**** BEGIN LOGGING AT Fri Oct 17 02:59:57 2008 Oct 17 07:07:30 morning. is there an intel atom machine configuration availible yet? Oct 17 07:11:08 Hello! I have compiled an OE image for the BeagleBoard. Now, I wish to make some changes to the kernel and build a new image. Is there some way to specify to OE to build only the kernel? Oct 17 07:22:55 Hello! I have compiled an OE image for the BeagleBoard. Now, I wish to make some changes to the kernel and build a new image. Is there some way to specify to OE to build only the kernel? Oct 17 07:27:03 yes it is Oct 17 07:27:26 bitbake -c rebuuild yourkernelbitbakefile Oct 17 07:28:00 http://bec-systems.com/oe/html/usage_workwithsinglerecipe.html Oct 17 07:32:49 ok. doesnt look like thers an intel atom machine yet Oct 17 07:46:42 ok .. will try it out. Oct 17 07:47:38 Getting an error which says: Nothing PROVIDES 'org.openembedded.dev/packages/linux/linux-omap_git.bb' Oct 17 07:48:50 What might be the reason? Oct 17 07:51:43 Checked the environment var's - seem to be ok. Oct 17 08:01:01 bitbake -c rebuild virtual/kernel Oct 17 08:04:24 Sorry ... first time I am using OE - is it bitbake -c rebuild virtual/name-of-kernel-bb-file ? Oct 17 08:05:25 pce: no, it's either "bitbake -c rebuild virtual/kernel", or if that one doesn't work then "bitbake -c rebuild linux-omap" Oct 17 08:07:58 ok, will try it out! Oct 17 08:09:11 RP: around ? Oct 17 08:25:51 Hey, ist libsdl-image-1.2.6 breaking for anyone else as well? Oct 17 08:49:47 alphaone: for me too Oct 17 08:50:23 alphaone: first conftest.cpp:21:28: error: ac_nonexistent.h: No such file or directory Oct 17 08:50:42 alphaone: then cc1: internal compiler error: in add_path, at c-incpath.c:362 Oct 17 08:55:30 "refusing to pull with rebase: your working tree is not up-to-date" Oct 17 08:55:30 Any idea what I should do in this situation? Oct 17 08:56:27 I'm guessing: commit your changes to the working tree? Oct 17 08:56:45 (disclaimer: git user since 2 weeks ago :) ) Oct 17 08:57:38 valhalla: you could also stash them (git-stash) Oct 17 08:58:42 right Oct 17 09:04:01 Kensan: 'git stash' is good Oct 17 09:04:15 Kensan: the trouble is that Fedora's git does not print any error message Oct 17 09:04:45 I copied the WD to a system with a newer git to actually find out what is wrong Oct 17 09:04:46 thebohemian: oh :( Oct 17 09:05:06 anyway, I documented the workaround: http://wiki.openembedded.net/index.php/GitPhraseBook#Upgrading_your_data_.28lurking.29 Oct 17 09:05:23 thebohemian: what git version is the Fedora one? Oct 17 09:05:51 1.5.5.1 Oct 17 09:05:59 F9 Oct 17 09:06:14 thebohemian: hm ok. Oct 17 09:07:26 florian_: good morning Oct 17 09:08:26 btw, this is a nice git screencast: http://excess.org/article/2008/07/ogre-git-tutorial/ Oct 17 09:08:27 good morning Oct 17 09:24:35 /win 17 Oct 17 09:26:09 indeed Oct 17 09:26:58 hi lrg_ Oct 17 09:27:30 morning pb_ Oct 17 09:41:51 03Koen Kooi  07org.openembedded.dev * re986034ae5 10openembedded.git/packages/duke3d/ (duke3d_svn.bb files/duke3d.desktop): Oct 17 09:41:51 duke3d: add svn version from icculus.org Oct 17 09:41:51 * You'll need gamedata to play this, so download the shareware version or use your game cd Oct 17 09:52:32 03Koen Kooi  07org.openembedded.dev * r0b00aa2bda 10openembedded.git/contrib/angstrom/build-feeds.sh: angstrom feed builder: add more apps Oct 17 10:02:14 hi! Oct 17 10:02:43 does anyone know why avahi.inc forces the RDEPENDS for avahi-daemon and avahi-autoipd? Oct 17 10:03:11 this results in avahi-daemon not having its shared libs depends, and it won't run (libdaemon.so.0 not found) Oct 17 10:13:21 that sounds like a bug. Oct 17 10:13:33 the rdepends on update-rc.d should be unnecessary anyway since update-rc.d.bbclass is meant to do that. Oct 17 10:14:36 ... oh, I see, it's being done that way to avoid update-rc.d dependencies on the other subpackages. Oct 17 10:15:43 to be honest, it looks like this one recipe is trying to do too much. it would perhaps be better to break it into separate bb files for the libraries and the programs. Oct 17 10:16:12 all that said though, I think your lack of shlib rdepends is caused by something else. they ought to be additive to what's already in RDEPENDS, not suppressed by it. Oct 17 10:16:47 I wonder if the ordering of PACKAGES is causing you a problem there. I don't remember exactly how many passes package.bbclass uses for shlibs but it might be necessary to mention the library packages before the binary ones to make everything work out right. Oct 17 10:23:11 03Jeremy Lainé  07org.openembedded.dev * rad538671ab 10openembedded.git/packages/avahi/ (5 files): Oct 17 10:23:11 avahi: do not force RDEPENDS for avahi-daemon and avahi-autoipd Oct 17 10:23:11 * append to RDEPENDS instead of forcing them Oct 17 10:23:11 * fixes problem where avahi-daemon cannot find libdaemon.so.0 Oct 17 10:36:29 mickeyl: good morning Oct 17 10:36:35 sorry to see you had to sleep on the sofa again Oct 17 10:37:59 pb__: good morning. don't worry, it's just a consequence of me watching TV for too long. My marriage is still fine :D Oct 17 10:38:19 heh :-) very good. Oct 17 10:39:22 03Michael 'Mickey' Lauer  07org.openembedded.dev * ra0a28ac182 10openembedded.git/ (5 files in 2 dirs): freesmartphone.org: update revs Oct 17 10:40:01 * mickeyl just installed his new "Time Capsule" Oct 17 10:40:13 what's that? Oct 17 10:40:13 amazing product. turn on and forget Oct 17 10:40:22 some apple backup gadget Oct 17 10:40:43 basically just a harddrive + AP with some glue logic Oct 17 10:40:52 ah right Oct 17 10:40:54 * pb__ looks at the webpage Oct 17 10:41:33 yeah, looks cute Oct 17 10:41:36 i realize i could have done the same with other hardware, but I'm getting old and lazy. Turning it one and completely forgetting about it has some merits Oct 17 10:41:55 s/one/on/ Oct 17 10:42:10 indeed, life sometimes seems too short to build these things from raw materials Oct 17 10:42:41 yep. and since it comes in a shiny case it's even wife|girldfriend|etc-compatible Oct 17 10:42:41 if other people have already done the integration then I don't think there's any shame in taking advantage of their work Oct 17 10:42:49 right Oct 17 10:43:11 mickeyl: hehe *-compatible ;) Oct 17 10:43:15 yeah, even better Oct 17 10:44:48 that reminds me, I need to do something about the lousy range on our wifi ap. my wife is complaining that her laptop doesn't work at all in her studio. Oct 17 10:45:14 which is a reasonable complaint, it's only about six metres from where the AP sits. so much for 100m range. Oct 17 10:46:04 oh yeah Oct 17 10:46:07 same problem here actually Oct 17 10:46:22 i've been thinking about turning on of my spare wrt54g into a repeater Oct 17 10:46:28 but there seems to be mixed reports about such an attempt Oct 17 10:47:01 right, I was wondering about a repeater as well but I don't really know how to do that. Oct 17 10:47:20 I have a feeling that if I just move the AP to another place then it will start working for my wife but stop working for me :-} Oct 17 10:47:31 hehe, yes, that's possible Oct 17 10:47:43 the most simple thing seems to be taking two of the same and bundle them as a WDS. Only advanced firmwares support that though Oct 17 10:47:49 morning Oct 17 10:48:27 yah Oct 17 10:48:45 I wonder whether the ap that I have can do wds. Oct 17 10:48:59 it's some kind of netgear thing. I wonder if I can find the model number in my email. Oct 17 10:49:06 hi blindman Oct 17 10:50:45 if all else fails I suppose I could install a lan port in the studio for her to use, but that'd be a real pain to do. lots of digging and drilling through concrete. Oct 17 10:51:08 pb__ & mickeyl: I've heard reports that the Airport Express does do range boost seamlessly. never tried it myself though. Oct 17 10:51:40 ah right, cool Oct 17 10:52:03 the express is that little plug-top one, right? Oct 17 10:52:11 pb__: yes Oct 17 10:52:12 I get mixed up between all the different apple stuff Oct 17 10:52:32 pb__: it's the small wlan thingy with audio connectors etc. Oct 17 10:52:51 right, I know the one Oct 17 10:53:49 Kensan: hmm, right. i should try whether the time capsule does that as well. perhaps I could get away with just one after all Oct 17 10:54:56 mickeyl: here's a source http://www.neotrivium.com/blog/2007/3/22/das_tao_der_technik Oct 17 10:55:00 german only though :( Oct 17 10:55:15 03Jeremy Lainé  07org.openembedded.dev * r514f49a3f5 10openembedded.git/ (conf/checksums.ini packages/python/python-django_1.0.bb): python-django: update to django 1.0 Oct 17 10:57:00 03Jeremy Lainé  07org.openembedded.dev * r71cfc3966e 10openembedded.git/packages/python/python-django_0.96.1.bb: python-django: drop version 0.96.1 Oct 17 11:14:09 alphaone: libsdl-image 1.2.6 do_qa_configure failure http://bugs.openembedded.net/show_bug.cgi?id=4728, worked for me Oct 17 11:28:28 03Robert Schuster  07org.openembedded.dev * rc6a02df9fe 10openembedded.git/packages/libsdl/ (libsdl-image.inc libsdl-image_1.2.3.bb): libsdl-image: Moved export of SDL_CONFIG location into inc, fixes #4728. Oct 17 11:42:14 03Koen Kooi  07org.openembedded.dev * rf1201f2ad7 10openembedded.git/packages/linux/ (linux-openmoko/defconfig-oe linux-openmoko_2.6.24+git.bb): Oct 17 11:42:14 linux-openmoko 2.6.24+git: build in BT and sound modules to work around toolchain bugs Oct 17 11:42:14 * various modules fail with "unknown relocation: 40", this works around that Oct 17 11:43:36 crazy koen, he should fix his modutils Oct 17 11:44:01 reloc 40 is V4BX, there's no reason that insmod shouldn't be able to handle that. Oct 17 11:51:29 is there an automated way to add a recipes' md5 and sha256 sum to checksums.ini? Oct 17 11:52:28 not fully automated, but there is a semi-automatic process. I think it was posted on the list a few days ago. Oct 17 11:52:40 see also yesterday's irc logs for why I think checksums.ini is a bad idea :-} Oct 17 11:53:17 pb__: :) Oct 17 12:20:26 what do i have to change that bitbake is rebuilding a package? Oct 17 12:21:17 i created a package, an changed some code, now i want to bitbake it Oct 17 12:32:12 pb__: And what was conclusion of yesterday's discussion about checksums.ini? I have prepared patch for base and shorten checksums.ini (98 dupl. filename - all with same checksums :)) Oct 17 12:32:44 pb__: I would attach it to bugzilla if checksums.ini aren't dead already :). Oct 17 12:36:30 Shanuson: bitbake -c rebuild Oct 17 12:36:34 or increase PR Oct 17 12:41:44 is always "git branch" the same as git-branch ? Oct 17 12:43:35 git-branch is used in base.bbclass but on gentoo is in last version of git only "git" as command, so "git branch" is needed and maybe better if git is everywhere and git-commands only in some versions.. Oct 17 12:50:13 Shanuson: bitbake -c packagename then bitbake it again? Oct 17 12:50:49 (... oh, am I late :-) Oct 17 12:51:05 JaMa: I don't think there was any real conclusion. It might be a good idea to post to the mailing list about that. Oct 17 12:57:33 thx exco Oct 17 12:57:58 and thebohemian Oct 17 13:01:59 all right, so the default kernel config when you bitbake an image for tosa is... not for tosa. finagled one that seemed to work and was small enough. Now the installer just ran all night just putting '.' onthe screen Oct 17 13:04:49 * otavio starts a full build of OE using final GIT repository; glibc was failing Oct 17 13:12:17 where/how to i tell bitbake to include libm so i can use math.h? Oct 17 13:18:30 pb__: I just sent e-mail to ML but too late noticed that the patch is almost 1MB :( Oct 17 13:18:58 * JaMa hopes that I'll be rejected by ML Oct 17 13:19:55 morning all Oct 17 13:20:12 hi thesing Oct 17 13:23:16 03Koen Kooi  07org.openembedded.dev * rada680d9ac 10openembedded.git/packages/xorg-xserver/xorg-xserver-common.inc: xorg-xserver: change hal to RRECOMMENDS Oct 17 13:23:18 03Koen Kooi  07org.openembedded.dev * r6a9f1c3128 10openembedded.git/conf/checksums.ini: Oct 17 13:23:21 checksums: add another 4.3.2 checksum Oct 17 13:23:23 * why does it include the URL in the checks? Oct 17 13:26:43 JaMa: doh, I received your mail :-} Oct 17 13:26:47 that was indeed quite a patch Oct 17 13:26:54 heh Oct 17 13:27:30 sorry Oct 17 13:28:03 nevermind, 1MB is not really all that much in this day and age. Oct 17 13:28:11 people send me word documents several times that size all the time :-} Oct 17 13:28:32 and with fewer text in it :) Oct 17 13:28:44 heh, indeed Oct 17 13:29:55 I would have attached it to bugzilla or sent only sed replace + patch for removing duplicty, if I would noticed size :) Oct 17 13:30:44 hi Oct 17 13:31:00 hi woglinde Oct 17 13:32:02 he pb Oct 17 13:32:21 I got uclibc compile with hash-style=gnu Oct 17 13:32:35 now I am building an image Oct 17 13:32:42 which I will than test Oct 17 13:34:18 very good Oct 17 13:34:58 can someone have a look at this http://pastebin.com/d56778468 and tell me how to fix the error? Oct 17 13:35:29 exco: just git update Oct 17 13:35:52 exco: libsdl-image 1.2.6 do_qa_configure failure http://bugs.openembedded.net/show_bug.cgi?id=4728, worked for me Oct 17 13:39:23 compiling mz package i get:" whets.c:(.text+0xa00): undefined reference to `sin' " i guess bitbake doesnt use the -lm flag, but how do i tell him that he need to use it? Oct 17 13:41:58 Shanuson its not bitbakes fault Oct 17 13:42:05 its the package fault Oct 17 13:42:24 which software is this and how looks your configuration? Oct 17 13:43:12 i used the helloworld examble and added #include and some line with sin(bla) Oct 17 13:43:33 haha lol Oct 17 13:43:41 hehe Oct 17 13:43:44 now think whos error this is Oct 17 13:43:52 mine Oct 17 13:43:58 *g* :) Oct 17 13:44:05 that it was mine i knew Oct 17 13:45:02 i just dont know what to add, i guess i need some configure script or something, but never used one before Oct 17 13:45:26 either you provide a sane Makefile Oct 17 13:45:32 or configure.ac Makefile.am Oct 17 13:45:33 hm Oct 17 13:45:42 or you put all in recipe Oct 17 13:45:52 which is not the preferred way Oct 17 13:46:27 a normale makefile is ok? (like i would use to compile for host) Oct 17 13:46:41 normale Makefile is okay Oct 17 13:46:44 in do_compile Oct 17 13:46:54 bitbake will run ${MAKE} Oct 17 13:47:22 ok, will try that, makefiles i used before :) Oct 17 13:47:58 thx Oct 17 13:59:28 hi raster Oct 17 13:59:35 moop Oct 17 14:02:36 JaMa: I still run into the same error after "git pull --rebase" http://pastebin.com/d2e582a2f , so I'll have a look at the bug report you posted Oct 17 14:04:24 since I ran into some problems bitbaking in Ubuntu 8.10 - do I open bug reports on every issue - or should I just put it somewhere on the wiki? Oct 17 14:05:01 hi, does anyone know why I am getting "fatal: The remote end hung up unexpectedly" on a git pull? Oct 17 14:05:03 03Koen Kooi  07org.openembedded.dev * r16cb512d4d 10openembedded.git/conf/distro/include/angstrom.inc: angstrom: enable oabi compat for gta01 to keep gllin working, schwartz doesn't seem to work 100% Oct 17 14:05:17 exco you could ask here first Oct 17 14:05:24 otherwise put in the bugtracker Oct 17 14:05:37 03Koen Kooi  07org.openembedded.dev * r6ec5c4c167 10openembedded.git/packages/linux/linux-openmoko_2.6.24+git.bb: linux-openmoko: bump PR for OABI change Oct 17 14:06:17 woglinde: btw thanks for patching e2fsprogs for gettext :) Oct 17 14:08:11 sweetlilmre no prob but I made some mistake patching it Oct 17 14:08:19 he ph5 Oct 17 14:08:49 hej woglinde Oct 17 14:08:50 sweetlilmre|work: I ran into that problem the other day (seems like some changes on the git server?) http://wiki.openembedded.net/index.php/GitPhraseBook Oct 17 14:09:17 woglinde: np, I used the content of your patch to fix a whole load of stuff temporarily on my side :) Oct 17 14:09:49 sweetlilmre hehe Oct 17 14:10:05 you could use the new patch frome jermie lane too Oct 17 14:10:15 he sed all statements Oct 17 14:10:19 I was to lazy for it Oct 17 14:10:32 sed is some crazy stuff :) Oct 17 14:11:04 woglinde: well.. so far I needed to make these changes: http://pastebin.com/d126f59f3 Oct 17 14:12:33 sed rocks Oct 17 14:12:35 03Koen Kooi  07org.openembedded.dev * rf93dd081ef 10openembedded.git/contrib/angstrom/build-feeds.sh: angstrom feed builder: remove minimo now we have fennec Oct 17 14:12:35 like awk Oct 17 14:12:46 03Koen Kooi  07org.openembedded.dev * r38f6b8d090 10openembedded.git/: Merge branch 'org.openembedded.dev' of git@git.openembedded.net:openembedded into org.openembedded.dev Oct 17 14:12:46 03Koen Kooi  07org.openembedded.dev * r836ec75f62 10openembedded.git/: Merge branch 'org.openembedded.dev' of git@git.openembedded.net:openembedded into org.openembedded.dev Oct 17 14:12:46 03Koen Kooi  07org.openembedded.dev * r0c14e65143 10openembedded.git/: Merge branch 'org.openembedded.dev' of git@git.openembedded.net:openembedded into org.openembedded.dev Oct 17 14:13:18 exco ah right Oct 17 14:13:26 zecke showed me the fix from ubuntu Oct 17 14:13:41 on posix when you open a file for writing Oct 17 14:13:50 you have to se the filemask too Oct 17 14:13:56 I do not know Oct 17 14:14:05 why I did not enter this bug here Oct 17 14:14:58 http://archive.ubuntu.com/ubuntu/pool/main/g/gettext/gettext_0.17-3ubuntu2.diff.gz Oct 17 14:15:23 bugger. the dev tree had moved. how do I point my current repo to the new url? Oct 17 14:16:12 sweetlilmre check out new Oct 17 14:16:46 woglinde: from scratch!?! Oct 17 14:17:51 sweetlilmre yes git is fast Oct 17 14:18:18 woglinde: yes, but the rebuild (of demo image) is agonising Oct 17 14:19:36 sweetlilmre I rebuild every 2 days Oct 17 14:19:44 we have changer linker options Oct 17 14:19:47 you need to rebuild anyways Oct 17 14:19:54 to get the new goodies Oct 17 14:20:14 mickeyl: well fair enough then. Oct 17 14:20:24 woglinde: you probably have a really fast machine :) Oct 17 14:20:26 mickeyl uclibc seems to work with that too Oct 17 14:20:31 sweetlilmre nopw Oct 17 14:20:40 athlon 2400+ Oct 17 14:21:08 woglinde: ah, but you are running Linux natively, I have to do so under a VM (tho its not too bad) Oct 17 14:21:12 mickeyl do you know why a package reports ther is no checksum entry is found Oct 17 14:21:16 but the entry is in? Oct 17 14:26:38 woglinde: is it exactly the same url? Oct 17 14:27:29 pb hm I even run the source checker Oct 17 14:27:37 its linux-2.6.24 Oct 17 14:27:41 per JaMa's lengthy :-) email to the list, checksums.ini currently has to match on full pathname, not just the leafname of the file. so, if you download from a different mirror, you will lose. Oct 17 14:29:24 woglinde: (checksum) no idea Oct 17 14:29:48 woglinde: (uclibc) that reminds me... I really want to try uClibc w/ fso Oct 17 14:30:15 that would be neat Oct 17 14:30:25 have you tried building fso as thumb? Oct 17 14:44:18 thumb? Oct 17 14:44:39 (as in arm thumb instructions?) Oct 17 14:45:00 moin Oct 17 14:45:04 heya z. Oct 17 14:45:08 woglinde: are you using up-to-date git? Oct 17 14:45:55 OT: Do we/should we reach a conclusion if we want to remove the bk history? Oct 17 14:46:03 woglinde: there was an issue where some linux .bb files had double slashes in the URL so the URL *looked* like the one in checksums.ini but was not Oct 17 14:46:14 zecke: why? Oct 17 14:46:38 zecke: (as in... why should we want taht?) Oct 17 14:46:48 mickeyl: see richards mails Oct 17 14:47:00 mmh, backlog in oe Oct 17 14:47:02 k Oct 17 14:49:49 hmm, is there an ETA on the "unbroken" BKCVS conversion? Oct 17 14:50:08 mickeyl: no idea Oct 17 14:50:14 i don't care too much about the pre-mtn dates Oct 17 14:50:23 there is this grafts approach but I have no idea Oct 17 14:50:23 i would have been fine with an extra repo for that Oct 17 15:06:30 i just commented Oct 17 15:06:37 if it's just me... throw it away Oct 17 15:06:41 stuff it in another repo Oct 17 15:06:43 and forget about it Oct 17 15:06:52 (or clean up once a better way comes up)) Oct 17 15:08:42 morning all - has anyone looked at using bitbake across several machines ala' distcc? Oct 17 15:10:02 mickeyl: well, this means we need to redo the conversion and then transplant every change so far? Oct 17 15:10:10 HopsNBarley: not only looked... icecream... Oct 17 15:11:15 zecke: seems so Oct 17 15:11:22 of course, my personal take on that is forget about it Oct 17 15:11:33 but RP seems to disagree Oct 17 15:12:07 i wouldn't even mind bogus history in our main tree that pre-dates the mtn switch Oct 17 15:12:13 but again... i'm pragmatic Oct 17 15:12:43 zecke, i knew i'd come to the right place (-: Oct 17 15:14:40 mickeyl: did you volunteer to do the work? Oct 17 15:16:05 no, like i said, i'm fine with ignoring the issue Oct 17 15:16:23 let him who's not do it ) Oct 17 15:16:24 ;) Oct 17 15:16:31 (or is that too mean?) Oct 17 15:16:58 to be honest I don't even know if he has a case :} Oct 17 15:17:41 mickeyl: right, as in arm thumb Oct 17 15:17:53 zecke: hail zecke Oct 17 15:18:03 pb__: didn't do that yet. what's the best way to enable it? Oct 17 15:18:22 mickeyl: it should just be a case of adding "-mthumb" to cflags Oct 17 15:18:23 zecke: my main question is do we need to act _now_since ignoring it will make it worse or can we ignore until it _really_ bugs us? Oct 17 15:18:45 Hello folks, I've mailed a small set of patches to the mailing list for acking. They're ready for merging from our POV but would be nice if someone else could ack it there. Oct 17 15:18:48 this is the bogus mtn history, right? I can't see any reason why waiting would make it worse. Oct 17 15:19:22 we already have changes in the git tree that would need to be re-grafted onto any new import, and I don't think it makes much difference whether you have to re-graft a hundred changes or ten thousand. Oct 17 15:19:22 pb__: i think it's the bokus BK part Oct 17 15:19:31 pb__: sounds reasonable, ya Oct 17 15:19:37 oh right, yeah, sorry, bk Oct 17 15:19:38 otavio: which mailing list? Oct 17 15:19:46 mickeyl: oe-devel Oct 17 15:19:56 pb__: will require conf/machine/include/tune-thumb.inc do that? Oct 17 15:19:57 mickeyl: is it the right one, right? Oct 17 15:20:02 otavio: sure. Oct 17 15:20:09 mickeyl: also, to be honest, I can't really see why it is such a big deal to have the broken history there. if it's just duplicated commits or things being munged together that shouldn't be, this doesn't really seem like it will do any harm to just ignore. Oct 17 15:20:11 otavio: i have a slight backlog on that list Oct 17 15:20:15 mickeyl: great. could you take a look on them? Oct 17 15:20:22 sure, let me check Oct 17 15:20:55 mickeyl: yes, it looks like tune-thumb.inc will do the business if you also set ARM_INSTRUCTION_SET = "thumb" Oct 17 15:21:28 mickeyl: awesome; i'll post a 20-patches set about xfce for review too; then the only missing ones will be the 4.2 update related patches Oct 17 15:21:52 pb__: ok, cool. last question on that... why do we not build thumb as default when including tune-arm920t? Oct 17 15:22:13 otavio: cool. give us a bit time, some patches need thorough review. xfce should not be any difficult though Oct 17 15:22:13 * otavio loves to be able to push those patches upstream :-) much less work to maintain them ;-) Oct 17 15:22:22 *nod* Oct 17 15:23:10 mickeyl: all those are simply ones; no class changes (except the xfce refactoring) and then should be easy to deal with. Oct 17 15:23:49 mickeyl: thumb isn't always a win, it depends on your system architecture and your application mix. Oct 17 15:24:18 pb__: ah, that's tricky Oct 17 15:24:36 pb__: so we need to benchmark, it won't be _visible_ right? Oct 17 15:24:44 mickeyl: the general rule of thumb (ha ha) is that it gets you about a 30% reduction in code size, at the cost of about a 50% slowdown in execution speed. however, on openmoko (and arm92x in general) the cache is so small that I suspect the speed loss will be offset, and maybe even completely cancelled out, by being able to fit more instructions in cache. Oct 17 15:25:02 a Oct 17 15:25:03 h Oct 17 15:25:10 mickeyl: it probably won't be a massive step change, no. I don't think it will be like suddenly turning on a switch and having everything be twice as good Oct 17 15:25:17 so yeah, need to benchmark and see what happens Oct 17 15:25:22 cool Oct 17 15:25:26 will give that a try Oct 17 15:25:30 thanks for the pointer Oct 17 15:26:03 I'm still struggling to understand why my gta01 has such worse benchmark results than my old ipaqs did. still seeing about a 60% increase in syscall overhead which seems inexplicably bad. Oct 17 15:26:27 it does seem that openmoko is running with round robin cache replacement, which is usually a performance loss, but again the impact from that ought to be small Oct 17 15:27:39 in theory a 266MHz arm9 with 133MHz memory ought to be about as fast, or faster than, a 200MHz strongarm with 100MHz memory. strongarm does have a slightly better microarchitecture but not that much. Oct 17 15:27:42 anyone with successfull build of xserver-xorg? my is stuck on DEBUG: runstrip: skip /home/projects/OE/build.dev.gta/tmp/work/armv4t-angstrom-linux-gnueabi/xserver-xorg-1_1.3.0.0-r6.1/image/usr/lib/X11/xserver/C/print/models/PS2PDFspooldir-GS/ps2pdf_spooltodir.sh at least for half an hour.. Oct 17 15:27:47 so, all rather weird Oct 17 15:28:01 and of course, I hope that the answer will allow me to make my own arm920 platforms run faster as well :-) Oct 17 15:28:19 JaMa: heh, I don't think it ought to be trying to strip that file. Oct 17 15:28:28 I guess you have encountered a bug in strip :-} Oct 17 15:28:38 pb__: cool Oct 17 15:28:43 woglinde: can you do me a favor? Oct 17 15:28:56 pb__: sounds like a good plan Oct 17 15:29:15 pb__: i too find the Neo very slow, but always thought it was the slow flash Oct 17 15:29:55 pb__: there is _skip_ before that file.. but still hanging.. maybe looping somewhere else without debug message :) Oct 17 15:30:09 JaMa: might be worth checking with "ps" to find out what it's doing. Oct 17 15:30:17 hey guys Oct 17 15:30:34 yo kergoth Oct 17 15:31:15 can anybody help for the error "Autoreconf execution failed". I am building libxcb . Oct 17 15:33:20 pb__: there is only │ │ └─bitbake /usr/bin/bitbake -v -D xserver-xorg │ │ └─bitbake /usr/bin/bitbake -v -D xserver-xorg │ │ └─(bitbake) Oct 17 15:33:34 without other childs :/ Oct 17 15:34:26 mickeyl: also related to thumb, in case you care: that change koen made earlier to avoid type-40 relocs would be better addressed by patching whatever modutils you are using to just ignore those relocs. they are harmless but useless on your platform. Oct 17 15:34:57 or you could patch your binutils to not output them in the first place, but that seems like a worse idea. it isn't really a toolchain bug, the relocs are being generated correctly. Oct 17 15:35:18 http://page.mi.fu-berlin.de/freyther/use_open_properly.patch could anyone please apply this to gettext? Oct 17 15:35:24 pb__: is there some ready-made patch i could just use? (this is all very low level for me :) Oct 17 15:35:37 no, but I could make one for you easily enough Oct 17 15:35:44 are you using module-init-tools, or busybox insmod? Oct 17 15:36:20 "the real thing" Oct 17 15:36:24 heh Oct 17 15:36:27 okay, let me have a look Oct 17 15:36:53 cool, thanks Oct 17 15:37:26 florian: didn't you have some kind of uber-upgrade-install initramfs? Oct 17 15:38:03 oh, right, all that code is in the kernel anyway nowadays Oct 17 15:39:28 mickeyl: I think all you need to do is add "#define R_ARM_V4BX 40" to include/asm-arm/elf.h, and then add "case R_ARM_V4BX: break; /* do nothing */" to the big switch statement in arch/arm/kernel/module.c:apply_relocate() Oct 17 15:39:32 if that makes sense Oct 17 15:40:35 thesing: I have been working on such a thing but its not really generic yet. The idea is to have a variant of the oe initramfs stuff that comes with update scripts. But these would depend on the platform... Oct 17 15:41:08 pb_: wow, sounds simple enough for me Oct 17 15:41:11 thesing: It would be an option to search storage devices for the device specific update scripts of course. Oct 17 15:41:14 pb__: will do that, thanks Oct 17 15:41:59 florian: Yes for updater.sh or so. Oct 17 15:42:18 mickeyl: the upstream kernel dudes probably wouldn't accept that patch, since you need something more complex on armv4, but it should be good enough for your local tree. Oct 17 15:44:31 right Oct 17 15:52:23 re Oct 17 15:53:32 zecke I will patch gettext Oct 17 15:54:48 wahahhaahaha Oct 17 15:54:51 this is stupid Oct 17 15:55:08 checksum init yields about / vs. // Oct 17 15:56:01 thesing: right Oct 17 15:58:35 autoconf-native-2.61 does not like || make if anyone is interested Oct 17 16:07:46 uhm how can I compile mplayer with tv support? Oct 17 16:08:22 hm Oct 17 16:08:29 what you mean with tv? Oct 17 16:08:31 dvb-t? Oct 17 16:08:39 digital or analog? Oct 17 16:10:34 I need it to be able to use -tv switch and v4l2 Oct 17 16:10:47 so I can playback data from an em28xx video caputre device Oct 17 16:11:30 ah analog tv then Oct 17 16:11:32 sorry Oct 17 16:13:32 woglinde, I think it doesn't matter. just want OE to build it with tv support Oct 17 16:14:08 I ran bitbake -c patch mplayer hope I can sort it then. the answer must be somewhere in the workdir :) Oct 17 16:21:45 pb__: xserver-xorg finally finished (about 1,5 hour only in strip phase..) with lots of scanelf: ELF has multiple DT_RPATH's !? Oct 17 16:21:48 scanelf: ELF has multiple DT_RUNPATH's !? Oct 17 16:22:44 that sounds bad Oct 17 16:23:15 03woglinde  07org.openembedded.dev * r73917d04aa 10openembedded.git/: Merge branch 'org.openembedded.dev' of git@git.openembedded.net:openembedded into org.openembedded.dev Oct 17 16:23:25 03woglinde  07org.openembedded.dev * r3e8df80fdf 10openembedded.git/packages/gettext/ (3 files in 2 dirs): Oct 17 16:23:25 gettext: add patch open() mode from ubuntu Oct 17 16:23:25 * add use_open_properly.patch from ubuntu Oct 17 16:23:25 gettext_0.17-3ubuntu2.diff Oct 17 16:23:27 * bump PR on native and target Oct 17 16:23:53 is it possible to disable || make on a specific .bb file _without_ editing the package i.e. as a bitbake command? Oct 17 16:26:18 HopsNBarley: heh your nick is the name of a good chain of pubs in SA :) Oct 17 16:27:04 woglinde, the standard configuration output by the .configure script seems to enable everything I need. but then it won't compile Oct 17 16:29:29 woglinde, you know an alternative player in OE that can do what I want? Oct 17 16:39:28 03Koen Kooi  07org.openembedded.dev * r28e4d0d45d 10openembedded.git/packages/duke3d/duke3d_svn.bb: duke3d: fix DEPENDS Oct 17 16:40:46 sweetlilmre, sa = south africa? Oct 17 16:41:11 i'm glad it's a "good" chain (-: Oct 17 16:46:17 JaMa: ah, hm, that doesn't sound too good Oct 17 16:46:32 even one DT_RUNPATH is generally undesirable, multiple ones must be even worse Oct 17 16:50:49 woglinde, still around? Oct 17 16:50:56 dcordes ah sorry Oct 17 16:50:59 I was at dinner Oct 17 16:51:13 pb is it useful to install ldd and ldconfig to staging? Oct 17 16:51:36 dcordes hm I could look at the debianmultimedia package Oct 17 16:52:43 woglinde, I just ran bitbake -c devshell and ran ./configure in that terminal. now it's compiling. hope it won't error out Oct 17 16:53:17 woglinde: not really Oct 17 16:53:26 pb okay uclibc is doing it Oct 17 16:53:37 and so I will squash it Oct 17 16:53:38 woglinde: I guess cross-ldd and cross-ldconfig might have some value, but the regular ones would be useless Oct 17 16:53:41 hm to be precies Oct 17 16:53:44 even the cross ones are probably not much good Oct 17 16:53:48 the uclibc.bb makes it Oct 17 16:54:08 that is a bit odd Oct 17 16:54:16 well, if it was done deliberately, I would be inclined to leave it alone. Oct 17 16:54:21 maybe it's useful for someone, and it won't do any harm Oct 17 16:54:29 bbl Oct 17 16:55:57 pb__ hm no it is even getting installed in usr/sbin/sbin Oct 17 16:56:28 btw. wasn't there a plan to split the packages into some sections like poky did? Oct 17 16:57:05 not that I had heard of, but there certainly might have been. Oct 17 16:57:20 you mean to split up the physical location of the files in the repository? Oct 17 16:57:59 yes. To sth. like core, noncore or so. Oct 17 16:58:18 thesing hm this was just an idea Oct 17 16:58:19 So that we have a subset of packages known to work. Oct 17 16:58:33 where would one find the source that gets compiled into updater.sh? Oct 17 16:59:00 how do i keep libsdl-image's configure script from looking for /usr/bin/sdl-config? Oct 17 16:59:16 hallyn: packages/zaurus-updater Oct 17 16:59:45 echelon hm Oct 17 16:59:50 thesing: thanks Oct 17 16:59:54 I will look at it this weekend Oct 17 17:00:11 using the dev branch Oct 17 17:02:34 is /linuxrc in rootfs normal? Oct 17 17:03:48 woglinde, that didn't enable tv support Oct 17 17:05:59 thesing: depends on your platform and distro. it's certainly not abnormal in itself. Oct 17 17:06:58 pb_: its in a console-image for collie and angstrom. I don't think that it used to be there. Oct 17 17:08:35 will "bitbake pname" touch configuration files after the source was manually configured with "bitbake -c devshell pname" ? Oct 17 17:09:25 thesing: might be best to check with #angstrom, see if anybody knows where it came from. Oct 17 17:10:36 pb_: in #angstrom are only users. I think only koen knows what goes in angstrom and he is not on irc. Oct 17 17:11:03 03Otavio Salvador  07org.openembedded.dev * rc8f54eb5d1 10openembedded.git/classes/xfce.bbclass: Oct 17 17:11:03 XFCE: allow to set a specific release to be used while grabbing the source Oct 17 17:11:03 This will be used to avoid duplicated code in many XFCE related Oct 17 17:11:03 recipes. Without that, all recipes that uses a different version then Oct 17 17:11:05 the global XFCE release need to specify the SRC_URI making harder to Oct 17 17:11:07 update and change the mirror to be used globaly. Oct 17 17:11:09 03Otavio Salvador  07org.openembedded.dev * rff3ecd7691 10openembedded.git/packages/xfce/xfce-mcs-manager.inc: Oct 17 17:11:12 xfce-mcs-manager.inc: fix .pc file to not have tmpdir on includedir field Oct 17 17:11:14 The includedir field was left with full path and this made the package Oct 17 17:11:16 to fail in sanity checks. This patch changes it to point to Oct 17 17:11:18 /usr/include as all other package does. Oct 17 17:11:20 03Otavio Salvador  07org.openembedded.dev * r5c4c08ba74 10openembedded.git/packages/xfce/gtk-xfce-engine_2.4.1.bb: gtk-xfce-engine: specify the required XFCE release Oct 17 17:11:23 03Otavio Salvador  07org.openembedded.dev * reaf2c4d027 10openembedded.git/packages/xfce/mousepad_0.2.12.bb: mousepad: specify the required XFCE release Oct 17 17:11:30 thesing: ah, I see Oct 17 17:11:34 03Otavio Salvador  07org.openembedded.dev * r2907b81706 10openembedded.git/packages/xfce/exo_0.3.2.bb: exo: specify the required XFCE release Oct 17 17:11:37 03Otavio Salvador  07org.openembedded.dev * rb5ab8f1745 10openembedded.git/packages/xfce/ (xfwm4.inc xfwm4_4.4.1.bb): xfwm4: use a single recipe since we don't have multiple versions packaged Oct 17 17:11:40 03Otavio Salvador  07org.openembedded.dev * r70463fe4fb 10openembedded.git/packages/xfce/ (xfce4-mixer.inc xfce4-mixer_4.4.1.bb): xfce4-mixer: use a single recipe since we don't have multiple versions packaged Oct 17 17:11:44 03Otavio Salvador  07org.openembedded.dev * rc23266091f 10openembedded.git/packages/xfce/ (xfce4-panel.inc xfce4-panel_4.4.1.bb): xfce4-panel: use a single recipe since we don't have multiple versions packaged Oct 17 17:11:50 03Otavio Salvador  07org.openembedded.dev * rc11c3affd3 10openembedded.git/packages/xfce/ (xfwm4-themes.inc xfwm4-themes_4.4.1.bb): xfwm4-themes: use a single recipe since we don't have multiple versions packaged Oct 17 17:11:50 :D Oct 17 17:11:58 03Otavio Salvador  07org.openembedded.dev * r8b559e8d48 10openembedded.git/packages/xfce/ (xfce4-appfinder.inc xfce4-appfinder_4.4.1.bb): xfce4-appfinder: use a single recipe since we don't have multiple versions packaged Oct 17 17:12:02 03Otavio Salvador  07org.openembedded.dev * r169bdf1e27 10openembedded.git/packages/xfce/ (libxfce4util.inc libxfce4util_4.4.1.bb): libxfce4util: use a single recipe since we don't have multiple versions packaged Oct 17 17:12:06 03Otavio Salvador  07org.openembedded.dev * r8995500351 10openembedded.git/packages/xfce/ (xfce-mcs-plugins.inc xfce-mcs-plugins_4.4.1.bb): xfce-mcs-plugins: use a single recipe since we don't have multiple versions packaged Oct 17 17:12:09 Koen acked them privately; that's why I've pushed them Oct 17 17:12:12 03Otavio Salvador  07org.openembedded.dev * r31c4bfb152 10openembedded.git/packages/xfce/ (libxfce4mcs.inc libxfce4mcs_4.4.1.bb): libxfce4mcs: use a single recipe since we don't have multiple versions packaged Oct 17 17:12:18 03Otavio Salvador  07org.openembedded.dev * r9f5ffb5475 10openembedded.git/packages/xfce/libxfce4util_4.4.1.bb: libxfce4util: readd pkgconfig class to fix stagging Oct 17 17:12:21 03Otavio Salvador  07org.openembedded.dev * r63fe642a20 10openembedded.git/packages/xfce/xfce4-dev-tools_4.4.0.bb: xfce4-dev-tools: update to 4.4.0.1 Oct 17 17:12:24 03Otavio Salvador  07org.openembedded.dev * r759bcabde9 10openembedded.git/packages/xfce/ (exo/configure.patch exo_0.3.2.bb): Oct 17 17:12:27 exo: update to 0.3.4 Oct 17 17:12:29 This release needs a configure patch to avoid testing for strftime Oct 17 17:12:33 modifiers that can't be ran while cross compiling. Oct 17 17:12:35 03Otavio Salvador  07org.openembedded.dev * re99596398f 10openembedded.git/packages/xfce/gtk-xfce-engine_2.4.1.bb: gtk-xfce-engine: update to 2.4.2 Oct 17 17:12:40 03Otavio Salvador  07org.openembedded.dev * rf7f51b0f44 10openembedded.git/packages/xfce/xfwm4-themes_4.4.1.bb: xfwm4-themes: update to 4.4.2 Oct 17 17:12:45 03Otavio Salvador  07org.openembedded.dev * rf61b447ce8 10openembedded.git/packages/xfce/xfce-mcs-manager_4.4.1.bb: xfce-mcs-manager: use a single recipe since we don't have multiple versions packaged Oct 17 17:12:49 03Otavio Salvador  07org.openembedded.dev * r500886ae9c 10openembedded.git/packages/xfce/xfce-utils_4.4.1.bb: xfce-utils: update to 4.4.2 Oct 17 17:12:52 03Otavio Salvador  07org.openembedded.dev * r79f6a1fdd8 10openembedded.git/packages/xfce/xfce-mcs-plugins_4.4.1.bb: xfce-mcs-plugins: update to 4.4.2 Oct 17 17:12:57 03Otavio Salvador  07org.openembedded.dev * re728bcbf4d 10openembedded.git/packages/xfce/xfce4-appfinder_4.4.1.bb: xfce4-appfinder: update to 4.4.2 Oct 17 17:13:00 (32 lines omitted) Oct 17 17:22:25 he robert Oct 17 17:22:57 03Otavio Salvador  07org.openembedded.dev * r28583f2a7b 10openembedded.git/packages/xfce/libxfce4mcs_4.4.1.bb: libxfce4mcs: update to 4.4.2 Oct 17 17:24:34 ok. /linuxrc comes from the new busybox version Oct 17 17:45:29 that is a bit weird. why does busybox install /linuxrc? Oct 17 17:46:41 probably some feature to make it ready for use in initramfs Oct 17 17:47:08 not like its that hard to create a linuxrc.. Oct 17 17:55:54 bye Oct 17 17:56:00 I am moving now to zecke Oct 17 18:11:34 03Koen Kooi  07org.openembedded.dev * r96b3f1d2e4 10openembedded.git/contrib/angstrom/build-feeds.sh: angstrom feed builder: add xfce stuff Oct 17 18:41:59 I'm getting a failure on python-pygobject-native: http://pastebin.com/d75dc758d any ideas? Oct 17 18:50:18 03Mike Westerhof  07org.openembedded.dev * r317c1e6932 10openembedded.git/conf/checksums.ini: checksums.ini: added apex 1.5.14 Oct 17 19:02:53 re Oct 17 19:30:03 03Jeremy Lainé  07org.openembedded.dev * rdb23121a07 10openembedded.git/packages/boost/ (boost-36.inc boost_1.36.0.bb): Oct 17 19:30:03 boost-1.36: build missing boost libraries Oct 17 19:30:03 * pass --with-LIBRARY to jam for each boost library Oct 17 19:30:03 * align package names with their Debian counterparts Oct 17 19:50:46 03Phil Blundell  07org.openembedded.dev * rbfd07fa2e8 10openembedded.git/packages/lirc/lirc_0.8.1+cvs20070307.bb: lirc: move .la files and other development things to -dev package Oct 17 19:50:57 03Phil Blundell  07org.openembedded.dev * r59aa6dc63c 10openembedded.git/: Merge branch 'org.openembedded.dev' of git@git.openembedded.net:openembedded into org.openembedded.dev Oct 17 19:52:36 03Phil Blundell  07org.openembedded.dev * re0eabc7eb8 10openembedded.git/classes/rootfs_ipk.bbclass: Oct 17 19:52:36 rootfs_ipk: add support for BAD_RECOMMENDATIONS and Oct 17 19:52:36 ONLINE_PACKAGE_MANAGEMENT configuration settings Oct 17 20:10:27 is OE packaged with any WM that doesn't require a mouse? Oct 17 20:12:38 echelon: imho matchbox can work without a mouse pretty well Oct 17 20:12:56 oh Oct 17 20:13:04 it does tiling? Oct 17 20:18:01 hi all Oct 17 20:18:34 what's the size of matchbox Oct 17 20:19:33 i starting just now use git repository, i would have two local dirs of metadata, stable and dev. What's the better solution clone the two branches separately or what ? Oct 17 20:20:06 stable Oct 17 20:20:16 echelon: no, it stacks windows at full size Oct 17 20:20:48 echelon: its quite small Oct 17 20:21:01 awesomewm only takes up 114kb Oct 17 20:22:26 gremlin[it]: if you want to base a stable 'product' on it use stable. if its for something you want to merge back upstream use dev Oct 17 20:24:35 flo_lap .. i need both i have some project based on stable some other (and maybe contribute too) based on dev ... with monotone i have two completely separate directory I take in sync separately too ... Oct 17 20:25:09 flo_lap: that's not what he's asking. he wants to know if they should be seperate local git clones. he can either do that, clone both from the remote or clone one to the other, or he can use git-new-workdir from the contrib dir in the git sources to make a second or more (symlinked) workdirs from the single .git dir (only safe if each dir is a seperate branch) Oct 17 20:25:54 hmm Oct 17 20:25:56 Is first time I use git (more than just checkout) so i'm figuring out if i can continue working in the same way as i did with monotone or there is a better way Oct 17 20:26:14 kergoth you partially answer me ... Oct 17 20:26:29 partially? Oct 17 20:28:01 yes ... seem last way is better (git-new-workdir) ... is what you suggest to do ? Oct 17 20:28:14 kergoth: ah ok, didn't get that Oct 17 20:28:22 :) Oct 17 20:29:06 gremlin[it]: i find that way easier to maintain, personally, since there's less pushing and pulling involved Oct 17 20:29:09 why not use branches? Oct 17 20:29:11 but either works, up to you Oct 17 20:29:23 Kensan: those are branches... hes' asking whether he should ahve as eperate clone for each branch Oct 17 20:29:30 I believe stable and dev are branches and in the same repos Oct 17 20:29:36 yes, and? Oct 17 20:29:49 you can't have two branches checked out in the same working copy Oct 17 20:29:53 he wants both available Oct 17 20:29:56 so what's the benefit of clonig the repos twice? Oct 17 20:30:02 so he has two working copies, one with each. Oct 17 20:30:11 then he doesnt have to check out repeatedly to switch back and forth Oct 17 20:30:15 ic Oct 17 20:30:33 gremlin[it]: I have set up now like with mtn Oct 17 20:30:42 kergoth ... mhhh or you are able to decode my bad English or you are the only other who use both stable and dev ;) Oct 17 20:30:48 gremlin[it]: I have two branches in separate dirs Oct 17 20:30:49 hehe Oct 17 20:30:58 i don't, in oe, but i have a number of working copies of the kernel Oct 17 20:31:17 i have one kernel git repo with a ton of remotes and then a number of working copies from it Oct 17 20:31:47 gremlin[it]: I cat provide small howto if you wanted :) Oct 17 20:32:25 Jay7 : one repo and two remote ? Oct 17 20:33:01 gremlin[it]: just two dirs with two repos (.git). Each repo traking one remote branch (stable and dev) Oct 17 20:33:08 kergoth : can you point me what git command is the key of your repo structure ? Oct 17 20:33:09 *only one Oct 17 20:33:18 seems working :) Oct 17 20:33:49 Jay7 : this way is simple ... just clone twice with different branch ... Oct 17 20:34:06 git-init && git-remote && git-pull Oct 17 20:34:43 I have considered about mtn-like schema with only one $GIT_DIR Oct 17 20:35:01 but this seems like not a true-git-way Oct 17 20:39:46 03Otavio Salvador  07org.openembedded.dev * r5e2aa7cd1f 10openembedded.git/packages/xfce/thunar_0.9.0.bb: Oct 17 20:39:46 thunar: properly install libraries into the staging area Oct 17 20:39:46 xfdesktop needs thunar-vfs to be able to show the created icons on the Oct 17 20:39:46 Desktop area and for it to work thunar needs to properly install the Oct 17 20:39:46 required headers on staging. Oct 17 20:40:58 03Otavio Salvador  07org.openembedded.dev * rc4b6c7940b 10openembedded.git/conf/checksums.ini: checksums.ini: add thunar 0.9.0 Oct 17 20:40:59 03Otavio Salvador  07org.openembedded.dev * rad1d933314 10openembedded.git/conf/checksums.ini: checksums.ini: add libpam 1.0.2 Oct 17 20:40:59 03Otavio Salvador  07org.openembedded.dev * r6a987cb2fa 10openembedded.git/conf/checksums.ini: checksums.ini: add xfdesktop 4.4.2 Oct 17 20:53:20 http://tipd.com/Stocks/what-really-happened-in-the-stock-market-comic/ Oct 17 20:55:15 kergoth ... how is the mood in US on the streets ? you are in south-west right ? Oct 17 21:25:32 gremlin[it]: what is the sa1100 usb gadget driver doing? Oct 17 21:26:56 thesing you mean the state of developing or what is used for ? Oct 17 21:28:05 gremlin[it]: the first. I pointed you at the balloon driver and you wanted to look at it. Oct 17 21:28:30 ghhh time ... i neeeed time ..... Oct 17 21:29:05 did you try it at all? Oct 17 21:29:17 no nothing ... Oct 17 21:29:55 ok. Oct 17 21:30:06 sorry Oct 17 22:25:21 ~seen RP Oct 17 22:25:24 rp is currently on #openmoko (2d 10h 31m 55s), last said: 'mmp: yes'. Oct 17 22:31:51 quick question - what's now the right way to specify additional packages to be included in the FS image? Oct 17 22:35:27 denix: create a new image. Oct 17 22:36:34 denix: there is ANGSTROM_EXTRA_INSTALL for some images as a dirty way. Oct 17 22:38:11 khem: RP is in here. (in case you didn't see him) Oct 17 22:47:16 thesing: so, the right way is to create a new image/task recipe... what about MACHINE_ESSENTIAL_EXTRA_RDEPENDS in minimal-image? Oct 17 22:48:59 denix: you could add stuff to it. but its only for machine specific stuff. Oct 17 22:49:15 denix: other images use it through task-base (I think) Oct 17 22:49:36 when i try to build cacao-initial i get this error... Oct 17 22:49:39 cp -fpPR /tmp/ZipitZ2/tmp-zipit2-angstrom/work/i686-linux/cacao-initial-0.98-r0/temp-staging//tmp/ZipitZ2/tmp-zipit2-angstrom/staging/i686-linux/lib/* /tmp/ZipitZ2/tmp-zipit2-angstrom/staging/i686-linux/lib Oct 17 22:49:45 cp: cannot stat `/tmp/ZipitZ2/tmp-zipit2-angstrom/work/i686-linux/cacao-initial-0.98-r0/temp-staging//tmp/ZipitZ2/tmp-zipit2-angstrom/staging/i686-linux/lib/*': No such file or directory Oct 17 22:50:02 thesing: sorry, you are right, it is in task-boot Oct 17 22:50:08 obviously /tmp/ZipitZ2/tmp-zipit2-angstrom/work/i686-linux/cacao-initial-0.98-r0/temp-staging//tmp/ZipitZ2/tmp-zipit2-angstrom/staging/i686-linux/lib/* isn't a directory, so where can i fix this? Oct 17 22:50:20 denix: np Oct 17 22:52:00 thesing: anyway, I was using this EXTRA_RDEPENDS from task-boot as a quick way of addind extra packages to the image, but now it's seems broken Oct 17 22:52:37 thesing: specifically, due to the shared libs packaged in own lib.ipk files... Oct 17 22:53:46 anyone? Oct 17 23:03:01 echelon: that might be a question for rschuster Oct 17 23:03:22 echelon: or your disk is full Oct 17 23:20:53 so, will RDEPENDS="zlib" work? Oct 17 23:22:02 flo_lap: no.. plenty of space left Oct 17 23:22:25 the problem is that it's trying to copy over the files to a non-existant directory Oct 17 23:26:59 can anybody help me with library dependencies? please :) Oct 17 23:29:46 denix: that shouldn't be necessary, runtime library dependencies are set automatically Oct 17 23:30:33 flo_lap: that's the problem - I need zlib in the filesystem, but nothing YET requires it Oct 17 23:31:18 denix: ah right, then you can do this in the image description like with the task packages. Oct 17 23:31:43 flo_lap: that's where I'm getting problems Oct 17 23:33:11 flo_lap: if I have zlib in RDEPENDS and try building minimal-image, it gets into IMAGE_INSTALL and complains about "Cannot satisfy the following dependencies for task-boot" Oct 17 23:34:18 denix: ah right, you ned to have the name of the resulting package in RDEPENDS, not the bb name like in DEPENDS Oct 17 23:34:49 it works with ncurses, as it produces ncurses.ipk, but doesn't work with zlib, as it produces libz1.ipk Oct 17 23:34:57 03Florian Boor  07org.openembedded.dev * r4c1483de8f 10openembedded.git/packages/unicap/ucview_0.22.bb: ucview 0.9.3: add image capture tool ucview Oct 17 23:34:58 03Florian Boor  07org.openembedded.dev * ra55c96741e 10openembedded.git/packages/unicap/ (unicap-0.9.3/pkgconfig.patch unicap_0.9.3.bb): unicap 0.9.3: video data api framework Oct 17 23:35:38 * flo_lap now speaks git :-) Oct 17 23:35:57 well... somehow ;) Oct 17 23:36:49 flo_lap: so, when I specify libz in RDEPENDS, it won't build it anymore, as there is no such .bb recipe... Oct 17 23:37:23 do I need to specify both DEPENDS="zlib" to build and RDEPENDS="libz/libz1" to include in the filesystem? Oct 17 23:37:50 03Jeremy Lainé  07org.openembedded.dev * r748fa1c06f 10openembedded.git/ (3 files in 3 dirs): wt: add wt 2.2.0, a C++ Web Toolkit Oct 17 23:37:58 denix: RDEPENDS = libz1 Oct 17 23:38:07 yes, if nothing DEPENDS on zlib you won't get this package Oct 17 23:42:17 flo_lap, thesing: that I understand so far :) I used MACHINE_ESSENTIAL_EXTRA_RDEPENDS from task-boot/minimal-image before and it worked, but not anymore. Oct 17 23:43:57 what is the error message? Oct 17 23:44:06 that allowed adding extra packages to RDEPENDS, but now I also need to add to DEPENDS, which is not available. Or is it a wrong way of doing it and I should just make my own task/image? Oct 17 23:44:53 thesing: the message is a dreaded "* ERROR: Cannot satisfy the following dependencies for task-boot: * zlib" Oct 17 23:47:08 thesing: which is based on what we just discussed is correct, RDEPENDS should have libz1 instead of zlib... And zlib should go into DEPENDS, which is not easily accessible from local.conf Oct 17 23:47:45 denix: I would grep for sth that RDEPENDS on zlib Oct 17 23:50:19 good night Oct 17 23:57:20 thesing: no working .bb which RDEPENDS on zlib or any other library I need (libpng, tslib, jpeg, freetype) Oct 17 23:58:28 thesing: looks like complete reliance on libs being picked up automatically Oct 17 23:59:32 does the machine-rdepends with libz1 work? Oct 17 23:59:51 thesing: the way task-base includes tslib and its libts.so, is to RDEPENDS on tslib-calibrate, which uses libts.so Oct 18 00:00:12 (I think we switched to debian package naming some time ago) Oct 18 00:00:31 JaMa: the patch from http://bugs.openembedded.net/show_bug.cgi?id=4728 got applied, but buiding libsdl-image still breaks with the same error Oct 18 00:00:45 Hello Oct 18 00:00:49 Is Otavio there? Oct 18 00:02:53 thesing: cannot try machine-rdepends yet, as I'm trying to find a simple way to add zlib to DEPENDS... Oct 18 00:03:24 denix: the simplest way is to create a new image. Oct 18 00:03:38 Otavio: it seems that xfce4-dev-tools checksum is missing in conf file... Oct 18 00:05:02 thesing: yeah, I know ;) I was trying to avoid it, but at this point it is probably easier... Oct 18 00:05:39 it is easy. copy an image and add additional packages to IMAGE_INSTALL Oct 18 00:06:56 03Thomas Kunze  07org.openembedded.dev * rc437c4dab4 10openembedded.git/packages/base-files/ (base-files/collie/fstab base-files_3.0.14.bb): base-files: remove custom fstab file for collie Oct 18 00:06:57 03Thomas Kunze  07org.openembedded.dev * r40c92bda9d 10openembedded.git/packages/pointercal/ (files/collie/pointercal pointercal_0.0.bb): pointercal: add file for collie Oct 18 00:06:57 03Thomas Kunze  07org.openembedded.dev * r869781d052 10openembedded.git/packages/linux/ (4 files in 2 dirs): Oct 18 00:06:58 linux-rp: update collie patches Oct 18 00:07:00 * touchscreen works now Oct 18 00:07:02 * use old hacked sharp flash driver Oct 18 00:07:04 03Thomas Kunze  07org.openembedded.dev * r958429dd9f 10openembedded.git/packages/kexecboot/ (3 files in 3 dirs): linux-kexecboot: use hacked sharp-flash driver Oct 18 00:07:28 wow git is really fast with pushing Oct 18 00:08:23 thesing: yes, I know how to do it. I tried to avoid it as there was way to do the same through local.conf Oct 18 00:08:35 thesing: up until it stopped working :) Oct 18 00:09:49 thesing: wow, thanks for touchscreen on collie! Oct 18 00:10:41 the driver was already there. But the chip wasn't enabled. Oct 18 00:10:49 denix: do you have a collie? Oct 18 00:11:20 btw, RDEPENDS="libz1" gives this: ERROR: '['/OE/packages/tasks/task-boot.bb']' RDEPENDS/RRECOMMENDS or otherwise requires the runtime entity 'libz1' but it wasn't found in any PACKAGE or RPROVIDES variables Oct 18 00:13:10 thesing: I was supposed to start testing your collie images, but haven't had time yet Oct 18 00:14:36 thesing: any special way to format an SD card for your collie images? Oct 18 00:15:02 I tested only ext2 Oct 18 00:15:52 do I need 2 ext2 partitions? or zImage can be in the root fs? Oct 18 00:19:33 you only need one. the kernel is in rootfs. Oct 18 00:20:14 collie uses kexecboot a LAB implementation now. Oct 18 00:20:45 gregoiregentil: I'll add the missing checksums Oct 18 00:21:02 zImage-kexecboot-collie.bin is an actual zImage, right? Oct 18 00:21:25 its a kernel with the bootloader. Oct 18 00:21:38 you need to flash it normally Oct 18 00:21:44 (reset+c+d Oct 18 00:22:31 that would reflash NOR? Oct 18 00:22:36 otavio: actually there is one more missing Oct 18 00:22:51 libxfce4-utils...something Oct 18 00:23:10 denix: yes. but only the kernel partition. Oct 18 00:23:35 thesing: any way to recover the original one later? Oct 18 00:24:04 just flash the original zImage later. Oct 18 00:24:58 do you have openzaurus installed? Oct 18 00:25:11 nope, Sharp ROM Oct 18 00:25:43 not sure if it works then. You might need to wipe the flash. Oct 18 00:26:14 you could try with a new linux-kexecboot kernel Oct 18 00:26:58 you can unpack the sharprom kernel from the Ospack file. Oct 18 00:27:22 thesing: what's missing now from the device driver perspective? Oct 18 00:28:27 audio and usb-gadget and PM. PM is should be ready soon. Oct 18 00:29:00 for usb-gadged there is a driver that worked when I tried last. Oct 18 00:30:31 CF should not be a problem, will CF WiFi work? Oct 18 00:31:28 yes. Oct 18 00:31:52 what about SDIO? Oct 18 00:32:32 Should not be too hard to get working. But I don't have any sdio card. Oct 18 00:35:17 Ospack - where I get that? was it on Sharp CD? Oct 18 00:35:41 you can get it on sharp hp Oct 18 00:36:11 http://www.myzaurus.com/ROMupdate4.asp Oct 18 00:36:41 oh, thanks Oct 18 00:37:11 can't believe it's over 6 yo :) Oct 18 00:42:25 do you have a procedure for your imgaes detailed somewhere? the readme file is very basic... Oct 18 00:43:42 03Thomas Kunze  07org.openembedded.dev * r5dad38b25b 10openembedded.git/conf/checksums.ini: checksums.ini: add kexecboot Oct 18 00:43:59 what do you mean? how to build them? Oct 18 00:44:50 or install instructions? Oct 18 00:45:09 install instructions Oct 18 00:46:16 I'll write some. Oct 18 00:46:37 thanks a lot! Oct 18 00:47:08 I sound like a noob, but I never flashed my Zaurus - had too many other devices to flash at work :) Oct 18 00:53:00 so wrote something more. Oct 18 00:54:33 denix: see http://wiki.openzaurus.org/Collie/Install for flashing instructions. Oct 18 00:55:19 btw. you can go back to sharprom if you put the ospack file on a cf and flash it. Oct 18 00:56:49 ok, I'll give it a try Oct 18 00:57:52 is OZ based on 2.4? Oct 18 00:58:20 thesing: git access is working for you ? Oct 18 00:58:29 yes. works fine. Oct 18 00:58:38 hmmm not for me Oct 18 00:59:11 * otavio is finishingto add the XFCE checksums Oct 18 00:59:40 I cant even git clone Oct 18 00:59:42 denix: yes. At http://oz.hentges.net/releases/3.5.4.1/T1/collie/ are the best versions. Oct 18 01:00:57 khem: works without any problem. You have the private ssh key on the pc right? Oct 18 01:01:28 thesing: yes same works for other open source projects so I dont think there is any issue on my end. Oct 18 01:03:29 thesing: I dont remember if I sent the right pub key\ Oct 18 01:04:40 khem: you have two keys there. Oct 18 01:04:59 one is rsa and another one dsa :) Oct 18 01:05:16 may be thats the problem Oct 18 01:06:03 I don't even know who is responsible for git. Oct 18 01:06:21 mickey|zzZZzz is zzzzzz Oct 18 01:10:33 thesing: the new image.bb does not work either - if I have zlib in IMAGE_INSTALL, it complains there is no such package, if I have libz1 in IMAGE_INSTALL, bitbake complains there is no such buildable target Oct 18 01:14:53 03Otavio Salvador  07org.openembedded.dev * r7788c6a5d4 10openembedded.git/packages/xfce/xfdesktop_4.4.2.bb: xfdesktop: depends on thunar to enable support for desktop icons Oct 18 01:14:54 03Otavio Salvador  07org.openembedded.dev * r1b37391a88 10openembedded.git/conf/checksums.ini: checksums.ini: add exo 0.3.4 Oct 18 01:14:54 03Otavio Salvador  07org.openembedded.dev * r420b6893be 10openembedded.git/conf/checksums.ini: checksums.ini: add gtk-xfce-engine 2.4.2 Oct 18 01:14:55 03Otavio Salvador  07org.openembedded.dev * r71f7348dd9 10openembedded.git/conf/checksums.ini: checksums.ini: add libxfce4mcs 4.4.2 Oct 18 01:14:57 03Otavio Salvador  07org.openembedded.dev * r4f08b156a7 10openembedded.git/conf/checksums.ini: checksums.ini: add libxfce4util 4.4.2 Oct 18 01:15:00 03Otavio Salvador  07org.openembedded.dev * r1e5e6c46ea 10openembedded.git/conf/checksums.ini: checksums.ini: add xfce4-dev-tools 4.4.0.1 Oct 18 01:15:03 03Otavio Salvador  07org.openembedded.dev * r5138a77083 10openembedded.git/conf/checksums.ini: checksums.ini: add libxfcegui4 4.4.2 Oct 18 01:15:06 03Otavio Salvador  07org.openembedded.dev * r37d4986734 10openembedded.git/conf/checksums.ini: checksums.ini: add xfce4-icon-theme 4.4.2 Oct 18 01:15:09 03Otavio Salvador  07org.openembedded.dev * r2c18f41640 10openembedded.git/conf/checksums.ini: checksums.ini: add xfce4-session 4.4.2 Oct 18 01:15:14 03Otavio Salvador  07org.openembedded.dev * r791d2d0491 10openembedded.git/conf/checksums.ini: checksums.ini: add xfwm4-themes 4.4.2 Oct 18 01:15:17 03Otavio Salvador  07org.openembedded.dev * rbf4c7ff81d 10openembedded.git/conf/checksums.ini: checksums.ini: add xfce-mcs-manager 4.4.2 Oct 18 01:15:20 03Otavio Salvador  07org.openembedded.dev * r0908a5429a 10openembedded.git/conf/checksums.ini: checksums.ini: add orage 4.4.2 Oct 18 01:15:24 gregoiregentil: solved; i think i've added all missed checksums regarting xfce Oct 18 01:15:25 03Otavio Salvador  07org.openembedded.dev * r42a9d8242a 10openembedded.git/conf/checksums.ini: checksums.ini: add xfce4-appfinder 4.4.2 Oct 18 01:15:28 03Otavio Salvador  07org.openembedded.dev * r5f2e800061 10openembedded.git/conf/checksums.ini: checksums.ini: add xfce4-mixer 4.4.2 Oct 18 01:15:31 03Otavio Salvador  07org.openembedded.dev * r3821e1b6d6 10openembedded.git/conf/checksums.ini: checksums.ini: add xfprint 4.4.2 Oct 18 01:15:36 03Otavio Salvador  07org.openembedded.dev * r8a8b914155 10openembedded.git/conf/checksums.ini: checksums.ini: add xfce4-panel 4.4.2 Oct 18 01:15:39 03Otavio Salvador  07org.openembedded.dev * re505c9b6d6 10openembedded.git/conf/checksums.ini: checksums.ini: add mousepad 0.2.13 Oct 18 01:16:52 thesing: anyway thanks for all your help, I'll keep trying... Oct 18 01:17:20 I think it works here. Oct 18 01:18:16 denix: http://www.pastebin.ca/1229844 Oct 18 01:19:16 thesing: that's strange, as nothing's changed except updating to the tip of .dev branch couple of weeks ago Oct 18 01:20:01 do you see the actual libz.so in the rootfs? Oct 18 01:20:10 yes Oct 18 01:21:57 hmm, that's strange... Oct 18 01:23:11 do you have bitbake 1.8.10 ? Oct 18 01:23:25 I do have a complex setup with external-toolchain/SDK, but it worked until I updated to the tip of oe-dev... Oct 18 01:23:50 1.8.11 Oct 18 01:24:08 ok. then its probably you toolchain stuff. Oct 18 01:24:28 denix: what problem is it something wrong in toolchain ? Oct 18 01:24:38 did you try a full rebuild after you updated ? Oct 18 01:25:10 thesing: clean rebuild won't help Oct 18 01:25:30 and you can't get rid of the external toolchain? Oct 18 01:25:31 thesing: not sure how toolchain may affect DEPENDS/RDEPENDS issue Oct 18 01:25:50 yes it sounds unlikely. Oct 18 01:25:56 thesing: I can definitely try it Oct 18 01:26:47 maybe its the bitbake version. or .dev was broken when you checked out. Oct 18 01:27:34 anyway. time for some sleep. Oct 18 01:27:48 night all. Oct 18 01:28:17 khem: no, toolchain is fine. I'm having problems installing zlib into the rootfs Oct 18 01:31:43 IMAGE_INSTALL += in your image file Oct 18 01:31:49 doesnt work ? Oct 18 01:32:03 khem: been there, won't work Oct 18 01:32:20 complains no such package zlib Oct 18 01:32:23 hmm what distro ? Oct 18 01:32:32 and it's true - it is called libz1.ipk :) Oct 18 01:32:55 it's customized/modified angstrom Oct 18 01:33:30 the interesting thing is that it used to work, until I updated oe-dev to the latest couple week ago... :) Oct 18 01:36:02 here is the command it runs: opkg-cl -f /oe/angstrom-tmp/staging/i686-linux/etc/opkg.conf -o /oe/angstrom-tmp/rootfs install task-boot util-linux-mount util-linux-umount angstrom-version zlib Oct 18 01:36:28 and here is the error message at the end: Collected errors: * Cannot find package zlib. Oct 18 01:37:02 denix can you add it to DEPENDS Oct 18 01:37:26 it builds, but won't install in rootfs Oct 18 01:39:16 what happens if you add it to RDEPENDS Oct 18 01:40:01 then it says no such package zlib Oct 18 01:44:25 it's probably due to one of my custom configurations, which worked before, but became incompatible with recent changes... Oct 18 01:44:49 trying now to figure out which one Oct 18 01:46:10 what changes do you have Oct 18 01:49:24 recent changes I meant in the oe-dev Oct 18 01:49:51 or do you mean my config? Oct 18 01:51:27 your config Oct 18 01:52:41 SDK/external-toolchain with CSL Oct 18 01:53:20 deploy outside of temp Oct 18 01:54:23 the first one also changes SDK_PREFIX and sets several ASSUME_SHLIBS Oct 18 01:55:32 ASSUME_SHLIBS could be problems Oct 18 01:55:43 why do u use external toolchain Oct 18 01:56:42 one of the requirements Oct 18 01:57:24 ok Oct 18 01:58:33 Otavio: Are you testing before committing? ;-) There are a dozen of missing cheksums for xfce* Oct 18 01:58:44 it's a weird setup to use binary CSL, but that's one of the reqs I have... and it used to work! :) Oct 18 01:59:18 I'll try disabling ASSUME_SHLIBS for now Oct 18 02:00:39 denix: it used to work thats the common phrase when you .dev :) Oct 18 02:00:47 khem: do you have zlib.ipk? Oct 18 02:01:35 khem: I know, that's why I don't say it's .dev problem, but mine, just having hard time figuring it out :) Oct 18 02:05:09 denix: My trees are busted atm Oct 18 02:05:25 I am trying to transition into git world Oct 18 02:32:11 khem: reverted back to when it worked - will try to figure out what's changed Oct 18 02:32:54 hmmm better you should upgrade step by step Oct 18 02:33:06 I dont know how old was your snapshot that worked Oct 18 02:33:18 khem: about 2 weeks old Oct 18 02:33:31 oh then it should not be that hard Oct 18 02:33:38 do binary builds Oct 18 02:33:52 khem: was traveling around that time, so was not following all the changes Oct 18 02:33:52 like upgrade to a week old snapshot and see Oct 18 02:34:13 denix: ok, r you in US west coast ? Oct 18 02:34:23 khem: will do Oct 18 02:34:32 khem: east coast Oct 18 02:34:39 ah ok Oct 18 02:35:32 * khem dinner time bbl Oct 18 02:35:39 khem: it's 10:30pm here and I'm still in the office... had only 1 bagel and lots of coffee today :) Oct 18 02:35:56 heh Oct 18 02:36:02 khem: time for me to go home Oct 18 02:36:06 good night Oct 18 02:36:15 ok drive safe and happy weekend Oct 18 02:36:25 u2 Oct 18 02:36:55 * denix is away: I'm not here **** ENDING LOGGING AT Sat Oct 18 02:59:56 2008