**** BEGIN LOGGING AT Wed Nov 30 02:59:57 2011
Nov 30 09:19:20 <s0undt3ch> khem: around?
Nov 30 09:19:31 <s0undt3ch> khem: remeber oe_filter_out
Nov 30 09:19:39 <s0undt3ch> does that accept regex?
Nov 30 09:19:45 <s0undt3ch> seems it doesn't
Nov 30 10:46:31 <dany> Hi All, whats the "correct way" of having yocto generate a kernel source package for development with meta-toolchain ?
Nov 30 12:02:02 <lianhao> RP__: ping
Nov 30 13:41:27 <RP__> lianhao: pong
Nov 30 14:11:03 <bluelightning> anyone tested a dpkg-based rootfs recently?
Nov 30 15:45:52 <RP__> bluelightning: no, I worry it might not be in a good state :/
Nov 30 15:47:12 <bluelightning> RP__: it fails for me at do_rootfs because dpkg-scanpackages can't find Dpkg.pm
Nov 30 15:47:44 <bluelightning> the file is there, it's just it's not where the perl paths are set to look
Nov 30 15:48:18 <RP__> bluelightning: something sounds wrong with the various path mangling :(
Nov 30 16:04:23 <s0undt3ch> hello ppl
Nov 30 16:04:48 <s0undt3ch> what's the good place to start reading if I want to create a stripped down distro with yocto?
Nov 30 16:06:03 <s0undt3ch> right now the machine in use is qemuarm
Nov 30 16:22:49 <khem> s0undt3ch: oe_filter_out does take regexp look in meta/recipes-core/uclibc/uclibc.inc
Nov 30 16:23:21 <s0undt3ch> khem: I think I as using it in the wrong place
Nov 30 16:23:24 <s0undt3ch> I was trying to do
Nov 30 16:23:41 <s0undt3ch> FILTERED_QT_TASK="${@oe_filter_out('qt4-embedded-([demos,examples,init,assistant]+)', '${task-qt4e-base}', d)}"
Nov 30 16:23:51 <s0undt3ch> if kept failing
Nov 30 16:24:12 <khem> s0undt3ch: ah right now it only works on space separated strings
Nov 30 16:24:16 <khem> e.g. CFLAGS
Nov 30 16:24:32 <s0undt3ch> so I just copied over what I needed from the task-qt4e-base
Nov 30 16:24:36 <khem> I have posted a page which it can take a separator
Nov 30 17:07:10 <msm> do we have any interest in a meta-benchmark layer?
Nov 30 17:12:24 <bluelightning> msm: I'm sure I remember someone asking on the oe-core mailing list about adding benchmarking tools a few months ago
Nov 30 17:12:52 <kergoth> might have been noor from mentor. we need either a recipes-benchmarks or a meta-benchmarks, I don't much care which as long as they have someplace to live
Nov 30 17:12:53 <kergoth> heh
Nov 30 17:13:17 <msm> me either
Nov 30 17:13:29 <msm> i just want something coordinated, and somewhere the recipes get accepted easily
Nov 30 17:13:38 * kergoth nods
Nov 30 17:13:40 <msm> did noor start work on this at all?
Nov 30 17:13:51 <kergoth> not sure, will ask
Nov 30 17:36:49 <pirut> hey I am getting a checksum mismatch for libX11-1.4.4.tar.bz2"
Nov 30 17:37:23 <pirut> where is the expected checksum stored
Nov 30 17:39:17 <JaMa> patch is on oe-core ML and now also in poky ML
Nov 30 17:39:34 <JaMa> I guess it's from libx11-trim..
Nov 30 18:09:20 <pirut> http://pastebin.com/0L9rNVRK
Nov 30 18:09:58 <pirut> meta/recipes-graphics/xorg-lib/libx11_1.4.4.bb   <- has the correct md5 checksum for what is downloaded
Nov 30 18:10:06 <pirut> but is compared against
Nov 30 18:10:19 <pirut> meta/recipes-graphics/xorg-lib/libx11-trim_1.4.4.bb
Nov 30 18:49:24 <s0undt3ch> how come I can't deselct packages using hob?
Nov 30 18:52:21 <incandescant> I had to disable it as I couldn't get it to work reliably in time to release
Nov 30 20:50:43 <s0undt3ch> incandescant: dam, no use for it like that :)
Nov 30 20:51:23 <s0undt3ch> incandescant: you can't also make the X axis of the window always hidden, I'd like to know what brought a package in, and that's usually a long string ;)
Nov 30 20:51:53 <incandescant> s0undt3ch: the former is definitely on the list of things to fix, would you mind filing a bug for the latter?
Nov 30 20:52:19 <s0undt3ch> incandescant: on yocto's bugzilla?
Nov 30 20:52:38 <incandescant> s0undt3ch: yes please, there's a category for it and everything :-)
Nov 30 20:53:05 <s0undt3ch> incandescant: any reason why the package selection is very slow? ie, if I select an image with more that 200 packages, it takes ages and the UI is just frozen
Nov 30 20:53:53 <incandescant> s0undt3ch: it's trying to calculate all the interdependencies of the 200 packages - I believe there's a plan in place to lessen some of that kind of pain
Nov 30 20:54:06 <s0undt3ch> ok
Nov 30 20:54:56 <s0undt3ch> would be cool to have the progress bar pulsing
Nov 30 20:55:02 <s0undt3ch> on that stage
Nov 30 20:57:36 <incandescant> yeah, I think I put code in for that but atm the package dependency calculation blocks the UI thread
Nov 30 20:58:19 <s0undt3ch> separate thread?
Nov 30 20:58:22 <s0undt3ch> just an idea
Nov 30 21:47:58 * incandescant nods
Nov 30 21:48:25 <incandescant> it's just a matter of finding the time to fix it :-)
Nov 30 21:52:55 <incandescant> i didn't have the time before we froze edison
Nov 30 21:58:26 <ka6sox> ~seen pidge?
Nov 30 22:01:04 * incandescant summons pidge
Nov 30 22:01:26 * pidge climbs out of the murky depths of license wrangling
Nov 30 22:01:36 <pidge> ka6sox: you rang?
Nov 30 22:02:01 <ka6sox> hi there....
Nov 30 22:02:17 <ka6sox> working on implementing the oe-core build machine :D
Nov 30 22:02:36 <ka6sox> have some infra questions...can I email them to you?
Nov 30 22:03:08 <ka6sox> trying to get this done and working.
Nov 30 22:03:34 <pidge> ka6sox: of course.
Nov 30 22:03:37 <ka6sox> I am also trying to remember if 96GB is enough RAM.
Nov 30 22:03:46 <ka6sox> or you needed more like 110GB
Nov 30 22:04:39 <ka6sox> is that something you can answer off the cuff?
Nov 30 22:05:36 <pidge> ka6sox: it's a more complex answer than yes or no. I would say it's enough depending on the images being built. If it isn't enough, you can:
Nov 30 22:05:41 <pidge> - limit your images
Nov 30 22:06:11 <pidge> - break up build sets and reuse your tmpfs
Nov 30 22:06:12 <ka6sox> qemu-*
Nov 30 22:06:29 <ka6sox> so common I suppose then arch specific.
Nov 30 22:06:29 <pidge> - move DL_DIR and maybe SSTATE_DIR to disk
Nov 30 22:07:06 <ka6sox> ya, I was hoping to avoid having to use local disk and use an NFS mount instead :9
Nov 30 22:07:07 <ka6sox> :(
Nov 30 22:07:27 <pidge> ka6sox: we currently use NFS for DL_DIR and SSTATE
Nov 30 22:07:33 <pidge> which makes things super nice
Nov 30 22:07:45 <ka6sox> pidge, which brings me to the next Q
Nov 30 22:07:47 <pidge> since my entire build cluster can share it
Nov 30 22:08:10 <ka6sox> and I don't know if this has been brought up in the Yocto Tech Group yet...
Nov 30 22:08:42 <ka6sox> sharing common bits to reduce overhead.
Nov 30 22:09:18 <ka6sox> I really don't want to be replacing HDs every 6months or less.
Nov 30 22:10:53 <pidge> we do a bit of that on yocto-autobuilder. if you look at it, we do, for example a world fetch (should be universe, reminds self to fix that) on the main nightly buildset. That buildset triggers other arches which utilize the fetch's DL_DIR. makes it so you don't have to have multiple DL_DIRS and don't have every target fetching. Since the sstate cache is also in NFS, you can share bits as they're built
Nov 30 23:10:02 <s0undt3ch> does yocto also provides a gtk direct frame buffer package ?
Nov 30 23:13:24 <incandescant> I seem to recall a patch for one recently
Nov 30 23:14:58 <s0undt3ch> incandescant: the sato gtk them is meant for x11? right?
Nov 30 23:15:19 <incandescant> yeah
Nov 30 23:15:34 <s0undt3ch> incandescant: my target device is http://www.lilliputuk.com/computers/panel-pcs/gk7000/
Nov 30 23:16:05 <s0undt3ch> that has an ARM9 @ 400MHz, and 128Mb of SD Ram
Nov 30 23:16:20 <s0undt3ch> since you might be more aquainted with emmbeded stuff
Nov 30 23:16:38 <s0undt3ch> incandescant: would running x11 on that kind of a machine be too heavy?
Nov 30 23:16:39 <incandescant> sato was developed for the Sharp Zaurus, so I'd expect you can run on that
Nov 30 23:17:53 <s0undt3ch> incandescant: zaurus now runs on Qt?
Nov 30 23:18:34 <s0undt3ch> hmm, zaurus has half the resources the gk7000 has....
Nov 30 23:19:56 <s0undt3ch> hmm SL-5500... guess there are a bunch of zaurus versions.... :)
Nov 30 23:19:59 <incandescant> specifically the zaurus akita and spitz, many years ago - that gk7000 is herculean in comparison
Nov 30 23:20:55 <s0undt3ch> incandescant: uff, guess I might not need that framebuffer support after all
Nov 30 23:21:31 <s0undt3ch> although I've almost convinced myself of using QtEmbedded instead of GTK
Nov 30 23:23:38 <incandescant> that should also be well supported in OE-core/Poky - the opie maintainer is a colleague of mine
Nov 30 23:23:50 <incandescant> he has a layer for opie somewhere
Nov 30 23:32:55 <ant____> meta-opie is in the main git server fwiw
Nov 30 23:32:59 <ant____> http://cgit.openembedded.org/
**** ENDING LOGGING AT Thu Dec 01 02:59:58 2011