuke tmp and start over Oct 05 18:31:35 huh Oct 05 18:31:46 is there a shared libx11 in the sysroot? Oct 05 18:31:57 notice its using the static, and statics can't link one lib against another Oct 05 18:32:05 the info is lost, would have to explicitly link to xau too Oct 05 18:32:29 ../tmp/sysroots/x86_64-linux/usr/lib/libX11.so Oct 05 18:32:50 I am thinking that I need to remove tmp and start over Oct 05 18:33:25 maybe the .so is unusabble. it should be using the .so by default unless there's a -Bstatic, afaik Oct 05 18:33:49 I am worried I have pieces bit with coipiler from pre linaro patches Oct 05 18:34:15 khem, my bitbake is up-to-date, but there's neither -E nor -o Oct 05 18:34:27 I am goign that route Oct 05 18:34:35 bb in a couple of hours :) Oct 05 18:36:04 http://www.angstrom-distribution.org/feeds/unstable/ipk/glibc/armv7a/base/libpixman-1-0_0.19.4-r5.0.6_armv7a.ipk Oct 05 18:36:11 this file is lost Oct 05 18:36:20 404 Oct 05 18:36:25 on opkg Oct 05 18:47:05 sigh playya__ dont worry I have the testcase here Oct 05 18:47:13 ok Oct 05 18:47:26 playya__: watchout for gcc upgrade later today try it and report back if it fixed your issue Oct 05 18:47:41 ok Oct 05 18:50:30 how can i add new library to meta-toolchain ? Oct 05 18:54:02 Tartarus: should we have the LIBTOOL_HAS_SYSROOT as a global variable ? Oct 05 18:54:16 Tartarus: I was thinking of defining it in bitbake.conf Oct 05 18:54:41 i.e. defaults to 'no' Oct 05 18:54:45 Sounds like a plan Oct 05 18:55:11 and then distros can set it to yes along with libtool versions Oct 05 18:55:18 pinned in distro.conf Oct 05 18:55:27 i'd think that would be better placed in autotools.bbclass or something, not everything uses libtool Oct 05 18:55:30 but /shrug Oct 05 18:56:03 kergoth: hmmm Oct 05 18:56:10 semantically yes Oct 05 18:56:27 just do a ?= in the class, distro can still override it Oct 05 18:56:34 yes Oct 05 18:57:09 kergoth: there is no way to set it automatically if one choses libtool 2.4 Oct 05 18:57:27 ah, right, if you need to set version prefs based on it, class won't do Oct 05 18:57:28 bitbake is does not know versions early enough to affect class parsing Oct 05 18:57:35 man, we have way too many global variables :) Oct 05 18:58:23 kergoth: as I think it should go away in future I will just do minimal Oct 05 18:58:50 and put it in autotools.bbclass ?= Oct 05 18:58:55 and use it there Oct 05 18:59:19 well, depends on what you're using it for. what will be done based on it? anything? or is it just a distro setting to affect class behavior? Oct 05 18:59:21 hmmm there are LDFLAGS_TARGET in bibtake.conf which should also use this var Oct 05 19:00:07 ideally its just DISTRO setting but also affects some global vars like LDFLAGS Oct 05 19:00:29 ah, in that case it would have to be global Oct 05 19:00:40 at least its temporary, indeed Oct 05 19:01:02 yeah I am not happy to add clutter but there need to be backward compatibility Oct 05 19:01:21 * kergoth nods, agreed Oct 05 19:01:47 Let me put it in bitbake.conf I think that way I have better control Oct 05 19:01:51 its nice to move default values to where they're used in cases where its possible. its not possible here. Oct 05 19:02:00 yes Oct 05 19:02:07 :) Oct 05 19:03:17 03Martin Jansa  07master * r1140d80172 10openembedded.git/recipes/vala/vala_0.11.0.bb: Oct 05 19:03:17 vala_0.11.0: fix checksums declaration Oct 05 19:03:17 * checksums were OK, but declared with name=archive which wasn't used in Oct 05 19:03:17 SRC_URI this time Oct 05 19:03:17 Signed-off-by: Martin Jansa Oct 05 19:17:50 03Martin Jansa  07master * rb558852f0a 10openembedded.git/recipes/qi/qi-ubi_git.bb: Oct 05 19:17:50 qi-ubi: remove PROVIDES it does provide virtual/bootloader as qi does Oct 05 19:17:50 * but it's ok to build both and user will download just the one he needs Oct 05 19:49:43 mickeyl, you there? Oct 05 19:50:04 kergoth, ? Oct 05 19:50:35 yo Oct 05 19:51:07 03Michael 'Mickey' Lauer  07org.openembedded.dev * r21f7b77529 10openembedded.git/recipes/libgee/libgee_0.5.2.bb: libgee: bring back 0.5.2; wasn't supposed to be removed Oct 05 19:51:07 03Scott Anderson  07org.openembedded.dev * rb3da649ee2 10openembedded.git/recipes/ipkg-utils/ (2 files in 2 dirs): (log message trimmed) Oct 05 19:51:07 ipkg-utils: Make arfile.py handle six digit UIDs Oct 05 19:51:07 Essentially, the problem is that arfile.py is splitting the ar header with Oct 05 19:51:07 white-space instead of fixed-width fields, so two fields would get treated Oct 05 19:51:08 as a single field. This makes things better than before as it now honors Oct 05 19:51:08 the fixed field widths. Oct 05 19:51:09 Signed-off-by: Scott Anderson Oct 05 19:51:43 mickeyl, just want your input on this patch http://pastebin.org/31831 Oct 05 19:52:25 mickeyl, its needed since eglibc has created forced eabi on later versions Oct 05 19:52:52 other than that i wouldn't compare strings with 'is' it looks good to me, but please run it through the toolchain experts Oct 05 19:53:54 boom :) Oct 05 19:54:09 so, where are we? Oct 05 19:54:10 mickeyl, who should I talk with? Oct 05 19:54:36 kergoth, khem and pb_ Oct 05 19:54:50 mickeyl, roger thx Oct 05 19:54:54 np Oct 05 19:55:23 hmmm Oct 05 19:55:43 kristoffer: what problem are you trying to solve Oct 05 19:57:00 loadammo: I am actually Oct 05 19:57:04 damn! Oct 05 19:57:11 :D Oct 05 19:57:13 khem: I am the one trying to solve Oct 05 19:58:08 khem: trying to dodge this: http://sourceware.org/bugzilla/show_bug.cgi?id=12097 Oct 05 19:59:11 khem, suggested patch by filip is at patch http://pastebin.org/31831 Oct 05 20:00:15 khem,binutils is not building for me Oct 05 20:01:26 filip: you can simply add PREFERRED_VERSION_eglibc_armv4 = "2.11" Oct 05 20:01:34 you dont need that python-foo Oct 05 20:01:51 filip: I have to look into the issue it seems interesting Oct 05 20:02:08 khem: well... 2.12 is fine for OABI Oct 05 20:02:14 khem: we need 2.11 for EABI only Oct 05 20:02:32 khem: so PREFERRED_VERSION_eglibc_armv4 = "2.11" would be too broad Oct 05 20:02:34 khem, http://pastebin.com/gHrd1eg3 Oct 05 20:03:01 filip: I know its ok to use 2.11 for both it makes it clear Oct 05 20:03:16 secondly its a workaround unless we figure out the solution Oct 05 20:03:25 there is no need to over complicate it Oct 05 20:03:29 Crofton|work: hmm Oct 05 20:03:47 khem: ok then, I'll just set PREFERRED_VERSION_eglibc_armv4 = "2.11" Oct 05 20:03:51 khem: thanks for the suggestion :) Oct 05 20:04:28 Crofton|work: cc1: warnings being treated as errors Oct 05 20:04:40 whats your host gcc version Oct 05 20:05:34 filip: ok Oct 05 20:06:25 filip: that bug needs to be addressed Oct 05 20:06:30 in glibc Oct 05 20:06:56 gcc version 4.4.4 20100630 (Red Hat 4.4.4-10) (GCC) Oct 05 20:08:08 khem, thx for feedback Oct 05 20:11:43 kristoffer filip btw I have a fix for you Oct 05 20:14:44 khem, fixes are always nice :) pastebin? Oct 05 20:14:56 khem: well I've reported it Oct 05 20:15:05 let me do it Oct 05 20:16:28 khem: unfortunately I don't have the knowledge to fix it Oct 05 20:18:17 filip: http://pastebin.com/KiMhk4hx Oct 05 20:18:20 try that fix Oct 05 20:18:25 and it should work Oct 05 20:27:38 khem, any thoughts on my problem? Oct 05 20:27:41 XorA|gone: ping Oct 05 20:28:07 Crofton|work: I asked your host gcc version Oct 05 20:28:46 filip: I attached the fix to the glibc bugzilla too for record Oct 05 20:28:48 gcc version 4.4.4 20100630 (Red Hat 4.4.4-10) (GCC) Oct 05 20:28:54 k Oct 05 20:28:55 :) Oct 05 20:29:10 Crofton|work: and its binutils-cross right Oct 05 20:30:01 khem: should I add it to the (e)glibc recipes? Oct 05 20:30:18 | make[1]: Leaving directory `/home/balister/oe/tmp/work/armv7a-angstrom-linux-gnueabi/binutils-2.20.1-r10.2/binutils-2.20.1/build.arm-angstrom-linux-gnueabi.arm-angstrom-linux-gnueabi' Oct 05 20:30:32 I am building a toolchain to run on the board Oct 05 20:31:06 Crofton|work: ok its target binutils Oct 05 20:31:41 yeah Oct 05 20:32:05 it worked last week Oct 05 20:33:45 Crofton|work: try this fix http://pastebin.com/xY4dgKDU Oct 05 20:34:07 Crofton|work: yes because last week probably you were not using gcc 4.5 :) Oct 05 20:34:07 03Michael 'Mickey' Lauer  07org.openembedded.dev * r64624b9cdf 10openembedded.git/recipes/libgee/libgee_0.6.0.bb: libgee 0.6.0: gobject-introspection-native seems to be enough to make it compile Oct 05 20:34:09 03Michael 'Mickey' Lauer  07org.openembedded.dev * r6e37929e8d 10openembedded.git/recipes/gobject-introspection/gobject-introspection_0.9.10.bb: gobject-introspection: add WIP NOTE Oct 05 20:34:14 are you using gcc 4.5 now ? Oct 05 20:35:02 cafe will close at 2 gotta go and fetch some food Oct 05 20:35:16 yes Oct 05 20:35:21 angstrom-2010.x Oct 05 20:39:29 03Michael 'Mickey' Lauer  07org.openembedded.dev * r3d476c259a 10openembedded.git/recipes/libgee/libgee_0.6.0.bb: libgee: darn... typo. sorry. Oct 05 20:41:05 khem, next issue http://pastebin.com/Pca6sqWC Oct 05 20:41:18 | libudev/libudev-queue.c:185:1: internal compiler error: in cond_exec_process_insns, at ifcvt.c:273 Oct 05 20:41:33 Crofton|work: known Oct 05 20:41:41 ok ... Oct 05 20:41:45 any fix? Oct 05 20:41:47 Crofton|work: use CFLAGS += " -Os" Oct 05 20:41:52 for that recipe Oct 05 20:43:21 ok, only needed there? Oct 05 20:43:31 yep Oct 05 20:43:48 there are couple of other recipes rarely used Oct 05 20:43:53 add it to them if you see same OCE Oct 05 20:43:54 ICE Oct 05 21:25:13 03Khem Raj  07master * r607ae31d3e 10openembedded.git/recipes/gcc/ (gcc-4.5.inc gcc-4.5/gcc-scalar-widening-pr45847.patch): Oct 05 21:25:13 gcc-4.5: Fix ICE found while compiling mplayer Oct 05 21:25:13 Signed-off-by: Khem Raj Oct 05 21:32:47 mickey|bbl: pong Oct 05 21:35:54 mickey|bbl: if its about the gobject introsprection stuff the answer will be "I have no idea" :-) Oct 05 21:39:22 Um, hi Oct 05 21:39:55 What tzdata package do I install for whole coverage? Oct 05 21:40:34 Even if I just wanted America and Europe I cannot install them both Oct 05 21:48:00 Why do tzdata packages conflict with each other when they don't share any files? Oct 05 21:49:03 Ah no I am a moron Oct 05 21:49:06 It's timezones Oct 05 22:33:09 03Tom Rini  07org.openembedded.dev * r8377b8ec57 10openembedded.git/classes/distutils-base.bbclass: Oct 05 22:33:09 distutils-base: Only RDEPEND python-core on target packages Oct 05 22:33:09 Signed-off-by: Tom Rini Oct 05 22:51:21 03Tom Rini  07org.openembedded.dev * r760e489ed4 10openembedded.git/recipes/shasum/shasum.inc: Oct 05 22:51:22 shasum.inc: NATIVE_INSTALL_WORKS = "1" Oct 05 22:51:22 Signed-off-by: Tom Rini Oct 05 23:00:41 khem, http://pastebin.com/u5mSPgJK Oct 05 23:01:29 after cleaning gcc gcc-cross gcc-cross-initial gcc-cross-intermediate Oct 05 23:11:07 i think the patch isn't compatible Oct 05 23:16:07 playya__: hmm thx Oct 05 23:16:13 committed a corrected one Oct 05 23:19:08 03Khem Raj  07master * r4e3f2321db 10openembedded.git/recipes/gcc/gcc-4.5/gcc-scalar-widening-pr45847.patch: Oct 05 23:19:08 gcc-4.5: Fix gcc-scalar-widening-pr45847.patch Oct 05 23:19:08 Signed-off-by: Khem Raj Oct 05 23:19:19 03Khem Raj  07master * rc3b2d9fac1 10openembedded.git/recipes/binutils/binutils_2.20.1.bb: Oct 05 23:19:19 binutils_2.20.1.bb: GCC 4.5 causes more warnings disable warnings during binutils build Oct 05 23:19:19 Signed-off-by: Khem Raj Oct 05 23:20:56 kergoth_: around ? Oct 05 23:25:45 03Martin Jansa  07master * r2dd7b7f9de 10openembedded.git/recipes/freesmartphone/msmcomm.inc: Oct 05 23:25:45 msmcomm: bump SRCREV for version compatible with latest vala Oct 05 23:25:45 Signed-off-by: Martin Jansa Oct 05 23:25:47 03Martin Jansa  07master * rbcda7b5c1c 10openembedded.git/recipes/dri/ (libdrm-2.4.22/glamo.patch libdrm_2.4.22.bb): Oct 05 23:25:48 libdrm: add 2.4.22, needed by new xf86-video-intel Oct 05 23:25:48 Signed-off-by: Martin Jansa Oct 05 23:25:53 03Martin Jansa  07master * recafbae0c9 10openembedded.git/recipes/freesmartphone/ (fso-specs_git.bb libfso-glib_git.bb): Oct 05 23:25:54 freesmartphone: bump SRCREV for new vala compatible versions Oct 05 23:25:54 Signed-off-by: Martin Jansa Oct 05 23:25:54 03Martin Jansa  07master * rb8ec0a9a73 10openembedded.git/ (44 files in 6 dirs): Oct 05 23:25:54 xorg: new versions for 2010-10-05 Oct 05 23:25:54 Signed-off-by: Martin Jansa Oct 05 23:25:55 03Martin Jansa  07master * r4f8ed5aa3b 10openembedded.git/recipes/freesmartphone/ (cornucopia.inc mdbus2_git.bb mioctl_git.bb): Oct 05 23:25:56 cornucopia: bump SRCREV for new vala compatible versions Oct 05 23:25:56 Signed-off-by: Martin Jansa Oct 05 23:25:57 03Martin Jansa  07master * r6b067c0cec 10openembedded.git/recipes/vala-dbus-binding-tool/vala-dbus-binding-tool_git.bb: Oct 05 23:25:57 vala-dbus-binding-tool: bump SRCREV for new vala compatible versions Oct 05 23:25:58 Signed-off-by: Martin Jansa Oct 05 23:27:01 khem, that still doesn't look right. the patch has a stray semicolon Oct 05 23:27:54 03Martin Jansa  07master * rb9dcc0a1e8 10openembedded.git/ (4 files in 3 dirs): Oct 05 23:27:54 EFL: bump SRCREV for latest bugfixes and beta release Oct 05 23:27:54 Signed-off-by: Martin Jansa Oct 05 23:27:57 03Martin Jansa  07master * r27583e6ae2 10openembedded.git/recipes/webkit/ (webkit-efl/install-efl-launcher.patch webkit-efl_svn.bb): Oct 05 23:27:57 webkit-efl: bump SRCREV for new API needed by newer eve Oct 05 23:27:57 Signed-off-by: Martin Jansa Oct 05 23:29:38 grg_: thx Oct 05 23:29:47 too much multitasking Oct 05 23:29:53 humans are not made for that Oct 05 23:29:53 :) Oct 05 23:30:07 just increase your timeslice Oct 05 23:30:16 haha Oct 05 23:30:51 that will not increase the day from being 24 hrs long Oct 05 23:31:56 grg_: how did bitbake world go Oct 05 23:32:10 did u finish to some level Oct 05 23:32:18 03Khem Raj  07master * rf94d9bcd25 10openembedded.git/recipes/gcc/gcc-4.5/gcc-scalar-widening-pr45847.patch: Oct 05 23:32:18 gcc-4.5: fix gcc-scalar-widening-pr45847.patch one last time Oct 05 23:32:18 Signed-off-by: Khem Raj Oct 05 23:32:34 khem, the last one i did got through 65000 or so tasks Oct 05 23:32:58 grg_: thats pretty good Oct 05 23:33:03 did u save the logs somewhere Oct 05 23:33:05 looks like a matter of fixing individual problems now, not too many big things in the way Oct 05 23:33:10 yeah, i have the world log Oct 05 23:33:12 I have fixed many fails in the run I had Oct 05 23:33:18 so it should be better now Oct 05 23:33:43 i'll start another one soonish, with gcc 4.5 and compare Oct 05 23:33:51 bigger problems are when conflicting packages are dragged in Oct 05 23:33:56 grg_: cool Oct 05 23:34:02 yeah, like cairo-directfb Oct 05 23:34:09 right Oct 05 23:34:38 i can actually go back and build stuff after the world has completed now, so that's a start Oct 05 23:34:45 we have to figure out a way to either amicably make them live side by side or choose few world profile Oct 05 23:38:39 khem: whats up? am around now Oct 05 23:39:39 khem, new error :( : http://pastebin.com/tz2u7w97 Oct 05 23:42:43 kergoth: I am trying to make PACKAGE_PREPROCESS_FUNCS conditional Oct 05 23:42:57 kergoth: but it seems its not working as I expect it to Oct 05 23:44:39 PACKAGE_PREPROCESS_FUNCS += "['autotools_prepackage_lamangler',''][bb.data.getVar('LIBTOOL_HAS_SYSROOT', d, 1) == "yes"]}" Oct 05 23:45:09 this is what I am trying to only append autotools_prepackage_lamangler when LIBTOOL_HAS_SYSROOT is not yes Oct 05 23:45:16 but it does not expand it Oct 05 23:45:34 and keeps sticking the unevaluated expression Oct 05 23:46:53 it works fine when I do same for TARGET_LDFLAGS Oct 05 23:47:07 this is a python var I think is that the difference ? Oct 05 23:50:04 nevermind now it seems to work Oct 05 23:51:59 you forgot a ${@ in the above Oct 05 23:52:02 heh Oct 05 23:52:14 you can also use '${LIBTOOL_HAS_SYSROOT}' Oct 05 23:52:18 it'll expand it inside of the snippet Oct 05 23:52:24 which i think is cleaner when using ${@}, personally Oct 06 00:02:53 khem, looks like my last bitbake world was with gcc 4.5. Comparing it with a previous log from gcc 4.4.4 shows almost no difference in the number of successful do_compile tasks. Oct 06 00:03:10 grg_: ok Oct 06 00:03:19 (differences probably accounted for by other changes in the tree) Oct 06 00:03:35 kergoth: you mean I dont have to bb.data.getVar LIBTOOL_HAS_SYSROOT Oct 06 00:04:48 yes Oct 06 00:04:54 kergoth: right right Oct 06 00:04:58 ['${LIBTOOL_HAS_SYSROOT}' == 'yes'] Oct 06 00:05:18 *slightly* less ugly that way ;) Oct 06 00:05:24 SYSROOTEXTRALIBDIRSED Oct 06 00:05:27 see that one Oct 06 00:05:37 now I need to conditionalize that Oct 06 00:11:13 kergoth: see if you can suggest some thing :) Oct 06 00:11:49 http://pastebin.com/xvVs5UZC Oct 06 00:15:58 oh, right Oct 06 00:16:01 in that case you have to use getVar Oct 06 00:16:12 because it has a ' in it, the surrounding '' gets screwed :) Oct 06 00:18:00 what i'd really like to do is make available the datastore as locals() Oct 06 00:18:17 so you could just FOO = "${@BPN + '-native'}" Oct 06 00:18:22 that'd be awesome. Oct 06 00:43:26 khem, are you still working on the gcc bug or could i go to bed? Oct 06 01:11:55 playya__, git pull. it was fixed Oct 06 01:14:38 nope Oct 06 01:16:18 yep Oct 06 01:18:09 grg_, this happend with HEAD about 1,5h ago: http://pastebin.com/tz2u7w97 Oct 06 01:18:39 playya__, git pull, it was fixed Oct 06 01:19:00 i already pulled f94d9bcd25 Oct 06 01:25:54 playya__, you probably just need to clean gcc-cross-initial Oct 06 01:26:39 this is what i did for testing: bitbake -c clean gcc gcc-cross gcc-cross-initial gcc-cross-intermediate && bitbake mplayer Oct 06 01:28:01 i just kicked off a bitbake gcc-cross, from scratch. So we'll see Oct 06 01:29:58 which machine? Oct 06 01:30:06 qemumipsel Oct 06 01:30:31 i think it's affiliated to armv7a Oct 06 01:31:27 well the patch applies to a file which looks target agnostic Oct 06 01:31:39 gcc-4.5/gcc/tree-vect-patterns.c Oct 06 01:34:07 I'm not a gcc expert Oct 06 01:39:03 Arm specific stuff would be in gcc-4.5/gcc/config/arm Oct 06 01:52:45 playya__, you need to clean something, i don't know what. gcc-cross-initial built fine for me Oct 06 01:53:39 i think initial built fine, too. but i used the same command Oct 06 01:55:11 your paste shows gcc-cross-initial failing at the exact line of the patch, with an error indicative of a stray semicolon at the point where the last commit fixed the stray semicolon Oct 06 01:57:11 hmm. then i missed the last commit Oct 06 01:57:23 i thought, i was up-to-date Oct 06 02:10:57 oh wow Oct 06 02:11:06 making the metadata variables accessible as locals in snippets was extremely easy Oct 06 02:11:40 def test_metadata_as_locals(self): Oct 06 02:11:40 self.d.setVar("foo", "bar") Oct 06 02:11:40 self.d.setVar("bar", "${@foo + '/baz'}") Oct 06 02:11:41 self.assertExpanded("bar", "bar/baz") Oct 06 02:12:58 thoughts? Oct 06 02:14:46 kergoth, what do you think about this patch: http://github.com/playya/tslib/commit/b53df58338421391576ffe54f0a26a1b6dc6f843 Oct 06 02:15:25 except the " " <-> \t issue Oct 06 02:20:11 looks reasonable Oct 06 02:23:30 btw. how to implement multitouch in tslib? Oct 06 02:24:11 every implementation i read request only 1 sample Oct 06 02:34:30 playya__: you need to clean gcc Oct 06 02:34:59 playya__: bitbake -c clean gcc-cross gcc-cross-initial gcc-cross-intermediate eglibc eglibc-initial Oct 06 02:35:18 then bitbake Oct 06 02:35:48 i think, i have to clean even more, because i pulled some more dependencies Oct 06 02:37:16 kergoth: interesting Oct 06 02:39:38 any thoughts on the direct access to vars as locals in snippets? Oct 06 02:39:40 * kergoth_ ponders Oct 06 02:51:49 good night **** ENDING LOGGING AT Wed Oct 06 02:59:57 2010