**** BEGIN LOGGING AT Mon Jun 16 02:59:57 2008 Jun 16 07:51:40 morning Jun 16 07:51:49 i see some new wiki Jun 16 08:00:47 are someone checking the MacOSX ? Jun 16 08:00:52 building? Jun 16 08:59:04 03koen 07org.oe.dev * r9cee85b1... 10/ (1 packages/pulseaudio/pulseaudio-alsa-wrapper.bb): pulseaudio alsa wrapper: add more modules Jun 16 09:09:51 03dirk 07org.oe.dev * rccd65c16... 10/ (1 classes/insane.bbclass): insane.bbclass: add linux-gnuspe as valid target Jun 16 09:15:32 good morning Jun 16 09:29:43 hi all Jun 16 09:30:06 any western european developer around, specificly those involved with the e.V. process? Jun 16 09:33:15 I'm the Akademy 2008 organiser and am looking for people who want to present at the Embedded and Mobile Day. Jun 16 09:44:03 * XorA is around Jun 16 09:45:21 XorA: hi Jun 16 09:45:40 hey Stecchino Jun 16 09:45:52 where is Akademy this year? Jun 16 09:46:10 Belgium Jun 16 09:46:24 ah, better than Glasgow :-) Jun 16 09:46:59 from 9/08 -> 15/08 Jun 16 09:47:42 and the Em. & Mob. day is on the 12th Jun 16 09:48:23 I dont think the eV guys are out of bed yet Jun 16 09:48:42 that would be rschuster, florian, mickeyl and zecke from memory Jun 16 09:48:47 I was hoping someone from OE would be intersted in giving a talk. Jun 16 09:49:10 Stecchino: more than certainly someone will be available Jun 16 09:49:12 I contacted zecke directly since he's a KDE developer as well Jun 16 09:49:28 Stecchino: zecke is in Taiwan at the moment :-) Jun 16 09:49:43 yeah, was last week though Jun 16 09:50:01 he only fell off irc a few mins ago Jun 16 09:50:29 mickey|zzZZzz: ping me when you wake up :) Jun 16 09:50:30 does the talk have to be Qt/KDE related? Jun 16 09:50:35 no, not at all Jun 16 09:51:05 then there are a million of us can talk I think, just depends who has time/money Jun 16 09:51:27 but I'm hoping the panel discussion will be about the possibilities of using Open Desktop code on embedded devices Jun 16 09:52:15 It might be a good opportunity to introduce the OE e.V. as well. There will be plenty of companies that would like to know. Jun 16 09:55:31 yeah Jun 16 09:55:42 * XorA shakes the channel to wake the sleepyheads Jun 16 09:55:59 there seems to be a lot of overlap with OpenMoko as well Jun 16 09:56:04 morning Jun 16 09:56:24 * XorA denies all knowledge of OpenMoko :-D Jun 16 09:56:29 * hrw will not go to Akademy Jun 16 09:56:44 just send a mail to michael@openmoko.org for the price of 10 freerunner for education, might be the same mickey? Jun 16 09:56:48 it conflict with other parties Jun 16 09:56:59 hrw: which ones? Jun 16 09:57:03 Stecchino: no, its mickey@openmoko Jun 16 09:58:25 hrw: you an expert on n810 yet? Jun 16 10:02:00 * Stecchino would like to get into more embedded development Jun 16 10:02:07 will after Akademy Jun 16 10:02:21 Stecchino: my brother wedding :) Jun 16 10:02:26 XorA: depends Jun 16 10:03:19 hrw: surely he'll get married again? You can go to that one. No reason to skip Akademy ;) Jun 16 10:03:26 hrw: try to answer a question a let us decide :p Jun 16 10:03:55 but Akademt is 7 days, most wedding parties don't last that long :) Jun 16 10:12:00 Stecchino: haha Jun 16 10:57:58 hi florian_, zecke Jun 16 10:58:30 hi all! Jun 16 11:09:38 florian, zecke: hi Jun 16 11:09:58 Remember the mail you might have received about the Akademy 2008 Embedded and Mobile Day? Jun 16 11:10:06 we are still looking for speakers Jun 16 11:10:29 goodmorning zecke BTW Jun 16 11:11:12 hmm or evening Jun 16 11:11:21 * Stecchino is confused by timezones Jun 16 11:12:13 * florian is quite sure he did not get this mail :-) Jun 16 11:12:33 Stecchino: I guess we have UGT here anyway. Jun 16 11:12:50 I've send it to various mailing lists but in CC instead of TO. means it probably got lost in moderation Jun 16 11:13:33 florian: UGT? Jun 16 11:13:44 ~UGT? Jun 16 11:13:45 i heard ugt is Universal Greeting Time. Created in #mipslinux, it is a rule that states that whenever somebody enters an IRC channel it is always morning, and it is always late when the person leaves. The local time of any other people in the channel, including the greeter, is irrelevant. http://www.total-knowledge.com/~ilya/mips/ugt.html Jun 16 11:14:04 lol Jun 16 11:19:42 03koen 07org.oe.dev * r1ca4c522... 10/ (1 conf/machine/hx4700.conf): hx4700: append jffs2 to fstypes Jun 16 11:19:42 03koen 07org.oe.dev * r1ca4c522... 10/ (1 conf/machine/hx4700.conf): hx4700: append jffs2 to fstypes Jun 16 11:19:42 03koen 07org.oe.dev * r1ca4c522... 10/ (1 conf/machine/hx4700.conf): hx4700: append jffs2 to fstypes Jun 16 11:20:00 03koen 07org.oe.dev * r1ca4c522... 10/ (1 conf/machine/hx4700.conf): hx4700: append jffs2 to fstypes Jun 16 11:20:00 03koen 07org.oe.dev * r1ca4c522... 10/ (1 conf/machine/hx4700.conf): hx4700: append jffs2 to fstypes Jun 16 11:20:00 03koen 07org.oe.dev * r1ca4c522... 10/ (1 conf/machine/hx4700.conf): hx4700: append jffs2 to fstypes Jun 16 11:20:03 03koen 07org.oe.dev * r1ca4c522... 10/ (1 conf/machine/hx4700.conf): hx4700: append jffs2 to fstypes Jun 16 11:20:05 03koen 07org.oe.dev * r1ca4c522... 10/ (1 conf/machine/hx4700.conf): hx4700: append jffs2 to fstypes Jun 16 11:20:07 03koen 07org.oe.dev * r1ca4c522... 10/ (1 conf/machine/hx4700.conf): hx4700: append jffs2 to fstypes Jun 16 11:30:05 hmmm Jun 16 11:30:06 CIA? Jun 16 11:30:10 slightly confused Jun 16 11:35:17 so mickeyl, florian, zecke or anyone else: would you be interested in presenting about openembedded at the Akademy 2008: Embedded & Mobile Day Jun 16 11:35:33 Stecchino: mail oe-devel ML? Jun 16 11:35:36 it doesn't have to be anything about Qt or KDE Jun 16 11:35:40 I will Jun 16 11:36:08 i'd love to, but my travel contigent is already eaten up this ear Jun 16 11:36:10 year, even Jun 16 11:37:51 perhaps the first OE e.V. sponsored trip? Jun 16 11:38:24 i asume your from DE, so it's not far for you. Belgium Jun 16 11:39:11 Stecchino: ah hey, I sadly have no time to guy aKademy, family duties in berlin :( Jun 16 11:39:35 zecke: even on the 12th? Jun 16 11:40:18 Stecchino: yes, my little cousin is coming from eindhoven to visit me, so I have no way of leaving him alone Jun 16 11:40:44 well... contigent as in the time I'm allowed to leave my wife alone Jun 16 11:40:45 zecke: have fun :) Jun 16 11:41:14 mickeyl: you'd be supprised how many wives and girlfriends are being dragged to Akademy this year Jun 16 11:41:49 *nod* Jun 16 11:41:59 will ask Sabine Jun 16 11:42:03 mickeyl: you can go there to catch up with ellen Jun 16 11:42:43 zecke: Good point to let sabine allow the trip. ;) Jun 16 11:43:01 haeh? I don't get you Jun 16 11:43:24 zecke: you could always bring the cousin along. Some other kids were coming. We were even planning daycare and trips to zoo and toy museum Jun 16 11:43:36 bbl, need to get food Jun 16 11:43:36 heh Jun 16 11:43:45 * mickeyl imagines Sabine going into the zoo Jun 16 11:43:50 Stecchino: hehe, I will have to ask him, he would be more interested in improving his python foo ;) Jun 16 11:43:51 zecke: So you have the feeling I normally have when chatting with you? Great. :) Jun 16 11:43:57 KDE dev's are starting to reproduce, so we need to keep kids in mind now :) Jun 16 11:44:13 www.plankendael.be Jun 16 11:44:16 very nice Jun 16 11:44:30 NL and FR only it seems Jun 16 11:44:32 to bad Jun 16 11:44:36 stefan_schmidt: hehe, I understand you, I just wanted to highlight why I'm single :) Jun 16 11:45:05 heh, and away he is Jun 16 14:24:16 03mickeyl 07org.oe.dev * r56c0efd3... 10/ (1 conf/distro/include/preferred-om-2008-versions.inc): preferred-om-versions: prefer pulseaudio 0.9.10 Jun 16 14:24:22 03mickeyl 07org.oe.dev * r3b2129cb... 10/ (1 conf/distro/include/sane-srcrevs.inc): opkg[-native|-sdk]: prefere revision 4488. should fix the segfault w/ unsigned repositories Jun 16 14:24:27 03xora 07org.oe.dev * r02ce8fba... 10/ (12 files in 3 dirs): firefox_2.0.0.14+3.0rc1.bb : first stab at firefox 3 recipe Jun 16 14:24:32 03koen 07org.oe.dev * r1ca4c522... 10/ (1 conf/machine/hx4700.conf): hx4700: append jffs2 to fstypes Jun 16 14:24:37 03mickeyl 07org.oe.dev * r9923a1db... 10/ (1 packages/freesmartphone/gsmd2_git.bb): gsmd2 git disable tests make it compile now Jun 16 14:24:42 03mickeyl 07org.oe.dev * r2d0e7ded... 10/ (1 conf/machine/om-gta01.conf conf/machine/om-gta02.conf): om-gta0[1|2]: allow local.conf to override PREFERRED_PROVIDER_virtual/kernel Jun 16 14:24:48 03koen 07org.oe.dev * r3b0348e0... 10/ (1 conf/distro/include/sane-srcrevs.inc): sane-srcrev: wind back opkg-native to 4452 *again* Jun 16 14:54:56 has anyone built OE distro using Code sourcery toolchain as an external toolchain? Jun 16 15:07:01 vsriram: It might be easier to build the Code Sourcery toolchain with OE. iirc we have some of them... Jun 16 15:13:20 you mean to say that, we can build code sourcery toolchain using oe and it will be same as the pre-built toolchain toolchain that they have Jun 16 15:13:42 are you sure that both are same. i would be interested in that. Jun 16 17:59:44 are people aware that currently /documentation and /wiki on the open embedded website currently give a 404, which itself gives those two links? Jun 16 18:03:46 I was just looking around to see if there was any sort of device specific stuff other than at handhelds.org Jun 16 18:13:06 I've been working on the iPAQ 2xx and have got most of it working now. I want to make sure no one else ends up doing it all again. Jun 16 18:13:39 I could also use some help with the higher level stuff, like how you setup ALSA properly. Jun 16 18:25:56 oliford: hm, that website seems to be a pretty sorry state indeed. there's the new wiki at wiki.openembedded.net, but I don't think it is meant to replace the old one yet. Jun 16 18:27:16 http://oe.linuxtogo.org/user-manual still works, kind of Jun 16 19:00:15 hello. i'm pretty new to oe, and i'm trying to understand what's the purpose of the qemu-native package that seems to be built by most recipes. can anyone please explain ? Jun 16 19:03:13 it's the emulator for running target binaries on your build host. Jun 16 19:03:39 nerochiaro: qemu-native is built and used to generate locale for the target machine Jun 16 19:03:55 is it necessary if all i need is to run the binaries on the target, or i am building packages to redistribute ? Jun 16 19:04:05 nerochiaro: it's used to build glibc's locales because the binary format is arch specific and there is no cross tool to build them from other architectures Jun 16 19:04:58 i see. it seems to drag along a lot of dependencies, though, like an older gcc and libsdl, if i'm not understanding wrong. is this correct ? Jun 16 19:05:16 does setting ENABLE_BINARY_LOCALE_GENERATION = "0" in local.conf disable the qemu-native build? Jun 16 19:05:54 yes Jun 16 19:07:07 nerochiaro: FYI, the older gcc (3.x) is built because qemu-native cannot be built with gcc4.x; and most distro dodays have gcc >= 4.0, so OE build the 3.x for you Jun 16 19:07:32 cyrilRomain: and SDL too ? Jun 16 19:07:45 nerochiaro: you could install gcc3 in your distro if it has it and set ASSUME_PROVIDED += "gcc3" to work around the gcc3 build. Jun 16 19:08:10 or you could try the ENABLE_BINARY_LOCALE_GENERATION = "0" if you don't care about building locales and just need glibc to build some libs/apps. Jun 16 19:08:11 nerochiaro: SDL is required by qemu iirc .. Jun 16 19:08:46 nerochiaro: you can also use ASSUME_PROVIDED for qemu, but I would follow pH5 suggestion: ENABLE_BINARY_LOCALE_GENERATION = "0" Jun 16 19:09:57 for now i think i won't need locales, so ok to disable them. but in case i need them later, can i build them once, then have OE fetch the binaries instead of rebuilding them ? Jun 16 19:11:22 i think i will only need UTF8 later on Jun 16 19:16:52 is it error-proof to share a same /tmp dir for .stable and .dev (same distro) ? Jun 16 19:25:19 if the kernel i'm building (fetched from a repository that is not kernel.org) is 2.6.23, why does OE also download one copy of the same kernel version from from kernel.org ? Jun 16 19:25:56 (i think it does that to build linux-libc-headers-2.6.23, but i don't understand why can't it use my own kernel) Jun 16 19:29:59 nerochiaro: the DISTRO determines which kernel to build Jun 16 19:30:14 hmm Jun 16 19:30:20 the OE-wiki appears quite broken Jun 16 19:30:31 DISTRO and MACHINE actually iirc Jun 16 19:30:37 looks like some wiki-markup is ignored Jun 16 19:30:49 cyrilRomain: yes, but why doesn't it just get that kernel once, instead of twice fom two repositories (mine and kernel.org) Jun 16 19:33:02 nerochiaro: hmm names differs ? Jun 16 19:33:56 cyrilRomain: they do, yes Jun 16 19:34:14 nerochiaro: that why both are downloaded then Jun 16 19:35:19 cyrilRomain: is there a way to tell OE that it can use the version in my repo safely for both tasks it need to do ? for "building" the gcc headers, i think any 2.6.23 will do, including the kernel from my own repo Jun 16 19:36:28 nerochiaro: there is something wrong because only one should be built Jun 16 19:37:46 cyrilRomain: you mean that both should have the same name ? Jun 16 19:38:23 nerochiaro: no .. Jun 16 19:38:42 then OE should notice automatically they are the same kernel Jun 16 19:38:43 ? Jun 16 19:39:00 nerochiaro: I meant, if they have different names, it is normal both are downloaded instead of only one Jun 16 19:39:24 but the wrong thing in your environment is that two kernels are being built be OE Jun 16 19:39:33 nerochiaro: and I think you want only one Jun 16 19:40:16 nerochiaro: which DISTRO and MACHINE do you use ? Jun 16 19:40:36 well, they are not both actually built. they are both downloaded, but only my kernel is built. the other kernel (the vanilla from kenrel.org) is only used for the linux headers i think Jun 16 19:40:50 i'm using DISTRO = "angstrom-2008.1" Jun 16 19:40:51 MACHINE = "neuros-osd2" Jun 16 19:41:00 nerochiaro: ah ok. it's ok then :) Jun 16 19:41:44 cyrilRomain: well, it's not really ok, because it seems quite a waste to me to download the whole vanilla kernel just to pick up headers which are already provided by the kernel in my repo Jun 16 19:42:24 oe doesn't notice that, but i think it should Jun 16 19:44:32 nerochiaro: yeah but since they have different name, OE cannot guess it's the same kernel sources Jun 16 19:45:55 nerochiaro: or you could localy patch ./linux-libc-headers/linux-libc-headers_2.6.23.bb to point to your non-vanilla kernel Jun 16 19:46:53 cyrilRomain: can i do that via the local.conf file ? Jun 16 19:47:31 or a similar mechanism that doesn't make me touch the actual oe recipes ? Jun 16 19:47:37 nerochiaro: no :-/ Jun 16 19:47:59 ok, fair enough Jun 16 19:48:03 thanks for the tips Jun 16 20:19:08 re Jun 16 20:23:36 does someone know if using the same /tmp dir for .stable and .dev is sane ? Jun 16 20:23:52 (same distro) Jun 16 20:25:05 * flo_lap would not do this Jun 16 20:25:22 * cyrilRomain would not to, but would like :-/ Jun 16 20:27:28 So I've been finding some interesting breakages with oe/bitbake when compiling from a pwd that contains a symlink, or from within a Makefile environment. Jun 16 20:28:07 I have ~/dev/gumstix-oe -> ~/dev/svn/myproj/gumstix-oe Jun 16 20:28:34 and if I try to cd ~/dev/gumstix-oe; bitbake myimage, I get an error on libtool Jun 16 20:28:54 If however I cd ~/dev/svn/myproj/gumstix-oe; bitbake myimage, it compiles fine Jun 16 20:29:11 Anyone care to speculate as to the behavior difference? Jun 16 20:32:59 Fidge1: the GettingStarted guide recommands not to choose a location with symlinks Jun 16 20:33:28 Fidge1: but tbh, I don't know the reason why Jun 16 20:35:22 probably because somewhere symlink is not resolved/followed Jun 16 20:40:10 cyrilRomain: glibc used to break horribly if there was a symlinkl Jun 16 20:40:16 cyrilRomain: it may be fixed now, but no-one is brave enough to check Jun 16 20:41:47 XorA|gone: thanks for the info. It seems that libtool currently complains :-/ But Fidge1 is maybe brave enough to check and reports issues ;) Jun 16 20:49:09 yes, there is a longstanding libtool issue with symlinks. I don't think anybody has ever bothered to fix it. Jun 16 20:50:12 iirc, it does something funny with comparing the result of `pwd` to a precomputed path, which goes wrong in the symlink case because pwd returns the real path. Jun 16 21:40:36 anyone can help my with this error ? http://www.pastie.org/216117 Jun 16 21:43:48 nerochiaro: /home/nerochiaro/neuros/git/oe/angstrom-tmp/work/i686-linux/git-native-1.5.3-r2/temp/log.do_compile.22317 Jun 16 21:44:11 aha... the log is there Jun 16 21:44:15 markus: it's pasted at the bottom of previous paste log Jun 16 21:44:20 looking at various recipes I see functions such do_compile, do_install - I haven't found documentation that describes how/when these are called. Is there a short list and specific order they are called in? Jun 16 21:45:01 tharvey: look at bitbake manual, poky manual can be handy, reading the bb-files. and on some wikis Jun 16 21:46:11 do_fetch -> do_unpack -> do_configure -> do_compile -> do_install -> do_staging -> do_deploy Jun 16 21:46:15 maybe, im not sure. Jun 16 21:46:27 ok thx Jun 16 21:46:28 not sure about install/staging.. Jun 16 21:47:09 fetch fetches sources to oe/build/sources or some download directory, unpack unpacks in work-dir Jun 16 21:47:16 where it's configured and compiled Jun 16 21:47:34 I'm trying to add uImage support to mx31ads kernel recipe... thought it would be really straightforward (ie define KERNEL_IMAGETYPE="uImage") and I'm done - but it would appear not... perhaps I'm finding the mx31ads kernel recipe is not very well written... Jun 16 21:47:44 and its installed into staging dir, and deployed to ipk-packages in deploy-dir? Jun 16 21:48:47 seems like uImage imagetype handling is in linux.inc which is not included by mx31ads kernel recipe and when I include it I end up with some chicken and egg issues with regards to do_deploy being called before kernel is built Jun 16 21:49:16 hmmm, sorry. maybe somebody else can help you Jun 16 21:49:28 i am an oe-noob :) Jun 16 21:50:09 I think the mx31ads recipe is just a poor recipe... I've been beating my head against the wall trying to understand how to make it do what I want and it doesn't seem to work like some of the other kernel recipes Jun 16 21:50:12 nerochiaro: i think just something is wrong with the bb-file or something. a bug? Jun 16 21:50:34 markus: i don't know. it's from oe's git, i just updated Jun 16 21:51:27 nerochiaro: yes... it can still fail as you learned Jun 16 21:52:09 tharvey: what is mx31 anyways? Jun 16 21:52:25 look at other recipes and modify those Jun 16 21:52:30 markus: sure, i was just wondering if anyone around here could help find out why it failed Jun 16 21:52:57 nerochiaro: sorry, im not helping. Jun 16 21:53:20 marcan, mx31 is a freescale cpu... mx31ads is freescale's dev board Jun 16 21:53:37 oops... mean for markus Jun 16 21:54:07 but is really recipes for building linux kernels so different from eachother, its the kernel, some patches and a config right? Jun 16 21:55:29 *are !!! omg Jun 16 21:55:32 and so on... Jun 16 21:55:36 right... should be very basic one would think... seems like the mx31ads kernel recipe is odd. ie the issue I have is related to 'addtask deploy before do_build after do_compile' - not sure I understand the diff between build and compile but it essentially calls it before kernel is built which is wrong Jun 16 21:56:19 isn't build a collection of actions? Jun 16 21:56:51 build = {fetch->....compile...} Jun 16 21:56:56 or...? Jun 16 21:57:00 what Jun 16 21:57:12 dunno... thats why I was asking where to look for description of these Jun 16 21:58:23 http://oe.linuxtogo.org/user-manual&dpage=usage_workwithsinglerecipe <--- short description Jun 16 21:58:27 but not enough Jun 16 22:06:48 ya, thats decent documentation for actions - I suppose classes can add new actions and *_prepend *_append will add items before/after already defined actions Jun 16 22:08:05 looks like do_deploy is something that mx31ads 'adds' and uses 'addtask deploy before do_build after do_compile' to put it in the action chain, but I haven't found where do_build comes from Jun 16 22:12:44 me neither... i don't even know where the which documenation i should read. Jun 16 22:12:58 if there is any. Jun 16 22:22:14 I'm still not understanding PREFERRED_PROVIDER evidently. I've created mymachine-kernel_2.6.22.6.bb and set PREFERRED_PROVIDER_virtual/kernel="mymachine-kernel" yet am still getting told to define PREFERRED_PROVIDER when building virtual/kernel Jun 16 22:22:30 Hmm Jun 16 22:22:35 COMPATIBLE_MACHINE problem? Jun 16 22:23:33 Ie for one of my arm targets, I have foomach.conf in conf/machines/ Jun 16 22:23:37 and then in the linux recipe Jun 16 22:23:46 COMPATIBLE_HOST = "arm.*-linux" Jun 16 22:23:51 COMPATIBLE_MACHINE = "foomach" Jun 16 22:23:52 probably... as I haven't understood COMPATIBLE_MACHINE yet..... do I have to define it in kernel recipe? Jun 16 22:24:20 And foomach.conf has (both to be safe): Jun 16 22:24:29 PREFERRED_PROVIDER_virtual/kernel = "linux-foomach" Jun 16 22:24:35 I think its strange that every time you create a new machine you have to go and edit a kernel recipe to add a COMPATIBLE_ for it? Jun 16 22:24:40 PREFERRED_VERSION_linux-foomach = "2.6.23.9" Jun 16 22:25:02 I dunno Jun 16 22:25:15 I think there's some magic you can do to say your new mach is compatible with some existing recipe Jun 16 22:25:33 you think that openembedded is designed to allow simple "plugging in" of custom stuff Jun 16 22:25:34 (and there's generic ones, which is why you get hte define a ... warning) Jun 16 22:25:36 it's not Jun 16 22:25:40 ;) Jun 16 22:25:47 When you see the define a preferred provider warni Jun 16 22:25:56 Does your recipe show up as a possibility? Jun 16 22:25:57 Or no? Jun 16 22:26:07 I think thats what I'm looking for, but am I missing something special about COMPATIBLE_? Do virtual package names get matched on something like $COMPATIBLE_HOST-$COMPATIBLE_MACHINE or something like that? Jun 16 22:26:40 It's not specific to virtual/ Jun 16 22:26:52 Any recipe can do COMPATIBLE_ stuff Jun 16 22:27:00 no, my recipe does not show as a possibility - thats where I'm missing the concept - I don't understand the criteria its using to generate the possibilities Jun 16 22:27:08 (I think there's a userland package or two that's x86 only, for example) Jun 16 22:27:26 I don't quite know the special casing around the kernel Jun 16 22:27:30 But, sec... Jun 16 22:28:09 Hmm, nope, dunno ;) Jun 16 22:29:28 Tartarus: are you the guy with dual boot Angstrom/Maemo? Jun 16 22:29:41 XorA|gone, on my N810? Jun 16 22:29:46 yes Jun 16 22:29:50 I've got it booting from external MMC, yes Jun 16 22:29:57 so maemo/lots of things :) Jun 16 22:30:13 do you know if you can run maemo fine on the Angstrom kernel? Jun 16 22:30:23 jilles: so what is it then? :> i really wonder. Jun 16 22:30:28 Angstrom, dunno. Jun 16 22:30:29 I want to keep the maemo on internal flash, but work on Angstrom on MMC Jun 16 22:30:33 I did poky for a few weeks Jun 16 22:30:40 only problem was the power button didn't work Jun 16 22:30:49 so I had to open terminal and sudo poweroff Jun 16 22:30:57 the poky/angstrom kernel is supposed to have a patch for that Jun 16 22:31:05 pressing the button led to endless not quite resuming cycles Jun 16 22:31:21 hmmm Jun 16 22:31:31 Didn't bug me that much, really Jun 16 22:31:42 I got the N810 to evaluate some more custom stuff Jun 16 22:32:01 markus, it seems to be ok if you're willing to accept distro specific mods pretty much all over the tree Jun 16 22:32:12 so my kernel recipe is not in the list of suggestions when it says I should define PERFERRED_PROVIDER_virtual/kernel - what would be mismatched between my machine conf and my kernel recipe? Jun 16 22:32:12 ah, I actually want to use mine, but also play with Angstrom Jun 16 22:32:21 until point Angstrom can replace maemo for me Jun 16 22:32:33 Ah Jun 16 22:33:04 tharvey, silly question Jun 16 22:33:12 you have inherit kernel in yours, right? Jun 16 22:34:21 I have 'inherit kernel' about half way through my kernel recipe... is there some order dependence? Jun 16 22:35:08 I think the basic problem is I'm not understanding how bb 'names' the recipes my recipe file is 'mymachine-kernel_2.6.22.6.bb' Jun 16 22:35:51 anyway Im about to go keyboard faced, so Im off Jun 16 22:38:06 my machine conf has 'PREFERRED_PROVIDER_virtual/kernel = "mymachine-kernel"' and PREFERRED_VERSION_mymachine-kernel = "2.6.22.6" - is this enough to tell bb that I want to use 'mymachine-kernel_2.6.22.6.bb' for my virtual/kernel? Jun 16 22:38:59 perhaps I'm having trouble seeing the forest through the trees... there are something like 4900 recipes that bb is going through - thats a pretty large example file base Jun 16 22:41:44 perhaps its something really stupid like my kernel recipe is not getting parsed by bb - how would I verify that its in the list of recipes? Jun 16 22:43:17 bitbake -v -D -g your_target 2>&1 | tee log Jun 16 22:43:34 (and you'll get the depends dot files for your_target, to boot) Jun 16 22:47:47 hmmm I think that perhaps is the issue - I don't think my kernel recipe is getting read. My BBPATH looks like '/home/tharvey/oe/build:/home/tharvey/oe/local:/home/tharvey/oe/stable' where stable is the OE tree, and I have my kernel recipe in /home/tharvey/oe/local/packages/linux/mymachine-kernel_2.6.22.6.bb which doesn't seem to be parsed Jun 16 22:48:16 Ah, but what's BBFILES look like? Jun 16 22:48:45 The proper way, roughly, is: Jun 16 22:48:57 * Tartarus pastebins Jun 16 22:49:30 I don't define BBFILES... just BBPATH Jun 16 22:49:39 http://pastebin.com/m518d5e4c Jun 16 22:49:46 I mean in the distro.conf, or your local.conf Jun 16 22:49:50 But that's why :) Jun 16 22:50:40 doh! Jun 16 22:50:56 damn thats it.... Jun 16 22:51:08 I guess I forgot about BBFILES as BBPATH was in my env Jun 16 22:51:31 when I went about adding a local path to add/override files in the oe stable tree Jun 16 22:52:38 why does bb require both BBFILES and BBPATH? Why not just look at *.bb in BBPATH? Jun 16 22:53:38 it's complicated, heh Jun 16 22:53:47 but BBPATH is used for one thing Jun 16 22:53:49 BBFILES another Jun 16 22:55:44 03mwester 07org.oe.dev * r6334a7f1... 10/ (1 packages/meta/slugos-packages.bb): SlugOS Package Feeds - demote ctorrent (do_install fails) Jun 16 22:56:04 I haven't come across the BBFILE_PATTERN_upstream stuff... is that required to prioritize precedence of the bb files? or is the order you place then in BBFILES what is used? Jun 16 22:56:23 It's for correct prioritization Jun 16 22:56:26 You can do without Jun 16 22:56:30 but it's not as correct Jun 16 22:56:46 (I personally went 6mo w/o doing it right, didn't hit a problem then foudn I wasn't doing it 100% correct) Jun 16 22:57:51 seems to be a lot of missing documentation.... Jun 16 22:58:02 http://oe.linuxtogo.org/user-manual&dpage=special_bitbake_collections is far from explanatory Jun 16 22:58:30 tharvey: yeah that's it Jun 16 22:58:59 I'm wondering why those line are not in local.conf.sample, even commented Jun 16 22:59:08 exactly what I was thinking... Jun 16 23:01:45 now my recipe is getting selected - thanks! Jun 16 23:16:58 so does bitbake handle dependencies for bb files? if you change a single recipe do all recipes that depend upon it get rebuilt? Jun 16 23:17:22 No Jun 16 23:19:03 so if you want something to be rebuilt rm -rf it from its place in the work dir? all knowledge of what actions have been performed on it are there right? Jun 16 23:19:44 Well, almost Jun 16 23:19:52 If you're on .dev and have packaged-staging in Jun 16 23:19:58 you can bitbake -c clean lib Jun 16 23:20:06 and i do believe that will back out the staged stuff Jun 16 23:22:27 what about rm -rf of the work dir.... that enough to cause all but the toolchain to rebuild? Jun 16 23:23:14 No Jun 16 23:23:24 If you delete the workdir, nothing Jun 16 23:23:36 If you're OK with just restaging thigns on top of the old build Jun 16 23:23:39 bitbake -c rebuild lib Jun 16 23:24:41 by 'lib' you mean the particular recipe or specifically a recipe called 'lib'? Jun 16 23:45:33 Tartarus, 'bitbake -c rebuild virtual/kernel' seemed to work to rebuild but spewed a lot of 'ERROR: QA Issue:' messages at the end from do_qa_staging **** ENDING LOGGING AT Tue Jun 17 02:59:57 2008