Conclusions and future work

We have shown that the appropriate categories of flow graphs up to bisimulation are the intial categories with certain structure.

As a corollary, providing a data flow / control flow semantics for a language is exactly the same as providing a semantics in this categorical framework.

We have presented some tentative notions of higher-order graph and graphical operational semantics, similar to Milner's. These need future work.

This semantics has been implemented as a Java applet (which drew all of the graphs in this talk).

http://www.cogs.susx.ac.uk/users/alanje/premon/