**** BEGIN LOGGING AT Sun Sep 02 02:59:57 2007 Sep 02 04:13:17 is the webkit-gtk blowing up in build for anyone as well? Sep 02 06:01:52 03mickeyl 07org.oe.dev * rb9087aed... 10/ (4 files in 4 dirs): netbase 4.21: add support for fic-gta02, restore formatting, bump PR Sep 02 06:01:58 03rpurdie 07org.oe.dev * r523a58e7... 10/ (1 classes/sdk.bbclass): sdk.bbclass: Remove uneeded code Sep 02 06:02:03 03rpurdie 07org.oe.dev * rde0f7d33... 10/ (4 files in 2 dirs): package_x.bbclass: Sync with poky preparing for individual package writing tasks Sep 02 06:02:08 03rpurdie 07org.oe.dev * r2fa0d3ae... 10/ (1 packages/makedevs/makedevs_1.0.0.bb): makedevs: Bump PR to fix autobuilders which didn't pickup the previous changes Sep 02 06:02:15 03rpurdie 07org.oe.dev * r3cb55f38... 10/ (1 classes/package_ipk.bbclass): package_ipk.bbclass: Fix whitespace Sep 02 07:14:53 hi koen ! Sep 02 07:17:33 ciao! Sep 02 07:22:17 hahah :) goete morgen! Sep 02 07:26:18 i got an erroro with eglibc, there's something wrong in package management, it name eglibc-intermediate as eglibc-intermediate-2.6+svnrNOT IMPLEMENTED-r0 ... so it fail to fetch anr/or decompress ... Sep 02 07:30:23 the last one i was able to build was 2007-08-29 ... after the date became 'NOT IMPLEMENTED' ... Sep 02 07:52:57 * * OE Bug 2910 has been created by autobuild(AT)openembedded.org Sep 02 07:52:59 * * gcc-cross-initial-4.1.2-r6-do_unpack Sep 02 07:53:01 * * http://bugs.openembedded.org/show_bug.cgi?id=2910 Sep 02 07:53:57 * * OE Bug 2908 has been created by autobuild(AT)openembedded.org Sep 02 07:53:59 * * libtool-native-1.5.10-r6-do_compile Sep 02 07:54:01 * * http://bugs.openembedded.org/show_bug.cgi?id=2908 Sep 02 07:54:57 * * OE Bug 2909 has been created by autobuild(AT)openembedded.org Sep 02 07:54:59 * * glibc-intermediate-2.5-r6-do_unpack Sep 02 07:55:01 * * http://bugs.openembedded.org/show_bug.cgi?id=2909 Sep 02 08:01:15 I got an error with eglibc, there's something wrong in package management, it name eglibc-intermediate as eglibc-intermediate-2.6+svnrNOT IMPLEMENTED-r0 ... so it fail to fetch anr/or decompress ... the last one i was able to build was 2007-08-29 ... after the date became 'NOT IMPLEMENTED' ... Sep 02 09:19:57 * * OE Bug 2908 has been RESOLVED (INVALID) by Sep 02 09:19:59 * *  libtool-native-1.5.10-r6-do_compile Sep 02 09:20:01 * * http://bugs.openembedded.org/show_bug.cgi?id=2908 Sep 02 09:20:57 * * OE Bug 2909 has been RESOLVED (INVALID) by Sep 02 09:20:59 * *  glibc-intermediate-2.5-r6-do_unpack Sep 02 09:21:01 * * http://bugs.openembedded.org/show_bug.cgi?id=2909 Sep 02 09:21:57 * * OE Bug 2910 has been RESOLVED (INVALID) by Sep 02 09:21:59 * *  gcc-cross-initial-4.1.2-r6-do_unpack Sep 02 09:22:01 * * http://bugs.openembedded.org/show_bug.cgi?id=2910 Sep 02 09:30:52 morning all Sep 02 09:35:40 hey RP Sep 02 09:36:18 koen: How did our patchsets compare in the end? Sep 02 09:36:41 lots of whitespace Sep 02 09:36:42 I get the feeling there are a ton more changes that need syncing :/ Sep 02 09:37:02 koen: I've noticed you keep breaking whitespace in files :-( Sep 02 09:37:04 I merged in my changes to get rootfs_deb working Sep 02 09:37:25 koen: Check the whitespace on that change ;-) Sep 02 09:37:28 I suspect my vi is horrendously misconfigured Sep 02 09:40:38 Should I update gnome-common from 2.4.0 to 2.18.0 ? Sep 02 09:40:52 yes Sep 02 09:41:08 and also sync the changes ross made to various recipes for that :P) Sep 02 09:41:14 s/P// Sep 02 09:41:27 The problem is I'm not sure that is backwards compatible Sep 02 09:42:03 in what way? Sep 02 09:43:25 Part of the changes to other recipes was to remove the gnome2-macros directory from gnome.bbclass Sep 02 09:43:52 gnome in OE needs to get cleaned anyway Sep 02 09:44:12 every 6 months I try to do it Sep 02 09:44:58 I'm not sure if we want to merge the poky 2.19 stuff or wait to 2.20 and then merge the poky stuff + updates to 2.20 Sep 02 09:45:36 I'll sort out gnome-common but I'll leave that choice to you :) Sep 02 09:46:01 I discovered the 'compare dir' in meld :) Sep 02 09:46:41 I have scripts that diff poky against OE... Sep 02 09:46:46 I should auotmate it more... Sep 02 09:48:33 RP: do you have any thoughts on how we can make OE be more aware of rm_work? Sep 02 09:48:48 koen: More aware in what way? Sep 02 09:49:26 if I do e.g bitbake virtual/kernel -c compile -f it will start with unpack, patch, etc Sep 02 09:49:39 (after completing a kernel build) Sep 02 09:50:22 Thats only to be expected... Sep 02 09:50:39 right now it won't do that Sep 02 09:51:08 OE doesn't know the workdir vanished because all the stamps are still there Sep 02 09:52:40 I guess the solution would be to have rm_work very carefully remove certain stamps. That won't work though since bitbake's build integrity tests will notice they're missing and always rebuild everything :/ Sep 02 09:53:06 hence my "any thoughts on how we can make OE be more aware of rm_work?" Sep 02 09:53:54 and would it be usefull to have a "workdir layout version" like bitbake has a cache version? Sep 02 09:54:09 so OE can tell people to wipe tmp (or do that automagically) Sep 02 09:54:55 The only thing I can think of would be to hack bitbakes tasks dependency checking to ignore certain issues under certain circumstances, or teach every task about checking the integrity of workdir before proceeding (make them aware of rm_work) Sep 02 09:55:26 I have wondered about workdir "PR" bumps... Sep 02 09:55:33 my remote builder is running short on diskspace for multimachine builds :( Sep 02 09:55:35 hey mickeyl Sep 02 09:55:41 morning mickeyl Sep 02 09:55:45 good morning boys and girls Sep 02 09:55:47 hi koen Sep 02 09:55:48 hi rp Sep 02 09:55:58 mickeyl: I found your makedevs problem Sep 02 09:56:03 RP: cool. what was it? Sep 02 09:56:13 mickey|mdd: Lack of a PR bump on makedevs Sep 02 09:56:21 heh Sep 02 09:57:34 hello Sep 02 09:59:23 RP: or a "rm_work" stamp? Sep 02 10:00:22 koen: I don't see how that would help :/ Sep 02 10:07:02 how can I create symlinks in do_install ? I need something like "ln -s ../init.d/S95powerd /etc/init.d/powerd" here is my .bb file http://pastebin.ca/678740 Sep 02 10:07:22 mr_nice: inherit update-rc.d Sep 02 10:08:04 mr_nice: http://www.openembedded.org/user-manual&dpage=update-rc-d_class Sep 02 10:08:30 koen: thx Sep 02 10:22:28 03rpurdie 07org.oe.dev * r06814600... 10/ (8 files in 5 dirs): gnome-common: 2.4.0 -> 2.18.0 + associated cleanup (from Poky) Sep 02 10:26:36 03rpurdie 07org.oe.dev * r17828248... 10/ (7 files in 6 dirs): Fix PKG_CONFIG_PATH references which should be PKG_CONFIG_DIR Sep 02 10:28:41 my current autobuilder script: http://rafb.net/p/EnwKb749.html Sep 02 10:45:03 * koen hears crashing sounds and cursing coming from the attic Sep 02 10:52:40 who looks after spectrum-fw ? Sep 02 10:59:19 * koen discovers batman lego Sep 02 10:59:44 rwhitby: my guess is hrw Sep 02 11:01:32 koen: my son has the bat dragster with catwoman on the motorbike Sep 02 11:01:44 nice! Sep 02 11:13:33 koen: did you only delete the ep93xx crunch things or did you investigate some time in why it breaks building ? Sep 02 11:17:14 stefan|mdd: did you have fun yesterday (http://www.spiegel.de/politik/ausland/0,1518,503388,00.html)? Sep 02 11:18:06 grma: only the 4.1.2 patches got removed Sep 02 11:18:11 the 4.2.1 patches are still there Sep 02 11:18:30 zecke: oh, not at Aalborg. Sep 02 11:19:00 zecke: But nice food and many blonde danish girls to look at. :) Sep 02 11:19:00 so building with 4.2.1 works ? Sep 02 11:19:12 mmmh blonde danish girls Sep 02 11:19:31 haha Sep 02 11:19:40 grma: building regular arm code works, haven't tried crunch yet Sep 02 11:19:56 ok, thanks Sep 02 11:21:21 * stefan|mdd is happy here are no girls to disatract. :) Sep 02 11:23:12 rwhitby: I've hacked on spectrum-fw in the past Sep 02 11:24:55 RP: the problem is that each time it builds, it goes and fetches some file by ftp into the workdir, instead of into DL_DIR. so next time it builds (due to rm_work), it fetches it all over again Sep 02 11:25:17 (there's a script in there which does the fetch, when it could just as easily be done in SRC_URI) Sep 02 11:26:47 rwhitby: Feel free to change that to use our standard fetchers Sep 02 11:34:31 RP: normal SRC_URI unpacks the file into ${WORKDIR}, right? how do I get it just to copy the file from DL_DIR to WORKDIR instead of unpacking it (cause the script which gets run in do_configure unpacks it in a special way) ? Sep 02 11:35:29 Hmm this '&' in the filename is causing problems too. Sep 02 11:36:33 mickey|mdd++ Sep 02 11:38:11 mickey|mdd++ Sep 02 11:38:23 rwhitby: you could try overriding do_unpack Sep 02 11:39:09 rwhitby: Override do_unpack. Koen beat me to it :) Sep 02 11:39:26 thx Sep 02 11:39:50 do_unpack is python iirc Sep 02 11:40:46 zecke: He is at a tutorial atm and his notebook is here Sep 02 11:41:01 hehe Sep 02 11:41:11 hopefully with a xlock ;) Sep 02 11:41:30 mickeyl runs neod on his laptop :) Sep 02 11:43:40 zecke: no :) Sep 02 11:45:25 Hmm - this filename with '&' in it is just too hard. do_fetch doesn't realise that it's just finished downloading the file and tries to download it again from the source mirror. Sep 02 11:45:30 * rwhitby gives up Sep 02 11:46:07 sounds like a bug in the fetcher and url escaping Sep 02 11:47:30 the test case is to add "ftp://symstore.longisland.com/Symstore/services_download/wirless_prod/MC\&DriverOnlyInstallers.zip to SRC_URI in packages/orinoco/spectrum-fw.bb Sep 02 11:48:20 the single \ before the & seems to cause the file to be fetched, but not successfully recorded as being fetched Sep 02 11:48:32 Someone needs to fix bitbake :/ Sep 02 11:49:05 * rwhitby wonders whether this is why spectrum-fw is the way it is, instead of using SRC_URI ;-) Sep 02 11:49:34 now you know why I have project doing 'wiki releases' Sep 02 11:49:35 rwhitby: I was just thinking that Sep 02 11:49:42 s/have/hate/ Sep 02 11:49:49 ~lart opensync for wiki releases Sep 02 11:49:50 * ibot whips out a hot clue gun and makes sure that opensync is stuck to the floor for wiki releases Sep 02 11:50:26 http://www.opensync.org/attachment/wiki/download/libopensync-0.31.tar.bz2?rev=&format=raw Sep 02 11:51:30 RP: you agree with committing the package_write changes on tuesday? Sep 02 11:51:44 koen: Its fine with me Sep 02 11:53:09 I wonder how many people will be able to decode the parse errors to "berlios down. AGAIN" next week Sep 02 12:00:21 03freyther 07org.oe.dev * r4234d014... 10/ (3 files in 3 dirs): Sep 02 12:00:21 qmake2,uicmoc4: Fix the LICENSE to only be GPL. QtopiaCore/QtE never was QPL licensed Sep 02 12:00:21 Also adjust indention of the SRC_URI Sep 02 12:00:27 03freyther 07org.oe.dev * r8729d47a... 10/ (1 conf/checksums.ini): conf/checksums.ini: Add md5,sha256 sums for QtopiaCore 4.3.1 Sep 02 12:19:49 hello, does anyone know of a shell tool that can highlight 'watched' words in the output in colour? Sep 02 12:39:45 likewise: sed? Sep 02 12:40:32 i.e. sed 's/(word)/\033[31m\1\033[0m/' Sep 02 12:43:06 NAbyss_: thanks looking into that Sep 02 13:27:09 hrw|gone: You asked me about SDHC on pxa. I just had a first try with a micrSDHC card on my ezx phone, but it failed. Kernel was 2.6.21 which have the SDHC patch in mainline. Sep 02 13:27:15 crap Sep 02 13:27:21 I checked in a SRCREV Sep 02 13:27:24 *again* Sep 02 13:27:57 hrw|gone: Perhaps I give it another in the upcoming week Sep 02 13:28:15 stefan|mdd: 2.6.22 has a batch of sd fixed iirc Sep 02 13:28:30 fixes* Sep 02 13:28:34 koen: And a bunch of break sd patches. :) Sep 02 13:28:43 that as well Sep 02 13:28:46 koen: I think we wait for .23 Sep 02 13:28:54 did pierre receive his devboard yet? Sep 02 13:29:04 koen: mickey|mdd also pestered me about thinking having -ezx on to of -rp Sep 02 13:29:28 koen: need to talk with RP about it, but not today. Sep 02 13:29:40 stefan|mdd: as long as it uses linux.inc :) Sep 02 13:29:58 and uses the ezx-patch script ao2 wrote Sep 02 13:30:07 koen: yeah, many stuff to clean up Sep 02 13:30:43 koen: You mean the one put it into SRC_URI? I'm not really happy with this. What was the reason for it again? Sep 02 13:31:05 getting patches from svn just breaks too often Sep 02 13:31:52 But pulling a whole patchset in looks really insane to me Sep 02 13:32:05 why? Sep 02 13:32:27 patches in OE are patches who can't/shouldn't/whatever be upstream Sep 02 13:32:28 Because that is what quilt is used for Sep 02 13:32:36 OE can just use pinpointed svn dates Sep 02 13:32:51 dragging patches in to OE metadata is unnecessary pollution given we have control about upstream Sep 02 13:33:17 heh Sep 02 13:33:28 and it didn't break a zillion times in the past months Sep 02 13:33:35 o wait, it did! Sep 02 13:33:47 yes Sep 02 13:33:51 because it was NOT pinpointed Sep 02 13:35:51 and running quilt by hand is better as patch.bbclass exactly why? Sep 02 13:38:09 *shrug* that's a different issue Sep 02 13:38:09 koen: Cares about the order and does not write it all in the bb? Sep 02 13:38:31 if we can rewrite patch.bbclass to use our series files, lets go for it Sep 02 13:38:38 s/our/downloaded/ Sep 02 13:40:17 03koen 07org.oe.dev * rfa59fdc5... 10/ (4 files in 2 dirs): tinymail: update to 0.0.1 prerelease, needs more work with packaging Sep 02 13:40:31 packages/linux/linux-ezx-2.6.21/update_patches.sh honours the series file Sep 02 13:41:24 03koen 07org.oe.dev * rd73aff45... 10/ (1 packages/tinymail/libtinymail_svn.bb): tinymail svn: it's not tuesday yet, so s/SRCREV/SRCDATE/ Sep 02 13:42:42 mickey|mdd: Have you any opinions on http://www.rpsys.net/openzaurus/temp/timestamp.patch ? Sep 02 13:43:05 mickey|mdd: The idea being to make this work for devices like the PXA which loose the RTC over halt Sep 02 13:44:02 TIMESTAMP=`awk '{ print substr($0,9,4) substr($0,1,4);}' /etc/timestamp` Sep 02 13:44:04 Good morning> I'd like to remove qemu from OE so it will compile on an x86_box with gcc-4.x. How can I remove it? (The RAID card on the i386 machin is failing, and I can't get through a build Sep 02 13:44:06 ). Sep 02 13:44:11 saves an invocation of busybox Sep 02 13:44:19 RP: looks good to me, lets give it a go Sep 02 13:45:33 koen: true, I'll fix that :) Sep 02 13:48:13 DukeOfURL: you can disable locale generation in local.conf "ENABLE_BINARY_LOCALE_GENERATION = "0" Sep 02 13:48:34 thank you Sep 02 13:48:59 DukeOfURL: apt-get install gcc-3.4 Sep 02 13:53:37 mickey|mdd: my biggest problem with pulling in complete dirs with patches is that you can't pick and choose updates Sep 02 13:53:42 mickey|mdd: if a single commits fixes one patch and breaks another, OE is screwed Sep 02 13:55:53 koen: You could check each patch out of SVN separately ;-) Sep 02 13:56:02 * RP hides Sep 02 13:56:21 only with very recent svn releases :) Sep 02 13:57:02 but that would be a good compromise Sep 02 13:57:26 koen: That's a valid problem. But still it brings the problem in that the stuff between OE and e.g. ezx tree drift away. I need to think about this a bit longer. Sep 02 13:57:45 koen: It also plays together with eventually base on -rp Sep 02 13:58:22 RP: You are still working on -rp all the time to have it in sync with mainline? Sep 02 13:58:25 stefan|mdd: the other side of that is OE people yelling at every commit on ezx-svn that breaks something Sep 02 13:58:57 a small drift isn't always a bad thing Sep 02 14:01:12 if the problem is developer laziness I volunteer to update the patches when needed Sep 02 14:01:12 koen: It's just that it costs time to take care about it, being not sure if it builds what I just checked in, etc.. Sep 02 14:01:34 see above :) Sep 02 14:01:51 koen: yeah, I know you would take care. Sep 02 14:02:27 koen: As I said. Let me think about it. I like hacking today not thinking. :) Sep 02 14:06:25 new zaurus material: http://www.sharp-world.com/corporate/news/070831.html Sep 02 14:13:36 anyway, my offer to update such patchsets stands Sep 02 14:14:42 so updating will at most cost you an email :) Sep 02 14:14:46 koen: I think about it on my 10hours train trip tomorrow, promised Sep 02 14:14:54 thanks Sep 02 14:18:02 mickey|mbb RP can you please have a look at oe bug 2598. RP I also had a look at zaurusd but I think it is oversized for just doing power off. SIMpad also does not use SW key events so I would have to massive rewrite the zaurusd. I also personaly don't linke to have a programm called after a button is pressed. so I don't know if it would be worth the efforts. Sep 02 14:18:16 oki Sep 02 14:18:22 thx Sep 02 14:19:26 mr_nice: I was more thinking of adding powerd to that. It was the hinge code I was pointing you at Sep 02 14:20:54 later all Sep 02 14:21:59 RP: yes I know. But I think it is oversized for it. I also want to blcok the event so that there are no conflicts with opie and it's own power off daemon Sep 02 14:22:39 mr_nice: One of the points in zaurusd was you didn't have to install any bits that aren't used Sep 02 14:22:47 mr_nice: Anyhow, it doesn't matter Sep 02 14:32:50 03rpurdie * r968 10/ (2 files in 2 dirs): BBHandler.py: Improve debug message Sep 02 14:46:14 03rpurdie 07org.oe.dev * r266638ec... 10/ (4 files in 3 dirs): Sep 02 14:46:14 initscipts: Improve RTC handling so /etc/timestamp is used if its timestamp is 1 Sep 02 14:46:14 day or more greater than the existing time. Also set the timestamp on Sep 02 14:46:14 halt/reboot so devices without a hardware RTC (or with a broken RTC like the Sep 02 14:46:14 PXA) can benefit. Sep 02 14:49:25 rwhitby: did you try URL escaping (%xx)? Sep 02 15:52:52 03rpurdie 07org.oe.dev * rb985ab87... 10/ (1 packages/zaurus-updater/zaurus-updater.bb): zaurus-updater: Fix task dependencies (no need to work around rm_work now Sep 02 15:52:57 03rpurdie 07org.oe.dev * r858dbc3e... 10/ (33 files in 4 dirs): linux-rp: Sync with Poky (adding 2.6.22 in the process) Sep 02 15:56:37 RP: you still get to sync linux-rp once and then w/ mainline? Sep 02 15:56:43 RP: how's pxa status nowadays? Sep 02 15:56:56 mickey|mdd: I don't have as much time as I'd like for that :-( Sep 02 15:57:26 mickey|mdd: PXA status is good but others in mainline are starting to do some major cleanups with a view to getting pxa3xx support merged in Sep 02 15:57:29 koen, Sorry to keep bothering you regarding the N800.. Would you recommend it, do you think it is worth getting, any major complaints? Sep 02 15:57:50 RP: i see. in consequence there might be some breakage ahead Sep 02 15:58:15 mickey|mdd: I don't think it will be too bad tbh but its something to be aware of Sep 02 15:58:26 mickey|mdd: Any particular reason? Sep 02 15:58:49 RP: yes. i want to base the exz kernel on linux-rp Sep 02 15:59:05 since i view this to be one of the best pxa kernels around Sep 02 15:59:07 :) Sep 02 15:59:18 :) Sep 02 15:59:20 mickey|mdd, 2.6.20 kernel seems very stable for the PXA but I (and others I have spoken with, cbrake for one) have had several stablility issues with kernels after 2.6.20 and the PXA. Sep 02 15:59:28 svolpe_gerrath: really? darn Sep 02 15:59:55 mickey|mdd: Samuel@OH is looking at sorting out htcuniversal support in linux-rp too and that then should be going mainline wards Sep 02 16:00:10 RP: oh good! Sep 02 16:01:00 svolpe_gerrath: FWIW, the Zaurus seems pretty solid with 2.6.22 and 2.6.23.rc4 Sep 02 16:01:12 mickey|mdd, I hope to do more testing on it later but for now I'm sticking with the 2.6.20.14, it works well. Sep 02 16:01:44 mickey|mdd, I had issues with jffs2 core dumping and I know cbrake had realtime threading issues. Sep 02 16:02:09 ok, hmmm Sep 02 16:02:49 mickey|mdd, again I did not have time to really shake out what the issue was so it could have been at my end but since 2.6.20.14 worked great for me I just patched what I needed from the 2.6.22/23 kernel into it. Sep 02 16:04:03 svolpe_gerrath: the n800 is a nice piece of hardware, but as I said, the software sucks Sep 02 16:04:05 * RP wishes people did have time to track these things down since if nobody does mainline just bitrots# Sep 02 16:05:01 RP, thanks.. I hope to go back to the 2.6.22/2.6.23 and find the issue. Sep 02 16:05:25 RP I know I agree and I will get back to it unfortunetly those darn schedules tend to get in the way :-/ Sep 02 16:05:43 svolpe_gerrath: I know the feeling all too well, believe me :-( Sep 02 16:06:33 RP, on a good note I did take a snap-shot of the source tree and archived it with some notes so I can go back later and recreate the issue :-) Sep 02 16:06:47 svolpe_gerrath: sounds good :) Sep 02 16:08:46 koen, I'm assuming there are no real issues with re-flashing it with a new rootfs (OE).. Sep 02 16:09:13 the wifi driver is troublesome, but it should work Sep 02 16:09:25 maybe some extra patches from poky are needed Sep 02 16:09:45 koen, what is the bootloader like to work with? Sep 02 16:11:11 koen, is it enough horse power to play videos, I know with the default code that the video is jerky but I was assuming that was do to the software and not hardware. Sep 02 16:12:17 mickey|mdd RP is the power off daemon ok for you? koen told me to ask you before it would be accepted. Sep 02 16:12:41 i need time to review that Sep 02 16:12:46 mr_nice: I haven't had a chance to look at it yet. Sep 02 16:12:54 ok Sep 02 16:13:49 please give us a bit Sep 02 16:14:20 yes, no problem Sep 02 16:17:32 I really wish poky hadn't diverged this much from OE, its giving me some real headaches... Sep 02 16:18:07 486kb of diff still remaining and in theory there shouldn't be much at all :/ Sep 02 16:22:54 RP: some diffs is being caused by poky being only poky/gtk/glibc, which is no good for .dev Sep 02 16:23:42 koen: I'm quite happy to support uclibc in poky, its just a question of merging back those fixes (now the underlying bitbake issue is resolved) Sep 02 16:24:45 I've added uclinc builds to the autobuilder, so hopefully OE support will keep in shape Sep 02 16:25:44 good morning! Sep 02 16:25:52 hey sakoman Sep 02 16:26:54 03koen 07org.oe.dev * r237851b0... 10/ (1 conf/distro/angstrom-2007.1.conf): angstrom: make distroversion consistent between 2007.x and 2008.1 Sep 02 16:27:02 I'm going to being using OE as the build system for several products. I want to use a unique feed URI per product for updates/upgrades. What is the "OE approved" way to override the AGNSTROM_URI? Sep 02 16:29:07 I guess I really meant angstrom FEED_URIS :-) Sep 02 16:29:52 ehm Sep 02 16:30:12 I guess it's really time for me to write a recipe for ipkg feeds :) Sep 02 16:30:49 I didn't mean to create more work for you :-) Sep 02 16:31:13 It *is* one of those things that every commercial user is going to need though Sep 02 16:31:28 heh Sep 02 16:31:42 up to now 'commerical' users where quite monolithic Sep 02 16:31:54 upgrade package -> reflash complete devince Sep 02 16:31:57 monolithic? Sep 02 16:32:01 Ah! Sep 02 16:32:08 * koen learns to type Sep 02 16:32:16 I agree it's needed :) Sep 02 16:32:35 even nokia doesn't get packagemanagement Sep 02 16:32:50 I see that monolithic thing changing now Sep 02 16:32:52 they use apt-get, but updates can only be done via reflashes Sep 02 16:33:32 I'll be using it on my stuff, and some of my consulting clients are also being influenced to do the same Sep 02 16:34:27 Requires major whack on the side of the head to get them to change their thinking Sep 02 16:35:36 People just get stuck in "that's the way we've always done it" Sep 02 16:36:04 indeed Sep 02 16:36:37 it'd be sooo nice if nokia opened the n770/800 for custom imagemaking Sep 02 16:39:07 evening Sep 02 16:42:03 someone runs openzaurus? Sep 02 16:42:13 Koen, one last N800 question, I noticed it has graphics acceleration on the processor, does the software take advantage of that? does it really make much of a difference? Sep 02 16:42:19 someone runs openzaurus? Sep 02 16:42:36 svolpe_gerrath: no driver for 3d and 2d goes through the dsp Sep 02 16:43:01 (this is where intel's mid is going to kill nokia if they're not careful... mid will have free 3d) Sep 02 16:43:32 ^ and of course more open to alt. images Sep 02 16:43:50 koen, so basically no real acceleration, ok.. that helps compare since other units I'm looking at have no acceleration as well. Sep 02 16:45:57 * * OE Bug 2911 has been created by autobuild(AT)openembedded.org Sep 02 16:45:59 * * angstrom-feed-configs-1.0-0-do_compile Sep 02 16:46:01 * * http://bugs.openembedded.org/show_bug.cgi?id=2911 Sep 02 16:47:57 * * OE Bug 2912 has been created by autobuild(AT)openembedded.org Sep 02 16:47:59 * * angstrom-feed-configs-1.0-0-do_install Sep 02 16:48:01 * * http://bugs.openembedded.org/show_bug.cgi?id=2912 Sep 02 16:51:57 * * OE Bug 2911 has been RESOLVED (FIXED) by Sep 02 16:51:59 * *  angstrom-feed-configs-1.0-0-do_compile Sep 02 16:52:01 * * http://bugs.openembedded.org/show_bug.cgi?id=2911 Sep 02 16:53:57 * * OE Bug 2912 has been RESOLVED (FIXED) by Sep 02 16:53:59 * *  angstrom-feed-configs-1.0-0-do_install Sep 02 16:54:01 * * http://bugs.openembedded.org/show_bug.cgi?id=2912 Sep 02 16:54:11 svolpe_gerrath: N800 uses mobile graphics chip from epson (hwa745), Nokia 770 uses slightly older chip hwa742, but it has full documentation available: http://www.erd.epson.com/index.php?option=com_docman&task=cat_view&gid=36&Itemid=40 Sep 02 16:54:21 03koen 07org.oe.dev * r590d2793... 10/ (1 conf/distro/include/angstrom.inc): angstrom: make ANGSTROM_URI overridable Sep 02 16:54:27 03koen 07org.oe.dev * r0a5045ef... 10/ (1 packages/angstrom/angstrom-feed-configs.bb): Sep 02 16:54:27 angstrom-feed-configs: a recipe that add config files for feeds Sep 02 16:54:27 * this will replace the feeds in the distro.conf Sep 02 16:54:27 * Derivative distro can provide their own version by overriding ANGSTROM_URI and FEED_BASEPATH Sep 02 16:54:28 * WIP, suggestions welcome Sep 02 16:54:47 svolpe_gerrath: it is an external chip and not part of OMAP2420 Sep 02 16:54:49 sakoman: there you go Sep 02 16:55:04 sakoman: largely untested Sep 02 16:55:35 sakoman: probably needs a few CONFFILES statements to make sure user additions get saved and all Sep 02 16:55:36 ssvb - i thought there was a powervr core in the n800? is it not wired up? Sep 02 16:56:53 happycube: yes, powervr core is there as a payload (part of OMAP2420), but it does not have any software support Sep 02 16:58:19 happycube: I don't know how all this graphics hardware is chained there, maybe powervr is even disabled in hardware Sep 02 17:00:15 happycube: only nokia knows better, but right now N800 works with the external graphics controller from epson Sep 02 17:00:51 hi all Sep 02 17:03:03 koen: btw, 2D does not go through DSP on nokia internet tablets, it is just framebuffer mapped to both ARM and DSP cores there and Nokia 770 used DSP to decode video directly to the framebuffer, that's not the case for N800 anymore Sep 02 17:07:13 by the way, can anybody do me a favour and help to benchmark some code on XScale? http://www.oesf.org/forums/index.php?showtopic=22280&st=45 Sep 02 17:07:25 ssvb: right, I meant movie decoding Sep 02 17:07:54 I really want to know if it is good for XScale now before submitting this code as a patch to ffmpeg Sep 02 17:09:16 koen: thanks! I'll give it a try later today Sep 02 17:09:16 instructions are in post #51 Sep 02 17:26:49 RP: ping? Sep 02 17:27:03 zecke: pong Sep 02 17:27:23 RP: is any one of o-hand here to answer basic clutter questions? Sep 02 17:27:57 zecke: Everyone on my roster is showing offline atm I'm afraid Sep 02 17:28:35 okay, I will try tomorrow again Sep 02 17:29:39 zecke: Are we talking in or out of working hours? Sep 02 17:29:52 mine or theirs? Sep 02 17:30:18 zecke: I'm just thinking you might stand more chance when they're at work... Sep 02 17:30:22 zecke: You might stand a chance of catching ebassi on some of the gnome channels Sep 02 17:30:40 will try it tomorrow during working hours Sep 02 17:31:28 RP: I'm having troubles coming up with a sane situation for the feed-config recipe Sep 02 17:32:01 RP: basically I want RPROVIDES = "feed-configs" and PREF_PROVIDER_feed-configs = "angstrom-feed-configs" Sep 02 17:32:04 is that valid? Sep 02 17:33:06 koen: We simply can't do that. The package manager has no knowledge of PREF_PROVIDER Sep 02 17:33:41 the trick would be to shift the burden of keeping the feeds same to the distro maintainer Sep 02 17:35:06 koen: Its the same problem as with XSERVER. You'll have to declare a variable which you insert into the DEPENDS list with the right feed config package name Sep 02 17:35:37 that's a good idea Sep 02 17:36:01 03rpurdie 07org.oe.dev * r8548c765... 10/ (6 files in 2 dirs): site: Sync with Poky Sep 02 17:36:08 03rpurdie 07org.oe.dev * rfb528dee... 10/ (167 files in 3 dirs): conf/machine/include: Rename .conf to .inc files Sep 02 17:36:11 * koen should aslso write a doc for that Sep 02 17:36:15 03rpurdie 07org.oe.dev * rc34acf5b... 10/ (8 files in 3 dirs): conf/machine: Various tweaks from Poky Sep 02 17:36:39 koen: so with this scheme would I need a custom distro per product? I don't necessarily want to share feeds between products. Sep 02 17:36:53 Nicer to override in the product image recipe Sep 02 17:38:20 Perhaps I'm just not thnking clearly though :-) Sep 02 17:42:23 you can override it in the image recipe Sep 02 17:43:45 my goal is to cut down on derivative distros Sep 02 17:44:50 Same here, I'd like to stick to just one for all products and do all customization in machine conf and image recipes Sep 02 17:47:12 some things can't (or shouldn't) be done in a machine and image recipes, but we can always try :) Sep 02 18:00:09 03rpurdie 07org.oe.dev * rd52a54d5... 10/ (1 conf/machine/hx2000.conf): hx2000: Add ushbost to MACHINE_FEATURES, don't include kernels in images, add tarballs to output types Sep 02 19:01:12 RP: another ping Sep 02 19:56:13 re Sep 02 20:20:38 flo_lap: please take a look at oe bug 2598. I uploaded the power off daemon as a patch. Sep 02 20:20:55 mr_nice: oh cool Sep 02 21:26:07 03mickeyl 07org.oe.dev * r6e64a28e... 10/ (3 files in 3 dirs): tslib: remove e680 directory, not necessary anymore Sep 02 21:26:14 03mickeyl 07org.oe.dev * r3fc5a8a6... 10/ (3 files in 3 dirs): Sep 02 21:26:14 xserver-kdrive-common: add support for the Motorola Ezx Platform Sep 02 21:26:14 also repair formatting (please use spaces only) Sep 02 21:26:20 03mickeyl 07org.oe.dev * r1daa04a2... 10/ (2 files in 2 dirs): xserver-kdrive-common: bump PR Sep 02 21:26:26 03mickeyl 07org.oe.dev * r9ad2cb66... 10/ (1 conf/machine/include/motorola-ezx-base.inc): motorola-ezx-base.inc: all Motorola EZX phones have 240x320 screen Sep 02 21:26:32 03mickeyl 07org.oe.dev * raaceb234... 10/ (4 files in 3 dirs): fic-gta01|fic-gta02|ezx.conf: set MACHINE_DISPLAY_ORIENTATION and MACHINE_DISPLAY_PPI Sep 02 21:26:38 03mickeyl 07org.oe.dev * r3044a9fe... 10/ (1 packages/tasks/task-openmoko.bb): task-openmoko: pragmatic approach to choosing themes Sep 02 21:26:45 03mickeyl 07org.oe.dev * r761183d5... 10/ (1 packages/openmoko2/openmoko-theme-standard2-qvga_svn.bb): add openmoko-theme-standard2-qvga Sep 02 21:26:52 03mickeyl 07org.oe.dev * r922cbf08... 10/ (6 files in 3 dirs): openmoko-terminal2: build new terminal found in OM svn Sep 02 21:27:24 i need to customize default /etc/inittab in angstrom image for my machine, what't a correct way to do so? Sep 02 21:29:05 add a directory under initscripts/initscripts-1.0 named as your $MACHINE Sep 02 21:29:10 then put your custom inittab into Sep 02 21:29:12 bump PR Sep 02 21:29:15 boom, done Sep 02 21:29:45 ah, thanks Sep 02 21:55:05 ls -al Sep 02 21:55:09 whoops Sep 02 21:55:10 hi all :-) Sep 02 21:59:24 i've put my cutom inittab in packages/initscripts/initscripts-1.0/$MACHINE, bumped PR but it don't work Sep 02 21:59:41 i'm still getting default inittab Sep 02 22:01:42 ah, nm it works but somethings appends some additional line into it somewhere Sep 02 22:03:14 anyway I wonder how this package/$MACHINE/file specific stuff works, since I can't find anything about it in initscripts.bb file Sep 02 22:21:36 hmm Sep 02 22:21:44 | /dev/initctl is not a fifo Sep 02 22:21:48 and then a Sep 02 22:21:55 BusyBox: Usage: run-parts .... Sep 02 22:21:58 new errors Sep 02 22:22:02 during startup Sep 02 22:26:44 mickeyl: Angstrom? Sep 02 22:26:53 kind of Sep 02 22:27:26 mickeyl: same here. I just built a angstrom-console-image. I already wondered where I broke it. Sep 02 22:27:42 a Sep 02 22:27:43 h Sep 02 22:27:53 mickeyl: No init system was installed here. Sep 02 22:27:54 i guess these are artefacts from RP's makedev changes Sep 02 22:28:56 Did RP change makedevs? (I added dir support, but forgot to bump the PV. Shoot me.) Sep 02 22:29:07 i think so, yes Sep 02 22:29:33 + a lot of poky sync's Sep 02 22:29:40 maybe one of these was to blame Sep 02 22:31:25 My revision is 2007-08-31T18:16:21 Sep 02 22:37:49 mickeyl: my checkout has no RP changes to makedevs. Sep 02 22:38:23 hmm, strange. then something else broke Sep 02 22:38:43 with the new /dev creating Sep 02 22:39:20 mickeyl: is your /dev broken? I cannot see anything strange with my rootfs Sep 02 22:39:36 mickeyl: except I do not see an init of some sort. Sep 02 22:40:06 well, psplash doesn't start, complaining about "unable to open initial console" Sep 02 22:40:29 only until i reverted the makedevs stuff + the changes in the two bbclasses it worked again Sep 02 22:42:43 mickeyl: you work against head? Sep 02 22:42:51 yes Sep 02 22:43:22 http://userweb.kernel.org/~kzak/util-linux-ng/ Sep 02 22:43:25 mickeyl: I doubt its makedevs at fault tbh, nothing changed to do with the actual device creation Sep 02 22:44:52 ok, i found the run-parts problem Sep 02 22:44:59 it's in opentapi, ignore that for now Sep 02 22:45:08 but | /dev/initctl is not a fifo is scary Sep 02 22:45:31 i've hit it also Sep 02 22:45:46 just deleted /dev/initctl fixed it on next reboot Sep 02 22:46:01 With udev or a static /dev? Sep 02 22:46:32 static Sep 02 22:47:09 can't confirm it on device, but i have this file in /dev in rootfs dir Sep 02 22:57:45 It is the makedevs changes Sep 02 22:58:00 The old pipe creation code did nothing, the new code creates a dummy file Sep 02 23:08:53 good nite all Sep 02 23:55:13 ~gettingstarted Sep 02 23:55:14 gettingstarted is probably http://www.openembedded.org/wiki/GettingStarted Sep 02 23:59:57 03rpurdie 07org.oe.dev * re672e671... 10/ (3 files in 3 dirs): makedevs: Fix named pipe creation and set file modes correctly Sep 03 00:47:46 can we field boot problems for ixp425 here for oe/slugos ? Sep 03 00:49:01 any ixp4xx system builders ? Sep 03 01:29:50 how to remove a package not needed , gstreamer fails to compile now Sep 03 01:30:16 just cut it out of the bb task file ? Sep 03 01:54:38 root password for angstrom distro ? anyone ? thx Sep 03 02:07:40 oe lucid ? how to root password ? **** ENDING LOGGING AT Mon Sep 03 02:59:56 2007