|
> The two calls to "eval" are identical, yet return
> different results. That breaks referential transparency.
You are right, I don't know much about the syntax of clojure, but the Scheme version works as I'd expect. Yes, the eval calls return different results, but then again, we'd expect to compute a different output for different inputs.What I tried to show is that calling (swapargs (swapargs '(a b c))) will always return the original list, that is, demonstrates referential transparency. In the case of '(modulo a b), result of evaluation returns a the same result when repeatedly given the same a, b inputs. The point of the macro was exchanging the second and third elements of the input list. Naturally for the modulo operation, the order of the inputs is significant, and exchanging the operands will give the "opposite" remainder as the result. In your example the two calls to (eval ...) are not identical and the "different results" are perfectly correct, without implications for referential transparency. Don't know what kind of applications you might have in mind, but of course, no PL is optimum in all domains. For the kinds of programs I've tackled, Lisp/Scheme has been a good fit. Or maybe it has to do with the way my brain works just as much as the purposes I am applying the language to. That wouldn't surprise me a bit. |