**** BEGIN LOGGING AT Tue Sep 07 02:59:57 2010 Sep 07 05:52:00 hey guys, i'm playing with OE's sdk build capability, I created a meta-toolchain-myboard.bb, and the task-myboard-toolchain-target which had glibc-dbg in it, now when i do bitbake meta-toolchain-myboard, it fails since do_populate_sdk finds unmet recommendations for glibc-locale stuff Sep 07 05:52:31 i removed that, just left task-sdk-bare and did bitbake -c clean and tried rebuilding but same error. any suggestions? Sep 07 06:01:15 htns: sorry, I can only propose to use the mailinglist Sep 07 06:04:18 htns: and what does your bb look like ? Sep 07 06:08:17 Hi all Sep 07 06:08:26 I am merging native and non native recipes Sep 07 06:09:18 In libfribidi recipe there is recipe named fribidi_0.19.2.bb Sep 07 06:09:48 and libfribidi_0.10.4.bb and libfribidi-native_0.10.4.bb Sep 07 06:10:05 we I use BBCLASSXTEND in fribidi.bb file Sep 07 06:10:51 try to build libfribidi-native package it says multiplt providers for libfribidi-native set preferred_provider Sep 07 06:11:25 should I also set the preferred_provider and add it in the patch ? Sep 07 06:13:39 greetings openembedders Sep 07 06:16:07 I've got a curious problem. I built a newer version of boost, then decided it wasn't for me -- now when I try to build my older version it complains 'Package version for package went backwards which would break package feeds from (${new-but-not-preferable} to ${old-and-works}) -- tried a distclean on package-index, doesn't seem to fix things. Sep 07 06:24:27 Noor: determine differences between these two recipes Sep 07 06:24:44 and probably only one will be needed Sep 07 06:25:21 nar: you need to delete the ipk from your feeds Sep 07 06:25:45 Noted. Thanks for the input. Sep 07 06:26:09 nar: plus you need to select the old version thru PREFERRED_PROVIDED Sep 07 06:26:19 PREFERRED_PROVIDER I meant Sep 07 06:26:22 then do Sep 07 06:26:34 bitbake -c clean boost; bitbake boost Sep 07 06:26:51 'k Sep 07 06:28:27 khem libfribidi recipe is downloading fribidi-0.10.4.tar.bz2 from sourceforce_mirrors Sep 07 06:29:05 Incidentally, I did something kinda neat the other day. I built what I call a 'solid image', which is a concatenation of a (u-boot script, a kernel+initramfs for rootpivot, rootfs in ext2 format) -- the u-boot script contains meta-data for offsets/size of the kernel+initramfs and the rootfs such that in u-boot you just need to ext2load /solidimage 0x0;go 0x0, and it takes care of the chain of booting/mounting automatically. Sep 07 06:29:17 and fribidi is downloading from http://fribidi.org/download the version is fribidi-0.19.2.tar.gz Sep 07 06:29:20 Is that worth writing up a recipe for to send up-stream you think? Sep 07 06:30:50 Noor: ok. seems fribidi is newer so try to see whats there in libfribidi which is different preferably get all into fribidi recipe Sep 07 06:31:07 Noor: and rename/delete old recipe if no one is using it Sep 07 06:31:32 but there is one comment in fribidi recipe Sep 07 06:31:36 it says Sep 07 06:31:57 nar: you could create a wiki page for this if you will Sep 07 06:32:24 Noor: you have to read thru the recipe Sep 07 06:32:32 # Slightly incompatible with 0.10.x, so: Sep 07 06:32:40 Yeah. I have diagrams and everything too. :p Sep 07 06:32:43 DEFAULT_PREFERENCE = "-1" Sep 07 06:32:47 DEFAULT_PREFERENCE_angstrom = "1" Sep 07 06:33:13 nar: nice it would be if you put it up in wiki Sep 07 06:33:23 Noor: hmmm Sep 07 06:33:40 zecke, khem , thx, i was able to get it to build by renaming everything to board2 and having only task-sdk-bare in task-board2-toolchain-target.bb Sep 07 06:34:28 it seems like bitbake -c clean meta-toolchain-myboard was not sufficient to wipe the dependencies it had generated Sep 07 06:34:36 secondly this recipe has been added since Aug 8th 2009 Sep 07 06:35:04 so its quite old recipe Sep 07 06:35:22 Aug 8th 2009 is not that old Sep 07 06:35:40 gm Sep 07 06:36:16 :) Sep 07 06:36:32 good morning eFfeM_work Sep 07 06:37:59 we have too many DP's and DP_something's in oe; especially in old recipes DP=-1 seems to be a fair indication that the recipe is probably never used Sep 07 06:38:27 noor and the multiple providers is probably because you still have the native recipe around Sep 07 06:39:01 Noor: it seems not many are using libfribidi recipe Sep 07 06:39:22 eFfeM_work: but when I built the target packagae for libfribidi fribidi_0.19.2.bb is being picked by bitbake Sep 07 06:39:29 Noor: you could consolidate the fribidi_0.19.2.bb Sep 07 06:39:58 Noor, I agree with khem and the fact that 0.19.2 is checked is due to default_preference Sep 07 06:40:19 but DEFAULT_PREFERENCE = "-1" is in the fribidi recipe Sep 07 06:40:24 not in the libfribidi recipe Sep 07 06:40:25 conf/compatibility-providers.conf has PREFERRED_PROVIDER_libfribidi ?= "fribidi" Sep 07 06:40:41 Noor: you can require conf/compatibility-providers.conf in your local.conf too Sep 07 06:40:55 but that will not solve the native problem I guess Sep 07 06:42:32 * khem has 15000/82500 tasks building with bitbake world before some thing in java/swt chickens out Sep 07 06:42:49 java needs love in OE Sep 07 06:43:07 I have fixed few things Sep 07 06:43:20 khem, cool Sep 07 06:43:57 khem when I tested it I did bitbake -k world to get a complete list of what would fail Sep 07 06:44:20 oh I dont do that I go one by one Sep 07 06:44:26 I dont have that much time anyway these days Sep 07 06:44:52 whats Roberts Schuster's nick Sep 07 06:44:59 and when posting my results I got some nasty remarks from a certain dev that we should disallow bitbake world Sep 07 06:45:03 khem no idea Sep 07 06:45:17 if I change any preferred_provider then do I have to upstream that as well Sep 07 06:45:21 heh disable bitbake world why Sep 07 06:45:36 Noor: no, Sep 07 06:45:45 Noor: its a distro choice Sep 07 06:45:59 then how OE maintainer will test it Sep 07 06:46:27 Noor: they need to select it Sep 07 06:46:33 I mean when he want to test it. It will get the same error that I am getting multiple provider error Sep 07 06:47:00 Noor: as I said try to figure out if we really need the old recipe Sep 07 06:47:00 OK. That means they will habdle it themselves I dont have to worry about it Sep 07 06:47:42 multiple providers are ok but there need to be a good reason to have two recipes for seemingly same source Sep 07 06:48:07 I can uderstand two different recipes providing same thing Sep 07 06:49:59 khem here is the message on bitbake world http://lists.linuxtogo.org/pipermail/openembedded-devel/2010-July/022221.html Sep 07 06:50:24 One more dumb question.. I'm using codesourcery toolchain, and I'm running into a problem with stack-traces that I'm unsure if it's related to CSL or not -- I compile my app with '-gstabs -O0 -rdynamic' (along with my platform specific stuff), and when I force a core-dump (kill -SIGSEGV), the threads in the core file (except for the main thread) always show up as 0x00000000 with no stacktrace. Am I missing something stupid here? Sep 07 06:52:12 nar: gstabs ??? Sep 07 06:52:22 nar: why not dwarf Sep 07 06:52:37 use -g -O0 Sep 07 06:52:44 Trying to use google-breakpad, I still have the same problem with vanilla -g Sep 07 06:52:45 and you will be happier man Sep 07 06:53:31 nar: is it arm Sep 07 06:53:40 indeed Sep 07 06:54:07 are you using CSL for all compilation jobs ? Sep 07 06:54:16 yes, 100% Sep 07 06:54:21 ok Sep 07 06:54:44 and you dont have -fomit-frame-pointer in CFLAGS Sep 07 06:55:08 I would suggest that you install the -dbg packages on rootfile system too Sep 07 06:55:14 and then retry Sep 07 06:55:46 Hmm. I tried that, I don't explicitly say -fomit-frame-pointer but.. I'm wondering now if there's something setting that serruptitiously. I should double check. Sep 07 06:57:17 god damn it Sep 07 06:57:21 Good call khem Sep 07 06:57:27 nar: you can add -fno-omit-frame-pointer to your global flags Sep 07 06:57:53 Where would I do that? local.conf? Sep 07 06:58:20 nar: -02 enables -fno-omit-frame-pointer on ARM by default Sep 07 06:58:27 good morning Sep 07 06:58:58 nar: yes FULL_OPTIMIZATION += "-fno-omit-frame-pointer" something on those lines Sep 07 06:59:04 mckoan: gm Sep 07 06:59:24 Well what I'm seeing is Sep 07 06:59:24 export TARGET_CXXFLAGS="-isystem/opt/build/tmp/staging/armv5te-none-linux-gnueabi/usr/include -fexpensive-optimizations -frename-registers -fomit-frame-pointer -O2 -ggdb3 -fpermissive" Sep 07 06:59:42 ahh, thank you Sep 07 07:00:09 nar: whats your distro Sep 07 07:00:31 custom at this point, started out with angstrom-2008.1, barely resembles it anymore. Sep 07 07:00:53 nar: ok do you still include the files that angstrom.conf included ? Sep 07 07:01:20 some of them yeah Sep 07 07:01:23 e.g. conf/distro/include/angstrom-eglibc.inc etc Sep 07 07:01:29 whats your libc ? Sep 07 07:01:39 glibc that shipped with codesourcery Sep 07 07:01:42 glibc or eglibc Sep 07 07:01:43 ok Sep 07 07:02:07 then you could also edit conf/distro/include/angstrom-glibc.inc Sep 07 07:02:10 and add it there Sep 07 07:02:21 infact remove -fomit-frame-pointer Sep 07 07:02:43 it may help it may not worth a try there may be other problems Sep 07 07:03:04 definitely. good input Sep 07 07:03:36 unless I see the build tree and logs its hard to deduce what could be wrong at your end Sep 07 07:03:50 its midnight sh**t Sep 07 07:03:58 time to sleep Sep 07 07:04:04 I think this is on the right path. I'm rebuilding my distro, should be easy from here. Sep 07 07:04:08 gn all Sep 07 07:04:13 Thank you for your input khem, zecke. Sep 07 07:04:14 night Sep 07 07:04:37 rschus: hey Sep 07 07:04:48 rschus: I have some java patches posted to ml Sep 07 07:04:58 I would love to hear your feedback Sep 07 07:06:12 khem: ok, will review them later (some stuff to do here still) Sep 07 07:06:42 rschus: thanks Sep 07 07:07:04 * khem really sleeps Sep 07 07:07:41 nite khem Sep 07 07:35:51 morning Sep 07 07:35:59 good morning Sep 07 08:11:50 good morning Sep 07 08:12:44 03Andreas Mueller  07org.openembedded.dev * r8d9d249f12 10openembedded.git/recipes/moblin/ (6 files in 2 dirs): (log message trimmed) Sep 07 08:12:44 moblin: update recipes to match latest clutter Sep 07 08:12:44 * nbtk: bump required to configure properly with clutter Sep 07 08:12:44 * clutter-imcontext: now required by nbtk Sep 07 08:12:44 * hornsey_git: bump required to configure properly with clutter Sep 07 08:12:45 * libccss_git: bump required to match new nbtk requirements Sep 07 08:12:46 bumping by SRCREV_pn-libccss = .. is a hack to get throgh the door Sep 07 08:12:46 03Andreas Mueller  07org.openembedded.dev * r8d64681eb7 10openembedded.git/recipes/clutter/ (21 files in 5 dirs): (log message trimmed) Sep 07 08:12:47 clutter: clean builds for clutter 1.0 Sep 07 08:12:47 * all 0.9 recipes (are actually 1.0rc) & files referenced were removed Sep 07 08:12:48 * all 0.8 recipes were renamed to clutter*-0.8 to ensure parallel Sep 07 08:12:48 installation with later versions Sep 07 08:12:49 * clutter-common.inc as container for helpers was created (also from Sep 07 08:12:50 clutter-fpu.inc) Sep 07 08:12:50 03Andreas Mueller  07org.openembedded.dev * r970c9cfb83 10openembedded.git/recipes/clutter/ (3 files): (log message trimmed) Sep 07 08:12:51 clutter (+box2d +gtk): Bump to latest (stable) versions Sep 07 08:13:12 * removed clutter*svn*: repositories no more available Sep 07 08:13:12 * removed clutter* with version < 0.8 Sep 07 08:13:12 * removed moblin-proto: repositories no more available Sep 07 08:13:12 * removed table: repositories no more available (now in toys/attic) Sep 07 08:13:12 03Andreas Mueller  07org.openembedded.dev * r712d5d30f5 10openembedded.git/recipes/gnome/ (3 files): (log message trimmed) Sep 07 08:13:12 libchamplain(-gtk): Bumped to 0.7.1 to meat actual clutter Sep 07 08:13:13 * builds separate packages for libchamplain and libchamplain-gtk Sep 07 08:13:13 * libchamplain_0.2.8 removed because it tries to configure against Sep 07 08:13:14 (3 lines omitted) Sep 07 08:15:25 morning Sep 07 08:20:56 morning Sep 07 08:59:11 hm, there is no busybox 'arch' applet ?? Sep 07 08:59:47 is there a package that delivers arch ? Sep 07 09:01:22 nevermind, uname -m does it Sep 07 09:14:48 Is there any instrinsic value of running real 'glibc' over 'eglibc' ? Sep 07 09:15:33 I assumed there would be.. But CSL's shitty toolchain has me at my wits end. Sep 07 09:40:53 hm makedevs refuses to create a fifo for me does this sound familar ? Sep 07 09:41:19 In the build environment or on the running embedded system? Sep 07 09:41:28 in the build env Sep 07 09:41:51 it does make the fakeroot entries for block and char devs Sep 07 09:42:44 hmm. Can't say I've run into that. Whats the error? Sep 07 09:43:15 | makedevs: /home/frans/workspace/gtd/tmp/rootfs/axon-bootmode-image/dev/log: file can not be created with mknod! Sep 07 09:44:29 Whats your device table entry look like? Sep 07 09:44:59 /dev/log p 640 0 0 - - - - - Sep 07 09:45:08 commented it out for now, the b and c's work fine Sep 07 09:45:15 hmm! Sep 07 09:45:27 actually i retract that comment Sep 07 09:45:45 the b and c also are not created but the error is not trapped Sep 07 09:45:51 ! Sep 07 09:47:00 i have a nice rfs actually I can even boot from it, but /dev is empty (actually would have expected init to fail if it cannot find /dev/console Sep 07 09:47:28 Do you control the hardware platform? Sep 07 09:47:33 yes Sep 07 09:47:41 will there be stuff hotplugging into it? Sep 07 09:47:51 or a static set of devices Sep 07 09:47:56 nope, that is why i'm using a static table Sep 07 09:47:59 heh Sep 07 09:48:11 makes things a little bit faster and no need to have udev or friends on the system Sep 07 09:48:26 Consider this the worst advice you'll get all day: tar up a 'good' dev directory and just untar it and don't mess around with mknod at all. :p Sep 07 09:49:08 Going through makedevices when you've got a static set of devices is pedantic IMHO. Sep 07 09:49:26 (and, wasted cycles during bootup) Sep 07 09:50:21 well oe does support devicetable, and devicetable makes it somewhat simpler if hthings need to be changed Sep 07 09:51:15 Otherwise, I'd investigate fakeroot not working correctly. Sep 07 09:51:16 makedevice's add_new_fifo() looks trivial.. it just tries mknod(path,mode,0) and craps out if it fails. Sep 07 09:51:43 I suppose. Sep 07 09:52:31 yes, well added a perror to it and that showed no such file or dir, apparently the dev dir does not exist yet Sep 07 09:53:00 interesting. Sep 07 09:53:38 ah, compared with another dev talbe that one indeed starts with Sep 07 09:53:43 /dev d 777 0 0 - - - - - Sep 07 09:53:55 ho ho Sep 07 09:54:20 guess that'll help Sep 07 09:55:23 Odd that /dev wouldn't exist until makedevs 'makes' it. Seems kinda like something that the image would force by default. Sep 07 09:55:33 would/should Sep 07 09:55:33 yes, will look at it Sep 07 09:56:45 I'm as giddy as a school girl. glibc 2.10.1 is compiling with gcc 4.4.3 for me under ARM. Sep 07 09:57:01 It pays to keep git sync'd I guess. Sep 07 09:57:30 non-corrupted thread stack-traces are in my near future. Sep 07 09:57:32 guess so, i had no problems with glibc and arm Sep 07 09:58:04 I used CSL as my toolchain.. a decision I've now kind of hastily got to reverse. Sep 07 09:58:29 With 2 weeks till customer ship! Sep 07 10:00:29 never used csl, i always build the std cross compiler Sep 07 10:00:45 what hw are you using ? Sep 07 10:01:00 custom board, Marvell Feroceon based Sep 07 10:01:06 (not pogoplug. :D) Sep 07 10:02:19 512 meg ram, 1GB NAND via SATA bridge, 1ghz processor. Not a bad little box. Sep 07 10:04:33 nar: Sounds cute... link? :) Sep 07 10:05:05 ohh.. well, it's not really general platform. It's specific to office/industrial light controls Sep 07 10:05:36 Quite some power for light controls... Sep 07 10:05:50 Indeed. It does lots of useful things. Sep 07 10:05:58 Things that make zigbee cry. Sep 07 10:06:08 nar, sounds cute Sep 07 10:06:13 * florian got a Dockstar recently Sep 07 10:06:40 The FreeAgent? Sep 07 10:06:50 actually compiled a lot of stuff for sheeva last few days, still need to test & configure Sep 07 10:06:54 * eFfeM_work also has an openrd client Sep 07 10:07:04 ahh yeah Sep 07 10:07:15 Man Sep 07 10:07:17 I'll tell you what Sep 07 10:07:19 I don't care who knows it Sep 07 10:07:23 I ****ing hate Marvell Sep 07 10:07:29 :-) Sep 07 10:07:59 I knew things were going to be bad when I had to crack their documentation PDF to remove the 4 seperate watermarks per page. Sep 07 10:08:11 Everything else has been down-hill. Sep 07 10:09:02 marvell and getting good documentation is a pain (also with NDA etc); things only work with them if you have a good technical contact Sep 07 10:10:21 We originally prototyped with TI OMAP-L137, who rolled a sweet JTAG interface + IDE into an Eclipse setup.. had heaps of stand-alone sample code.. heaps of great docs.. Sep 07 10:10:43 To Marvell, who have no JTAG solution, or IDE.. they suggested we use 'makeroot' for building our distro.. Sep 07 10:11:40 I don't blame Marvell tho, we're small potatos right now.. so they don't care about us that much. Hopefully once we get into production we'll be dealing with numbers that get them excited. Sep 07 10:12:56 The real irony is, the pogoplug has better u-boot _and_ kernel than our 'commercial' processor. Sep 07 10:13:40 The version of u-boot they provided is so old it should have come on a punchcard. Sep 07 10:14:14 they don't care about the small guy, actually I used to work for quite a large company and even there I had problems getting info Sep 07 10:14:19 1.1.4 ? Sep 07 10:14:24 yeah Sep 07 10:14:41 I had to backport a bunch of junk just to make it bareable. Sep 07 10:15:01 i see that with openrd too, they promise upgrades, but they hardly deliver Sep 07 10:15:31 oe u-boot works for sheeva, but misses some functionality that is in their 1.1.4 Sep 07 10:15:43 i think sd support and maybe bubt Sep 07 10:15:59 heh, bubt. I remember that. Sep 07 10:16:52 I had problems with bubt and u-boot's netconsole I think. Sep 07 10:17:22 I just wrote a bunch of u-boot script that does basically what bubt does to compensate. Sep 07 10:17:26 the u-boot they use is indeed not that good Sep 07 10:17:45 Our board is kinda tricky.. no JTAG or UART pins Sep 07 10:18:45 no jtag, that is nasty if your flash gets corrupted Sep 07 10:18:48 nar: heh, yes, working with marvell is just a bad idea. Sep 07 10:18:54 it is only four wires Sep 07 10:19:16 The hardware guys are all ex-cisco, so they're front-loading the cost-reduction. :D Sep 07 10:20:10 I've got a couple dev boards to experiment on, but.. it's 'no mistakes allowed' when I push stuff live. Sep 07 10:20:20 ah ok, had to make a mindset change as well when I changed from high vol to low vol electronics Sep 07 10:20:46 I fought for months against not having serial port. :( And lost. Sep 07 10:20:57 but not having jtag is not really a cost saver, you just need to have a few lines Sep 07 10:21:08 same for serial, do the level shifting and buffering off-board Sep 07 10:21:14 oh, we have it.. just no header Sep 07 10:21:21 ah ok Sep 07 10:21:40 So nuking in the field is an RMA event, nuking in the lab is 'go cry to a hardware guy' event. Sep 07 10:22:42 hrm. glibc-2.10.1 didn't build for me. rats. Sep 07 10:23:25 nar, i recall that, there the device used drm and the user should not be able to hack that Sep 07 10:23:34 nar glibc for me builds for sheeva Sep 07 10:23:53 What GCC version you using/ Sep 07 10:23:54 ? Sep 07 10:29:09 I think 4.3.3, I've done that at home, just select sheevaplug as machine and you'll see what it builds Sep 07 10:30:39 hmm! ok. I was trying 4.4.3 Sep 07 10:30:45 Perhap this will improve my odds! Sep 07 10:31:17 I really want to get google-breakpad working by 8am Sep 07 10:31:31 It's all integrated into the app, but the stack-trace for threads is junk. Sep 07 10:31:41 So hopefully this will fix it. Sep 07 10:42:29 [STABLE] I would like to include in our SDK the possibility to create ipks, but unfortunately ipkg-build isn't there. And ipkg-utils-native doesn't install anything. Sep 07 10:42:38 How do I get ipkg-build? Sep 07 10:44:54 ipkg-build is part of ipkg-utils, but.. if your OE is setup right, it should be able to make packages on it's own. Sep 07 10:45:22 You don't normally have to run ipkg-build manually Sep 07 10:47:43 hrm. So, if I use 'internal' GCC for cross-compiling.. Is there a simple way to provide use of it to my developers? With CSL it's just 'add this to your path', which was an attractive feature. Sep 07 11:11:49 nar: there is the possibility to make an sdk, but I never tried it myself Sep 07 11:13:02 03Stefan Schmidt  07org.openembedded.dev * rd70dc036b7 10openembedded.git/recipes/netbase/ (netbase/imote2/interfaces netbase_4.21.bb): Sep 07 11:13:02 netbase: Add interface file for imote2 and bump PR Sep 07 11:13:02 We need to set 'auto usb0' here to avoid some random kernel oops when trying to Sep 07 11:13:02 to gadget_ethernet. Sep 07 11:16:47 hmm Sep 07 11:21:39 Building SDKs works fine in stable/2009, I use it to provide cross-compiler for my colleagues Sep 07 11:22:25 opkg is part of the SDK but 1) libssl and libcrypto are missing and 2) opkg can't CREATE ipks anyway Sep 07 11:23:03 ipkg-utils-native doesn't have a do_install() method in stable/2009 so there is no ipk to add to my SDKs Sep 07 11:23:43 Result: my colleagues can't provide ipks to install on our hardware Sep 07 11:25:47 Could dpkg-buildpackage from debian-lenny build packages for embedded devices? Will try... Sep 07 11:36:01 hi, if got a small problem: I want to use in a machine the current cvs-kernel of hh.org. so I defined: Sep 07 11:36:03 PREFERRED_PROVIDER_virtual/kernel = "linux-handhelds-2.6" Sep 07 11:36:05 PREFERRED_VERSION_linux-handhelds-2.6 = "cvs" Sep 07 11:36:07 but the preferred version doesn't call the linux-handhelds-2.6_cvs.bb - instead linux-handhelds-2.6_2.6.16-hh9.bb is used. why?? Sep 07 11:36:09 if I define e.g. "2.6.21" this will work. Sep 07 11:37:21 hi all Sep 07 11:38:13 I have a question if someone can answer Sep 07 11:38:36 when merginig native and target recipes, what to do if they are both inheriting from different recipes Sep 07 11:39:31 "inheriting"? Depending or requiring? Sep 07 11:41:25 inherit Sep 07 11:42:29 vps: using the *inherit* keyword Sep 07 11:42:48 Sorry, no idea. I thought it was only possible to inherit from classes Sep 07 11:43:14 yes, from calsses, not recipes Sep 07 11:44:57 dnjl: I think you have to match PV in your preferred version: PV = "${K_MAJOR}.${K_MINOR}.${K_MICRO}-hh${HHV}+cvs${SRCDATE}" Sep 07 11:45:42 vps: in non-native recipe, i have "inherit autotools pkgconfig binconfig" and in native, it is "inherit autotools pkgconfig native gettext" Sep 07 11:48:38 fahad, i'm not too sure on binconfig, but if the native recipe requires gettext, I would expect it in the nonnative recipe too Sep 07 11:51:45 eFfeM_work: okay, then i think i should just add gettext to non-native as well, and see if it builds fine Sep 07 11:51:52 yes Sep 07 12:00:33 vps: tnx, this works if I define K_* and HHV before Sep 07 12:09:46 03Koen Kooi  07org.openembedded.dev * rd5b3850cb6 10openembedded.git/recipes/bash/bash_3.2.bb: bash 3.2: bump PR Sep 07 12:09:46 03Koen Kooi  07org.openembedded.dev * r16d7a4c8a3 10openembedded.git/conf/distro/include/ (4 files): Sep 07 12:09:46 angstrom: switch to ggdb2 to save space, noone seems to really exploit the things -ggdb3 provides Sep 07 12:09:46 Signed-off-by: Koen Kooi Sep 07 12:09:46 Acked-by: Graeme Gregory Sep 07 12:09:46 Acked-by: Philip Balister Sep 07 12:09:54 03Koen Kooi  07org.openembedded.dev * rc13e53be1a 10openembedded.git/recipes/mozilla/ (16 files in 2 dirs): firefox: update to 3.6.8 Sep 07 12:12:15 hrm. I keep getting 'This configuration not matched in ../abi-tags' Sep 07 12:12:19 when I try to build glibc Sep 07 12:13:12 hm, never seen taht Sep 07 12:14:49 #define __ABI_TAG_OS 0 Sep 07 12:14:49 #ifndef __ABI_TAG_VERSION Sep 07 12:14:49 # define __ABI_TAG_VERSION 2,0,0 Sep 07 12:14:49 #endif Sep 07 12:15:05 is the contents of tmp/work/armv5te-angstrom-linux-gnueabi/glibc-2.10.1-r36.1/build-arm-angstrom-linux-gnueabi/csu/abi-tag.h Sep 07 12:16:21 no idea, I haven't seen this with the sheeva build I did last weekend (but that was on my system at home, so cannot check what I have in that file) Sep 07 12:16:26 rgr Sep 07 12:18:09 I'm glad I use source control.. Because this dev tree is a disaster at this point. Sep 07 12:24:34 ahh. __ABI_TAG_VERSION should be the kernel sources it's built against.. 2,0,0 hardly seems right. Sep 07 12:25:49 Using "Debian `dpkg-deb' package archive backend version 1.14.29 (i386)" I were able to create a .deb which installs on our ppc603e-based machine. :-) Sep 07 12:26:16 yay Sep 07 12:26:51 Hi Sep 07 12:27:01 greets Sep 07 12:27:12 ls Sep 07 12:27:16 I am trying to boot beagle board rev C3 Sep 07 12:27:20 no such file! Sep 07 12:27:30 and I haven't ben able to boot it Sep 07 12:27:59 Any body who has worked on beagle board plz respond thanks Sep 07 12:28:27 waiting!!!!!!!!!! Sep 07 12:29:08 lol Sep 07 12:29:29 No need to wait more : I've worked on a beagle board. It was booting. :-D Sep 07 12:29:49 like a charm Sep 07 12:30:04 waiting too :-) Sep 07 12:30:07 Please insert $100 per hour! Sep 07 12:30:13 I m tryiing to boot it from MMC card and it shows only initial logo Sep 07 12:30:18 http://elinux.org/BeagleBoard Sep 07 12:30:19 1,000 hours minimum Sep 07 12:30:32 anand_: have you read the channel topic ? Sep 07 12:30:39 ~topic Sep 07 12:30:44 help topic Sep 07 12:30:54 where is it? Sep 07 12:31:06 just type /topic ;> Sep 07 12:31:08 give me the link Sep 07 12:31:12 you get it when you join the channel (at least on decent clients Sep 07 12:31:24 ok Sep 07 12:31:41 ok, now what is the part of the last sentence that you did not understand ? Sep 07 12:32:39 See I need help regarding beagle board, if u can plz respond Sep 07 12:33:12 I copied MLO and u-boot.bin into MMC card and tried to boot borad from MM card Sep 07 12:33:16 anand_: check : http://elinux.org/BeagleBoard and here : http://elinux.org/BeagleBoardBeginners Sep 07 12:33:23 anand_ have you tried #ubuntu that might be equally inappropriate Sep 07 12:33:25 ok thanks' Sep 07 12:33:38 and things like (02:28:05 PM) anand_: waiting!!!!!!!!!! Sep 07 12:33:44 does not help to get answers either Sep 07 12:33:48 No actually i want to try symbian Sep 07 12:33:56 you might want to try #beagle Sep 07 12:34:33 I m trying symbia on beagle Sep 07 12:34:41 symbian Sep 07 12:34:50 eFfeM_work: I think he does not get the idea that #xxxxx means "channel on IRC" Sep 07 12:35:19 Jin^eLD: nope Sep 07 12:35:29 anand_: please type Sep 07 12:35:33 /join #beagle Sep 07 12:35:41 ok Sep 07 12:36:51 it is a much more appropriate place for your Q Sep 07 12:37:01 or try this link : http://lmgtfy.com/?q=beagleboard+symbian Sep 07 12:37:11 * eFfeM_work wonders how people interested in symbian on beagle end up in #oe anyway Sep 07 12:37:16 ericben: :P Sep 07 12:39:52 http://members.fortunecity.com/therealpuppet/annoy.html Sep 07 12:40:03 doesn't mention the use of excessive exclamation marks though Sep 07 12:41:31 ehm, how to I remove all old kernel-builds from a build root? I don't want delete the whole build env and start from scratch. Sep 07 12:41:33 I've built some experimental kernels and now, if I build a image, it will every time include a newer kernel (and modules), which I doesn't use anymore. Sep 07 13:09:21 03Koen Kooi  07org.openembedded.dev * r56a7be995b 10openembedded.git/conf/distro/ (angstrom-2008.1.conf angstrom-2010.x.conf): Sep 07 13:09:21 angstrom: bump DISTRO_PR Sep 07 13:09:21 See http://thread.gmane.org/gmane.linux.distributions.angstrom.devel/4426 Sep 07 13:13:41 glib-2.0 - 2.24.1 - r0 : configure: error: Could not find a glib-genmarshal in your PATH ... it seems glib-2.0 needs glib tools to be installed on the host Sep 07 13:15:31 ericben: what distro do you use on host Sep 07 13:15:47 * eFfeM_work never had that message (ubuntu 8.04, 10.04 and opensuse 11.2) Sep 07 13:16:02 I think the customer has a kubuntu ! Sep 07 13:16:14 ubuntu & opensuse may already have glib installed Sep 07 13:16:38 kubuntu may not install this, I'm investigating Sep 07 13:36:56 can anyone hint me on bbclasses? looking at the autotools bbclass as example; does "EXPORT_FUNCTIONS do_configure do_install" mean that the defined functions autotools_do_configure and autotools_do_install will be called instead of the "default" ones? Sep 07 13:39:01 ~topic Sep 07 13:39:21 vps it is /topic Sep 07 13:39:47 thank you! I always wondered whether this would work in pidgin Sep 07 13:40:31 yw Sep 07 13:55:56 what is native machine conf supposed to be doing? a lot of packages fail because they attempt to install stuff to / on the build host (where they of course get permission denied) Sep 07 14:08:56 hey Sep 07 14:08:58 oops wrong window Sep 07 15:05:49 hello. i used to be able to point a bb recipe to a folder on disk using file://path-to-folder as SRC_URI. however nowadays it seems that does not work anymore; is this intentional? Sep 07 15:09:19 hi! I compiled opie-image with angstrom to EZX Phone A1200..it work but with wrong colours..any idea? my device uses 18bpp... Sep 07 15:09:28 CMoH: sounds like you aren't using a current version of OE Sep 07 15:10:04 kergoth_, so is this not intentional? Sep 07 15:10:14 it was broken, it was fixed Sep 07 15:10:18 i see; thanks Sep 07 15:10:44 CMoH: http://git.openembedded.org/cgit.cgi/openembedded/commit/?id=dd478c59e5ec36bc5f43ed9721323ad8171b5ef4 Sep 07 15:11:13 thanks Sep 07 15:11:33 that's a pretty recent commit Sep 07 15:12:04 a week ago, yes Sep 07 15:12:23 any help? Sep 07 15:13:45 something wrong with the server? Getting permission denied when cloning repository Sep 07 15:14:21 i ever clone as root,after i do chmod -R 777 openembedded Sep 07 15:22:27 Im setting up my pushing environment on a different server (development server). All I need is to move my keys to that server right? Sep 07 15:23:04 hi! I compiled opie-image with angstrom to EZX Phone A1200..it work but with wrong colours..any idea? my device uses 18bpp... Sep 07 15:43:07 hi! I compiled opie-image with angstrom to EZX Phone A1200..it work but with wrong colours..any idea? my device uses 18bpp... Sep 07 16:43:39 I realize this isn't the place for it, but I've got a weird issue.. With redundant u-boot env variables, when I use fw_setenv -- when the system resets, it drops directly to the u-boot prompt after the countdown.. doesn't run the bootcmd or anything. Sep 07 16:43:45 I'm baffled. Sep 07 16:43:57 Resetting power seems to 'fix it' Sep 07 17:08:38 03Khem Raj  07org.openembedded.dev * r62b89c2652 10openembedded.git/recipes/udev/ (udev-compat141_141.bb udev.inc udev_151.bb udev_154.bb): Sep 07 17:08:38 udev,udev-compat141: Fix machine specific RDEPENDS Sep 07 17:08:38 Signed-off-by: Khem Raj Sep 07 17:08:50 03Khem Raj  07org.openembedded.dev * r8df63c817f 10openembedded.git/recipes/xml-commons/ (jaxme_0.5.2.bb xpp3_1.1.3.4.O.bb): Sep 07 17:08:51 jaxme_0.5.2.bb, xpp3_1.1.3.4.O.bb: Remove do_stage_append Sep 07 17:08:51 Signed-off-by: Khem Raj Sep 07 17:08:51 Signed-off-by: Robert Schuster Sep 07 17:08:51 03Khem Raj  07org.openembedded.dev * r3b019733d8 10openembedded.git/classes/java-library.bbclass: Sep 07 17:08:51 java-library.bbclass: do_stage is not needed anymore Sep 07 17:08:52 Signed-off-by: Khem Raj Sep 07 17:08:52 Signed-off-by: Robert Schuster Sep 07 17:26:32 nar_: are you doing saveenv ? Sep 07 17:27:10 nar_: usually when you modify the params the stay in ram unless you write them flash using saveenv Sep 07 17:27:34 and when you reset the old values are read from rom again Sep 07 17:37:42 hmm, I have ASSUME_PROVIDED += "virtual/gcc-initial virtual/gcc-intermediate virtual/gcc virtual/g++" - why is it still building gcc-cross-intermediate? Sep 07 17:38:10 is there a reason you're not including 'cross' anywhere in your provided? Sep 07 17:38:44 uhm, maybe I missed it? Sep 07 17:38:54 there is no "virtual/gcc" Sep 07 17:39:05 i suggest you look at the recipes, or various config files that set preferences Sep 07 17:39:18 I copied the line from native machine conf I think Sep 07 17:39:21 bitbake -e gcc-cross-intermediate|grep \^PROVIDES Sep 07 17:39:51 basically I was trying to come up with a configuration that would not build anything, and that should only be capable of produceing "all"-arch ipk's by repackaging some tar files, nothing more Sep 07 17:40:36 an assume provided like that is intended to not build the crosscompiler. that isn't going to magically make it not build anything else Sep 07 17:40:47 it'll just try to build things without having a crosscompiler, and die Sep 07 17:40:59 I know, I tried adding more assume/provided things but that did not really work out Sep 07 17:41:38 assume provided will make bitbake not bulid that recipe *at all*. i guess you can pick the recipes you don't want to build one by one.. have fun Sep 07 17:41:39 but I guess there is no real way to turn all that stuff off, so I'll probably still end up with building at least some minimal set of packages even if I do not need them Sep 07 17:41:50 yeah, thats what I figured Sep 07 17:42:03 so I ended up with trying to avoid at least the "big ones" Sep 07 17:42:12 i really don't see the point. Sep 07 17:42:46 very very few recipes will work at all without a crosscopmiler, much less without their dependencies. you'll have to assume provided every recipe but a few tasks or something Sep 07 17:43:13 well, I want to make use of the convenience that comes with OE/recipes for creating ipk's of some web sources Sep 07 17:43:34 okay, so create a recipe that doesn't depend on anything. Sep 07 17:43:36 which are all noarch and are not complie Sep 07 17:43:41 INHIBIT_DEFAULT_DEPS = "1" Sep 07 17:43:44 in the recipe Sep 07 17:43:47 then bkae the recipe Sep 07 17:43:53 you don't need to use assume provided for that at all Sep 07 17:44:26 hmm; I also have one recipe that basically adds the configuration for opkg, i.e. the stuff in /etc/opkg for a feed for those Sep 07 17:44:27 Jin^eLD: its gcc-cross-intermediate and gcc-cross-initial Sep 07 17:45:10 from what I see that package does not have any depends either, still it was trying to build gcc/glibc and all the other stuff, I guess I must have something in my local.conf then Sep 07 17:46:39 Jin^eLD: bitbake -g is your friend Sep 07 17:46:45 roger that, thanks Sep 07 17:49:00 Jin^eLD, not setting a varaible in a recipe doesn't mean its empty. Sep 07 17:49:08 Jin^eLD, are you forgetting that classes exist? Sep 07 17:49:19 i just told you how to disable the default dependencies Sep 07 17:49:57 well that recipe did not inherit anything.. but yes, I am about to try that variable Sep 07 17:50:02 :) Sep 07 17:50:11 yes, it did. Sep 07 17:50:28 the "base" class along with all classes listed in INHERIT are loaded for all recipes Sep 07 17:51:18 aah, I did not think about that... Sep 07 17:51:21 doh :> Sep 07 17:51:32 if base wasn't loaded, no tasks would exist Sep 07 17:51:35 including configure, compile, install.. Sep 07 17:51:39 the recipe would just do nothing at all Sep 07 17:53:10 oh, I see.. Sep 07 17:53:26 well next step for me will be to write a bbclass, that will handle this custom packaging that I have in mind Sep 07 17:56:18 kergoth_: does "EXPORT_FUNCTIONS do_configure do_install" mean that the defined functions autotools_do_configure and autotools_do_install will be called instead of the "default" ones? Sep 07 17:56:50 yeah, it basically creates wrappers that call the class named ones Sep 07 17:57:06 its a convenience, to make it easier to name class tasks such that they can be run directly from overridden versions in recipes Sep 07 17:57:33 so any foo_do_install in a bbclass will replace the default function? ok.. thx Sep 07 17:57:38 if autotools just defined do_confjigure, there'd be no way to create a do_configure that runs the autotools one :) Sep 07 17:58:13 :) Sep 07 17:59:06 i can't remember if EXPORT_FUCNTIONS is a global thing, or if you have to do it in each class that creates a foo_do_bar Sep 07 17:59:10 drawing a blank Sep 07 18:13:40 03Koen Kooi  07org.openembedded.dev * r3967caeb51 10openembedded.git/recipes/nodejs/nodejs_0.2.0.bb: Sep 07 18:13:40 nodejs: fix recipe Sep 07 18:13:40 * fix configure Sep 07 18:13:40 * fix do_install Sep 07 18:13:40 * fix packaging Sep 07 18:13:50 03AJ ONeal  07org.openembedded.dev * r8a9ba8cde8 10openembedded.git/recipes/nodejs/ (4 files in 2 dirs): Sep 07 18:13:51 added recipe for nodejs Sep 07 18:13:51 Signed-off-by: Koen Kooi Sep 07 18:34:15 03Koen Kooi  07org.openembedded.dev * r241b8865f2 10openembedded.git/recipes/gcc/gcc-package-target.inc: Sep 07 18:34:15 gcc: trade QA errors for broken runtime dependencies Sep 07 18:34:15 * this fixes the missing deps on libstdc++ and libgcc after building gcc Sep 07 18:41:02 khem, got your email, got any other ideas?.. Sep 07 18:59:38 khem: Yeah.. I do the saveenv.. (twice) Sep 07 18:59:47 I'm betting this is a u-boot bug with marvell's crappo u-boot Sep 07 18:59:59 or I have to do saveenv twice or something ridiculous Sep 07 19:51:22 khem how many builds do you run simultaneously on garnet? Sep 07 20:02:56 [Help Please] Why is MACHINE in my local.conf file not recognized? The file contains the line MACHINE = "at191sam9260ek" and the environment variable BBPATH is properly set. However when I run bitbake I get ""Please set a valid MACHINE in your local.conf". But when I set MACHINE in the environment, it works fine. Sep 07 20:07:40 Is there a at191sam9260ek.conf file? Sep 07 20:08:14 Or was that a typo? Sep 07 20:08:37 at91sam9260ek exists Sep 07 20:08:40 not 191 Sep 07 20:11:13 It's typo ... in my local.conf file. Thanks!! Sep 07 20:42:35 hi, can someone confirm bitbake -g works fine ? Here I get a lot of lines like described in this bug report : http://bugs.openembedded.net/show_bug.cgi?id=5002 Sep 07 21:25:15 evening Sep 07 21:25:29 hi Sep 07 21:28:35 hi pb__ Sep 07 21:29:28 How's going wrt house? Sep 07 21:30:26 Monsoon season is coming on Uk... Sep 07 21:32:09 khem, I need to give garnet more RAM...its OOM'd 2X today Sep 07 21:39:52 <[Rui]> kergoth: hi, I noticed you had to revert that fix about the paths of file:// urls, have you come up with a better fix? Sep 07 21:51:18 ka6sox: hi Sep 07 21:51:25 ka6sox: only one build Sep 07 21:51:31 but it runs like 4 jobs Sep 07 21:51:35 in parallel Sep 07 21:52:55 ericben: should be fixed in latest bitbake as it seems Sep 07 21:53:56 Hi pb__ how is your house construction coming up Sep 07 21:56:45 [Rui]: apparently you didn't look closely. it was reverted, then re-applied with an additional fix Sep 07 21:57:03 ka6sox-work: you can bring garnet down if you want Sep 07 21:57:06 <[Rui]> kergoth: maybe it was after I checked the log! Sep 07 21:57:13 I can restart the builds Sep 07 21:57:23 i think it was that day, but perhaps Sep 07 21:57:39 <[Rui]> kergoth: I tend to not update so frequently because then I'll have to wait over a day before I can start doing my packages to test the application :| Sep 07 21:57:42 khem, then I should give it like 2.5GB... Sep 07 21:57:44 * kergoth nods Sep 07 21:57:51 <[Rui]> kergoth: thanks :) Sep 07 21:58:51 ant_mob, khem: well, we have some walls on the ground floor now. Sep 07 21:58:55 ka6sox-work: ok Sep 07 21:59:17 next week, hopefully a first floor as well. Sep 07 21:59:24 pb__: cool. Sep 07 21:59:39 khem, its back up...go ahead and start up all 4 Sep 07 21:59:39 pb__: can you reside in it at same time now Sep 07 21:59:44 and lets see if it oom's Sep 07 21:59:44 ok Sep 07 22:00:07 khem: no, not yet. Sep 07 22:00:28 probably another 8 weeks at least until it's habitable. Sep 07 22:00:39 hmm ok Sep 07 22:01:14 ground floor walls means literally that. no roof, no ceiling. Sep 07 22:01:23 also no plumbing or electricity yet. Sep 07 22:01:43 and no doors or windows :-} Sep 07 22:02:55 pb__: heh got it. but skeleton is ready Sep 07 22:03:35 he .."barebox" Sep 07 22:04:46 khem: thx for fixing klibc static renaming fallout Sep 07 22:04:59 I had no ti Sep 07 22:05:01 ant_mob: np Sep 07 22:05:41 time to test the initramfs imgs Sep 07 22:05:41 pb__: Sounds like you're making progress at least Sep 07 22:06:02 * RP is having trouble finding a builder to come and sort out the garage Sep 07 22:06:11 * RP is giving up on doing that himself Sep 07 22:07:06 khem: about udev-compat, it looks like 2.6.36 will soon obsolete the need for it Sep 07 22:07:28 (fingers crossed) Sep 07 22:07:45 ant_mob: yeah and also the scsi headers will be sorted out Sep 07 22:07:55 so .36 will be nice Sep 07 22:08:46 NameError: global name 'providerlog' is not defined Sep 07 22:08:51 hmmm latest bitbake Sep 07 22:09:29 kergoth: I guess its the logging branch stuff .. Sep 07 22:10:08 hmm, haven't seen that one Sep 07 22:10:10 * kergoth looks Sep 07 22:12:08 kergoth: http://pastebin.com/sJe8fhp7 Sep 07 22:12:18 has more info Sep 07 22:12:44 khem: fixed, thanks Sep 07 22:12:59 yeah lib/bb/taskdata.py doesnt seems to define it Sep 07 22:13:02 ok Sep 07 22:15:13 kergoth: thanks seems to work now Sep 07 22:15:33 cool, thanks for the report. downside to python, if you don't actually hit each codepath, errors can lurk :) Sep 07 22:15:47 yes Sep 07 22:15:51 code coverage is needed Sep 07 22:16:09 yeah Sep 07 22:16:15 kergoth: http://bugs.openembedded.net/attachment.cgi?id=13797 Sep 07 22:16:22 this seems to be lurking in master Sep 07 22:16:34 I mean http://bugs.openembedded.net/show_bug.cgi?id=5002 Sep 07 22:16:58 I guess we need that patch on master Sep 07 22:18:24 my comment from last year seemed to imply it was fixed in master, but differently. can you reproduce that? Sep 07 22:19:03 kergoth: yes Sep 07 22:20:00 kergoth: il seems for rdepend in depgraph["rdepends-pkg"][package]: Sep 07 22:20:21 gives letters and not strings Sep 07 22:20:22 was that supposed to be english? Sep 07 22:20:35 yes, i know what the behavior is, the patch implies it, and the bug explains it Sep 07 22:20:48 sorry this was line 443 of bitbake/lib/bb/cooker.py Sep 07 22:21:37 yes, but the bug was marked as solved and it seems to still exist and the patch doesn't seems to work anymore Sep 07 22:21:43 so re-open the thing Sep 07 22:21:45 it's not that hard Sep 07 22:21:50 if its broken, reopen the bug Sep 07 22:22:15 sorry, I was just trying to understand the thing before Sep 07 22:35:18 kergoth: while talking of bitbake, do you think this patch is ok ? : http://patchwork.openembedded.org/patch/2883/ Sep 07 22:36:27 i'd question the wisdom of using chdir. the caller might not be expecting its current directory to change. it'd be better to do that in the subprocess it spawns off to run the fetch command Sep 07 22:40:45 if I understand well hg.py, it's changing the dir just after runfetchcmd so is that a problem too change dir just before runfetchcmd ? Sep 07 22:42:20 for the bug report, it's assigned to bitbake-dev@lists.berlios.de as a default, should I add openembedded's dev list in copy ? Sep 07 22:50:23 ok bug submited : http://bugs.openembedded.net/show_bug.cgi?id=5471 Sep 07 23:00:34 when installing python-modules, python-dev is installed which pulls several dev packages. Is this an expected behaviour or should we remove python-dev from RDEPENDS_python-modules ? Sep 07 23:04:28 ericben, hi, quick thing: Sep 07 23:04:37 the wesnoth demo problem was an sdl bug Sep 07 23:04:40 hi GNUtoo|laptop Sep 07 23:04:59 ok this was demo effect :) Sep 07 23:05:24 Are you back in Italia ? Sep 07 23:06:07 not yet Sep 07 23:06:13 I'm using my vpn Sep 07 23:06:20 but I will very soon Sep 07 23:09:40 and as soon as I've the time....I'll try to port serial on an upstream kernel for my bug device Sep 07 23:09:54 btw you know wifi? Sep 07 23:10:34 there was a thread on the wireless mailing list about wl1271 and omap zoom Sep 07 23:10:43 it was about the virtual card detect thing Sep 07 23:11:14 if the card is wired, do I need a similar hack? Sep 07 23:11:24 for instance if it's powered by gpios Sep 07 23:12:19 the driver is libertas_sdio but it's not on the bug device Sep 07 23:12:57 on one board we also have libertas_sdio and no card detect as it's soldered on the board Sep 07 23:15:40 I mean virtual card-detect Sep 07 23:15:50 but maybe it's not needed Sep 07 23:17:17 I don't know what is this virtual card-detect Sep 07 23:18:45 ok Sep 07 23:19:04 it's because certain driver like wl1271 are powerd down by defualt Sep 07 23:19:16 on wl1251 it's the same Sep 07 23:19:35 if you don't force a card-detect I assume that it won't find the card Sep 07 23:21:03 I'll try without card detect then Sep 07 23:21:05 thanks a lot Sep 07 23:24:56 for libertas, we are using it as a module and that works fine. I must check but I think for this SD slot we don't register a card detect in the platform driver Sep 07 23:26:00 on i.MX we can have the card detect using DAT3 pin so you don't always need a GPIO Sep 07 23:26:25 do yoou know what is python-distutils ? is it only for development or also for the target ? Sep 07 23:27:31 ok thanks a lot! Sep 07 23:27:33 I think it's for compiling python stuff Sep 07 23:27:34 not sure Sep 07 23:27:34 http://www.mail-archive.com/linux-omap@vger.kernel.org/msg31698.html is the thread about virtual card detect btw Sep 07 23:28:11 http://www.spinics.net/lists/linux-omap/msg33767.html here theyy say this is no more needed Sep 07 23:28:59 ok Sep 07 23:30:08 thanks a lot Sep 07 23:31:37 I'll go to sleep very soon Sep 07 23:31:53 bye, and thanks a lot Sep 07 23:31:59 btw about distutils Sep 07 23:32:00 same here , bye Sep 07 23:32:12 you know better perl or python? Sep 07 23:32:18 none :-D Sep 07 23:32:31 ok Sep 07 23:32:36 both don't use makefiles Sep 07 23:32:41 but a perl or python script Sep 07 23:32:50 ok Sep 07 23:32:52 for "compiling" or installing a module Sep 07 23:33:08 so distutils is necessary for that to work Sep 07 23:33:13 from what I understand from t he doc : distutils is used to "build" the modules Sep 07 23:33:21 but is it also needed at runtine on the target ? Sep 07 23:33:37 http://docs.python.org/distutils/setupscript.html Sep 07 23:33:42 for the reference Sep 07 23:33:44 because in the end, that's a 6MB package only for python-distutils Sep 07 23:33:47 not at all Sep 07 23:33:49 yes I was also here Sep 07 23:33:55 unless you want to build the modules on target Sep 07 23:34:24 ok thanks Sep 07 23:35:22 kergoth: I see these huge messages with rm_work enabled Sep 07 23:35:32 I'll submit to patches : one to remove python-dev from python-modules and an other to remove python-distutils. python experts will judge but with these pathces my image went from 256MB to 56 ! Sep 07 23:36:33 kergoth: http://pastebin.com/6SBcSgVy Sep 07 23:37:02 no idea Sep 07 23:37:33 I did not enable packaged staging but still those messages are confusing Sep 07 23:37:47 may be rm_work brought those in Sep 07 23:39:07 do_qa_staging fails got packages and when I run bitbake then it works Sep 07 23:39:09 weird Sep 07 23:39:48 its for recipe called recipes/check/check_0.9.5.bb Sep 07 23:39:55 hmm, I have my own bbclass, which for testing purposes looks like this: http://pastebin.mozilla.org/784934 and I see that do_check is being run, but why does bb.fatal does not abort? I also tried bb.note and did not get any output either Sep 07 23:49:27 Jin^eLD: you didn't add do_check to EXPORT_FUNCTIONS Sep 07 23:49:35 so it's running a nonexistent empty task, if anything Sep 07 23:49:40 oh hmm Sep 07 23:49:52 I saw that some classes do not use that EXPORT_FUNCTIONS thing at all Sep 07 23:49:55 so I was kind of unsure Sep 07 23:50:03 i already told you what it does Sep 07 23:50:09 it handles the different naming Sep 07 23:50:21 it creates a wrapper for which runs classname_ Sep 07 23:50:30 if you don't want to use it, don't, but don't name your task addons_do_check Sep 07 23:50:32 just make it do_check Sep 07 23:50:54 ok I guess I was confused then, got it now Sep 07 23:52:35 yep now it works, thanks again ) I can finally go to sleep now :) Sep 07 23:52:52 ericben|away: the manifest is generated, iirc. see if you need to change the scripts that generate them, too Sep 07 23:52:55 np Sep 08 00:11:45 nite Sep 08 00:15:35 hello Sep 08 00:15:45 i keep getting this error when getting to python: Sep 08 00:15:48 ./libpython2.6.so: undefined reference to `_PyParser_Grammar' Sep 08 00:16:09 is this somehow available? Sep 08 00:16:48 jeez it's late Sep 08 00:16:55 is this behavior avoidable? :) Sep 08 00:17:28 Yeah, rebuild Sep 08 00:17:30 race condition :( Sep 08 00:17:51 you mean the parallel build between packages? or within the python package? Sep 08 00:19:11 well since no deps get pulled in, i assume its in-package race condition :( Sep 08 00:19:30 within python itself Sep 08 00:25:17 03Khem Raj  07org.openembedded.dev * r079ae22e05 10openembedded.git/recipes/midpath/midpath-backend-sdl_0.2+0.3rc2.bb: Sep 08 00:25:17 midpath-backend-sdl_0.2+0.3rc2.bb: Fix QA errors and remove do_staging Sep 08 00:25:17 Signed-off-by: Khem Raj Sep 08 00:25:26 03Khem Raj  07org.openembedded.dev * rb095bd2073 10openembedded.git/recipes/ffmpeg/ffmpeg_0.5.bb: Sep 08 00:25:27 ffmpeg_0.5.bb: On ARMv7 it needs to be compiled with -fPIC Sep 08 00:25:27 * building non-PIC shared libraries fails under ARMv7 Sep 08 00:25:27 * Also reported here Sep 08 00:25:27 * https://bugs.launchpad.net/ubuntu/+source/gcc-4.4/+bug/503448 Sep 08 00:25:27 Signed-off-by: Khem Raj Sep 08 00:25:32 03Khem Raj  07org.openembedded.dev * r24dc165e69 10openembedded.git/conf/bitbake.conf: Sep 08 00:25:32 bitbake.conf: Redefine STAGING_DIR_JAVA to point to target sysroot by default Sep 08 00:25:32 Signed-off-by: Khem Raj Sep 08 00:25:49 03Khem Raj  07org.openembedded.dev * r0c1a8195b2 10openembedded.git/recipes/classpathx/gnumail_1.1.2.bb: Sep 08 00:25:49 Revert "gnumail_1.1.2.bb: Point inetlib to java data dir in target staging area." Sep 08 00:25:49 This reverts commit 65c8f0a20ea3fe4d22c04d3a0c3a9f85d4ca1e53. Sep 08 00:25:49 Signed-off-by: Khem Raj Sep 08 00:31:00 lets see Sep 08 00:31:10 if I pull in non-legacy libtool-cross from poky Sep 08 00:31:16 nano on a few diff machines in one tmpdir enough? Sep 08 00:31:58 also making sure adding inherit cross is safe Sep 08 00:32:11 (so that pstaging mangling can happen, and then making the mangle big enough) Sep 08 00:32:47 ... or not since that happens w/ the sysroot post hook now Sep 08 00:32:51 Tartarus: no :) Sep 08 00:32:57 (that need for..) Sep 08 00:33:03 khem, heh, what else do you want baked? Sep 08 00:33:06 nano is autotools Sep 08 00:33:18 Tartarus: buiild few images Sep 08 00:33:25 like gcc and friends Sep 08 00:33:33 may be a native-sdk-image Sep 08 00:33:46 ok Sep 08 00:34:11 Tartarus: btw. are you bringing in all libtool changes or only parts Sep 08 00:36:03 just dropping legacy staging Sep 08 00:36:21 since that drops the need to have inherit cross in libtool-cross (or some gross kludges to packaged-staging) Sep 08 00:36:50 ty Tartarus; rebuilding python-only worked this time Sep 08 00:43:57 where i found the downloaded sources of OE ? Sep 08 00:44:07 unpacked... Sep 08 02:13:01 03Khem Raj  07org.openembedded.dev * rb389158b11 10openembedded.git/recipes/midpath/ (4 files in 2 dirs): Sep 08 02:13:01 midpath-core_0.2+0.3rc2.bb: Fix compilation error and QA HASH error on ARM Sep 08 02:13:01 * Remove legacy staging Sep 08 02:13:01 Signed-off-by: Khem Raj Sep 08 02:13:28 angelox_123: sources are in your build tree Sep 08 02:13:47 unpacked? Sep 08 02:13:51 angelox_123: $tmpdir/work Sep 08 02:13:56 angelox_123: yes Sep 08 02:14:12 it unpacks them Sep 08 02:14:14 when building Sep 08 02:14:26 if you do not enable rm_work it should be there Sep 08 02:16:23 ok thanks..i enabled.... Sep 08 02:16:32 if i disable i need re-build all? Sep 08 02:16:33 thanks Sep 08 02:19:15 khem you have a hung python process....chewing up CPU time. Sep 08 02:19:36 ka6sox-work: lemme check Sep 08 02:21:08 ka6sox-work: this is the parent process of the build task its not hung Sep 08 02:21:16 it will be running for long hours Sep 08 02:21:46 okay...I've just had them hang for 30+hrs before. Sep 08 02:22:18 I have started builds which should go for next two days continuous Sep 08 02:22:29 okay sounds good... Sep 08 02:22:49 I'll keep a loose watch on it to make sure it doesnt' OOM. Sep 08 02:23:03 ka6sox-work: ok thanks **** ENDING LOGGING AT Wed Sep 08 02:59:59 2010