**** BEGIN LOGGING AT Tue Sep 04 02:59:56 2007 Sep 04 05:56:36 *yawn* Sep 04 05:56:37 morning Sep 04 06:21:29 hi mickeyl Sep 04 06:21:58 mickeyl: rewind log to check our libmokopanelui2 discussion Sep 04 06:24:06 hrw: I couldn't fix that libsdl problem yesterday, so I reverted it to what it was. Sep 04 06:24:48 the trouble is the STAGING_BIN_CROSS is a directory that doesn't end in /bin, but the place in scummvm which used it (--with-sdl-prefix) expected a directory to which it would append /bin Sep 04 06:29:33 and it does not like my i586 target ;( Sep 04 06:31:54 hrw: DEPENDS is correct, it _is_ a build time dependency. Sep 04 06:32:09 it's a header file Sep 04 06:32:44 ah. Sep 04 06:36:24 good morning all Sep 04 06:37:46 hi Stelios Sep 04 06:39:49 hrw : hi ! Sep 04 06:42:58 I tried to use GLIS case from IKEA today - this plastic is too hard and breaks too easy ;( Sep 04 06:50:42 hrw : Did you use dremel to cut it or something else ? Sep 04 06:51:14 sharp knife Sep 04 06:51:29 We are thinking of designing a plexiglass case for Efika, as there is a request from a customer, but we are not sure we want to get into that kind of work Sep 04 06:52:03 I lack dremel currently Sep 04 06:52:06 Use a dremel cutter at low speed and should cut fine Sep 04 06:52:49 how's alix performance ? any benchmarks ? Sep 04 06:53:14 steliosk: I use it as nfs now so did not made any benchmarks other ten iperf Sep 04 06:54:32 i wonder how it will perform with float point. i want to test g729 codec on it Sep 04 06:56:49 is it in OE? I will be able to build/run it then Sep 04 07:00:49 03koen 07org.oe.dev * rc21d99da... 10/ (1 conf/sanity.conf): bitbake: require 1.8.8 Sep 04 07:00:55 03koen 07org.oe.dev * r1581de48... 10/ (1 packages/webkit/webkit-gtk_svn.bb): webkit: switch to SRCREV Sep 04 07:01:01 03koen 07org.oe.dev * rdc0a9b26... 10/ (1 classes/openmoko2.bbclass): openmoko: switch to SRCREV Sep 04 07:01:08 03koen 07org.oe.dev * r843a2e60... 10/ (1 packages/uclibc/uclibc_svn.bb): uclibc: switch to SRCREV Sep 04 07:01:15 03koen 07org.oe.dev * r2b01a0fd... 10/ (1 packages/dfu-util/dfu-util_svn.bb): dfu-utils: switch to SRCREV Sep 04 07:01:21 03koen 07org.oe.dev * r4ee9744d... 10/ (1 packages/gnuradio/gnuradio_svn.bb): gnuradio: convert to SRCREV Sep 04 07:01:28 03koen 07org.oe.dev * r2a487f85... 10/ (1 packages/gsm/libgsmd_svn.bb): libgsmd: convert to SRCREV Sep 04 07:01:36 03koen 07org.oe.dev * r9d496fc2... 10/ (5 files in 3 dirs): hildon-1: add svn version which builds against stock gtk Sep 04 07:01:44 03koen 07org.oe.dev * re28ef080... 10/ (25 files in 6 dirs): openmoko*: convert to SRCREV Sep 04 07:01:52 03koen 07org.oe.dev * r9c10fc79... 10/ (1 packages/ezx/opentapi_svn.bb): opentapi: convert to SRCREV Sep 04 07:02:00 03koen 07org.oe.dev * rdb3553a5... 10/ (5 files in 2 dirs): gtk-webcore: convert to SRCREV Sep 04 07:02:26 mickeyl: back in FF? Sep 04 07:02:40 yep, finally Sep 04 07:02:42 good morning all :) Sep 04 07:02:43 morning koen Sep 04 07:02:49 ah! Sep 04 07:02:51 SRCREV day Sep 04 07:02:55 * mickeyl dances the SRCREV dance Sep 04 07:03:08 * koen waits for RP to wake up and commit the pkgdata and friends Sep 04 07:03:23 Poky has lot of SRCREV stuff Sep 04 07:03:46 I have a sed script for that :) Sep 04 07:03:49 ;D Sep 04 07:04:13 but I committed only the recipes I actually build Sep 04 07:04:22 koen: does it also reset PR? Sep 04 07:04:31 mickeyl: RPs next change will require a build from scratch Sep 04 07:04:42 hrw: I decided not to Sep 04 07:04:52 koen: why? Sep 04 07:05:15 oki Sep 04 07:05:33 hrw: I didn't see the need for it Sep 04 07:05:37 who will tell openmoko abusers? Sep 04 07:06:09 koen: ok Sep 04 07:06:16 we reset them in Poky Sep 04 07:07:23 * mickeyl calls alice dsl technical hotline Sep 04 07:07:38 back home and started to have net problems? Sep 04 07:07:58 you'll like net problems now Sep 04 07:07:58 right. in the meantime i have upgraded to alice deluxe, but they didn't release my limits, i.e. i have still the same performance as before :/ Sep 04 07:08:06 it results in parse errors Sep 04 07:08:28 mickeyl: so what speed you ave and whic you should? Sep 04 07:08:35 see http://bugs.openembedded.org/show_bug.cgi?id=2897 Sep 04 07:08:54 i have 8000kbit/256kbit, but should have 16000kbit/1000kbit Sep 04 07:09:47 hmm.. when here will be ethernet speed at good price Sep 04 07:11:05 but my 3072/512 works quite good recently Sep 04 07:11:22 * koen is still on 1 mbit cable Sep 04 07:11:41 koen: my cable was 1mbit before Sep 04 07:12:21 the isp is stopping business in december, so I'm now busy hunting down places where I use koen@dominion as email :( Sep 04 07:12:34 (and reading "semiconductor devices" of course) Sep 04 07:13:03 koen: I'm in process of switching email from @hrw.one.pl to @haerwu.biz Sep 04 07:13:53 http://laughingsquid.com/roadworks-steamroller-prints-and-street-fair/ Sep 04 07:15:26 koen: consider switching angstrom to binutils 2.18 Sep 04 07:15:37 BFD: /home/hrw/devel/build/angstrom/tmp/work/armv4t-angstrom-linux-gnueabi/alsa-lib-1.0.14-r1/image/usr/lib/.debug/libasound.so.2.0.0: warning: sh_link not set for section `.ARM.exidx' Sep 04 07:15:38 I was waiting for you to say that :) Sep 04 07:15:42 those are gone Sep 04 07:15:56 koen: poky builds from scratch were ok during nigt Sep 04 07:16:36 added, will do a buildrun after the pkgdata changes Sep 04 07:17:27 mickeyl: What does that speed cost? Sep 04 07:17:57 * CM pays about 35EUR/month for 100/10 mbit Sep 04 07:22:33 how to freeze thunderbird - give 9K mails to move from imap to another imap Sep 04 07:25:34 hrw: how to freeze outlook - Move 9 mails with 300k attatchment to a personal folder :( Sep 04 07:27:45 how to freeze outlook select 'mark all as read' Sep 04 07:32:12 rwhitby: scummvm searc for sdl-config in path given and in path given/bin/ Sep 04 07:33:13 rwhitby: NOTE: package scummvm-0.9.1-r1: task do_compile: started Sep 04 07:36:34 CM: 60 eur, included is a voice flatrate Sep 04 07:36:36 aaaaaargh Sep 04 07:36:37 heh Sep 04 07:36:42 they didn't send me new hardware Sep 04 07:36:48 and as it turns out i still have a ADSL 1 modem Sep 04 07:36:51 *sigh* Sep 04 07:36:56 no wonder Sep 04 07:37:04 :-/ Sep 04 07:37:10 new hardware underway now. Sep 04 07:45:36 weird bug. When using icecc on a mixed i686/amd64 environment everything compiles well except i686-generic targets Sep 04 07:46:03 removing the amd64 machines from the icecc cluster makes i686-generic build fine Sep 04 07:47:11 all other targets (ppc,x86,xscale) build without problems Sep 04 07:47:23 rwhitby: NOTE: package scummvm-0.9.1-r1: task do_build: completed Sep 04 07:51:04 patched 'configure'? Sep 04 07:51:17 no Sep 04 07:53:28 scummvm 0.10 has been out for a while Sep 04 07:55:43 feel free to update Sep 04 07:57:48 03hrw 07org.oe.dev * rfebd6fbe... 10/ (1 packages/scummvm/scummvm.inc): Sep 04 07:57:48 scummvm: give proper path to sdl-config Sep 04 07:57:48 configure script check for GIVEN_PATH and GIVEN_PATH/bin Sep 04 08:03:53 hrw: aha, you took off the /.. Sep 04 08:05:28 rwhitby: I read configure script and thats was it Sep 04 08:06:04 yeah, my interest level wasn't that high :-) Sep 04 08:07:31 neither do mine Sep 04 08:07:48 but I prefer to fix instead of discussing and discussing Sep 04 08:09:39 I wonder how people use thunderbird mailer.. Sep 04 08:10:11 you can have sorted mails or threaded.. but not both Sep 04 08:10:17 hrw: thx for fixing Sep 04 08:10:44 rwhitby: no problem Sep 04 08:11:33 so today's the SRCREV day? Sep 04 08:11:59 yes Sep 04 08:12:23 morning Sep 04 08:13:26 Hi Sep 04 08:17:36 morning all Sep 04 08:18:35 steliosk: so what do we do with i586/i686 generics? Sep 04 08:18:55 hrw : I am building them to test then now Sep 04 08:19:08 that's why i posted the problem with icecc above Sep 04 08:19:14 ok Sep 04 08:20:38 btw does anyone know of a place for 1u cases that can be customized ? Our supplier stopped bringing cases Sep 04 08:21:28 hrw: thunderbird sucks ass, but it's the least bad mailer with threaded display on osx Sep 04 08:34:26 RP: bitbake 1.8.8 is in, time for the remaining poky stuff :) Sep 04 08:34:42 koen: Looking at it now Sep 04 08:44:09 hey lrg Sep 04 08:45:19 hey koen Sep 04 08:46:20 ~fish lrg Sep 04 08:46:21 * ibot slaps lrg around with a large trout Sep 04 08:46:41 koen: pushed Sep 04 08:47:07 thanks Sep 04 08:47:08 * lrg goes to work Sep 04 08:47:30 and will probably get fished again by XorA Sep 04 08:56:11 good morning Sep 04 08:57:31 03mwester 07org.oe.dev * r6e5c49eb... 10/ (1 packages/madwifi/madwifi-ng_r2702-20070903.bb): Sep 04 08:57:31 madwifi-ng: added recipe for the latest snapshop in order to provide support for more Sep 04 08:57:31 recent kernels. Default preference = -1, since this version will not build on BE kernels. Sep 04 08:57:43 03rpurdie 07org.oe.dev * rc43c60af... 10/ (1 classes/base.bbclass): base.bbclass: Wipe install directory before running install to make rebuilding packages safer (from poky) Sep 04 08:57:51 03rpurdie 07org.oe.dev * rb77696cb... 10/ (4 files in 3 dirs): classes/conf: Remove bitbake < 1.8.8 compatibility code Sep 04 08:57:59 03rpurdie 07org.oe.dev * r190270f4... 10/ (9 files in 2 dirs): package*.bbclass: Switch to separate tasks for each packing type (from poky) Sep 04 08:58:04 03rpurdie 07org.oe.dev * ra9a8c4a4... 10/ (1 classes/package.bbclass): package.bbclass: Turn install path into a variable (PKGDATA) and cleanup some unneeded complexity (from poky) Sep 04 08:58:10 03rpurdie 07org.oe.dev * r209ccafa... 10/ (1 classes/multimachine.bbclass): Sep 04 08:58:10 multimachine.bbclass: Use a new PKGDATA_DIR which includes the multimachine path Sep 04 08:58:10 to prevent cross contamination between machines. This change requires all Sep 04 08:58:10 install tasks to rerun (remove the do_install stamps from tmp/stamps) Sep 04 09:09:28 autobuilder restarted Sep 04 09:13:21 nice Sep 04 09:14:05 and I need to clean my tmp ;( Sep 04 09:14:28 or keep on not updatng for today Sep 04 09:15:23 mickeyl: the autobuilder also builds openmoko images for all OE phones :) Sep 04 09:16:13 * XorA took the excuse to make a multimachine build Sep 04 09:23:03 ~lart adsl customer support Sep 04 09:23:03 * ibot whips out a shotgun, trudges over to adsl customer support, and goes postal Sep 04 09:26:56 http://pastebin.ca/680863 - to not having libc6-dev conflict with linux-libc-headers-dev Sep 04 09:27:29 objections? other then 'ARGH.. why I need to rebuild glibc again!!!' Sep 04 09:28:17 does libc6-dev depend on linux-libc-headers-dev? Sep 04 09:28:40 yes Sep 04 09:28:52 ok with me then Sep 04 09:30:49 configure: error: Need linker with .init_array/.fini_array support. Sep 04 09:30:54 in glibc Sep 04 09:31:06 -itermediate Sep 04 09:31:29 XorA: 'unset MACHINE' Sep 04 09:31:35 and rebuild binutils Sep 04 09:31:49 hrw: fine by me Sep 04 09:33:11 koen: Please set a valid MACHINE in your local.conf Sep 04 09:34:14 I set it in auto.conf Sep 04 09:42:32 hrw : the optimizations for i585 and i686 build here without any problem for uclibc Sep 04 09:42:39 hrw : trung glibc now Sep 04 09:42:48 s/trung/trying Sep 04 09:43:04 great Sep 04 09:46:42 I am updating alix to geode packages now (-mcpu=k6-2) Sep 04 09:58:14 hrw : any idea how debian handles the different x86 machines ? Sep 04 09:58:24 steliosk: not at all Sep 04 09:58:32 steliosk: Debian is i486 Sep 04 09:58:38 hmmm, i'll have a look and we could follow that Sep 04 09:59:08 i think there are more arches supported Sep 04 09:59:10 steliosk: libc-i686 is provided, few packages are provided in optimised packages, kernels are for misc subarchs Sep 04 09:59:43 you will hate me - glibc to rebuild Sep 04 10:04:36 ~hate hrw Sep 04 10:06:29 ~hurt hrw Sep 04 10:06:30 * ibot takes a large hammer and plays whack-the-mole on hrw's hands, laughing all the while Sep 04 10:06:52 Dokuro-chan ;-) Sep 04 10:07:12 ;D Sep 04 10:07:15 hrw: I bumped eglibc as well :) Sep 04 10:07:50 * XorA wonders if we could implement subtask PR bumps :-) Sep 04 10:10:25 XorA: Insert a "PR" value into the stamps ;-) Sep 04 10:11:21 hey mallum Sep 04 10:11:39 hey koen Sep 04 10:13:37 RP: that would be part of the way, but for example glibc has a small change in packaging, so doesnt really need a rebuild Sep 04 10:13:47 RP: just a do_install done again Sep 04 10:13:49 XorA: but repackaging Sep 04 10:14:22 XorA: which can be hard if rm_work is in use Sep 04 10:14:47 XorA: Just making it apply to a package's subtask would be a nightmare :) Sep 04 10:16:00 are you all done pushing things that will make me rebuild tmp? Sep 04 10:16:37 RP: can I just move some dirs in staging to get it more-or-less compatible? Sep 04 10:17:04 hrw: You could copy pkgmaps to each of the machine specific directories Sep 04 10:17:55 like poky has? Sep 04 10:18:14 hrw: Yes, its poky's change that just went into OE Sep 04 10:18:41 RP: thats why we have the RP super genius :-) Sep 04 10:24:28 Angstrom psplash looks nice on 20" 1680x1050 ;) Sep 04 10:28:59 * XorA needs some sort of testing rack for zauruses :-) (and neos) Sep 04 10:30:15 Hi all.. Is the 'recent checkins' feature on openembedded.org broken? It's 12 weeks behind.. Sep 04 10:31:02 * RP too Sep 04 10:37:28 XorA: a debug board with the reset switch wired up to a serial controlled relay works well for me. Sep 04 10:38:35 rwhitby: yeah, but that doesnt allow me to flash 5x zaurus in one go :-) Sep 04 10:59:43 03hrw 07org.oe.dev * r0e25c866... 10/ (14 files in 2 dirs): glibc: do not package /usr/include/scsi as it is in libc-linux-headers-dev (from Poky) Sep 04 10:59:51 03koen 07org.oe.dev * r3fed2353... 10/ (3 files in 2 dirs): eglibc: apply glibc fix to eglibc as well Sep 04 11:07:08 argh.. Sep 04 11:07:26 fsck on 420GB partition takes eons Sep 04 11:26:01 uf. pass 2 started Sep 04 11:29:06 * hrw removes DL_DIR/cvs/ Sep 04 11:39:57 zecke: wake up :-D Sep 04 12:48:37 mickeyl RP did you found some time to look at oe bug 2598? Sep 04 12:52:59 03koen 07org.oe.dev * r29af1af8... 10/ (1 packages/ffmpeg/ffmpeg_svn.bb): ffmpeg: convert to SRCREV Sep 04 12:53:04 03koen 07org.oe.dev * rdcf81de9... 10/ (1 packages/libxosd/libxosd_svn.bb): libxosd: convert to SRCREV Sep 04 12:53:10 03koen 07org.oe.dev * r59643594... 10/ (6 files in 5 dirs): matchbox*: convert to SRCREV Sep 04 12:53:20 03koen 07org.oe.dev * r0240cbca... 10/ (1 packages/tinymail/libtinymail_svn.bb): tinymail: convert to SRCREV Sep 04 12:53:24 ~oebug 2598 Sep 04 12:53:26 03koen 07org.oe.dev * r385c0fc9... 10/ (1 packages/libowl/libowl_svn.bb): libowl: convert to SRCREV Sep 04 12:53:31 !oebug 2598 Sep 04 12:53:31 03koen 07org.oe.dev * rc459d2db... 10/ (1 packages/sphyrna/sphyrna_svn.bb): sphyrna: convert to SRCREV Sep 04 12:53:33 * * Bug 2598, Status: NEW, Created: 2007-07-04 07:06 Sep 04 12:53:34 * * : angstrom simpad gpe: metabug suspend with power button does not work Sep 04 12:53:35 * * http://bugs.openembedded.org/show_bug.cgi?id=2598 Sep 04 12:53:37 03koen 07org.oe.dev * r659ab8c0... 10/ (1 packages/xoo/xoo_svn.bb): xoo: convert to SRCREV Sep 04 13:04:44 hi al Sep 04 13:04:51 s/al/all/ Sep 04 13:05:26 in what package belongs .la file? Sep 04 13:05:31 -dev Sep 04 13:05:41 see conf/bitbake.conf Sep 04 13:05:48 thanks koen Sep 04 13:08:30 morning Sep 04 13:08:30 hmm. is there currently a way to get an overview of recent changes? (with 1-click diffs etc) :) Sep 04 13:11:17 koen: angstrom-version no longer actually makes a .ipk Sep 04 13:14:36 XorA: indeed Sep 04 13:14:50 XorA: and I know why Sep 04 13:14:52 koen: makes image building a touch difficult :-) Sep 04 13:14:57 koen: ah cool Sep 04 13:15:01 koen: new changes? Sep 04 13:15:31 yes Sep 04 13:15:35 see commit message :) Sep 04 13:20:00 03koen 07org.oe.dev * r617f82b6... 10/ (1 packages/angstrom/angstrom-version.bb): angstrom-version: ${D} gets deleted on do_install, so install the files in do_install instead of do_compile Sep 04 13:20:08 XorA: there you go Sep 04 13:20:29 koen: cheers Sep 04 13:25:49 it had me wondering for a while Sep 04 13:28:58 anyone know much about gumstix? Sep 04 13:29:09 like how to assemble them? Sep 04 13:39:15 Crofton|home: the should fit just on top of eachother Sep 04 13:40:09 they do, but I hae no power cable, only usb :) Sep 04 13:40:13 hello Sep 04 13:40:37 I have a card with a usb piort, but it suggests I cannot power it this way Sep 04 13:41:43 hi cbrake Sep 04 13:42:17 djay-il: hello Sep 04 13:42:32 its been awhile for me... Sep 04 13:42:40 I wasn't around :-) Sep 04 13:42:41 djay-il: what are you doing these days? Sep 04 13:42:57 cbrake: I started a new job 2 days ago Sep 04 13:43:59 Crofton|home: judging from http://gumstix.com/store/catalog/product_info.php?cPath=26&products_id=170 you can power it via usb Sep 04 13:45:09 http://gumstix.com/store/catalog/product_info.php?products_id=196 Sep 04 13:45:49 cbrake:besides, I'm getting married :-) Sep 04 13:46:06 djay-il: contrats -- lots of changes :-) Sep 04 13:46:18 cbrake: indeed Sep 04 13:46:28 and I just moved to a new appartment Sep 04 13:46:55 and sort of made myself an independent business Sep 04 13:47:07 sort of freelance/consulting... Sep 04 13:47:32 koen, I did get a screen with the gusmtix Sep 04 13:48:49 http://gumstix.com/store/catalog/product_info.php?products_id=195 Sep 04 13:49:09 no touch screen Sep 04 13:51:42 "Samsung LCD panel with resistive touchscreen overlay" Sep 04 13:53:14 doh Sep 04 13:54:19 good, we should be able to get openmoko gooing :) Sep 04 13:59:24 what I do not have is a serial cable ..... Sep 04 14:00:04 good morning! Sep 04 14:00:27 gm Sep 04 14:00:28 hi sakoman Sep 04 14:00:46 Crofton|home: have you figured out how to assemble your gumstix? Sep 04 14:01:00 well, I have stacked up the boards Sep 04 14:01:06 but I do not have the serial adapter Sep 04 14:01:13 or a power supply Sep 04 14:01:18 board or cable? Sep 04 14:01:54 So they didn't send you the wall wart? Sep 04 14:02:00 no wall wart Sep 04 14:02:04 Argh!!!! Sep 04 14:02:08 yeah Sep 04 14:02:11 That sucks! Sep 04 14:02:18 also he sent a USB cable, but it does not fit Sep 04 14:02:30 What boards did they send? Sep 04 14:02:31 I think the serial boad may be newish Sep 04 14:02:33 or something Sep 04 14:02:35 not sure Sep 04 14:02:55 I just assembled the board stack Sep 04 14:02:56 How many mini-din connectors on the serial board? Sep 04 14:03:17 3 Sep 04 14:04:04 To start out you probably just want to use the serial board (console vx) and the processor board (verdex) Sep 04 14:04:20 Did they send a board with ethernet? Sep 04 14:04:23 yeah Sep 04 14:04:31 it has wired and wireless Sep 04 14:04:49 I add that after you get the basics working Sep 04 14:05:19 And they didn't send a serial cable either? Sep 04 14:05:56 I need to drop Craig a note Sep 04 14:06:09 Did they send a "screws & spacers" kit? Sep 04 14:07:12 yes Sep 04 14:07:19 that seems pretty important :) Sep 04 14:07:20 If not you might want to use a little double stick foam tape to hold things together more securely than just with the connector Sep 04 14:07:59 The screws and spacers will try your patience if you do much assembly/disassembly :-) Sep 04 14:08:03 based on what I see, the should not ship anything without the S&S kit Sep 04 14:08:13 they already have .... Sep 04 14:08:48 I use screws on my product, but double stick foam on the bench Sep 04 14:09:50 I'll drop Craig a note and see if we are missing some info Sep 04 14:11:01 I think he'll be embarrassed that you are missing a couple of the basics to get started Sep 04 14:11:44 Too bad you aren't closer, I have extras of both missing pieces in the drawer Sep 04 14:12:35 I'm sure Sep 04 14:12:57 but, these are just things to be worked out Sep 04 14:13:26 I'm jealous, you've got the cool new stuff! Sep 04 14:13:42 yes, but I am sure it needs more power :) Sep 04 14:13:55 Yeah, and costs more :-) Sep 04 14:15:14 There was something on the "internet" last week about solar really starting to take off Sep 04 14:15:46 Yeah, the gov incentives are helping Sep 04 14:16:07 bbiab Sep 04 14:18:11 all ... what did everyone decide was the cleanest way to NOT include the uImage in the final image. I know I had a rootfs preprocess command that deleted it, but there was another way of setting the kernel iamge to "" but I can't recall the syntax Sep 04 14:24:49 christopher: look at zaurus config Sep 04 14:32:30 hrw: thanks. RDEPENDS_kernel-base = "" got it Sep 04 14:39:52 christopher, that goes in machine.conf, right? Sep 04 14:42:59 Crofton|home: it was in a zaurus*.inc file. I was going to put mine in my machine file Sep 04 14:44:41 I am going to do the same for the gumstix Sep 04 15:07:12 so RDEPENDS_kernel-base = "" is preferred over FILES_kernel-image = "" Sep 04 15:07:48 in the kernel recipe. Looks like I should probably update a few of my machines/kernels ... Sep 04 15:07:58 cbrake, I have been gone a few days ... Sep 04 15:08:21 but I think this should stop the dangling symlink also .. Sep 04 15:08:28 I should know later today Sep 04 15:24:08 Crofton: ping Sep 04 15:24:31 Jin^eLD, pong, but I'll be gone for a bit Sep 04 15:25:08 Crofton|work: I am still trying to build the davinci stuff, but can't because git is failing to get things Sep 04 15:25:20 uboot_git can not be fetched Sep 04 15:25:36 and also problem with the kernel Sep 04 16:01:34 XorA|gone: yes? Sep 04 16:04:25 XorA|gone: wake up :) Sep 04 16:05:58 zecke: could you make feedread switch tabs if I double tap on an entry? Sep 04 16:06:32 it's counter intuitive to see nothing happen when tapping on an entry and having to manualy got back and forth between the first and second tab Sep 04 16:09:36 koen: seeing nothing? is highlight not working? Sep 04 16:09:48 it hightlights Sep 04 16:09:53 koen: I agree with automatically switching tabs, how would one switch back? Sep 04 16:10:08 switching back can be done manually Sep 04 16:10:19 koen: for switch back I have the <- and -> buttons as replacement, then I need some clever way to merge 'feeds' Sep 04 16:10:28 AFAIK lifera can do that Sep 04 16:10:48 sakoman: the feed configs are going to be done in a less annoying way Sep 04 16:11:22 sakoman: the plan is to have ANGSTROM_FEED_CONFIG somewhere that specifies which _recipe_ to use, so you can add your own recipes Sep 04 16:12:59 03koen 07org.oe.dev * r8c69b6f5... 10/ (1 conf/distro/include/angstrom.inc): angstrom.inc: include a var that selects which package will provide the feed configs Sep 04 16:13:05 03koen 07org.oe.dev * re6b5c0b0... 10/ (1 packages/angstrom/angstrom-version.bb): angstrom-version: and bump PR Sep 04 16:13:11 03koen 07org.oe.dev * ra0ad096e... 10/ (1 packages/angstrom/angstrom-feed-configs.bb): angstrom-feed-configs: mark everything as conffiles so user edits get preserved Sep 04 16:14:04 is anyone creating a custom UI with gtk using OE ? I am wondering how to compile using OE ...even if it is one of the examples. when I do a make on one of those in the gtk folder, it fails due to not finding a pango library. Sep 04 16:19:51 koen: check how that change works and if it conflicts with scrolling or other abnormal behaviour Sep 04 16:22:15 will do after I get a toolchain up Sep 04 16:22:22 thanks for looking into it Sep 04 16:27:01 koen: OK, thanks for the "heads up" Sep 04 16:27:11 are those the changes that just went in? Sep 04 16:27:23 sakoman: yes, but they don't do anything yet Sep 04 16:27:33 OK :-) Sep 04 16:32:45 RP: does that prelink patch solve the problem we saw with prelinking wget earlier? Sep 04 16:41:18 03pfalcon 07org.oe.dev * r91c7d956... 10/ (1 packages/qte/qte-mt-static_2.3.10.bb): Sep 04 16:41:18 qte-mt-static: Set DEFAULT_PREF=-1, this is hightly adhoc target. Sep 04 16:41:18 * To help a bit people who try to build OPIE w/o reading instructions. Sep 04 16:42:45 christopher: Hmm, what kind of example do you need? Your app is using standard autotools? Sep 04 16:42:52 christopher: pkgconfig to find gtk+? Sep 04 16:43:06 03koen 07org.oe.dev * rf187242e... 10/ (1 conf/distro/angstrom-2007.1.conf): angstrom 2007, 2008: switch to binutils 2.18 to get rid of the warning spew in stripping and gdb Sep 04 16:45:00 koen: if you have time to read, read the gcc summit 2003 proceedings Sep 04 16:49:00 * koen makes a note Sep 04 17:08:30 are there any language translation applications in oe? Sep 04 17:10:06 Crofton: did you get a console-lcd board in the package? Sep 04 17:10:13 it has 3 mini-din connectors on it Sep 04 17:10:22 yeas Sep 04 17:10:27 if not, sounds like the packing/shipping team really messed up your shipment Sep 04 17:10:30 oh ok Sep 04 17:10:34 but no wall wart Sep 04 17:10:44 and no mini-din to serial Sep 04 17:10:57 ah ok no minidin to serial Sep 04 17:10:59 I see Sep 04 17:11:01 I assembled it and admired the assembly :) Sep 04 17:11:08 I thought you meant the baord was missing not the sable Sep 04 17:11:14 ah Sep 04 17:11:43 ok, I'll get them to send the serial null modem and wall wart then, and the correct USB cable this time... Sep 04 17:16:21 * koen tries to find the snag in the 20Mbit/s for €20/month subscription Sep 04 17:16:47 "offer only valid in Alaska on thursdays" Sep 04 17:17:03 * koen checks Sep 04 17:17:58 I found the "after 6 months €24/month" clause Sep 04 17:18:11 and the "need a phoneline clause" Sep 04 17:18:15 ah Sep 04 17:18:17 found it Sep 04 17:18:30 "you live in an 8 mbit/s area" Sep 04 17:18:39 stupid adsl Sep 04 17:19:52 heh, my mind isn't fully focused atm Sep 04 17:19:52 my aunt is in hospital Sep 04 17:20:14 I hope she gets well soon Sep 04 17:20:22 sable? Sep 04 17:22:33 s/sable/cable/ Sep 04 17:24:12 gah, I detest these soft math courses Sep 04 17:24:34 anything that can be explained with marbles, coins or dice is not math Sep 04 17:25:24 hughescr, that sounds good Sep 04 17:25:51 bummer, I though I learned a secret code name Sep 04 17:25:54 * mwester dimly remembers arithmetic with marbles, coins, dice, and other small inanimate objects... Sep 04 17:26:14 phdcomics.com is bashing humanties majors atm .... Sep 04 17:26:22 koen, probability? Sep 04 17:26:30 easy targets for bashing, though. Sep 04 17:26:31 Crofton: yes Sep 04 17:28:07 koen, probability is very important to communication engineers Sep 04 17:28:07 there are guys in my office who look at BER graphs all day Sep 04 17:28:11 and claim you are not a comm engineer unless you are creating new BER plots all day Sep 04 17:28:13 they are wrong Sep 04 17:28:31 Crofton: I get all the stuff in our com theory courses Sep 04 17:28:39 but I'm made to endure the basics Sep 04 17:28:53 (basically I forget the exam date each year) Sep 04 17:29:58 Crofton: send some of the BER-creators on a critical customer network problem, armed with some test gear -- a taste of the real world is often a good cure for PHD-itis. Sep 04 17:30:40 take "Stochastic Signal Processing" :) Sep 04 17:31:02 Crofton: I want that "laser in use sign" :) Sep 04 17:31:19 http://www.phdcomics.com/comics/archive.php?comicid=906 Sep 04 17:31:22 the yellow one Sep 04 17:31:56 hehehe Sep 04 17:32:02 It took a moment, but I figured out what koen is going on about Sep 04 17:34:57 * * OE Bug 2917 has been created by autobuild(AT)openembedded.org Sep 04 17:35:00 * * angstrom-console-image-1.0-r0-do_rootfs Sep 04 17:35:01 * * http://bugs.openembedded.org/show_bug.cgi?id=2917 Sep 04 17:35:29 christopher: many of my project use custom gtk apps -- the best way I've found is create a project in svn using standard autotools build system. Get it all working on a x86 system. Sep 04 17:36:09 christopher: then create a recipe that fetches it from svn and builds it -- much like any of the other recipes. If you can get it working on x86, and have the dependencies built, it usually just works. Sep 04 17:36:54 christopher: this is where you really start to appreciate autotools, and "inherit autotools" Sep 04 17:37:25 and for libs 'inherit autotools pkgconfig' :) Sep 04 17:37:56 koen, you should talk to sakoman about an image in between minimal and console Sep 04 17:37:57 * * OE Bug 2917 has been RESOLVED (WONTFIX) by Sep 04 17:37:59 * *  angstrom-console-image-1.0-r0-do_rootfs Sep 04 17:38:01 * * http://bugs.openembedded.org/show_bug.cgi?id=2917 Sep 04 17:39:39 RP: I see a lot of errors when using nfs to store sources when using latest bitbake Sep 04 17:41:00 RP : http://rafb.net/p/FqKQkY86.html Sep 04 17:52:32 bah Sep 04 17:52:49 the day I receive the orange filter for my camera I get clear skies Sep 04 17:54:14 is it possible to disable this 'DEBUG: SVN fetcher hitting network for svn://gtk-webcore.svn...' when updating BB's cache? Sep 04 17:54:42 i think it's src_uri or so Sep 04 17:54:51 SRCREV Sep 04 17:55:07 ok whatever, it's slow as hell :) Sep 04 17:58:28 it's about 10 min. for 50% of packages now Sep 04 18:01:30 koen, now I understand why are you looking for fast adsl connection :) Sep 04 18:05:27 koen: Just a bitbake question. What is this lock thing that's suddenyl required? http://rafb.net/p/QVljk184.html Sep 04 18:05:50 bitbake -c rebuild libdmx works Sep 04 18:06:14 Not all recipies fail either, so far only the X stuff Sep 04 18:12:09 does this dir exists? /home/moko/build/tmp/work/armv4t-angstrom-linux-gnueabi/libdmx-1_1.0.2-r1/install/ Sep 04 18:15:04 ynezz: I just rebuilt libdmx, but now it failed on libpcap. I'll look at the same line Sep 04 18:16:03 ynezz: No, it doesn't exist. Instead of install there's a temp dir Sep 04 18:24:15 * CM gives up and does something productive instead Sep 04 18:38:39 Crofton, Koen, trying to do an angstrom-minimal-image build for gumstix-verdex, it died on task-boot and said missing dependency update-modules, so i did a bitbake update-modules, everything went OK, then I tried a bitbake angstrom-minimal-image again and I am getting missing dependency module-init-tools-depmod, any ideas? Sep 04 18:38:45 http://pastebin.com/m397eebc4 Sep 04 18:39:58 not yet Sep 04 18:40:19 I have my auto builder running, so it should appear in bugzilla sometime Sep 04 18:40:24 if it happens for me Sep 04 18:40:47 cbrake: do you know of any tutorials that are helpful for doing that? that is, building and using OE as a framework? Sep 04 18:42:27 zecke: ping Sep 04 18:42:57 µpong Sep 04 18:43:27 asmola_, I am running an omap5912osk build atm Sep 04 18:43:46 zecke: Have you still a bit space for the nights at OEDEM? After some rework it seems OEDEM would fit nice between vacation and my .tw trip. ;) Sep 04 18:43:49 we'll see if the problem is global, is specific to gumstix Sep 04 18:43:54 sounds good Sep 04 18:44:03 I am on glibc atm Sep 04 18:44:21 stefan_schmidt: It is a Hilbert-Room. There is always place for one more chair Sep 04 18:44:30 once I finish the first machine, the rest go fast Sep 04 18:44:33 zecke: :D Sep 04 18:44:45 zecke: Good to know. Sep 04 18:45:02 CM: remove tmp and rebuild Sep 04 18:45:04 zecke: At least all the 'old' guys will go for a hotel anyway. ;) Sep 04 18:45:37 stefan_schmidt: oh, let me reread but the answer doesn't change Sep 04 18:45:53 * koen also reserves some space in the hilbert room Sep 04 18:46:03 heh Sep 04 18:46:36 koen: btw, I had some thoughts about the kernel patch stuff between ezx and oe. Sep 04 18:46:56 koen: Your goal is to have an always buildable kernel in oe, right? Sep 04 18:47:03 we will have enough beds for the poor students employed by certain .tw based companies even if bjoern and doku are not aware of their luck :} Sep 04 18:47:18 koen: My goal is to be able to build the latest svn stuff even if it would fail. Sep 04 18:47:33 stefan_schmidt: an always buildable kernel being the default, yes Sep 04 18:47:42 zecke: heh. I have tp use my money to drink with you. ;) Sep 04 18:48:13 koen: hmm, just one bb with known good SRCREV and one with latest and default -1? Sep 04 18:48:36 koen: Once we one the newest version works fine we can just update the SRCREV Sep 04 18:48:47 I still extremely dislike fetching patches from svn for the reasons mentioned Sep 04 18:48:54 koen: Would make it easier for you and we would not need to pull the patches in. Sep 04 18:49:07 if we are going to have a stable and toy kernel we can put hte stable patches in OE Sep 04 18:49:17 koen: Then only build releases with miorrored tar balls? Sep 04 18:49:38 koen: hmm, how is this handled with linux-rp? Sep 04 18:49:48 linux-rp is not really a good example Sep 04 18:50:14 koen: But I consider to base newer ezx stuff on it. :) Sep 04 18:50:29 After a lot poking from mickeyl :) Sep 04 18:51:57 zecke: I had some chats with your headhunter at MDD. Nice guy. He was excited about OpenMoko with sound on the A780. :) Sep 04 18:52:46 * zecke checks for his head Sep 04 18:53:13 zecke: Hard to find, eh? Sep 04 18:54:00 christopher: no, I don't know of anything offhand. I you want to send me an email (cbrake@bec-systems.com) with some specific questions, I'll consider them for my next blog post. I think how to set up a custom application would be a good next topic. Sep 04 18:54:41 christopher: one of the biggest benefits of OE is having a somewhat standard way to do things -- you can focus on real work, instead of creating yet another build system. Sep 04 18:54:44 cbrake: ok. I can do that. OE can be overwhelming to a newbie and I am trying to understand what part to customize and how to use what is already there. Sep 04 18:55:19 cbrake: yeah, exactly. I'd like to have a base system with X, and gtk support and just do a hello world example. and then, I can follow that around and go from there Sep 04 18:56:05 cbrake: I have a couple of images that mount etc, but they are running a lot more things that I need in my system and end up not working or crashing. Sep 04 18:56:55 christopher: angstrom-console-image has never given me a lot of problems, but I've not built a fresh image in the last week or two. Sep 04 18:57:19 hmm, I'm the only one with failing glibc_2.5 for a new build from scratch? Sep 04 18:57:53 * stefan_schmidt searches bugzilla Sep 04 19:00:07 cbrake: does that have x-windows support? Sep 04 19:00:28 cbrake: or kdrive I should say Sep 04 19:00:53 cbrake: and that has the gtk libraries for running gtk apps? if so, that would be the place to start. Sep 04 19:01:36 christopher: no, but you can extend angstrom-console-image pretty easily: http://bec-systems.com/web/content/view/79/9/ Sep 04 19:02:49 christopher: the more difficult task is how to slim down agnstrom-console-image, but that should be doable with a custom image recipe as well. Sep 04 19:03:21 cbrake: ok. what size are most of your images with x and gtk support? 6-8 mb Sep 04 19:03:22 ? Sep 04 19:03:25 christopher: I'm still running in bloat mode on most of my current projects -- no time for optimization right now :-\ Sep 04 19:04:03 cbrake: ok, I can understand that. Sep 04 19:04:18 christopher: 18-30MB JFFS2, but some of those contain quite of bit of custom application stuff Sep 04 19:04:54 cbrake: oh wow. That's bigger than I thought. i think the gpe x11 image I built earlier was around 26 or so Sep 04 19:05:12 christopher: but, with 128MB of NAND flash I don't really care what the flash footprint is :-) Sep 04 19:05:52 christopher: at least not right now. Yeah, I think you could build a stripped down X+GTK image in about 15-20MB JFFS2 size. Sep 04 19:06:12 cbrake: ok. I'd be fine with that. if I can get it to work. :) hehe Sep 04 19:06:44 christopher: well, pretty soon things will start to click and you'll be glad you stuck with it :-) Sep 04 19:07:35 cbrake: hope so. so, I see the bb file you are showing on your site. would you then add gtk depends to that file? Sep 04 19:08:18 christopher: yes, that should work. Sep 04 19:09:33 stefan_schmidt: linux-rp is a great kernel due to the patches it has, but not a great recipe Sep 04 19:09:38 not a bad recipe either Sep 04 19:10:19 does it looks ok? http://ynezz.true.cz/ts72xx-oe.patch Sep 04 19:11:06 just rewritten it to use linux_.bb Sep 04 19:11:12 does linux/ts72xx/2.6.22.6/ actually work? Sep 04 19:11:19 yes Sep 04 19:11:30 I'd use linux-2.6.22.6/ts72xx/ Sep 04 19:11:32 i've rebuild both and they boots Sep 04 19:11:35 koen: moment, I try to find out why glibc header files are ending on .x instead of .h here. Sep 04 19:12:00 stefan_schmidt: there's some sed magic in packages/glibc iirc Sep 04 19:12:18 ok, I'll rewrite it Sep 04 19:13:05 uh, shell. Anybody can tell what this should do? h=`echo $r|sed -e's,\.x$,.h,'` Sep 04 19:13:19 * stefan_schmidt feels as he would read perl. Guilty. ;) Sep 04 19:13:21 could you put DEFAULT_PREFERENCE = "-1" next to the DEFAULT_PREFERENCE_ts72xx = "(-)1" lines? Sep 04 19:13:33 ynezz: apart from that it looks ok Sep 04 19:13:42 ok, fine Sep 04 19:13:46 attach it to the bugtracker with that change and it looks ready to commit Sep 04 19:14:18 I'll be away tomorrow, so if you want it in fast, poke someone else ;) Sep 04 19:16:00 OK, it seems this take me longer to understand. Sep 04 19:16:08 koen: Back to the patches. Sep 04 19:17:48 koen: I still don't think pulling all patches, even only the known good ones, into OE is such a great idea. If you dislike pulling it from svn you would need to pull a lot more stuff in oe. Tarballs for the stable version are not enough? Sep 04 19:18:08 koen: In the end we should do tarball more often anyway. :) Sep 04 19:18:19 s/tarball/tarballs/ Sep 04 19:30:03 stefan_schmidt: I'd like to have a controlled way to tweak single patches Sep 04 19:30:44 stefan_schmidt: linux-rp uses single patches :) Sep 04 19:31:17 stefan_schmidt: the toy kernel can pull patches from svn directly if it wants, but the user kernel won't change often Sep 04 19:32:41 koen: You would like to tweak our patches? They are perfect. ;) Sep 04 19:33:27 koen: Is there another kernel OE do it the same way? Sep 04 19:33:52 koen: I mean you can alway have 'distro' patches inside of OE Sep 04 19:38:58 * * OE Bug 2918 has been created by ynezz(AT)true.cz Sep 04 19:38:59 * * Support for Technologic Systems TS-72xx SBC Sep 04 19:39:01 * * http://bugs.openembedded.org/show_bug.cgi?id=2918 Sep 04 19:39:09 * * OE Bug 2919 has been created by craig(AT)gumstix.com Sep 04 19:39:11 * * Create a distro for gumstix Sep 04 19:39:13 * * http://bugs.openembedded.org/show_bug.cgi?id=2919 Sep 04 19:39:49 sakoman: my patches are now all in OE bugs Sep 04 19:48:37 crofton, how is the osk build going? Sep 04 19:54:59 stefan_schmidt: linux_*.bb :) Sep 04 19:55:57 * * OE Bug 2920 has been created by cwixon(AT)usa.net Sep 04 19:55:59 * * New Recipe: Free42 RPN Calculator Sep 04 19:56:01 * * http://bugs.openembedded.org/show_bug.cgi?id=2920 Sep 04 20:03:17 NOTE: the following files were installed but not shipped in any package: Sep 04 20:03:17 NOTE: /usr/lib/preloadable_libiconv.so Sep 04 20:03:20 hmmm Sep 04 20:03:44 (yes, I'm building uclibc, woglinde, you can relax now) Sep 04 20:04:16 crunching along Sep 04 20:04:27 * koen launches second autobuilder Sep 04 20:04:31 I think I set it to only do one process Sep 04 20:04:36 this time for .deb packages Sep 04 20:05:08 now I have 2 machines that will spam bugzilla for new packages I want to add :) Sep 04 20:05:19 koen, do we have any "policies" on distros derived from angstrom? Sep 04 20:06:11 yes Sep 04 20:06:19 "Don't do it*" Sep 04 20:06:25 heh Sep 04 20:06:30 * unless you have a _very_ good reason Sep 04 20:06:43 how can we accomodate Craig's gumstix issues? Sep 04 20:07:14 I am going to work though the gumstix patches and knock out the easy stuff first Sep 04 20:07:57 I suspect there will be "loss of control" issues moving from a buildroot system they had complete control over Sep 04 20:08:25 * Crofton observes that so many problems are really "social issues" not technical :) Sep 04 20:08:32 so put someone from gumstix (i.e. craig) in the angstrom-core-team Sep 04 20:09:28 then the "core team" grows every time someone wants to do real work with it Sep 04 20:09:43 I guess so Sep 04 20:09:49 and becomes dominated by commericial interests Sep 04 20:09:51 the core team thingy wasn't my idea Sep 04 20:09:58 * * OE Bug 2921 has been created by autobuild(AT)openembedded.org Sep 04 20:10:00 * * ltrace-0.4-r1-do_compile Sep 04 20:10:02 * * http://bugs.openembedded.org/show_bug.cgi?id=2921 Sep 04 20:10:26 These are general issues, that hopefully become more common Sep 04 20:10:30 is my thinking Sep 04 20:10:57 so developing systems to support vendors who want to use OE is really important Sep 04 20:11:18 right Sep 04 20:11:56 experience learns that having your own distro is great for the first week and then it bitrots Sep 04 20:12:00 rumour has it I have a new PC here I can install OE on Sep 04 20:12:09 having a derivative distro jsut delays that bitrot :) Sep 04 20:12:13 so I stop doing builds on the SCM/webserver Sep 04 20:12:40 this is one of my concerns Sep 04 20:12:50 IMNSHO the "too many vars in local.conf" can be fixed by providing a sample local.conf Sep 04 20:13:23 since uclibc is practically unmaintained nowadays, I foresee a rising interest in using glibc on gumsticx Sep 04 20:13:30 especially since OE makes it so easy Sep 04 20:13:50 Sep 04 20:13:58 sadly, he did not send me the lowest end gumstix Sep 04 20:14:14 apparently they have disk space issues Sep 04 20:14:37 right Sep 04 20:14:43 I don't debate that Sep 04 20:14:59 and I suspect a number of his problems are related to underlying issues we should fix anyway Sep 04 20:15:07 OTOH I've been known to overestimate the reading skills of users Sep 04 20:15:10 * khem chuckles Sep 04 20:16:21 koen, http://ossie.wireless.vt.edu/trac/ticket/58 Sep 04 20:16:54 from a fellow who has declared that he "is not a Linux developer" Sep 04 20:17:13 apparently sudo make install is testing some people's limits Sep 04 20:17:14 koen: Sorry, distracted. OE is really pulling all that kernel patches in? It should not be me to stop this as long as I have my toy recipe to play with the newest stuff. Any opinions on others about that? Should we write a mail to oe-devel? Sep 04 20:17:43 asmola_, I am on 504 of 584 tasks for the omap5912osk Sep 04 20:17:57 * * OE Bug 2922 has been created by  Sep 04 20:17:59 * * qpe-mahjongg - replace some sick code Sep 04 20:18:01 * * http://bugs.openembedded.org/show_bug.cgi?id=2922 Sep 04 20:18:08 stefan_schmidt: write to oe-devel and keep in mind that not everyone is a leet developer Sep 04 20:18:29 koen: heh, will do. Sep 04 20:18:57 * * OE Bug 2923 has been created by  Sep 04 20:18:59 * * libsdl_net doesnt compile - looking for host includes Sep 04 20:19:01 * * http://bugs.openembedded.org/show_bug.cgi?id=2923 Sep 04 20:19:11 Crofton, mine died right around that point Sep 04 20:19:18 "oe should only have patches that " translates to "I'm soooo leet and lazy" for me Sep 04 20:19:35 * koen takes of the oe-user hat and goes to sleep Sep 04 20:19:38 'night all Sep 04 20:19:45 gn Sep 04 20:19:51 then i started punching in some random bitbake commands and now it looks to be at 250 or so Sep 04 20:19:54 gn koen Sep 04 20:19:58 Crofton: I think gumstix is going to stress the OE conventions Sep 04 20:20:04 night Sep 04 20:20:06 yeah Sep 04 20:20:15 There's not really a single "machine" Sep 04 20:20:31 lots of combinations of hw Sep 04 20:20:50 And each gumstix OEM is gonna want to control which kernel modules get installed and autoloaded Sep 04 20:20:55 we have the goal of one distro yo unite them all Sep 04 20:21:12 I suspect we will stress monotone Sep 04 20:21:28 it should be possible for someone to maintain a local branch in mtn Sep 04 20:21:35 Yeah, not sure what the right answer is Sep 04 20:21:47 pull from main OE server and move stuff into the branch in a controlled fashion Sep 04 20:22:02 That's what I am thinking Sep 04 20:22:08 arguably buildroot has the same issue, but with the gui config, users are able to easily pick and choose what things they want / don't want Sep 04 20:22:16 also some people just stop pulling for certain projects Sep 04 20:22:16 right Sep 04 20:22:28 likewise does that I think Sep 04 20:22:42 For most commercial products that is always the case Sep 04 20:22:54 Too bad we do't have the OEDNAM Sep 04 20:23:02 You branch at some point in the release cycle Sep 04 20:23:25 you also need to work with your SCM system Sep 04 20:23:36 And do any future updates by hand and only after really careful benefit analysis :-) Sep 04 20:23:58 * Crofton has had really stupid conversations about release and development management with people who do understand these things lately Sep 04 20:24:16 well, we are supposed to have tools that make this better :) Sep 04 20:24:35 a secret: no one does this well :-) Sep 04 20:24:57 lets just say, I have heard people say really dumb things Sep 04 20:25:18 hopefully I haven't been one of them :-) Sep 04 20:25:21 no Sep 04 20:25:27 this is school related stuff Sep 04 20:25:32 ah Sep 04 20:25:47 electrical engineers make lousy software guys Sep 04 20:25:55 generally speaking of course Sep 04 20:26:02 and vice versa! Sep 04 20:26:06 yep Sep 04 20:26:18 and this is what makes software defined radio hard Sep 04 20:26:49 indeed Sep 04 20:27:15 http://www.openembedded.org/wiki/OEDEM2007 Sep 04 20:27:27 we should enter some of this discussion here Sep 04 20:28:21 try to get some discussion going on how people use OE for real, and how OE can better support the "customer" base Sep 04 20:28:39 are you going? Sep 04 20:29:47 probably not Sep 04 20:30:16 there is an outside chance, but it is very complicated Sep 04 20:30:50 hi Sep 04 20:30:53 koen: Mail sent. Should clear the positions. If I missed something don't hesitate to clarify it. Sep 04 20:30:54 unless monay falls from the sky Sep 04 20:31:16 understood, it would be an expensive trip! Sep 04 20:31:24 there is a way to get multiple gumstix configs supported in OE\ Sep 04 20:31:50 also vendors which use OE and gumstix can make own configs (for own usage) Sep 04 20:31:55 hrw, let mw get mine booting and I can start trying to work though the issues Sep 04 20:32:24 we need to describe the "best" way to do that, and end up with a bitrotted database Sep 04 20:32:39 hrw: yes, that's what I assume I will do "worst case" Sep 04 20:33:17 it is always good to attempt to do the right thing, then compromise to make deadlines Sep 04 20:33:25 exactly Sep 04 20:33:29 now we have gumstix-connex and gumstix-verdex - rigt? Sep 04 20:33:34 yes Sep 04 20:34:25 asmola_, omap5912osk built Sep 04 20:34:29 if there is gumstix-verdex-umbaumba which adds pcmcia/cf and wifi then it can be done in a way which will use gumstix-verdex stuff but be different to it Sep 04 20:34:41 OVERRIDES can be set in machine configs Sep 04 20:35:39 crofton, ok, good to know, mine is cranking away again, we'll see how far it goes Sep 04 20:36:00 hrw: so gumstix-verdex-umbaumba.conf would "require" gumstix-verdex and then set the overrides? Sep 04 20:36:30 sakoman: yes, and will add "pcmcia" to MACHINE_FEATURES Sep 04 20:37:08 obviously the solution is for gumstix to give us all slightly different hw to test on :) Sep 04 20:37:38 so there are a bunch of gumstix-verdex-*, but then how are these combined if you are using say pcmcia and network? Sep 04 20:37:41 roa -> Berlin is $758 :) Sep 04 20:38:09 asmola_work: network is smc91xx connected to pxa directly or pcmcia? Sep 04 20:38:12 a week in Berlin, not so cheap Sep 04 20:38:25 hrw: in that case, it would be best if gumstix-verdex just handles the bare verdex cpu module Sep 04 20:38:34 I think directly, not sure xactly what pins are on the connector though Sep 04 20:39:04 and then there were also gumstix-verdex-goliath, etc override conf files Sep 04 20:39:13 or we can add MACHINE_GUMSTIX_* variables whic user will set in local.conf Sep 04 20:39:31 MACHINE_GUMSTIX_NETWORK = "1" # enable smc91xx module Sep 04 20:39:47 MACHINE_GUMSTIX_PCMCIA = "1" # enable CF extension Sep 04 20:39:48 etc Sep 04 20:39:51 hrw, i think directly as well, i have the schematics and eagle on my home computer but not here Sep 04 20:40:10 I am wondering if the connector is a form of pcmcia ... Sep 04 20:40:13 asmola_work: as long as it is not cf/pcmcia it does not matter how it is Sep 04 20:40:50 but if it is cf/pcmcia but not allow to connect other card then it is to be handled in a bit other way Sep 04 20:41:21 crofton, i don't beleive so because i know that on the hybrid board you have, the wireless is cf/pcmcia Sep 04 20:42:14 which is why the hubrid board (ethernet & wireless) doesnt have the cf card slot Sep 04 20:42:34 asmola_work: wireless is marvel 8385? Sep 04 20:44:29 hrw: yes Sep 04 20:44:49 although they don't have a board released with wireless for the verdex ( although I think crofton may have one ) Sep 04 20:45:03 :) Sep 04 20:46:25 I need to extract libertas update for 2.6.23-rc3 from libertas tree and check that card in zaurus ;D Sep 04 20:50:40 crofton, yeah made it to 553 of 557 again, grrr Sep 04 20:51:06 can you go ahead and pastebin the error while I am waiting on git? Sep 04 20:51:20 yeah, doing that rightnow Sep 04 20:51:44 http://pastebin.com/m4143fde5 Sep 04 20:52:04 i got that last time so i did a bitbake update-modules Sep 04 20:52:21 angstrom-2007.1 and gumstix-verdex Sep 04 20:52:24 yes Sep 04 20:52:26 when did you update last? Sep 04 20:52:37 noon or so today Sep 04 20:52:48 i have enable_binary_locale_generation=0 Sep 04 20:52:53 heh Sep 04 20:52:54 me too Sep 04 20:53:07 we don't need no steenkin locale Sep 04 20:53:23 :) Sep 04 20:55:30 I wonder if the module_autoload lines are pulling in update-modules .... Sep 04 20:55:50 if i build update-modules then i get this... Sep 04 20:56:07 drum roll Sep 04 20:56:28 http://pastebin.com/m3b4fd756 Sep 04 20:56:50 hrw, ping Sep 04 20:58:01 what happened to angstrom-console-image? Sep 04 20:58:27 ? Sep 04 20:58:45 can you look at asmola_work 's pastebins? Sep 04 20:58:46 Crofton: you do from scratch? Sep 04 20:58:52 I'm hoping you know the answer Sep 04 20:58:58 asmola_work, did Sep 04 20:59:03 hrw, that was from scratch Sep 04 20:59:10 I am building omap5912osk, then gumstix Sep 04 20:59:22 I am almost to the same point on gumstix Sep 04 20:59:30 my from scratch build is on glibc now ;( Sep 04 20:59:37 if it doesn't break on your computer i'm thrrowing mine out the winodw Sep 04 20:59:43 heh Sep 04 20:59:54 asmola_work: is it fast? Sep 04 20:59:55 except there aren't any windows here (probably not an accident) Sep 04 21:00:09 we try really hard to isolate from the host, but it is really hard .... Sep 04 21:00:18 is my computer fast? Sep 04 21:00:58 im running multiple threads on the build since i am on a dual core Sep 04 21:02:09 asmola_work: so throw it under my door ;D Sep 04 21:02:15 heh Sep 04 21:02:17 :) Sep 04 21:02:23 long throw Sep 04 21:02:41 looks like I will have to do two new builds ;( Sep 04 21:05:21 i am going to run home, i'll be back online in about 20 Sep 04 21:06:52 ok Sep 04 21:07:32 I have a machine here I can set up to do builds Sep 04 21:07:43 and it will be sort of dedicated to this stuff Sep 04 21:07:45 hopefully Sep 04 21:10:05 Crazy thing: http://nsa.unaligned.org/ Sep 04 21:11:25 for machine in fic-gta01 alix gumstix-verdex c7x0 ;do echo "MACHINE='$machine'" >conf/auto.conf;bitbake angstrom-console-image;bitbake openmoko-devel-image;bitbake angstrom-x11-pimlico-image;bitbake mc strace less task-proper-tools cvs subversion lighttpd monotone;bitbake task-native-sdk;done Sep 04 21:11:36 should be enough for more then night Sep 04 21:17:09 bye Sep 04 21:20:43 bother, gumstix succeeded for me ... Sep 04 21:20:46 hi Sep 04 21:20:51 I need to build from scratch Sep 04 21:28:13 woglinde: hi Sep 04 21:29:12 hi mr_nice Sep 04 21:29:28 woglinde: do you have some time for me? Sep 04 21:30:32 hm yes Sep 04 21:30:42 woglinde: :) Sep 04 21:31:02 woglinde: can you have a look at oe bug 2598 Sep 04 21:31:45 woglinde: I wrote a patch for that simpad power off problem Sep 04 21:32:27 hm whats powerd doing? Sep 04 21:33:08 hm ah Sep 04 21:33:17 woglinde: it wathes the given evdev file for KEY_POWER. if it is pressed it does apm -s Sep 04 21:34:01 hm why the heck the kernel cant see this himself Sep 04 21:34:06 crofton, ii'm back Sep 04 21:34:21 woglinde: good question Sep 04 21:34:36 it built for me, but it was not a clean build Sep 04 21:34:53 mr_nice I will check it in tomorrow okay? Sep 04 21:35:32 yeah everything had been working fine for me but in th interest of the greater good last week i decided to build from scratch and thats when i hit this bump and then i dropped it and picked it up again today and seeing the same issue Sep 04 21:35:44 woglinde: :) Sep 04 21:35:50 I am going to make a build machine at school Sep 04 21:36:03 and try there, but it will take me a while to get it going Sep 04 21:36:12 I have started a one cpu clean build at home Sep 04 21:37:27 ok, sounds good, let me know how it goes Sep 04 21:37:41 will do, sorry about this Sep 04 21:37:52 it is a nuisance to replicate .... Sep 04 21:37:54 no need to be sorry, i can't thank you enough Sep 04 21:38:43 thank my spare clock cycles Sep 04 21:38:44 with all of this ossie/angstrom stuff i have been able to fool everyone at work into thinking that i know something :) Sep 04 21:39:05 well, we expect you to pay us back :) Sep 04 21:39:20 haha. uh oh Sep 04 21:39:55 bye Sep 04 21:42:57 asmola_, make some cool shit with what you have and show it to big shots Sep 04 21:43:24 I'm going to walk home and burn an install dvd Sep 04 21:43:50 ok, cool, talk to you later Sep 04 22:09:21 stefan_schmidt: for the ixp4xx-kernel, I pull in patches from a SRCREV on the nslu2-linux svn server. Sep 04 22:09:39 it makes it easier to keep patches in sync between OE, OpenWrt, and Debian. Sep 04 22:10:19 rwhitby: So fixed rev and no separate patches in OE? That's also the way I see it. Sep 04 22:10:36 rwhitby: Thanks. Let's wait for more comments. Sep 04 22:11:18 rwhitby: In the end we both are more the 'upstream' guys in this problem. So it make sense we have the same opinion, as it makes our life easier. :) Sep 04 22:11:57 zecke, ping? Sep 04 22:11:58 yeah, and I've had the same difference of opinion with koen on this issue too. Sep 04 22:12:04 polyonymous_: pong Sep 04 22:12:08 did I break something? Sep 04 22:12:12 zecke, yeah :) Sep 04 22:12:22 something important? ;) Sep 04 22:12:26 stefan_schmidt: I see his point, from the point of view of an OE-only viewpoint. Sep 04 22:12:30 zecke, it turns out you can't EXPORT_FUNCTION from the file containing dash in the name Sep 04 22:12:37 (qmake-base) Sep 04 22:13:15 I've even forced bitbake to emit qmake-base_do_configure function, but it only works if run.do_configure has #!/bin/bash :) Sep 04 22:13:21 stefan_schmidt: but to take his point of view to the extreme, then all downloads would need to be on OE-hosted servers just in case an upstream server goes away (and this somewhat is the case with the sources mirror logic) Sep 04 22:13:25 rwhitby: yeah, lets see how others think about this Sep 04 22:13:27 polyonymous_: for sure file a bitbake bug (even if that code in bitbake is ugly...) Sep 04 22:13:36 zecke, so, basically, the whole opie is broken. Sep 04 22:13:41 rwhitby: indeed Sep 04 22:13:43 polyonymous_: so we are back to copying this code to qmake and qmake2? Sep 04 22:13:46 zecke, well, how would you want to fix it? Sep 04 22:13:52 zecke, that's the question. Sep 04 22:14:00 zecke, if I were sure of it I wouldn't poke you :) Sep 04 22:14:24 zecke, this can be done, I think with one more ugly workaround in bitbake like replacing '-' with '_'... Sep 04 22:14:43 Or renaming qmake-base to qmake_base (but that breaks conventions) Sep 04 22:14:44 polyonymous_: file a bitbake bug and then move the code to qmake and qmake2.bbclass Sep 04 22:14:56 qmakebase ;) Sep 04 22:15:18 zecke, well, as a temporary workaround I'll add another .bbclass instead of duplicating code. Sep 04 22:15:30 to be inherited by qmake and qmake2 Sep 04 22:15:45 polyonymous_: copy+bug report and once bitbake is fixed and we depend on it we can clean it up Sep 04 22:16:05 Offhand does anyone know a good graphical patch reject resolver? Sep 04 22:16:05 zecke, ok. But what fix would you expect? Sep 04 22:16:25 zecke, I can do it myself, but I'm not sure which way to go - it's not just bitbake's problem, but also shell's. Sep 04 22:16:48 polyonymous_: oh shell :} Sep 04 22:16:50 RP, offhand, only gvim. Sep 04 22:17:01 zecke, yeah, that's the problem. I *did* fix the bitbake part :) Sep 04 22:17:06 zecke, but it only works for bash. Sep 04 22:17:23 polyonymous_: rename functions on the fly Sep 04 22:17:35 zecke, s/-/_/? Sep 04 22:17:36 polyonymous_: as you suggested and make sure that the other version is currently not present Sep 04 22:17:56 zecke, ok, thanks. I think I'll try to fix it in bitbake then. Sep 04 22:18:04 awesome! Sep 04 22:18:21 * polyonymous_ dives into bitbake :) Sep 04 22:19:33 zecke: btw, have you any idea why export FOO ?= "value" doesn't work properly? Sep 04 22:19:45 zecke: and should we support "export FOO" ? Sep 04 22:21:12 stefan_schmidt: I've also replied in email. Sep 04 22:21:14 RP: IIRC export FOO is already used by metadata :) Sep 04 22:21:19 rwhitby: thanks Sep 04 22:21:34 zecke: export FOO = "" works, export FOO does not afaik Sep 04 22:21:49 RP: for ?= we would need to look into the feeder and I don't feel like doing that :} Sep 04 22:22:48 zecke: damn ;-) Sep 04 22:22:51 * zecke fights this pos Qtopia... (did I say that) *oops* Sep 04 22:36:57 * * OE Bug 2924 has been created by autobuild(AT)openembedded.org Sep 04 22:36:59 * * openmoko-image-1.0-r0-do_rootfs Sep 04 22:37:01 * * http://bugs.openembedded.org/show_bug.cgi?id=2924 Sep 04 22:40:00 zecke: export FOO ?= "" works as expected, we just have some strange default values :/ Sep 04 22:46:36 03rpurdie * r969 10/ (4 files in 4 dirs): Add support for the syntax "export VARIABLE" Sep 04 22:53:53 koen: hope my comments on bug #2919 don't come across as too harsh -- I didn't re-read before hitting "submit" Sep 04 22:53:56 /me wonders about the cursing zecke does about gtk Sep 04 22:54:30 ljp: hi :) Sep 04 22:54:37 koen: I'm just trying to keep everything absolutely as drop-dead simple as possible for gumstix users trying to figure out OE Sep 04 22:54:53 there are over 7,000 of them, and 99% of those are not very tech-savvy Sep 04 22:55:05 ljp: ask my co-workers for quotes Sep 04 22:55:14 they can just about figure out how to follow the buildroot setup/build instructions, which is 3 command line instructions Sep 04 22:55:33 If I were to try and get them using OE today, it'd be a support nightmare for me Sep 04 22:56:06 But I desperately want to get away from buildroot and onto OE asap Sep 04 22:56:36 and having a gumstix-specific distro/set of images seems like the easiest way to smooth many of these edges Sep 04 22:56:44 zecke: You mean something like: "It's wonderful. This sane API. Fast to learn." ? *caught* Sep 04 22:57:27 stefan_schmidt: ah right, I complained about the GtkTreeIter up/down/forward/backward navigation at the camp :) Sep 04 22:57:44 and randomly say Gtk+ devs get paid for LOC they produce Sep 04 22:57:56 heh Sep 04 22:58:13 Duplicating work brings money in... Sep 04 22:58:51 frameworks are a threat to the money flow ;) Sep 04 22:59:24 zecke: Not really. Just create your own. It's hip these days. :) Sep 04 22:59:37 using vala? Sep 04 23:00:08 isnt that like all guis build in c? :) Sep 04 23:00:15 zecke: For example. And if that is boring too, go for another language. Sep 04 23:00:43 hughescr, we understand your problems Sep 04 23:00:53 I think we are still grappling with how to deal with them Sep 04 23:01:07 stefan_schmidt: what I find funny about vala is. If Mr. de Icaza would have developed a C# frontend for gcc vala would not be needed... Sep 04 23:01:12 without creating millions of distros and special cases Sep 04 23:01:27 hughescr, also, did you sort out my cable issue? Sep 04 23:01:45 ljp: but my complain means I take a look at Qtopia... Sep 04 23:02:54 hughescr: I'm playing around with an "expert" gumstix machine type that might work better for guys like me who want fine grain control Sep 04 23:03:11 zecke: Perhaps. But sometimes it's easier to start another language. You know when people have this great vision about how the future should look like... Sep 04 23:03:44 stefan_schmidt: hehe, it is too early to argue Sep 04 23:03:53 hughescr: . . . and are willing to put up with a little extra work :-) Sep 04 23:04:22 zecke: BTW, you need to rant more Qtopia now then before or people will think you opinion is paid. ;) Sep 04 23:05:05 zecke: And of course rant about OpenMoko Sep 04 23:05:30 zecke: Talking about that. GSoC finished fine? Sep 04 23:05:58 stefan_schmidt: if I manage to upload code... Sep 04 23:06:32 zecke: oh, sometimes the real problems came at the end... Sep 04 23:07:40 lol so Palm just throw away 10 million $ of revenue... Sep 04 23:08:56 hello Sep 04 23:09:12 Crofton: cable issue yes, they're going overnight today -- should be there tomorrow Sep 04 23:09:18 thanks Sep 04 23:09:35 can you send a tracking number if possible? Sep 04 23:09:49 sakoman: yes I was planning on doing that basically with like one example and then some wiki docs on how to roll ones own Sep 04 23:10:17 I'm hoping this sort of thing is discussed some at OEDEM in early October Sep 04 23:10:43 I'll show you what I've put together after I verify that it actually works :-) Sep 04 23:11:12 Crofton: tracking number emailed Sep 04 23:11:20 hughescr, also, there have been suggestions that koen be sent to school to learn "social skills" :) Sep 04 23:11:26 lisppaste7, Sep 04 23:11:37 so don't worry about seeming harsh, just as long as the argument is good Sep 04 23:11:44 Crofton: social skills no problem with me; I prefer bluntness :) Sep 04 23:11:58 you two will get along fine then :) Sep 04 23:12:07 open source devs need thick skins Sep 04 23:12:18 yeah Sep 04 23:12:24 all devs need thick skin :-) Sep 04 23:12:39 Holding the sweet-and-happiness in reserve for when you *really* need something is sometimes useful too :) Sep 04 23:12:49 well, paid devs can chose to surround themselves with yes men :) Sep 04 23:12:58 industry fights just happen behind closed doors :-) Sep 04 23:13:02 yeah Sep 04 23:13:23 No less fireworks though! Sep 04 23:13:34 yakov pasted "migration error" at http://paste.lisp.org/display/47237 Sep 04 23:13:41 Just a smaller audience :-) Sep 04 23:14:04 yakov, what are you trying to do? Sep 04 23:14:31 I think maybe you are trying to migrate from a higher version mtn db using a lower version of mtn Sep 04 23:14:34 migrate db from here http://www.openembedded.org/snapshots/OE.mtn.bz2 Sep 04 23:14:34 I suspect Sep 04 23:14:42 in order to use with my current mtn Sep 04 23:14:46 yakov, try upgrading mtn Sep 04 23:14:56 there is 31 snapshot Sep 04 23:15:11 i'm using it on debian also Sep 04 23:15:18 err ubuntu Sep 04 23:15:22 Crofton, 0.31 is old?? Sep 04 23:15:22 yakov, upgrade mtn, or get .31 snapshot Sep 04 23:15:31 ynezz, yes im on ubuntu too... Sep 04 23:15:32 yakov, right :) Sep 04 23:15:42 http://www.openembedded.org/snapshots/OE-this-is-for-mtn-0.31.mtn.bz2 Sep 04 23:15:43 i see ): Sep 04 23:16:07 I have 0.36 on my fedora 7 box Sep 04 23:16:08 ynezz, thx Sep 04 23:17:01 Crofton, built from source? Sep 04 23:17:06 rpm Sep 04 23:17:10 i think :) Sep 04 23:17:23 trying to check Sep 04 23:18:02 heh they provide staticaly linked one just for lazy me :) Sep 04 23:18:29 yakov, I think they got tired of answering questions about boost versions .... Sep 04 23:18:48 yeah, looks like 0.36 is available as an rpm and yum Sep 04 23:19:53 heh, boost... Sep 04 23:20:04 in bitbake, I want to force a re-deploy stage. ie basically want to do "bitbake linux -c rebuild" but instead of rebuild, redeploy Sep 04 23:20:07 anyway migrates using 0.36 fine.. Sep 04 23:20:07 any way to do that? Sep 04 23:20:14 not sure Sep 04 23:21:03 I'll just live through -c rebuild I guess :) Sep 04 23:21:17 gooooooooo, ccache! Sep 04 23:21:40 heh, you touch machine.conf? Sep 04 23:22:10 No, just trying to get my uImage to use the linux compress/decompress instead of u-boot's Sep 04 23:22:16 ah Sep 04 23:22:21 u-boot is about 20x slower then the kernel's own decompressor Sep 04 23:22:23 how do you do that? Sep 04 23:22:31 roughly speaking Sep 04 23:23:08 -c deploy ? Sep 04 23:23:29 basically http://pastebin.com/m5a4202cf Sep 04 23:23:40 ynezz: doesn't do the deploy if it was already deployed Sep 04 23:23:47 -f Sep 04 23:23:54 to force? Sep 04 23:24:04 ynezz: ah, maybe Sep 04 23:24:05 dunno if it works for deploy Sep 04 23:24:27 what if you use the make uImage target when building the kernel? Sep 04 23:24:31 Crofton: basically use the gzipped vmlinuz and -C none for mkimage Sep 04 23:24:47 you might need to use -b /path/to/linux.bb also Sep 04 23:24:50 Crofton: then it embeds something in a uImage which has to be gunzipped before execution Sep 04 23:25:22 Crofton: if you use the compressed/vmlinuz instead, then it's an executable image with its own decompression routine in it Sep 04 23:25:23 I am wondering if make uImage generates the better decompression Sep 04 23:25:40 u-boot decompression routine is really really really slow Sep 04 23:25:47 linux's is really really fast Sep 04 23:26:28 At some point I would like to see linux.inc use "make uImage" rather than do the mkimage manually Sep 04 23:26:29 note: my pastebin there doesn't actually work (not tested yet) Sep 04 23:26:52 Crofton: iirc linux's make uImage does the same wrong thing that linux.inc is doing now Sep 04 23:26:56 there is an issue I have not completely worked through though Sep 04 23:26:59 ok Sep 04 23:27:04 ie compresses the executable instead of using linux's uncompressor Sep 04 23:27:07 I can check when I worry about this Sep 04 23:27:09 huh, nice. statically linked monotone was unable to resolve monotone.openembedded.org :/ Sep 04 23:27:28 though simply given IP worked fine.. Sep 04 23:29:56 ok, time to head home Sep 04 23:37:15 zecke, RP - I haven't tested _much_, but I think that should do: http://rafb.net/p/bi0zOc32.html Sep 04 23:46:57 * * OE Bug 2925 has been created by craig(AT)gumstix.com Sep 04 23:46:59 * * Use linux decompressor instead of u-boot' s for uImage kernels Sep 04 23:47:01 * * http://bugs.openembedded.org/show_bug.cgi?id=2925 Sep 05 00:06:13 night Sep 05 00:06:21 polyonymous_: sorry, I'm too tired... Sep 05 00:16:29 Crofton: created bug #2925 with the working uImage decompressor fix Sep 05 00:22:17 saw that Sep 05 00:22:30 cia is useful Sep 05 00:24:04 I will try and test/benchmark on the OSK this evening Sep 05 00:24:20 I need to get my sleep schedule back to normal ... Sep 05 00:26:42 What I do not like about linux.inc is that it touches a lot of recipes, so I need to be careful what I push Sep 05 00:43:02 I need some form of stopwatch for linux Sep 05 00:43:05 any suggestions Sep 05 00:43:16 I need to try to measure kernel uncompress times ... Sep 05 00:46:45 hughescr, you there? Sep 05 00:49:52 asmola_, know of any stopwatch programs for linux? Sep 05 00:50:21 no, sorry :( Sep 05 00:52:54 did the gumstix clean build complete or is it still crunching away? Sep 05 00:55:37 I have replicated the error :) Sep 05 00:56:24 :) Sep 05 00:57:10 can you post to bugs.openembedded..org? Sep 05 00:57:24 sure Sep 05 00:58:19 thanks, I'll add a comment Sep 05 00:58:29 trying to figure out how to time kernel decompress Sep 05 01:00:00 www.casio.com Sep 05 01:00:08 time cat Sep 05 01:00:13 ctrl-c when done Sep 05 01:02:27 ok, now to try hughescr patch Sep 05 01:02:56 yeah i was gonna try all that stuff once i got the build competed :) Sep 05 01:04:35 ill prolly just grab one of my old directories of working stuff and then try and patch that stuff in Sep 05 01:09:51 asmola: here now Sep 05 01:10:17 hello, i saw your comments above in regards to making it as simple as possible Sep 05 01:10:42 Crofton: stopwatch you can do it internally or externally Sep 05 01:10:51 internally, turn on "timing of printks" in the kernel config Sep 05 01:11:07 externally, wrap something around your console viewing thing which adds a timestamp to every line Sep 05 01:11:25 i definitely agree with you, in watching the gumstix-users mailing list that it has to be really easy, in light of that and in trying to not hack open-embedded? i can't help but think a GUI interface of some kind would be a good approach Sep 05 01:11:35 interestingly, this is a topic for the upcoming OE conference Sep 05 01:11:41 ie if you're connecting to your linux console via a serial port, just use socat or somethign to munge the output and add a timestamp prefix Sep 05 01:11:55 the OE eclipse plugin Sep 05 01:12:15 I'm using time cat and ctrl-c Sep 05 01:12:18 :) Sep 05 01:12:37 yeah, that'll work :) Sep 05 01:12:58 basically, I want to provide some documentation this does no harm :) Sep 05 01:13:10 asmola: UI actually would almost certainly introduce all kinds of extra build host dependencies and make the problem worse ;) Sep 05 01:13:28 some project use a makefile to hide the OE work from teh dev Sep 05 01:13:30 asmola: nothing wrong with command line, as long as it's like 1 command they need to type which is hard to fuck up Sep 05 01:13:39 needless to say, this is "controversial" Sep 05 01:13:41 eg buildroot process is literally 3 short commands Sep 05 01:13:51 Crofton: yeah, I can imagine Sep 05 01:14:10 yeah, and they are prompted to enter their config or they can go through the buildroot menu config to tweak, def easier than OE from an end user peerspectrive Sep 05 01:14:16 we hate debugging other peoples problems when their language is based on a makefile Sep 05 01:14:46 but you move into complex apps and build root starts to fall over Sep 05 01:14:55 also, there seem to be a lot of forks Sep 05 01:15:03 * Crofton|home has never used buildroot Sep 05 01:15:53 maybe gui isn't neccesarily the best word, but just a friendly user interface for them to select which modules they want, and maybe this could be something that hooks into OE such that when the hardware specifc files are arranged as per a particular directory structure users are able to select which ones and then the options are catted into the gumstix-verdex.conf for example Sep 05 01:16:07 bother, OSK kernel is not based on linux.inc Sep 05 01:16:50 asmola: I don't want them to have to choose which modules they want Sep 05 01:16:55 asmola: they don't know which modules they want Sep 05 01:17:14 asmola: def. would be nice to let advanced people start choosing Sep 05 01:17:41 asmola: but they can do that today by just creating a new image which requires the existing image and adds some PACKAGES Sep 05 01:17:53 or creating a new distro which requires the existing distro Sep 05 01:18:10 it's not really "creating a new distro" so much as making minor revisions to an existing one Sep 05 01:18:24 i see what youre saying, one command, you're done, it works regardless of what you have plugged into your gumstix Sep 05 01:20:07 yup Sep 05 01:20:31 basically, I think we need a better way to tune things Sep 05 01:20:49 except instead of "works regardless of what you have plugged in" I'd define it as "build your the same binary image that was factory-installed" Sep 05 01:20:54 crap Sep 05 01:21:08 can't do the timing test as easy as I thought Sep 05 01:21:19 need to move the OMAP kernels to linux.inc Sep 05 01:21:31 asmola_, did you crete a bug? Sep 05 01:21:57 typing it up right now Sep 05 01:24:40 I am going to check the gumstix kernel sizes, hopefully tomorrow evening I can do teh uncompress times ... Sep 05 01:25:35 Bug 2926 Sep 05 01:25:57 * * OE Bug 2926 has been created by adam.smola(AT)gmail.com Sep 05 01:25:59 * * Clean build of Angstrom for Gumstix-Verdex Fails Sep 05 01:26:01 * * http://bugs.openembedded.org/show_bug.cgi?id=2926 Sep 05 01:39:17 asmola: heading to dinner, but I suspect that bug is related to in-flux gumstix machine.conf which was not in sync with distro/image changes or something Sep 05 01:39:41 I'm hopeful hrw looks at it in the morning Sep 05 01:40:03 Crofton: re decompress times; to see it's faster you don't really need a timer. It's visibly faster :) Sep 05 01:40:05 ok, thanks for your help Sep 05 01:40:09 I need to arrange for my script to rotate which machine builds first to catch this sort of issue Sep 05 01:40:21 u-boot decompress sits there for ~15 seconds, the linux decompress takes ~1 second or so Sep 05 01:40:23 hughescr, I am curious :) Sep 05 01:40:36 :) Sep 05 01:40:50 Off to dinner, ttyls Sep 05 01:41:01 later Sep 05 01:41:03 hughescr, on OSK, I am seeing 3.6 seconds Sep 05 01:41:19 from end of u-boot timer to uncompress complete Sep 05 01:41:37 but that uses result from make uIamge ... Sep 05 01:44:14 crofton|home, is there any easy way to look at older versions of files in the monotone repository? Sep 05 01:44:31 the subversion setup you guys have the ossie stuff under is really slick in that respect Sep 05 01:46:56 oh, if you build first for omap it works Sep 05 01:47:01 oh i am just going to do that **** ENDING LOGGING AT Wed Sep 05 02:59:56 2007