**** BEGIN LOGGING AT Wed Feb 28 03:00:01 2007 Feb 28 03:15:06 HopsNBarley: Good evening Feb 28 03:37:01 hi guys Feb 28 03:37:12 what package name would I depend on if i needed libgl and libglu? Feb 28 03:37:16 they don't seem to exist themselves Feb 28 03:39:00 jselect: see the openembedded email list archives, the last few topics discuss this. Feb 28 03:41:07 awesome, already looking there and I think I may be blind :) Feb 28 03:50:10 hmm or not. I can't see anything about libgl in the email list archives Feb 28 04:14:20 ouch. That's got to be something bad.. I specified libgl and libglu RDEPENDS and suddenly gcc is failing with a command line that includes libs in libs/.libs/ Feb 28 04:35:31 hmm. Now if i remove the libglu and libgl rdepends it seems oeconf is sure it needs includes in libs/.libs instead of just libs/ Feb 28 05:25:13 well that's broken. I had a problem where oe was looking for libs in libs/.libs/ and the only way to cleanly fix it (other than redirecting all libs to "libs/" was to clean world Feb 28 05:25:17 nasty stuff.. Feb 28 05:40:30 Hmmm.. libsdl-x11 and xorg-x11 together compile properly Feb 28 05:53:38 but xorg-x11 does not provide a symlink to gl.h from outside the xorg headers dir Feb 28 05:59:52 good day Feb 28 06:19:25 morning all Feb 28 06:20:28 rwhitby : madwifi-ng builds ok now on powerpc Feb 28 06:20:58 cool. so we're done? Feb 28 06:21:35 i think yes. will try to use it also.probably in the afternoon Feb 28 06:21:55 btw could you have a look at the asterisk package also Feb 28 06:22:09 Ifaistos: imagine that - two OE developers broke each other's distros, and fixed it without even shouting at each other :-) Feb 28 06:23:08 shouting never fixed anything Feb 28 06:23:36 indeed. let's hope this start a new era of cooperation in OE ;-) Feb 28 06:24:09 heh - asterisk_1.2.14.bb looks pretty xscale-centric, doesn't it? Feb 28 06:24:58 export CROSS_PROC="arm" is a giveaway Feb 28 06:25:59 * mwester goes to look at the fuse .bb recipes once again before he risks breaking a distro somewhere... ;) Feb 28 06:26:11 has it always been like that, or did it become arm-centric at some point? Feb 28 06:27:02 but also look at the asterisk.patch. one thing i noticed is that it disables optimazation for other archs x86,ppc mainly Feb 28 06:27:09 not sure Feb 28 06:27:52 1.0.9 uses ${TARGET_ARCH}, so someone has armicised it somewhere along the line. Feb 28 06:29:50 I think we will soon need a kind of "repository marshal" that looks what get's in Feb 28 06:30:19 arch specific patches, and definitions are all over Feb 28 06:31:50 Ifaistos: no-one is listed as asterisk maintainer, so you should just try and make it cross-architecture again. Feb 28 06:33:13 looks like asterisk went bad between 1.2.12.1 and 1.2.13 - find out who checked in 1.2.13 and you have the culprit. Feb 28 06:33:55 * Ifaistos prepares tar and feathers Feb 28 06:34:12 I'm not even sure of an easy way to do that in monotone ... Feb 28 06:34:29 :) Feb 28 06:34:39 rwhitby: mtn log? Feb 28 06:35:34 * rwhitby hopes his name doesn't show up ... Feb 28 06:37:55 * mwester ducks behind a nearby trash can, and tries to remember if he ever committed any asterisk changes ... Feb 28 06:41:16 asterisk: added 1.2.12.1 on behalf of Corneliu Doban Feb 28 06:45:50 anyway, i'll have a look at it, but could you please let me know why asterisk depends on all that packages ? is it used with them on slugos ? Feb 28 06:46:44 i think zlib , openssl and ncurses would be enough Feb 28 06:47:25 to build. Feb 28 06:47:41 the others provide extra functionality Feb 28 06:48:56 I believe the other packages are used for things like music on hold and so forth Feb 28 06:49:31 hmm - my "mtn log" command is still running ... Feb 28 06:50:04 looks like OE devs are "busy bees" :) Feb 28 06:51:22 mtn 34 (or thereabouts) was supposed to make the log command useful once again, I was told some weeks ago. Feb 28 06:54:21 you mean 'log '? Feb 28 06:55:39 yeah. I have an old version, so I can't complain. Feb 28 06:55:54 (0.31) Feb 28 06:56:42 ah, yeah, the restricted log speedups landed in 0.32 :-) Feb 28 07:00:57 * rwhitby does a grep for CIA and asterisk in #oe irc logs and gets the information faster than 0.31 :-) Feb 28 07:03:46 Ifaistos: I've tracked down the culprit, and will have words with him. Feb 28 07:12:17 :) Feb 28 07:12:54 madwifi is all ok for me as well (not using arm though, just sh4 and i486) Feb 28 07:16:36 rwhitby : have you by any chance tried asterisk and speex with fixpoint on xscale ? Feb 28 07:18:18 Ifaistos: nope, I'm not a big asterisk user. Feb 28 07:18:33 talk to osas Feb 28 07:21:55 * oxo is wondering: is het possible to specify the gcc-version to use for kernel (cross) compilation? Feb 28 07:26:14 oxo: Check out the jlime distro - it uses gcc 3.x for kernel and 4.x for everything else. Feb 28 07:26:24 ah, tnx Feb 28 07:37:25 v8jlene: you mean PREFERRED_PROVIDERS += "virtual/${TARGET_PREFIX}gcc-3.3.3:gcc-cross-kernel" ? Feb 28 07:37:36 it seems to have no effect at all Feb 28 07:37:58 'gcc-cross-kernel' is never referenced outside its declaration in conf/distro/jlime-donkey.conf Feb 28 07:38:54 oxo: check also the KERNEL_CCSUFFIX in packages/linux/linux-jlime-jornada7xx_2.6.19rc6.bb for example Feb 28 07:39:39 * oxo sees the light Feb 28 07:41:30 oxo: Jlime is handled by Kristoffer / Ken|JLime so he's the best to ask about this. I just know about it because I also build for sh where gcc 4.x can't be used for kernels - I use 3.x for everything, while Ken uses 4.x for everything except the kernel. Feb 28 07:44:14 * oxo did some tweaks and created a linux-krups_2.6.6.bb Feb 28 07:44:17 it might just work Feb 28 07:47:22 2.6.6 is a bit, ehm, "old" Feb 28 07:47:40 but I guess all those sparc hackers moved on to ultrasparcs Feb 28 07:50:38 My ultrasparcs even feel old... OE on a U5 or an X1 might make a nice firewall/router though... maybe I should work try it at some stage ;) Feb 28 08:01:24 Ifaistos: Checking in fix for asterisk now. I got the .bb for 1.2.13 and 1.2.14 from a slug user and didn't do good enough QC. Sorry. Feb 28 08:02:29 koen: yeah, but it is the latest confirmed working kernel with "our" patches Feb 28 08:02:42 (vga text mode, mtd/jsflash, sound) Feb 28 08:03:26 when I really have seas of spare time, I'll try to port those patches to 2.6-recent, perhaps trying to get them upstream Feb 28 08:03:34 but I'm not the original author Feb 28 08:05:42 03nail 07org.oe.dev * r7f12c033... 10/ (3 files in 2 dirs): Unbreak asterisk. Silly arm-specific variables Feb 28 08:07:57 morning Feb 28 08:08:06 dzien dobry Feb 28 08:08:45 koen: sooner or later Polish will be official here? :D Feb 28 08:08:56 :) Feb 28 08:10:04 koen: at novotel I told lrg/xora/robtaylor that it will happen Feb 28 08:10:47 damn, I hate html mail Feb 28 08:14:59 NAiL : np. will check them a bit latter. just a quick note. The patches also need some review. especially the asterisk.patch Feb 28 08:24:03 03koen 07org.oe.dev * rff3d5f3c... 10/ (3 files in 2 dirs): openmoko-panel-demo(simple): fix packaging Feb 28 08:30:50 03koen 07org.oe.dev * r23c25869... 10/ (3 files in 3 dirs): openmoko-session: fix depends, make startup 2 seconds faster and add some matchbox-panel-2 tweaks Feb 28 08:30:56 03koen 07org.oe.dev * r1366964b... 10/ (1 packages/tasks/task-openmoko.bb): task-openmoko: switch back to matchbox-panel-2 now openmoko-session can handle it Feb 28 08:48:43 03koen 07org.oe.dev * r060ef98f... 10/ (1 conf/bitbake.conf): bitbake.conf: add matchbox-panel debug dirs Feb 28 08:49:41 koen: hmm, fyi, the monotone_ciabot_hookversion.py script in mtn mainline just got a real revision parser, so those random "1"s go away if you upgrade to it Feb 28 08:50:14 which random "1"s? Feb 28 08:50:36 koen: in " koen org.oe.dev * r060ef98f... / (1 conf/bitbake.conf)" Feb 28 08:50:43 in the fiel list Feb 28 08:50:47 not that big a deal, but hey :-) Feb 28 08:50:47 ah, right Feb 28 08:50:53 never even noticed it :) Feb 28 08:51:06 * koen adds to todo Feb 28 08:51:25 I wonder what 0.33 will break Feb 28 08:51:42 everything! Feb 28 08:51:44 or nothing Feb 28 08:51:47 it's a mystery Feb 28 08:51:58 no 'db migrate' needed? Feb 28 08:52:01 ("anything that parses log output" is a start) Feb 28 08:52:11 db migrate needed, but no regenerate_caches this time IIRC Feb 28 08:52:19 ah, hmmm Feb 28 08:53:06 time to add yet another version to http://www.openembedded.org/snapshots/ Feb 28 08:53:25 or people could just run db migrate after they download, I guess Feb 28 08:53:30 heh Feb 28 08:53:37 migrate is not nearly as slow, time it first :-) Feb 28 08:53:38 just overestimate users Feb 28 08:54:11 mm, if you have any suggestions on how the helpful error messages could be more helpful, let us know... Feb 28 08:54:41 users don't read error messages Feb 28 08:55:11 or rather, the people that start reporting the 'error' never do Feb 28 08:55:27 can't help you with that :-/ Feb 28 08:55:30 nope Feb 28 08:56:29 you could try to convince JustinP to shave that last 80% speed-up out of initial pull so you can stop using the seeding databases, now that he's a mtn internals expert ;-) Feb 28 09:11:27 03koen 07org.oe.dev * r265fc70b... 10/ (3 files in 3 dirs): openmoko-sesson: scale icons to 96px instead of 100px Feb 28 09:12:30 hmm, how often should I sync my OE.mtn in order to have recent stuff and not overload the server? Feb 28 09:12:47 ie: how often do/should most people sync? Feb 28 09:13:11 oxo: you can sync to ewi546.ewi.utwente.nl as often as you want :) Feb 28 09:13:50 the thing is: m.o.o seems somewhat faster then ewi546 to me Feb 28 09:14:13 does ewi have the updates pushed to it? Feb 28 09:14:25 or is it pull-synced? Feb 28 09:14:37 both are synced every 10 minutes Feb 28 09:15:01 ah Feb 28 09:31:23 03koen 07org.oe.dev * r75d78b08... 10/ (3 files in 3 dirs): openmoko-session: start some panel applets by default Feb 28 09:46:03 morning all Feb 28 09:46:36 good morning RP Feb 28 09:46:37 Ello Feb 28 09:54:10 RP: morning:) Feb 28 10:30:47 gm Feb 28 10:37:17 http://www.flickr.com/photos/32615155@N00/405228461/in/set-72157594561002629/ Feb 28 10:37:37 lots of people Feb 28 10:38:17 nerds! Run for your loves Feb 28 10:39:23 http://www.flickr.com/photos/32615155@N00/405229708/in/photostream/ <- zecke's slug in pieces Feb 28 10:41:11 So I hvae noticed that if I have a program that links against liba, and liba is linked against lib Feb 28 10:41:32 I do not need to have libb in the link statement, only liba Feb 28 10:42:00 if libtool is working, yes Feb 28 10:42:04 Should my autofoo (for the program) check fr lliba (since libb won't build without it it)? Feb 28 10:43:27 stylistically speaking what is the "right" thing to do? Feb 28 10:45:17 I guess so Feb 28 10:45:25 if your own app doesn't need libb Feb 28 10:45:39 so don't check unless I need it Feb 28 10:45:41 in case liba decides to work without libb, but your app still needs it Feb 28 10:46:06 well, that isn't likely in this case (major change) Feb 28 10:46:07 ok Feb 28 10:47:14 I noticed this last nght and realized is wasn't being consistent Feb 28 10:49:40 hi guys, is there a way for me to specify to bitbake another path for the gcc compiler when I am building something? Feb 28 10:53:25 hi zecke Feb 28 10:56:05 donato_home: somewhere in the .conf file, you can set ASSUME_PROVIDED, so it will expect some cross compiler to be available on system... Feb 28 10:56:28 donato_home: like using a precompiled toolchain? Feb 28 10:57:04 donato_home: in that case read the manual Feb 28 11:05:00 03koen 07org.oe.dev * r2db9e92f... 10/ (3 files in 3 dirs): openmoko-session: use openmoko-footer instead of second panel (thanks Mickey!), start mbimputmgr by hand Feb 28 11:10:02 http://pastebin.ca/375557 Feb 28 11:11:39 How can I apply a patch to a package, for onlu one machine and not others? Feb 28 11:12:03 SRC_URI_machinename_append = "..." Feb 28 11:12:15 ops. append_machinename Feb 28 11:12:50 SRC_URI_append_omap5912osk Feb 28 11:13:19 What I really need is SRC_URI_append_usb1.0 ... Feb 28 11:17:58 my pastebin above has a QA error, non debug package include .debug Feb 28 11:19:20 FILES_${PN}-dbg += "${libdir}/python2.4/site-packages/.debug" Feb 28 11:20:53 This is some form of python debugging info? Feb 28 11:21:04 I guess I should rad the book I lugged back from FOSDEM Feb 28 11:21:12 After I tape something over the snake Feb 28 11:21:18 no, OE copy the debugging symbols into a .debug dir Feb 28 11:21:22 * Crofton|home fears snakes Feb 28 11:21:28 ah Feb 28 11:21:29 ok Feb 28 11:22:04 does /python*/site... work? Feb 28 11:22:45 I think it should Feb 28 11:22:56 yeah the python p[ackage is creatd that way Feb 28 11:23:05 we'll know in a minute Feb 28 11:23:56 http://www.phdcomics.com/comics.php Feb 28 11:29:28 hmm Feb 28 11:29:57 * oxo just created a 'task-krups.bb' with 2 packages in it: 'krups-task-base' and 'krups-task-net' Feb 28 11:30:13 but BB claims that nothing provides my fresh 'krups-task-base' Feb 28 11:30:40 do I have to register the bb somewhere else than packages/task/task-krups.bb ? Feb 28 11:39:08 Feature request: spell checker for bb files Feb 28 12:02:09 I had some trouble earlier on trying to install and use wget Feb 28 12:02:25 the problem is that wget is built with --enable-ipv6 Feb 28 12:02:41 which makes connections use ipv6 by default Feb 28 12:02:55 my kernel did not have ipv6 support Feb 28 12:03:22 can kernel modules/config options be specified as dependencies? Feb 28 12:04:04 tkp: Yep.. RDEPENDS += "kernel-module-foo" Feb 28 12:04:08 tkp: you can pull in any modules you want by setting them up as dependencies for your image. Feb 28 12:04:15 NOTE: package ossiecf-0.0.0+svn20070228-r0: task do_fetch: completed Feb 28 12:04:15 NOTE: package ossiecf-0.0.0+svn20070228-r0: task do_unpack: started Feb 28 12:04:15 NOTE: Unpacking /home/balister/bb_sources/ossie_ossie-dev.mprg.org_.repos.ossie.ossie.trunk__20070228.tar.gz to /home/balister/oe/tmp/work/ppc603e-angstrom-linux/ossiecf-0.0.0+svn20070228-r0/ Feb 28 12:04:19 NOTE: package ossiecf-0.0.0+svn20070228-r0: task do_unpack: completed Feb 28 12:04:19 NOTE: package ossiecf-0.0.0+svn20070228-r0: task do_patch: started Feb 28 12:04:21 NOTE: package ossiecf-0.0.0+svn20070228-r0: task do_patch: completed Feb 28 12:04:24 NOTE: package ossiecf-0.0.0+svn20070228-r0: task do_configure: started Feb 28 12:04:25 NOTE: exceptions.TypeError:float() argument must be a string or a number while evaluating: Feb 28 12:04:27 ${@base_less_or_equal("ROOT_FLASH_SIZE", "16", "", "en-gb", d)} Feb 28 12:04:30 NOTE: package ossiecf-0.0.0+svn20070228-r0: task do_configure: completed Feb 28 12:04:31 NOTE: package ossiecf-0.0.0+svn20070228-r0: task do_qa_configure: started Feb 28 12:04:33 NOTE: Checking sanity of the config.log file Feb 28 12:04:35 NOTE: package ossiecf-0.0.0+svn20070228-r0: task do_qa_configure: completed Feb 28 12:04:37 NOTE: package ossiecf-0.0.0+svn20070228-r0: task do_compile: started Feb 28 12:04:39 NOTE: exceptions.TypeError:float() argument must be a string or a number while evaluating: Feb 28 12:04:41 ${@base_less_or_equal("ROOT_FLASH_SIZE", "16", "", "en-gb", d)} Feb 28 12:04:43 NOTE: package ossiecf-0.0.0+svn20070228-r0: task do_compile: completed Feb 28 12:04:47 NOTE: package ossiecf-0.0.0+svn20070228-r0: task do_install: started Feb 28 12:04:50 NOTE: exceptions.TypeError:float() argument must be a string or a number while evaluating: Feb 28 12:04:51 ${@base_less_or_equal("ROOT_FLASH_SIZE", "16", "", "en-gb", d)} Feb 28 12:04:53 NOTE: package ossiecf-0.0.0+svn20070228-r0: task do_install: completed Feb 28 12:04:55 NOTE: package ossiecf-0.0.0+svn20070228-r0: task do_package: started Feb 28 12:04:56 way to go Feb 28 12:04:57 NOTE: exceptions.TypeError:float() argument must be a string or a number while evaluating: Feb 28 12:04:59 ${@base_less_or_equal("ROOT_FLASH_SIZE", "16", "", "en-gb", d)} Feb 28 12:05:02 NOTE: Multiple libraries (libossieparser.so.0, libossiecf.so.0, libossieidl.so.0) found and LEAD_SONAME not defined Feb 28 12:05:04 NOTE: DO PACKAGE QA Feb 28 12:05:06 NOTE: Checking Package: ossiecf-dbg Feb 28 12:05:08 NOTE: Checking Package: ossiecf Feb 28 12:05:10 N Feb 28 12:05:12 Does ack Feb 28 12:05:14 crap, sorry Feb 28 12:05:19 http://pastebin.ca/375588 Feb 28 12:05:20 Does this error make sense to anyone? Feb 28 12:05:22 er exception maybe Feb 28 12:05:24 CoreDump|afk, thanks Feb 28 12:05:25 CoreDump|afk: so, would you normally specify kernel module dependencies in the image.bb file, or in a package that actually depends on it (wget in this case)? Feb 28 12:05:26 I swear I copied the link into the paste buffer Feb 28 12:05:40 ~lart inconsistency between cut/paste and the center button Feb 28 12:05:40 * ibot hauls inconsistency between cut/paste and the center button up by the scruff of the neck and spanks him until he waddles Feb 28 12:05:43 or in something like task-mydistro Feb 28 12:06:16 tkp: having wget RDEPEND on the ipv6 module would break images where this module is not / can not be compiled for whatever reasons Feb 28 12:06:17 NAbyss_: never EVER RDEPEND on a kernel-module Feb 28 12:06:38 koen: Really? Feb 28 12:06:39 koen: that was my thinking.. and is why I am asking Feb 28 12:06:44 think of what happens when ipv6 is built-in Feb 28 12:06:48 koen: Does that break stuff badly? Feb 28 12:07:05 Ah, good point Feb 28 12:07:07 tkp: RRECOMMENDS = "kernel-module-ipv6" is the best option Feb 28 12:07:08 but wget with --enable-ipv6 DOES rdepend on a kernel module, or kernel support at least Feb 28 12:07:12 Crofton: sounds like ROOTFS_FLASH_SIZE is unset for your machine? Feb 28 12:07:12 I've just been using it as a local hack for my images Feb 28 12:07:17 koen: in wget? Feb 28 12:07:22 tkp: yes Feb 28 12:07:58 ok Feb 28 12:08:10 that would make sense Feb 28 12:08:20 tkp: or create a default config in /etc (if that is possible) to tell wget to use IPV4 only Feb 28 12:09:20 |'ll submit a patch for wget with the addition of the RRECOMMENDS Feb 28 12:11:05 * CoreDump|afk is off to work Feb 28 12:13:54 patch in the bug tracker Feb 28 12:15:06 so, if something is in RRECOMMENDS, does that force it to get built when building up an image, just like RDEPENDS? Feb 28 12:15:19 yes Feb 28 12:15:28 (depending on your bitbake version) Feb 28 12:16:27 really? what aboutt version 1.6.3? Feb 28 12:19:55 * tkp assumes that in the latest version it does... and upgrades his bitbake Feb 28 12:21:28 koen, is root flash size set for efika? Feb 28 12:21:44 * Crofton|home just realized this machine makes efika builds ... Feb 28 12:25:49 morning Feb 28 12:26:54 I got opencv lib to compile under OE and now I want to be able to use the resulting libraries with the OE produced cross compiler to build test software manually Feb 28 12:27:57 (without a bb package). cross compiling my code using these libs doesn't work as the host tries to get the native libs. Feb 28 12:31:09 so in short: how could I procude binaries for an ixp target on an x86 host using ixp opencv libs outside of OE? Feb 28 12:31:20 Crofton|home i think its 65MB Feb 28 12:31:24 64 Feb 28 12:31:43 I don't remember, but we need to add it to the machine file Feb 28 12:33:53 Or does a cross compiler find libs for it's target? It can't use the host's ldconf ones right? Feb 28 12:33:57 03koen 07org.oe.dev * r7f4176e2... 10/ (1 conf/bitbake.conf): bitbake.conf: ensure root flash size is set and default to '0' Feb 28 12:43:07 signed first half of keys... Feb 28 12:46:35 Ifaistos, root flash size is user defined for efika :) Feb 28 12:46:46 depends what you use in the ide slot :) Feb 28 12:47:42 Crofton|home : :) sure. but i think it was creating 64MB images (but maybe i don't remember correctly) Feb 28 12:48:17 The "problem" is I get messages when ROOT_FLASH_SIZE is not defined Feb 28 12:48:23 not fatal, but annoying Feb 28 12:49:25 Crofton|home: mtn up :) Feb 28 12:49:44 heh Feb 28 12:52:10 03crofton 07org.oe.dev * r136c8765... 10/ (1 packages/usrp/usrp_0.12.bb): Feb 28 12:52:10 usrp-0.12.bb : Fix QA problems with python packages. Feb 28 12:52:10 Apply usb1.1 patch only for omap5912osk. Feb 28 13:13:35 hello! little question I've built angstrom-bootstrap-image on x86, booting stops at "unble to open an initial console". Any quick way to enable the console? Feb 28 13:13:50 someone here played with Xen? Feb 28 13:14:23 hrw: not yet Feb 28 13:14:38 hrw: requires x86 and all my experimentation machines are !x86 Feb 28 13:14:48 koen: right Feb 28 13:17:48 koen: rememer that guy which had to work on f-b? Feb 28 13:18:07 feed browser? Feb 28 13:18:13 koen: I just realised that I was talking with him on fosdem... Feb 28 13:18:15 koen: yes Feb 28 13:30:25 cool, dri seems to initialize properly now on the efika Feb 28 13:31:32 we need name tags at fosdem Feb 28 13:32:28 Crofton: badges? Feb 28 13:32:39 yes Feb 28 13:32:53 Crofton: also poster(s), flyers Feb 28 13:33:04 yes Feb 28 13:33:20 I mean badges for all attendees, so we can put faces to names better Feb 28 13:38:45 hrw, http://jaaksi.blogspot.com/ Feb 28 13:39:59 Crofton: I read about n800 challenge already Feb 28 13:40:12 ok, wasn't sure if you knew about that blog Feb 28 13:40:28 Crofton: I found it on other place Feb 28 13:44:02 koen : can you build gcc (not cross) for efika ? Feb 28 13:44:26 koen: task-native-sdk for each platform :D Feb 28 13:44:41 Ifaistos: http://www.angstrom-distribution.org/unstable/feed/ppc603e/base/gcc_4.1.1-r9_ppc603e.ipk Feb 28 13:45:04 hrw: I already did that :) Feb 28 13:46:09 koen: then rebuild python Feb 28 13:48:45 http://paulgraham.com/anybots.html Feb 28 13:51:31 wow, nice link hrw Feb 28 13:59:10 dexter should knock the other robot over! Feb 28 14:14:28 re Feb 28 14:27:23 how could I force a package to be rebuilt every time I build an image? Feb 28 14:27:50 I had been using do_build[nostamp] = 1, but this doesn't seem to work quite right Feb 28 14:28:35 tht does force do_build to happen again, but I seem to end up with the ipk being lost Feb 28 14:33:07 koen: can you add glib_cv_uscore=${glib_cv_uscore=no} to site/sparc as well? Feb 28 14:34:41 sure Feb 28 14:34:44 oh wait, there is more stuff to be added Feb 28 14:34:59 something with pthread too Feb 28 14:35:50 * koen stabs matlab again Feb 28 14:37:19 03koen 07org.oe.dev * rcbe569b7... 10/ (1 site/sparc-linux): sparc-linux: add glib_cv_uscore=${glib_cv_uscore=no} Feb 28 14:38:23 I see some QA issues in gcc-cross Feb 28 14:39:00 http://pastebin.ca/375699 Feb 28 14:41:35 how do I get this new Madwifi to apply the wackelf patch to my armv5te machine build? Feb 28 14:44:16 koen: checking size of pthread_t... configure: error: cannot run test program while cross compiling Feb 28 14:44:30 I really can't find what hint to stash in site/sparc for this one Feb 28 14:47:04 I think it's something with ac_cv_sizeof_pthread_t in it Feb 28 14:47:11 but it's not defined for other archs Feb 28 14:52:40 and I do not dare to guess it myself :-) Feb 28 14:54:28 on sparc64 it's 4 Feb 28 15:02:30 Can anyone help me out with this: Feb 28 15:02:43 bin/sh: line 1: 25677 Floating point exception(core dumped) ../src/util/makekeys ks_tables_h Feb 28 15:02:44 make[1]: *** [ks_tables.h] Error 136 Feb 28 15:02:46 make[1]: Leaving directory `/media/sda1/oebuild/tmp/work/arm-linux/diet-x11-X11R7.1-1.0.1-r4/libX11-X11R7.1-1.0.1/src' Feb 28 15:07:55 Glombool: I was getting this error a few days ago, but i broke something else, so i cant get back to the point where its trying to build diet-X11 Feb 28 15:08:03 Ouch Feb 28 15:09:36 i was looking through the oe tickets and it looked like ticket 1192 was for a diet-X11 patch, and someone else said they had got the package to build on feb 17, 2007 Feb 28 15:10:07 Hm Feb 28 15:10:18 I'll have to check it out... Feb 28 15:10:51 I'm assuming that's in the bugtrackker... Feb 28 15:10:55 in Feb 28 15:11:24 i'm not sure this is related, but makes me think it could be a new problem since feb 17 Feb 28 15:11:44 Yeah Feb 28 15:11:45 http://bugs.openembedded.org/show_bug.cgi?id=1192 Feb 28 15:12:10 right... i saw it. Feb 28 15:14:13 ...I'm trying a different provider. Feb 28 15:14:13 see what happens. Feb 28 15:14:33 It's making me mad because I just need to compile a custom gpe-soundbite to work with my collie. Feb 28 15:14:49 Stupid non-standard pda. Feb 28 15:17:29 Jin^eLD: morning :) Feb 28 15:17:36 hey hvontres|poodle :) Feb 28 15:18:48 Jin^eLD: I did some more testing on qemu and it seems that qemu on fedora still has a problem..... which reminds me, I need to post some glibc_install.inc related quetions... Feb 28 15:19:18 to be honest I did not follow this issue, so did not do any further tests yet Feb 28 15:20:34 ...now I *REALLY* broke it :( Feb 28 15:20:50 Jin^eLD: well, I just got a little curious. The binary locale generation cmdline still looks fishy to me though. I'll have to do some more poking around. Feb 28 15:21:52 I was working on another platform with another distro since then, so I did not stumble across this anymore Feb 28 15:26:42 oh, i didn't break it :) Feb 28 15:27:00 you got it working? :0) Feb 28 15:27:39 not sure yet Feb 28 15:27:47 will be a minute or so Feb 28 15:30:40 hmm. nope Feb 28 15:33:29 does bitbake have a "clean" feature? Feb 28 15:36:10 ah, yes it does :) Feb 28 15:36:13 bitbake -c clean Feb 28 15:37:05 i used bitbake -c clean world, and that got me to the point where i am now, cant build anything cuz its looking for some i686 gnuconfig stuff Feb 28 15:37:25 ouch Feb 28 15:37:59 ok.... "consider defining PREFERRED_PROVIDER_virtual/arm-linux-libc-for-gcc" Feb 28 15:38:13 dspencek: you can also remove you tmp dir and start over Feb 28 15:38:14 wait, the error now is about m4-native Feb 28 15:38:25 i'll try that Feb 28 15:38:38 if i add PREFERRED_PROVIDERS .= virtual/arm-linux-libc-for-gcc:diet-x11" Feb 28 15:38:48 is that what it means? Feb 28 15:40:06 koen, I'm still unable write to the OE rep. If you have not added me yet that is fine I just wanted to make sure I'm not doing something wrong. Feb 28 15:40:48 If my machine is an armv5te, do I need to have ixp4xxle as a PACKAGE_EXTRA_ARCHS as well? I'm not sure how the two are related. Feb 28 15:41:54 apaulsen: armv5te != ixp4xx Feb 28 15:42:21 svolpe_gerrath: It's kinda hard to add you when you don't send a key..... Feb 28 15:42:49 koen, I did send a key.. Feb 28 15:42:58 it's not in my inbox Feb 28 15:43:09 Perhaps I should try org.openembedded.oz354x instead... Feb 28 15:43:17 since I'm developing for openzaurus... Feb 28 15:43:31 Glombool: what were you using before? Feb 28 15:43:35 koen, uhm, well I sent it using the example on the OE website.. I will just email it to you direct. Feb 28 15:43:51 the org.openembedded.dev Feb 28 15:43:54 just follow http://www.openembedded.org/monotonephrasebook Feb 28 15:43:57 ahh Feb 28 15:44:12 makes a difference I suppose :) Feb 28 15:44:25 koen, that is what I followed.. I will try again. Thanks. Feb 28 15:44:37 hrw: ok, I'm still trying to get madwifi to build for my machine and it looks like the wackelf patch only applies to ixp4xx machines Feb 28 15:45:31 is there any documentation on adding native bitbake recipes? I found http://www.openembedded.org/user-manual&dpage=recipes_native, but nothing is there yet Feb 28 15:46:11 Glombool: i donno, i'm using dev too though, i thought the oz354x was like a freeze on the dev for building the 354x distro and packages Feb 28 15:46:38 oh Feb 28 15:47:07 Ah, I don't want that one anyway Feb 28 15:47:36 dspencek: .oz354x is obsolete Feb 28 15:47:51 right Feb 28 15:47:56 ok Feb 28 15:50:00 koen, I just sent the key using my gmail since you did not recieve it using mail.. I'm using svolpe@openembedded.org as my username. Feb 28 15:53:00 hrw: so are you planning to make oz 3.5.5 based on .dev then? (When you can find some time of courese :) ) Feb 28 15:55:10 QOTD: I'm going back to windoze, too many dammed options with linux ;) -- Blumf Feb 28 15:59:29 ok, i'm back Feb 28 15:59:37 koen, just let me know when you get my email :-) Feb 28 15:59:57 I'm deving for 3.5.4 at the moment. No collie support for newer versions yet :( Feb 28 16:00:07 * Glombool gives collie a dirty look Feb 28 16:04:20 fscking RBLs Feb 28 16:04:30 I cant sent mail from my home machine Feb 28 16:04:43 hrw: ouch Feb 28 16:05:11 and my exim knowledge suxx Feb 28 16:05:46 hrw: I installed this www.citadel.org and it works great Feb 28 16:15:25 koen, RP: thanks for the multimachine tips. works awesome. Feb 28 16:24:37 anyone got a minimal loal.conf for building angstrom? Feb 28 16:26:50 hrw, hi Feb 28 16:27:43 i saw your comment about SDHC; this means that mainline kernels now support it? or the kernels OE is building? Feb 28 16:27:46 (or both :) Feb 28 16:27:53 or... work in progress, still Feb 28 16:32:49 raduga: its in mainline kernel Feb 28 16:38:28 hrw, should I set ROOT_FLASH_SIZE for the osk? Feb 28 16:39:24 to get madwifi to work I had to add WACKELF_SRC_URI_compulab-pxa270 = " file://20-xscale-VFP-wackelf.patch;patch=1" Feb 28 16:39:25 to the madwifi-ng.inc. This still doesn't seem like the most elegant solution, but I don't have any better ideas either. Feb 28 16:40:30 Crofton__: if it has flash onboard then you better set it Feb 28 16:40:56 the filesystem partition in flash is 30464K Feb 28 16:41:13 FSCK! Feb 28 16:41:17 looks like the var takes units of M Feb 28 16:41:18 I HATE SMTP Feb 28 16:41:22 heh Feb 28 16:41:27 RBLs etc SHIT Feb 28 16:41:51 to send one fscking mail from my box to somewhere I need to have MTA installed Feb 28 16:41:57 KURWA Feb 28 16:43:04 hrw, do I round up or down for root flash size? Feb 28 16:43:13 I am kind of guessing down .... Feb 28 16:43:59 Crofton__: I would go down... since writing to the "rounded up" flash may cause IO errors :) Feb 28 16:44:09 :) Feb 28 16:44:11 yeah Feb 28 16:44:15 I set to 30K Feb 28 16:44:19 er 30 Feb 28 16:44:39 hmmm... 30K seems like a bit much rounding :) Feb 28 16:44:45 :-) Feb 28 16:45:09 I am going to see if I can use angstrom on the OSK instead of openomap Feb 28 16:45:48 Crofton__: 29 might be better: 30464K/1024=29.73M Feb 28 16:47:20 good point Feb 28 16:48:35 hrw: thanks Feb 28 16:48:37 apaulsen: Agreed - it's ugly. As I consider it further, it seems to me that the correct place is in the autoconf logic -- that's where it knows such low-level details. Feb 28 16:49:11 apaulsen: but if you just pastebin your fixed .bb file, I'll commit it so that it works for everyone (even though its ugly). Feb 28 16:51:56 apaulsen: BTW, there was an assumption made (by me, originally) that the wackelf only applied to ixp4xx --- it seems it also applies to the pxa device. My bad - sorry! Feb 28 16:52:04 cu Feb 28 16:53:32 mwester: http://www.pasteserver.net/292 Feb 28 16:58:14 does anyone know if it's possible to specify the timeout the ipkg uses with wget? Feb 28 17:18:12 is there a bitbake option to rebuild all packages? Feb 28 17:19:17 not all packages in the tree, but all packages previously built? Feb 28 17:19:40 I want to change my glibc and gcc versions, then recompile my packages Feb 28 17:21:29 hhm, why does bootstrap-image try to build qemu-native? Feb 28 17:21:49 probably to build the locale bits, if you're using that option Feb 28 17:22:13 hooray netatalk compiles... Feb 28 17:22:55 Crofton: ENABLE_BINARY_LOCALE_GENERATION = "0" Feb 28 17:22:58 that helped me Feb 28 17:23:37 in machine or local.conf ? Feb 28 17:23:46 local Feb 28 17:24:08 theres nothing machine specific about that afaik, i'd expect local or distro (if anything) Feb 28 17:36:23 hi kergoth Feb 28 17:37:53 I have kernel-module-ipv6 in RRECOMMENDS for wget Feb 28 17:38:06 and I have wget in RDEPENDS for another package Feb 28 17:38:17 which is included in my root image Feb 28 17:38:31 yet my root image ends up without the kernel module Feb 28 17:39:00 isn't RRECOMMENDS supposed to be included in a root image? Feb 28 17:42:56 03mwester 07org.oe.dev * rff5998e4... 10/ (1 packages/madwifi/madwifi-ng_r.inc): madwifi-ng: The wackelf patch also applies to compulab-pxa270 Feb 28 17:43:30 how can I get things in RRECOMMENDS into the root image? Feb 28 17:44:41 RDEPENDS? (but I seem to recall that it's dangerous to depend on kernel modules. Feb 28 17:44:45 ) Feb 28 17:44:48 right Feb 28 17:44:50 hey chouimat Feb 28 17:44:55 so..?! Feb 28 17:47:24 xumbi8: rm -rf tmp ? Feb 28 17:49:14 it gets included when installing the wget package on the target Feb 28 17:49:51 ie, `ipkg install wget` installs the kernel module first Feb 28 17:54:01 here's a weird one: mkfs.jffs2 is chokeing with: tmp/rootfs/etc/ipkg/arch.conf: Not a directory Feb 28 17:54:23 yet the file is there. if i run the exact mkfs.jffs2 command from the run.* script, it works. any ideas? Feb 28 18:06:25 hi all Feb 28 18:06:42 hi mr_nice Feb 28 18:07:24 woglinde: hi, may I ask you something about cross compiling and libraries? Feb 28 18:07:31 sure Feb 28 18:09:44 woglinde: I hava a programm (xynth) which needs expact (xml parser) for its widget-tool. I have cross compilde the library and added LD_PATH=path/to/lib to the Makefile. but I will get a ld error that it can't find it. what do I need to tell make to know about the lib? Feb 28 18:10:38 mr_nice hm you crosscompiled the expatlib and installed it on the target device? Feb 28 18:11:39 woglinde: no, I crosscompiled expactlib and want to compile xynth which needs the lib to compile. Feb 28 18:11:49 ah Feb 28 18:12:06 LD_PATH is only for runtime Feb 28 18:12:18 you need a compiler flag Feb 28 18:12:34 woglinde: which is called? Feb 28 18:13:31 mr_nice -llibname specfied the libaryname Feb 28 18:13:48 and -LPathtosomewehre the searchpath for this library Feb 28 18:14:20 look in one of the logs in work Feb 28 18:14:29 woglinde: ah many thx! Feb 28 18:14:30 dspencek: the diet-x11 compile prob is just the dev branch Feb 28 18:16:09 mr_nice hm I am wrong its no compiler flag its linke flag ;) Feb 28 18:16:26 but they are they right options when gcc is linking Feb 28 18:17:16 woglinde: hehe, thats the thing which matters Feb 28 18:23:00 koen, did my key arrive in your email yet? Feb 28 18:32:14 if i ./run.image_cmd.12345, it works. if it let bitbake run that file, it doesn't. this somehow has to involve multimachine, cause that is the only thing that has changed. Feb 28 18:35:38 Can someone explain binary locale generation to me, in a few words Feb 28 18:38:26 Crofton__: If I understand this correctly, the glibc locale information needs to be compiled into binary form. This can be done on the target(ususlly slow) or in qemu (binary-locale generation), asuming qemu-native builds and works Feb 28 18:38:27 Crofton__: locale generation takes ~100MB ram, which most machines don't have, so we do it off-line using qemu Feb 28 18:38:43 03koen 07org.oe.dev * r8f6b6651... 10/ (1 packages/ohand-tasks packages/ohand-tasks/tasks_0.2.bb): o-hand tasks: add 0.2 Feb 28 18:39:04 I suspect I never install locale stuff on the OSK anyway ... Feb 28 18:39:31 koen, any idea on my wierd one? debugging tips? Feb 28 18:40:14 HopsNBarley: no idea, sorry Feb 28 18:40:14 this whole locale stuff is confusing to a native lowest common denominator speaker Feb 28 18:40:21 koen: could you take a peak at the glibc-package.bbclass stuff I posted to the ml? I might have a fix at home I can throw into the bug tracker when I get home tonight. Feb 28 18:40:49 thanks. Feb 28 18:41:17 hvontres|poodle: well, there are already 2 or 3 bugs filed about that... Feb 28 18:42:33 I found the business cards ! Feb 28 18:42:55 koen: I was just trying to see if there was a reason to WANT to use stuff in /usr/share that I was not aware of. Feb 28 18:43:20 hvontres|poodle: it's a bug in os.path.join Feb 28 18:43:48 koen: what about the second issue of using datadir instead of treedir? Feb 28 18:43:52 Koen: is it okay if I put myself up as contact for press and donation related issues and that when something comes in I report to core team ? Feb 28 18:44:51 gerwinin: sure Feb 28 18:46:39 Koen: I will contact nokia for vouchers for the n800 for 20 oe developers Feb 28 18:46:55 20? Feb 28 18:47:09 Woglinde : should it be more ? Feb 28 18:47:22 gerwinin, I like the way you think :) Feb 28 18:47:47 Woglinde: a voucher means that we can purchase the device for 99 euro Feb 28 18:48:00 gerwinin hm right Feb 28 18:48:29 hm I have the n770 from zecke, but dont play with it anymore Feb 28 18:48:45 exit Feb 28 18:49:24 Woglinde: the n800 is their new device Feb 28 18:50:16 gerwinin yes I know Feb 28 18:50:29 I am reading linuxdevices every day Feb 28 18:56:36 Hello Feb 28 18:56:45 hi sirfred Feb 28 19:02:16 koen : did you fix the problem with efika ? Feb 28 19:02:33 koen : just in haven't look at any commits yet Feb 28 19:06:50 * oxo crosses his fingers Feb 28 19:07:07 my Javastation Krups is going to try booting the image I created with OE Feb 28 19:07:10 first time! Feb 28 19:07:13 oxo hehe Feb 28 19:07:20 Oxo :) Feb 28 19:07:51 it's somewhat more complicated than handheld jffs-images, because I have to split it into a romfs with SILO/PROLL/kernel and a jffs2 containing the FS Feb 28 19:08:13 Oxo cool Feb 28 19:08:13 part of the SILO-steps do not like to be done on a non-sparc box, but I'll try to make a cross-solution to it Feb 28 19:08:28 there is a whopping 7M of space left for JFFS2 :-) Feb 28 19:08:45 ieehls silo Feb 28 19:10:21 ieehls indeed, it doesn't feel like booting today Feb 28 19:11:47 hi all Feb 28 19:13:03 hi gremlin Feb 28 19:13:22 hi woglinde Feb 28 19:16:15 re Feb 28 19:16:40 hi hvontres|poodle Feb 28 19:16:55 hi pgfeller :) Feb 28 19:17:28 sorry, I haven't had a chance to try your new stuff yet. I might get a chance tonight when I get home Feb 28 19:17:49 hvontres|poodle: absolutely no problem Feb 28 19:19:00 hvontres|poodle: just let me know if you need additional details to try things - or if there are unexpected problems Feb 28 19:19:18 RP: Quick qemu-arm question: what does "-r 2.6.16" do ? Feb 28 19:43:29 grrr why XFS spend about 15minutes to wipe out all /tmp dir ??? Feb 28 19:44:00 gremlin which kernel? Feb 28 19:44:14 2.6.19-1.2911.fc6 Feb 28 19:44:24 args Feb 28 19:44:36 2.6.19 has problems with xfs Feb 28 19:44:43 uff :( Feb 28 19:44:54 better use 2.6.20 Feb 28 19:45:06 dont know if fc fixed it back Feb 28 19:48:28 sigh Feb 28 19:58:07 z72ka: greetings! Feb 28 19:58:37 Sleep_Walker: Hi :) Feb 28 19:59:32 somebody noticed? : http://www.linuxdevices.com/articles/AT3411564713.html Feb 28 20:00:33 I need help, I have error in all of packages opie-*, error is (concretelly - for example): ERROR: QA Issue: non debug package contains .debug directory: opie-style-phase path /work/armv5te-angstrom-linux-gnueabi/opie-style-phase-1.2.2-r0/install/opie-style-phase/opt/QtPalmtop/plugins/styles/.debug/libphasestyle.so Feb 28 20:00:34 - can anybody help me? Feb 28 20:00:49 z72ka pull and update Feb 28 20:01:00 pskovloksy fixed it yesterday Feb 28 20:01:30 woglinde: Aaaa :) thx... I try it Feb 28 20:17:42 RP: you're an LTTng god, arent you?.. Feb 28 20:28:37 getting ready to test Angstrom image on the OSK Feb 28 20:34:47 koen, Angstrom boots on the OSK Feb 28 20:35:20 Crofton: cool Feb 28 20:35:26 yeah Feb 28 20:35:41 I can work toward orphaing openomap Feb 28 20:36:09 and create an openSDR image :) Feb 28 20:36:27 yeah Feb 28 20:36:34 I already did that for efika Feb 28 20:38:00 well, built an sdr with angstrom Feb 28 20:38:32 * koen wishes the usrp wasn't so expensive Feb 28 20:38:43 it is all relative :) Feb 28 20:39:40 It is probably cheaper than the data acquisition card in the utwente design Feb 28 20:39:53 yeah Feb 28 20:40:17 hi Crofton koen Feb 28 20:40:23 hey gremlin[it] Feb 28 20:40:23 gm Feb 28 20:45:33 should I create bugzilla entries for packages with QA issues? Feb 28 20:45:55 crofton why not Feb 28 20:48:10 I'm not sure if we are supposed to ignore some QA ERRORs :) Feb 28 20:48:20 like http://bugs.openembedded.org/show_bug.cgi?id=1785 Feb 28 20:52:57 he zecke Feb 28 20:55:15 Crofton: since last week all qa errors should be errors instead of attention points Feb 28 20:55:25 hang on Feb 28 20:55:52 koen hm whats with natvie .la which contains special paths Feb 28 20:55:56 ? Feb 28 20:56:02 native Feb 28 20:56:22 it means they are broken Feb 28 20:56:33 http://pastebin.ca/376138 Feb 28 20:56:51 woglinde: hey Feb 28 20:57:08 okay they shouldnt have the path set because oe set the right -L path? Feb 28 20:57:16 woglinde: they should be installed and should not point to workdir Feb 28 20:58:18 zecke please have a look at libz.la Feb 28 20:58:46 woglinde: paste it please Feb 28 20:59:13 there is nothing to paste Feb 28 20:59:26 contains no work dir but QA bubble up Feb 28 20:59:58 it's empty? Feb 28 21:00:09 woglinde: find . -name 'libz.la' | wc -l Feb 28 21:00:14 woglinde: is that one or two? Feb 28 21:01:46 mickeyl: ping? so the devices were produced and I can decide if you throw them away, or I will pay custom and throw it away? Feb 28 21:03:27 zecke because QA dont say if it means native .la or target .la Feb 28 21:03:36 only libz.la ;) Feb 28 21:04:37 woglinde: it looks into STAGING_LIBDIR, I should print the complete path again... Feb 28 21:05:20 zecke: yes, that's more or less it Feb 28 21:05:38 i mean there's a whole lot of stuff you can do with the device Feb 28 21:05:43 it just isn't "mobile" Feb 28 21:06:03 mickeyl: I'm just interested in the cost point of view Feb 28 21:06:36 mickeyl: so the devices were produced, so only the shipping cost might be saved? Feb 28 21:08:01 mickeyl: same battery life as a ezx phone ;) Feb 28 21:08:27 I need a little help with a bitbake file. Feb 28 21:08:33 Oh... nevermind. It's working now :) Feb 28 21:08:46 see, we are telapathic Feb 28 21:08:55 Heh heh. Feb 28 21:09:19 zecke: yes. the devices are already done. Feb 28 21:09:43 I was getting frustrated earlier. Trying to develop for OZ 3.5.4 using the dev branch. Then I realized "DUH! Use the OZ branch!" Feb 28 21:10:51 mickeyl: http://burtonini.com/computing/screenshots/tasks-0.2.png did ross mail you? Feb 28 21:11:06 not yet :) Feb 28 21:11:21 I guess I do have a bitbake question anyhow. I compiled a package. Then I hacked the code a little and am recompiling it. It seems to be going through the entire dependency process again (Just not having to recompile). Is this a case where I would use the -b attribute? Feb 28 21:11:22 mickeyl: "merge insane.bb" hmm Feb 28 21:11:54 Glombool, parsing? Feb 28 21:12:38 Crofton: um... what do you mean? Feb 28 21:12:59 I mean you are wondering about the parsing stage of running bitbake? Feb 28 21:13:04 Oh, no. Feb 28 21:13:17 It's actually baking now :) Feb 28 21:13:30 zecke: how about "Write bbclass to remove .la files" ? Feb 28 21:14:15 hvontres|poodle: I think libtool's concept is okay, just not GNU libtool Feb 28 21:14:23 Crofton It's going through each dependancy and unpacking the source, but the sources are all already compiled and in the build directory. Feb 28 21:14:26 I should some time to play with jlibtool Feb 28 21:14:34 g'night Feb 28 21:14:51 zecke: Who is the main glibc person around here? Feb 28 21:15:15 Glombool, I'm not sure why that is happening Feb 28 21:15:26 hvontres|poodle: pb_ Feb 28 21:15:27 Oh, ok. Feb 28 21:15:30 I thought it was strange. Feb 28 21:15:44 I guess I'll see what happens ... Feb 28 21:15:50 ~seen pb_ Feb 28 21:16:08 * rwhitby is going to use the p-1 phone as an asterisk<->gsm sms gateway permanently connected to power. Feb 28 21:16:13 ...maybe i deleted something on accident. Feb 28 21:16:14 pb_ is currently on #gpe (2d 7h 49m 57s) #oe (2d 7h 49m 57s) #handhelds.org (2d 7h 49m 57s). Has said a total of 10 messages. Is idling for 3h 5m 36s, last said: 'hi zap'. Feb 28 21:16:37 bitbake -c clean doesn't clean deps too right? Just the package you tell it to? Feb 28 21:16:48 zecke: I disabled the cleandir code in package.bbclass Feb 28 21:17:07 zecke: the problems I was seeing disappeared as well Feb 28 21:17:58 Oh, I deleted the tmp/cache folder... Feb 28 21:18:21 Glombool: next time try bitbake -c rebuild Feb 28 21:18:55 hvontres|poodle ah, ok. Works with revisions too then I assume? Feb 28 21:19:03 BTW: anyone used stgit here? Feb 28 21:19:54 * Glombool thinks: did I type that out loud? Feb 28 21:20:25 anyway... gotta run. Thanks for the help. Feb 28 21:25:59 Crofton : Might be of interest -> http://www.em.avnet.com/evs/home/0,1707,RID%253D%2526CID%253D35679%2526CCD%253DUSA%2526SID%253D32214%2526DID%253DDF2%2526LID%253D32233%2526PRT%253D0%2526EFL%253D%2526PVW%253D%2526ASM%253Dl%2526BID%253DDF2%2526CTP%253DEVS,00.html Feb 28 21:27:43 interesting Feb 28 21:27:53 the closest they get to me is about 4 hours Feb 28 21:28:40 Crofton : Well i am not so lucky... probably UK is the safest bet Feb 28 21:29:05 ouch Feb 28 21:29:06 Crofton : at least i will be able to understand the language ;) Feb 28 21:30:48 Crofton : the funny thing is it costs me less to fly to UK than most of the Greek cities Feb 28 21:30:54 yeah Feb 28 21:31:01 airline fares can be strange Feb 28 21:32:23 i was thinking of trying to add the ml403 to the OE targets Feb 28 21:33:12 interesting Feb 28 21:33:32 do you have a board yourself ? Feb 28 21:33:37 no Feb 28 21:34:20 does that have a hard ppc core? Feb 28 21:34:30 soft one Feb 28 21:34:39 can take up to 2 Feb 28 21:34:46 there is an XUP board also with hard core Feb 28 21:37:31 hi Feb 28 21:37:53 haven't done any research on the differences between soft/hard cores Feb 28 21:38:18 well, one uses fpga resources, the hard one is in silicon Feb 28 21:38:37 any differences in speed ? Feb 28 21:39:05 http://www.xilinx.com/univ/xupv2p.html Feb 28 21:39:16 btw i plan to evaluate this also -> http://www.impulsec.com/ Feb 28 21:39:17 I'm not sure about performance differences Feb 28 21:40:08 There is so much to learn Feb 28 21:41:01 yesssss Feb 28 21:41:12 hm Feb 28 21:41:42 XUP industrial customers $1599, academics $299 Feb 28 21:41:42 okay Feb 28 21:41:56 angstroem boots on simpad Feb 28 21:41:58 :) Feb 28 21:42:00 irda segfaults Feb 28 21:42:08 and opie didnt start up Feb 28 21:42:28 hmmm an order of 100 boards is it considers "academic" ? ;) Feb 28 21:43:33 I'm trying to understand why this recipie seems not to package/install. http://www.pastebin.ca/376196 any guidance? Feb 28 21:44:05 so errors seem to be generated, which is confusing Feb 28 21:44:44 what are the errors? Feb 28 21:45:06 why does is depend on tiff and libxml2? Feb 28 21:46:08 ah tiff for fax Feb 28 21:46:17 Crofton: no errors are reported. That's what's confusing Feb 28 21:46:33 the dependencies listed are those used as you say, for fax and tone processing Feb 28 21:46:37 but when you compile packages that need it you get errors? Feb 28 21:46:57 says can't find spandsp.h Feb 28 21:46:57 I think you need to add a staginh task Feb 28 21:47:07 hang on Feb 28 21:47:21 as if it's not installing into staging/rootfs or wherever Feb 28 21:47:28 do_stage () { Feb 28 21:47:28 autotools_stage_all Feb 28 21:47:28 } Feb 28 21:47:36 Add that Feb 28 21:47:41 K Feb 28 21:47:51 is that not automatic as part of autotools? Feb 28 21:48:05 not staging Feb 28 21:48:40 autotools will install into runtime, but not staging Feb 28 21:48:51 staging is where libraries and includes go Feb 28 21:48:59 roughly :) Feb 28 21:51:32 K Feb 28 21:52:42 is packaging of *-dev ipkgs relatively automated or should I assume all the libs will be in spandsp, since in this case all it is, is a library? Feb 28 21:53:09 testing the change now... Feb 28 21:53:11 the runtime libs get automaticaly installed in images Feb 28 21:53:36 K Feb 28 21:53:42 installing the spandsp package should instal lthe libs Feb 28 21:53:53 -dev would pick up headers etc, I expect Feb 28 21:57:51 hillct: Are you building this for asterisk ? Feb 28 21:58:06 for openpbx Feb 28 21:58:12 a fork of asterisk Feb 28 21:58:48 spadsp needs asterisk headers Feb 28 21:59:00 aparently not... Feb 28 21:59:09 I believe it used to... Feb 28 21:59:32 testing now... Feb 28 21:59:43 I'll keep that in mind though. thanks Feb 28 22:04:36 hmm for openpbx you need also php and a bunch of other stuff to work the gui stuff Feb 28 22:06:03 nope Feb 28 22:06:10 you're thinking of FreePBX Feb 28 22:06:19 which is a front end for Asterisk Feb 28 22:06:32 OpenPBX is a fork of asterisk Feb 28 22:07:01 ok, new problem Feb 28 22:07:03 right Feb 28 22:07:10 a fork of asterisk? why? Feb 28 22:07:29 gpl disputes... Feb 28 22:07:31 for better open hardware support Feb 28 22:07:33 clearner code Feb 28 22:07:45 FAR superior config file format Feb 28 22:07:51 and yes, GPL issues Feb 28 22:07:57 Ifaistos: I see political crap :) Feb 28 22:08:03 somewhat Feb 28 22:08:21 but also the code looks a lot better Feb 28 22:09:26 seen this error when trying to put together other recipes Feb 28 22:09:37 when compiling OpenPBX, I see: Feb 28 22:09:39 CROSS COMPILE Badness: /usr/include in INCLUDEPATH: /usr/include/openssl Feb 28 22:09:39 cc1: internal compiler error: in add_path, at c-incpath.c:362 Feb 28 22:10:12 chouimat: Digium that developed asterisk requires developers to sign an agreement that their code can be used in the non-gpl version of asterisk Feb 28 22:10:12 can I get around this by simply setting the enviroment variable INCLUDEPATH ? Feb 28 22:10:15 or is this something deeper? Feb 28 22:10:23 chouimat: A lot of people did not like that Feb 28 22:10:44 hillct, maybe configure script includes hard coded paths Feb 28 22:10:49 K Feb 28 22:10:58 will check it out Feb 28 22:11:00 hillct: Can you post the .bb file Feb 28 22:11:03 this is bad for cross compiling Feb 28 22:11:10 Ifaistos: sure. one sec Feb 28 22:11:12 Ifaistos: it's their right (both group) personnaly if I don't want to sign such thing I will not release my code :) Feb 28 22:16:18 Ifaistos: http://www.pastebin.ca/376254 Feb 28 22:16:34 note the patch needs to be separated out at the bottom Feb 28 22:22:41 ok Feb 28 22:22:50 what arch are you trying to build it ? Feb 28 22:24:54 ~lart building OE using nfs disks Feb 28 22:24:54 * ibot wallops building OE using nfs disks with a main rotation server that needs rehubbing. It won't take long Feb 28 22:35:35 Ifaistos: ARM Feb 28 22:36:29 using SlugOS/BE Feb 28 22:38:42 hmmm by builds fail at libpcap Feb 28 22:40:39 I had that issue when building against slugos-3.10-beta but against HEAD that was not an issue Feb 28 22:40:57 seemed to be related to capability.h Feb 28 22:42:22 i am using gcc 4.1.1 glibc 2.5 on x86 Feb 28 22:49:47 hi Feb 28 22:50:55 Hmm, anoncvs.gnome.org seems to have moved which breaks libmimedir Feb 28 22:51:24 alphaone: hehe, cvs->svn Feb 28 22:52:16 Yeah, I just read it. Feb 28 22:53:01 hillct: i'll look at it tomorrow. Feb 28 22:53:09 night everyone Feb 28 22:53:16 night Ifaistos Feb 28 22:53:32 night Ifaistos Feb 28 22:56:16 zecke: You want me to submit a bug report? Feb 28 22:56:33 alphaone: probably, best with a patch attached :) Feb 28 22:56:54 what's the name of the variable that contains the target platform? I'm trying to point to the target platform staging directory Feb 28 22:57:41 zecke: I'm already looking into it :-) Feb 28 23:00:28 zecke: The code now lies in /svn/libmimedir/trunk What is the best way to handle that in the .bb file? Feb 28 23:01:12 alphaone: Just change the path in the bb file? Feb 28 23:01:13 alphaone: is that trying to fetch a specific revision, or always fetches the latest/HEAD? Feb 28 23:01:21 Do I just change S ? Feb 28 23:01:50 alphaone: http://bitbake.berlios.de/manual/ch03s06.html Feb 28 23:02:04 alphaone: you change the SRC_URI from the cvs protocol to svn and adjust Feb 28 23:02:09 k, thanks Feb 28 23:02:24 It is fetching HEAD Feb 28 23:02:50 Otherwise there would be rev= in the SRC_URI, right? Feb 28 23:40:43 zecke: I'm not awake enough to add a patch right now. I will look into it tomorrow. Feb 28 23:46:44 is the mtn server down? Feb 28 23:48:33 Crofton|home: use one of the mirrors ;) Feb 28 23:49:30 it pulled eventually Feb 28 23:52:12 03crofton 07org.oe.dev * r9523a6c4... 10/ (1 packages/ossie/ossie-usrp-tx-control_svn.bb): ossie-usrp-tx-control_svn.bb : Update path in subversion repository. Feb 28 23:56:44 03crofton 07org.oe.dev * r7d4cc290... 10/ (1 packages/ossie/ossie-usrp-tx-control_svn.bb): ossie-usrp-tx-control_svn.bb : Fix path in ossie subversion server. Mar 01 00:10:58 heh same change pushed from two different machines Mar 01 00:11:21 I do not think mtn went insane :) Mar 01 00:25:14 ALright, I can't seem to figure out the .bb file format. I'm trying to get it to look at a local directory for the source package. What do I need to add to the .bb file? Mar 01 00:26:59 SRC_URI = "file://blahh.tar.gz" Mar 01 00:27:09 place blahh.tar.gz in blahh/files/ Mar 01 00:27:30 ah, ok Mar 01 00:27:36 thanks Mar 01 00:27:51 np Mar 01 00:27:51 was the bitbake manual not clear enough about teh syntax? Mar 01 00:28:03 nope Mar 01 00:28:11 i read it :) Mar 01 00:28:55 http://bitbake.berlios.de/manual/ch03s02.html Mar 01 00:29:03 page titled "Local File Fetcher" Mar 01 00:29:12 what about that isnt obvious? Mar 01 00:29:42 That's not the one I read. Mar 01 00:29:50 That one is obvious :) Mar 01 00:48:49 Glombool, which one did you read? Mar 01 00:50:44 gah. build still broken. i'm beginning to think i've installed something (on gentoo) that is somehow borking the run.image_cmd... phase of my image. ring a bell with anyone? Mar 01 00:51:31 HopsNBarley: well, pastebin your error and let us see ;) Mar 01 00:55:09 http://www.pastebin.ca/376466 Mar 01 00:56:16 i think i've backed all the way up to before where this started... but it persists. Mar 01 01:12:29 Crofton: (late reply) I read info on the Opie site Mar 01 01:21:35 HopsNBarley: wierd indeed Mar 01 01:34:09 03coredump2 07org.oe.dev * r77c7c9f8... 10/ (1 classes/rootfs_ipk.bbclass): (log message trimmed) Mar 01 01:34:09 rootfs_ipk.bbclass: Implements a sensational new feature called 'oe-feed' Mar 01 01:34:09 which, for the first time ever, actually lets you use your ipk deploy dir as a Mar 01 01:34:09 feed in ipkg.conf./mtcommit ../classes/rootfs_ipk.bbclass Oh wait......we Mar 01 01:34:09 already could do that in the past, now couldn't we?? Set DEPLOY_ENABLE_OEFEED=1 Mar 01 01:34:11 in your shell env or local.conf to activate fast oe-feed/Packages creation, set Mar 01 01:34:13 DEPLOY_ENABLE_OEFEED=2 to enable traditional (and slooooow) oe-feed/Packages Mar 01 02:29:41 Alright, I read the bitbake manual, I still don't understand why my .bb file is not working properly. Mar 01 02:30:07 I have a package I want to compile in /home/user/sources/package.tar.gz Mar 01 02:30:45 The .bb file has FILESPATH = "/home/user/sources" and Mar 01 02:31:03 SRC_URI= "file://package.tar.gz" Mar 01 02:31:24 No matter what I do, it still tries to find the package through ftp Mar 01 02:32:59 anyone still here? Mar 01 02:41:57 You need to read the bitbake manual. There should be something about setting up a local source area there. Mar 01 02:44:55 Glombool: as I said, mkdir PACKAGE_NAME/files ; cp sources.tar.gz PACKAGE_NAME/files. Then SRC_URI = "file://source.tar.gz" Mar 01 02:45:11 Okay... i know. Mar 01 02:45:13 hold on Mar 01 02:45:17 Glombool: see altboot/altboot_0.0.0.bb as an example Mar 01 02:45:27 Alright, thank you. I'll look at it. Mar 01 02:46:04 The question. the dir I make. Where do I need to place it though. In the sources folder? Mar 01 02:47:12 nahh Mar 01 02:47:28 packages/$YOUR_PACKAGE_NAME/ Mar 01 02:47:39 ok Mar 01 02:47:44 like packages/altboot/files Mar 01 02:47:52 Right Mar 01 02:47:54 That makes sense Mar 01 02:50:11 03coredump2 07org.oe.dev * r6a45fed3... 10/ (1 classes/rootfs_ipk.bbclass): rootfs_ipk.bbclass: Work around PATH'ing problems in other image-related bbclasses and / or .bb's Mar 01 02:51:09 (unb0rks slugos-image again) Mar 01 02:55:41 GRRR! It's still trying to download from ftp.handhelds.org for some reason. Mar 01 02:57:18 Glombool: perhaps there's a misunderstanding -- are you trying to add your own files in addition to the normal sources for the package? Or are you trying to have the package select your source *instead* of the normal sources for the package? Mar 01 02:57:38 I'm adding my own files. Mar 01 02:57:58 And it's trying to download your files from ftp.handhelds.org? Mar 01 02:58:05 Yup Mar 01 02:58:53 Here's what I'm doing... I hacked the gpe-soundbite source so it will work properly on my collie. Mar 01 02:59:18 I gave it a different name... gpe-soundbite-collie **** ENDING LOGGING AT Thu Mar 01 02:59:57 2007