On Tuesday, Jun 22, 2004, at 02:48 Australia/Sydney, der Mouse wrote:

>> I did test negative scaling (inversion) which would succeed in giving
>> me a headache pretty quickly.
>
> For normal use, yes.  But there are plenty of special purposes for
> which reversal is reasonable.  (As simple examples: it allows you to
> compensate for the mouse being physically reversed, or for its
> generating deltas that are reversed from how the software expects.)
>
> I would be tempted to make it a full 2x2 (or 3x3, or whatever) linear
> transform - in particular, such that it is possible to switch axes.
> That may be overkill, though.

While flexible to the extreme, I'm thinking it would be overkill. And
rather confusing to the novice user.

I think I'll settle for separate x,y,z scaling, and remove the check
for 0 numerator. And add read access. Unless anyone objects?

--
stix.