**** BEGIN LOGGING AT Tue Feb 20 02:59:57 2007 Feb 20 03:23:16 okay, wasn't really broken, just needed to remove work and stamps Feb 20 03:44:34 anyone tell me what this means? Feb 20 03:44:35 ERROR: QA issue: libSDL.la failed sanity test (reference to workdir or installed) Feb 20 03:46:29 I'm not getting that on my checkout from yesterday Feb 20 03:46:52 but I'd say it means that libSDL or part of it went outside the working area Feb 20 03:47:27 odd, there are a lot of those Feb 20 03:47:36 I'm building from .dev, angstrom-gpe-image Feb 20 03:47:51 different libs, lot of same error Feb 20 03:48:23 maybe your oe build area setup is broken? Feb 20 03:48:54 didn't change since install, and it just started dropping this error Feb 20 03:50:05 interesting Feb 20 03:50:43 I'm currently trying to get some binary drivers (under an NDA) working with a custom recipe but they're being a pain. I could have a look later but you're better off waiting for someone with more experience with OE Feb 20 03:51:17 jselect, what driver? Feb 20 03:52:06 sigma designs rmf drivers Feb 20 03:52:39 and the binary driver is for the same arch as you are using oe for? x86 or arm or such? Feb 20 03:53:33 sure is. I have it working under an old transmeta build system called midori Feb 20 03:53:47 it's just a few niggling things with the module loading Feb 20 03:53:58 jselect, what is the issue, just doesn't load cleanly? Feb 20 03:54:43 doesn't take the correct parameters from my bootloader to the kernel requesting reserved memory areas and the module itself doesn't use the reserved memory areas Feb 20 03:54:48 I've fixed it before on an other system Feb 20 03:55:14 eventually had to hardcode the values into kernel because it was being stripped out and I had no time to find out why. now I do :) Feb 20 03:55:25 heh Feb 20 03:55:27 good luck Feb 20 03:55:37 thx Feb 20 03:57:44 Zero_Chaos: QA issue is described in an email on the oe-devel mailing list. You're subscribed to that, right? Feb 20 03:58:28 rwhitby, yeah but I'm out of town and my email checking computer is off Feb 20 03:58:48 rwhitby, is there an online version of the list I can search? Feb 20 04:00:08 http://www.openembedded.org/more-quality-assurance-tests Feb 20 04:00:56 rwhitby, thanks man Feb 20 04:01:20 Zero_Chaos: np Feb 20 04:03:19 short of it, not my fault Feb 20 04:03:22 yes? :-) Feb 20 05:00:07 a quick question: how would one go about changing from read-only root image generation to read-write and vice versa? Feb 20 05:01:05 jselect: If it's read only on the target.. remove ro from the flags for / in /etc/fstab? Feb 20 05:02:42 morning all Feb 20 05:08:34 NAbyss_: there's a bit more to read-only root than just mounting read only Feb 20 05:08:46 such as read-only patches for misbehaving applications Feb 20 05:08:50 and a var tmpfs mount Feb 20 05:08:52 or somesuch Feb 20 05:10:42 Jin|away: Well, this was too good to be true...... NOTE: Task failed: localedef returned an error (command was PATH="/home/henry/oe/dev/tmp/staging/i686-linux/bin/arm-angstrom-linux-gnueabi:/home/henry/oe/dev/tmp/staging/i686-linux/bin:/home/henry/oe/dev/tmp/cross/bin:/usr/lib/qt-3.3/bin:/usr/kerberos/bin:/usr/lib/ccache:/usr/local/bin:/usr/bin:/bin:/usr/X11R6/bin:/home/henry/bin" I18NPATH="/usr/share/i18n" qemu-arm -r 2.6.16 Feb 20 05:10:50 NOTE: package glibc-2.5-r4: task do_package: failed Feb 20 05:35:13 jselect: converting RW to RO is a lot more difficult, yeah.. sounded like you wanted to go RO -> RW, not the other way Feb 20 06:25:20 03rwhitby 07org.oe.dev * rf8ba681a... 10/ (1 conf/distro/ucslugc.conf): ucslugc: Make sure THUMB_INTERWORK is set, and add the missing :slugos OVERRIDE Feb 20 07:04:56 morning Feb 20 07:06:17 Ello Feb 20 07:11:36 morming hrw, NAbuss_ Feb 20 07:12:33 good morning all Feb 20 07:13:57 hrw: do we want to push the ipkg sort patch today or watch for more feedback? Feb 20 07:14:05 morning koen Feb 20 07:15:07 koen: add backward compatibility and decide. for me it can be today Feb 20 07:15:40 hrw: http://scap.linuxtogo.org/files/bd97cb0350dcac3426fc905727b534c9.png Feb 20 07:15:41 i vote also to push it :) Feb 20 07:16:35 koen: libipkg based or other? Feb 20 07:16:48 hrw: libipkg Feb 20 07:17:03 it requires you to select a section before it will show a list Feb 20 07:17:21 the section list for the angstrom feeds is ehm.... "long" Feb 20 07:17:29 shit.. I have to make local angstrom-2007.1 distro config ;( Feb 20 07:21:31 hrw: why? Feb 20 07:21:55 koen: insane == extra step for each package Feb 20 07:22:09 and I do not have time to rebuild everything Feb 20 07:23:17 koen: package-index.bb needs rewriting too Feb 20 07:25:53 will send patch Feb 20 07:28:09 already patched it Feb 20 07:28:13 testing it now Feb 20 07:28:32 03koen 07org.oe.dev * rdfe542a1... 10/ (1 classes/package_ipk.bbclass classes/rootfs_ipk.bbclass): Feb 20 07:28:32 package_ipk, rootfs_ipk: Add patch from Marcin and Koen to sort ipkgs into per architecture subdirs of deploy/ipk Feb 20 07:28:32 see: http://lists.linuxtogo.org/pipermail/openembedded-devel/2007-February/001462.html Feb 20 07:29:09 so now I will have to merge my own changes Feb 20 07:29:33 rootfs_ipk_do_indexes () Feb 20 07:29:57 03koen 07org.oe.dev * r16a968c3... 10/ (1 packages/meta/package-index.bb): package-index: catch up with recent package_ipk changes Feb 20 07:33:11 koen: ugly version you did ;D Feb 20 07:34:20 and incorrect Feb 20 07:34:40 I want package-index to index each subdir, not the ones for this machine Feb 20 07:34:50 * koen will think about that after breakfast Feb 20 07:38:57 koen : good news. i think the libc problem for the ppc with softfloats is solved :) Feb 20 07:39:08 Ifaistos: cool Feb 20 07:39:41 koen: pull/merge/update check my ver Feb 20 07:40:18 03hrw 07org.oe.dev * r77c944a3... 10/ (1 classes/rootfs_ipk.bbclass packages/meta/package-index.bb): rootfs_ipk/package-index: created function to update deploy dirs and made package-index use it Feb 20 07:49:39 hi Rod Feb 20 07:53:03 ~pastebin Feb 20 07:53:15 methinks pastebin is a place to paste your stuff without flooding the channel - try http://pastebin.ca, or http://channels.debian.net/paste, or http://rafb.net/paste/, or http://pastebin.com is usually painfully too slow and unresponsive to use, use one of the other pastebin sites Feb 20 07:55:59 hrw,koen: Does this look like a problem with qemu or possibly rm_work: http://rafb.net/p/Sqbgz379.html Feb 20 07:56:46 qemu Feb 20 07:57:44 hrw: thanks. Grrr... I guess the compat-34 rpms for FC6 don't quite work then...:( Feb 20 07:58:41 hrw,koen: do you know of any quick sanity checks for testing qemu-native? re-building glibc is kinda tedious Feb 20 07:59:07 has anyone build apache2 lately ? Feb 20 07:59:36 i am getting a very strange error -> configure: error: Size of "void *" is less than size of "long" Feb 20 07:59:36 ~lart CoreDump|afk for HARDCODING sharp zaurus into e17 login manager Feb 20 07:59:37 * ibot executes killall -TERM CoreDump|afk for HARDCODING sharp zaurus into e17 login manager Feb 20 07:59:37 Ifaistos: Attempted and failed Feb 20 07:59:46 Ifaistos: Yeah, same thing here Feb 20 07:59:49 hvontres|home: amd64? Feb 20 08:00:08 NAbyss_ : what target ? Feb 20 08:00:11 hrw: boy, I wish :) nope, old IA32 (PII) Feb 20 08:00:27 Ifaistos: spitz (arm) Feb 20 08:00:50 NAbyss_ : it fails for x86 also... Feb 20 08:01:11 Ifaistos: That's weird.. I would've expected x86 to work.. who's the maint? Feb 20 08:01:35 the funny things is site files defines both void * and long as the same size... 4 bytes Feb 20 08:02:21 03hrw 07org.oe.dev * ra1525fff... 10/ (1 packages/bash/bash_3.0.bb packages/bash/bash_3.2.bb): bash: create /etc/shells before using it Feb 20 08:02:25 03hrw 07org.oe.dev * r8443ed9b... 10/ (4 files in 2 dirs): bash: unified, bumped PR Feb 20 08:03:29 JustinP: ping Feb 20 08:04:45 fsck. e17/entrance dies Feb 20 08:05:20 <_law_> morning all Feb 20 08:05:36 <_law_> btw http://lawskynet.blogspot.com/2007/02/vda-ii-linux.html ;-) Feb 20 08:10:10 ok, night all. Feb 20 08:13:11 enlightenment: symbol lookup error: /usr/lib/libecore_x.so.1: undefined symbol: Xutf8TextListToTextProperty Feb 20 08:13:36 did you build it against full x11? Feb 20 08:14:26 hard to get that info now.. feeds are very mixed here Feb 20 08:15:02 * koen wonders how to add stuff to openmoko-pim Feb 20 08:15:50 I'm starting to like the moko gui now I discovered some non-obvious features :) Feb 20 08:16:13 hm. rebuilding whole progear feeds will take day or two... Feb 20 08:16:53 anyone want to donate a few buildboxes + hosting? Feb 20 08:17:00 would save hrw some time :) Feb 20 08:17:07 03pfalcon 07org.oe.dev * r9f76011f... 10/ (1 packages/linux/linux-hackndev-2.6_svn.bb): linux-hackndev-2.6 svn: Hackndev's SVN trunk now is 2.6.20. Feb 20 08:18:02 haha Feb 20 08:20:37 koen: That Japanese on Maemo stuff is very interesting. Would you be able to get it to work on your Nokia with the instructions given on http://makotonen.livejournal.com/14747.html ? Feb 20 08:21:28 Laibsch: maybe with the packages Feb 20 08:21:38 Laibsch: but all parts are available in OE as well Feb 20 08:21:58 Laibsch: but the author, in true .jp style, forked everything Feb 20 08:22:14 Laibsch: instead of making a patch and sending that upstream Feb 20 08:23:58 NAbyss_ : found what the problem is.... it passed config Feb 20 08:24:37 morning Feb 20 08:24:42 hey XorA Feb 20 08:24:58 koen: be careful, you overreact in your .jp bashing. Feb 20 08:25:03 Ifaistos: How'd you fix it? Feb 20 08:25:08 koen: That guy is .fi Feb 20 08:25:20 that guy is Engineering Manager at nokia Feb 20 08:25:26 Laibsch: it happened to every 'project' I've seen done by .jp dudes Feb 20 08:25:48 koen: and it is not that Japanese do this on purpose Feb 20 08:25:49 XorA: right, nokia is fork-happy as well Feb 20 08:26:04 Laibsch: source land 'by accident' in svn? Feb 20 08:26:15 koen: ? Feb 20 08:26:45 NAbyss_ : the ./configure will bailout if cross_compile gets set. it does not check the size of the * null and * long its justs exits. Feb 20 08:26:58 Ifaistos: Ah, interesting Feb 20 08:27:21 NAbyss_ : it compiled. i'll patch it and push it in a while Feb 20 08:27:39 Ifaistos: Cool :) Feb 20 08:27:39 koen: he is also a black belt in judo, so we will let you shout at him :-) Feb 20 08:28:09 XorA: You're reading up, he? Where? Is that guy by chance also coming to FOSDEM? Feb 20 08:28:48 NAbyss_ : check it and let me know if it runs for you.. you will also need to add in the site file ac_cv_sizeof_long_p=${ac_cv_sizeof_long_p=X} where X is the size in your arch Feb 20 08:29:07 XorA: judo is for sissies, karate is da bomb Feb 20 08:29:16 Laibsch: Im pretty sure thats a guy I met at Linux World London Feb 20 08:29:18 morning all Feb 20 08:29:21 hey RP Feb 20 08:29:43 * RP suspects he knows the person XorA is talking about Feb 20 08:29:44 Hi RP Feb 20 08:30:13 RP: linked on mallums blog Feb 20 08:31:17 XorA: I'd seen it before but not realised who it was until now... Feb 20 08:31:37 RP: nor did I until I saw Liam in coconuts Feb 20 08:31:46 * XorA shudders Feb 20 08:32:28 Sadly, I know exactly the photo to which you refer without looking... Feb 20 08:33:34 RP,XorA: clue me in. Sounds like you had a bad experience. Feb 20 08:34:01 koen: What did you mean "source land 'by accident' in svn? Feb 20 08:34:44 Laibsch: one day the developer wakes up and he finds all kind of forked sources in his svn Feb 20 08:36:07 Laibsch: http://flickr.com/photos/rossburton/185522033/in/pool-guadec2006/ Feb 20 08:47:48 hi all Feb 20 08:48:28 hi niv_one_three Feb 20 08:48:28 03xora 07org.oe.dev * r6fc02959... 10/ (1 packages/msmtp packages/msmtp/msmtp_1.4.10.bb): msmtp_1.4.10.bb : a sendmail replacement for use in clients like mutt. Feb 20 08:48:41 I hope I will have a chance to see for real how the open embedded work at FOSDM Feb 20 08:49:02 I mean see the development enviroment Feb 20 08:49:12 koen: BTW, what did you mean the parts are in OE? I guess *his* work is not in or is it? Feb 20 08:49:15 and how to understand compilation errors Feb 20 08:49:43 I am new, and hope this travel could lead me to use my simpad Feb 20 08:49:45 Laibsch: his little private forks aren't in, but all the seperate parts should be available Feb 20 08:49:58 now I can't really realy on the familier image Feb 20 08:50:10 koen: hi Feb 20 08:50:21 is it possible to have a package DEPEND on a specific version of another package ? Feb 20 08:50:25 hrw: hi back at you Feb 20 08:50:38 hey niv_one_three Feb 20 08:50:49 koen: looking forward to meeting you Feb 20 08:50:50 koen: I wonder how it works. And I think he made a major hack usability wise. He must have emulated a keyboard with one input method and put another input method (keyboard-based Japanese input) on top of it. Feb 20 08:51:36 koen: While that would work for example for openmoko (where I thought about what good input method could be used except for HWR) I think this is not something I would want to do. Feb 20 08:52:04 ah that hwr... Feb 20 08:52:14 koen: If my assumption is correct this is also not just some minor "glueing" together but bigger stuff. Feb 20 08:52:26 hrw: ? Feb 20 08:52:39 hrw: Nunome still does not compile. Feb 20 08:54:35 Laibsch: I have 'hwr' in hilight list Feb 20 08:54:38 hi Liam Feb 20 08:55:11 koen@bitbake:/data/build/koen/OE/build/tmp/angstrom/deploy/glibc/ipk$ ls Feb 20 08:55:11 a780 all armv4t armv5te ep93xx fic-gta01 h2200 Packages Packages.filelist Packages.gz Feb 20 08:55:13 yay! Feb 20 08:56:22 hrw: Ah, well that makes sense. Incidentally I made extra sure not to spell hrw to not disturb you ;-) Feb 20 08:57:17 hi Marcin, koen, all Feb 20 08:57:28 hey Liam Feb 20 08:58:57 * Laibsch is off for some shabu-shabu Feb 20 09:06:49 good morning Feb 20 09:07:41 is it possible to have a package DEPEND on a specific version of another package ? Feb 20 09:07:42 03ifaistos 07org.oe.dev * r0e1d8ab5... 10/ (1 site/ix86-common): site/ix86-common: Add ac_cv_sizeof_long_p Feb 20 09:07:46 03ifaistos 07org.oe.dev * re515321b... 10/ (1 site/powerpc-linux): site/powerpc-linux: Add ac_cv_sizeof_long_p Feb 20 09:07:55 03ifaistos 07org.oe.dev * r03ed420d... 10/ (1 site/powerpc-linux-uclibc): site/powerpc-linux-uclib: Add ac_cv_sizeof_long_p Feb 20 09:07:55 03ifaistos 07org.oe.dev * r457c634a... 10/ (3 files in 3 dirs): Feb 20 09:07:55 packages/apache2/apache2-2.2.3: Fix wrong assuption about ptr length during configure Feb 20 09:07:55 To be able to compile you need to add ac_cv_sizeof_long_p in the site file of your arch (if not allready there) Feb 20 09:08:53 NAbyss_ : give a try and let me know Feb 20 09:13:09 03koen 07org.oe.dev * r609905b6... 10/ (1 site/arm-linux): arm-linux site: add ac_cv_sizeof_long_p=${ac_cv_sizeof_long_p=4} Feb 20 09:19:28 koen : Sorry to ask again and again, but need to figure out if is it possible to have a package DEPEND on a specific version of another package ? Feb 20 09:20:10 it's not really possible Feb 20 09:22:29 koen : so both versions of the packages would have to be defined in advance... Feb 20 09:55:50 greetings Feb 20 10:11:05 koen : have a look at this-> http://rafb.net/p/2estvm68.html its from a clean build with today's repository Feb 20 10:13:19 hrw: http://bec-systems.com/web/index.php?option=com_content&task=view&id=65&Itemid=9 - nice. I'm glad that someone find cpio.gz useful ;-). but we still need better support for building initrd's in OE. Feb 20 10:14:19 psokolovsky_: its cbrake's article Feb 20 10:14:37 cool Feb 20 10:16:10 hi Feb 20 10:17:08 saladino: hi Feb 20 10:38:57 lrg: ping Feb 20 10:39:51 hi wookey_ Feb 20 10:43:57 anyone build php recently ? Feb 20 10:46:47 Ifaistos: mtn sync && mtn up Feb 20 10:47:12 qa staging shouldn't die anymore Feb 20 10:47:20 if it does, fix the offending package :) Feb 20 10:50:53 Ifaistos: http://www.openembedded.org/more-quality-assurance-tests Feb 20 10:54:44 Ifaistos: ERROR: Architecture did not match (20 to 3) on /work/ppc603e-angstrom-linux/udev-092-r15/install/udev/sbin/scsi_id Feb 20 10:54:57 Ifaistos: could you run 'file' on that binary? Feb 20 10:57:33 ~spell symptom Feb 20 10:57:36 'symptom' may be spelled correctly Feb 20 11:01:36 03koen 07org.oe.dev * r40506635... 10/ (1 classes/insane.bbclass): Feb 20 11:01:36 insane.bbclass: we really don't want packages with the wrong architecture, so do bb.fatal instead if bb.error Feb 20 11:01:36 symptom: Feb 20 11:01:36 ERROR: Architecture did not match (20 to 3) on /work/ppc603e-angstrom-linux/udev-092-r15/install/udev/sbin/scsi_id Feb 20 11:01:36 This means OE detected type 3 (i386) while it expected type 20 (powerpc), most likely caused by some fool hardcoding 'gcc' as compiler Feb 20 11:03:12 koen: its a link to /lib/udev/scsi_id Feb 20 11:04:35 koen: its a ppc file, no error in there Feb 20 11:05:09 koen: -> http://rafb.net/p/bLM3VI36.html Feb 20 11:05:54 Ifaistos: I've mailed QA meister zecke about it Feb 20 11:07:39 me gets back to matlab Feb 20 11:09:02 Ifaistos, whatz up ?!! Feb 20 11:09:12 Ifaistos, pos pai to wifi? Feb 20 11:11:05 kala. poios ? Feb 20 11:12:19 Ifaistos, join #cacko plz gia na min kano identify apo edo Feb 20 11:12:38 koen|away: I need to look through things again BUT Feb 20 11:12:46 koen|away: insane.bbclass is used to find ALL errors Feb 20 11:13:17 koen|away: this is why all methods set sane = False conditionally, it will never be set to sane = True from within a test Feb 20 11:19:54 If I build angstrom, should I get warnings/erros from teh QA stuff? Feb 20 11:20:42 Crofton|home: depends on how much fatal error koen disabled Feb 20 11:20:52 heh Feb 20 11:21:06 what is the best way to keep a log of the bitbake output? Feb 20 11:22:34 Crofton|home: check for the qa_log in tmp Feb 20 11:22:59 hmm, do not see one Feb 20 11:23:54 temp of $WORKDIR Feb 20 11:24:47 ah Feb 20 11:24:59 I am nuking tmp and rebuilding my images Feb 20 12:05:17 hrw: just read your article about USB-wifi drivers -- very helpful! Feb 20 12:05:31 cbrake: thx Feb 20 12:06:37 hi hrw Feb 20 12:13:26 he zecke Feb 20 12:13:39 woglinde: hi Feb 20 12:14:41 NOTE: Not creating empty archive for glibc-2.5-r4 Feb 20 12:14:43 ARGH Feb 20 12:22:01 koen: something is fscked in angstrom Feb 20 12:22:49 koen: bitbake glibc does not give libc6 package Feb 20 12:23:35 hi all Feb 20 12:23:39 hi gremlin[it] Feb 20 12:25:30 gremlin[it], Hi! Did you know that angstrom-gpe-iamge was plucked to fit in 16Mb for you? Feb 20 12:26:16 is the packaging subdirectory stuff supposed to happen when multimachine is not selected? Feb 20 12:27:01 (our autobuilder, which is not using multimachine), just put stuff in an armv5te subdirectory to where it used to put it. I thought these changes were only for multimachine ... Feb 20 12:28:49 oops, I tell a lie - we do have multimachine set. ignore what I said :-) Feb 20 12:29:41 rwhitby: they are non multimachine related Feb 20 12:30:26 so the structure of tmp/deploy/ipks will change whether I have multimachine or not? Feb 20 12:30:42 (and correspondingly, my rsynced feeds will change?) Feb 20 12:31:16 yes Feb 20 12:31:19 (and if that happens, all existing deployed systems will break?) Feb 20 12:31:52 hmm - guess I should have read those RFC's more closely - for some reason I thought this was only going to be enabled in the multimachine case. Feb 20 12:32:00 php in order to config needs to execute a script from libxml (xml2-config) where should this script reside ? in cross/bin or staging/bin ? Feb 20 12:32:43 rwhitby: the changes are backward compatible Feb 20 12:33:27 koen: for the non-multimachine case? So if I remove multimachine all my ipks will once again end up in a *single* directory with no subdirectories? Feb 20 12:33:55 no Feb 20 12:33:59 they still will get sorted Feb 20 12:34:02 psokolovsky_, ohhh thanks !!! i'll try this evening ... but i'm not the only owner of an h3600 :) :) :) Feb 20 12:34:11 koen: then how is that backward compatible? Feb 20 12:34:19 but nothing build related will break Feb 20 12:34:29 gremlin[it], sure, I meant for you to test ;-) Feb 20 12:34:41 I'm not discussing build here, I'm discussing structure of tmp/deploy/ipk Feb 20 12:35:03 gremlin[it], but you'll need to set DISTRO_TYPE="release" for that. ping me when you'll be doing that. Feb 20 12:35:34 I'm not against the change, but *don't* try and tell me that the location of the resulting ipks in the directory structure is backwards compatible, cause it clearly is not. Feb 20 12:35:37 ah ok ... i'll go to swimming pool this evening i'll try at 200:00 about ... Feb 20 12:35:48 ops ... 22:00 i mean Feb 20 12:36:46 rwhitby: I'm not saying the structure is backward compatible, I'm saying OE will keep working as expected instead of requiring a rebuild Feb 20 12:37:30 koen: I don't care what *you're* saying. I'm saying that the structure of the tmp/deploy/ipk directory has changed. And that is going to break the nslu2-linux feeds unless I make it backwards compatible again. Feb 20 12:37:56 rwhitby: that problem is outside the scope of OE Feb 20 12:38:34 koen: that's simply weaseling out of it. just admit that it's an incompatible change and tell me what variables need to be changed to put all the ipks back in the one directory again. Feb 20 12:38:57 there isn't such a variable Feb 20 12:39:46 koen: Wouldn't it be a good idea to have such a variable? Feb 20 12:40:10 NAiL: IMO not Feb 20 12:40:22 I'm sorry I didn't reply to the RFC. I didn't think it was relevant at the time. Feb 20 12:40:46 NAiL: "that's outside the scope of OE" - koen feels that OE fscking around with feed locations is not a problem for anyone. Feb 20 12:41:02 it's now our problem to clean up the mess. Feb 20 12:41:30 my upload script didn't break Feb 20 12:41:36 * koen shrugs Feb 20 12:42:05 yeah, well that's not a very good test of the non-multimachine case when *you're* already using multimachine. Feb 20 12:42:25 this isn't multimachine related Feb 20 12:42:57 * NAiL notes koens first mail to the ml regarding this: Feb 20 12:42:57 Hi, Feb 20 12:42:57 When using multimachine and ipkg deploy/ipk gets awfully crowded after a while. I want to Feb 20 12:43:01 propose sorting it by subarch. Attached is a patch that adds support for that to Feb 20 12:43:15 right Feb 20 12:43:17 And it isn't multimachine related, you say? Feb 20 12:43:18 What it is is that all the ipks used to be in one directory, where you could run make-index over them and then rsync them to one feed directory. Now that has changed incompatible in the structure of tmp/deploy/ipk. Feb 20 12:43:23 the patch isn't Feb 20 12:43:34 bitbake package-index still works Feb 20 12:44:13 koen: admitting a problem is just something that you're unable to do, isn't it? Feb 20 12:44:23 add an variable to get the old beahviour, let someone fix the nslu2 deployment/upload scripts and get over it Feb 20 12:44:52 zecke: Don't get me wrong, I agree with the change. What I don't agree with is someone trying to tell me that nothing has changed. Feb 20 12:45:12 I dind't say nothing has changed Feb 20 12:45:40 * XorA realises his feed sync script needs to be fixed Feb 20 12:45:51 XorA: exactly. Feb 20 12:46:02 mine relies on only one feed Feb 20 12:46:21 XorA: as I expect does anyone else who wasn't already using multimachine like koen was. Feb 20 12:46:36 rwhitby: ehm Feb 20 12:46:45 rwhitby: this is not multimachine related Feb 20 12:47:01 rwhitby: multimachine has nevert split deploy/ipk Feb 20 12:47:12 XorA: you also need to go to all your current users, and tell them to add more feeds to their ipkg dir, cause the feeds are no longer a single dir. Feb 20 12:47:25 que? Feb 20 12:47:40 what have user to do with this? Feb 20 12:47:47 users* Feb 20 12:47:53 * XorA will find tmp/deply/ipkg *.bb -exec cp \{\} feed/ Feb 20 12:48:06 koen: already flashed images, and users that do ipkg upgrade Feb 20 12:48:15 XorA: of course, that is the other alternative. Feb 20 12:48:17 koen: I don't think /etc/ipkg/* is installed by packages? Feb 20 12:48:17 zecke: that's nonsense Feb 20 12:48:29 it is not (in some cases) Feb 20 12:48:35 zecke: it is Feb 20 12:48:44 koen: I'm insane I may do that Feb 20 12:48:46 zecke: only koen's case matters, it seems. Feb 20 12:49:06 zecke: OE removes references to deploy before writing out the final ipkg.conf Feb 20 12:49:36 zecke: only FEED_URIs get added, which don't change Feb 20 12:50:18 zecke: the rootfs before and after the change is *exactly* the same Feb 20 12:50:36 koen: right, it boils down to the feed sync and if the seperation of feeds is only done in OE land or is used for distribution as well Feb 20 12:51:23 zecke: right Feb 20 12:51:48 zecke: we have been quite adamant in the past that deploy/ipk is *not* a feed Feb 20 12:54:04 XorA: I'll borrow your script when you're done please. It's midnight here, and this problem is something which will need to be solved tomorrow. I'll simply turn off feed updating until I find the time to clean up the new mess. Feb 20 12:54:08 zecke: and shifting the work to your users because you can't fix your upload script is just plain insulting Feb 20 12:54:33 koen: not being able to admit a change which impact your developer community is insulting. Feb 20 12:55:15 It's a good change, but a change nonetheless. Feb 20 12:55:51 I said it didn't impact *build* nor *mu* upload scripts Feb 20 12:56:03 I don't know how your upload scripts work Feb 20 12:56:17 and I said that it *did* impact tmp/deploy/ipk structure, but you haven't admitted that yet. Feb 20 12:56:28 I did Feb 20 12:56:40 Ifaistos: which check followed symlinks? Feb 20 12:56:50 rwhitby: find tmp/deploy/glibc/ipk/ -name "*.ipk" -exec cp \{\} feed/ \; Feb 20 12:57:03 "[14:10] koen: rwhitby: I'm not saying the structure is backward compatible, I'm saying OE will keep working as expected instead of requiring a rebuild" Feb 20 12:57:04 rwhitby: then run your normal upload script Feb 20 12:57:42 XorA: don't forget to re-run bitbake package-index after the move. Feb 20 12:57:51 * koen rsyncs and runs http://www.angstrom-distribution.org/unstable/feed/unsorted/sort.sh.txt Feb 20 12:58:03 rwhitby: I use ipkg-make-index directly Feb 20 12:58:09 03freyther 07org.oe.dev * r4386c713... 10/ (1 classes/insane.bbclass): Feb 20 12:58:09 classes/insane.bbclass: No use to make it bb.fatal if we return False... Feb 20 12:58:09 The idea of insane.bbclass is to report all issues found at Feb 20 12:58:09 once. This is why the checks return True if no error was found Feb 20 12:58:09 and False if at least one was found. Feb 20 12:58:10 At the end of all check runs we will see if at least one check Feb 20 12:58:12 failed and call bb.fatal Feb 20 12:58:57 zecke : posted a link a little bit back but closed irc and lost it Feb 20 12:59:07 zecke : maybe koen has it Feb 20 12:59:37 zecke: udev ships a symlink, see the commit mail I forwared Feb 20 12:59:41 forwarded* Feb 20 13:00:04 koen: "... keep working as *you* expected, not keep working the same way as it did before" Feb 20 13:00:37 did OE break? Feb 20 13:00:38 no Feb 20 13:00:47 your upload script did Feb 20 13:01:08 anyway, no point in discussing it more. you've admitted it was an incompatible change to the tmp/deploy/ipk directory structure which impacted some developers, and I agree with the reason for the change, so now it's simply an extra job to fix it. Feb 20 13:01:17 as one might have expected when deploy/ikp changed Feb 20 13:01:33 the patch was sent to the mailinglist and received no objections Feb 20 13:02:07 heh - it was prefaced with "When using multimachine ..." Feb 20 13:03:45 yes, since multimachine highlighted the problem Feb 20 13:04:57 XorA: thanks for the find command, will add it tomorrow. Feb 20 13:05:53 rwhitby: unfortuneately it goes and empties all the morgues as well Feb 20 13:07:13 Hmm - good point. Maybe I need to add a "backwards-compatibility" option to package-index.bb ... Feb 20 13:08:21 re Feb 20 13:08:31 koen: Ifaistos : okay will leave torwards no Feb 20 13:08:33 w Feb 20 13:08:53 zecke, koen: I suspect insane class when it comes to glibc packaging Feb 20 13:09:15 hrw: I suspect that as well, but I cannot see how Feb 20 13:09:41 03tmbinc 07org.oe.dreambox * re47513d6... 10/ (1 packages/enigma/enigma_cvs.bb): enigma: on dm600pvr, install 'showshutdownpic' Feb 20 13:09:48 03tmbinc 07org.oe.dreambox * r345802a6... 10/ (1 packages/dreambox/dreambox-dvb-modules.bb): dreambox-dvb-modules/dm600pvr: update drivers to 20070220 Feb 20 13:09:49 03tmbinc 07org.oe.dreambox * r0c7d2887... 10/ (1 packages/initscripts/initscripts-opendreambox_1.0.bb): initscripts-opendreambox: workaround for dm600pvr to not unmount /tmp for displaying the shutdown picture Feb 20 13:15:58 hrw: remember pocketworkstation? Read http://www.maemopeople.org/index.php/jaffa/2007/02/20/running_debian_proper_alongside_maemo and weep Feb 20 13:17:03 ;P Feb 20 13:17:09 powerpc in OE is 32bit? Feb 20 13:17:17 yes Feb 20 13:17:49 I want to update insane.bbclass with extra field: 32bit or 64bit Feb 20 13:22:20 morning all Feb 20 13:23:15 night all Feb 20 13:23:27 koen: nothing wrong with a good bit of chrooted goodness Feb 20 13:24:30 XorA: nothing wrong with that Feb 20 13:24:56 XorA: it's just the X-over-vnc running under X on the same machine.... Feb 20 13:33:26 CIA-20: Feb 20 13:42:43 03hrw 07org.oe.dev * r46f94374... 10/ (1 classes/insane.bbclass): insane.bbclass: add i{4,5,6}86 Feb 20 13:42:48 03hrw 07org.oe.dev * r2c67c0e3... 10/ (1 classes/insane.bbclass): insane.bbclass: reformatting Feb 20 13:42:52 03hrw 07org.oe.dev * re7714ad5... 10/ (1 classes/insane.bbclass): insane.bbclass: remove commented code to unbreak builds Feb 20 13:42:57 03hrw 07org.oe.dev * r00269cb6... 10/ (1 classes/insane.bbclass): Feb 20 13:42:57 insane.bbclass: unbreak it for 64bit targets Feb 20 13:42:57 Added extra entry into dict to mark 32bit/64bit and use it for manipulating Feb 20 13:42:57 ELF headers. Feb 20 13:42:57 Some architectures would need checking (not OE ones). Feb 20 13:50:07 koen: http://projects.linuxtogo.org/frs/?group_id=14 could need an update :-) Feb 20 13:50:23 indeed Feb 20 13:50:34 but everytime I think that berlios goes down :( Feb 20 13:51:02 koen: heh, indeed Feb 20 13:51:14 * koen crosses fingers Feb 20 13:52:25 03ifaistos 07org.oe.dev * rca58070f... 10/ (3 files in 3 dirs): packages/php/php-5.1.4.bb: Patch to allow php-5 to compile for x86 (at least) Feb 20 13:52:30 03ifaistos 07org.oe.dev * r019782b0... 10/ (1 packages/libxml/libxml2_2.6.26.bb): Feb 20 13:52:30 packages/libxml/libxml2_2.6.26.bb: copy xml2-config to staging/bin Feb 20 13:52:30 This is need by php during install to very that libxml2 is installed Feb 20 13:52:38 03ifaistos 07org.oe.dev * r4729b71e... 10/ (5 files in 3 dirs): /packages/linux: Add kernel 2.6.20 for dht-walnut machine Feb 20 13:52:40 florian: can't reach berlios :( Feb 20 13:52:45 03ifaistos 07org.oe.dev * r8c287818... 10/ (4 files in 3 dirs): packages/linux: Add kernel 2.6.20 for x86 Feb 20 13:53:02 Ifaistos: inherit binconfig was not enough for libxml2? Feb 20 13:53:12 Ifaistos: doesn't 'inherit binconfig' take care of that? Feb 20 13:53:31 no its not set to executable Feb 20 13:54:01 Ifaistos: then do_compile_append() { chmod 755 xml2-config} Feb 20 13:54:05 koen: http://prdownload.berlios.de/bitbake/bitbake-1.6.6.tar.gz, the first mirror seems to be alive Feb 20 13:54:57 Ifaistos: binconfig.bbclass mangle files content Feb 20 13:56:03 hrw : not sure what you mean Feb 20 13:56:48 Ifaistos: all *-config files are staged by binconfig.bbclass (inherit binconfig). during staging they have all paths changed to point to staging Feb 20 13:57:30 Ifaistos: so if you have to make it executable to get it used by binconfig.bbclass then do it in extra step Feb 20 13:57:54 ahhh ok Feb 20 13:58:55 basically both patches are king of a hack to get it compiled and see how it works Feb 20 13:59:19 php needs *alot more work* Feb 20 14:00:07 so at least we can get apache+php working on x86 Feb 20 14:00:52 florian: http://projects.linuxtogo.org/frs/?group_id=14&release_id=44 Feb 20 14:01:15 I also worry a bit when see all those 2.6.20 kernels recipes Feb 20 14:01:42 all tested to at least boot ;) Feb 20 14:01:47 2.6.20 progear, simpad, efika, dht-walnut, rp Feb 20 14:01:52 first two are joined Feb 20 14:03:15 since the api changed somewhat in 2.6.20 you need them to keep in sync with varius external modules Feb 20 14:04:30 koen: many thanks! Feb 20 14:04:45 np Feb 20 14:04:53 the one thing i can not decide is if the x86 should be a "minimal" one or have everything defined as modules (except basic drivers need it for booting) Feb 20 14:05:02 I wonder if we maybe want a generic kernel project at ltg Feb 20 14:06:31 more fragmentation! Feb 20 14:07:26 koen: or less if we convince enough projects to join... Feb 20 14:08:01 e.g. if i would have to work on a kernel i could not use the hh.org tree. Feb 20 14:08:13 packages/linux/linux_2.6.20.bb was added as generic kernel Feb 20 14:08:42 what is the easiest way to tell angstrom to produce a tar.gz rootfs rather than a jffs2 image? Feb 20 14:09:13 Jin^eLD: In your local.conf, IMAGE_TYPES Feb 20 14:09:17 ah, thx Feb 20 14:09:32 Er IMAGE_FSTYPES Feb 20 14:10:06 got that, thanks Feb 20 14:10:46 will be my first test with angstrom on that hardware :> Feb 20 14:10:57 Good luck :) Feb 20 14:10:58 and actually my first angstrom test at all, first time I built it now :) Feb 20 14:11:07 thx =) Feb 20 14:14:15 koen : would you be interested for binary images for angstrom - for efika and x86 ? Feb 20 14:15:06 koen : something that would fit in a 128MB usb stick or CF Feb 20 14:15:28 Ifaistos: my progear run angstrom Feb 20 14:16:15 i mean something people can download, write to usb stick and run it, without the...pain Feb 20 14:16:44 of having to master OE Feb 20 14:24:49 Hmm... Anyone familiar with PC1 encryption? Feb 20 14:25:05 I'm trying to find something that can en/decrypt it on the cmdline Feb 20 14:25:26 what is /var/lib/dbus/machine-id?! Feb 20 14:26:11 florian: some id dbus needs, iirc via dbus-uuidgen Feb 20 14:26:29 Ifaistos: sure, you can upload those to unstable/images/ if you want Feb 20 14:31:40 Ifaistos: do you know how to get agp working properly on the efika? Feb 20 14:31:53 so that I can use DRI/DRM on my radeon? Feb 20 14:31:58 Koen: is the pressrelease good enough to send out ? Feb 20 14:32:52 koen : What do mean by "properly" ? Feb 20 15:13:43 gerwinin: no objections from me, dunno what the others think Feb 20 15:13:57 Ifaistos: it's using some weird agp-over-pci thing Feb 20 15:14:13 koen: Nokia have released the cx3110x source, there's a .bb in poky ;-) Feb 20 15:14:24 RP: :) Feb 20 15:14:38 koen : its agp actually. agp x1 @66Mhz Feb 20 15:15:22 Ifaistos: not quite Feb 20 15:15:29 the 'not quite' is the problem Feb 20 15:16:41 koen : the pci bus of the power pc can be 33 or 66 Mhz. when you plug the raiser the bios switches it to 66 Feb 20 15:17:04 well, agp still doesn't work properly :( Feb 20 15:17:15 koen : haven't done much work on grafix modes, either than running Angstrom Feb 20 15:17:20 I remember some weird agp configs from the pegasos board Feb 20 15:17:50 koen : well its a big hack... having a vm to initialize the bios of the card... Feb 20 15:20:22 koen: which pegasos? there were two versions Feb 20 15:20:26 both suxx Feb 20 15:21:33 no idea Feb 20 15:21:41 the one I saw in the wild a while ago Feb 20 15:24:26 RP: merged into OE and source mirrored :) Feb 20 15:25:00 03koen 07org.oe.dev * r0133c0f2... 10/ (1 packages/c3110x packages/c3110x/cx3110x_1.1.bb): c3110x: add wifi driver for connexant chips (from poky) Feb 20 15:46:53 JustinP: http://pastebin.ca/365019 is needed to build edje-native Feb 20 15:49:07 * koen builds efika from scratch to fix insane problems Feb 20 15:49:20 * koen stabs FIR filterdesign in the z-plane Feb 20 15:53:06 I need to add a newer version of libxml-native for the latest version of php (5.2) to work. What is the cleanest way to submit patches? Do I submit a patch for libxml-native and a patch for php-native_5.2 or just submit one mtn patch for both? Feb 20 15:54:31 two patches Feb 20 15:54:55 clean builds with the new rootfs for 2 targets (efika/dht) with uclibc build succesfully Feb 20 15:55:34 hrw, thanks. Feb 20 15:55:53 Gerrath : Have you tested php 5.2 ? on which target ? Feb 20 15:56:01 hrw, I'll just set the php bug dependent on the libxml-native bug.. Feb 20 15:56:14 good thinking Feb 20 15:56:21 Ifaistos, I have tested it on xscale (arm), I added in your patch as well :-) Feb 20 15:57:36 Ifaistos, I have patches to add in the latest php4 and php5 to OE, I just need to clean up the native files which is what I'm doing now. I would love for you to test them on your system if you have time. Feb 20 15:57:47 hrw: I'll know in ~2 hours if insane causes the glibc problem Feb 20 15:58:55 Gerrath : sure. just ping me when you submit them on the bugtracker Feb 20 15:59:08 Ifaistos, will do, thanks. Feb 20 15:59:26 koen: ok Feb 20 16:08:02 ibot change 350 usd to ils Feb 20 16:08:07 350.00 United States Dollar (USD) makes 1,465.46 Israel New Shekel (ILS) (from http://www.xe.com/) Feb 20 16:12:02 btw.. someone has spare pcmcia -> cf (type I) adapter? Feb 20 16:15:14 cbrake: I just read your latest article Feb 20 16:22:15 hi Feb 20 16:22:58 cu Feb 20 16:29:29 Ifaistos, your php patch actually breaks php for other builds, can you try this instead on your board and see if it still works: --with-libxml-dir=${STAGING_BINDIR_CROSS} Feb 20 16:32:31 xumbi8: yes, things are moving along. I am now booting into the nor partition, and then switch_root'ing into the jffs2 on NAND. Feb 20 16:32:45 xumbi8: so far things seem to be working. Feb 20 16:33:17 xumbi8: it takes 20seconds to mount a 512MB NAND parition, so I'll have to trim down the partition size some. Feb 20 16:36:28 Gerrath : will try later, as i am going now. In any case make any changes you want to get a working system and then i'll see how to get it going for x86 and the ppc platforms Feb 20 16:37:07 Ifaistos, ok, I will have patches ready in an hour or two, should I email you at the email address in bug tracker? Feb 20 16:37:46 Gerrath : send to stelios@stelioscellar.com Feb 20 16:38:31 Ifaistos, ok thanks. Feb 20 16:46:14 xumbi8: a few other things to work out like CONFIG_MTD_CMDLINE_PARTS is broke in the cm-x270 NAND driver ... Feb 20 16:51:33 hrw|gone: 1.6M libc6_2.5-r4_ppc603e.ipk Feb 20 17:04:20 cbrake: ah very cool, 20 seconds is fast, when I had a jffs2 file system running in just 128MB of NAND it took about a minute to mount Feb 20 17:08:23 koen: Do we have any sort of snity tests for qemu-native (other than rebuilding glibc) ? Feb 20 17:08:24 not really Feb 20 17:09:11 koen: ok. I was thinking of digging into my segfault from earlier some more... Feb 20 17:11:28 " So it turns out that OpenOffice exports RTF which neither it or Microsoft office is able to read." Feb 20 17:11:43 from http://blogs.gnome.org/view/uraeus/2007/02/20/0 Feb 20 17:12:08 wow... what a usefull feature :) Feb 20 17:12:20 It's a new form of security. Feb 20 17:12:36 does anybody still use RTF?????? Feb 20 17:12:59 it's the only sanely spec'ed format out there Feb 20 17:13:09 I bump into it all the time with my M$ customers. Feb 20 17:13:40 write once...read never :) Feb 20 17:13:53 Many of them think that they are being "open" by putting "README.RTF" in a folder instead of "README.DOC". :( Feb 20 17:14:39 Ifaistos: hey, that sounds like the perfec format for my new WORM based filesystem :) Feb 20 17:15:38 err, I Mean WORN based of course :) Feb 20 17:15:42 xumbi8: mount time may depend how full it is as well -- I'm currently only 6% full. Feb 20 17:16:48 http://academics.vmi.edu/ee_js/Research/IC_Datasheets/digital_cmos/Write%20Only%20Memory.pdf Feb 20 17:23:52 03koen 07org.oe.dev * r0385e8bd... 10/ (1 classes/insane.bbclass): insane.bbclass: fix ERROR: Exception:exceptions.TypeError Message:not enough arguments for format string by removing a spurious '%s' Feb 20 17:25:32 hrw|gone: ok, gcc packaging goes wrong now :( Feb 20 17:39:46 hi all Feb 20 17:40:09 hey mr_nice Feb 20 17:40:15 hrw|gone: see bug 1892 Feb 20 17:40:20 ni Feb 20 17:40:50 !oebug 1892 Feb 20 17:40:52 * * Bug 1892, Status: NEW, Created: 2007-02-20 09:55 Feb 20 17:40:53 * * koen(AT)dominion.kabel.utwente.nl: Bitbake decides to not package stuff Feb 20 17:40:54 * * http://bugs.openembedded.org/show_bug.cgi?id=1892 Feb 20 17:43:41 koen: hi Feb 20 17:48:14 03florian 07org.oe.dev * r2aa77368... 10/ (4 files in 3 dirs): gpe-bootsplash-theme-gpephone: New theme for gpephone images. Feb 20 17:48:20 03florian 07org.oe.dev * r571d3cdf... 10/ (1 packages/gpephone/task-gpephone.bb): task-gpephone: Use specific theme. Feb 20 17:49:25 hi Crofton__, have you had time to take a look at the omap tty problem ? Feb 20 17:53:03 Sorry, I won't get a chance this week. My plane leaves for FOSDEM tomorrow afternoon Feb 20 17:53:11 and the locals are beating on me :) Feb 20 17:53:38 and I need to go see a guys defense on UWB and position location Feb 20 17:53:41 ok, i'll try to investigate a little bit on my own Feb 20 17:54:13 but do you know- if i clear the line console=ttyS0 from bootargs Feb 20 17:54:42 is it possible to accesss the flash or should i repair the flash for that ? Feb 20 17:54:52 * koen stabs bitbake Feb 20 17:54:56 * koen stabs matlab Feb 20 17:55:06 with that out of the way, time for dessert Feb 20 17:55:16 :D Feb 20 17:57:39 if there is only one char to change in a textfile is it better to create a patchfile or write the sed command in a bugreport? Feb 20 17:57:55 putumutukas, it is OK to clear that from the kernel args, you should still be able to talk to u-boot via serial Feb 20 17:58:11 ok, i'll try that then Feb 20 17:58:55 Cofton, and just in case i mess it up again- do you happen to have any good tools to repair Rev D flash ? Feb 20 18:01:58 putumutukas, I have my laptop in the defense ... Feb 20 18:02:48 or better to ask are sed commands accepted in a bugreport? Feb 20 18:03:41 Crofton, i have mine just in front of me :P but did i understand right that you have some stuff in your laptop ? Feb 20 18:04:01 no Feb 20 18:04:10 I can just chat some while I watch the talk Feb 20 18:04:23 a ok Feb 20 18:04:45 then i go and try what happens if i remove the console line from the bootargs Feb 20 18:05:02 ok Feb 20 18:29:31 mickeyl: good morning! Feb 20 18:29:45 mickeyl: how's the defense preparation coming along? Feb 20 18:31:28 mmh Feb 20 18:31:31 not feeling well about it Feb 20 18:32:02 i definitely started way too late Feb 20 18:32:19 so now 24h left to prepare and i'm very nervous Feb 20 18:32:46 mickeyl: make sure you get enough sleep the night before Feb 20 18:33:01 koen: I'll try. can't tell that I'm sleeping very good these days though Feb 20 18:33:16 i'll probably look like a Zombie @ fosdem :) Feb 20 18:33:33 Crofton, it works- removing the console=ttyS0 gave me back my long awaited serial port :) Feb 20 18:33:36 'thanks' to my cold I'm sleeping 10 hours a day Feb 20 18:33:50 * koen is wearing a mickeyl style woolen cap Feb 20 18:33:57 hehe, cool Feb 20 18:34:08 get well soon Feb 20 18:34:13 you need to be fit in 72h Feb 20 18:35:52 * koen unplugs the cat5 cable and waits for irc to catch up Feb 20 18:36:46 thanks for the info putukas Feb 20 18:37:20 mickeyl, I'm watching a guys defense right now, and i keep nodding off Feb 20 18:37:30 hehe Feb 20 18:37:43 The commitee guys aren't though :) Feb 20 18:37:46 can he answer the questions? Feb 20 18:37:55 he seems to be doing ok Feb 20 18:37:59 good Feb 20 18:38:03 mickeyl, gete to work :) Feb 20 18:38:10 been all day. need a rest Feb 20 18:38:11 mickeyl, just relax, breathe, and you'll do fine. i totally blew one question and they let me by... (-; Feb 20 18:38:20 HopsNBarley :) Feb 20 18:39:09 Crofto you're welcome, but if you happen to create a new kernel image maybe you enable the ttyS1 and ttyS2 too which are available on the 40pin connector ;) Feb 20 18:42:17 on the OSK? Feb 20 18:42:54 yes Feb 20 18:43:35 regular serial ports? do they need hw setup? Feb 20 18:43:36 the ttyS1 and S2 are available on hardware, they just have to be connected via the 40pin connector Feb 20 18:43:47 I don't recall a kconfig option for this Feb 20 18:44:00 you have a mate to the connector? Feb 20 18:45:06 yes Feb 20 18:46:04 i'm building some sensors to the rs485 line which is made from rs232 Feb 20 18:46:37 I'll try and look into this Feb 20 18:46:42 but since by default the rs485 has a limit of adresses (31) then it would be very easy to expand Feb 20 18:46:53 Feel free to send patches :) Feb 20 18:47:23 i'm not so familiar with linux yet :P Feb 20 18:47:39 i'm more the hardware guy- thatswhy i also asked ilvez to help me a bit Feb 20 18:48:15 mickeyl: Don't forget to get some sleep too :) Feb 20 18:49:32 sleeping is overrated ;) Feb 20 18:50:09 mickeyl: sure...but not in the middle of your own defense :) Feb 20 18:50:15 hehehe, true Feb 20 18:50:39 lart windblows per process memory limits Feb 20 18:50:45 ~lart windblows per process memory limits Feb 20 18:50:45 * ibot keeps mailing windblows per process memory limits free America Online CDs until he drowns Feb 20 19:02:29 mickey_bbl, make sure you talk about all previous work ..... Feb 20 19:26:02 WHat is this? Is this something i should fix? How do i fix it ? http://pastebin.ca/365220 Feb 20 19:34:13 goxboxlive: I think that is the result of the new and improved insanity checker in angstrom. IIRC, those errors are informative only. koen or zecke should be able to tell you more Feb 20 19:34:40 ok, i c thankyou Feb 20 19:35:00 does no one read oe-devel or look at http://openembedded.org??!?!?!?!?!?!??! Feb 20 19:35:27 do we actually need to post everything to IRC every 5 minutes via a bot?!?!?!?!?!! Feb 20 19:36:40 goxboxlive: http://www.openembedded.org/more-quality-assurance-tests Feb 20 19:37:42 yes i saw. thank you (koen: i'll read oe-dev before i ask next time) . Feb 20 19:39:06 goxboxlive: Don't worry too much... koen is just not the best at level-1 tech support :) Feb 20 19:39:37 :-) i know Feb 20 19:43:30 03pfalcon 07org.oe.dev * r0cdadd2b... 10/ (4 files in 2 dirs): linux-handhelds-2.6 2.6.19: Drop old releases. Feb 20 19:43:43 03pfalcon 07org.oe.dev * ree1a305d... 10/ (5 files in 2 dirs): linux-handhelds-2.6: Drop old releases. Feb 20 19:43:48 03pfalcon 07org.oe.dev * rcf5a4fa4... 10/ (1 conf/distro/angstrom-2007.1.conf): Feb 20 19:43:48 angstrom-2007.1.conf: Bump PREFERRED_VERSION_linux-handhelds-2.6 to use Feb 20 19:43:48 current release instead of old, removed. Feb 20 19:43:58 03pfalcon 07org.oe.dev * r9925d98b... 10/ (1 packages/linux/linux-handhelds-2.6_cvs.bb): linux-handhelds-2.6 cvs: Trunk is 2.6.20-hh0 now. Feb 20 19:50:41 gah Feb 20 19:51:08 how do I set the path to ipkg.conf at compile time? It defaults to /etc no matter what --prefix-stuff I set. Feb 20 20:00:38 03pfalcon 07org.oe.dev * r59b994b5... 10/ (6 files in 2 dirs): handhelds-pxa 2.4.19, handhelds-sa 2.4.19: Drop few older versions. Feb 20 20:02:56 psokolovsky__: why did you break angstrom? Feb 20 20:03:17 koen, how did I? Feb 20 20:03:21 yes Feb 20 20:03:33 please add back 2.6.16-hh9 Feb 20 20:03:42 since PREFERRED_VERSION_linux-handhelds-2.6 ?= "2.6.16-hh9" is breaking without that Feb 20 20:03:52 hi all Feb 20 20:04:21 koen, false alarm. 2.6.16-hh8 ws dropped. 2.6.19-hh9 too. Feb 20 20:04:44 was Feb 20 20:04:48 damn Feb 20 20:04:52 I'm getting dylexic Feb 20 20:05:00 psokolovsky__: sorry Feb 20 20:05:14 koen, np, I try to be careful, but can miss something too ;-) Feb 20 20:11:19 Does anyone mind if I commit an updated swig? Feb 20 20:12:03 what's swig? Feb 20 20:12:33 some wrapper extraction program :) Feb 20 20:12:47 sounds 1337 Feb 20 20:12:51 http://www.swig.org/ Feb 20 20:13:16 There is already a bb file, I just need a new version Feb 20 20:13:16 sometimes I regret not choosing CS Feb 20 20:13:56 CS is for lusers Feb 20 20:14:03 you will have a real degree Feb 20 20:14:08 EE, right? Feb 20 20:14:10 yes Feb 20 20:14:17 I have EE Feb 20 20:14:26 Crofton,koen: bah..ME's rule..:) Feb 20 20:14:56 I bet half the people here have non CS degrees Feb 20 20:15:09 for us, computers are tools Feb 20 20:15:50 hi Feb 20 20:16:12 what's EE ? Feb 20 20:16:21 non-CS here... I got into the field because the blasted things weren't doing what I needed them to do, so I had to write the code myself. Feb 20 20:16:29 Electrical Eng Feb 20 20:16:43 (EE also) Feb 20 20:16:51 mwester, where the blasted things are people with CS degrees? Feb 20 20:16:59 :) Feb 20 20:17:38 in italy we have Electric Eng, Electronic Eng, Computer Software Eng, Telecom Eng ... i'm computer software Eng, about but not the same of CS ... Feb 20 20:17:38 hehehe! CAD stations, at the time. great concept, weak software. Feb 20 20:17:54 hmmm, you must be old .... Feb 20 20:18:01 I'm ancient. Feb 20 20:18:10 Megatek? Calma? Feb 20 20:18:18 Medusa. Feb 20 20:18:20 I think those are words Feb 20 20:19:41 anyway, koen, see/ not such a bad major Feb 20 20:19:51 yeah Feb 20 20:19:58 somebody knows why tcl package can be complaining about not finding hypo symbol? Feb 20 20:21:04 hypot Feb 20 20:21:07 not hypo Feb 20 20:21:08 xD Feb 20 20:23:42 koen, we have swig versions 1.3.24,29,31 Should I delete .24 now? Feb 20 20:26:05 if you want Feb 20 20:26:37 I don't see any reason to keep three versions Feb 20 20:36:05 03crofton 07org.oe.dev * r8380c633... 10/ (3 files in 2 dirs): swig_1.3.31bb, swig-native_1.3.31 : Update to latest version of swig. Feb 20 20:36:12 03crofton 07org.oe.dev * r44bd9a4a... 10/ (3 files in 2 dirs): swig_1.3.24.bb, swig-native_1.3.24.bb : Remove old files. Feb 20 20:38:12 gremlin[it], conf/angstrom-*.conf Feb 20 20:39:20 greentux, conf/distro/angstrom-*.conf Feb 20 20:39:45 they son't contain the word 'insane' :( Feb 20 20:39:54 psokolovsky__: :) nice try. Feb 20 20:40:52 gremlin[it], why don't you make friends with grep -r? ;-) Feb 20 20:41:07 greentux, sorry ;-) Feb 20 20:41:17 gremlin[it], conf/distro/include/angstrom.inc Feb 20 20:41:26 psokolovsky__: 1 beer left @ fosdem beer event :) Feb 20 20:42:11 greentux, too bad I won't be there ;-) Feb 20 20:42:13 psokolovsky__ ... distro/include/angstrom.inc Feb 20 20:42:22 yup Feb 20 20:42:28 * psokolovsky__ psokolovsky Feb 20 20:42:33 damn Feb 20 20:42:59 time to sleep %) Feb 20 20:43:16 hahah :) Feb 20 20:43:19 good night psokolovsky Feb 20 20:43:23 'night psokolovsky Feb 20 20:44:24 gremlin[it], nope, that happens everyday ;-), i'm still online for some time, hope to get confirmation of good jffs2 size ;-) Feb 20 20:44:50 ok ... Feb 20 20:47:29 libc ok Feb 20 20:53:52 hrw|gone: go ahead Feb 20 20:56:36 I have what I'm sure is a dumb question... but I must have changed something, now all my ipks are ending up in tmp/deploy/ipk/armv5te/ instead of tmp/deploy/ipk, any ideas? Feb 20 20:58:57 psokolovsky got result ... still big 17M (instead of 18M of few dasys ago) ... Feb 20 20:59:39 gremlin[it], you killed task-base*.ipk? Feb 20 21:00:08 16.75Mbytes exactly ... Feb 20 21:00:16 mhhh seem yes .. retry ... Feb 20 21:01:09 gremlin[it], 16252928 Feb 18 18:29 Angstrom-gpe-image-test-20070218-h3600.rootfs.jffs2 Feb 20 21:01:41 gremlin[it], that's 15872K Feb 20 21:01:52 that's what I have Feb 20 21:02:51 gremlin[it], please build tar.bz2 too and be ready to send me its /usr/lib/ipkg/status Feb 20 21:07:52 could something have changed in the way OE builds packages in the last day or two? I can't figure out what I could have changed to cause this Feb 20 21:10:17 psokolovsky where i send ? Feb 20 21:10:43 gremlin[it], pmiscml at gmail com Feb 20 21:11:14 xumbi8, yep, there were changes. it's faq. see maillist Feb 20 21:11:31 psokolovsky: oh awesome, thanks a lot, I was going nuts! Feb 20 21:22:28 * cbrake wonders if there is a way to prepend a binary image with a 32-bit int that contains the size of the image in a shell script ... Feb 20 21:40:20 * koen fights with apache, php and mmcache Feb 20 21:41:36 mickeyl: are the openmoko-pim apps supposed to work? Feb 20 21:42:05 * koen can't get them to do something usefull, but that might be due to different base system Feb 20 21:43:08 no idea frankly Feb 20 21:43:14 there were so many things goin on Feb 20 21:43:21 i had no time to look into PIM Feb 20 21:43:37 but due to o-hand working on it I pretty much rely on them Feb 20 21:43:58 I figured Feb 20 21:44:04 looks nice, though Feb 20 21:44:21 I can confirm OH has them working... Feb 20 21:44:25 hi all Feb 20 21:44:38 Is there a 'best practices' document for package submission? Feb 20 21:45:06 RP: they work and communicate to EDS, but I can't seem to be able to either enter or import stuff Feb 20 21:45:32 koen: Built against omoko or OE? Feb 20 21:45:34 hillct: attach patches to bugzilla, people will comment on best practices from there :) Feb 20 21:45:40 heh Feb 20 21:45:46 RP: are you interested in a patch that get pxa2xx_pcmcia working with cpufreq? Feb 20 21:45:54 RP: OE metadata Feb 20 21:45:56 mreimer: yes :) Feb 20 21:45:56 what is the script that checks your bb files? Feb 20 21:46:10 koen: ok, but is there anyting i should read as a starting point? Feb 20 21:46:12 Crofton: contrib/oe-stylize.py ? Feb 20 21:46:30 hillct, run them through the script koen mentioned Feb 20 21:46:53 note that that script can mess up your recipes by reordering stuff Feb 20 21:47:01 :) Feb 20 21:47:08 K Feb 20 21:47:12 so back things up first Feb 20 21:47:14 RP: sent. it's simple. Feb 20 21:47:16 will do Feb 20 21:47:28 the thing that counts is that your stuff ends up attached to bugzilla Feb 20 21:47:33 Crofton: I'm a physicist by training btw, to add weight to your previous comment :) Feb 20 21:47:33 that makes sure it doesn't get lost Feb 20 21:47:41 yeh Feb 20 21:47:45 K Feb 20 21:47:56 CS guys study weird stuff Feb 20 21:47:57 hillct: we like plain-text and patches, so avoid tar if possible Feb 20 21:48:37 so if I have secondary files besides the recipies and patches, add as separate attachments? Feb 20 21:48:40 K Feb 20 21:48:40 but tarballs are fine if you have more than 2 or 3 files Feb 20 21:48:49 ah Feb 20 21:49:17 mreimer: Looks good and removes a nice chunk of code :). We should try and get cpufreq into mainline really if its stable and working... Feb 20 21:49:48 Oh, I should mention here. ARM kexec made mainline with a big but. The syscall number changed Feb 20 21:49:51 RP: Do you have any ideas what might be causing the framebuffer corruption on resume I am seeing in 2.6.20 on poodle ? Feb 20 21:50:04 hvontres|poodle: No :-( Feb 20 21:50:05 RP: agreed Feb 20 21:50:07 RP: so we need to patch kexec-tools? Feb 20 21:50:19 koen: right, and its them incomatible with previous kernels Feb 20 21:50:41 * koen gets out OLDEST_KERNEL Feb 20 21:50:53 g'night all ... Feb 20 21:51:02 'night gremlin[it] Feb 20 21:51:13 'night gremlin[it] Feb 20 21:51:27 mickeyl: http://scap.linuxtogo.org/files/bd97cb0350dcac3426fc905727b534c9.png :D Feb 20 21:51:37 OE's sections are insane, though Feb 20 21:51:54 I thought it was better to get the code into mainline despite the syscall change. We could have had that problem at any time anyway Feb 20 21:56:21 I take it oe-stylize.py just takes the bb filename as an argument. Feb 20 21:56:33 It would be handy if it generated a usage message Feb 20 21:57:32 hillct: don't take the output as gospel truth, though Feb 20 21:57:53 Its best thought of as a rough indicator... Feb 20 22:04:46 RP, so where did it get to? .21-rc? Feb 20 22:05:19 psokolovsky: you mean kexec? Feb 20 22:05:27 RP, yes Feb 20 22:05:49 psokolovsky: 2.6.20-gitsomething. It'll be in the as yet unreleased .21-rc1 Feb 20 22:06:06 RP, ok, thanks Feb 20 22:06:21 psokolovsky: I'm afraid I have some backlight changed queued which will cause some work to merge int hh.org too Feb 20 22:06:42 (changes you'll hopefully understand and agree with though) Feb 20 22:07:01 RP: to corgi_bl? Feb 20 22:07:12 koen: should new packages start at r0 or r1? Feb 20 22:07:51 r0 Feb 20 22:08:03 you can omit PR from new packages Feb 20 22:08:08 (it defaults to r0) Feb 20 22:08:14 psokolovsky: To the whole class - see git.o-hand.com linux-rpurdie-backlight Feb 20 22:08:34 K Feb 20 22:09:23 RP: ok, will look. I just hope that corgi_bl will lose dependency on sharp-sl.{h/etc}, like the patch we submitted to you ;-) Feb 20 22:09:23 Docs say MAINTAINER is a valid field/variable to be included in recipies, but the test script says no Feb 20 22:09:34 RP: and will be renamed eventually ;-) Feb 20 22:09:48 hillct: it got moved to a seperate file Feb 20 22:10:09 hillct: http://www.openembedded.org/repo/org.openembedded.dev/MAINTAINERS Feb 20 22:10:20 psokolovsky: Its still in my inbox although I wanted to get the core fixed up first. Now I have a way to push this stuff more easily, I stand a better chance of looking at that ;-) Feb 20 22:10:23 ah Feb 20 22:10:36 RP, nice! Feb 20 22:10:53 K Feb 20 22:11:06 psokolovsky: I doubt it will get renamed though, I kind of like the current name and in reality, the name doesn;t mean much ;-) Feb 20 22:11:39 RP, but people find it confusing ;-) Feb 20 22:11:50 Udev is failing suddenly. It happens during QA checking status. Anyone noticed simular problem? : http:pastebin.ca/365402 Feb 20 22:12:11 psokolovsky: Why make things totally easy? ;-) Feb 20 22:12:36 RP, I really hope you're joking ;-) Feb 20 22:12:54 psokolovsky: My main concern is apps in opie and other places which depend on that name. I've tried to get people to code generically but have failed Feb 20 22:13:03 goxboxlive: how recent is your OE checkout? Feb 20 22:13:11 3 minutes Feb 20 22:13:41 it's shouldn't do bb.fatal on that anymore Feb 20 22:13:54 RP, well, then it's nice chance to have userspace fix its stuff too ;-) Feb 20 22:15:27 psokolovsky: I just don't really want to break it for the sake of a name Feb 20 22:15:40 * koen can't remember how many times he said "no, don't hardcode /dev/input/event" Feb 20 22:16:15 koen: Probably a similar number of times to be saying make the applet use /sys/class/backlight/*/brightness Feb 20 22:16:19 s/be/me/ Feb 20 22:16:21 RP: ok, it's change not for tomorrow, and we can review if it's really hardcoded much Feb 20 22:16:39 RP: good thing I wrote the minilite code which does just that :) Feb 20 22:16:48 6 hours for ~10 lines of C.... Feb 20 22:17:04 psokolovsky: I can think of several places, some more public than others... Feb 20 22:17:09 RP: I don't think kexec is actually *used* in slugos yet. Feb 20 22:17:23 NAiL: I know some of your users use it? Feb 20 22:17:44 If they're using head, it's their problem ;-) Feb 20 22:17:55 NAiL: If its not official, thats a help :) Feb 20 22:18:18 the kexec patch wasn't in use by the last release ('bout a year ago) Feb 20 22:18:51 NAiL: I've given rwhitby a head up on it, lets see what he says... Feb 20 22:19:00 so if head kinda breaks, it's ok with me, as long as we can fix it before the next release. Feb 20 22:19:20 NAiL: As long as you match the kernel to the userspace tools, its fine Feb 20 22:20:02 Then there's no problem. We can fix the syscall number in the kernel, and the userspace stuff comes later on. Feb 20 22:20:39 NAiL: ok, I just want to make sure people know about it :) Feb 20 22:21:50 hi Feb 20 22:21:55 hi CoreDump|home Feb 20 22:28:54 wow. you're right. That script really hoses things up Feb 20 22:30:25 rwhitby: morning Feb 20 22:30:34 I was just discussing kexec with RP Feb 20 22:31:06 cool Feb 20 22:31:23 and CoreDump|home for the altboot support of kexec? Feb 20 22:31:47 * CoreDump|home just got online and still doesn''t have the serial cable Feb 20 22:32:06 Basically, we just need to agree when and what to change Feb 20 22:32:33 I know all the zaurus kernels have the kexec patch, what else does? Feb 20 22:33:30 RP: the main problem in the past was that the arm kexec tools ( or was that the kernel patch itself?) didn't support a custom CMDLINE. Which limited kexec at least on the Z's a lot Feb 20 22:34:14 CoreDump|home: That hasn't been addressed yet. Even if kexec support had that, its a bit problem on the Zs due to the bootloader being broken Feb 20 22:34:16 RP: hh kernels have some old version, consider not really used. Feb 20 22:35:00 CoreDump|home: I'd probably look at implementing it but we simply can't use it so I haven't... Feb 20 22:35:28 no problem there =) Feb 20 22:35:52 Getting something into mainline was my main aim at the moment. Extending the implementaiton in mainline is much easier :) Feb 20 22:36:10 RP true true Feb 20 22:36:18 indeed Feb 20 22:36:36 RP russell is constantly demanding changes that fit his world view :) Feb 20 22:37:13 Kristoffer: To be fair, if he asks for something to be changed, he normally has good reason Feb 20 22:37:24 RP I know, thats the downside. Feb 20 22:38:23 the problem with know-it-alls and smart-asses is that usually they _really_ know "it" all and _are_ smarter (in some cases) than you. Bloody annoying at times =) Feb 20 22:38:49 :D True Feb 20 22:39:58 The thing is I'm now telling people off for the things Russell used to NAK my patches for :) Feb 20 22:40:12 haha :) Feb 20 22:41:07 RP: as I said, it can be annoying when they are right ;) Feb 20 22:41:34 CoreDump|home: :) Feb 20 22:42:06 I only recently returned back to update the 2.6.20 for superh-3 (jornada processor) and found the entire IRQ structure had changed. So instead of fixing what I was suppose to, I had to start fixing alot of undeclared stuff. Hate when that happens Feb 20 22:42:12 I know what I'm talking about. People keep calling me smart-ass at work all the time :\ Feb 20 22:42:42 :) Feb 20 22:43:42 I'm surrounded by people who know a lot more than me about certain areas. Of course I know certain areas better than them too so it kind of cancels out :) Feb 20 22:44:56 Kristoffer: That is the big benefit in merging things into mainline. I had that with the Zaurus code at first, now most of it is in mainline, it gets automatic conversion Feb 20 22:45:12 Yes, getting it in was tough, but very much worthwhile from that PoV Feb 20 22:46:16 RP yes. I pointed it out to paul(Linuxsh maintainer) that the build was messed up and he simply stated that machines that receive maintenance builds Feb 20 22:46:42 I took it as a sign to start sending him lots of patches :) Feb 20 22:49:00 RP btw, may I ask a favor? I'm most likely going to send some patches to Russell tommorow night (or night after that). Got time before then to look at some patches, for obvious no-no's? Feb 20 22:49:31 I can have a quick glance, maybe not a full on review Feb 20 22:51:29 hi Feb 20 22:51:41 RP would greatly appreciate it. Will talk to you tommorow Feb 20 22:53:32 Kristoffer: Worst case, mail me. Tomorrow would be better than the day after as I'll be packing for FOSDEM Feb 20 22:54:31 RP: ixp4xx kernel has kexec patch in it Feb 20 22:55:41 RP roger Feb 20 22:57:27 rwhitby: Any problem with changing the syscall number and updating the kexec-tools PR at the same time? Feb 20 22:58:06 RP: happy to change at any time. Feb 20 22:58:34 matching upstream-to-be is always worthwhile. Feb 20 22:59:00 rwhitby: I'm wondering about now? :) Feb 20 22:59:22 now is good. Feb 20 23:00:11 ok, lets get on and do it. I think nslu2 and the zaurus are the two main users, anyone else will have to play catchup Feb 20 23:00:16 RP: if you point me to your commit, I'll do the same change to our ixp4xx kernel svn repo. Feb 20 23:00:31 (which will then flow through to OE kernel builds) Feb 20 23:00:38 rwhitby: Expect something in a few minutes :) Feb 20 23:01:32 I've got all day :-) Feb 20 23:03:10 rwhitby: I'm going to sleep in a minute ;-) Feb 20 23:05:54 rwhitby: I'm going to have to unpack some of these kernels to redo the diff so its going to talk slightly longer than that... Feb 20 23:11:47 Yay, Linus pulled the backlight tree :) Feb 20 23:12:04 is it possible to have more that one pacth applied to a src using the SRC_URI += "file://" stuff ? Feb 20 23:12:17 RP: quick question, are you planning on updating just .dev or is this going to go into .354x as well? Feb 20 23:12:43 hvontres|poodle: Just .dev atm. .354x will be a question for hrw Feb 20 23:12:47 im trying to add another patch to package adding another SRC_URI += file:// to the bb file but it fails Feb 20 23:13:18 saladino: Try file://some.patch;patch=1 Feb 20 23:13:28 saladino: You should be able to separate them by newlines, and add ;patch=1 to the end of the URI, from memory.. i.e. SRC_URI += "file://blah;patch=1 \ (newline) file://foo;patch=1" Feb 20 23:13:49 RP: and just so I understand this correctly, the syscall change only affects the kernel that kexec is run on, not the one that gets loaded next, correct? Feb 20 23:14:29 ahm Feb 20 23:14:35 newline is the key Feb 20 23:14:40 i'll try Feb 20 23:14:57 hvontres|poodle: Yes Feb 20 23:15:04 the 1 of patch= is like doing patch -p1 ? Feb 20 23:15:31 RP: ok... change away then :) Feb 20 23:15:55 saladino: No, that would be ;pnum=1;patch=1 Feb 20 23:15:58 saladino: patch=1 just means its a patch Feb 20 23:16:05 ahm Feb 20 23:16:05 ok Feb 20 23:17:08 RP: One other thing, you mentioned a while back you were having trouble getting the serial console on poodle to work. is that still broken? I am about to build a serial-cable to help with some more kerel stuff (poser levels and some of the 2.6.20 wiredness) Feb 20 23:17:54 ~s/poser/power Feb 20 23:18:11 hvontres|poodle: Yes, mine doesn't work properly. Not sure if thats my lead or the device Feb 20 23:19:21 RP: ok.... that'll make hardware debugging fun....:) Feb 20 23:19:59 rwhitby: http://www.rpsys.net/openzaurus/temp/kexec_convert.patch is basically what you need (this is an old kernel so lots of missing calls to fill in) Feb 20 23:20:12 RP: if mine works, would you like one too? I have enough parts to build two boards. and unlike the standard cable, you also get USB :) Feb 20 23:21:16 hvontres|poodle: I actually have the bits to make one too, just not the time to actually do it but thanks for the offer :) Feb 20 23:22:32 humm, i think something is not working well with various patches to patch, i have 2, when i put only one, no matter what, it works, when i put 2 and one specifically first it fails with a python exception, and with the other ones first it fails the patch but the file is pacthed Feb 20 23:22:42 SRC_URI += "file://error_print_progname.patch;patch=1 \ Feb 20 23:22:44 file://more_math.patch;patch=1" Feb 20 23:23:24 saladino: Do the patches overlap? Feb 20 23:24:17 nop Feb 20 23:24:40 diferent files Feb 20 23:25:44 saladino: It should work. I think the record for the most patched build is held by a kernel with over 50 patches... Feb 20 23:26:09 RP: that wouldn't be yours, would it ? :) Feb 20 23:26:19 hvontres|poodle: yes, 2.6.16 Feb 20 23:26:26 i think that i should work, but i cant find my error Feb 20 23:26:28 im in it Feb 20 23:26:47 saladino: What happens if you swap the order of the patches.. is it the patch that fails, or having 2 patches in SRC_URI? Feb 20 23:27:41 when the second pacth fails(the one that does not cause a python excep) the system gives me a console, in the working dir, i do patch -p1 NAbyss i don't understand Feb 20 23:28:34 with 2 patches in SRC_URI 2 behaviors Feb 20 23:28:39 depends on the order Feb 20 23:28:40 saladino: Trying to isolate the problem - see if it's a broken patch, or a broken SRC_URI Feb 20 23:28:53 or a python exception or a patch fail Feb 20 23:29:03 very simple patches Feb 20 23:29:13 the second one is mine Feb 20 23:29:19 not mine, but i have put it there Feb 20 23:31:20 i'll pastebin things Feb 20 23:35:14 hum Feb 20 23:35:16 wait Feb 20 23:35:22 something happened Feb 20 23:35:41 CoreDump|home: I might have a first cut of the "SD bootmangaer" patch for altboot next week. Feb 20 23:36:12 hvontres|poodle: nice Feb 20 23:36:21 RP: got a patch against 2.6.20 handy? Feb 20 23:37:25 i think its fixed Feb 20 23:37:31 some cleans and started working Feb 20 23:37:40 :? Feb 20 23:38:34 CoreDump|home: I just need to figure out how to "remember" the last kernel selected so it can auto-boot the the previous setup. Feb 20 23:38:54 rwhitby: If only you'd asked earlier. Not easily now :/ Feb 20 23:39:31 rwhitby: I can give you one with broken whitespace? Feb 20 23:39:58 hvontres|poodle: that can be tricky indeed and wasn't propperly implemented in altboot at all Feb 20 23:40:33 rwhitby: http://www.rpsys.net/openzaurus/temp/kexec_convert1.patch Feb 20 23:42:46 CoreDump|home: well, maybe I can "borrow" some of the stuff from the main menu. Feb 20 23:43:17 hvontres|poodle: that should work yeah Feb 20 23:43:42 RP: thx Feb 20 23:44:25 CoreDump|home: of course that only applies if you have more than one kernel on your SD card....:) Feb 20 23:44:36 hehe Feb 20 23:46:55 RP: could the display corruption be connected to EABI at all? Feb 20 23:47:05 hvontres|poodle: I'd doubt it Feb 20 23:47:37 RP: didn't think so...but I guess I could try rebuilding with OABI instead... Feb 20 23:55:01 How do I nuke a head don't want in monotone? Feb 20 23:55:34 RP: "Merge" it without taking any of the changes? Feb 20 23:56:34 NAbyss: No, the whole point is I don't want to upset koen ;-) Feb 20 23:57:22 RP: not upsetting koen is overrated :-) Feb 20 23:57:24 RP: Is it a local head, or a head someone else's pushed? Feb 20 23:57:59 NAbyss: Its local. I've found a rather nasty command which appears to have nuked it... Feb 20 23:58:09 rwhitby: :) Feb 20 23:58:12 Yeah, you'd have to use the db backend commands to nuke it Feb 20 23:58:39 RP: building with that patch now ... Feb 20 23:59:07 I'll give it the same amount of testing that I've given the previous kexec patch .... Feb 20 23:59:19 ... (none) Feb 20 23:59:49 (the work CoreDump|home is doing will be the first supported use of kexec in SlugOS) Feb 20 23:59:49 rwhitby: How much testing do you think the zaurus kernels are getting? ;-) Feb 21 00:00:10 tsk tsk tsk... Feb 21 00:02:06 RP: Hey, I do try to build one every couple of months...:) Feb 21 00:02:38 hvontres|poodle: I do test them a fair bit really, I'm just not testing this specific change on all of them... Feb 21 00:03:15 kexec change pushed Feb 21 00:03:42 03rpurdie 07org.oe.dev * r4345baf0... 10/ (61 files in 4 dirs): linux-rp: Drop 2.6.18, 2.6.19, patch correct kexec syscall number for other versions Feb 21 00:03:46 03rpurdie 07org.oe.dev * r833bbc29... 10/ (12 files in 11 dirs): kexec-tools: Update ARM patch for new official syscall number. Distro maintainers be warned, this needs to be changed in kernels.. Feb 21 00:06:52 Totally screwed that up :/ Feb 21 00:08:00 03rpurdie 07org.oe.dev * radcbf2a7... 10/ (12 files in 11 dirs): disapproval of revision '833bbc29d20408f96387089c35fe96e31685a167' Feb 21 00:11:47 and that is what happens if you type mtn commit -m "" .. when you mean . Feb 21 00:12:19 RP: time to catch some zzZZzz's :) Feb 21 00:12:26 03rpurdie 07org.oe.dev * r635d3a24... 10/ (3 files in 3 dirs): kexec-tools: Update ARM patch for new official syscall number. Distro maintainers be warned, this needs to be changed in kernels Feb 21 00:13:02 hvontres|poodle: yes :) Feb 21 00:13:21 RP: BTW, nice bike :) Feb 21 00:13:49 hvontres|poodle: Which one are you looking at? Feb 21 00:14:21 RP: I think it was the VFR Feb 21 00:15:43 hvontres|poodle: I thought you might be. I got a Triumph Daytona 675 in the middle of last year ;-) Feb 21 00:16:36 * hvontres|poodle is jealous... and desperately needs to find time to fix up his '78 CX500 Feb 21 00:16:41 I still like the VFR but the 675 has its attractions too ;-) Feb 21 00:17:43 I have a '72 MGB which is more than enough for fixing up. Since the car can't be counted upon I need a bike that can! Feb 21 00:18:19 The VFR is a '90 so getting on a bit I guess although it doesn't look it Feb 21 00:19:53 RP: hehe... I always thought it was ironic that a british company could come up with electrical parts that don't tolerate moisture very well...:) Feb 21 00:20:32 RP: I refer of course to Lucas, Lords of Darkness :) Feb 21 00:21:28 n8 Feb 21 00:21:37 hvontres|poodle: I know. Thankfully the MG has a more modern wiring loom from when it was converted to RH drive (it was originally an american export, reimported) Feb 21 00:21:42 'night CoreDump|afk Feb 21 00:22:45 RP: That sounds almost as daft as the bloke who re-imported a 1990 Chevy Blazer to the states in 2005 :) Feb 21 00:23:19 hvontres|poodle: It made sense at the time, trust me ;-) Feb 21 00:27:02 RP: Ok... now go get some sleep :) Feb 21 00:31:22 hvontres|poodle: Just going ;-). 'night all **** ENDING LOGGING AT Wed Feb 21 02:59:57 2007