**** BEGIN LOGGING AT Fri Nov 24 03:00:00 2006 Nov 24 03:36:03 03kristoffer 07org.oe.dev * r1c3aaf97... 10/ (1 conf/distro/jlime-donkey.conf): Nov 24 03:36:03 conf/distro/jlime-donkey.conf: Addition of more RDEPENDS Nov 24 03:36:03 * Adding tslib-calibrate console-tools and tslib-tests Nov 24 03:55:25 hey Nov 24 04:08:23 can anyone confirm my irc-client not using colors now? Nov 24 04:27:39 eugenebiz, looks ok Nov 24 04:27:49 10x Nov 24 04:28:04 try using my nick in a line Nov 24 04:28:24 10Crofton:01 Hi Nov 24 04:29:05 hmm, maybe it is colourizing my nick? Nov 24 04:29:15 eugenebiz, is this a diff color Nov 24 04:29:43 Crofton: Oh, now i understand, i am copy-pastign nick witch is colored :-) Nov 24 04:30:08 heh Nov 24 04:30:27 other peoples colours are annoying in irc ... Nov 24 04:30:32 Crofton: Now i know my mistake, 10x for pointing out. Nov 24 04:30:53 np Nov 24 04:31:10 good night Nov 24 04:31:20 good night Nov 24 07:05:35 morning ... Nov 24 07:48:53 hi Nov 24 07:58:55 I'm trying to compile task-bootstrap for the first time but I'm getting an error. ERROR: function do_install failed ERROR: log data follows (/usr/src/openEmbedded/build/tmp/work/ipaq-pxa270-linux/initscripts-1.0-r82/temp/log.do_install.7144) | /usr/src/openEmbedded/build/tmp/work/ipaq-pxa270-linux/initscripts-1.0-r82/temp/run.do_install.7144: line 530: do_install: command not found ERROR: TaskFailed event exception, aborting Nov 24 07:59:15 is this a known bug or have I done something wrong/left something out? Nov 24 08:05:51 NOTE: package initscripts-1.0: failed Nov 24 08:11:52 zezom: both 'task-bootstrap' and 'ipaq-pxa270' are depricated Nov 24 08:12:26 koen, oh, I got task-bootstrap from the wiki tutorial Nov 24 08:12:33 what should I be using? Nov 24 08:12:47 task-base Nov 24 08:12:57 also ipaq-pxa270 is what I though was the closest to my ipaq rx4540 Nov 24 08:13:15 thanks Nov 24 08:13:38 sorry to be a pain but what would be a good type to use for a ipaqrx4540? Nov 24 08:13:53 I was told to just find one what has the same cpu. Nov 24 08:15:49 and if I change from ipaq-pxa270 to something else will I have to recompile everything? Nov 24 08:41:15 morning Nov 24 08:44:06 hey XorA Nov 24 08:44:10 hey pH5 Nov 24 08:44:25 hey koen Nov 24 08:46:35 * XorA finally remebers some C Nov 24 08:53:38 * hvontres|poodle waits for angstrom opie image to build....... Nov 24 08:54:08 * hvontres|poodle goes to bed ... too much food :) Nov 24 09:16:09 does OE/bitbake team has plan to create a GUI tool to maintain OE meta data? Nov 24 09:18:07 moring Nov 24 09:19:05 it's not easy for new user to get familiar with OE, because it has so many features Nov 24 09:19:12 O_Neil, : zecke has some code Nov 24 09:19:29 hey , OE is for developer so ... Nov 24 09:19:49 yes Nov 24 09:20:43 When I recommend it to another department, they think it's too complex Nov 24 09:22:08 lazyness, its not hard to write like 4-5 lines of the form XXX="uuu" Nov 24 09:22:23 sure Nov 24 09:22:42 they said, " it has too many keywords" Nov 24 09:22:57 I have half a dozen recipes that are literally SRC_URI = "blah" inherit autotools, now thats not complex Nov 24 09:23:46 XorA, : i spend more that a week to try to make a recipe for gnash , and i still on it Nov 24 09:24:02 it's not so easy Nov 24 09:24:31 maybe a GUI tool may make OE more simple Nov 24 09:24:36 but the real pb i think is autotools miswritter :) Nov 24 09:25:07 Genesis: you have a special cash, in englesh we would say Gnash is Gash :-) Nov 24 09:25:18 Gash meaning rubbish Nov 24 09:26:11 * XorA wonders what a GUI would deliver that gvim doesnt Nov 24 09:26:31 there is afterall a template .bb in the docs Nov 24 09:26:58 sure, for experienced OE developer Nov 24 09:28:33 the department I said is Cyclades :) Nov 24 09:28:39 good morning folks Nov 24 09:29:14 they have a build system written purely in makefile Nov 24 09:29:59 of course it's not good as OE, but it's written by themselves Nov 24 09:30:19 morning all Nov 24 09:32:00 It's hard to persuade them to adopt OE because it's complex for new user Nov 24 09:33:25 christ, if someone can build something in the hideuosness that is make then OE should be easy :-) Nov 24 09:34:25 :-) Nov 24 09:35:03 * XorA once wrote a complex build system in make/nmake/ant because he couldnt work out how to get bitbake going on windowz Nov 24 09:35:07 never again Nov 24 09:48:20 good morning Nov 24 09:48:47 why "bitbake -e pkgname" only returns global environment variables? , I think it should return variables related to the package Nov 24 09:51:46 O_Neil: Try it with -b pkgname Nov 24 09:51:56 I know Nov 24 09:52:56 sometimes you don't know the filename Nov 24 10:00:26 bbl Nov 24 10:03:43 can any one else compile the qemu-qop-nogfx-native-0.8.0+svn20061124-r0 package? I keep getting ssl errors.... and when I go to the web site their is a ssl cert mis match... is their a way to get around it? Nov 24 10:04:26 * koen uses qemu-native 0.8.2 Nov 24 10:05:19 koen, ok, qemu-qop-nogfx-native was a dep of task-base... Nov 24 10:05:27 ? Nov 24 10:05:39 I'll try installing qemu-native and see if it realises it's installed Nov 24 10:06:16 when I did a bitbake task-base it tried to install qemu-qop-nogfx-native Nov 24 10:06:54 koen, is hx4700 ok to use? Nov 24 10:07:49 that should work Nov 24 10:07:49 qemu-native failed for the same reasons Nov 24 10:07:49 "same reasons" Nov 24 10:07:49 * koen gets a crystal ball Nov 24 10:08:29 NOTE: Fetch svn://nowt.dyndns.org/svn/qemu;module=trunk;proto=https svn: PROPFIND request failed on '/svn/qemu/trunk' svn: PROPFIND of '/svn/qemu/trunk': SSL negotiation failed: SSL disabled due to library version mismatch (https://nowt.dyndns.org) Nov 24 10:08:39 ssl issues Nov 24 10:08:55 thats not qemu-native Nov 24 10:09:39 bitbake qemu-native, not qemu-qop-nogfx-native Nov 24 10:09:57 NOTE: package qemu-qop-nogfx-native-0.8.0+svn20061124: failed ERROR: Build of qemu-native failed I did bitbake qemu-native Nov 24 10:10:27 why would it be trying to installing the nogfx version if I specified qemu-native? Nov 24 10:10:53 sounds like your distro is lacking the defines for that Nov 24 10:11:04 is your distro conf only picking on that version Nov 24 10:11:17 ../../conf/distro/include/angstrom.inc:28:PREFERRED_PROVIDER_qemu-native = "qemu-native" Nov 24 10:11:48 I'll check Nov 24 10:12:36 what DISTRO setting do you have now? Nov 24 10:13:41 DISTRO = "familiar-unstable" Nov 24 10:13:53 right, that's unmaintained Nov 24 10:14:13 I'm still really new to open embedded so I don't really know anything that's out side of the wiki tutorial Nov 24 10:14:20 can you suggest a better one? Nov 24 10:14:26 koen: might be an idea to drop the qop-nogfx version, it was only put in as an experiment for gcc4 Nov 24 10:14:29 try DISTRO="generic" or DISTRO="angstrom-2007.1" Nov 24 10:14:44 thanks Nov 24 10:15:00 are their any significent diffrents between them? Nov 24 10:16:18 yes, those distro actually work :) Nov 24 10:16:48 I'm glad to hear that :) Nov 24 10:16:56 but I was meaning between the two that you listed Nov 24 10:18:21 Ok, a newbie qustion :-) When openmoko hits oe.dev, will we be able to build it for other platforms than the openmoko phones it self? Like will I be able to run it on my HTC Universal? Nov 24 10:23:19 goxboxlive: well... i guess so, in theory at least. i guess you need to touch some hardware interfaces and the uis Nov 24 10:24:06 florian_kc : cool. Nov 24 10:24:17 03koen 07org.oe.dev * r5c22db77... 10/ (1 removal.txt): removal.txt: add task-bootstrap* Nov 24 10:54:26 Good Morning Nov 24 10:54:46 I'm having problems with GCC building, as always Nov 24 10:55:12 It fails with a "file not found" error during installation Nov 24 10:55:31 Checking it tells me that "fix-header" was not built Nov 24 10:55:53 And as such, is missing Nov 24 10:56:18 Compilation went on great, no apparent errors Nov 24 10:56:57 Searching the gcc source directory gives me the fix-header source code, but no .o or binary files Nov 24 10:57:15 That tells me that it was just skipped, for some reason Nov 24 10:58:10 The makefiles contain enough references to fix-header to warrant building it Nov 24 11:06:27 So, has anyone encountered anything like this? Nov 24 11:07:22 Compiling it manually is pretty much out of the question, since I don't know what to link against Nov 24 11:31:38 re Nov 24 11:33:09 meetings suxx Nov 24 11:33:18 hey hrw :) Nov 24 11:36:56 hi schurig Nov 24 11:37:40 hrw|work: hi! Nov 24 11:37:43 hrw|work: the card arrived Nov 24 11:38:14 unfortunately I don't have much time today. A database server crashed at a customers site and I have to help setting up a new one ... Nov 24 11:38:28 schurig: prpplague also worked on this driver Nov 24 11:38:58 hrw|work: shouldn't we then make some irc-channel and a svn repository for our work? Nov 24 11:39:33 schurig: looks like Nov 24 11:39:48 hrw|work: any better idea than Berlios? Nov 24 11:40:05 schurig: ltg? Nov 24 11:40:14 ltg? Nov 24 11:40:17 ~ltg Nov 24 11:40:40 linuxtogo.org Nov 24 11:40:57 that would be fine Nov 24 11:48:56 hrw|work: or marvell8385 ? Nov 24 11:49:39 also good name for me Nov 24 11:49:51 maybe even better Nov 24 11:50:41 03koen 07org.oe.dev * rc09d5f93... 10/ (1 packages/tinymail/tinymail_svn.bb): tinymail: drop eds-dbus, since tinymail has its own camel now Nov 24 11:50:41 #gmen the superheroes bringing G to PDA Nov 24 11:51:14 XorA: G-spot? Nov 24 11:51:53 hrw|work: heh heh Nov 24 11:52:10 hrw|work: but most men are apparently unable to find that, so thats failing before you start Nov 24 11:56:44 hrw|work: I've submitted the project Nov 24 11:57:34 ok Nov 24 11:57:35 brb Nov 24 11:58:18 koen: Maybe pvanhoof should rename his camel to "dromedary" ;) Nov 24 12:01:00 * koen looks at http://projects.linuxtogo.org/projects/marvell8385/ Nov 24 12:03:30 koen: webpage of project looks familiar Nov 24 12:03:42 s/familiar/like something known Nov 24 12:04:00 * koen looks at florian_kc Nov 24 12:04:44 ah cool Nov 24 12:06:01 schurig: will we keep patches in SVN or import marvell sources into it (which can be not necessary good because of lack of license in them)? Nov 24 12:06:07 i can add a vhost or you could just use marvell8385.home.linuxtogo.org which uses the htdocs in /home/groups/marvell8385 Nov 24 12:20:30 hrw|work: I would keep the sources there Nov 24 12:20:34 hrw|work: but I am not a lawyer Nov 24 12:21:12 hrw|work: as far as I understand it: if I can download something without a password, without an EULA or anything other that restricts me in usage, them I'm free to use it like I want Nov 24 12:21:52 hrw|work: I know that there are exceptions, e.g. pictures, maps and art are always protected my copyright law (at least in Germany), but I'm very sure that this is not the case for software Nov 24 12:22:01 hrw|work: but again, IANAL Nov 24 12:22:15 s/my/by/ Nov 24 12:22:44 schurig: server is also in germany Nov 24 12:23:39 code will have to have a total re-write before it ever goes in mainline Nov 24 12:23:40 maybe I should ask at the german branch of the FSF ? Nov 24 12:24:05 XorA: this will rather never go into mainline Nov 24 12:24:15 schurig: if you can Nov 24 12:24:18 hrw|work: thats a pity Nov 24 12:24:21 XorA: I agree. And I think this speaks more for the "keep sources in SVN" than for the "keep patch in SVN" approach, because otherwise we'd have 200 patches in SVN :-) Nov 24 12:24:55 with no license code would have to 100% written by you two guys so you could prove ownership Nov 24 12:33:41 hmm, actually the code *HAS* a copyright. It says "Copyright (c) Marvell Semiconductor, Inc., 2003-2005". Nov 24 12:33:51 And the PDF file says "Confidential". Nov 24 12:34:14 I always thought that the driver is totally without anything Nov 24 12:35:22 in UK law I think having the word confidential there would make you pretty screwed to use that code at all, unless you can find a license that overrides that Nov 24 12:36:11 best bet is probably an email to Marvell asking for a license, suggesting GPL, when they answer yes, put that email into the svn alongside the code Nov 24 12:37:24 XorA: I agree Nov 24 12:38:11 XorA: and additionally, we migth contact OLPC people. At Marvells website you can download drivers. And the search box explicitly mentions "OLPC" as a category. So I guess there are some (good) relationships between this company and the OLPC project Nov 24 12:39:16 schurig: contacting OLPC is probably good as well, as rumour has it Marvell actually talk to them Nov 24 12:39:33 OLPC use USB version: 8388 iirc Nov 24 12:42:01 hrw|work: yes, they use USB. I just downloaded their USB driver. And all files that start with wlan_* are similar to the CF card driver. But they bear a GPL license. So we could start with the WLAN_* files and write our CardServices stuff around that ?!? Nov 24 12:42:32 sounds like a safe bet Nov 24 12:43:03 I'll join in few moments Nov 24 12:46:04 btw, prpplague is in the process of moving back to DFW Nov 24 12:46:57 DFW? Nov 24 12:47:08 Dallas-Fort Worth area Nov 24 12:51:09 There where an article in a Norwegian internet web paper abot OPenmoko. Openembedded where also mentioned. Nov 24 12:51:27 newpaper i mean Nov 24 12:51:30 +s Nov 24 12:51:39 cool Nov 24 12:51:50 :-) Nov 24 12:52:22 goxboxlive: which one? Nov 24 12:52:31 itavisen Nov 24 12:52:55 someone should start linking these articles on OE.org Nov 24 12:53:32 goxboxlive: hmm.. Can't see it. Nov 24 12:53:34 NAiL: No sorry, digi: http://www.digi.no/php/art.php?id=358522 Nov 24 12:53:38 ah, ok ;) Nov 24 12:54:31 XorA: o yes Nov 24 12:54:57 "Ideen er hentet pakkehåndteringsverktøyet apt-get, først og fremst kjent fra Debian-baserte Linux-distribusjoner. " Nov 24 12:55:09 heh Nov 24 12:55:17 I guess sean's presentation was a bit more confusing than he wanted to Nov 24 12:55:55 "OpenMoko-plattformen er også basert på OpenEmbedded, et rammeverk å lage Linux-distribusjoner rettet mot integrerte enheter." Nov 24 12:56:05 koen: for some reason everywhere keeps posting the OE produces debian based distros Nov 24 12:56:11 anything odd about the last quote? Nov 24 12:56:11 I guess 'rammeverk' means framework Nov 24 12:56:17 koen: indeed Nov 24 12:56:27 'raamwerk' in dutch Nov 24 12:56:30 heh Nov 24 12:56:53 koen: Do you understand the article? Nov 24 12:59:53 most of it Nov 24 13:00:41 hmm, impressing. Nov 24 13:06:05 norwegian is quite similar to german and dutch Nov 24 13:08:54 python-egenix-mx-base requires patch "gcc4.patch", which can't be found Nov 24 13:09:28 I remember something similar some time ago, but the patch was merely in the wrong folder Nov 24 13:09:33 Now, I can't find it anywhere Nov 24 13:11:51 I consider this pretty much serious, since it halted my "bitbake -k world" process Nov 24 13:11:55 :| Nov 24 13:14:31 XorA hrw|work: Okay, I've made an inquiry to Marvell. Maybe I have to do another one in written form, too. Nov 24 13:14:41 Someone should consider doing something about this Nov 24 13:15:07 B_Lizzard: you like filing a bug report? Nov 24 13:15:19 Ok... Nov 24 13:15:22 I'll do so Nov 24 13:15:43 BTW, it built ok with gcc 4.1.1 when I commented the patch out Nov 24 13:15:54 So it might not be required Nov 24 13:16:40 I'll just add a note about it in the bug tracker Nov 24 13:17:43 Also, can I resume the build process if something causes it to abort? Nov 24 13:18:00 Cause currently it has to go through the whole package list again Nov 24 13:18:12 And fail those that fail, etc, etc Nov 24 13:26:02 hrw|work: any idea? I think you should also know the list passwords :-) Nov 24 13:36:05 hey mickeyl Nov 24 13:36:30 hi guys Nov 24 13:40:19 hi mickeyl Nov 24 13:41:41 hey mickeyl Nov 24 13:50:22 ~lart fontconfig for calling host gcc with arm arguments Nov 24 13:50:23 * ibot breaks out the Hoover and sucks up fontconfig for calling host gcc with arm arguments Nov 24 13:51:05 morning all Nov 24 13:51:11 hey RP Nov 24 13:51:32 yo RP Nov 24 13:52:05 mickeyl, hrw|work: Have either of you any objections to a namechange on linux-openzaurus. At the moment, the best I can come up with is linux-rp although I'm open to suggestions Nov 24 13:52:27 linux-openRP Nov 24 13:53:02 linux-allthepatchesthatmakekernelusefullbutwererejected.bb Nov 24 13:53:15 RP: not at all. linux-rp sounds appropriate. go for it Nov 24 13:53:29 linux-openzaurusandipaqsandcompulabandqemuarmandmaybesomemore Nov 24 13:54:13 koen: you missed out poky (which is part of the reason a change is needed) Nov 24 13:54:34 ;-) Nov 24 13:54:49 * koen digs out the "told you so" Nov 24 13:54:53 RP: linux-rp is ok Nov 24 13:55:12 I'd like openezx use linux-rp eventually Nov 24 13:55:22 once that stuff gets sent to mainline *cough* Nov 24 13:55:29 linux-oe :-D Nov 24 13:55:43 XorA: too generic Nov 24 13:56:00 hm.. linux-progear or add progear into linux-rp ;d Nov 24 13:56:01 hrw|work: its getting very generic Nov 24 13:56:17 I've only delayed this as long as I don't like the name linux-rp that much. It does describe it reasonably accurately though :) Nov 24 13:56:53 I guess it is turning into linux-oe :) Nov 24 13:57:10 packages/linux/linux_2.6.18.bb Nov 24 13:57:27 linux-notcrappix Nov 24 13:57:29 the amount of kernel has always amazed me Nov 24 13:57:43 OE should be able to handle 1 .bb per kernel version Nov 24 13:57:53 and have machine + distro overrides sort the rest Nov 24 13:58:18 koen: Some patches prevent that but the current situtation is a bit worse that it should be Nov 24 13:58:21 kernel guys are fussy Nov 24 13:58:23 linux-supercalifragalisticexpialodotious.bb Nov 24 13:58:46 XorA: DISTRO=marypoppins-1.0"? Nov 24 13:58:59 koen: parse error Nov 24 13:59:15 linux-rp reminds me Nov 24 13:59:22 damn koen watches far too much tv Nov 24 14:01:03 RP: is arch/arm/mach-pxa/akita-ioexp.c usable as example? Nov 24 14:01:19 hey Nov 24 14:01:20 suihkulokki: an example of what? Nov 24 14:01:23 hi Liam Nov 24 14:01:31 * XorA hides from lrg Nov 24 14:01:33 hey Richard Nov 24 14:01:46 XorA: what's happened Nov 24 14:02:05 XorA: Its ok, we won't say anything ;-) Nov 24 14:02:05 lrg: no idea, just on principal Nov 24 14:02:11 heh Nov 24 14:02:43 mickeyl: the bottom comments of http://blog.gustavobarbieri.com.br/2006/11/canola-i-said-its-fast.html might interest you Nov 24 14:02:54 * lrg is having to deal with power cuts at his hotel :( Nov 24 14:02:56 RP: we have a generic I2C led/gpio chip connected to thecus n2100 Nov 24 14:03:05 hey lrg Nov 24 14:03:09 lrg: met Harald yet? Nov 24 14:03:13 hey mickeyl Nov 24 14:03:26 RP: but the how the pins are connected is completly device specific, ie some pin is led, another is powerdown, etc Nov 24 14:03:35 mickeyl: not yet, I was too busy to attend today. I'm going tomorrow Nov 24 14:04:07 suihkulokki: I'd try and write a generic i2c driver and then some machine specific glue if possible Nov 24 14:04:14 koen: yeah. the more i see EFL progressing, the more I know that it'll might be more useful for certain UI styles than Gtk Nov 24 14:04:32 suihkulokki: akita-ioexp was a bit of a hack (but there was no other way to do it) Nov 24 14:07:03 RP: ok, thanks. Nov 24 14:18:15 ~seen kergoth Nov 24 14:18:37 kergoth is currently on #wowi-lounge (2d 20h 5m 4s) #oe (2d 20h 5m 4s) #wowace (2d 20h 5m 4s) #elinux (2d 20h 5m 4s). Has said a total of 1320 messages. Is idling for 6h 54m 41s, last said: 'ugly, but what can ya do'. Nov 24 14:40:35 morning Nov 24 14:41:11 hi Nov 24 14:42:14 hi chouimat, CoreDump|home Nov 24 14:49:23 is anyone else having problems w/ zaurusd in the .oz354x branch? Someone reported it b0rking for c7x0 builds Nov 24 14:53:42 Genesis: I succesfully installed mplayerplug-in but it has some heavy hick up Nov 24 14:56:29 koen: Can you remember the paths ncurses installed arm binaries too if they ended up in PATH? Nov 24 15:13:39 RP: no, sorry Nov 24 15:14:58 koen: I'm a bit worried if it happened with the standard PATH :-/ Nov 24 15:16:29 koen: are you ok with commiting the mx31 fix ? Nov 24 15:16:31 * hrw|work wants do_configure_pci() Nov 24 15:16:52 lrg: no, since the code works in other kernel recipes as well Nov 24 15:16:58 hrw|work: If we do that, we loose multimachine Nov 24 15:17:10 lrg: linux-oz and linux-handhelds use the same if check Nov 24 15:17:10 gerwinin, : it works on my board the m*plugin Nov 24 15:17:18 koen: ah I see, so its not a bug :) Nov 24 15:17:32 lrg: it could be a bashism Nov 24 15:17:35 RP: if distro use pci (on all machines) then we do not loose Nov 24 15:17:36 gerwinin, : which FF ? Nov 24 15:17:37 hehe Nov 24 15:18:53 RP: so having DISTRO_FEATURES only in overrides gives us working multimachine anyway Nov 24 15:19:05 normally people do not run multidistro builds Nov 24 15:19:27 hrw|work: Right, I thought you were referring to COMBINED_FEATURES Nov 24 15:19:53 RP: firstly yes but I has some thinking then Nov 24 15:20:18 I'm still uneasy about all the possible combinations of options this could introduce though... Nov 24 15:21:16 All of a sudden my parse_pci_function_ids(): function gets renamed... Nov 24 15:22:11 RP: this will add some stuff yes, but also allow to keep wpa-supplicant (example) not duplicated into pci and !pci ver Nov 24 15:22:49 hrw|work: Can you not look in DISTRO_FEATURES without using overrides or just add it to overrides locally? Nov 24 15:23:40 DEPENDS = "gnutls \ Nov 24 15:23:40 ${@base_contains("DISTRO_FEATURES", "pci", "madwifi", "",d)}" Nov 24 15:23:41 it would be sort of ok if do_configure_ would trigger PACKAGE_ARCH = ${MACHINE} Nov 24 15:23:43 that way? Nov 24 15:24:24 hrw|work: That kind of thing, yes Nov 24 15:24:31 I can Nov 24 15:24:42 koen: Thankfully DISTRO_FEATURES aren't machine specific Nov 24 15:24:45 * chouimat wants for xmas a linux based portable mediaplayer with a 500GB HDD :) Nov 24 15:24:46 and when find time I will push my wpa-supplicant 0.5.5 which use such thing Nov 24 15:25:04 chouimat: running for 2 weeks on 2 AA batteries? Nov 24 15:25:13 hrw|work: 6 months ;) Nov 24 15:25:24 chouimat: be realistic - 2 weeks Nov 24 15:25:31 nonstop Nov 24 15:25:32 RP: hrw's example would break angstrom Nov 24 15:25:50 RP: since MACHINE=guinness needs pci, while other machines have pci disabled Nov 24 15:25:52 hrw|work: ok 2 weeks ... I don't care if it work on fuel cell ;) Nov 24 15:26:21 koen: but you do not have wpa-supplicant for same ARCH for pci/nocpi rather? Nov 24 15:26:38 hm. nslu2 == pci and arm. so you have Nov 24 15:26:52 loft has pci Nov 24 15:27:04 and madwifi doesn't build if you have pci support disabled in the kernel Nov 24 15:27:19 I know Nov 24 15:27:28 so machine=c7x0 would build madwifi, since DISTRO_FEATURE has pci Nov 24 15:27:31 and that would break Nov 24 15:27:58 So DISTRO_FEATURE is incorrect there, it needs to be MACHINE_FEATURE or COMBINED_FEATURE Nov 24 15:28:09 exactly Nov 24 15:28:19 or wpa-supplicant-nopci + wpa-supplicant-pci Nov 24 15:28:56 the question is.. will wpa-supplicant-madwifi^Wpci run on !pci machine Nov 24 15:29:09 not solution it is ofcourse Nov 24 15:29:36 COMBINED_FEATURE + triggering PACKAGE_ARCH=MACHINE when true Nov 24 15:30:02 that would give a wpa-sub_armv5te + wpa-sub_loft Nov 24 15:34:05 http://pastebin.ca/256731 kind of? Nov 24 15:35:28 that's what I had in mind Nov 24 15:35:37 lets hope it works Nov 24 15:36:00 oh, and it needs to get stylized Nov 24 15:36:22 can you test it? Nov 24 15:36:35 not right now Nov 24 15:36:38 maybe tonight Nov 24 15:37:01 no problem Nov 24 15:41:32 o... Nov 24 15:41:35 http://fedoraproject.org/extras/6/i386/repodata/repoview/bitbake-0-1.6.0-1.fc6.html Nov 24 15:41:52 ~lart me for not checking it before Nov 24 15:41:53 * ibot pries hrw|work's back open with a screwdriver and flashes a new bootldr to hrw|work for not checking it before Nov 24 15:43:10 "Bitbake is a simple tool" ... Simple? Nov 24 15:43:34 a powerful tool would be better :-) Nov 24 15:43:35 it is. the complexity comes from OE Nov 24 15:43:54 schurig: http://developer.berlios.de/projects/bitbake say so Nov 24 15:44:28 "a simple tool for complex people"? Nov 24 15:44:33 * mickeyl chuckles Nov 24 15:44:39 naa, it's not only for you Nov 24 15:44:41 * schurig ducks Nov 24 15:44:59 or "takes cover", not sure what is better english Nov 24 15:45:18 mickeyl: bitbake is not simple. Anyone who's studied the dependency graphs will realise this ;-) Nov 24 15:46:15 "simple to use" Nov 24 15:46:22 not "simple algorithms" Nov 24 15:46:29 "simple to confuse" Nov 24 15:47:04 everyone knows that it's very complex to make it simple to use. and bitbake does a good job at that imo Nov 24 15:47:15 mickeyl: right :) Nov 24 15:47:20 it just needs a gui Nov 24 15:47:32 and sound effects Nov 24 15:47:35 the gui needs time Nov 24 15:47:40 it has: vi, emacs, joe, kate, whatever :-) Nov 24 15:47:56 and mistical module to order pizza Nov 24 15:48:09 schurig: Have you looked at the output from bitbake multithreading? Nov 24 15:48:30 bitbake -c do_pizza Nov 24 15:48:50 RP: no, not really. I'm still with single-core CPUs of the 2.4 GHz class Nov 24 15:49:08 schurig: We need a gui to make any sense of it... Nov 24 15:49:18 RP: usually not even make -j makes much sense there Nov 24 15:50:07 schurig: multithreading will make sense even on single core for things like do_fetch once we write a decent threading algorithm Nov 24 15:51:51 RP: yeah, but I mostly never kill my downloads directory. For a new project, I symlink another downloads directory into the current tree. And then sometimes I did just something like "bitbake -c do_fetch" in one console while doing stuff in another console Nov 24 15:52:57 Hello Nov 24 15:53:11 schurig: me too, but its a nice improvment if we don't have to worry about that :) Nov 24 15:53:16 hrw|work: the original makefiles are a mess Nov 24 15:53:17 hi sirfred Nov 24 15:53:31 RP: I agree. Maybe I'm just old-fashioned :-) Nov 24 15:54:03 hrw|work: should we just use a 2.6 makefile? Or do you need this driver also for 2.4 kernels? Nov 24 15:54:53 schurig: collie ;( Nov 24 15:55:11 schurig: but it can be done separatelly Nov 24 15:55:51 Okay, then I'm making a makefile with the usual "make modules -C $(KERNELDIR) SUBDIRS=`pwd`" approach then Nov 24 15:56:08 hrw|work: have you seen my readme? Already in SVN. Nov 24 15:56:13 not yet Nov 24 15:56:26 schurig, RP: I use DL_DIR in site.conf - simpler Nov 24 15:56:59 hrw|work: I do as well ;-) Nov 24 15:57:00 hrw|work: probably, but I like the symlink of it in my current dir. Shorter paths if I want to "ls" something Nov 24 15:58:27 All my machines have /usr/oe/sources. On the LAN this is an nfs drive so its shared between all machines... Nov 24 15:58:27 schurig: nice readme Nov 24 15:58:58 In /usr as at the time it was the only drive with space on the nfs server :} Nov 24 15:59:14 mickeyl: we have 2 ciabots for OE now: one for email and one for xmlrpc Nov 24 15:59:28 mickeyl: since the xmlrpc server has been misbehaving a lot lately Nov 24 15:59:58 koen: ok, thanks for letting me know. I'm still sad we didn't go the async threads route, but ok ;) Nov 24 16:23:17 bye Nov 24 16:33:33 I've just seen a qtopia-core 4.2.0 package under oe. Is it under GPL license ? Nov 24 16:33:52 sirfred: qtopia-core is qtembedded Nov 24 16:34:15 ~qtopia core Nov 24 16:34:26 rumour has it, qtopia core is qt/embedded. please don't support stupid marketing tricks. Nov 24 16:34:33 :) Nov 24 16:34:45 greetings to all Nov 24 16:35:16 56Mb of source code Nov 24 16:39:24 is there a way to force oe to use uclibc-svn instead of a specific version ? Nov 24 16:46:09 hi Nov 24 16:48:33 Ifaistos, PREFERRED_VERSION = ... Nov 24 16:49:38 Ifaistos: yes put preferred_version_uclibc = 0.9.28+{srcdate} style Nov 24 16:53:08 crofton : I have set PREFERRED_VERSION="svn" but it ignores it Nov 24 16:53:25 ade|desk : I'll try it Nov 24 16:57:04 Ifaistos: just use the bitbake shell and try 'which uclibc' then set the var to what it needs to be and try again, i can't remember which {SRCthing} is needed after the 0.9.28 Nov 24 16:58:35 ade|desk : shoult it be something like REFERRED_VERSION_uclibc = "0.9.28+cvsxxxxxx" ? Nov 24 17:00:00 bbl Nov 24 17:05:59 uclibc is svn not cvs Nov 24 17:07:38 Try using PREFERRED_VERSION_uclibc = "0.9.28+svn${SRCDATE}". That then forces you to set SRCDATE_uclibc = "2006xxyy" which you will also need, Nov 24 17:08:53 PREFERRED_VERSION_uclibc = "0.9.28+svn${SRCDATE}" Nov 24 17:08:53 PREFERRED_VERSION_uclibc-initial = "0.9.28+svn${SRCDATE}" Nov 24 17:09:01 should work Nov 24 17:11:57 RP : Hi ! Thanks. it worked ;) Nov 24 17:11:58 but you need to remove a header file once uclibc is built as it conflicts with a define from the linux headers, can't remeber which one it is atm Nov 24 17:12:37 I am having problems with uclibc 0.9.28 and gcc 4 on powerpc Nov 24 17:14:02 is the config for uclibc valid for powerpc Nov 24 17:14:55 * ade|desk needs a powerpc, one of the only archs i've not played with Nov 24 17:14:59 it compiles with 3.4.4. its some issue that gcc 4.x has with powerpc Nov 24 17:15:29 http://bugs.busybox.net/print_bug_page.php?bug_id=370 Nov 24 17:15:29 remember 0.9.28 came out before gcc4 did Nov 24 17:17:01 hopefully svn uclibc will fix that for you Nov 24 17:17:21 I hope so :) Nov 24 17:20:35 RP: I will probably by in London for Christmas... Wife wants to go shopping.... Nov 24 17:21:11 03koen 07org.oe.dev * r59e9ba26... 10/ (3 files in 2 dirs): Nov 24 17:21:11 poppler: update to 0.5.4 Nov 24 17:21:11 poppler-data: add 0.1, needed for proper cjk stuff, weird license Nov 24 17:28:30 does anyone have any experience with u-boot? if my machine is properly defined - I just bitbake u-boot and then flash the image? Nov 24 17:32:47 Ifaistos: Sadly London is 350 miles away from where I am :-( Nov 24 17:36:03 03koen 07org.oe.dev * reba4ce25... 10/ (1 packages/evince/evince_0.6.1.bb): evince: update to 0.6.1 Nov 24 17:38:09 btw, task-base does not create the rootfs image, right? Nov 24 17:38:39 is it correct that task-bootstrap is deprecated and bootstrap-image should be used? Nov 24 17:39:31 what? bootstrap-image is for an image. task-base is a task., you're comparing apples and orangges. Nov 24 17:39:46 oh ok, that means I do not understand much of what I am doing ;) Nov 24 17:40:20 kergoth: if I created a distro, and wanted everything to get build and the result of it be a rootfs image - what would I run? Nov 24 17:57:07 when bitbake bootstrap-image fails with "task-bootstrap: unsatisfied recommendation for xxx", but xxx package is already been built.. what is the reason for that? Nov 24 18:05:04 hmm Nov 24 18:05:06 not sure Nov 24 18:05:26 Jin^eLD, u-boot should just build, but you may want to look over the bb file Nov 24 18:05:53 Crofton: u-boot wants a machinename_config file, so the build fails Nov 24 18:05:58 but thats another issue.. Nov 24 18:06:05 first I'd like to figure out why I can not create a rootfs Nov 24 18:06:10 hmm Nov 24 18:06:21 boostrap-image fails? Nov 24 18:06:23 bitbake bootstrap-image should do it? Nov 24 18:06:28 yep, it fails Nov 24 18:06:33 that will make a basic image Nov 24 18:09:42 http://pastebin.ca/256839 Nov 24 18:09:45 that's what it says Nov 24 18:11:03 this is your own distro? Nov 24 18:11:42 pastebin your distro file if that is the case Nov 24 18:13:21 http://pastebin.ca/256841 Nov 24 18:14:21 REFERRED_PROVIDER_task-bootstrap = "task-bootstrap" Nov 24 18:14:24 don't do that Nov 24 18:14:44 oh ok Nov 24 18:14:50 also you inherit package_ipk and debian a couple of times Nov 24 18:14:58 that may cause problems :) Nov 24 18:14:59 oh yes, indeed Nov 24 18:15:25 ./include/angstrom.inc:9:MACHINE_TASK_PROVIDER = "task-base" Nov 24 18:16:05 ok Nov 24 18:17:03 thanks.. trying this out Nov 24 18:18:13 koen: regarding uboot.. when doing "bitbake u-boot" it stars compiling with some parameter that sets the config file to machinename_config Nov 24 18:18:21 where exactly is it searchinf for that file? Nov 24 18:18:28 and how is this supposed to look? is there any example? Nov 24 18:18:37 no idea on that, I don't have machine that use uboot (yet) Nov 24 18:18:50 * koen looks at mr uboot Crofton Nov 24 18:18:56 :> Nov 24 18:19:03 Jin^eLD, look in the uboot makefile Nov 24 18:20:27 Crofton: well, someone passes this option to uboot, under "machine_config" I mean - actually the machine name that I am using in OE Nov 24 18:22:33 Crofton: ok it is the bb file that is setting it... UBOOT_MACHINE ?= "${MACHINE}_config" Nov 24 18:22:51 the question is how that file looks and where it's supposed to be... Nov 24 18:23:00 whaich file? Nov 24 18:23:05 er which file? Nov 24 18:23:11 koen: after changes to the distro conf it now says: | task-base: unsatisfied recommendation for kernel-module-input Nov 24 18:23:18 the _config isn't a file Nov 24 18:23:28 hmm Nov 24 18:23:32 not sure about that one Nov 24 18:23:32 Crofton: aaah! it's the machine setting? Nov 24 18:23:56 in the uboot makeifle, those targets just configure u-boot Nov 24 18:24:02 Crofton: ok now I get it Nov 24 18:24:31 Jin^eLD: sounds line a missing kernel recipe Nov 24 18:26:16 koen: I do have the kernel.. I derived it from compulab-pxa270_2.6.16.bb Nov 24 18:26:46 kernel is already built successfully, I have the vmlinux file and a bunch of module ipk's Nov 24 18:27:11 Could anyone do me a favour and see if they are able to build 'btsco' from oe.dev? Nov 24 18:30:15 goxboxlive: it does not find bluetooth headers and fails Nov 24 18:30:37 The same problem that i have. Thanks for trying out Nov 24 18:30:53 I have bluez-dev installed, but it failes anyway Nov 24 18:31:01 np Nov 24 18:48:01 koen: in task-base.bb task-base-kernel26-extras-rrecommends = "kernel-module-input kernel-module-uinput", who is setting those "extra recommends" things? because that's what I am missing now Nov 24 18:48:17 my device does not have any input.. Nov 24 18:48:24 it shouldn't bail out on missing rrecommends Nov 24 18:49:22 hmm, maybe its somthing different then.. Nov 24 18:50:18 http://pastebin.ca/256872 Nov 24 18:50:25 could you take a look? what about "Error 4" Nov 24 18:50:26 ? Nov 24 18:50:39 and then, it cannot satisfy some dependencies Nov 24 18:50:50 but the question is - why? Nov 24 18:51:00 for example busybox... it has been compiled Nov 24 18:51:14 tried a clean rebuild? Nov 24 18:51:34 did you remove stuff from tmp/deploy? Nov 24 18:51:42 no, I did not touch tmp/deploy Nov 24 18:51:55 I started of completely clean, without anything Nov 24 18:51:59 then managed to build a kernel Nov 24 18:52:06 and then at some point task-base finished Nov 24 18:52:16 after that I tried to get an image and here I am.. Nov 24 18:52:21 what happens if you do "bitbake base-files base-passwd busybox initscripts netbase sysvinit sysvinit-pidof tinylogin modutils-initscripts fuser setserial udev sysfsutils module-init-tools samba mediatomb" ? Nov 24 18:53:18 ok.. here we go.. hmm. sysvinit-pidof Nov 24 18:53:29 it says nothing provides dependency sysvinit-pidof Nov 24 18:53:44 who is setting that? Nov 24 18:54:15 that is RDEPENDS_task-base-core-default Nov 24 18:55:14 right Nov 24 18:55:29 bitbake sysvinit says Attempted builds: 0, so I guess it has been built? lemme check deploy.. Nov 24 18:55:46 yep, its there Nov 24 18:55:51 even the pidof ipkg Nov 24 18:56:01 so why can't it find it? Nov 24 18:57:53 it also reports and unsatisfied recommendation for dropbear, but dropbear ipk is in the deploy directory as well Nov 24 18:57:56 I don't get it Nov 24 18:58:03 ah Nov 24 18:58:05 I think I know why Nov 24 18:58:14 could you pastebin your machine.conf? Nov 24 18:58:54 http://pastebin.ca/256883 Nov 24 18:59:44 that's what I expected Nov 24 18:59:47 :) Nov 24 19:00:02 you got hit by http://www.openembedded.org/bonsai/view/rev/21604/ Nov 24 19:00:02 what is it? Nov 24 19:00:08 lemme check Nov 24 19:00:32 try this: http://pastebin.ca/256887 Nov 24 19:01:22 k.. moment Nov 24 19:02:31 it will trigger a reparse, but not a rebuild Nov 24 19:02:49 yep, waiting for the cache Nov 24 19:04:12 03koen 07org.oe.dev * r14c3e0e9... 10/ (1 packages/btsco/btsco_0.42.bb): Nov 24 19:04:12 btsco: stop configure from putting -I/usr/include into the commmand line, fixes the problems people where seeing in #oe Nov 24 19:04:12 * config.log is a handy debug tool Nov 24 19:06:26 koen: seems like it worked :) Nov 24 19:06:29 thanks a lot! Nov 24 19:07:13 btsco builds again as well Nov 24 19:07:50 :> Nov 24 19:11:19 koen: what would be the proper way to setup a customized inittab? is there a way via variables, or would I have to make a patch for my stuff or something similar? Nov 24 19:17:35 mkdir packages/sysvinit/sysvinit/ ; cp my-inittab packages/sysvinit/sysvinit//inittab Nov 24 19:18:11 thx Nov 24 19:18:21 and rebuild sysvinit Nov 24 19:31:34 RP: Any updates on http://bugs.openembedded.org/show_bug.cgi?id=1553 ? Nov 24 19:35:26 thx koen: btsco where buildt. Nov 24 19:36:00 CoreDump|home: iirc that patch is in the latest ASoC Nov 24 19:36:15 http://opensource.wolfsonmicro.com/~lg/asoc/ Nov 24 19:36:54 koen: I'll try that, thanks =) Nov 24 19:39:03 CoreDump|home: lrg is in india right now, so if it isn't in, it'll have to wait till he gets back Nov 24 19:39:29 yeah, no hurry. Just cheking Nov 24 19:39:37 checking even Nov 24 19:44:14 koen: Id be able to guess the status if I could actually see that bug Nov 24 19:46:19 * koen hands XorA|gone some extra goggles Nov 24 19:46:48 this tv-show is creepy Nov 24 19:46:52 'dexter' Nov 24 19:47:36 koen just watched the first episode and found it interesting but not enough to dowload the others episodes ... will wait to see it on tv :) Nov 24 19:49:01 internet is so slow that I queued up the whole batch last week Nov 24 19:49:25 halfway through the first episode now Nov 24 19:49:46 hi koen Nov 24 19:49:53 hey nicolasfr Nov 24 19:50:20 koen: I am having a strange pb for more than a week and cannot find solution... Nov 24 19:50:48 I made a fresh install of Edgy and installed OE, and now gcc-cross-4.1.1-r7 won't build Nov 24 19:51:13 target 586 architecture with distro=angstrom Nov 24 19:51:54 haven't tried building under edgy yet Nov 24 19:52:03 only dapper and debian stable Nov 24 19:52:36 Maybe I should try again under Dapper in vmware to see if it's a Edgy problem or OE problem... Nov 24 19:54:10 I don't know exactly where it breaks, there is nothing in logs except 4 lines here: http://pastebin.ca/256928 Nov 24 19:56:12 nicolasfr: try copying the glibc lines in the angstrom files for the guinness machine Nov 24 19:56:49 ./angstrom-2007.1.conf:155:PREFERRED_VERSION_linux-libc-headers_guinness ?= "2.6.18" Nov 24 19:56:49 ./angstrom-2007.1.conf:160:PREFERRED_VERSION_glibc-intermediate_guinness ?= "2.4" Nov 24 20:00:16 koen: trying to rebuild with changes Nov 24 20:02:24 hmm, when threads fail.. what could be the reason? did I miss some parameter when building or is it a kernel thing? that's the first time I use this system with a self built and not with a prebuilt kernel Nov 24 20:03:31 The thump patches fail when applied in uclibc-0.9.28-svn Nov 24 20:05:20 Ifaistos: what thumb patches? Nov 24 20:05:33 there are no patches in the uclibc_svn.bb Nov 24 20:05:37 dropbear -F just exists without even printing something.. doh.. Nov 24 20:07:56 hrw|gone: RP : qemu needs a patch for 2.6.17 kernels Nov 24 20:07:56 hrw|gone: RP : I have no linux/compiler.h anymore Nov 24 20:08:04 hey zecke Nov 24 20:08:34 koen: which two angstrom machines should I regulary compile for? Nov 24 20:08:52 koen : http://pastebin.com/832242 Nov 24 20:08:54 h2200 and ep93xx Nov 24 20:09:32 Ifaistos: is your checkout of .dev up to date? Nov 24 20:10:12 koen : Not sure on this machine. Probably not Nov 24 20:10:23 koen: which target(s)? Nov 24 20:10:55 zecke: angstrom-bootstrap-image should cover the basics Nov 24 20:11:36 koen: so 2x angstrom-bootstrap-image + 2x xfce-image or gpe-image or opie-image? Nov 24 20:11:58 zecke: only bootstrap-image for now Nov 24 20:12:09 okay Nov 24 20:12:57 koen: any prefered version of buttbake? Nov 24 20:13:11 I use trunk Nov 24 20:13:59 but 1.6.x should work as well Nov 24 20:16:57 koen: now with the changes glibc-intermediate-2.4 fails with error "ld: cannot find -lgcc_eh" Nov 24 20:17:36 hrm Nov 24 20:17:49 I'm pretty much out of ideas Nov 24 20:20:41 koen, your check in earlier about /usr ... paths roughly how does that work? Nov 24 20:21:03 I think there are similar evil macros in omniorb and xerces ... Nov 24 20:21:28 Crofton: it seds out the bit that introduces -I/usr/include Nov 24 20:21:47 is it likely to work on other configure scripts? Nov 24 20:22:00 or is it specific to the one package? Nov 24 20:22:21 Crofton: http://www.openembedded.org/viewmtn/diff.psp?id1=eba4ce2543ebf7d29b3e9a78c4eda7758bbb42b8&id2=14c3e0e9b46fd8a7eea8892c338da559032a5f30 Nov 24 20:22:30 sed -i 's:-I${bluez_prefix}/include::g' configure Nov 24 20:22:36 that pretty specific Nov 24 20:22:48 yeah Nov 24 20:22:49 ok Nov 24 20:22:55 normally I'd try to patch configure.ac, but that wasn't to problem Nov 24 20:23:04 but, config.log reveals all? Nov 24 20:23:35 config.log showed 'Cross Compile badness: /usr/include in include path' Nov 24 20:24:01 hmmm Nov 24 20:24:16 which is a pretty obvious error message :) Nov 24 20:28:43 koen: which DISTRO to set (I'm lazy) Nov 24 20:29:06 zecke: DISTRO="angstrom-2007.1" Nov 24 20:33:19 thanks Nov 24 20:34:18 RP: oh it took that long to recognize why you want an autobuild.bbclass Nov 24 20:34:22 no, thank you! Nov 24 20:45:00 koen: it runs nows Nov 24 20:45:17 zecke: danke schön Nov 24 20:45:43 koen: sadly I won't find the time during the next two weeks to get the new tinderbox in a usable shape Nov 24 20:46:24 Koen: I made this mplayer plugin work but video quality still sucks Nov 24 20:46:51 Koen: many kudo's go to Genesis although Nov 24 20:47:37 Koen: it is not stable enough to put in Nov 24 20:48:19 nicolasfr: I just did 486/generic and guiness/angstrom builds. The angstrom build failed building cups, but a lot of things worked before that. Nov 24 20:48:45 cbrake: my build for both works Nov 24 20:50:05 cbrake: I needed patches although for firefox and stuff Nov 24 20:50:29 gerwinin: great. I have not debugged the cups issue. Are you running 64 or 32 bit host? Nov 24 20:50:47 cbrake: thanks. are use running Ubuntu Edgy ? Nov 24 20:50:58 nicolasfr: no, gentoo Nov 24 20:51:05 nicolasfr: on amd64 Nov 24 20:52:58 Well that must be a problem with Edgy. Everything used to work fine under Dapper. The logs won't show exactly where it breaks, I need to digg. Nov 24 21:35:49 Nicolasfr : Did you succeed in compiling already ? Nov 24 21:45:44 gerwinin: I did with Dapper, but I moved to Edgy two weeks ago and since then, no luck with angstrom and gcc-4.1.1-cross Nov 24 21:50:38 CoreDump|home: The status of that bug is it should be fixed in later revisions of ASoC. I've not had a chance to merge and test those later revisions yet Nov 24 21:50:54 zecke: qemu is a problem :-/ Nov 24 21:53:02 RP: yes, but RedHat has a patch Nov 24 21:53:21 zecke: Which works for non redhat systems as well? Nov 24 21:53:35 http://rpmfind.net/linux/RPM/fedora/devel/x86_64/qemu-0.8.2-4.fc7.x86_64.html Nov 24 21:53:54 dunno, I only found this.. I have no clue about source rpm's or the quality of the patch :) Nov 24 21:56:00 * zecke builds rpm... Nov 24 22:27:25 zecke: Have you any thoughts on an unexport or unset option for variables? Nov 24 22:28:13 zecke: Basically, I want to be able to trigger scripts to have "unset MACHINE" Nov 24 22:28:51 I have not had many thoughts lately Nov 24 22:29:02 abusing __anonfunc to kill it Nov 24 22:29:15 bb.data.delVar('MACHINE') Nov 24 22:29:48 but this doesn't kill it from os.environ Nov 24 22:30:30 zecke: no. Locally I have code to allow an unexport flag to be set which triggers an "unset MACHINE" in emit_env Nov 24 22:32:05 but that is more like unexport MACHINE Nov 24 22:33:00 zecke: http://sam.rpsys.net/poky-bindir.patch See data.py and base.bbclass Nov 24 22:33:51 I didn't go as far as adding unexport support to the parser but I'm wondering about it... Nov 24 22:34:55 just one question Nov 24 22:35:00 are you sure you want to return 0? Nov 24 22:35:18 the other if does not return from within Nov 24 22:36:16 RP: I wouldn't mind an unexport in the parser, it feels symmetric Nov 24 22:36:43 zecke: We also have what I feel is an important use case... Nov 24 22:37:07 let us keep it in mind Nov 24 22:37:13 oh the Red Hat patch is good Nov 24 22:37:18 -#include Nov 24 22:37:30 zecke: I should drop the \n and return 1 :) Nov 24 22:38:14 RP: have you changed libreadline in poky? Nov 24 22:38:17 Actually, I should keep the \n and return 1. Its just whitespace though :) Nov 24 22:38:32 zecke: no, why? Nov 24 22:38:55 http://ewi546.ewi.utwente.nl/tinderbox/showlog.pl?machine_id=141&logfile=20061124201907.log Nov 24 22:39:14 back to firefighting Nov 24 22:43:57 zecke: Strange. The error looks somehow familiar yet I've not seen readline fail before :-/ Nov 24 22:45:28 CoreDump|home: RP: will someone tell me the ASoC bug you talk about as I still dont have DNS for bugs.oe.org Nov 24 22:46:23 XorA|gone: add 64.136.190.225 bugs.openembedded.org to your /etc/hosts Nov 24 22:47:10 XorA|gone: OSS mixer not working due to the mappings being incorrect Nov 24 22:47:23 RP: ah right that one Nov 24 22:47:55 stop using bloomin OSS :-) Nov 24 22:48:48 I havent tested a change like that, Liam been sneaking in fixes without telling me Nov 24 22:49:08 XorA|gone: Liam said you'd tested it :} Nov 24 22:49:41 RP: when I talked to him about the OSS mixer stuff, he said we wouldnt do it as OSS is nixed soon anyway Nov 24 22:49:50 something has slipped through here Nov 24 22:50:07 well, unsurprisingly the latest asoc doesn't apply to 2.6.17 Nov 24 22:50:49 CoreDump|home: Drop the Makefile hunk and use the makefile patch on rpsys.net Nov 24 22:51:17 RP url? Nov 24 22:51:59 just add the soc/ diretory to Makefile and quilt refresh Nov 24 22:52:15 * XorA|gone has that on autopilot now Nov 24 22:52:23 CoreDump|home: http://www.rpsys.net/openzaurus/patches/archive/asoc_makefile-r0.patch is for 2.6.17 iirc, r1 applies for 2.6.18+ Nov 24 22:52:40 RP: thanks Nov 24 22:52:50 XorA|gone: I'm a quilt n00b =) Nov 24 22:53:20 CoreDump|home: so am I Nov 24 22:53:37 CoreDump|home: but just type quilt gives me enough instructions to work with Nov 24 22:54:06 quilt is one of my favourite tools :) Nov 24 22:54:19 I looked at the gnome GUI, couldn't get it to work Nov 24 22:54:34 yup, its good, quite a few patches for smdk2440 kernel managed with quilt Nov 24 22:55:38 quilt isn't a GUI program... Nov 24 23:02:52 would a 4 GB SDHC card be capable of doing swap and/or compiling? Nov 24 23:03:03 or is there still the whole limit on # of writes? Nov 24 23:03:52 * XorA|gone used a normal SD card for swap for months without problems Nov 24 23:05:10 * XorA|gone finds trolltech fudding OpenMoko Nov 24 23:05:21 where? Nov 24 23:06:00 http://news.zdnet.co.uk/communications/0,1000000085,39284632,00.htm Nov 24 23:07:18 3rd and 2nd last paragraphs Nov 24 23:10:10 hmm good nite :) Nov 24 23:11:33 OpenMoko? Nov 24 23:11:40 good night as well Nov 24 23:12:54 Luke-Jr: you been under a rock? Nov 24 23:12:59 ~openmoko Nov 24 23:13:07 openmoko is probably a joint venture of FIC and the open source community creating a complete open source phone platform stack. see http://openmoko.com Nov 24 23:13:49 * CoreDump|home drops poodle_audio-r4.patch Nov 24 23:15:51 NOTE: package linux-openzaurus-2.6.17-r29: task do_compile: started Nov 24 23:16:41 Why does the 2.6.18 kernel assign my serial_cs card ttyS0 (on a spitz), when ttyS0 and ttyS1 are already taken? Nov 24 23:17:40 XorA|gone: as far as OE goes, quite :) Nov 24 23:17:57 doesn't TrollTech already have an open source phone thing? Nov 24 23:18:20 Luke-Jr: *cough* Nov 24 23:18:58 *cough cough* Nov 24 23:18:59 Necronom: its not really ttyS0 its ttyS3 Nov 24 23:19:23 Necronom: dmesg lies Nov 24 23:19:27 Ahh. Nov 24 23:19:35 XorA|gone: does that mean "no"? Nov 24 23:19:59 Luke-Jr: there has been some "friendly" debate on the "Open"ness of Greenphone from trolltech Nov 24 23:22:44 Is anyone else here thinking that a module versioning of "kernel-module-blahh-2.6_2.6.17-r29_poodle.ipk" is deeply flawed? shouldn't that be kernel-module-blahh-2.6.17_2.6.17-r29? Nov 24 23:24:16 XorA|gone: so it's not really open? Nov 24 23:24:34 Luke-Jr: its Qt, not a world I know Nov 24 23:24:44 ... Nov 24 23:24:49 Luke-Jr: I hear rumours it has not been released as GPL Nov 24 23:24:55 Qt is GPL... Nov 24 23:25:00 hm Nov 24 23:25:21 Luke-Jr: QtopieCore4 or something isnt I beleive, and thats whats on the phone Nov 24 23:29:10 CoreDump|home: I asked for the -2.6 to be added, but cant for the life of me remeber why, I think it might have been so you can distinquish 2.4 and 2.6 kernel builds easilly Nov 24 23:29:23 thats a long time ago in history Nov 24 23:30:34 XorA|gone: which is nice yeah. I was more wondering why the "2.6" wasn't "2.6.17" so a kernel 2.6.18 wouldn't be automatically pulled in during upgrades Nov 24 23:31:17 with this system kernel-upgrades via ipkg are a pain at best, impossible at worst Nov 24 23:31:28 CoreDump|home: no idea, I barely remeber, maybe RP does Nov 24 23:36:05 I'm thinking about abusing updater.sh to install a modules.tgz ripped from bootstrap straight into the rootfs, then flash the new kernel Nov 24 23:36:32 then let the user ipkg upgrade the rest of his modules Nov 24 23:37:55 I thought the 2.6 was for PARALLEL_MODULES_INSTALL? Nov 24 23:38:05 (which IMO should be ripped out) Nov 24 23:38:56 whats PARALLEL_MODULES_INSTALL? Nov 24 23:39:14 XorA|gone: Allows you to install a 2.4 and a 2.6 kernel at once - the ipaqs used it Nov 24 23:39:26 If you unset it, the sufix disappears IIRC Nov 24 23:39:44 koen: Can angstrom drop that at least? Nov 24 23:39:45 RP: that was basically what I was doing on Z at one point Nov 24 23:39:53 RP: angstrom has I think Nov 24 23:39:59 yay :) Nov 24 23:40:17 RP: I seem to remeber no -2.6 on my angstrom modules anyway Nov 24 23:40:34 * RP should "fix" openzaurus too Nov 24 23:41:07 I bounced back and forth between kernel to track down that X11 suspend bug Nov 24 23:42:56 Anyone needing that could just untar modules.tgz into the filesystem... Nov 24 23:43:23 RP: modules.tgz appears to contain _all_ modules which makes it rather large Nov 24 23:43:53 CoreDump|home: People complain if the kernel doesn't build everything. We can't have it all ways... Nov 24 23:44:27 * CoreDump|home is not complaining Nov 24 23:44:49 modules.tgz was created purely so didn't have to install a zillion ipks when testing new kernels Nov 24 23:44:52 However, we should really establish clean kernel update path w/o flashing the rootfs Nov 24 23:45:44 CoreDump|home: ipkg install new modules - reboot, reflash kernel Nov 24 23:46:13 The kernels should always boot without any modules even if things like sound doesn't work Nov 24 23:47:08 you could statically link an ipkg-cl and install modules from updater.sh if you really wanted some fun Nov 24 23:47:12 joe sixpack can't unb0rk his system if the modules are fubar Nov 24 23:47:27 yuck Nov 24 23:47:38 CoreDump|home: What stops them? Nov 24 23:47:39 man, whatis it with people calling other people joe six-pack today!! Nov 24 23:47:44 hello all Nov 24 23:48:05 RP: their unleetness mostly =) Nov 24 23:48:09 hi Greg2 Nov 24 23:49:03 hi CoreDump|home RP Nov 24 23:49:30 will the latest snapshot of oe..mnt 24-nov-2006 work with monotone .31? Nov 24 23:54:07 hi Greg2 Nov 24 23:55:54 'night all Nov 24 23:56:23 'night RP Nov 24 23:56:40 n8 RP Nov 24 23:57:06 RP: the latest kernel w/ new asoc ooses rather badly on boot *cough* Nov 24 23:57:14 Coredump|home: I'm upgrading my system, should I use monotone .31 or .30? Nov 24 23:57:53 no idea, I'm still using 0.29 =) Nov 24 23:57:59 CoreDump|home: what zaurus? Nov 24 23:58:04 :) Nov 24 23:58:06 XorA|gone: Poodl Nov 24 23:58:16 CoreDump|home: it works on corgi and spitz :-) Nov 24 23:59:08 :( Nov 24 23:59:10 ooses? Nov 24 23:59:25 CoreDump|home: although ASoC isnt built in so it cant be my fault :-) Nov 25 00:02:01 ok, this kernel is fubar Nov 25 00:02:40 I cant test poodle as I cant get filesystem small enough Nov 25 00:03:03 poodle runs off SD natively these days =D Nov 25 00:03:55 might have to persude wolfson to buy some then, all mine are tied up in other machines Nov 25 00:04:19 anyway I shall go sleep, bye Nov 25 00:04:19 no hurry Nov 25 00:04:25 n8 XorA|gone Nov 25 00:04:36 I just picked up an 1GB SD card for $13.47 USD Nov 25 00:04:51 Black Friday here Nov 25 00:12:39 is any body using monotone .31 ? Nov 25 00:15:15 Greg2: I am Nov 25 00:15:44 with the latest snapshot? Nov 25 00:16:08 I am not sure if I did an update of OE today Nov 25 00:16:15 but I did one yesterday evening I think Nov 25 00:16:40 ok... thanks Nov 25 01:13:21 g-night all Nov 25 02:46:30 allo darlins Nov 25 02:46:59 well we have another device for the list, to play with (at least as a pda - a pretty damn expensive one) Nov 25 02:47:04 the ipaq hw6915 Nov 25 02:47:29 http://wiki.xda-developers.com/index.php?pagename=Ipaq6915 **** ENDING LOGGING AT Sat Nov 25 02:59:57 2006