True, but in this case, the template serves a different purpose from the macro. It's there to make zap polymorphic, and would be unnecessary in a dynamic or ty

True, but in this case, the template serves a different purpose from the macro. It's there to make zap polymorphic, and would be unnecessary in a dynamic or type-inferred language. The main point is that pg's macro can be implemented as an ordinary function in a language that has references.