here are some mathematical questions for you to complement your work
on the coordination aspect and aspect weaving. They are important
for dealing with the evolution of class graphs and strategies.
We would like to know how well traversal graphs reflect
changing path sets.

Notation:
We have a strategy S and a class graph C. They determine the traversal
graph T_S(C) and path set PathSet_S(C).

Prove or find a counter example for the following statements:

For all strategies S1 and S2 and class graphs C:
PathSet_S1(C) = PathSet_S2(C) iff T_S1(C) is isomorphic to T_S2(C)

For all strategies S and class graphs C1 and C2:
PathSet_S(C1) = PathSet_S(C2) iff T_S(C1) is isomorphic to T_S(C2)

(use default name map)

-- Karl
PS. The strategy paper is at:
ftp://ftp.ccs.neu.edu/pub/people/lieber/strategies.ps

================

Boaz conjectures that traversal graph isomorphism is sufficient
for equal path sets, but probably not necessary.

Boaz thinks that we can find two non-isomorphic traversal graphs
which define the same path set.

Boaz also thinks that the question of class graph evolution
is best studied for primitive transformations applied to class
graphs (see Walter Huersch's thesis). The problem:

INPUT: a strategy S and two class graphs C1 and C2,
OUTPUT: is PathSet_S(C1) = PathSet_S(C2) ? 

may be computationally hard.

Instead, use C1 and PrimitiveTrans(C1) and ask whether the
path set has changed.