To: Jansson
Subject: questions

Hi Patrik:

You write:
Provability: we get reusable proofs, adaptive proofs, ...

How do you express the correctness of an adaptive program?

You write:
Some algorithms are polytypic by nature.

I would say that most algorithms are polytypic.
Consider an algorithm that uses f(x). We can make
it polytypic by using fetch(x, "from X to F").

Larsson writes:

Polytypic programming is based on parameterization of method code
on types. A polytypic object is instantiated to a family of 
specific objects that share a common structure.

==

In AP, the polytypic objects also share a common structure that
is expressed by the class graph of the aspectual collaboration.

I like Larsson's view. Is it different from your view?

Larsson also talks about Galois connections and embedding relations.
Do you know an example of a Galois connection involving 
an AP concept. I don't know enough about Galois connections.
Can we gain anything by expressing graph embeddings as Gallois connections?

-- Karl