**** BEGIN LOGGING AT Fri Dec 04 02:59:57 2009 Dec 04 03:46:40 mwester, would it be acceptable to add your fork/exec ("gunzip") patch to opkg as an #ifdef, and a configure option, e.g. --no-fork for it? Dec 04 03:47:02 are there any other outstanding problems that would prevent you from using an unpatched opkg? Dec 04 03:47:07 Time. Dec 04 03:47:33 as in, you don't have time to test? Dec 04 03:48:13 SlugOS is locked down to svn ver 160 of opkg -- we tried a few versions beyond that; they all failed for one reason or another, and tick pretty much indicated he wasn't able to help debug/resolve those at the time, so we locked it down at 160. Dec 04 03:49:06 I have no time to be able to devote to the exhausting task of testing opkg -- there are so very many scenarios that need to be tested, most of which require that I re-image an NSLU2 and start over again and again for the tests (no software simulator exists AFAIK) Dec 04 03:49:22 well of course its difficult debugging problems when there is no error checking... Dec 04 03:49:46 So I simply (naively it seems) assumed that I could create that separate recipe, and lock SlugOS down to 160, and happily let the rest of the world go by! :) Dec 04 03:51:15 i would really like to see the default opkg be a low memory usage variant Dec 04 03:51:28 I completely realize that a 32MB machine is not an "interesting" target for anyone in OE, or even (I suspect) for opkg any more... so it is not a great concern to me that we remain behind. I just want the old version to build reliably... :) Dec 04 03:51:42 anyone around familiar with the gcc recipes? Dec 04 03:51:44 low memory these days seems to be 1GB or less :D Dec 04 03:52:03 i'm targetting a 64mb device Dec 04 03:52:03 i'm seeing some qa issues with the target gcc package (4.3.3) Dec 04 03:53:13 but running out of memory with do_rootfs on a 1gb ram host was happening for large images Dec 04 03:53:44 wow Dec 04 03:54:05 who has a 1gb ram host these days anyway? Dec 04 03:54:12 (apart from me) Dec 04 03:55:07 jerm^, post your QA issues to a pastebin and somebody will probably look at them Dec 04 03:55:16 if not, then try the mailing list Dec 04 03:55:17 I have a number of virtual machines I've used for years on company-issues windows laptops while traveling -- 768MB. They used to build SlugOS very nicely during tedious company meetings :) Dec 04 03:55:36 * rwhitby is considering moving webos-internals from ipkg to opkg ... Dec 04 03:57:05 I am pretty much exclusively restricted to VMs Dec 04 03:57:13 How much RAM does a palm pre have, anyway? Dec 04 03:57:18 lol Dec 04 03:57:40 mwester: 256MB Dec 04 03:57:48 not too shabby! Dec 04 03:58:37 i think that openwrt targets mostly 32mb devices and they've been using opkg for a while Dec 04 03:59:26 But the fork() / OOM Killer issue (the one grg hasn't accepted my patch for) will strike even the Openmoko devices, which have a lot more memory... Dec 04 04:00:08 i'm not really convinced that changing fork to vfork will change much Dec 04 04:00:19 And I think that openwrt changes the OOM behavior by default, if I recall correctly -- that dodges the OOM killer to a great extent. Dec 04 04:00:21 linux has had copy on write for some time now... Dec 04 04:00:43 grg - you really must read up on the issue, then. Dec 04 04:00:48 i did Dec 04 04:01:05 does koen ever hang around here? Dec 04 04:01:06 Then, no offense please, but you do not understand what's really happening. Dec 04 04:01:14 maybe not Dec 04 04:01:22 got some links? Dec 04 04:01:26 copy-on-write does not solve it. Dec 04 04:01:39 koen appears to be in a different time zone Dec 04 04:01:43 I will dig through my file on the OOM killer and this issue, and try to come up with some links. Dec 04 04:02:48 yeah...he's "across the pond" Dec 04 04:02:48 The fundamental problem, though, is that the OOM killer computes the *total possible* (i.e. most pessimistic) number of pages, not the the most optimistic -- so copy-on-write is meaningless for the default OOM Killer behavior. Dec 04 04:02:56 just wondered if he ever hung out in the channel Dec 04 04:03:21 he appears to be the most active in the maintenance of the gcc recipes Dec 04 04:03:40 So the moment a fork() happens, the OOM killer can spring into action, based on nothing but the fact that the sum of all possible VM requirements exceeds the threshold. Dec 04 04:03:53 vfork() exists *only* to solve this problem. Dec 04 04:03:55 I was just wondering why, if the soft float libraries are found, those are packaged instead of the hard float Dec 04 04:04:17 if fork() was not an issue, then logically there would be no reason for vfork() to exist. Dec 04 04:04:28 well, vfork exists for BSD compatibility Dec 04 04:04:35 No. Dec 04 04:04:37 not for OOM killer work arounds Dec 04 04:04:41 wron Dec 04 04:04:42 g Dec 04 04:04:51 #define vfork fork Dec 04 04:04:58 would suffice for bsd (well, almost). Dec 04 04:05:13 when was the oom killer introduced? Dec 04 04:05:38 vfork exists despite a number of attempts to remove it because it fills a need -- that need is exactly the issue we are discussing -- the problem that fork() is too heavyweight. Dec 04 04:06:11 Dunno when oom killer came around... it was a sad day, though. Dec 04 04:06:56 http://linux-mm.org/ Dec 04 04:07:59 http://linux-mm.org/OverCommitAccounting <-- useful link for the "sledgehammer" solution to the fork()/vfork() issue. Dec 04 04:12:52 Sigh Dec 04 04:13:54 Why do I feel that despite proving a solution works, and is the only viable one out there, that it still ends up kicked to the curb by those upstream so often? Dec 04 04:16:53 :( Dec 04 04:16:53 "Any particular process leader may be immunized against the oom killer if the value of its /proc//oomadj is set to the constant OOM_DISABLE" Dec 04 04:17:12 mwester, im trying to work with you on this one Dec 04 04:17:21 I already did the work. Dec 04 04:17:26 And tested it. Dec 04 04:17:34 And hundreds of users are using it. :) Dec 04 04:18:01 so, i should merge it without understanding it? Dec 04 04:18:09 You're trying to get me to do more work, this time in a manner that is more pleasing to you. (you have that right, you are the upstream maintainer - I don't argue that). Dec 04 04:18:53 i'm not suggesting that you send me a patch for integration as #ifdef'd code. i will do that Dec 04 04:18:54 I am aware of the "immunization" thing; not sure if that actually worked with opkg or if there was some other issue. Dec 04 04:19:28 but i really would prefer having less code paths Dec 04 04:19:52 http://sourceware.org/ml/libc-help/2009-06/msg00035.html <--- middle of thread that pretty much rehashes this same discussion we are having :) Dec 04 04:20:50 Right, I agree less code paths are good -- so get rid of the system() and fork() stuff altogether; I can't see why that should be ifdef'd or even run-time selected. Dec 04 04:22:48 i did get rid of system Dec 04 04:23:01 the only fork left is the one for unzip Dec 04 04:23:17 which can't be a vfork because its not exec'ing Dec 04 04:24:02 one option would be to follow your example and remove all the libbb code. Replacing it to exec("gunzip") Dec 04 04:24:09 right Dec 04 04:24:41 let me think about that for a while. removing libbb altogether has its merits Dec 04 04:26:06 the other option is to update libbb to a more recent version which doesn't require separation with a fork() Dec 04 04:26:27 i've looked at that, and busybox have redesigned things quite a lot Dec 04 04:27:25 Could anyone pastebin its log.do_compile of boost_1.36.bb? As reported to the mailing list, I don't manage to compile this package and I would like to compare log.do_compile and log.do_configure with a working env. Thanks Dec 04 04:30:12 gregoiregentil: i can try Dec 04 04:30:26 gregoiregentil: will take me a sec to get that system up Dec 04 04:30:31 Thanks. I would appreciate. Dec 04 04:38:11 gregoiregentil: running...slowly but surely :) Dec 04 04:38:20 OK ;-) Dec 04 04:39:20 i've never used boost before...mostly low level code for me... Dec 04 04:40:24 what does the boost-jam native tool do? Dec 04 04:40:27 that just completed Dec 04 04:42:24 The way I understand it: boost-jam is a native tool needed for the compilation of boost. It' the primary tool used during the compilation of boost Dec 04 04:42:31 ahh Dec 04 04:42:38 i figured as much... Dec 04 04:43:06 compile of the target libraries is running...did you want the configure log too? Dec 04 04:46:31 yes if possible Dec 04 05:03:50 jerm^: so did you manage to do it? Dec 04 05:11:37 gregoiregentil: http://pastebin.com/m15d431b7 Dec 04 05:11:48 yep Dec 04 05:11:48 :) Dec 04 05:11:52 beat me to it Dec 04 05:11:55 sorry Dec 04 05:12:01 this system is slooooow Dec 04 05:12:56 gregoiregentil: this was from a Dec 1 build Dec 04 05:12:57 i see machine/conf/mtx-1.conf has 'PREFERRED_VERSION_yamon'...was there at some time support for building yamon from within oe? Dec 04 05:13:14 http://pastebin.com/m6e30c3f6 Dec 04 05:13:52 sorry it took so long, just built a few minutes ago...repo updated a few days ago Dec 04 05:14:02 hope all of this data helps! Dec 04 05:17:41 Many thanks to jerm^ and sakoman Dec 04 05:17:45 I'm comparing now Dec 04 05:21:35 I still think that there is a problem in the recipe for gcc-4.4.1 (there is a recent patch). My compilation is with gcc-4.4.1 and it's not working. I reported it to the mailing list but nobody replied Dec 04 05:25:18 i can try to bump up to 4.4.1 Dec 04 05:25:26 what architecture are you running? Dec 04 05:25:31 target arch Dec 04 05:25:38 armv7a Dec 04 05:26:02 I'm trying with gcc4.3 at the same time Dec 04 05:41:43 gregoiregentil: my build is going to take forever...i gotta run for the night Dec 04 05:41:52 want me to send you results via email? Dec 04 05:43:17 actually...anyone around that knows the history of the gcc recipes? Dec 04 05:44:25 guess not...i'll try again later... :) Dec 04 05:44:46 gregoiregentil: if you want, I'll stay connected...pm me your email addr Dec 04 06:26:44 in my boost: log_do_compile, I have this: http://pastebin.com/d6a52b29b. Why do I have gcc-4.4.1? Where is calculated the version of gcc? I have 4.3 everywhere so I don't understand Dec 04 07:22:56 03Sebastian Spaeth  07org.openembedded.dev * ra5f31b42d1 10openembedded.git/conf/distro/shr.conf: Dec 04 07:22:56 shr.conf: set DISTRO_apm = fso-apm Dec 04 07:22:56 Signed-off-by: Sebastian Spaeth Dec 04 07:22:56 03Sebastian Spaeth  07org.openembedded.dev * r49d5f70658 10openembedded.git/recipes/ (5 files in 5 dirs): Dec 04 07:22:56 task-base: Introduce DISTRO_apm and convert recipes to use it. Dec 04 07:23:00 * This allows distros to override the apm provider with an DISTRO_apm variable Dec 04 07:23:02 * The change should be transparent and invisible for other distros Dec 04 07:23:04 * This could also be done with a virtual/apm package, and a PREFERED_PROVIDER set to apm Dec 04 07:23:06 Signed-off-by: Sebastian Spaeth Dec 04 07:31:58 03Sebastian Spaeth  07org.openembedded.dev * r80739fde7f 10openembedded.git/recipes/ (5 files in 5 dirs): Dec 04 07:31:58 Revert Introduce DISTRO_apm patch until I get permission Dec 04 07:31:58 * This was committed accidentally Dec 04 07:31:58 * This reverts commit 49d5f70658ab0386bee5869656d0405a6d416e35. Dec 04 07:46:39 good morning Dec 04 07:47:20 hrw|gone: thanks for adding simone, i have one as well :) Dec 04 07:48:36 http://elinux.org/CELF_Open_Project_Proposal_2010 Dec 04 07:48:40 morning Dec 04 07:59:30 Jay7: morning, thanks for the link Dec 04 08:04:25 I do confirm at least that classpath-native is broken with gcc-4.4 while it's working with gcc-4.3 Dec 04 08:07:19 * Jay7 will propose kexecboot to CELF :) Dec 04 08:28:09 03Robert Nelson  07org.openembedded.dev * r8435956a8f 10openembedded.git/recipes/linux/ (2 files in 2 dirs): Dec 04 08:28:09 linux-omap 2.6.32: mach to plat change for ehci-omap.c Dec 04 08:28:09 drivers/usb/host/ehci-omap.c:40:22: error: mach/usb.h: No such file or directory Dec 04 08:28:19 03Robert Nelson  07org.openembedded.dev * r46a9116310 10openembedded.git/recipes/linux/linux-omap-2.6.32/beagleboard/defconfig: linux-omap 2.6.32: disable CONFIG_CPU_IDLE Dec 04 08:28:20 03Robert Nelson  07org.openembedded.dev * r9c7c5ce888 10openembedded.git/recipes/linux/linux-omap-2.6.32/beagleboard/defconfig: linux-omap 2.6.32: enable mmc power, CONFIG_TWL4030_POWER replaces CONFIG_TWL4030_MADC Dec 04 08:49:52 03Sebastian Spaeth  07org.openembedded.dev * ra2d188c697 10openembedded.git/recipes/calc/ (calc_0.0.2.bb calc_0.0.3.bb calc_git.bb): Dec 04 08:49:52 calc: bump to v 0.0.4. Move to openmoko-3rdparty dir Dec 04 08:49:52 Signed-off-by: Sebastian Spaeth Dec 04 09:03:41 mickey|office: good morning Dec 04 09:27:10 pb_: morning Dec 04 09:27:38 hi ant_work Dec 04 09:28:37 jo Dec 04 09:28:51 moin moin woglinde Dec 04 09:28:56 he ant Dec 04 09:29:39 woglinde: rebuild from scratch fixed that nasty ldap issue... Dec 04 09:29:46 hm Dec 04 09:29:51 why? Dec 04 09:29:59 he Dec 04 09:30:24 I can't imagine why opkg-cl was linked against libldap.xx Dec 04 09:30:35 can't yet... Dec 04 09:30:56 you should have examie config.log Dec 04 09:31:06 but as some wrote on the ml or here Dec 04 09:31:08 its wascurl Dec 04 09:31:10 curl Dec 04 09:31:13 because of that broken opkg-cl I could not rebuild opkg-native Dec 04 09:31:45 woglinde: for me is clear what I did: I gor rid of openldap on my Gentoo buildhost Dec 04 09:31:52 link against curl.la which says you need ldap.la Dec 04 09:31:55 and that broke opkg-* Dec 04 09:32:13 ah,ok Dec 04 09:32:23 abiword now has the similar problem Dec 04 09:32:27 it drags in libwmf Dec 04 09:32:40 which is not in oe and then using host system one Dec 04 09:32:45 argh Dec 04 09:32:48 2 ways to solve Dec 04 09:32:57 disable wmf support or package libwmg Dec 04 09:33:02 args libwmf Dec 04 09:33:16 iirc pb_ has a snipped of code for that situations..perhaps was only for auto-tooled recipes? Dec 04 09:33:22 libwmf is fairly old Dec 04 09:33:24 http://wvware.sourceforge.net/libwmf.html#news Dec 04 09:34:11 wow..2002 Dec 04 09:34:21 yeah Dec 04 09:41:08 jo florian Dec 04 09:44:27 good morning Dec 04 09:44:57 03David-John Willis  07org.openembedded.dev * rfdcd0cf2ee 10openembedded.git/conf/machine/omap3-pandora.conf: omap3-pandora: Update machine config to define versions of x-load/u-boot to use. Dec 04 09:44:59 03David-John Willis  07org.openembedded.dev * rcbca9b79b7 10openembedded.git/MAINTAINERS: MAINTAINERS: Add myself as omap3-pandora machine maintainer. Dec 04 09:44:59 03David-John Willis  07org.openembedded.dev * rbf3f7a4d6f 10openembedded.git/recipes/x-load/x-load-omap3pandora_git.bb: x-load-omap3pandora: Bump SRCREV to include early start for batt charging. Dec 04 09:45:00 03David-John Willis  07org.openembedded.dev * r040fed82ad 10openembedded.git/recipes/u-boot/u-boot-omap3pandora_git.bb: u-boot-omap3pandora: Bump GIT SRCREV (USB charge fix) Dec 04 09:45:03 03David-John Willis  07org.openembedded.dev * rf0efa83cf3 10openembedded.git/ (3 files in 2 dirs): Dec 04 09:45:06 libsdl: Add 1.2.14 (and native). Note: Does not use libsdl.inc as that has hacks for older SDL versions. Dec 04 09:45:08 * Only tested on the OpenPandora under Angstrom currently. Dec 04 09:45:10 03David-John Willis  07org.openembedded.dev * rd528b9fca9 10openembedded.git/recipes/ttf-fonts/ttf-freefonts_20090104.bb: ttf-freefonts: Add GNU FreeFont (Free UCS Outline Fonts) package. Dec 04 09:45:13 03David-John Willis  07org.openembedded.dev * rcf32e41887 10openembedded.git/recipes/xorg-driver/ (xf86-input-evdev_2.2.6.bb xf86-input-evdev_2.3.0.bb): xf86-input-evdev: Add 2.2.6 and 2.3.0 (2.3.0 disabled for now). Dec 04 09:45:22 03David-John Willis  07org.openembedded.dev * rba9ab8c783 10openembedded.git/ (3 files in 2 dirs): libpng: Add target and native recipes for 1.2.40 with checksums. Dec 04 09:45:24 03David-John Willis  07org.openembedded.dev * r32acd42a04 10openembedded.git/recipes/tslib/ (3 files in 2 dirs): tslib: Add config files for the omap3-pandora. Dec 04 09:45:29 03David-John Willis  07org.openembedded.dev * rc3316cf01d 10openembedded.git/recipes/zsh/zsh_4.3.10.bb: Dec 04 09:45:32 zsh: Add recipe for 4.3.10 (latest upstream). Dec 04 09:45:34 * Checksum was already in OE. Dec 04 09:45:36 03David-John Willis  07org.openembedded.dev * r29ee333bea 10openembedded.git/recipes/xfce-extras/gigolo_0.3.2.bb: gigolo: Add build and runtime depends on gvfs (without it, the app is not a lot of use). Dec 04 09:45:46 03David-John Willis  07org.openembedded.dev * r7d8f5bda03 10openembedded.git/ (conf/checksums.ini recipes/devilspie/devilspie_0.22.bb): Dec 04 09:45:49 devilspie: Add the devilspie daemon - version 0.22. Dec 04 09:45:51 * devilspie is a small daemon for managing X window behaviour based on rules. Dec 04 09:45:53 03David-John Willis  07org.openembedded.dev * r1df0bde673 10openembedded.git/conf/distro/include/angstrom-2008-preferred-versions.inc: angstrom-2009.X: Bump SDL upto 1.2.14 Dec 04 09:45:58 03David-John Willis  07org.openembedded.dev * rc33a047eed 10openembedded.git/recipes/xorg-font/xorg-font-common.inc: xorg-font-common: Update include to deal with newer fonts installed to /usr/share and add INC_PR. Dec 04 09:46:02 03David-John Willis  07org.openembedded.dev * r9112ebc764 10openembedded.git/recipes/xorg-font/ (font-adobe-100dpi_1.0.1.bb font-adobe-75dpi_1.0.1.bb): font-adobe-*dpi: Add 1.0.1 versions of adobe fonts. Dec 04 09:46:10 03David-John Willis  07org.openembedded.dev * r303ae98b5f 10openembedded.git/recipes/xorg-font/ (41 files): xorg-font: Add INC_PR to all the Xorg fonts. Dec 04 09:46:13 03David-John Willis  07org.openembedded.dev * r8fce963836 10openembedded.git/recipes/sudo/libgksu_2.0.12.bb: libgksu: Add --disable-gtk-doc to recipe. Dec 04 09:46:16 03David-John Willis  07org.openembedded.dev * r197b0cb11c 10openembedded.git/recipes/xorg-font/font-alias_1.0.2.bb: font-alias: Add 1.0.2 Dec 04 09:46:21 03David-John Willis  07org.openembedded.dev * rf591b28ed9 10openembedded.git/recipes/xorg-font/ (33 files): xorg-font: Add latest versions of the XOrg font packages. Dec 04 09:46:24 03David-John Willis  07org.openembedded.dev * r0c3c0ca02c 10openembedded.git/recipes/xorg-font/ (encodings_1.0.2.bb encodings_1.0.3.bb): encodings: Add latest Xorg font encodings package 1.0.3. Dec 04 09:46:27 03David-John Willis  07org.openembedded.dev * re18e677774 10openembedded.git/recipes/xchat/xchat_2.8.6.bb: xchat 2.8.6: Drop TCL support from build. Dec 04 09:46:32 03David-John Willis  07org.openembedded.dev * rb5261869ab 10openembedded.git/conf/distro/include/angstrom-2008-preferred-versions.inc: angstrom-2009.X: Add util-macros 1.3.0 as a pref. version. Dec 04 09:46:35 03David-John Willis  07org.openembedded.dev * ra2a6fd04c5 10openembedded.git/recipes/xorg-util/util-macros-native_1.3.0.bb: util-macros: Add 1.3.0 native. Dec 04 09:46:38 03David-John Willis  07org.openembedded.dev * ra0fb85891d 10openembedded.git/recipes/xorg-font/ (font-util-native_1.1.1.bb font-util_1.1.1.bb): font-util: Add 1.1.1 target and native. Dec 04 10:02:55 ~seen jama Dec 04 10:02:58 jama is currently on #openmoko-cdevel (15h 44m 19s). Has said a total of 55 messages. Is idling for 17m 48s, last said: 'mrmoku: you can.. it should be the same'. Dec 04 10:03:12 Hi Dec 04 10:03:43 hi sgh Dec 04 10:03:52 Mesa-dri fails to compile. The error starts with this: "../Makefile.template:106: depend: No such file or directory" Dec 04 10:03:58 03Martin Jansa  07org.openembedded.dev * rc88bf809f3 10openembedded.git/recipes/xorg-font/ (font-util-native_1.1.1.bb font-util_1.1.1.bb): Dec 04 10:03:58 font-util: Fix cross building, use PKG_CONFIG_PATH in MAPFILES_PATH test Dec 04 10:03:58 Signed-off-by: Martin Jansa Dec 04 10:04:18 sgh mesa 7.4? Dec 04 10:04:54 woglinde: pong Dec 04 10:05:08 woglinde: yes Dec 04 10:05:11 jama how is xorg 7.5 going? Dec 04 10:05:14 hi esbenh Dec 04 10:05:26 woglinde: good for gta02 and spitz Dec 04 10:05:58 did you upgrade all x-packages to versions of xorg 7.5 now? Dec 04 10:05:59 woglinde: but I have some issues with spitz now.. not sure if its from 1.7.2 but 1.7.3 didn't resolve it now.. Dec 04 10:06:27 woglinde: yes when xorg-7.5 was announced and then also some upgrades later.. Dec 04 10:06:56 jama ah by the way as I saw it yesterday Dec 04 10:07:03 when you do inherit autotools Dec 04 10:07:14 you dont need an extra do_stage anymore Dec 04 10:07:50 woglinde: in which package? Dec 04 10:08:05 hm didnt remember Dec 04 10:08:35 maybee I will find it with qgit Dec 04 10:10:45 hm Dec 04 10:10:48 cannt find Dec 04 10:10:57 or it was mrmoku Dec 04 10:12:08 I haven't add inherit to autotools in same commit as adding do_stage but if it was recipe from someone else and I just pushed it than maybe I wasn't carefull enough to remove it... Dec 04 10:12:39 ;) Dec 04 10:13:58 args Dec 04 10:14:03 jama please fix 0d7c70004bf85ee59d8ce48d6d83c68b313859bd Dec 04 10:14:12 that will broke micro Dec 04 10:14:43 be broken Dec 04 10:15:33 pb_: compatability-providers.inc sounds ok to me. Although I think it is really something like "default-providers.inc" which sets a default provider that is used unless distros override it... Dec 04 10:16:02 woglinde: whats broken there? Dec 04 10:16:20 woglinde: how can mokoeightball break sth? Dec 04 10:16:48 pb_: as mickey and koen commented on the patch now, are you fine with me committing this now (to fix angstrom) and work towards a default-providers.inc then? Dec 04 10:17:09 woglinde: me? Dec 04 10:18:47 spaetz: another posiblillity is to add an shr override to the RPROVIDES Dec 04 10:19:03 JaMa: what do you think about including default providers in bitbake.conf as koen proposed? Dec 04 10:19:24 03Koen Kooi  07org.openembedded.dev * r99f78d4002 10openembedded.git/recipes/xorg-font/ (font-util-native_1.0.1.bb font-util_1.0.1.bb): font-util: fix merge damage Dec 04 10:19:29 Heinervdm: if that works.... I still think virtual/apm might be the best solution Dec 04 10:19:35 03Koen Kooi  07org.openembedded.dev * rbcf9669002 10openembedded.git/: Merge branch 'org.openembedded.dev' of git@git.openembedded.org:openembedded into org.openembedded.dev Dec 04 10:19:59 jam installing to /usr stuff Dec 04 10:20:09 /usr stuff dont exist on micro Dec 04 10:20:09 spaetz: yes a virtual is the best solution Dec 04 10:20:20 Heinervdm I correct me Dec 04 10:20:24 03Koen Kooi  07org.openembedded.dev * rea85fd4e1f 10openembedded.git/conf/distro/include/sane-srcrevs.inc: efl: bump SRCREV to fix some crashes Dec 04 10:20:28 it will broken on micro Dec 04 10:20:52 woglinde: ah ok, now i understand, so using $datadir will fix it? Dec 04 10:22:38 ${datadir} ${bindir} Dec 04 10:22:40 and so on Dec 04 10:22:45 dont hardcode paths Dec 04 10:23:05 but maybee the package it self needs to be told to search other paths Dec 04 10:24:38 woglinde: probably, but that app will never run on another device then the om-gta0{1,2} because it has hardcoded sysfs paths in it Dec 04 10:27:07 heinervdm whats prevent me to use micro on gta's? Dec 04 10:27:24 woglinde: nothing :) Dec 04 10:27:50 wouldnt it be better to abstract the sysfs Dec 04 10:28:02 so it can run on other devices with accels? Dec 04 10:28:05 like n900 Dec 04 10:31:29 yes, but i think that should do the author Dec 04 10:33:07 spaetz: if mickey is happy, I am delighted. so yes, go ahead. Dec 04 10:33:25 morning pb Dec 04 10:33:30 hi woglinde Dec 04 10:35:04 spaetz: looks good too, but I need to check when is bitbake.conf processed (same problem with weak assignments as with sane-providers but cannot change when bitbake.conf will be included..) Dec 04 10:38:41 hi cbrake Dec 04 10:40:20 03Martin Jansa  07org.openembedded.dev * r26396d98ae 10openembedded.git/conf/checksums.ini: checksums.ini: add checksums for newer xorg recipes from xorg-7.5 branch Dec 04 10:44:44 hm what was the BBEXTEND status now? Dec 04 10:44:49 can we use it? Dec 04 10:45:07 it would simplify x recipes a lot Dec 04 10:47:04 03Koen Kooi  07org.openembedded.dev * r04da0cb263 10openembedded.git/conf/distro/include/sane-srcrevs.inc: opkg: bump to r446 to make unarchive errors warnings again Dec 04 10:51:18 03Marco Cavallini  07org.openembedded.dev * r3512b7de2d 10openembedded.git/conf/checksums.ini: checksums.ini added checksum for wxWidgets-2.8.10 Dec 04 10:52:51 morning Dec 04 10:53:06 mickeyl: good to know. any plans to make something with sim.one? Dec 04 10:55:35 hi hrw Dec 04 11:00:12 03Sebastian Spaeth  07org.openembedded.dev * r3943c0dac3 10openembedded.git/ (5 files in 5 dirs): Dec 04 11:00:12 task-base: Introduce DISTRO_APM and convert recipes to use that Dec 04 11:00:12 Signed-off-by: Sebastian Spaeth Dec 04 11:01:25 time to unsubscribe from oe-commits Dec 04 11:01:40 lol Dec 04 11:01:59 ? Dec 04 11:02:08 nevermind Dec 04 11:03:29 if I do a "bitbake -f -c compile foo" and "bitbake foo" afterwards, wil the previously changed sources get removed, if I inherit "rm_work" in my local.conf? Dec 04 11:03:50 rm_work will delete it Dec 04 11:03:56 so be aware Dec 04 11:04:01 to copy the patch Dec 04 11:04:08 I ofen enough was biten by this Dec 04 11:04:40 ic, but bitbake foo will create a new ipk? Dec 04 11:04:43 you can add "do_rm_work() { ; }" into recipe to not get it removed Dec 04 11:05:17 Weaselweb sure but you cannt reproduce it Dec 04 11:05:25 woglinde: yep, that's for sure Dec 04 11:05:33 without changing source code again Dec 04 11:06:28 03Sebastian Spaeth  07org.openembedded.dev * r9b1bb99f54 10openembedded.git/recipes/freesmartphone/fso-apm_git.bb: Dec 04 11:06:28 fso-apm: remove R|PROVIDES=apm Dec 04 11:06:28 Signed-off-by: Sebastian Spaeth Dec 04 11:07:12 hrw: sorry for the commit noise. Dec 04 11:07:45 spaetz: no problem - I jsut do not have a time to check all commits Dec 04 11:08:02 heh, I guess so Dec 04 11:10:34 hrw: dunno, i didn't think about a usecase yet. what are you going to use it for? Dec 04 11:10:45 jama ping Dec 04 11:11:01 jama could you push in xorg 7.5 recipes with default-pref -1? Dec 04 11:11:12 so we can test it easily against dev Dec 04 11:12:41 woglinde: as I pointed out there is no need for that, you can fix your favourite distro at 7.4 by including 1 or two files Dec 04 11:13:02 xora no need for 7.5? Dec 04 11:13:05 mickey|office: no use Dec 04 11:13:16 woglinde: no need to make it def_pref -1 Dec 04 11:13:21 ah Dec 04 11:13:22 *g* Dec 04 11:13:28 mickey|office: so far I jsut booted it, got 2.6.31 and 2.6.32 booted Dec 04 11:13:32 hrw: *nod* oh well, will find something Dec 04 11:13:51 woglinde: just include preferred-xorg-X11R7.4 and X11R4.5udates Dec 04 11:14:13 mickey|office: http://marcin.juszkiewicz.com.pl/tmp/pic14.jpg Dec 04 11:14:14 4.5 was supposed to be 7.4 Dec 04 11:14:37 xora hm yes understand it now Dec 04 11:14:47 will change micro back Dec 04 11:14:48 hrw: nice Dec 04 11:15:09 palm pre should be fast enough to run qt Dec 04 11:15:17 then again, i love the webos UI Dec 04 11:15:27 mickey|office: next step in UI is making 3 windows instead of 3 columns Dec 04 11:15:31 but don't fancy coding w/ web technologies... Dec 04 11:15:43 mickey|office: and finding out how maemo5 icons are named Dec 04 11:15:46 Do you not find it horribly slow? Dec 04 11:16:03 no, i find it quite snappy Dec 04 11:16:04 I played with some in shops and they were fairly painful. Dec 04 11:16:15 * XorA hopes that webos is at least better than google wave for speed Dec 04 11:16:15 Very long startup times for apps Dec 04 11:16:24 http://marcin.juszkiewicz.com.pl/tmp/pic15.jpg is desktop ver Dec 04 11:16:35 is relative Dec 04 11:16:44 compared to the freerunner it's quite snappy Dec 04 11:16:50 yes, it's slower than the iphone Dec 04 11:16:59 Compared to the freerunner anything is fast. Dec 04 11:17:01 but it's fast enough for my taste Dec 04 11:17:14 yep, that's why the FR is a very good development platform Dec 04 11:17:36 if soft is fast on FR then fly on other? Dec 04 11:18:00 hrw do you use qtdesigner? Dec 04 11:18:17 sometimes I wish QT wasnt so alien Dec 04 11:18:24 xora? Dec 04 11:18:40 woglinde: yes Dec 04 11:18:54 woglinde: I do not like apps which makes UI in code Dec 04 11:19:12 woglinde: UI Designer allows me to quickly change UI without touching code Dec 04 11:19:18 hrw yeah Dec 04 11:19:25 it rockz Dec 04 11:19:39 woglinde: I find it a wierd sudo language, all that MOC stuff Dec 04 11:19:40 but it was hard to me to understand the layout concept Dec 04 11:19:50 woglinde: it may be thats died since I last looked at Qt Dec 04 11:19:55 so resizing was as I wanted Dec 04 11:20:04 woglinde: http://marcin.juszkiewicz.com.pl/2006/04/27/mileage-hires/ Dec 04 11:23:14 hm I will cleanup sdl-recipe Dec 04 11:23:18 its looks so nasty Dec 04 11:24:31 woglinde: layouts are not so bad when properly used Dec 04 11:26:43 the rootfs is build from ipks? Dec 04 11:27:20 yes Dec 04 11:27:21 weaselweb yes Dec 04 11:42:46 woglinde: sorry. It is mesa-dri-7.0.3 Dec 04 11:46:24 woglinde: you can test if if you git merge that branch.. I've merged oe.dev few mins ago.. so no conflicts should be there.. Dec 04 11:48:05 btw koen bumped EFL_SRCREV.. Dec 04 11:48:21 and I merged oe.dev to shr.. so hopefully it wont break sth Dec 04 11:48:28 ahh wrong channel Dec 04 11:48:42 spaetz: mrmoku: it was for you :) Dec 04 11:49:16 to change the busbyox config i need to modify defconfig in ${WORKDIR}? i guess there is no make menuconfig support for that Dec 04 11:49:44 right Dec 04 11:50:08 JaMa: ok :) Dec 04 11:50:41 are the targets for bitbake (like compile, clean, ...) predefined? or can I extend them on one package? Dec 04 11:51:03 hm change the config in recipe dir Dec 04 11:51:24 weaselweb you can extend them Dec 04 11:51:37 look at the various .bbclass cmake for instance Dec 04 11:51:52 you can do _append _prepend to each state Dec 04 11:52:00 too Dec 04 11:52:27 i would add a new command for busybox to create a modified config using menuconfig Dec 04 11:54:30 not sure how ie reacts when in defined BB_THREADS Dec 04 11:55:31 badly, I suspect, but the kernel already has a menuconfig task so this would not be a new problem Dec 04 11:56:14 ~blame Nokia for lack of QToolBar in Qt Designer Dec 04 11:56:16 * ibot blames simon42 and perhaps Nokia for lack of QToolBar in Qt Designer and the end of the world as we know it Dec 04 11:57:21 hrw hm Dec 04 11:57:33 hrw are you using the newest? Dec 04 11:57:36 pkg-config is OE is strange. without changing PKG_CONFIG_PATH to "${STAGING_LIBDIR_NATIVE}/pkgconfig" the path from buildhost is returned... Dec 04 11:57:37 yes Dec 04 11:57:41 4.6.0rc1 Dec 04 11:57:54 ok, 4.6.0 was released but that does not changed Dec 04 11:57:59 I take the responsibility for the "end of the world" thingy... but I never, ever touched something in Qt (-: Dec 04 11:58:10 hrw you could ask on qt Dec 04 11:58:19 maybee there is a reason Dec 04 11:58:24 woglinde: I can code it by hand too Dec 04 12:24:31 woglinde: can you compile mesa-dri? Dec 04 12:26:49 sgh beaglboard and mesa 7.4 works Dec 04 12:28:10 sgh: ok ... strange .... does mesa-dri depend on a specific kernel-configuration ? Dec 04 12:28:10 sgh: ok ... strange .... does mesa-dri depend on a specific kernel-configuration ? Dec 04 12:28:10 sgh: ok ... strange .... does mesa-dri depend on a specific kernel-configuration ? Dec 04 12:28:16 sorry ... Dec 04 12:29:02 sgh hm Dec 04 12:29:15 mesa depends on libdrm Dec 04 12:29:25 which might me depends on some kernel versions Dec 04 12:30:09 woglinde: I'll investigate. I will paste a build error Dec 04 12:30:27 we are behind with libdrm anyway Dec 04 12:40:15 03Martin Jansa  07org.openembedded.dev * r6a70394463 10openembedded.git/recipes/tasks/task-base.bb: Dec 04 12:40:15 task-base: remove quotes around DISTRO_APM Dec 04 12:40:15 Signed-off-by: Martin Jansa Dec 04 12:44:15 jo crofton Dec 04 12:46:02 gm Dec 04 12:47:17 did someone build the usermanual from doc subdir recently? i got error like this a lot: chapters/introduction.xml:20: parser error : Entity 'Aring' not defined Dec 04 12:47:17 OpenZaurus, Ångström, Familiar and SlugOS. Dec 04 12:52:24 woglinde: http://pastebin.com/d331c3624 Dec 04 12:52:31 woglinde: that is the error Dec 04 13:05:27 let me check Weaselweb Dec 04 13:08:23 sgh which libdrm? Dec 04 13:08:31 is used Dec 04 13:08:37 looks like the libdrm is to new Dec 04 13:08:42 Weaselweb: with simple 'make' I got: html, xhtml, pdf, ps, dvi. rtf, tex versions of usermanual Dec 04 13:08:55 hrw: mh Dec 04 13:09:22 Weaselweb: on Debian 'sid' Dec 04 13:09:34 coult that be the reason? "./usermanual.xml:31: warning: failed to load external entity "http://www.oasis-open.org/docbook/xml/4.2/docbookx.dtd"]>" Dec 04 13:09:41 Weaselweb: yes Dec 04 13:09:41 could Dec 04 13:09:43 woglinde: libdrm_2.4.11.bb Dec 04 13:09:50 hrw: you know how to fix that? Dec 04 13:10:01 yeah Dec 04 13:10:02 Weaselweb: but I am not quite sure about it. I am not docbook expert Dec 04 13:10:04 try older libdrm Dec 04 13:10:26 woglinde: ok ... is is enought to clean the curent libdrm and compile abnother afterwards ? Dec 04 13:10:54 to be sure delete header and lib from staging yourself Dec 04 13:11:04 ok Dec 04 13:11:07 woglinde: ok Dec 04 13:14:00 re ant Dec 04 13:14:19 hrw: another package was missing. thanks though Dec 04 13:20:56 woglinde: apparently it did not work..... I will try to kill tmp and ensure that libdrm is locked to 2.3.0 and then build mesa-dri only. Dec 04 13:49:21 03Koen Kooi  07org.openembedded.dev * rd843cfb729 10openembedded.git/recipes/linux/linux-omap_2.6.32.bb: linux-omap 2.6.32: bump SRCREV Dec 04 13:51:39 My current breakage after coming back to compiling for OE after a few weeks is http://pastebin.com/f6a3e7654 Dec 04 13:51:44 What's happening there? Dec 04 13:53:42 nevermind Dec 04 13:53:51 I had a patch for testing applied Dec 04 13:53:56 it turned out that patch was faulty Dec 04 14:02:46 morning Dec 04 14:06:46 morning kergoth Dec 04 14:11:46 03Koen Kooi  07org.openembedded.dev * r4252b6f04c 10openembedded.git/recipes/efl1/ecore_svn.bb: ecore: maxrev iconv patch Dec 04 14:27:21 mwester: ...I read your reply... :/ Dec 04 14:28:18 and btw we have to support some Zaurus with 32M ram too... Dec 04 14:36:34 ant_work: poodle/corgi? Dec 04 14:37:02 yes... Dec 04 14:37:58 and oneday bcm634x routers my next target...) Dec 04 15:02:47 how do I get git to show me both author and committer? Dec 04 15:03:11 use qgit Dec 04 15:03:19 good idea Dec 04 15:03:25 not possible in this case Dec 04 15:03:30 why? Dec 04 15:03:31 remote machine Dec 04 15:03:38 hm Dec 04 15:03:40 git show ---format==full Dec 04 15:03:45 broonie: thanks Dec 04 15:15:50 re raster Dec 04 15:24:01 woglinde: wogzaaa! Dec 04 15:24:15 hm was my efl patch applied upstream Dec 04 15:24:23 or why koen maxreved it Dec 04 15:29:25 ant_work, I'm glad to know that I do not have the only tiny-memory devices! Dec 04 15:30:13 ant_work, I know that the *right* solution is to test the latest opkg sources, and add patches for the latest versions -- but I have no time to do so right now :( Dec 04 15:30:56 03Martin Jansa  07org.openembedded.dev * r1d3881e24b 10openembedded.git/recipes/linphone/files/b64_assert.patch: Dec 04 15:30:56 linphone: merge missing patch from shr/import Dec 04 15:30:56 Signed-off-by: Martin Jansa Dec 04 15:42:27 hi Dec 04 15:42:53 I'm trying to build angstrom minimal-image, based on the angstrom webpage instructions Dec 04 15:43:37 bitbake stops with the following: Patch gcc43-c++-builtin-redecl.patch does not apply (file not found) Dec 04 15:43:54 I pulled OE repository today Dec 04 15:44:32 the task is gcc-cross-initial_4.3.1.bb do_patch Dec 04 15:44:38 any idea what can I do? Dec 04 15:48:33 http://tinderbox.openembedded.net/packages/gcc-cross-initial/ looks good Dec 04 15:48:59 is that for .dev or the stable branch Dec 04 15:53:55 twinsen ^^ Dec 04 15:55:45 bye Dec 04 15:57:23 I checked out 2009-stable Dec 04 15:59:31 i don't see on the tinderbox page 4.3.1 only .2 and .3 Dec 04 16:00:14 i did this: git checkout origin/stable/2009 -b stable/2009 Dec 04 16:00:38 I'm not sure the stable branch is well supported Dec 04 16:00:51 actually I'm quite new to OE _and_ git... Dec 04 16:00:55 I'm surprised to see something so fundamental break, though Dec 04 16:01:09 wow :) that sentence rocks! Dec 04 16:01:36 OE is trying hard but so far doing a fairly bad job at delivering consistent quality Dec 04 16:01:45 what are u guys using? Dec 04 16:01:49 .dev Dec 04 16:02:01 ok, how to checkout that? Dec 04 16:02:07 It's where fixes happen Dec 04 16:02:25 But it's also bleeding edge and sometimes/very often things break Dec 04 16:02:39 twinsen: should be in the wiki, I think Dec 04 16:02:54 sounds fair :) I used stable because the angstrom tutorial said to, i'm ok with dev Dec 04 16:03:36 I only want to build a beagleboard kernel, and the sgx drivers/libs (all other stuff is handled by gentoo on the board) Dec 04 16:03:44 ok, I'll check the wiki Dec 04 16:03:47 thanks! Dec 04 16:26:19 this is not good... bitbake dies with an error: NameError: global name 'os' is not defined (File "", line 18, in check_sanity) Dec 04 16:26:48 (i switched to org.openembedded.dev branch) Dec 04 16:28:06 check the mailing list archives; something like that was mentioned not too long ago. Dec 04 16:29:10 ok :) Dec 04 16:41:41 03Martin Jansa  07org.openembedded.dev * reebba7062b 10openembedded.git/conf/checksums.ini: Dec 04 16:41:41 checksums.ini: add linphone checksum and move wxWidgets-2.8.10.tar.bz2 checksum to right place Dec 04 16:41:41 Signed-off-by: Martin Jansa Dec 04 17:05:12 03Martin Jansa  07org.openembedded.dev * rd39f508544 10openembedded.git/recipes/linphone/files/b64_assert.patch: linphone: patch needs CR+LF to apply, sorry for that, now it compiles fine Dec 04 17:08:07 hi otavio Dec 04 17:11:46 hey...anyone here familiar with the gcc recipes for the target? Dec 04 17:11:57 I'm suddenly seeing a QA error in packaging Dec 04 17:13:20 which appears to be caused by the symbolic link created here: http://cgit.openembedded.org/cgit.cgi/openembedded/tree/recipes/gcc/gcc-package-target.inc#n129 Dec 04 17:13:58 that symlink gets sucked into the non -dev gcc package, causing the QA error.... Dec 04 17:14:23 jerm hm I see only warnings Dec 04 17:14:33 but you are right gcc package is broken somehow Dec 04 17:14:38 hmm...i'm getting an error Dec 04 17:14:51 there is also an issue that causes a conflict with meta-toolchain Dec 04 17:15:14 a decision is made that installs the nof (soft-float) libgcc if it is there...otherwise it installs the hard-float Dec 04 17:15:20 in base_libdir Dec 04 17:16:10 hm ah this problem Dec 04 17:16:13 when this package (libgcc1) gets installed using meta-toolchain it overwrites the hard-float libgcc installed by the gcc-cross-sdk package Dec 04 17:16:50 then libstdc++, which is the hard-float version, cannot link against it b/c of GCC symbol versioning on some of the functions it needs from libgcc Dec 04 17:17:12 qmul, etc is GCC_4.2 for soft-float versions, but GCC_4.1 for hard-float versoins Dec 04 17:17:14 versions Dec 04 17:17:22 this is all on powerpc Dec 04 17:18:35 so I was trying to figure out why the symlink is being created and why the soft-float version of libgcc is preferred in this recipe Dec 04 17:18:52 when making the libgcc1 package Dec 04 17:19:25 hm in which file it is? Dec 04 17:19:32 unfortunately I have to go to the office...where IRC is banned...ughh Dec 04 17:19:38 hm Dec 04 17:19:42 even through web? Dec 04 17:19:44 woglinde: i am building gcc_4.3.3 Dec 04 17:19:47 write to the ml again Dec 04 17:19:49 yep...even through web Dec 04 17:20:05 I was planning on it...just wanted to see if anyone had any ideas Dec 04 17:20:16 I think you localzied Dec 04 17:20:19 it Dec 04 17:20:25 Koen was the last to touch the gcc recipe so I thought he may have ideas Dec 04 17:20:25 hm but I will look into it maybee Dec 04 17:20:29 or ask khem later Dec 04 17:20:39 ok...thanks! Dec 04 17:20:48 cya Dec 04 17:40:15 are there any docs that explain how to go about creating custom machine setup? Dec 04 17:40:38 hm? Dec 04 18:47:26 03Alan Campbell  07org.openembedded.dev * r08dcecdc77 10openembedded.git/recipes/ti/files/mp3dec_cs1omap3530.patch: (log message trimmed) Dec 04 18:47:26 ti-cs1omap3530: add MP3 decode support to OMAP3 DVSDK 3.0 GA DSP codec-combo Dec 04 18:47:26 Prerequisites : Dec 04 18:47:26 - Download MP3 Decoder package from OMAP35x DVSDK 3.0 Download page Dec 04 18:47:26 - install, untar it Dec 04 18:47:27 - add the mp3dec codec to the codec-server build via either copying the Dec 04 18:47:31 directory (recursively), symbolic link to additional repository, or Dec 04 18:47:33 03Koen Kooi  07org.openembedded.dev * rb788499815 10openembedded.git/ (77 files in 8 dirs): (log message trimmed) Dec 04 18:47:36 recipes/ti: merge in updates Dec 04 18:47:38 * New codec recipes from Arago team in Germantown Dec 04 18:47:40 * Updated dm6446 and dm355 from Brijesh Singh Dec 04 18:47:43 * cmem, gst-ti and dmai fixes from Brijesh Singh Dec 04 18:47:44 angstrom: introduce SOC_FAMILY and start using it TI recipes as a testing ground Dec 04 18:47:46 * SOC_FAMILY tells you which system on chip is used in the MACHINE, e.g. omap3, dm355, pxa25x, s3c24xx. Dec 04 18:47:48 03Koen Kooi  07org.openembedded.dev * ra16e99914c 10openembedded.git/conf/machine/ (da850-omapl138-evm.conf hawkboard.conf): hawkboard, da850-omapl138-evm: use l138 include Dec 04 18:47:53 03Koen Kooi  07org.openembedded.dev * re23f7ee8b6 10openembedded.git/recipes/powervr-drivers/ (libgles-omap3.inc omap3-sgx-modules_1.4.14.2514.bb): libgles-omap3, omap3-sgx-modules: switch to new style installjammer unpacker Dec 04 18:48:39 re Dec 04 18:50:49 03Roger Monk  07org.openembedded.dev * r9700bd3af7 10openembedded.git/conf/machine/dm355-evm.conf: dm355-evm: fix UBOOT machine config Dec 04 19:08:35 ERROR: Multiple .bb files are due to be built which each provide opkg (/oe/org.openembedded.dev/recipes/opkg/opkg-nogpg-nocurl_svn.bb /oe/org.openembedded.dev/recipes/opkg/opkg-nogpg_svn.bb). Dec 04 19:08:40 bbl Dec 04 19:11:27 Ouch. Dec 04 19:11:44 * mwester goes to check his build logs to see if also encountered that error. Dec 04 19:12:50 Nope, I did not encounter that one, just a compile failure for opkg-nogpg-nocurl_svn.bb Dec 04 20:12:04 hi all Dec 04 20:14:08 hi thesing Dec 04 20:15:12 hi thesing Dec 04 20:37:09 gm to all Dec 04 20:37:24 hello khem Dec 04 20:42:28 hi khem Dec 04 20:45:59 hi khem Dec 04 20:47:34 * khem is running a build after weeks Dec 04 20:47:36 re xora Dec 04 20:47:42 khem good luck Dec 04 20:47:46 opkg might break Dec 04 20:47:55 hi florian woglinde and XorA Dec 04 20:48:08 and here it breaks uclibc/nptl Dec 04 20:48:11 khem: are you still intending to stand for TSC? Dec 04 20:48:44 XorA: with all you guys at helm :) Dec 04 20:48:58 khem: there isnt that many all you guys :-( Dec 04 20:49:06 infact Im surprised at the lack of response Dec 04 20:49:06 XorA: I am interested though Dec 04 20:49:20 currently I count only 4 candidates Dec 04 20:50:31 khem: see the topic, Ill be starting the voting on Sunday Dec 04 20:50:45 XorA: ok Dec 04 20:50:57 I will express my interest on ml Dec 04 20:51:05 and in general kick any other people you think should apply Dec 04 20:51:38 that goes for everyone :-) Dec 04 20:52:14 ok Dec 04 20:52:19 * XorA shall be in USA timezone tomorrow, so that gives people an extra 5 hours :-) Dec 04 20:52:29 XorA: cool .. Dec 04 20:52:40 will you be in east coast or west Dec 04 20:52:42 or canada Dec 04 20:52:45 Salt Lake City Dec 04 20:53:09 Utah hmm Dec 04 20:53:18 not far from California Dec 04 20:53:44 poor khem Dec 04 20:53:55 or rather, X0rA Dec 04 20:54:03 hm Dec 04 20:54:08 4 people is good Dec 04 20:54:11 or 5 Dec 04 20:55:00 ideally we want 5 Dec 04 20:55:50 yes Dec 04 20:55:58 hm Dec 04 20:56:04 zecke didnt stand up? Dec 04 20:56:15 not yet Dec 04 20:56:22 you can see all candidates on oe-members Dec 04 21:02:38 re hrw Dec 04 21:03:08 XorA: according to your mail candidates needs to delegate themself and cannot be proposed by other members - am I reading right? Dec 04 21:03:30 hrw: yes, this is what the voting policy says Dec 04 21:03:43 hrw: or my understanding of it Dec 04 21:04:08 "The candidacy period lasts two weeks. All members who declare their candidacy in a statement sent to the membership list become candidates for the election. " part of 3.2 point? Dec 04 21:04:17 yes Dec 04 21:05:10 who are the candidates thus far? Dec 04 21:05:22 hrw: it kinds of makes sense, prevents you voting in people by random Dec 04 21:05:30 kergoth: you, RP, koen me Dec 04 21:05:39 ah, cool Dec 04 21:05:55 at least thats the only emails I have seen hit the list Dec 04 21:06:17 with khem its 5 Dec 04 21:06:23 that would be good Dec 04 21:06:32 woglinde: but Khem did not candidated Dec 04 21:06:47 hrw: he said he would above :-) Dec 04 21:07:05 I know, but so far he did not Dec 04 21:07:17 with the amount of OE fanatics I expected 10 names Dec 04 21:07:21 one thing is sure - I will not candidate Dec 04 21:08:15 hrw *g* Dec 04 21:08:34 hrw: all those years of OZ you be perfect Dec 04 21:08:37 I have experience of using OE, Poky, maintaining forks/branches of them, distros built from them, handling different machines and their needs etc but prefer to be developer Dec 04 21:10:16 XorA: I do not know when we lost many devs - I do not see Rod activity for example Dec 04 21:10:23 hm anyone knows the clean solution to deal in autotools with -config shells? Dec 04 21:10:58 hm maybee check for it with the macros Dec 04 21:11:16 hrw: good point, as a board member we should try and remedy that Dec 04 21:11:26 and then in Makefile.am `-config --cflags` `-config --ldflags` Dec 04 21:11:41 this seems reasonable Dec 04 21:12:45 need to get a awesome stand display for FOSDEM this year, OE has been named in many many devices over the last year Dec 04 21:13:08 XorA: and from TSC member I would expect knowledge of recent OE changes. I lack such one because during last months I live in Poky and stable/2009 land Dec 04 21:13:12 XorA: o yes! Dec 04 21:13:46 XorA: Palm Pre, Montavista, BUG, Archos Dec 04 21:13:54 hrw: reporting every single branch is a PITA, it means I miss many many commit Dec 04 21:14:16 XorA: I am unsubscribing from oe-commits Dec 04 21:14:42 hm why nobody fixes it Dec 04 21:14:52 woglinde: lack of empowerment I assume Dec 04 21:14:59 woglinde: which TSC can fix I guess Dec 04 21:15:15 xora give me sysadmin rights Dec 04 21:15:17 XorA: btw - can we get planet.openembedded.org for OE devs blogs? planet.ltg is LTG Dec 04 21:15:26 I will fix it this evening Dec 04 21:15:26 one of the things we discussed in Berlin was making people feel more empowered Dec 04 21:15:42 woglinde: cbrake and Laibsh have admin I think, I dont Dec 04 21:16:12 hrw: mail the members list, sounds like a good idea Dec 04 21:17:41 I guess an OE drink and gripe at FOSDEM might be an idea :-) Dec 04 21:17:55 I vote in the rum bar :-) Dec 04 21:18:34 http://www.psixpda.com/ Dec 04 21:19:09 hmm, same keyboard fault as n810, but looks nice Dec 04 21:20:39 * florian wonders which kernel to use for his htc universal Dec 04 21:21:04 hrw: nice a PDA that can run WoW Dec 04 21:21:35 03Klaus Kurzmann  07org.openembedded.dev * r30d737f92f 10openembedded.git/ (3 files in 2 dirs): Dec 04 21:21:35 shr-wizard_git.bb: new recipe Dec 04 21:21:35 Signed-off-by: Klaus Kurzmann Dec 04 21:21:43 hrw: oh... well, PC based... crazy Dec 04 21:22:28 03Klaus Kurzmann  07org.openembedded.dev * re340492e1e 10openembedded.git/recipes/shr/shr-wizard_git.bb: Dec 04 21:22:28 shr-wizard_git.bb: SRCREV instead of SRCPV Dec 04 21:22:28 Signed-off-by: Klaus Kurzmann Dec 04 21:25:17 03Andrea Adami  07org.openembedded.dev * rdaf52c62bb 10openembedded.git/recipes/kexecboot/kexecboot_git.bb: Dec 04 21:25:17 kexecboot_git: bump to 456e052144f58d3c1116a38cbb5caa44c070bce8. Bump Dec 04 21:25:17 PR. Dec 04 21:25:48 http://3.bp.blogspot.com/_HQtYHYy2Hpc/SadRjeFQkTI/AAAAAAAAAuA/17Nm4Z_TV-8/S259/google.jpg Dec 04 21:27:45 mwester: it's just opie-image having that multiple thing issue... Dec 04 21:51:53 hmmm gcc-cross is not buildable for uclibc Dec 04 21:52:06 did someone else have this issue Dec 04 21:52:25 make[1]: *** [configure-target-libiberty] Error 1 Dec 04 21:54:41 hmm no change seems to cause it Dec 04 21:54:48 I guess I will clean it up Dec 04 21:54:50 and retry Dec 04 21:55:06 khem Dec 04 21:55:09 which distro? Dec 04 22:00:14 sleep time Dec 04 22:01:34 nite hrw Dec 04 22:07:28 03Klaus Kurzmann  07org.openembedded.dev * rd9be32b1c7 10openembedded.git/recipes/tasks/task-shr-minimal.bb: Dec 04 22:07:28 task-shr-minimal: add shr-wizard Dec 04 22:07:28 Signed-off-by: Klaus Kurzmann Dec 04 22:15:50 khem: hi, which machine are building for? Dec 04 22:17:16 khem: never mind, I was afraid I was to blame, but my toolchain patches are not pushed yet :-) Dec 04 22:25:19 likewise *g* Dec 04 22:25:25 re jama Dec 04 22:28:39 urg Dec 04 22:29:59 hugh Dec 04 22:32:32 03Andrea Adami  07org.openembedded.dev * ra5139b280f 10openembedded.git/recipes/kexecboot/ (2 files in 2 dirs): linux-kexecboot_2.6.32: initial commit. Dec 04 23:04:19 woglinde: mesa-dri-7.4 compiles .... why is it -1 as default preference, when mesa-dri-7.0.3 is not? Dec 04 23:06:56 sgh hm because distro includes preffered-x-versions Dec 04 23:07:11 so the defpref dont matters Dec 04 23:08:32 woglinde: ok, but still mesa-dri-7.0.3 is chosen if nothing else applies. And the current fact is that it does not compile. I would think that every mesa-dri should have a default preference of -1 to force an explicit version to be selected. Dec 04 23:10:02 sgh rais this to list Dec 04 23:10:20 when TSC is formed TSC will give advises Dec 04 23:10:25 woglinde: will do, Dec 04 23:10:30 to solve it Dec 04 23:10:37 woglinde: thanks for the help earlier today :D Dec 04 23:10:44 he no prob Dec 04 23:23:01 sgh: def pref doesnt quite work like that Dec 04 23:23:31 sgh I see nothing wrong with promoting 7.4 to prefferred anyway Dec 04 23:24:08 xora yes but nobody does it Dec 04 23:24:38 hm has anyone a site where I can look up what differten ifdef for linux darwin and bsd is set? Dec 04 23:24:38 sgh can do it :-) Dec 04 23:25:09 XorA: ok ... is'nt is correct that a defaul_preference of -1 will avoid the version to be build _unless_ specified by a PREFERRED_VERSION_whatever in a conf-file Dec 04 23:25:18 whole mesa directory could use some cleanup :/ Dec 04 23:26:05 there is about 4 generations of mesa (with own includes) and at least 4 variants (-dri, -full, -xlib, normal) Dec 04 23:26:23 sgh: def pref is just a weigting, if they are all equal they all have equal weigth Dec 04 23:27:26 jama *g* not only mesa Dec 04 23:27:44 XorA: yes yes :D ..... mesa-dri-7.0.3 has no default preference ..... soeee that'll be 1 or how ? Dec 04 23:28:17 XorA: or at least "higher" that -1 Dec 04 23:28:21 zero be defauklt I think Dec 04 23:28:26 yes exactly Dec 04 23:28:29 jepp Dec 04 23:29:39 anyway I really should go sleep got to get up in 3 hours Dec 04 23:30:43 xora uh why? Dec 04 23:31:17 jama sdl needs cleanups too Dec 04 23:31:20 woglinde: plane to catch# Dec 04 23:31:32 hm oh Dec 04 23:31:38 where will you fly Dec 04 23:31:53 Salt Lake City Dec 04 23:31:57 woglinde: sdl doesn't scare me as much as mesa, because its not blocking my xorg-7.5 merge :) Dec 04 23:33:54 03Florian Boor  07org.openembedded.dev * rebd05b466b 10openembedded.git/recipes/linux/linux-handhelds-2.6_cvs.bb: Dec 04 23:33:54 linux-handhelds-2.6: Update cvs recipe to use the patches from 2.6.21-hh20 recipe. Dec 04 23:33:54 This makes it compile again. Dec 04 23:36:44 jama clean it up Dec 04 23:44:06 gnite Dec 05 00:01:46 nite all Dec 05 00:04:01 woglinde: gcc is now linking libgcc with static uclibc I think its after your changes Dec 05 00:04:09 to uclibc Dec 05 00:04:13 recipes Dec 05 00:04:42 the solution is to have a /usr/lib/libc.so Dec 05 00:04:46 which is a linker script Dec 05 00:06:32 khem aehm Dec 05 00:06:35 I fixed it Dec 05 00:06:43 why you dont have it Dec 05 00:07:29 I have latest master Dec 05 00:07:40 actually linker looks for libc in usr/lib Dec 05 00:07:47 and now it finds libc.a Dec 05 00:07:50 and uses it Dec 05 00:08:02 so we need to have usr/lib/libc.so Dec 05 00:08:15 which it will prefer over libc.a if found Dec 05 00:08:30 and this will be a stub to linker script Dec 05 00:08:40 which will pull in the correct libc.so to link with Dec 05 00:10:24 f56a544c49284c258c5c46520752b7be0c0f2379 I see removed the symlinks from do_stage Dec 05 00:10:33 right Dec 05 00:10:46 because I fixed it for 0.9.30.1 Dec 05 00:11:02 sorry I dont have the time to fix all recipes Dec 05 00:11:21 if master dont installs the libc.so right Dec 05 00:11:24 its broken Dec 05 00:11:54 dont like the nasty workarounds in oe Dec 05 00:12:18 what do you mean ? Dec 05 00:12:49 there should be a version of libc.so in usr/lib Dec 05 00:13:05 right if you compile 0.9.30.1 Dec 05 00:13:05 + # Need to overwrite the version from -initial Dec 05 00:13:05 + #if [ ! -e ${D}${libdir}/libc.so ]; then Dec 05 00:13:05 + # ln -s ../../lib/libc.so.0 ${D}${libdir}/libc.so Dec 05 00:13:05 + #fi Dec 05 00:13:15 there is a libc.so Dec 05 00:13:38 what does it look like Dec 05 00:13:44 is it a copy or symlink ? Dec 05 00:13:46 or a linker scripy Dec 05 00:13:47 its the linker script Dec 05 00:13:52 ah Dec 05 00:13:59 cool Dec 05 00:14:02 yes Dec 05 00:14:16 hm which distro you are using? Dec 05 00:14:21 installfix.patch Dec 05 00:14:22 I will try it Dec 05 00:14:26 and fix it for you Dec 05 00:14:27 is prolly whats missing Dec 05 00:14:37 I am using uclibc-nptl Dec 05 00:15:05 yeah Dec 05 00:15:07 ah Dec 05 00:15:13 its parallel make error Dec 05 00:15:19 so basically you broke all but 0.9.30.1 :) Dec 05 00:15:28 NOPE!!!!!!!!!! Dec 05 00:15:33 read the commit msg Dec 05 00:15:41 uclibc is broken for parallel make Dec 05 00:15:58 dont play with me Dec 05 00:16:04 cost me some time to find it out Dec 05 00:16:34 now we have libc.so in place Dec 05 00:16:39 what does that mean Dec 05 00:16:50 *sigh* Dec 05 00:16:51 who puts it in place makefiles or recipes ? Dec 05 00:16:59 okay Dec 05 00:17:06 let us hunt it Dec 05 00:17:10 as I did Dec 05 00:17:12 please open Dec 05 00:17:23 Makefile.in Dec 05 00:17:27 I am not applying ./uclibc-0.9.30.1/installfix.patch to nptl recipes Dec 05 00:17:33 probably thats the issue ? Dec 05 00:17:58 yeah Dec 05 00:18:06 but I will show you where the root is Dec 05 00:18:07 this patch is only applied to one recipe Dec 05 00:18:14 thats ok Dec 05 00:18:25 sure I left it to you to test it Dec 05 00:18:28 but currently only 0.9.30.1 would work Dec 05 00:18:31 and others wont Dec 05 00:18:41 thats what I meant earlier Dec 05 00:18:45 its in since 4 weeks Dec 05 00:18:48 or 2 weeks Dec 05 00:18:53 nobody complains Dec 05 00:19:04 open Makefile.in Dec 05 00:19:13 because 0.9.30.1 is default for uclibc Dec 05 00:19:24 go to line install: install_runtime install_dev Dec 05 00:19:31 and prolly no one uses others except me Dec 05 00:19:50 woglinde: I understand the change Dec 05 00:20:02 as you see install_runtime install_dev can be made parallel Dec 05 00:20:07 with parallel make Dec 05 00:20:14 now go to install_runtime Dec 05 00:20:46 @if [ -x lib/$(UCLIBC_LDSO_NAME)-$(VERSION).so ] ; then Dec 05 00:20:54 remind it Dec 05 00:21:01 now go to install_runtime install_dev Dec 05 00:21:13 args Dec 05 00:21:16 go to install_dev Dec 05 00:21:26 the issue it genuine I see that Dec 05 00:21:39 but it should have been applied to other uclibc versions too Dec 05 00:21:48 probably posted upstream as well Dec 05 00:21:52 yeah Dec 05 00:22:00 as I said I left it to you Dec 05 00:22:00 thats my point :) Dec 05 00:22:06 I know its late in Germany Dec 05 00:22:16 becaue I dont have the time to fix all recipe and test them Dec 05 00:22:21 so I fix what I can do Dec 05 00:22:28 thats fine Dec 05 00:22:33 I fix what is default Dec 05 00:22:34 I will munge others Dec 05 01:23:23 re raster Dec 05 01:25:57 woglinde: wogz! how goes? Dec 05 01:26:48 hm autotools Dec 05 01:26:59 I wrote support for a lib Dec 05 01:27:06 from scratch Dec 05 01:27:54 doing autotools froms cratch is a bit of a steep learning curve Dec 05 01:28:52 hm I already knew much Dec 05 01:28:55 and forgot some Dec 05 01:29:00 and learned some new Dec 05 01:29:09 now its in good shape Dec 05 01:29:23 I think Dec 05 01:29:31 but its hard to test for moment **** ENDING LOGGING AT Sat Dec 05 02:59:57 2009