Conceptual question: how is this language different than other prologs? about xl

[c3d]

Don't let the superficial syntax similarity fool you. XL is not evaluating things the way Prolog does, at least not by default. This is because Prolog's backtracking, while very practical and useful in some cases, is not necessarily expressive in others. I would not do heavy duty number-crunching in Prolog, for example.

It was, however, important to me to design a language that would make it possible, if not easy, to implement Prolog-style inference.

The way XL approaches the problem is to focus on the process of translation. At the root, there is a very simple parse tree, and the is operator describes how you transform the parse tree. This is why a pattern in the form Pattern is Implementation is called a rewrite. Conceptually, it rewrites code that looks like Pattern as code that looks like Implementation. This is true even for builtin operations, where the implementation will be builtin Something.

This means that if you can find a nice syntax to "tag" some specific rewrites as being Horn clauses, then you can completely control how the rewrite for these happens. We could even use the Prolog syntax for that, since it's not used in XL.

move 1,X,Y,Z :-  
    writeln "Move top disk from ", X, " to ", Y

move N,X,Y,Z :- 
    N>1                     // May backtrack here
    M = N-1 
    move M,X,Z,Y
    move 1,X,Y,0
    move M,Z,Y,X

The implementation could use automatic returns as currently envisioned for the error types, i.e. boolean value would cause backtrack / return if false.