Toby Bartels wrote:
... Certainly much of what is in the patent application is obvious, but perhaps not all of it; were these diagrams of diagrams a new idea?, or was applying them to computer system specifications a new idea?. ...
Dear Toby, The idea of treating specifications as colimits is a few decades old now. Burstall and Goguen used it in their categorical account of their specification language Clear, with a specification used to construct a new theory as colimit of others. The hierarchical step, diagrams of diagrams, was studied by Catherine Oriat in her thesis and (I believe) a TCS paper in 2000. My own student Gillian Hill investigated a variant of this (PhD Thesis 2002; also two papers with me, 2001, 2006), replacing the category of finite diagrams over a base category C by the equivalent category of finitely presented presheaves. Both are finite cocompletions, but a presheaf presentation by generators and relations comes over neatly as a "configuration by components and sharing". For obvious reasons the iterated construction "flattens" back down to the single one (the construction is a KZ-monad in the 2-category of categories). Gillian also investigated a multi-level configuration language that maintains the hierarchical structure without flattening (configurations of configurations of configurations of ...) and includes cross-level specification morphisms. However, we did not persevere to work out the categorical semantics of this, nor did we make a computer implementation. Regards, Steve Vickers. [For admin and other information see: http://www.mta.ca/~cat-dist/ ]