**** BEGIN LOGGING AT Thu Sep 14 02:59:56 2006
Sep 14 05:55:13 <natescape> Anyone know what would cause an OpenSlug 2.5 system to have really slow file read/write performance?  It seems like it's running at USB 1.X speeds.
Sep 14 15:17:00 <kitno455> NAiL, ping
Sep 14 15:23:39 <kitno455> have found a couple bugs in slugos HEAD, and dont have perms to fix.
Sep 14 15:23:49 <kitno455> i will detail for anyone logging...
Sep 14 15:24:03 <koen> attach patches to bugs.openembedded.org
Sep 14 15:24:20 <kitno455> slugos specific too?
Sep 14 15:24:26 <koen> yes
Sep 14 16:04:33 <kitno455> ok, have posted two bugs.
Sep 14 16:05:59 <kitno455> eno- hello!
Sep 14 16:06:25 <eno> morning kitno455 
Sep 14 16:06:49 <eno> i'm packaging erlang for openslug
Sep 14 16:07:13 <eno> is there an equivalent of optware-check-package?
Sep 14 16:07:25 <kitno455> not already in openembedded?
Sep 14 16:07:59 <eno> i don't think there's erlang.bb
Sep 14 16:08:48 <kitno455> dont see one...
Sep 14 16:09:04 <kitno455> what does optware-check-package doo?
Sep 14 16:09:08 <kitno455> s/doo/do/
Sep 14 16:09:08 <apt> kitno455 meant: what does optware-check-package do?
Sep 14 16:09:38 <eno> it does some sanity check, make sure binaries are stripped, rpaths are correct
Sep 14 16:10:19 <kitno455> i do not know. i think all bins are stripped by default....
Sep 14 16:10:25 <eno> k
Sep 14 16:10:55 <kitno455> ask on #oe? koen just left here
Sep 14 16:11:04 <kitno455> haha
Sep 14 16:11:09 <kitno455> koen just returned
Sep 14 16:11:59 <eno> i got the erlang small_build working
Sep 14 16:12:38 <eno> if a package is not already in oe, how to add?
Sep 14 16:12:38 <kitno455> i have never played with erlang, so i dont know what that means :)
Sep 14 16:13:27 <eno> quite interesting language
Sep 14 16:13:48 <eno> deal with concurrency nicely, as nice as you can get
Sep 14 16:13:55 <kitno455> what do you use it for personally?
Sep 14 16:14:43 <eno> not for anything in particular, but it's just cool to run it on slug
Sep 14 16:15:06 <kitno455> that seems to be the rationale for most packages on the slug :)
Sep 14 16:16:37 <kitno455> erlang sounds like 3/4 of an OS rather than a prog lang.
Sep 14 16:16:45 <kitno455> much like emacs :)
Sep 14 16:17:22 <eno> that's true
Sep 14 16:17:59 <eno> although to use OS thread is actually heavier than spawning erlang processes
Sep 14 16:18:22 <eno> so its approach has some reason behind it
Sep 14 16:18:25 <kitno455> i would like to learn more about the in-place code upgrades.
Sep 14 16:18:31 <kitno455> that sounds sweet
Sep 14 16:18:48 <eno> you should watch the erlang movie
Sep 14 16:18:57 <kitno455> they have a movie?
Sep 14 16:19:11 <eno> funny in a geeky way
Sep 14 16:19:58 <eno> http://video.google.com/videoplay?docid=-5830318882717959520
Sep 14 16:21:29 <kitno455> wow
Sep 14 16:22:50 <kitno455> i love the hair
Sep 14 16:22:58 <eno> lol
Sep 14 16:25:27 <joshin> Has anyone been able to build gcc 4.1.x recently?
Sep 14 16:25:54 <joshin> Its been dying for me on libgcc2.c for debianslug and openslug builds.
Sep 14 16:25:55 <kitno455> joshin, i built HEAD yesterday, so i think so
Sep 14 16:26:12 <joshin> kitno455, what's your build machine?
Sep 14 16:26:34 <joshin> Mine is a amd64 gentoo linux machine.
Sep 14 16:29:33 <joshin> http://pastebin.ca/169409
Sep 14 16:30:37 <kitno455> FC4 amd32
Sep 14 16:32:22 <joshin> Thanks.  Going to try it on my older P3 1.1Ghz Gentoo machine and see if its a 64bit issue.  Sigh...
Sep 14 16:32:22 <kitno455> eno- i just wasted 11.5 minutes of my life to watch these guys fix a typo :)
Sep 14 16:35:21 <eno> where can i find oe packaging style guide? i have a 11M erlang.ipk, i don't think that's the way to go
Sep 14 16:37:27 <kitno455> most packages seem to split docs and devel headers out
Sep 14 16:38:01 <kitno455> some also split libs from executables, though in your case that might make no sense
Sep 14 16:38:02 <eno> /usr/lib/erlang is probably 10M
Sep 14 16:38:17 <kitno455> run file on it, and make sure its stripped
Sep 14 16:38:47 <kitno455> its an os inside a binary, you may have to leave it that big
Sep 14 16:38:50 <kitno455> :)
Sep 14 16:39:49 <eno> yes they are all stripped properly
Sep 14 16:40:17 <kitno455> how big is it when built on x86?
Sep 14 16:40:57 <eno> about the same
Sep 14 16:40:57 <koen> eno: and there are no .debug directories left in the main package?
Sep 14 16:41:52 <eno> $ find tmp/work/erlang-R11B.1-r0/install/erlang-dbg/ | less
Sep 14 16:41:53 <eno> $ find tmp/work/erlang-R11B.1-r0/install/erlang-dbg/ tmp/work/erlang-R11B.1-r0/install/erlang-dbg/
Sep 14 16:42:16 <eno> so nothing under erlang-dbg
Sep 14 16:42:51 <eno> most of ipk are libs
Sep 14 16:43:22 <eno> for optware, i have erlang.ipk and erlang-libs.ipk
Sep 14 16:43:30 <eno> with erlang.ipk including just the core libs
Sep 14 16:44:09 <kitno455> how is their size?
Sep 14 16:44:45 <eno> erlang about 6M, actually erlang-libs 29M
Sep 14 16:45:55 <kitno455> can you prune the erlang.ipk down to the same files as optware package?
Sep 14 16:46:23 <eno> yeah, i'll probably do the same
Sep 14 16:51:41 <eno> koen: $ find tmp/work/erlang-R11B.1-r0/install/erlang/ -type d -name .debug
Sep 14 16:51:41 <eno> tmp/work/erlang-R11B.1-r0/install/erlang/usr/lib/erlang/erts-5.5.1/bin/.debug
Sep 14 16:51:41 <eno> tmp/work/erlang-R11B.1-r0/install/erlang/usr/lib/erlang/bin/.debug
Sep 14 16:51:41 <eno> tmp/work/erlang-R11B.1-r0/install/erlang/usr/lib/erlang/lib/tools-2.5.1/bin/.debug
Sep 14 16:53:30 <koen> looks like you need something like FILES_${PN}-dbg += "${libdir}/erland/*/.debug"
Sep 14 16:54:00 <koen> and make sure the PACKAGES statement is correct
Sep 14 17:05:20 <eno> k
Sep 14 19:19:49 <kitno455> ok, does udev not process all 'RUN' rules during bootup?
Sep 14 19:20:27 <kitno455> i can get it to mount my flashdrive at boot time, but it wont run any of my other custom rules
Sep 14 19:21:08 <kitno455> but if i hotplug the drive once the machine has booted, or if i run udevsynth, the custom rules get run
Sep 14 19:21:37 <kitno455> i guess that / is not mounted yet?
Sep 14 20:21:40 <kitno455> ahh- udev IS running my custom scripts, but they are blowing up because the root fs is mounted ro
Sep 14 20:21:54 <NAiL> aha
Sep 14 20:22:28 <kitno455> but there apear to be two / mounts?
Sep 14 20:22:41 <kitno455> rootfs on / type rootfs (rw)
Sep 14 20:22:42 <kitno455> and
Sep 14 20:22:50 <kitno455> /dev/root on / type jffs2 (ro)
Sep 14 20:23:07 <NAiL> that is after booting to external disk right?
Sep 14 20:23:28 <NAiL> mtab doesn't get updated when you do the pivot_root bit
Sep 14 20:23:30 <kitno455> no, that is during the boot process, before syslog even starts
Sep 14 20:23:36 <NAiL> oh... that's odd
Sep 14 20:23:48 <NAiL> I guess it's time for me to try to dig up a serial cable again
Sep 14 20:23:59 <kitno455> hold
Sep 14 20:24:12 <kitno455> yes- this is tru after boot as well
Sep 14 20:24:19 <kitno455> s/tru/true/
Sep 14 20:24:21 <apt> kitno455 meant: yes- this is true after boot as well
Sep 14 20:24:39 <kitno455> but i dont think its an issue.
Sep 14 20:25:18 <NAiL> it's about time the slugos boot process got reworked
Sep 14 20:25:18 <kitno455> after boot, /dev/root is rw
Sep 14 20:25:55 <kitno455> if udevsynth gets run a second time, after / is mounted rw, it works ok
Sep 14 20:26:26 <kitno455> any idea where that re-mount happens?
Sep 14 20:26:28 <joshin> Nail, kitno455, did either of you see the troubles that eFfeM and I were having with the current bitbake branch?    
Sep 14 20:26:50 <kitno455> the AMD64 issues?
Sep 14 20:27:12 <joshin> No, the "ImportError: cannot import name msg" when trying to run bitbake
Sep 14 20:27:23 <NAiL> that's a bitbake issue I think?
Sep 14 20:27:29 <kitno455> no, have not seen it, but i have a day-old checkout
Sep 14 20:27:35 <joshin> don't update.  
Sep 14 20:27:57 <joshin> Though someone in #oe was able to run just fine after updating their repository.
Sep 14 20:28:02 <NAiL> bbiaw
Sep 14 20:28:08 <kitno455> adios
Sep 14 20:28:13 <joshin> take care NAiL 
Sep 15 00:34:51 <joshin> Aha, I understand...
Sep 15 00:36:00 <joshin> The current revision of bitbake on the nslu2 monotone server is broken.  It was made at a time when bitbake was being updated with a new msg module and somehow that is missing from the current revision.
Sep 15 00:36:48 <joshin> So, if you revert back to d209100e92a2764e16ae183a398a48a64bcc96d5, you'll be fine.  (Or you could just grab bitbake directly from OE's repository)
Sep 15 02:11:56 <jacques> getting this from bitbake:
Sep 15 02:11:59 <jacques> Traceback (most recent call last):
Sep 15 02:11:59 <jacques>   File "/home/packages/nslu2/mtn/debianslug/bitbake/bin/bitbake", line 26, in ?
Sep 15 02:11:59 <jacques>     import bb
Sep 15 02:11:59 <jacques>   File "/home/packages/nslu2/mtn/debianslug/bitbake/lib/bb/__init__.py", line 73, in ?
Sep 15 02:11:59 <jacques>     from bb import msg
Sep 15 02:12:01 <jacques> ImportError: cannot import name msg
Sep 15 02:20:53 <joshin> The current revision of bitbake on the nslu2 monotone server is broken.  It was made at a time when bitbake was being updated with a new msg module and somehow that is missing from the current revision.
Sep 15 02:20:58 <joshin> So, if you revert back to d209100e92a2764e16ae183a398a48a64bcc96d5, you'll be fine.  (Or you could just grab bitbake directly from OE's repository)
Sep 15 02:21:42 <joshin> Or you could pester the right person to resync the NSLU repository with the OE one
Sep 15 02:21:50 <joshin> I just don't know who the right person is.
Sep 15 02:22:20 <jacques> joshin, thanks
Sep 15 02:22:44 <joshin> No problem.  Cost me a few hours to figure it out so glad it is helping someone else too.  :)
Sep 15 02:23:02 <jacques> yeah, that looked like a painful one to figure out
Sep 15 02:23:18 <jacques> I had already decided to give up for a while :-)
Sep 15 02:23:52 <jacques> machine is now building buildroot, so oe missed its chance ;-)
Sep 15 02:24:27 <joshin> :)
**** ENDING LOGGING AT Fri Sep 15 02:59:57 2006