**** BEGIN LOGGING AT Mon Sep 17 02:59:57 2007 Sep 17 04:22:56 anyone know if CONFIG_AEABI is needed for ixp425 ? Sep 17 04:33:43 ~seen mwester Sep 17 04:34:16 mwester is currently on #oe (21d 2h 34m 15s) #openmoko (21d 2h 34m 15s) #nslu2-linux (21d 2h 34m 15s) #openslug (21d 2h 34m 15s). Has said a total of 345 messages. Is idling for 4h 39m 48s, last said: 'Same way - understand and compare the dependencies, so that you can add a dependency on the ixp4xx modules into the ... Sep 17 06:32:19 sakoman: koen@lieve:/media/hda4/OE/build/tmp-new/angstrom/deploy/uclibc/ipk$ du usr/lib/libgd.so.2.0.0 -hs: 216K usr/lib/libgd.so.2.0.0 Sep 17 06:38:08 sakoman: so now we have to figure out why your gd is 1.1MB and mine 200k Sep 17 08:06:07 03rpurdie * r975 10bitbake/ (ChangeLog lib/bb/parse/parse_py/BBHandler.py): BBHandler: Don't add duplicates to BBPATH Sep 17 08:39:03 morning Sep 17 08:39:11 morning all Sep 17 08:39:34 good morning all Sep 17 08:40:37 morning folks Sep 17 09:00:24 rwhitby: have a look at gdk-pixbuf-csource-native_2.10.14.bb Sep 17 09:14:30 koen: ah, so it's just a missing dependency? Sep 17 09:14:34 thx Sep 17 09:15:32 it started building it for me yesterday, so at least one recipe depends on it :) Sep 17 09:16:12 bug #3000 was filed thanks to pixbug-native Sep 17 09:16:21 should that dependency be added to libwnck, or the gnome.bbclass? Sep 17 09:20:00 I'm looking for a very small embedded device for audio streaming. all it needs is cf slot or something for wifi or better onboard 802.11g and audio jack and it needs to be cheap and small ^^ Any idea? Sep 17 09:23:14 GTA02 :-D Sep 17 09:23:35 no it needs to live in my bathroom Sep 17 09:23:43 03koen 07org.oe.dev * ra7779f83... 10/ (1 packages/gtk+/gtk+.inc packages/gtk+/gtk+_2.12.0.bb): gtk+ 2.12.0: also build tiff pixbug loader Sep 17 09:23:43 and has to be cheap.. Sep 17 09:23:52 03rpurdie 07org.oe.dev * r4f00154c... 10/ (3 files in 3 dirs): pkgconfig-0.22: Add sysroot support patch from poky Sep 17 09:27:05 what about those gumstix boards? would they be any good for the intent? Sep 17 09:29:14 dcordes: http://corporate.reciva.com/products_and_technology/internet_radio_modules.asp Sep 17 09:30:33 thanks Sep 17 09:34:26 rwhitby: add it to gtk*.inc and bump PR for affected gtk packages Sep 17 09:35:46 XorA: any idea where I can order? Sep 17 09:36:16 dcordes: no idea, just happen to have a couple of the full devices in work Sep 17 09:36:32 dcordes: I think the whole thing + case is 100GBP in high street shops in UK Sep 17 09:37:11 hm I only need the internet radio board and possibly lcd Sep 17 09:40:13 I will make a case myself because it has to be resistant against high humdity rate Sep 17 09:47:27 * koen discovers OE doesn't have procmail Sep 17 09:49:50 any other idea for small audio streaming board? Sep 17 09:58:32 Gah, there are no longer direct flights Newcastle <-> Berlin :-( Sep 17 10:15:30 XorA: Do you have a direct email address for that hotel? Sep 17 10:16:08 RP: honigmond@t-online.de Sep 17 10:16:37 XorA: thanks. Anything in particular I need to mention to them when contacting them? Sep 17 10:16:51 RP: I just used the web form Sep 17 10:17:02 RP: Im in the Restaurant Hotel Sep 17 10:19:19 XorA: Thanks, I was just wondering what RH was :) Sep 17 10:25:06 * XorA digs into the imageon patch for kdrive Sep 17 11:25:39 i see there are gnome packages in oe. i can't see gnome-python packages though, is someone working on them? Sep 17 11:26:24 mickeyl: good morning! Sep 17 11:26:33 *yawn* Sep 17 11:26:34 morning koen Sep 17 11:27:08 morning it is Sep 17 11:28:02 (openmoko and gnome mobile would need gnome-python) Sep 17 11:29:39 RP, mickeyl: where should zaurus asound.state file be in? Sep 17 11:29:58 RP. mickeyl: I'd say alsa-state.bb, but it seems zaurusd also has its own Sep 17 11:37:03 koen: zaurusd did this long before alsa-state existed Sep 17 11:37:42 RP: the problem is that alsa-state exists now and includes state files for a few (but now all) zaurusses Sep 17 11:37:54 koen: zarusud makes certain assumptions about the state file and manipulates it so the zaurus ones should perhaps remain there, at least until I get time to sort things out properly Sep 17 11:38:01 RP: and I just discovered that OE disabled state handling in zaurusd :( Sep 17 11:39:11 mickeyl: have you seen http://bugs.webkit.org/show_bug.cgi?id=14998 ? Sep 17 11:39:42 RP: I don't really care where the statefiles go for zaurus machines, I just want #2667 fixed Sep 17 11:40:36 koen: I don't have time to look at it at the moment Sep 17 11:41:12 * nud wonders why there are stuffs like avahi on a console image Sep 17 11:41:14 yes, zecke told me about 14998 just yesterday Sep 17 11:42:54 nud: to have your devboards show up on mdns so you don't have to hunt dhcp logs for IPs Sep 17 11:43:36 03koen 07org.oe.dev * re14f4387... 10/ (5 files in 5 dirs): alsa-state: update c7x0 state file (from #2667), add akita one (from #2667) Sep 17 11:43:41 03koen 07org.oe.dev * re693b2bf... 10/ (3 files in 2 dirs): alsa: enable fixed-point PCM conversion for TARGET_FPU=soft Sep 17 11:54:39 huhu Sep 17 12:30:07 hi i get an error when building gst-plugins-base within my bootstrap image. its a custom distro with a debian buildhost latest version. OE gcc ist 4.1.2, binutils 2.17.50.05, glibc 2.5. here is a log http://pastebin.com/m428541eb Sep 17 12:30:16 anyone able to help please? Sep 17 12:30:37 it's the CROSS COMPILE Badness: /usr/include in INCLUDEPATH: /usr/include Sep 17 12:30:38 error Sep 17 12:35:11 you need to find out more what drags in your local headers Sep 17 12:35:22 do this by inspecting Makefile, Makefile.am, config.log etc. Sep 17 12:40:30 thought the same (Makefiles) in almost every Makefile there is an includedir = /usr/include Sep 17 12:40:43 but the gst-plugin-base package is the only one that's complaining Sep 17 12:41:11 should i point it from includedir = /usr/include to /mytoolchainpath/usr/include? Sep 17 12:44:55 the autotools generated do.configre file also does a --includedir=/usr/include Sep 17 12:45:02 now im getting confused Sep 17 13:05:30 where can I find a recipe for the squashfs kernel module? Sep 17 13:05:46 if any :) I saw one ipk of it in the angstrom repo Sep 17 13:06:18 dcordes: packages/linux/linux*.bb Sep 17 13:06:58 same is happening with libsdl_net by the way. Sep 17 13:07:17 nik0n: i'm having the same problem with libsdl_net Sep 17 13:08:53 koen|away: nothing squashish there Sep 17 13:09:39 so? Sep 17 13:09:47 don't know :) Sep 17 13:10:10 It's just that there is a kernel-module-squashfs in the angstrom repo and I take from it that there has to be a recipe Sep 17 13:10:29 yes, like I said: packages/linux/linux*.bb Sep 17 13:11:15 oh so it is in the kernel already? Sep 17 13:11:52 torpor: im searching over 8hours now but havent found anything. Sep 17 13:12:04 when i compile angstroem dist with same versions i use, it works fine Sep 17 13:12:14 with the same PREFERED_VERSION etc. Sep 17 13:14:42 koen|away: ? Sep 17 13:16:29 dcordes: if the defconfig enables it, yes Sep 17 13:17:02 so there is no seperate recipe? Sep 17 13:18:01 no Sep 17 13:20:37 why is it in the angstrom repo then? Sep 17 13:26:18 oh cbrake, you come just as I need you :p Sep 17 13:26:21 hello, how are you? Sep 17 13:28:53 more seriously, to anyone: I was wondering where you are supposed to configure your kernel options Sep 17 13:29:21 because machine features seems to be a somewhat limited way to do this Sep 17 13:29:38 dcordes: it's in the repo because linux*.bb builds it Sep 17 13:34:30 koen|away: thing is openmoko doesn't build it and I would like to install it Sep 17 13:34:57 15:14 < koen|away> dcordes: if the defconfig enables it, yes Sep 17 13:35:57 so the only way to build it for openmoko is to enabled squashfs in the defconfig? Does this create an ipkg of squashfs? Sep 17 13:36:05 or will it be in the kernel Sep 17 13:37:05 03mickeyl 07org.oe.dev * rb8e71a10... 10/ (5 files in 2 dirs): libwnck: remove old ones, add dependency to gdk-pixbuf-csource-native Sep 17 13:37:46 s/ipkg/ipk Sep 17 13:38:31 dcordes: s/openmoko/fic-gta01/ and yes Sep 17 13:39:16 03mickeyl 07org.oe.dev * rb8e71a10... 10/ (5 files in 2 dirs): libwnck: remove old ones, add dependency to gdk-pixbuf-csource-native Sep 17 13:40:29 koen|away: does it create an ipk? Sep 17 13:40:38 yes Sep 17 13:43:34 koen, what distro are you using with the N800? How is the Nokia boot loader to work with, is there any good docs for it? Sep 17 13:51:57 nud: hello! Sep 17 13:52:19 cbrake: hello! Sep 17 13:52:26 svolpe: still the nokia one, haven't had time to reflash Sep 17 13:52:54 cbrake: is your default kernel setup for cm-x270 targetted to what you do with it ? because it seems to be a bit light Sep 17 13:53:09 like: there is apparently no sound device Sep 17 13:53:12 koen|away, ahh, I saw that N800 was in OE and thought you were working on it. I wondered what distro (in OE) was best to use with it. Sep 17 13:53:25 koen|away, thanks anyway. Sep 17 13:53:35 nud: no, its just a random default -- I would be glad to expand it -- feel free to send me patches Sep 17 13:54:06 svolpe: the best distro is of course angstrom :) Sep 17 13:54:14 cbrake: ok as soon as I understand how I'm supposed to configure the kernel :p Sep 17 13:54:25 koen|away, that is what I have set right now. Sep 17 13:54:37 koen|away, 2008.1 Sep 17 13:54:50 nud: mostly done in the defconfig file. I usually do my kernel builds outside OE, and then when I get things working, add the defconfig, and patches to OE. Sep 17 13:55:59 nud: we may also want to include some kernel modules in the OS image by default, but we probably need to figure out how to tie that to machine/distro features Sep 17 13:56:18 cbrake: well, don't you have a defconfig that contains sound, usb and all ? ;-) Sep 17 13:56:44 does using the kernel uncompressor instead of the uboot one work for anyone? Sep 17 13:57:04 nud: yes I do -- let me pastebin it, and I'll add it at some point ... Sep 17 13:57:06 it hangs right after "done. booting the kernel" here Sep 17 13:58:02 cbrake: ah thanks Sep 17 14:00:00 nud: ftp://bec-systems.com/pub/defconfig Sep 17 14:00:33 nud: 2.6.23-rc6 has high res timer support for PXA270 :-) Sep 17 14:00:56 nud: not sure when 2.6.23 will be released ... Sep 17 14:01:03 cbrake: so the best way: copy everything from kernel/compulab-* into my local directory Sep 17 14:01:14 cbrake: I don't think I need high res timer at all ;-) Sep 17 14:02:56 mickeyl: zImages work for me on virtually any platform Sep 17 14:03:09 hmm yeah Sep 17 14:03:10 mickeyl: add the printascii() hack to see where it hangs Sep 17 14:03:11 uImage this is Sep 17 14:03:21 i suspect some problems with load/entry point Sep 17 14:03:30 but since i have 0 knowledge about uboot, thiis si hard to debug for me Sep 17 14:03:42 i'm afraid i need to go back to using the uboot uncompressor for now Sep 17 14:03:47 no time for experiments :/ Sep 17 14:03:57 or use printascii to find out where it hangs Sep 17 14:04:13 nud: I check the kernel out with git, branch to the release point, and then apply each patch and check it into git. Sep 17 14:04:25 koen|away: is that fast to enable? Sep 17 14:04:47 add 'printascii();' and recompile Sep 17 14:05:18 nud: you need to set up a few env variables to use the OE toolchain Sep 17 14:05:43 yep, ARCH, CC, LD Sep 17 14:05:46 koen|away: to kernelcommandline ? Sep 17 14:06:09 mickeyl: http://www.arm.linux.org.uk/mailinglists/faq.php Sep 17 14:06:27 nud: http://pastebin.ca/700699 Sep 17 14:13:45 hi all! Sep 17 14:14:03 I am currently having a weird issue with kdrive and have been looking around trying to figure it out. Sep 17 14:14:09 I ran xdpyinfo... Sep 17 14:14:37 and for the most part, everything looks ok, but I did notice that the largest cursor was set to 1024x768 (resolution of lcd) Sep 17 14:14:48 is that changeable in the kernel or in the kdrive settings? Sep 17 14:23:16 cbrake: is there something to do to have the kernel modules in the image ? Sep 17 14:23:53 nud: RDEPENDS the module package Sep 17 14:24:06 nud: or add it to your custom image file Sep 17 14:24:16 this isn't done by default ? Sep 17 14:24:21 nud: no Sep 17 14:24:35 because I just saw there is no module installed in my image (which could explain the lack of sound support) Sep 17 14:25:00 ok I will add the module package into my image RDEPENDS Sep 17 14:26:37 nud: I need to study the right way to do this first -- it is probably ok to add sound/usb modules for all machines. Sep 17 14:27:06 nud: err, for the compulab-pxa270 machine Sep 17 14:27:19 cbrake: yes they all have usb support Sep 17 14:27:46 and if they don't use it... well, I don't use framebuffer and such things :p Sep 17 14:28:36 nud: plenty of NAND flash ..., but we want to keep the kernel modules out of the NOR flash images -- that is the only concern really Sep 17 14:29:08 yes I see Sep 17 14:29:59 can't you do something like RDEPENDS_jffs2 ? Sep 17 14:31:46 koen: what image were you building for your 200K libgd experiment? uclibc or libc? Sep 17 14:32:17 sakoman: uclibc Sep 17 14:32:41 cbrake: never, ever RDEPEND on a kernel-module Sep 17 14:32:47 nud: there are a bunch of EXTRA_RDEPENDS variables, so may be something there -- needs a little more thought Sep 17 14:32:55 koen: did your image contain libiconv.so? Sep 17 14:33:20 sakoman: I didn't build an image containing GD Sep 17 14:33:27 koen: where do kernel modules get included then -- can you point me to an example to follow? Sep 17 14:33:41 cbrake: *RECOMMENDS Sep 17 14:33:57 koen: yes that makes sense Sep 17 14:33:59 cbrake: if you compile in a module you put in RDEPENDS, hell breaks loose Sep 17 14:34:07 koen: agreed Sep 17 14:34:53 koen: I've determined that my libgd is oversize because libiconv is being statically linked with it Sep 17 14:34:58 so if I wish to have my modules in my image I can just add IMAGE_INSTALL += "kernel-modules-2.6.22" ? Sep 17 14:35:32 sakoman: ouch Sep 17 14:36:22 nud: I usually explicitly pull in just the ones I need, as there are quite a few built. You look at that ipk and see if it has dependencies for all the other modules. Sep 17 14:36:38 No other package in my image uses libiconv. But as part of task-base it gets built so it is in staging. Config for libgd of course picks up the libiconv in staging and thats what causes the issue. Sep 17 14:37:03 sakoman: I wonder why it statically includes it Sep 17 14:38:11 koen: Good question! As an experiment I deleted libiconv.h in staging and rebuilt libgd. As expected 200K-ish. A bit smnaller than the buildroot version in fact! Sep 17 14:39:35 sakoman: does gd have a --without-iconv or --without-nls switch? Sep 17 14:39:41 koen: I also tried to alter the gd recipe a bit by including --without-libiconv-prefix but no joy Sep 17 14:40:47 koen: don't see a --without-nls option in configure Sep 17 14:41:06 * koen ponders about patching gd autofoo Sep 17 14:43:34 koen: for my application I also need to use --with-freetype and --without-jpeg Is there an OE approved way to do this since standard OE gd has different options? Sep 17 14:44:00 I think freetype can be enabled globally Sep 17 14:44:11 sakoman: copy the .bb file and exit EXTRA_OECONF in it (using bbcollections) Sep 17 14:44:26 and add freetype to DEPENDS Sep 17 14:44:32 nud: not an option at this point Sep 17 14:45:34 koen: and would the global enable override the explicit --without-freetype in the recipe? Sep 17 14:45:58 I meant removing that and adding freeptype to depends :) Sep 17 14:47:10 koen: ah, works for me. I always assume I can't necessarily cause the official recipe to change :-) Sep 17 14:52:39 is the current bbfile path available in an environment variable ? Sep 17 14:52:44 during build I mean Sep 17 14:53:29 I'd like to use files in the receipt path during IMAGE_PREPROCESS_COMMAND Sep 17 14:55:50 ok, found FILE_DIRNAME Sep 17 14:56:50 nud: I really want to solve that in a different way Sep 17 14:57:05 koen: solve what ? Sep 17 14:57:15 what you are trying to do Sep 17 14:57:56 well usually people add a bunch of commands in IMAGE_PREPROCESS_COMMAND Sep 17 14:58:10 but I've a script of a dozen lines Sep 17 14:58:23 (to edit a few config files in the image) Sep 17 14:58:57 so IMAGE_PREPROCESS_COMMAND += "IMAGE_ROOTFS=${IMAGE_ROOTFS} ${FILE_DIRNAME}/cet-vphone-image/tweak-config.sh;" seems to work quite well Sep 17 14:59:42 I didn't say it didn't work Sep 17 14:59:52 You really want to solve that in a different way Sep 17 15:00:01 koen: how would you do that ? Sep 17 15:00:31 not with a script that bypassed the package manager Sep 17 15:02:06 koen: I really don't want to patch every single package to customize the configuration Sep 17 15:04:29 koen: but if you have a cleaner idea on how to do it, I'm ok with cleaning up my system Sep 17 15:04:38 s/am ok/would be glad/ Sep 17 15:04:39 koen: in which syntax do I add squashfs module to /linux-openmoko/defconfig-2.6.22.5-fic-gta01? Sep 17 15:05:00 printascii() -> http://bugzilla.openmoko.org/cgi-bin/bugzilla/show_bug.cgi?id=874 Sep 17 15:05:56 or in more general terms- how do I add any module to a defconfig? Sep 17 15:06:54 dcordes: you edit the defconfig. Sep 17 15:07:06 and then rebuild the kernel Sep 17 15:07:43 mickeyl: the point is where in the defconfig is it added and in what syntax Sep 17 15:08:05 dcordes: aha. you probably want to use make menuconfig then Sep 17 15:08:09 and diff afterwards Sep 17 15:08:26 mickeyl: I added a comment Sep 17 15:08:32 thanks Sep 17 15:08:34 mickeyl: 99% sure it's a race Sep 17 15:09:35 ok, this is the place where it gets Laf0rge's task. Sep 17 15:09:38 * mickeyl forsees a jtag session :D Sep 17 15:09:56 mickeyl: make menuconfig? Sep 17 15:10:00 sakoman: http://www.flickr.com/photos/koenkooi/1396654535/ Sep 17 15:10:59 dcordes: make menuconfig. Sep 17 15:11:06 in the kernel directory Sep 17 15:11:13 you might want to add ARCH=arm as well Sep 17 15:12:36 mickeyl: I type ARCH=arm by reflex nowadays, which doesn't work for avr32 and bfin :) Sep 17 15:13:31 hehe Sep 17 15:13:55 koen: nice! great depth of field and really nicely framed Sep 17 15:14:48 mickeyl: I don't understand. what is make menuconfig? Sep 17 15:15:11 dcordes: before tweaking defconfig's you might want to learn about how to do that. make menuconfig is one option to do that Sep 17 15:15:51 koen: I like your portrait too. did you use the timer or did someone else take it? Sep 17 15:16:13 sakoman: my gf made that photo Sep 17 15:17:24 koen: she's good! Sep 17 15:17:30 mickeyl: I thought adding a module to the defconfig would be as easy as addinga single line Sep 17 15:18:03 dcordes: that depends. some configuration entries trigger others. i would rather go the menuconfig/diff route Sep 17 15:18:55 I tried running make menuconfig in packages/linux/linux-openmoko but doesn't work Sep 17 15:20:31 am I supposed to run it inside a folder with the kernel sources? Sep 17 15:21:08 yes Sep 17 15:21:16 ARCH=arm make menuconfig Sep 17 15:22:34 so I fetch the source, apply the changes I would like to make to the spare defconfig which is in the source and then diff it to the moko/gta01 defconfig? Sep 17 15:24:37 bitbake -c configure -b ; cd /work/-angstrom-// ; ARCH=arm make menuconfig Sep 17 15:27:59 koen: -c makes bitbake fetch,configure the kernel so I can edit the defconf, right? Sep 17 15:28:36 does it also patch so the result is my ready to use gta01 defconfig? Sep 17 15:35:56 dcordes: yes, configure runs after patching Sep 17 15:36:21 raster filed a cute bug Sep 17 15:36:45 so all tasks until with -c defined one are ran? Sep 17 15:37:09 yes Sep 17 15:39:24 koen: OE/build/tmp/openmoko/work/fic-gta01-angstrom-linux-gnueabi/linux-openmoko-2.6.22.5-moko11+svnr2937-r2/linux-2.6.22.5 that's where I want to make menuconfig, right? Sep 17 15:39:41 looks like it Sep 17 15:41:42 koen: it errors- could you take a look? http://www.pastebin.ca/700794 Sep 17 15:44:40 dcordes: apt-get install libncurses5-dev Sep 17 15:47:18 koen, mickeyl: I taught poky "bitbake virtual/kernel -c menuconfig" earlier on today ;-) Sep 17 15:47:30 :) Sep 17 15:47:54 If we make it copy the files back that would be a nice final touch :) Sep 17 15:48:12 OE already does quite a lot defconfig munching in linux.inc Sep 17 15:48:17 we could extend that :) Sep 17 15:48:19 * koen runs Sep 17 15:48:46 koen: muchos gracias seƱor Sep 17 15:48:56 de nada Sep 17 15:49:43 * koen makes a note to pack warm raingear for OEDEM Sep 17 15:51:00 XorA: you arrive the 5th, right? Sep 17 15:51:51 koen: any hint on how to add the module? Sep 17 15:52:10 device drivers -> filesystems -> misc IIRC Sep 17 15:53:13 RP: yay! Sep 17 15:55:16 there is no jfs support in the moko kernel by default? Sep 17 15:56:41 03mickeyl 07org.oe.dev * r98d3a190... 10/ (1 packages/linux/linux-openmoko-devel_svn.bb): Sep 17 15:56:43 linux-openmoko-devel: revert back to using the u-boot uncompressor, since Sep 17 15:56:45 the kernel uncompressor triggers a hang (see http://bugzilla.openmoko.org/cgi-bin/bugzilla/show_bug.cgi?id=874) Sep 17 15:57:36 could somebody give me a hint on where to find squashfs in menuconfig if any? Sep 17 15:57:47 i would look under the "file systems" directory Sep 17 15:57:50 koen: yeah Sep 17 15:58:01 if you can't find it, it may not be available in your particular kernel Sep 17 15:58:07 so you'd need to patch it in Sep 17 16:00:01 mickeyl: I can't see it in the gta01 2.6.22-5 kernel Sep 17 16:00:09 then it may be missing Sep 17 16:00:14 iirc squashfs is not upstream et Sep 17 16:00:15 yet Sep 17 16:01:07 Yes, its not upstream Sep 17 16:01:10 meaning it is not in the kernel.org kernel? Sep 17 16:02:46 03mickeyl 07org.oe.dev * r98d3a190... 10/ (1 packages/linux/linux-openmoko-devel_svn.bb): Sep 17 16:02:50 linux-openmoko-devel: revert back to using the u-boot uncompressor, since Sep 17 16:02:52 the kernel uncompressor triggers a hang (see http://bugzilla.openmoko.org/cgi-bin/bugzilla/show_bug.cgi?id=874) Sep 17 16:03:08 cia is broken again... Sep 17 16:03:38 ok stupid question. next one: how can I "patch it in"? Sep 17 16:05:03 03mickeyl 07org.oe.dev * r98d3a190... 10/ (1 packages/linux/linux-openmoko-devel_svn.bb): Sep 17 16:05:03 linux-openmoko-devel: revert back to using the u-boot uncompressor, since Sep 17 16:05:03 the kernel uncompressor triggers a hang (see http://bugzilla.openmoko.org/cgi-bin/bugzilla/show_bug.cgi?id=874) Sep 17 16:05:15 koen, any idea who has done the porting work for the N800 into OE? I checked the MAINAINERS list but there is nothing there for the nokia800? Sep 17 16:05:42 svolpe: almost all of it is done by o-hand Sep 17 16:06:26 03daniel 07org.oe.dev * r4291007b... 10/ (3 files in 3 dirs): Sep 17 16:06:26 matchbox-keyboard-inputmethod: Add a patch to delay mb_kbd_ui_hide. Sep 17 16:06:26 This reduces flicker (i.e. when switching tabs on openmoko-terminal2) Sep 17 16:06:42 koen, o-hand? Sep 17 16:07:06 koen, ahh, just googled them :-) Sep 17 16:07:23 svolpe: in other words: RP :) Sep 17 16:08:01 mickeyl: + file://printascii.patch;patch=1 \ Sep 17 16:12:18 I have the squashfs diff file. To what do I apply it in order to enable squashfs support? Sep 17 16:13:23 koen: aarghs Sep 17 16:15:42 :( Sep 17 16:26:05 03mickeyl 07org.oe.dev * rd1d4ee2c... 10/ (3 files in 3 dirs): linux-openmoko-devel: add printascii patch, disable it Sep 17 16:34:14 I just tried building gst-plugins-base_0.10.14.bb, it built after I fixed a couple of things, which I will push soon, but I also noticed that it did not build any of its dependancies, what would cause that? Is there a variable that might be set that disables dependancies? This is a plug-in so I could see were that might be the case.. Sep 17 16:37:12 maybe it is related to this? http://permalink.gmane.org/gmane.comp.handhelds.openembedded/15882 Sep 17 16:37:43 although that seems like a different issue Sep 17 16:40:47 who is the maintainer of gstreamer? Sep 17 16:54:43 any mdio manager ? Sep 17 17:19:05 * Crofton checks flights to Berlin again .... Sep 17 17:22:31 still just under $800 Sep 17 17:27:16 Crofton : Its 900Euro from Athens to Austin :( Sep 17 17:27:19 * Crofton stops dreaming and looks at the paper that must be finished before he works on the thesis again Sep 17 17:27:27 urg Sep 17 17:27:44 still, it was $1200+ from here to Munich in June Sep 17 17:28:32 Crofton : well, the same ticket, same company was 1200E a couple of weeks ago.... Sep 17 17:28:38 :) Sep 17 17:29:03 723 for the same timeframe as OEDEM Sep 17 17:29:16 Crofton : maybe US come closer to Europe in the meanwhile due to tectonic plates.... Sep 17 17:30:14 I think we are moving apart ... Sep 17 17:30:24 Reunite Pangea Sep 17 17:30:32 hehe Sep 17 17:35:45 steliosk, should we use edk 8.2? Sep 17 17:35:57 we have ise 9.?, but no edk for it Sep 17 17:36:02 this is what we are using Sep 17 17:36:08 8.2? Sep 17 17:36:19 8.2 something let me check Sep 17 17:36:25 I would assume major version needs to match at least Sep 17 17:36:42 * Crofton needs to teach thesheriff to ask his own questions .... Sep 17 17:36:58 8.2.02i Sep 17 17:37:12 thanks Sep 17 17:37:27 get the service packs also Sep 17 17:52:27 any clue if a marvel switch chip should work on the ixp4xx platform.. or is the npe driver not right for this Sep 17 17:52:50 the network mdio is going up and down Sep 17 18:19:51 gm Sep 17 18:44:58 03mickeyl 07org.oe.dev * r50f023d7... 10/ (1 packages/gtk+/gdk-pixbuf-csource-native_2.10.14.bb): gdk-pixbuf-csource-native: DEPENDS on libpng-native and jpeg-native Sep 17 19:02:05 03mickeyl 07org.oe.dev * rbbe95c67... 10/ (1 packages/openmoko2/openmoko-sound-system2_0.1.0.bb): openmoko-sound-system2: rreplace openmoko-sound-system Sep 17 19:02:12 03mickeyl 07org.oe.dev * r334f1ee8... 10/ (1 packages/openmoko2/openmoko-sound-theme-standard2_svn.bb): openmoko-sound-theme-standard2: rreplaces openmoko-sound-theme Sep 17 19:16:53 hello - can anyone tell me where bitbake logs do_fetch commands? Sep 17 19:21:38 koen: hey, what do you plan todo tomorrow Sep 17 19:23:15 03crofton 07org.oe.dev * r3338e793... 10/ (1 conf/machine/omap5912osk.conf): omap5912osk : Add usb module to image. Sep 17 19:23:31 03crofton 07org.oe.dev * re9a6786e... 10/ (1 conf/machine/include/gumstix.inc): Sep 17 19:23:31 gumstix.inc : autoload smc91x module. Provide a commented out example of how Sep 17 19:23:31 to install all modules in the image. Sep 17 19:58:49 can someone tell me the difference between xserver-kdrive and xserver-kdrive-fbdev ? Are they needed in conjunction or one or the other? Sep 17 20:02:06 somebody knows a way to prevent gpsdrive's window from jumping around in vga? Sep 17 20:31:26 christopher: xserver-kdrive supports different hardware, the fbdev version supports the linux framebuffer. xserver-kdrive is probably a common package shared between the servers Sep 17 20:53:04 hmm Sep 17 20:53:05 *sigh* Sep 17 20:53:20 * mickeyl realizes after building for some hours that setting ARM_ABI has no effect Sep 17 20:53:32 so much for my oabi build Sep 17 20:54:04 ~curse angstrom-glibc.inc Sep 17 20:54:05 May you be reincarnated as a Windows XP administrator, angstrom-glibc.inc ! Sep 17 20:56:37 RP: ok. right now, I am just using a default Framebuffer, so I think all I would want is the fbdev version. Sep 17 20:57:13 RP: I have that in my image file for the XSERVER, but the distro (angstrom), shows it as xserver-kdrive... Sep 17 20:57:25 RP: and when I look at what is installed via the package manager, both are. Sep 17 20:58:00 RP: so, I am just not sure what is needed etc. I am seeing some weird stuff with some of the colors and use of the mouse so i was trying to track down what the issue was and what configuration was being used. Sep 17 21:01:37 mickeyl: angstrom-2007.1-legacy.conf Sep 17 21:02:27 hi gang - i'm entering "embedded linux with a screen" for the first time. Should I consider Qtopia Platform? Sep 17 21:03:05 of course Sep 17 21:03:13 HopsNBarley: depends on what you want to do... But it is an option. Sep 17 21:03:51 the device is a casually interacted with NAS, essentially. but it does "other stuff". not a handheld. Sep 17 21:08:25 XorA: yeah, i did include that. but TARGET_OS is silently overriding my ARM_ABI Sep 17 21:08:30 is it possible to use GPE with kdrive and xserver-nodm? Sep 17 21:08:41 currently there is no way to build OABI for a machine that's included in the hardcoded list :/ Sep 17 21:08:47 I don't like the gpe login Sep 17 21:08:53 yeah, Qtopia is not bad Sep 17 21:09:03 but there are alternatives Sep 17 21:09:15 such as X11+ Sep 17 21:09:28 depends on what you need Sep 17 21:09:34 hehe Sep 17 21:09:47 today is diplomatic day, right? Sep 17 21:09:48 :D Sep 17 21:09:49 * XorA is still stuck in a 2008.1 build Sep 17 21:09:55 * zecke is playing the new moko click sound to threaten mickeyl Sep 17 21:10:17 heh Sep 17 21:10:20 * XorA debates dunking his CPU in liquid notrogen and overclocking it Sep 17 21:10:26 dcordes: should be possible yes Sep 17 21:10:49 ftp://ftp.trolltech.com/qtopia/tech-preview/qtopia-opensource-src-4.3.0-preview.tar.gz Sep 17 21:11:01 hey! Sep 17 21:11:04 new stuff Sep 17 21:11:06 how much stuff is GPL? Sep 17 21:11:16 :) all of it Sep 17 21:11:16 oh, well it is a tech-preview... Sep 17 21:11:26 but how much source is contained? Sep 17 21:11:34 all of it Sep 17 21:11:45 hmm, fun, pkg-config-native 0.22 doesn't build here Sep 17 21:11:49 * XorA detects a time loop here Sep 17 21:11:53 ljp: hehe, so some da soon I will have to do the Qt4 dance? Sep 17 21:12:08 zecke: so I hear Sep 17 21:12:21 | main.c:428: error: 'VERSION' undeclared (first use in this function) Sep 17 21:13:04 ljp: congratulation! Sep 17 21:13:22 thanks Sep 17 21:13:53 98mb... so the buildsystem is still not optimal Sep 17 21:14:04 theres more announcements soon Sep 17 21:14:43 depends on what you mean optimal., it contains everything you need to build qtopia Sep 17 21:14:49 thanks guys. Sep 17 21:15:06 ljp: everything? compilers, libc? Sep 17 21:15:12 HopsNBarley: good luck with it. Sep 17 21:15:22 And don't forget to use Openembedded Sep 17 21:15:28 don't worry (-; Sep 17 21:19:50 no toolchains but you donthave to find 15 different packages on 15 different servers just to build it Sep 17 21:21:14 ljp: so Qtopia is now a Toolkit? Sep 17 21:21:43 qt is a toolkit Sep 17 21:22:46 hehe, the license header says "Qtopia Toolkit" Sep 17 21:24:55 hmmm Sep 17 21:26:19 qtopia core is a toolkit Sep 17 21:27:14 ljp: I can only say what I read... Sep 17 21:27:29 ljp: and it is under GPL already, I assumed it would be under the TP license Sep 17 21:28:41 wait 8.5 hours for the announcement Sep 17 21:29:20 at 1:00am EST? Sep 17 21:29:25 err EDT Sep 17 21:30:06 * zecke is sitting in mickeyls "something like a cellar" and is poking around stuff Sep 17 21:30:13 8 am oslo time Sep 17 21:30:21 2am here Sep 17 21:30:21 hah Sep 17 21:30:28 office... read my lips.. O F F I C E Sep 17 21:30:40 and now let's go for a wlk Sep 17 21:30:43 walk, even Sep 17 21:32:29 oh, thats zecke for the punishment :-) Sep 17 21:33:09 * Philippe puts cellar + punishment together... Sep 17 21:33:18 working for mickeyl must be dangerous :-) Sep 17 21:34:45 * XorA hides Sep 17 21:35:26 Hi all, we are in trouble with dirty staging direcoty and the bitbake -c clean is not cleaning it... Does any body have any idea how to clean it? we are thinking to implement a kind of log of installed files of each .bb in the staging to remove it... what do you think? ( zecke koen XorA RP ??? any idea? ) Sep 17 21:36:16 vivijim: finish packaged staging, there is a patch in bugtracker that gets it nearer completion Sep 17 21:36:32 vivijim: then all files in staging are under package manager control Sep 17 21:37:06 vivijim: http://bugs.openembedded.org/show_bug.cgi?id=2948 Sep 17 21:40:20 XorA: it is using a sync with ipkg database isn't it? because we are not using ipkg packages, just the deb ones... Sep 17 21:40:40 vivijim: that was one of the fixes koen mentioned needed doing Sep 17 21:49:44 vivijim: As XorA said, packages staging got so far but finishing Sep 17 21:49:50 but needs finishing Sep 17 21:52:50 RP: I guess with your sysroot staging thoughts packaged staging might actually start to use -dev packages? Sep 17 21:53:30 XorA: I'm currently still of the opinion staging should be separate packages tbh Sep 17 21:53:53 XorA: Maybe in time, yes, who knows... Sep 17 21:54:34 RP: suppose it depends who finishes it :-) Sep 17 21:54:52 XorA: yes :) Sep 17 21:57:05 re Sep 17 21:57:28 hi zecke Sep 17 22:04:13 hi, i have a strange problem, anytime when a patch for a package should be applied, i get an quilt error? Anybody an idea? Sep 17 22:05:47 ljp: so there is more to comeĆ Sep 17 22:06:56 ljp: I wonder if people do a ls in devices/ ... Sep 17 22:08:53 * kergoth thinks sysroot style staging is just adding more complexity to the layout for something that's largely a convenience gain... but he's biased, since he decided to do it the way it is right now rather than as a filesystem layout, back in the day Sep 17 22:10:05 kergoth: I have really mixed feelings about this Sep 17 22:12:21 kergoth: Part of me doesn't like that we configure with "libdir=/usr/lib", then install into something /lib. If that causes real problems for programs, you have to blame the build system really... Sep 17 22:14:19 if a buildsystem depends on a certain layout, rather than accepting seperate arguments for locating libs vs headers vs binaries, then that's a flaw in the buildsystem of the package that needs those, imo. with regard to bininstall/etc, i'd rather see that go away in favor of a manifest or FILES type patterns defining what files from the install phase go into staging packages Sep 17 22:14:49 course i havent touched OE in like a year, so take what i say with a grain of salt :) Sep 17 22:15:16 * kergoth rereads the mail laying out the pkgconfig issue(s) Sep 17 22:18:24 kergoth: The problem thats making me rethink things is pkgconfig. Taking dbus as an example it adds session_bus_services_dir=/media/data1/builds/poky/eabi/tmp/staging/arm-poky-linux-gnueabi/share/dbus-1/services to the .pc file (with our current pkgconfig hacking) Sep 17 22:19:22 ah, and that path is getting compiled into apps using dbus? Sep 17 22:19:27 kergoth: Or worse still, extensiondir=${libdir}/evolution-data-server-1.2/extensions from eds Sep 17 22:19:33 kergoth: right Sep 17 22:19:37 that's ugly Sep 17 22:19:38 :( Sep 17 22:21:34 kergoth: I don't know what to suggest other than a sysroot solution to solve it Sep 17 22:22:43 kergoth: The ${libdir} in extentiondir is particularly nasty since we have to change libdir to make pkgconfig give sane compiling values. Sep 17 22:24:28 how's the sysroot pkgconfig option work, how does it know to prepend the sysroot for requests for libdir, but not for extensiondir? Sep 17 22:24:45 kergoth: It only does it when returning -L and -I options Sep 17 22:25:00 ah. Sep 17 22:25:40 kergoth: I tried to shoot it down in flames but the builds I tried using it worked rather too smoothly ;-) Sep 17 22:26:12 well, it's uglier, but we could just patch in our s/// from target paths to staging paths into pkgconfig for -L and -I. not as clean code wise as a simple prepend, but.. Sep 17 22:26:14 hehe Sep 17 22:28:51 kergoth: Thats another option. It just feels a lot uglier Sep 17 22:29:08 kergoth: We have the same kind of problem with the .la files too Sep 17 22:29:45 having multiple libdirs and target fs style layout for an area that doesnt need them is ugly too, just in a different way :) Sep 17 22:32:43 kergoth: There are other fringe ideas that could sway the balance. Using qemu scratchbox style for configure for example... Sep 17 22:33:02 and packages staging if people want to reuse -dev packages Sep 17 22:33:10 packaged Sep 17 22:34:51 kergoth: Having experimented switching poky to use target fs layout for staging, it is surprisingly more complex I admit... Sep 17 22:35:48 What really annoys me is the recipes all make hardcoded assumptions about staging though :) Sep 17 22:36:05 well, if it fixes that many issues, i guess it'd be worth it. we certainly wouldnt be the first buildsystem to use target layout for shared stuff when crosscompiling. Sep 17 22:36:28 heh, yeah, i tried to fix that a long time ago, i ran a world build overriding every path variable, prefix, libdir, etc.. and watched things go explodey Sep 17 22:36:43 one of those things that should probably be tested periodically by an automated build Sep 17 22:37:35 kergoth: The target layout assumptions are one thing, the staging layout ones are much more engrained again :/ Sep 17 22:38:01 We've made a lot of progress with target layout assumptions over time, its better than it used to be Sep 17 22:38:18 ah, thats cool Sep 17 22:42:49 My main annoyance atm is that I can't rely on "libdir" to return "/usr/lib" for use in STAGING_DIR* variables since native.bbclass sets prefix to STAGING_DIR :/ Sep 17 22:45:06 ick Sep 17 22:47:56 layout_exec_prefix = "/usr" layout_libdir = "${layout_exec_prefix}/lib" export libdir = "${layout_libdir}" seems a bit like overkill... Sep 17 22:50:26 kergoth: While you're awake, whats the historic reason for cross being separate to staging? Lack of multimachine? Sep 17 22:52:06 i don't recall. i'm thinking perhaps not all packages prefixed their binary names with the target prefix when they were built targeting that, or their 'share' data was target specific but unprefixed, so the overlap with -native was problematic, or something along those lines Sep 17 22:52:25 so yeah, i think you're right, the lacko fm ultimachine :) Sep 17 22:52:31 been a long time Sep 17 22:52:34 heh Sep 17 22:52:36 no kidding Sep 17 22:52:47 4 years and rocking... Sep 17 22:54:49 kergoth: I split staging binaries into three camps recently, native, cross and target with path set appropriately so if both the cross and native "freetype-config" was present, it gets the right one. Based on that I'd guess cross could be merged now Sep 17 22:55:00 yep, i'd say so Sep 17 22:55:30 I ask since currently we install glibc twice, once into cross, once into staging and it takes an ice age. Having to install once would be a good speedup Sep 17 22:55:42 ah, no doubt Sep 17 22:56:21 By the sounds of it, nobody would object to the removal of cross so I might try it at some point :) Sep 17 22:57:31 well, what % of the oe userbase even messes with -cross packages.. probably a small group thatd be objecting even if something was wrong :) Sep 17 22:58:45 how are things at ohand nowadays? Sep 17 22:59:40 kergoth: I keep wondering how long it would take for someone to notice if I proposed something totally hatstand on the bitbake or oe dev list :) Sep 17 23:00:09 kergoth: ok thanks. Somehow it manages to keep me rather busy :) Sep 17 23:00:21 hehe Sep 17 23:00:24 i'm not surprised Sep 17 23:01:14 03mickeyl 07org.oe.dev * rcc4ff44a... 10/ (3 files in 2 dirs): uboot-openmoko: set localversion to PV Sep 17 23:02:07 i need a device, phone or pda or micro laptop, to manage my damn todo list. what are the best contenders for that nowadays? havent touched embedded devices in a while. saw sony's micro laptop thingy, that was cute, but pricey Sep 17 23:03:46 if you'd ask in 1 year, i'd recommend a Neo1973... Sep 17 23:04:31 hehe Sep 17 23:05:33 i do some working from home nowadays and some working in the office, so todo lists are a bit of an issue, as are alarms and calendar.. need something thatll yell at me no matter where i am :) Sep 17 23:05:35 kergoth: a palm? Sep 17 23:08:35 kergoth: I hack on my devices too much to leave any of them usable :/ Sep 17 23:08:50 hehe Sep 17 23:09:29 i hack less nowadays, kinda just want things to -work- Sep 17 23:14:18 * RP knows the feeling Sep 17 23:15:44 koen: ping? Sep 17 23:22:08 kergoth: nokia n800 is nice hardware. no phone though Sep 17 23:24:01 honestly, i'm tempted to ditch my cell entirely and get a land line. 90% of the time what people want to talk to me about can wait till i get home or to work :P Sep 17 23:24:59 heh, yeah Sep 17 23:25:48 * RP keeps him mobile switched off. Its strictly for my convinience, not other peoples ;-) (unless I'm travelling which is different) Sep 17 23:29:47 kergoth, just be selective Sep 17 23:30:39 RP: hehe, I have it on but I don't hear it anyway... Sep 17 23:30:46 RP: and people can use jabber... Sep 17 23:31:39 * Crofton|home is seriously pissed off at people he has to "work" with Sep 17 23:36:03 mithro: hi! have fun at your job Sep 17 23:36:28 zecke: cool thanks :P Sep 17 23:36:34 zecke: did I tell you what they do? Sep 17 23:36:42 oh, I also work with rwhitby :) Sep 17 23:36:48 mithro: no, you didn't Sep 17 23:36:51 wow! Sep 17 23:37:22 mithro: I surfed to the astc site and it looked like hardware building Sep 17 23:38:02 zecke: they build simulators is guess is the three word description Sep 17 23:59:23 s/is/I/ Sep 18 00:15:11 I built gst-plugins-base_0.10.14.bb it built fine but it did not build any of its DEPENDS. I can build the DEPENDS manually and they work. Other packages with DEPENDS work just fine. What could be wrong with gst-plugins-base? Sep 18 00:15:50 did it not build them, or did it not package them? Sep 18 00:16:19 rwhitby, it did not build them. Sep 18 00:17:39 gst-plugins.inc: DEPENDS = "gstreamer libmikmod libmad liboil libogg tremor libvorbis libid3tag esound" Sep 18 00:17:39 none of those packages built or even tried to build. Sep 18 00:18:09 if I do bitbake gstreamer (or any other package in the list) they build just fine. Sep 18 00:28:58 svolpe: does bitbake -g tell you the same thing? Sep 18 00:48:12 The following is in two of the .inc files called for gst-plugins-base: #OE_LT_RPATH_ALLOW = "any" Sep 18 00:48:12 #OE_LT_RPATH_ALLOW[export]="1" Sep 18 00:48:34 with out the #'s as I commented them out. Sep 18 00:48:43 once I commented them out then the DEPENDS built. Sep 18 00:48:52 what is OE_LT_RPATH_ALLOW = "any" Sep 18 00:48:53 and Sep 18 00:48:59 OE_LT_RPATH_ALLOW[export]="1" Sep 18 01:02:00 03mickeyl 07org.oe.dev * rab258c42... 10/ (1 conf/distro/include/moko-autorev.inc): moko-autorev.inc: add missing packages Sep 18 01:02:05 03mickeyl 07org.oe.dev * r3ecc9177... 10/ (4 files in 3 dirs): linux-openmoko-devel: add squashfs 3.2 patch and enable in defconfig as module Sep 18 01:02:16 03daniel 07org.oe.dev * r4136e006... 10/ (1 conf/distro/include/sane-srcrevs.inc): matchbox-keyboard: Update sane-srcrevs.inc for improved handling of shift. Sep 18 02:20:57 03daniel 07org.oe.dev * re456a6f4... 10/ (5 files in 4 dirs): navit: Add Navit, the car navigation system with routing engine. Sep 18 02:48:28 Crofton: ping! **** ENDING LOGGING AT Tue Sep 18 02:59:59 2007