Hi Mira:

it is important to make a clear distinction between structure-shy
and structure-generic. At the same time we need to keep all our 
definitions consistent.

I don't know yet what the best solution is.

In
http://www.ccs.neu.edu/home/lieber/connection-to-aop.html

I write:
Adaptive Programming is the special case of Aspect-Oriented Programming where one of the building
blocks is expressible in terms of graphs and where the other building blocks refer to the graphs using
traversal strategies. A traversal strategy is a partial specification of a class diagram pointing out a few
cornerstone classes and relationships. Traversal strategies may be viewed as regular expressions specifying
a traversal through a graph. 

Therefore, programming with APPCs is still adaptive programming
altough it is a better form of adaptive programming.

But at one point you say in the paper: 
AP is structure-shy programming.
APPCs are structure generic 
(and therefore not a part of AP (that could be one possible interpretation)). 

Here is a comparison of structure-shy and structure-generic:

structure-shy			structure-generic
class graph - strategy          class graph - ICG - strategy
adaptive to a family of 	adaptive to a family of
  class graphs			  interface class graphs
				adaptive to a family of
				  class graphs 

Maybe we could do the following:

structure-generic = structure-shy with an interface
                  = structure-shy with one more level of indirection
structure-generic = structure-shyness + specializing strategies	


I view the definition of the structure-generic requirement R1
as generic description of what structure-shyness is about.

On page three we could say:
The structure-generic requirement goes further in that it implies
that the software is adaptive to a family of interface class graphs
(each interface class graph = a family of class graphs) rather
than only to a family of class graphs.

But also this does not sound right.

The use of an interface class graph does not really improve structure-shyness
but it improves the range of class graphs to which the APPC
can be applied to using _manual_ intervention.

structure generic is about
mapping abstract strategies to concrete ones when
a class graph is given.


I think we need to talk about this. Are you in?

-- Karl