**** BEGIN LOGGING AT Mon Feb 07 02:59:56 2011 Feb 07 08:01:56 03Steffen Sledz  07org.openembedded.dev * r59b44e5f05 10openembedded.git/contrib/testing/testscript.sh: Feb 07 08:01:57 testscript.sh: update to bitbake 1.10.2 Feb 07 08:01:57 Signed-off-by: Steffen Sledz Feb 07 08:08:14 is there a way to say -b bbfile -c clean -b bbfile .. to clean and rebuild in one step ? Feb 07 08:10:55 rob_w: theoretically, there is -crebuild, but AFAIK doesn't work Feb 07 08:11:02 good morning Feb 07 08:12:11 rob_w: -c rebuild starts with compile.. so you you need -c clean to ie refresh applied patches then -c rebuild won't be enough Feb 07 08:12:26 s/you you/if you/ Feb 07 08:13:22 k thnanks Feb 07 08:57:16 hi all, do we have any commit policies about adding a bunch of BBCLASSEXTEND = "native" to recipes Feb 07 08:58:57 i want to add it e.g. to glibmm Feb 07 09:02:02 mrnice|work: usually is a good habit to post them to ML before. also http://wiki.openembedded.net/index.php/Commit_Policy Feb 07 09:03:12 mckoan: ok, I will post them before pushing thx Feb 07 09:08:46 hm.. I see no my mail about tinderclient cleanup.. Feb 07 09:10:02 ah.. it's gmail Feb 07 09:14:33 03Martin Jansa  07master * ra766a43a35 10openembedded.git/recipes/tasks/task-shr-feed.bb: Feb 07 09:14:33 task-shr-feed: add samba Feb 07 09:14:33 Signed-off-by: Martin Jansa Feb 07 10:09:16 03Koen Kooi  07org.openembedded.dev * r673040f31b 10openembedded.git/recipes/powervr-drivers/ (libgles-omap3.inc libgles-omap3/cputype): Feb 07 10:09:16 libgles-omap3: catch up with devmem2 changes Feb 07 10:09:16 Signed-off-by: Koen Kooi Feb 07 10:29:59 03Koen Kooi  07org.openembedded.dev * r2b7f6e73c7 10openembedded.git/contrib/angstrom/source-mirror.txt: Feb 07 10:29:59 angstrom source mirror: blacklist git2/ dir as well Feb 07 10:29:59 Signed-off-by: Koen Kooi Feb 07 10:36:47 can someone tell me why tslib based setup inside Xorg is not able to update newly calibrated into a running Xorg ? Feb 07 10:43:09 03Eric Bénard  07org.openembedded.dev * raa5ea12073 10openembedded.git/recipes/busybox/busybox_1.18.2.bb: (log message trimmed) Feb 07 10:43:09 busybox-1.18.2: add latest fixes Feb 07 10:43:09 ping : fix IPv6 pinging Feb 07 10:43:09 1bb52a9b73772d47a49bcaee0eb4eeb0c68b9b4c Feb 07 10:43:09 udhcpc: fix a problem with binary-encoded options Feb 07 10:43:09 fbea22056014a05a0c6619eb458d4955869b5e93 Feb 07 10:43:09 tar: update TAR_AUTODETECT Feb 07 10:48:38 rob_w: because you have to use xinput_calibrate Feb 07 10:50:29 ohnoes Feb 07 10:51:08 yet , if ts_calibrate updates its info and stuff .. and as its below Xorg . .why would this work ? Feb 07 11:13:50 rob_w: there is a msg in openembedded-devel (1 Feb) Feb 07 11:13:54 The problem seems to be that Xorg is started before ts_calibrate runs for the first time. I found a fairly simple hack that seems to fix the problem. Feb 07 11:13:56 # cp /etc/X11/Xinit.d/89xTs_Calibrate /etc/rc5.d/S89xTs_Calibrate Feb 07 11:14:18 read it all Feb 07 11:15:50 Hi all , I want to test this kernel image ,kernel-image-2.6.29-omap1 which I downloaded from the link http://www.angstrom-distribution.org/repo/?pkgname=kernel-image-2.6.29-omap1 Feb 07 11:15:52 hmm but that doesnt sound to my problem .. see , Xorg is running -> but user doesnt like the calibration ,, so he starts ts_calibrate -- but then the new coordinates or not activve till Xorg runs Feb 07 11:16:38 This is a kernel-image-2.6.29-omap1_2.6.29-r46.1_beagleboard.ipk ..... which is a opkg Image , how can I get the uImage of this to be loaded using a tftp server ? Feb 07 11:16:57 yes Feb 07 11:17:30 for instance opkg install will install this on my host machine Feb 07 11:17:50 well how to put this in the target machine Feb 07 11:18:40 maybe I can format the sdcard and make the uboot load this image from there but , how to get the uImage again Feb 07 11:18:57 becoz it will uboot will want a uImage Feb 07 11:40:42 sudo tar -xzvf kernel-image-2.6.29-omap1_2.6.29-r46.1_beagleboard.ipk -xjvf gives not in gzip and not in bzip2 format... Feb 07 11:43:48 03Koen Kooi  07org.openembedded.dev * raabea6ba22 10openembedded.git/recipes/powervr-drivers/ (libgles-omap3.inc libgles-omap3/rc.pvr): Feb 07 11:43:49 libgles-omap3: fix other devmem2 references and remove stale comment about non-TSPA releases Feb 07 11:43:49 Signed-off-by: Koen Kooi Feb 07 11:44:43 woglinde: hi, these patches from yesterday haven't helped, bootstrap-image still needs 3372 tasks done Feb 07 11:45:07 filip try bitbake -g bootstrap-image Feb 07 11:45:12 and look for avahi Feb 07 11:45:23 in pn-bla.log Feb 07 11:47:51 woglinde: avahi -> or -> avahi ;P? Feb 07 11:48:05 woglinde: "gnome-vfs" -> "avahi" Feb 07 11:48:15 woglinde: the only one this way Feb 07 11:48:40 woglinde: "gst-plugins-base" -> "gnome-vfs" Feb 07 11:48:53 woglinde: "bluez4" -> "gst-plugins-base" Feb 07 11:49:12 woglinde: and finally "task-base" -> "bluez4" [style=dashed] Feb 07 11:49:19 ah right yes Feb 07 11:49:39 you must disable bluez stuff Feb 07 11:49:42 hm, somethimes I think it would be nice for OE to provide something like USE in portage Feb 07 11:49:55 that was discussed mayntimes Feb 07 11:50:09 and always didnt find a majaority Feb 07 12:14:15 plz give me some directions to how to use the ipk extention kernel image I got from the http://www.angstrom-distribution.org/repo/?pkgname=kernel-image-2.6.29-omap1 Feb 07 12:14:24 I need a uImage to load Feb 07 12:15:58 justdoit_wg compile it yourself Feb 07 12:20:45 I see I was thinking I ll test this then compile Feb 07 12:20:58 thanks for the help ! Feb 07 12:21:03 I ll compile Feb 07 12:21:28 03Koen Kooi  07org.openembedded.dev * r9b3d83a326 10openembedded.git/contrib/angstrom/source-mirror.txt: Feb 07 12:21:29 angstrom source mirror: blacklist *.done as well Feb 07 12:21:29 Signed-off-by: Koen Kooi Feb 07 12:37:23 hi likewise Feb 07 13:19:02 woglinde: have you checked dietlibc-status lately ? Feb 07 13:19:22 ant_work no Feb 07 13:23:44 ./oebb.sh config da850_omapl138_evm http://pastebin.com/ESVDfG9q it says unknown machine ! Did things go right Feb 07 13:24:50 it has created a build and sources dir . sources has openembedded dir Feb 07 13:25:24 I am following this link http://www.angstrom-distribution.org/building-angstrom Feb 07 13:26:19 03Bernhard Guillon  07master * r82c07ebb7b 10openembedded.git/recipes/libglademm/libglademm_2.4.1.bb: Feb 07 13:26:19 libglademm_2.4.1.bb: remove legacy staging, bump PR Feb 07 13:26:19 *removing legacy staging also solve missing libglademm-2.4.pc problem Feb 07 13:26:19 Signed-off-by: Bernhard Guillon Feb 07 13:31:59 ant_work: there are needed a lot of ARM EABI related patches to make dietlibc work: https://github.com/ensc/dietlibc/commits/rebase Feb 07 13:33:02 the OE recipe does not work nice either Feb 07 13:33:45 heh, at some point it was ok (2009 ?) Feb 07 13:34:00 then when I tried I got gcc segfault Feb 07 13:35:09 ah, I see now you are the author ;) Feb 07 13:35:20 ant_work: afair, it never worked for eabi. e.g. stack was misaligned in start.o and some function do not have a return and jump into the next code placed by the linker Feb 07 13:35:34 the 2011-01-09 Feb 07 13:35:43 patches are the most important ones Feb 07 13:35:49 ok, that's why I could compile a binary but ..not working Feb 07 13:36:17 test rtarget is armv5te Feb 07 13:37:04 hi , all I think the ./oebb.sh config da850_omapl138_evm has already done the steps for me .. as I follow the link http://wiki.openembedded.org/index.php/Getting_Started Feb 07 13:37:30 Now I can further follow from Updating OpenEmbedded section Feb 07 13:37:55 Sorry if i am bothering with smthing basic Feb 07 13:39:04 ensc|w make it better Feb 07 13:39:07 the recipe Feb 07 13:39:27 ensc|w: btw we use the binary in an initramfs so abi/eabi is really a non-issue provided we compile the kernel Feb 07 13:39:57 well, disabling oabo-compat will save some kb... Feb 07 13:40:18 err.. oabi Feb 07 13:41:01 whahah mov pc, Feb 07 13:41:04 ensc? Feb 07 13:41:19 bx lr please for all armv >=6 Feb 07 13:42:04 woglinde: https://www.cvg.de/people/ensc/dietlibc-cross.bb https://www.cvg.de/people/ensc/dietlibc.inc Feb 07 13:42:24 woglinde: afais, dietlibc is armv4 too Feb 07 13:43:15 but recipe uses my git tree and I am not sure whether I want to fork dietlibc ;) Feb 07 13:43:34 ensc sure Feb 07 13:43:41 thats why we can use ifdef Feb 07 13:44:36 and we could maybee steal the neon memcpy routine from glibc Feb 07 13:47:12 and seems my ccache patch isnt that 100% stable Feb 07 13:47:22 seems to fail for ant all the time Feb 07 13:49:47 possibly after I disabled Gentoo's CCACHE Feb 07 13:49:48 hm okay only in the md5sum its used with mov pc Feb 07 13:50:37 ah no Feb 07 13:50:49 there are a lot of mov pc Feb 07 13:53:15 woglinde: about diet (and klcc) wrappers, do you think we just use them as 'crosscompilers'? Do you think we should have a (diet : k)libc-sdk ? Feb 07 13:53:46 * ant_work missed an 'or' Feb 07 13:54:29 ieeks __test_and_set uses swp Feb 07 13:57:36 ensc -> https://wiki.ubuntu.com/ARM/Thumb2PortingHowto Feb 07 13:58:13 woglinde: I know... I develop for a year on cortex m3 Feb 07 13:58:19 okay Feb 07 13:58:31 should be only monkey work for us Feb 07 13:58:59 but uclibc works fine for thumb2/m3 so there was no need (and no time) to start the dietlibc port Feb 07 13:59:20 yeah uclibc has all the ifdefs Feb 07 14:00:18 no... there is a swp (or a arm32) in the thread functions which prevents build for thumb2 Feb 07 14:00:27 ah Feb 07 14:00:36 why didnt you send a patch? Feb 07 14:01:01 because I commented out the code only (I do not need threading on -m3) Feb 07 14:01:19 *sigh* Feb 07 14:01:42 * woglinde wonders why khem wasnt stumbled over it yet Feb 07 14:01:48 clean fix might be the usage of the gcc intrinsics but I am not sure about he implications Feb 07 14:01:58 why you dont need threading? Feb 07 14:02:01 thats a bit odd Feb 07 14:03:25 because I do not run Linux on the Cortex-M3 platform (192 KiB ROM is a little bit to small ;) Feb 07 14:03:34 hm ah okay Feb 07 14:03:59 ensc|w: and you use OE to build the stuff for the -m3? Feb 07 14:04:52 ynezz: nearly... Feb 07 14:05:15 woglinde: https://www.cvg.de/people/ensc/thumb2.patch is my patch for the ulibc issue (but probably wrong) Feb 07 14:08:57 ah linuxthread.old Feb 07 14:09:00 okay Feb 07 14:09:13 that isnt used anymore since ntpl is in Feb 07 14:09:44 woglinde: yes, but built by default by the uclibc recipe Feb 07 14:09:54 ? Feb 07 14:10:09 hm I am always using the git recipe Feb 07 14:10:16 that is using nptl Feb 07 14:11:49 ynezz: https://www.cvg.de/people/ensc/0001-cortex-m3-adjusted-target-cflags.patch https://www.cvg.de/people/ensc/0002-gcc-4.5-added-thumb2-build-patch.patch are most important fixes for -m3 Feb 07 14:13:29 woglinde: ok; will try it. When I started with -m3 I had problems I had problems with the git recipe and because I need not very much from it (only base gcc function and memcpy and so) I stayed with .31 Feb 07 14:25:19 yawn Feb 07 14:25:24 Anyone looking for me yet? Feb 07 14:25:36 Or want to run away from mips64/arm uclibc failure logs? :) Feb 07 14:25:44 (qt4+uclibc+those arches) Feb 07 14:27:03 * ajb_oe is confused, whatever he does the -dev packages won't go into the base SDK tarball, just as additional ipks Feb 07 14:27:27 I don't really want to tell my users to do an opkg install for all the dev files after they untar the SDK Feb 07 14:29:22 You shouldn't Feb 07 14:29:37 What does your custom meta-toolchain recipe look like? Feb 07 14:31:59 Tartarus: http://pastebin.com/hczLc6u2 Feb 07 14:34:53 http://www.openembedded.org/index.php/Kernel_Building how do i choose a particular kernel version like i want to choose 2.6.33-rc4/ Feb 07 14:34:55 have I been confused by host/target? re-reading it implied the -dev packages go into the target task Feb 07 14:35:00 plz guide Feb 07 14:38:44 Tartarus: minimal - console-image build stopped repeatedly because of races Feb 07 14:38:54 this time perl seems not guilty Feb 07 14:40:36 I guess this is the answer PREFERRED_VERSION_linux-rp = "2.6.32" Feb 07 14:42:24 justdoit_wg: I don't know but you could look at a custome .bb file like the jlime ones which seem to do "special" versions Feb 07 14:42:28 justdoit_wg: see recipes/linux/linux-jlime-jornada7xx_2.6.32+git.bb: Feb 07 14:42:50 Tartarus: iirc curl-native and/or libxml2-pyton-native Feb 07 14:43:32 thanks ajb_oe Feb 07 14:44:35 ant_work, that's just odd Feb 07 14:45:03 ajb_oe, yes, you have it backwards Feb 07 14:45:23 you should have TOOLCHAIN_TARGET_TASK = "task-sdk-bare libxml2-dev ...." Feb 07 14:45:45 justdoit_wg: ajb_oe DEFAULT_PREFERENCE in the recipe is evaluated in case there is'nt any PREFERRED_VERSION set Feb 07 14:46:09 Tartarus: I did reproduce the falure twicw Feb 07 14:46:11 TOOLCHAIN_HOST_TASK would be for, if you have a custom runs on the host tool for allowing the user to generate special files for the radio to feed into other tools Feb 07 14:46:40 ant_work, what distro/machine and what host? and -j/BB_NUM_THREADS? Feb 07 14:46:53 Only problem I had over the weekend w/ our builder was running out of space ;( Feb 07 14:46:54 Tartarus: the second run fixes all Feb 07 14:47:04 Need to switch to bitbake gcc-cross && bitbake real-target I think Feb 07 14:47:42 (Do have rm_work, but with some of these big images like qt4e-demo-image, a whole bunch gets unpacked while doing the toolchain dance) Feb 07 14:48:13 I have 4 BB_THREADS and make -j5 Feb 07 14:49:23 ok, and MACHINE/DISTRO? Feb 07 14:49:25 Tartarus: failing image is console-image / minimal distro Feb 07 14:49:32 target is armv5te Feb 07 14:49:43 machine c7x0 Feb 07 14:49:44 so qemuarm Feb 07 14:49:45 oh Feb 07 14:49:46 k Feb 07 14:49:58 yep, I can test binaries ;) Feb 07 14:50:37 I had metadata of saturday night Feb 07 14:50:54 and have kept the log (at home) Feb 07 14:51:00 ok, what githash? Feb 07 14:51:12 will tell you later... Feb 07 14:51:19 can't ssh from here now Feb 07 14:51:25 k Feb 07 14:52:14 seems pure race issue / missing deps Feb 07 14:52:53 I'd say that having more bb threads explicits the issues Feb 07 14:53:40 Well, I have more threads and don't see that Feb 07 14:54:12 np, I have the logs Feb 07 14:54:42 sec Feb 07 14:55:05 http://imgur.com/L73MM Feb 07 14:55:09 is my weekend status Feb 07 14:55:16 -j6 and 6 bb threads Feb 07 14:55:35 some of the 'surprise' failures are disk space :( Feb 07 14:55:46 rest are things i'll either dig at now or post to the ML Feb 07 14:56:13 ant_work, I am wondering where to find DEFAULT_PREFERENCE, I want to choose the kernel 2.6.33-rc4 and for da850_omapl138_defconfig I cnt see this in the local.conf in the build dir ? Feb 07 15:00:00 if it's not in the machine.conf and related includes you'll end-up building the highest version in tree with DEFAULT_PREFERENCE=1 (or the override for you machne) Feb 07 15:01:29 in your case linux-davinci it seems Feb 07 15:03:28 and being that linux-davinci_2.6.31+2.6.32-rc2-psp03.01.00.37.bb doesn't have a negative DEFAULT_PREFERENCE, voilà, it's chosen Feb 07 15:04:47 ah, no, wait Feb 07 15:05:12 linux-davinci_git.bb has explicit DEFAULT_PREFERENCE = "1" Feb 07 15:05:51 so this one should be chosen Feb 07 15:05:55 k Feb 07 15:06:00 Let me look Feb 07 15:06:41 actually the ones with negative preference are 'masked' Feb 07 15:07:12 (unless you force that in your conf using PREFERRED_VERSION) Feb 07 15:07:45 K Feb 07 15:08:06 afaik most devs tend to let the kernel release float, just adapting DEFAULT_PREFERENCE instead of touching conf files Feb 07 15:08:30 so I do "bitbake linux-davinci-hawkboard /virutal/kernel " something like that Feb 07 15:09:10 yes, one or the other Feb 07 15:09:58 I dont find a .bb file for hawkboard Feb 07 15:10:33 wasn't da850_omapl138 ? Feb 07 15:10:48 in the recipes/linux/linux-davinci/hawkboard Feb 07 15:11:01 No actually I wanted hawkboard Feb 07 15:11:57 well.. root/conf/machine/hawkboard.conf Feb 07 15:12:22 ok Feb 07 15:12:52 MACHINE = "hawkboard" in your local.conf Feb 07 15:13:19 (no idea about status of support, sorry) Feb 07 15:14:02 No problem thank you for the help , Really Feb 07 15:14:23 now I must check how to get the bitbake to compile for hawkboard Feb 07 15:14:37 is anybody using upstart with oe? Feb 07 15:14:41 do you know any good tutorial that I may be missing out Feb 07 15:14:52 [16:12] MACHINE = "hawkboard" in your local.conf Feb 07 15:15:08 then follow the docs Feb 07 15:15:28 http://www.openembedded.org/index.php/Getting_started Feb 07 15:16:05 MACHINE=hawkboard bitbake u-boot arago-gst-image Feb 07 15:16:11 I read through that Feb 07 15:16:28 that's all :) Feb 07 15:17:04 thanks Feb 07 15:17:10 yw Feb 07 15:18:30 I guess I ll have to use Building a single package, option"bitbake -b /stuff/openembedded/recipes/blah/blah.bb" to compile a kernel for hawkboard Feb 07 15:18:50 but no .bb file in that davinci dir for hawkboard Feb 07 15:22:13 what is built doing 'bitbake virtual/kernel' ? Feb 07 15:23:50 hey , still doing it Feb 07 15:25:25 ant_work: hey Feb 07 15:25:29 path: root/recipes/linux/linux-davinci/hawkboard/defconfig Feb 07 15:25:37 pwgen_8: hey there Feb 07 15:26:47 Jay7 I'm sorry that neither of us got back to you. What can we do for you? Feb 07 15:27:26 ant_work. i try to build with oe for an wrt54 router. but bitbake throws a lot of ":EOL while scanning string literal (TARGET_LDFLAGS," errors, any idea ? Feb 07 15:28:01 bitbake master? Feb 07 15:28:28 I use bitbake-1.10.2 Feb 07 15:30:49 pwgen_8: I'd suggest to use bitbake from git. I've read about strange issues with older versions / buildhosts Feb 07 15:31:21 unless you have some broken file / disk full ... Feb 07 15:32:31 pwgen_8: goole has an hit : http://wiki.openmoko.org/wiki/OpenEmbedded Feb 07 15:32:54 doh Feb 07 15:34:39 hi stefan Feb 07 15:34:58 ant_work: 1.10.2 seems the last official release .. Feb 07 15:36:05 btw it would seem this is about the use of SRCPV Feb 07 15:36:22 * ant_work doesn't trust wiki's Feb 07 15:36:30 hi woglinde Feb 07 15:39:08 ant_work: bitbake 1.11 seems also broken .. i think the wrt54 isnt working .. Feb 07 15:39:42 pwgen_8, sounds like you've got a mis-match " or ' in your local.conf Feb 07 15:39:49 or some other file you've edited Feb 07 15:40:02 pwgen_8: it could be that old thread...[oe] [PATCH] bitbake.conf: SRCPV always upgradeable paths for git recipes Feb 07 15:42:25 Tartarus: or a tzpe somewhere in the oe stuff wehne i select the wrt54 target. because same setting with DISTRO and TARGET for angstrom/x86 is at least parsing and starting Feb 07 15:42:56 http://pastebin.com/YZwvvaEC Feb 07 15:43:23 Little dissapointment Feb 07 15:44:24 justdoit_wg: I don't want to be rude, but... Feb 07 15:44:26 angstrom-setup-scripts$ ./oebb.sh config hawkboard Feb 07 15:44:29 ^^^^^^^ Feb 07 15:44:38 this is not the standard Feb 07 15:45:14 yes but I followed the instructions Feb 07 15:45:23 thats how this was made Feb 07 15:45:44 http://www.angstrom-distribution.org/building-angstrom Feb 07 15:45:49 in OE you really just need a 3 lines local.conf and a short file to source for the env Feb 07 15:47:27 K , http://pastebin.com/ah9pqyUm Feb 07 15:47:33 is the local.conf Feb 07 15:47:52 K I could edit this file to keep the relevent Feb 07 15:48:09 my env is also right Feb 07 15:48:38 did I go wrong anywhere ? Feb 07 15:49:48 The program 'bitbake' is currently not installed Feb 07 15:49:51 Tartarus: did you have a chance to talk to khem about release manager hat for 2011-03? Feb 07 15:50:29 Tartarus: personally I would think rotating it makes sense. Given the knowledge gets transferred from old to new Feb 07 15:50:50 That way we would have more people being capable of doing such a thing for OE :) Feb 07 15:51:18 justdoit_wg: try to avoid other tmpdirs . it should have the OE path ending with /tmp. otherwise some packages will fail Feb 07 15:51:50 I but I see the bitbake in my sources Feb 07 15:52:17 ok pwgen I ll check that Feb 07 15:52:20 eh? you don't need the tmpdir to end in /tmp, i do builds all the time without that being the case Feb 07 15:52:51 TMPDIR = "/home/sukrut/angstrom-setup-scripts/build/tmp-angstrom_2008_1" Feb 07 15:53:03 is my tmp here in the local.conf Feb 07 15:53:04 * ant_work is amazed by the overcomplicated oebb.sh Feb 07 15:53:13 oh ho Feb 07 15:53:22 this is direclty via the link Feb 07 15:53:45 kergoth: wrong. try to build sdk-native-image and zoutr build will fail. because ther is a bad patch that needs to have this dir named /tmp . Feb 07 15:54:05 shall I scrap it all and begin new Feb 07 15:54:30 pwgen : so I just name it as TMPDIR = "/home/sukrut/angstrom-setup-scripts/build/tmp right Feb 07 15:54:44 right ? Feb 07 15:55:19 justdoit_wg: yup Feb 07 15:55:37 ok thanks Feb 07 15:56:55 pwgen_8: that's not an aspect of oe, that's one particular broken recipe. Feb 07 15:57:06 don't start telling people they have to do shit they don't, fix the problem instead Feb 07 15:58:03 ant_work : shall I install bitbake with sudo apt-get install bitbake , Now I see the http://pastebin.com/SZiaANPd Feb 07 15:58:58 as I already see the bitbake source is located in the sources dir under the angstrom-setup-scripts Feb 07 15:59:47 stefan_schmidt: haven't seen khem yet, heh Feb 07 16:00:08 But yeah, march is coming up quick Feb 07 16:00:17 Tartarus: indeed Feb 07 16:00:22 justdoit_wg: it looks like bitbake is not in your path Feb 07 16:00:43 Tartarus: Let me write a mail about it to get some reactions Feb 07 16:00:48 k Feb 07 16:02:40 http://pastebin.com/YZwvvaEC Feb 07 16:03:00 ant_work my PATH is set correctly Feb 07 16:03:28 no Feb 07 16:03:30 The program 'bitbake' is currently not installed. Feb 07 16:05:33 oh ! Feb 07 16:05:55 sukrut@sukrut:~/angstrom-setup-scripts$ echo $PATH Feb 07 16:05:56 # Feb 07 16:05:58 /angstrom-setup-scripts/sources/bitbake/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games Feb 07 16:06:00 ^^ no bitbake there Feb 07 16:06:04 ? Feb 07 16:06:22 I mean in ..sources/bitbake/bin ? Feb 07 16:07:09 * ant_work refuses to read that setup script...it's just much easier to clone bitbake locally... Feb 07 16:07:19 forgive my ignorance but I do see bitbake in sources/bitbake/bin Feb 07 16:07:28 I see Feb 07 16:09:05 "sh oebb.sh bitbake nano" you know it starts bitbake with this command Feb 07 16:09:18 Tartarus: If I want to keep the split tasks as seperate task .bb files should the dev dependancies be in PACKAGE or RDEPENDS? What does TOOLCHAIN_TARGET_TASK get explanded into? Feb 07 16:10:19 I guess now I ll have to use something like "sh oebb.sh bitbake virtual/kernel " to get it to work acc to this link http://www.angstrom-distribution.org/building-angstrom Feb 07 16:10:48 let me try Feb 07 16:11:21 let see Feb 07 16:11:35 thanks for the clue ant_work Feb 07 16:11:44 yw Feb 07 16:11:47 bbl Feb 07 16:33:41 Hi, I have added a new recipe for a custom package called "visstrim_m10-utils Feb 07 16:33:51 but when I try to build it I get the following Feb 07 16:33:57 ERROR: '['/home/javier/GIT/stuff/openembedded/recipes/images/micro-image-visstrim.bb']' RDEPENDS/RRECOMMENDS or otherwise requires the runtime entity 'visstrim_m10-utils' but it wasn't found in any PACKAGE or RPROVIDES variables Feb 07 16:34:27 do I have something left to do apart from creating a .bb file? Feb 07 16:35:09 the recipe is located in "recipes/visstrim_m10-utils/visstrim_m10-utils_0.1.bb" Feb 07 16:49:05 Helio_2000: the _m10 might be confusing bitbake Feb 07 16:49:44 bitbake looks for ${PN}_${PV}.bb Feb 07 16:50:00 package name and package version Feb 07 16:50:47 no, it doesn't. bitbake doesn't care what the filename is, it just happens that the *default* PN, PV, and PR *can* come from the filename Feb 07 16:50:54 they can also be set just as well from the recipe itself Feb 07 16:51:02 you can name it _.bb just fine ;) Feb 07 16:51:10 I've just added: RPROVIDES_${PN} ="visstrim_m10-utils" Feb 07 16:51:14 and now it works Feb 07 16:51:39 technically, you can name a recipe .bb -- i've done it before when using srctree Feb 07 17:29:25 rick-work: oestats-client problem is closed for some time I hope Feb 07 17:29:35 so, no interaction is needed from your side Feb 07 17:50:47 Jay7 cool. Can you give me a short version of what you did so that I can log it ??? Feb 07 17:51:26 rick-work: client have assert, that checks for empty fields before sending data Feb 07 17:51:37 it start to fail somehow Feb 07 17:52:27 so I've removed it w/o any visible problems Feb 07 17:53:21 Anybody using Hudson to autobuild OE? Feb 07 17:53:46 03Chris Larson  07master * r78b6d4cb26 10bitbake.git/bin/bitbake-layers: (log message trimmed) Feb 07 17:53:46 Add initial bitbake-layers script Feb 07 17:53:46 This script has subcommands which operate against your bitbake layers, either Feb 07 17:53:46 displaying useful information, or acting against them. Currently, it only Feb 07 17:53:46 provides a show_appends command, which shows you what bbappends are in effect, Feb 07 17:53:46 and warns you if you have appends which are not being utilized. Feb 07 17:53:48 Currently, a bug exists when using this due to the DataContext stuff, but I'm Feb 07 17:55:31 I've set up Hudson to autobuild "testing next" but OE seems sometimes to build forever when started from Hudson but build fine otherwise. A OE build issue or Hudson issu? Feb 07 17:59:09 hard to say Feb 07 17:59:18 how are the slaves configure? Feb 07 17:59:22 whats local.conf look like? Feb 07 17:59:34 did you take hudson overhead (memory) into account on the slaves? Feb 07 18:00:05 03Chris Larson  07master * r2cc6b6951b 10bitbake.git/bin/bitbake-layers: Feb 07 18:00:05 bitbake-layers: fix bug with env vars leaking in Feb 07 18:00:05 I forgot that bin/bitbake is what does the environment filtering based upon Feb 07 18:00:05 BB_ENV_WHITELIST, etc. Feb 07 18:00:05 Signed-off-by: Chris Larson Feb 07 18:01:46 AndersT, oh and do you have any output ie console or logs w/ hudson? Feb 07 18:02:09 http://imgur.com/L73MM is what I got done w/ hudson+OE over the weekend, fwiw :) Feb 07 18:03:45 Nothing special in the output from Hudson other than the usual bitbake output and that it has reached a certain build step. Feb 07 18:04:27 and then just stops? Feb 07 18:04:38 How many bitbake threads do you have going? Feb 07 18:04:45 And did you take into account the hudson memory overhead? Feb 07 18:04:54 Right now it has reached "NOTE: Running task 6086 of 6215 (ID: 109, /home/hudson/jobs/OE-testing_next/workspace/openembedded/recipes/eglibc/eglibc_2.12.bb, do_package_stage) " since this morning Feb 07 18:05:08 And you aren't out of disk space? Feb 07 18:05:19 is the VM heavy into swap? Feb 07 18:05:30 (s/VM/slave/) Feb 07 18:05:34 250 GB disk left Feb 07 18:06:26 "top" reports a system load of 1.8 and only used 5 GB of th 6GB RAM Feb 07 18:06:54 One python process runs at 100% on one of the cpus Feb 07 18:07:18 looks like that got wedged somehow Feb 07 18:07:30 is this the first time you've seen this? Feb 07 18:07:34 or happens repeatidly? Feb 07 18:08:02 I have configured bitbake to use 5 threads and "-j5" also. Which works just fine if I start the builds outside of Hudson. Feb 07 18:08:33 Yeah, but hudson will suck up another 1gb or so of memory, at leat thats what we've seen here Feb 07 18:08:37 are you into swap at all? Feb 07 18:09:14 I set up Hudson last week and configured it to build 3 machines and 3images= 9 combinations. Almost every time it has ended up like this. Feb 07 18:09:22 ok Feb 07 18:09:29 So, is the machine into swap at all? Feb 07 18:09:37 and what does your local.conf look like? Feb 07 18:09:40 ~pastebin Feb 07 18:09:41 [~pastebin] A "pastebin" is a web-based service where you should paste anything over 3 lines so you don't flood the channel. Here are links to a few : http://www.pastebin.com , http://pastebin.ca , http://channels.debian.net/paste , http://paste.lisp.org , http://bin.cakephp.org/ , http://asterisk.pastey.net/ , or install pastebinit with yum or aptitude. Feb 07 18:09:43 400MB of used swap Feb 07 18:10:07 jama ping Feb 07 18:10:37 AndersT, can you see what process is really wedged on the machine? That might help figure out whats going on Feb 07 18:11:04 I'll paste my config to pasebin in a few secs Feb 07 18:11:07 k Feb 07 18:13:22 The local.conf is in the pastebin. Can you see it? Feb 07 18:13:55 what's the url? Feb 07 18:14:14 http://pastebin.com/x0xPeW3F Feb 07 18:15:06 "python /home/hudson/jobs/OE-testing_next/workspace/bitbake-1.10.2/bin/bitbake x11-image" is the process that is using 99% on one CPU. Feb 07 18:15:22 ok Feb 07 18:15:36 but what else is still alive? Feb 07 18:15:39 what's the easiest way to disable all the mirrors and premirrors and force everything to be downloaded from original SRC_URIs? Feb 07 18:16:03 Tartarus, from the build you mean? Feb 07 18:16:06 denix mom Feb 07 18:16:28 woglinde_: what's that? Feb 07 18:16:30 AndersT, well, pastebin everthing from ps auxwwww | grep hudson Feb 07 18:16:39 CVS_TARBALL_STASH = "" Feb 07 18:16:43 mom == moment Feb 07 18:16:54 CVS_TARBALL_STASH = "" for scm stuff Feb 07 18:17:00 should proably renamed Feb 07 18:17:03 to SCM Feb 07 18:17:08 kergoth? Feb 07 18:18:16 Tartarus, http://pastebin.com/7uffCfa6 Feb 07 18:18:39 CVS_TARBALL_STASH is still supported, but is deprecated, the old var, iirc Feb 07 18:19:03 ah whats the new? Feb 07 18:20:21 re stefan Feb 07 18:21:48 ah, right, the tarball stash vars aren't used at all anymore, its all mirrors/premirrors Feb 07 18:24:17 AndersT, that's all of it? Feb 07 18:24:18 hm Feb 07 18:24:57 Yes. I've set up Hudson to build the matrix sequencially. Feb 07 18:25:12 kill 29671 and see if it unsticks Feb 07 18:26:10 also, i assume you have no networking issues on the box, since it all works normally Feb 07 18:26:33 Still stuck. Feb 07 18:26:40 Networking works fine. Feb 07 18:27:15 Yesterday when the same happened I started a OE build of the samt things manually in a shell and that finished fin. Feb 07 18:27:58 all I can suggest is trying a thread less Feb 07 18:28:00 4 not 5 Feb 07 18:28:06 see if things just start getting happier Feb 07 18:28:09 kergoth: so then, what's the easiest way to disable all mirrors/premirrors? Feb 07 18:28:27 MIRRORS = "", PREMIRRORS = "" Feb 07 18:28:27 I remember there was some discussion on the list, but can't find it now... Feb 07 18:28:28 :P Feb 07 18:28:58 :) Feb 07 18:29:05 I'll try that. And maybe also restart Hudson. I do have stopped some builds, maybe there are som inconsitences in Hudson after that. Feb 07 18:29:53 Tartarus, how many threads and parallel makes dou you run on and on which HW? Feb 07 18:36:19 I do -j8 and 6 BB threads, but it's some real beefy HW under the hood (I've got virtual machines and wrote this to help manage stuff: http://wiki.jenkins-ci.org/display/JENKINS/Lab+Manager+Plugin Feb 07 18:40:28 kergoth: will you have some time today to prototype oestats2-client? :) Feb 07 18:40:47 * Jay7 is kind of reminder :) Feb 07 18:45:08 kergoth: MIRRORS is set in a class, can it be modified in local.conf? Feb 07 18:46:29 denix: why wouldn't you be able to? of course, the classes are loaded after the local.conf, so you'll need to use an override Feb 07 18:46:33 MIRRORS_local = "" Feb 07 18:48:27 thanks! I guess my multitasking abilities are broken today... :) Feb 07 19:05:25 Tartarus: here the log of curl-native failing on first run http://www.scribd.com/doc/48366410/aa Feb 07 19:06:03 made it to PHL Feb 07 19:06:31 ant__ githash pls :) Feb 07 19:06:35 Crofton, yay :) Feb 07 19:06:38 configure: error: Unable to link function send Feb 07 19:09:00 Tartarus: hm... I pulled afterwards Feb 07 19:09:39 I see METADATA_REVISION = "93686d8" on 5th feb Feb 07 19:10:47 and there I had curious issue aka ERROR: Logfile of failure stored in: /dev/null Feb 07 19:11:00 hmmm, I wish all the iphone users would hurry up and ove to Verizon Feb 07 19:11:01 my 3g tethering is a bit sluggish atm Feb 07 19:11:06 on feb 6th was better, just curl-native ;) Feb 07 19:11:53 ant__, you need be21179c5321bd0afb9221f020ac12ad75c86a3b Feb 07 19:18:05 hmm Feb 07 19:39:19 kdialog --title "OE build" --passivepopup "Build complete!" 10 Feb 07 19:39:43 way to notify yourself via pretty kde plasma popup message :) Feb 07 19:40:25 * kergoth chuckles Feb 07 19:42:21 Tartarus: retrying now with fresh pull 82c07ebb7bc876bd2898c4bfbd769809a33f5727 Feb 07 19:45:22 gm all Feb 07 19:45:30 someone nabbed my nick Feb 07 19:45:52 * khem ghosts someone Feb 07 19:50:18 hi khem Feb 07 19:55:06 re effem Feb 07 19:59:07 hi woglinde_all Feb 07 19:59:22 that is, hi woglinde, all Feb 07 19:59:32 had some network issues still not fully resolved Feb 07 19:59:43 so might bump in and out a few times Feb 07 20:00:18 Tartarus: NOTE: package curl-native-7.21.2-r4.0: task do_compile: Failed Feb 07 20:00:26 brb Feb 07 20:04:36 hello woglinde_ Feb 07 20:05:23 anyone using make-3.82 or F14 and seeing failure compiling linux-libc-headers for ppc ? Feb 07 20:05:31 2.6.31 that is Feb 07 20:05:49 one way would be to use later version of kernel headers Feb 07 20:06:08 but I think its better to fix 2.6.31 if possible Feb 07 20:06:27 ppc603e-oe-linux-uclibc/linux-libc-headers-2.6.31-r5/linux-2.6.31/arch/powerpc/Makefile:169: *** mixed implicit and normal rules. Stop. Feb 07 20:06:32 hm khem can you enable usb wacom tablet support on all qemu kernel recipes? Feb 07 20:06:48 woglinde_: it should be enabled I think Feb 07 20:07:03 hm no Feb 07 20:07:10 hrmm ok Feb 07 20:07:19 * khem notes down Feb 07 20:08:37 no Feb 07 20:08:56 checked it Feb 07 20:16:13 khem: http://patchwork.ozlabs.org/patch/60728/ Feb 07 20:16:52 khem, and note a follow-up patch to fix a uImaged -> uImage typo Feb 07 20:20:38 03Chris Larson  07master * rb2f1e746e9 10openembedded.git/recipes/ (eglibc/eglibc.inc gcc/gcc-common.inc glibc/glibc.inc): Feb 07 20:20:38 gcc, {e,}glibc: work around pstage install order issue Feb 07 20:20:38 Note that this is dealt with differently in the sstate world, so this should Feb 07 20:20:38 *not* go into the meta-oe/oe-core universe. Feb 07 20:20:38 Signed-off-by: Chris Larson Feb 07 20:22:38 Tartarus, khem: yocto have runtime testing framework already Feb 07 20:23:02 (WRT ML) Feb 07 20:23:09 Jay7: yes I know Feb 07 20:23:35 Jay7: I have a rudimentary script here which builds all qemu images and boots them Feb 07 20:23:42 but its not yet fully automatic Feb 07 20:24:00 imho there is no need to reinvent it for this OE release because we will got it after Feb 07 20:24:02 Tartarus: we should grab those parches Feb 07 20:24:25 Tartarus: would you be interested to port them back Feb 07 20:24:31 or gimme the second link too Feb 07 20:24:52 khem: I have some parts developed for my testbuilder but very rudimentary too Feb 07 20:25:14 dont have link for the 2nd, jusy saw a mention Feb 07 20:25:40 Tartarus: now it fails on | url.c:73:2: error: #error "We can't compile without socket() support!" Feb 07 20:26:26 Tartarus: OK I will make that note in patch lemme try it out Feb 07 20:26:41 do anybody have something against filip's patches in ML/patchwork? Feb 07 20:27:20 do we have some 'timeout' for this? Feb 07 20:28:22 hi Jay7 Feb 07 20:28:27 ant__: hey Feb 07 20:28:59 ant__, clean TMPDIR? Feb 07 20:29:10 yes, I'm posting log Feb 07 20:29:28 * filip bothering Feb 07 20:29:32 http://www.scribd.com/doc/48372004/bb Feb 07 20:30:06 previously clean build failed always on curl_native, but on doi_configure Feb 07 20:30:07 http://www.scribd.com/doc/48366410/aa Feb 07 20:30:35 git bisect time? Feb 07 20:31:17 mah, on second run it usually builds Feb 07 20:31:18 you've got a good candidate for git bisect run Feb 07 20:31:21 Jay7: no protests mean yes, right ;)? Feb 07 20:32:35 Tartarus: in both cases cmake_native and libxml2_native were still building Feb 07 20:32:54 try doing bitbake -c compile curl-native Feb 07 20:32:57 and see if it fails Feb 07 20:33:03 and if so, pop into git bsiect run Feb 07 20:34:30 see, if I bitbake an image, it always works. I'll try now building a single package Feb 07 20:34:51 i mean from clean Feb 07 20:34:54 to see where this came in from Feb 07 20:35:41 checking for send... no configure: error: Unable to link function send Feb 07 20:35:47 this is strange Feb 07 20:37:27 hm sdl needs with-libtool-sysroot too Feb 07 20:39:37 Tartarus: just bitbake -c clean curl-native and bitbake curl-native is ok, even without finishing the 2 a.m. Feb 07 20:39:46 hm ah no Feb 07 20:39:53 alright this works well Feb 07 20:39:56 its need libtool2.4 support at all Feb 07 20:39:58 I'll retry from scratch now Feb 07 20:40:20 03Khem Raj  07master * r9f1368c60d 10openembedded.git/recipes/linux-libc-headers/ (2 files in 2 dirs): Feb 07 20:40:20 linux-libc-headers_2.6.31.bb: Fix build with make 3.82 for powerpc Feb 07 20:40:20 Signed-off-by: Khem Raj Feb 07 20:40:46 03Chris Larson  07master * rbef07c6dc7 10bitbake.git/ (bin/bitbake lib/bb/__init__.py): Feb 07 20:40:46 Bump version from 1.9.0 to 1.10.0 Feb 07 20:40:46 Signed-off-by: Chris Larson Feb 07 20:40:54 03Richard Purdie  07master * red50160a15 10bitbake.git/lib/bb/runqueue.py: Feb 07 20:40:54 Gross hack to fix packaged-staging with BBCLASSEXTEND packages Feb 07 20:40:54 (From Poky rev: 589df80ca1b825afd138531bb796aa8617ed0991) Feb 07 20:40:54 Signed-off-by: Richard Purdie Feb 07 20:40:57 03Chris Larson  07master * rf432c71b34 10bitbake.git/lib/bb/runqueue.py: Feb 07 20:40:57 Fix IndexError encountered when a 'depends' flag references a nonexistant task Feb 07 20:40:57 Signed-off-by: Chris Larson Feb 07 20:40:59 03Jeff Dike  07master * r47482fb9f1 10bitbake.git/bin/bitbake: (log message trimmed) Feb 07 20:40:59 create cooker object after cleaning the environment Feb 07 20:40:59 Previously, the cooker object was created before the environment was Feb 07 20:40:59 cleaned, saving everything that was in the environment and dumping Feb 07 20:40:59 into the run scripts. Feb 07 20:40:59 The patch ensures that the cooker gets a cleaned environment when it's Feb 07 20:41:00 created. Feb 07 20:41:02 03Chris Larson  07master * rd6951f72ac 10bitbake.git/lib/bb/parse/__init__.py: Feb 07 20:41:02 Make the file not found error actually useable. Feb 07 20:41:02 Signed-off-by: Chris Larson Feb 07 20:41:05 03Eric BENARD  07master * r0922a86574 10bitbake.git/lib/bb/fetch/hg.py: (log message trimmed) Feb 07 20:41:05 bitbake: lib/bb/fetch/hg: fix fetching from a mercurial repository Feb 07 20:41:05 * without this fix, we get : Feb 07 20:41:05 updating working directory Feb 07 20:41:17 cooker: fix UnboundLocalError Feb 07 20:41:18 Signed-off-by: Chris Larson Feb 07 20:41:18 03Richard Purdie  07master * r479716be4a 10bitbake.git/lib/bb/server/none.py: Feb 07 20:41:19 bitbake/server: Raise a shutdown event if we're seeing exceptions Feb 07 20:41:19 Signed-off-by: Richard Purdie Feb 07 20:41:20 03Chris Larson  07master * rf5c086fd1a 10bitbake.git/lib/bb/runqueue.py: Feb 07 20:41:20 Fix exit code display for task failure Feb 07 20:41:21 Per the python documentation, os.waitpid returns the exitcode shifted up by 8 Feb 07 20:41:21 bits, and we weren't compensating, resulting in a display of 'failed with 256' Feb 07 20:41:22 when a worker process exits with a code of 1. Feb 07 20:41:22 Signed-off-by: Chris Larson Feb 07 20:41:23 03Chris Larson  07master * r75875baa5b 10bitbake.git/lib/bb/ (cooker.py parse/parse_py/ConfHandler.py): Feb 07 20:41:23 Move the unset BBPATH sanity check after the bblayers processing Feb 07 20:41:24 This way we can fully utilize bblayers, you can do everything in bblayers.conf Feb 07 20:41:25 and avoid setting any environment variables at all. Feb 07 20:41:25 Signed-off-by: Chris Larson Feb 07 20:41:26 03Richard Purdie  07master * r0964af191f 10bitbake.git/lib/bb/cooker.py: Feb 07 20:41:26 Fix handling of empty variables in local config files Feb 07 20:41:27 (From Poky rev: dbd9308) Feb 07 20:41:27 Signed-off-by: Richard Purdie Feb 07 20:41:34 kergoth: what happened to the autoconf cleanup you were doing ? Feb 07 20:41:45 03Chris Larson  07master * r0adcb7c375 10bitbake.git/ChangeLog: Feb 07 20:41:45 Add 1.10.0 entry to ChangeLog Feb 07 20:41:45 Signed-off-by: Chris Larson Feb 07 20:41:52 03Chris Larson  07master * rfd40c85cb2 10bitbake.git/lib/bb/cooker.py: Feb 07 20:41:53 Search up the directory tree for bblayers.conf Feb 07 20:41:53 Signed-off-by: Chris Larson Feb 07 20:41:56 khem: it's on a branch, need to rebase, there were some autotools.bbclass changes that result in a bunch of conflicts Feb 07 20:42:04 03Chris Larson  07master * rd237fa4e03 10bitbake.git/lib/bb/msg.py: Feb 07 20:42:05 Switch the debug_level dictionary to a defaultdict Feb 07 20:42:05 This already exists in master, but in a different form. Feb 07 20:42:05 Signed-off-by: Chris Larson Feb 07 20:42:49 kergoth: you do so much stuff and leave it getting thru the last yard :) Feb 07 20:43:25 khem: yeah, i've been working on that recently, but its an ongoing effort. with oe, things always stall after sending to the list, since most people don't want to touch the core classes with a ten foot pole Feb 07 20:43:46 (aside: we need to get a smarter cia notification script for oe/bitbake, this one sucks) Feb 07 20:44:29 none of the above was new, it was just a merge of 1.10 to master so that git logs make more sense and git describe works properly Feb 07 20:44:32 heh Feb 07 20:44:50 hmm Feb 07 20:44:57 * ant__ pulled and then inspected with git log to be sure... Feb 07 20:45:03 heh Feb 07 20:45:54 kergoth: do u know if python puts length limit on stdin Feb 07 20:46:13 its just a file descriptor Feb 07 20:46:15 i can't imagine it would Feb 07 20:46:16 If I had longer than 1024 chars it just truncated the paste Feb 07 20:46:36 it is *buffered*, but that shouldn't result in any loss of data Feb 07 20:46:41 this script was using readline() from stdin Feb 07 20:47:09 oh you mean I wont see it pasted but script should have got complete data ? Feb 07 20:47:27 not sure what you're referred to re: pasted, etc Feb 07 20:47:34 * kergoth lacks context here Feb 07 20:47:51 I had the content in clipboard Feb 07 20:48:03 and ran script in a shell Feb 07 20:48:08 and it waits for stdin Feb 07 20:48:20 so then I paste the clipboard into this shell Feb 07 20:48:29 and I see it did not paste all Feb 07 20:48:34 but only first 1024 chars Feb 07 20:48:48 the paste works ok if I just paste it in the shell Feb 07 20:48:56 i'd read a lot more data than 1024 chars from stdin before :) Feb 07 20:48:58 so its not a shell problem or my clipboard Feb 07 20:49:04 so its likely a bug in your script Feb 07 20:49:10 hrmm Feb 07 20:49:13 khem: would you care to look @ http://patches.openembedded.org/patch/465/ and http://patches.openembedded.org/patch/475/ to say whether these look ok for you? Feb 07 20:49:15 test case: import sys Feb 07 20:49:15 for l in sys.stdin.readline(): print(l) Feb 07 20:49:24 dd if=/dev/zero bs=1024 count=1024 | python test.py Feb 07 20:50:16 filip: looks ok Feb 07 20:50:45 khem: so, do we have your Ack? :) Feb 07 20:50:45 Tartarus: good news NOTE: package curl-native-7.21.2-r4.0: task do_compile: Succeeded Feb 07 20:50:45 from scratch Feb 07 20:51:17 Jay7: Yes but I need to run it thru to be sure Feb 07 20:51:30 Jay7: he said it! Feb 07 20:51:34 khem: thanks Feb 07 20:51:37 ok, can you please check all filip's patches then? Feb 07 20:52:05 all 6 from pw Feb 07 20:52:47 Jay7: http://patches.openembedded.org/patch/471/ needs a PR bump. You could apply it (dpkg doesn't work atm anyway) or I can resend it later with a bumped PR Feb 07 20:53:05 Jay7: I will, and please supercede any patches that should not be looked at Feb 07 20:53:42 filip: you should send all the latest patches you have and receieved feedback and corrected Feb 07 20:54:00 filip: resend and supersede old Feb 07 20:54:21 khem: what about my tinderclient removal patch? Feb 07 20:54:39 i really hate conflicts that result surrounding moving code from one file to another Feb 07 20:54:49 royal pain, since the conflict markers are only in one o the files Feb 07 20:55:05 wait wait, afaik the admindir is the only one still uncorrected Feb 07 20:55:09 am I missing something? Feb 07 20:56:21 mckoan|away: have a look please at http://patches.openembedded.org/patch/477/ Feb 07 20:56:54 Jay7: that you have my ack Feb 07 20:57:03 and I would prefer having Ack from Koen because of angstrom.inc Feb 07 20:57:16 khem: ok Feb 07 20:57:29 Jay7: yes definitely distro maintainers should ack them too Feb 07 20:58:23 my patch isn't time-critical like filip's :) Feb 07 21:05:04 k Feb 07 21:05:12 * khem out for lunch Feb 07 21:25:46 Jay7: how is mine time critical :P? Feb 07 21:27:56 hm was DEPENDS_avr32 += correct? Feb 07 21:34:21 Tartarus: about curl-native from scratch Feb 07 21:34:46 checking for LDAP libraries... cannot find LDAP libraries Feb 07 21:35:20 while on image-build I see checking for LDAP libraries... -lldap Feb 07 21:35:30 some contamination Feb 07 21:35:59 still can't see how < checking for send... no can happen Feb 07 21:36:18 when from scratch I get > checking for send... yes Feb 07 21:36:18 > checking types of args and return type for send... int,const void *,size_t,int,int Feb 07 21:36:50 * ant__ is rediffing confs Feb 07 21:43:32 hm.. conftest.c:176: error: conflicting types for 'send' Feb 07 21:43:32 /usr/include/sys/socket.h:141: note: previous declaration of 'send' was here Feb 07 21:44:42 recipes/tasks/task-base.bb:34 - why isn't task-base-bluetooth conditional as for example task-base-screen ? Feb 07 21:45:02 it raises task amount from 1600 to around 3400 when building bootstrap image Feb 07 21:45:20 on a machine that has no builtin bluetooth ;p Feb 07 21:45:35 filip: no, you're not reading the task correctly Feb 07 21:45:41 it always *creates* that package Feb 07 21:45:47 but it doesn't always *install* it Feb 07 21:45:49 see line 109 Feb 07 21:46:13 kergoth: I know Feb 07 21:46:22 so how exactly is creating the task a problem? Feb 07 21:46:27 kergoth: I just don't want to wait for useless stuff to be compiled Feb 07 21:46:29 creating a package doesn't increase build time Feb 07 21:46:33 kergoth: it slows down my development Feb 07 21:46:38 no, that isn't hte root cuase Feb 07 21:46:39 aaah Feb 07 21:46:42 once again, you're misunderstanding the task Feb 07 21:46:59 afaik, anyway, i could be wrong Feb 07 21:47:05 kergoth: well I remember waiting on do_compile of gtk+ for example Feb 07 21:47:14 i'd strongly suggest using bitbake -g to see what's pulling things in Feb 07 21:47:27 kergoth: and now I don't have it at all in pn-depends.dot Feb 07 21:47:27 i'm not seeing what gtk has to do with this Feb 07 21:47:30 kergoth we already did Feb 07 21:47:34 sure Feb 07 21:47:39 bluez-audio Feb 07 21:47:39 kergoth: bluez needs gst-plugins Feb 07 21:47:42 drags in gst Feb 07 21:47:47 thats it Feb 07 21:48:02 the other stuff I fixed yesterday Feb 07 21:48:22 kergoth: in RDEPENDS this task is optional, in PACKAGES not Feb 07 21:48:31 kergoth: maybe we could make it optional in RDEPENDS as well? Feb 07 21:48:34 may be split bluez to recipe w/o audio and recipe with audio only Feb 07 21:48:55 Jay7: why bother on machines not bluetooth capable? Feb 07 21:49:13 filip hm its a machine feature Feb 07 21:49:20 remove bluez at all from machine then Feb 07 21:49:33 yup but it's not conditional in recipes/tasks/task-base.bb:34 Feb 07 21:49:34 what machine dont have a usb? Feb 07 21:49:44 woglinde_: jornada 7xx unhacked Feb 07 21:49:50 it's not only a machine feature, by the way..its a combined feature, if it's not listed in the distro features it shouldn't be built either Feb 07 21:49:50 ah hehe Feb 07 21:50:03 yeah so its a bug Feb 07 21:50:05 i expect this is a weakness of our packaging :\ Feb 07 21:50:18 ok, could we just prepend it on line 34 as on line 109? Feb 07 21:50:23 ideally it would see that the emitted package isn't actually used anywhere, and not build its rdepends/rrecommends Feb 07 21:50:38 but for now, yeah, likely best to conditionalize the packages entries Feb 07 21:50:42 * kergoth ponders Feb 07 21:50:55 so, should I send a patch? Feb 07 21:50:55 maybe oe-lite has the right idea, no build vs runtime separation Feb 07 21:50:59 someone knows if udev-165 runs with .39 kernel? Feb 07 21:51:06 filip: yep Feb 07 21:51:31 hmmm Feb 07 21:51:45 woglinde_: .29? Feb 07 21:51:51 jay7 yes Feb 07 21:52:00 toshiba ac100 is stuck at it Feb 07 21:52:16 besides it was first byable tegra2 product Feb 07 21:53:26 but there is another place to fix - the last one: classes/kernel.bbclass Feb 07 21:53:41 unconditional bluez-dtl1-workaround in DEPENDS Feb 07 21:53:47 or maybe it doesn't pull in stuff... Feb 07 21:53:52 we still need to bring in pb's bluez-libs vs bluez-apps for bluez4, i never finished undoing the naming change on my bluez branch Feb 07 21:54:02 filip seems at the end of the week you will have commit rights Feb 07 21:54:17 hehe Feb 07 21:54:19 kergoth gogo Feb 07 21:54:23 * kergoth mutters Feb 07 21:54:35 woglinde_: kristoffer used to push my stuff but he works @ ER now ;p Feb 07 21:54:44 kergoth: If you will do this and autoconf changes I will run testbuilder on your branch ;) Feb 07 21:54:48 * woglinde_ calls kergoth's boss Feb 07 21:55:30 and may be even run results in qemu ;) Feb 07 21:55:33 filip so its time to get rw-access on your own Feb 07 21:57:11 yeah.. better to have rw access and don't use it than have no rw access ;) Feb 07 21:57:12 well it's up to you guys, I can keep working up my credibility if you need it Feb 07 21:57:46 filip sent enough patches Feb 07 21:57:56 woglinde_: pretty sure i started the bluez branch at mvista ;) Feb 07 21:57:57 filip: btw, have you tested apt-get on your jornada? Feb 07 21:58:17 khem: ynezz: it looks like you'r the last two touching curl recipe...well it looks like a ldap check is needed Feb 07 21:58:17 Jay7: not yet, I still have some update-alternatives fixes to do Feb 07 21:58:31 filip: I'm interested in RAM amount it eats Feb 07 21:58:36 Jay7: unless you mean debian ;p Feb 07 21:59:01 filip: underlying OS it's not important for me here :) Feb 07 21:59:37 Jay7: well I had gentoo running on my jornada Feb 07 21:59:56 Jay7: remember tossing it under the bed for a couple of days to build a whole system, with distcc and stuff Feb 07 22:00:18 hehe.. my friend have gentoo on his Zaurus Feb 07 22:00:26 with distcc and stuff too :) Feb 07 22:00:58 I still should have an image of the CF card with it somewhere Feb 07 22:02:39 okay good nite Feb 07 22:04:40 ok I need a bit of help here... why does RDEPENDS use {@base_contains('COMBINED_FEATURES'... and PACKAGE uses MACHINE_FEATURES and DISTRO_FEATURES ? Feb 07 22:05:15 wouldn't it be more coherent to use COMBINED_FEATURES in both? Feb 07 22:06:37 it was intended that the distro features would control what went into the image, while machine features controls what gets built (e.g. available in feeds), afaik Feb 07 22:06:44 but i don't recall, that task has been around forever Feb 07 22:14:37 ok, sent the patch Feb 07 22:33:08 there seems to be some kind of a problem parsing recipes in bitbake - children hang on reading from a pipe descriptor Feb 07 22:35:41 filip: hm.. seems your patch should help us (Zaurus) too Feb 07 22:35:59 clamshells have no bluetooth in MACHINE_FEATURES Feb 07 22:36:48 Jay7: the gain is even bigger than I've described Feb 07 22:36:58 Jay7: 3375 -> 1342 Feb 07 22:37:00 but it may use CF or USB BT -> so should go into feeds Feb 07 22:40:52 So... Feb 07 22:41:07 I think switching to gtk+ 2.20.x and glib-2.26.x is breaking me with a crazy race sometimes Feb 07 22:41:18 In that you shouldn't be mixing them like that.. Feb 07 22:43:51 Tartarus: I built from scratch cmake-native, libxml2-native and curl-native - all fine Feb 07 22:44:09 Tartarus: exactly the two versions :/ Feb 07 22:44:18 glib-2.0-2.26.1-r3 Feb 07 22:44:31 glib-2.0-2.26.1-r3 Feb 07 22:44:37 gtk+-2.20.1-r10.4 Feb 07 22:46:07 Tartarus: I can retry with BB_NUMBER_THREADS = 2 and see Feb 07 22:46:21 Your problem went away yes? Feb 07 22:46:27 with4 is repeatable here Feb 07 22:46:43 still? Feb 07 22:46:53 it was one hour ago Feb 07 22:47:09 so, no, is there Feb 07 22:47:20 let see with less bbthreads Feb 07 22:48:01 I don't think it is because of ldap Feb 07 22:48:16 but still... Feb 07 22:49:04 disable parallel make for glib or gtk :) Feb 07 23:48:02 Tartarus: with 2 threads all is slower and the order of tasks differs (before curl-native do_configure was failing with task id 1450 and now I am at 1723 and hasn't yet started) Feb 07 23:49:36 no races until now Feb 07 23:54:04 git bisect run time Feb 08 00:09:07 oh, is building right now (got task 2158 of 5637) Feb 08 00:12:24 and yes, finds ldap.h Feb 08 00:52:33 gn Feb 08 01:56:37 * Tartarus takes own advice, puts git bisect on this strange gst-plugins-base failure **** ENDING LOGGING AT Tue Feb 08 02:59:57 2011