Proposed response to gnopernicus/DecTalk problem

[Bill Haneman <Bill Haneman Sun COM>]

Here's my take on what we've learned so far, and my current suggestions.

I think that in the absence of a reliable means of autosensing both 
braille and non-braille devices on the serial ports, we should take 
Kenny's advice, and have gnopernicus make no assumptions about serial 
connections.  (The current defaults don't seem to be useful, and as 
we've seen, can be harmful).

This would mean that Gnopernicus' default braille settings would specify 
neither serial port nor braille driver (unless, possibly, we used the 
BrlTTY driver as our default). When gnopernicus detects that braille 
support has been enabled, by whatever means, and no serial port has been 
explicitly specified, it should post a dialog notifying the user, and 
allowing the user to either select serial port for the braille device or 
turn braille
support off again.

A similar change could be backported to the currently-UI-frozen branches 
of gnopernicus, without the notification dialog, by posting the Braille 
configuration UI whenever braille is turned on without explicitly 
configured port settings.

In the future the braille configuration dialog could probe the serial 
ports (if and when a safe method of probing become available 
technically) and report its result in the dialog (i.e. "device 'Vario 
80' detected on port 1" ).

The theoretical issue of USB port renumbering for braille-and-speech 
users is one we will need to explore further - we don't know that this 
can happen with USB serial devices, but I have seen it happen with other 
USB devices such as switches, mice, and head trackers, so we should 
investigate further.

regards,

Bill