**** BEGIN LOGGING AT Thu Aug 30 02:59:57 2007 Aug 30 05:09:58 * * OE Bug 2895 has been created by autobuild(AT)openembedded.org Aug 30 05:10:00 * * angstrom-x11-gpe-image-1.0-r0-do_rootfs Aug 30 05:10:02 * * http://bugs.openembedded.org/show_bug.cgi?id=2895 Aug 30 05:11:58 * * OE Bug 2896 has been created by autobuild(AT)openembedded.org Aug 30 05:12:00 * * angstrom-x11-pimlico-image-1.0-r0-do_rootfs Aug 30 05:12:02 * * http://bugs.openembedded.org/show_bug.cgi?id=2896 Aug 30 05:38:35 mickey|ICE676: should task-openmoko-feed actually be meta-openmoko-feed, so that BUILD_ALL_DEPS is set to 1? Aug 30 05:38:58 rwhitby: yes, i think you're correct. feel free to change Aug 30 05:39:21 will do Aug 30 05:39:26 thanks Aug 30 05:44:25 mickey|ICE676: meta/meta-openmoko-feed.bb or just meta/openmoko-feed.bb ? Aug 30 05:44:42 (I prefer the latter) Aug 30 05:59:58 ya, go for the latter Aug 30 06:00:55 mickey|ICE676: I made the meta package build task-openmoko-debug, task-openmoko-native-sdk and task-openmoko-feed Aug 30 06:01:23 (so it wasn't a change, but an addition of a meta package to build the tasks) Aug 30 06:03:14 ok, cool Aug 30 06:03:38 that seemed to be the way that the other meta packages did it. Aug 30 06:03:50 (I should change slugos-packages over to that too) Aug 30 06:04:07 yeah, consistency good Aug 30 06:17:37 03rwhitby 07org.oe.dev * r4a5dce36... 10/ (1 packages/meta/openmoko-feed.bb): openmoko-feed: Added meta package to actually build all the packages in task-openmoko-* Aug 30 06:18:40 mickeyl_: there it is Aug 30 06:18:55 yay Aug 30 06:25:19 mickeyl_: bitbake openmoko-feed now does the right thing Aug 30 06:28:13 nice! Aug 30 06:28:52 morning Aug 30 06:51:38 mickeyl_: as I said, you want meta-openmoko with meta.bbclass Aug 30 06:52:23 mickeyl_: and if openmoko@oe.org isn't going to use --author on commits is is going to get blocked Aug 30 06:54:25 so you would rather see a fork instead of acting more diplomatic? Aug 30 06:54:47 I asked twice already Aug 30 07:12:01 gm Aug 30 07:13:31 good morning all Aug 30 07:13:37 rwhitby, mickey|zzZZzz: BUILD_ALL_DEPS doesn't do anything with bitbake 1.8.x Aug 30 07:13:41 morning all Aug 30 07:15:12 hey RP Aug 30 07:15:49 RP: doesn't meta.bbclass set a rdeptask, does it "just work" because of debian.bbclass? Aug 30 07:16:22 koen: Yes, meta.bbclass has some tasks dependencies that make it work Aug 30 07:16:46 I should go and remove all BUILD_ALL_DEPS references... Aug 30 07:16:52 koen: Would you take care of my eglibc patch ? Aug 30 07:19:34 khem: yes, after breakfast :) Aug 30 07:20:02 koen: Cool ! I gotta go to bed now . g'night Aug 30 07:20:10 'night Aug 30 07:42:58 * * OE Bug 2894 has been RESOLVED (FIXED) by Aug 30 07:43:00 * *  eglibc should provide glibc packages when chosen as default libc Aug 30 07:43:02 * * http://bugs.openembedded.org/show_bug.cgi?id=2894 Aug 30 07:46:53 03rpurdie 07org.oe.dev * rc6c228c0... 10/ (6 files in 2 dirs): classes: Remove bitbake 1.6 compatibility code Aug 30 07:47:00 03raj.khem 07org.oe.dev * rd0ad1bdf... 10/ (4 files in 3 dirs): eglibc: (r)provide glibc stuff, since it is compatible, closes #2894 Aug 30 07:55:03 yay, berlios down again Aug 30 08:03:44 morning Aug 30 08:09:15 morning xora Aug 30 08:09:35 hey XorA Aug 30 08:11:15 oooh, a good bitchslapping fest :-) Aug 30 08:12:04 steliosk: Can you give me an advice please, which gcc version is the best to compile a linux ppc kernel. I used gcc-3.4.4 (Kernel works) and gcc-4.1.1 (Kernel didn't boot) so far. Aug 30 08:12:57 cyberdeck : gcc 4.1.1 from OE ? Aug 30 08:12:58 cyberdeck: what about 4.1.2 or 4.2.1? Aug 30 08:14:04 steliosk: i used the gcc from OE and versions forced with (PREFERRED_VERSION_{gcc-cross,cross-initial} = "4.1.1"). Aug 30 08:14:15 koen: didn't test them yet. Aug 30 08:15:19 RP: re: BUILD_ALL_DEPS: yep, that's why I went with inheriting meta, and making it a meta package which built other tasks, Aug 30 08:15:20 cyberdeck : i am using 4.1.1 with 2.6 kerneks on ppc405 without problems for sometine now Aug 30 08:16:10 (I'm assuming meta packages only building tasks is stylisticly preferred) Aug 30 08:16:24 steliosk: did you know if never versions will work / won't work? Aug 30 08:16:47 rwhitby: single apps is fine as well Aug 30 08:17:54 koen: so openmoko-feed.bb should subsume task-openmoko-feed (since task-openmoko-feed does nothing more than just add things to task/openmoko-feed) ? Aug 30 08:18:31 (openmoko-feed would still just refer to any other useful openmoko (or any other) tasks) Aug 30 08:18:35 cyberdeck : We tried 4.1.2 but haven't test it extensivly. There are some issues with float point and that we stayed with 4.1.1 Aug 30 08:19:04 cyberdeck : which kernel did you try to build ? Aug 30 08:19:14 morning Aug 30 08:19:36 cyberdeck : btw you can have a different gcc version for kernel and for the rest of the build Aug 30 08:19:54 rwhitby: meta packages are pretty much used by distro maintainers only, so the contents (tasks vs packages) doesn't matter much Aug 30 08:20:28 insert 'only' somewhere above Aug 30 08:20:31 steliosk: I used the secretlabs.ca git for a 2.6.22 kernel Aug 30 08:20:52 cyberdeck : add something like that to the kernel KERNEL_CCSUFFIX = "-3.4.4" Aug 30 08:21:43 steliosk: so i can use REFERRED_VERSION_{gcc-cross,cross-initial} = "4.1.1" and KERNEL_CCSUFFIX = "-3.4.4" both at once? Apps built with 4.1.1 and Kernel with 3.4.4? Aug 30 08:21:58 cyberdeck : yes Aug 30 08:50:56 anyone having problems with recent mtd-utils ? - http://rafb.net/p/KO5KrU53.html Aug 30 08:58:53 steliosk: testing Aug 30 09:02:32 RP : What does that mean ? -> NOTE: Task failed: Unknown fetch Error: [Errno 37] No locks available Aug 30 09:02:45 good morninggood morning Aug 30 09:02:48 nfs ? Aug 30 09:02:49 woops Aug 30 09:03:08 koen : yes Aug 30 09:03:10 florian: good mornigng^2 to you to! Aug 30 09:04:18 koen: :-) Aug 30 09:07:47 steliosk: lack of statd on NFS server Aug 30 09:15:05 but it was working fine till 5 minutes ago.... Aug 30 09:15:32 steliosk: because you did not had anyting new to fetch? Aug 30 09:16:22 steliosk: mtd-utils-native 1.0.0+git fails here too Aug 30 09:16:23 hrw : i'll check, because its pulling git stuff there also Aug 30 09:17:06 i think in general we should use stable versions by default Aug 30 09:17:38 pulling from dev branches has issues Aug 30 09:20:11 mtd-utils-native-1.0.0+git-r4 fails to build for me when trying to do a TOSA build Aug 30 09:26:31 another happy OE customer :) http://developer.irexnet.com/pub/ Aug 30 09:26:50 yes Aug 30 09:28:41 RP: can SRCREV fetch errors (e.g berlios being down) be taught to raise skippackage instead of raising a parse error? Aug 30 09:29:21 NOTE: package mtd-utils-native-1.0.0+git-r4: task do_build: completed Aug 30 09:30:13 need to check does will it also work in OE Aug 30 09:30:42 hrw: it seems for tosa it attempts to build ./src/ubiupdatevol.c but not for other machines Aug 30 09:31:19 XorA: mtd-utils-native does not depend on machine rather Aug 30 09:31:45 hrw: something different between tosa and neo/terrier triggers the problem though Aug 30 09:32:52 make -C ./ubi-utils is called when I build for tosa, not for other machines Aug 30 09:32:54 steliosk, XorA: http://pastebin.ca/675589 please test Aug 30 09:33:12 its ugly Aug 30 09:33:27 or is this a case of a floating git Aug 30 09:33:32 with no proper $PV Aug 30 09:35:11 hrw: NOTE: package mtd-utils-native-1.0.0+git: completed Aug 30 09:37:43 anyone from the core devs worked on the iliad tablet ? Aug 30 09:41:18 steliosk: I dont think so, but we all have drooled over it Aug 30 09:41:45 mickeyl bought one a while ago iirc Aug 30 09:41:57 03koen 07org.oe.dev * re70f7117... 10/ (4 files in 2 dirs): angstrom-x11--image: fix basename Aug 30 09:42:01 03koen 07org.oe.dev * r64776f34... 10/ (1 packages/mtd/mtd-utils_1.0.0+git.bb): mtd-utils: lock down git rev Aug 30 09:42:22 ~prod mickey|zzZZzz Aug 30 09:42:23 * ibot zaps mickey|zzZZzz with a high voltage cattle prod Aug 30 09:42:35 it seems I forgot to push that mtd patch.... Aug 30 09:43:41 * XorA wants an iliad for RPG Aug 30 09:44:01 hmmm 433 inc VAT Aug 30 09:44:09 * koen wants an iliad for college work Aug 30 09:44:18 koen: liar, you do no work :-) Aug 30 09:44:31 it's still holiday over here :) Aug 30 09:44:39 for 4 more days :) Aug 30 09:45:55 man, this is the second day of wikipedia being in the dutch news because the royal family vandalised it Aug 30 09:46:00 slow news week Aug 30 09:46:12 college work is a contradiction in terms :) Aug 30 09:46:36 wassup with the server guys? Aug 30 09:46:36 svn: Can't connect to host 'svn.berlios.de': Connection refused Aug 30 09:47:37 down Aug 30 09:48:36 darn Aug 30 09:52:35 [09:55] koen: yay, berlios down again Aug 30 09:52:49 lets see if it takes them another 2 days to get it fixes Aug 30 09:52:53 fixed Aug 30 09:53:45 and then one day we will do org.openembedded.bitbake :( Aug 30 10:27:25 cp Angstrom-x11-image-glibc-test-20070830-tosa.rootfs.jffs2 Angstrom-x11-image-glibc-test-20070830-tosa-installkit/initrd.bin Aug 30 10:27:25 cp: cannot stat `Angstrom-x11-image-glibc-test-20070830-tosa.rootfs.jffs2': No such file or directory Aug 30 10:27:42 hmm, tosa installkit doesnt work because file is called *.jffs2.bin Aug 30 10:28:37 +++ conf/machine/include/zaurus-2.6.inc 139d86a1bf9e6a09586996371439ca744c3a4c8e Aug 30 10:28:38 -IMAGE_CMD_jffs2 = "mkfs.jffs2 -x lzo --root=${IMAGE_ROOTFS} --output=${DEPLOY_DIR_IMAGE}/${IMAGE_NAME}.rootfs.jffs2.bin ${EXTRA_IMAGECMD}" Aug 30 10:28:38 +IMAGE_CMD_jffs2 = "mkfs.jffs2 -x lzo --root=${IMAGE_ROOTFS} --output=${DEPLOY_DIR_IMAGE}/${IMAGE_NAME}.rootfs.jffs2 ${EXTRA_IMAGECMD}" Aug 30 10:30:43 koen: cheers, I thought I was searching for a tosa special case and overlooked that Aug 30 10:35:26 XorA: did you try bluez-gnome 0.14 yet? Aug 30 10:36:09 koen: no, Ive not had a device working for ages Aug 30 10:37:29 koen: does it render upwards yet? Aug 30 10:37:40 haven't tried that Aug 30 10:37:53 I was hindered by the openmoko tap-n-hold screw up for days Aug 30 10:40:31 * XorA has not bar to hold icons on openmoko anyway Aug 30 10:40:37 s/not/no/ Aug 30 10:45:41 03xora 07org.oe.dev * r249c76d0... 10/ (1 conf/machine/include/zaurus-2.6.inc): machine/include/zaurus-2.6.inc : dont put .bin on end os jffs2 images Aug 30 10:48:34 03xora 07org.oe.dev * r249c76d0... 10/ (1 conf/machine/include/zaurus-2.6.inc): machine/include/zaurus-2.6.inc : dont put .bin on end os jffs2 images Aug 30 10:49:05 ~lart CIA-17 Aug 30 10:49:06 * ibot hurls dozens of incontinent, insomniac, hungry kittens with tiny little razor-sharp claws and a wide variety of contagious intestinal parasites at CIA-17 Aug 30 10:49:28 XorA: zaurus images are not jffs2 Aug 30 10:49:50 they are prepended with 16bytes header Aug 30 10:49:53 header + jffs2 Aug 30 10:50:07 hrw: but the clamshells set it to .jffs as well Aug 30 10:50:08 03xora 07org.oe.dev * r249c76d0... 10/ (1 conf/machine/include/zaurus-2.6.inc): machine/include/zaurus-2.6.inc : dont put .bin on end os jffs2 images Aug 30 10:50:17 so at least now it's consistent Aug 30 10:50:22 ok Aug 30 10:50:27 so do we reverse that change and fix install kit Aug 30 10:50:36 ah, too late Aug 30 10:51:12 03xora 07org.oe.dev * r249c76d0... 10/ (1 conf/machine/include/zaurus-2.6.inc): machine/include/zaurus-2.6.inc : dont put .bin on end os jffs2 images Aug 30 10:51:21 CIA-17: SHUT UP Aug 30 10:51:23 03koen 07org.oe.dev * rf7b8feca... 10/ (1 conf/machine/include/zaurus-2.6.inc): zaurus-2.6.inc: consistently suffix jffs2 images with jffs2 Aug 30 10:54:06 * koen wonders why people attempt to cross compile stuff outside of OE for angstrom without checking the feeds or even using OE Aug 30 10:55:18 what feeds? Aug 30 10:55:46 http://www.angstrom-distribution.org/ and click on "package browser" Aug 30 10:59:45 Yesterday I had posted asking for suggestions on handhelds good for Linux hacking and I got several good recommendations, however after checking them out I realized none of them had a USB Host port, does anyone know of a nice handheld that has a USB Host? Aug 30 11:00:28 the new zaurus machines have one Aug 30 11:00:57 hey zecke Aug 30 11:01:17 ho Aug 30 11:01:24 I'm soo hungry :} Aug 30 11:06:08 * florian hands zecke a cookie Aug 30 11:06:35 cookies box is what I consider for alix case Aug 30 11:06:59 http://www.epiacenter.com/modules.php?name=Sections&sop=viewarticle&artid=11 like that Aug 30 11:07:19 but without lcd Aug 30 11:07:22 Koen, Aug 30 11:07:55 Koen, what model number is that for the Zaurus? Aug 30 11:08:29 sl-c1000 and sl-c3xxx Aug 30 11:08:39 the 3xxx ones also have a microdrive Aug 30 11:08:49 (akita and borzoi in OE speak) Aug 30 11:08:55 (I don't have any) Aug 30 11:09:38 koen, the SL-C1000 looks nice, I think the sl-c3xxx is overkill with the hardrive for what I'm looking for, thanks for the info. Aug 30 11:10:06 morning all Aug 30 11:11:57 koen : for the fetch error what did you say is the problem ? Aug 30 11:12:17 lack of statd on NFS according to hrw Aug 30 11:13:13 3700 ? Ss 0:00 /sbin/rpc.statd --port 32765 --outgoing-port 32766 Aug 30 11:13:45 stad seems to be running Aug 30 11:15:30 we did not have such a problem before, since we are using nfs storage for the sources Aug 30 11:15:37 from the begging Aug 30 11:16:30 steliosk: yes, but recent bitbake use locking Aug 30 11:16:36 hey ppl. Aug 30 11:16:51 steliosk: to control several bitbake instances not fetc the same Aug 30 11:17:22 hrw : but stad is running on the nfs server Aug 30 11:17:24 I am using nfsv3 for few months to store DL_DIR. from some time I have to run statd for it too Aug 30 11:17:43 steliosk: then I have no idea - my knowledge about nfs is very small Aug 30 11:18:42 RP : Any idea why we get this error ? NOTE: Task failed: Unknown fetch Error: [Errno 37] No locks available Aug 30 11:18:56 * hrw -> afk Aug 30 11:20:56 morning all Aug 30 11:23:44 if I change s.th. in the machine description, say add a package, how do I make shure bitbake includes the new package during next build of the image ? Aug 30 11:25:04 I am sking because I have noticed that bitbake just creates a new image w/o adding the new package. Aug 30 11:27:47 stupid question ? If its in the manual, please could anyone point me to the location ? Aug 30 11:28:09 I'd say many idlers. Aug 30 11:29:02 jan_js: it depends which parts you change Aug 30 11:29:17 jan_js: rebuild task-boot task-base ipkg-collateral first - then image Aug 30 11:29:58 I have added a package required for this machine via MACHINE_EXTRA_RDEPENDS. Aug 30 11:30:50 hrw: Thanks anyway ! Aug 30 11:32:12 rwhitby: can mokomakefile be used like "make update; make setup; bitbake openmoko-devel-image" so we can switch most of users of it to use bitbake directly instead of magic like 'make package-rebuild-umbaumba;make package-umba-nfs-utils' way? Aug 30 11:32:42 the makefiles should stop its scope at 'make setup' Aug 30 11:33:14 koen: "make update" can be good too Aug 30 11:33:26 and ofcourse "make update-makefile" :) Aug 30 11:33:45 right Aug 30 11:33:50 hrw: your usecase is exactly how I use MokoMakefile Aug 30 11:33:54 hrw: here's the thing. if you can work out a way that I can move from on OE build directory to another, and just run bitbake, without having to remember to source a setup file first for that directory, then I'll consider it. Aug 30 11:34:31 make environment Aug 30 11:34:44 XorA: make environment can't affect the parent environment Aug 30 11:35:05 rwhitby: but you can fork into a shell in the correct directory Aug 30 11:35:42 XorA: but you never come out again. Not clean enough. Aug 30 11:35:49 rwhitby: it was just an idea, no proper research behind it Aug 30 11:36:11 That "source setup-env" is the only reason why I (personally) wrap bitbake commands. Aug 30 11:36:26 rwhitby: in local.conf you need to set TMPDIR thats sure Aug 30 11:36:29 "Because our metadata was indeed bound to embedded, it was renamed was well." Aug 30 11:36:35 at http://www.openembedded.org/ Aug 30 11:36:57 make bitbake-gui :-D Aug 30 11:36:58 oh its an old message Aug 30 11:37:13 heh, news in reverse order, just got me confused Aug 30 11:37:15 :) Aug 30 11:37:28 The correct way to tell we have a fix for a bug is making it blocking 2194, right ? Aug 30 11:38:04 (after attaching the patch of course) Aug 30 11:38:05 My personal build philosophy is that you should be able to cd into a directory, and type a single command to build. If that single command is bitbake, fine. At the moment, it's two commands: source setup and then bitbake. That's why I put a convenience target in the makefile. Aug 30 11:38:35 openwrt, for instance, meet's the criteria of cd into a dir and type a single command to build. Aug 30 11:39:31 nodens: really? I submitted a patch yesterday but did not know about the blocking thing, where did you read that? Aug 30 11:39:37 so in fact BBPATH is the problem? Aug 30 11:40:21 Jin^eLD, on http://bugs.openembedded.org/show_bug.cgi?id=2194 :) Aug 30 11:40:38 but as I see it nowhere else, I ask here. Just to be sure :) Aug 30 11:41:25 rwhitby: so let 'make setup' calls "bash --init-file setup-env"? Aug 30 11:45:18 hrw: no good. I don't run make setup before each build Aug 30 11:45:26 don't even run make update before some builds Aug 30 11:45:55 can I just ignore mokomakefile users? Aug 30 11:46:04 so I'm very open to discussing how to change bitbake or OE to meet that criteria, and if that criteria is met, then I fully agree that the single bitbake command should not be hidden behind a single makefile command Aug 30 11:46:07 but at the moment, the single makefile command wraps two other commands (source setup-env ; bitbake ...) Aug 30 11:46:32 I don't see why you need to make it an us and them situation? Aug 30 11:46:48 for me it is three commands: cd ~/devel; . setup.sh build/angstrom;bitbake Aug 30 11:47:20 Seriously, do you think you would be getting *less* new-openmoko-end-user questions without mokomakefile? Aug 30 11:47:58 rwhitby: I am trying to get best of both - mokomakefile to setup/update + bitbake to build Aug 30 11:48:28 hrw: so am I. let's talk about how to make it a single command rather than your three. Aug 30 11:48:42 rwhitby: now I cant tell them 'bitbake this' because most of them does not have any idea what bitbake is and how to use it Aug 30 11:49:06 rwhitby: it cant be done in one command Aug 30 11:49:13 if you only tell them "bitbake this", then they will forget to source setup-env and get it wrong anyway Aug 30 11:49:24 need BBPATH to default to . and then conf/local.conf to do BBPATH="REALBBPATH" Aug 30 11:49:33 "letsbuild nfs-utils" will not work. Aug 30 11:50:05 looks like my firefox on fedora 7 is missconfigured, because on openembedded.org i see: eg at Documentation Development ... or is there a problem on the page/wiki ? Aug 30 11:50:28 as letsbuild command need to know where bitbake/oe are, where to build, how to update and then run bitbake Aug 30 11:50:28 s/openembedded.org/www.openembedded.org Aug 30 11:50:58 hrw: "make build-package-foo" is one command. It's guaranteed to run from the correct place (cause the Makefile is only in one place). Aug 30 11:51:37 rwhitby: make prepatch-linux-gta01-2.6.21.6 works too? Aug 30 11:52:26 grma: I either dont understand what you mean or it doesnt happen Aug 30 11:52:29 hrw: nope, but most openmoko application developers never need that command. And that's probably the crux of the problem. OE (a distro build system) is being required to be used by openmoko application developers. Aug 30 11:52:58 I also would prefer openmoko guys to ship external toolchain Aug 30 11:53:02 XorA: yes, i think my firefox on fedora7 is mssconfigured, because i see html tags on the page ... Aug 30 11:53:09 especially when OE can generate one easily Aug 30 11:53:24 the guy who is writing a social networking gui application should *never* need to know a bitbake command line argument. Aug 30 11:53:40 make external-toolchain :-) Aug 30 11:53:42 grma: drupal on OE site is missconfigured too. Aug 30 11:54:00 rwhitby: such guys should not touch OE at all Aug 30 11:54:08 to build own app Aug 30 11:54:27 hrw: tell rwhitby how to build external toolchain, that make a good addition to MokoMakefile Aug 30 11:54:31 hrw: correct, but they are being forced to *right* *now*. hence mokomakefile exists Aug 30 11:54:35 grma, on /usr/local/bin:/usr/local/sbin:/usr/bin:/usr/sbin:/bin:/sbin Aug 30 11:54:37 oups Aug 30 11:54:44 grma, on http://www.openembedded.org/wiki/OeFaq ? Aug 30 11:55:00 seems there is a module (html inclusion or something) missing Aug 30 11:55:19 nodens: yes, there i also see html tags Aug 30 11:55:30 grma: nodens ah I see now thanks Aug 30 11:55:57 looks like input changed from Filtered HTML to Text for that Page Aug 30 11:56:04 XorA: iirc in OE it was 'bitbake meta-toolchain' Aug 30 11:57:11 ofcourse for openmoko it needs to be extended to ship openmoko libs with all deps Aug 30 11:57:50 * XorA sees more items on OM TODO list Aug 30 11:59:46 * hrw has a luck to not having access to OM TODO list Aug 30 12:00:52 hrw: I think its like a black hole, you can keep putting stuff in, but you cant see what is there :-) Aug 30 12:01:15 anyway guys, that's my rationale behind mokomakefile. you can see that it's not some evil plot to hide bitbake commands or trivialise the build process. there is a real criteria which mokomakefile (and other build systems like openwrt) meets - one command to build from scratch. I'm very open to talking about how to meet the criteria. Please don't ask me to remove the criteria, cause it makes very good sense to my usage. Aug 30 12:02:03 I will probably wait for openmoko maintainer. Aug 30 12:02:21 in mean time I have some texts to write Aug 30 12:02:21 I'm also very open to putting the real bitbake commands alongside the mokomakefile commands in any and all documentation, so people who do want to learn how the build process works can do so. Aug 30 12:03:25 hrw: thx for the discussion. it's always good to discuss these things rationally and understand the motivations. Aug 30 12:03:26 we need fuzzy target matching in OE Aug 30 12:03:26 the problem is that someone started writing openmoko-use-oe docs in suc way that it breaks too many rules Aug 30 12:03:56 ~curse the one which suggested DISTRO_EXTRA_RDEPENDS in local.conf Aug 30 12:03:56 May you be reincarnated as a Windows XP administrator, the one which suggested DISTRO_EXTRA_RDEPENDS in local.conf ! Aug 30 12:04:09 hrw: mokomakefile simply automated the original build instructions, which I assumed mickey|zzZZzz wrote. Aug 30 12:04:50 * hrw -> afirc Aug 30 12:04:51 hrw: mokomakefile didn't cause the DISTRO_EXTRA_RDEPENDS fiasco. Aug 30 12:05:06 I know Aug 30 12:05:19 somewhere, the right OE documentation didn't get to the application developers who needed to know how to build their apps. Aug 30 12:05:56 but DISTRO_EXTRA_RDEPENDS + -crebuild == -cclean+-cbuild with ignoring deps === om users problems Aug 30 12:05:58 So OE developers (including myself) should just fix the places where there is wrong information and point to (or write) the tutorials that say how to do it properly, Aug 30 12:06:00 the missing step is actually how to make a feed, a lot of the applications programmers think the only way to install software is to force it into an image Aug 30 12:06:14 * hrw bb in 10-15 Aug 30 12:06:54 instead, we have OE developers insulting application developers for the simple reason that they have no desire to become build maintainers (cause their desire is to be application developers) Aug 30 12:07:51 calling them stupid and clueless, when they are quite clearly not so, cause they are writing applications rather than building build systems. Aug 30 12:16:24 yep Aug 30 12:18:01 * zecke appreciates the makefile... Aug 30 12:27:05 stefan_schmidt: please poke mickeyl Aug 30 12:31:33 and the reverse om users insulting OE devs that just say "that page is wrong" isn't helping either Aug 30 12:33:03 koen: agreed Aug 30 12:34:01 anyway, it's my wedding anniversary tomorrow, so I can't stay up late on IRC. night all. Aug 30 12:34:09 * koen fixes some more bugs in the oe website Aug 30 12:34:12 rwhitby: congratulations Aug 30 12:34:23 thx. 11 years. Aug 30 12:36:24 * koen seriously wants PackageKit support in angstrom Aug 30 12:36:45 dbus abstraction layer ? Aug 30 12:36:56 yes Aug 30 12:36:56 koen: if you have some time, maybe take a look at 2892, I attached a patch to fix the imagemagick problems that I was reporting yesterday Aug 30 12:37:07 just some stuff in the .bb file was missing Aug 30 12:37:33 autotools_stage_all didn't work? Aug 30 12:37:47 tried that as well but did not work for me for some reason Aug 30 12:38:02 and also there is a line telling which modules go into the package Aug 30 12:38:07 i.e. the module-Q16 stuff Aug 30 12:38:16 and there - if the *.la are missing it just does not work Aug 30 12:38:25 koen, looks neat. Aug 30 12:38:40 only *.so was installed in the original .bb, and there convert was not able to find any formats when testing on the target system Aug 30 12:38:49 can't be worse than gpe package manager anyway ;) Aug 30 12:39:01 further, binconfig was missing - so Wand-config/Magick-config were not staged Aug 30 12:41:11 * cbrake suspects its past time to update his blog entry on the _correct_ way to add applications to OE ... Aug 30 12:41:48 like http://blog.haerwu.biz/2007/08/23/why-using-of-distromachine-variables-in-localconf-is-wrong/ ? Aug 30 12:42:01 * koen hugs ANGSTROM_EXTRA_INSTALL Aug 30 12:43:52 koen: yup Aug 30 12:44:57 Jin^eLD: testing the patch irght now Aug 30 12:45:06 koen: thx Aug 30 12:45:06 damn, 30M of u-boot crap Aug 30 12:45:30 and a gazillion lowlevel foo files in deploy Aug 30 12:47:38 ~lart svn.berlios.de Aug 30 12:47:38 * ibot hits svn.berlios.de with an anvil and laughs with a contralto voice ... Haha Ha HA Ha Aug 30 12:51:57 * * OE Bug 2892 has been RESOLVED (FIXED) by Aug 30 12:51:59 * *  ImageMagickis not correctly staged, module libraries not correctly installed Aug 30 12:52:01 * * http://bugs.openembedded.org/show_bug.cgi?id=2892 Aug 30 12:54:32 hrw: http://bugs.openembedded.org/show_bug.cgi?id=2897 Aug 30 12:57:26 So what is it with uboot? A bootloader should be a few thousand lines of source... how does that end up to be such a large tarball? Aug 30 12:58:08 board specific support Aug 30 12:59:02 it's not possible to checkout just gta support Aug 30 13:00:53 uboot doesn't depend on anything, so it might be better to just download precompiled image Aug 30 13:01:06 which is probably not present in mainline compared to amount of patches they use Aug 30 13:01:48 anyaway who would hack on uboot without debrick board Aug 30 13:02:23 03jin 07org.oe.dev * rf4febba2... 10/ (1 packages/imagemagick/imagemagick_6.2.9.bb): image-magick: fix staging and packaging, closes #2892 Aug 30 13:03:38 ynezz: the problem is that uboot running machines needs kernel created with uboot-mkimage which is only in uboot Aug 30 13:04:01 ah Aug 30 13:04:40 but we have a version in OE now that builds mkimage and has its patches in OE and git revision locked down Aug 30 13:15:10 * XorA tries to remeber how to use a zaurus and get tosa running Aug 30 13:18:37 * hrw forgot that XorA has tosa access Aug 30 13:19:28 hrw: tosa has wolfson touchscreen Aug 30 13:20:17 XorA: and audio - I know Aug 30 13:20:53 hrw: at least I get to fix it on wolfson time before I leave :-) Aug 30 13:21:49 any progress somewhere on the tosa usb driver and kernel >2.6.17? Aug 30 13:22:30 arse bollox, standard USB cable doesnt fit tosa? Aug 30 13:23:29 XorA: tosa use sharp i/o for usb device Aug 30 13:23:44 koen: none rater Aug 30 13:23:56 koen: thx for the hint, I'll pester the guy who wrote the actual driver. Aug 30 13:24:01 hrw: whats that usb shaped hole? Aug 30 13:24:44 usb host Aug 30 13:25:45 what about the hole on the sync cradle? is that just the client bit broken out of the sharp IO adapter? Aug 30 13:26:08 right Aug 30 13:26:22 * hrw do not use tosa cradle Aug 30 13:27:01 not good start, tosa oopses on resume Aug 30 13:27:24 2.6.17 kernel? Aug 30 13:27:42 linux-rp-2.6.17-r35 Aug 30 13:28:31 I take it tosa patches arent synced upto 2.6.22 yet Aug 30 13:29:31 the ohci driver lacks suspend support Aug 30 13:30:08 I suspect this module isn't unloaded during suspend Aug 30 13:30:32 do13: how badly wrong will a 2.6.22 kernel go? Aug 30 13:33:14 I suspect all parts execpt ohci needs small to be refreshed Aug 30 13:34:28 morning Aug 30 13:34:37 should be small changes Aug 30 13:34:57 ohci needs a lot of work Aug 30 13:34:58 usb 3-1: configuration #2 chosen from 2 choices Aug 30 13:34:58 cdc_subset: probe of 3-1:2.0 failed with error -22 Aug 30 13:34:58 rndis_host: probe of 3-1:2.0 failed with error -110 Aug 30 13:35:19 tosa? Aug 30 13:35:54 do13: PC end of the link Aug 30 13:36:34 I haven't check usb client support since ages Aug 30 13:37:03 no wireless at work Aug 30 13:37:09 bad Aug 30 13:37:19 cf ethernet? Aug 30 13:37:50 do13: I got PCMCIA ethernet, but zaurus kernels do not like it Aug 30 13:38:21 I take it tosa isnt standard pxa27xx USB then Aug 30 13:38:32 which ethernet chipset? Aug 30 13:38:45 pxa255 usb Aug 30 13:40:11 Similar to cxxx but slightly changed. Can't remember the differences offhand Aug 30 13:41:10 ~lart touchscreen Aug 30 13:41:10 * ibot whips out a hot clue gun and makes sure that touchscreen is stuck to the floor Aug 30 13:42:35 btw. you plan to leave wolfson? Aug 30 13:42:45 do13: already have Aug 30 13:42:50 do13: working out notice Aug 30 13:43:24 do13: so if you fancy working from .de get CV to Liam :- Aug 30 13:43:27 ) Aug 30 13:43:35 I wasn't aware of that Aug 30 13:43:58 Can you talk about the reason? Aug 30 13:44:15 do13: change of scene Aug 30 13:44:43 XorA: you are leaving them or they remove you? Aug 30 13:45:24 hrw: leaving under own steam Aug 30 13:46:26 XorA: would you recommend working for wolfson? Aug 30 13:46:36 do13: its cool here Aug 30 13:46:48 what is the purpose of the DEPENDS variable in an image recipe? I don't see a lot of consistency to how DEPENDS is used. Aug 30 13:47:03 cbrake: pull in builtime deps Aug 30 13:47:06 +d Aug 30 13:47:21 XorA: can you talk about your new job? Aug 30 13:47:35 do13: no Aug 30 13:47:54 do13: but there is nothing negative about Wolfson that I am leaving Aug 30 13:48:02 the plot thickens Aug 30 13:48:15 do13: and why I helping them find replacement Aug 30 13:48:27 ~praise WinCE Aug 30 13:48:28 All hail WinCE! Aug 30 13:49:26 * hrw -> out Aug 30 13:49:33 koen: why does angstrom-console-image DEPEND on task-base-extended, but angstrom-x11-image DEPENDS on task-base? They both RDEPENDS task-base-extended. Aug 30 13:50:00 only the RDEPENDS is needed Aug 30 13:50:09 the rest is historical cruft Aug 30 13:50:18 do13: our wince guys had a CF ethernet card Aug 30 13:50:42 cool Aug 30 13:50:58 ~seen prpplague Aug 30 13:51:02 prpplague is currently on #edev (34m 41s) #openjtag (34m 41s) #uclibc (34m 41s). Has said a total of 4 messages. Is idling for 19s, last said: 'greenphone?'. Aug 30 13:54:27 koen: can we take DEPENDS out of the image files then, or do we risk breaking things? Aug 30 13:54:44 that should be sage Aug 30 13:54:49 safe* Aug 30 13:56:06 koen: I'm just thinking it would help reduce confusion. Do you mind if I run some tests and then do that? Aug 30 13:56:22 not at all Aug 30 13:56:26 <- lazy Aug 30 13:56:59 koen: :-) I'm converting one of my projects over to not abuse DISTRO_EXTRA_RDEPENDS, so after I get that done I'll make the change in the other image files. Aug 30 13:59:37 I use ANGSTROM_EXTRA_INSTALL as an ad-hoc hack for that Aug 30 13:59:52 for other stuff I just make a new image Aug 30 14:00:54 cyberdeck : Did your kernel boot when build with gcc 3.4.4 ? Aug 30 14:01:20 RP: how do the poky ipkg/deb changes interact with rm_work? Aug 30 14:02:39 new project device for OE: http://www.elinux.org/Hammer_Board Aug 30 14:15:10 koen : nice. price ? Aug 30 14:25:50 not known yet Aug 30 14:27:22 koen: which company sells these boards? Aug 30 14:27:42 afaik AMLTD Aug 30 14:36:18 koen: greetings Aug 30 14:36:50 prpplague: ave my Lord Aug 30 14:39:50 steliosk, do13: prpplague is the hammer overlord Aug 30 14:42:41 one more day and FIC will announce who is going to work on OE? Aug 30 14:42:51 koen : Does gcc-cross-sdk work ? Aug 30 14:43:11 steliosk: that's florian's territory Aug 30 14:43:40 florian : Does gcc-cross-sdk work ? Aug 30 14:44:17 hi, how to force bitbake to rebuild bzImage ? Aug 30 14:44:41 bitbake -c rebuild virtual/kernel Aug 30 14:44:56 koen: Thanks ! Aug 30 14:45:26 steliosk: I haven't tried for some days, but I guess so Aug 30 14:45:44 zecke: you have news? Aug 30 14:46:13 XorA: not yet, I wonder when to congratulate this guy... ;) Aug 30 14:47:08 florian : have you used it/deploy it for a project etc ? Aug 30 14:48:10 I hope he fixes the abortions called linux-gta01* Aug 30 14:49:02 steliosk: yes, I built some binaries with the 4.1.2 i have installed on my box here Aug 30 14:49:16 koen: hehe unique descrition Aug 30 14:49:28 florian : nice. Anything to watch out ? Aug 30 14:52:20 steliosk: hmm... it might be a good idea to try if the g++ works properly, I had some trouble ith it some time ago. Aug 30 14:52:43 florian : which arch ? Aug 30 14:54:15 steliosk: some arm target, but it might have been related to the fact that I imported that toolchain into Scratchbox Aug 30 14:56:54 03koen 07org.oe.dev * r1f260b57... 10/ (34 files in 3 dirs): gcc 4.2.1: add patches for maverick crunch on ep93xx (That's an FPU) Aug 30 14:56:59 03koen 07org.oe.dev * r4fb1a744... 10/ (50 files in 3 dirs): gcc 4.1.2: add patches for maverick crunch (that's an FPU) Aug 30 14:57:36 now fftw can run twice as fast on an ep93xx :) Aug 30 14:58:57 koen: does that mean you can run Arm Bru :-D Aug 30 15:00:57 :D Aug 30 15:01:34 hey mgross Aug 30 15:01:41 mgross: how's the robot doing? Aug 30 15:02:07 oh great - all the .pc files in the -dev packages are wrong Aug 30 15:04:36 I'm doing pretty good Aug 30 15:04:46 who are things? Aug 30 15:04:57 * * OE Bug 1903 has been RESOLVED (FIXED) by Aug 30 15:04:59 * *  Packaging for pkg-config 0.21 Aug 30 15:05:01 * * http://bugs.openembedded.org/show_bug.cgi?id=1903 Aug 30 15:05:24 my robot has been getting ignored for the past few months ;( Aug 30 15:05:32 poor robot Aug 30 15:05:39 I've been busy with too many other things. Aug 30 15:06:21 reminds me, really should setup mindstorms Aug 30 15:06:23 I have a bunch of v4L and PIL tweaking to do on the slug before its ready for its next run. Aug 30 15:06:24 * hrw|gone urinates on git Aug 30 15:06:45 so, I did a new pull yesterday and now when I boot an angstrom x11 or console image, I get the following: Aug 30 15:06:47 VFS: Mounted root (jffs2 filesystem). Aug 30 15:06:47 Freeing init memory: 132K Aug 30 15:06:47 Warning: unable to open an initial console. Aug 30 15:06:52 Why does video capture suck so bad under linux? Aug 30 15:07:04 that is with, or without, init=/bin/sh Aug 30 15:07:21 christopher: well, what is console? Aug 30 15:07:56 zecke: I thought it was bash by default? will it matter if it is ash? Aug 30 15:08:13 christopher: you mix up words shell and console Aug 30 15:08:29 how can I make shure support for ext2 is compiled into the kernel ? Aug 30 15:08:42 christopher: you probably passed console=xxxS? to kernel, did you get that right Aug 30 15:08:55 on fedora7 mtd-utils-native, fails to compile how is the correct way to solve/mention this at OE ? http://pastebin.com/m55487e88 Aug 30 15:08:58 ... instead of beeing provided as module ? Aug 30 15:08:59 oh right. gotcha now. yeah, it worked yesterday before the pull. Aug 30 15:09:00 jan_js: alter defconfig Aug 30 15:09:04 same as always Aug 30 15:09:16 console=ttyS0,115200 Aug 30 15:09:20 grma: mtn pull; mtn update Aug 30 15:09:24 ok Aug 30 15:09:29 so, maybe that means that the dev wasn't created Aug 30 15:09:32 let me check that Aug 30 15:09:44 XorA: MACHINE_EXTRA_RRECOMMENDS = "kernel-modules" has no influence ? Aug 30 15:09:49 christopher: or you ttyS0 s no longer known as that Aug 30 15:10:07 jan_js: not on kernel building Aug 30 15:10:10 yep, nothing in /dev Aug 30 15:10:18 how can I rebuild the dev directory in the rootfs? Aug 30 15:10:28 christopher: it should be built with udev Aug 30 15:10:31 I did a rebuild of the image and it seemed to go through and do the rootfs...but nothing under dev Aug 30 15:10:37 christopher: which wont happen if you do init=/dev/sh Aug 30 15:10:42 Question: what is a good FS to use for a home directory located on a 4 gig SD card? I need it to be robust against system failure more than anything else. Aug 30 15:10:44 s/dev/bin/ Aug 30 15:10:48 ok. hmmm. Aug 30 15:11:03 but the rootfs on my machine (that is, the directory), should have the dev entries, right? Aug 30 15:11:15 christopher: not until udev runs Aug 30 15:11:19 I passed in init=/bin/sh yesterday and it worked Aug 30 15:11:27 ok. udev on the pc or udev on the unit? Aug 30 15:11:41 on where ever you need a ./dev populated Aug 30 15:12:09 yeah, I need those on the pc before it gets put in a jffs2 image, right? Aug 30 15:12:38 christopher: shouldnt do Aug 30 15:12:50 mgross: I would use ext2/ext3 Aug 30 15:12:54 re Aug 30 15:12:59 XorA: Which defconfig ? I have got quite a bunch of them under build/tmp ... Aug 30 15:13:20 jan_js: read manuals Aug 30 15:13:26 build/tmp/rootfs/dev ...isn't yours populated? Aug 30 15:13:37 I'm confused even more now. :) hehe Aug 30 15:13:50 christopher: no, its empty, udev builds /dev on boot Aug 30 15:13:57 ok Aug 30 15:15:13 well, even without init=bin/sh, it doesn't find a console Aug 30 15:15:41 then I guess you serial driver isnt using ttyS0 as a name Aug 30 15:15:59 christopher: The kernel won't use a fs to open the console Aug 30 15:16:13 that's weird. it works with a rootfs created with buildroot and one created with OE two days ago Aug 30 15:16:20 I haven't changed that in months Aug 30 15:16:31 christopher: if it can't open the console you have either no tty,serial in your kernel, or the name of the serial is wrong Aug 30 15:16:44 zecke: ok, I'll double check the kernl Aug 30 15:16:45 kernel Aug 30 15:17:07 ext2 is not robust against system failures. Aug 30 15:17:34 mgross: mount card in sync mode, its much better then Aug 30 15:17:44 XorA: udev doesn't work with an empty /dev, since the kernel needs /dev/console and some other nodes to start init and launch udev Aug 30 15:18:09 sync mode. thanks! Aug 30 15:18:09 koen: so, I should have something in the dev directory, right? Aug 30 15:18:25 koen: but it doesnt need ttyS0 in it Aug 30 15:18:58 I am making a clean angstrom build, and do_rootfs aborted with a makedevs error: build/angstrom/rootfs/dev/initctl: file can not be created Aug 30 15:19:08 any idea? Aug 30 15:19:22 the whole build/angstro etc. dirs were created by Angstrom during the build process Aug 30 15:19:30 so I dod not mess around with permissions or anything Aug 30 15:28:37 i have now successfully build, how do now freeze this version in OE ? with CVSDATE ? Aug 30 15:29:58 morning kergoth Aug 30 15:30:11 hey Aug 30 15:53:00 hello Aug 30 15:53:44 anyone have any idea on this error? Aug 30 15:53:46 ERROR: log data follows (/home/mranostay/openembedded/tmp/work/centrality-linux-gnueabi/linux-2.6.21-r7/temp/log.do_deploy.18509) Aug 30 15:53:46 | tar: lib: Cannot stat: No such file or directory Aug 30 15:53:46 | tar: Error exit delayed from previous errors Aug 30 15:54:00 that is on the do_deploy step Aug 30 15:56:09 mranostay_work: kernel without modules? Aug 30 15:59:24 koen: i ahve successfully build a ep93xx with crunch, is then glibc also build with crunch supoort ? Aug 30 15:59:34 s/ahve/have Aug 30 16:00:11 or how do i find that out ? Aug 30 16:01:07 03florian 07org.oe.dev * r969fb791... 10/ (3 files in 2 dirs): intltool: It needs some perl modules to work at all - latest ones at least. Aug 30 16:01:13 03florian 07org.oe.dev * re6bdcad4... 10/ (3 files in 2 dirs): intltool: increment PR Aug 30 16:02:06 Anyone who does not agree with the intltool dependencies? Aug 30 16:02:29 grma: no - base system is armv4t normal for ep93xx Aug 30 16:06:11 hrw: that means i don't booted a crunch glibc image? i only booted a image where gcc can produce crunch code, but glibc is not builded with chrunch compiler? Aug 30 16:07:15 grma: right Aug 30 16:07:24 ok thanks Aug 30 16:07:28 florian: libxml-parser-perl is not needed for it? Aug 30 16:08:01 grma: basically we build base system with softfloat and fpu intensive apps with fpu optimizations Aug 30 16:09:42 ok, thanks, then i have to try to build a fpu intensive application with fpu support ... thanks a lot for make this clear to me. Aug 30 16:17:51 hrw: I'm quite sure it is needed but it was commented in the bb files. Aug 30 16:18:45 florian: I will switch alix to geode rootfs in few days so will be able to do some testing under angstrom on machine more usable then simple pda Aug 30 16:20:08 hrw: ah cool, that might be a chance to fix more similar issues Aug 30 16:21:06 koen: should the rootfs/dev have anything in it when I look on my pc (before jffs2 image is built) or is this done at run time. I never checked when it was working and now that it isn't, I am not sure how it should be Aug 30 16:22:05 koen: is there anything permissions-wise that I should know? for example, should root own those files for the rootfs? Aug 30 16:22:17 that is what buildroot would do is make root own all files Aug 30 16:23:24 christopher: we build images under fakeroot Aug 30 16:23:34 christopher: and /dev/ is populated on device during boot Aug 30 16:25:31 hrw: ok, so /dev/ttyS0 would be built before trying to attach as a console? Aug 30 16:25:45 hrw: my issue is that it won't open the console. same kernel, same command line works fine on other rootfs Aug 30 16:25:53 and in fact, it worked yesterday. Aug 30 16:26:01 christopher: attaching as cmdline=console=/dev/ttyS0 passed to kernel? Aug 30 16:26:05 I am trying another clean build to see if it makes a difference Aug 30 16:26:09 hrw: correct. Aug 30 16:26:26 christopher: iirc kernel do not look into rootfs for it ;) Aug 30 16:26:45 christopher: kernel know sooner what ttyS0 is then how to mount rootfs Aug 30 16:27:25 hrw: yeah, guess that's true. :) so, when it says it failed to open initial console after mounting root, it just means it can't run the scripts? Aug 30 16:27:37 hrw: I tried with init=/bin/sh to bypass all of that, but same message Aug 30 16:28:23 christopher: you built image with OE or by hand? Aug 30 16:28:41 hrw: kernel built outside of oe Aug 30 16:28:46 guys... angstrom is failing on do_rootfs: http://pastebin.ca/675904 Aug 30 16:29:04 as I see the rootfs/dev directory does not exist yet, so something wrong in do_rootfs? Aug 30 16:29:17 Jin^eLD: /OE/org.openembedded.dev/files/device_table-chekov.txt is present? Aug 30 16:29:24 yes it is Aug 30 16:29:32 it is trying to create the first entry from it Aug 30 16:29:50 which is /OE/build/angstrom/rootfs/dev/initctl Aug 30 16:29:57 and that fails because dev/ is not yet there Aug 30 16:30:00 its a clean build Aug 30 16:30:10 Jin^eLD: add "d /dev" into table Aug 30 16:30:28 /dev/initctl p 600 0 0 - - - - - Aug 30 16:30:51 thats how it looks right now: /dev/initctl p 600 0 0 - - - - Aug 30 16:30:56 Jin^eLD: look at device-table-minimal in OE Aug 30 16:31:04 ok let me compare Aug 30 16:31:56 oooh, the standalone d dev line! Aug 30 16:31:58 thanks hrw Aug 30 16:32:01 np Aug 30 16:32:10 it was recent addon Aug 30 16:34:40 guys .. is there a really easy way to set up internet on bluetooth? i'm using ajzaurusUSB on Mac OSX and getting kernel panics (on my macbook) whenever i try to do big transfers .. so i want to swap to bluetooth and see if that gives me less hassle... anyone know if this is working on openmoko yet, and if so, how so? Aug 30 16:35:20 hrw: do I need to add more stuff? now it complains that angstrom/rootfs/bin/tinylogin can not be created, so also a line for rootfs/bin ? Aug 30 16:35:28 that would seem odd in a devtable Aug 30 16:37:31 Jin^eLD: only /dev/ entries should be there Aug 30 16:37:48 torpor: bluez.org -> PAN-HOWTO Aug 30 16:37:51 torpor: it is easy Aug 30 16:38:09 hrw: oh ok Aug 30 16:38:10 my bad Aug 30 16:38:23 hrw thanks Aug 30 16:38:25 i'll check it out Aug 30 16:38:44 np Aug 30 16:39:29 oh this won't work on mac osx .. its linux only. damn. Aug 30 16:40:17 torpor: dont say that osx lack PAN Aug 30 16:40:32 yup Aug 30 16:41:02 torpor: http://www.karppinen.fi/2007/04/10/bluetooth-pan-supported-in-mac-os-x-1049/ Aug 30 16:41:24 doh! Aug 30 16:41:26 start pand server on phone and connect mac to it Aug 30 16:44:29 hrw: should I just do a completely new build or is there a way to track down why it can't see a console? Aug 30 16:58:43 christopher: try to use OE built kernel? no new build should be needed as long as image was done with OE Aug 30 16:58:51 and sorry but I have to go Aug 30 17:01:52 RP: I just made some changes to dpkg in OE with my fingers crossed, I can't see how dpkg works in poky will all the dependency loops Aug 30 17:03:25 koen :) Aug 30 17:03:51 like dpkg-native and u-a-native depending on eachother Aug 30 17:03:58 I did have a hard time with the dpkg stuff when I tried it ;) Aug 30 17:04:03 and failed in the end... Aug 30 17:07:42 angstrom console image compiles cairo and x11 stuff? what's the X11 image then? Aug 30 17:09:32 build depends or runtime depends? Aug 30 17:11:09 aah.. ok, that could be it, probably just the build Aug 30 17:11:19 I just noticed that during building Aug 30 18:36:47 hrw: what is different about the kernel that OE builds? Aug 30 18:48:40 ~lart network manager Aug 30 18:48:40 * ibot pries network manager's back open with a screwdriver and flashes a new bootldr to network manager Aug 30 18:59:33 hmm Angstrom midori package isnt very helpful Aug 30 19:00:41 i tried to rebuild, and now i get: Aug 30 19:00:43 http://pastebin.com/m1080fe9c Aug 30 19:00:48 someon an idea ? Aug 30 19:01:44 i see only warnings ... Aug 30 19:01:54 grma: new one on me Aug 30 19:02:37 XorA|gone: sorry i don't understand, what ? Aug 30 19:02:54 grma: never seen that before Aug 30 19:03:40 yes, before the crunch patches it worked ... Aug 30 19:04:00 grma: ooops, sounds like koen broke gcc :-) Aug 30 19:04:10 :-) Aug 30 19:05:47 but this are only warnings ... Aug 30 19:06:40 ahh, the error is above: Aug 30 19:06:44 | gtkwidget.c:4715: fatal error: internal consistency failure Aug 30 19:07:47 urgs Aug 30 19:07:58 that's nasty Aug 30 19:25:25 Hello! I finally got my hands on a Zaurus SL-C3200, alas for a few days only; what's the best ROM to put on it? It's Japanese right now and that's not quite useful for me ;-) Aug 30 19:53:11 re Aug 30 19:53:14 http://www.vanityfair.com/ontheweb/features/2007/02/autumn200702?printable=true¤tPage=all Aug 30 19:53:30 zap: it depends Aug 30 20:07:21 hrw: I understand :) Aug 30 20:07:58 zap: and there is no such thing as 'ROM' Aug 30 20:09:13 I was meaning a ROM image, but I'm not sure how Zaurus flashes its ROM Aug 30 20:09:21 what I have seen until now is some tar files Aug 30 20:09:41 I wonder if Zaurus' boot loader unpacks them right on the root device? Aug 30 20:09:53 zaurus does not have a rom Aug 30 20:10:18 And where's the initial bootloader, that appears when you press OK on startup? Aug 30 20:11:04 in flash? Aug 30 20:11:15 ah ok, I was meaning ROM as flash Aug 30 20:11:29 I understand they aren't strictly speaking the same :) Aug 30 20:12:17 most of the flash is formatted as the root partition? Aug 30 20:13:29 koen: I don't see any dependency loops in poky which is strange Aug 30 20:14:50 RP: poky prefers update-alternative-cworth and thats why it work Aug 30 20:15:03 RP: angstrom probably do not set it so it got the loop Aug 30 20:15:10 hrw: ah yes, I think I cheated ;-) Aug 30 20:15:11 zap: not quite Aug 30 20:15:30 zap: it depends on model and distro Aug 30 20:15:38 zap: and kernel version Aug 30 20:16:04 zap: and here we can suggest you Ångström Aug 30 20:16:45 Ångström sounds Japanese :) Aug 30 20:17:07 ah, Angstrom Aug 30 20:17:14 for some reason xchat mangled the name Aug 30 20:17:32 you lack utf8 handling Aug 30 20:17:47 perhaps thats because my /codepage is set to 1251 Aug 30 20:17:58 ok never mind Aug 30 20:18:02 Angstrom is ex-oe? Aug 30 20:18:07 zap: russian you are? Aug 30 20:18:15 yep Aug 30 20:18:29 angstrom is OE-derived distro Aug 30 20:18:38 a fork? Aug 30 20:19:42 sorry I think I'm asking stupid questions, I'm out of PDA business for a couple of years ;) Aug 30 20:28:07 zap: OE is a tool to build software/distros Aug 30 20:30:27 How stable Angstrom is? Aug 30 20:30:34 can't find a release, just unstable Aug 30 20:30:53 try it Aug 30 20:31:00 for me it is stable for use Aug 30 20:31:15 does it support 100% of the SL-C3200 hardware? Aug 30 20:31:20 or something won't work? Aug 30 20:31:44 iirc it is working ok Aug 30 20:31:53 good Aug 30 20:38:23 hi zap Aug 30 20:38:28 hi Aug 30 20:38:52 zap: It was the cxx00 and c7x0 series that got the backlight class pushed into the kernel, hardware support is pretty good Aug 30 20:39:13 nice Aug 30 21:00:51 gn Aug 30 21:21:12 SL-C3200 has no USB host? Aug 30 21:32:26 zap: it does Aug 30 21:34:25 bye Aug 30 21:41:47 morning Aug 30 21:41:57 rwhitby: greetings Aug 30 21:42:41 morning rwhitby Aug 30 21:51:27 rwhitby: go to your wife and celebrat Aug 30 21:51:28 e Aug 30 21:57:56 hrw|gone: she's still asleep Aug 30 21:58:10 I'm giving my daughter breakfast :-) Aug 30 22:12:55 rwhitby, breastfeeding? Aug 30 22:13:21 polyonymous_: we don't need to hear about your fantasies ;-) Aug 30 22:13:33 I just asked. Aug 30 22:14:33 which notebook is "better" (that is - not for gaming) amd or intel? Aug 30 22:14:38 no, I'm not equipeed for that Aug 30 22:15:19 rwhitby, did I guess it right that you have a brand new daughter? Aug 30 22:15:38 no, she's 2.5 years Aug 30 22:15:58 ah. I just thought that's what you had to celebrate Aug 30 22:16:24 ah, sorry - now I see your train of thought from hrw|gone's message. Aug 30 22:16:28 mine is 3 y.o. and tomorrow morning she goes to kindergarten for the first time. I'm scared like hell. Aug 30 22:16:32 11 yrs wedding anniversay Aug 30 22:16:46 ah. Congrats then. Should console hrw now ;-) Aug 30 22:18:10 but seriously does anybody have idea what are pro and contras for intel and amd notebooks? speed/bettery life, etc.? Aug 30 22:20:43 polyonymous_: 1 minute after you leave the kindergarten gate, watching her crying as you leave, she will be happy again and have forgotten all about you :-) Aug 30 22:20:57 polyonymous_: depends on the processor and chipset Aug 30 22:21:15 amd geode laptop would last longer then an intel p4 Aug 30 22:21:25 rwhitby, I doubt she'll be crying ;-) The problem is that she doesn't speak German and we're in Germany... She'll learn fast, but I'm worried. After all, I'm her dad ;-) Aug 30 22:21:40 wirelessdreamer, I'm talking about something more highendish. Aug 30 22:21:41 intel i810 chipset wih integrated vodeo would last longer then amd 64bit Aug 30 22:21:54 aha Aug 30 22:22:20 probably go core solo if you want longer batter life less power, core duo if you want more [ower at the expense of battery life Aug 30 22:22:22 so, if I'm thinking about dual core options, going for intel would probably be better? Aug 30 22:22:29 amd's dual core's are battery hogs too Aug 30 22:23:06 sec Aug 30 22:23:06 of course single core last longer, but I'm planning to use it a lot on AC, so I want compromise ;) Aug 30 22:25:13 http://www.dynamism.com/r6/specs.shtml Aug 30 22:25:25 intel core2 with intel graphics Aug 30 22:25:42 if you get nvidia or ati you can game but your battery life goes down by half Aug 30 22:26:07 I want to vi, some gcc, etc. ;-) Aug 30 22:26:16 "some" gcc, I'd say ;-) Aug 30 22:26:41 also mp3, etc., but only on AC, so that doesn't matter. Aug 30 22:26:47 look for any core duo or core2 duo with an intel chipset for graphics and your set then Aug 30 22:27:02 So, you think intel is the way to go, in short. Aug 30 22:27:03 amd doesn't have the power saving laptop niche, thats intel Aug 30 22:27:03 Thanks. Aug 30 22:27:07 def Aug 30 22:27:24 Any particular model, keeping linux (no dualboot crap) in mind? Aug 30 22:27:46 price point? Aug 30 22:28:28 the cheaper the better, but I consider all options.... Don't want to waste money, so I may buy an expensive one. Aug 30 22:28:42 And the bigger screen (in terms of pixels, not inches) the better. Aug 30 22:29:04 the one i just linked is really nice, its a cousin to the toughbook line Aug 30 22:29:45 i really like subnote though, but i use a zaurus c3000 still so go figure Aug 30 22:29:47 ah, I opened window and forgot ;-) Aug 30 22:30:04 I already have c3100, so that's another story ;-) Aug 30 22:30:15 That's not quite a replacement. Aug 30 22:30:58 no matter how nice everything else is i look at nothing else fits in my pocket like a z and costs under 2 grand Aug 30 22:31:04 hmm.. I think I want bigger screen than that one, but thanks for the info - that's something to start from. Aug 30 22:31:33 dynamism has a lot of nice laptops on their site, and its well organized Aug 30 22:31:37 Yeah, Z is great, but not for all purposed. Aug 30 22:31:41 i always research there even if i don't get it Aug 30 22:31:44 purposes too. Aug 30 22:31:52 Yup. thanks. Aug 30 22:32:03 did you hear about the new via ulv proc? Aug 30 22:32:27 nah, I'm not following hardware. Until I have to research ;-) Aug 30 22:32:45 1 watt 500mhz x86 Aug 30 22:33:01 i'm really hopping something in the formfactor of the z comes out for that Aug 30 22:33:22 sounds interesting but not for this one. This is more of desktop replacement. Not that I'm going to retire my desktop. Aug 30 22:33:54 thats more along the z line anyway, but still interesting :) Aug 30 22:34:04 yup. Aug 30 22:35:23 Why on SL-C3200 I would want to have on the internal HDD a third VFAT partition? Aug 30 22:35:34 wish there was a site somewhere that would let me specify parameters and see what notebooks satisfy my criteria along with their prices... Well, anyway, thanks again and I'm gone to sleep. That awful kindergarten tomorrow morning ;-) Aug 30 22:35:47 zap, dunno why would you want it, but I don't have it. Aug 30 22:35:58 so you can have just root and home / Aug 30 22:35:59 ? Aug 30 22:36:05 zap I have a single partition. Aug 30 22:36:14 on wiki for some reason it tells me to create three Aug 30 22:36:39 and third to format as vfat (?) Aug 30 22:36:45 zap - I dunno. Maybe for some compatibility... anyway, I don't have it and can't think of any reason why do I HAVE to have it. Aug 30 22:37:06 maybe that partition can be exposed via the USB-storage-device interface? Aug 30 22:37:09 vfat may be of use to act as an usb storage for windows machine. Aug 30 22:37:14 aha Aug 30 22:37:16 yes Aug 30 22:37:32 and how linux/sharp is sharing that partition with windows? Aug 30 22:37:33 you can, of course, do it in the file too. Aug 30 22:37:47 I mean, do I have to unmount it first? Aug 30 22:37:50 over gadget module by the name I can't remember. Aug 30 22:37:56 ah, yeah, I'd think yes. Aug 30 22:38:17 I mean, is it useable or still a "developer-only" thing? Aug 30 22:38:21 if you don't want to unmount you're probably safer with samba. Aug 30 22:38:27 aha Aug 30 22:38:30 samba sounds fine Aug 30 22:38:34 I'm a "developer-only", so I dunno ;-) Aug 30 22:38:39 :) Aug 30 22:38:41 but I'm happy with it. Aug 30 22:38:56 ok thanks for hints, I'll make one big partition Aug 30 22:39:15 sure, if you think you can handle consequences you should always go your own way ;-) Aug 30 22:39:24 Again, sorry, I'm out to sleep ;-) Aug 30 22:39:34 g'nigh :) Aug 30 22:39:45 thanks. ciao Aug 30 22:39:47 * polyonymous_ is gone Aug 30 22:59:57 * * OE Bug 2898 has been created by carbonfreeze(AT)gmail.com Aug 30 22:59:59 * * broken keymap in spitz image 20070828 Aug 30 23:00:01 * * http://bugs.openembedded.org/show_bug.cgi?id=2898 Aug 30 23:00:09 sorry for idiocy, but how do I remove an applet off the GPE panel? I have three light bulbs there, and I would like just one :) Aug 30 23:00:32 zap there is panel editor Aug 30 23:00:44 You should be able to remove them with that Aug 30 23:01:03 aha Aug 30 23:01:07 thank you Aug 30 23:01:12 also long tap on the panel might give you a remove option I think Aug 30 23:02:00 03mickeyl 07org.oe.dev * recfd6561... 10/ (1 packages/efl1/imlib2_1.4.0.002.bb): imlib2: set OE_LT_RPATH_ALLOW to make imlib find its modules Aug 30 23:02:09 03mickeyl 07org.oe.dev * r16ddc63c... 10/ (1 packages/openmoko2/openmoko-theme-standard2_svn.bb): openmoko-theme-standard2: use SVNREV Aug 30 23:02:15 03mickeyl 07org.oe.dev * rd8e98da9... 10/ (1 packages/openmoko2/openmoko-firststart2_svn.bb): add openmoko-firststart2 Aug 30 23:02:21 03koen 07org.oe.dev * r7a23814b... 10/ (5 files in 3 dirs): dpkg: merge fixes from poky, update to 1.3.25 Aug 30 23:02:26 03koen 07org.oe.dev * r08a806e6... 10/ (3 files in 2 dirs): angstrom-package*: select proper u-a Aug 30 23:02:31 03koen 07org.oe.dev * rcfd70352... 10/ (1 packages/dpkg/dpkg-native_1.13.25.bb): dpkg-native: do not DEPENDS on u-a-native, since u-a-native already depends on this -> loop Aug 30 23:02:36 03koen 07org.oe.dev * r8a70ab8f... 10/ (15 files in 5 dirs): apt: sync with poky Aug 30 23:02:44 03koen 07org.oe.dev * ra713e30a... 10/ (3 files in 2 dirs): bluez: update to 3.17 Aug 30 23:04:19 ah I wonder if that bluez update will fix the problem I've been having compiling gnome-bluez Aug 30 23:06:13 hmm http://bugs.openembedded.org/show_bug.cgi?id=2889 the patch attached to this is wrong though I believe Aug 30 23:06:24 is there an OE way to avoid packaging kernels in the rootfs? Aug 30 23:06:53 Philippe: what do you mean by "wrong" Aug 30 23:07:20 craigh: well it is machine specific line in a generic file Aug 30 23:08:09 yeah, I guess -- but just above that there's about 200 lines of machine-specific patches in that generic file... Aug 30 23:09:01 craigh: yeah you seem to be right Aug 30 23:09:22 I just want more experienced devs opinion before commiting it Aug 30 23:09:50 Probably there should be a separate file for each machine which includes the common one Aug 30 23:09:58 and put all those patches and stuff in each machine-specific file Aug 30 23:10:31 whether or not to put the kernel image into the rootfs is really a distro thing too, not really something that should be in package/linux probably Aug 30 23:10:41 craigh: that was the plan at least as there has been lots of complaints about machine specific stuff being all over the place Aug 30 23:11:01 craigh: exactly my point Aug 30 23:11:55 one step at a time I guess :) Aug 30 23:12:03 craigh: that is how it goes. Aug 30 23:12:32 well and I need to get that image smaller. It does not fit on my old connec Aug 30 23:12:37 err connex Aug 30 23:12:44 Eventually, what'd be nice is to actually build an ipkg for the kernel image which will use mtd-utils on the system to put the uImage into the right place in flash Aug 30 23:13:35 going from ipkg to mtd-utils would not be ideal :-) Aug 30 23:14:08 craigh: there just must be another way to handle the kernel thing as I just remember openmoko also has a seperate kernel Aug 30 23:14:16 but it might be distro specific Aug 30 23:14:30 mickey|zzZZzz: or koen should know Aug 30 23:15:40 sorry about the dc Aug 30 23:15:43 I'll ping them about it tomorrow Aug 30 23:15:52 stupid DSL link keeps dropping me when the weather gets hot here Aug 30 23:16:27 I was just saying that angstrom-minimal-image should be small enough to fit on a non-XM gumstix Aug 30 23:17:13 hughescr: exactly the issue I am struggeling with Aug 30 23:18:04 I mean "it is small enough" Aug 30 23:18:12 the one I built was iirc 3.2MB or so Aug 30 23:18:17 oh, duh Aug 30 23:18:23 that's not small enough with a 1MB kernel Aug 30 23:18:58 hughescr: well and I want angstrom-console-image to fit on it :-) Aug 30 23:19:16 yeah, you might have some trouble there Aug 30 23:19:26 I would have bought an XM if they had not announced them the day after I ordered mine :-) Aug 30 23:19:37 unless you turn off some of the MACHINE_FEATURES -- that might help Aug 30 23:24:22 anyway Aug 30 23:24:27 * Philippe is off to sleep Aug 31 00:04:28 hmm Aug 31 00:04:43 is bluez-gnome_0.14-r1 broken? Aug 31 00:04:53 it's failing with a very odd msg during the install phase for me Aug 31 00:05:06 | make[3]: execvp: /home/craig/code/gumstix/oe/build/tmp/work/iwmmxt-angstrom-linux-uclibcgnueabi/bluez-gnome-0.14-r1/image/usr/share/applications: Permission denied Aug 31 00:05:20 that /usr/share/applications is a directory Aug 31 00:05:34 so... not entirely surprising that execvp is having some issues trying to execute it **** ENDING LOGGING AT Fri Aug 31 02:59:56 2007