**** BEGIN LOGGING AT Fri Apr 20 02:59:58 2012 Apr 20 06:59:13 morning Apr 20 07:17:58 hi :) i almost made it! but now i see: http://pastebin.com/raw.php?i=u8cASit8 Apr 20 07:18:45 how can i fix that? where can i uncomment to build that package? i do bitbake angstrom-x-image Apr 20 07:32:47 ericbutters: It looks like there is a checksum mismatch for obexd-0.34.tar.gz. Either you've got the wrong tar ball or the recipe got the wrong hashes. Apr 20 07:34:58 kenws yes you are right. i fixed it with the correct values in the .bb file Apr 20 08:51:08 good morning Apr 20 08:52:17 hi :) Apr 20 09:20:23 how long takes a bitbake with x11 on your machines? on mine it takes a life. ;) Apr 20 09:21:39 ericbutters: a build from scratch on my quad-core takes up to 4/5 hours Apr 20 09:22:46 i run in a virtualbox .. i see NOTE: Running task 4513 of 10437 Apr 20 09:25:04 I usually have about 7000 tasks Apr 20 09:27:10 mckoan: thats my usual also Apr 20 09:28:01 hi, udev-182 recipe (meta-oe) does not seem to include sysvinit scripts. Does it depend on having systemd? Is there some intent that systems using meta-oe should have systemd now? Is meta-oe abandoning sysvinit? Should everyone move towards systemd? Is there risk that people stop testinng recipes against sysvinit? Apr 20 09:28:04 many questions :) Apr 20 09:28:10 is there some statement of intent around this? Apr 20 09:28:17 yes. i do bitbake angstrom-x-image with firefox added to recipe Apr 20 09:53:46 jonasdn: it's fair to say that meta-oe has been organised around systemd because the person who maintains it is focused upon supporting systemd Apr 20 09:53:46 hello, i'm trying to build oe, i got this error :FATAL: base-passwd-3.5.20: ftp://ftp.debian.org/debian/pool/main/b/base-passwd/base-passwd_3.5.20.tar.gz cannot check archive integrity Apr 20 09:53:47 the package is missing in the ftp Apr 20 09:53:47 what can i do to fix it? Apr 20 09:54:24 jonasdn: but we definitely want sysvinit to remain supported Apr 20 09:56:25 thank you for that answer Apr 20 09:57:10 but am I mistaken when I say that udev-182 seem to depend on systemd? at least it would'nt work on my system since there where no sysvinit scripts. Shouldn't the recipe then have a depend on systemd? Apr 20 10:08:46 jonasdn: I'm not sure, but I think the systemd authors have committed to continue supporting sysvinit recently, so it ought to work Apr 20 10:13:08 hmm, ok, but the recipe udev-182 does not install the sysvinit script, like udev-173 did, and does not inherit update-rc.d Apr 20 10:13:47 so when I upped the rev of meta-oe in my tree and it switched to 182, everything built fine and I booted up target and all was fine except udev wasn't running :) Apr 20 10:24:24 jonasdn: this seems to be where it got removed - http://comments.gmane.org/gmane.comp.handhelds.openembedded.scm/49741 Apr 20 10:42:00 morning all Apr 20 10:48:41 hi pb_ Apr 20 10:58:39 yeah, and with negative DEFAULT_PREFERENCE since it was a 'huge change', then I guess noone had time to get the scripts back or think about another way :) Apr 20 11:04:06 udev-182 removed the DEFAULT_PREFERENCE = "-1" Apr 20 11:04:44 http://patches.openembedded.org/patch/23905/ Apr 20 11:18:09 jonasdn: I'm not very happy about it either :/ Apr 20 11:18:47 The message says that patches to re-instate the sysvinit scripts are welcome but will be scrutinized, maybe I'll give it a go Apr 20 11:19:09 jonasdn: I know I'd appreciate that :) Apr 20 11:40:42 hi bluelightning ... How r u going Apr 20 11:41:01 Noor: good thanks and you? Apr 20 11:41:15 I am doing great as well Apr 20 11:41:24 Thanks bluelightning Apr 20 11:44:53 Hi Apr 20 11:45:27 How to extends the meta-toolchain? i need to add libboost-dev into it Apr 20 11:45:52 I have error like ERROR: Nothing PROVIDES 'virtual/x86_64-angstromsdk-linux-gcc' Apr 20 12:15:10 Does someone can help me about Nothing PROVIDES 'virtual/x86_64-angstromsdk-linux-gcc' ? Apr 20 12:15:55 Delkia: sure it should be virtual/x86_64-angstromsdk-linux-gcc? Tried without the virtual/? Apr 20 12:16:47 It's come from depends, i don't write that Apr 20 12:17:30 do you have the angstrom meta? Apr 20 12:18:06 yes Apr 20 12:18:12 my bb file are here http://pastebin.com/Hg0U1fju Apr 20 12:18:35 the task-pegase-toolchain-target works Apr 20 12:19:56 I think the name of your virtual sdk seems weird, I'll update my angstrom meta and see if I can find it Apr 20 12:21:45 mertsas: Thanks Apr 20 12:22:43 can't really find it, would be good to know which file depends on it Apr 20 12:24:11 when greping for angstromsdk, all I find is the provider of the sdk in the conf file Apr 20 12:40:47 mertsas: Which package can provide virtual/x86_64-angstromsdk-linux-gcc ? gcc ?? Apr 20 13:06:35 Delkia: guess that x86_64-angstromsdk thingy is just adding gcc, gdb etc, but don't know what it is Apr 20 13:08:41 it's more easy to create toolchain with the classic tree ;) Apr 20 13:12:36 Does anyone know how the initial cross gcc (which gets built using a different prefix than the binutils) finds the appropriate assember for the target? Apr 20 13:19:51 the gcc driver usually looks under the given prefix but falls back to the host system bin dir which may lead to pickup the wrong assembler Apr 20 13:20:15 ah, the initial gcc gets configured using --bindir and friends Apr 20 13:21:06 Did they have a way to add a librairie to the meta-toolchain ? Apr 20 13:21:20 boost-dev in my case Apr 20 13:22:07 * kenws doesn't know about meta-toolchain Apr 20 13:31:11 hm, I still think that building the (initial) gcc to a different prefix (from the bintulils) without specifying (--with-as and friends) where to find the assembler is not supported Apr 20 13:32:48 khem: ^^ Apr 20 13:39:14 The configury of gcc 4.6.3 is somehow able to find the cross assembler (even it's not in the prefix) and creates the following symlink to be used by the driver: Apr 20 13:39:16 /sysroots/x86_64-linux/usr/libexec/armv7a-vfp-oe-linux-gnueabi.gcc-cross-initial/gcc/arm-oe-linux-gnueabi/4.6.3/as Apr 20 13:40:52 while gcc 4.6.4 does not detect the cross assembler and the driver thenn picks up the one from the host system which obviously calls for trouble when actually building things (like eglibc-initial) Apr 20 13:57:08 Delkia: Isn't there already a recipe for boost that outputs a boost-dev package? Apr 20 13:58:10 i thinks yes Apr 20 13:58:51 In the OE classic tree, we need to define a host task, not in the layers tree Apr 20 13:58:59 to build the toolchain Apr 20 13:59:40 i try to rebuild it just with a target task with boost-dev package in RDEPENDS Apr 20 14:00:14 I'm afraid I can't help. I've never touched classic-oe Apr 20 14:07:32 khem: I just noticed that gcc-package-cross.inc creates the symlinks into libexec - that's how the driver of the initial gcc finds the target asm Apr 20 14:08:34 I didn't work for as I forgot to set the BINV to 4.6.4 in my gcc recipe - user error : ) Apr 20 14:10:39 but inserting symlinks into the libexec dir sounds nasty - I suppose --with-as and friends could be used if the prefix differs (I think this is the case for the initial gcc only) Apr 20 16:40:39 anyone confident with recipes-devtools/python/python-twisted_10.2.0.bb Apr 20 16:45:08 looks like it does want to ship a file in more than one package. Looking at package.bbclass, if I get it right, this is prevented and the file ends up being shipped in the first package declearing it in its FILES_ variable? Apr 20 16:46:35 gizero76: yes Apr 20 16:48:55 XorA: are you familiar with recipes-devtools/python/python-twisted_10.2.0.bb ? Apr 20 16:51:57 gizero76: no, you are correct that a file can only be in one package Apr 20 17:05:58 XorA: Ok, thx! I need to figure out how to fix this. What I see here is: python-twisted_10.2.0.bb does declare a few packages. Some of them are also listed in its RDEPENDS_${PN}. One of these dependency package, ${PN}-bin, ends up being empty, since the only file it should contain is already shipped in ${PN}-protocols which is listed first in PACKAGES. Now I have another recipe I'm creating which should RDEPENDS on python-t Apr 20 17:06:00 wisted. Since python-twisted-bin is in my recipe's dependency chain but it is not created (because it would be empty) I get an error in do_rootfs for my image Apr 20 17:08:22 I can solve this by switching package order in PACKAGES, but I think this is not a clean solution. Any suggestion how to fix this in a cleaner way? Apr 20 17:22:10 i got error creating the rootfs for angstrom-x-image see here: http://pastebin.com/raw.php?i=1rUAeHpf Apr 20 17:22:56 gizero76: it sounds like changing order in PACKAGES is correct thing to do if the file it should contain ends up in wrong package Apr 20 17:23:42 gizero76: if this is oe-core/meta-oe based I think there were some subtle re-orderings to the defaults so maybe no-one uses twisted so never noticed the change Apr 20 17:31:06 XorA: well, this is with oe-core/poky/yocto plus python-twisted imported verbatim from meta-oe. Maybe the check that triggers the error is yocto specific: I have no oe-core/meta-oe to test with. There is another suspect point here since the package unexpectedly being empty also declare another file that is missing in my tree (${libdir}/${PYTHON_DIR}/site-packages/twisted/spread/cBanana.so). One missing file + One file "stol Apr 20 17:31:08 en" by a previously populated package = "Not creating empty RPM package for python-twisted-bin". Apr 20 17:32:09 XorA: Reordering does fix but the recipe is still "incorrect" since it does implicitly request to stuff a file in two different packages Apr 20 17:33:16 gizero76: issue a patch then :-D Apr 20 17:34:39 gizero76: there is a high chance twisted has a very low user count so no-one ever noticed the error Apr 20 17:37:07 Issuing a patch is my intent! I need to get more confident with twisted build, since is actually just a dependent package for me. Thanks for the help Apr 20 17:48:54 gizero76: you can also use ALLOW_EMPTY = "1" but better fix is to sort out packaging which I think you have **** ENDING LOGGING AT Fri Apr 20 18:23:12 2012 **** BEGIN LOGGING AT Fri Apr 20 18:24:15 2012 **** ENDING LOGGING AT Fri Apr 20 18:25:00 2012 **** BEGIN LOGGING AT Fri Apr 20 18:26:19 2012 Apr 20 20:36:39 Hi, is there a way to build a single package only with openembedded? I first created a complete image second one I would like to modify one project which is included in oe Apr 21 00:03:27 * mr_science hits the road... **** ENDING LOGGING AT Sat Apr 21 02:59:59 2012