**** BEGIN LOGGING AT Mon Aug 24 02:59:57 2009 Aug 24 04:40:12 wow dhl sucks Aug 24 06:36:16 zecke: I was sleep deprived. :) but I gave you due credit in follow up email Aug 24 06:37:23 gn Aug 24 06:37:29 * khem -> sleep () Aug 24 06:54:53 khem: :} Aug 24 07:58:31 Goodmorning Aug 24 08:02:28 03Elena Grandi  07org.openembedded.dev * r4b9209b155 10openembedded.git/ (2 files in 2 dirs): xapian-bindings-python: new recipe Aug 24 08:02:38 03Elena Grandi  07org.openembedded.dev * r5e3196b053 10openembedded.git/ (conf/checksums.ini recipes/xapian/xapian-core_1.0.14.bb): xapian-core: updated to version 1.0.14 Aug 24 08:02:39 03Elena Grandi  07org.openembedded.dev * rfec8c1fb53 10openembedded.git/recipes/sugar/sugar-datastore_0.84.0.bb: sugar-datastore: added runtime dependency on the xapian bindings Aug 24 08:13:51 03Steffen Sledz  07org.openembedded.dev * ra701cc1e0a 10openembedded.git/recipes/linux/linux-2.6.24/hipox/defconfig: linux-2.6.24: enable initrd support for hipox machine Aug 24 08:16:43 good morning Aug 24 08:37:37 What application / initscript is normally responsible of spawning an console on an attached screen? Aug 24 08:40:17 03Elena Grandi  07org.openembedded.dev * r0066ccb831 10openembedded.git/recipes/xapian/ (5 files): xapian-core: moved common lines in xapian-core.inc Aug 24 08:40:49 lrg_: morning, I don't have a problem with you adding your company to the list of OE friendly companies Aug 24 08:42:25 zecke: morning. thanks Aug 24 08:42:42 zecke, I looking at budget airlines now for Berlin.... Aug 24 08:43:12 lrg_: we are in need for more marketing though, and Angstrom deserves a face lift and some list of supporting companies too Aug 24 08:44:10 zecke, I agree, the more the better. Gives Angstrom more credibility with pointy haired bosses ;) Aug 24 08:44:23 lrg_: oh, you can always use the "proxy" form to ask someone to go to berlin Aug 24 08:44:51 zecke: XorA|gone and I are discussing this atm Aug 24 08:45:42 lrg_: bosses, right. The website should tell what it is, why one needs it, and whom to pay for support Aug 24 09:00:33 What is the appropriate way to start a terminal (console) on my screen (as it doesn't show up with a console image). Aug 24 09:03:59 tsjsieb: sorry? you are talking about the target? Aug 24 09:04:07 yes Aug 24 09:04:09 tsjsieb: and you want to have a password prompt (getty) running? Aug 24 09:04:30 tsjsieb: is your console= kernel cmdline correct? is your inittab entry correct too? Aug 24 09:04:30 yes correct, normally it starts automattically on screen isn't it? Aug 24 09:05:28 console in kernel cmdline is to my serial port on my own pc, Aug 24 09:06:04 tsjsieb: What do you mean by screen? GNU screen? Aug 24 09:06:23 No, the 'hardware' screen wich is attached to the device Aug 24 09:07:09 I would like to start getty there with login screen, but I can't find wich argument I should give to getty Aug 24 09:08:30 I can use the screen for framebuffer applications, but I don't know how I can start getty for the screen? that should be possible, isn't it? Aug 24 09:09:04 tsjsieb: /etc/inittab declares where/when to start the login process Aug 24 09:09:42 tsjsieb: e.g. 1:2345:respawn:/sbin/getty 38400 tty1 will start it on the first tty... Aug 24 09:10:04 tsjsieb: e.g. if console=/dev/ttySAC0 then tty1 will be the console attached to /dev/ttySAC0 Aug 24 09:16:35 zecke: thnx, but I don't completely understand the last part. Is there standard a tty on which the 'hardware screen' should be? Or can I find it in some way Aug 24 09:18:15 tsjsieb: If you don't use your serial as console it should use the framebuffer device, IIRC you can also type console=/dev/fb0? Aug 24 09:24:49 (or simply omitting console and let the kernel figure it out) Aug 24 09:26:51 I did the last, but then I have no serial console, and also no console on the hardware screen (only ssh acces) Aug 24 09:29:55 03Robert Schuster  07org.openembedded.dev * r4f2d35ae98 10openembedded.git/recipes/ecj/libecj-bootstrap_3.5.bb: libecj-bootstrap 3.5: Set DEFAULT_PREFERENCE to -1. Aug 24 09:32:37 tsjsieb: well, for serial you will need a correct entry in the inittab... Aug 24 09:32:56 tsjsieb: for you "hardware screen", I have no idea what it is, so I can't comment (the options are just way too many) Aug 24 09:34:11 O.k. thnx for your help. Aug 24 09:36:50 tsjsieb: In your custom machine, did you set SERIAL_CONSOLE correctly? Aug 24 09:38:00 tsjsieb: SERIAL_CONSOLE is used by recipes/sysvinit/sysvinit_2.86.bb to fix/change the inittab Aug 24 09:38:36 tsjsieb: and there is also USE_VT Aug 24 09:39:33 yes, and the serial console works fine :) only the console on the hardware screen is the problem (I want to use fbida to display a picture for testing purposes, but that programs does seem to work when run from ssh terminal and frambuffer as target) Aug 24 09:52:39 I'm having some trouble with building rootfs on an NFS share. It takes about 1 hour (vs. 5 minutes on my local machine), as far as I can see it's already done the "meat" of the unpacking but then it goes on for an hour with messages like these: http://pastebin.ca/1541395 Aug 24 09:53:11 morning Aug 24 09:53:16 it seems to me that whatever it's trying to do is wrong regardless. Aug 24 09:53:20 morning hrw Aug 24 09:59:07 good morning Aug 24 10:42:19 hi all Aug 24 10:44:33 hm Aug 24 10:44:38 is mirggi usable now/ Aug 24 10:45:30 almost.... Its not perfect but works Aug 24 10:47:34 I didn't manage to find the trick how to convince it that i want to change the channel window Aug 24 10:49:13 ah... "cancel" ... Clever :) Aug 24 11:31:03 ~curse ulf linux-2.6.30.2 recipe Aug 24 11:31:55 hrw: be more friendly :) Aug 24 11:32:01 bbl, buying a faster machine now Aug 24 11:32:22 zecke: his simple recipe with zillions of files generated is ugly Aug 24 11:32:44 heh, hrw is the new koen :-) Aug 24 11:35:26 ~curse ulf linux-2.6.30.2 recipeNOTE: Handling BitBake files: / (5432/7083) [76 %]ERROR: Could not include required file linux-2.6.30/dummy-arch/2.6.30.2/patch-sets/SRC_URI_append.inc while parsing /home/hrw/devel/oe/openembedded/recipes/linux/linux_2.6.30.2.bb Aug 24 11:35:31 etc Aug 24 11:53:21 I have an issue with GNU_HASH when building my own Makefile-based recipe Aug 24 11:53:45 The case is that in Makefile I used assignment to LDFLAGS Aug 24 11:54:36 I took a look at the patchwork and there seem to be quite a few recipes which were being corrected lately due to GNU_HASH QA error Aug 24 11:55:40 but they sometimes used the different approach: changing CC=${CC}, removing LDFLAGS, etc. Aug 24 11:56:34 what is the current best practice of using "special" (i.e. for implicit rules) variables in Makefiles to avoid clash with OE? Aug 24 11:57:37 in makefile you need to add to LDFLAGS not replace it Aug 24 11:57:47 thats all basically Aug 24 11:58:06 does it work even when make -e option is passed? Aug 24 11:58:52 thats how I fixed that in some apps Aug 24 11:59:07 I saw I guess 3 approaches: Aug 24 11:59:42 TARGET_CC_ARCH += "${LDFLAGS}" Aug 24 12:00:16 hack Aug 24 12:01:02 oe_runmake 'CC=${CC}' 'LD=${LD}' 'LDFLAGS=${LDFLAGS}' Aug 24 12:02:00 and: export LD = "${CC} ${LDFLAGS}" Aug 24 12:02:17 3rd - patch makefile to do LDFLAGS="${LDFLAGS} something-what-was-set-in-that-makefile"? Aug 24 12:03:08 sure. I'll probably go with appending to LDFLAGS in Makefile Aug 24 12:04:52 but is there any clue on what 'make' variables OE sets and wants to keep them? I want to avoid overwriting something like CXXFLAGS, CPPFLAGS, etc. Is appending always a good solution? Aug 24 12:07:15 rather better then overwriting them Aug 24 12:09:17 OK, thanks Aug 24 12:16:56 good morning Aug 24 12:19:40 good afternoon :) Aug 24 12:20:19 ~ugt Aug 24 12:20:27 ah.. ibot on vacations... Aug 24 12:20:30 hi rkirti Aug 24 12:21:43 :) hello hrw Aug 24 12:22:00 th1_: http://www.total-knowledge.com/~ilya/mips/ugt.html Aug 24 12:22:59 I already found it from hrw's comment ;) Aug 24 12:23:22 via wikipedia.. Aug 24 12:25:35 interesting, I did'n know that Aug 24 12:26:10 mckoan: ronetix boards use at91bootstrap to boot to u-boot? Aug 24 12:30:30 * broonie wonders WTF mkimage does that is so resource intensive. Aug 24 12:31:03 broonie: iirc, it only converts zImage to uImage format,which is acceptable to uboot Aug 24 12:31:55 rkirti: I know what the goal is. Aug 24 12:32:25 What I want to know is how it's managing to implement something that should be very simple so that it ends up being the most resource intensive part of a kernel build. Aug 24 12:32:45 broonie: ah, sorry then..but it takes hardly a second here :) Aug 24 12:32:49 use the source luke^Wbroonie Aug 24 12:32:58 broonie: result has few GB? Aug 24 12:35:17 hrw: they don't use at91bootstrap Aug 24 12:35:44 lucky ones Aug 24 12:36:03 I am working on getting my at91sam9263ek working again Aug 24 12:36:11 too much played with at91bootstrap and uboot Aug 24 12:38:56 hrw: are the original ones working on at91sam9263ek? Aug 24 12:44:03 kermit 00:23:5A:A6:B2:85 Aug 24 12:44:33 oops, wrong chan Aug 24 12:44:40 ok, I have working combo Aug 24 12:44:53 linux4sam at91bootstrap 1.14 + OE u-boot 2009.01 Aug 24 12:47:59 When building gnome-games-2.24.0-r5, I'm getting "Error: unrecognized option '-/SDL'". Anyone else seeing this? I'm on a Debian lenny amd64. Aug 24 12:49:02 hrw: It's only 1.8M Aug 24 13:17:52 03ghost  07org.openembedded.dreambox.release_25 * rb6439f661d 10openembedded.git/packages/enigma2/enigma2-plugins.bb: enigma2-plugins.bb: push srcdate because of a bugfix in genuine dreambox plugin Aug 24 13:17:53 03ghost  07org.openembedded.dreambox.release_25 * r94868243dc 10openembedded.git/packages/gstreamer/gst-plugin-dvbmediasink.bb: gst-plugin-dvbmediasink: push srcdate (added a workaround for a decodebin2 bug, dont poll in flush and locked mode) Aug 24 13:17:53 03ghost  07org.openembedded.dreambox.release_25 * refdc5c86c5 10openembedded.git/packages/dreambox/dreambox-dvb-modules.bb: dreambox-dvb-modules.bb: another dm800 driver update (deinterlacer auto handling was not working okay) Aug 24 13:17:57 03Felix Domke  07org.openembedded.dreambox.release_25 * re3a6688f3d 10openembedded.git/: Merge branch 'org.openembedded.dreambox.release_25' of git://git.opendreambox.org/git/ghost/opendreambox-1.5 into org.openembedded.dreambox.release_25 Aug 24 13:18:01 03ghost  07org.openembedded.dreambox.release_25 * r2003c2108a 10openembedded.git/packages/dreambox/dreambox-dvb-modules.bb: dreambox-dvb-modules.bb: update dm8000 drivers ... fix a deadlock in 4501 driver when tune to a invalid DVB-S2 FEC Aug 24 13:18:07 03ghost  07org.openembedded.dreambox.release_25 * r8aa5785b96 10openembedded.git/packages/enigma2/enigma2.bb: enigma2.bb: push srcdate to fix a zaptimer bug Aug 24 13:18:12 03ghost  07org.openembedded.dreambox * r9a6ea59c68 10openembedded.git/packages/dreambox/dreambox-dvb-modules.bb: dreambox-dvb-modules.bb: another dm800 driver update (deinterlacer auto handling was not working okay) Aug 24 13:18:16 03ghost  07org.openembedded.dreambox * r8d55d6578d 10openembedded.git/packages/dreambox/dreambox-dvb-modules.bb: dreambox-dvb-modules.bb: update dm8000 drivers ... fix a deadlock in 4501 driver when tune to a invalid DVB-S2 FEC Aug 24 13:18:20 03ghost  07org.openembedded.dreambox * r08dce6ac44 10openembedded.git/packages/gstreamer/gst-plugin-dvbmediasink.bb: gst-plugin-dvbmediasink: push srcdate (added a workaround for a decodebin2 bug, dont poll in flush and locked mode) Aug 24 13:18:26 03Felix Domke  07org.openembedded.dreambox * r15cf870fc6 10openembedded.git/: Merge branch 'org.openembedded.dreambox' of git://git.opendreambox.org/git/ghost/opendreambox-1.5 into org.openembedded.dreambox Aug 24 13:18:30 03ghost  07org.openembedded.dreambox * rbae63090c4 10openembedded.git/packages/gstreamer/ (2 files in 2 dirs): packages/gstreamer: hotfix for refcounting problem (not from gstreamer team) Aug 24 13:21:43 When building gnome-games-2.24.0-r5, I'm getting "Error: unrecognized option '-/SDL'". Anyone else seeing this? I'm building on a debian lenny amd64. Aug 24 13:22:04 * * OE Bug 4578 has been RESOLVED (FIXED) by nicola.mfb(AT)gmail.com Aug 24 13:22:06 * * uicmoc4-native_4.4.0 does not compile Aug 24 13:22:08 * * http://bugs.openembedded.net/show_bug.cgi?id=4578 Aug 24 13:25:04 does anyone know why midori segfaults? I think it's missing something, but I can't figure out why. I tried with both 0.1.18 and 0.1.2 Aug 24 13:55:38 When building gnome-games-2.24.0-r5, I'm getting "Error: unrecognized option '-/SDL'". Anyone else seeing this? I'm building on a debian lenny amd64. Aug 24 13:57:54 bgat: look at http://tinderbox.openembedded.net/packages/gnome-games/ Aug 24 13:57:55 bgat, not many off us build gnome-games :) Aug 24 13:58:39 Crofton|work: he...you need a full working gpe-environment to have a bit of time for games :) Aug 24 13:58:46 hrw: ping Aug 24 13:59:10 pong Aug 24 13:59:21 hello Aug 24 13:59:47 it seems you could perhaps push som epackaged-staging patches... Aug 24 14:00:15 nobody else (apart RP and kergoth) using it... Aug 24 14:00:21 ant_work: sorry but I refuse to add anything new in this week - have to finish own stuff Aug 24 14:00:38 np, just I guess I found a couple of bugs Aug 24 14:03:14 ant_work: I don't have time for gnome-games either, but it's part of the beagleboard-demo-image. :( Aug 24 14:03:46 I see in the failed do_compile ccache was used...are you using it too? Aug 24 14:04:13 I suppose so, since I didn't tell it not to... Aug 24 14:04:50 just an idea... Aug 24 14:05:43 the thing is, the Makefile actually has "-/SDL" everywhere, suggesting that whatever builds that file got something wrong. Aug 24 14:06:21 I'm really new to oe, so I'm not sure where to look. Aug 24 14:07:24 I do notice that the tinderbox machine is i686, but I'm amd64. I can't imagine that's a problem, tho. Aug 24 14:08:29 Here's an offending line from the Makefile: Aug 24 14:08:58 GNOME_GAMES_CFLAGS = ... ... ... -/SDL -D_GNU_SOURCE=1 ... ... ... Aug 24 14:09:21 SDL_CFLAGS = -/SDL ... Aug 24 14:09:29 SDL_MIXER_CFLAGS = -/SDL ... Aug 24 14:10:30 check makefile.in or makefile.am Aug 24 14:11:59 bgat: perhaps try gnome-games_2.6.1.bb Aug 24 14:12:34 commit 43100bfa8ec1d7eebb9cecb491719e3491ef8f77 Aug 24 14:12:34 Author: Roger Monk Aug 24 14:12:34 Date: Fri Aug 7 09:45:57 2009 +0100 Aug 24 14:12:34 gnome-games: fix paths to host SDL, bump PR Aug 24 14:12:45 do_configure_append() { Aug 24 14:12:45 for i in $(find ${S} -name "Makefile") ; do Aug 24 14:12:45 - sed -i -e s:'-I /usr/include -I /usr/local/include'::g $i Aug 24 14:12:45 + sed -i -e s:'I/usr/include'::g -e s:'-I /usr/include -I /usr/local/include'::g $i Aug 24 14:13:18 Is that line turning "-I /usr/include -I /usr/local/include" into "" ? Aug 24 14:13:59 ant_work: how do I do that? Aug 24 14:14:03 bgat: the key error (in the logs) was CROSS COMPILE Badness: /usr/include in INCLUDEPATH: /usr/include/SDL Aug 24 14:14:22 ant_work: which means what? Aug 24 14:14:34 I cross-compile all the time, didn't realize I was engaging in "badness". :) Aug 24 14:14:58 Those include paths are obviously wrong, of course. Aug 24 14:15:08 please pastebin your log Aug 24 14:16:47 ant_work: you mean the log files in temp, or my console log? Aug 24 14:18:09 the console log should point to the do_compile I suppose Aug 24 14:18:22 console output: http://pastebin.com/m48728748 Aug 24 14:19:01 $ find tmp -name "SDL.h" Aug 24 14:19:02 tmp/staging/armv7a-angstrom-linux-gnueabi/usr/include/SDL/SDL.h Aug 24 14:19:02 tmp/staging/x86_64-linux/usr/include/SDL/SDL.h Aug 24 14:19:02 tmp/work/armv7a-angstrom-linux-gnueabi/libsdl-x11-1.2.11-r7/SDL-1.2.11/include/SDL.h Aug 24 14:19:02 tmp/work/armv7a-angstrom-linux-gnueabi/libsdl-x11-1.2.11-r7/image/usr/include/SDL/SDL.h Aug 24 14:19:02 tmp/work/armv7a-angstrom-linux-gnueabi/libsdl-x11-1.2.11-r7/staging-pkg/staging/armv7a-angstrom-linux-gnueabi/usr/include/SDL/SDL.h Aug 24 14:19:05 tmp/work/armv7a-angstrom-linux-gnueabi/libsdl-x11-1.2.11-r7/install/libsdl-x11-dev/usr/include/SDL/SDL.h Aug 24 14:19:07 tmp/work/x86_64-linux/libsdl-native-1.2.11-r2/SDL-1.2.11/include/SDL.h Aug 24 14:19:09 tmp/work/x86_64-linux/libsdl-native-1.2.11-r2/staging-pkg/staging/x86_64-linux/usr/include/SDL/SDL.h Aug 24 14:21:46 hm, at line #32 -/SDL is passed Aug 24 14:23:22 yep Aug 24 14:25:18 bgat: I don't have the makefiles here...I'd look at the Roger's patch and see if/how that sed apply Aug 24 14:25:54 the patch tries to solve the a.m. Badness Aug 24 14:26:42 ok, I'll give that a try. Aug 24 14:27:16 [16:13] Is that line turning "-I /usr/include -I /usr/local/include" into "" ? probably yes Aug 24 14:28:03 I just turned that line into "echo foo", will see what happens. :) Aug 24 14:28:57 I guess you want tmp/staging/armv7a-angstrom-linux-gnueabi/usr/include/SDL/SDL.h Aug 24 14:29:35 yep Aug 24 14:29:36 bye Aug 24 14:32:46 ok, I can at least reproduce the CROSS COMPILE Badness message now. That means I'm hacking in the right place. Aug 24 14:34:57 I'll try turning it into '${STAGING_INCDIR}'... Aug 24 14:36:02 bgat: I'm sorry I don't have the sources here and cannot be precise (lacking specific gtk / sdl know-how), but generally speaking the right approach is to patch the makefile.(in:am) Aug 24 14:36:38 Well, since Roger hacked the .bb file, I'll hack his hack and then send the diffs to him and see what he wants to do with them. Aug 24 14:36:51 I get this error while trying to build chromium: scons: *** No SConstruct file found. Aug 24 14:37:23 I think I might have fixed the problem, since bitbake is running a LOT longer than it did before. :) Aug 24 14:40:02 that's nice! please post the patch so someone could push the fix Aug 24 14:40:13 (when you're sure it's ok :) Aug 24 14:42:50 well, it's still grinding so I think I'm good. Aug 24 14:43:28 http://pastebin.com/m45aef4ea Aug 24 14:44:47 I just emailed that to Tony, too. Aug 24 14:47:04 th1_: sorry, no idea as I'm an oe noob. But it sounds like something didn't create the SConstruct file that scons needs. Disk full, perhaps? Aug 24 14:47:18 ant_work: NOTE: Tasks Summary: Attempted 2710 tasks of which 2361 didn't need to be rerun and 0 failed. Aug 24 14:47:20 woo! Aug 24 14:47:58 bgat, no, not full :) but thanks Aug 24 14:48:15 bgat: fine, do you see any QA Log? Aug 24 14:48:40 ant_work: heh, I just rm -rf'ed the tmp directory so I could do a clean build. Aug 24 14:48:46 oops! Aug 24 14:48:48 :) Aug 24 14:49:09 np, I suppose only acouple of fixes are needed (http://tinderbox.openembedded.net/public/logs/qa/125201.txt) Aug 24 14:49:35 you can forget about those Aug 24 14:49:58 I don't care if gnome-games actually _works_, I just want it to build. :) Aug 24 14:50:14 that's a valid reason :) Aug 24 14:50:27 bgat, rofl Aug 24 14:54:22 alright, I'm late for a meeting. bbl Aug 24 15:20:44 03Elena Grandi  07org.openembedded.dev * r111ed2d525 10openembedded.git/recipes/sugar/sugar-base_0.84.1.bb: sugar-base: fixed renamed dependency openssh-keygen Aug 24 15:23:13 does anyone have a list of ports you need open to do OE builds? Aug 24 15:23:20 we should put that on the wiki Aug 24 15:25:31 well, none, if you've done your fetching ahead of time ;) but that's a good idea Aug 24 15:56:16 hi kergoth Aug 24 15:56:26 hey Aug 24 16:30:40 03Michael Smith  07org.openembedded.dev * rdfe2bb428b 10openembedded.git/recipes/mesa/ (mesa-tls.inc mesa-common.inc): Aug 24 16:30:40 mesa-dri: move tls configure logic to a class Aug 24 16:30:40 Will allow reuse by xorg-xserver. Aug 24 16:30:40 Signed-off-by: Michael Smith Aug 24 16:30:51 03Michael Smith  07org.openembedded.dev * rdc929c2ebc 10openembedded.git/recipes/mesa/ (4 files): Aug 24 16:30:51 mesa-dri: set PACKAGE_ARCH to ${MACHINE} Aug 24 16:30:51 ${MACHINE_DRI_MODULES} is passed to configure, so the package output Aug 24 16:30:51 depends on MACHINE. As foretold by commit f06119fc :) Aug 24 16:30:53 Also switch to INC_PR and fix built of mesa-xprogs when prefix != /usr. Aug 24 16:30:55 Signed-off-by: Michael Smith Aug 24 16:30:57 03Michael Smith  07org.openembedded.dev * rfaac35c7aa 10openembedded.git/recipes/xorg-xserver/ (8 files): Aug 24 16:31:00 xorg-xserver-common: pass --enable-glx-tls Aug 24 16:31:02 Uses the same inherit as mesa-dri, so they both get the same answer. Aug 24 16:31:04 Also switch to INC_PR. Aug 24 16:31:06 Signed-off-by: Michael Smith Aug 24 18:12:42 what's the commit policy on core changes again, 2 acks? Aug 24 18:13:55 soemthing like that :) Aug 24 18:14:12 do you have some that lack acks? Aug 24 18:18:20 good question, i have a pile in my inbox that i need to go through and check on their states Aug 24 18:18:26 i really should play with patchwork more Aug 24 18:23:55 :) Aug 24 18:24:00 mickey|bbl, http://www.ossie.wireless.vt.edu/trac/wiki/BeagleBoard Aug 24 18:24:02 doh Aug 24 18:24:05 try Aug 24 18:24:15 mickey|bbl, http://typedia.com/ Aug 24 18:25:56 i really really need to find a workflow I'm happy with for getting stuff upstream into OpenEmbedded from local more minimal collections/overlays Aug 24 18:27:07 git branches? Aug 24 18:27:50 03Koen Kooi  07org.openembedded.dev * re0041be7a7 10openembedded.git/ (4 files in 4 dirs): linux-omap 2.6.29: update holes patch, make iommu modular for beagle Aug 24 18:28:33 doh, micro doesn't seem to build anymore since khem switched it to eglibc. Aug 24 18:28:55 * pb_ switches back Aug 24 18:29:01 Crofton: I'm using them, but there are a few problems. the fact that the repo is more minimal means its slightly more complex.. you really dont want to manage a standalone collection by git rm'ing half the tree, you'd get conflicts for just about every commit.. and, if you're doing cherry picks back and forth, git can lose track of what has and hasn't gone up or come down Aug 24 18:29:35 k Aug 24 18:29:37 ga Aug 24 18:31:10 anyone knows if double-reference of virtual packages work? Aug 24 18:31:43 double-reference? Aug 24 18:32:14 something like PREFERRED_PROVIDER_virtual/foo = virtual/bar, PREFERRED_PROVIDER_virtual/bar = blah Aug 24 18:34:44 and besides setting PROVIDES = "virtual/foo" in a recipe, is there anything else to define a virtual package? Aug 24 18:34:57 no, you can't do that. the rhs of PREFERRED_PROVIDER needs to be a concrete package. Aug 24 18:35:52 I didn't quite understand your second question but I am fairly sure the answer is no to that one as well. :-} Aug 24 18:37:26 pb_: thanks! :) so, for the 2nd Q - what's the proper way to make a recipe provide a virtual package then? Aug 24 18:38:11 just what you said: set PROVIDES. Aug 24 18:45:05 pb_: thanks again. too bad one virtual cannot prefer another virtual though... :) Aug 24 20:20:24 pb_: what is the build error Aug 24 20:22:49 pb_: Lets fix the problem instead of switching Aug 24 20:30:30 khem: I get a vast number of undefined symbols linking nscd. Aug 24 20:33:09 pb_: that's interesting... I get a QA error about nscd some days ago. But building glibc the next time it was gone. Aug 24 20:36:14 florian: was that regular glibc, or eglibc? Aug 24 20:38:45 pb_: regular glibc Aug 24 20:46:05 good morning folks Aug 24 20:47:09 hi ant__ Aug 24 20:47:27 is the heat wave 'vorbei' in DE? Aug 24 20:48:51 here in IT it was impossible to work (evening 34°) I had two friends with half-fried notebooks crashing! Aug 24 20:53:19 florian: ah right, my problem was with eglibc Aug 24 20:53:29 ant__: today was warm here but not really hot Aug 24 20:53:36 Crofton|work: are you administering patchwork? Aug 24 20:54:06 pb_: ok, this seems to be a problematic binary :) Aug 24 20:54:10 "yes" Aug 24 20:54:15 what is the problem? Aug 24 20:54:17 hi, I somehow dislike the 'Carnivore' instinct: it happens to mangle the patched Aug 24 20:54:27 emails Aug 24 20:54:41 florian: heh, it seems Aug 24 20:54:49 I forgot It's observing me, and I fired two mails to ML Aug 24 20:55:05 :) Aug 24 20:55:09 then I posted some fixes but the patches shows ugly Aug 24 20:55:20 not much I can do about it Aug 24 20:55:31 probably I'll have to archive these and start anew Aug 24 20:55:34 you can remove the stuff you do not want in there Aug 24 20:55:39 yep Aug 24 20:56:31 Crofton|work: and the two hooks Download mbox | patch Aug 24 20:56:42 mbox -> scrambled Aug 24 20:56:48 patch - > dead Aug 24 20:56:51 RIP Aug 24 20:57:39 hmmm Aug 24 20:58:18 yes, no raw here Aug 24 21:00:16 Crofton|work: could we please get ENABLE_XMLRPC turned on for patchwork? Aug 24 21:00:19 it might allow for some hotness Aug 24 21:00:30 then again, it might not :) Aug 24 21:00:31 heh Aug 24 21:00:40 these are beyond me Aug 24 21:00:48 I didn't install it Aug 24 21:01:24 gotcha. Aug 24 21:01:35 msmith_: hey, while you're there...I remember you did one commit 'desesperate try to deploy u-boot' and you changed do_stage in do_install..or..what was the trick? Aug 24 21:02:32 i think that may have been someone else, I am normally working with x86 stuff, nothing cool enough to need u-boot :) Aug 24 21:04:02 oh..sorry..it was Ulf indeed Aug 24 21:04:03 http://cgit.openembedded.org/cgit.cgi/openembedded/commit/?h=ulf/linux-2.6.30.2&id=30a1560d4b65d17fb879fe711c5b230c90c1b123 Aug 24 21:05:46 I see some 'special' packages appear to fail on rebuild from packaged-staging, just looking at possible fixes Aug 24 21:07:54 03Chris Larson  07org.openembedded.dev * r743aaf30cc 10openembedded.git/classes/distutils-common-base.bbclass: Aug 24 21:07:54 distutils-common-base.bbclass: only raise missing python-native error if python-native isn't in DEPENDS. Aug 24 21:07:54 Signed-off-by: Chris Larson Aug 24 21:07:54 Acked-by: Michael 'Mickey' Lauer Aug 24 21:09:14 03Jeremy Puhlman  07org.openembedded.dev * rd149d261d3 10openembedded.git/conf/bitbake.conf: (log message trimmed) Aug 24 21:09:14 bitbake.conf: Add export for cross NM. Aug 24 21:09:14 When not providing a cross nm, the configure for libtool-cross falls back to Aug 24 21:09:14 the system nm. This can lead to empty "global_symbol_pipe" and Aug 24 21:09:14 "global_symbol_to_cdeclvarble" variables in the generated libtool script. Aug 24 21:09:18 Signed-off-by: Jeremy Puhlman Aug 24 21:09:20 Signed-off-by: Chris Larson Aug 24 21:12:18 03Chris Larson  07org.openembedded.dev * r278cec5e0e 10openembedded.git/classes/kernel.bbclass: Aug 24 21:12:18 kernel.bbclass: Install & package symvers as /boot/Module.symvers-. Aug 24 21:12:18 Signed-off-by: Chris Larson Aug 24 21:12:18 Acked-by: Michael Smith Aug 24 21:13:10 03Dale Farnsworth  07org.openembedded.dev * r34cd6c2a37 10openembedded.git/classes/kernel.bbclass: (log message trimmed) Aug 24 21:13:10 kernel.bbclass: Set EXTRA_OEMAKE to null for kernel Aug 24 21:13:10 Currently the only thing in EXTRA_OEMAKE is '-e MAKEFLAGS='. We don't Aug 24 21:13:10 want to overide the kernel's Makefile variables from the environment. Aug 24 21:13:10 It caused the passed -j parameter from PARALLEL_MAKE to be ignored. Aug 24 21:13:14 Signed-off-by: Dale Farnsworth Aug 24 21:13:16 Signed-off-by: Chris Larson Aug 24 21:15:27 03Stanislav Brabec  07org.openembedded.dev * rc62f14e1aa 10openembedded.git/ (6 files in 3 dirs): subversion: Updated to version 1.6.5. Deleted old unused versions. Aug 24 21:15:27 mickey|bbl, I am closing http://bugs.openembedded.org/show_bug.cgi?id=5296 minimal now uses a recent eglibc which solved the issue Aug 24 21:15:37 03Stanislav Brabec  07org.openembedded.dev * r140ad9e1ac 10openembedded.git/: Merge branch 'org.openembedded.dev' of git@git.openembedded.org:openembedded into org.openembedded.dev Aug 24 21:16:35 03Dale Farnsworth  07org.openembedded.dev * r97cd982275 10openembedded.git/classes/patch.bbclass: (log message trimmed) Aug 24 21:16:35 patch.bbclass: Add "git" patchtool mechanism, which uses git-apply Aug 24 21:16:35 It can be selected by setting PATCHTOOL = "git". Aug 24 21:16:35 This is useful because git-apply honors the permissions information Aug 24 21:16:35 produced by git-format-patch. Aug 24 21:16:36 Signed-off-by: Dale Farnsworth Aug 24 21:16:38 Signed-off-by: Chris Larson Aug 24 21:19:05 * * OE Bug 5296 has been RESOLVED (FIXED) by ospite(AT)studenti.unina.it Aug 24 21:19:07 * * libfsobasics_git.bb fails to configure Aug 24 21:19:09 * * http://bugs.openembedded.net/show_bug.cgi?id=5296 Aug 24 21:25:14 03Evan Doiron  07org.openembedded.dev * r571084dd3f 10openembedded.git/ (conf/checksums.ini recipes/nano/nano_2.0.9.bb): Aug 24 21:25:14 nano: add v2.0.9 Aug 24 21:25:14 Signed-off-by: Evan Doiron Aug 24 21:25:14 Signed-off-by: Michael Smith Aug 24 21:25:14 Acked-by: Khem Raj Aug 24 21:25:25 03Evan Doiron  07org.openembedded.dev * r48c52b45a0 10openembedded.git/ (conf/checksums.ini recipes/dmidecode/dmidecode_2.10.bb): Aug 24 21:25:25 dmidecode: add v2.10 Aug 24 21:25:25 Signed-off-by: Evan Doiron Aug 24 21:25:25 Signed-off-by: Michael Smith Aug 24 21:25:27 Acked-by: Khem Raj Aug 24 21:28:10 pb_: strange I did not get such errors Aug 24 21:28:45 pb_: what version of gcc/binutils/eglibc combination is it pulling Aug 24 21:30:33 gcc 4.2.4, binutils 2.18, eglibc 2.10 Aug 24 21:31:34 Sounds reasonable although I havent tried 4.2.4 lately Aug 24 21:31:49 what symbols does it complain about Aug 24 21:34:14 accept4, inotify_init1, mkostemp, __nss_next2, __nss_group_lookup2, qsort_r, __asprintf_chk, plus sundry others Aug 24 21:34:46 I also get a "R_ARM_TLS_LE32 relocation not permitted in shared object" error which is rather weird. Aug 24 21:37:00 hmm seems static libraries getting pulled into shared objects Aug 24 21:38:41 ah yes, it does seem to be trying to use libc.a for some reason. Aug 24 21:39:26 let me try with this combo and which machine do you chose Aug 24 21:39:51 s/chose/choose/ Aug 24 21:40:48 __nss functions are private functions in libc it should have got them when building nscd Aug 24 21:40:55 qemuarm Aug 24 21:41:37 ok let me try it out I usually used osk5912 Aug 24 21:44:02 pb_ what linux version does it use for linux-libc-headers Aug 24 21:44:09 I used 2.6.29 Aug 24 21:45:28 2.6.23 Aug 24 21:49:32 pb_: I just started a build I will know in couple of hours Aug 24 21:52:22 righto Aug 24 21:54:20 pb_: is it compiling for OABI I wonder Aug 24 21:54:37 because in micro.conf TARGET_OS = "linux" Aug 24 21:54:50 should be "linux-gnueabi" for arm Aug 24 21:56:17 no, this is an eabi configuration. Aug 24 21:56:43 include/sane-toolchain-eglibc.inc has some not-so-sane logic which forces eabi for platforms that it thinks ought to be getting it. Aug 24 21:56:56 heh Aug 24 21:57:07 and, as it happens, that is producing the result I wanted so I left it be. Aug 24 21:57:10 I thought distro conf overwrote it Aug 24 21:57:22 but if it is picking from eglibc.conf then it should get it right Aug 24 21:57:31 (it's not just the eglibc one, sane-toolchain-uclibc and sane-toolchain-glibc both do the same crazy thing) Aug 24 21:57:50 yes python at its cracking best Aug 24 21:58:32 no, the definition in micro.conf comes first, so it's overridden by the later ones in "sane"-toolchain-xx. Aug 24 21:58:53 micro definition is prolly not needed then Aug 24 21:59:03 it's needed for non-arm platforms Aug 24 21:59:35 the complex expression in toolchain. incs will set it for non arm too Aug 24 22:00:05 ah, so it will Aug 24 22:20:04 * * OE Bug 5303 has been created by hyc(AT)symas.com Aug 24 22:20:06 * * perl ExtUtils/typemap and ExtUtils/xsubpp in the wrong package Aug 24 22:20:08 * * http://bugs.openembedded.org/show_bug.cgi?id=5303 Aug 24 22:20:47 * archae0pteryx is frustrated by patchwork Aug 24 22:24:31 archae0pteryx: why Aug 24 22:25:43 I can figure out how to use it for my immediate needs, but if for example I wanted to use patchwork from both OE and kernel.org that Aug 24 22:25:51 that's not completely obvious Aug 24 22:31:01 you can git-am patches from it Aug 24 22:31:17 there is a script in contrib for oe which is handy Aug 24 22:37:04 Yeah, I see that, and I'm guessing that if you want to use pwclient w/ .pwclientrc that doesn't work for more than one url Aug 24 23:05:48 pb_L eglibc build fine with the combo you suugested Aug 24 23:12:43 can anyone remember the motivation behind coreutils-native? was it just to get the patched 'install' binary? Aug 24 23:17:31 generally coreutils was not cross compilation friendly Aug 24 23:17:47 To get rid of our hacked up versions of install.sh Aug 24 23:17:49 so probably one reason to nail the beast Aug 24 23:19:16 khem: that's highly unlikely. most of coreutils doesn't give a shit about cross. install is the issue, since an install -s has, historically, resulted in corrupt binaries Aug 24 23:19:58 kergoth: I did not mean cross compiling itself I meant when used in cross compiling process Aug 24 23:20:04 so did i Aug 24 23:20:05 yes thats one of the problems Aug 24 23:20:17 I'd be curious to hear about anything other than install that's a problem Aug 24 23:20:28 outside of a Linux host? Aug 24 23:20:34 No, it was just for a good install Aug 24 23:20:45 and to kill the install.sh stuff Aug 24 23:20:48 okay, thought so. thanks Aug 24 23:21:30 wondering why we didn't do a recipe based on the coreutils-native one that -only- builds install, to avoid building crap we don't end up needing Aug 24 23:28:00 atleast we use a single version Aug 24 23:28:39 otherwise every distro will have its own coreutils issues being injected into OE build Aug 24 23:29:31 you'd be far more likely to hit problems with some of the doc generation tools that users already have to install on their build machines, than anything to do with coreutils Aug 24 23:29:35 *far* more likely Aug 24 23:51:09 how about doing a ccache-native build :) Aug 25 00:37:24 ccache is evil Aug 25 00:38:25 I don't know if it's ccache bugs or interactive with filesystem issues or just bad luck, but every once in a while it'll cache something bad Aug 25 00:38:43 And that's much more annoying than the shasum-native won't build issue Aug 25 00:40:01 really? Tridgewell claims that doesn't happen Aug 25 00:40:09 * grg shrugs Aug 25 00:40:28 i've never had an issue Aug 25 00:40:54 what filesystem is your $HOME (or the location of your cache)? Aug 25 00:40:55 Yeah, I used to go "that can never happen" until it bit me Aug 25 00:40:59 GFS Aug 25 00:41:12 i dare say that doesn't get lots of testing Aug 25 00:41:17 yeah Aug 25 00:41:26 It can be funny at times too Aug 25 00:41:30 GFS that is Aug 25 01:46:11 hi Aug 25 01:46:24 openembedded is angstrom ? Aug 25 01:46:30 they are same ? Aug 25 01:46:37 no Aug 25 01:46:46 you build Angstrom with Openembedded Aug 25 01:47:01 Angstrom is an extension of OpenEmbedded so ? Aug 25 01:52:20 OpenEmbedded/BitBake is the tool/framework you use to build Linux distributions for embedded Linux devices. Angstrom is one of the distributions you can select. Aug 25 02:00:00 ok thanks for the information Aug 25 02:00:26 i want to change the splash screen on angstrom , that is compass image and loading panel , how can i modify them ? Aug 25 02:02:47 hmm Aug 25 02:03:10 i ve the same problem , how can i change angstrom compass startup picture ? Aug 25 02:04:26 * kergoth experiments with the 'recipe within the scm checkout/clone' model Aug 25 02:12:50 Deviator, KadirB you will need to look at the psplash recipe Aug 25 02:13:39 the psplash package will need to be rebuilt Aug 25 02:14:20 ok thnx alot Aug 25 02:14:35 hmm Aug 25 02:16:43 i c... Aug 25 02:16:48 ok i ll look there too Aug 25 02:16:57 i ll be back tomorrow i hope Aug 25 02:17:02 thanks for all helps Aug 25 02:17:08 bye friends. Aug 25 02:24:43 hmmm Aug 25 02:25:58 i hope i told them the right thing. if i wanted to change my mind, its too late now :D Aug 25 02:30:58 NOTE: package quilt-native-0.48-23-gedd8cf1-dirty-r0: task do_fetch: Started Aug 25 02:30:59 NOTE: package quilt-native-0.48-23-gedd8cf1-dirty-r0: task do_fetch: Succeeded Aug 25 02:31:00 progress Aug 25 02:49:31 hmm, we hsould change the bitbake event handling to let you specify a 'events' flag on an eventhandler to only fire it for those events **** ENDING LOGGING AT Tue Aug 25 02:59:56 2009