[orca-list] Performance enhancements for Firefox 3

[Joanmarie Diggs <joanmarie diggs gmail com>]

Hey folks.

We're in the process of making access to Firefox 3 peppier.  And we're
doing it in phases rather than as one massive change.  I have just
committed a couple of patches to trunk which should make life better.  A
brief bit of background on the changes:  

When we first implemented support for Firefox 3, there were a number of
Firefox bugs that made it very difficult to reliably present the
contents of any given line without laboriously examining it character by
character.  So we laboriously examine each line character by character
to be accurate, but the price that gets paid is speed.  We often were
doing this examination twice too, once for speech and once for braille.
The end result is that on a page that consists of long lines of text,
Orca would slow down considerably.  So....

Change one is to only do the laborious examination once rather than
twice. <grin>  It should make navigation a bit faster and should not
come with other side effects.

Change two is a bit more significant than that.  Thanks to a number of
fixes made by the Firefox a11y team (thanks guys!), we are now able to
reliably obtain line contents in a far less laborious way.  The new
method does things completely and utterly differently.  As such, you'll
see some changes which are good like an improvement in speed, and items
that are all on one line (e.g. in a table) being presented as such.
However, it's hard to anticipate every last condition we'll come across.
We've tested it, and Michael Whapples saw our work on bug 500016 and
tested it too -- and he found a couple such conditions which we've since
fixed.  Thanks Michael!!  Now we need y'all to test it so that we can
have speedy *and* accurate access in the GNOME 2.22 release.

If, while testing, you come across an issue and are not sure if it was
an issue prior to the change, we've added a setting for you.  If you add
the following lines to your orca-customizations.py file:

import orca.Gecko
orca.Gecko.performanceEnhancements = False

Orca will use the old way of obtaining line contents.  (But it still
will only do it once rather than twice.) As you might have guessed,  

orca.Gecko.performanceEnhancements = True 

will cause the new way to kick in.  (Note that capitalization of True
and False count.)  The new way is the default way, so if you want to
throw caution to the wind, you don't have to put anything in
orca-customizations.py. <smile>

We look forward to your feedback on these changes and will continue to
work on additional improvements over the next couple of weeks (including
form-related problems).

Thanks everyone!!
--Joanie