**** BEGIN LOGGING AT Fri Jan 18 02:59:57 2008 Jan 18 07:35:42 moin Jan 18 07:48:59 hi zecke Jan 18 07:50:43 moin Jan 18 07:51:10 pb__: all these years and I still know the gcc build system so little Jan 18 07:51:28 pb__: but If you don't beat me with it, I will try to clean our recipes for gcc4.3 and llvm Jan 18 08:10:36 Hello Jan 18 08:16:24 I have a question about clean up tmp files. For a fast rebuild I would like to "save" the native and cross toolchain in a tar archiv and extract the archiv befor a new build. But the archiv is tooo big. I don't need to save the src, .o's and install files. So which files can i delete? I need to take the stamps to suppress a rebuild. Can I delete the whole tmp/work folder? Jan 18 08:18:52 should i keep the cache and deploy folder? Jan 18 08:22:49 pf_: I'd just clean out everything Jan 18 08:23:10 or you are likely to run into the same trouble that made you want to clean out tmp in the first place Jan 18 08:23:33 I only keep deploy Jan 18 08:25:13 ah scratchbox is stealing 11gb from my disk :) Jan 18 08:25:57 I need to reduce the buildtime. If you only keep deploy, the next build will rebuild the native and cross toolchain Jan 18 08:26:18 I think i need the tmp/stamps and tmp/deploy Jan 18 08:27:32 pf_: you are substitung your own fiddling time with build time Jan 18 08:27:47 I'd let my computer work more when I have to work less just about any time Jan 18 08:27:59 And I have a VERY slow computer for compilation Jan 18 08:31:03 no the problem is that we make a build every day and if we make little changes, we need to wait too long for a new build from our buildserver Jan 18 08:33:39 pf_: You know about PR? Jan 18 08:33:48 what changes are you making Jan 18 08:34:00 Du ziehst das Pferd vom flaschen Ende auf Jan 18 08:34:13 tar cvzf toolchain.tar.gz tmp/stamps tmp/cross tmp/staging Jan 18 08:34:20 Laibsch: warum? Jan 18 08:34:22 well, try it Jan 18 08:34:26 It might work Jan 18 08:34:37 I'd be more prudent with my time Jan 18 08:34:52 pf_: You know about PR? Jan 18 08:35:03 What changes are you making? Jan 18 08:35:04 Laibsch: PR? No Jan 18 08:35:16 Das könnte das richtige Ende vom Pferd sein Jan 18 08:36:08 we never change the toolchain. Jan 18 08:36:20 Laibsch: was ist PR? Jan 18 08:37:15 pf_: http://www.openembedded.org/user-manual Jan 18 08:37:57 PR is increased by 1 when source code has not changed but the bb file has Jan 18 08:38:30 bitbake will then go through the motions again for that particular package Jan 18 08:39:12 pf_: You should familiarize yourself with the data in a bb file Jan 18 08:40:05 Ok this "PR", i know it Jan 18 08:43:30 ASSUME_PROVIDED += " virtual/${TARGET_PREFIX}gcc " Jan 18 08:43:36 this is it Jan 18 08:52:39 * Laibsch still wonders what kind of changes you make that require daily clean-out of tmp Jan 18 08:53:15 but I guess there are infinite ways to shoot yourself in the foot, make your life more miserable than it needs to be and generally waste time ;-) Jan 18 08:59:15 you are rigth, we don't need to rebuild the whole image from scratch, but we want to make sure that we can rebuild our images every time, it would be a big problem if we experience in a few month the your image X can't be build up from scratch. Now the whole tmp will be only deleted at WE Jan 18 09:21:04 morning Jan 18 09:21:10 morning all Jan 18 09:21:14 pf_: you know about "bitbake -c rebuild"? Jan 18 09:21:14 hi all ! Jan 18 09:21:17 hi RP Jan 18 09:21:17 re Jan 18 09:21:24 hey RP Jan 18 09:21:25 good morning RP & hrw! Jan 18 09:21:35 good morning, gremlin[it] Jan 18 09:21:47 wow, everybody waking up at the same time Jan 18 09:28:16 hahah really i'm in office since 08 am :) Jan 18 09:31:11 * Laibsch subtracts 2,5 hours from gremlin[it] work hours since he forgot to stamp time card Jan 18 09:31:13 ;-) Jan 18 09:31:31 hahahha Jan 18 09:32:53 what's the -c command for build a package and it's dependencies ? Jan 18 09:33:03 for bitbake Jan 18 09:33:04 buildall Jan 18 09:34:15 i'm not able (now trying with buildall) to obtain xserver-xorg-xinit ... and other script to automatic startup X at boot Jan 18 09:40:59 no ... -c buildall xorg-xserver dont produce xorg-xserver-xinit :( Jan 18 09:51:44 has anyone seen something like that? Jan 18 09:51:45 http://pastebin.mozilla.org/306507 Jan 18 09:52:20 if I manually copy *curl* from ../arm-angstrom-linux-gnueabi to all-angstrom-linux-gnueabi then the task runs through Jan 18 10:13:37 lo Jan 18 10:23:27 what's is the way to report broken urls in .bb files? bugzilla from topic? Jan 18 10:24:00 The_Niz: yes Jan 18 10:24:11 hi florian_ Jan 18 10:24:18 okay tnx Jan 18 10:24:39 hi all Jan 18 10:26:38 is anyone using busybox 1.7.2? Jan 18 10:27:02 it seems that the udhcpc does not get/provide the DNS server ips anymore Jan 18 10:33:49 I have 1.8.2 on alix Jan 18 10:34:14 figured it out, was my fault - permission badness of the script that takes care of the resolv.conf stuff Jan 18 10:36:10 a info ... angstrom-200.12 is just a distro version or a stable branch/tag of OE metadata too ? Jan 18 10:36:38 stable branch Jan 18 10:37:30 hrw name ? Jan 18 10:37:38 on OE wiki i don't find nothing about Jan 18 10:38:12 org.openembedded.angstrom-2007.12-stable Jan 18 10:38:25 I found the reference to it in angstrom-2008.1.conf Jan 18 10:38:35 and it's also documented on the angstrom website Jan 18 10:38:59 gremlin[it]: I can say one thing about OE website but it is not political correct Jan 18 10:39:05 :) Jan 18 10:39:23 hahaha :) Jan 18 10:39:41 i found some info in Angstrom page Jan 18 10:41:04 can i use the same OE.mtn to handle both sable and dev branch right ? Jan 18 10:42:26 yes Jan 18 10:52:06 bonjour :) Jan 18 11:06:32 avahi Genesis Jan 18 11:08:12 nooooooooooooooo avahi nooooooooooooo broadcast and multicast !!! Jan 18 11:08:20 ^^ Jan 18 11:08:24 yo hrw :) Jan 18 11:10:38 gremlin[it]: ;D Jan 18 12:09:14 How is it that http://rafb.net/p/c18jcd64.html can happen? I even rebuilt busybox just seconds before trying to compile that image. Jan 18 12:09:20 Any ideas? Jan 18 12:10:27 mine also failed, but looks very much different from yours Jan 18 12:10:44 anyway, no clue, neither for my failure nor for your log :) Jan 18 12:15:25 I'm passing a string in CXXFLAGS_append to a pre-compiler define but bitbake adds \ before each " how do I escape a qoute in a bb file so bitbake does not add \? Jan 18 12:26:47 will it stay with .net? Jan 18 12:27:04 yes Jan 18 12:27:12 .org may be repaired eventually Jan 18 12:27:16 but it will stay with .net as well Jan 18 12:27:23 .net is under my control :) Jan 18 12:27:54 hi mickeyl Jan 18 12:27:59 hey hrw Jan 18 12:28:02 how's life? Jan 18 12:28:07 mickeyl: quite ok Jan 18 12:28:11 (except waiting for the little one ;) Jan 18 12:28:24 mickeyl: busy with few projects and testing poky under RPM based craps Jan 18 12:28:31 centos, opensuse, fedora Jan 18 12:28:55 good that we do not have clients using mandriva (or how it is called in this week) Jan 18 12:29:29 mandriva is fairly straightforward, didn't have any problems as long as I was using it Jan 18 12:32:13 I need more desk space Jan 18 12:32:23 and as always: faster machines ;) Jan 18 12:36:55 when it comes to network installer nothing beats Debian Jan 18 12:37:16 fedora just installed firefox... as part of development utisl Jan 18 12:47:29 Hi Jan 18 12:48:54 Did anyone realize that changes in the compose keys in the kernel keymap doesn't affect the compose combinations under kdrive? Is this expected? Jan 18 14:36:39 mickey|bbl: Do you think that http://oz.leggewie.org/wip/wifi-suspend.patch is the right way to move this into a separate package? Jan 18 14:36:54 morning Jan 18 14:39:55 Salutations Jan 18 14:48:58 Are tasks supposed to be "bitbake-able"? Jan 18 14:51:35 Does anybody know where are the key compose combinations for libX11 stored in oe ? Jan 18 14:54:26 sirfred: no, I don't really Jan 18 14:54:39 but there is a meta-bug for keyboard-related issues in the tracker Jan 18 14:54:54 Maybe you can quickly find the relevant files being mentioned there? Jan 18 14:55:42 Laibsch: Will give it a look, thanks! Jan 18 14:57:36 bug 1908? Jan 18 14:58:02 Laibsch: Running into difficulties to connect to the bugs site Jan 18 14:59:16 * Laibsch points to topic Jan 18 14:59:24 .net Jan 18 14:59:26 not .org Jan 18 15:00:11 oops Jan 18 15:02:47 do we have enough people for fosdem to finally create OE fundation? Jan 18 15:03:12 hrw: wiki seems to be lacking in attendees Jan 18 15:03:32 I'm afraid the page has a lot of internal links to the .org domain Jan 18 15:03:45 form name="login" action="http://bugs.openembedded.org/" method="POST"> and so on Jan 18 15:04:41 sirfred: use the IP and fix you /etc/hosts :-) Jan 18 15:05:01 s/fix/hack Jan 18 15:05:40 XorA: there were problems with editing wiki Jan 18 15:05:54 sirfred: call all hacks fixes, that way you sound like a better coder :-) Jan 18 15:05:59 hrw: there were? Jan 18 15:06:03 XorA: :D Jan 18 15:06:33 I'm going to use the show_bug.cgi page directly. :-( Jan 18 15:08:09 XorA: I do not know more Jan 18 15:08:42 hrw: and we are down 2 core developers at least Jan 18 15:08:46 Hmm, 1908 is not very helpful. Just a set of people discussing if it is actually a bug. :) Jan 18 15:09:49 The fact is that I'm able to insert special characters using the compose feature, but I don't know how to change the compose map in X applications. The kernel compose combinations are not used by x Jan 18 15:10:00 s/kernel/console Jan 18 15:10:24 XorA: the problem with website (for me) is drupal which I do not understand and lack time to learn it Jan 18 15:10:51 * XorA runs drupal Jan 18 15:11:10 sirfred: I have already taken care of that Jan 18 15:11:11 XorA: Have you setup a wiki on drupal? Jan 18 15:11:20 It should be fixed now, pointing back to .net Jan 18 15:11:33 XorA: I find the mediawiki support a little poor Jan 18 15:11:33 If not, "vi /etc/hosts" Jan 18 15:11:37 sirfred: no, I just use it for my blog on xora.org.uk Jan 18 15:11:43 XorA: AH, ok Jan 18 15:12:02 sirfred: I tried the forums modules, but they were all crap, so I use FUDforum Jan 18 15:12:12 XorA: I was looking for something with a lot of features (forums, blogs, wiki,...) and most of them are not very good Jan 18 15:12:20 I mean, the drupal modules Jan 18 15:12:32 The forums are a hell, yes. :) Jan 18 15:12:34 sirfred: yeah, drupal forums models were shit Jan 18 15:12:35 sirfred: But did you see the russian support thing? Jan 18 15:12:51 Laibsch: Yes, about the xkbd packages? Jan 18 15:13:01 Look through that file, come on, be a little creative after I got you going in $some direction Jan 18 15:13:04 mainly I use drupal as wordpress didnt run on postgres Jan 18 15:13:08 Laibsch: Curiously, I'm afraid that xkb is not enabled in libx11. Jan 18 15:13:33 And I'm not sure what that actually means, related with compose stuff Jan 18 15:13:44 What a complex world! Jan 18 15:14:02 I just want to type easily my ñ's and €'s and so Jan 18 15:14:22 XorA: Down 2 core devs? You mean for FOSDEM? Jan 18 15:14:32 Laibsch: yes Jan 18 15:14:57 who is not coming? hrw and ? Jan 18 15:15:14 hrw is busy forking :-) Jan 18 15:15:48 sirfred: If I had known the answer, I'd have given it to you. All I can do is point you to related stuff I have seen in the past Jan 18 15:16:01 XorA: I hope he won't forget to exec, after that Jan 18 15:16:09 hrw: Forking what? Forking out cash? Jan 18 15:16:13 * Laibsch gets in line Jan 18 15:16:30 Laibsch: And I thank you Jan 18 15:16:34 Laibsch: my daughter cames to real world anyday soon Jan 18 15:16:46 oh, that kind of fork? Jan 18 15:16:55 Now, that indeed is a joyous fork! Jan 18 15:16:59 So, exec'ing soon. Congrats Jan 18 15:17:07 Let's all fork and be happy ;-) Jan 18 15:21:41 * XorA runs a forkless OS Jan 18 15:22:25 XorA: unless unofficial fork will gets you ;D Jan 18 15:25:02 this is dangerous ... but so fun :) Jan 18 15:26:55 * hvontres|work has two child proceses running around the house....and running and running .... Jan 18 15:32:34 hvontres|work: fork bomb hit you then :-D Jan 18 15:33:33 and using kill -9 isn't allowed :D Jan 18 15:35:21 XorA: does yaffs2-native from OE builds ok for you? Jan 18 15:35:34 hrw: yes Jan 18 15:41:23 in Poky I had to patch it a bit Jan 18 15:42:59 XorA: hmmm... well that would explain the way my house looks :) Jan 18 15:43:34 chouimat|work: yes, even if sometimes it seems like a good idea... Jan 18 15:45:07 hvontres|work: I know the feeling (even if I don't have kids) the next door couple has 5 of them and sometimes I want to put them out of my misery ... and they also seems to have no control over them ... loud music and shouting, yelling until 3 in the morning Jan 18 15:45:10 Ah, but sending a SIGSTOP is a good thing now and again. Jan 18 15:47:15 * mwester finds that the child processes seem to consume all available resources at times. Jan 18 15:48:10 can anybody tell me where i can find the ksyms file if /proc/ksyms not exist? ksymoops will not take /proc/kallsyms Jan 18 16:05:03 guys, how can it be that "bitbake -c rebuild task-base-extended" runs fine, but that "bitbake console-image" complains about unsatisfiable dependencies for task-base-extended in do_rootfs? http://rafb.net/p/pvZmuJ25.html Jan 18 16:06:19 Laibsch: have you tried -DDD, maybe there will be clues in the debug output? Jan 18 16:06:20 just a guess.. Jan 18 16:06:20 * hrw pushed another set of crap Jan 18 16:06:51 Jin^eLD: I'll take a look Jan 18 16:23:43 Anyone with qt4-x11-free experience? It builds fine by itself, but if I try to put it in an image, I get: ERROR: Cannot satisfy the following dependencies for qt4-x11-free: libqtcore4, etc. Jan 18 16:24:02 tnb: I get similar problems now Jan 18 16:24:10 but not for qt4 Jan 18 16:24:15 overall Jan 18 16:24:43 tnb: When do the errors occur? in do_rootfs? Jan 18 16:24:48 Laibsch: any workaround that you know of? Jan 18 16:24:49 yes Jan 18 16:25:06 Jin^eLD: Unfortunately, I don't see anything: http://oz.leggewie.org/wip/bb.log Jan 18 16:25:18 tnb: are you up2date? Jan 18 16:25:41 i tried building the initramfs-image, which results in a 3MB image... which is probably because it uses the glibc Jan 18 16:25:55 tnb: do you actually have a libqtcore4.ipk on hand in your deploy directory? Jan 18 16:25:57 Laibsch: I didn't try with .dev, only with angstrom2007.12-stable branch Jan 18 16:26:11 is it possible to use uclibc for an image, or do i need to use a separate build tree? Jan 18 16:26:40 pb_: I am getting similar errors since a few hours ago, see my earlier pastebin Jan 18 16:26:45 tnb: what happens if your try to put a user of Qt4 into the image (e.g. qsvn?) Jan 18 16:27:11 zecke: I am getting the feeling this is not qt4 problem Jan 18 16:27:56 pb_: oops, nice naming on "qtcoree": libqtcoree4_4.3.3-r1_armv5te.ipk Jan 18 16:28:09 heh Jan 18 16:28:35 tnb: debian package renaming, libQtCoreE -> libqtcoree Jan 18 16:28:36 yeah, they *all* have an extra "e" Jan 18 16:28:51 tnb: this allows to install QtopiaCore and Qt/X11 next to each other Jan 18 16:29:21 oh, so the "e" version is the embedded version? Jan 18 16:29:52 zecke: Was there some changes to this recently? Jan 18 16:29:58 if so, then building the X11 version is not creating the libqtcore4 Jan 18 16:30:26 zecke: because I can build task-base fine, but the image complains about unsatisfiable deps for the image (task-base) Jan 18 16:31:18 Laibsch: and do the packages that are unsatisfiable dependencies actually exist? Jan 18 16:33:51 i.e. the libqtcore4 ipk. only the libqtcoree4 ipk exists Jan 18 16:35:04 * * OE Bug has been marked as DUPLICATE of bug 2806 by Jan 18 16:35:06 * *  Openldap fails to build Jan 18 16:35:08 * * http://bugs.openembedded.org/show_bug.cgi?id= Jan 18 16:35:11 tnb: yeah, looks like you have somehow ended up with the -e version of qt being built rather than the x11 version Jan 18 16:35:46 might be worth trying a manual rebuild of the qt-x11 package and see what happens Jan 18 16:36:57 tnb: how did you end up with the qtopiacore version? We don't have something like virtual/qt in place Jan 18 16:40:14 zecke: I built qtopia-core when this problem first showed up; i.e., the missing libs occur when just qt4-x11-free is built Jan 18 16:40:48 I cleaned qtopia-core and rm'd all of the *qt* ipks and now I'm rebuilding qt4-x11-free Jan 18 16:40:57 we'll see soon. Jan 18 16:41:08 tnb: do you try to directly install libqtcore? or is it in a task and you install the task? Jan 18 16:41:35 I created a new image based from one of the x11 images Jan 18 16:42:16 also tried a new task approach. It did work fine in angstrom-2007.12-rc1 Jan 18 16:43:56 bye Jan 18 17:06:51 pb_: yes, the ipk are there: http://rafb.net/p/ajvSGt76.html Jan 18 17:07:14 tnb: Did you find the root of your problem? Jan 18 17:14:45 Laibsch: no, looking at what is built, ipks for libqtcore4-dev (and doc, etc) are created, but libqtcore4 is not Jan 18 17:15:40 tnb: not created? any other package missing? Jan 18 17:16:15 zecke: right, the -dev and -dbg packages exist in deploy but not the main ones Jan 18 17:16:40 tnb: weird, Jan 18 17:16:50 tnb: I will check once I'm back in the hotel Jan 18 17:17:06 zecke: cool, thanks Jan 18 17:20:46 tnb: what machine are you building for? Jan 18 17:34:13 Laibsch: nothing in OE. I'll try for hx4700. Jan 18 17:34:41 in case there's some var that's not set or something goofy like that Jan 18 17:37:26 OK Jan 18 17:37:37 I thought maybe this was due to the latest collie changes Jan 18 17:37:44 But it does not look like it, then Jan 18 17:39:16 Laibsch: and which architectures does ipkg.conf admit as installation candidates? Jan 18 17:39:51 pb_ Sorry, I don't understand your question? Jan 18 17:39:56 PACKAGE_ARCH? Jan 18 17:40:07 for collie that would be arm-oabi Jan 18 17:43:16 what Architectures does the ipkg.conf file mention? Jan 18 17:53:06 pb_: Is http://rafb.net/p/gSwXbM46.html the file you were looking for? Jan 18 17:53:44 No Arches, but I think that was always like that and did not interfere with image creation in the past Jan 18 17:53:53 Sonkei does not yet have feeds defined Jan 18 17:55:06 No, that's the installed ipkg.conf file. You need the one that ipkg is actually using during image construction. Jan 18 17:55:12 It should be mentioned on the ipkg command line. Jan 18 17:56:56 Where would that be in your case? Jan 18 17:57:01 In what directory Jan 18 17:57:11 $DEPLOY Jan 18 17:57:12 ? Jan 18 17:57:43 I only have ipkg.conf files in $WORKDIR Jan 18 17:57:56 And one in staging Jan 18 17:58:20 What does the ipkg command line say? Jan 18 17:58:44 the one in staging looks like this: http://rafb.net/p/EGgmyH52.html Jan 18 17:59:00 I never saw any ipkg command line during the build Jan 18 17:59:10 The one that produced the ipkg error you pasted earlier. Jan 18 18:07:00 pb_: I guess that would be the link above Jan 18 18:07:05 http://rafb.net/p/EGgmyH52.html Jan 18 18:07:15 the arches look OK to me Jan 18 18:08:31 I'd have thought you'd want arm-oabi in there, if that's what you're using as PACKAGE_ARCH. Jan 18 18:09:07 I can only assume that is handled automatically Jan 18 18:09:14 But maybe it is not Jan 18 18:09:23 I'll try for spitz and compare Jan 18 18:20:39 It should indeed be automatic, but it doesn't seem to be working correctly for you. Jan 18 18:28:47 zta: I think it does, you could check using intel tools Jan 18 18:28:50 oops wrong dindow Jan 18 18:34:36 anyone familiar with the git fetcher? Jan 18 18:34:40 how can i specify a branch? Jan 18 18:35:30 mickeyl: I never worked that out Jan 18 18:35:37 oh Jan 18 18:35:38 bummer Jan 18 18:35:38 mickeyl: I ended up making my branch a new git repo Jan 18 18:35:41 i need one now Jan 18 18:35:45 but Im not very clever sometimes Jan 18 18:35:48 heh Jan 18 18:36:01 ok, that means go and check the git fetcher... Jan 18 18:36:05 * mickeyl does Jan 18 18:37:04 why is the a files/device_table* in the root of OE? Jan 18 18:37:17 i think that has historic roots Jan 18 18:37:27 but i honestly forgot Jan 18 18:37:37 Author: lenehan@openembedded.org Jan 18 18:37:37 | Date: 2006-07-06T14:58:36 Jan 18 18:37:54 looks like a cockup no-one ever fixed Jan 18 18:37:54 check bitkeeper, it's been there for years Jan 18 18:41:48 hmm Jan 18 18:41:58 why are we using git checkout-index as opposed to git checkout? Jan 18 18:43:34 * XorA|gone looks towards RP Jan 18 18:45:38 Hmm, xkbd is just a "onscreen" keyboard Jan 18 18:52:23 hi mickeyl Jan 18 18:53:13 sirfred: what were you expecting? Jan 18 18:54:41 pb_: Something related with x11 key management Jan 18 18:55:40 pb_: I'm trying to figure out where keys are processed for composing, for example. Jan 18 18:58:14 sirfred: xmodmap is what controls the compose settings Jan 18 18:58:29 pb_: I was not able to find an example of composing with xmodmap Jan 18 18:59:01 Can you be more specific about what you are trying to do? Jan 18 18:59:07 Of course. Jan 18 18:59:45 I have to go home now, but I'll read the scrollback later. Jan 18 19:00:00 Currently, pressing the compose key (if defined) and then 'a' and 'a' again generates a å Jan 18 19:00:19 Or Compose + a + e = æ Jan 18 19:00:42 I've changed that in the kernel keymap, but X seems to use different source to configure that Jan 18 19:01:12 So, in my ubuntu, while stracing a x client, it looks for a Compose file in the locale dirs. Jan 18 19:01:31 With diet-x11 it doesn't happen Jan 18 19:01:56 But anyway, compose works, with what I think are some default, stored in some place unknown for me Jan 18 19:02:01 That's the problem Jan 18 19:02:10 pb_: Ok, Thanks for listening. :) Jan 18 19:08:37 pb_: spitz compiles fine Jan 18 19:08:54 could indeed be the changes to machine configs Jan 18 19:09:00 which were done lately Jan 18 19:19:54 hi all! Jan 18 19:20:22 getting back to my openmoko image tries to get it working into my HTC Magician Jan 18 19:21:28 i've now got a silly conflict : ERROR: The following packages conflict with libgsmd:libgsmd-devel Jan 18 19:21:38 any ideas? Jan 18 19:22:04 looks like you tried to build/install both libgsmd versions Jan 18 19:22:07 choose one Jan 18 19:22:50 anyone working on HTC Touch support? Jan 18 19:23:03 mickeyl: i've just done a simple 'bitbake openmoko-image' Jan 18 19:23:28 which DISTRO ? Jan 18 19:24:04 RP: i'm afraid i need a hand on branch support for the git fetcher ;) Jan 18 19:24:27 mickeyl: Angstrom Jan 18 19:24:33 how are things in the oe universe lately? Jan 18 19:24:37 mickeyl: needs help with git? Jan 18 19:24:42 otavio: yeah Jan 18 19:24:48 mickeyl: I'm used to git Jan 18 19:24:53 mickeyl: what you want? Jan 18 19:25:04 kergoth: good so far. could you point bugs.openembedded.org to 76.74.156.109, please? Jan 18 19:25:08 kergoth: how's your life? Jan 18 19:25:14 will do Jan 18 19:25:19 otavio: i'd like to have support for branches in the git fetcher Jan 18 19:25:20 kergoth: thanks Jan 18 19:25:28 not bad, thinking i need a new hobby since i hardly ever code anymore Jan 18 19:25:29 heh Jan 18 19:25:30 you? Jan 18 19:25:53 mickeyl: where is the current one code to me look at? Jan 18 19:25:53 good so far. still with OpenMoko Inc. as main client and slowly writing my next book Jan 18 19:26:07 nice Jan 18 19:26:10 what's the book? Jan 18 19:26:12 otavio: lib/bb/fetch/get.py Jan 18 19:26:30 mickeyl: any idea for the syntax? Jan 18 19:26:31 kergoth: softwaredeveloping on linux smartphones and pdas ( no surprise here ;) Jan 18 19:26:33 mickeyl: for the uri? Jan 18 19:26:39 ah, nice Jan 18 19:26:51 otavio: I'd like to select the branch with ;branch= Jan 18 19:26:55 within SRC_URI Jan 18 19:27:14 the git fetcher is doing so many commands i'm a bit overwhelmed by that : Jan 18 19:28:01 kergoth: do you still read oe-related mailing lists? Jan 18 19:28:25 mickeyl: it's on bb source? not OE code Jan 18 19:28:33 hi mickeyl kergoth Jan 18 19:28:34 no. the fetchers are part of bitbake Jan 18 19:28:40 hi chouimat|work Jan 18 19:28:42 think i'm still subscribed, keep an eye out for anything particularly interestnig, but for the most part i ignore it Jan 18 19:28:47 mickeyl: will take a look at it, hold Jan 18 19:28:51 kergoth: righto Jan 18 19:28:54 otavio: cool, thanks. Jan 18 19:29:25 * chouimat|work is currently having fun with a marwell ARM926EJ-Sid based cpu Jan 18 19:29:45 hey chouimat Jan 18 19:30:02 kergoth: I'm still enjoying my new job :) Jan 18 19:30:16 nice Jan 18 19:30:23 kergoth: www.axentra.com Jan 18 19:30:37 i still like what i'm doing at mvista, just hoping they'll give me a different customer to support, so i have real bugs to diagnose instead of just backporting upstream fixes Jan 18 19:30:40 bored outof my mind half hte time Jan 18 19:30:57 otavio: my first attempt was http://pastebin.ca/861004, but that doesn't really cut it Jan 18 19:31:32 kergoth: hehe ... currently abusing yesterday software build on the new device Jan 18 19:32:56 mickeyl: you looks to have it Jan 18 19:33:02 mickeyl: looks like you can use tags Jan 18 19:34:31 mickeyl: try to use tag ... obviously, it would be better to use branch keyword, but it could be work out later Jan 18 19:34:41 mickeyl: looking at the source code, it looks it will work Jan 18 19:35:13 branch= sounds a little odd, naming wise, since checkout can take any tree-ish/ref Jan 18 19:35:56 hmm Jan 18 19:36:07 * mickeyl tries Jan 18 19:36:36 mickeyl: you shouldn't fix it to origin remote Jan 18 19:36:45 mickeyl: as you done in your diff Jan 18 19:36:51 mickeyl: better to use the tag ... Jan 18 19:37:04 so tag=origin/branchname ? Jan 18 19:37:14 mickeyl: maybe the code will need changes to support it... Jan 18 19:37:17 mickeyl: yes ... Jan 18 19:37:26 mickeyl: but looks like the if won't allow it Jan 18 19:37:31 hmm, no that doesn't work Jan 18 19:37:46 you can't give a tag with a slash within Jan 18 19:38:03 the current code parses this into a filename for reusing checkout Jan 18 19:38:29 and besides this... Jan 18 19:38:37 how would i use tag for a branch _and_ a revision? Jan 18 19:40:09 mickeyl: if you use a revision, you don't need a branch Jan 18 19:40:22 okay Jan 18 19:40:28 mickeyl: since a change with a different parent, gets a different sha1sum Jan 18 19:40:38 ah, branche independent Jan 18 19:40:43 interesting Jan 18 19:40:43 mickeyl: yes. Jan 18 19:40:54 mickeyl: what means is the parent object Jan 18 19:41:06 mickeyl: that's why if you sign a tag, you're safe on git Jan 18 19:41:18 mickeyl: if you change anything, it'll change the sum Jan 18 19:41:40 mickeyl: now, to allow it to use the branch ... do: Jan 18 19:43:25 - ud.localfile = data.expand('git_%s%s_%s.tar.gz' % (ud.host, ud.path.replace('/', '.'), ud.tag), d) Jan 18 19:43:25 + ud.localfile = data.expand('git_%s%s_%s.tar.gz' % (ud.host, ud.path.replace('/', '.'), ud.tag.replace('/', '.')), d) Jan 18 19:43:26 ? Jan 18 19:43:45 Laibsch: I changed qt_packaging.inc to have NONDEV_PACKAGES be += and I added QT_LIBINFIX = "" to qt4-x11-free.inc and now it builds all of the required packages Jan 18 19:45:07 pb_: I'm not sure which of those fixed it, but it seems to be building everything now Jan 18 19:45:15 mickeyl: http://paste.debian.net/47164 Jan 18 19:45:48 hmm, right Jan 18 19:45:49 tnb: what's the problem are you having? I'm working on another qt building fix too Jan 18 19:45:55 the problem with this approach is Jan 18 19:46:04 it will completely confuse checkouts with different branches Jan 18 19:46:11 as the branch name is not encoded in the filename Jan 18 19:46:14 mickeyl: it won't Jan 18 19:46:18 mickeyl: doesn't matter Jan 18 19:46:33 otavio: the problem was that qt4 was not building all of the packages. the -dev and -dbg packages built, but not the nondev Jan 18 19:46:36 mickeyl: if you had a change on the branch, it'll get a new hash Jan 18 19:46:42 mickeyl: so, new checkout Jan 18 19:46:56 tnb: how I test it? Jan 18 19:47:03 that is, qt4-x11-free when I say qt4 Jan 18 19:47:32 tnb: I think I know what's Jan 18 19:47:41 tnb: I'll provide you a patch .. so you can try Jan 18 19:47:43 otavio: add qt4-x11-free to one of the x11 images Jan 18 19:47:43 tnb: can be? Jan 18 19:48:01 mickeyl: got it? Jan 18 19:48:16 mickeyl: it changes the object pointed by the branch Jan 18 19:48:16 otavio: spoke too soon. still unresolved dependencies: libqtopengl4 qt4-examples qt4-plugins-accessible qt4-plugins-sqldrivers Jan 18 19:48:21 mickeyl: so the hash changes Jan 18 19:48:36 mickeyl: so it makes a new checkout Jan 18 19:48:44 ya, sounds convincing Jan 18 19:48:56 mickeyl: does it works for you? Jan 18 19:49:06 looks like it's checking out the proper branch Jan 18 19:49:07 i need to do something about the filename though Jan 18 19:49:10 PR even Jan 18 19:49:19 linux-openmoko-devel-2.6.24-rc8+git3+origin/andy-r0 Jan 18 19:49:32 otavio: more specifically: | ERROR: Cannot satisfy the following dependencies for qt4-x11-free: Jan 18 19:49:33 | libqtopengl4 qt4-examples qt4-plugins-accessible qt4-plugins-sqldrivers Jan 18 19:50:09 mickeyl: sorry ... don't get your problem Jan 18 19:50:24 SRCREV now has a strange value Jan 18 19:50:31 since the tag is not a hash Jan 18 19:51:14 * mickeyl tries something different Jan 18 19:53:16 no Jan 18 19:53:37 i don't think we can overload ud.tag Jan 18 19:53:53 this breaks PR Jan 18 19:53:58 PR? Jan 18 19:54:01 well, SRCREV Jan 18 19:54:02 otavio, pb_: for now, I'll just remove those from qt_packaging.inc so I can make progress. I don't think the build is set up for opengl, etc. and I certainly don't want qt4-examples on my handheld :) Jan 18 19:54:18 SRCREV is usually computed out of the tag or the latest revision Jan 18 19:54:30 in the branch case this does not work Jan 18 19:54:36 mickeyl: we just need a way to get the revision ... hold Jan 18 19:55:26 mickeyl: yes ... Jan 18 19:55:35 mickeyl: we need a way to get the hash for the branch tip Jan 18 19:55:41 mickeyl: so it can be use Jan 18 19:55:44 yep Jan 18 19:55:44 used Jan 18 19:56:11 how lastest_revision works there? Jan 18 19:56:14 i didn't get Jan 18 19:56:41 fetches the HEAD version using git ls-remote Jan 18 19:56:42 ahhhhhhh Jan 18 19:56:45 holds Jan 18 19:59:03 got it Jan 18 19:59:19 http://paste.debian.net/47175 Jan 18 19:59:22 mickeyl: try this way Jan 18 19:59:54 ops ... hold Jan 18 20:00:06 now you don't fetch the branch :) Jan 18 20:00:10 err Jan 18 20:00:11 wait Jan 18 20:00:16 you do Jan 18 20:00:20 since tags are branch independent Jan 18 20:00:23 which i learned ;) Jan 18 20:00:38 if branch needs to be before tag Jan 18 20:01:29 mickeyl: http://paste.debian.net/47176 should work Jan 18 20:03:13 ok, but it needs to check whether ud.branch is given Jan 18 20:03:20 mickeyl: could you try this one? Jan 18 20:03:21 mickeyl: no Jan 18 20:03:29 mickeyl: it'll be empty Jan 18 20:03:39 else: ... Jan 18 20:03:42 sorry Jan 18 20:03:44 hehe Jan 18 20:03:45 tnb: Thanks for letting me know. But It looks like you ran into a seemingly similar, but ultimately different issue. For me, all packages had been built. Jan 18 20:03:47 it won't be set at all Jan 18 20:03:54 so you need to set it in the if clause at the top Jan 18 20:03:58 if it's not given Jan 18 20:04:15 Laibsch: what are you trying? Jan 18 20:05:36 mickeyl: http://paste.debian.net/47177 Jan 18 20:05:48 mickeyl: works? Jan 18 20:05:56 let me try Jan 18 20:07:36 hmm, no Jan 18 20:07:42 :( Jan 18 20:07:45 why? Jan 18 20:07:45 File "/local/pkg/oe/bitbake/lib/bb/fetch/git.py", line 136, in _latest_revision Jan 18 20:07:45 return output.split()[0] Jan 18 20:07:45 IndexError: list index out of range Jan 18 20:08:31 how's the output of git ls remote differently when you submit a branch? Jan 18 20:09:08 otavio@neumann:~/hacking/ossystems/liboslicense$ git ls-remote . otavio/devices Jan 18 20:09:08 f00b82c4fad5de4dacf0efee2740ab5ac946ae5b refs/remotes/otavio/devices Jan 18 20:09:09 example Jan 18 20:09:49 mickeyl: please, print output Jan 18 20:11:00 have you wrote the branch name right? Jan 18 20:11:18 thesing: I have trouble building images for collie Jan 18 20:11:19 otavio@neumann:~/hacking/ossystems/liboslicense$ git ls-remote . otavio/devicesERR Jan 18 20:11:19 otavio@neumann:~/hacking/ossystems/liboslicense$ Jan 18 20:11:47 http://rafb.net/p/ajvSGt76.html Jan 18 20:12:08 aaah Jan 18 20:12:09 heh Jan 18 20:12:15 thesing: It might be that my latest changes to collie config broke PACKAGE_ARCH or something related to that Jan 18 20:12:24 with git ls-remote you are _not_ supposed to submit origin/ Jan 18 20:12:26 but just Jan 18 20:12:37 so much for consistency... Jan 18 20:13:30 Laibsch: I can still build collie images but I didn't try a clean build. Jan 18 20:14:03 mickeyl: in case of origin, yes ... Jan 18 20:14:10 mickeyl: but others you need to put Jan 18 20:14:20 mickeyl: otavio/ is a remote Jan 18 20:14:31 wow that looks good Jan 18 20:14:36 very good Jan 18 20:14:41 awesome, thanks a lot! Jan 18 20:14:43 otavio@neumann:~/hacking/ossystems/liboslicense$ git ls-remote . origin/master Jan 18 20:14:44 19648018b5f99e11c0ef9c24bc7d0c20151fa932 refs/remotes/origin/master Jan 18 20:14:45 this will be a nice addition for the git fetcher Jan 18 20:14:53 i will run some more tests Jan 18 20:14:57 to see whether we broke something Jan 18 20:14:59 mickeyl: this works here Jan 18 20:15:01 but if not, commit then Jan 18 20:15:16 mickeyl: what's the git version you're using? Jan 18 20:15:33 mickey@buildhost:/local/pkg/oe/bitbake/lib/bb/fetch$ git --version Jan 18 20:15:33 git version 1.4.4.4 Jan 18 20:15:38 ahhhhhh Jan 18 20:15:49 mickeyl: _very_ old! Jan 18 20:15:49 debian edge Jan 18 20:15:51 well Jan 18 20:15:53 debian Jan 18 20:15:54 :) Jan 18 20:16:07 mickeyl: use git 1.5 from backports.org Jan 18 20:16:12 mickeyl: muuuuch! better Jan 18 20:16:34 mickeyl: with 1.5, origin/master or origin/foo works too Jan 18 20:16:45 i see Jan 18 20:16:51 mickeyl: mine is 1.5.4-rc3 Jan 18 20:17:05 mickeyl: and you're welcome :-) Jan 18 20:17:20 ya, thanks again. pretty cool. Jan 18 20:17:24 didn't think it could be that easy Jan 18 20:17:41 tried to apply the concepts of other SMCs Jan 18 20:17:42 mickeyl: git is easy ;-) Jan 18 20:17:44 SCMs Jan 18 20:17:54 ya, it looks ok Jan 18 20:18:02 my only grief is the lack of incrementing version numbers Jan 18 20:18:03 mickeyl: no problem Jan 18 20:18:11 mickeyl: ahh Jan 18 20:18:19 that's something i like in svn Jan 18 20:18:21 mickeyl: well ... indeed but you get used to it Jan 18 20:18:24 you can easily update packages Jan 18 20:18:28 mickeyl: it's not a big deal Jan 18 20:19:26 mickeyl: but you basically have it: git-describe (again, 1.5) Jan 18 20:26:49 03mickeyl 07bitbake-1.8 * r998 10/ (ChangeLog lib/bb/fetch/git.py): add branch support to the git fetcher -- thanks Otavio Salvador. Jan 18 20:27:11 ok, enough for today. Jan 18 20:27:11 cu Jan 18 20:30:03 sirfred: oh, right, I see what you mean now. That behaviour is, unfortunately, somewhat client specific. For example, gtk does all its own compose processing and doesn't use the xim stuff from what I remember. Qt might be the same. Jan 18 20:30:40 diet-x11 does indeed disable some of the locale bits, though I didn't think it actually turned off the Compose file handling. Jan 18 20:34:19 hi lrg Jan 18 20:36:28 03mickeyl * r999 10bitbake/ (ChangeLog lib/bb/fetch/git.py): bitbake 1.9.x: add branch support to the git fetcher -- thanks Otavio Salvador. Jan 18 20:44:58 hey pb_ Jan 18 20:45:54 guys, i've finally managed to build an openmoko-image (angstrom) which actually boots on my Magician Jan 18 20:46:15 however, the touchscreen is not working :( Jan 18 21:03:09 Hi. Question about ARM linux boot sequence. After decompressing kernel, next code turning on MMU and jump to a virtual address of next function, start_kernel() end etc. I can't understand, how virtual address correspond physical address. Kernel in virtual addr - is this copy of original, or original directly visible from virtual addresses? Jan 18 21:10:06 pb_: So, it's a gtk issue? Jan 18 21:10:24 pb_: Well, at least I have a place where to start investigating. Thanks! Jan 18 21:13:24 pb_: After a fast search it seems that they are hardcoded. (https://help.ubuntu.com/community/GtkComposeTable) :-( Jan 18 21:37:00 to all, 100% we'll stay at Astrid Hotel ? Jan 18 21:42:04 damn ... no one here that come to fosdem ... Jan 18 23:11:45 ~seen koen Jan 18 23:11:47 koen was last seen on IRC in channel #oe, 36d 13h 5m 24s ago, saying: 'I forgot to commit that portion yesterday'. Jan 18 23:13:59 hvontres|work: koen is not doing irc Jan 18 23:14:09 hvontres|work: catch him via email or jabber Jan 18 23:15:51 bye **** ENDING LOGGING AT Sat Jan 19 02:59:56 2008