The Evaluator

It is often convenient to automatically create parts of a model from parameters or based on other parts and known functions. The evaluator is a function that interprets AIDDL terms similar to the Lisp programming language for this purpose. Unlike Lisp, however, terms will not be automatically evaluated. It is up to the user to decide if or when this happens.

The evaluator takes any AIDDL term as an argument and recursively evaluates it according to the following cases:

• List: evaluate all arguments and return new list with result

• Set: evaluate all arguments and return new set with result

• Key-value pair: evaluate key and value and return new key-value pair

• Entry Reference: resolve reference (default) resolve reference and evaluate recursively evaluate result (must be enabled manually)

• Tuple (x1, x2, ..., xn):

  • x1 is not a function URI: evaluate all arguments and return new tuple with result

  • x1 is a lazy function f: return f((x1, ... xn))

  • x1 is a function f: evaluate x2 to y1 and so on and create tuple (y1, ..., yn-1) of n-1 evaluated arguments

    • if n == 1: return f(())

    • if n == 2: return f(y2)

    • else: return f((y2, ..., yn))

• _: return argument unchanged

Any time it comes across a tuple with a known function as first element the function will be called on the remainder of the arguments evaluated. Under the hood there are a few more cases considered. Functions flagged for lazy evaluation are assumed to evaluate their own arguments. This is important to allow functions like a Boolean and or or to avoid evaluating all arguments in case a false or true result is found early.