[aosd-discuss] Inheritance hierarchies

[Ashley at Metamaxim]

Raffi

> But how are aspects compositional? The problem with reasoning about AO
> programs is that doing so generally requires whole-program analysis.

At least in the approach we have been using, a well defined form of local 
reasoning is possible, so whole-program analysis is not required. Primarily 
this is because we are using the parallel composition operator (P||Q) of 
Hoare's CSP to combine aspects with their base. The form of local reasoning 
this enables is fully described in the paper I referenced.

I find it hard to understand how it is possible to use aspects to any 
significant degree unless at least some degree of local reasoning is 
possible.

Rgds
Ashley