**** BEGIN LOGGING AT Wed Jul 04 02:59:56 2007 Jul 04 06:47:01 morning Jul 04 07:03:34 good morning all Jul 04 07:05:29 re Jul 04 07:05:50 hey zecke|tv Jul 04 07:12:34 hi guys Jul 04 07:15:49 looks like we need to expand sanity a bit Jul 04 07:16:00 * * OE Bug 2561 has been RESOLVED (FIXED) by Jul 04 07:16:02 * * qemu links with zlib, but zlib is compiled after qemu Jul 04 07:16:04 * * http://bugs.openembedded.org/show_bug.cgi?id=2561 Jul 04 07:16:13 to check does host tools from list are available and if yes then add them into ASSUME_PROVIDED Jul 04 07:19:13 steliosk: hi, got my info about magicbox? Jul 04 07:19:15 wow I hate my uni Jul 04 07:25:41 hrw: I had the same idea yesterday :) Jul 04 07:27:15 koen: then you was first and you will write it Jul 04 07:27:16 lD Jul 04 07:27:27 crap Jul 04 07:27:38 * koen passes it on to zecke Jul 04 07:28:05 03hrw 07org.oe.dev * rf3281034... 10/ (1 packages/qemu/qemu-native.inc): qemu-native: added zlib-native to DEPENDS - close #2561 Jul 04 07:33:54 good morning all Jul 04 07:34:51 hrw : yes, i did. We should "ping" them a 2-3 weeks time again Jul 04 07:35:06 koen : hi Jul 04 07:36:30 hey steliosk Jul 04 07:37:53 steliosk: I would rather 3-4 so they will have a time to mail us Jul 04 07:51:27 hmm, this ain't good: http://rafb.net/p/AERW7279.html Jul 04 07:51:54 seems some "please install into staging"-part is missing Jul 04 07:53:43 another proof why one shouldn't run OE as root Jul 04 07:53:54 yes, indeed :-) Jul 04 07:54:06 * oxo considered the option for some nanoseconds Jul 04 07:54:23 brb, driving lessons~ Jul 04 07:55:37 koen: add that to sanity? Jul 04 07:56:10 if `rm -rf /` == success then echo 'do not use OE as a root'; Jul 04 07:56:46 hrw: Nice :) Jul 04 07:58:37 good morning Jul 04 08:04:16 morning Jul 04 08:05:11 hey XorA Jul 04 08:05:45 XorA: +44.........58 is still valid? Jul 04 08:06:25 koen: yes Jul 04 08:08:33 Do not use Bitbake as root. Jul 04 08:08:33 root@home:~/devel/OH/poky/build# Jul 04 08:13:25 Anyone else seeing this effect that glibc-intermediate_2.3.2+cvs20040726 is pulled in for no obvious reason? Jul 04 08:14:56 hi florian Jul 04 08:15:07 florian bitbake bla -g Jul 04 08:15:13 depends.dot Jul 04 08:15:43 woglinde: that's an idea... Jul 04 08:15:52 yeah Jul 04 08:37:12 03hrw 07org.oe.dev * r96de2576... 10/ (1 classes/sanity.bbclass): sanity: if OE is used by root user (userid 0) then raise error (from Poky) Jul 04 08:48:00 * * OE Bug 2567 has been RESOLVED (FIXED) by Jul 04 08:48:02 * * Patch to generate atk version 1.19.3 Jul 04 08:48:04 * * http://bugs.openembedded.org/show_bug.cgi?id=2567 Jul 04 08:49:31 hrw: what nutter ran OE as root? Jul 04 08:50:32 XorA: look one hour back Jul 04 08:52:14 xora lol Jul 04 08:58:03 03Joaquim 07org.oe.dev * r0e150443... 10/ (1 packages/atk/atk_1.19.3.bb): atk: added 1.19.3 - close #2567 Jul 04 09:01:54 morning ! Jul 04 09:10:47 I have just found the reason why qt4-x11 does not build on x86. The configure script gets the wrong qt_arch from bitbake. Jul 04 09:11:35 I have put the correct string into qt4_arch.inc and it builds now. Jul 04 09:12:26 Anybody who can do the modification in the original file ? Jul 04 09:12:29 jan_js: please update the bugtracker with that info Jul 04 09:12:36 o.k. Jul 04 09:13:37 hmm, but does anyone know how I could fix http://rafb.net/p/AERW7279.html ? Jul 04 09:13:40 jan_js: what is given to configure. Didn't I say to look at it? Jul 04 09:15:30 hrm... oe dependencies are too complex ;) Jul 04 09:15:33 http://linuxtogo.org/~florian/img/oe-complexity.png Jul 04 09:16:01 woglinde: it looked too easy ;) Jul 04 09:16:06 florian where is glibc now camee from? Jul 04 09:16:20 florian hm bitbake bla -e | grep glibc Jul 04 09:16:26 could help too Jul 04 09:18:14 zecke: bitbake gives x86 to configure, which seemed sane to me on first sight. However, There is a (relatively new) patch, that circumvents the complete architecture check of the configure script. Jul 04 09:18:43 * florian earching ;) Jul 04 09:18:44 so now the configure script uses the x86 directly, but it only knows I386. Jul 04 09:18:48 eh searching Jul 04 09:19:15 florian when is viewmtn working again? Jul 04 09:19:31 * woglinde crawls fast away Jul 04 09:19:46 florian: scary ! Jul 04 09:20:05 jan_js thats the dependcy graph Jul 04 09:20:08 so what Jul 04 09:20:19 you hive 6 or more tasks per package Jul 04 09:20:22 have Jul 04 09:27:17 Well, actually that patch removes the architecture check, probably for good reason. I think thats o.k. as long as qt4_arch.inc maps the architecture correctly. Jul 04 09:27:51 woglinde: Just my first impression ;-) Jul 04 09:30:53 jan_js: there is one .inc file that maps from OE HOST_ARCH to Qt's one, you need to change that Jul 04 09:31:07 jan_js: ah okay, you found that Jul 04 09:33:02 anyone here who tried Angstrom on a 770 lately? Jul 04 09:33:21 I am adding a comment to that bug now. Jul 04 09:33:33 florian are there now all kernel-modules awailable? Jul 04 09:34:27 woglinde: i'm not sure - the last time i tried was with Familiar and the original initrd. Jul 04 09:41:49 re mickeyl Jul 04 09:51:21 is there articles for new distro? Jul 04 09:51:33 hi Jul 04 09:52:15 archer: such as 'how to create new distro fromOE because angstrom does not fits me'? Jul 04 09:56:58 * oxo has another question: Jul 04 09:57:36 is there a less complex way to 'regenerate ipk indices' than 'PATH=/bla/tmp/staging/$HOST_ARCH/bin ipkg-make-index ...' ? Jul 04 09:58:03 my distro can't be built with command: bitbake xxx Jul 04 09:58:28 oxo: bitbake package-index Jul 04 09:58:34 hrw, is there any links? Jul 04 09:58:40 archer: but it is built with OE? Jul 04 09:58:51 hrw: nice! tnx! Jul 04 09:59:12 yes Jul 04 10:00:00 archer: you can create meta recipe which will include list of packages which you want to build for distro and then do "bitbake archer-distro-all-supported-packages" Jul 04 10:00:14 archer: I have 'created' a new distro based on the generic one - if you like i send you the config file. Jul 04 10:01:00 jan_js: i do it the same Jul 04 10:01:42 I would base own distro on angstrom Jul 04 10:01:44 archer: All right. So whats the problem ? Jul 04 10:01:50 or just use angstrom even Jul 04 10:02:30 it reports ERROR: Nothing provides dependency xxxx Jul 04 10:03:16 what is "it"? Jul 04 10:03:39 basically you just put a line like : require conf/distro/generic.conf into your new config file, and it should work. Jul 04 10:04:11 archer: sorry, have to go to lunch now. Talk to you later! Jul 04 10:05:13 ok, i am trying Jul 04 10:07:53 Guys, is it possible to fetch only the Kernel sources for Akita SL-C1000 out of the OE tree ? Jul 04 10:08:18 bitbake -c fetch virtual/kernel Jul 04 10:08:47 or bitbake -b ../org.openembedded.dev/packages/linux-rp_2.6.*.bb -c fetch Jul 04 10:09:04 awesome. Jul 04 10:09:08 Thank you XorA Jul 04 10:09:29 By the way, is it correct to say that the most up-to-date SL-C1000 are in there ? Jul 04 10:09:40 SL-C1000 kernel sources I meant, sorry Jul 04 10:09:52 I guess yes Jul 04 10:10:06 Ecco: depends what you think as 'most up-to-date' Jul 04 10:10:16 Ha. Well, good question Jul 04 10:10:40 2.6.22-rc7 is more up to date then 2.6.21? but what if 2.6.21 patchset has more stuff then 2.6.22-rc7? Jul 04 10:10:53 You got a point Jul 04 10:11:10 Let's say I'm more interested in features Jul 04 10:11:12 But Jul 04 10:11:23 What really matters for me is being able to "update" pretty simply Jul 04 10:12:01 (for instance I don't mind if the tree is still at 2.6.20 for instance, as long as a future "bitbake update" or something alike will update my tree in a couple month) Jul 04 10:12:22 And I'm not looking for a totally bleeding-edge kernel. Stability is more important Jul 04 10:13:46 Ecco: 2.6.21 is current testing kernel Jul 04 10:17:02 Oh, one last question Jul 04 10:17:14 what's the difference between virtual/kernel Jul 04 10:17:24 and ../org.openembedded.dev/packages/linux-rp_2.6.*.bb Jul 04 10:18:25 linux-rp provides virtual/kernel and if you MACHINE=blah is right, it will select the right kernel Jul 04 10:18:27 Ecco: thats long story - read OE manual Jul 04 10:18:41 Hmm, ok, thanks a lot guys Jul 04 10:19:45 Well Jul 04 10:19:54 Actually I've been using OE in the past Jul 04 10:20:07 And it's a pretty massive thing Jul 04 10:20:23 I mean, personnally I'm only interested in kernel sources Jul 04 10:20:38 Do I have to go through the whole OE installation ? Jul 04 10:20:58 no Jul 04 10:21:05 look at the .bb file Jul 04 10:21:11 and took all per hand Jul 04 10:21:22 look at SRC_URL Jul 04 10:21:31 bzw URI Jul 04 10:22:13 Hmm, nice idea ^^ Jul 04 10:22:27 I already did this in the past, but the problem is it's not automated Jul 04 10:22:36 Well Jul 04 10:22:45 I'm gonna try to put up a nice little script Jul 04 10:22:49 yes Jul 04 10:22:52 Someone might be interested in this as well Jul 04 10:23:13 (Personnally I'm trying to use Debian on my Zaurus) Jul 04 10:23:25 another one... Jul 04 10:23:34 what people see in Debian on Zaurus? Jul 04 10:24:07 other then 'I need to build my kernel by hand' and 'I have no idea how to get X working like it works in OZ/Angstrom/pdaX' Jul 04 10:24:32 hrw hm inventing new wheels Jul 04 10:24:41 wheels with corners Jul 04 10:24:51 square wheels Jul 04 10:24:53 ^^ Jul 04 10:24:56 No, actually Jul 04 10:25:05 * XorA doesnt see a problem with running armel on Zaurus Jul 04 10:25:08 people have to much time Jul 04 10:25:08 I really gave OZ/OE/Angstrom a run Jul 04 10:25:12 debian armel Jul 04 10:26:03 XorA: I know that is should work - atleast on c3x00 as others will need sd/cf for rootfs Jul 04 10:26:10 I have two problems with OE : Jul 04 10:26:21 -> First, it's really hard to customize Jul 04 10:26:26 lol Jul 04 10:26:29 It's kinda like "GPE or nothing" Jul 04 10:26:34 lol Jul 04 10:26:37 And I personnally don't like GPE Jul 04 10:26:48 Of course you can install another custom WM Jul 04 10:26:55 But did you guys really tried it ? Jul 04 10:27:02 no Jul 04 10:27:04 opie runs Jul 04 10:27:07 I mean, the IceWM package is bugged to death Jul 04 10:27:15 than fix it Jul 04 10:27:19 Xfce doesn't even run Jul 04 10:27:20 Ecco: you know why icewm package is bugged? Jul 04 10:27:23 sure, I run OpenMoko, no sign of GPE Jul 04 10:27:27 hrw : I've no idea Jul 04 10:27:38 Actually it was already pretty hard to get it working Jul 04 10:27:51 Ecco: because no one use it. and those which use icewm refuse to report bugs Jul 04 10:28:01 Yeah, you got a point Jul 04 10:28:29 But anyway, I don't even wanna use it Jul 04 10:28:30 Ecco: so if you know how to use icewm then help us fixing it Jul 04 10:28:37 Well Jul 04 10:28:47 the first problem was to actually start it instead of GPE Jul 04 10:29:00 same with xfce. we have it because one day I updated it to newest version. but not tested on device Jul 04 10:29:16 Ecco: look at fluxbox-gpe recipe to know how to replace matchbox in gpe Jul 04 10:29:32 ok, thanks Jul 04 10:29:48 But now let's say I wanna use enlightenment Jul 04 10:30:00 ipkg install e-wm Jul 04 10:30:12 Well Jul 04 10:30:19 How comes it's not mentionned there, then : Jul 04 10:30:23 that uses update-alternatives to replace matchbox-wm Jul 04 10:30:43 http://www.angstrom-distribution.org/repo/?action=section§ion=x11%2Fwm Jul 04 10:30:55 Hmm Jul 04 10:31:08 Hmm, well, sounds nice then Jul 04 10:31:17 Ok, I'm gonna give up installing Debian for now Jul 04 10:31:52 Another question, then : what policy do you guys suggest about CF cards use ? Jul 04 10:32:05 I.E. : Is rootfs on CF a bad idea ? Jul 04 10:32:16 (I don't use any other CF-card) Jul 04 10:32:52 I would not use rootfs on CF Jul 04 10:32:55 I usually have rootfs on SD for my machines (non-zaurus) Jul 04 10:33:04 so I can use the CF slot for other stuff Jul 04 10:33:05 rootfs on CF card work fine if you dont use CF slot Jul 04 10:33:13 what XorA says Jul 04 10:33:25 C3200 is rootfs on CF card :-D Jul 04 10:33:57 Yeah XorA, you're right Jul 04 10:35:19 What's the recommended way to install the root FS on a CF card, then ? Jul 04 10:36:17 RP: ping Jul 04 10:38:58 florian: pinged Jul 04 10:41:10 oh Jul 04 10:41:22 * florian has his own theory Jul 04 10:41:33 which is? Jul 04 10:41:47 moin Jul 04 10:42:36 ha no Jul 04 10:43:18 o.O Jul 04 10:43:47 $TMPDIR/deploy/glibc/ipk/Packages is the wrong file if you want to populate a SDK Jul 04 10:44:08 If you re using Angstrom at least Jul 04 10:44:10 that one should be empty Jul 04 10:44:26 right... but iirc this is correct for Poky Jul 04 10:44:49 florian: because Poky does not split ipk/ dir like OE do Jul 04 10:45:54 hrw: ok, then it's quite obvious that one of the changes merged in from poky broke building sdks Jul 04 10:47:23 didn't I fix that a while ago? Jul 04 10:48:57 koen: I remeber I built Angstrom SDKs successfully some time ago. Jul 04 10:50:34 ah, I only fixed meta-toolchain Jul 04 10:51:24 koen: ok, found - i'll try if it work for meta-sdk as well Jul 04 10:51:45 florian: http://rafb.net/p/iiY61249.html Jul 04 10:51:47 hrw, still would not work Jul 04 10:51:52 Is bc required software for OE now? Jul 04 10:52:10 laibsch? Jul 04 10:52:14 Laibsch: the kernel.bb's have it used since years Jul 04 10:52:17 even predates OE Jul 04 10:52:37 ya Jul 04 10:52:44 koen: thanks! Jul 04 10:52:49 that's from the collie kernel Jul 04 10:52:53 in OZ Jul 04 10:52:57 well, I only received an error today Jul 04 10:53:03 And for 2.6.21 Jul 04 10:53:12 bc is req only for collie kernels Jul 04 10:53:22 http://www.openembedded.org/wiki/RequiredSoftware mentions only 2.4 kernels in relation to bc Jul 04 10:53:36 hrw: Including 2.6 kernels? Jul 04 10:53:40 archer: hard to tell why it does not work when you do not show what you work on Jul 04 10:53:46 Laibsch: look at linux-rp.inc Jul 04 10:54:48 * Laibsch assumes that is a yes Jul 04 10:55:23 we talk about it before Jul 04 10:55:46 * Laibsch updated http://www.openembedded.org/wiki/RequiredSoftware Jul 04 10:55:46 my distro can't be built with command: bitbake xxx Jul 04 10:56:02 so use angstrom, that works Jul 04 10:56:31 xxx, pr0n0 linux Jul 04 10:56:36 koen, i write one myself Jul 04 10:56:45 archer: and you did not show distro config, local.conf or error output Jul 04 10:56:48 which doesn't work Jul 04 10:57:04 ergo, angstrom is better, since that works Jul 04 10:57:20 *g* Jul 04 11:01:44 MACHINE = "awg2410" Jul 04 11:01:44 DISTRO = "instrument" Jul 04 11:01:44 BUILD_ARCH = "i686" Jul 04 11:01:44 it is local.conf Jul 04 11:01:44 the distro config is almost the same as openmoko Jul 04 11:03:48 bye Jul 04 11:04:34 I say again, use angstrom if you don't want to give use usefull info to debug Jul 04 11:04:38 koen: Can I use angstroem to build for a x86 machine then ? Jul 04 11:04:44 jan_js: yes Jul 04 11:04:59 x86, sparc, arm, mips, avr32, etc Jul 04 11:06:03 koen: And is it possible to derive my distro from angstroem , e.g. to add more packages and maybee enlightenment or qt ? Jul 04 11:06:20 koen: x86 is ok. Jul 04 11:06:38 why do you want to 'derive' it? Jul 04 11:06:42 just use it as is? Jul 04 11:06:51 a distro doesn't define packages, only policy Jul 04 11:07:46 koen: Then I would have to install additional packages later, by hand. Jul 04 11:08:16 no, you just write janjs-image.bb Jul 04 11:08:18 koen: Which is all right, if I have to do it once or twice ... Jul 04 11:08:30 images aren't tied to distros Jul 04 11:09:03 http://www.angstrom-distribution.org/repo/?action=details&pnm=e-wm <- enlightenment http://www.angstrom-distribution.org/repo/?action=details&pnm=qt4-x11-free <- qt4 Jul 04 11:09:20 (no qt4 for x86, due to the bug you hit) Jul 04 11:09:35 no, that should be fixed now. Jul 04 11:10:11 jan_js, would you please send me your conf file? my conf file still can't work Jul 04 11:10:30 koen: So I am going to define my own image, but use angstroem distro, right ? Jul 04 11:10:57 yes Jul 04 11:11:16 koen: Thanks for the hint, I am going to try that one. Jul 04 11:11:52 archer: Looks like that might be the way to for you, two ? Jul 04 11:13:47 how does oe handle defconfigs ?? it seems my defconfig is overwritten at one point Jul 04 11:15:00 rob_w|mis: depends on the kernel Jul 04 11:15:14 rob_w|mis: some kernels munge some entries and run oldconfig over it Jul 04 11:15:20 hmm seems my symbols are missing Jul 04 11:18:24 archer: If you want to have it by mail, please give me your mail adress. Jul 04 11:23:50 koen: short question: glibc2.5 killed linuxthreads, uses ntpl only which requires tls? Jul 04 11:24:20 afaik that is correct Jul 04 11:24:39 you can cheat by copying the linuxthreads dir and hacking configure a bit Jul 04 11:24:57 xydarcher@126.com Jul 04 11:25:18 2.6.XX kernels always run make oldconfig when you change .config Jul 04 11:25:32 so if i have an ancient kernel from a no longer existing company, I should use 2.4 or 2.3. How should I override angstrom to build a older glibc? Jul 04 11:26:24 try 2.4 with PREFERRED_VERSION_glibcfoobarsomething_dnard = "2.4" Jul 04 11:26:32 see angstrom-2007.1.conf Jul 04 11:26:59 I know how I can do it, I just wondered about policy Jul 04 11:27:25 put it in angstrom-legacy, that's a free-for-all :) Jul 04 11:30:23 thx Jul 04 11:30:23 i am considering using angstrom, but i don't know whether it would do Jul 04 11:36:19 archer: what should it do? Jul 04 11:38:24 "trying to assign nonexistent symbol" says configure of my kernel , how do i make it known ? Jul 04 11:39:33 rob_w|mis: your config use values which were removed Jul 04 11:39:58 those are new values .. how do i populate them so config know them Jul 04 11:41:41 koen: ok, it was a little bit more complex because of the changed architecture names for the sdk packages - but it should work now. Jul 04 11:41:54 florian: cool Jul 04 11:43:37 * florian pushes and tries the same with the other mata-sdk* packages Jul 04 11:48:57 CONFIG_SERIAL_PXA_ALTMAJOR hmm, someone finally did it then Jul 04 11:49:46 RMK promised to do it in linux 2.5.x Jul 04 11:51:26 florian: where do you push to? none of the OE mirrors have it? Jul 04 11:52:21 koen: ah had a minor problem updating... one sec Jul 04 11:57:05 archer: I just did a little reading and found that http://bec-systems.com/web/content/view/59/9/ best describes the method koen suggests. Jul 04 11:57:41 archer: actually you could apply that to the generic distro as well... Jul 04 11:58:22 03florian 07org.oe.dev * r359ad4a8... 10/ (1 classes/sdk.bbclass): sdk.bbclass: Add function to create sdk pacakge indices. Jul 04 11:58:26 03florian 07org.oe.dev * r1b058511... 10/ (1 packages/meta/meta-sdk.bb): meta-sdk: Update to support new sdk arch names and splitted package directories. Jul 04 12:00:43 btw. I think that the preferred method of adding packages to a distro should be at a more prominent location in the oe doku! Jul 04 12:01:24 jan_js: distro does not contain packages Jul 04 12:01:41 distro builds packages Jul 04 12:02:00 o.k., to an image, of course. Jul 04 12:02:14 florian: could you have a look at the rest of #2393 to fix meta-toolchain? Jul 04 12:02:17 jan_js: image creation is in OE manual Jul 04 12:03:05 Yes, thats what I assumed, but i could onlx find it at becks site ;-( Jul 04 12:03:34 which site? Jul 04 12:03:41 OE manual is on OE website Jul 04 12:03:44 http://www.openembedded.org/user-manual Jul 04 12:04:28 koen: i'll try Jul 04 12:04:55 ah, bec-systems site : http://bec-systems.com/web/content/view/59/9/ Jul 04 12:05:02 florian: once that works we can merge external-toolchain.conf from poky Jul 04 12:05:40 florian: that way you can hand people an SDK + oe checkout and have them start building right away instead of waiting on glibc Jul 04 12:06:02 koen: ok, sounds useful... i had quite some people asking about external toolchains lately Jul 04 12:06:47 people asking about external toolchains should cancel their contract with mvista or timesys if they are going to use OE anyway Jul 04 12:07:32 hrw: o.k., there is a 6 - liner on page 10, but that was probably too concise for me ;-) Jul 04 12:08:48 jan_js: look at angstrom-console-image.bb as example Jul 04 12:08:54 koen: yes sure, but that provides them an easier way to stay compatible for a while. Jul 04 12:09:06 writing docs never was my strong side Jul 04 12:09:30 hrw: thanks, I'll have a look into that and see if I can learn ... Jul 04 12:10:49 hrw: No, it's great that there are two manuals out there allready ! Jul 04 12:16:42 hrw: However, the method of adding packages seems basically simple, so it should be in there. Jul 04 12:29:15 += adds space, _append does not - right? Jul 04 12:35:15 hrw: right Jul 04 12:35:29 hence the foo_append = " bar " Jul 04 12:35:54 re Jul 04 12:41:43 hi [lala] - long time no see Jul 04 12:42:03 <[lala]> hi hrw, that's sadly true Jul 04 12:42:15 <[lala]> almost no time for oe lately Jul 04 12:50:22 tried to build angstrom-e-image and get errors like this: Jul 04 12:50:27 Could not inherit file classes/rootfs_${IMAGE_PKGTYPE}.bbclass while parsing /stuff/org.openembedded.dev/packages/angstrom/angstrom-e-image.bb Jul 04 12:50:42 What mistake did I make ? Jul 04 12:53:43 jan_js: sounds like you did not set a distribution or your distribution does not define which package type to use. Jul 04 12:54:27 hmm.. I set angstrom as distribution in my local.conf Jul 04 12:55:24 "angstrom-2007.1" ? Jul 04 12:56:21 no, just angstrom. I will try angstrom-2007.1 then . Jul 04 12:57:54 03florian 07org.oe.dev * r62723979... 10/ (1 packages/meta/meta-sdk-gpe.bb): meta-sdk-gpe: Update to support new sdk arch names and splitted package directories. Jul 04 12:58:02 03florian 07org.oe.dev * r5f6bdd72... 10/ (1 packages/meta/meta-sdk.bb): meta-sdk: Some more directory layout related fixes. Jul 04 12:58:10 florein. thanks! Jul 04 12:58:21 ähm florian Jul 04 12:58:37 jan_js: yw Jul 04 13:03:52 I am trying to build angstrom-minimal-image and do_rootfs is failing... http://www.deadlock.dhs.org/jin/log.do_rootfs.22123.txt Jul 04 13:03:54 any ideas? Jul 04 13:05:15 uh... looks nasty... parse error Jul 04 13:06:49 yes.. so I wonder what went wrong Jul 04 13:11:27 Jin^eLD: which bitbake do you use? Jul 04 13:11:59 1.8.6 Jul 04 13:12:18 you think my bitbake installation may be broken? Jul 04 13:12:39 ~lart git for being a worthless peice of crap Jul 04 13:12:39 * ibot_ chops git in half with a free AOL CD for being a worthless peice of crap Jul 04 13:12:51 just a guess but 1.8.6 should be fine Jul 04 13:14:04 Jin^eLD: what version is /usr/local/bin/bbimage? is that from 1.8.6 as well? Jul 04 13:14:17 oops, let me check Jul 04 13:15:13 I indeed have two bbimage files, in /usr/bin and in /usr/local/bin Jul 04 13:15:16 moment Jul 04 13:15:27 * XorA wonders why the solution to all problems in git is just to delete all and start again, what the hell is the point in it Jul 04 13:15:40 ok, the one in local is from who konws where, sorry Jul 04 13:16:11 *retrying* Jul 04 13:16:47 zecke: thanks, problem solved :) Jul 04 13:23:47 Jin^eLD: don't forget to add FEED_ARCH_as532x = "armv5te" to angstrom-2007.1.conf Jul 04 13:23:58 roger that Jul 04 13:24:09 what does it do, btw? Jul 04 13:24:21 that makes the online feeds work Jul 04 13:24:34 aah, cool! Jul 04 13:25:26 http://www.openembedded.org/repo/org.openembedded.dev/conf/distro/angstrom-2007.1.conf <- the FEED_URIs bit Jul 04 13:25:33 koen: Well, so OPIE code analysis shows that what can be realistically done is to put its binaries to /usr/bin, major libraries to /usr/lib, and all the rest to /usr/share/opie. Would this work for Angstrom? Jul 04 13:25:45 I want to make a package out of the feed config so 'ipkg upgrade' can poke at them Jul 04 13:26:08 psokolovsky: that should work Jul 04 13:26:23 anything FHS\{/opt} works for me Jul 04 13:26:44 FHS is stupid, it favors reiserfs ;) Jul 04 13:26:59 psokolovsky, maybe /usr/lib/opie? Not that it matters, but there may be arch-depended data which theoretically should go into /usr/lib :) Jul 04 13:27:23 no standards are sane, but parts of FHS make sense. Jul 04 13:27:31 koen: thanks. because otherwise, opie just made with assumption that there's single root for all its stuff, instead of multiple roots like $bindir, $libdir, etc. Changing that would mean patches each app separately, and that's completely different story. Jul 04 13:28:15 polyonymous: well, let's make choice. yes, there're plugins to be there. Jul 04 13:28:17 /usr/lib/(lib-exec)/opie/{plugins,applets,apps} Jul 04 13:28:37 and this completely different story is actually the way it should be done in the long run. Jul 04 13:28:51 psokolovsky, I'd suggest lib if everything should go in one place. Jul 04 13:28:54 polyonymous: yes. any news from bipolar? Jul 04 13:29:03 the long run is qt4(-x11) Jul 04 13:29:03 polyonymous: ok Jul 04 13:29:36 psokolovsky, not sure if it was bipolar or ljp, but one of them said he'd expect "some news" maybe "next week". Well, not much, in short :) Jul 04 13:30:19 polyonymous: ;-). so it's exactly why it's "completely different story" ;-) Jul 04 13:30:20 koen, yes, speaking of which - I have qt4-kdepimpi, but right now my unfinished bb pulls it from git repository on my host, which doesn't sound suitable for inclusion into OE to me :) Jul 04 13:30:53 polyonymous: only if you promiss to send the changes back upstream to kde-pim (not /pi) :) Jul 04 13:30:56 psokolovsky, I didn't argue with that, I only said that this "completely different storry" it's the way to go in the long run. :) Jul 04 13:31:21 koen, I doubt they want these changes. It's a dirty hack. Jul 04 13:31:51 koen: it is no use to send these patches upstream Jul 04 13:31:54 koen, besides, the changes based on kdepimpi. Jul 04 13:32:04 absolutely, zecke. Jul 04 13:32:31 kdepim/pi is too different to kdepim 3.x Jul 04 13:32:43 and way away from sanity too :) Jul 04 13:32:44 and kdepim 3.x is in phase 'no new stuff rather' Jul 04 13:33:20 and it seems like kde4 would be the way to go, but I got an impression it's overbloated. Jul 04 13:35:18 polyonymous: kde4 is abstraction on top of abstraction on top of abstraction Jul 04 13:35:36 koen, exactly. Jul 04 13:35:58 e.g phonon is completely unnecessary Jul 04 13:36:47 koen: last time I checked FreeBSD, PC-BSD, OpenSolaris, Darwin, Windows Vista, Windows Mobile didn't ship with HAL preinstalled Jul 04 13:36:53