Re:

[Tim Klinger <klinger@xxxxxxxxxx>]

At 11:57 AM 8/24/98 -0700, you wrote:
>Hmm,
>This idea of saving state in component-based pattern-speak is known as a
Memento
>pattern.  It bothers me that your solutions are all inheritance based.  I
have
>objects that are already of a type and this solution will drive me to
multiple
>inheritance which has other problems.  A component based solution along the
>Memento line frees me of these constraints.  Comments on this?

That's a good point.  I played around with a containment solution today.
It works on my test cases and seems like it could be made to work in
general, but it's tricky.  I'll gladly send you the code that I wrote but
you might have to be the guinea pig in this experiment since I don't have a
lot of time to work on it.

>To Tim's interesting point that they designed the 'revertable' data
structures so
>as not to incur the psychological barrier of not trying algorithms because
they
>might take too long, I can add that in my experience the boredom of actually
>getting the incremental computing right was enough to put me off.  8-) 

Actually, it was David's point but, of course, I agree :).

Tim