**** BEGIN LOGGING AT Tue Jun 15 02:59:57 2010 Jun 15 04:10:39 03Khem Raj  07org.openembedded.dev * rfdf077747e 10openembedded.git/recipes/libtiff/tiff_3.9.2.bb: Jun 15 04:10:39 tiff_3.9.2.bb: Use default do_configure from autotools. Jun 15 04:10:39 Signed-off-by: Khem Raj Jun 15 04:10:49 03Khem Raj  07org.openembedded.dev * r869ee7b239 10openembedded.git/recipes/libsdl/ (libsdl-x11-1.2.14/libtool-update.patch libsdl-x11_1.2.14.bb): Jun 15 04:10:49 libsdl-x11-1.2.14: Update libtool macros and remove legacy staging Jun 15 04:10:49 Signed-off-by: Khem Raj Jun 15 04:11:44 03Khem Raj  07org.openembedded.dev * r6f2200ef7c 10openembedded.git/recipes/gcc/ (gcc-4.5.inc gcc-4.5/gcc-flags-for-build.patch): Jun 15 04:11:44 gcc-4.5: Bump SRCREV Jun 15 04:11:44 Signed-off-by: Khem Raj Jun 15 06:20:37 hi Jun 15 06:20:59 i build a new gnome anstrom image but i dont see the configuring output lines from the gnome packages Jun 15 06:21:14 where is the output written to? Jun 15 06:21:38 i load the kernel with - console=ttymxc0,115200 console=tty1 root=/dev/mmcblk0p1 rw debug noinitrd psplash=false rootdelay=1 Jun 15 06:22:06 i boot from a sd card Jun 15 06:36:27 gm Jun 15 07:07:09 ls Jun 15 07:11:54 while trying to build angstrom for beagleboard with uclibc I get: Jun 15 07:11:56 ERROR: Information not available for target 'arm-linux-uclibceabi' Jun 15 07:12:03 anyone an idea what is wrong/how to fix Jun 15 07:16:10 nevermind, picked up an old siteinfo.bbclass from an overlay Jun 15 07:21:50 morning Jun 15 07:23:09 hi hrw Jun 15 07:26:47 i build a new gnome anstrom image but i dont see the configuring output lines from the gnome packages Jun 15 07:26:52 where is the output written to? Jun 15 07:26:55 i load the kernel with - console=ttymxc0,115200 console=tty1 root=/dev/mmcblk0p1 rw debug noinitrd psplash=false rootdelay=1 Jun 15 07:27:42 puchu this is not an angstrom support channel Jun 15 07:29:26 eFfeM_work where do i have to ask to get support? Jun 15 07:29:57 i thought it included in oe so there is someone in here who can tell me how to get the old behaviour where the output was written to the console Jun 15 07:30:18 puchu, there is #angstrom Jun 15 07:30:26 ohh thanks Jun 15 07:31:13 puchu and your info is hardly enough to say anything about it, e,g. what hw you have, might well be your console= line is wrong, but configuring output lines from gnome packages does not tell too much about the problem Jun 15 07:45:38 eFfeM_work its an arm cpu Jun 15 07:45:49 a board develops by the compony im working at Jun 15 07:45:52 developed Jun 15 07:46:06 a board developed by the company im working at Jun 15 07:49:04 eFfeM_work but the system boots and i cant see the configure output but i can see gnome running - i also see the init process - gnome runs on a fb and i also have serial console access over usb Jun 15 07:59:08 puchu: configure output? do you see the kernel log? do you see the messages from init on the console ? Jun 15 08:17:51 Hello all Jun 15 08:18:50 I'm looking for an existing .bb file to add Alsaplayer in my openembedded distro Jun 15 08:19:03 Can anyone may help me ? Jun 15 08:20:36 eFfeM_work yes Jun 15 08:23:31 puchu: then it does not seem to be a kernel issue but something in the app area (i'm not a gnome wiz so can't advise there) Jun 15 08:25:27 eFfeM_work without the message it is like the system crashed so the output has to be reenabled Jun 15 08:25:37 to let the user know the system is still doing something Jun 15 08:41:37 03Koen Kooi  07org.openembedded.dev * r45000bb54a 10openembedded.git/recipes/opencv/opencv_svn.bb: opencv: add bzip2 dep Jun 15 09:14:26 fjanot: if alsaplayer is aplay then openembedded/recipes/alsa/alsa-utils_1.0.20.bb Jun 15 10:10:06 ericben: Unfortunatly, alsaplayer is not aplay. alsaplayer is able to play mp3 and to be remotly controlled. aplay cannot do that Jun 15 10:10:10 hi woglinde Jun 15 10:10:43 fjanot: so you're on your own here, you need to write your recipe Jun 15 10:12:17 I had tried but it's not so easy for me (i'm an oe/python newbie). It's the reason why I asked for it to avoid doing the job twice Jun 15 10:12:48 But sure it will be a good exercise for me ! Jun 15 10:18:07 hi all Jun 15 12:21:09 I just pulled the latest oe from git yesterday and tried to build bootstrap-image for an x86 target, but it keeps failing on building ltrace. It says that it cannot find libiberty. There are several libiberty.a throughout the build/ dir. Anyone have an idea what I can do to track this down? Jun 15 12:27:07 wm? Jun 15 12:27:58 soccer Jun 15 12:28:33 heh Jun 15 12:28:42 rday is annoyed by the world cup :) Jun 15 12:29:24 Hrm? Jun 15 12:44:19 greetings. Jun 15 12:59:36 hi MWe Jun 15 12:59:37 err Jun 15 12:59:40 hi mwester Jun 15 13:00:56 hello all Jun 15 13:05:36 I have a problem to compile bitbake qt4-embedded. Jun 15 13:05:38 Are the right place here for that. Jun 15 13:05:44 ? Jun 15 13:06:55 right place, but I can't help (and with thsi amount of info I think others cannot either) Jun 15 13:07:01 Don't ask to ask - just ask your question, with as much detail as required (use pastebin if necessary), and perhaps someone who knows will see your question when they get a moment and be able to respond. Jun 15 13:16:44 my error is : http://pastebin.com/PdHkiD04 Jun 15 13:19:26 loupium: which distro/machine ? Jun 15 13:21:29 angstrom-2008.1 Jun 15 13:22:14 loupium: using Openbmedded dev or stable ? Jun 15 13:22:46 dev Jun 15 13:22:56 loupium: and which machine ? Jun 15 13:23:38 MACHINE = "hawkboard" Jun 15 13:27:40 loupium: I'm relaunching compilation for an armv5 target and will let you know if that fails Jun 15 13:29:45 I'm unable to build ltrace for an x86 target from a fresh oe dev pull from yesterday. ld fails looking for libiberty. Build error log: http://pastebin.com/BCguqytq Jun 15 13:29:58 Anyone have a clue how to approach fixing this? Jun 15 13:38:14 Hey, http://pastebin.com/b0uCwyw8 angstrom distro and omap5912osk board , Please help.! Jun 15 13:43:48 I'm looking for the list "TARGET_ARCH =" ? Jun 15 14:22:49 03Khem Raj  07org.openembedded.dev * r4dd5818c12 10openembedded.git/recipes/iproute2/ (iproute2-2.6.34/configure-cross.patch iproute2_2.6.34.bb): Jun 15 14:22:49 iproute2_2.6.34: Make configure to honor cross environment. Jun 15 14:22:49 * The configure is hand made and it calls gcc directly to poke for Jun 15 14:22:49 stuff which is not correct when build in cross enviroment so Jun 15 14:22:49 change gcc calls to $CC and pass the sysroot as a parameter to Jun 15 14:22:49 search libraries in correct sysroot and not in /usr/lib or /lib Jun 15 14:22:50 Signed-off-by: Khem Raj Jun 15 14:22:51 03Khem Raj  07org.openembedded.dev * r9b54b9315e 10openembedded.git/recipes/binutils/ (binutils-2.20.1/libtool-update.patch binutils_2.20.1.bb): (log message trimmed) Jun 15 14:22:51 binutils-2.20.1: Update libtool files so cross native builds are correct. Jun 15 14:22:51 * Currently libtool adds -L/usr/lib to the link command which directs Jun 15 14:22:52 linker to pull from /usr/lib. Many distro's dont have stuff which is Jun 15 14:22:52 conflicting in /usr/lib with the target libraries but some do. Like Jun 15 14:22:53 archlinux has libgcc_s.so.1 in /usr/lib due to the above -L it gets Jun 15 14:22:53 picked over the target libgcc and the link command fails. -L/usr/lib Jun 15 14:38:31 rf_: hmm ltrace needs libiberty it seems which is not staged yet Jun 15 14:38:43 khem: in order to build ltrace, I had to revert "bf1756a by Khem Raj, 8 days ago, message: gcc-cross: Don't stage libiberty.a in sysroot." Jun 15 14:38:48 then it seemed to build ok Jun 15 14:38:50 rf_: you can bitbake binutils first and then bitabke ltrace Jun 15 14:39:28 ok. Shouldn't that have happened automatically? I pulled a fresh oe then just tried to build an image. Jun 15 14:40:11 rf_: if it fixes then we can add to depends Jun 15 14:48:24 khem: hello there Jun 15 14:54:29 khem: our JaMa did his best to bother Gentoo devs. Finally they say it's binutils upstream bug... Jun 15 14:54:31 http://bugs.gentoo.org/show_bug.cgi?id=323319 Jun 15 15:07:17 Did anyone tried working with Packagekit with qt frontend on ARM Jun 15 15:10:49 khem: bitbake binutils; bitbake ltrace works, so it seems we should add binutils to ltrace DEPENDS Jun 15 15:20:21 03Dmitry Eremin-Solenikov  07org.openembedded.dev * r9078128528 10openembedded.git/classes/icecc.bbclass: Jun 15 15:20:21 icecc: also set environment in do_install Jun 15 15:20:21 Libtool in some packages (e.g. curl) tries to relink libraries during Jun 15 15:20:21 installation stage. If the Icecc environment isn't set in the do_install Jun 15 15:20:21 such relinking fails. Fix this by adding do_install_prepend to Jun 15 15:20:22 icecc.bbclass with set_icecc_env() call. Jun 15 15:20:23 Signed-off-by: Dmitry Eremin-Solenikov Jun 15 15:20:23 03Dmitry Eremin-Solenikov  07org.openembedded.dev * r0c38732598 10openembedded.git/contrib/angstrom/sort.sh: Jun 15 15:20:24 angstrom feed sorter: add 486sx arch and drop armv7 from the arches list Jun 15 15:20:24 Add 486sx (vortex86sx machine) and drop armv7 (no such thing exists Jun 15 15:20:25 anymore). Jun 15 15:20:25 Signed-off-by: Dmitry Eremin-Solenikov Jun 15 15:20:26 03Dmitry Eremin-Solenikov  07org.openembedded.dev * r163cf0f103 10openembedded.git/recipes/gdb/gdb-cross-sdk.inc: Jun 15 15:20:26 gdb-cross-sdk: drop bfd&opcodes messages from the archive Jun 15 15:20:27 As a way to restore meta-toolchain generation, drop bfd.mo and Jun 15 15:20:27 opcodes.mo from gdb-cross-sdk archives, as these files clash with ones Jun 15 15:35:08 rf_: yes thanks for testing. I have a local patch for that here Jun 15 15:37:48 khem: great, glad to help. I'm new to oe, hope to help more as I continue learning... Jun 15 15:39:13 03Khem Raj  07org.openembedded.dev * r83a48026dd 10openembedded.git/recipes/ltrace/ltrace_0.5.3.bb: Jun 15 15:39:13 ltrace_0.5.3.bb: Add binutils to DEPENDS Jun 15 15:39:13 * ltrace needs libiberty which is staged by Jun 15 15:39:13 binutils. Earlier it was staged via gcc-cross Jun 15 15:39:13 Signed-off-by: Khem Raj Jun 15 15:45:28 meh Jun 15 15:49:40 hi kergoth Jun 15 16:03:53 ericben : I'll be there tomorrow ; bye Jun 15 16:16:32 cbrake: made any progress with the srctree/gitver stuff? think i caught a bug of some sort in vegas this weekend, so a bit out of it, not sure if i'll have the energy to look into it today, but figured i'd check on the status Jun 15 16:16:35 have a nice rest of day Jun 15 16:20:41 pb_: hey Jun 15 16:21:49 RP: http://bit.ly/b1rHZJ Jun 15 16:30:20 re Jun 15 16:30:26 hi kergoth and khem Jun 15 16:31:40 woglinde: hello Jun 15 16:31:55 khem hm how can I sovle the problem when I link libgcc with gcc-cross and libgcc_s.so says I cannt find libc.so Jun 15 16:32:19 args s/I/it Jun 15 16:32:36 kergoth: I still need to solve the issue bitbake was having. RP replied that we need to get to the bottom of the problem and not mask it. Jun 15 16:32:45 kergoth: so I need to capture another log, which I have not done yet Jun 15 16:32:46 woglinde: hmmm I guess your toolchain is broken in some respect Jun 15 16:32:47 hmm, okay Jun 15 16:32:56 khem oe gcc-4.3.3? Jun 15 16:34:38 woglinde: is glibc/eglibc/uclibc staged ? Jun 15 16:34:46 glibc Jun 15 16:35:06 woglinde: and you can verify that libc.so existi Jun 15 16:35:12 in the sysroot Jun 15 16:35:51 yes Jun 15 16:35:57 the ugly script Jun 15 16:36:14 OUTPUT_FORMAT(elf32-littlearm) Jun 15 16:36:15 GROUP ( /lib/libc.so.6 /usr/lib/libc_nonshared.a AS_NEEDED ( /lib/ld-linux.so.3 ) ) Jun 15 16:38:57 could anyone try to fetch expat-2.0.1? Jun 15 16:39:34 and in another note, are checksum disabled? Jun 15 16:40:55 woglinde: ok Jun 15 16:41:00 zecke: fetched ok, checksums are not disabled Jun 15 16:41:37 JaMa: which sf.net mirror was used? Jun 15 16:41:48 I am getting CRC errors on unpack Jun 15 16:41:49 woglinde: can you reproduce it always say even on hello world Jun 15 16:41:52 zecke: don't you have problem with unpacking? (known bug in ubuntu gzip) Jun 15 16:41:58 khem mom Jun 15 16:42:05 zecke: check ubuntu bugtracker Jun 15 16:42:05 JaMa: ah thanks, I am on fedora Jun 15 16:42:31 JaMa: thanks Jun 15 16:42:41 zecke: hmm are fedora and ubuntu bug compatible :) Jun 15 16:43:16 maybe they upgraded gzip as ubuntu did.. Jun 15 16:43:41 khem hm args was problem with g++ and linking Jun 15 16:43:48 nice... :( Jun 15 16:46:22 woglinde: so its problem with g++ linking ? Jun 15 16:47:10 khem hm -> http://pastebin.com/sAY9xxi4 Jun 15 16:47:16 seems gcc too Jun 15 16:47:54 problem with -isystem? Jun 15 16:50:24 woglinde: and in libc.so in /usr/lib ? Jun 15 16:50:43 /devel/arm/oetmp-ang/sysroots/armv7a-angstrom-linux-gnueabi/usr/lib/libc.so Jun 15 16:52:43 cbrake: will look over the bitbake process handling, never know, maybe can spot something via inspection Jun 15 16:53:14 woglinde: ok can you add a -v switch to link command and paste the output for me Jun 15 16:53:18 kergoth: trying to pull away from other tasks to collect the log info ... Jun 15 16:54:07 heh, know how that is, thats why i haven't had a chance to look at it yet Jun 15 16:54:33 been working on the bitbake integration of my variable reference tracking / signature generation code, and the resurrection of my bitbake python logging module bits Jun 15 16:55:14 * kergoth goes for a walk with the dog for a bit Jun 15 16:56:35 khem hm -> http://pastebin.com/11Zqaee4 Jun 15 16:56:51 khem maybe this socket error is tje problem Jun 15 16:58:52 woglinde: this is in libc.so.6 so if that refernce is solved propertly then this error would go Jun 15 16:59:14 woglinde: I wonder if adding -lc solves it for a test Jun 15 16:59:21 khem hm yes Jun 15 16:59:28 thats what I thought too Jun 15 16:59:33 lets try again Jun 15 17:00:05 nope Jun 15 17:00:07 *sigh* Jun 15 17:02:20 zecke: yes, I've had trouble with expat. Are you seeing this issue: https://bugs.launchpad.net/ubuntu/+source/gzip/+bug/524366 Jun 15 17:03:02 khem hm -> http://pastebin.com/dGge2yLg Jun 15 17:03:05 thats with -v Jun 15 17:03:29 hi stefan_schmidt Jun 15 17:03:35 hi woglinde Jun 15 17:03:37 hi all Jun 15 17:04:00 khem maybee the problem is the first -lgcc_s Jun 15 17:04:02 woglinde: incomplete ? Jun 15 17:04:08 no Jun 15 17:04:29 I only wanted to show the seroius part Jun 15 17:05:23 cbrake: exactly Jun 15 17:05:45 khem dont understand it either Jun 15 17:05:59 why collect is using three times -lgcc_s Jun 15 17:09:51 woglinde: thats ok Jun 15 17:10:10 woglinde: but why it cant reach libc.so is issue Jun 15 17:11:11 hm binutils are 2.18 Jun 15 17:17:44 03Khem Raj  07org.openembedded.dev * rffd3912629 10openembedded.git/recipes/patchelf/patchelf_0.5.bb: Jun 15 17:17:44 patchelf: New recipe Jun 15 17:17:44 * Its used to modify rpaths in ELF files. Replacement for chrpath. Jun 15 17:17:44 Signed-off-by: Khem Raj Jun 15 17:18:10 interesting. does patchelf work better, or just an alternative? Jun 15 17:20:12 kergoth: its maintained so I suppose it will Jun 15 17:20:18 ah, that's a plus :) Jun 15 17:20:50 kergoth: plus I have to shoe this in into OE's relocatable.bbclass yet Jun 15 17:20:56 * kergoth nods Jun 15 17:21:08 I am tired of those chrpath warnings Jun 15 17:21:36 kergoth: btw. your ORIGIN patch was interesting why did you not push it Jun 15 17:21:41 was there some technical snag Jun 15 17:22:03 don't think so, think it just got stalled in patchwork limbo :) Jun 15 17:22:16 kergoth: it had few acks Jun 15 17:22:24 you may want to revive it Jun 15 17:22:26 there's more to it than just that though, since there are binaries in non-standard locations that need different relative paths Jun 15 17:22:40 those have to be handled case by case to do it that way Jun 15 17:22:42 woglinde: can you link a small hello world ok ? Jun 15 17:23:09 kergoth: yes I thought of that. but they must be small in number Jun 15 17:23:28 yeah, mostly binary plugins that need to link back to the library Jun 15 17:23:33 i.e. python, perl, etc Jun 15 17:23:47 iirc there's a couple headaches in gcc too, tom dealt with those Jun 15 17:24:02 Yeah, gcc/gdb/binutils are a minor headache Jun 15 17:24:06 perl is the sucky one Jun 15 17:24:11 python iirc isn't a big deal Jun 15 17:25:33 maybe a combination of $ORIGIN for standard and patchelf/chrpath for non-standard would be a way to do things Jun 15 17:26:01 perhaps Jun 15 17:26:27 * kergoth shrugs, his brain isn't at 100% today Jun 15 18:28:03 hi woglinde Jun 15 18:28:27 I'm trying to do something usable for navit on 1.4.3 Jun 15 18:28:39 but I wonder where is libnavit_la_LINK defined Jun 15 18:28:56 s/libnavit_la_LDFLAGS/libnavit_la_LDFLAGS// Jun 15 18:30:14 that regex is a no-op, no? :) Jun 15 18:32:51 lol Jun 15 18:33:11 s#/$## Jun 15 18:33:40 that doesn't change the fact that you're replacing libnavit_la_LDFLAGS with libnavit_la_LDFLAGS :) Jun 15 18:33:55 ah ok I didn't see that Jun 15 18:33:57 lol Jun 15 18:34:10 hehe Jun 15 19:14:52 03Tom Rini  07org.openembedded.dev * rd7074fcd05 10openembedded.git/recipes/oprofile/ (oprofile.inc oprofile/opcontrol-eval-fix.patch): Jun 15 19:14:52 oprofile: busybox shell needs we need a bit more quoting in opcontrol Jun 15 19:14:52 Signed-off-by: Tom Rini Jun 15 19:17:19 khem 0:0 Jun 15 19:17:40 woglinde, hi Jun 15 19:19:03 re Jun 15 19:19:08 khem 0:0 Jun 15 20:01:54 ERROR: Exception: Message:Variable 'do_fetchall' not captured by the signature for 'do_populate_sysroot' Jun 15 20:01:54 gah Jun 15 20:02:08 damnit, i can't figure out where this one is coming from Jun 15 20:03:23 kergoth: I've seen this before, but trying to remember what causes this: http://pastebin.com/iTDfsa5d Jun 15 20:03:54 (for this srctree recipe/src: http://cgit.bec-systems.com/cgit.cgi/autotools-demo/) Jun 15 20:04:38 git describe returns 1.1 Jun 15 20:04:45 hmm Jun 15 20:08:27 I may have updated my Ubuntu OS since I last played with this, so tracing through gitver now Jun 15 20:13:10 this is returning None: ref = popen(["git", "symbolic-ref", "HEAD"]) Jun 15 20:14:27 so it returned non-zero Jun 15 20:14:29 cbrake@happy:/scratch/oe/oe-build/openembedded.custom/recipes/autotools-demo$ git symbolic-ref HEAD Jun 15 20:14:31 on a detached head? Jun 15 20:14:32 fatal: ref HEAD is not a symbolic ref Jun 15 20:14:48 kergoth: yup, submodules :- Jun 15 20:14:50 kergoth: yup, submodules :-\ Jun 15 20:15:02 it shouldn't be failing like that Jun 15 20:15:05 look at gitver.bbclass Jun 15 20:15:12 oh, i see Jun 15 20:15:16 found it Jun 15 20:15:21 line 41 Jun 15 20:15:27 it uses ref before the 'if ref' check Jun 15 20:15:40 likely nobody tested it in a detached head :) Jun 15 20:15:56 heh, those crazy people using submodules ... Jun 15 20:16:16 hm Jun 15 20:16:28 seems brazil won Jun 15 20:16:31 just need to tweak that logic, move the reffile = into the if ref , and add the if based on reffile into there as a nested if, or so Jun 15 20:17:50 kergoth: ok, tweaking and testing Jun 15 20:17:58 cool Jun 15 20:20:44 kergoth: can you review: http://pastebin.com/EcimaHyV Jun 15 20:20:49 it seems to work Jun 15 20:21:17 looks reasonable to me Jun 15 20:22:09 kergoth: ok, I probably need two sign-offs on this Jun 15 20:22:19 so I guess I'll send to the mail list Jun 15 20:22:25 don't bother, its not a class that's core or high impact Jun 15 20:22:32 ok Jun 15 20:22:34 don't need double acks on every commit Jun 15 20:22:51 kergoth: guess I'm probably the only one using it right now :-) Jun 15 20:23:03 one or two others do, i think Jun 15 20:23:04 thats about it Jun 15 20:24:06 cbrake you have my ack too Jun 15 20:24:13 if its matter Jun 15 20:24:16 *g* Jun 15 20:24:59 03Cliff Brake  07org.openembedded.dev * r1a8df1e6dd 10openembedded.git/classes/gitver.bbclass: gitver.bbclass: fix issue with detached heads Jun 15 20:25:08 hm Jun 15 20:25:10 http://lists.freedesktop.org/archives/xcb/2010-June/006111.html Jun 15 20:26:57 gah! Jun 15 20:26:58 grr Jun 15 20:28:52 hmm Jun 15 20:29:00 yes? Jun 15 20:29:24 kergoth is making funny noises. Jun 15 20:29:28 indeed Jun 15 20:29:46 kergoth: you mentioned merging the exception handling fix to bitbake: http://cgit.bec-systems.com/cgit.cgi/bitbake/commit/?h=cbrake/1.10-srctree-fixes&id=15cd67608dd5a7564186d29ebb936bd110ace64f Jun 15 20:29:58 kergoth: has this been pushed to any of the public repos yet? Jun 15 20:30:48 cbrake: http://git.openembedded.org/cgit.cgi/bitbake/commit/?id=acca7eb8450cbe30d65b5b266f1a642e7965df8a Jun 15 20:31:02 instead of manually printing it, i changed it to let the toplevel handler do it Jun 15 20:31:05 just re-raised Jun 15 20:31:12 should give the same functionality Jun 15 20:31:20 kergoth: ok, I'll test that Jun 15 20:31:25 cool Jun 15 20:31:27 now that I'm back to that error Jun 15 20:31:39 master has a number of exception handling fixes, but there's still a lot to be done Jun 15 20:36:12 woglinde_: Did you try to reproduce the issue you are seeing with gcc 4.4.4+binutils 2.20.1 Jun 15 20:37:10 making funny noises because I've begun integration of variable reference tracking into bitbake, initially only constructing a signature for the metadata that a task uses, and raising an error if getVar() is run against something that wasn't captured by it.. and I can't figure out where all the references are coming from Jun 15 20:37:12 khem no Jun 15 20:37:17 no time for it now Jun 15 20:37:25 this time CHRPATH_BIN wasn't captured by the signature for do_populate_sysroot for some reason Jun 15 20:37:27 very strange Jun 15 20:37:46 it captures sysroot_binaries_preprocess, yet doesn't get the call to rpath_replace which is run in sysroot_binaries_preprocess Jun 15 20:45:18 kergoth: http://pastebin.com/Gs6cJqP7 it is better -- we don't get a stack trace though -- even with -D Jun 15 20:45:41 try using master. Jun 15 20:46:28 this is the master Jun 15 20:46:41 master doesn't suppress exceptions at toplevel at all Jun 15 20:46:48 debug or otherwise Jun 15 20:47:04 * kergoth looks over the code Jun 15 20:47:21 hmm Jun 15 20:47:29 * cbrake injects some exceptions Jun 15 20:48:22 khem ping Jun 15 20:48:29 khem the problem is accept4 Jun 15 20:48:32 kergoth: add exceptions in bitbake works: http://pastebin.com/KdY9vv5b Jun 15 20:48:43 which dont seemed defined in my gcc libc Jun 15 20:48:46 combonation Jun 15 20:48:57 kergoth: perhaps its only exceptions in modules that don't get traced Jun 15 20:49:06 that doesn't make any sense Jun 15 20:49:11 it doesn't care where it comes from Jun 15 20:49:28 it could be a concern if the client/server split was enabled, since the UI would be in its own process, but it isn't Jun 15 20:49:32 i dunno offhand Jun 15 20:50:11 03Koen Kooi  07org.openembedded.dev * r9191c2a222 10openembedded.git/recipes/efl1/evas_svn.bb: evas: disable neon for the time being, see http://thread.gmane.org/gmane.comp.window-managers.enlightenment.devel/25194 Jun 15 20:50:12 woglinde_: hmmm thats a syscall Jun 15 20:50:37 khem another example why dont like lennart Jun 15 20:50:38 hm.. Jun 15 20:50:59 woglinde_: which package is that ? Jun 15 20:51:12 systemd Jun 15 20:51:36 khem accept4 can open sockets nonblocked Jun 15 20:51:44 woglinde_: yes I know :) Jun 15 20:51:50 thats the only difference to accept Jun 15 20:51:55 ls Jun 15 20:53:35 kergoth: the finally statement is what causes the backtrace to get dropped Jun 15 20:54:02 that's.. odd. finally is for cleanup, it shouldn't change that behavior as far as i'm aware Jun 15 20:54:04 khem its not yet in oe Jun 15 20:54:07 pwgen started Jun 15 20:54:12 I am nearly finished Jun 15 20:54:26 woglinde_: ok Jun 15 20:54:33 kergoth: so it essence, the raise does not do anything Jun 15 20:54:42 whahahahahahahaaaa Jun 15 20:54:45 dman it Jun 15 20:54:55 why he calls systemd in the build process Jun 15 20:55:11 to catch its own tail Jun 15 20:55:24 ieehks for xml documentation Jun 15 20:55:34 ./systemd --introspect=org.freedesktop.systemd1.Manager > org.freedesktop.systemd1.Manager.xml Jun 15 20:55:38 03Khem Raj  07org.openembedded.dev * r473cd5f616 10openembedded.git/recipes/gnu-config/ (4 files in 2 dirs): Jun 15 20:55:38 gnu-config: Add recipe for git version. Jun 15 20:55:38 * Rename gnu-config to files folder. Jun 15 20:55:38 Signed-off-by: Khem Raj Jun 15 20:55:43 lets see if I can disbale doc Jun 15 20:56:22 no Jun 15 20:56:24 what a dork Jun 15 20:56:52 khem, did you have a chance to review my configury? Jun 15 20:57:02 blindvt: yes Jun 15 20:57:17 blindvt: I am still looking for pb to comment more Jun 15 20:57:35 kergoth: the following gets the traceback working if you decide that is desirable: http://pastebin.com/6AYkP1Ha Jun 15 20:57:56 i'd rather understand why it's behaving this way, but that'll do for now Jun 15 20:58:05 I am not so excited because I would have liked the kconfig features somehow mapped better to DISTRO and MACHINE features Jun 15 20:58:34 khem, pb_ is on vacation, is he? Jun 15 20:58:42 kergoth: yeah, thats odd, the finally makes it impossible to re-raise an exception Jun 15 20:58:42 oh is it ? I dont know Jun 15 20:59:07 blindvt: although your patch does not worsen the situation I think Jun 15 20:59:49 cbrake: hmm, mind testing something? Jun 15 20:59:55 kergoth: sure Jun 15 21:00:10 hi blindvt Jun 15 21:00:19 cbrake: try changing it to a nested try block. try/finally, and under that first try, try/except/else Jun 15 21:01:46 * kergoth 'll have to double check the python documentation Jun 15 21:02:34 khem, btw, i said i'm 50% through the nptl make nightmare. Yesterday i continued an add got another (!) 60% of the whole thing. By now i think i'm way in the second iteration and that one starts to look somewhat clean ;) Jun 15 21:02:47 kergoth: nope, still gets caught Jun 15 21:02:52 (in finally) Jun 15 21:02:57 very weird Jun 15 21:05:57 cbrake: "When an exception has occurred in the try clause and has not been handled by an except clause (or it has occurred in a except or else clause), it is re-raised after the finally clause has been executed." Jun 15 21:06:59 hmmm Jun 15 21:09:21 why the hell lennart introduced rootdir Jun 15 21:09:23 *sigh* Jun 15 21:09:36 ooh, coreutils-native:do_configure is executing Jun 15 21:09:40 woot Jun 15 21:10:39 hm which oevar was /usr or / ? Jun 15 21:10:49 prefix Jun 15 21:11:12 hmmm Jun 15 21:11:38 thanks Jun 15 21:12:21 i think the forced re-export of env vars might be about to bite me in the ass Jun 15 21:12:45 woglinde_, hi Jun 15 21:12:59 ieehks lennart did avahi too Jun 15 21:13:07 one point more to hate him Jun 15 21:13:31 exec_prefix is also independent of prefix and set to / or /usr Jun 15 21:15:25 kergoth: yeah, I really don't understand it yet -- I wrote some similar test code that behaves as expected Jun 15 21:15:35 hmm Jun 15 21:15:36 odd Jun 15 21:17:06 blindvt: | libm/libm_so.a(e_lgamma_r.os): In function `gamma_r': Jun 15 21:17:07 | e_lgamma_r.c:(.text+0x36): undefined reference to `__signbitl' Jun 15 21:17:18 thats the latest master on uclibc Jun 15 21:18:09 oh. Let me have a look Jun 15 21:19:50 kergoth: found it -- the return value in the finally is the culprit. This seems to fix it: http://pastebin.com/Zcu8B6bE Jun 15 21:20:08 aha Jun 15 21:20:11 good call Jun 15 21:20:18 khem, do you have long double math enabled? Jun 15 21:20:21 that makes sense, actualy Jun 15 21:20:29 kergoth: nod Jun 15 21:20:30 since exceptions are *reraised* after the finally Jun 15 21:20:33 but that finally never ends Jun 15 21:20:43 sheesh, going to have to remember that one Jun 15 21:22:19 khem, sed -i -e s/__signbitl/signbit/ libm/e_lgamma_r.c Jun 15 21:22:26 blindvt: its on arm Jun 15 21:22:43 cbrake: shouldn't need that return_value = None, its set to 1 earlier on Jun 15 21:23:07 cbrake: want to go ahead and do a format-patch of that fix and i'll apply it? alternatively, can add you to the bitbake committers if you feel comfortable with that Jun 15 21:24:07 blindvt: yes that will fix it. Care to commit ? Jun 15 21:24:22 khem, in a second, mompls Jun 15 21:30:04 hm did _files = "/etc" covers the /etc dir? Jun 15 21:30:10 +whole Jun 15 21:31:01 kergoth: ok, patch sent to the bitbake list -- please ignore the patches with return_value = None, a corrected one should follow Jun 15 21:31:30 woglinde_: generally you should only package file by file for perfect results Jun 15 21:31:49 khem for systemd it make sense Jun 15 21:33:33 woglinde_: use ${sysconfdir}/* Jun 15 21:34:19 khem right Jun 15 21:34:22 FILES_${PN} = " ${sysconfdir} should do it Jun 15 21:35:12 CONFFILES_${PN} = "${sysconfdir}" even better :) Jun 15 21:35:17 hm /lib Jun 15 21:39:01 cbrake: k, thanks Jun 15 21:49:16 cbrake: k, applied to master. Jun 15 21:51:47 kergoth: thanks Jun 15 21:52:08 gah, wtf.. somehow exec_func of staging_helper is resulting in an access to the CFLAG variable in openssl-native, yet CFLAG isn't exported, as far as i can see Jun 15 21:53:44 this makes no sense. Jun 15 21:56:17 hmm Jun 15 21:56:19 oh, right.. Jun 15 21:56:43 setscene is shell, do_configure is shell, bitbake emits all shell functions in case they call one another Jun 15 21:56:55 that explains how cflag is being used, but i thought i had already handled this case.. grr Jun 15 22:02:58 damn Jun 15 22:03:15 PACKAGES += "${PN}-gui" Jun 15 22:03:27 ${PN}-gui_files = "${bindir}/systemadm" Jun 15 22:03:29 working Jun 15 22:04:15 woglinde_: FILES_${PN}-gui = "${bindir}/systemadm" Jun 15 22:04:24 ags Jun 15 22:04:28 args Jun 15 22:04:34 thanks khem Jun 15 22:05:45 gr Jun 15 22:07:01 getting so, so close on this Jun 15 22:07:18 03Khem Raj  07org.openembedded.dev * r4721a47280 10openembedded.git/recipes/uclibc/uclibc_git.bb: Jun 15 22:07:18 uclibc_git: Bump SRCREV Jun 15 22:07:18 Signed-off-by: Khem Raj Jun 15 22:07:27 kergoth: what improvements are expected with var ref ? Jun 15 22:07:30 at the moment i'm actually being bitten by the fact that the old code doesn't track paths during expansion the way the new stuff does :| Jun 15 22:07:43 khem: long term it should make it so we can touch local.conf without reparsing the recipes Jun 15 22:07:49 my immediate goal, however, is signature generation Jun 15 22:07:56 creating a hash of the metadata or part of hte metadata Jun 15 22:07:57 kergoth: ah data consistency Jun 15 22:08:01 for e.g. pstage Jun 15 22:08:07 as well as potentially scrapping stamps entirely Jun 15 22:08:17 in favor of tracking the directories tasks touch + the metadata they use Jun 15 22:08:36 http://github.com/kergoth/OE-Signatures#readme Jun 15 22:09:38 i have a number of tasks successfully executing while only allowed to use variables the signature for the task included, but keep finding odd corner cases Jun 15 22:10:24 kergoth: may be good that you run into them we can find and fix many latent bugs Jun 15 22:10:28 khem, pb said this to the overall idea (IIUC), but better ask him for an explicit statement if you read that differently: http://www.mail-archive.com/openembedded-devel@lists.openembedded.org/msg07039.html Jun 15 22:10:43 khem: yeah, this is true, there is some crazy shit floating around the metadata Jun 15 22:11:32 blindvt: ok will pull in those patches and run some builds this week Jun 15 22:11:50 PACKAGES += "${PN}-gui" isnt working Jun 15 22:12:22 http://www.rt-embedded.com/blog/archives/pcd/ Jun 15 22:12:35 khem, TIA. Please let me know if you experience unforeseen trouble Jun 15 22:12:52 blindvt: sure. Jun 15 22:17:33 http://paste.pocoo.org/show/225865/ Jun 15 22:17:42 I have started to see this very recently Jun 15 22:24:27 PACKAGES += " ${PN}-gui " Jun 15 22:24:35 woglinde_: that was for you Jun 15 22:25:26 woglinde_: also make sure that this variable is not overwritten after this Jun 15 22:28:44 * kergoth grumbles Jun 15 22:30:37 this makes no sense Jun 15 22:30:41 hmm Jun 15 22:31:30 siteinfo_endianness is only used in do_configure here, and my expand() patch is bypassing the patch for do_configure's expansion.. so i can't see how this could possibly be called against siteinfo_endianness.. yet it is Jun 15 22:31:33 * kergoth scratches head Jun 15 22:32:36 siteinfo_endianness is also used in some distro.conf or machine.conf files IIRC Jun 15 22:33:43 only slugos from what i can see Jun 15 22:33:48 which isn't what i'm using Jun 15 22:33:55 * kergoth ponders Jun 15 22:34:01 yeah Jun 15 22:35:06 the annoying thing is that the current expand implementation doesn't track the path Jun 15 22:35:10 so i can't go back and see how i got there Jun 15 22:35:37 kergoth: good to always have debug prints Jun 15 22:35:54 * khem ponders food Jun 15 22:36:17 hmm Jun 15 22:37:28 * kergoth tests Jun 15 22:38:23 03Dmitry Eremin-Solenikov  07org.openembedded.dev * r024ce7a16f 10openembedded.git/recipes/lowpan-tools/ (lowpan-tools.inc lowpan-tools_0.2.1.bb lowpan-tools_git.bb): Jun 15 22:38:23 lowpan-tools: add 0.2.1 release Jun 15 22:38:23 Signed-off-by: Dmitry Eremin-Solenikov Jun 15 22:38:26 03Dmitry Eremin-Solenikov  07org.openembedded.dev * raa45596db9 10openembedded.git/recipes/opkg/ (opkg-nogpg-nocurl_svn.bb opkg-nogpg_svn.bb opkg_svn.bb): Jun 15 22:38:26 opkg, opkg-nogpg, opkg-nogpg-nocurl: make ${PN}-dev provide virtual/opkg-dev Jun 15 22:38:26 This will be used in task-sdk-base (instead of ipkg-dev). Also bump all Jun 15 22:38:26 PRs to force rebuild. Jun 15 22:38:26 Signed-off-by: Dmitry Eremin-Solenikov Jun 15 22:38:26 03Dmitry Eremin-Solenikov  07org.openembedded.dev * r7f86f9029c 10openembedded.git/recipes/gdb/gdb-cross-sdk.inc: Jun 15 22:38:27 gdb-cross-sdk: drop more files conflicting with binutils-cross-sdk Jun 15 22:38:27 Signed-off-by: Dmitry Eremin-Solenikov Jun 15 22:38:28 03Dmitry Eremin-Solenikov  07org.openembedded.dev * reeffe0a6ee 10openembedded.git/recipes/tasks/task-sdk-base.bb: Jun 15 22:38:28 task-sdk-base: replace ipkg-dev with virtual/opkg-dev Jun 15 22:38:29 ipkg is long EOL. However task-sdk-base still tries to pull it in via Jun 15 22:38:29 ipkg-dev dependency. Then it clashes with opkg pulled in via Jun 15 22:38:46 ooh Jun 15 22:38:49 i got past that issue :D Jun 15 22:38:50 DISTRO_FEED_CONFIGS and so building of meta-toolchain-gpe,-opie, etc. Jun 15 22:38:51 fails. This fix allows me to build those sdks w/o any problems. Jun 15 22:38:51 Signed-off-by: Dmitry Eremin-Solenikov Jun 15 22:40:39 Exception RuntimeError: 'maximum recursion depth exceeded in __subclasscheck__' in ignored Jun 15 22:40:42 that looks rather ominous Jun 15 22:41:09 whee, ncurses-native is compiling Jun 15 22:43:14 gah Jun 15 22:43:19 * kergoth grumbles Jun 15 22:43:42 spoke too soon, runqueue order changed on me Jun 15 22:43:43 hrmph Jun 15 22:44:36 hm dont understand it Jun 15 22:44:59 maybee the inherits overwriting it Jun 15 22:45:02 but PACKAGES = " ${PN} ${PN}-gui ${PN}-dbg " Jun 15 22:45:06 isnt working Jun 15 22:45:33 bitbake -e foo | grep \^PACKAGES, then check for the bits that aren't default with grep? Jun 15 22:45:33 heh Jun 15 22:47:07 meh, stomach and head out of whack again Jun 15 22:47:10 * kergoth goes to lay down a bit Jun 15 22:47:22 nite kergoth Jun 15 23:28:17 how come http://meego.com has shiny linux foundation logo but not http://openembedded.org Jun 15 23:30:35 because we have no association with the linux foundation, perhaps? Jun 15 23:40:38 * cbrake wishes the USB subsytem on his workstation would quit crashing :-\ Jun 15 23:40:48 getting as bad as windows ... Jun 15 23:51:20 kergoth: do you think it would be good to have an association with the linux foundation ? Jun 16 00:06:10 bbl Jun 16 00:34:23 cbrake: which distro ? Jun 16 00:34:49 dcordes: in what way should be associate **** ENDING LOGGING AT Wed Jun 16 02:59:56 2010