**** BEGIN LOGGING AT Tue May 20 02:59:56 2008 May 20 04:01:18 hello there. Does anyone know any trick on how to build a rootfs without a kernel? I mean, I am trying to bitbake console-image here but since the kernel for palmld is broken, it wont build May 20 04:01:49 As I talk I am trying to adjust my build so it pull the linux-handheld kernel but I am afraid I will need to do something else May 20 04:02:43 Actually, I've already compiled the kernel by hand and I was jut wondering it there is a way to tell bitbake about it May 20 07:11:38 morning May 20 07:40:39 Laibsch: thanks for the mmap hint from yesterday. May 20 07:40:48 likewise: hey May 20 07:40:56 khem: hey khem May 20 07:41:06 likewise: I was reading through the messages May 20 07:41:15 so did you solve mmap thing May 20 07:41:23 it is not something I have seen May 20 07:41:24 likewise: you're very welcome May 20 07:41:28 khem: sorry for spamming you :-) yes, it was the sysctl setting that hit me May 20 07:41:49 heh you must have installed ubuntu 8.04 May 20 07:42:30 khem: cool, you can sniff my OS from OE issues :-) May 20 07:42:42 likewise: sort of :) May 20 07:43:03 hey I am trying to make eglibc separate out from glibc package dir May 20 07:43:21 khem: yes, I am interested in that, as well as configuring eglibc afterwards May 20 07:43:31 and I am seeing that something is pullin in glibc too when I chose eglibc to be my C library May 20 07:43:57 I tries to use -v and see who is pulling it in but I am not seeing it being pulled in for any reason May 20 07:44:17 any further ideas how can debug this .... May 20 07:44:58 actually I have made eglibc provide all the packages that glibc provides so it should not pull glibc ideally May 20 07:45:26 how can I find out which package is culprit here that's still needing something from glibc :) May 20 07:47:44 re May 20 07:48:54 khem: try bitbake -g and look at *depends.dot May 20 07:49:08 bonjour May 20 07:49:52 khem: I am starting off some experiments off the HEAD of .dev so if there is anything I can reproduce let me know. I'm currently targetting ixp4xx (arm xscale) and mpc8313e-rdb (powerpc) with HEAD. May 20 07:50:23 likewise: For that I need to commit the new package May 20 07:50:25 :) May 20 07:50:34 it will be broken though May 20 07:50:46 mtn diff . to pastebin is ok as well May 20 07:51:04 or tar up the glibc/eglibc dirs May 20 07:51:26 likewise: good idea I will do that May 20 07:51:39 likewise: meanwhile I think I am seeing the culprit now May 20 07:51:44 khem: :-) May 20 07:51:54 likewise: it's the glibc-dbg package May 20 08:21:10 likewise: are dbg package special May 20 08:25:03 hrw|gone: really gone? May 20 08:27:00 yes, really May 20 08:27:16 hrw|gone: heh :) May 20 08:27:31 hrw|gone: I have a dev version of oestats where I introduce the concept of a "package" May 20 08:27:32 w8 15m May 20 08:28:01 it will allow us to have 3 states for a package: Started / Failed / Succeeded May 20 08:28:16 the downside is it means trashing the current DB as I've changed the structure quite a bit May 20 08:47:17 hi May 20 08:47:21 morning etc May 20 08:47:35 moin May 20 08:58:18 hi zecke May 20 08:58:29 zecke: We should find a few minutes to talk about some patches May 20 08:58:37 zecke: and the git trial May 20 08:58:40 hehe May 20 08:59:04 zecke: Did you see my hacked up version of your environment one? May 20 08:59:12 yes, give me an hour to settle May 20 08:59:22 ok :) May 20 09:18:07 hi all May 20 09:21:05 someone know Sharma Nishant or Tsai Rex? May 20 09:23:54 no May 20 09:42:10 ~seen jeremy_laine May 20 09:42:11 jeremy_laine was last seen on IRC in channel #oe, 1h 13m 55s ago, saying: 'the downside is it means trashing the current DB as I've changed the structure quite a bit'. May 20 10:07:53 03rpurdie 07org.oe.dev * r055be031... 10/ (1 classes/gtk-icon-cache.bbclass): gtk-icon-cache.bbclass: Remove '-' characters from function names fixing issues with recent bitbakes (from poky) May 20 10:16:17 hrw: back May 20 10:16:45 hrw: ok to get proper a proper package status I need to understand what BB_DEFAULT_TASK represents May 20 10:17:55 jeremy_laine: Its the name of the task thats run when you run bitbake foo May 20 10:18:13 hrw: when I do "bitbake -c clean dropbear" for instance, BB_DEFAULT_TASK is still "build" May 20 10:18:51 hrw: so how do I record the fact that once we hit "do_clean", package "dropbear" is completed? May 20 10:20:21 thats interesting question May 20 10:21:57 jeremy_laine: the problem is that you do have to know which task is final target. "bitbake -ccompile dropbear" will have finaltask=compile, finalpackage=dropbear but BB_DEFAULT_TASK still 'build' because it can have to build gcc, libc and... May 20 10:22:02 thats how I see that May 20 10:22:30 hrw: ok, is there a variable for this "final task"? May 20 10:22:56 no idea :( May 20 10:23:15 jeremy_laine: No, there is not May 20 10:23:25 RP: ah :) May 20 10:23:43 brb May 20 10:24:04 As I've said previously, bitbake has no idea what constitues a package. It just has a list of tasks where A has to run before B May 20 10:24:53 RP: would not 'finaltask' be one of ends of taskstree? May 20 10:25:15 hrw: Not really May 20 10:25:16 addtask do_a_little_dance_now after finaltask May 20 10:32:15 RP: is self.configuration.cmd the right command? May 20 10:34:29 RP: is there any way I can access this from a bbclass? May 20 10:36:54 RP: bitbake question for you, I wrote this quick task http://git.openmoko.org/?p=openmoko.git;a=blob;f=classes/om-utils.bbclass;h=5c0464ef856f6d504488d522e3d602f971f65d8f;hb=HEAD to output the WORKDIRS we used for a build, so I can autoclean old ones May 20 10:37:11 RP: but it doesnt print the WORKDIR for stuff like git-native May 20 10:37:54 RP: can I get it to do this in a simple fashion or do I cheap and blacklist git-native and such from autocleaning May 20 10:41:19 RP: what would you think of http://pastebin.com/m769f084 ? May 20 10:49:02 jeremy_laine: Thats the task that will be run for the targets. It may or may not be the task thats run for anything built as an intermediate step May 20 10:49:50 jeremy_laine: Its not intended to be visible to class files May 20 10:50:35 XorA: Which bitbake are you using? May 20 10:50:48 I did make a change in bitbake fixing a bug in recrdeptask recently May 20 10:51:04 RP: svn, maybe not the latest latest May 20 10:51:31 jeremy_laine: Why do you need this? May 20 10:52:10 RP: following yesterday's discussion I have introduced a "Package" model in oestats (server side) so I can store the status for a given package May 20 10:52:36 RP: retesting May 20 10:52:42 jeremy_laine: It simply isn't going to work May 20 10:52:47 RP: so far, a "package" is marked as "Succeeded" unless it has a "Failed" task, which does not reflect the fact that the packing can be "in progress" May 20 10:53:02 RP: ok, not a problem, the DB needed the cleanup anyhow May 20 10:53:11 (for scaling / pagination) May 20 10:54:12 jeremy_laine: The problem is that the task that is the end of the chain for a given package varies. In some cases it will be populate_staging, in others it may be packaged_stage, it others, package_write May 20 10:54:30 hrw: is it OK with you if I trash the DB or did you have some builds you wanted to examine? May 20 10:54:51 RP: ok, I'd rather do nothing than add something unreliable on this point May 20 10:54:54 so do_build might be the end of package bar but it only requires git-native to complete to populate_stagging May 20 10:55:42 anyone seeing issues building avahi due to some libtool relink message on 64 bit systesms? May 20 10:55:51 jeremy_laine: Trust me, adding the idea of packages isn't going to work well since bitbake has no concept of this and we've kept it clear of the concept for a reason (flexibility) May 20 10:56:27 jeremy_laine: feel free to do whatever you want. I did build only to populate your service with data for your tests May 20 10:56:33 RP: not a problem, I basically use the "package" notion to group tasks pertaining to a given build / PN / PV /PR May 20 10:57:08 hrw: would be interested in filling the DB again afterwards, your builds highlighted the fact we *need* pagination :) May 20 10:57:12 Crofton: not with poky and libtool 2.2.x May 20 10:57:31 jeremy_laine: remind me during evening - I will run same set again May 20 10:57:38 I should also add fro armv6 May 20 10:57:49 Crofton: Whats the error exactly? May 20 10:57:51 hrw you need to add a ":)" to that :) May 20 10:57:59 rp, let me paste May 20 10:58:07 Crofton: I did builds for i586, armv4t, armv5te, armv6 May 20 10:58:27 I need to make notes on what builds, I've been trying combinations over the last 24 hours and have forgot :) May 20 11:00:06 http://rafb.net/p/mrhegr20.html May 20 11:00:19 gmm arnv5 May 20 11:00:54 the is a path into x86_64 staging ... May 20 11:01:45 I'm going to update bitbake and try again .... May 20 11:01:59 Crofton|work: That looks nasty. I'd suspect a corrupt .la file May 20 11:02:19 rp, agreed about nasty May 20 11:02:31 Crofton|work: To debug I'd add set -x to the top of the libtool script and see where that path comes from May 20 11:02:38 * RP hugs strace May 20 11:02:57 03jeremy_laine 07org.oe.dev * r3e1c95b8... 10/ (1 classes/oestats-client.bbclass): oestats-client.bbclass: add the DEPENDS variable to task reports May 20 11:02:57 jeremy_laine: http://oestats.jerryweb.org/builders/5/ - please improve 404 a bit May 20 11:03:13 let me try building with newer bb and nuking tmp May 20 11:03:18 just in case May 20 11:03:28 hrw: what would you like to see on the 404 page? May 20 11:03:29 then I'll come back and ask about inserting the -x May 20 11:03:43 jeremy_laine: link to / would be nice May 20 11:04:09 I can build what I need on laptop and have some other work I need to do first May 20 11:09:14 hrw: ok, added to TODO file, would rather focus on usability for the moment May 20 11:09:50 RP, looks like problem exists for armv5 and armv6 build May 20 11:09:56 hrw: can you do an "mtn pull" before your next build so that we feed the DEPENDS field? May 20 11:10:05 rp, only occurs in the 64 bit install May 20 11:10:17 hrw: other than that you're good to run builds whenever you want, have moved to the new DB schema May 20 11:11:07 hrw: what would you think of having the OESTATS_BUILDER value used as the primary key for "builders"? May 20 11:11:35 hrw: that way you could have a /builders/hrw/ url May 20 11:14:54 sure May 20 11:16:23 jeremy_laine: interesting idea May 20 11:16:46 jeremy_laine: build 3 is mine May 20 11:21:41 hrw: http://oestats.jerryweb.org/builders/Hrw_Poky_build/ May 20 11:31:40 Crofton: strange. I don't see it in poky but that setup is a lot different... May 20 11:34:06 you have libtool 2 in poky? May 20 11:35:29 Crofton: yes May 20 11:35:47 How hard will it be to switch .dev to libtool 2? May 20 11:35:54 painful May 20 11:36:01 bother May 20 11:36:26 I suspect part of the issue is the ancient libtool May 20 12:06:20 ~curse php pear May 20 12:06:21 May you be reincarnated as a Windows XP administrator, php pear ! May 20 12:06:56 each time when I look at pear code it reminds me why I tried to avoid it May 20 12:18:05 * * OE Bug 4282 has been created by mad_soft(AT)inbox.ru May 20 12:18:07 * * x11vnc misses libxtst dependency May 20 12:18:09 * * http://bugs.openembedded.net/show_bug.cgi?id=4282 May 20 13:03:56 03Otavio 07org.oe.dev * rd5f229bf... 10/ (5 files in 2 dirs): mtools: update 3.9.9 > 3.9.11 May 20 13:04:01 03Otavio 07org.oe.dev * rf128d8d5... 10/ (3 files in 2 dirs): syslinux: update 3.36 > 3.63 May 20 13:04:06 03Otavio 07org.oe.dev * r6d84f779... 10/ (1 packages/acpid/acpid_1.0.6.bb): acpid: Add 1.0.6 May 20 13:04:10 03Otavio 07org.oe.dev * rc1e1fa87... 10/ (1 packages/startup-notification/startup-notification_0.8.bb): startup-noticiation: Add missing DEPENDs on libsm May 20 13:04:15 03Otavio 07org.oe.dev * r25aecfbd... 10/ (1 packages/glibc/glibc-package.bbclass): glibc: Fix binary locale generation for i586/i686 TARGET_ARCH May 20 13:17:46 03Otavio 07org.oe.dev * r3622dbee... 10/ (3 files in 3 dirs): xserver-xorg-1.3.0.0: Fix compile error by adding a missing header May 20 13:20:00 sorry otavio May 20 13:22:53 hrw: I noticed you've been busy. No problem May 20 13:22:59 hrw: I talked to rp ;-) May 20 13:23:19 RP: thanks a lot May 20 13:23:37 RP: I'll refresh the other ones here, clean up them and ping you again May 20 13:23:50 otavio: np and sounds good :) May 20 14:27:56 03likewise 07org.oe.dev * rb580e935... 10/ (1 packages/mtd/mtd-utils_1.1.0.bb): mtd-utils-1.1.0: Removed DEFAULT_PREFERENCE=-1 later than scheduled. May 20 14:31:18 03mickeyl 07org.oe.dev * r854af845... 10/ (14 files in 2 dirs): efl cvs bump PV in compliance with upstream May 20 15:07:11 I hope I can get some help with opie touchscreen problem I have (bug 4212). May 20 15:07:11 Don't you say "d'oh", I heard you sighing ;-) May 20 15:07:14 I progressed on bug triaging, but now I need your help May 20 15:07:57 Interestingly enough, .stable does not have this problem. I am now trying to bisect where the problem occurs, but of course that is not an easy task. May 20 15:09:27 DISTRO=angstrom-2007.1 against .stable works. I have prepared a diff between the two branches at http://oss.leggewie.org/wip/bug4212.diff.bz2 May 20 15:10:29 DISTRO=angstrom-2007.1 is not present anymore in .dev (or so it seems) May 20 15:10:54 but you can copy it from .stable May 20 15:11:09 I think the angstrom legacy distro file may match .stable May 20 15:11:33 OK, I'll give that a shot. May 20 15:11:39 I believe, I'll have to be back May 20 15:11:41 we do not want people building stuff labaled with the same distro from different branches ... May 20 15:11:49 Expect more from me on this May 20 15:12:17 Crofton: I believe I tried legacy as well, but that broke May 20 15:12:22 the build broke May 20 15:14:19 I found it interesting to see that .stable has a newer (!) qte than .dev May 20 15:14:30 But apparently that does not explain the difference May 20 15:14:41 It kind of struck me nonetheless May 20 15:17:15 bye May 20 15:19:32 Laibsch: if I had to hazard a guess, I'd say it was removed in .dev since it is unused (and probably never will be) May 20 15:20:07 as 2.3.10 is the recommended qte ver to use with Opie May 20 15:21:26 Thanks, understood May 20 16:42:07 I have a quick question or two about monotone and angstrom stable. I backported the fix from 2.6.25 for pxa2xx_udc (zaurus clamshells usb networking) and it now exists as a patch, and a slight tweak to linux-rp_2.6.24.bb I can create a mtn diff for the .bb but I am unsure of how to handle the patch. I assume I don't just (can't) commit it May 20 16:52:31 03koen 07org.oe.dev * r1b4c6de8... 10/ (3 files in 3 dirs): linux-omap2 git: update to rc3 and remove v4l2 from defconfig since that's broken in rc3 May 20 16:52:35 03Laibsch 07org.oe.dev * ra35e2ad2... 10/ (1 packages/vnc/x11vnc_0.9.3.bb): x11vnc: add libxtst to DEPENDS. Closes 4282. May 20 16:56:37 ++ May 20 17:02:42 RP: would ~10 minutes be okay to start? May 20 17:13:32 okay, cya tomorrow May 20 17:17:19 Laibsch: 2.3.12 qt/e was based on snapshot as there was no official 2.3.12 release at all. And one day that snapshot got removed from ewi when it stopped being used as mirror for few services May 20 17:17:37 Laibsch: and 2.3.12 was never used by anyone for any serious use or testing. May 20 17:17:38 bye May 20 17:40:08 re May 20 17:41:23 flo_lap, thanks for organizing Linuxtag May 20 17:42:33 Crofton: yw :) May 20 17:46:18 re May 20 17:59:06 * * OE Bug 4283 has been created by mas(AT)ecs.soton.ac.uk May 20 17:59:08 * * Resume corrupts screen if in landscape mode May 20 17:59:10 * * http://bugs.openembedded.net/show_bug.cgi?id=4283 May 20 18:13:29 I wonder how to approach the fb bug. Trouble is so much knowledge so little time. May 20 18:25:40 03koen 07org.oe.dev * r37205cd4... 10/ (8 files in 3 dirs): linux-omap2 git: add patches to convert mcbsp to a platform device and add omap3 ASoC driver based on that (still missing the actual TWL4030 codec driver though) May 20 18:43:43 * broonie really should have a trawl through OE and see what ASoC drivers are being carried in there... May 20 18:58:58 03koen 07org.oe.dev * r560eb1fe... 10/ (1 packages/dsplink/dsplink_1.50.bb): dsplink: fix pnum due to changed S May 20 19:00:22 so it would seem that base-image is supposed to be the smallest possible image yet its got things like perl in it? it seems huge in fact... at least it takes forever to build May 20 19:01:37 tharvey: pastebin /usr/lib/ipkg/status from image May 20 19:02:24 re May 20 19:02:50 ~pastebin May 20 19:02:51 [~pastebin] A "pastebin" is a web-based service where you can paste anything over 3 lines without flooding the channel. Here are links to a few : http://www.pastebin.com , http://pastebin.ca , http://channels.debian.net/paste , http://paste.lisp.org , http://www.rafb.net/paste May 20 19:03:11 where is that relative to my work dir? May 20 19:03:20 tmp/rootfs/ May 20 19:03:54 I thought I would start simple... try to build angstrom for lite5200 base-image - after 1.5 hours bails out with an error building linux-bd-neon May 20 19:05:07 according to that there is only 38 packages (before it bailed out) - not sure what makes the build so long? May 20 19:05:36 is there a way to blow away all tmp except for toolchain? May 20 19:40:03 03crofton 07org.oe.dev * r0e517dab... 10/ (3 files in 3 dirs): May 20 19:40:03 u-boot_git.bb : Add support for building u-boot for the Lyrtech Small May 20 19:40:03 Form Factor Software Defined Radio (SFFSDR) board. May 20 19:40:53 Crofton: How was the .ie meeting/conference? May 20 20:26:39 03likewise 07org.oe.dev * r635b1919... 10/ (1 contrib/mtn-bisect.sh mtn:execute true): mtn-bisect.sh May 20 22:04:32 hrw: thanks for providing some date to oestats :) May 20 22:05:32 hrw: I have added pagination to the "builds" view and the "builds for a given builder" view May 20 22:33:51 * hrw -> sleep May 20 22:35:47 hrw|gone: good night May 20 22:53:51 03mickeyl 07org.oe.dev * rb2a4a7fd... 10/ (1 packages/python/python-efl.inc): python-efl.inc: remove bashisms to make the Ubuntu crew happy May 21 00:03:51 mickey|zzZZzz: thx **** ENDING LOGGING AT Wed May 21 02:59:56 2008