Teleconference minutes

[Mitch Bradley]

OLPC System Software Teleconference, 2007-09-04

Participants: Jim, Mitch, Richard, Kim, Quanta, Bernie, J5, cjb, Michael

== Status ==

Resume: cjb: wireless bug has been pushed to the back because of the
power bug (1835).  The current status of the power bug is that we
see 8-bytes of instruction fetch on the LPC bus after the initial
jump instruction in the reset vector.  We think that the problem
may be related to an "out" instruction that occurs early in the
process.

Activation/security:  We need to add the signed kernels and initrds
to the build, beside the old-style ones, as a way of testing the
firmware security (#2964).  The firmware will continue to behave in the
old way unless the "wp" tag is set in manufacturing data.  OFW
has commands to set and clear that tag, so an XO can be switched
back and forth from secure booting mode.

Interoperability problem vis a vis build 567 needs to get into the 
release notes.

wad gave Kim a working school server.

#2122 show-source keybindings needed - will get into a build probably 
tomorrow.

#2833 camera fails in os 547 - we don't know why the new gstreamer is 
failing.
For now we are sticking with the old gstreamer.  Leave as blocker but 
move to
later milestone.  ? Do we need to get Jon working on implementing new 
interface ?

#3018 - booting with Ethiopian locale - we might have to pull in the F8 
package.
The problem could be in the i18n of the initscripts.  Bernie suggests 
setting
the locale to C during boot, and then switching the locale for X init.
AI: Bernie to fix it.  (cjb: perl and other tools get really slow with
international locales because chars stop being bytes.)

#3075 - could be a number format issue.  Perhaps csound or TamTam don't 
handle
non-US number formats.

#3131 - done

#447 - reassign from behdad to bernie

#2804 - The problem is at a lower level than X11 - kernel or EC.  
Possible temperature related.
Saw some packets that were not 6-bytes long.  The problem comes and 
goes.  Is the packet format
mode changing, or are bytes being lost?  Problem also seen in console 
mode using gpm.

Sometimes the game keys generate different key codes?  Is it a PS2 
issue, or old EC firmware?
We want at least C25.

#1547 - microbenchmarks have been run, but we need sugar benchmarks

#2198 - workaround works for 1.3, should "just work" when we get Xorg 1.4.
        Time is running out on 1.4

We need a way to document the procedure for making test images that
differ from official builds and documenting/capturing the process.

#2708 - fixed

#2188 - needs packaging - we should put the common stuff in the base system.
Potential licensing issues with samples.

#3074 - wrong MIME type in browser.  We have registered a MIME type.
J5 will fix it, and ping Marco