Eivind Sandstrand (c473bp@ms)

cse473 spring 1991

May 29th

*******************************************************************************






1)

a]

The one thing I had wished to be able to model in Prodigy was a timing of 

the different operations.  Then, at the end of the planner be able to see

which goal-/operator-/sc-rules-combinations would produce te best times for

the robot.  Also, along this line, would be other measuremenmts of the 

operations like chance of damage to parts, robot wear etc.  If I had done this

my domain would have been much more interesting as a model than it is today.

I would also have liked to have been able to experiment more with sc-rules.

This could potentially have increased the performance greatly, and given room 

for larger and more real-life problems.  Currently, the problems has to be 

kept to a minimum even with the sc-rules working.


b]

I was afraid the sc-rules would be difficult to encode, but they proved to be

(to me, that is!) more straightforward than the regular operators.  


c]

The fact that it runs under/with Lisp which enables emacs interaction and in

turn the possibility of creating a domain with only one window but using the

emacs' Mr.Buffer.  It would have been a major pain if was to jump in and out

of editor and Lisp to debug code.  


d]

Better debugging tools - being able to step through a domain without haveing

to run it first.

It's own integrated editor - would eliminate the need for memory-hogging emacs.







