Updates for Update.1 from Joyride

[Jim Gettys]

Bert, all,

Here's what is going on:

We have to have something for the first systems and G1G1 that have a few
key bugs fixed; the wireless driver, and a key fix for the browser
(turns out that our browser, and any current firefox development
version, including the Firefox 3.0 beta released today chokes on TMobile
hot spots).  And we need to fix the embedded controller problem that was
causing suspend/resume instability, that Richard Smith found last week
(whew, we had thought we still had hardware trouble, but thankfully, the
production machines should be fine once they've been updated).

This bug is at:
https://bugzilla.mozilla.org/show_bug.cgi?id=404059
If anyone has expertise in the bowels of the firefox code base who can
help chase this bug, it would be appreciated.

Ship.1 has shipped.  It is build 623.  The first production machines
(beyond the first machine off the production line Mary Lou hand carried
back) arrived in Cambridge today.

So rather than trying to mix a "features" release with  a "bug fix"
release, we're planning an Ship.2 build that will most likely based on
the 623 stable point, and let Update.1 take however long it will take to
stabilize.  We did this sort of thing with the 406.x series as you
remember. We are hoping for a fix for the browser.  The wireless driver
problems have at least been temporarily fixed, though work will continue
on it.

The new Update.1 build (637 built last night) is the first post 623
build that verges on usable, though it will likely exhibit
suspend/resume issues on B4's (you can turn OHM off for greater
stability).  It has activity isolation running by default, the new X
server, activity updates, etc. There are issues with the DCON being
powered off that can cause trouble with the more aggressive power
management work Chris Ball is getting running with OHM.  Our
hope/expectation is that tonight's update.1 build will be pretty fully
usable. We will likely inhibit turning off the DCON chip in suspend on
B4 hardware as a result, which will hurt power consumption in suspend on
the prototypes.  This hardware problem is fixed in mass production
hardware.

So in short, we're screwing down the lid on Update.1.  But we likely
have to do a Ship.2 build, and that on top of a feature release is a bad
idea, so we'll let Update.1 slip and be sane about letting it be ready
when it is ready, rather than having to throw it over the wall on
December 1, ready or not.

Don't anyone get too many ideas about adding features or invasive fixes
at this date for Update.1.  We have lots of new stuff in it already, and
people need to be patient.
          Best regards, and thanks to everyone for their great help!
                                   - Jim



On Tue, 2007-11-20 at 21:40 +0100, Bert Freudenberg wrote:
> On Nov 20, 2007, at 20:43 , Jim Gettys wrote:
> 
> > Joyride and Update.1 were synchronized as of Monday evening,  
> > November 21.
> 
> I guess you made a sign error when subtracting 1 from today and meant  
> Nov. 19th.
> 

Brain fault, sorry.

> But, more importantly - could you let us know what the plan is with  
> Update.1? I see that Ship.1 is scheduled to be shipped next week, and  
> Ship.2 one day later. Update.1 appears to have been pushed out by 2  
> weeks. Is it still the case that Update.1 will be put on each mass- 
> produced laptop before it reaches the end users? If so, then what is  
> Ship.2 for?
> 
> Thanks,
> 
> - Bert -
> 
> 
-- 
Jim Gettys
One Laptop Per Child