From: jdb9608@cs.rit.edu (John D Beutel)
Message-Id: <9110210329.AA18621@junior.rit.edu>
Subject: Re: Interfaces to VR devices
To: yanek@mthvax.cs.miami.edu (Yanek Martinson)
Date: Sun, 20 Oct 91 23:44:51 EDT
Cc: glove-list@karazm.math.uh.edu
In-Reply-To: <9110202011.AA16502@mthvax.cs.miami.edu>; from "Yanek Martinson" at Oct 20, 91 4:11 pm
> We could have some set of operations/functions that all interfaces
> should support, either using the hardware feature or simulating it in
> software.

It's difficult to plan the union of these fundamental functions
when constraints like speed are so important and most of these
devices are unknown.

Take the PowerGlove, for example.  I have to take back my
suggestion of a standard interface function called pre_glove().
Originally it was for the interrupt driven version of the
hi-res driver I was working on, to free up the CPU for graphics.
I thought that starting the glove poll about 70 ms before the
data was needed would be a crucial optimization, before it
occured to me that the important data should be gotten in the
first 1 ms so the poll could be started and the important
data returned in one function call.  Now, pre_glove() isn't
needed.

Since we're just starting to learn how to interface with the
glove, there may be more fundamental changes in the way we do
it.  So, we shouldn't make a standard yet.  On the other hand,
if we could nail down a standard interface before we had much
experience making programs for it, then the programs would be
compatable, which is desirable.  It's a catch-22.

The various techniques of using the PowerGlove have differences which
must be small compared to the ones of different devices, and I've never
used any of the other VR input devices, so I can't suggest what
the fundamentals should be.  Judging by the PowerGlove so far,
I do expect more differences than there at first seems to be.
Francis, I second the request for more info on what the
HIT Lab has learned on standards for the variety of VR devices.
Without practical experience with these other devices,
I think making a standard for their use would be largely
a waste of time.

I'd really like to see a nice PowerGlove standard, at least
on the fundamentals like how it'll be polled or interrupt
driven.  For example, people using fast polling with
interrupts, or especially people who get 68HC11 boards working
and send serial data only when it's ready, will probably want
a standard function that will be called by an interrupt whenever
the next data packet arrives.  That hasn't been mentioned
for the glove standard yet because nobody's implemented that
kind of interface yet.  But they will...

--
J. David Beutel  11011011  jdb9608@cs.rit.edu  "I am, therefore I am."