(2004-04-21) Coplien Pattern Meta

Jim Coplien on the theoretical underpinnings of any Pattern Language. Patterns go to the fundamental nature of what structure is. By definition, structure is symmetry; so patterns are based in symmetry. By definition, there is no perfect symmetry in nature; therefore, patterns embrace symmetry breaking and local symmetries. By definition, you add structure to a system in a way that minimizes energy; that's where the rules come from for building a language out of sequences.

Ralph Johnson weighs in. I suppose I ought to read your paper. But if you define structure as symmetry, it will be a useless paper. It will not correspond to any natural definition of structure... I agree that patterns are about structure, but they are not structures. They are transformations of a system that produce a structure. I don't like the word "function" because the systems are almost never well defined enough to formalize the patterns, hence we can't really define the transformations precisely.