**** BEGIN LOGGING AT Sun Mar 18 02:59:56 2007 Mar 18 03:17:37 v8jlene: seems to me the best way that I can possibly fix this will be to add options to set the include/lib dirs of the other extensions... Mar 18 03:18:28 is anyone actually using angstrom? I'm getting errors on freetype now... :-( Mar 18 03:48:59 JustinP: I'm using angstrom(got the binary dist working) Mar 18 03:52:19 SirManiac: well, I mean compiling...freetype is failing now... Mar 18 03:53:52 I've compiled angstrom-x11-image, and nut working on other programs Mar 18 03:54:33 Just checked, I got freetype working Mar 18 03:56:38 hmmm... Mar 18 03:56:52 it's giving me this | checking for suffix of native executables... ./configure: line 2943: i686-pc-linux-gnu: command not found Mar 18 03:57:04 Don't remeber any problem... Mar 18 03:58:05 that's wierd, sounds like it trying for i686-pc-linux-gnu-gcc, but forgetting the gxx Mar 18 03:58:30 I've got to proofread more before I press enter. Mar 18 04:00:02 JustinP: what is your target? Mar 18 04:04:38 SirManiac: arm (spitz) Mar 18 04:05:19 JustinP: ok, same as mine Mar 18 04:06:31 try a clean(bitbake -c clean -b packages/freetype/freetype_2.3.0.bb), and restart bitbake as before, it may be a glitch Mar 18 04:07:12 ls Mar 18 04:07:26 oops, wrong window :-[ Mar 18 04:07:40 I've tried a -c rebuild, no difference Mar 18 04:07:58 and this is a completely clean build as of today Mar 18 04:08:28 hmmm Mar 18 04:10:25 searching for that error in my tree.. Mar 18 04:11:43 do you have the gcc executable in your path? Mar 18 04:13:20 just after that error in the configure scripts is "error: native C compiler is not working" Mar 18 04:14:23 yep Mar 18 04:14:37 heh, of course gc is in my path....normally at least Mar 18 04:14:41 everything else built fine Mar 18 04:14:55 and just above that I see | checking for i686-pc-linux-gnu-gcc... i686-pc-linux-gnu Mar 18 04:15:16 thought i'd check with me it's the dumb things like that that stump me the most. ;) Mar 18 04:18:00 check the tmp/work/armv5te-angstrom-linux-gnueabi/freetype-2.3.0-r0/temp/run.do_configure... script, and look for CC= , and see what it has. Mar 18 04:18:43 while in that file, see if PATH includes the location of your native gcc Mar 18 04:19:53 CC="ccache arm-angstrom-linux-gnueabi-gcc -march=armv5te -mtune=xscale" Mar 18 04:20:04 BUILD_CC="ccache gcc" Mar 18 04:20:35 That's probably the problem, ccache is confusing the configure script Mar 18 04:20:42 PATH includes /usr/bin which is where gcc is Mar 18 04:20:43 had some trouble with that myself Mar 18 04:21:06 so...I should turn off ccache somewhere? Mar 18 04:21:14 add the CCACHE="" to your local.conf Mar 18 04:21:36 trying Mar 18 04:21:45 to skip the re-parsing, you can explicitly run the individual bb file ro see if that works. Mar 18 04:21:54 true Mar 18 04:23:41 same error... Mar 18 04:24:16 no ccache in the CC vars Mar 18 04:24:20 just for sanity's sake check the latest run.do_configure and make sure the ccache is gone. Mar 18 04:24:28 yeah, it is ;-) Mar 18 04:24:29 ahh, ok Mar 18 04:26:52 in that configure script, there is a function called oe_runconf, what do the --host= and --target= say? Mar 18 04:27:16 :q Mar 18 04:27:54 really got to watch the window focus Mar 18 04:28:30 --host=arm-angstrom-linux-gnueabi --target=arm-angstrom-linux-gnueabi Mar 18 04:29:38 and --build? Mar 18 04:31:17 --build=i686-linux Mar 18 04:32:03 found the bug report: http://bugs.openembedded.org/show_bug.cgi?id=1906 Mar 18 04:33:18 * JustinP applies cyril's patch Mar 18 04:34:42 funny, wonder why I didn't find this bug, must be the particular host system. Mar 18 04:36:10 I'm on gentoo as well... Mar 18 04:36:48 i'm using ubuntu Mar 18 04:37:17 well, one of the commenters on the bug say's they're using gentoo Mar 18 04:38:18 I know.. just wondering (aloud) if that's the difference that causes the problem. Mar 18 04:39:12 well....it's unlikely to be gentoo's "fault". It's more likely autoconf or freetype's configure script's fault and gentoo happens to make the bug show itself ;-) Mar 18 04:40:11 yes of course, it's clearly a problem in the configure.ac Mar 18 04:40:59 didn't mean to badmouth gentoo Mar 18 04:41:15 ;-) Mar 18 04:41:26 ughugh....patch fails now Mar 18 04:43:32 if you can't get tha patch to work, it's a simple edit job and I think the configure script will run autoconf to make the new configure script Mar 18 04:56:45 looks like the configure.patch isn't needed any more Mar 18 04:57:30 hmm, noticed that extra patch Mar 18 05:00:41 it's still happening... Mar 18 05:01:04 did you clean it? Mar 18 05:07:47 maybe the patch didn't work right...I'm still playing with it Mar 18 05:09:10 configure worked...now it can't find libtool Mar 18 05:11:26 * SirManiac is trying to figure out the libtool detection in the configure script... Mar 18 05:13:27 ok, I got it to install....bit I had to symlink arm-angstrom-linux-gnueabo-libtool to libtool Mar 18 05:14:11 it's good that you figured it out. Mar 18 05:14:24 I did it manually, though Mar 18 05:14:29 the bb isn't fixed Mar 18 05:15:04 I noticed that.. Mar 18 05:15:32 hopefully they will add that soon... Mar 18 05:16:01 I can check in what I have now but I'm not sure it's right at all Mar 18 05:16:49 I too new at this to know who to talk to. Mar 18 05:17:08 maybe the mailing list? Mar 18 05:17:15 heh Mar 18 05:19:00 yeah, I don't think the patch fully applied Mar 18 05:19:19 was it for a different version? Mar 18 05:21:19 I'm still playing with it ;-) Mar 18 05:38:40 03justinp 07org.oe.dev * reff88718... 10/ (4 files in 3 dirs): Mar 18 05:38:40 freetype: add 2.3.1 from bug #1906 Mar 18 05:38:40 - Fixes problems with building that 2.3.0 had Mar 18 05:38:44 03justinp 07org.oe.dev * reefec481... 10/ (1 conf/distro/angstrom-2007.1.conf): angstrom-2007.1: update freetype to 2.3.1 Mar 18 05:39:11 cool, guess you got it working. Mar 18 05:42:35 having problems with glibc with MACHINE=native, i'm guessing not very many people use that build Mar 18 05:43:15 nslu2 people use it I think Mar 18 05:44:07 hmmm Mar 18 05:45:25 I've never tried it myself, no use for it Mar 18 05:46:34 I'm trying to set up a native environment of GPE so I can port an application, and use Xnest to test's it's GUI. Mar 18 05:46:56 figured it would be easier then the MACHINE=x86 Mar 18 05:47:08 don't know Mar 18 05:47:12 Hi, I sill have the problem that I cannot build gcc,( CROSS COMPILE Badness: /usr/include in INCLUDEPATH: /usr/include/libffi) I have even tried to rename /usr/include :P , would it be a good idea to switch to a non-dev branch (known wokring config)? Mar 18 06:48:40 03justinp 07org.oe.dev * rdd877a68... 10/ (1 packages/musicpd/gmpc_svn.bb): gmpc: add missing DEPENDS, curl Mar 18 07:49:22 good morning all Mar 18 07:50:45 cyrilRomain: morning. I applied part of your freetype patch Mar 18 07:51:00 JustinP: I saw that, thanks :) Mar 18 07:52:47 JustinP: btw, what is the policy for package removal ? Mar 18 07:54:39 cyrilRomain: well.....I think it should be added to REMOVAL and then removed after a certain period of time. People outside of OE core may use it (i.e. corporate projects) Mar 18 07:55:52 JustinP: ok. So next times I'll just submit fix patch, and suggest a removal in comment, if any Mar 18 07:57:17 :-) Mar 18 08:00:03 hmm today is going to be a bug day for me: the following list of package does not build: firefox, erlang, apt, irssi, libqpe-opie, python-fuse ! Mar 18 08:02:00 wow, fun Mar 18 08:10:56 * koen|lisboa stabs TAP air Mar 18 08:28:35 03koen 07org.oe.dev * r65373f0b... 10/ (5 files in 4 dirs): libgsmd: merge with openmoko svn, but put the device-specific initscript in a device specific directory Mar 18 08:28:39 03koen 07org.oe.dev * rdc7b99f9... 10/ (3 files in 3 dirs): libgsmd: contrary to popular belief at FIC, you can't actually make calls with a GPS, you need a GSM Mar 18 08:43:15 does the dropping shell still works for you guys ? since some weeks when I get dropped into the shell I don't have the prompt anymore and typing command to nothing ? Mar 18 08:53:04 morning all Mar 18 08:53:09 hi Ifaistos Mar 18 08:53:37 I think perl in OE has a bug Mar 18 08:53:39 ldconfig: /usr/lib/libperl.so.5.8.7 is not an ELF file - it has the wrong magic bytes at the start. Mar 18 08:53:59 this is on the target, while trying to install file Mar 18 08:54:27 and also trying to run some perl stuff fails Mar 18 08:55:00 ->http://rafb.net/p/c4deXA40.html Mar 18 09:12:49 interesting -> root@efika:/usr/lib$ file /bin/busybox Mar 18 09:12:49 /bin/busybox: Berkeley DB (Hash, version 131092, native byte-order), for GNU/Linux 2.4.0, dynamically linked (uses shared libs), stripped Mar 18 09:13:41 all my rootfs binary files are Berkelet DB :) Mar 18 09:14:53 8-/ Mar 18 09:16:41 cyrilRomain : probably stripping is stripping too many things... Mar 18 09:17:47 scary Mar 18 09:18:50 yep that' it Mar 18 09:19:07 i recompiled file with debug and no stripping Mar 18 09:19:13 and here are the results Mar 18 09:19:15 root@efika:/usr/lib$ file /bin/busybox Mar 18 09:19:15 /bin/busybox: ELF 32-bit MSB executable, PowerPC or cisco 4500, version 1 (SYSV), for GNU/Linux 2.4.0, dynamically linked (uses shared libs), stripped Mar 18 09:21:11 will try to rebuild perl the same way Mar 18 09:21:29 still scary, you set 'no stripping' and it is 'stripped' ? Mar 18 09:22:08 * cyrilRomain wonders if it is possible that strip command is called twice Mar 18 09:23:38 no if i build file with no stripping it works as it should, if stripping is enabled then everything is "Berkeley DB" Mar 18 09:23:42 root@efika:/usr/lib$ file /usr/bin/file Mar 18 09:23:42 /usr/bin/file: ELF 32-bit MSB executable, PowerPC or cisco 4500, version 1 (SYSV), for GNU/Linux 2.4.0, dynamically linked (uses shared libs), not stripped Mar 18 09:25:06 Ifaistos: i still wonder why it is now right and 'not stripped', but it was right and 'stripped' two line above Mar 18 09:26:14 file reports wrong format if its stripped Mar 18 09:26:23 10:19 < Ifaistos> /bin/busybox: ELF 32-bit MSB executable, PowerPC or cisco 4500, version 1 (SYSV), for GNU/Linux 2.4.0, dynamically linked (uses shared libs), stripped Mar 18 09:26:48 if its build (file) without stripping it reports the correct format of the files Mar 18 09:27:24 which brings up the question... what other programs are not working as they should when the get stripped ? Mar 18 09:27:51 that's the really scary part Mar 18 09:28:57 ah ok, I misunderstood. good question .. :[ Mar 18 09:31:02 really strange behaviour anyway Mar 18 09:32:35 i have seen it before but with 3.4.4 Mar 18 09:33:07 asterisk for examples was sigfaulting when stripped Mar 18 11:04:55 03nail 07org.oe.dev * r1ba0cb9a... 10/ (1 packages/dtc/dtc_20070216.bb packages/dtc/dtc_git.bb): dtc: Add cross-compiled version, thanks to timtimred Mar 18 11:22:43 hi all Mar 18 11:51:15 mickeyl: good morning! Mar 18 11:51:32 morning Mar 18 11:59:36 hey koen|lisboa, mickeyl Mar 18 11:59:43 hey cyrilRomain Mar 18 11:59:58 mickeyl: it seems that python-2.4.4-manifest.inc need to be regenerated by Mar 18 11:59:58 generate-manifest.py. I've had QA issues http://rafb.net/p/erArqx12.html that disappeared with a renegerated Mar 18 12:00:01 manifest (python-misc is indeed not here anymore) Mar 18 12:01:16 some guys unfortunately started to ignore my header warning (regenerated, don't do manual edits) Mar 18 12:01:31 which means we have to take a close look on next update Mar 18 12:01:39 to make the changes appear via generate-manifest.py as well Mar 18 12:02:03 putting "*" into python-misc is broken by design anyway Mar 18 12:09:07 03koen 07org.oe.dev * r4657aa1d... 10/ (3 files in 3 dirs): libgsmd: unbreak initscript some more, thanks to rwhitby for spotting that typo Mar 18 12:11:54 koen|lisboa: seppuku works for the OE bugtracker, should we setup a angstrom autobuild on ewi again? Mar 18 12:12:04 koen|lisboa: or poke greentux and ask for hardware Mar 18 12:12:24 zecke|hangover: I can look at setting up an autobuilder next week Mar 18 12:20:13 zecke|hangover: did you see how OLPC wants to implement legacy support for pulseaudio? Mar 18 12:21:48 no Mar 18 12:22:43 zecke|hangover: a devnode presented via FUSE that talks to pulseaudio Mar 18 12:23:05 so apps see a devnode and won't need any config to use pulse Mar 18 12:23:10 pretty neat Mar 18 12:23:25 koen|lisboa: But linux 2.4 has no fuse Mar 18 12:23:42 zecke|hangover: actually.... Mar 18 12:23:51 but neat, I wonder about the latency of this redirection Mar 18 12:23:52 zecke|hangover: FUSE has a 2.4 backport.... Mar 18 12:24:04 * zecke|hangover hands a cookie to koen Mar 18 12:24:16 thanks Mar 18 12:24:32 only 2 hours left..... Mar 18 12:24:46 I hate 'C' Mar 18 12:25:29 what is the best supported target/distro? Mar 18 12:25:34 zecke|hangover: zack said the same a few days ago Mar 18 12:26:23 which triggered a QT vs STL discussion Mar 18 12:26:41 koen|lisboa: I'm toying with path handling in C Mar 18 12:27:12 koen|lisboa: and your string might contain ./../../foo/bla/../../../ and I need to walk that and must make sure my target string is big enough... Mar 18 12:36:47 koen|lisboa : Hi ! Is gcc (not cross) packaging still broken ? Mar 18 12:37:14 Ifaistos: it shouldn't be Mar 18 12:40:23 koen|lisboa : hmmm i am trying to build perl on efika itself but when testing for the compiler it reports that as is missing Mar 18 12:40:54 koen|lisboa : powerpc-linux-gcc: error trying to exec 'as': execvp: No such file or directory Mar 18 12:41:18 Ifaistos: 'as' is in binutils, right? Mar 18 12:41:28 (or rather 'binutils-symlinks') Mar 18 12:41:33 koen|lisboa : yes its installed Mar 18 12:42:03 Ifaistos: btw, I don't know if it is related, but there is a typo in perl-native: http://rafb.net/p/nHeNmj14.html Mar 18 12:47:18 cyrilRomain : hmmm could this be the reason that the oe build perl libs have problem ? Mar 18 12:47:32 Ifaistos: probably Mar 18 12:47:40 cyrilRomain : not linked correclty Mar 18 12:47:43 not sure though Mar 18 12:48:33 * Ifaistos wonders how many hours it will take to build perl on efika... Mar 18 13:18:23 apaulsen, psokolovsky_, slapin, XorA|gone: I am still looking for a quick test case to determine what part of my glibc changes breaks things. quick meaning not involving complete recompilation of glibc every time. Any idea? Mar 18 13:28:08 Things I tried but which failed: "bitbake -g", "bitbake -vvv --dry-run|grep -I path" and others I cannot remember now but which did not show the version being chosen. Mar 18 13:44:50 Laibsch, build gtk+ Mar 18 13:45:28 hmmmm Mar 18 13:46:01 * koen|lisboa wonders how to start boarding without the airplane being present Mar 18 13:46:55 aha Mar 18 13:47:22 the TAP website is more up to date than the display in front of me Mar 18 13:49:25 zecke: vchroot seems like a nice idea Mar 18 13:55:10 psokolovsky_: Building gtk+ apparently first build version 2.5 of glibc and only builds runs into trouble later on. Mar 18 13:55:32 Laibsch, exactly Mar 18 13:55:33 I was not yet able to verify 100% since it always takes so long and because of weekend commitments. Mar 18 13:56:05 psokolovsky_: Well, then every test run takes me very long. I was looking to short-cut that. Mar 18 13:56:14 Maybe there is no way. Mar 18 13:57:19 Laibsch, for me, after I habe glibc-2.5 built, then glibc built, running "bitbake gtk+" caused glibc-2.3.5 to be started to built Mar 18 13:57:34 so, It kinda doesn't require rebuild to reproduce Mar 18 13:58:01 (but I have no idea if youre changes trying to fix that would require rebuild) Mar 18 13:58:07 I am not following you here. "after I habe glibc-2.5 built, then glibc built," ?? Mar 18 13:58:45 psokolovsky_: No, the changes don't require a rebuild per se. But adding line by line of the changes and looking for which one breaks things does. Mar 18 13:59:23 remember, this was all meant to be 1 trivial packaging change and 2 semantically equivalent unifications Mar 18 13:59:30 Laibsch, to which files to you add lines? Mar 18 13:59:56 koen|away: yes, but C sucks (path handling) Mar 18 14:00:02 The three commits which I reverted all applied to packages/glibc/ Mar 18 14:00:29 Laibsch, I know ;-) that's why I say that we all should be wary touching such core and tangled stuff like glibc ;-). Don't take that as dicouragement to have this fixed though! Mar 18 14:00:39 koen|away: with python/Qt you would just do str.split('/') and then apply each '..' as cd.up Mar 18 14:00:57 koen|away: strtok and making sure the string is big enough makes not too much fun Mar 18 14:01:06 Laibsch, as I told, please apply -dbg change, make srue it's ok (it must be!), commit it, and you have only 2 changes Mar 18 14:01:12 Ok, I need some help here. I'm trying to build thttpd with php-support. This requires compiling both in one .bb. Is there any examples i could look at there? Mar 18 14:01:42 (php patches thttpd, compiles, then thttpd should be compiled) Mar 18 14:01:46 psokolovsky_: Things break. I now believe that your call for an RFC for what I did would have been over-the-top. This was not meant to change anything. Mar 18 14:02:13 psokolovsky_: I can only commit after -dbg has been tested. Which I have a hard time to do. Mar 18 14:02:16 NAiL: not that I know of. Obviously you need a different compile,configure method and add both to SRC_URI Mar 18 14:02:58 zecke: yeah, timtimred got as far as fetching both, and compiling php correctly. but how can I add a different OECONF to both? Mar 18 14:03:20 NAiL: define do_compile_httpd() {} Mar 18 14:03:22 Laibsch, it could be anything. including bug in bitbake. again, idea is not "can it break" or not, idea that core compenents requier special attention. as for "things break" - agree 100%. I break thinsg myself too Mar 18 14:03:33 NAiL: and do_compile_php() {} Mar 18 14:03:46 zecke: will it work to define EXTRA_OECONF inside those? Mar 18 14:04:02 NAiL: python do_compile() { bb.build.exec_func('do_compile_php')\n bb.build.exec_func(other) } Mar 18 14:04:12 NAiL: and then use the OVERRIDES Mar 18 14:04:33 NAiL: EXTRA_OECONF_do_compile-zhttpd = "FOO" Mar 18 14:04:43 NAiL: EXTRA_OECONFIG_do-compile-php = "BLA" Mar 18 14:05:05 zecke: ok, I think I got most of that. How does OVERRIDES work? Mar 18 14:06:11 NAiL: when BitBake executes a task, the taskname gets prepended to the OVERRIDES Mar 18 14:06:15 psokolovsky_: special attention had been paid. I had no problems rebuilding glibc and after that I felt confident to commit. I still believe these changes were not RFC-material. That they apparently did break is another story. This is "things break". Mar 18 14:06:38 NAiL: so it is like EXTRA_OECONF_arm, but for a task (do_compile_thttpd) Mar 18 14:06:55 Laibsch, ok, that's not important. there's no scape goat. what' Mar 18 14:07:01 NAiL: BTW: don't use php :) Mar 18 14:07:34 Laibsch, what's important is how to make your changes work Mar 18 14:07:56 zecke: why not? Mar 18 14:08:52 NAiL: hehe, don't look here for non-emotional arguments :) Mar 18 14:08:57 aha, yeah Mar 18 14:09:33 NAiL: it is inherently insecure but your 'customer' might be Danger Mouse Mar 18 14:10:03 I know. Mar 18 14:12:09 zecke: btw, won't "addtask" work, instead of OVERRIDES? Or am I not getting this? :) Mar 18 14:13:33 NAiL: you can use addtask, I would just use do_compile(), do_configure() as dispatchers, but adding tasks will work as well Mar 18 14:13:50 NAiL: and this will be the same, you won't touch OVERRIDES directly, they will just work for you Mar 18 14:14:18 aha Mar 18 14:14:49 NAiL: sorry to confuse you Mar 18 14:15:42 np, bb is very flexible Mar 18 14:16:26 and as such has several ways to solve stuff ;) Mar 18 14:17:44 oh.. Is there a simple way to define two ${S} variables, one for do_[configure|compile]_php and one for do_[*]_thttpd? Mar 18 14:18:45 S_compile_php = ... S_configure_php Mar 18 14:19:07 lol userfriendly.org Mar 18 14:19:09 hi zecke Mar 18 14:20:17 zecke: uh, like so: S_compile_php = ${WORKDIR}/thttpd-2.25b S_configure_php Mar 18 14:20:46 * NAiL feels like a complete newbie again Mar 18 14:21:52 NAiL: http://svn.berlios.de/viewcvs/bitbake/branches/bitbake-1.8/lib/bb/build.py?rev=784&view=markup and check OVERRIDES Mar 18 14:22:08 thanks Mar 18 14:22:17 aha, zecke, nice to see you Mar 18 14:22:22 OVERRIDES is used to make stuff like FOO_arm FOO_zaurus work Mar 18 14:22:46 Henryk: hehe, I think I have input for your last mail besides that I (mostly) agree with richard Mar 18 14:23:31 zecke: yeah, i was at implementing and running against some bitbake limitations Mar 18 14:25:12 uh, i just see: everything has changed in bitbake-1.8 Mar 18 14:25:48 Henryk: richard said that in the initial response to harald :) Mar 18 14:26:09 Henryk: I like the git log | wc -l idea, it is quite clever Mar 18 14:26:26 i hope so, took me some time Mar 18 14:30:06 mickeyl: what do you think about this: http://en.pastebin.ca/400131 Mar 18 14:36:36 zecke: hehe, nice. some of the changes in bitbake 1.8 are very similar to what I had already done locally Mar 18 14:50:13 I guess that bb.data.getVar('FILE', d, 1) is the bitbake recipe currently executed, yes? Mar 18 14:52:50 Henryk: yes, while parsing it might point to the .inc, .bbclass that is currently parsed Mar 18 14:53:38 ok Mar 18 14:55:17 v8jlene or NAiL or anyone else, bugs 2011 and 1958 are awaiting checkin Mar 18 15:11:48 pH5: looks good to me Mar 18 15:13:44 hillct: ok, there's still some stuff to be done on 2011, but it's minor ;) Mar 18 15:13:58 oh? Mar 18 15:14:04 what'd I miss? Mar 18 15:14:28 I'd thought I'd cleaned it up far more than was asked Mar 18 15:15:11 the only thing I didn't do was move the INIT_SCRIPTS variables to the bottom because it makes no sense to stick them in the most inconvenient place possible Mar 18 15:15:25 they should fe quite near the top Mar 18 15:16:00 in my opinion, that falls into the category of something that needs to be changed in oe-stylize Mar 18 15:16:39 but what are the areas that are of concern to to you? Mar 18 15:16:42 hillct: It works in its current incarnation. It's just that naming the patches with ${PV} is kinda awkward. Better to remove the ${PV} and put the patch in a version-subdirectory instead ;) Mar 18 15:16:55 I've changed that in the one I'm committing to 3.10 Mar 18 15:17:26 ah Mar 18 15:17:30 I see Mar 18 15:17:32 makes sense Mar 18 15:18:14 I saw that done in other recipes but I figured that version directories would only be needed if there were many version-specific patches Mar 18 15:18:29 * hillct makes a note for next time Mar 18 15:19:07 It's easier when adding a new version to just do "cp -R version-dir new-version-dir", instead of renaming individual patches Mar 18 15:19:13 mickeyl: Ok. If you don't mind, I'll commit that and bump the PR. Mar 18 15:19:17 I have more problems with pygtk2. It seems like the do_install task is missing half of the files... Mar 18 15:19:24 NAiL: I suppose it is, yes Mar 18 15:19:53 it's not a "major" problem, it's just a bit of future-proofing ;) Mar 18 15:20:05 zecke: did you already try to build python-openmoko with OE / using python-pygtk2? Mar 18 15:20:27 pH5: no, I have only done it on my workstation with preinstalled pygtk2 Mar 18 15:20:46 pH5: is configure complaining about missing files? Mar 18 15:21:11 NAiL: just so I understand the process, are packages customized for each distro or will your changes to the recipe be available to all distros? Mar 18 15:21:26 zecke: no, but compile is complaining about a missing pygtk.h (which is never installed by OE's pygtk2) Mar 18 15:21:46 that is bad, pyglib.h is installed? Mar 18 15:21:58 I've just finished writing a .bb file for gtk+-directfb. It is very simular to the gtk+-x11 so I planned on using gtk-2.10.inc but I had to change the following lines in gtk-2.10.inc: Mar 18 15:22:08 oe_libinstall -so -C gtk libgtk-directfb-2.0 ${STAGING_LIBDIR} Mar 18 15:22:08 oe_libinstall -so -C gdk libgdk-directfb-2.0 ${STAGING_LIBDIR} Mar 18 15:22:08 Mar 18 15:23:11 in the do_stage.. they were originall oe_libinstall -so -C gtk libgtk-x11-2.0, is there some way to set up a condition or do I just need to make a gtk-directfb-2.10.inc? Mar 18 15:23:17 hillct: packages will be available to build for all distros. I'll be committing to oe head shortly Mar 18 15:23:45 NAiL: note that OpenPBX depends on SpanDSP (bug 1958) and recommends the logrotate package so they'd also need to be present in any distro where OpenPBX is offered Mar 18 15:24:59 hillct: if openpbx is added to a distro, it'll pull spandsp as well Mar 18 15:25:12 cool Mar 18 15:25:24 but my point is spendsp hasn't yet been checked in Mar 18 15:25:37 er spandsp Mar 18 15:25:59 zecke: is my assumption about the version comparison correct, e.g. that "uboot-gta01-1.2.0+svn10" is more than "uboot-gta01-1.2.0+svn2"? (aka 'natural' sort ordering, and not lexical sort ordering) Mar 18 15:25:59 svolpe_gerrath: introduce a variable :) Mar 18 15:27:13 and set the variable in the package calling the gtk-2.10.inc.. I guess I could also parse the package file name.. Mar 18 15:27:24 Henryk: good question, code is law and we should have a vercmp in lib/bb/__init__.py Mar 18 15:27:54 hillct: I'm checking in both at the same time. Mar 18 15:27:57 zecke: that's why i was asking. last time i checked there were about three different version comparison functions with different results Mar 18 15:28:19 zecke: no, pygtk2 doesn't install any headers in do_install. this is the usual autotools magic, I don't understand it at all. Mar 18 15:28:58 !oebug 1958 Mar 18 15:28:59 * * Bug 1958, Status: RESOLVED, Created: 2007-03-09 10:26 Mar 18 15:29:00 * * hillct(AT)pinnacledigital.com: SpanDSP library for fax and voice VOIP services Mar 18 15:29:01 * * http://bugs.openembedded.org/show_bug.cgi?id=1958 Mar 18 15:29:04 !oebug 2011 Mar 18 15:29:05 * * Bug 2011, Status: RESOLVED, Created: 2007-03-16 23:00 Mar 18 15:29:06 * * hillct(AT)pinnacledigital.com: New Package: OpenPBX.org - VOIP Soft Switch/PBX Mar 18 15:29:07 * * http://bugs.openembedded.org/show_bug.cgi?id=2011 Mar 18 15:29:51 hmm I'm about to run :} Mar 18 15:30:38 NAiL: thanks Mar 18 15:31:04 03nail 07org.oe.dev * r72025f26... 10/ (11 files in 6 dirs): openpbx/spandsp: Add packages openpbx and spandsp, thanks to hillctl Mar 18 15:31:54 zecke, the only down side is gtk+ for X11 is just named gtk+_2.10.9.bb and not gtk+-X11_2.10.9.bb so I guess I can't just parse the file name and will have to add the variable to all the packages calling gtk-2.10.inc or put a condition to default to X11. Mar 18 15:31:56 zecke: and one important question: what would be the right[tm] way for me to maintain the local version numbers. In which directory should the local state be kept and maybe there's already an infrastructure in place for that? Mar 18 15:32:26 Henryk: hmm okay utils.py and __init__.py both offer vercmp and providers.py is using utils.py Mar 18 15:33:11 Henryk: I think we should add a new variable to bitbake.conf and assign ${TMPDIR}/your-favorite-name to it Mar 18 15:33:48 ok Mar 18 15:34:17 Henryk: make sure to change the right/all bitbake.conf Mar 18 15:34:54 pH5: this autotools magic is just calling make install with the staging directory as destination Mar 18 15:35:45 ah cool, I like the utils.py version comparison much better than the __init__.py one Mar 18 15:36:37 Henryk: grep through classes in OE to check if vercmp is used and then kill the __init__.py copy after transfering the test case (wow is it obvious that I try to offload work?) Mar 18 15:36:44 zecke: but I can't read/understand gtk/Makefile Mar 18 15:36:52 zecke: Makefile.am looks fine to me. Mar 18 15:36:59 *g* Mar 18 15:38:19 pH5: hmm, I would whisper strace and check if pygtk.h comes along Mar 18 15:38:46 pH5: or find pygtk.h in the Makefile.am and figure out which target is using it Mar 18 15:40:25 pH5: http://page.mi.fu-berlin.de/~freyther/python-openmoko-0.3.tar.bz2 just a dist of my current version, it covers all classes Mar 18 15:40:30 later :) Mar 18 15:40:30 zecke: hmm, I don't like ${TMPDIR}. I believe local version numbers should survive removing that directory Mar 18 15:41:06 zecke: yay, now if I only could get the prerequisites to work :) Mar 18 15:41:25 Henryk: DL_DIR is inside the TMPDIR by default as well Mar 18 15:41:59 Henryk: alternatively create a dir inside the DL_DIR Mar 18 15:43:07 zecke: oh, it is? interesting. I only know the openmoko setup where DL_DIR is one level and beneath TMPDIR. Yes, DL_DIR would be ok Mar 18 15:43:36 ok, thanks Mar 18 15:43:47 Henryk: is this the reason to use a variable, it allows people to reassign it :) Mar 18 15:43:55 yeah Mar 18 15:44:04 so either join our geek cooking or cya tomorrow Mar 18 15:45:03 fine, see you tomorrow. will code now Mar 18 15:54:11 ooh. I tried to build pygtk2 2.10, but openmoko still uses gtk 2.6. instead of an error configure just decided not to include the gtk subdir in the build. Mar 18 15:54:46 s/an error/throwing an error/ Mar 18 15:56:22 nice Mar 18 16:18:45 psokolovsky_: It seems that moving r equire glibc-package.bbclass to .inc was creating problems: http://oz.leggewie.org/wip/286930242424d10b2503eb290dbc31c7975bab74.patch The other two patches which I had applied first do not trigger any problems with the build of gtk+. Would you be so kind to verify? Mar 18 16:19:15 You find the other two patches in the same directory on my server. Mar 18 17:00:51 apaulsen, slapin: Would be nice if the two of you (and anybody else for that matter) applied http://tinyurl.com/3awwdu and http://tinyurl.com/2qufql to try and verify that they don't trigger the problem you saw earlier. Mar 18 17:32:02 NAiL: it appears I've included a bunch of dev files in the main openpbx package Mar 18 17:32:06 correcting now Mar 18 17:32:16 will have an update shortly Mar 18 18:08:27 Anybody have an idea what kind fo machine the nas100d is? Mar 18 18:10:22 Laibsch: http://www.nslu2-linux.org/wiki/NAS100d/HomePage Mar 18 18:11:09 Runs SlugOS very nicely :) Mar 18 18:13:42 mwester: Can you take a look at bug 822? The DEPENDS for openssl has been fulfilled. Mar 18 18:14:07 Sorry, wrong number Mar 18 18:14:11 !oebug 823 Mar 18 18:14:12 * * Bug 823, Status: NEW, Created: 2006-04-08 09:11 Mar 18 18:14:13 * * evanchsa(AT)gmail.com: hostap-daemon package updates Mar 18 18:14:14 * * http://bugs.openembedded.org/show_bug.cgi?id=823 Mar 18 18:14:32 The proposed DEPENDS on openssl has already been done. Mar 18 18:14:56 Ok, so hostap-daemon should now build and work correctly? Mar 18 18:14:57 The only thing left before closing this one is DEPENDS_nas100d_append = "madwifi-modules" Mar 18 18:15:24 mwester: I assume so. Take a look at the current version of the bb file in the OE tree. Mar 18 18:15:34 !oebug 1998 Mar 18 18:15:35 * * Bug 1998, Status: NEW, Created: 2007-03-16 00:10 Mar 18 18:15:36 * * jbs(AT)bandos.homelinux.com: lirc-modules fails to build Mar 18 18:15:37 * * http://bugs.openembedded.org/show_bug.cgi?id=1998 Mar 18 18:15:58 I wonder if nas100d is included either in mtx-1 or mtx-2. mwester, do you happen to know? Mar 18 18:15:59 Ok. I'll sync up and see if I can test (I have the D-Link DSM-G600, almost identical to the NAS100d). Mar 18 18:16:07 mwester: Nice. Mar 18 18:16:11 * mwester wonders what mtx-1 might be? Mar 18 18:16:16 me too Mar 18 18:16:33 Could be an independent device or a device class. Mar 18 18:16:48 `lart openwrt for being openwrt Mar 18 18:17:03 rwhitby might know about the mtx stuff? Mar 18 18:17:06 chouimat|nap: Well said Mar 18 18:17:23 mwester: He is not here ATM, it seems. Mar 18 18:17:48 So it seems, but he reads the log. Mar 18 18:17:53 Laibsch: just went to their channel with a question ... and now I just want to get postal in it ;) Mar 18 18:18:03 postal? Mar 18 18:18:14 Laibsch: shoot them all Mar 18 18:18:18 asking questions in #openwrt is almost NEVER a good idea ;-) Mar 18 18:18:28 and hang'em high ;-) Mar 18 18:19:23 mwester: mtx-1 is the meshcube Mar 18 18:19:46 Laibsch: the problem is they are giving a bad name to linux ... but they doesn't seem to care ... they change something and they don't document it anywhere but they kind of expect you to read their private emails :) Mar 18 18:19:55 Ah - I don't know about the meshcube; I've not heard of the nas100d in respect to that. Mar 18 18:20:20 is there any way from inside a .inc file to determine the name of the file (.bb) that called it? Example a.bb calls b.inc is there a way to determine a from inside of b? or do I just need to set up a global variable inside of a.bb before it includes b.inc? Mar 18 18:20:45 svolpe_gerrath: ${PN} ? Mar 18 18:21:19 * Laibsch points mwester to conf/machines/ Mar 18 18:21:31 Laibsch, tried it but it returns b (name of inc file) Mar 18 18:21:33 good information but no nas100d.conf Mar 18 18:21:40 * chouimat|nap need to find time for his wrtoe stuff :) Mar 18 18:21:50 chouimat|nap: Please do Mar 18 18:22:03 chouimat|nap: And let me know when you have something to test|show Mar 18 18:22:11 * Laibsch cannot wait to dump openwrt Mar 18 18:22:42 svolpe_gerrath: Then I don't know. Mar 18 18:22:58 Laibsch: ok ... I think I will start by installing qemu and some other emulator so I can play with it while my wrt is getting a small hardware upgrade Mar 18 18:23:48 mwester: What machine are you using when compiling for the nas100d? Mar 18 18:24:06 * Bernardo uses dd-wrt - was going to try openwrt, but not anymore Mar 18 18:24:12 Host? x86 32-bit running Fedora Core 6 Mar 18 18:24:31 chouimat|nap: a sd card? :) Mar 18 18:24:50 Bernardo: serial port and jtag Mar 18 18:25:04 That might be my next mods Mar 18 18:25:07 Bernardo: I have the wrtsl54gs Mar 18 18:25:14 I have the gl Mar 18 18:25:21 wrt54gl Mar 18 18:25:24 mwester: Do you know anything about foonas? Is that the same as angstrom but for all NAS-type devices? Mar 18 18:25:45 mwester: machine I meant MACHINE in OE Mar 18 18:26:07 Bernardo: I want to be able to do rm -r openwrt-checkouts Mar 18 18:26:19 I also have a smc2804wbr v.2 that I'd like to remove vxworks from and install something like dd-wrt micro, but I can't even find the jtag pinout for it Mar 18 18:26:33 chouimat|nap: Emulartor is a good idea since I do not intend to HW mod my router. But that way we can collaborate right from the start Mar 18 18:26:36 NAiL is the guy who knows about foonas; I think it is Angstrom for a set of NAS devices but very early in development. Mar 18 18:27:05 more or less Mar 18 18:27:14 In OE, the MACHINE is usually ixp4xx Mar 18 18:27:19 google does not know either about nas100d and openembedded. Mar 18 18:27:20 Laibsch: maybe ... ok I really have to go take a nap ... if I don't get there soon she will kill me ... later ;) Mar 18 18:27:28 lol Mar 18 18:27:29 bye Mar 18 18:27:46 bye Mar 18 18:27:50 (S) Mar 18 18:28:56 mwester: btw, foonas is more or less working on the qnap ts-[12]01, linkstation hg/hd, thecus n2100 ;) Mar 18 18:29:01 NAiL: Can you confirm that the nas100d needs madwifi and that DEPENDS_nas100d_append = "madwifi-modules" would be the way to do this? Mar 18 18:29:07 * Laibsch points to bug 823 Mar 18 18:29:20 Laibsch, I'm thinking there probably is not a nice way to do it. which is a shame since a lot of times .inc files are used for many different packages. I guess I will use a variable set in my main package.. Mar 18 18:29:27 Laibsch: DEPENDS_nas100d += "madwifi-modules" ? Mar 18 18:29:44 NAiL: I wonder if the _nas100_append will do anything since there is no machine configuration for it. Mar 18 18:29:58 ah, you're right Mar 18 18:29:58 NAiL: Your line should be equivalent Mar 18 18:30:06 Laibsch, I'll scourer through other OE packages first just to make sure I'm not missing anything. Mar 18 18:30:08 yeah Mar 18 18:30:37 svolpe_gerrath: What are you trying to do exactly? Mar 18 18:30:38 Well, making a machine.conf for nas100d shouldn't be a probem Mar 18 18:30:58 NAiL: Well, I am just wondering if 823 can be closed Mar 18 18:31:10 If we don't have the nas100d to begin with, then yes Mar 18 18:31:52 hmm Mar 18 18:32:33 Laibsch, I'm adding a package gtk+-directfb and want it to use gtk-2.10.inc but gtk-2.10.inc copies libgtk-x11 and libgdk-x11, gtk+-directfb will need it to copy libgtk-directfb and libgdk-directfb instead. Those are the only changes. I wanted to have gtk-2.10.inc to make the choice based on the package that included it but I think I might have to do it off a variable set in the package. Mar 18 18:32:42 how are nas100d images built anyway? Are they just "spinoffs" of the slugos image? Mar 18 18:32:50 Yes Mar 18 18:33:24 NSLU2, nas100d (aka Lude), dsmg600, loft, etc are all supported from the same slugos image Mar 18 18:33:25 ok, I guess a machine conf called nas100d.conf, which includes ixp4xx(le?).conf should do the trick? Mar 18 18:33:35 hmm Mar 18 18:34:00 Does it hurt anything to make it dependent on the image in general (which is ixp4xx AFAICT) Mar 18 18:34:02 ? Mar 18 18:34:20 Laibsch, I wonder if I can parse the WORKINGDIR since it would be set by the main .bb file and not the .inc? Mar 18 18:35:50 of course, adding a nas100d.conf which works with foonas shouldn't be more than half an hour work ;) Mar 18 18:35:58 (including an image-making bit) Mar 18 18:42:31 Laibsch, same issue, I looked in bitbake.conf in all the information variables are just pthon calls that get the information for the current file.. Mar 18 18:42:45 * mwester , after searching other bb files, thinks the dependency should be on both "nslu2be" and "nslu2le", but perhaps "nslu2" will refer to both. But will defer to rwhitby for enlightenment. Mar 18 18:45:54 I'm having difficulty specifying additional files for my -dev package. Wat's the valid equivelent to this? Mar 18 18:45:56 FILES_${PN}-dev += "${libdir}/openpbx.org/modules/*.la \ Mar 18 18:45:56 ${libdir}/openpbx.org/*.la" Mar 18 18:46:47 or if I need to add things to the -dev pacakge do I need to completely override the automatic file selection process? Mar 18 18:47:31 I could do something like just create a -libs package but that seems wrong to me Mar 18 18:56:33 * hillct wonders if you have to explicitly add ${PN}-dev to the PAXKAGES list, for FILES_${PN}-dev values to be processed, despide -dev already being in the PACKAGES array by default Mar 18 19:08:36 hillct: you don't have to. Mar 18 19:16:18 pH5: yet it doewsn't work if I don't, so I guess I'm doing something else wrong... Mar 18 19:22:13 hillct: perhaps you overwrote PACKAGES or there is some ordering issue? Mar 18 19:24:01 I'm appending to PACKAGES Mar 18 19:24:42 to add other sub-packages, which worked fine until I discovered that the contents of -dev were not being properly identified and I needed to add a few files to it Mar 18 19:24:53 hillct: PACKAGES =+ or += ? Mar 18 19:25:30 += Mar 18 19:25:34 so prepending i guess Mar 18 19:25:38 there is a note in documentation about that Mar 18 19:25:43 is that the issue? Mar 18 19:25:48 will check Mar 18 19:26:01 hillct: I think in some case, especially PACKAGES, =+ should be used Mar 18 19:26:46 ok Mar 18 19:27:10 will try that after this current test build Mar 18 19:27:12 =+ is prepending, += is appending. do the files end up in some other package instead of -dev? Mar 18 19:27:14 += append and =+ for prepend Mar 18 19:28:17 the files are ending up in the main package as though my FILES_ statements are being ignored and the default -dev package file selection method is being used Mar 18 19:30:22 hillct: that's because the main package is processed before -dev (look at the order in PACKAGES), so either overwrite FILES_${PN} to exclude those files or put -dev in the beginning: PACKAGES =+ "${PN}-dev" Mar 18 19:31:48 yup Mar 18 19:31:51 makes sense Mar 18 19:32:01 I thought += was the other way around Mar 18 19:34:28 brb Mar 18 19:38:23 why in the name of all that is holy is wifi not set as a zaurus machine feature?? Mar 18 19:38:48 I know it's not build-in but we kind of need wireless-tools et al to use wireless.... Mar 18 19:59:21 !seen zautrix Mar 18 19:59:22 Bernardo, I don't remember seeing zautrix. Mar 18 20:01:02 Hello Mar 18 20:04:04 pH5: still doesn't work. The problem appears to be the matching expression: Mar 18 20:04:12 FILES_${PN}-dev = "${libdir}/openpbx.org/modules/*.la \ Mar 18 20:04:12 ${libdir}/openpbx.org/*.la \ Mar 18 20:04:12 ${includedir}/openpbx/*" Mar 18 20:04:37 it likes * at the end of a matching line but not in the middle Mar 18 20:04:57 is there some other way to match *.la ? Mar 18 20:07:04 hillct: *.la should work. many other packages use that Mar 18 20:07:12 hillct: I don't understand why that wouldn't work. The python glob module is used to resolve those wildcards. Mar 18 20:07:37 * JustinP tests angstrom-e-image Mar 18 20:07:45 the docs say it should work, and as you say, it's working in other packages... Mar 18 20:07:51 strange Mar 18 20:08:46 While building gpe-image, I got in log.image_cmd. a lot of lines with "Wrong data_crc" and "Wrong bitmask" errors. THe image builds anyway, but it fails when I try to boot my Husky with it. Any hint? Mar 18 20:10:36 When I use the gpe-image-c7x0-*.rootfs.summary.jffs2 instead of the gpe-image-c7x0-*.rootfs.jffs2, I don't get those errors on boot, but kernel complains about not finding init. Mar 18 20:13:37 well, I guess I can use the other approach of specifying the files for the main package explicitly Mar 18 20:20:24 hi zecke Mar 18 20:20:43 DoKu: hey :) Mar 18 20:20:56 had a good trip home? Mar 18 20:21:42 DoKu: well, I didn' meet anyone but besides that it was okay Mar 18 20:21:58 DoKu: do you know where manu works? Mar 18 20:22:33 no, but i haven't tried to figure out neither Mar 18 20:27:59 DoKu: hmm, http://linuxtv.org/hg/~manu/stb0899-c5?fa=785a07c29ff5;file=linux/include/linux/dvb/frontend.h;style=gitweb annotate does not work Mar 18 20:29:52 DoKu: oh SVGGlobal is not inside WebKit, but I just saw a #ifdef MOBILE define Mar 18 20:36:47 in python, does * match 0 or more characters or 1 or more characters? Mar 18 20:37:17 Will this ${libdir}/*.so* match files ending in .so ? Mar 18 20:38:45 well, wifi is in DISTRO_FEATURES but wireless-tools isn't installed.... Mar 18 21:13:02 "morning" all Mar 18 21:13:07 RP: wb Mar 18 21:13:35 First time I've sat in front of a computer for any length of time outside work hours for quite a few days... Mar 18 21:14:40 RP: welcome, then ;-) I hope you've been relaxing Mar 18 21:15:09 JustinP: I wish ;-). Running around doing things, just not computer realted for a change... Mar 18 21:15:19 ah Mar 18 21:16:52 I did find time to remember what my motorbike looks like yesterday which counts as relaxing I guess. Sadly its showing some kind of engine fault. If only I could connect to the ECU... Mar 18 21:17:22 :( Mar 18 21:17:28 RP: did you search for a house yet? Mar 18 21:18:08 zecke: No, too little time atm... Mar 18 21:35:41 DoKu: is your driver working again? Mar 18 21:55:54 RP: I haven't had time to try to generalize my sharpsl-rc driver Mar 18 21:56:09 RP: would it be ok to push it into OZ? It works fine for spitz at least Mar 18 21:56:16 RP: I mean OE Mar 18 22:02:48 JustinP: yes, I'm ok with that Mar 18 22:03:24 RP: of course I probably need to test with the newest kernel first... Mar 18 22:05:44 JustinP: Would be wise, I think the patch will need tweaking for 2.6.20 iirc Mar 18 22:19:10 Good night. Mar 18 22:52:46 03nail 07org.oe.dev * rce6c16b8... 10/ (7 files in 2 dirs): linkstation/turbostation/n2100 machine support: Clean up and rename Mar 18 22:52:50 03nail 07org.oe.dev * r89bba421... 10/ (1 conf/distro/foonas.conf): foonas: Update defines for linkstations, n2100 and turbostation Mar 18 22:59:02 Laibsch: re madwifi for nas100d - look in task-mokogateway for the correct way to do it based on availability of pci or not. Mar 18 23:09:12 rwhitby: Thanks for responding. Mar 18 23:09:32 I am mainly interested in understanding if the mentioned bug can be closed or not. Mar 18 23:09:38 there should be no need for an nas100d machine Mar 18 23:09:43 which bug is it again? Mar 18 23:10:10 823 Mar 18 23:10:21 I think Mar 18 23:10:24 !oebug 823 Mar 18 23:10:26 * * Bug 823, Status: NEW, Created: 2006-04-08 09:11 Mar 18 23:10:27 * * evanchsa(AT)gmail.com: hostap-daemon package updates Mar 18 23:10:28 * * http://bugs.openembedded.org/show_bug.cgi?id=823 Mar 18 23:10:32 yup Mar 18 23:11:02 03nail 07org.oe.dev * r40982544... 10/ (1 conf/distro/include/foonas.inc conf/distro/foonas.conf): foonas: Clean up further, move lots to include/foonas.inc Mar 18 23:11:07 03nail 07org.oe.dev * r9e3f5936... 10/ (1 conf/distro/foonas.conf packages/images/foonas-image.bb): foonas: Update image config, move crap from distro config Mar 18 23:11:21 hey, cool feature Mar 18 23:13:43 * v8jlene is back Mar 18 23:13:47 rwhitby: Anything like DEPENDS_nas100d += "madwifi-modules" necessary? Mar 18 23:14:23 it should use the pci switch like in task-mokogateway Mar 18 23:15:34 Laibsch: like this: http://pastebin.ca/400854 Mar 18 23:15:48 (see the HAS_PCI stuff) Mar 18 23:16:04 that should match what wpa_supplicant does Mar 18 23:17:54 Ok, too late for me to fix now. Mar 18 23:18:06 Maybe tomorrow (but probably not since I am away) Mar 18 23:19:04 Laibsch: /msg me an email address and I'll email you that patch Mar 18 23:19:22 Can you just attach it to bug 823? Mar 18 23:19:31 ok Mar 18 23:19:37 Or just commit it Mar 18 23:19:43 Thanks. Mar 18 23:19:48 no, I can't commit it. Mar 18 23:20:05 I couldn't get hostap-deamon to work with madwifi-ng Mar 18 23:22:21 I don't understand. The patch you will post is not yet functional? Or is it just that your meta-data is so different from .dev that you cannot easily create a patch against .dv? Mar 18 23:22:24 .dev Mar 18 23:22:46 the patch applies fine, and builds the packages correctly. the packages (hostapd and madwifi-ng) don't work together. Mar 18 23:22:53 some ioctl error Mar 18 23:25:09 OK, I understand. The current hostapd and madwifi in .dev do not have that problem, do they? Mar 18 23:26:33 mickey|bbl: why is the file called callbacks.h and not closures.h? Mar 18 23:28:45 because using the scientificly incorrect term is sometimes preferable... the term callback is clear to most people, the term closure is not :/ Mar 18 23:29:55 mickey|bbl: then we need to teach them better :) Mar 18 23:30:15 mickey|bbl: the gtk+ docu puts an emphasis on closure, I always need to (re)read the wikipedia article :) Mar 18 23:30:31 i agree... but i can only save the world step-by-step ;) Mar 18 23:31:00 03nail 07org.oe.dev * rebcce100... 10/ (6 files in 3 dirs): foonas: Move image-stuff to classes for lsppchg, lsppchd, n2100 and turbostation Mar 18 23:31:08 mickey|bbl: GtkTextBuffer is horrible compare to QSimpleRichText+QMimeSource :} Mar 18 23:31:29 mickey|bbl: and GdkPixbuf has issues with loading data from memory Mar 18 23:35:45 don't turn the knife in the wound Mar 18 23:35:58 i know it sucks in a lot of areas Mar 18 23:36:04 but it gets better... Mar 18 23:36:15 mickey|bbl: yeah, I wonder how to cache images Mar 18 23:36:43 mickey|bbl: I could save them as 8bit andn then use gdk_pixbuf_new_from_data... Mar 18 23:36:59 mickey|bbl: but chordmaster is a good example (hopefully) Mar 18 23:38:12 ya, i always wanted to write a more comprehensive example Mar 18 23:38:17 *sigh* time Mar 18 23:38:59 see also stylus-demo for how to embed your navigation part to get the auto-fullscreen trigger button Mar 18 23:39:23 hmm Mar 18 23:39:33 OT: how can get a traceback instead of a ParseError? Mar 18 23:39:55 NOTE: Handling BitBake files: - (0344/4348) [ 7 %]NOTE: Retrieved remote revisions: [''] Mar 18 23:39:55 ERROR: Error in executing: /local/pkg/openmoko/openembedded/packages/s3c2410-utils/sjf2410-linux-native_svn.bb Mar 18 23:39:55 ERROR: Exception: Message:local variable 'date' referenced before assignment Mar 18 23:40:04 i'd like to know where exactly this bails out Mar 18 23:40:04 mickey|bbl: oh hmm MokoTreeView vs. MokoNavigationList, hmm where is the difference? Mar 18 23:40:19 mickey|bbl: don't touch my wound :) Mar 18 23:40:32 mickey|bbl: go to the except and uncomment it :} Mar 18 23:40:47 zecke: please only use MokoTreeView. MokoNavigationList is some encapsulation from the Shanghai guys to get a border that rather needs to done by the theme. Please ignore MokoNavigationList Mar 18 23:41:00 if I would know which except I'd do that Mar 18 23:41:09 there's a whole lot of try/except clauses in BitBake :) Mar 18 23:41:12 mickey|bbl: look in cooker.py Mar 18 23:41:36 k Mar 18 23:41:38 bb.parse is a good indicator as well Mar 18 23:42:41 mickey|bbl: parse_bbfiles Mar 18 23:43:00 :645 Mar 18 23:44:14 hmm Mar 18 23:44:27 sure? Mar 18 23:44:30 mickey@allanon:/local/pkg/oe/bitbake$ find . -name "*"|xargs grep parse_bbfiles Mar 18 23:44:30 mickey@allanon:/local/pkg/oe/bitbake$ Mar 18 23:44:51 i didn't touch all this for quite a while Mar 18 23:44:55 can't find anything Mar 18 23:45:29 mickey|bbl: which version of bitbake is this? Mar 18 23:45:32 hm collect_bbfiles ? Mar 18 23:45:34 1.6.6 Mar 18 23:46:19 yeah, collect_bbfiles looks good Mar 18 23:46:49 yep, got it Mar 18 23:46:50 mickey|bbl: right Mar 18 23:46:50 thanks Mar 18 23:47:02 psokolovsky: ping Mar 18 23:51:29 mickey|bbl: I sketched my deliveres and ideas for GSoc together and submitted something, don't hesitate to turn it down :) Mar 18 23:51:36 ;) Mar 18 23:51:41 thanks Mar 19 00:15:57 nite Mar 19 00:59:03 http://rafb.net/p/fSEvaL84.html means I need liboil-native (which is not yet in OE), right? Mar 19 01:09:16 Laibsch: No, you'd only need the native version if you were building a -native package, which your not doing. Mar 19 01:09:34 v8jlene: I am not sure I am convinced. Mar 19 01:09:57 gst-plugins is looking for liboil in the configure phase Mar 19 01:10:22 Thus I assume the liboil stuff needs to be in staging at that time. Mar 19 01:10:52 liboil itself will not suffice for compiling the package I believe. Am I mistaken? Mar 19 01:11:10 Yep, it needs to be in staging. Mar 19 01:11:29 But liboil itself is sufficient. Mar 19 01:11:32 Well, it is a lib Mar 19 01:11:42 You only need -native to build native packages. Mar 19 01:11:53 v8jlene: Try to compile it and you will see it is not ;-) Mar 19 01:12:01 it = gst-plugins Mar 19 01:12:29 First thing to check would be if the liboil pc file is correctly installed into tmp/staging//lib/pkgconfig Mar 19 01:14:02 no, there is nothing there. Mar 19 01:14:08 Let me check the liboil bb Mar 19 01:14:32 -rw-r--r-- 1 lenehan lenehan 493 2007-03-18 10:46 tmp/staging/i486-linux/lib/pkgconfig/liboil-0.3.pc Mar 19 01:23:37 Laibsch: I can't actual test it here because it depends on tremor and the SRC_URI and/or the server it points at is broken. Mar 19 01:24:42 Can you keep an eye on the URL so we can fix it if the server does not come back up? Mar 19 01:27:03 http://svn.xiph.org/ itself works Mar 19 01:30:08 v8jlene: I have the liboil stuff in ../tmp/openzaurus/staging/arm-linux/lib/pkgconfig/ Mar 19 01:31:08 Laibsch: Yeah, looks like it's been split into two versions... looks like this is what we want: http://svn.xiph.org/branches/lowmem-branch/Tremor/ Mar 19 01:31:42 The other branch is for "platforms whose smallest data unit is larger than 8 bits" Mar 19 01:32:15 That should not really matter as the latest svn stuff is pref -1 Mar 19 01:32:30 I take that all back.. they are branches ;) It looks like it's all there just fine... Mar 19 01:33:08 Yeah, but even the other version is being checked out from svn. Mar 19 01:34:06 yes, but even they had rearranged stuff in the meantime it would not mean we needed an update to SRC_URI since the server itself was obviously still in place and the rearrangement would not have effect the old revision. Mar 19 01:34:15 even if they had Mar 19 01:35:21 Back on your other problem, so you have the .pc file. So the next question would be why isn't the configure for gst-plugins finding it? Mar 19 01:35:37 I am not sure it was in place earlier. Mar 19 01:35:48 I did a bitbake - c rebuild liboil Mar 19 01:36:09 Now I am trying again for gst-plugins. Let's see if it masters do_configure now Mar 19 01:37:40 yes, do_configure went fine now. gst-plugins is at do_compile now. Mar 19 01:37:47 Thanks v8jlene **** ENDING LOGGING AT Mon Mar 19 02:59:57 2007