**** BEGIN LOGGING AT Mon Nov 09 02:59:56 2009 Nov 09 03:02:58 yeah i must be on crack.. i'll have to look again when i'm more awake :) Nov 09 03:05:41 as long as the second class is parsed before the third, should be fine. *shrug* Nov 09 07:48:24 good morning Nov 09 08:08:48 gm Nov 09 08:08:55 mckoan, good trip home? Nov 09 08:11:26 ok, off to lhr Nov 09 08:11:32 ci Crofton, fine thx Nov 09 08:50:15 I asked this question on friday but left the office and never saw if I got an answer so I'll try again. I need to modify the /etc/sudoers a bit. I looked in the sudo recipe saw that sudoers is bundled in the tar.gz. So what would be the way to replace that one with my custom sudoers? Nov 09 08:55:08 gm Nov 09 08:56:46 The reason I want to replace sudoers is becuase I need my web application to be able to run ifconfig. Maybe editing the sudoers from my recipe isn't they right way? Nov 09 08:57:18 jovox: look for example at initscripts recipe Nov 09 08:58:03 hi ynezz Nov 09 08:58:05 ok Nov 09 08:58:15 woglinde: hi Nov 09 08:58:25 ynezz hms I was to fast I will revert the geoge patches this evening Nov 09 08:58:34 args geode Nov 09 08:58:46 ok :) Nov 09 08:59:21 jovox: the more traditional way would be to provide a dedicated setuid wrapper as part of your package, rather than relying on sudoers. Nov 09 09:00:38 but, if you do want to stick with sudo, you should be able to just fiddle /etc/sudoers from your postinst. Nov 09 09:01:04 pb_: Hm.. you're talking to a beginner :-) I'll see what I can find about setuid wrapper Nov 09 09:01:29 jovox: can you use some sed and an include into sudoers? Would that not be nicer? Easy to remove when you uninstall the package also :) Nov 09 09:01:48 woglinde: anyway I think, that the glibc fix is correct, without you can't build geode based image Nov 09 09:03:04 woglinde: I'm just writting mail to Otavio about it Nov 09 09:04:07 pb_: you mean overwrite /etc/sudoers in the postinst of my package? Nov 09 09:05:51 Would anyone have an issue with upgrading sudo in dev? I just noticed that I have a newer version locally (1.7 based so you can use includes into /etc/sudoers). Makes the type of thing jovox is doing nicer :). Nov 09 09:14:01 hi woglinde Nov 09 09:14:06 hi marco Nov 09 09:15:46 03Martin Jansa  07shr/merge * r2c30714f7c 10openembedded.git/recipes/linux/ (3 files in 2 dirs): Nov 09 09:15:46 linux-openmoko-2.6.31: drop patches already applied upstream Nov 09 09:15:46 Signed-off-by: Martin Jansa Nov 09 09:20:15 jovox: not so much overwrite, but modify Nov 09 09:21:47 DJWillis: good idea Nov 09 09:22:47 Ok. I'm trying to find something about the "postinst" you mentioned. Actually I tried to find a list of all functions like do_install, do_compile, etc etc but didn't find anything in the bitbake manual. Could you point out a document where I can read about them and when to use them? Nov 09 09:23:12 morning Nov 09 09:28:33 pb_: just trying to workout why my local sudo recipe is broken, been spending time this weekend trying to get PAM working with stuff like SUDO, SHADOW etc. and made a bit of a mess ;-) Nov 09 09:31:56 hi ant Nov 09 09:41:16 woglinde: hey Nov 09 09:47:58 jovox: http://docs.openembedded.org/usermanual/html/recipes_installation_scripts.html Nov 09 09:48:24 hi zecke Nov 09 09:48:46 pb__: master! Nov 09 10:05:30 jo zecke Nov 09 10:05:39 zecke you missed some pints Nov 09 10:05:41 *g* Nov 09 10:09:04 woglinde: in which way? Nov 09 10:09:11 in the pubs Nov 09 10:17:09 -> bbl Nov 09 10:17:28 hi lrg Nov 09 10:17:37 how is the swine flu? Nov 09 10:22:05 pb_: morning Nov 09 10:22:29 dont think it the pig flu anymore, it was mild compared to others Nov 09 10:22:54 feeling better now though Nov 09 10:23:05 hi lrg Nov 09 10:23:09 lrg: good-o Nov 09 10:23:10 jo jay7 Nov 09 10:23:17 woglinde: wo won the Beer contest? UK or DE guests? Nov 09 10:23:30 hey woglinde Nov 09 10:24:00 morning :) Nov 09 10:24:20 is there photo-reports from OEDEM? ;) Nov 09 10:36:38 sadly we forgot to take an official OEDEM photo. I think mckoan might have some photos though. Nov 09 10:36:53 no photos? Nov 09 10:37:02 no report? no nothing for me? *sad panda* Nov 09 10:37:15 heh. there will be a report. Nov 09 10:37:46 mickeyl has the official minutes of the ga. I think he is going to deal with them in some appropriate way. Nov 09 10:38:17 I think rp has the notes from the rest of the oedem but they will need a bit of editing in order to have them make sense for non attendees. Nov 09 10:38:43 we did plan to take an official photo of all attendees but kgilmer slipped away before we could do so :-} Nov 09 10:40:02 zecke: I just sent a summary of some of the topics as a start Nov 09 10:40:28 pb__: Reading through the notes, these are going to need some editting :/ Nov 09 10:41:04 kergoth: sure? Nov 09 10:41:17 RP: ah great Nov 09 10:41:31 RP: I wonder who has board duty now? Nov 09 10:41:45 kergoth: indeed; fixing it Nov 09 10:42:07 zecke: Mickey, Florian, Crofton and XorA Nov 09 10:42:53 RP: right, I suspected as much Nov 09 10:43:01 (the editing, not the board) Nov 09 10:43:40 pb__: I guess I'll email them to the attendees and suggest people try and summarise each session? Nov 09 10:43:51 yes, good plan Nov 09 10:43:52 kergoth: now it does: git clone git://projetos.ossystems.com.br/git/users/otavio/bitbake.git Nov 09 10:44:12 pb__: Can you check I didn't miss or misrepresent anything in the summary I just sent out? :) Nov 09 10:45:31 RP: that summary looks good to me Nov 09 10:52:32 robert has photos too Nov 09 10:52:37 but he is on tour Nov 09 10:53:35 ah yes, right, his pilgrimage Nov 09 10:53:40 *g* Nov 09 10:53:49 ~lart openstreetmap Nov 09 10:53:50 * ibot frags openstreetmap with his BFG9000 Nov 09 10:56:46 args Nov 09 10:56:48 my fault Nov 09 10:56:53 have to parse the maps Nov 09 10:59:39 pb__: good. I've sent the notes around now too Nov 09 11:00:01 good-o Nov 09 11:00:28 I'll have a look at those once I get fed up with debugging this stupid curl problem. :-} Nov 09 11:02:26 I want to keep my own scripts and programs in a /apps-folder. Do I just create the dir and install to it in do_install? I tried doing a "install -d ${D}/apps", but did't get a new dir in the finished rootfs. Nov 09 11:03:03 tasslehoff: you need to add it to FILES as well. Nov 09 11:04:07 pb__: a "FILES += "in my recipe? or somewhere else Nov 09 11:04:29 OT: has anyone run into GSlice performance problems? Nov 09 11:04:58 tasslehoff: in your recipe Nov 09 11:05:24 zecke: it works pretty well for me so long as you don't have multi threading Nov 09 11:05:40 performance does suck somewhat in the presence of multiple threads Nov 09 11:08:04 pb__: i have threads and pthread_getspecific is #1 in the profile Nov 09 11:08:18 zecke: ah Nov 09 11:08:30 yeah, if you have threads then you will lose Nov 09 11:08:34 pb__: is this more of a problem in glibc, glib, ARM? Nov 09 11:08:40 though, if you are using tls, pthread_getspecific() is pretty lightweight Nov 09 11:09:14 pb__: any recent EABI should have tls enabled... so it is the vast majority of calls then :( Nov 09 11:09:44 zecke: yeah. you could improve matters by patching glib to use __thread, rather than pthread_getspecific/setspecific Nov 09 11:10:09 pb__: my understanding is... I have pretty large Gtk widget hierachy and propagating a paint requires creating and destroying GdkRegion's (which uses gslice) Nov 09 11:10:11 that has been sort of on my to-do list for a while but it has never risen to the top Nov 09 11:10:24 (partly because I have an ambition to eliminate the use of threads from my application anyway) Nov 09 11:11:05 zecke: yes, I think your understanding is correct. gtk and gdk are quite heavy users of GSlice. Nov 09 11:11:09 pb__: hehe, so for __thread GCC will directly generate a tls call/lookup (I don't remember exactly) Nov 09 11:11:34 it is a shame that gslice doesn't allow you to have completely separate arenas for different usages. this would allow you to remove the thread overhead altogether in many cases. Nov 09 11:11:37 zecke: right Nov 09 11:11:53 pb__: it can't do the same for getspecific? Nov 09 11:12:05 zecke: yeah, it can, but it is wrapped in an extra function call then Nov 09 11:12:41 also, depending on your cpu and exact abi details, the tls operation may be fairly heavyweight in itself. Nov 09 11:13:41 have a look at what __aeabi_read_tp() is doing on your system Nov 09 11:17:17 pb__: seems to jump into a page exposed by the kernel Nov 09 11:21:07 right, that should be ok Nov 09 11:21:16 assuming you don't have a hard tls register, this is about as good as it gets Nov 09 11:24:00 pb__: I don't see any other implementation of this function in glibc?! Nov 09 11:24:33 ah, hm. I thought there was an mrc-based one in glibc but maybe that was never actually implemented. Nov 09 11:25:40 pb__: not in 2.9 Nov 09 11:25:54 okay Nov 09 11:26:05 I'm pretty sure it hasn't been done since 2.9 so I guess it just never happened. Nov 09 11:26:31 hard-tp equipped hardware is still moderately rare so it is probably not all that big a deal Nov 09 11:27:18 arm1176 has it, and I guess at least some of the cortex-a series do. Nov 09 11:27:31 it wasn't in arm1136 or anything earlier, iirc Nov 09 11:27:51 pb__: this target here is good ole xscale.. Nov 09 11:28:01 right, definitely no hard tp on xscale Nov 09 11:43:11 otavio: Hi, regarding "[oe] [PATCH] libgsf: fix compile error, add missing dependency gtk+" - my original patch with DEPENDS="gtk+" was commited already, with dependency on gdk-pixbuf there were some troubles as I wrote in email, like Abiword breakage, honestly I don't know what kind of patch should I cook :) Nov 09 11:43:47 ynezz: to fix gdk-pixbuf? Nov 09 11:44:07 ynezz: so we can avoid gtk+ depends Nov 09 11:44:14 just by removing gtk+'s pkgconfig files from gdk? Nov 09 11:44:49 honestly, I don't know much about GTK :) Nov 09 11:44:55 ynezz: or updating them? or updating the recipe that does it to work with newer versions Nov 09 11:45:23 ah, update gdk to gtk version, right? Nov 09 11:45:32 ynezz: first of all; nice to meet you :-) Nov 09 11:45:48 ynezz: might be; the idea is to reduce compilation time Nov 09 11:45:58 nice to meet you too Nov 09 11:46:22 yes I got it, it's better to compile gdk then whole gtk+ Nov 09 11:51:59 pb__: did you look at how you would replace GPrivate with using __thread? Nov 09 11:52:53 03Petr Štetiar  07org.openembedded.dev * r29bc82428b 10openembedded.git/recipes/linux/ (linux-2.6.31/iei-nanogx-466/defconfig linux_2.6.31.bb): Nov 09 11:52:53 linux: Linux 2.6.31.5 for IEI Nano GX-466 Nov 09 11:52:53 Signed-off-by: Petr Štetiar Nov 09 11:52:53 Acked-by: Otavio Salvador Nov 09 11:53:01 re Nov 09 11:53:04 03Petr Štetiar  07org.openembedded.dev * rbec85b495e 10openembedded.git/ (conf/checksums.ini recipes/linux/linux_2.6.31.bb): Nov 09 11:53:04 linux: Linux 2.6.31.5 Nov 09 11:53:04 Signed-off-by: Petr Štetiar Nov 09 11:53:04 Acked-by: Otavio Salvador Nov 09 11:53:06 03Petr Štetiar  07org.openembedded.dev * rdbee1acac7 10openembedded.git/conf/machine/iei-nanogx-466.conf: Nov 09 11:53:09 Machine configuration for IEI Nano GX-466 Nov 09 11:53:11 Signed-off-by: Petr Štetiar Nov 09 11:53:13 Acked-by: Otavio Salvador Nov 09 11:53:45 hi otavio btw. Nov 09 11:53:54 woglinde: hi :-) Nov 09 11:54:11 otavio I will revert the geode patches this evening Nov 09 11:54:24 ynezz: about the i586 x i486. Can you test it for performance? Nov 09 11:54:31 woglinde: which ones? Nov 09 11:56:08 is Phil Blundell around? Nov 09 11:56:16 otavio: I would like to do some test, but I don't know when :) Nov 09 11:56:50 ynezz: it is better todo tests before commiting stuff ;-) Nov 09 11:57:14 it wasn't slow for me, so I didn't care Nov 09 11:57:43 which benchmark should I use? Nov 09 11:58:21 ynezz: an AMD engenier, when I was doing it, suggested to use i486 in absense of geode. Might be nice to test to confirm it. Nov 09 11:58:31 ok Nov 09 11:58:58 ynezz: dunno; something massive and that do not depends on X ;-) Nov 09 11:59:37 pb__: I should be able to add FILES_${PN} += "/apps" in more than one recipe? Nov 09 12:00:11 ynezz: "git grep bench" shows nice things Nov 09 12:00:15 ok Nov 09 12:00:28 tasslehoff: yes Nov 09 12:01:10 zecke: not in any detail, it just seemed like this would be a better way for gslice to do its thread pools Nov 09 12:02:04 pb__: do you know how pthread_mutex_lock is implemented on ARM? could you assume it is moderately okay on armv5t? Nov 09 12:02:08 pb__: it worked on the first one, but when I did the same on another to move it from bindir to appsdir it said: ERROR: QA Issue with loader: non debug package contains .debug directory: Nov 09 12:05:44 ah, FILES_${PN}-dbg += "${appsdir}/.debug" solved it Nov 09 12:14:13 03Richard Purdie  07rpurdie/work-in-progress * r7213df0bcd 10openembedded.git/classes/base.bbclass: Nov 09 12:14:13 base.bbclass: Fix staging paths for native recipes (from Poky) Nov 09 12:14:13 Signed-off-by: Richard Purdie Nov 09 12:14:15 03Richard Purdie  07rpurdie/work-in-progress * r25600181da 10openembedded.git/classes/base.bbclass: Nov 09 12:14:15 base.bbclass: Drop import bb and import os as no longer needed Nov 09 12:14:18 Signed-off-by: Richard Purdie Nov 09 12:14:20 03Richard Purdie  07rpurdie/work-in-progress * r82c6bf2656 10openembedded.git/conf/sanity.conf: Nov 09 12:14:23 sanity.conf: Bump minimum bitbake version requirement Nov 09 12:14:25 Signed-off-by: Richard Purdie Nov 09 12:14:27 03Richard Purdie  07rpurdie/work-in-progress * rfc32427d98 10openembedded.git/classes/base.bbclass: Nov 09 12:14:30 base.bbclass: Drop legacy rebuild mangling code Nov 09 12:14:32 Signed-off-by: Richard Purdie Nov 09 12:14:51 hrw|gone: The above commit should fix that native problem you were having Nov 09 12:15:00 hrw|gone: and I pushed a fix to the icu recipe in poky too Nov 09 12:41:04 morning Nov 09 12:41:07 RP: thx Nov 09 12:41:15 hrw: welcome back Nov 09 12:41:16 hi hrw - trip back ok? Nov 09 12:41:23 RP: yes Nov 09 12:41:30 RP: gm Nov 09 12:41:37 hi denix Nov 09 12:41:39 from Berlin to home I travelled alone in microbus Nov 09 12:43:32 03Richard Purdie  07org.openembedded.dev * rcb23f95780 10openembedded.git/conf/sanity.conf: Nov 09 12:43:32 sanity.conf: Bump minimum bitbake version requirement Nov 09 12:43:32 Signed-off-by: Richard Purdie Nov 09 12:43:33 03Richard Purdie  07org.openembedded.dev * r29c7d3351f 10openembedded.git/classes/package.bbclass: Nov 09 12:43:35 package.bbclass: Change to operate on a copy of the install directory instead of directly. Also take the opportunity to rename the temp directories so their function is clear (from Poky) Nov 09 12:43:36 RP: bitbake-dev and master req python2.6? Nov 09 12:43:38 Signed-off-by: Richard Purdie Nov 09 12:43:40 03Richard Purdie  07org.openembedded.dev * r353ba77e54 10openembedded.git/classes/package.bbclass: Nov 09 12:43:43 package.bbclass: Move variable checks to one place making the code more readable (from Poky) Nov 09 12:43:45 Signed-off-by: Richard Purdie Nov 09 12:43:47 03Richard Purdie  07org.openembedded.dev * r0d722f5b12 10openembedded.git/classes/base.bbclass: Nov 09 12:43:49 hrw: yes, at present Nov 09 12:43:50 base.bbclass: Drop legacy rebuild mangling code Nov 09 12:43:52 Signed-off-by: Richard Purdie Nov 09 12:43:54 03Richard Purdie  07org.openembedded.dev * r9a5937f6c6 10openembedded.git/classes/base.bbclass: Nov 09 12:44:01 base.bbclass: Drop import bb and import os as no longer needed Nov 09 12:44:03 Signed-off-by: Richard Purdie Nov 09 12:44:05 03Richard Purdie  07org.openembedded.dev * rc41e3631d3 10openembedded.git/classes/package.bbclass: Nov 09 12:44:08 package.bbclass: Remove hardcoded use of PKGDEST variable (from Poky) Nov 09 12:44:10 Signed-off-by: Richard Purdie Nov 09 12:44:12 03Richard Purdie  07org.openembedded.dev * r6111d7132a 10openembedded.git/classes/package.bbclass: Nov 09 12:44:15 package.bbclass: Add the option to declare package preprocess function hooks (from Poky) Nov 09 12:44:17 Signed-off-by: Richard Purdie Nov 09 12:45:10 ok Nov 09 12:49:22 /usr/lib/gcc/x86_64-linux-gnu/4.3.4/../../../../lib/libSM.so: undefined reference to `uuid_unparse_lower@UUID_1.0' Nov 09 12:49:26 /usr/lib/gcc/x86_64-linux-gnu/4.3.4/../../../../lib/libSM.so: undefined reference to `uuid_generate@UUID_1.0' Nov 09 12:49:34 RP: thats dbus-native in recent poky trunk Nov 09 12:49:41 gm all Nov 09 12:49:51 hrw: After that fix? Nov 09 12:50:01 hrw: Was that with a clean tmp? Nov 09 12:50:07 not clean Nov 09 12:50:13 have to rip tmp out? Nov 09 12:50:23 hrw: yes, for that fix :( Nov 09 12:50:38 hrw: Its a one line bug but a serious one Nov 09 12:50:49 hrw: er, sorry, not for poky Nov 09 12:50:55 hrw: for OE it is Nov 09 12:51:32 hrw: Its just dbus-native and icu-native that failed? Nov 09 12:52:18 gnome-keyring do_configure too Nov 09 12:52:42 configure.in:68: error: possibly undefined macro: AM_GCONF_SOURCE_2 Nov 09 12:53:02 anyway it is time to do some work for me finally Nov 09 12:53:14 * hrw -> October poky turnk then Nov 09 12:53:25 hrw: You're worrying me as I'm not seeing these Nov 09 12:53:44 hrw: Have you tested Poky trunk recently without the new changes? Nov 09 12:54:08 RP: I did build on oedem and now just did ;git pull; and tried to continue build Nov 09 12:54:24 RP: but it was your branch... not trunk Nov 09 12:55:28 hrw: ok, the icu failure should be fixed, the other issues worry me Nov 09 12:57:38 RP: so Poky/master or Poky/master-work-in-progress? Nov 09 12:59:02 hrw: I'd like to get wip working if we can please but it'd be intersting to know if master works too or not Nov 09 13:02:20 03Richard Purdie  07rpurdie/work-in-progress * r96b090d1b4 10openembedded.git/classes/autotools.bbclass: Nov 09 13:02:20 autotools.bbclass: Remove pointless inherit (from Poky) Nov 09 13:02:20 Signed-off-by: Richard Purdie Nov 09 13:02:21 03Richard Purdie  07rpurdie/work-in-progress * r6f481bcc54 10openembedded.git/classes/autotools.bbclass: Nov 09 13:02:21 autotools.bbclass: Allow for leading whitespace in gettext and intltool searches (from Poky) Nov 09 13:02:25 Signed-off-by: Richard Purdie Nov 09 13:02:27 03Richard Purdie  07rpurdie/work-in-progress * rf6c3cb05c0 10openembedded.git/classes/autotools.bbclass: Nov 09 13:02:30 autotools.bbclass: Create m4 directory earlier avoding issues with intltoolize (from Poky) Nov 09 13:02:32 Signed-off-by: Richard Purdie Nov 09 13:02:34 03Richard Purdie  07rpurdie/work-in-progress * rc942cea699 10openembedded.git/classes/ (base.bbclass packagedata.bbclass): Nov 09 13:02:37 packagedata.bbclass: Split package data handling functions out from base.bclass (from Poky) Nov 09 13:02:39 Signed-off-by: Richard Purdie Nov 09 13:02:41 03Richard Purdie  07rpurdie/work-in-progress * ra4959a6371 10openembedded.git/classes/binconfig.bbclass: Nov 09 13:02:44 binconfig.bbclass: Convert do_install_append() into a PACKAGE_PREPROCESS_FUNCS (from Poky) Nov 09 13:02:54 Signed-off-by: Richard Purdie Nov 09 13:02:56 03Richard Purdie  07rpurdie/work-in-progress * rdb67608163 10openembedded.git/classes/autotools.bbclass: Nov 09 13:02:59 autotools.bbclass: Convert .la file mangling to package preprocessing hook Nov 09 13:03:01 Signed-off-by: Richard Purdie Nov 09 13:03:03 03Richard Purdie  07rpurdie/work-in-progress * rf985c5f5a0 10openembedded.git/classes/ (autotools.bbclass base.bbclass): Nov 09 13:03:06 autotools.bbclass: Separate out useful staging functions into base.bbclass and call from autotools classes (from Poky) Nov 09 13:03:09 Signed-off-by: Richard Purdie Nov 09 13:03:11 03Richard Purdie  07rpurdie/work-in-progress * r070cb3ccd3 10openembedded.git/classes/ (base.bbclass binconfig.bbclass pkgconfig.bbclass): Nov 09 13:03:16 binconfig/pkgconfig.bbclass: Convert staging functions into SYSROOT_PREPROCESS_FUNCS operating on SYSROOT_DESTDIR Nov 09 13:03:20 Signed-off-by: Richard Purdie Nov 09 13:03:22 03Richard Purdie  07rpurdie/work-in-progress * rfb84242b1d 10openembedded.git/classes/canadian-cross.bbclass: Nov 09 13:03:25 canadian-cross.bbclass: Fix circular references after layout_ changes Nov 09 13:03:27 Signed-off-by: Richard Purdie Nov 09 13:03:31 03Richard Purdie  07rpurdie/work-in-progress * r4b7aa60600 10openembedded.git/classes/native.bbclass: Nov 09 13:03:34 native.bbclass: If do_stage isn't overridden, allow do_install to run for native packages (from Poky) Nov 09 13:03:36 Signed-off-by: Richard Purdie Nov 09 13:03:38 03Richard Purdie  07rpurdie/work-in-progress * r828e7db904 10openembedded.git/classes/ (base.bbclass packaged-staging.bbclass): Nov 09 13:03:43 packaged-staging.bbclass: Use a variable for the location of the staging lock file (from Poky) Nov 09 13:03:47 Signed-off-by: Richard Purdie Nov 09 13:03:49 03Richard Purdie  07rpurdie/work-in-progress * r9b00fa2c61 10openembedded.git/classes/base.bbclass: Nov 09 13:03:52 base.bbclass: Note legacy staging packages in debug output Nov 09 13:03:54 (30 lines omitted) Nov 09 13:03:58 03Richard Purdie  07org.openembedded.dev * rf6c3cb05c0 10openembedded.git/classes/autotools.bbclass: Nov 09 13:04:03 autotools.bbclass: Create m4 directory earlier avoding issues with intltoolize (from Poky) Nov 09 13:04:05 Signed-off-by: Richard Purdie Nov 09 13:04:07 03Richard Purdie  07org.openembedded.dev * rc942cea699 10openembedded.git/classes/ (base.bbclass packagedata.bbclass): Nov 09 13:04:10 packagedata.bbclass: Split package data handling functions out from base.bclass (from Poky) Nov 09 13:04:12 Signed-off-by: Richard Purdie Nov 09 13:04:16 03Richard Purdie  07org.openembedded.dev * r96b090d1b4 10openembedded.git/classes/autotools.bbclass: Nov 09 13:04:19 autotools.bbclass: Remove pointless inherit (from Poky) Nov 09 13:04:21 Signed-off-by: Richard Purdie Nov 09 13:12:22 hrw: The build explodes here too - looking into what I broke :) Nov 09 13:12:32 hrw: Worryingly it does break differently Nov 09 13:18:27 pb__: I've found an interesting problem with DEPENDS_virtclass-native :/ Nov 09 13:18:52 pb__: Specifically when we use that construct it makes the DEPENDS_prepend in base.bbclass useless Nov 09 13:20:24 re Nov 09 13:20:30 hi rp and hrw Nov 09 13:20:30 hi woglinde Nov 09 13:20:34 RP: hm, why so? I thought _prepend and _append were processed last, after all the other overrides had run. Nov 09 13:20:46 so, in that case, I would have expected it to still apply to the overridden value Nov 09 13:23:35 pb__: Hmm. That isn't what is happening at least Nov 09 13:26:42 RP: thanks for the merge Nov 09 13:27:32 denix: I've not merged fully yet - trying to clean up the series a bit and make it a logical transition first :) Nov 09 13:27:59 rp how much time you have to work on it? Nov 09 13:28:11 re Nov 09 13:28:17 pb__: FOO = "a", FOO_append = "b" and FOO_qemuarm = "c" results in c, assuming my MACHINE is qemuarm Nov 09 13:29:01 woglinde: I will find the time to do what it needed Nov 09 13:29:18 okay Nov 09 13:29:19 RP: ah, drat. Nov 09 13:30:08 pb__: On the postive side it means its not some obsure issue with BBCLASSEXTEND's handling of the override :/ Nov 09 13:30:16 could replace that DEPENDS_prepend with an anonymous python func that does the prepending, which I guess would work but is not exactly pretty. Nov 09 13:30:24 RP: heh, I guess that is true Nov 09 13:30:40 pb__: Jay7 I have any photo, I asked to other to send me them, IIRC Crofton|work and Robert took many Nov 09 13:30:54 mckoan: ah, right Nov 09 13:31:04 hi gnutoo Nov 09 13:31:41 pb__: I think it has to be promoted to an anonymous function :/ Nov 09 13:31:59 * RP curses at the irc client for segfaulting Nov 09 13:32:11 rp urgs Nov 09 13:32:20 I get only psi segfaulted Nov 09 13:35:18 RP: right, I was just saying that when your irc client went away :-) Nov 09 13:35:56 pb_: anonymous python functions run before override expansion Nov 09 13:36:18 pb__: of this particular override anyway Nov 09 13:36:56 Since virtclass native does insane things with DEPENDS anyway I think I'll just adjust that code Nov 09 13:37:13 oh, right, I guess they do. that is a bit annoying, unless you are prepared to have an anonymous function that spots DEPEND_virtclass-native and prepends a spare copy to that as well. Nov 09 13:37:23 yeah, that might be the easiest solution Nov 09 13:37:26 pb__: I tried to do a echo "#Testing" >> /etc/sudoers from sudo.inc in my overlay recipe but nothing was written to sudoers Nov 09 13:37:50 ~lart psi Nov 09 13:37:50 * ibot plops psi into a giant vat of herring Nov 09 13:37:55 jovox: you'd need to show me the actual file, it is hard to say anything useful based on that snippet Nov 09 13:38:01 03Liam Girdwood  07org.openembedded.dev * r2f9a991fd5 10openembedded.git/ (8 files in 3 dirs): Nov 09 13:38:01 pulseaudio: added pulseaudio-0.9.19 Nov 09 13:38:01 Signed-off-by: Liam Girdwood Nov 09 13:38:02 03Liam Girdwood  07org.openembedded.dev * r53c5c0c34d 10openembedded.git/: Merge remote branch 'origin/org.openembedded.dev' into upstream Nov 09 13:38:02 lunchtime now, bbiab Nov 09 13:38:09 bye pb Nov 09 13:38:22 uh commits from liam Nov 09 13:38:24 03Martin Jansa  07shr/merge * rc87a750a79 10openembedded.git/conf/distro/include/shr-autorev.inc: Nov 09 13:38:24 shr: pin libfsobasics revision Nov 09 13:38:24 Signed-off-by: Sebastian Spaeth Nov 09 13:38:28 they are rar Nov 09 13:38:33 rare Nov 09 13:38:54 woglinde: heh, very true. just got my new key from mickey|away Nov 09 13:39:04 I managed to delete my old key :( Nov 09 13:39:08 pb__: http://pastebin.com/m7537a5f7 Nov 09 13:39:18 lrg hm yeah knows that Nov 09 13:39:23 or forgetting the passwd Nov 09 13:39:45 woglinde: heh, I was hoping people would have forgotten ;) Nov 09 13:40:30 woglinde: now I have glibc-2.10.1 patch to email to the list Nov 09 13:40:54 I meant I know the feeling loosing the key or forgetting the key Nov 09 13:41:59 its not a good feeling :-/ Nov 09 13:42:27 I was moving to a new PC at the time and copied stuff over before nuking the old HD Nov 09 13:42:44 ah Nov 09 13:43:11 I once forgot over night how my LDAP passwd was typed Nov 09 13:43:17 at work Nov 09 13:43:24 thats not nice either Nov 09 13:44:01 woglinde: XorA|gone has never let me forget the fact I deleted my key ;) Nov 09 13:44:01 it's easy to copy key to phone Nov 09 13:44:19 that way I have it all the time with me Nov 09 13:44:29 ynezz: heh, but I had gta01 at the time ;) Nov 09 13:44:32 :D Nov 09 13:44:47 lrg: well you did delete it about 30secs after I said have you got you keys backed up :-) Nov 09 14:01:02 gm florian_kc Nov 09 14:03:16 RP: got some parsing fall-out after the 1.8.16 against .dev head: http://www.pastebin.ca/1663401 Nov 09 14:03:17 pb__: looking for an UK web radio to listen at office I came across this :-D http://www.radiofeeds.co.uk/reciva.asp Nov 09 14:04:11 likewise: ah, let me fix that ;-) Nov 09 14:04:52 RP: glad to have you on the channel during such commits :-) Nov 09 14:07:41 03Richard Purdie  07org.openembedded.dev * r0fc2395bd6 10openembedded.git/classes/base.bbclass: Nov 09 14:07:41 base.bbclass: Promote packagedata inherit to a direct inherit due to ordering issues Nov 09 14:07:41 Signed-off-by: Richard Purdie Nov 09 14:08:15 likewise: I'm trying to be around for reasons like that :). The fix was in the wip branch in this case, its just a patch ordering issue - let me know if that resolves that problem Nov 09 14:09:05 pb__: The order we resolve _prepend and _append in causes some serious headaches :/ Nov 09 14:09:34 Somewhat off topic but given there are at lease a few eclipse users on here. Does anyone know how to limit the number of break points? Nov 09 14:09:56 pb__: It appears to be immediate meaning I can't play the trick I hoped of chaging "DEPENDS" as the core variable early enough Nov 09 14:10:19 RP: tricks are no longer allowed in programming anyway :-) Nov 09 14:10:41 likewise: Don't look at the core OE code then ;-) Nov 09 14:13:12 RP: My lack of python-language-completeness makes me oversee all tricks anyway Nov 09 14:13:25 When I need is a hook pre data expansion in bitbake Nov 09 14:14:19 RP: that fixes this one. I'll let the build proceed on this one. Nov 09 14:15:04 likewise: good, thanks for testing - its an artifact of the fact I'm kind of drip feeding things and didn't catch that dependency in the patch series Nov 09 14:16:05 RP: no problem, thanks for dripping goodies on us. Nov 09 14:16:29 pb__: I feel tempted to add a anonymous_preexpansion() hook but that could appear a little evil ;-) Nov 09 14:16:30 (ok that sounded gross) Nov 09 14:16:49 likewise: it did :) Nov 09 14:16:52 What's this mean: *** %n in writable segment detected *** Nov 09 14:17:22 likewise: Its a compiler warning I think? Nov 09 14:18:25 it's actually a glibc warning Nov 09 14:18:33 i.e. you get it at run time, not compile time Nov 09 14:18:47 pb__: That makes more sense ;-) Nov 09 14:19:23 morning Nov 09 14:19:49 %n is a particularly toxic format parameter because it causes printf() to write back to memory rather than reading from it. Nov 09 14:21:06 if you have the hardening options enabled, libc will not permit format strings with %n in them except where the string resides in read-only memory, to reduce the risk of security problems due to unchecked user input Nov 09 14:22:04 I think fedora, and maybe a couple of other distros, turn that on by default nowadays. Nov 09 14:22:32 good morning Nov 09 14:23:14 pb__: I can't come up with a none evil way of solving this :( Nov 09 14:23:23 or a way without adding new options to bitbake Nov 09 14:23:29 RP: yah, I can't immediately think of one either. Nov 09 14:23:58 I am a bit surprised (though not completely shocked) that bitbake is behaving the way that it apparently is. Nov 09 14:24:29 I guess that requiring a new bitbake version at this point would not be such a big deal so now would be the time to add that feature if it is needed :-) Nov 09 14:24:42 pb_ did you see my paste? Nov 09 14:25:22 jovox: oh, yeah, sorry. that does look ok to me. so, you're saying that you do "ipkg install sudo", or whatever, but your appended line doesn't show up in /etc/sudoers? Nov 09 14:25:34 and presumably you don't get an error either Nov 09 14:26:41 bbiab, gotta read some powerpoint presentations for my boss now Nov 09 14:26:46 sudo is "included" in my image. And as you said, the /etc/sudoes has not changed Nov 09 14:28:47 pb__: The best solution might be an -= or "_delete" extension to the metadata as then I'd not have to play with depends anyway... Nov 09 14:29:03 We've talked about one for long enough, the ordering issues just scare me Nov 09 14:33:00 RP: if you wanted to go down the deletion route then presumably an anonymous function would work for that without any bitbake changes. Nov 09 14:33:17 since in that case, what you want to operate on is the overridden data anyway Nov 09 14:33:30 pb__: I mean adding that to the syntax in general Nov 09 14:33:54 pb__: Adding a special case for this use is probably the best short term fix thoug Nov 09 14:38:08 yeah Nov 09 14:38:17 adding _delete in general might indeed be a good thing Nov 09 14:38:25 morning kergoth Nov 09 14:38:53 hi kergoth Nov 09 14:47:24 pb__: http://git.pokylinux.org/cgit.cgi/poky/commit/?h=master-work-in-progress&id=e3264eed5de2496f45402b2f3290915f1fd030bf - not ideal but probably the least nasty thing I can find Nov 09 14:47:54 pb__: Its not like we're going to have a large number of these things and this buys us some time... Nov 09 14:50:52 yeah, fair enough Nov 09 14:50:56 I think that is fine for now Nov 09 14:53:48 florian: good morning Nov 09 14:55:19 hey all Nov 09 14:55:23 * florian made it back home Nov 09 14:55:27 hi kergoth Nov 09 14:55:35 * kergoth yawns Nov 09 14:55:55 glad to hear oedem went well Nov 09 14:56:54 kergoth: I hope that next one will be with you being present Nov 09 14:57:32 i'll have to set aside some cash so i can do it out of pocket if MV won't send me Nov 09 15:01:37 and then reimburst it from e.V. Nov 09 15:06:37 * kergoth mulls over possibilities for persistence/storage of build output (i.e. pkgdata) for subsequent builds Nov 09 15:07:38 kergoth: you mentioned something about locking used versions/revisions for OE. was it merged? and how to use it? Nov 09 15:08:05 i don't think my implementation ever went in, someone elses did, but i don't recall how you enable it offhand Nov 09 15:08:29 ok Nov 09 15:08:59 I just have builds now with ~20% of recipes hitting customer svn Nov 09 15:09:04 very slow svn... Nov 09 15:09:22 13 minutes and not yet parsed 831 recipes ;( Nov 09 15:11:14 hrw There is a BB_SRCREV_POLICY = cache or clear variable Nov 09 15:12:28 k Nov 09 15:13:48 thx Nov 09 15:15:55 RP: so 'cache' value means 'do not check revs until you have them in cache dir'? Nov 09 15:16:36 hrw: It means do not check if they are in the cache, yes Nov 09 15:17:49 love it Nov 09 15:20:49 * kergoth wonders what it would take to add better traceability to the system Nov 09 15:22:01 hi kergoth Nov 09 15:23:12 i wonder if you could dump make's internal database to analyze what it will do when you run a given make target.. Nov 09 15:25:58 though, i guess you'd have to parse shell commands to make use of that.. not so trivial Nov 09 15:27:01 RP: how does the e2factory staging stuff work? Nov 09 15:29:20 kergoth: Firstly that always use a chroot and require root privs. Each time they build anything they populate a chroot from staging package and build within that entirely Nov 09 15:29:41 ah, interesting Nov 09 15:29:53 kergoth: Each build is associated with a checksum which represents the contents of the chroot Nov 09 15:29:57 that would certainly make things deterministic, if hefty.. Nov 09 15:30:11 Right, that was what their objective was Nov 09 15:30:43 hrw: I've pushed fixes to poky's wip branch. It'd be interesting to know if things work better now Nov 09 15:30:49 rp hm I saw some of your autotools.bbclass changes Nov 09 15:31:07 rp what are you doing if configure.ac poiting to different m4-dir Nov 09 15:31:18 than m4 Nov 09 15:31:23 hmm. Nov 09 15:31:28 woglinde: Not a different one, just moing the mkdir command around Nov 09 15:31:44 woglinde: I found a weird corner case in poky Nov 09 15:32:04 hm Nov 09 15:32:08 s/moing/moving/ Nov 09 15:32:16 woglinde: What concerns you about that? Nov 09 15:32:30 RP: that sounds a lot like what i was playing with with private staging areas, only even more so. i too was thinking about using a checksum to cache those staging areas for other recipes to use Nov 09 15:32:33 hmm Nov 09 15:32:37 I think there will be some problems when the m4 macros are in a diffrent dir Nov 09 15:32:54 hm Nov 09 15:32:58 or nevermind Nov 09 15:33:01 should work Nov 09 15:33:05 RP: ok Nov 09 15:33:06 kergoth: yes, I think we could do something like this Nov 09 15:33:24 woglinde: We already have the mkdir there, I just moved it before another call Nov 09 15:33:44 i like the chroot idea on top of it, but could even make that piece optional, or leverage qemu to avoid the need for root.. Nov 09 15:33:46 * kergoth ponders Nov 09 15:34:30 kergoth: All compiles under qemu? Nov 09 15:34:42 rp hm I think the correct solution is search configure.ac/in if m4-macro-dir points to something then create m4 or not Nov 09 15:34:50 heh, point taken, that would be pretty slow, though not as slow as it used to be :) Nov 09 15:34:58 ah well, something to think about Nov 09 15:34:58 kergoth: :) Nov 09 15:35:08 rp otherwise its little bit confusing if there is an empty m4 dir Nov 09 15:35:11 kergoth: yes, I liked their use of checksums Nov 09 15:35:19 RP: within same build or from scratch one? Nov 09 15:35:19 woglinde: Its always been there though ;-) Nov 09 15:35:34 woglinde: Whether it should be is a different question than my patch which just moved the mkdir ;-) Nov 09 15:35:37 hm yeah now I know why Nov 09 15:35:45 woglinde: me too Nov 09 15:36:12 woglinde: too many other problems ahead of that one in my queue to worry :/ Nov 09 15:37:05 hmm iptables doesnt run. Could not load /lib/modules/2.6.30/...... the reason is becuase I'm running 2.6.25.20 Nov 09 15:37:15 rp I can fix it Nov 09 15:37:22 should be easy for me Nov 09 15:37:41 bye till later meeting Nov 09 15:42:22 florian: sorry, but I should continue to bother you about wiki.. Nov 09 15:43:51 Jay7: yes you are right, i'll do my best but I have a pile of extremely urgent jobs at the office right now :-( Nov 09 15:43:58 but i didn't forget :-) Nov 09 15:44:07 florian: ok :) Nov 09 15:45:14 RP: | /home/hrw/devel/OH/poky/build/tmp/work/x86_64-linux/icu-native-3.6-r4/temp/run.staging_helper.12529: line 1050: unexpected EOF while looking for matching `'' Nov 09 15:45:20 RP: do from scratch? Nov 09 15:45:27 RP: should I do from scratch? Nov 09 15:50:13 RP: same from scratch Nov 09 15:51:13 RP: its sysroot_stage_libdir() function Nov 09 15:51:41 but I do not see error there Nov 09 15:54:10 found I think Nov 09 15:56:48 nope Nov 09 16:01:57 hrw: I'm running a build from scratch here, hopefully I'll hit that Nov 09 16:05:08 RP: 'bitbake icu-native' on scratch is enough Nov 09 16:05:28 hrw: yes, fix pushed Nov 09 16:05:32 re Nov 09 16:05:40 hrw: and something else I'd forgotten to push :( Nov 09 16:05:46 hrw: one of those days, sorry Nov 09 16:05:50 happens Nov 09 16:07:43 NOTE: package icu-native-3.6-r4: task do_build: completed Nov 09 16:13:10 hi cbrake Nov 09 16:23:47 RP: dbus-native lacks dep on libsm-native Nov 09 16:24:20 *g* oe-pbuilder Nov 09 16:25:15 hrw: thats why it used -lSM and -lSomething from buildhost? thanks for catching it.. Nov 09 16:26:18 jama jupp there are some more stuff of this Nov 09 16:27:19 I've seen it because buildhost's libSM.*.so depends also on libuuid.so.1 which was missing in oe build Nov 09 16:29:01 jama jupp Nov 09 16:30:57 yep Nov 09 16:31:47 hrw: We should disable adt_audit instead :) Nov 09 16:32:49 do what you want Nov 09 16:33:15 hrw: We have no users of the audit stuff though, right? Nov 09 16:33:22 and this means no libx11-native ;-) Nov 09 16:36:11 hrw: fix pushed Nov 09 16:37:56 has anyone had to deal with the use of ext2 extended attributes in an oe built filesystem? Nov 09 16:38:31 File "populate_packages", line 11, in populate_packages Nov 09 16:38:31 UnboundLocalError: local variable 'os' referenced before assignment Nov 09 16:38:31 ERROR: Task 3713 (/home/hrw/devel/OH/poky/meta/packages/gnome/gnome-mime-data_2.18.0.bb, do_package) failed Nov 09 16:38:38 RP: next one Nov 09 16:39:00 ~curse libtool & imagemagick Nov 09 16:39:01 May the fleas of a thousand camels infest your most sensitive regions, libtool & imagemagick ! Nov 09 16:40:19 kergoth: in what way? Nov 09 16:41:12 CosmicPenguin: well, consider someone that wants to use acls or something, and wants a read only rootfs. Nov 09 16:41:28 you can't add them in a rootfs postprocess, because it isnt ext2 yet, and i doubt genext2fs supports attributes Nov 09 16:42:07 kergoth: what about own IMAGECMD_ext2 which will make image and then call command which will ACLs? Nov 09 16:42:28 yeah - isn't that exactly like making an ext3 filesystem? Nov 09 16:42:28 can those commands run against an image file rather than files in a mounted ext2 filesystem? Nov 09 16:42:38 hrw: that error should never happen :/ Nov 09 16:42:56 hrw: Trying to get my build up to that... Nov 09 16:43:27 hrw: Reading the function I can't see anything obviously wrong with its use of os ... Nov 09 16:44:36 * kergoth reads up on more details of xattrs Nov 09 16:45:08 RP: add "import os" to populate_packages and it will work Nov 09 16:47:49 and yes, I know - it should work without it Nov 09 16:48:22 hrw: right, the interesting question is why/how it broke Nov 09 16:49:35 03Otavio Salvador  07org.openembedded.dev * r636a0d3acb 10openembedded.git/ (3 files in 3 dirs): Nov 09 16:49:35 xf86-video-cirrus: add 1.3.2 and set it as prefered for X11 7.5 Nov 09 16:49:35 Signed-off-by: Otavio Salvador Nov 09 16:50:51 hrw: the xattr commands are run against files in the mounted filesystem, as far as i can tell. we can't mount the image as root, so how to add attributes to files within an image? :) there's postinsts, but of course that only works if the rootfs is rw on the target Nov 09 16:51:02 * kergoth scratches head Nov 09 16:51:04 * kergoth reads more Nov 09 16:51:10 kergoth: no idea, never played with acl Nov 09 16:51:14 me neither Nov 09 16:51:16 :) Nov 09 16:51:26 kergoth: Can some of the utils not work on a file? Nov 09 16:51:42 not sure what you mean Nov 09 16:52:25 kergoth: You can't be the first person who wants to change an unmounted system. Often the utils can take a /dev/xyz0 option and can equally take a file Nov 09 16:52:41 kergoth: hmm - yeah, that might be trouble Nov 09 16:52:48 the utils don't operate against a device, RP Nov 09 16:52:53 they run against a file in a mounted filesystem Nov 09 16:53:07 kergoth: :( Nov 09 16:53:12 agreed Nov 09 16:53:29 kergoth: Like I said though, you can't be the first person who wants to do this... Nov 09 16:53:33 * kergoth had a prospective customer ask if we knew how to do such a thing, and got confused :) Nov 09 16:53:35 agreed Nov 09 16:53:50 but, often other less nice buildsystems will require you build as root Nov 09 16:53:52 hmm Nov 09 16:54:30 * kergoth googles some more Nov 09 16:54:43 * hrw -> find OE -name old-libtool-get-rid-of.patch Nov 09 16:55:05 ynezz: your work with IEI Nano GX-466 looks very cool, thank you Nov 09 16:55:28 ynezz: which image have you been testing? Nov 09 17:11:14 arghhhh Nov 09 17:11:23 I hate libtool Nov 09 17:11:39 hrw *g* Nov 09 17:11:47 its not libtool fault Nov 09 17:11:53 its dev fault who uses this Nov 09 17:12:16 woglinde: I have old app which used libtool 1.x and need to build it with 2.x libtool Nov 09 17:12:38 yeah so dont ship the m4 macros Nov 09 17:14:29 how do you get a new bitbake? Nov 09 17:14:50 good morning all Nov 09 17:14:55 svn looks not updated Nov 09 17:14:59 mckoan: git clone it Nov 09 17:15:02 kergoth: I think you could build something on top of debugfs to do that. Nov 09 17:15:10 ("that" being your xattr thing) Nov 09 17:15:14 mckoan: its hosted on oe now Nov 09 17:15:31 have a nice rest of day Nov 09 17:15:34 khem: I supposed that, but I can't find it in oe tree Nov 09 17:15:37 mckoan: with qt-embedded Nov 09 17:15:54 ynezz: thx, I'm giving it a try ;-) Nov 09 17:16:07 mckoan: cgit.openembedded.org Nov 09 17:16:21 git clone git://git.openembedded.net/bitbake Nov 09 17:16:50 khem: thx, I'll update GettingStarted page then Nov 09 17:17:14 kergoth: or patch genext2fs to read xattrs from some kind of manifest file during image construction. I don't think that'd be very hard. Nov 09 17:17:22 mckoan: I had to hack qt-embedded a lot to get it compile, I was testing it with qt-embedded-4.5.3 Nov 09 17:18:05 i thought about patching genext2fs as well.. would have to learn more about ext2 first.. :) Nov 09 17:18:32 patching genext2fs is not a terrible idea Nov 09 17:19:20 ynezz: qt-e = fb0 ? Nov 09 17:19:38 hmm, wonder why no one else seems to have done such a thing Nov 09 17:19:39 p Nov 09 17:19:45 probably dont mind doing shit as root.. silly people Nov 09 17:19:57 mckoan: yep Nov 09 17:20:30 of course, could create an x86 chroot, run qemu in it, and mount it as root in there Nov 09 17:20:31 hehe Nov 09 17:20:45 ynezz whats was wrong in qt-emebedded Nov 09 17:20:57 kergoth: looks like debugfs "mi" command does do what you want, although not in a convenient format Nov 09 17:21:00 ynezz: so are you using a customized image? Nov 09 17:21:21 kergoth: it probably wouldn't be beyond the wit of man to hack that to provide a suitable batch interface to it. Nov 09 17:22:23 mckoan: I was using custom image, I decided for beagleboard instead of nano for this project :) Nov 09 17:22:59 ah Nov 09 17:23:03 hmm Nov 09 17:23:27 woglinde: 4.5.3 for example don't provide tar.gz source archive, I'll try to pastebin my modifications Nov 09 17:23:55 coreutils-native_7.2.bb failed Nov 09 17:24:10 ynezz: ok thx Nov 09 17:24:14 ynezz ah so you mean there is no new version? Nov 09 17:25:03 ynezz: BTW I'm talking about Machine IEI Nano GX-466 Nov 09 17:25:05 mckoan qt-embedded has some more output modules then fb0 Nov 09 17:25:48 woglinde: uh, good to know, I used qt/e years ago Nov 09 17:26:54 khem: Sorry, python 2.6 or later is required for this version of bitbake Nov 09 17:27:13 mckoan: what version is that, master? Nov 09 17:27:20 I think 1.8 should be happy with python 2.5 Nov 09 17:27:21 pb yes Nov 09 17:27:25 khem: In Lenny there is not python 2.6, does this mean I have to build it from src? Nov 09 17:27:42 mckoan: I think you can use the version from ubuntu Nov 09 17:27:43 mckoan you can take it from unstable or ubuntu Nov 09 17:27:51 args experimental Nov 09 17:27:55 but this topic at oedem Nov 09 17:27:58 +was Nov 09 17:27:59 woglinde: hm, I wipped out whole dir and removed that git branch :( Nov 09 17:27:59 pb__: the problem is that since today I'm no longer able to bitbake, yahooo! Nov 09 17:28:15 woglinde: but it's easy to reproduce, try to update 4.5.2 package to 4.5.3 Nov 09 17:28:18 mckoan: what goes wrong, exactly? Nov 09 17:28:30 I'm not sure that upgrading to bitbake master is going to be the solution to your problem. Nov 09 17:29:13 woglinde: I meant, that there's no ftp://ftp.trolltech.com/qt/source/qt-embedded-linux-opensource-src-${PV}.tar.bz2 for 4.5.3 Nov 09 17:29:23 woglinde: there's only tar.gz Nov 09 17:30:41 woglinde: and qt-embedded.inc relies on tar.bz2 Nov 09 17:32:21 mckoan: or use the bitbake releases: wget http://download.berlios.de/bitbake/bitbake-1.8.16.tar.gz Nov 09 17:32:52 ynezz: after the require qt-embedded.inc, just override the SRC_URI, maybe? Nov 09 17:34:41 pb__: initally I had "Bitbake version 1.8.16 is required and version 1.8.13 was found" because I was stuck at svn version Nov 09 17:34:47 ynezz so whats the problem do it as likewise say Nov 09 17:34:51 own SRC_URI Nov 09 17:34:57 likewise: yep, that would fit Nov 09 17:35:34 woglinde: ok, I would do it next time I'll play with qt-embedded Nov 09 17:36:31 * mckoan bitbaking again, phew! Nov 09 17:36:41 mckoan: :-) Nov 09 17:36:53 pb__: damn.. debugfs can change the i_file_acl field in the inode, but that's not the actual data, it points to a block that holds the attributes.. and i don't see commands to mess with that :) itd probably be easier to add them to that than to add support to genext2fs though. thanks for the pointer Nov 09 17:37:03 mckoan: IEI Nano GX-466 is x86? Nov 09 17:37:24 so whatdo you think if I put this solution (wget 1.8.16) in GettingStarted? Nov 09 17:37:31 likewise: it's geode, so x86 Nov 09 17:37:34 likewise: yep is geode Nov 09 17:37:59 mckoan: that's cool. Also mentioned the git (khem mentioned) for people who want both bitbake and .dev head. Nov 09 17:38:06 wtf is going on, this unbound variable error only occurs if my function is called populate_packages Nov 09 17:38:26 it does not happen with another function name Nov 09 17:38:34 pb__: it is scriptable, btw, it can accept a command on stdin Nov 09 17:38:37 pb__: :) Nov 09 17:39:33 mckoan: right, 1.8.16 is what you want. I think you can get that from the 1.8 branch in git or from the tarball. Nov 09 17:39:38 git master is almost certainly not what you want. Nov 09 17:39:47 kergoth: ah, jolly good Nov 09 17:44:04 So an "import os.path" later in the function causes os to become a locally unbound variable Nov 09 17:44:14 (at the start) Nov 09 17:46:03 This testcase is insane :/ Nov 09 17:46:06 I think that my new entry would be useful for search engines too when you look for those specific errors you may face to http://wiki.openembedded.net/index.php/Getting_started#Obtaining_BitBake Nov 09 17:48:04 ynezz: pastebin your image if you can Nov 09 17:48:48 kergoth: apropos the e2factory thing: rather than running all compiles in qemu, I wonder if you could patch fakeroot to give you a sort of "fakechroot" kind of function. Nov 09 17:49:10 right now fakeroot doesn't wrap enough functions to do chroot but it would probably not be especially hard to arrange. Nov 09 17:49:51 mckoan: I can't, I've deleted that branch :( I thought, that nobody would be interested Nov 09 17:50:10 mckoan: but I can dd my CF card and put it somewhere if it helps Nov 09 17:50:58 hrw|gone: git pull should give you a fix Nov 09 17:52:00 pb__: someone did start a fakechroot project, not sure how well it works. there's also a fakeroot-ng, but it uses ptrace rather than LD_PRELOAD, so it's slower Nov 09 17:52:04 ynezz: CF ? what do you mean? the target image? Nov 09 17:52:22 mckoan: yes, that's only thing I've left now... Nov 09 17:53:26 kergoth: I guess the overhead from ptrace would not be too bad for compilation since gcc does not make many syscalls. Nov 09 17:53:31 it'd be fairly awful for make, though. Nov 09 17:53:39 * kergoth nods Nov 09 17:53:52 i like that you don't have to worry about LD_PRELOAD being lost or overridden though Nov 09 17:55:36 true. of course, could wrap setenv() and suchlike to make that less likely. Nov 09 17:56:09 ynezz: no prob, I try some images now to see what happens Nov 09 17:56:17 RP: after your modifications my build stops at coreutils-native, I've now reverted to your pre tag and it builds Nov 09 17:56:22 actually, you could just wrap exec* and fiddle them to invoke ld.so with an appropriate --library-path argument, forget about LD_PRELOAD in the environment. Nov 09 17:56:29 RP: I did rm -fr tmp always Nov 09 17:56:36 ynezz: how did it break? Nov 09 17:56:36 have a nice rest of the day Nov 09 17:57:23 RP: in do_compile Nov 09 17:57:43 ynezz: Can you pastebin the log please Nov 09 17:58:15 RP: yes, I'll do, I have to rebuild again Nov 09 17:58:24 pb__: good idea Nov 09 17:58:50 ynezz: failures are only useful if I have the info to fix them ;-) Nov 09 17:59:05 * kergoth wants to see an oe self hosted setup at some point, get it to build itself a filesystem to build for the target in.. Nov 09 17:59:49 kergoth: The nativesdk stuff may make that a lot easier Nov 09 18:00:15 true, build enough of those and you could chroot right into the sysroot, eh? :) Nov 09 18:00:58 kergoth: yes Nov 09 18:01:18 kergoth: All the tools have their own libc and so on Nov 09 18:01:19 kergoth: right, that'd be neat. Nov 09 18:01:21 that could be fun to experiment with Nov 09 18:02:13 * pb__ go home now Nov 09 18:02:44 later all Nov 09 18:02:49 * kergoth thinks of a recipe chrooting into sysroot and running bitbake in there.. bitbake under bitbake.. yikes :) Nov 09 18:02:53 later pb__ Nov 09 18:04:50 course, if you could, programmatically from a python task tell bitbake to parse the new recipes and all and produce a second cooker for the rest, that could be pretty slick Nov 09 18:05:06 have a task in the first cooker fire off the second :) Nov 09 18:09:20 RP: I've two cases, it's always in different place http://pastebin.ca/1663691 http://pastebin.ca/1663685 Nov 09 18:13:08 03Martin Jansa  07shr/merge * r04b763dabb 10openembedded.git/recipes/tasks/task-shr-minimal.bb: Nov 09 18:13:08 task-shr-minimal: bump PR for shr-theme-gry Nov 09 18:13:08 Signed-off-by: Martin Jansa Nov 09 18:18:00 bye I will going home too Nov 09 18:28:44 ynezz: what distro is your host system running Nov 09 18:33:06 khem: ubuntu Nov 09 18:33:12 which version Nov 09 18:35:24 interpid, so 8.10? Nov 09 18:35:38 don't know where to get version :) Nov 09 18:35:46 update manager offers me 9.04 Nov 09 18:35:47 yeah Nov 09 18:36:27 ynezz: I have 8.04 and it seems to be building coreutils-native successfully Nov 09 18:36:34 hmpf Nov 09 18:36:51 khem: did you rm -fr tmp before? Nov 09 18:38:38 but if I reset to pre-nov2009-core-updates it's now building task 1064 of 7285 for beagleboard-demo-image, so no problem Nov 09 18:40:34 ynezz: I do that atleast twice in a week Nov 09 18:40:53 and the one I build from scratch is from yesterday uptodate .dev Nov 09 18:41:08 it should be today, after all the RP's changes Nov 09 18:41:16 ah Nov 09 18:43:33 trying now on debian-testing box Nov 09 18:45:58 do_package_stage_all and do_rootfs are now executed in parallel? with 2 or more BB_THREADS? Nov 09 18:46:56 ynezz: some of your config tests seems to fail Nov 09 18:47:21 ynezz: do you have config.log from the failed run Nov 09 18:47:26 yes Nov 09 18:47:37 paste it somewhere Nov 09 18:47:53 configure: WARNING: sys/bitypes.h: present but cannot be compiled Nov 09 18:48:04 this is a potential problem Nov 09 18:49:01 khem: http://pastebin.ca/1663750 Nov 09 18:49:27 thats not config.log Nov 09 18:49:34 ah, sorry :) Nov 09 18:49:47 config.log will be inside the coretutils dir Nov 09 18:52:33 http://filebin.ca/vtpatj/config.log Nov 09 18:55:31 your host installtion of gcc seems to be culprit here Nov 09 18:55:43 or may be its something RP added Nov 09 18:55:52 let me take a quick look Nov 09 18:56:31 I've build that image succesfully yesterday Nov 09 18:57:23 lets see ...can you compile a simple program with your host CC Nov 09 18:57:26 #include Nov 09 18:57:34 int main () { Nov 09 18:57:41 int x = sizeof (ssize_t *) + sizeof (ssize_t); Nov 09 18:57:46 return !x; Nov 09 18:57:56 } Nov 09 18:58:07 good morning #oe Nov 09 18:58:15 gcc -std=gnu99 -c test.c Nov 09 19:00:01 khem: http://pastebin.ca/1663763 Nov 09 19:00:08 no error Nov 09 19:00:24 gcc (Ubuntu 4.3.2-1ubuntu12) 4.3.2 Nov 09 19:00:29 ynezz: ok now add this -isystem/media/data/devel/oe/beagle.git/tmp/staging/i686-linux/usr/include Nov 09 19:00:36 to compiler commandline Nov 09 19:00:56 no error either Nov 09 19:01:31 ok ... now edit the testcase and add #include at top Nov 09 19:01:37 and redo Nov 09 19:02:19 it's ok Nov 09 19:03:48 ynezz: http://khem.pastey.net/127989 Nov 09 19:03:54 try this test instead Nov 09 19:06:15 it's ok Nov 09 19:06:51 gcc -std=gnu99 -c -isystem/media/data/devel/oe/beagle.git/tmp/staging/i686-linux/usr/include -O2 -g -isystem/media/data/devel/oe/beagle.git/tmp/staging/i686-linux/usr/include test.c Nov 09 19:07:03 * kergoth ponders Nov 09 19:07:47 it's ok Nov 09 19:07:48 ynezz: what happens if you also use ccache Nov 09 19:08:00 prefix the above command with ccache Nov 09 19:08:04 it's ok Nov 09 19:08:12 ccache gcc -std=gnu99 -c -isystem/media/data/devel/oe/beagle.git/tmp/staging/i686-linux/usr/include -O2 -g -isystem/media/data/devel/oe/beagle.git/tmp/staging/i686-linux/usr/include test.c Nov 09 19:08:43 http://pastebin.ca/1663778 Nov 09 19:09:09 otavio: fyi, there was a typo in your bitbake branch email - s/uses/users/ ;) Nov 09 19:11:02 khem: but this test fails http://pastebin.ca/1663780 Nov 09 19:11:43 khem: but it's because of missing sys/types.h right? Nov 09 19:12:21 * kergoth is kind of sick of bb.fetch's half-assed OO Nov 09 19:13:40 ynezz: http://paste.ubuntu.com/314367/ Nov 09 19:13:44 last chance Nov 09 19:13:49 try this test Nov 09 19:14:01 yes Nov 09 19:14:07 thats the problem Nov 09 19:16:10 khem: http://pastebin.ca/1663787 Nov 09 19:16:12 problem Nov 09 19:17:16 ynezz: somehow this test gets prepared wrongl Nov 09 19:17:17 y Nov 09 19:17:35 hm Nov 09 19:20:20 03Martin Jansa  07shr/merge * r8ad8b37db1 10openembedded.git/recipes/gpe-icons/gpe-icons_0.25.bb: Nov 09 19:20:20 gpe-icons: install to directory managed by update-alternatives Nov 09 19:20:20 Signed-off-by: Martin Jansa Nov 09 19:21:30 on my debian box it's ok Nov 09 19:24:35 ynezz: hmmm this fails on different ubuntu's and also on fedora as I try Nov 09 19:25:39 strange Nov 09 19:30:43 has anyone ever looked into a mod_python recipe for OE? Nov 09 19:39:33 hi Nov 09 19:39:46 hey hrw Nov 09 19:41:24 ynezz: one thing you could try is to autoreconf during do_configure if it is not run Nov 09 19:41:37 or disable it if it is being run Nov 09 19:44:52 mm, looks like bitbake 1.10 is coming along nicely Nov 09 19:50:28 Ok, noob question: I built an x11-image successfully, and it runs. Then I ran bitbake meta-toolchain-qt4, which ran fine. I then put the x11 image on my device, and used the SDK to make the test executable. But when I try to run it, it's missing basic QT libs (libQtGuiE.so.4). What step did I miss to get the QTE stuff into the x11 image? Nov 09 19:53:06 kergoth: indeed Nov 09 19:53:49 etalvala: your image does not contain the qt libraries Nov 09 19:53:51 otavio: looks pretty sane to me, but someone who knows the srcrev stuff may want to look at it instead, i'm not very familiar with that part of bb.fetch :) Nov 09 19:54:01 which is quite normal for an x11-image Nov 09 19:54:20 * kergoth tries to remember what issues he hit with the bb ast Nov 09 19:54:43 florian: may be x11-gpe-image will have them ? Nov 09 19:55:40 florian: Right, I assumed the meta-toolchain bit took care of that - do I just do some opkg installs to fix it on the target? Nov 09 19:55:54 kergoth: maybe the refactoring code might be merged in meanwhile? Nov 09 19:56:04 seems reasonable Nov 09 19:56:25 i'll do it if i find the time, otherwise ask richard, he's heading up the 1.10 work :) Nov 09 19:56:34 kergoth: this avoids diverging for no reason Nov 09 19:56:37 khem: eh no, gpe is gtk based :) Nov 09 19:56:41 * kergoth nods Nov 09 19:56:55 oh, i remember what i didn't like about ast.. the way it handles inclusion Nov 09 19:57:01 * kergoth wonders where zecke is Nov 09 19:57:19 florian: yeah indeed :) neuros-osd-base-image seems to be qt based Nov 09 19:58:17 kergoth: taipei :-} Nov 09 19:58:34 heh :) Nov 09 19:58:48 etalvala: meta-toolchain-qt4 would do this - but for the toolchain only. it does not influence the images. you could easily define your own image that pull in qt as well or if you just want a fast solution for testing you could copy the files from the toolchain. Nov 09 19:59:04 khem: good to know if you need soemthing for testing :) Nov 09 19:59:34 currently, the Node object for inclusion waits until eval() time to parse the included files.. so toplevel recipes are parsed right away and return a statement group, but files they include don't get parsed until later. would be better to have the included files also parsed right away, and return a statement group inside of the main statement group to be evaluated.. just wanted to run it by him before trying to implement it Nov 09 19:59:35 heh Nov 09 20:00:25 florian: Thanks! So I could just scrape all the libs from the toolchain tarball, or add the meta task as an RDEPEND in my image build? Nov 09 20:01:44 florian: Or would it be recipes/tasks/task-qte-toolchain-target.bb as an RDEPEND Nov 09 20:01:47 kergoth: sort of lazy evaluation Nov 09 20:02:13 yeah, that's the point behind the ast work that zecke did on the parser Nov 09 20:02:29 * kergoth glares at the clock Nov 09 20:07:01 khem: will try Nov 09 20:09:31 kergoth: like C compilers this seems like pre-processor sort of function Nov 09 20:10:10 so parsing include files recursively seems familiar to me Nov 09 20:11:23 * kergoth nods Nov 09 20:11:43 * kergoth knows very little about parsing and compiler stuff, need to read the dragon book or something.. Nov 09 20:12:17 * khem has done learn by doing :) Nov 09 20:12:38 dragon book is nice though I got hold of it late Nov 09 20:17:48 etalvala: you should add the libraries there only... or just a simple qt app that pulls in the dependencies, that's much easer. Nov 09 20:18:15 * florian wonders where meta-toolchain-qt4 is from, imho we do not have this in .dev Nov 09 20:18:36 florian: Here's the doc for it: http://docs.openembedded.org/usermanual/html/ch05s08.html Nov 09 20:20:19 doc says it qte not qt4 which is in sync :) Nov 09 20:20:47 khem: I thought those were the same thing in QT4... Nov 09 20:21:21 which has absolutely nothing to do with the name of the recipe Nov 09 20:22:02 kergoth: sorry - I keep getting the name backward Nov 09 20:23:27 Thanks for the help, all - I think this is enough to get me in the right direction Nov 09 20:45:51 re Nov 09 20:50:44 hm what is the best way to revert a commit? Nov 09 20:51:33 http://www.kernel.org/pub/software/scm/git/docs/user-manual.html#fixing-mistakeswoglinde: Nov 09 20:51:38 omg.. Nov 09 20:51:41 woglinde: http://www.kernel.org/pub/software/scm/git/docs/user-manual.html#fixing-mistakes Nov 09 20:53:01 I want to revert 2c92869662d1f0b28229cdba9e9852ba642e0733 and 7e28606c9432c4fcb7e633c2dfb699eee14094c6 Nov 09 20:54:20 git revert then imho Nov 09 20:54:39 hm git revert uuid? and then push? Nov 09 20:54:47 iirc yes Nov 09 20:54:59 can you git revert a pushed change? I think not Nov 09 20:54:59 look man git-revert before to be sure :) Nov 09 20:55:18 git revert is good only if you have messed your working tree Nov 09 20:55:19 it will create new commit that reverting changes Nov 09 20:55:21 khem hm? I saw many reverts pushed? Nov 09 20:55:45 woglinde: yes but you will still have the history Nov 09 20:56:01 aeh yes Nov 09 20:56:18 but you dont have the stuff in actual state Nov 09 20:56:24 *sigh* Nov 09 20:57:34 you can do something like git push -f Nov 09 20:58:12 no Nov 09 20:58:22 I want to do it as they other has done Nov 09 21:03:42 woglinde: prolly git revert Nov 09 21:03:46 and then git push Nov 09 21:04:06 prolly? Nov 09 21:04:06 actually git revert -s Nov 09 21:04:19 shorthand for probably ;) Nov 09 21:07:24 hm okay I will push Nov 09 21:07:49 03Henning Heinold  07org.openembedded.dev * r8fb41edf83 10openembedded.git/conf/machine/ (geodegx.conf geodelx.conf): Nov 09 21:07:49 Revert "geodegx: Fix glibc configure error" Nov 09 21:07:49 This reverts commit 7e28606c9432c4fcb7e633c2dfb699eee14094c6. Nov 09 21:07:49 Signed-off-by: Henning Heinold Nov 09 21:07:50 yes, git revert is the right thing to do. Nov 09 21:07:59 "git push -f" is definitely not a good idea Nov 09 21:08:00 03Henning Heinold  07org.openembedded.dev * r8e41a1e490 10openembedded.git/ (4 files in 4 dirs): Nov 09 21:08:00 Revert "geodegx: Linux 2.6.31.4" Nov 09 21:08:00 This reverts commit 2c92869662d1f0b28229cdba9e9852ba642e0733. Nov 09 21:08:26 hm seems to work Nov 09 21:08:31 very good Nov 09 21:08:49 geode stuff was to fast Nov 09 21:08:54 have to lart myself Nov 09 21:09:26 heh Nov 09 21:09:29 ~lart woglinde Nov 09 21:09:29 * ibot DoSes woglinde Nov 09 21:09:45 thanks Nov 09 21:10:09 any time Nov 09 21:10:24 hi bluelightning Nov 09 21:10:26 woglinde: is geode an amd machine ? Nov 09 21:10:38 khem yes arm processor Nov 09 21:10:47 mostly found on wrap and alix boards Nov 09 21:10:47 hi woglinde, all Nov 09 21:10:53 args amd Nov 09 21:10:58 its i486 Nov 09 21:10:59 woglinde: amd, not arm Nov 09 21:11:00 heh Nov 09 21:11:40 it was cyrix/natsemi originally, but I think amd bought that division at some point. Nov 09 21:11:48 jo Nov 09 21:12:14 I made a faxmachine out of wrap board Nov 09 21:12:31 sweet :) Nov 09 21:12:50 yeah with a isdn card it would be cooler Nov 09 21:12:53 pb_: I see mips technology has OCI tools for those chips Nov 09 21:13:01 http://www.mips.com/products/software-tools/legacy/system-navigator-probe-for-amd-geode-gx-and-lx-processors/ Nov 09 21:13:05 that confused me Nov 09 21:13:09 because I can easily integrate voic answer Nov 09 21:13:16 I thought is it like alchemy or some such Nov 09 21:13:24 woglinde: is it a voip fax machine or pstn? Nov 09 21:13:25 but minipic isdn cards are fucking pricy Nov 09 21:13:43 smith normal minipci modem/fax card Nov 09 21:13:58 unfornatly there arent many minipci modem/fax/voice cards Nov 09 21:14:15 khem: yeah, that is a bit odd Nov 09 21:14:24 intel made some chips which can do it but seems no one really made a card Nov 09 21:14:37 only normal pci cards Nov 09 21:14:43 *sigh* Nov 09 21:15:02 singel mini isdn pci cards start at 100 euros Nov 09 21:15:41 03Martin Jansa  07shr/merge * r47e94ea0ea 10openembedded.git/conf/distro/include/shr-autorev.inc: Nov 09 21:15:41 libfsobasics: unpin git revision, resolved upstream Nov 09 21:15:41 Signed-off-by: Martin Jansa Nov 09 21:16:09 unfornatly the faxmachine runs debian instead of oe based distribution Nov 09 21:16:19 but hylafax suckz at building Nov 09 21:17:22 100 eur does sound a bit steep for just a single bri card Nov 09 21:17:33 minipci Nov 09 21:17:36 I wonder whether any of the regular pci ones could be made to work on a minipci interface Nov 09 21:17:37 you dont get it cheaper Nov 09 21:17:49 pb case is to small Nov 09 21:17:53 ah Nov 09 21:17:59 03Khem Raj  07org.openembedded.dev * r189fc6375f 10openembedded.git/recipes/mono/ (2 files in 2 dirs): Nov 09 21:17:59 mono: Replace cross compilation workaround for x86 architecture. Nov 09 21:17:59 * Replace the workaround with backported patches from mono 2.4 branch Nov 09 21:17:59 Signed-off-by: Khem Raj Nov 09 21:18:02 usb? Nov 09 21:18:04 case looks like 2,5 disk Nov 09 21:18:12 usb draws more power Nov 09 21:18:29 2,5 external disk Nov 09 21:18:35 pretty neat Nov 09 21:20:13 ah, oh well. isdn bri is probably only interesting for germans anyway :-} Nov 09 21:20:21 *g* Nov 09 21:20:37 you can make good money with asterisk in de Nov 09 21:20:42 yeah Nov 09 21:21:05 here in the uk, virtually nobody has bri. companies have isdn pri, and domestic users have pots. Nov 09 21:21:23 in north america there's hardly any isn Nov 09 21:21:25 *isdn Nov 09 21:21:40 except for people with overpriced^Wawesome videoconferencing setups Nov 09 21:21:45 which is a pain for us, actually, because pri is way more capacity than we actually need. if there was a way for us to have, say, two bri interfaces instead then that'd be fine. Nov 09 21:21:49 but, no carrier will sell that to us. Nov 09 21:22:19 pb yeah he normal standard is 2 bri lines Nov 09 21:22:24 here Nov 09 21:22:25 args Nov 09 21:22:51 you mean two channels, right? Nov 09 21:23:01 i.e. one physical pair of wires Nov 09 21:23:07 right Nov 09 21:23:21 Continued noobism: Is meta-toolchain-qte meant to build an SDK for QT-without-X, only? Is there anything comparable to make an SDK for QT under X? Nov 09 21:23:30 yeah, you can get that here as well for domestic service Nov 09 21:23:41 it's about 5x the price of pots, though, so very few people do Nov 09 21:24:46 hm pots is normal analog line? Nov 09 21:25:25 yeah Nov 09 21:25:36 actually, it seems that bt don't even offer isdn bri for domestic use anymore Nov 09 21:25:45 http://bt.custhelp.com/cgi-bin/bt.cfg/php/enduser/cci/bt_adp.php?p_sid=&p_faqid=8337&cat_lvl1=346&p_cv=1.346&p_cats=346&s_cid=con_FURL_homehighway Nov 09 21:25:51 apparently it was discontinued in 2005 :-} Nov 09 21:25:52 hihi bt suckz Nov 09 21:26:01 anyway Nov 09 21:26:10 if read the blog posts from mathew Nov 09 21:26:22 you can still order isdn2e as a business service but it costs about 600EUR per year Nov 09 21:26:53 whereas an analog phone on the domestic tariff is way less than that Nov 09 21:27:32 hm we are paying 35 euros for isdn/dsl Nov 09 21:27:41 where inet connection is flat Nov 09 21:27:45 oh yeah, and isdn2e as deployed by bt is incompatible with dsl Nov 09 21:27:46 heh Nov 09 21:28:06 so, if you want a dsl connection you need to have a separate analog phone line anyway Nov 09 21:28:13 lol Nov 09 21:28:23 yeah you can do it here too Nov 09 21:28:32 but isdn is only 5 euros more Nov 09 21:28:41 so better 2 isdn lines with 3 numbers Nov 09 21:28:45 actually, that's not quite fair. it's rather the case that dsl as deployed by bt is incompatible with isdn. Nov 09 21:28:56 i.e. it is the adsl that's different here, not the isdn Nov 09 21:29:10 *g* Nov 09 21:29:16 this suckz anyway Nov 09 21:29:20 yeah Nov 09 21:29:32 the upshot is that it is very unattractive to have isdn for phone use Nov 09 21:29:34 what you have at home pb? Nov 09 21:29:40 just analog phone and dsl Nov 09 21:29:44 ah Nov 09 21:29:45 okay Nov 09 21:29:48 at the office we have a pri circuit Nov 09 21:29:59 and a second fibre connection for internet Nov 09 21:29:59 yeah buisness is different Nov 09 21:30:19 but you can oder here vdsl with 50 mbit and inet-tv Nov 09 21:30:22 for 50 euros Nov 09 21:30:32 order Nov 09 21:34:38 yah Nov 09 21:34:46 I'm not sure if we even have vdsl at all in the uk yet Nov 09 21:35:12 yeah here too Nov 09 21:35:21 I only have 2 mbit adsl Nov 09 21:41:31 woglinde: I can suggest business-idea about faxing :) Nov 09 21:41:42 jay7 lol Nov 09 21:41:46 I know them all Nov 09 21:41:53 just create service that will recieve fax and send ot via email to customer :) Nov 09 21:41:59 lol Nov 09 21:42:01 lol Nov 09 21:42:05 thats already here Nov 09 21:42:22 well.. that is known idea then :) Nov 09 21:42:41 jay7 hm hylafax can do this since email exists Nov 09 21:42:52 hm.. I should consider that for my customers.. Nov 09 21:42:53 efax.com Nov 09 21:56:10 03Tom Hacohen  07shr/merge * r55c853620b 10openembedded.git/recipes/shr/ (3 files): Nov 09 21:56:10 fixed the gtk themes Nov 09 21:56:10 Signed-off-by: Martin Jansa Nov 09 22:00:04 * * OE Bug 5333 has been created by quickx(AT)hotmail.com Nov 09 22:00:06 * * at91sam9261ek Build Fails at sqlite3 Nov 09 22:00:09 * * http://bugs.openembedded.org/show_bug.cgi?id=5333 Nov 09 22:04:00 re kergoth Nov 09 22:12:33 khem: looks that I fixed it, I had full home dir... :P thanks a lot for your help anyway, learned something :) Nov 09 22:16:01 hmm Nov 09 22:16:17 is there a way to get the working directory of one package (i.e. apache2) from inside another package (i.e. modpython)? Nov 09 22:16:17 03Frans Meulenbroeks  07shr/merge * rbab585ab7b 10openembedded.git/ (conf/checksums.ini recipes/help2man/help2man_1.36.4.bb): help2man: added recipe created by Lynn Lin (bugzilla #4774) Nov 09 22:16:19 03Frans Meulenbroeks  07shr/merge * ra7531d1ef4 10openembedded.git/: Merge branch 'org.openembedded.dev' of git.openembedded.org:openembedded into org.openembedded.dev Nov 09 22:16:20 03Frans Meulenbroeks  07shr/merge * r4b03e3165f 10openembedded.git/: Merge branch 'org.openembedded.dev' of git.openembedded.org:openembedded into org.openembedded.dev Nov 09 22:16:21 03Petr Štetiar  07shr/merge * rbec85b495e 10openembedded.git/ (conf/checksums.ini recipes/linux/linux_2.6.31.bb): Nov 09 22:16:24 linux: Linux 2.6.31.5 Nov 09 22:16:26 Signed-off-by: Petr Štetiar Nov 09 22:16:28 Acked-by: Otavio Salvador Nov 09 22:16:30 03Frans Meulenbroeks  07shr/merge * rce0f6d41d8 10openembedded.git/ (conf/checksums.ini recipes/grun/grun_0.9.3.bb): grun: added recipe (created by tovarish, bugzilla 5067) Nov 09 22:16:33 03Frans Meulenbroeks  07shr/merge * r8ff7186c01 10openembedded.git/ (conf/checksums.ini recipes/texi2html/texi2html_1.82.bb): texi2html: added recipe contributed by Lynn Lin (bugzilla #4759) Nov 09 22:16:40 03Petr Štetiar  07shr/merge * rdbee1acac7 10openembedded.git/conf/machine/iei-nanogx-466.conf: Nov 09 22:16:43 Machine configuration for IEI Nano GX-466 Nov 09 22:16:45 Signed-off-by: Petr Štetiar Nov 09 22:16:47 Acked-by: Otavio Salvador Nov 09 22:16:49 03Frans Meulenbroeks  07shr/merge * rbb4e450ad8 10openembedded.git/recipes/mysql/mysql5_5.1.40.inc: mysql5: fixed packaging Nov 09 22:16:54 03Alexandre Tisserant  07shr/merge * rc78f8925f4 10openembedded.git/recipes/gnome-mplayer/ (2 files in 2 dirs): Nov 09 22:16:57 gecko-mediaplayer: make ai-flash.js work with midori as well Nov 09 22:16:59 * midori only supports scripts in ~/.local/share/midori/scripts so users will have to do Nov 09 22:17:01 mkdir -p ~/.local/share/midori/scripts ; cp /usr/lib/mozilla/extensions/flash@alwaysinnovating.com/chrome/content/ai-flash.js ~/.local/share/midori/scripts/ Nov 09 22:17:03 dejavu Nov 09 22:17:10 03Richard Purdie  07shr/merge * r0d722f5b12 10openembedded.git/classes/base.bbclass: Nov 09 22:17:13 base.bbclass: Drop legacy rebuild mangling code Nov 09 22:17:15 Signed-off-by: Richard Purdie Nov 09 22:17:16 ynezz: anothere branch.. Nov 09 22:17:17 03Richard Purdie  07shr/merge * r6111d7132a 10openembedded.git/classes/package.bbclass: Nov 09 22:17:24 package.bbclass: Add the option to declare package preprocess function hooks (from Poky) Nov 09 22:17:24 ah Nov 09 22:17:26 Signed-off-by: Richard Purdie Nov 09 22:17:28 03Petr Štetiar  07shr/merge * r2d8e4dfc56 10openembedded.git/recipes/wv/wv_1.2.0.bb: Nov 09 22:17:30 wv: fix compile problem Nov 09 22:17:32 base_do_compile() checks if there's Makefile or makefile in source directory, Nov 09 22:17:34 and then it runs oe_runmake. Since there's GNUmakefile generated, change Nov 09 22:17:38 do_compile accordingly. Nov 09 22:17:44 Signed-off-by: Petr Štetiar Nov 09 22:17:46 03Richard Purdie  07shr/merge * r9a5937f6c6 10openembedded.git/classes/base.bbclass: Nov 09 22:17:49 base.bbclass: Drop import bb and import os as no longer needed Nov 09 22:17:51 Signed-off-by: Richard Purdie Nov 09 22:17:55 03Richard Purdie  07shr/merge * rf6c3cb05c0 10openembedded.git/classes/autotools.bbclass: Nov 09 22:18:01 autotools.bbclass: Create m4 directory earlier avoding issues with intltoolize (from Poky) Nov 09 22:18:02 Signed-off-by: Richard Purdie Nov 09 22:18:04 03Richard Purdie  07shr/merge * rc942cea699 10openembedded.git/classes/ (base.bbclass packagedata.bbclass): Nov 09 22:18:07 packagedata.bbclass: Split package data handling functions out from base.bclass (from Poky) Nov 09 22:18:11 Signed-off-by: Richard Purdie Nov 09 22:18:12 wow, huge commit going on? Nov 09 22:18:16 (50 lines omitted) Nov 09 22:18:19 merge... Nov 09 22:18:46 msmith jupp Nov 09 22:18:49 I wonder if there is a way to prevent shr/merge from sending email & CIA updates? Nov 09 22:18:56 that will be fixed after the split Nov 09 22:18:59 of git stuff Nov 09 22:19:07 Oh is there another tree coming? cool Nov 09 22:19:10 look at the oedem notes Nov 09 22:19:13 sorry for that noise.. Nov 09 22:19:22 jama dont worry Nov 09 22:19:36 yeah my delete button works pretty well :) Nov 09 22:19:43 shr devs hopes to start pushing stuff to oe.dev and drop shr/merge soon :) Nov 09 22:19:51 jama cool Nov 09 22:20:01 .dev will be master in the near future Nov 09 22:20:14 so stay tooned Nov 09 22:20:15 *g* Nov 09 22:20:25 ynezz: btw. coreutils-native built just fine after I pulled in rp's changes Nov 09 22:20:28 ah, that will be easier to spell :) Nov 09 22:20:57 khem pitty you didnt make it to oedem Nov 09 22:21:02 woglinde: hmm haven't seen it in RP's oedem notes, are there any other notes? Nov 09 22:21:08 jama hm Nov 09 22:21:18 woglinde: yeah Nov 09 22:21:21 than we were tired to write it Nov 09 22:21:24 I regret too Nov 09 22:21:26 but it was discussed Nov 09 22:21:37 khem maybee next year Nov 09 22:22:00 woglinde: ah, ok.. looking forward to master :) Nov 09 22:22:07 hehe okay Nov 09 22:22:13 gobby is real cool Nov 09 22:22:15 woglinde: oe.dev is so long to type :) Nov 09 22:22:22 jama jupp Nov 09 22:22:29 and its a branch Nov 09 22:22:31 yeah or build a bigger base for oe devs in the US and host one here Nov 09 22:22:34 rather than master Nov 09 22:22:36 in kernel git Nov 09 22:23:05 khem hm will be cheaper to fly the few americans over Nov 09 22:23:13 than the europeans Nov 09 22:23:33 Is OE really that Euro focused? Nov 09 22:23:38 heh yes Nov 09 22:23:53 DJWillis: most of devs as of now are in eu Nov 09 22:24:17 *g* at least one and half is in asia Nov 09 22:24:27 lol Nov 09 22:24:27 .it, .de, .pl :) Nov 09 22:25:21 DJWillis: I don't think oe is all that euro-focussed; there are plenty of developers in NA. but, oedem is meant to be the oe developer's european meeting, so there is a certain amount of euro bias implicit in that event :-) Nov 09 22:25:39 Jay7: it would be easier to write EU states without OE developers :) Nov 09 22:26:41 hehe :) Nov 09 22:26:46 pb there is no european in oedem Nov 09 22:27:24 let's do next oedem in australia :) Nov 09 22:27:44 some parts of the carabiean still count as europe :-) Nov 09 22:27:47 hm at raster? Nov 09 22:27:56 DJWillis: of course, the americans are welcome to hold an oedam. Nov 09 22:28:14 XorA|gone: put me down for that one ;). Nov 09 22:29:38 actually, at the recent oedem, something like 25% of the attendees were from north america. which, considering the relative distances involved, is quite a high percentage. Nov 09 22:29:57 pb_: very true, would have liked to have gone this year but personal life got in the way (In UK). US, ASIA and EU Oeconfs ;-) Nov 09 22:36:18 ynezz, khem: Did you reach a conclusion over that problem? Nov 09 22:38:46 RP: yes, it was problem on my side, fixed now Nov 09 22:38:54 RP: thoughts on copying bb.parse.BBHandler.vars_from_file into base.bbclass and deprecating the one in bitbake, for the PN/PV/PR stuff? there's no reason for that to be in the core Nov 09 22:39:16 kergoth: I'd prefer bb.utils Nov 09 22:39:32 kergoth: I don't like having a large library of functions in base.bbclass Nov 09 22:39:38 why? bitbake doesn't care about the filename format Nov 09 22:39:45 nor should it Nov 09 22:40:07 RP: I think its prolly not related to your changes Nov 09 22:40:13 well, moving to bb.utils would be better than nothing, poking into a specific parser is disgusting Nov 09 22:40:32 kergoth: No, but it seems crazy to put a whole load of functions into parsed metadata too - I'd prefer base.bbclass to be a clean set of data more about tasks Nov 09 22:40:46 btw is sideefect of this change http://cgit.openembedded.net/cgit.cgi/openembedded/commit/?h=shr/merge&id=68cdc74da781c0755b26bb336dd666cb58104dce that gps-utils now depends on libgps instead of libgps17 and it creates libgps without 17? Nov 09 22:40:54 kergoth: but yes, poking into the parser is crazy too Nov 09 22:41:59 kergoth: Personally I'd prefer to dump a load of the base_* functions into utils.py, or adjust things so we can extend the bb library of functions with one from OE Nov 09 22:41:59 well, where do you draw the line? when do you consider a utility sufficiently generic that it belongs in bitbake rather than specific to OE? Nov 09 22:42:26 ynezz I reverted the geode stuff Nov 09 22:42:39 that would be nice. I'd love to be able to tell bitbake to import a .py from a collection/overlay Nov 09 22:42:54 would be lovely for defining classes to be used by tasks Nov 09 22:43:39 kergoth: Perhaps as we have the bb.* namespace, oe should add an oe. namespace Nov 09 22:43:50 and then we could have an oe.utils Nov 09 22:43:52 rp yeah Nov 09 22:44:28 RP: that's a good idea. create an oe module in globals or builtins and import specified python scripts into it Nov 09 22:44:49 I'd be happy with that as long as it doesn't get abused Nov 09 22:45:34 in what way? Nov 09 22:45:40 kergoth: and I'd like to stop abusing __builtins__ and start using some custom context Nov 09 22:45:43 agreed Nov 09 22:46:01 woglinde: ok, thanks Nov 09 22:46:06 kergoth: I'd just not like to see things in oe. that should be bb. but people were lazy Nov 09 22:46:51 again, where do you draw the line? Nov 09 22:47:04 kergoth: difficult but you know what I mean ;-) Nov 09 22:47:11 this is the problem with the current split, it's unclear, since use of it by anyone else is entirely theoretical Nov 09 22:48:10 I guess the question is that if someone added a new set of tasks using the bb syntax, how likely would they be to need these functions? Nov 09 22:48:42 i.e. are they specific to the way the code in oe works or more inherent to bitbake itself Nov 09 23:01:23 that seems like a reasonable rule of thumb Nov 09 23:05:24 hrw|gone: one more fix pushed, poky-image-sato works ;-) Nov 09 23:15:12 hmm, nobody's noticed I broke locale generation in OE yet? :) Nov 09 23:15:27 nobody besides koen uses locale Nov 09 23:17:24 jo raster Nov 09 23:19:34 woglinde: yo Nov 09 23:30:21 hm last naruto episode before bed Nov 09 23:31:23 ahem..there was a discussion OpenEmbedded vs. openembedded. Pls note the files in contrib/artwork are all lowercase Nov 09 23:38:04 03Michael 'Mickey' Lauer  07org.openembedded.dev * rce3802e207 10openembedded.git/ (7 files in 2 dirs): vala: add unofficial point release 0.7.8.1 Nov 09 23:44:18 RP: ant__ klibc does not build for x86 Nov 09 23:44:29 ant__: ups was not meant for RP Nov 09 23:44:49 khem: my bad? Nov 09 23:45:33 http://khem.pastey.net/128012 Nov 09 23:45:57 ant__: dont know so far I did not know that it was pulled in into console-image Nov 09 23:48:24 could be this 6f36bd6e28bce49b72e5a3cff77e0c74b8f65e41 Nov 09 23:50:09 most likely seems so Nov 09 23:54:27 khem: there was some discussion on klibc ML Nov 09 23:56:59 khem: http://www.zytor.com/pipermail/klibc/2009-September/thread.html Nov 10 00:02:27 03Richard Purdie  07org.openembedded.dev * re6ee9ab2e4 10openembedded.git/recipes/glibc/glibc-package.bbclass: Nov 10 00:02:27 glibc-package.bbclass: Operate on PKG, not D after package.bbclass updates Nov 10 00:02:27 Signed-off-by: Richard Purdie Nov 10 00:44:24 bitbake 1.8.16 is now required? Nov 10 00:44:33 debian still has only 1.8.12 Nov 10 00:44:47 Is Jan Lübbe around? Nov 10 00:45:24 Laibsch: yes Nov 10 00:45:43 Hi Florian Nov 10 00:45:47 (refers to the first question) Nov 10 00:45:52 hi Laibsch Nov 10 00:46:14 what requires it and when are we going to see a new package in Debian? Nov 10 00:51:24 Laibsch: Latest .dev changes. RP sent an OEDEM summary mail to the list which has a quite good summary. Nov 10 00:51:43 I see Nov 10 00:51:49 * Laibsch looking through ml Nov 10 00:52:59 well, bitbake still doesn't need to be 'installed'..just unpacked, isnt? Nov 10 00:53:43 once the instructions were: stay away from packaged versions Nov 10 00:54:19 ant__: yes you can just add it to PATH and feel happy Nov 10 00:54:49 Yes, I think you can still use the svn version Nov 10 00:54:57 I do the same on my build machines because some trees need different versions. Nov 10 00:55:02 I just really want software on my system to be packaged. period Nov 10 01:00:56 Laibsch: the majority of people out there too, probably ;) Nov 10 01:01:45 apt-het clean Nov 10 01:01:50 woops! Nov 10 01:02:01 wrong window :P Nov 10 01:02:59 who are the six members running for positions on the TSC? Nov 10 01:10:46 iirc its in the wiki Nov 10 01:16:45 maybe Nov 10 01:16:56 at least I couldn't find it on the obvious pages Nov 10 02:30:09 I'm having some issues building an omap3-console-image, would this be the proper place to ask about it? **** ENDING LOGGING AT Tue Nov 10 02:59:57 2009