**** BEGIN LOGGING AT Mon Feb 23 02:59:58 2009 Feb 23 03:13:25 03Mike Westerhof  07org.openembedded.dev * r24b3cbc8c8 10openembedded.git/conf/distro/slugos.conf: SlugOS: slugos.conf - add IMAGE_INITSCRIPTS, which apparently became requires suddenly. Feb 23 07:45:21 good morning Feb 23 07:45:27 morning Feb 23 07:45:56 morning Feb 23 08:02:25 Morning. Feb 23 08:07:58 I've made a new package and added it to the list of packages (IMAGE_INSTALL) that my image should install. But this fails for some reason. Before I added the package, the image builds perfectly. What have I done wrong? http://pastebin.com/d1e629f3 Feb 23 08:08:21 The new package is "my-music" and listed in the url above. Feb 23 08:17:41 OE says it can't find my package "my-music", but how can that be? The my-music directory (with the recipe in it) is located in the same directory as my-setup, which installs fine. Feb 23 08:22:12 FILES_${PN} = "${WORKDIR}/They_All_Laughed.mp3" is totally invalid Feb 23 08:22:20 so you made empty packages Feb 23 08:22:58 FILES_${PN} = "/tmp/music/They_All_Laughed.mp3" matches when you do in do_install Feb 23 08:27:58 hello, does someone knows where to find a 2.6.10-davinci kernel? it seems that montavista didn't relase it but prefered to port the driver to a newer kernel and then release the newer kernel...it's for making a diff to see the difference between a machine kernel and the davinci one...in order to see how hard it is to port it to a kernel at least >= 2.6.16 that has EABI Feb 23 08:29:03 hm.. Feb 23 08:29:14 ok Feb 23 08:29:33 FILES_${PN} should match what's installed by do_install? Feb 23 08:30:02 with absolute paths, as seen by the final image? Feb 23 08:32:07 XorA? Feb 23 08:33:44 Is the "install" command executed with proper path variables? Feb 23 08:41:03 Zta: looks like it Feb 23 08:42:26 this still seems to generate an empty package, though..: http://pastebin.com/m26445d Feb 23 08:43:21 you'll have to check what the logs say Feb 23 08:43:26 I have to go out now Feb 23 08:44:50 No, it works =) Feb 23 08:44:55 thanks for the help Feb 23 08:50:05 by the way...what's the correct way to force OABI for angstrom? Feb 23 08:50:34 /tmp might not be the best place to install anything.. Feb 23 08:51:45 Zta, look at the dev mailing list...there was a thread explaning how to install in home...but I must go now Feb 23 08:52:10 ok Feb 23 08:56:26 not easy to find =) Feb 23 09:07:16 I did find this, though =) http://thread.gmane.org/gmane.comp.handhelds.openembedded/21201/focus=21203 Feb 23 09:14:40 good morning Feb 23 09:57:42 morning Feb 23 09:59:00 good morning Feb 23 09:59:36 time to do yab... Feb 23 09:59:40 yet another build Feb 23 10:02:58 morning hrw :-) Feb 23 10:03:03 hi, does someone knows how to force OABI in angstrom Feb 23 10:03:14 (kernel 2.6.10) Feb 23 10:03:33 Gnutoo: grep collie conf/distro/ -r |grep ^angstrom Feb 23 10:05:02 hrw, ah ok it returns that...angstrom-2008.1.conf:ARM_ABI ?= "${@['','oabi'][bb.data.getVar('MACHINE',d) in ['collie','h3600', 'h3800', 'simpad', 'htcwallaby']]}" I've tried so far: Feb 23 10:05:11 *adding a new machine to the list Feb 23 10:05:19 *changing ARM_ABI in local.conf Feb 23 10:05:55 (ARM_ABI = "oabi") Feb 23 10:07:07 *adding PACKAGE_ARCH = "arm-oabi" and BASE_PACKAGE_ARCH = "arm-oabi" to the machine config Feb 23 10:08:45 in theory ARM_ABI = "oabi" in local.conf should be enough Feb 23 10:11:56 hrw, I have it but it still says: TARGET_OS = "linux-uclibcgnueabi" Feb 23 10:13:57 $ grep ABI local.conf => ARM_ABI = "oabi" Feb 23 11:02:26 hi Feb 23 11:04:59 hi woglinde Feb 23 11:05:14 *sigh* Feb 23 11:05:28 udev and hal and libvolumeide gives me headaches Feb 23 11:06:47 im trying to compile xwindows, there are too many providers for xwindows like x.org and xfree86 what is the difference between these, how does one choose one Feb 23 11:07:14 s4 please use a proper distribution Feb 23 11:07:17 like angstroem Feb 23 11:07:23 and you dont have to care Feb 23 11:07:32 about choosing the right thing Feb 23 11:07:43 yeah, but i want to know what is the difference Feb 23 11:07:52 why there are so many providers Feb 23 11:08:02 what differenciates them and when to use what Feb 23 11:08:31 hm you should try wikipedia for fork Feb 23 11:08:44 or even X self Feb 23 11:11:54 wikipedia says xfree86 is the root of all x windows and all other xwindows (x.org) are forks of xfree86 Feb 23 11:12:12 because of licensing issues Feb 23 11:12:24 :) Feb 23 11:17:11 s4: minor nitpick ... it's an X Window System; no s in Window :) Feb 23 11:17:56 blinder hihi Feb 23 11:18:11 thanks Feb 23 11:19:40 s4: we are not running windows, right? :) Feb 23 11:20:46 wrong. im on windows Feb 23 11:20:48 ;-) Feb 23 11:21:20 ~lart maemo folks Feb 23 11:21:21 * ibot hurls dozens of incontinent, insomniac, hungry kittens with tiny little razor-sharp claws and a wide variety of contagious intestinal parasites at maemo folks Feb 23 11:21:58 i like that Feb 23 11:24:59 s4: you are doing something wrong then :) Feb 23 11:28:45 yeah i know Feb 23 11:29:46 okay, i want to build a xwindow system from scratch even though i can do it through oe just to understand. Feb 23 11:30:08 i find that xfreedesktop also gives you xlibraries and xserver Feb 23 11:59:30 hi XorA Feb 23 11:59:44 hey florian Feb 23 12:03:28 florian: any luck with libertas? Feb 23 12:04:07 XorA: not yet... i was ill last week and have to catch up now first. Feb 23 12:05:00 florian: ah, ive got 2.6.28 running on the tosa minus a backlight Feb 23 12:05:13 florian: I shall have to try libertas again Feb 23 12:06:11 XorA: yes that would be pretty interesting. i started to try 2.6.28 here too, but it takes a little while. Feb 23 12:09:27 03ghost  07org.openembedded.dreambox * rea5aea549a 10openembedded.git/packages/gstreamer/ (gst-plugins-ugly_0.10.9.bb gst-plugins.inc): gst-plugin.inc: add libcdio to DEPENDS... pump PR of gst-plugins-ugly Feb 23 12:09:32 03Felix Domke  07org.openembedded.dreambox * r6772f58177 10openembedded.git/: Merge branch 'org.openembedded.dreambox' of git://git.opendreambox.org/git/ghost/opendreambox-1.5 into org.openembedded.dreambox Feb 23 12:09:33 03ghost  07org.openembedded.dreambox * ra7f9cff8f5 10openembedded.git/packages/gstreamer/gst-plugin-dvbmediasink.bb: gst-plugin-dvbmediasink.bb: push SRCDATE Feb 23 12:09:34 03ghost  07org.openembedded.dreambox * r7a832e12ef 10openembedded.git/packages/tuxbox/ (2 files in 2 dirs): tuxbox-tuxtxt-32bpp.bb: no more crash when press text key in e2 video player when non ts file ist playbacked Feb 23 12:09:37 03ghost  07org.openembedded.dreambox * r71f8c9c19e 10openembedded.git/: Merge branch 'org.openembedded.dreambox' of git://git.openembedded.net/openembedded into org.openembedded.dreambox Feb 23 12:09:40 03ghost  07org.openembedded.dreambox.release_25 * r929490f266 10openembedded.git/packages/tuxbox/ (2 files in 2 dirs): tuxbox-tuxtxt-32bpp.bb: no more crash when press text key in e2 video player when non ts file ist playbacked Feb 23 12:09:46 03ghost  07org.openembedded.dreambox.release_25 * rf3e628d5d0 10openembedded.git/packages/gstreamer/ (gst-plugins-ugly_0.10.9.bb gst-plugins.inc): gst-plugin.inc: add libcdio to DEPENDS... pump PR of gst-plugins-ugly Feb 23 12:56:07 I want to ship a set of config files that overwrites the package standard files when building an image. Is there some smart way to do this? If I just create a configs package, is there some way I can be sure that the config package is installed last and therefore overwrites the old configs? Feb 23 13:00:26 hi jay7 Feb 23 13:00:40 woglinde: hi :) Feb 23 13:24:10 03Stefan Schmidt  07fso/milestone5 * rcd6df32016 10openembedded.git/ (conf/checksums.ini packages/cellhunter/cellhunter_0.3.1.bb): cellhunter: Add new recipe. Game that collects GSM cell tower locations. Feb 23 13:24:16 03Stefan Schmidt  07fso/milestone5 * r65d4eac944 10openembedded.git/packages/tasks/task-openmoko-feed.bb: task-openmoko-feed: Add cellhunter Feb 23 13:25:00 he mickeyl Feb 23 13:32:42 hi Feb 23 13:40:21 re Feb 23 13:40:25 morgen mickeyl Feb 23 13:44:36 morgen hre^hw Feb 23 13:44:40 how are things? Feb 23 13:44:50 mickeyl: good Feb 23 13:45:01 have some interesting toys to work on Feb 23 13:45:05 but also some crap ones Feb 23 13:48:47 03Elena Grandi  07org.openembedded.dev * rd5ab9dee38 10openembedded.git/packages/tuxtype/tuxtype_1.7.3.bb: Feb 23 13:48:47 tuxtype 1.7.3: new recipe Feb 23 13:48:47 Tuxtype is a typing game Feb 23 13:48:52 ~curse dmp for 'inventing' vortex86sx Feb 23 13:48:53 May you be reincarnated as a Windows XP administrator, dmp for 'inventing' vortex86sx ! Feb 23 13:56:25 amazing... Feb 23 13:56:46 device use ddr2 chips which can run at 800MHz Feb 23 14:00:08 hrm, mips with oe anyone? Feb 23 14:00:18 like working einvorns or somesuch? Feb 23 14:00:36 environs even Feb 23 14:14:38 hey thesing Feb 23 14:16:40 hi XorA Feb 23 14:16:42 morning all Feb 23 14:17:03 thesing: I got collie with MMC_DEBUG set, just working out how to dump serial port to a file Feb 23 14:18:53 great. Feb 23 14:19:07 anybody tried to build recently ? Feb 23 14:19:09 XorA: cat /dev/ttyS0 | tee collie.log? Feb 23 14:19:12 python_2.6.1.bb seems broken Feb 23 14:19:21 * hre do poky builds today Feb 23 14:19:32 hre: makes nonsense because serial port doesnt know the params :-) Feb 23 14:19:51 my Packages.filelist is empty. How to fill it? Feb 23 14:19:53 XorA: setserial first? Feb 23 14:20:06 booxter: rm Packages*;bitbake package-index Feb 23 14:20:20 XorA: or use minicom - it has logging Feb 23 14:20:31 hre, tnx Feb 23 14:21:40 * XorA has now put the battery in backwards twice in a row Feb 23 14:22:15 ahah, the joys of stty Feb 23 14:22:32 argh.. just 2 cores in use during build ;( Feb 23 14:22:38 XorA: if you use minicom it should be quite easy to log everything too. Feb 23 14:22:54 thesing: that means the pain of minicom :-) Feb 23 14:23:00 09:22 in nyc... soon this machine will get used to other things ;( Feb 23 14:23:14 * hre -> food Feb 23 14:29:41 thesing: http://www.xora.org.uk/~dp/mmc.txt Feb 23 14:33:27 XorA: is this a regular sd card? Feb 23 14:34:10 thesing: microSD class 4, same happens on two regular sized SD as well Feb 23 14:34:44 but not sdhc cards or so? Feb 23 14:34:55 two are SD one is SDHC Feb 23 14:35:01 that one is SD Feb 23 14:36:48 Wel I'll look into it. Feb 23 14:37:00 thesing: cheers Feb 23 14:37:19 thesing: 2 are sandisk cards so shouldnt be cheapo knockoffs Feb 23 14:38:13 gah, tosa is even more broken in 2.6.29-rc5 :-( Feb 23 14:38:45 what doesn't work any more? Feb 23 14:38:53 framebuffer, backlight Feb 23 14:39:05 in 2.6.28 in OE the only non working device was backlight Feb 23 14:39:17 after I stole a couple of 2.6.29-rcX patches Feb 23 14:41:53 guys, are you interested in qt-embedded-4.5.0-rc1 packages? Or don't you need this rc-stuff? Feb 23 14:42:22 booxter: updates are always wanted Feb 23 14:45:09 03Jan Lübbe  07fso/milestone5 * ra190ef96a4 10openembedded.git/ (3 files in 2 dirs): pycd: add recipe Feb 23 14:48:42 booxter: do one for the latest snapshot instead :) Feb 23 14:48:53 XorA: do you know which commit broke it? Feb 23 14:49:06 erbo, nightly thing you mean? Feb 23 14:49:14 thesing: nope Feb 23 14:49:44 booxter: yes Feb 23 14:49:59 I would do a git-bisect and write a mail do arm-ml if you can't fix it yourself. Feb 23 14:54:34 03Frans Meulenbroeks  07org.openembedded.dev * rd28f108c6c 10openembedded.git/packages/bluez/bluez4_4.30.bb: bluez: add dependency on libusb Feb 23 15:01:55 03Fraxinas  07org.openembedded.dreambox * re109121cbf 10openembedded.git/packages/sqlite/sqlite3-3.6.11/ (4 files): remove superfluous files Feb 23 15:01:59 03Fraxinas  07org.openembedded.dreambox * r1a658cc2f6 10openembedded.git/packages/ (11 files in 5 dirs): Feb 23 15:01:59 adds dependencies for database search engine Feb 23 15:01:59 handle zlib modification in python-2.5.1-manifest.inc with care - i'm not entirely sure whether that has bad side effects Feb 23 15:02:14 03Fraxinas  07org.openembedded.dreambox * rc1692cc4af 10openembedded.git/: Merge branch 'org.openembedded.dreambox' of git://git.openembedded.net/openembedded into org.openembedded.dreambox Feb 23 15:10:44 any docs for openembedded qtopia/qt-embedded development? Feb 23 15:11:02 I mean - how to use toolchain generated by OE Feb 23 15:12:28 I only built opie stuff so did it with OE Feb 23 15:15:17 03Stanislav Brabec  07org.openembedded.dev * ra7653f0e9c 10openembedded.git/packages/matchbox-panel/ (2 files in 2 dirs): matchbox-panel: Prevent crashes in system-monitor applet (just a work-around). Feb 23 15:15:28 03Stanislav Brabec  07org.openembedded.dev * r2598f7b63e 10openembedded.git/packages/mc/ (10 files in 2 dirs): mc: Updated to 4.6.2 + openSUSE and Debian UTF-8 patches. Feb 23 15:16:23 hre, opie is not a qt-embedded thing Feb 23 15:16:33 it's a qtopia2 I think Feb 23 15:17:05 anyway, is there any doc about using OE toolchains for development? Feb 23 15:17:09 opie use qt/e 2.3 and is qtopia1.5/1.6/2.x Feb 23 15:17:34 booxter: http://marcin.juszkiewicz.com.pl/2008/07/17/using-poky-sdk-to-build-software/ maybe? Feb 23 15:31:07 Okay, I officially fail at Bitbake. I have a git tree with a 2.6 kernel forked from linux-omap. I have an overlay set up, and I'd like to overlay linux-omap. However, bitbake insists on building both kernels, saying one provides something that the other doesn't. Any ideas? Feb 23 15:36:02 MostAwesomeDude: MACHINE/DISTRO and 'bitbake what-exactly-you-gave' please Feb 23 15:36:39 hre: Angstrom, MACHINE is "oswald" but it's just a copy of "beagleboard" Feb 23 15:37:13 And the command is "bitbake linux-omap" Feb 23 15:38:47 MostAwesomeDude: and linux-omap recipe does not have your machine in list of compatible_machines Feb 23 15:38:59 MostAwesomeDude: you need to edit linux-omap recipe Feb 23 15:48:38 Hmm, can I use /build/tmp/cross as a toolchain in e.g. Eclipse CDT? Feb 23 15:49:36 (I know it's a very lame question) Feb 23 15:49:49 it is not recommended Feb 23 15:50:05 better build 'bitbake meta-toolchain', install result and use it Feb 23 15:55:53 thanks Feb 23 16:11:57 anybody know about libosso and libhildon-1 for nokia800 for OE? Feb 23 16:13:56 udovdh: chinook/diablo ones? Feb 23 16:14:13 yes Feb 23 16:14:23 libosso i can bitbake i see Feb 23 16:14:26 udovdh: if OE lack them then check http://git.pokylinux.org/cgit.cgi/experimental/meta-maemo/ and get it from there Feb 23 16:14:29 libhildon too? Feb 23 16:14:36 thanks Feb 23 16:14:44 how to include them in a build of navit? Feb 23 16:14:44 meta-maemo was my work on adding maemo components into Poky. I never merged it into OE Feb 23 16:14:52 udovdh: fetch, copy Feb 23 16:15:31 hre yes... Feb 23 16:15:41 libhildon is in the meta-maemo git tree I understand? Feb 23 16:16:58 check, I do not remember Feb 23 16:17:42 I see the git tree is consisting of recipes Feb 23 16:17:44 inetersing Feb 23 16:17:49 interesting Feb 23 16:18:14 so now (assuming I cna build them) I only have to have the libs avilable when navit is ./configured... Feb 23 16:18:16 ok thanks Feb 23 16:19:34 udovdh: why you do not check does navit is available in 'maemo extras devel' repository? Feb 23 16:20:01 I bitbaked navit svn myself for h2200 and now nokia800 Feb 23 16:20:20 want to be doing that for newer versions etc Feb 23 16:23:46 so I built libosso Feb 23 16:24:01 I wan to make sure that ./configure for `my` project finds it Feb 23 16:24:09 what is the preferred way to do that? Feb 23 16:24:16 like DEPENDS = "gtk+ libglade" ? Feb 23 16:27:31 hre: Thanks for the advice. I'm just defining a new kernel providee "linux-oswald" and will use that until my patches find their way upstream. Feb 23 16:29:15 MostAwesomeDude: I do not like you Feb 23 16:29:16 ;D Feb 23 16:29:30 why people create own recipes for kernels... Feb 23 16:30:53 hre, is my DEPENDS guess right? Feb 23 16:32:59 basically yes Feb 23 16:39:25 hre, thanks Feb 23 16:39:39 sop I move the i10n directory to my packages dir Feb 23 16:39:45 so... Feb 23 16:51:47 03Jeremy Lainé  07org.openembedded.dev * rb27d058757 10openembedded.git/packages/u-boot/ (u-boot-1.3.2/boc01/007-090112-CAPSENSE.patch u-boot_1.3.2.bb): u-boot-1.3.2: update keypad support for boc01 Feb 23 16:58:34 hre: I apparently hate myself. :c Feb 23 16:58:52 But we need to have "up and running" before "submit patches upstream." Feb 23 17:02:18 :) Feb 23 17:03:25 If it makes you feel better, I'm not some evil corporate drone, and we're planning on moving everything onto omap, and eventually onto vanilla. Feb 23 17:06:17 I know Feb 23 17:06:30 btw - hispeed usb is 480 and fullspeed is 12? Feb 23 17:07:45 hre: correct Feb 23 17:10:48 too much cables on desk... Feb 23 17:11:00 and I need to add 2 ethernets to this mess ;( Feb 23 17:11:44 * mwester solved the problem of too many cables on the desk (just don't look on the floor _under_ the desk!) Feb 23 17:17:41 ;D Feb 23 17:18:24 * hre curses badly to opensync, kdepim etc for lack of syncing with syncml phones Feb 23 17:19:03 * hre -> off Feb 23 17:25:04 digital widescreen monitors are ideal things to cover cables ... Feb 23 17:33:49 03Otavio Salvador  07org.openembedded.dev * raaf8a26454 10openembedded.git/packages/initrdscripts/files/80-ext3.sh: initramfs-module-ext3: use mkdir -p to create mount path Feb 23 17:33:49 03Otavio Salvador  07org.openembedded.dev * r3881ea7112 10openembedded.git/packages/initrdscripts/files/98-aufs.sh: initramfs-module-aufs: use mkdir -p to create mount path Feb 23 17:33:50 03Otavio Salvador  07org.openembedded.dev * r83ab7056b3 10openembedded.git/packages/initrdscripts/files/init.sh: Feb 23 17:33:50 initramfs-uniboot: add support to start a shell and enable debug Feb 23 17:33:52 To start a shell, you just need to use 'shell' boot param while to Feb 23 17:33:54 enable shell debugging use 'debug'. Feb 23 17:33:56 03Otavio Salvador  07org.openembedded.dev * r3d377399ab 10openembedded.git/ (conf/checksums.ini packages/e2fsprogs/e2fsprogs_1.41.4.bb): e2fsprogs: add 1.41.4 Feb 23 17:34:01 03Otavio Salvador  07org.openembedded.dev * r6fb58774bc 10openembedded.git/packages/initrdscripts/ (3 files in 2 dirs): Feb 23 17:34:04 initramfs-module-psplash: allow psplash to be disabled using boot param Feb 23 17:34:06 Add support to disable psplash using boot param as done for regular Feb 23 17:34:08 psplash package. To disable it, use 'psplash=false' as boot param. Feb 23 17:34:10 03Otavio Salvador  07org.openembedded.dev * r0b7903ed22 10openembedded.git/packages/initrdscripts/files/init.sh: initramfs-uniboot: call depmod if the image lack a dep file Feb 23 17:34:15 03Otavio Salvador  07org.openembedded.dev * r0ba9afbe37 10openembedded.git/packages/psplash/ (files/psplash-init psplash_svn.bb): Feb 23 17:34:18 psplash: improve how to detect boot param to avoid the psplash starting Feb 23 17:34:20 It was done using a looping through cmdline contents but it failed Feb 23 17:34:22 depending on params order; using a grep makes it works more reliably. Feb 23 17:58:24 03Frans Meulenbroeks  07org.openembedded.dev * rf2092fad18 10openembedded.git/packages/gtk-webcore/midori_git.bb: midori: changed PV in recipe to make sure ordering is ok Feb 23 17:58:25 03Frans Meulenbroeks  07org.openembedded.dev * re1dd20e04f 10openembedded.git/: Merge branch 'org.openembedded.dev' of git@git.openembedded.net:openembedded into org.openembedded.dev Feb 23 18:00:57 who is Frans Meulenbroeks? Feb 23 18:01:26 eFfeM if I recall correctly Feb 23 18:01:57 * XorA needs to deliver a swift boot to the nads for the bluez4 checkin Feb 23 18:02:14 eFfeM: ping? Feb 23 18:02:36 03Otavio Salvador  07org.openembedded.dev * r40a5267e28 10openembedded.git/packages/psplash/psplash.inc: psplash.inc: set PR for force the updating of branded packages Feb 23 18:03:32 XorA: pong Feb 23 18:03:46 eFfeM: you checkin that bluez4 change? Feb 23 18:03:53 yes Feb 23 18:04:02 eFfeM: its 100% wrong Feb 23 18:04:21 eFfeM: you destroy DEPENDS variable Feb 23 18:04:22 ouch, why ? Feb 23 18:04:36 so it should have been += ? Feb 23 18:05:00 and libusb-compat which is libusb0 API is already in the DEPENDS Feb 23 18:05:13 from where ? Feb 23 18:05:26 eFfeM: top of the file Feb 23 18:05:28 xora hehe yes Feb 23 18:05:32 saw that too Feb 23 18:05:41 actually came to this because I saw that one of the packages i needed faile because it missed a usb function Feb 23 18:05:57 yes please fix the commit Feb 23 18:06:32 woglinde: XorA will do Feb 23 18:06:56 should I retract the commit one way or another, or just undo the change, increase pr and commit that Feb 23 18:07:07 03Jeremy Lainé  07org.openembedded.dev * rd87a82f5a2 10openembedded.git/packages/linux/ (3 files in 2 dirs): linux-2.6.27: update keypad support for boc01 Feb 23 18:07:18 eFfeM: well see if you can work out what went wrong for you Feb 23 18:07:37 eFfeM: as I guess you saw a real bug, maybe libusb-compat misses functionality or something Feb 23 18:07:39 for me nothing went wrong locally but on tinderbox there was an error Feb 23 18:08:12 eFfeM: better undo and increment PR then, just in case someone built it in meantime Feb 23 18:08:24 the issue is that i see several of the packages fail on tinderbox with all kind of odd reasons and locally they build fine Feb 23 18:08:39 and of course that are pacakges i want to pull from the feed Feb 23 18:09:58 eFfeM: irritating heisenbugs :-) Feb 23 18:10:20 there's the fix Feb 23 18:10:38 sorry not yet, one more sec Feb 23 18:12:29 there you go Feb 23 18:12:39 03Frans Meulenbroeks  07org.openembedded.dev * rd81d509d27 10openembedded.git/: Merge branch 'org.openembedded.dev' of git@git.openembedded.net:openembedded into org.openembedded.dev Feb 23 18:12:54 03Frans Meulenbroeks  07org.openembedded.dev * rcc30078fd4 10openembedded.git/packages/bluez/bluez4_4.30.bb: blues4: undo previous (faulty) commit; apologies for any inconvenience! Feb 23 18:13:27 btw eFfeM comes from FM (my initials) but most systems think that is too short or it is already taken, so I picked something that pronounces the same Feb 23 18:13:39 hence the capitalisation Feb 23 18:16:21 eFfeM: I remeber the FOSDEM conversation now :-) I just forgot your real name :-) Feb 23 18:16:34 this one http://tinderbox.openembedded.net/public/logs/3511771.txt totally drives me nuts, this builds for me, configure on tinderbox apparently things docbook is there and builds doc; i do not have docbook and it is in packages/nonworking. Apparently there are some remnants of docbook on tinderbox Feb 23 18:16:39 XorA: np Feb 23 18:17:23 03Otavio Salvador  07org.openembedded.dev * rd950d0886e 10openembedded.git/packages/psplash/psplash-openmoko_svn.bb: psplash-openmoko: fix package compilation since it lacked logo-math.patch at SRC_URI Feb 23 18:18:05 The problem with libusb is that some recipes require libusb1 and other libusb-compat -- the two conflict by installing different archive libs to the same filename in staging. Thus it becomes order-dependent if your build succeeds or not. Feb 23 18:18:30 hmm; not too nice Feb 23 18:18:58 I "fixed" it by creating a dummy version "0.0.0" usb package -- if you set PREFERRED_VERSION it will pull in the newer libusb1 stuff instead. Feb 23 18:19:03 * Crofton|work needs to work out how to make gnuradio build with libusb1 Feb 23 18:19:10 btw this one: http://tinderbox.openembedded.net/public/logs/3511771.txt is also pretty odd; it is for boost Feb 23 18:19:22 This leaves a small number of apps that REQUIRE the old libusb breaking, but at least you can build deterministically that way. Feb 23 18:19:25 they peek at libusb-0.xx private data Feb 23 18:19:34 * XorA always blames Crofton|work for boost :-D Feb 23 18:19:54 still trying to find who had the libusb error Feb 23 18:19:58 * mwester doesn't even try to get the latest libboost working for SlugOS -- too arcane. Feb 23 18:20:25 mwester: consider it a challenge :-) Feb 23 18:20:34 * Crofton|work blames Eric Blossom for boost Feb 23 18:20:46 I have challenges sufficient, Mr. eFfeM ! :D Feb 23 18:20:52 well, he added the later version boost dependency to gnu radio Feb 23 18:21:07 gnu radio depends on boost? Horrors! Feb 23 18:21:09 mwester: LOL Feb 23 18:21:14 * Crofton|work wonders how many packages use boost .... Feb 23 18:21:41 and old libusb Feb 23 18:21:48 well, for usrp support Feb 23 18:21:52 frans@linux-suse:~/oe/openembedded/packages> grep -l boost */* |wc Feb 23 18:21:52 42 42 1027 Feb 23 18:23:18 libusb does not seem that pop Feb 23 18:26:02 what is the preferred libusb package ? Feb 23 18:28:53 libusb1 with libusb-compat Feb 23 18:34:23 !seen otavio Feb 23 18:34:24 woglinde, otavio is right here! Feb 23 18:34:32 otavio are you there? Feb 23 18:34:50 hi Feb 23 18:35:32 woglinde: Looking around for me? :-) Here am I :-) Feb 23 18:35:40 woglinde: shoot :-) Feb 23 18:35:46 eFfeM: and set PREFERRED_VERSION_libusb ?= "0.0.0" in your conf file to ensure that anything depending on the old libusb will ALSO use libusb and libusb-compat instead. Otherwise you will still have both libs built, and it will be random as to which one actually ends up in staging. Feb 23 18:36:20 otavio you checked the new efsprogs in wit PR=14 Feb 23 18:36:35 whee Feb 23 18:36:41 and you the download is not from a sf.mirror Feb 23 18:36:45 args Feb 23 18:36:48 fixed up the dpkg/apt stuff so it actually builds right now and doesn't rely on some host stuff Feb 23 18:36:49 woglinde: oh :( heck! I copied the old recipe and fixed it to compile it Feb 23 18:36:51 will push shortly Feb 23 18:36:55 woglinde: and forgot to fix the PR Feb 23 18:36:56 need to get my desk assembled :) Feb 23 18:37:04 woglinde: yes, it is Feb 23 18:37:16 debiab qa? Feb 23 18:37:19 debian qa Feb 23 18:37:21 woglinde: it fetched it by itself Feb 23 18:37:46 woglinde: debian qa is a script to hide the mirror Feb 23 18:37:54 woglinde: otherwise you get random paths Feb 23 18:38:05 woglinde: it is changed in my bitbake.conf Feb 23 18:42:39 woglinde: fixed, sorry Feb 23 18:42:39 03Otavio Salvador  07org.openembedded.dev * r0f5b41fdc7 10openembedded.git/packages/e2fsprogs/e2fsprogs_1.41.4.bb: e2fsprogs (1.41.4): fixes PR to r1, sorry Feb 23 18:44:07 * * OE Bug 5037 has been created by di(AT)fh-wedel.de Feb 23 18:44:09 * * module_strip.bbclass uses ${KERNEL_MAJOR_VERSION} but does not inherit its definition Feb 23 18:44:11 * * http://bugs.openembedded.net/show_bug.cgi?id=5037 Feb 23 18:49:51 Hello, is there a list of packages that are included in OE? Feb 23 18:52:01 hm, i do an opkg install of apache2 then i do opkg list_installed and apache2 is not there ? Feb 23 18:52:25 opkg remove apache2 does not remove anything but httpd -l says it is from apache2 Feb 23 18:52:28 how come ? how fix? Feb 23 18:53:23 and httpd from apache clashes with httpd from busybox Feb 23 18:59:42 Steffanx: run opkg list Feb 23 19:00:49 i'll give it a try Feb 23 19:02:18 eFfeM: opkg is busted. Feb 23 19:02:34 ? Feb 23 19:02:59 eFfeM: and busybox httpd should be broken out to a separate package that can have its own RCONFLICTS (like mdev/udev). Feb 23 19:03:52 Steffanx: the list is far too long to begin to discuss here, but opkg is broken in little ways, big ways, developer ways, and user ways. Feb 23 19:04:03 hmm ok Feb 23 19:04:22 so.. i should no use it Feb 23 19:05:06 Steffanx: Sorry, you have no choice. Feb 23 19:05:35 Those in charge of OE have determined that you must, except that someone is working on making dpkg work, I think. Feb 23 19:06:54 mja, i first have to get it work on ubuntu(8.10) and it looks like that sucks Feb 23 19:06:56 opkg mostly works, as long as you don't do anything outside of the unusual with it. And apparently, eFfeM is doing something unusual (!!???) Feb 23 19:07:37 That's just one of the things that makes you crazy; opkg list_installed should be such a simple operation, but it doesn't always work right. Feb 23 19:07:59 re Feb 23 19:08:14 OE supports building images from deb packages Feb 23 19:08:14 hi hrw! Feb 23 19:08:39 mwester: it can even support rpm packages but I need to find time to finish merging it from poky Feb 23 19:09:07 * * OE Bug 5038 has been created by di(AT)fh-wedel.de Feb 23 19:09:09 * * kernel object suffix not calculated correctly in module-base.bbclass Feb 23 19:09:11 * * http://bugs.openembedded.net/show_bug.cgi?id=5038 Feb 23 19:09:16 mwester: Who exactly was it that made this determination? I have a hard time believing that it can genuinely be the official oe policy that only opkg is permitted, and even more so that its use is obligatory. Feb 23 19:09:29 hrw: yes, but can it create .deb packages, and build the feeds, and all that? There was some discussion a few days ago about that, and it seemed that someone had determined that it had bit-rotted (at least that was what I took from that conversation, perhaps I am mistaken) Feb 23 19:10:14 mwester: from what I know no one worked on getting apt repositories working. but I think that it can be done in just few hours Feb 23 19:10:41 pb__: we had that "discussion" when ipkg was ripped out and the broken opkg dumped in. The core team met in private, and the response back was that it was a mistake to have done it that way, but that it was done now, and opkg would stay (hardcoded into the bbclass files). Feb 23 19:11:37 hi all Feb 23 19:11:49 hrw: That would translate to a few days for me, then, I guess. It's on my list of things to do, but I am one user sucess report away from releasing SlugOS 5, and have no time to research right now. Feb 23 19:12:13 I am hoping that modifying dpkg to make it smaller will be easier than trying to make opkg work. Feb 23 19:12:23 I'm playing with Qt Embedded 4.4.3 on h2200 and wonder... http://nie.ath.cx/qte-qa.log <- shouldn't those files be packaged as well? Feb 23 19:13:12 mwester: I'm still a bit baffled as to where these hard-coded references actually are. I tried "grep -i opkg classes/*" and didn't get any results that looked obviously bogus. Feb 23 19:13:20 pocek: should be rather. patches are welcome Feb 23 19:13:46 dummy question: to test the latest myckey patch "mickey/sane-updates-and-defaults" i have to change branch in my OE repo ... right ? Feb 23 19:13:54 gremlin[it]: yes Feb 23 19:14:39 mwester: ONLINE_PACKAGE_MANAGEMENT = "ipkg ipkg-collateral" in local.conf - test it Feb 23 19:14:45 pb__, meta-toolchain.bb is hard-coded Feb 23 19:14:52 pb__: if someone removed the hard-coded references since that discussion, it would delightful. But I would maintain that there should be NO references to opkg hardcoded; all should be virtualized so that one could revert to ipkg (which I'm sure had bugs too, but at least we _knew_ what those bugs were!) Feb 23 19:14:58 03Jan Lübbe  07fso/milestone5 * r73cf91df44 10openembedded.git/ (conf/checksums.ini packages/cellhunter/cellhunter_0.3.1.bb): cellhunter: update to 0.4.0 Feb 23 19:15:01 hrw: Oh I know that one!! Feb 23 19:15:11 mwester, I hope you haven't already done the fun of fixing apt-native/dpkg-native Feb 23 19:15:13 hrw: it broke SlugOS yesterday. :( Feb 23 19:15:19 I've got those things fixed, but haven't pushed yet Feb 23 19:15:31 been busy w/ installer folks and building a desk this am Feb 23 19:16:04 Tartarus: I've done nothing; I'm near to releasing a distro and have no time right now. I've just been discovering more and more problems with opkg as testing continues. Feb 23 19:16:10 mwester: regarding the core team decision that you mention, was the policy specifically that opkg should stay no matter what, or simply that the change in question should not be reverted? Feb 23 19:16:11 mwester, ah, ok Feb 23 19:16:18 mwester, yes, dpkg/apt stuff is a tiny bit bitrotted Feb 23 19:16:20 but not horriblt Feb 23 19:16:23 *horribly Feb 23 19:16:30 and I've got it fixed now Feb 23 19:16:37 only outstanding issue is meta-toolchain Feb 23 19:16:44 if it's just a question of parameterising the hard-coded references to opkg in package_ipk.bbclass and the like, it would surely be trivial to replace opkg with ${IPKG_EQUIVALENT} in those places. Feb 23 19:16:57 (and that package_deb.bbclass doesn't have the nice make a feed functions that package_ipk has) Feb 23 19:17:55 pb__: I would think so, but that didn't seem to be an option back when the issue first surfaced. I would have expected that someone would have suggested it, if it were so simple. Feb 23 19:18:09 from what I see in classes/rootfs_ipk.bbclass it should be possible to get ipkg instead of opkg via ONLINE_PACKAGE_MANAGEMENT variable Feb 23 19:18:19 Tartarus: ah right, yeah, the meta-toolchain stuff does look a bit suboptimal. Feb 23 19:18:35 ok, opkg is used to generate rootfs but it also adds links to make ipkg work with opkg generated rootfs Feb 23 19:18:55 But it doesn't matter right now; opkg is what is demanded by the current implementation, and if a lot of work has to be done, I think Tartarus's suggestion of dpkg is the better solution. Feb 23 19:19:29 heh.. I will check it Feb 23 19:20:00 mwester: Yes, that's probably true. But your earlier comment seemed to suggest that all non-opkg package managers were somehow forbidden by policy (rather than being simply unimplemented), and as far as I can tell that claim seems not to be correct. Feb 23 19:20:48 If you don't personally want to do the work to make ipkg work, that's obviously fine. But it's a bit of a stretch to say that "those in charge have determined" that one cannot. Feb 23 19:20:50 pb__: let's not get into a religious argument. Feb 23 19:21:28 pb__: "policy" is a non-argument -- Policy can be that monkeys are permitted to fly. "what works" is reality -- monkeys can't fly. Feb 23 19:22:22 pb__: I saved the email threads, including the private ones, but I'd really rather not go into that "war" all over again. it was made VERY clear to me that ipkg was NOT going to be used in OE again, and that if I wished to, I would do it myself on my own branch. Feb 23 19:23:10 Therefore, it seems very obvious that the policy-by-action of the OE core team is that opkg is the correct package manager to use. Feb 23 19:23:21 policy-by-words is just religion. Feb 23 19:23:25 mwester: well, the interesting question from my point of view is whether or not a change to make ipkg (or $other_package_manager) usable would be accepted into the oe tree or not, if someone were to write the patch. That is clearly a matter of policy. Feb 23 19:23:57 pb__ how about an even simpler patch -- one I suggested and was shot down? Feb 23 19:24:02 "git revert " Feb 23 19:24:44 * mwester is growing angry over this once again, and decides this is stupid and this discussion will end. Feb 23 19:24:55 mwester: would that reversion solve the issue of making the package manager configurable? Feb 23 19:25:04 I don't really know what the change is that you're talking about. Feb 23 19:25:19 pb__: If it makes you happy, then sure -- I retract everything I say that uses the word "policy" -- and pigs can fly too. Feb 23 19:25:28 If it would just swap back to having ipkg hard-wired then I guess that might be preferable to opkg, but it doesn't really seem like it would solve the underlying problem. Feb 23 19:25:59 mwester: I don't think you ever actually used the word "policy", that was me. Feb 23 19:26:22 Now, how about YOU make ME happy, and make ipkg work again? Then I will beleive it. Otherwise, it's a decision that was forced on all distros by the OE core team -- call it policy, or not, it was forced on all distros regardless. Feb 23 19:27:03 Well, I would be happy to have ipkg work again as well. Feb 23 19:27:09 mwester: at the time when we did ipkg->opkg switch ipkg was over year dead and opkg had paid developers Feb 23 19:27:14 I am perfectly happy to try making the change that I suggested earlier and see what happens. Feb 23 19:27:15 * mwester decides to go back to the incredibly difficult task of keeping a non-Angstrom distro working in OE, instead of arguing about unilateral edicts issued from on high. Feb 23 19:27:38 hrw: if someone had looked at the code, they would NOT have paid the developer! Feb 23 19:27:41 heh! Feb 23 19:27:54 mwester: ipkg code was already not nice too Feb 23 19:27:57 anyway, I'm done with this discussion. It changes nothing. Feb 23 19:28:02 was afk Feb 23 19:28:11 But, before trying to actually write the patch, I am trying to understand what exactly the core team did or didn't say to you before. Feb 23 19:28:13 mwester: and I think that doing release directly from .dev is not sane Feb 23 19:28:24 mwester: i know opkg is not that good, but foir now it seems i have to live with it Feb 23 19:28:34 I don't particularly want to spend time on reinventing something that has already been tried and failed. Feb 23 19:28:54 hrw, if opkg pays developers, they should pay good developers Feb 23 19:29:12 * eFfeM detected a few memory leaks within a minute after looking at the source Feb 23 19:30:23 afk again Feb 23 19:34:17 hrw, which distro to test mikey patch 'generic' ? Feb 23 19:34:46 all really Feb 23 19:34:51 mwester: actually, looking at the number of places that need to be parameterised, maybe it would be easier just to make a wrapper for ipkg to allow it to be called as "opkg". Would that work for you? Feb 23 19:35:13 I can't think of any other obvious ways in which the two of them are api-incompatible. Feb 23 19:36:08 gremlin[it]: generic probably Feb 23 19:36:23 gremlin[it]: minimal Feb 23 19:36:39 generic was renamed to minimal a while ago Feb 23 19:36:57 ok thanks all ! Feb 23 19:37:08 other distros should also still build :) Feb 23 19:37:16 heh, one would hope Feb 23 19:37:30 :) Feb 23 19:38:15 mickey|bbl, why does PR go backwards in http://cgit.openembedded.org/cgit.cgi?url=openembedded/commit/&h=mickey/sane-defaults-and-updates&id=ed1d18ed62f0f9a8cdb3d3daabd9e47cae410d40 Feb 23 19:41:56 ah, that's a mistake. originally i tried to update to a newer netbase, but none of it was an straightforward update Feb 23 19:42:02 good catch Feb 23 19:43:34 mickey|bbl: I think that our netbase is too much our now to be easy upgradable Feb 23 19:44:11 yep Feb 23 19:45:08 thanks :) Feb 23 19:45:22 I am trying to skim the changes Feb 23 19:46:19 hi all Feb 23 19:48:49 gm Feb 23 20:07:18 mwester: IPKG_VARIANT = "ipkg ipkg-collateral" in image recipe or distro config will give you image with ipkg instead of opkg Feb 23 20:07:23 mwester: just checked Feb 23 20:08:15 mwester: my image got opkg-collateral and ipkg-collateral inside but both are simple text packages and I think that you can manage to get rid of one of them Feb 23 20:08:49 mwester: all I needed to find it was looking at run.do_rootfs, rootfs_ipk.bbclass and angstrom distro config Feb 23 20:10:48 mwester: it was added by Koen on last Sunday^W^Wyesterday Feb 23 20:11:03 * hrw -> off Feb 23 20:26:19 i want to submit a new package, what would be the easiest way? or is there a guide for creating the git patch so that it can be easily applied? Feb 23 20:29:36 cpo__: send git format-patch to ML Feb 23 20:29:54 Jay7 how to create this git format patch? Feb 23 20:30:04 man git-format-patch :) Feb 23 20:31:09 http://git.or.cz/gitwiki/GitCheatSheet Feb 23 20:31:20 look here at 'Publish' Feb 23 20:32:09 kthx Feb 23 20:40:34 ok its out Feb 23 20:49:31 re Feb 23 20:50:48 hi florian Feb 23 20:53:06 * florian shoots his telephony provider - they sent a new IAD which asks for a PIN Feb 23 20:54:03 doh Feb 23 20:55:22 Stupid Alice - internet access works with a random DSL modem but for telephony there is a pile of strange things to do. Feb 23 20:55:47 florian hm? Feb 23 20:56:00 you mean the ne grps/utms flatrate? Feb 23 20:56:02 new Feb 23 20:56:48 woglinde: no... my Spairon IAD broke. Feb 23 20:57:46 woglinde: I got this new Fritzbox-like thingy as a replacement which now asks for a PIN. Feb 23 20:58:57 woglinde: I have no idea if this the the initial PIN for activating telephony or some PIN specific for the device. Feb 23 21:09:17 florian: call to support :) Feb 23 21:11:06 hm Feb 23 21:11:10 Jay7: i love their support... even it was 100% clear that the iad was broken they sent some technican for measurements Feb 23 21:11:12 never heard of that Feb 23 21:11:22 lol Feb 23 21:11:27 did you let him in Feb 23 21:11:49 hehe.. Feb 23 21:11:58 woglinde: yes... was a kind and fearless guy (everyone ill this day here) Feb 23 21:12:08 show me a good support service :) Feb 23 21:12:50 woglinde: he lookd at my replacement modem and the n810 connected and wondered what his customer wants him to do ;) Feb 23 21:23:19 hi, all! Feb 23 21:23:51 hello slapin Feb 23 21:24:18 could anybody recommend some code samples for 64-bit division using 32-bit instructions, so no undefined reference to `__umoddi3' Feb 23 21:24:28 and the like Feb 23 21:24:55 slapin: look at kernel Feb 23 21:25:12 do_div... Feb 23 21:29:11 hi khem Feb 23 21:29:36 woglinde: hello Feb 23 21:30:00 hi woglinde, khem, and all Feb 23 21:30:10 likewise: how are you long time Feb 23 21:35:58 khem`: bought a house, working on PCIe device driver and playing with CUDA/OpenCL programming. Feb 23 21:36:11 khem`: How're you doing? Feb 23 21:59:44 hi kergoth Feb 23 22:25:55 likewise: all is well. Feb 23 22:26:09 khem: good :-) Feb 23 22:26:11 its been raining here in CA for few days now Feb 23 22:26:19 every weekend :) Feb 23 22:26:21 khem: ah well not any worse than it is here Feb 23 22:26:43 yeah we needed this rain anyway otherwise we could have faced drought :) Feb 23 22:27:00 it will make the spring flourish again soon Feb 23 22:27:04 * khem remembers CUDA is the nvidia thingy Feb 23 22:27:34 khem: yeah... it's vendor-specific (yuk) but it's my fastest way to OpenCL Feb 23 22:28:05 isnt it for GPU programming Feb 23 22:30:52 khem: yup Feb 23 22:31:09 khem: OpenCL targets PowerPC Cell as well, or even FPGA, or ... Feb 23 22:31:32 khem: not that I suspect a good C->FPGA compiler in this era. Feb 23 22:31:46 sus->ex Feb 23 22:31:57 gotta go now (zzz) talk to you later Feb 23 22:32:26 cya all Feb 23 22:34:59 if i install dhclient, wireless-tools & wpa_supplicant into a embedded system will it need anything more to get a working network connection ? Feb 23 22:37:17 likewise: interesting. ttyl Feb 23 23:31:51 Whoo, more stupid questions. How to build x-load? It complains that my gcc doesn't know the -march armv7a ... Feb 23 23:33:19 gn all Feb 24 01:13:15 03Tom Rini  07org.openembedded.dev * re5b8428020 10openembedded.git/packages/apt/ (7 files): apt-native: Add perl-native to depends, change from += to =, bump PR Feb 24 01:13:27 03Tom Rini  07org.openembedded.dev * r072d455056 10openembedded.git/packages/dpkg/ (6 files): Feb 24 01:13:27 dpkg-native: Add dpkg-native.inc, set PERL_LIBDIR, fixup dpkg-architecture, bump PRs. Feb 24 01:13:27 Setting PERL_LIBDIR means that Dpkg.pm, etc, get put into the searchpath of Feb 24 01:13:27 perl-native now. Force dpkg-architecture to call our perl not host. We also Feb 24 01:13:29 didn't have a consistent set of DEPENDS but we should have. Feb 24 02:18:00 MostAwesomeDude: what machine did you build for ? Feb 24 02:29:14 khem`: Doesn't matter; turns out the hardware guys forgot to give me a few patches. Feb 24 02:30:03 hmm **** ENDING LOGGING AT Tue Feb 24 02:59:57 2009