**** BEGIN LOGGING AT Mon May 09 02:59:58 2011 May 09 07:40:42 good morning May 09 08:22:40 03Martin Jansa  07master * re1f165605e 10openembedded.git/recipes/omgps/ (omgps/fix.capability.patch omgps_svn.bb): May 09 08:22:40 omgps: fix build with capabilities enabled May 09 08:22:40 Signed-off-by: Martin Jansa May 09 08:22:41 03Martin Jansa  07master * rc830d2a852 10openembedded.git/recipes/linux/ (linux-kexecboot_git.bb linux_git.bb): May 09 08:22:41 linux(-kexecboot)_git: bump SRCREV to rc6 May 09 08:22:42 Signed-off-by: Martin Jansa May 09 08:22:45 03Martin Jansa  07master * r8ab2bb3f27 10openembedded.git/recipes/linphone/ (linphone-3.1.0/fix.unused.variable.patch linphone_3.1.0.bb): May 09 08:22:45 linphone: fix build without NLS enabled May 09 08:22:45 Signed-off-by: Martin Jansa May 09 08:22:55 03Martin Jansa  07master * rbecea0796b 10openembedded.git/recipes/tasks/task-shr-feed.bb: (log message trimmed) May 09 08:22:55 task-shr-feed: drop emacs-x11, cacao, phoneme-advanced-foundation which currently don't build and blocks build for long May 09 08:22:55 * will reenable them after fixing those May 09 08:22:55 * emacs-x11 - qemu gets killed while running temacs May 09 08:22:55 the same with older qemu May 09 08:22:55 | LC_ALL=C qemu-arm -cpu cortex-a8 -s 1048576 -L /OE/shr-unstable/tmp/work/armv7a-oe-linux-gnueabi/emacs-x11-23.1-r3/qemu-treedir temacs -batch -l loadup dump May 09 08:23:00 | qemu: uncaught target signal 11 (Segmentation fault) - core dumped May 09 08:32:21 HI everybody! I'm new here.... I was looking for someone who could answer a couple brief question about the openembedded releases (tags like Release-2011.03 or Release-2010.12). Can someone help me please? May 09 08:32:53 wiki is good for newbies May 09 08:33:25 lorenzo: http://wiki.openembedded.org/index.php/Release-2011.03 May 09 08:35:28 Thank you :) but I needed the opinion of someone who actually used one of them since I'm part of a developer team interested in using such technology in commercial products May 09 08:39:14 then read also about yocto May 09 08:45:58 lorenzo: what's the matter? May 09 08:46:05 hi mckoam May 09 08:46:09 ups mckoan May 09 08:46:21 woglinde: moin May 09 08:47:08 hi rp May 09 08:48:52 The developer team I work for is trying to build a new image for our platform (based on a dm365 from texas) based on one of the release (angstrom - Release-2011.03 or Release-2010.12). I just wanted to know if anybody know the difference and to know what kind of confidence do you have in such releases (e.g for how long they'll be supported) May 09 08:49:36 lorenzo only 2011.03 is supported May 09 08:50:01 and it will be supported until oe-core and meta-oe are fully functional May 09 08:50:12 but thats not 100% defined May 09 08:50:39 lorenzo: usually each release is supported until the next will be added :-D May 09 08:51:06 lorenzo: ask to you PM he already knows this problem ;-) May 09 08:53:19 thank you do much. and what should I choose between angstrom-2008.1 and angstrom-2010.x? In my experience choosing the newest version is often the best solution since it makes easier to upgrade to possible future releases... What's your opinion? May 09 08:58:36 lorenzo: I use KaeilOS, I favour stability so I'd use angstrom-2008.1 May 09 09:00:15 s/favour/prefer May 09 09:02:20 Ok. I already did try to compile the qt-demo-images based on angstrom-2008.1 and angstrom-2010.x and they do seem to work both. The only big difference to me seems to be the use of the eglibc instead of glibc in the 2010 version. Do someone what's best? do eglibc work better in an embedded environment? Are they reliable as the glibc? May 09 09:04:06 yes and yes May 09 09:04:22 lorenzo: the latter is better, IIRC uses top-class Linaro patches, I planned to switch KaeilOS to that ASAP May 09 09:08:08 he hrw May 09 09:08:15 ho May 09 09:08:31 uclibc works better May 09 09:08:39 woglinde: does router applies to German law? May 09 09:08:39 but thats only my opinion May 09 09:08:54 hrw to be honest I didnt switched it on May 09 09:09:51 no problem May 09 09:09:51 lorenzo: it is planned that 2010 will suffer upheaval and breakage without warning May 09 09:10:00 lorenzo: thats why its on the development branch May 09 09:10:45 Ok... and do someone know if the eglibc are reliable? They're new to me and I didn't really understand the different and why they became the default in the new distribution May 09 09:10:47 hm but only 2011 has newer qt May 09 09:10:52 with quick and qml May 09 09:10:58 let me see May 09 09:11:02 lorenzo__: eglibc is glibc + extras May 09 09:11:13 lorenzo__: Debian and Ubuntu switched to it long time ago May 09 09:11:26 lorenzo__: so for me it is granted that it is reliable May 09 09:11:58 hm only 4.7.1 May 09 09:11:59 main difference seems to be eglibc doesnt reject patches for silly reasons like no-one runs arm cpus May 09 09:12:18 XorA: LOL May 09 09:12:21 So wich branch should I look for if I want the most stable/supported release? May 09 09:12:32 try 2011 May 09 09:12:48 maintenance-2011.3 May 09 09:16:16 When I did a quick search i found out that eglibc should be an Embedded GLIBC... didn't talk about extras May 09 09:16:38 thats only marketing May 09 09:19:18 So the angstrom 2010.x will be considered stable when the 'x' disappears? May 09 09:20:20 o.O May 09 09:21:25 It's just that the 2008 version is now marked as angstrom-2008.1 while 2010 has an 'x' vesrion May 09 09:21:26 when we next fork out a stable branch May 09 09:21:40 .dev will never be considered stable May 09 09:21:58 in fact there will be no more Angstrom releases from .dev anyway, it will be oe-core based May 09 09:23:25 what do you mean by oe-core? May 09 09:23:48 lorenzo__: OE and Yocto are switching to layered system of recips May 09 09:24:04 oe-core is base set of recipes and users/vendors will add own ones May 09 09:24:22 and oe-core has strict rules on applying changes May 09 09:24:31 more strict that .dev ever had May 09 09:25:11 03Koen Kooi  07org.openembedded.dev * r903e9b928e 10openembedded.git/recipes/powervr-drivers/libgles-omap3.inc: May 09 09:25:11 libgles-omap3: prune source a bit better May 09 09:25:11 Signed-off-by: Koen Kooi May 09 09:25:15 Do you know how this will affect the angstrom distro in OE? May 09 09:25:23 03Koen Kooi  07org.openembedded.dev * rd721dd947e 10openembedded.git/recipes/powervr-drivers/ (libgles-omap3.inc libgles-omap3/rc.pvr): May 09 09:25:23 libgles-omap3: make initscript work with both busybox and bash May 09 09:25:23 From the bash man page: May 09 09:25:23 "The reserved word function is optional. If the function reserved word is supplied, the parentheses are optional." May 09 09:25:23 Signed-off-by: Koen Kooi May 09 09:26:43 lorenzo__: just told you, future Angstrom releases will be oe-core based May 09 09:27:41 gm obi May 09 09:27:51 gm :) May 09 09:28:14 Ok thanks... I'll just investigate more the impact of the new oe-core approach May 09 09:37:12 03Koen Kooi  07org.openembedded.dev * r85f1bf1888 10openembedded.git/recipes/powervr-drivers/ (libgles-omap3_4.03.00.02.bb omap3-sgx-modules_1.6.16.3977.bb): May 09 09:37:12 libgles-omap3 4.03.00.02: make it the default May 09 09:37:12 Signed-off-by: Koen Kooi May 09 09:37:32 lorenzo yes you should May 09 09:37:47 lorzenzo google for yocto and oe-core May 09 09:38:44 morning woglinde May 09 09:38:49 uhm May 09 09:38:52 ka6sox May 09 09:38:57 go to sleep May 09 09:39:06 don't I wish May 09 09:49:37 gm florian May 09 09:52:42 hi woglinde, florian, ka6sox-away, all May 09 09:53:08 gm pb May 09 10:06:54 hi mickey|office May 09 10:07:06 heho mickeyl May 09 10:08:52 good morning pb_ May 09 10:08:57 how's the house progressing? May 09 10:09:15 mickeyl will you now come to linuxtag? May 09 10:09:48 mickey|office: pretty good. fermacell about 75% done, the carpenters are installing window cills this morning. May 09 10:10:55 nice, when is the ETA for completion? May 09 10:11:15 woglinde: i plan to come on friday, _if_ Lara Marie doesn't arrive :) May 09 10:11:29 mickey|office: about three months ago :-} May 09 10:11:32 pb_: oops ;) May 09 10:11:36 mickeyl oh May 09 10:11:39 good products are always late ;) May 09 10:11:44 heh, yeah May 09 10:11:54 updated eta, not sure, but hopefully weeks not months now May 09 10:12:07 right. i guess you're tired of camping by now aren't you? May 09 10:12:11 mickeyl only friday? May 09 10:12:16 yeah, though it isn't so bad in the summer May 09 10:12:29 we will have a nice summer this year May 09 10:12:37 woglinde: yes. Sabine doesn't allow me to stay overnight... May 09 10:12:44 and i don't want to miss the birth May 09 10:12:45 and we can use bits of the house already, so that's nice too May 09 10:12:54 right May 09 10:13:15 mickeyl umh then better stay at frankfurt May 09 10:13:41 well, i think it's ok if i come on friday, but on the first sign of any progress i will take the next train back May 09 10:13:59 that might be to late May 09 10:14:15 mickey|office: heh, you are a brave man May 09 10:15:15 lunch May 09 10:15:15 actually... i'm the best man for a very good friend's wedding on 3th of June. It is my belief that she will come on 3th to spoil the wedding ;) May 09 10:15:26 heh May 09 10:15:42 the one who marries will also be her godmother, so she's going to introduce herself very well :D May 09 10:25:20 morning pb_ mickey|office May 09 10:25:32 hi ka6sox- May 09 10:26:43 mickey|office, can you point patchwork.openembedded.net to melo.openembedded.org as a cname? May 09 10:28:05 pb_, I think github will be enough. May 09 10:28:43 righto, cool May 09 10:29:10 03Martin Jansa  07master * r980407c047 10openembedded.git/recipes/shr/libphone-ui-shr_git.bb: May 09 10:29:10 libphone-ui-shr: bump PR to rebuild against new glib May 09 10:29:10 Signed-off-by: Martin Jansa May 09 10:30:42 so #25 can be marked "resolved" May 09 10:31:20 done May 09 10:34:02 pb_, thanks...it took you 7-8hrs to finish what I've wanted done for 6months...Bravo May 09 10:34:36 uhm? May 09 10:38:10 speaking of github, we should make some notes on using it May 09 10:38:28 so people know to clone and not just push the entire repo again May 09 10:52:10 Good Plan(tm) May 09 11:03:45 03Koen Kooi  07org.openembedded.dev * r74d3a98616 10openembedded.git/recipes/linux-firmware/linux-firmware_git.bb: May 09 11:03:45 linux-firmware: update SRCREV and make the wl12xx subpackage work with a wider range of kernels May 09 11:03:45 Signed-off-by: Koen Kooi May 09 11:55:58 XorA, did you send anything out? May 09 11:56:25 ka6sox-away: no, realised cant do the test way I wanted to then Crofton|work fell asleep again May 09 11:57:04 ur May 09 11:57:07 what? May 09 11:58:47 XorA, you really found switching versions of 78000 series regualtors? May 09 11:59:28 Crofton|work: http://uk.rs-online.com/web/search/searchBrowseAction.html?method=getProduct&R=6727155 May 09 11:59:35 Crofton|work: plugin replacements May 09 11:59:46 crazyiness May 09 12:06:06 Crofton|work: you have devcices with LM78xx? May 09 12:06:24 no May 09 12:06:36 But, back in the good old days I would use them :) May 09 12:06:53 Crofton|work: I plan to replace the ones in zx spectrum to make them run cool May 09 12:07:15 is the inductor internal? May 09 12:08:12 must be May 09 12:58:01 When I run something like "bitbake virtual/kernel -c configure", would I then find it's configured source code in: oe_come/tmp/work// ? May 09 12:58:30 bioster: yes May 09 12:58:54 in the git subdirectory of the target_machine? May 09 12:59:36 if kernel sources are taken from git repo (not release tarball) then usually oe_come/tmp/work///git May 09 13:00:12 and that would be the patched version, if it needed patches? May 09 13:00:16 if you add -v to bitbake call, you should see where are the sources unpacked May 09 13:00:21 yes May 09 13:00:36 Ok, thanks... May 09 13:00:42 -c unpack is without patches applied, -c patch is with patches and -c configure with patches and configured ;) May 09 13:01:19 Now if I have a kernel module that I need to compile against that version, I'd still need to use the cross compile toolchains, right? May 09 13:01:33 right May 09 13:01:42 yeah, that's where I'm getting stuck May 09 13:02:02 kernel headers are staged in sysroot May 09 13:02:04 I can't seem to get the recipe right (I'm a newb to this) May 09 13:02:20 so you can still build against it see recipes for examples May 09 13:03:03 but then -c configure is not enough.. you have to build and stage that kernel first May 09 13:03:14 can I use the toolchain I built without having to get a recipe working? May 09 13:04:07 yes, ie you can build devshell and use that May 09 13:04:27 devshell? Ok, I'll look that up May 09 13:04:43 recipes/addons/devshell.bb May 09 13:12:00 hmm... maybe I should try to get this recipe right a little more before diving into a new tool May 09 13:12:19 are there any kernel module recipes that I could look at, to base my recipe on? May 09 13:12:35 I found something on google and adapted it, but it's not really working for me May 09 13:16:28 03Koen Kooi  07org.openembedded.dev * re89f1115e4 10openembedded.git/recipes/connman/ (connman.inc connman_0.73.bb): May 09 13:16:28 Connman: add dnsmasq-dbus to RRECOMMENDS to fix DNS proxy May 09 13:16:28 Connman writes '127.0.0.1' to resolv.conf by default expecting dnsmasq to be there May 09 13:16:28 Signed-off-by: Koen Kooi May 09 13:16:28 Acked-by: Martin Jansa May 09 13:46:00 hey all. May 09 13:50:05 03Koen Kooi  07org.openembedded.dev * r788de34b9b 10openembedded.git/recipes/xorg-lib/ (5 files in 2 dirs): May 09 13:50:05 pixman: add 0.22.0 - latest stable release May 09 13:50:05 Signed-off-by: Koen Kooi May 09 14:23:17 Hi May 09 14:23:53 Anyone using linux-omap3 (sakoman) 2.6.36 and UBIFS ? May 09 14:25:13 davidlt: just read your email -- I think there are still upstream issues with ubifs in 2.6.36 May 09 14:26:03 I'm testing 2.6.39 this week -- there are some patches that look like they might address the uifs issue May 09 14:26:20 sakoman: I am Googling again this issue and it looks like people have problem on 37 too (e.g. http://plugcomputer.org/plugforum/index.php?topic=4324.0) May 09 14:26:45 right, that is why I didn't even mention .37 or .38 :-) May 09 14:27:44 This is not that great May 09 14:29:09 .39 is still far away from being stable? May 09 14:29:14 indeed. it is amazing that it has gone broken for so many kernel revs May 09 14:29:32 shows number of real users I guess May 09 14:29:45 who are not stuck with older kernel.. May 09 14:29:52 davidlt: .39 is at rc6 May 09 14:31:30 Any good alternatives for UBIFS? May 09 14:34:52 not that I am aware of May 09 14:35:17 hm, I wonder what kind of ubifs issues are you guys talking about May 09 14:35:29 you could try 2.6.39 and see if things are working now May 09 14:38:18 ynezz: on boot system fails to load from UBIFS partition: VFS: Cannot open root device "ubi0:overo-rootfs" or unknown-block(0,0) May 09 14:38:45 I think this started to happen in .35/.36 May 09 14:39:13 ah May 09 14:39:46 so it's "just" related to omaps, right? May 09 14:40:40 we had problems with ubi images since previous mtd-utils upgrade, while still on 2.6.34... May 09 14:41:31 and only when it was reflashed with dfu-util, when written with nandwrite everything worked fine May 09 14:41:43 .34 worked fine for me, haven't noticed any problems May 09 14:42:18 does it work if you downgrade kernel on the same image? May 09 14:42:57 Should be, I worked on .34 but I had to add WiFi AP, which required .36. Moving to .36 broke UBIFS. May 09 14:45:01 But mine OE is files are outdated for 2-3 months now. May 09 15:01:35 ok, so I'm trying out devshell, but it seems like it's still trying to compile for x86 rather than arm May 09 15:01:49 if I cd into: /oe_home/tmp/sysroots/beagleboard-angstrom-linux-gnueabi/kernel May 09 15:02:23 and then run make, it complains about no Makefile in the arch/x86/ directory May 09 15:05:01 run oe_runmake May 09 15:05:03 not make May 09 15:05:11 oh, ok... thanks May 09 15:05:27 nah, same thing May 09 15:06:01 yeah, looks like when I run make it's actually doing oe_runmake May 09 15:06:36 03Koen Kooi  07org.openembedded.dev * r0e59a0832f 10openembedded.git/recipes/nmap/nmap_3.81.bb: May 09 15:06:36 nmap 3.81: fix license May 09 15:06:36 Signed-off-by: Koen Kooi May 09 15:07:31 do I need to do anything interesting to get devshell to use the correct target arch and such? I was under the impression that it did that automagically May 09 15:07:40 it shoudl do May 09 15:07:52 I havent used it for years though May 09 15:08:18 hmm, well basically I'm trying to compile a kernel module for a cross compiled kernel (for the beagleboard) May 09 15:08:39 honestly I don't care too much about how I get it done, but I've been having trouble figuring out how to write a recipe for it May 09 15:08:54 so I was thinking I'd try "just compiling" it May 09 15:08:56 devshell is no different from any other task. its context is exactly the same as every other task, wrt variables May 09 15:09:54 bioster: used the cross compiler in the sysroots to compile the modules against the stage version of the kernel source May 09 15:12:05 ok, keeping in mind I'm a complete newb... I just exited the devshell, and I'm in tmp/sysroots/beagleboard-angstrom-linux-gnueabi May 09 15:12:11 I only see the kernel directory May 09 15:12:43 * XorA is just about to leave work so wont have time to get you through it May 09 15:12:57 is the cross compiler in the x86_64-linux directory instead of the beagle direactory? May 09 15:13:57 well, I appreciate the time you've taken already May 09 15:14:38 * bioster is strongly considering just posting this to elance or something. May 09 15:18:36 bitbake -c devshell foo .. echo $CC in the devshell terminal May 09 15:19:40 well, I was starting up devshell manually rather than through bitbake May 09 15:19:48 but when 'echo $CC', I get: May 09 15:19:55 arm-angstrom-linux-gnueabi-gcc -march=armv7-a -mtune=cortex-a8 -mfpu=neon -mfloat-abi=softfp -mthumb-interwork -mno-thumb --sysroot=/home/bioster/oe_home/tmp/sysroots/armv7a-angstrom-linux-gnueabi May 09 15:47:44 ack May 09 15:47:55 kergoth: so I tried starting up devshell the way you suggested May 09 15:48:05 and I get: oe_runmake: command not found May 09 15:48:23 when I try to oe_runmake after starting up devshell May 09 15:55:32 hi zecke May 09 15:57:20 pb_: hey! May 09 16:04:31 bioster: devshell is a subprocess, a child. shell functions like oe_runmake can't make it from one process to another, as they aren't environment variables. May 09 16:04:45 bioster: you can just "make $EXTRA_OEMAKE" or so May 09 16:36:05 Has anyone had the problem of fastjar building invalid jar files (and taking way too long for that)? May 09 16:45:28 Fastjar is used in compilestepone of icedtea6, where it updates rt.jar from icedtea6-native with newly built classes. May 09 16:46:41 This step has taken ~20 min. on my not-so-slow laptop and generated a broken rt.jar. When updating the jar with the "real" jar command from icedtea6-native, the build can continue. May 09 16:48:07 is anyone in .eu using github to pull and pushing back to git.oe.org? May 09 16:50:00 what do you mean? May 09 16:51:12 ah, sorry, re-read it again :) May 09 16:52:30 used do to it like in 10/2010, but I think, that it was opposite oe->github May 09 16:52:51 dunno, it should work in both directions, it's git May 09 16:53:18 I do it with kernel.org<->github May 09 16:53:26 if it helps somehow May 09 16:54:14 ynezz, the hope is that folks will pull from the larger mirrors of github, especially european devs. May 09 16:55:01 but the master is and for the forseeable future will be git.oe.org May 09 16:55:07 ka6sox-away: did you update wiki to reflect the mirror May 09 16:55:09 it's fine with me May 09 16:55:22 since there're mirrors now May 09 16:55:35 khem, I wanted testers first before making a general announcement. May 09 16:55:53 ok May 09 16:55:53 then we can put it on page1 May 09 16:55:57 I can test it later today if you want to May 09 16:56:04 ynezz, please May 09 16:56:09 but it should work, I can't imagine it won't May 09 16:56:26 it will 100% work, I'm that sure May 09 16:56:28 :) May 09 16:56:39 their might be some latency due to populating mirrors. May 09 16:56:48 but not more than a minute or 2 May 09 16:57:00 depending on which hook pb_ used May 09 16:57:03 ah, this one May 09 16:58:27 well, there might be some kind of an race and you would need to wait little bit and then try to push again, no big deal I think May 09 16:58:38 khem, we are phasing out of org.oe right? May 09 16:58:49 at least dev.oe.org right? May 09 17:00:01 if everyone pushes/pulls to git.oe.org then the mirrors being RO won't have a race...only latency May 09 17:00:13 yes, latency May 09 17:00:31 so worst case is another repull May 09 17:00:35 yep May 09 17:00:50 ka6sox-away: yes master is what people should use May 09 17:01:31 but he means probably this scenario, you've wip branch tracking github/master and you push to git.oe.org/master May 09 17:01:42 thats ok May 09 17:01:53 we all do that now May 09 17:02:03 if you clone oe.org it checks out master May 09 17:02:18 org.oe.dev is just alias May 09 17:02:22 and you can hit some period that there's new commit in git.oe.org/master May 09 17:02:31 but not in the github/master yet May 09 17:02:46 ynezz, correct May 09 17:02:48 so it's desynced and push to git.oe.org/master would fail May 09 17:02:56 right May 09 17:03:01 if you don't use -f :) May 09 17:03:12 you have to re-pull and then repush May 09 17:03:41 but forced push shouldn't be allowed, so it's ok May 09 17:03:57 correct..sure way to make a mess May 09 17:04:40 khem, I meant the "openembedded" repo would slowly fade away as the layers take over. May 09 17:05:02 ka6sox-away: I dont know May 09 17:05:07 that will not happen May 09 17:05:20 we still have a long term release maintained on it May 09 17:05:48 I think the slowly fade away model is fine May 09 17:05:59 where slowly is determined by usage May 09 17:06:59 right as they become replaced by meta-oe then I suspect it will become an orphan. May 09 17:30:48 03chase maupin  07org.openembedded.dev * r5502a2c568 10openembedded.git/recipes/ncurses/ (ncurses/ncurses-5.7-20110115.patch.gz ncurses_5.7.bb): (log message trimmed) May 09 17:30:48 ncurses: add missing patch file May 09 17:30:48 * The 20110115 patch file has gone missing. Add it to the May 09 17:30:48 OE repository like the previous patch file. May 09 17:30:48 * Bump the PR May 09 17:30:49 Signed-off-by: Chase Maupin May 09 17:30:50 Acked-by: Paul Menzel May 09 17:54:14 likewise: hey, did you prepare the poster for the LinuxTag? or am i confused with nicknames again? :( May 09 17:57:05 ka6sox: I am willing to help test the git setup for you. May 09 17:57:40 * darknighte does a little dance for the missing ncurses patch being added by chase.... May 09 19:20:44 Hello; I just sent the fixes patches for oe-core and meta-oe May 09 19:24:02 otavio, are they all broken up? May 09 19:24:19 ka6sox-away: broken up? May 09 19:24:30 separate submittals. May 09 19:24:36 not one big fix. May 09 19:25:27 ka6sox-away: yes May 09 19:25:30 ka6sox-away: sure May 09 19:25:44 they are 1 big one or separate/ May 09 19:25:45 ? May 09 19:25:51 ka6sox-away: broken up May 09 19:25:54 heh May 09 19:26:22 okay...usually when my cellphone truncates things I worry on the ML that patchwork will be stuffed up. May 09 19:27:00 oh no May 09 19:27:07 khem sent an 24mb patch May 09 19:27:18 why the ml didnt reject it May 09 19:27:18 OMG May 09 19:27:19 rs May 09 19:27:30 hehe May 09 19:28:09 because that my kmail "Received a problem" May 09 19:29:46 * ka6sox-away shakes his finger @ khem....hoping that patchwork isn't going to choke. May 09 19:31:10 ka6sox-away: you got the testing you needed on git? May 09 19:31:32 I'm trying to figure out why my sdk built from OE doesn't work for autoconf projects yet the poky sdk does. can anyone explain in a nutshell what './configure --host=' does and how it relates to things in the PATH or other vars set by environment-setup in the SDK? May 09 19:31:37 darknighte, not yet... May 09 19:32:09 ka6sox-away could please install a filter May 09 19:32:11 I see poky also defines CONFIGURE_FLAGS but this doesn't seem to be used in configure scripts that I can see and I assume its more of a suggestion than anything else (poky sdk documentation does not specify using it) May 09 19:32:11 woglinde, you are in .de right? May 09 19:32:17 for the email May 09 19:32:21 ka6sox yes May 09 19:32:45 woglinde, I thought that khem did May 09 19:33:11 woglinde, can you please pull from github for the oe-core repo? May 09 19:33:41 and then make a change (assuming you can write back to that one) May 09 19:33:48 otherwise make it meta-oe May 09 19:33:56 ka6sox-sick: yes I thought twice May 09 19:34:05 but gave it a try May 09 19:34:20 ka6sox later maybee May 09 19:34:23 * ka6sox-sick looks @ process times for hung python process. May 09 19:34:29 woglinde, anytime you have time. May 09 19:34:51 the idea is that we should be able to pull from github and push back to git.oe.org May 09 19:35:09 darknighte, I need testing in US too please. May 09 19:42:47 ka6sox-sick: so pw hung again May 09 19:44:16 kick khem May 09 19:44:25 you can fix it :D May 09 19:44:45 ok I punted the huge message May 09 19:56:40 this breakup of recipes to different layers makes finding recipes difficult May 09 19:57:27 has building with the new layer based repo paths been documented yet? May 09 20:01:17 tharvey: for starts one only needs oe-core May 09 20:01:37 then one can mix and match the layers if needed May 09 20:01:59 best example is so far I know is angstrom setup May 09 20:01:59 you mean openembedded-core? May 09 20:02:04 yes May 09 20:02:49 http://www.angstrom-distribution.org/building-angstrom May 09 20:03:35 you need to checkout oe-core branch of setup-scripts repo once you clone May 09 20:03:53 ok, I'll check that out and see if it makes sense May 09 20:03:57 ah...ok, good to know May 09 20:04:00 sure May 09 20:04:27 speak up if you run into issues May 09 20:05:51 khem, think your the resident toolchain expert... could you give me any info on how an autoconf project utilizes info from sdk environment-setup? I'm trying to understand why my oe-based sdk fails on autoconf projects and for example poky's does not - the 'only' think poky sdk does is 'configure --host=arm-poky-linux-gnueabi' May 09 20:06:28 trying to understand in general what autoconf does to know where to look and make sure my environment-setup is correct May 09 20:06:59 tharvey: paths I would think May 09 20:07:12 can you paste the two environment-setup scripts ? May 09 20:07:45 thats what I'm thinking - I would guess that my pkg-config is not getting used appropriately as autoconf is still resorting to looking at / for things May 09 20:08:25 tharvey: it has to use pkg-config from SDK May 09 20:09:02 if it uses this from the host then it will be broken since we use sysroot capabilities of pkg-config May 09 20:09:06 before I posted anything I wanted to build meta-toolchain from latest and verify against that to make sure I didn't bork anything myself (likely) as I create a custom sdk May 09 20:09:39 ok May 09 20:09:39 should I be building release-2012-03 or trunk based on new repo layout? May 09 20:09:58 meta-toolchain should be a lot better with oe-core+meta-oe ansgstrom May 09 20:10:25 tharvey: the new one May 09 20:17:30 khem, regarding pkg-config do you know why something like 'pkg-config gstreamer-0.10 --cflags' would report back proper -I but a --libs flag would not report back proper -L? May 09 20:17:58 proper you mean pointing into sysroot ? May 09 20:18:23 well... using --libs doesn't report 'any' -L, just -l May 09 20:18:46 enabling --debug says 'Package GStreamer has -L/usr/lib in Libs May 09 20:18:46 Removing -L/usr/lib from libs for gstreamer-0.10 May 09 20:18:46 ' - not clear what that means May 09 20:18:51 usually -I and -L are prepended May 09 20:18:54 with sysroot May 09 20:19:08 somebody have a tzdata2011e.tar.gz to upload to me? the server of it is down :\ May 09 20:19:40 khem, right, but take this output for example (from poky sdk): pkg-config --libs gstreamer-0.10 May 09 20:19:40 -pthread -lgstreamer-0.10 -lgobject-2.0 -lgmodule-2.0 -lxml2 -lgthread-2.0 -lrt -lglib-2.0 May 09 20:19:49 why isn't there any '-L' there? May 09 20:20:16 ok are you using libtool 2.4 May 09 20:20:37 on my host, or is libtool part of the sdk? May 09 20:20:53 sdk May 09 20:21:43 hmm... poky sdk has no libtool that I can find May 09 20:24:22 generally -L paths should be returned pointing into sysroot May 09 20:24:25 if all is well May 09 20:24:58 makes sense, but not seeing that with poky's sdk which otherwise appears to do the right thing May 09 20:25:15 meaning it properly cross-compiles something like libgd for example... what I'm testing with May 09 20:26:19 if you dont use -L and just have -lc May 09 20:26:38 then also it will work since gcc should be able to find the library in sysroot May 09 20:26:52 is sysroot fed into linker command ? May 09 20:29:22 in the poky sdk case, I see it using -L, just not clear where it gets this if pkg-config --libs doesn't reportit May 09 20:30:07 I'm not sure I understand where libtool factors in honestly - from what I can tell pkg-config is used for 'using libs' and libtool is used for 'building libs' right? May 09 20:30:33 in the case of libgd, we are building a lib that uses other libs (freetype, libz, etc) which is why I selected it for a test case May 09 20:36:57 ok May 09 20:37:04 whats failing for you May 09 20:37:11 which package ? May 09 20:37:34 either libtool is used to link May 09 20:37:37 no package... I'm taking an sdk that I built off a customized meta-toolchain.bb and trying to build libgd against it May 09 20:37:51 then it takes care of findin the libs and addding paths to linker commandline May 09 20:38:03 ok May 09 20:38:29 which pkg-config is beingused May 09 20:39:23 pkg-config is in the SDK path, thus the SDK's pkg-config is being used, and it does report the right thing for --cflags but I don't really agree wiht what it reports for --libs as I mentioned (I could simply not understand what its supposed to do) May 09 20:39:45 in the case of libgd I don't know what its configure script is looking for... autoconf is pretty daunting May 09 20:40:32 I'm sitting here comparing libgd's configure output from poky's sdk which appears to build it properly, and my OE-derived sdk which doesn't May 09 20:41:23 in general... i'm not familiar with what makes an sdk work once you get away from simple gcc/ld makefiles to autoconf which is why I'm having such difficulty debugging this May 09 20:42:11 --libs puts all linker flags May 09 20:42:20 you could use --libs-only-l May 09 20:42:24 or --libs-only-L May 09 20:42:34 to get specific flags May 09 20:44:05 right... I understand that. but 'pkg-config --libs-only-L gstreamer-0.10' for example produces 'nothing' and I can't understand how yet this sdk still can build libgd May 09 20:44:35 it can if sysroot is correct May 09 20:44:40 fed into linker May 09 20:44:48 (gstreamer is ahorrible example as libgd doens't use it) May 09 20:45:10 so where does sysroot come from then when autoconf is used? May 09 20:45:47 --sysroot option or builtin default May 09 20:46:48 options to configure? those don't appear valid via libgd's configure --help May 09 20:48:14 I better get a 'stock' meta-toolchain built so that I can compare with that and post some results May 09 20:49:07 tharvey: talk code May 09 20:49:17 post the errors and logs some where May 09 20:49:19 so going back to using release-2011-03 vs oe-core - still trying to understand the new repos: are you saying that old repo 'openembedded' has basically been changed to 'openembedded-core' May 09 20:49:20 i am confused May 09 20:49:38 no oe-core is different May 09 20:49:40 ya, thats what I want to do but for the life of me I can't seem to follow the changes that have been done to the oe repos May 09 20:49:54 OE's website has not been updated to explain this changes that I can find May 09 20:50:04 yocto is now using oe-core which is pokylike May 09 20:50:17 its still under work May 09 20:50:56 ok so I should simply stick with git://git.openembedded.org/openembedded May 09 20:51:10 depends what u want May 09 20:51:13 as I said May 09 20:51:33 oe-core might give you better experience with SDKs May 09 20:52:28 well I may be a bit stuck as I'm providing a SDK that is based off OE so I lag behind OE releases a bit or should I say I pin to the latest stable. As thats 2011-03 that is what I would prefer to compare against May 09 20:53:15 stable if you want classsic oe May 09 20:53:19 if I need to build up oe-core to learn its differences I'm all for that, but I'm still not understanding what your telling me to build - oe-core is not a repo that I can find - is it a branch of a repo? May 09 20:53:43 its openembedded-core May 09 20:54:01 but I dont want to waste few seconds to type all that May 09 20:54:08 ok, which is the poky-like one that needs some work - sounds like I should stay away from that :) May 09 20:54:10 in short its called oe-core May 09 20:54:15 ok, understood May 09 20:58:37 i'm having this problem with a recipe, the source zip file extracts in the current folder and doesn't contain a folder itself May 09 20:59:02 is there any way to force the unzip command to run in a different directory then WORKDIR? May 09 20:59:23 rcf, ran into that the other day myself... not sure if this is the best approach but I gave the url a 'unpack=false' and unpacked it myself where I wnated it May 09 21:00:08 and how did you that exactly? tried it as well but failed to get the "local" SRC_URI file to use with unzip May 09 21:00:35 SRC_URI="http:///.tar.gz;unpack=false" May 09 21:01:21 the file will be downloaded (or copied if file://) to WORDIR, then in do_install() you can do a 'tar -xvf ${WORKDIR}file -C ${D}/whereyouwantit May 09 21:01:38 oh i see :) gonna give it a try May 09 21:03:36 is this a tar a binary blob May 09 21:05:19 rcf, ya, if its simply a single file it will just be unpacked into $WORKDIR, in my case it was a directory of files that I wanted to copy everything from and needed it in a subdir May 09 21:07:21 is there any way to access the downloaded zip file so I can just do something like unzip -q ${SOURCESDIR}/file.zip ? May 09 21:07:45 if you know what the filename is that gets extracted it will be that name May 09 21:10:05 khem, maybe there is a pre-built sdk somewhere I can download and use as reference for my questions? going to take forever to build new May 09 21:10:23 does angstrom have modern downloadable sdk's? May 09 21:10:38 dont know try narsissus May 09 21:12:00 angstroem suffers the say as other oe distros May 09 21:13:34 yes, seems that way May 09 21:16:38 ${DL_DIR} was the variable i needed :) May 09 21:20:36 good nir May 09 21:20:41 ups good nite May 09 21:21:45 schlaf gut May 09 21:27:48 hrm, i have a very simple regex based lexer for the bitbake format.. sadly going to have to implement INDENT/DEDENT tracking to really handle the def'd functions properly May 09 22:22:18 khem, looks like my sdk issue is that my gcc is built with invalid configuration. ie --includedir points to /opt/angstrom/angstrom-devel/arm/include instead of /opt/angstrom/angstrom-devel/arm/arm-angstrom-linux-gnueabi/usr/include/ May 09 22:22:40 many of the built-in vars are set to the wrong place like that May 09 22:24:51 not clear yet where that config came from but seeing that so much of autoconf's behavior relies on those internal defaults isn't there a better way to override built-ins such that an sdk can be easily relocated to a diff path? May 09 23:48:13 hmm. May 09 23:48:37 wonder if i could use PLY or codetalker with this custom lexer May 09 23:53:56 khem, earlier regarding libtool in sdk - are you saying that an sdk version of libtool should be provided in the SDK's path? May 10 00:27:02 tharvey: yes libtool-sdk May 10 00:41:23 https://gist.github.com/963723 - not the prettiest code around, first pass May 10 01:43:29 anybody here familiar with the mono recipe? I've built it and installed as part of my image, but I only see the 'mono' runtime executable, and not 'gmcs' which is the compiler May 10 01:56:24 FILES_mono-gmcs = "/usr/bin/gmcs2 \ May 10 01:56:25 /usr/lib/mono/2.0/gmcs.exe*" May 10 02:00:10 GNUtoo|laptop: ah, is there a separate mono-gmcs package I need to include? May 10 02:00:20 ? May 10 02:00:46 ah you're making an image May 10 02:00:48 then yes May 10 02:00:54 yes, sorry May 10 02:04:01 thanks :) does bitbake (or anything else) provide a way to list the packages that are available to be included in an image? May 10 02:26:02 liamMT, yes look into the directory where the ipk go May 10 02:26:33 (deploy/ipk) May 10 02:30:26 ah - and I had not understood that I need to re-run the python script to generate the list of files for the mono recipe **** ENDING LOGGING AT Tue May 10 02:59:58 2011