**** BEGIN LOGGING AT Tue Sep 15 02:59:57 2009 Sep 15 06:02:25 03Angus Ainslie  07shr/import * r5384c4ad8e 10openembedded.git/recipes/openmoko-projects/paroli_git.bb: Sep 15 06:02:25 paroli: mostly fixed paroli install Sep 15 06:02:25 sound is missing during calls probably due to missed files Sep 15 06:02:25 paroli-calculator gets installed in the base install Sep 15 06:02:35 03Angus Ainslie  07shr/import * rb1b6257b1b 10openembedded.git/conf/distro/include/shr-autorev.inc: shr-autorev : autorev paroli Sep 15 06:14:40 names Sep 15 06:16:05 anybody who handles projects.linuxtogo.org hanging around here ? Sep 15 06:16:21 the downloads are all empty tar files Sep 15 06:16:35 mickey|zzZZzz: wake up :) Sep 15 06:36:22 03Khem Raj  07org.openembedded.dev * r2aaa77d10a 10openembedded.git/recipes/xorg-xserver/ (3 files in 2 dirs): (log message trimmed) Sep 15 06:36:22 xserver-xorg-1.5.3: Fix the build with nptl Sep 15 06:36:22 * These two patches are ported from Sep 15 06:36:22 http://bugs.gentoo.org/show_bug.cgi?id=263679#c13 Sep 15 06:36:22 it fixes the following error. Sep 15 06:36:24 ../../glx/.libs/libglx.a(glapi.o): In function `_glapi_set_dispatch': Sep 15 06:36:28 glapi.c:(.text+0x117ac): undefined reference to `pthread_once' Sep 15 06:37:37 good morning Sep 15 06:38:34 03Khem Raj  07org.openembedded.dev * r7c8f742ecc 10openembedded.git/classes/cross.bbclass: Sep 15 06:38:34 cross.bbclass: Override BASEPKG_HOST_SYS Sep 15 06:38:34 * BASEPKG_HOST_SYS from bitbake.conf is set as if packages Sep 15 06:38:34 were being built for target. So we need to override it Sep 15 06:38:34 for cross packages. Sep 15 06:38:36 Signed-off-by: Khem Raj Sep 15 06:38:40 Acked-by: Andrea Adami gmail.com> Sep 15 06:39:16 03Khem Raj  07org.openembedded.dev * rb230286329 10openembedded.git/recipes/eglibc/ (eglibc_2.10.bb eglibc_2.9.bb eglibc_svn.bb): Sep 15 06:39:16 eglibc: Bump the SRCREV to get new fixes. Sep 15 06:39:16 Signed-off-by: Khem Raj Sep 15 07:33:04 * * OE Bug 5318 has been RESOLVED (FIXED) by raj.khem(AT)gmail.com Sep 15 07:33:06 * * unable to bitbake ortp Sep 15 07:33:08 * * http://bugs.openembedded.net/show_bug.cgi?id=5318 Sep 15 07:35:04 * * OE Bug 5315 has been RESOLVED (FIXED) by raj.khem(AT)gmail.com Sep 15 07:35:07 * * helloworld-image fails on 'opkg_unarchive.patch' Sep 15 07:35:08 * * http://bugs.openembedded.net/show_bug.cgi?id=5315 Sep 15 08:15:41 03Khem Raj  07org.openembedded.dev * r2bc7690bb2 10openembedded.git/ (4 files in 2 dirs): (log message trimmed) Sep 15 08:15:41 libcgicc: Add new recipe for 3.2.8 release Sep 15 08:15:41 * Move common parts to libcgicc.inc Sep 15 08:15:41 * Add new recipe for 3.2.8 release Sep 15 08:15:42 * Fixes bug 5314 Sep 15 08:15:44 * Thanks to Jim Sheldon for reporting and providing Sep 15 08:15:48 a recipe for 3.2.7 release. Unfortunately that Sep 15 08:43:16 03Khem Raj  07org.openembedded.dev * rfbe1669439 10openembedded.git/ (conf/checksums.ini recipes/vagalume/vagalume_0.7.bb): Sep 15 08:43:16 vagalume: Rename vagalume_0.7.bb as vagalume_0.7.1.bb Sep 15 08:43:16 * Fixes PR 5146 Sep 15 08:43:16 * Credit to bugs@vdm-design.de for reporting and posting Sep 15 08:43:17 recipe Sep 15 08:43:19 Signed-off-by: Khem Raj Sep 15 09:06:05 * * OE Bug 5314 has been RESOLVED (FIXED) by raj.khem(AT)gmail.com Sep 15 09:06:07 * * libcgicc 3.2.7 recipe Sep 15 09:06:09 * * http://bugs.openembedded.net/show_bug.cgi?id=5314 Sep 15 09:17:55 03Klaus Kurzmann  07shr/import * r10da2820e8 10openembedded.git/conf/distro/include/shr-autorev.inc: Sep 15 09:17:55 shr-autorev.inc: update openmoko kernel for testing Sep 15 09:17:55 Signed-off-by: Klaus Kurzmann Sep 15 09:17:55 03Klaus Kurzmann  07shr/import * rceb0a43066 10openembedded.git/: Merge branch 'shr/import' of ssh+git://git@git.openembedded.net/openembedded into shr/import Sep 15 09:32:36 I am tryng to output the stream of my webcam to an URL... is it more cpu consuming to encode from RAW to MP4 than from MJPEG to MP4? is the difference that great? Sep 15 09:39:04 * * OE Bug 5146 has been RESOLVED (FIXED) by raj.khem(AT)gmail.com Sep 15 09:39:06 * * Update Vagalume to Version 0.7.1 Sep 15 09:39:08 * * http://bugs.openembedded.net/show_bug.cgi?id=5146 Sep 15 09:47:30 03Klaus Kurzmann  07shr/import * r00f98412c4 10openembedded.git/conf/distro/include/shr-autorev.inc: Sep 15 09:47:30 shr-autorev.inc: set advancedcaching to AUTOREV for testing Sep 15 09:47:30 Signed-off-by: Klaus Kurzmann Sep 15 09:51:38 03Graeme Gregory  07org.openembedded.dev * rf111ae953c 10openembedded.git/ (3 files in 3 dirs): Sep 15 09:51:38 linux-libc-headers_2.6.31.bb : add newest kernel headers, this has the Sep 15 09:51:38 0001-implement-TIF_RESTORE_SIGMASK-support-and-enable-the.patch enabled Sep 15 09:51:38 which is getting to be needed for udev. Sep 15 09:51:40 DEF_PREF = -1 for now as I havent tested this properly yet Sep 15 09:53:11 03Graeme Gregory  07org.openembedded.dev * rdd6f2e9c43 10openembedded.git/recipes/transmission/ (4 files): transmission : make use .inc file as all recipes are pretty identical. Sep 15 10:10:13 03Martin Jansa  07shr/import * ra7914014d1 10openembedded.git/recipes/openmoko-projects/paroli_git.bb: Paroli, fix typo and include icon from icons directory Sep 15 10:11:57 hi, i get: install: cannot stat `/home/kimrhh/OE/tmp/work/armv7a-angstrom-linux-gnueabi/glibc-2.9-r35.2/glibc-2.9/sunrpc/rpcsvc/bootparam_prot.h': No such file or directory Sep 15 10:12:05 trying to build console-image Sep 15 10:12:26 from origin/org.openembedded.dev Sep 15 10:12:49 kimrhh: use DISTRO=minimal and rebuild everything Sep 15 10:14:29 khem: ok Sep 15 10:15:04 khem: and then set DISTRO to what i need afterwards and rebuild again? Sep 15 10:15:58 minimal is distro which you will than use Sep 15 10:16:07 do you need a particular one Sep 15 10:16:25 angstrom-2008.1 Sep 15 10:17:08 is there any angstrom-2009.1? Sep 15 10:17:11 image ? Sep 15 10:17:21 dunno, let me check Sep 15 10:17:27 k Sep 15 10:19:04 * * OE Bug 1380 has been RESOLVED (FIXED) by raj.khem(AT)gmail.com Sep 15 10:19:06 * * HEAD uClibc build fails on uClibc install Sep 15 10:19:08 * * http://bugs.openembedded.net/show_bug.cgi?id=1380 Sep 15 10:20:51 julemore: don't seem so from http://cgit.openembedded.org/cgit.cgi/openembedded/tree/conf/distro Sep 15 10:29:29 ok ty Sep 15 10:49:38 hi lrg Sep 15 10:49:53 03Martin Jansa  07shr/import * rdda73de178 10openembedded.git/conf/checksums.ini: Missing checksum midori-0.1.10 Sep 15 10:54:08 hey pb__ Sep 15 11:06:30 03Koen Kooi  07org.openembedded.dev * rff47733554 10openembedded.git/recipes/linux/ (linux-omap-2.6.31/arch-has-holes.diff linux-omap_2.6.31.bb): linux-omap 2.6.31: add memhole patch Sep 15 11:11:56 anyone have a copy of DirectFB-1.4.2.tar.gz they can send me? Sep 15 11:12:14 did not seem to make it to our mirror before the site went away Sep 15 11:16:48 cbrake, there have been a couple of patches with a new src_uri over the past few months Sep 15 11:16:55 did the entire site go away? Sep 15 11:17:35 Crofton|work: seems that way Sep 15 11:18:34 Crofton|work: downloads from projects.linuxtogo.org are also screwed up they turn up empty tar files e.g. atd Sep 15 11:19:42 urg Sep 15 11:19:53 florian, is the ltg guy :) Sep 15 11:19:56 I think Sep 15 11:20:54 khem: oh right, sounds we are missing a link there Sep 15 11:21:06 * cbrake tries directfb irc channel ... Sep 15 11:22:09 cbrake: I tried already its knows the site is down weather it will come up no one knows yet Sep 15 11:22:41 florian: would be nice if you could fix it Sep 15 11:27:02 * florian shoots gforge Sep 15 11:28:07 should be fixed Sep 15 11:53:18 florian, thanks Sep 15 12:02:04 03shr build user  07shr/import * r86982203ae 10openembedded.git/recipes/freesmartphone/ (14 files in 2 dirs): update FSO recipes Sep 15 12:02:05 03Michael 'Mickey' Lauer  07shr/import * r395b318518 10openembedded.git/recipes/pulseaudio/libcanberra_0.14.bb: libcanberra: add 0.14 Sep 15 12:02:12 03shr build user  07shr/import * r1f67527ee5 10openembedded.git/recipes/freesmartphone/fsousaged_git.bb: fix fsousaged Sep 15 12:02:14 03shr build user  07shr/import * re768501b75 10openembedded.git/ (2 files in 2 dirs): add autorev to fsotimed, fix misc-vapi Sep 15 13:13:31 kgilmer, ping Sep 15 13:27:40 03Steffen Sledz  07org.openembedded.dev * rbdb0b97558 10openembedded.git/recipes/gob2/gob2-native_2.0.14.bb: gob2-native: missing glib-2.0-native dependency added Sep 15 13:48:20 ~seen cbrake Sep 15 13:48:24 cbrake is currently on #oe #openmoko #edev #nslu2-linux #elinux. Has said a total of 23 messages. Is idling for 2h 30m 49s, last said: 'Crofton|work: seems that way'. Sep 15 13:51:05 zhm did you ask in angstroem channel? Sep 15 13:54:20 good morning Crofton|work Sep 15 13:54:29 kgilmer, gm Sep 15 13:54:37 hi kgilmer Sep 15 13:54:40 hi crofton Sep 15 13:54:44 hi woglinde :) Sep 15 13:54:53 have you ever thought about setting up eclipse to use the tool chains and libraries that are in staging? Sep 15 13:55:01 gm woglinde Sep 15 13:55:53 Crofton|work, yes actually some of that is already done but not exposed. Sep 15 13:56:04 CDT, the C part of Eclipse is kind of a jungle. Sep 15 13:56:10 heh Sep 15 13:56:40 CDT already supports setting up your own toolchain from some arbitrary location. Sep 15 13:56:48 but it's not very easy to figure out Sep 15 13:56:57 at least it wasn't last year when I tried to do that the first time. Sep 15 13:57:00 why do you ask? Sep 15 13:57:02 I don't know anything about eclipse, other than what I saw you present :) Sep 15 13:57:10 and that this sounds like something that should work Sep 15 13:57:22 they release a new major version every year Sep 15 13:57:26 so it's hard to keep up Sep 15 13:57:46 * woglinde knows something about osgi Sep 15 13:58:02 still thinking about oedem but not sure. Crofton|work will you be there? Sep 15 13:58:09 kilgimer robert btw. too Sep 15 13:58:16 this is just something I keep thinking about, but figuring it all out is hard :) Sep 15 13:58:21 I hope so Sep 15 13:58:27 need to check some plane fares Sep 15 13:58:29 woglinde, did you notice yet another embedded osgi device on linuxdevices.com last week? Sep 15 13:58:46 aehm Sep 15 13:58:54 hm didnt read it for aeh 2 weeks or so Sep 15 13:59:29 they have a novel approach to device shielding Sep 15 13:59:30 http://www.linuxfordevices.com/c/a/News/Eurotech-Helios/ Sep 15 13:59:34 pretty cool imho Sep 15 14:00:28 hm ibms j2me Sep 15 14:01:01 yeah don't know what that is Sep 15 14:01:18 why cldc on a x86 Sep 15 14:01:42 yeah ibm had for long time an embbedded java stack Sep 15 14:01:56 workrd even for powerpc Sep 15 14:02:10 but dont in which states it is these times Sep 15 14:02:17 hm j9 was the name Sep 15 14:02:23 if I remeber correctly Sep 15 14:02:33 hi udovdh Sep 15 14:02:36 ah i've heard of that Sep 15 14:02:57 hello woglinde Sep 15 14:04:15 morning Sep 15 14:04:24 gotta head to a meeting. ttyl Crofton|work woglinde . Sep 15 14:04:34 bye] Sep 15 14:04:39 bye kgilmer Sep 15 15:21:57 morning kergoth Sep 15 15:22:26 hey Sep 15 15:32:37 khem: the problem with gcc-cross wasn't because of that patch.. on arm4t it commpiles fine with or without.. Now I'm rebuiling it on arm5te but with ccache disabled (first difference in logs from arm4/arm5 builds) Sep 15 15:44:18 JaMa: yes I am also having issues with ccache, I have disabled it Sep 15 15:44:52 I actually clean my ~/.ccache and I am hopeful it will start working again Sep 15 16:10:59 khem: hi. did you have a look at my kernel build problem? Sep 15 16:11:42 hi mickeyl Sep 15 16:12:54 hey p Sep 15 16:12:55 b_ Sep 15 16:15:07 Who should I talk to about seeing about getting collie working again (I used to know at least what was going on, then I stopped using my Z for several years, and the wiki seems to say everything is over on angstrom now, but refers to OE/OZ a lot.) Sep 15 16:18:16 james_lan: probably depends on what you are trying to run on the collie. is there a particular thing you want to get working? Sep 15 16:18:34 kernel-wise I don't think there's been a lot of collie development lately but there should still be kernels in the repository that work. Sep 15 16:19:06 I suppose your best option is probably to just try building a collie image and see what breaks :-} Sep 15 16:21:22 Well, I've determined the only test image in the angstrom feed that even flashes is the alt-boot, which can't apparently see the sd or cf card (or boot) So I suspect I have work to do. Sep 15 16:21:27 I'll be back later. Sep 15 16:25:52 hmm, i don't think i like the way the UIs are handled in bitbake master Sep 15 16:29:54 how does that work? I've never looked at it. Sep 15 16:31:56 morning Sep 15 17:03:42 pb__: spawns the daemon into the background, starts the requested ui via "exec" (imports it dynamically), then starts up the ui and ui xmlrpcserver (which receives events from the server), then sends a call to the server xmlrpcserver registering the host/port of the ui xmlrpcserver with its event subsystem. all status comes into the ui via the ui xmlrpcserver, bb.event objects coming across Sep 15 17:03:47 could be worse Sep 15 17:06:09 not sure how i feel about using the same event mechanism for the ui communication as for the metadata, though Sep 15 17:06:11 * kergoth ponders Sep 15 17:07:35 i don't really like the use of xmlrpc for the events, or the fact that the server has to know about the UIs, not just vice versa Sep 15 17:09:00 xmlrpc was my idea Sep 15 17:09:12 i think it matches perfect as abstraction Sep 15 17:09:20 server knowing about UIs seems backwards, ya Sep 15 17:09:50 xmlrpc to talk to the server makes sense Sep 15 17:09:55 xmlrpc to send events back to the UI, less so Sep 15 17:10:00 it's using both right now Sep 15 17:11:42 well, yeah Sep 15 17:11:53 guess what Sep 15 17:12:00 nowadays I'd use dbus Sep 15 17:12:12 but anyways, the problem is that you have a command language client/server Sep 15 17:12:19 but also event delivery server=>client Sep 15 17:12:24 yeah Sep 15 17:13:15 for me it looked natural to make the client receive events from server in the same RPC way Sep 15 17:13:27 but we can talk about that Sep 15 17:13:34 well, consider from the perspective of someone writing a bitbake frontend in a different language, independent project Sep 15 17:54:30 sheesh, i have so many oddball proof of concepts laying around my drive, should either fix them up or trash em Sep 15 17:57:00 in the git fetcher, can I use tag to fetch a branch? Sep 15 17:57:16 yes Sep 15 17:57:21 thanks Sep 15 17:57:35 tags and branches as well as revision numbers are equal in git Sep 15 17:58:42 yeah, I sense that from the fetcher docs in the bb manual Sep 15 18:00:16 What I really need is vars for the repo and branch in the recipe :) Sep 15 18:04:00 * kergoth chuckles at the thought of a recipe being a configobj ini file Sep 15 18:05:07 hmm, what rev does it fetch by default? Sep 15 18:06:01 Crofton: the remote's HEAD I suppose Sep 15 18:06:14 i.e. the branch that is set to default in the remote repository Sep 15 18:06:22 usually master Sep 15 18:06:25 svn grabs r1 :) Sep 15 18:06:35 uh Sep 15 18:12:34 * kergoth wonders if there are any best practices for exception usage in python Sep 15 18:15:57 hmm, no zecke Sep 15 18:19:10 kergoth: exceptions should be used in exceptional situations Sep 15 18:19:14 pb__: can you remind me what the name of that deal was where variables were versioned, references using specific versions, again? I'm drawing a blank Sep 15 18:19:21 akheron: i'd certainly hope so :) Sep 15 18:19:33 :) Sep 15 18:21:05 i'm more curious about when and how one should re-raise, and how the actual error message should be handled / passed up. for example, if you subclass Exception, should you call the Exception constructor and pass a basic string version of this exception in that way, or should you override __str__? Sep 15 18:21:28 Question: I have a "EXTRA_IMAGEDEPENDS += u-boot" in my machine.conf file, however, the u-boot doesn't get deployed when I "bitbake myimagerecipie". It does build u-boot as I see it in the tmp/work directory just doesn't do the last few steps, mainly deploy, unless I call "bitbake u-boot" directly. How do I tell it to build/deploy u-boot with "myimagerecipie" if u-boot isn't already built? I see DEPENDS = "task-base" and IMAGE_INSTALL = "" in the x11-imag Sep 15 18:22:24 IMAGE_INSTALL = "...(lots in here)..." that is Sep 15 18:36:47 hmm, i should finish moving bits into the ast module on the parsing branch Sep 15 18:39:02 kergoth: I usually subclass RuntimeError and pass the string in constructor Sep 15 18:39:23 hmm, okay Sep 15 18:39:29 if the error message is the only useful information stored in the exception object, there's no need to override __str__ Sep 15 18:39:54 class MyError(RuntimeError): pass is enough Sep 15 18:39:57 and if it isn't? Sep 15 18:40:06 * kergoth nods Sep 15 18:40:42 then you probably want to somehow format the error message, and that needs to be done in __str__ Sep 15 18:41:02 also in __init__ you take the extra data in and save it into member variables Sep 15 18:41:08 * kergoth nods Sep 15 18:41:14 okay, that makes sense Sep 15 18:41:16 thanks Sep 15 18:44:24 * kergoth thinks Sep 15 18:45:03 dont think.... feeeeel Sep 15 18:47:48 * khem is trying to revive bitbake world Sep 15 18:48:22 there are recipes who have DEPENDS on other recipes which are in nonworking/ Sep 15 18:48:27 what should we do about it Sep 15 18:48:35 move these also to non working ? Sep 15 18:51:34 re Sep 15 18:55:59 Does anyone know the best method to install a package on top of an image that overwrites some files in the image? Sep 15 18:56:55 Currently I'm using a opkg-cl call to do a "-force-overwrite install" within the ROOTFS_POSTPROCESS_COMMAND for the image, Sep 15 18:57:06 that's probably your best bet Sep 15 18:57:23 But the opkg-cl command returns an error that the package isn't found... Sep 15 18:58:05 I've stuck the package in EXTRA_IMAGEDEPENDS so it gets built, but for some reason the opkg-cl command doesn't find it. Sep 15 18:58:05 03Koen Kooi  07org.openembedded.dev * rd861047e32 10openembedded.git/recipes/linux/ (32 files in 2 dirs): linux-omap 2.6.29: add support for the omap3-touchbook Sep 15 18:58:06 03Koen Kooi  07org.openembedded.dev * r66c257096a 10openembedded.git/conf/machine/omap3-touchbook.conf: omap3-touchbook: add machine config based on the beagleboard one Sep 15 18:58:06 03Koen Kooi  07org.openembedded.dev * ra6d3ba856d 10openembedded.git/recipes/x-load/x-load_git.bb: x-load: add omap3-touchbook support Sep 15 18:58:07 03Koen Kooi  07org.openembedded.dev * r72c1fd8a2b 10openembedded.git/recipes/u-boot/ (4 files in 2 dirs): u-boot git: add omap3-touchbook support Sep 15 18:58:10 03Koen Kooi  07org.openembedded.dev * r143a138985 10openembedded.git/: Merge branch 'org.openembedded.dev' of git@git.openembedded.org:openembedded into org.openembedded.dev Sep 15 19:00:03 radhermit: take a look at rootfs_ipk.bbclass. it probably calls opkg-cl with arguments that you'd need. maybe theres a variable for that that you can use. Sep 15 19:00:12 Is there something I'm missing in the opkg call? Currently I'm using the command "opkg-cl ${IPKG_ARGS} -force-depends -force-overwrite install radix-extras". Sep 15 19:00:34 I've looked that that class and it calls opkg with the same arguments. I'll look at it again... Sep 15 19:02:45 hm, odd Sep 15 19:06:06 kergoth: didn't you work on makedev back then? Sep 15 19:06:17 kergoth: I'd like someone to ACK the patch from Antonio Ospite Sep 15 19:06:27 [the count bug] Sep 15 19:06:32 i played with some device table and makedevs type stuff ages ago, yeah Sep 15 19:06:46 just take a brief look over it Sep 15 19:06:48 it should be good Sep 15 19:06:51 he sent it last week Sep 15 19:07:40 k Sep 15 19:10:10 mickeyl: looks fine to me. Sep 15 19:10:33 righto Sep 15 19:10:35 will commit then Sep 15 19:10:36 thanks Sep 15 19:12:16 mickeyl: the recipe for python-phoneutils_git.bb has unmet DEPENDS on libphone-utils Sep 15 19:13:22 there is no recipe for libphone-utils in OE Sep 15 19:13:33 in .dev Sep 15 19:14:03 oh, missed to import that from shr/import Sep 15 19:14:03 thanks Sep 15 19:14:05 will fix asap Sep 15 19:14:12 thx Sep 15 19:14:33 mickeyl: there are recipes who have DEPENDS on nonworking/ recipes Sep 15 19:14:44 that's bad. Sep 15 19:14:45 should these recipes also be moved to nonworking/ Sep 15 19:14:51 yep Sep 15 19:14:55 or fix the nonworkings ;) Sep 15 19:15:02 hmm Sep 15 19:15:23 I can try that its all maemoooo stuff Sep 15 19:16:26 sugar also has a DEPENDS on a non existing recipe Sep 15 19:17:04 sugar-fructose DEPENDS on sugar-web-activity Sep 15 19:17:14 which does not exist Sep 15 19:17:20 such commits should be avoided Sep 15 19:17:31 agreed. Sep 15 19:17:39 because sugar-fructose is not included in any image this goes unnoticed Sep 15 19:46:46 kergoth: trying http://dev.bec-systems.com/svn/pub/qt-overo-palo-gpio/qt-overo-palo-gpio.bb Sep 15 19:47:19 kergoth: and getting http://pastebin.ca/1567403 Sep 15 19:47:34 kergoth: still waiting for qt-embedded to finish building and will then look some more at what is going on Sep 15 19:49:32 hmm, getting a 404 on the pastebin Sep 15 19:49:47 ah, works with 'www.' Sep 15 19:50:22 huh, thats odd. we really need to re-implement prepend/append wrt python tasks Sep 15 19:50:30 kergoth: re: the server needing to know about UIs - all the server knows about is that there is a given xmlrpc client that wants an event feed Sep 15 19:50:32 turn it into a list of python code objects to execute Sep 15 19:50:36 RP: i know, i read the code Sep 15 19:50:48 I'd be fine with adding other ways of getting events from the server too Sep 15 19:51:11 kergoth: Its been coded such that client and server on different machines is possible which is why dbus wasn't chosen Sep 15 19:51:28 i didn't propose dbus :) Sep 15 19:51:31 but fair enough Sep 15 19:51:42 kergoth: Just mentioning why :) Sep 15 19:52:07 kergoth: I'm not overly keen on some of it and would be interested in other suggestions Sep 15 19:52:58 i don't have any specific ideas just yet, just an impression that we could make it easier for third party UIs (MV might end up writing one, and it wouldn't necessarily be in python) Sep 15 19:53:11 will have to give it more thought Sep 15 19:54:08 cbrake: try changing the intentation in the do_populate_staging in srctree.bbclass? I guess you have a different set of classes loaded than i did in my tests, so didn't notice Sep 15 19:54:39 kergoth: ok, will do Sep 15 20:44:21 kergoth: mickey|bbl : about makedev / start... in device_table_minimal.txt there are devices going from 1 to 8 (see mmc, 7 is max). Sep 15 20:44:26 http://dl.getdropbox.com/u/112715/Documents/OpenEmbedded%20Usability%20Concerns.html/index.html - heh, *trying* to get my thoughts together Sep 15 20:45:47 if antonio's patch is applied almost all device_table* need to be fixed afterwards Sep 15 20:47:47 btw it seems to be 1yr old known issue...http://lists.uclibc.org/pipermail/busybox/2008-July/032533.html Sep 15 20:47:55 :/ Sep 15 21:02:22 khem: about that $TMPDIR/usr dir, it comes from ${TMPDIR}${layout_libdir} Sep 15 21:02:42 see packaged-staging Sep 15 21:17:47 jo Sep 15 21:21:37 khem: gcc-cross compiled fine without ccache on that arm5te.. Sep 15 21:51:27 abnd woglinde Sep 15 21:51:48 jo ant Sep 15 21:57:08 wb mickeyl Sep 15 21:57:14 hi ant__ Sep 15 21:57:30 hm arm is now member if linux foundation Sep 15 21:57:33 jo mickeyl Sep 15 21:57:40 hey woglinde Sep 15 21:58:32 mickeyl did you book hotel for oedem already? Sep 15 21:58:55 cambridge? not yet, haven't had time to look around Sep 15 21:58:57 do you? Sep 15 21:59:37 mickeyl: when I added more mmcblk0p devices I wanted 7 of them but had to set count=8 Sep 15 21:59:46 that was the wrong way to fix it :) Sep 15 22:00:06 ant__: indeed. we should fix that now Sep 15 22:00:13 the patch from Antonio is a good chance Sep 15 22:00:32 yes, was in uclibc ML too Sep 15 22:00:51 mickeyl nope Sep 15 22:00:51 last year :D Sep 15 22:00:58 the youth hostel is very pricy Sep 15 22:01:03 but I have no credit card Sep 15 22:01:42 did pb__ send a list of hotels? Sep 15 22:02:21 hm I didnt see one Sep 15 22:02:28 but he mentioned youth hostel too Sep 15 22:03:44 * timtimred will try and go to oedem! Sep 15 22:04:06 i work in the east end (of london) - but i might be working... Sep 15 22:04:16 timtimred: cool! Sep 15 22:04:31 wellll east bit of the city. gherking :) Sep 15 22:04:35 oops gherkin Sep 15 22:04:43 http://www.yha.org.uk/find-accommodation/east-of-england/hostels/Cambridge/index.aspx Sep 15 22:05:30 mickeyl there are 2 and 4 bed rooms or a hall with more beds Sep 15 22:09:04 lalalaaaaalaa Sep 15 22:11:07 hm security fix for bugzilla Sep 15 22:11:20 mickeyl who is maintaining bugzilla? Sep 15 22:11:31 hmm, Laibsch Sep 15 22:12:33 okay Sep 15 22:13:10 wtf kernelmodule in haskel Sep 15 22:15:38 meh Sep 15 22:17:03 he kergoth Sep 15 22:22:42 hmm, oedem Sep 15 22:22:54 Hey, is there any RECIPE for libx264? Sep 15 22:24:26 rp yes? Sep 15 22:24:31 julemore no Sep 15 22:25:00 guess why I patched ffmpeg not to install the presets if no libx264 is around Sep 15 22:25:14 woglinde: Have hotels been discussed somewhere? Sep 15 22:29:51 RP not yet Sep 15 22:30:06 as I said youth hostel is pricy Sep 15 22:32:26 woglinde: right :/ Sep 15 22:33:30 RP: do you have access to a poky tree? Pls check one staging- ipk like udev and see the arch, pls Sep 15 22:34:29 JaMa: cool, also you can try to delete ~/.ccache Sep 15 22:34:37 ant__: x86_64-linux :/ Sep 15 22:34:39 it helped me to get ccache to behave Sep 15 22:35:00 So close with this 32 bit sdk... Sep 15 22:35:10 RP: that's bad news :/ but at least is consistent with OE Sep 15 22:35:26 it means you can safely sync packaged-staging Sep 15 22:37:17 ant__: yes that seems to be the place where its happening staging_helper Sep 15 22:38:08 khem: In the good news, I have this sdk stuff basically working, just a few cleanups to make... Sep 15 22:38:17 RP: cool Sep 15 22:38:29 So there should be some nice commits to poky soon (and death to sdk.bbclass) Sep 15 22:38:35 post it I will like to have a look Sep 15 22:38:44 ah cool Sep 15 22:39:28 nie mickeyl Sep 15 22:39:33 n8 Sep 15 22:39:35 mickey|zzZZzz: night night Sep 15 22:40:07 someone should promote an official oedem hotel Sep 15 22:40:59 yeah, not a bad idea. I can suggest some hotels for anyone who wants to pick one. Sep 15 22:41:16 I think I posted some ideas on the list previously, or maybe it was here in irc. Sep 15 22:41:46 pb was here Sep 15 22:42:12 hey is there any Bitbake libh264? Sep 15 22:42:12 ah right. Sep 15 22:42:19 RP: this is my merged packaged-staging.bbclass sync with poky http://fr.pastebin.ca/1567615 just some doubts (and comments) about PSTAGE_EXTRAPATH Sep 15 22:43:11 well, executive summary recap: cheapest option is probably youth hostel (~20GBP/night), then there are several bed & breakfast type places in the £35/night range, then full-service hotels start at about £60/night. Sep 15 22:44:54 £1 = about €1.13, or $1.64 at today's exchange rates Sep 15 22:45:09 julemore no Sep 15 22:45:25 julemore do a find nextime yourself Sep 15 22:46:15 pb__: That doesn't sound so bad given its cambridge... Sep 15 22:46:43 pb__: Do we have a venue yet? or a preferred beer dispensing establishment? Sep 15 22:47:03 i'll be out in camden the night before Sep 15 22:47:10 :) Sep 15 22:47:19 or, somewhere in london Sep 15 22:47:19 woglinde: could you give me an example of a berlin postal code? Sep 15 22:47:29 12439 Sep 15 22:47:32 its mine Sep 15 22:47:33 thanks Sep 15 22:47:41 stupid fedex won't quote a rate without one Sep 15 22:47:53 hm? Sep 15 22:48:41 pb____: four of you in here - isn't that greedy? :) Sep 15 22:49:08 woglinde: just trying to find out roughly how much it costs to ship something from cambridge to berlin, but the fedex quote tool insists on having an exact postcode for both ends Sep 15 22:49:09 RP: heh Sep 15 22:49:17 yeah, I guess that is a bit much Sep 15 22:49:18 pb ah Sep 15 22:49:37 down to three now :-} Sep 15 22:49:46 *g* Sep 15 22:50:03 http://www.parcel2go.com is convenient for shipping stuff from UK Sep 15 22:50:15 pb____: :) Sep 15 22:50:17 RP: venue probably either my office or th1's but I'm not quite sure which yet Sep 15 22:50:19 it comes out cheaper than the carriers do directly in some cases Sep 15 22:51:02 timtimred: does it reach destination safely too Sep 15 22:51:16 sure, you can choose from carriers and you get tracking Sep 15 22:51:17 i was once in cambridge in 2001 Sep 15 22:51:29 DHL, parcel farce, shitty link Sep 15 22:51:37 erm, parcel force, city link Sep 15 22:51:39 beer dispensing, there are lots to choose from. preferred one near the city centre probably http://www.kingston-arms.co.uk/ Sep 15 22:51:50 pb____: They're both near the cinty centre? Sep 15 22:51:51 etc Sep 15 22:52:14 (the offices) Sep 15 22:52:19 i used them to ship out przes for nas-central and ebay-europe stuff Sep 15 22:52:24 RP: well, cambridge isn't very big, you are always fairly close to the city centre Sep 15 22:52:51 th1's office is on the science park, which is on the north side near the sewage plant :-} Sep 15 22:52:57 my office is on the east side near the airport Sep 15 22:52:57 pb____: well, they are actually in cambridge then :) Sep 15 22:53:12 yeah Sep 15 22:53:29 * RP describes himself as Newcastle for ease but in reality is 10 miles away... Sep 15 22:54:03 as in Newcastle... Sep 15 22:54:31 time for bed now Sep 15 22:54:33 sleep. Too much toolchain pain today... Sep 15 22:54:35 good nite Sep 15 22:54:38 'night woglinde Sep 15 22:54:42 * RP -> Zzzz Sep 15 22:54:49 good night RP Sep 15 22:54:53 I have java again tommorrow Sep 15 22:54:54 and woglinde Sep 15 22:54:56 night chaps Sep 15 22:55:02 woglinde: ah, bad luck Sep 15 22:55:13 pb hm I am fine with java Sep 15 22:55:28 pb____: you and th1 work at same place ? Sep 15 22:55:36 khem: no, that's why we have different offices Sep 15 22:56:02 oh I thought it was different building but same company Sep 15 22:56:09 no, different companies Sep 15 22:56:21 just two options for oedem Sep 15 22:56:28 ok Sep 15 22:58:13 pb____: I will make changes per your comments to the patches Sep 15 22:59:24 pb____: prismstumbler seems you have written a lot of it if not all Sep 15 23:00:15 recipes/prismstumbler/prismstumbler-0.7.3+0.7.4pre1/wireless.patch is not needed on newer systems where we use kernels 2.6.18+ Sep 15 23:00:24 should I just punt it out Sep 16 00:43:01 howdy folks Sep 16 00:43:57 what desktop environment are folks using on their OE platforms nowadays/ Sep 16 01:00:47 hey Sep 16 01:03:23 Soopaman: e! :) Sep 16 01:06:00 enlightenment? Sep 16 01:06:47 know any other e's? :) Sep 16 01:07:24 ecstasy Sep 16 01:07:29 :P Sep 16 01:09:53 grg: well i was intending "withing open source software you might find as a desktop env in oe" :) Sep 16 01:11:08 radhermit, is there an e app called ecstasy yet? Sep 16 01:11:18 oopse, that was supposed to be raster, Sep 16 01:11:31 grg: not... yet! Sep 16 01:11:32 :) Sep 16 01:11:46 hey raster, how's it going? Sep 16 01:11:54 haven't been that bold. yet... :) Sep 16 01:12:09 kergoth: yo! not bad. collecting new toys Sep 16 01:12:16 heh, nice Sep 16 01:12:40 unfortunatly oe isnt used to build for them.l i need to patch up this build system t 'suck less' Sep 16 01:13:01 at least this time i get a git tree of it so i can sanely send back patches Sep 16 01:14:05 what build system? Sep 16 01:15:38 diy Sep 16 01:16:02 fun fun Sep 16 01:16:08 yah Sep 16 01:16:16 it has bad bits Sep 16 01:16:25 no good isolation from the hot environment Sep 16 01:16:43 softare is finding locally pacagesd software and enabling features etc.base don it Sep 16 01:17:00 heh, everybody hates autotools, and think they can do better, but they don't realize how much it was doing for them... Sep 16 01:17:07 (eg they apend to pkg_config_path, not prepend, so personal pkg_config_path elements take priority) Sep 16 01:18:55 ohy god - none of their software uses autotools Sep 16 01:19:00 literally projects have build.sh's Sep 16 01:19:10 the odd hand-made Makefile Sep 16 01:20:04 nred to improve this project generator. so far it generates the autofoo for you so it has a semblance of being "right" out of the box with some simple inputs Sep 16 01:20:26 but i can understand it not being used. its a pita to use and fiure out. it adds a massive barrier-of-entry Sep 16 01:21:18 biggest problem with autotools is trying to fix someone else's broken garbage Sep 16 01:21:47 grg: that's the biggest problem without autotools too. Sep 16 01:22:23 but so few projects actually have the complexity that requires autotooling Sep 16 01:22:35 that isn't all autotools solves. Sep 16 01:22:44 grg: not surprised its broken tho. autotools is a struggle to deal with Sep 16 01:23:04 grg: its not the complxity. its simple things like "make distcheck" Sep 16 01:23:15 somethnig that makes sure your stuff works ith DESTDIR Sep 16 01:23:24 and ytghe tarball is clean without build/tmp junk Sep 16 01:23:35 and contains all u need to build and disseminate Sep 16 01:23:38 if you don't use autotools, every distro that needs to build your app has to go through and read every makefile in your system to figure out what all you've screwed up, and how to pass their variables in, how to get it to crosscompile, etc. you occasionally have to do that for autotools systems too, but it's usually devs not having a clue about build stuff, and it's usually minor Sep 16 01:23:59 having some semblance of standardisation - eg --prefix. Sep 16 01:24:05 at least using autotools, or qmake, or some existing system means your build behaves consistently Sep 16 01:24:11 indeed Sep 16 01:24:52 at this stage tho autotools tho its standardised, helps and works, does no favors to peolpe trying to use it to get them in fast Sep 16 01:25:07 ie i want an app. i ant a nice uild tree and want to produce an executable binary Sep 16 01:25:12 and its called "myapp" Sep 16 01:25:24 true, there's quite a bit of overhead Sep 16 01:25:26 it needs ti link to lib a, b and c (and thus also have headers for those libs) Sep 16 01:25:29 42 build files, one .c file Sep 16 01:25:39 heh Sep 16 01:25:44 hmm Sep 16 01:25:51 most dev s - thats all they want to care about when starting their project. the amount of autofoo needed to do that simple thing is rather hefty Sep 16 01:26:29 there needs to be autofoo generators that do this kind of thing Sep 16 01:26:41 quagmire looks cute. it replaces the majority of the makefile generation stuff, the automake stuff, with a pure make implementation that's still consistent, and easy to set up, but uses no libtool and crap Sep 16 01:26:41 i've had my own templates i use myself for my projects Sep 16 01:26:44 that's why a 3 line Makefile will always win me over Sep 16 01:26:54 autoconf + quagmire instead of autoconf + automake + libtool Sep 16 01:26:55 quagmire? Sep 16 01:27:15 found it Sep 16 01:27:32 http://code.google.com/p/quagmire/source/browse/trunk/example/simple/Quagmire.mk Sep 16 01:27:36 hmm still young it seems Sep 16 01:27:39 http://code.google.com/p/quagmire/source/browse/trunk/example/simple/configure.ac Sep 16 01:27:40 aye Sep 16 01:27:59 quite interesting approach though Sep 16 01:28:10 grg: totally underwstand. even a build script is a bit simpler than a makefile. if i'm in a hurry thats all i do. Sep 16 01:28:12 keeping autoconf means you've still got --prefix and all Sep 16 01:28:33 yeah, that's a good starting point. just good to move to something nicer when you release Sep 16 01:28:51 sure Sep 16 01:28:56 tho something that "emulates" autoconf Sep 16 01:29:01 ie provides a ./configure Sep 16 01:29:07 and accepts your usual autofoo params Sep 16 01:29:10 and env vars Sep 16 01:29:15 and otherwise works "with make" Sep 16 01:29:29 but branches off to use somthing else early on in the build would be good Sep 16 01:29:57 but for now, autofoo is all over the place, and i think its good if we get people using it more and properly. lower the barrier Sep 16 01:31:31 i'd love it if it set up all the stuff for u like a .desktop file was crated with exe and appname already done Sep 16 01:31:57 a "starting" icon .png was put in the tree, autofoo done to install it in the right place (ie PREFIX/share/applications + icons) Sep 16 01:32:16 so when peopl write apps... they are proprly made available in menus/launchers Sep 16 01:32:27 just to help developers write apps on/for linux better Sep 16 01:33:18 could write a wrapper m4 macro to do most of the footwork on that, but it's still hard to control the actual makefile generation aspect from autoconf, automake does too much of it Sep 16 01:33:22 too convoluted Sep 16 01:34:32 ah well Sep 16 01:35:54 i'm more thinking a small gui app that generates the initial autofoo for u Sep 16 01:36:02 so whatit gates is a clean starting build tree Sep 16 01:36:13 once started i'd hope devs can take it from there Sep 16 01:36:28 just a gui that pops up "enter name of project" Sep 16 01:36:32 thatd be cute. there's something like that for generating template trees for python projects, hands you a template setup.py for setuptools and all Sep 16 01:36:41 "enter executabls to generate" (a list u add them) Sep 16 01:36:44 or libs Sep 16 01:36:45 etc. etc. Sep 16 01:36:55 buildwsa a clean project tre with all the right things in the right places Sep 16 01:37:08 isn't that how IDEs behave? (i haven't used one in 10 years...) Sep 16 01:37:20 grg: bingo. but that sucks u into the whole ide Sep 16 01:37:37 i'm thinking just the "set up the project at the start" thing Sep 16 01:37:41 an xterm is a good ide Sep 16 01:37:49 hehehe Sep 16 01:37:57 same. all ides piss me off Sep 16 01:38:09 but i get tired of generating autofoo template trees by hand Sep 16 01:38:29 and as u dont do it often enough to memorise cmd-line optiosn to a script to do this Sep 16 01:38:45 a little gui to just give projct name, the exes + libs to pump out Sep 16 01:38:46 etc. Sep 16 01:39:09 all the mvl products include eclipse, including mvl6. i work for mv, and i haven't run it once. Sep 16 01:39:15 * kergoth shudders at the thought of that beast Sep 16 01:39:43 "include eclipse" ? Sep 16 01:39:57 it includes an eclipse based ide, i'm sure you've seen those before Sep 16 01:39:58 like u literally ship a copy of it as part of the sdk? Sep 16 01:40:07 where they wrap it up and include proprietary addins to it or whatever Sep 16 01:40:12 not just have them get it separately and u provide modules? Sep 16 01:40:22 ugh Sep 16 01:40:46 the last place i worked at went further than that, they provided a shoddy ubuntu based vmware image with the sdk+eclipse installed Sep 16 01:40:51 i guess some embedded folk like that sort of thing, they probably get used to those fancy UIs some of the jtags have :) Sep 16 01:41:12 grg: that's not a bad way to go, at least the environment is consistent that way. we'd get a lot less trouble if OE shipped a chroot or live cd Sep 16 01:41:17 heh Sep 16 01:41:34 problem was, no one ever maintained it :) Sep 16 01:41:39 hehe Sep 16 01:41:40 t Sep 16 01:41:40 yp Sep 16 01:41:40 ica Sep 16 01:41:43 er, typical Sep 16 01:41:47 make it once, dont touch it again Sep 16 01:41:48 grg: not sure thats much worse. Sep 16 01:41:59 http://dl.getdropbox.com/u/112715/Documents/OpenEmbedded%20Usability%20Concerns.html/index.html Sep 16 01:42:03 any comments? Sep 16 01:42:09 as u'd need to ship the jvm too to be sure. and then that is dependant on that jvm being for a particular os target Sep 16 01:42:16 a vmware image fixes that problem Sep 16 01:42:17 :) Sep 16 01:43:01 kergoth: a chroot for oe would indeedx solve much. personally i lik the idea. Sep 16 01:43:21 a chroot should contain the minimal set of build tools. oe build itself a chunk of those anyway (toolchain etc.) Sep 16 01:43:25 it's extremely difficult to do cross builds and never have anything make decisions based on the build machine's installed packages and crap, short of using an LD_PRELOAD wrapper to catch /usr/include or /usr/lib pokes or something Sep 16 01:43:32 so a consistent build environment would help a fair bit Sep 16 01:43:37 i would say awhy not build a minimal chroot host os too Sep 16 01:44:01 that's true, could ditch a lot of the native stuff if you could make more assumptions about the build environment Sep 16 01:44:06 would only need to keep the patched stuff Sep 16 01:44:15 i kno i'td be "more to build" but at least your oe chroot would need to populate itself just once Sep 16 01:44:32 there is a task-self-hosting recipe Sep 16 01:44:35 i keep meaning to take a look at it Sep 16 01:44:36 and package up the oe chroot too Sep 16 01:44:41 u can use a pre-packaged one Sep 16 01:44:44 or uild your own + packages Sep 16 01:45:07 we really need to improve the prebuilt binaries support. polish up pstage. shouldn't need to build what's already been built, and is unchanged Sep 16 01:45:07 hmm cant comment about all of that Sep 16 01:45:22 but recipes could do with being asier to find and use Sep 16 01:45:31 especialyl knowing which bloody one is being used in the image u are pumping out Sep 16 01:45:32 :) Sep 16 01:46:05 aye. bitbake -g helps to see why things are getitng pulled into the build, but it's not nearly enough information Sep 16 01:46:15 anyway - i already advised my homies here with their diy system to consider chroot Sep 16 01:46:25 where are you working nowadays? Sep 16 01:46:27 i think they'd be just happy populating it with binaries Sep 16 01:46:49 tho as long as they are able to compile those binaries... whenever they ant'need to Sep 16 01:46:54 these days i contract Sep 16 01:47:06 so i "walk the streets" whoring myself out Sep 16 01:47:18 ahh, nice Sep 16 01:47:37 i've been thinking about moving to that eventually, maybe leave montavista and sell my services back to them Sep 16 01:47:43 heh Sep 16 01:47:44 tho i seem to be getting a lot of work from here in korea :) Sep 16 01:48:11 (im kr atm - back to sydney tonight) Sep 16 01:48:41 ah Sep 16 01:49:23 i've been trying to get them to use oe Sep 16 01:49:28 just to make my life easier Sep 16 01:49:38 so far, no potatoe. Sep 16 01:49:43 potato Sep 16 01:49:44 i mean Sep 16 01:49:51 they havent bitten. Sep 16 01:50:00 so i have to keep fixing their diy job. Sep 16 01:50:11 i fear the day they release their sdk Sep 16 01:50:12 :) Sep 16 01:50:30 fixing peoples diy crap is REALLY annoying Sep 16 01:50:35 i know Sep 16 01:51:00 especially when it s "oh.. THAT problem. if u had used oe (properly) this problem wouldnt happen" Sep 16 01:51:00 kergoth, re usability concerns, the biggest surprise i had when starting with OE (just a couple months ago) was that VARIABLE_foo = bar could be an operator on VARIABLE. Sep 16 01:51:17 hell its morphed oer time from a single flash image now to using opkg and having instalable packages Sep 16 01:51:21 ah, yes, OVERRIDES for conditionals Sep 16 01:51:26 wont be long before libc and every element is a package Sep 16 01:51:28 you're right, not exactly intuitive Sep 16 01:51:36 so they'll end up re-creating oe Sep 16 01:52:31 grg: been mulling over possibilities for an alternative file format. Sep 16 01:52:56 it would be quite an effort to change that now though, wouldn't it? Sep 16 01:54:15 i'm thinking it would be additional Sep 16 01:54:33 no need to completely switch anytime soon, better to move incrementally Sep 16 02:39:30 hmmmm Sep 16 02:41:51 * kergoth thinks about what synchronization primitives we'd need to convert runqueue to use threads Sep 16 02:48:29 possibly not many, if you're careful Sep 16 02:50:53 as long as it's given a COW version of the dict.. Sep 16 02:51:11 * kergoth is toying around with running bitbake under ironpython, but ironpython lacks a lot of the subprocess type stuff Sep 16 02:53:36 BackgroundWorker seems to make it pretty painless. would probably want to lock around the task event firing though, since the handlers won't expect to run in parallel **** ENDING LOGGING AT Wed Sep 16 02:59:57 2009