**** BEGIN LOGGING AT Wed Jun 24 02:59:57 2009 Jun 24 04:52:55 does anyone here know if i can have two different versions of gcc Jun 24 04:53:19 userland was done w/ 4.1.1 but im trying to do a later kernel and i ran into the __weak problem Jun 24 04:53:25 rather the test for it Jun 24 04:53:36 i'd prefer not to recompile all of userland as well, if possible Jun 24 04:58:12 Yes, you can do a kernel-only gcc Jun 24 04:58:26 grep around in the machine conf dir for KERNEL_SUFFIX Jun 24 04:59:54 hmm Jun 24 05:00:16 well i did a dry run of 4.4.0 and it didnt look like it'd rm my current (4.1.1) from staging, but it did Jun 24 05:00:28 i guess switching back would be quite easy Jun 24 05:00:41 i see what you're saying though Jun 24 05:00:51 it can automate the switch specifically for the kernel compile..? Jun 24 05:01:08 Yes Jun 24 05:01:21 It'll give you a gcc used just for the kernel Jun 24 05:01:32 it does require you to have, roughly, gcc-cross-kernel-4.4.0_4.4.0.bb Jun 24 05:01:36 if you wanted to jump to 4.4.0 Jun 24 05:01:48 There's a few example recipes for the gcc, and it's really easy Jun 24 05:01:52 (to create a new one) Jun 24 05:02:02 cool thanks Jun 24 05:02:23 i did a rough 'port' of the dht-walnut machine type to 2.6.30 Jun 24 05:02:33 it was at like 2.6.23 with .20 preferred i think Jun 24 05:02:38 Yeah Jun 24 05:03:02 Could you do us a huge favor? Try building for angstrom, but modify the conf file to use the normal gcc version, and report back if it works or not Jun 24 05:03:30 angstrom drops ppc way back, and things build fine w/ newer gccs, but no one has HW around to test with right now and see if things boot still Jun 24 05:03:40 rather than 4.1.1? Jun 24 05:03:43 yeah Jun 24 05:03:43 yea i can try that Jun 24 05:03:47 thanks! Jun 24 05:03:55 theres a couple packages i've fixed Jun 24 05:03:59 slang was one of them Jun 24 05:04:05 that wasnt gcc related though Jun 24 05:04:14 makefile.in missing $(LDFLAGS) Jun 24 05:04:29 post that too please :) Jun 24 05:05:13 hey, real quick, do you know if anything has been tested on native sparc/solaris? Jun 24 05:05:29 ie. w/ sun c compilers Jun 24 05:06:00 Lets see, some time back someone was trying to use solaris as a host, hit a problem or two and didn't have the skills to further debug Jun 24 05:07:34 hmm Jun 24 05:07:41 i might put gentoo on it or something Jun 24 05:08:02 itd work nicely as a build machine if i threw a 146g fc-al in it Jun 24 05:09:29 bbt, good luck Jun 24 05:10:09 thanks. once i achieve success with the newer builds/compilers/etc, is there a preferred format? Jun 24 05:10:25 should i just send a diff of the whole openembedded git repo? Jun 24 05:50:58 03Klaus Kurzmann  07shr/import * r5dffee7f99 10openembedded.git/conf/distro/include/shr-autorev-unstable.inc: Jun 24 05:50:58 shr-autorev-unstable.inc: set fixed rev for intone, Jun 24 05:50:58 because current AUTOREV does not build Jun 24 05:50:58 Signed-off-by: Klaus Kurzmann Jun 24 06:48:21 good morning Jun 24 06:49:00 morning Jun 24 06:54:30 what is the simplest way to build for a different machine in the same tree ? Jun 24 06:54:48 just MACHINE=whatever bitbake whatever ? Jun 24 06:55:59 ah actually already found the answer to my own Q, that is indeed the way, building now Jun 24 06:58:08 eFfeM: fine :-) Jun 24 07:02:49 I have problems building x11-native ... It end up this a floating point exception.while executing ../src/util/makekeys. Any indeas? Jun 24 07:08:22 Tartarus : i added the gcc-cross-kernel for 4.4.0, with a slightly modified do_stage() (no libgcc install, seems deprecated?), and removed the old 4.4.0, reinstalled 4.1.1 Jun 24 07:08:37 the kernel compile is using the gcc-cross-kernel now Jun 24 07:09:08 KERNEL_CCSUFFIX = "-4.4.0" Jun 24 07:09:08 PREFERRED_PROVIDER_virtual/${TARGET_PREFIX}gcc-4.4.0 = "gcc-cross-kernel-4.4.0" Jun 24 08:26:18 pb_: are you awake, yet? Jun 24 08:42:15 pb_: can we enable bluez-libs-4.x? Jun 24 08:42:30 good morning Jun 24 08:55:06 03Rolf Leggewie  07org.openembedded.dev * r6bd858afbb 10openembedded.git/recipes/openssl/openssl.inc: Jun 24 08:55:06 openssl: add support for compilation on Via C3 host. Jun 24 08:55:06 Many thanks to Phil Blundell. Jun 24 09:03:26 Laibsch: hello Jun 24 09:03:32 hi Jun 24 09:03:37 hello all, anyone here is playing with the AT91SAM9260 CPU ? Jun 24 09:03:42 Laibsch: yes, but there isn't yet any corresponding bluez-apps that works. Jun 24 09:03:53 I might see if I can fix that today if I get a spare moment. Jun 24 09:04:00 which isn't needed when only using -libs, right? Jun 24 09:04:14 I just updated you bluez-libs recipe to 4.42 Jun 24 09:04:14 hi pb and laibsch Jun 24 09:04:15 CruX|: 9263, sorry Jun 24 09:04:27 pb_: and unified it Jun 24 09:05:16 Laibsch: right, it's not strictly necessary. it just means that anybody wanting the apps will end up building the libs twice. Jun 24 09:05:24 okay, very good Jun 24 09:05:25 hi woglinde Jun 24 09:05:34 did you prepare the recipes/bluez/bluez-libs/avinfo-link.patch Jun 24 09:05:35 ? Jun 24 09:05:42 what is its upstream status? Jun 24 09:05:56 no, I don't think that was one of mine. Jun 24 09:06:03 let me check the git log Jun 24 09:06:08 your recipe is the only one using it Jun 24 09:06:13 ah, hm Jun 24 09:06:16 this bluez situation is killing me Jun 24 09:06:33 oh yes, right, that is one of mine Jun 24 09:06:33 I tried hard to build an image without any bluez, but ultimately failed Jun 24 09:06:45 applicable to upstream? Jun 24 09:06:49 I don't think it's needed upstream, it's just a consequence of splitting the package into -libs and -apps Jun 24 09:07:00 but, equally, it wouldn't do any harm upstream either. I might send it to them and see what they say. Jun 24 09:07:03 mckoan: i need driver for rs485, is it enough to only enable rs485 mode in usart mode registers ? Jun 24 09:07:03 OK, I'll update the patch with that information Jun 24 09:07:15 mckoan: and use it as a rs232 ? Jun 24 09:07:29 yeah, it is a bit sad that you can't build an image without any bluez. there are plenty of non-bluetooth-capable machines out there. Jun 24 09:07:48 *sigh* why this crap works under cygwinshell but not from explorer Jun 24 09:14:06 03Rolf Leggewie  07org.openembedded.dev * r851c3641b5 10openembedded.git/recipes/bluez/ (bluez-libs_4.42.bb bluez-libs_4.40.bb): bluez: update bluez-libs to 4.42. unify. Jun 24 09:17:24 03Rolf Leggewie  07org.openembedded.dev * r079c832e00 10openembedded.git/recipes/bluez/bluez-libs/avinfo-link.patch: bluez: document upstream status of avinfo-link.patch as not applicable upstream. Jun 24 09:20:45 pb_: would the bluez bb and your bluez-libs bb stpe on each other's toes? Jun 24 09:21:03 otherwise, just let the bluez bb (R)PROVIDE bluez-apps Jun 24 09:21:05 and be done Jun 24 09:21:36 Laibsch: they will, but not in any way that really matters. Jun 24 09:21:53 in what way will that be? Jun 24 09:21:59 I'm still stuck with Jun 24 09:22:15 well, they will overwrite each other's libraries in staging, but the libraries will be basically identical so I doubt you will notice the difference. Jun 24 09:22:32 if bluez and bluez-libs are the same version, the libraries should be precisely identical Jun 24 09:23:10 !oebug 5133 Jun 24 09:23:11 * * Bug 5133, Status: CONFIRMED, Created: 2009-05-28 19:26 Jun 24 09:23:12 * * matt(AT)genesi-usa.com: do_rootfs hangs when opkg-cl asks for input Jun 24 09:23:13 * * http://bugs.openembedded.org/show_bug.cgi?id=5133 Jun 24 09:28:40 CruX|: is is enough to play with registers from userspace if you want Jun 24 09:36:11 pb_: what about bluez-utils, don't we need a package for that, too? Jun 24 09:36:55 Laibsch: ah, sorry, bluez-apps and bluez-utils are the same thing. just me mis-remembering the name. Jun 24 09:37:11 ok Jun 24 09:37:26 I'm in even deeper sh*t than I thought Jun 24 09:37:38 There is a dependency on -utils somewhere Jun 24 09:37:58 oh dear Jun 24 09:52:14 hello, where is the file to put some bash commands or path variable exports in angstrom? (like .bash_profile file) Jun 24 09:52:38 could someone help me? Jun 24 10:10:11 03Rolf Leggewie  07org.openembedded.dev * rcb429c4b7a 10openembedded.git/recipes/obexftp/obexftp_0.23.bb: obexftp: version 0.23 compiles fine against bluez-libs 3.x. Relax DEPENDS accordingly. Jun 24 10:10:49 I have problems building libx11-native ... It end up this a floating point exception.while executing ../src/util/makekeys. Any indeas? Jun 24 10:17:02 sgh: would be useful to pastebin the error ;-) Jun 24 10:18:31 mckoan: http://pastebin.ca/1472329 Jun 24 10:18:32 pb_: FYI http://paste.debian.net/40085/ Jun 24 10:18:40 maybe there is something important in there Jun 24 10:20:09 sgh: which triplet distro/machine/image ? Jun 24 10:21:09 sgh: which branch ? Jun 24 10:21:14 mckoan: angstrom_2008.1 but libx11-native from within bitbake -i Jun 24 10:21:21 mckoan: master i guess Jun 24 10:22:15 sgh: better ask to #angstrom too Jun 24 10:22:47 Laibsch: which package was that? Jun 24 10:23:04 the bluez-libs 4.42 Jun 24 10:23:04 sgh: what's the architecture of your build host? Jun 24 10:23:16 pb_: x86 Jun 24 10:23:24 Laibsch: ah right, that's fine then. we don't want those things to be packaged. Jun 24 10:23:36 I'm not quite sure why they're getting built in the first place, but it's harmless to not ship them. Jun 24 10:23:42 ok Jun 24 10:23:52 sgh: ah, hm. I've seen that on amd64 but I thought it worked on x86. Jun 24 10:24:03 I guess it must be broken on all architectures then. Jun 24 10:24:25 pb_: I wanted to try to gdb makesys but I don't know how the debug an application that need input from stdin :( Jun 24 10:24:40 sgh: you can run it with redirects at the gdb prompt. Jun 24 10:24:48 gdb) run < /path/to/file Jun 24 10:25:05 pb_: I'll try ... Jun 24 10:25:35 iirc, the hash table calculation is going wrong somehow and this causes it to crash with divide by zero. Jun 24 10:25:50 I didn't have the patience to debug it properly before, I just turned off makekeys in my build. Jun 24 10:25:56 it'd be nice to get it fixed though Jun 24 10:30:21 pb_: what is the purpose of makesys. The reason for me compiling it natively is that I'm compiling VTK for the target and vtk needs to compile and execute some stuff. That stuff is compiled with the cross-compiler and therefore not runable on my buildhost :(. So I wabt to natively compile VTK and its dependencies and the move the files into my targets staging area. Jun 24 10:31:25 pb_: If I don't need it - how do I turn it off then? Jun 24 10:31:36 ah, makesys? I thought we were talking about makekeys. I guess I read your earlier comment wrong. Jun 24 10:31:40 I don't know what makesys is. Jun 24 10:32:23 pb_: sorry your right ... it is makekeys ... my eyes are failing I guess :) Jun 24 10:33:02 ah, okay. it generates something to do with the keymaps. I think you can just patch it out of the makefile, it doesn't seem to be necessary for correct operation. both diet-x11 and regular (target) libx11 disable it, you could look at their .bb files for clues. Jun 24 10:33:42 libx11-trim does leave it enabled for some reason. I don't quite know what that recipe is good for. Jun 24 10:36:21 pb_: thanks ... I will check it out. Jun 24 10:37:29 hmm is there a xvncserver on oe ? Xvnc equivalent ? Jun 24 10:37:48 * BusError is building a sheevaplug distro Jun 24 10:54:33 re Jun 24 11:16:36 hi otavio Jun 24 11:20:37 hi gnutoo Jun 24 11:20:42 woglinde, hi Jun 24 11:21:58 woglinde, someone duplicated my work...that means that the same work was done 2 times...this isn't very optimal...that's why I have to talk to otavio Jun 24 11:22:25 woglinde, do you go to linuxtag? Jun 24 11:33:52 gnutoo yes on friday and satuarday Jun 24 11:34:06 ok Jun 24 11:34:20 unfortunately I can't go to linuxtag...I had exams Jun 24 11:34:37 maybee next year Jun 24 11:36:24 gah, i'm still struggling with the same error message Jun 24 11:36:49 arm-angstrom-linux-gnueabi-ld: ERROR: Source object .../build/tmp/cross/arm/lib/gcc/arm-angstrom-linux-gnueabi/4.3.3/libgcc.a(_udivdi3.o) has EABI version 4, but target u-boot has EABI version 0 Jun 24 11:37:33 i tried those eabi fixes that are under openembedded/recipes/u-boot/files but those are for some different arch Jun 24 12:12:17 anybody use git-sendmail with gmail imap? I need to know how to configure it Jun 24 12:18:15 mckoan: imap? maybe smtp?:) Jun 24 12:18:49 imap is only for reading Jun 24 12:19:52 mckoan: http://pastebin.com/m7ee39b9d Jun 24 12:19:59 here is my settings Jun 24 12:20:35 morning Jun 24 12:21:40 Gnutoo: I have tested your boost patches (1.36 only), I had to adapt them a little bit, but now it works (tested at running too) Jun 24 12:21:47 hey booxter Jun 24 12:22:01 Longfield, wow thanks Jun 24 12:22:27 Longfield: morning Jun 24 12:26:25 hi, Jun 24 12:28:45 I used the newest openembedded to build Angstrom, when it was running on my B5 beagleboard, i can't use mouse && keyboard. Can you give me some cute? Jun 24 12:29:03 anyone working on the sheevaplug port ? Jun 24 12:29:20 cute = clue, thank Jun 24 12:35:23 favor: how do you connect your mouse/keyboard to BB? Jun 24 12:35:41 via a power usb hub, Jun 24 12:36:27 my mouse/keyboard works well before I moved to newer kernel. Jun 24 12:37:43 my newer kernel version is 2.6.29-r37, I builded it in oe. Jun 24 12:38:14 03Klaus Kurzmann  07shr/import * rdf690bcd0c 10openembedded.git/recipes/connman/connman_git.bb: Jun 24 12:38:14 connman_git.bb: enable the plugins via EXTRA_OECONF Jun 24 12:38:14 Signed-off-by: Klaus Kurzmann Jun 24 12:38:15 03Klaus Kurzmann  07shr/import * r59ebc1abf3 10openembedded.git/conf/distro/include/shr-autorev.inc: Jun 24 12:38:15 shr-autorev.inc: actually really build connman from git Jun 24 12:38:19 Signed-off-by: Klaus Kurzmann Jun 24 12:40:20 CIA-3: excuse me? Jun 24 12:41:26 booxter: I use imap with gmail :-( Jun 24 12:41:39 CIA-3: are you talking to me? Jun 24 12:41:55 booxter: I'll try, thanks for the hint Jun 24 12:41:56 mckoan: imap is not used for sending anyway Jun 24 12:41:57 cia is just a bout Jun 24 12:42:00 aeh bot Jun 24 12:42:10 favor did you check that usb stuff is enabled Jun 24 12:42:14 mckoan: you use smtp for both pop3 and imap4 Jun 24 12:42:14 in the kernel config? Jun 24 12:42:43 booxter: you're right, so it must work :-) Jun 24 12:42:51 hi, I tried several things for adding the symbols to uclibc and it doesn't work Jun 24 12:43:06 gnutoo? Jun 24 12:43:08 woglinde: I just know to use 'dmesg | grep -i usb' when boot in to system console. Jun 24 12:43:25 woglinde, I tried: Jun 24 12:43:53 gnutoo boost agaon? Jun 24 12:43:59 * DEBUG_BUILD = "1" INHIBIT_PACKAGE_STRIP = "1" LDFLAGS =+ "-g -ggdb2" CFLAGS =+ "-Os -pipe -mips32 -mtune=mips32 -funit-at-a-time" Jun 24 12:44:02 woglinde, no Jun 24 12:44:29 woglinde, mips rootfs where every program that use shared libs segfault Jun 24 12:45:02 woglinde, it's the fault of ld-uClibc.so.0 Jun 24 12:45:05 woglinde: I checked the kernel config settings, I can see some usb stuff in input driver is enabled. Jun 24 12:49:09 gutoo hm did you tried with -O2 instead of -Os? Jun 24 12:49:15 I'll try Jun 24 12:49:36 is O2 the default? Jun 24 12:51:30 thanks Jun 24 12:54:49 guys :) Jun 24 12:55:08 is it really necessary to call oe_sha256sum and not use the native package? Jun 24 12:55:53 the same for shasum (sha1sum for every other linux distro since modern times). Just trying to cut down on the custom tools floating around here since Fedora had 100% of the pre-cross-toolchain-build dependencies installed by default Jun 24 12:57:18 bye toll later Jun 24 12:58:26 mmm...it's in bitbake.conf Jun 24 13:05:12 booxter: if I do git send-email I get this message back Need MIME::Base64 and Authen::SASL todo auth at /usr/bin/git-send-email line 775, line 1. Jun 24 13:05:34 any idea? Jun 24 13:05:43 mckoan: maybe you don't have these perl (?) modules? Jun 24 13:08:41 is there a way to bypass the "ERROR: QA Issue with tunctl: No GNU_HASH in the elf binary:" -- I keep running into these Jun 24 13:09:36 yeah Jun 24 13:11:07 there is some INSANE_package-virtual thing you can set.. I don't recall it exactly but check insane.bbclass for a hint Jun 24 13:11:28 but it would be better if you fixed the package to compile properly with gnu-hash style hashes Jun 24 13:16:47 booxter: I added those packages but still remain an error: Command unknown: 'AUTH' at /usr/bin/git-send-email line 775, line 1. Jun 24 13:17:08 booxter: I hope to do not annoy you. Did you ever seen this error? Jun 24 13:18:41 mckoan: no :) Jun 24 13:19:03 I think you should look at the problematic line to get the reason of the problem Jun 24 13:19:36 mckoan: btw your git is maybe quite old Jun 24 13:19:43 git --version ? Jun 24 13:39:32 booxter: sorry phone call. git version 1.5.6.5 Jun 24 13:39:59 mckoan: you should really update your git version... Jun 24 13:43:41 uh it is the debian lenny default one Jun 24 13:44:17 booxter: what is your version? Jun 24 13:44:57 mckoan: 1.6.3.3 Jun 24 13:45:20 mckoan: don't use distro git (at least debian), build it from git Jun 24 13:45:35 mckoan: (or use gentoo) :) Jun 24 13:50:14 NekoXP: yeah, that oe_sha256sum thing is a bit weird. I don't know why it's like that; I can't think of any reason why the host sha256sum wouldn't be fine. Jun 24 14:06:29 booxter: git version 1.6.3.3 worked! thank you very much :-D Jun 24 14:09:38 mckoan: I got some problems with send-email when using ubuntu version too. Build from git helped me too. Jun 24 14:09:58 there are a small set of low-hanging-fruit patches on there : git://repo.or.cz/openembedded/mini2440.git for-upstream-dev --- } have tested them for a few weeks on the -stable fork I have, but I also rebased them on -dev and they all work (on arm). anyone care to pull ? Jun 24 14:10:38 webgit view : http://repo.or.cz/w/openembedded/mini2440.git?a=shortlog;h=refs/heads/for-upstream-dev Jun 24 15:16:26 what is the recipe flag to prevent make -j ? Jun 24 15:17:07 BusError: PARALLEL_MAKE = "" Jun 24 15:17:13 thanks Jun 24 15:38:12 mmm...ld-uClibc.so.0 doesn't include debug info when other part of uclibc do... Jun 24 15:38:33 I'll have to debug at asm level Jun 24 15:38:49 I'll hope I'll remember enough mips assembly Jun 24 15:58:04 can anyone explain the whole libusb/libusb-compat vs libusb1 situation? Jun 24 16:07:19 Do you mean the upstream situation, or the way the recipes are setup/handled in OE? Jun 24 16:10:27 both? :) Jun 24 16:12:57 As I understand, the upstream folks created libusb1 with libusb-compat to replace libusb, but some apps rely on internal knowlege of the old libusb, and will not work with libusb-compat. Both libusb and libusb-compat provide libusb.a, so you can only have one on any given host. Jun 24 16:13:29 Problematic apps included (and may still include) gnuradio, dfu-util Jun 24 16:14:41 In OE, various recipes depend on libusb or libusb-compat, so unless a distro is very careful, they will build both -- the results are unpredictable, as things depend on build order at that point. Jun 24 16:15:14 ugh, thats ugly Jun 24 16:15:28 * kergoth noticed he had both in his overlay, and was concerned Jun 24 16:15:36 (as i should be, clearly) Jun 24 16:15:54 For SlugOS, I "resolved" the problem by creating a dummy libusb recipe with version number "0" that just depends on libusb-compat -- so setting PREFERRED_VERSION_libusb to that version will not actually result in the old libusb being built. Jun 24 16:16:20 But it does preclude some apps from building, the ones that don't like the new libusb-compat. Jun 24 16:16:53 I guess that technique qualifies as a "horrible hack", but it has worked very well for SlugOS... Jun 24 16:16:57 i should really try to implement a prototype of private staging areas Jun 24 16:17:08 That would help somewhat. Jun 24 16:17:34 I seem to recall that there is a run-time .so conflict as well, but perhaps not. Jun 24 16:17:52 ah. well, one issue at a time :) Jun 24 16:18:13 But even so, a private staging area would allow static linking, which would be a reasonable solution for the old apps. Jun 24 16:19:08 that's true Jun 24 16:19:41 is there a way to alter COMPATIBLE_MACHINE vars for packages globally (outside the recipe)? as COMPATIBLE_MACHINE is typically assigned with a hard '=' I think not unless there is some global conf files that are included 'after' the recipe Jun 24 16:20:06 mwester: I had once wondered if we could contstruct staging on the fly from DEPENDS using some form of unionfs Jun 24 16:20:11 COMPATIBLE_MACHINE_pn-somerecipe, tharvey Jun 24 16:20:17 mwester: that would clear a lot of these issues Jun 24 16:20:21 XorA: that's the sort of thing we're talking about Jun 24 16:20:38 kergoth, ah... so that would take precedence - thanks! Jun 24 16:20:50 kergoth: I did actually start looking at it, but I couldnt find a reliable way to do it in userspace without root access Jun 24 16:20:51 better actually making separate staging areas though, linking or overlaying on a global one still has conflicts if multiple versions of something are built, for example Jun 24 16:21:11 pb suggested just ipkg installing the dep pstage packages into a private staging area for each recipe ,and that seems viable Jun 24 16:22:36 I've struck by the fact that every "major-league" SCM/build-system I've worked with inevitably comes to the conclusion that they need a "smarter" filesystem. There is so much that can be done when you can ask questions from the FS about what *really* happened. Jun 24 16:22:59 sounds like you want clearcase ;) Jun 24 16:23:08 * XorA hides Jun 24 16:23:28 No, I'm past that! But ClearCase work paid my mortgage and put my kids through school. Jun 24 16:23:30 clearcase, particularly w/ clearmake really is pretty nifty in some ways, and frightening in others, of course Jun 24 16:23:35 ah :) Jun 24 16:23:45 okay, the svn fetcher requiring module= is just fucking stupid Jun 24 16:23:48 * kergoth adds to his list Jun 24 16:23:51 Most recently, I worked with the product from ElectricCloud -- awesome build tool. Jun 24 16:24:04 ah, nice, i read about that one a bit Jun 24 17:00:18 g'day kergoth Jun 24 17:00:25 hey pb__ Jun 24 17:00:26 how's it going? Jun 24 17:00:46 it's my birthday today, so feeling old, but pretty good apart from that Jun 24 17:01:16 how old is old? :D Jun 24 17:01:27 oh, and I'm hating xorg-xserver for not supporting multiple cards anymore Jun 24 17:01:59 I stupidly upgraded the x on my newly-repaired desktop machine and, bam, back to one screen again. suck. Jun 24 17:02:02 Neko: 33 Jun 24 17:02:09 hmm that's not too bad :D Jun 24 17:02:11 ah, happy birhtday Jun 24 17:02:24 happy birthday. at least you're not 63 :D Jun 24 17:02:28 heh, true enough Jun 24 17:02:34 heh, i did that, somehow screwed up the x on my workstation, no longe rhave the patience to fuck with that stuff Jun 24 17:02:38 i just reinstalled jaunty :P Jun 24 17:03:01 yeah, it's annoying. seems that something has changed in the xserver to mean that you can only have one video card nowadays Jun 24 17:03:17 so, if you want dualhead, it has to be one card with multiple crtcs. Jun 24 17:03:23 grah vista is going batshit >.< Jun 24 17:03:34 (which I do have on my desktop at the office, but that doesn't seem to work either for undisclosed reasons) Jun 24 17:03:42 btw guys is there any point in reporting a ton of bugs regarding ASSUME_NATIVE for native packages? Jun 24 17:03:43 so, all in all, frustrationsville. Jun 24 17:04:26 for instance using the system automake doesn't work because @INC doesn't include the native /usr/share/automake/ directory so it misses the autom4te/whatever.pm it looks for Jun 24 17:05:01 shasum is sha1sum on modern systems (after Fedora 2008 :) and oe_sha256sum is kind of retarded. Jun 24 17:05:29 Neko: you really dont want to use the native autoconf/automake/libtool, we patch them. Jun 24 17:05:30 also for some reason if I have ccache installed, quilt-native does not build anyway (compiler cannot create executables), so I have to use the native quilt Jun 24 17:05:37 ok Jun 24 17:05:50 Neko: someone hsould really go through and document which ones are unmodified, and thus safe to assume the host's tools will work.. Jun 24 17:07:01 pretty sure we can stop using shasum anyway Jun 24 17:07:04 python has an sha module Jun 24 17:07:08 heh Jun 24 17:07:42 the idea is, as I have had this plan for a while now, I want to put up a pretty sweet little build farm, a couple Xeon 5500 servers with 8-16 cores each, running OpenEmbedded, but to get the best performance and not waste developer time (as fast as they would be) I would much prefer to have the compiler toolchains "preinstalled" (we only build for ppc and arm9 and cortex-a8) so when someone starts from scratch they don't run through 890 tasks of download/configure Jun 24 17:07:42 /build/stage etc. Jun 24 17:08:05 well, you can always leverage packaged staging packages Jun 24 17:08:20 the same goes for all the little utils like quilt. I guess though it would not be too big a deal to make native rpms of the patched OE stuff either. Jun 24 17:08:58 if you use packaged staging packages, you wouldnt have to mess with external toolchains and asssume provided and stuff Jun 24 17:09:20 how do I make OE pick that up? Jun 24 17:09:35 if someone gets an account and gets started, their stamps dir will be empty and OE will churn through regardless Jun 24 17:10:01 it sounds like you've never used packaged staging Jun 24 17:10:06 the pstage packages include the stamps Jun 24 17:10:17 you're right I never used it :D Jun 24 17:10:31 if its available, it unpacks it into tmp, into staging and stamps, and uses that to produce the packages and all Jun 24 17:11:04 oh.. question #297 would be.. why is it such a big problem to build using an x64 host? :( Jun 24 17:11:08 by default it puts the packages into tmp, but if you move them out... Jun 24 17:11:10 absolutely nothing works Jun 24 17:11:14 i build on x64 hosts every day Jun 24 17:11:20 in fact, all my machines are Jun 24 17:11:29 I get random segfaults, python errors... Jun 24 17:11:42 sounds like you have a problem with your desktop Jun 24 17:11:43 packages do not build, toolchains do not build Jun 24 17:12:12 what I considered was.. Fedora royally f**ked their 64-bit testing regime Jun 24 17:12:16 INHERIT += "packaged-staging" to use it, change the bbclass to use ?= for PSTAGE_PKG and then you can point it to somewhere outside of tmp Jun 24 17:12:28 hmm okay Jun 24 17:12:55 I am considering using a 32-bit OS on there just so I can leverage psyco Jun 24 17:13:10 64-bit python is fast but it is nowhere near as good as the same box in 32-bit with psyco Jun 24 17:13:38 03Chris Larson  07org.openembedded.dev * r614b2f8ee4 10openembedded.git/classes/packaged-staging.bbclass: Jun 24 17:13:38 packaged-staging.bbclass: use ?= for PSTAGE_PKG. Jun 24 17:13:38 Signed-off-by: Chris Larson Jun 24 17:14:13 huh kergoth is that a coincidence or you nudging the bot? :D Jun 24 17:14:27 neither, i just got sick of it not using it, and pushed the change Jun 24 17:14:46 is this going into stable/2009 or do I have to move to dev? Jun 24 17:15:13 * kergoth knows nothing about stable, ask a stable developer Jun 24 17:15:15 (not that I can't hack it but, I hate having an unclean OE tree, makes git pulls a nightmare) Jun 24 17:15:47 I already had to bitchslap autoconf.bbclass or so because --exclude=autopoint is not a valid option (!?!) Jun 24 17:16:20 its valid for our autoconf Jun 24 17:16:29 I was using your autoconf Jun 24 17:16:30 thats why i said you shouldnt be using the one on your machine Jun 24 17:16:32 no, you weren't Jun 24 17:16:40 the autoconf-native in OpenEmbedded has the patches for that option Jun 24 17:16:44 this was before I was playing with assume_provided Jun 24 17:17:01 well, dunno what to tell you. the OpenEmbedded autoconf-native supports that Jun 24 17:17:01 it's one of the things that made me want to try it :D Jun 24 17:19:00 are there any docs on packaged-staging? Jun 24 17:22:38 i doubt it. i doubt any are necessary Jun 24 17:25:00 question #8900 - if I use inherit icecc even though native is blocked as a class, every time it builkds it is giving me which: no gcc in (null)) and then finally /oe/icecc-env.sh: line 83: echo: write error: Broken pipe Jun 24 17:25:23 it's all working but it seems it's not so much blocked as intentionally passed something broken so it doesn't get used Jun 24 17:42:24 pff and as usual poky is totally broken and doesn't build Jun 24 17:42:34 i thought intel buying those guys would stop the gentooism. Jun 24 18:14:44 03Chris Larson  07org.openembedded.dev * r151de3a580 10openembedded.git/ (classes/base.bbclass conf/bitbake.conf): Jun 24 18:14:44 bitbake.conf, base.bbclass: Changed build config summary header to be a variable. Jun 24 18:14:44 Default in bitbake.conf for BUILDCFG_HEADER is "Build Configuration". This Jun 24 18:14:44 lets distributions make their builds ever so slightly prettier :) Jun 24 18:14:44 Signed-off-by: Chris Larson Jun 24 18:14:48 Acked-by: Marcin Juszkiewicz Jun 24 18:31:33 what are you transforming into? Jun 24 18:31:35 good evening Jun 24 18:34:41 maybe in a bug device? Jun 24 18:35:16 it's seems a transformable hardware Jun 24 18:36:02 mickey|linuxtag: how's linuxtag? Jun 24 18:40:33 * khem rejoiced old days in germany Jun 24 19:04:21 re Jun 24 19:10:36 woglinde, hi is it normal that woglinde have no debug info but the other part of uclibc do? Jun 24 19:10:53 gnutoo what? Jun 24 19:11:00 I am not a ki Jun 24 19:11:01 woglinde, oops Jun 24 19:11:07 woglinde,lol sorry Jun 24 19:11:28 I meant ld-uClibc.so.0 Jun 24 19:11:48 Gnutoo: yes Jun 24 19:11:50 I copy pasted the lib name than your name and you bet what happend... Jun 24 19:12:02 gnutoo khem is back Jun 24 19:12:11 so you can ask him about uclibc too Jun 24 19:12:16 so I bet I've to debug the assembly part... Jun 24 19:12:17 he knows more than me Jun 24 19:12:18 Gnutoo: ld overrides the CFLAGS Jun 24 19:12:23 ok Jun 24 19:12:57 I'll look at bitbake.conf Jun 24 19:13:35 so DEBUG_OPTIMIZATION = "-O -fno-omit-frame-pointer -g -ggdb3" is CFLAGS? Jun 24 19:13:47 s/?// Jun 24 19:14:08 Gnutoo: yes should be if you use them Jun 24 19:14:15 ok Jun 24 19:14:29 gnutoo hm you have to watch out what uclibc really uses Jun 24 19:14:35 I just saw lower in bitbake.conf that it is Jun 24 19:14:42 because the normal gcc call is hide Jun 24 19:14:48 you dont really see the options Jun 24 19:14:49 ok Jun 24 19:15:12 03Tom Rini  07org.openembedded.dev * r33615f3f32 10openembedded.git/ (5 files in 3 dirs): alsa: Add alsa-lib/plugins/utils 1.0.19 Jun 24 19:15:13 03Tom Rini  07org.openembedded.dev * rc90998b875 10openembedded.git/recipes/pulseaudio/ (6 files): Jun 24 19:15:13 pulseaudio: Switch to INC_PR, add openssl to DEPENDS. Jun 24 19:15:13 Also remove dbus from the version specific DEPENDS lines as it's in the inc file Jun 24 19:15:16 khem V=1 or something for the real gcc calls? Jun 24 19:15:17 03Tom Rini  07org.openembedded.dev * r0b3740da98 10openembedded.git/recipes/gdb/ (gdbserver.inc gdbserver_6.7.1.bb gdbserver_6.8.bb): Jun 24 19:15:20 gdbserver: Switch to INC_PR, add LICENSE, only build gdbserver, drop DEPENDS Jun 24 19:15:22 To build just gdbserver, we don't need ncurses or readline, so drop the DEPENDS Jun 24 19:15:24 and pass --without- for configure. Also, we don't need to build anything more Jun 24 19:15:26 than 'gdbserver'. Jun 24 19:15:28 03Tom Rini  07org.openembedded.dev * rcdfc5e8896 10openembedded.git/recipes/meta/meta-toolchain.bb: meta-toolchain: Oops, we want layout_libdir not layout_base_libdir for opkg Jun 24 19:15:34 I'll modify bitbake.conf Jun 24 19:16:14 woglinde: yes V=1 you can see the gcc commandline Jun 24 19:16:26 I need to buy new battery for my MBP Jun 24 19:16:36 mbp? Jun 24 19:16:42 macbook pro Jun 24 19:16:58 it runs like 10 mins these days Jun 24 19:17:16 hehe Jun 24 19:17:53 Its 3 yrs old now Jun 24 19:23:52 hello Jun 24 19:24:20 I am doing a kernel recipe in it, I need to create some files in the kernel tree Jun 24 19:24:41 I started crteing diffs and trying to patch the original sources Jun 24 19:25:07 lisandropm look at quilt Jun 24 19:25:08 but bitbake fails complaining that the file to patch doesn't exists (which is the idea, I need to create it :-) ) Jun 24 19:25:12 its easy to use Jun 24 19:25:17 woglinde: quilt? ok! Jun 24 19:25:27 and provides you easy patches Jun 24 19:25:34 short workflow Jun 24 19:25:42 quilt new mypatch Jun 24 19:26:03 quilt edit filexy.c Jun 24 19:26:09 after editing Jun 24 19:26:33 quilte refresh Jun 24 19:26:42 then the patch is under patches Jun 24 19:26:46 in oe Jun 24 19:27:11 excellent, I'll stw for it Jun 24 19:28:03 put the patch under recipes/linux/linux-versionumber/yourmachine/ Jun 24 19:29:56 I was trying creating a new recipe (hardware not under oe... yet :-) ) Jun 24 19:30:06 but basing myself on it :-) Jun 24 19:30:34 why a new kernel recipe? Jun 24 19:30:41 do you have such big patches? Jun 24 19:30:50 or your kernel under git? Jun 24 19:30:55 no, just a newbie here :-D Jun 24 19:31:16 the kernel has modifications from the vendor of the set Jun 24 19:31:33 but has a horrible toolchain Jun 24 19:31:54 so I'm porting it to OE :-) Jun 24 19:32:16 okay Jun 24 19:32:36 send your patches to the oe-devel mailinglist for review Jun 24 19:32:57 I will, but take into account that I didn't do those myself Jun 24 19:33:20 meaning: I do not have a precise idea of what they do (but learning them during the process :- ) Jun 24 19:37:28 hm if you want if offical in oe you have to provide patches Jun 24 19:37:29 *g* Jun 24 19:41:22 I know, I just don't want to send patches I cannot say what they do :-) Jun 24 19:41:29 but will do my best for ot Jun 24 19:41:32 *it Jun 24 19:42:00 ??? Jun 24 19:42:06 no you didnt understand me Jun 24 19:42:11 fully Jun 24 19:42:18 not only the kernel patches Jun 24 19:42:26 the patches for oe itself Jun 24 19:42:30 machine.conf Jun 24 19:42:43 ah, yes, of course Jun 24 19:42:46 and the linux---.bb's Jun 24 19:52:33 so how do I fix a "bad rpath" ?? Jun 24 19:52:55 look at linking Jun 24 19:53:08 seems the buildsystem of the soft made mistakes Jun 24 19:53:23 it's some autoconf mess Jun 24 19:55:50 buserror do you have some pastebins for me? Jun 24 19:56:35 well I'll post the .bb and the log hold on Jun 24 19:59:11 woglinde, http://pastebin.com/m571fbbea Jun 24 19:59:52 the old libftdi faisl completely, so I tought I'd update it since I need support for the new IC anyway. it /almost/ works :> Jun 24 20:00:10 hm ftdi Jun 24 20:00:15 there was something Jun 24 20:00:27 but its to far away I remember Jun 24 20:00:43 the log is not enough Jun 24 20:01:16 I'm fishing the full log Jun 24 20:01:21 hm wait Jun 24 20:01:40 0.16? Jun 24 20:01:44 version Jun 24 20:01:45 yes Jun 24 20:01:47 the latest? Jun 24 20:01:48 okay Jun 24 20:02:04 they are conservative with version numbers :> Jun 24 20:02:54 which log you want ? do_compile, do_configure ? Jun 24 20:05:57 do_compile Jun 24 20:06:00 and do_install Jun 24 20:08:17 http://oomz.net/log.txt and Jun 24 20:08:21 http://oomz.net/logi.txt Jun 24 20:09:28 buserror where did you buy the sheeva? Jun 24 20:09:43 I'll do the debugging of the mips rootfs later...now I have to help someone make a recipe Jun 24 20:09:54 globalscale website. took forever to arrive... Jun 24 20:11:37 buserror do you really need the examples? Jun 24 20:11:49 the examples making the errors Jun 24 20:12:17 nope, not at all. Jun 24 20:12:45 oh ? hmm maybe there's a --without-examples Jun 24 20:12:49 so they should end up in the normal package anyway Jun 24 20:12:56 aeh shouldnt Jun 24 20:13:10 nope Jun 24 20:13:16 patching Makefile.am Jun 24 20:13:23 and remove examples Jun 24 20:13:39 otherwise you have to add a m4 macro to configure.in Jun 24 20:13:52 and an ifdef in Makefile.am Jun 24 20:14:04 choose whats easier Jun 24 20:14:27 heh well. I rather suspect nobody would /install/ examples for that anyway Jun 24 20:14:55 examples could be installed in another package Jun 24 20:14:57 *g* Jun 24 20:15:11 depends on you how much effort you want invest Jun 24 20:16:23 well, I already broke my teeth on wvdial earlier, I want an easy out on this one :> Jun 24 20:16:39 hehe Jun 24 20:16:56 then patch Makefile.am and remove examples Jun 24 20:18:14 btw, wvdial won't work on mipsel and arm* -- the wvstreams library uses "getcontext" api that is nor implemented on these. So I spent forever fixing new upsteam version in, gor it to work, and it faisl at runtime... Jun 24 20:30:33 wow Jun 24 20:30:37 this is cool Jun 24 20:30:42 besides the sheeva Jun 24 20:30:47 http://www.globalscaletechnologies.com/t-openrdcdetails.aspx#component Jun 24 20:33:10 woglinde, it's not out, yet, I think Jun 24 20:33:37 http://www.globalscaletechnologies.com/c-2-globalscale-technologies-products.aspx?currencysetting=EUR& Jun 24 20:37:15 woglinde, niceeeeeeeeeeeeeee Jun 24 20:37:40 but i'll like more something with a raw LCD interface ;) Jun 24 20:38:03 gremlin right Jun 24 20:38:14 but that would cost some more I think Jun 24 20:38:53 woglinde, should not ... Jun 24 20:39:02 gremlin should Jun 24 20:39:08 but they have to design it new Jun 24 20:39:12 for you Jun 24 20:39:14 at the moment Jun 24 20:39:19 and that would cost Jun 24 20:39:49 ahh ... in this sense true ! Jun 24 20:45:04 anyone here tried 2.6.30 w/ 4.4.0? Jun 24 20:45:32 m4t not yet Jun 24 20:46:18 it worked fine as my main compiler, but there is a CONFIG_FRAME_WARN=1024 directive Jun 24 20:46:47 that uses some compiler flag, but it seems when i try to use it as a gcc-cross-kernel.bb style package, it isnt compiled in Jun 24 20:46:49 fwiw, I don't think it'd be all that hard to implement getcontext()/setcontext() for arm if there is software that needs them. Jun 24 20:51:02 also, is there a preferred way to remove a compiler package from ./tmp/cross? Jun 24 20:51:11 bitbake -c clean doesnt do it Jun 24 20:53:32 this is what i am seeing now, btw, with 4.4.0: Jun 24 20:53:33 | cc1: error: unrecognized command line option "-Wframe-larger-than=1024" Jun 24 20:53:34 | cc1: error: unrecognized command line option "-fno-dwarf2-cfi-asm" Jun 24 20:54:14 i found the mailing list post when it was added to the kernel tree Jun 24 20:54:30 but i dont know why the compiler doesnt support those flags Jun 24 20:54:41 something about the gcc-cross-kernel.bb, i think Jun 24 20:54:57 maybee they are removed? Jun 24 20:55:19 i just reinstalled/recompiled it Jun 24 20:55:35 I meant the removed the options in gcc-4.4 Jun 24 20:55:51 oh Jun 24 20:55:53 yea Jun 24 20:56:22 the kernel compile succeeded when i simply used gcc_cross-4.4.0.bb though Jun 24 20:56:44 rather than 'wrapping' it in the gcc-cross-kernel package Jun 24 20:57:30 hm gcc-cross-kernel we only needed for older kernel which could only be compiled with gcc-3 Jun 24 20:58:15 oh Jun 24 20:58:25 well, angstrom/ppc405 is stuck at 4.1.1 Jun 24 20:58:32 and that is what my userland is done with Jun 24 20:58:33 Well, hang on Jun 24 20:58:42 k Jun 24 20:58:48 gcc-cross-kernel is generally useful Jun 24 20:58:48 03Rolf Leggewie  07org.openembedded.dev * r14b1edc935 10openembedded.git/conf/distro/minimal.conf: conf: remove bluetooth from DISTRO_FEATURES in minimal.conf Jun 24 20:58:57 We used it for cortex a8 stuff too :) Jun 24 20:59:14 And angstrom could move to 4.3.3 for ppc, if someone boot tests it :) Jun 24 20:59:34 i tried 4.3.3 for the 2.6.30 compile Jun 24 20:59:44 and it fails at the __weak test saying my compiler is too old Jun 24 20:59:52 i think my cross/ directory might be screwy at this point Jun 24 21:00:04 ok Jun 24 21:00:16 now, whats your gcc-cross-kernel-4.4.0_4.4.0.bb look like? Jun 24 21:00:17 just: Jun 24 21:00:22 require gcc-cross-initial_${PV}.bb Jun 24 21:00:22 require gcc-cross-kernel.inc Jun 24 21:00:23 ? Jun 24 21:00:34 i had to add a do_stage() because install-libgcc doesnt exist Jun 24 21:00:42 it doesnt exist in 4.3.3 apparently either Jun 24 21:00:50 ah yes, thats true Jun 24 21:00:52 i tried substituting it with install-multilib as well Jun 24 21:01:06 i think you can just drop the line, but i forget Jun 24 21:01:32 gcc -v shows a different thread model than my *initial* 4.1.1 had too Jun 24 21:01:38 single vs. posix Jun 24 21:01:43 and much less configure options Jun 24 21:04:20 ???? Jun 24 21:04:29 why do_stage for cross_compiler? Jun 24 21:05:28 i dunno that's what is defined in gcc-cross-kernel.inc Jun 24 21:17:27 [repost] there are a small set of low-hanging-fruit patches on there : git://repo.or.cz/openembedded/mini2440.git for-upstream-dev --- } have tested them for a few weeks on the -stable fork I have, but I also rebased them on -dev and they all work (on arm). anyone care to pull ? Jun 24 21:19:11 denix: ping Jun 24 21:41:16 03Rolf Leggewie  07org.openembedded.dev * ra4ee92d0a8 10openembedded.git/recipes/tasks/task-opie.bb: task-opie: make inclusion of irda and bluetooth tasks contingent upon COMBINED_FEATURES Jun 24 21:41:19 03Rolf Leggewie  07org.openembedded.dev * r3dede602a9 10openembedded.git/recipes/images/opie-image.bb: opie-image: make inclusion of irda and bluetooth tasks contingent upon COMBINED_FEATURES Jun 24 21:42:03 Laibsch: ping Jun 24 21:42:18 no need to ping, just ask your question Jun 24 21:42:24 saves one round-trip Jun 24 21:42:56 is a bit OT Jun 24 21:43:08 about sharprom-compatible Jun 24 21:43:37 sounds onT to me Jun 24 21:43:45 I'm testing a patch allowing to eliminate the silly gcc-2.95 thing from build logs Jun 24 21:44:11 still I have yet to see if it doesn't break sharprom distro Jun 24 21:44:31 I'll show you the patch soon Jun 24 21:44:48 hm eclipse now with native coca on mac Jun 24 21:44:51 ..building collie now (the other 5 are done :) Jun 24 21:46:01 ok, let me know when there is something to look at Jun 24 21:46:14 Laibsch: pls remember Pavel Machek asked for a serial cable Jun 24 21:46:30 surely worth to send :) Jun 24 21:47:45 Have him send me a mail Jun 24 21:47:49 I'm back in Europe Jun 24 21:47:57 fine, thx Jun 24 21:48:18 ah..collie still NOTE: Runtime target 'virtual/arm-angstrom-linux-gcc-2.95' is unbuildable, removing... Jun 24 21:48:22 I need to dig it out of the relocation cartons Jun 24 21:48:28 ok, let me finish drastically... Jun 24 21:49:04 wow..collie has so many kernels in OE... Jun 24 21:49:13 so many old kernels... Jun 24 22:20:19 ant__: pong Jun 24 22:21:09 hi denix, as I said, I'm editing sharprom-compatible Jun 24 22:22:47 ant__: any problems with it? Jun 24 22:23:06 hopefully not.. I'm finishing the build for collie Jun 24 22:23:12 (angstrom) Jun 24 22:23:36 I'm eradicating the gcc-2.95 legacy messages Jun 24 22:23:48 see: Jun 24 22:26:29 http://fr.pastebin.ca/1473352 Jun 24 22:26:40 denix: Laibsch: ^^ this is the idea Jun 24 22:27:41 ah, nice! Jun 24 22:28:45 there was a similar post some hours ago Jun 24 23:29:59 why do we need gcc-cross-kernel recipe Jun 24 23:40:17 m4t is trying to not jump his whole project from 4.1.1 to 4.4.0 Jun 24 23:40:35 But in general there's times where it's useful / needed to have a different compiler for the kernel than the rest of the world Jun 24 23:41:25 Tartarus: IMO building a full cross-compiler and then doing kernel build is a little sane thing to do in already insane cross compilation world Jun 24 23:41:43 Tartarus: what does it buy Jun 24 23:41:49 compilation time ? Jun 24 23:42:18 Hmm? Jun 24 23:42:22 Over what? Jun 24 23:42:35 complete gcc build Jun 24 23:42:45 Er Jun 24 23:42:48 or rathar toolchain build Jun 24 23:42:54 We don't otherwise allow for gccs to co-exist Jun 24 23:43:20 And yes, sometimes you need one gcc version for the kernel and one for the rest of the world Jun 24 23:44:51 IMO using same gcc for kernel and rest is a better integration when you think of interface between kernel and libc Jun 24 23:45:01 and changes like EABI Jun 24 23:54:13 Not always possible Jun 24 23:54:16 That's the problem Jun 24 23:55:43 may be then create a kernel SDK Jun 24 23:55:52 for kernel and modules Jun 25 00:04:32 03Andrea Adami  07org.openembedded.dev * rf3a000b477 10openembedded.git/recipes/zaurus-updater/zaurus-installer.bb: zaurus-installer: set LICENSE Jun 25 00:04:34 03Andrea Adami  07org.openembedded.dev * re141e38cd9 10openembedded.git/ (5 files in 2 dirs): Jun 25 00:04:34 embedix kernels: set COMPATIBLE_MACHINE = "none" Jun 25 00:04:34 - silence the notes like: Jun 25 00:04:35 - ERROR: '[]' RDEPENDS/RRECOMMENDS or otherwise Jun 25 00:04:37 - requires the runtime entity 'virtual/arm-angstrom-linux-gcc-2.95' Jun 25 00:04:39 - but it wasn't found in any PACKAGE or RPROVIDES variables Jun 25 00:04:41 - COMPATIBLE_MACHINE is set in sharprom-compatible Jun 25 01:31:18 hmm in stable/2009 libx11 doesn't build.. even though x11_disable_makekeys.patch is applied, it tries to run makekeys Jun 25 01:31:41 http://tinderbox.openembedded.org/public/logs/task/908421.txt Jun 25 01:59:03 sod it back to dev we go... **** ENDING LOGGING AT Thu Jun 25 02:59:57 2009