**** BEGIN LOGGING AT Thu Jul 30 02:59:57 2009 Jul 30 03:30:53 * kergoth grumbles Jul 30 03:32:45 * rkirti thinks ibot should have a cheer command for such situations :-) Jul 30 03:32:51 is there a way to determine the order that tasks (the ones from bitbake -c listtasks) are run? Jul 30 03:34:53 i suppose that doesnt really make sense, considering not all tasks are run automatically... but still... Jul 30 03:36:01 if i do 'addtask foo before a after b' and then 'addtask bar before a after b' how do i know whether foo or bar come first? Jul 30 03:44:15 you don't. Jul 30 03:44:20 that's not defined, or guaranteed Jul 30 03:49:59 ok Jul 30 03:54:19 so what about if there is an _append or _prepend ? are they inserted "closer" to the task than an addtask? or is it undefined too (presumably because they use addtask internally)? Jul 30 03:57:44 rkirti: one way to get _svn to get default preference is to give DEFAULT_PREFERENCE larger value than other recipes say DEFAULT_PREFERENCE = "10" in the bb file itself Jul 30 04:10:06 khem: yes, thats what I ended up doing. Also, I couldnt get away without specifying a rev. number in the SRC_URI, isnt there a way to specify it to pick up whatever rev. is there in trunk ? Jul 30 04:10:23 rkirti: mindate and maxdate are the dates that are checked against srcdate for the patch to be applied Jul 30 04:10:52 rkirti: yes you can use AUTOREV Jul 30 04:12:40 khem: thanks. that would be nice. the manual says the AUTOREV option can be added to a conf file. Any way to specify it in the recipe itself ? Jul 30 04:13:57 if you dont tie a recipe to a rev either in recipe itself or in sane-srcrev.inc bitbake will always pick latest Jul 30 04:14:26 but generally it is preferred to use a locked down rev Jul 30 04:15:14 ok Jul 30 04:15:27 if you do SRCREV=${AUTOREV} in the recipe it should do it Jul 30 04:15:44 ok Jul 30 04:15:57 grg: what do you mean? Jul 30 04:16:23 grg: _append/_prepend just append/prepend to the contents of a variable. it has nothing to do with tasks or their dependencies Jul 30 04:17:10 if you have for e.g. do_install_append, but there's also an addtask foo after install_append Jul 30 04:17:27 'after install' i meant Jul 30 04:17:30 ? Jul 30 04:17:35 addtask just sets a flag on the variable. Jul 30 04:17:37 thats all Jul 30 04:17:42 it doesn't give a shit about the contents of the variable Jul 30 04:17:51 nor does _append/_prepend care about the task flags Jul 30 04:18:05 i thought these were functions Jul 30 04:18:14 foo = "bar" Jul 30 04:18:17 foo_append = " baz" Jul 30 04:18:20 foo is now "bar baz" Jul 30 04:18:45 foo (){ echo "asdf" } Jul 30 04:18:56 foo_append = { echo "qwert"} Jul 30 04:19:04 yes, and? Jul 30 04:19:08 foo is now {echo "asdf" ; echo "qwert" } ? Jul 30 04:19:28 first, foo_append = {} isn't valid syntax Jul 30 04:19:33 second, you didn't include the ; separator Jul 30 04:19:35 so no Jul 30 04:19:44 uggh... sorry foo_append (){ echo "qwert"} Jul 30 04:20:08 ok, so there probably going to be a \n or two in there.. Jul 30 04:20:13 yup Jul 30 04:20:20 itd be two lines with the two echos, newline separated Jul 30 04:20:35 in bitbake, everything is key=value metadata, including functions and tasks Jul 30 04:20:44 foo = "echo asdf" Jul 30 04:20:47 foo[func] = "1" Jul 30 04:20:54 is very similar to foo () { echo asdf } Jul 30 04:21:05 because thats all the () {} syntax does, it flags it as a function Jul 30 04:21:41 i'm still a little sketchy on the moving between bourne shell syntax and python syntax Jul 30 04:21:55 python foo () { Jul 30 04:21:55 } Jul 30 04:21:57 is the same as Jul 30 04:21:59 foo () { Jul 30 04:21:59 } Jul 30 04:22:03 foo[python] = "1" Jul 30 04:22:20 its again, just another flag, that tells bitbake how to execute that function Jul 30 04:22:20 oh. neat Jul 30 04:22:39 maybe i should brush up on my python and read some bitbake code Jul 30 04:23:27 cheers kergoth Jul 30 04:24:04 to slide back to the original question, _prepend/_append just concatenate, the same way .= and =. do, except that _append/_prepend happen at the end of the parse rather than immediate. addtask manipulates flags on the task/function, iirc it messes with the 'depends' flags Jul 30 04:24:09 np Jul 30 04:24:59 is there variable scope? Jul 30 04:25:04 or is everything global? Jul 30 04:25:15 i.e. functions have local vars? Jul 30 04:25:41 or does it depend on whether its a python/shell function? Jul 30 04:26:08 perhaps i'm still confused Jul 30 04:26:33 shell functions get emitted as shell scripts and then those shell scripts get executed. each variable flagged as func and that isn't python gets emitted into that shell script, regardless of which shell func is being run right now (since we don't analyze the shell contents, we don't track what functions call which other functions) Jul 30 04:27:05 if you read ${T}/run.., thats the emitted shell script Jul 30 04:27:13 may help you to see how the shell functions work Jul 30 04:27:51 python functions get evaluated from bitbake's code as python functions, so they have their own scope, yes. they do pick up any globally defined python functions in the metadata that use the 'def' syntax Jul 30 04:28:29 bitbake/lib/bb/build.py is what does the function execution if you want to brush up on your python, its pretty straightforward Jul 30 04:29:38 the task dependency stuff is substantially more complex, since bitbake operates at the task level, a task in one recipe can depend on specific tasks in other recipes, etc. the bitbake taskdata/runqueue code figures out what to execute in what order to satisfy everything Jul 30 04:30:27 yes, i've seen that task dependencies are very complex from looking at the massive dependency graphs Jul 30 06:41:59 I'm attempting to build icedtea with uclibc using openembedded and jalimo. Jul 30 06:42:41 I started with the angstrom distro (as was suggested on some jalimo page), but ran into some trouble related to KERNEL_HEADERS Jul 30 06:43:28 I found that that might be (partially) fixed in uclibc-0.9.30.1, but angstrom didn't use that. Then I read about minimal-uclibc and that it should always compile, so I switched to that. Jul 30 06:44:20 And now util-linux is not linking because libuuid.so needs tls and uclibc doesn't have it Jul 30 06:45:26 Is there any uclibc build that has a higher chance of building successfully? I'm doing this to get a uclibc-based icedtea package, not to get an entire system. Jul 30 07:58:38 godo morning Jul 30 07:59:18 morning :) Jul 30 08:28:16 hi Jul 30 08:49:00 03Steffen Sledz  07org.openembedded.dev * r0501fbe7b6 10openembedded.git/recipes/linux/linux-2.6.24/hipox/hipox-nand.patch: linux-2.6.24: NAND flash size fixed for hipox machine Jul 30 09:01:31 morning everyone Jul 30 09:02:18 yo Jul 30 09:30:46 hi Jul 30 09:31:21 03Koen Kooi  07org.openembedded.dev * rf2e6ca6bfa 10openembedded.git/recipes/linux/ (148 files in 3 dirs): linux-omap-pm git: add patch for fix musb oops and dss2 patches Jul 30 09:31:50 hello woglinde Jul 30 09:36:53 hi, I have a problem with libx11-native_1.1.5.bb, it always fails to build in the same place, I've even tried a completely clean rebuild, my output is here: http://pastebin.ca/1512029 Jul 30 09:37:17 it's always this "xlibi18n/.libs/libi18n.a: member xlibi18n/.libs/libi18n.a(XDefaultIMIF.o) in archive is not an object" Jul 30 09:40:06 I'm really stuck at this problem Jul 30 09:49:00 hi rkirti Jul 30 09:58:06 can anyone help me figure out why my build crashes at libx11-native? Jul 30 10:08:27 ~lart evolution for being a beast to build Jul 30 10:08:28 * ibot explains, ever so gently, that if evolution doesn't give the channel more information, they can't help for being a beast to build Jul 30 10:09:11 ibot is losing its ever-so-appropriate larting sense :-( Jul 30 10:37:36 rkirti, I did give some info as in pastebin log ;) Jul 30 10:37:54 but, it was before you joined Jul 30 10:37:56 hi, I have a problem with libx11-native_1.1.5.bb, it always fails to build in the same place, I've even tried a completely clean rebuild, my output is here: http://pastebin.ca/1512029 Jul 30 10:37:56 it's always this "xlibi18n/.libs/libi18n.a: member xlibi18n/.libs/libi18n.a(XDefaultIMIF.o) in archive is not an object" Jul 30 10:39:15 I even disabled parallel make and extra jobs .. Jul 30 10:39:19 th1_: I am really sorry, I think you misunderstood. I asked ibot to lart a package I was trying to build Jul 30 10:39:26 ok Jul 30 10:39:30 th1_: ibot printed an inappropriate message Jul 30 10:39:41 ok ;) Jul 30 10:39:43 th1_: I had a look at your pastebin. I am sorry I cant help here Jul 30 10:39:55 I thought you meant the bot to tell me off ;) Jul 30 10:40:57 th1_: no, its ~lart , the object here was evolution Jul 30 10:41:42 ok Jul 30 10:46:24 yyayy ftw Jul 30 10:46:26 ccache -C Jul 30 10:46:29 did the job Jul 30 10:46:39 it must have had some stale badness in the ccache Jul 30 10:57:19 03Koen Kooi  07org.openembedded.dev * r48c0428ab3 10openembedded.git/recipes/linux/ (3 files in 3 dirs): linux-omap-pm git: fix mtd and tweak defconfig Jul 30 11:22:47 03Marc Olzheim  07org.openembedded.dev * rf88d3e68e3 10openembedded.git/recipes/libtool/ (libtool_2.2.4.bb libtool_2.2.6a.bb): Jul 30 11:22:47 libtool: 2e0fa4918d643ce6ec0299cd31cfe1c346e4bce3 broke do_configure. Jul 30 11:22:47 * See http://lists.linuxtogo.org/pipermail/openembedded-devel/2009-July/012754.html for more. Jul 30 11:27:35 03Koen Kooi  07org.openembedded.dev * redbe31b3ac 10openembedded.git/: Merge branch 'org.openembedded.dev' of git@git.openembedded.org:openembedded into org.openembedded.dev Jul 30 11:27:35 03Koen Kooi  07org.openembedded.dev * re6a18b4392 10openembedded.git/recipes/linux/ (3 files in 3 dirs): linux-omap-pm: make u-boot flashable from within linux, first attempt at making ehci work Jul 30 11:52:11 is it possible to write a recipe that will fetch a repository and then add a task to check if a specific file is in that repository, and if it is then it adds another repository to SRC_URI and fetches that? Jul 30 11:52:40 I know I can add a task after the first fetch but can I make it process the added SRC_URI afterwards? Jul 30 11:56:44 can I even dynamically change SRC_URI by script? Jul 30 11:58:49 where do I place a new recipe for an image? Can I place it in my "local" recipe"-dir together with my own recipes ? Jul 30 11:59:31 anywhere in BBFILES Jul 30 12:04:13 hrw: aaaaahh ... that's why .... BBPATH is != BBFILES. thanks Jul 30 12:07:50 https://aradia.panaceas.org/pb/f2b2f2518 Jul 30 12:08:17 replace system.conf Jul 30 12:12:03 then install qt4-dev-tools and run qdbusviewer Jul 30 13:28:54 morning Jul 30 13:30:28 hi chouimat Jul 30 13:46:13 hi is anyone working on integrating the palm pre patches in or Jul 30 13:46:17 s/or/oe? Jul 30 14:03:46 GNUtoo: afaik no one Jul 30 14:03:55 GNUtoo: you can ask Palm for them Jul 30 14:03:59 ok thanks a lot Jul 30 14:04:41 because I had my work duplicated twice so now I ask everytime I do something that someone could duplicate Jul 30 14:24:49 I try to do bitbake x11-image but it fails at the end although it's not trying to compile any packages, it says that packages are missing but if I do e.g. bitbake xserver-xorg (one of the missing ones) it says it's already done. Pastebin: http://pastebin.ca/1512254 Jul 30 14:25:41 GNUtoo: I think there is a lot of interest in the PRE but being CDMA and not avalable outside the US that tends to seriously limit it's hacker base ;-) Jul 30 14:25:55 yes Jul 30 14:26:08 it was for helping someone to free his pre Jul 30 14:26:15 I don't own a pre myself Jul 30 14:58:00 please look at my paste? I don't understand wtf is going on since it seems it's built all the packages it says it can't find. Jul 30 15:01:56 th1_: the "unsatisfied recommendation" still in that log imply a high degree of fuckage somewhere :(. No symlinks in your path or stuff like that as a first port of call. I assume this is still for the same custom board you had been working on the other day? Jul 30 15:04:09 th1_: it looks like you forgot to set the PACKAGE_ARCHES for your machine Jul 30 15:05:02 DJWillis, yes it's the same one. I solved a snafu by ccache and did a clean rebuild Jul 30 15:05:32 XorA, uhm, that sounds likely ;) Jul 30 15:06:14 my machine config: http://pastebin.ca/1512296 Jul 30 15:06:36 I didn't find PACKAGE_ARCHES in the template I copied from Jul 30 15:07:34 where should it go? Jul 30 15:07:52 the tune-pentium* has them Jul 30 15:07:59 ok.. Jul 30 15:08:07 then I'm back to square one Jul 30 15:08:36 except pentum is i686p4c and yoru making i686 Jul 30 15:09:48 ignore me Jul 30 15:10:05 hut it will be something around that area Jul 30 15:10:05 e.g. d201gly2.conf has the same "combo" as mine? Jul 30 15:10:36 look at the ipkg/opkg arch.conf for the rootfs installation and check the aches it knows are the same as the ones you built Jul 30 15:10:47 what's the difference between TARGET_ARCH and PACKAGE_ARCH? Jul 30 15:11:58 XorA, what's the ipkg/opkg arch.conf? Jul 30 15:12:02 I mean which file ;) Jul 30 15:12:17 youll have to look in the logs, I forget where it is put Jul 30 15:12:27 what am I looking for? Jul 30 15:12:33 arch.conf I think Jul 30 15:13:27 nothing in the log with that Jul 30 15:14:23 # Jul 30 15:14:23 | + for arch in '$ipkgarchs' Jul 30 15:14:23 # Jul 30 15:14:23 | + echo 'arch all 1' Jul 30 15:14:23 # Jul 30 15:14:25 | + echo 'arch x86_64-all-sdk 1' Jul 30 15:14:27 # Jul 30 15:14:29 | + echo 'arch x86_64-linux-sdk-all 1' Jul 30 15:14:31 yeah there is Jul 30 15:15:59 all my ipk files in deploy are *_i686p4c.ipk Jul 30 15:16:13 so what I need is tell rootfs builder that somehow? Jul 30 15:20:26 where does ipkgarchs='all any noarch i686 x86 i386 i486 i586 xenclient-dom0' get set? Jul 30 15:20:43 it looks like PACKAGE_EXTRA_ARCHS without BASE_PACKAGE_ARCH prepended Jul 30 15:20:49 th1_: some recipes emit packages which will work on any arch, some emit packages which will work on anything thats a certain architecture, some binary packages are actually machine specific... ipkgarchs is used to instruct ipkg/opkg on which archs it shuld be able to install Jul 30 15:20:59 * kergoth wonders if anyone's written a script to create an initial recipe based on debian/* Jul 30 15:21:04 ok Jul 30 15:21:16 but all my packages are either "all" or "i686p4c" Jul 30 15:21:54 but in tune-pentium4c.inc are BASE_PACKAGE_ARCH = "i686p4c" and PACKAGE_EXTRA_ARCHS += "x86 i386 i486 i586" Jul 30 15:22:11 but it only looks from the log like it uses PACKAGE_EXTRA_ARCHS in the searchpath Jul 30 15:23:06 kergoth, my log is here btw http://pastebin.ca/1512254 (you joined after I pasted it) Jul 30 15:24:40 th1_: package_ipk.bbclass's package_update_index_ipk sets ipkgarchs to PACKAGE_ARCHS Jul 30 15:24:47 (<3 grep) Jul 30 15:24:48 :P Jul 30 15:25:23 so its ignoring BASE_PACKAGE_ARCH basically? Jul 30 15:26:19 a quick grep shows that bitbake.conf defines the default PACKAGE_ARCHS to all any noarch ${TARGET_ARCH} ${PACKAGE_EXTRA_ARCHS} ${MACHINE} Jul 30 15:27:17 I've got a question about the stable branch of oe Jul 30 15:27:38 Specifically, I read on the wiki that only a subset of all metadata is supported Jul 30 15:28:20 the stable wiki page indicates a few files where those sets might be specified Jul 30 15:28:27 I can't seem to find them though Jul 30 15:28:41 ok then I just add i686p4c to PACKAGE_EXTRA_ARCHS in tune-pentium4c and see what happens :) Jul 30 15:39:48 bye Jul 30 15:44:15 03Holger Hans Peter Freyther  07org.openembedded.dev * rc5322a7b31 10openembedded.git/recipes/xorg-lib/libxext_1.0.5.bb: Jul 30 15:44:15 libxext-1.0.5: Depend on xdmcp as configure wants it Jul 30 15:44:15 libxext 1.0.5 want to have xdmcp, hand it over. prior Jul 30 15:44:15 versions do not have this requirement. Jul 30 15:44:19 03Holger Hans Peter Freyther  07org.openembedded.dev * r6d38e36cda 10openembedded.git/ (2 files in 2 dirs): openbsc: Add openbsc and build it from git. Jul 30 15:44:20 03Holger Hans Peter Freyther  07org.openembedded.dev * r838abf1993 10openembedded.git/ (4 files in 2 dirs): libdbi: Introduce libdbi 0.8.2 Jul 30 15:48:12 arse I can forsee some pain at work if CentOS collapses Jul 30 15:50:08 XorA: Are they likely to? There's other RH knockoffs out there, most of them claim binary compatibility... Jul 30 15:51:28 broonie: more likely they will rename I guess Jul 30 15:52:42 Oh, I found the story. Jul 30 15:55:18 although phoning UK Linux isnt going to help, theyve been dead for years Jul 30 15:57:05 kergoth, XorA thanks for your help so far I got much further now but still breaks: http://pastebin.ca/1512356 - 2 issues I can see: opkg tries to create /var/lib/opkg and also some dependencies are missing still (at the end of the paste) Jul 30 15:57:21 th1_: cool every step forward Jul 30 15:58:16 th1_: that log is invisible to me :-) Jul 30 15:58:18 yes I hope I'm getting close but these are still quite weird. Jul 30 15:58:50 re Jul 30 15:58:56 what arches do the packages that are missing have? Jul 30 15:59:08 * broonie hadn't realised Lance was involved in that. Jul 30 15:59:23 XorA, weird, I can open it from 2 separate locations Jul 30 15:59:34 broonie: I recognised the name, didnt realise it was the same guy until I read the Open Letter Jul 30 16:00:24 th1_: stupid firefox :-) Jul 30 16:00:35 it's a pretty long paste :/ Jul 30 16:00:56 * XorA remebers the days computers could handle ascii text Jul 30 16:01:14 well in fact there are plenty of, e.g., glibc-*.ipk files for example, but no glibc-2.x.ipk Jul 30 16:01:19 EBDIC for ever! Jul 30 16:01:46 broonie nah Jul 30 16:04:18 XorA, if your firefox barfs maybe this will work http://pastebin.ca/raw/1512356 Jul 30 16:05:44 th1_: ahah Jul 30 16:05:58 doesn't have nasty javascript and css Jul 30 16:07:01 to disable a package 'foo' in an image that doesn't build for a machine, shouldn't I be able to add DEFAULT_PREFERENCE_pn-foo = "-1" in a machine or distro.conf? Jul 30 16:10:30 XorA, can you guess anything from it ;) Jul 30 16:11:24 for example I have tmp/deploy/glibc/ipk/i686p4c/dbus_1.2.1-r12.3_i686p4c.ipk in my deploy dir, woudln't that satisfy dependency of dbus-lib? or is the -lib a separate package that didn't get built for some weird reason? Jul 30 16:11:30 th1_: no, except it looks like the packages of those versions are really missing, which would be odd Jul 30 16:12:32 output of "find tmp/deploy/ -name \*.ipk" in http://pastebin.ca/raw/1512382 Jul 30 16:14:48 th1_: it looks like packaging never finished for some reason, or debian renaming only half done Jul 30 16:15:10 th1_: dbus.bb kicks out dbus dbus-lib and dbus-x11 Jul 30 16:15:33 so why then is it requesting dbus? Jul 30 16:15:45 can I redo the packaging step? Jul 30 16:15:51 th1_: well its built libdbus but is looking for dbus-lib Jul 30 16:16:04 th1_: did you change distro half way into build? Jul 30 16:16:14 no I never touched the distro setting Jul 30 16:16:23 tmp/deploy/glibc/ipk/i686p4c/libdbus-1-3_1.2.1-r12.3_i686p4c.ipk Jul 30 16:16:49 is there, but obviously the dbus-lib -> libdbus rename never got to the RDEPENDS handling Jul 30 16:17:00 whih DISTRO btw? Jul 30 16:17:15 DISTRO = "angstrom-2008.1" Jul 30 16:17:18 I left it at default Jul 30 16:17:20 really odd Jul 30 16:18:03 tmp/deploy/glibc/ipk/i686p4c/libc6_2.9-r32.0.3_i686p4c.ipk Jul 30 16:18:03 is there as well Jul 30 16:18:20 which is glibc Jul 30 16:18:28 so it's just that they don't have debianized names? Jul 30 16:18:28 * XorA is puzzled Jul 30 16:18:36 th1_: yes Jul 30 16:18:38 should they even have that when I use angstrom distro? Jul 30 16:18:46 yes, they should Jul 30 16:18:49 ok Jul 30 16:18:59 well I can try to clean and rebuild one and see what happens Jul 30 16:19:00 like dbus Jul 30 16:19:20 clean and rebuild xserver-xorg Jul 30 16:19:37 after clean and rebuilding the missing packages Jul 30 16:19:51 but this is all very odd Jul 30 16:19:56 never seen this before Jul 30 16:20:54 I'll just try dbus first to see if it makes a difference? dbus -c clean dbus && bitbake dbus Jul 30 16:23:22 http://pastebin.ca/1512396 is the rebuild log and still no \*dbus-lib\*.ipk to be found Jul 30 16:23:37 yes, there never will be Jul 30 16:23:42 its libdbus Jul 30 16:24:00 ok that was already there Jul 30 16:24:03 and libc6 etc. Jul 30 16:24:19 the fault is in xserver-xorg which hasnt picked up the debian renames Jul 30 16:24:23 for some reason Jul 30 16:24:26 ok I will rebuild that then Jul 30 16:51:14 XorA, it's the same after the rebuild :( Jul 30 16:51:35 anyone have bits for creation of full disk images from oe, rather than just partition/filesystem images? Jul 30 16:51:41 syslinux or grub, preferably :) Jul 30 16:51:43 * kergoth looks at poky Jul 30 16:51:48 he kergoth Jul 30 16:52:02 th1_: something crazy is happing which is beyond my power to fix Jul 30 16:52:09 th1_: and only to you Jul 30 16:52:20 :( Jul 30 16:54:53 I think I have all the packages but it's asking for the wrong dependencies i.e. XXX-whatever instead of libXXX Jul 30 16:54:58 hey woglinde Jul 30 16:55:15 th1_: was this a clean build from scratch? Jul 30 16:55:19 yes Jul 30 16:55:32 on a decent unix file sytem? Jul 30 16:55:34 except I added the i686pc4 package type Jul 30 16:55:52 ext4 Jul 30 16:56:03 on ubuntu 9.04 x64 Jul 30 16:56:16 I wonder, I guess the missing package type might still cause this issue, you might need to build from scratch again Jul 30 16:56:28 try a console-image for speed Jul 30 16:57:31 dinner time :-) Jul 30 16:57:54 th1 are other .ipk created? Jul 30 16:58:19 so mv tmp tmp.old ; bitbake console-image? Jul 30 16:58:33 th1_: yes, and sorry this is causing you so much hassle Jul 30 16:58:39 woglinde, output of `find deploy -name \*.ipk` is pasted here: http://pastebin.ca/raw/1512382 Jul 30 16:58:49 XorA, no worries I appreciate the help Jul 30 16:59:25 how can I prevent it from redownloading stuff already in my download-dir? Jul 30 16:59:29 that would speed it up a lot Jul 30 16:59:39 th1 what you are bulding for? Jul 30 16:59:56 its for a xen dom0 for a specific appliance Jul 30 17:00:14 th1_: keep download dir outside of tmp/ Jul 30 17:00:19 XorA, it is already Jul 30 17:00:19 th1_: or copy it Jul 30 17:00:30 but last time I did this it redownloaded a lot of stuff Jul 30 17:01:47 th1_: it doesn't redownload files already in the DL_DIR. Jul 30 17:02:29 th1_ then you set your archive dir wrong Jul 30 17:04:09 I have DL_DIR = "${HOME}/src/oe/downloads" Jul 30 17:04:16 and thats also where it puts the tar files Jul 30 17:04:26 but it didn't redownload all of them last time just some Jul 30 17:04:34 hm Jul 30 17:04:39 maybee some was new Jul 30 17:05:01 lunch now Jul 30 17:05:02 it's possible though I was doing x11-image both times Jul 30 17:05:04 ok :) Jul 30 17:05:34 you can/should delete some old .tars from time to time Jul 30 17:05:35 now I have enabled PARALLEL_MAKE and BB_NUMBER_THREADS again since it didn't make a difference last time and it will be faster Jul 30 17:05:43 yes Jul 30 17:05:45 defenitily Jul 30 17:05:48 but now lunch Jul 30 17:05:53 or my wife will be angry Jul 30 17:05:53 yes but I have only had OE for a week :) Jul 30 17:05:58 ok later and thanks Jul 30 17:34:19 re Jul 30 18:15:13 hmm, someone just asked on the list how to stop OE building a kernel for a machine Jul 30 18:15:22 any remember if we figured it out? Jul 30 18:18:53 I bet poky has something to do that Jul 30 18:19:10 I need to look through the archives Jul 30 18:19:15 if I remember very old posts Jul 30 18:19:18 ok Jul 30 18:19:22 or "fix" the ml403 kernel so it builds Jul 30 18:19:26 even if it is useless Jul 30 18:19:29 ASSUME_PROVIDED += "virtual/kernel" ised to work, but it seems to have broken recently Jul 30 18:19:34 I remember something reently Jul 30 18:19:41 ah Jul 30 18:19:52 I suppose I will try that route .... Jul 30 18:19:54 * XorA hasnt tested Jul 30 18:20:06 I think I have spare cycles atm on a machine Jul 30 18:25:12 re Jul 30 18:25:24 jo florian Jul 30 18:33:45 hi is there a good place for getting pre-built toolchain for armv6? Jul 30 18:33:50 s/v6/v7 Jul 30 18:34:02 maybe I'll look at codesourcey Jul 30 18:34:03 CodeSourcery Jul 30 18:34:07 ok thanks Jul 30 18:37:44 mmm...code sourcey seems to give only the compilers...not the libc Jul 30 18:38:04 oops Jul 30 18:38:17 I didn't look carefully Jul 30 18:38:21 but they use newlib...strange Jul 30 18:38:44 tmp/cross :) Jul 30 18:39:35 ok Jul 30 18:40:22 it's for someone that: Jul 30 18:40:29 *doesn't want to use my binaries Jul 30 18:40:37 *only want busybox ans lsof Jul 30 18:40:45 statically compiled Jul 30 18:44:01 Gnutoo: use the linux one they have two variants *-*-linux and *-*-eabi Jul 30 18:44:14 ok Jul 30 18:54:41 minimal uses tinylogin should it be replaced Jul 30 19:08:21 btw we should bump lsof Jul 30 19:08:35 I didn't find the source at the location in the recipe Jul 30 19:09:01 ah there is an old directory Jul 30 19:09:07 we should update SRC_URI then Jul 30 19:28:49 kergoth, XorA, woglinde thanks for your help, the console-image built, now starting x11-image :) Jul 30 19:28:58 grats, good luck ;) Jul 30 20:02:52 hmm, sqlite is a total piece of junk :/ Jul 30 20:03:58 how so? Jul 30 20:04:14 http://pastebin.ca/1512714 Jul 30 20:04:34 :( Jul 30 20:04:40 I create the table, I sync is in every way I see possible and then it doesn't exist Jul 30 20:05:11 The whole point of using sqlite was to avoid issues like that Jul 30 20:07:09 Even sync, close and reopening the DB connection doesn't seem to be enough to actually force it to look like its supposed to :/ Jul 30 20:07:20 As in my deleted tables reappear Jul 30 20:12:44 heh, even after I revert this to known to work code, I have to blow away the sqlite db before it works properly :/ Jul 30 20:13:01 odd Jul 30 20:13:11 if it was that broken, you'd think people would notice :) Jul 30 20:13:25 kergoth: Yes :/ Jul 30 20:13:50 kergoth: It only happens when you try and rename a table and then recreate an empty verison of the original Jul 30 20:13:59 interesting Jul 30 20:14:17 That shouldn't be such a crazy thing to do... Jul 30 20:14:20 * kergoth can't be very helpful, sadly, as he hasn't had to mess with sqlite in some time Jul 30 20:14:24 you wouldn't think so Jul 30 20:14:48 I don't have the time to debug into this as it could by python's sqlite wrapper or it could be sqlite itself Jul 30 20:15:08 The fact the DB seemed corrupted somehow suggests sqlite itself but who knows Jul 30 20:16:13 ive only used the c stuff with sqlite and not seen anything like that Jul 30 20:16:24 couldn't you try to reproduce using the sqlite commandline interface? Jul 30 20:16:52 hrw|gone, http://www.cnbc.com/id/26789471?slide=4 Jul 30 20:17:03 kergoth: Yes, I really should make a standalone python test case, then see if the same commands in the CLI break Jul 30 20:20:10 yeah Jul 30 20:21:56 CLI doesn't reproduce the problem Jul 30 20:38:33 hi,codesourcey's toolchain is for armv7+glibc right? Jul 30 20:38:40 we need armv6 Jul 30 20:38:45 maximum Jul 30 20:40:09 Been a while since i poked their public stuff Jul 30 20:40:21 But iirc they give glibc and uclibc and lots of different options for prebuilt libs Jul 30 20:40:49 Gnutoo: you can chose that with -march Jul 30 20:41:11 and I am sure by default it should generate code for armv5 or something Jul 30 20:43:25 *** %n in writable segment detected *** <--- whats this from? seeing it since I updated my dev system from Ubuntu 8.04 to 9.04 - believe its an autoconf issue? Jul 30 20:46:55 that's what I did Jul 30 20:47:11 but when I do for armv4 it says illegal instruction Jul 30 20:47:17 on the target machine Jul 30 20:47:46 Gnutoo: hmmm Jul 30 20:47:56 so with -march=armv4 Jul 30 20:48:08 you get illegal instr Jul 30 20:48:14 yes Jul 30 20:48:22 I did -isystem and -I too Jul 30 20:48:32 when disassembling it seems arm assembly Jul 30 20:48:41 file also say arm Jul 30 20:48:48 The gcc libs the toolchain comes with are probably compiled for armv7 Jul 30 20:49:08 RP: I would doubt that Jul 30 20:49:27 Gnutoo: you can use my simple shell script to create a toolchain if you want Jul 30 20:49:42 yes but it will take hours Jul 30 20:50:01 I need to have lsof running on an armv6 board before my tester go away Jul 30 20:50:10 and before tommorow Jul 30 20:50:34 Gnutoo: I could tar it up but downloading might take time Jul 30 20:50:46 ok Jul 30 20:50:48 or rathar uploading Jul 30 20:50:52 I'll ask my tester Jul 30 20:57:15 by the way thanks a lot Jul 30 21:06:50 ASSUME_PROVIDED += "virtual/kernel" Jul 30 21:07:02 looks like that did stop the image from building the kernel Jul 30 21:09:50 03Philip Balister  07org.openembedded.dev * rb73732a25f 10openembedded.git/conf/machine/xilinx-ml403.conf: Jul 30 21:09:50 xilinx-ml403 : Do not build a kernel for this machine. Jul 30 21:09:50 * Kernel config depends on the fpga bitstream. No sense building Jul 30 21:09:50 kernel we can't use. Jul 30 21:10:32 my bitstream? :) Jul 30 21:11:49 heh Jul 30 21:12:04 you familiar with xilinx virtex 5? Jul 30 21:13:17 ppc stuff Jul 30 21:15:00 yeah Jul 30 21:34:34 guten morgen Jul 30 21:38:16 morning Martin-B :-) Jul 30 21:39:12 morning florian :-) Jul 30 21:42:55 is arm in openembedded normally big-endian or little-endian? Jul 30 21:44:14 xxiao: you can have both, but most devices use little Jul 30 21:45:42 florian: thanks. debian prefers to LE, openwrt prefers to BE, i'm just starting to learn openembedded with beagleboard, I assume it's LE then. Jul 30 21:46:24 xxiao: yes... not sure if the omap supports BE at all Jul 30 21:47:33 xxiao: openwrt has e.g. Intel IXP4xx among the supported ARM devices. These work BE too. Jul 30 21:48:15 florian: once using ixp4xx, i was told BE is more efficient with its NPE(ethernet) micro-engine, which is one reason they use BE on it Jul 30 21:48:26 florian: nbd just told me Orion in openwrt will be LE though Jul 30 21:49:58 xxiao: yes right Jul 30 21:50:56 Question if I may. Has anyone here used the "udev-static-devices" bb package before? It copies devices.tar.gz to "/lib/udev/", and it contains the correct items, but I never see my devices in "/dev/*" upon boot. My image is just a basic "x11-image" type image. Is there something special I have to do to make the "udev_static_devices_tarball" (devices.tar.gz)? Currently I've tried the "tar -cf" and "tar -zcf", both don't result in udev adding the items. I've t Jul 30 22:22:34 florian: all arms support BE and LE Jul 30 22:23:23 Networking devices use BE generally because packet processing is big endian Jul 30 22:23:38 for a LE processor a conversion is needed Jul 30 22:31:47 khem: try BE with a StrongARM ;) Jul 30 22:38:40 hi, how non-standard audio is handled in oe? Jul 30 22:38:53 is there a patch for each thing that access to the audio? Jul 30 22:39:07 such as: mplayer,vlc,gstreamer,pulseaudio etc... Jul 30 22:45:15 florian: hmm ages ago when I used a SA1110 based Ref board it worked ok Jul 30 22:47:21 khem: really... if i rember there were some peripherals that didn't work in be mode. there was somethgin in the errata document about this Jul 30 22:48:18 oh I wouldnt know about errata so much Jul 30 22:48:38 and least peripherals Jul 30 22:48:54 I only worrried about compiler testsuites Jul 30 22:49:10 :-) Jul 30 22:50:15 i was happy to have a toolchain that produced basically useful binaries at this time :-) Jul 30 22:50:22 time to get some sleep Jul 30 22:50:25 good night Jul 30 22:50:29 good nite Jul 30 22:50:31 mee too Jul 30 22:50:35 gute nacht Jul 30 22:50:38 tharvey: woglinde sleep well Jul 30 22:51:33 zecke: Do we still use tinylogin ? Jul 30 22:59:09 "we" as in angstrom? or "we" as in oe-based distros in general? Jul 30 23:37:08 mwester: in general Jul 30 23:37:20 mwester: I was trying to build a stock minimal-uclibc distro Jul 30 23:37:28 and it was trying to include it Jul 31 01:15:20 Mentor Graphics announced it has acquired Linux development firm Embedded Alley, plans to port Android to PowerPC, and will work with ARM, Freescale, Marvell, MIPS, RMI, and TI on Linux and/or Android projects. Jul 31 01:19:40 True. :) Jul 31 02:09:09 xxiao: I'm a little confused because I don't exactly think of Mentor Graphics as a SW provider Jul 31 02:12:33 archae0pteryx: looks like it's moving that way Jul 31 02:15:39 xxiao: well when I tried to use their EDA tools they weren't as user friendly as orcad or some others I can't think of, maybe that's why Jul 31 02:18:08 their nucleus RTOS seems doing well though **** ENDING LOGGING AT Fri Jul 31 02:59:57 2009