The following is an abstract of the draft paper that my talk at the Edinburgh-Glasgow workshop (of November 1994) was based on. The current draft is sometimes available from my home page.
Two abstractions from the branching discrete to the linear discrete model are given. They shed some light on the `consistency problem' and on the discrepancy between intuitive simulator behaviour and the linear discrete model. The stronger of the two depends on a notion of dynamic type for processes which ensures deadlock freedom.
A reasonably strong abstraction from the branching real to the branching discrete model is given. This depends on a finer notion of type which is a physically plausible formalisation of the timing properties of certain real components.