On 2011-06-15 20:29, Charles McAnany wrote: > Ah, so does the compiler figure out which ones are strongly and weakly pure > and then optimize as appropriate? Is there a way to indicate that a > function is strongly pure? Because it would seem odd to call a function > you thought was pure and wind up with a mutated argument. -Charles
It is not odd to have a function which is pure wind up with a mutated argument. All pure by itself does is indicate that the function cannot access mutable, global variables. That's it. Now, extra calls to a strongly pure function can be optimized out within an expression, because it is not possible for a strongly pure function to alter its arguments, and so it is guaranteed to have the same result every time that it is called, and it is guaranteed to have no side effects. So, yes, calls to strongly pure can be optimized out. What makes the difference between a weakly pure function and a strongly pure function is their parameters. All of the parameters of a strongly pure function are either immutable or implicitly convertible to mutable. The parameters of a weakly pure function can be anything (just so long as they aren't all immutable or implicitly convertible to immutable, since that would make the function strongly pure). The advantage to weakly pure functions over impure functions (aside from the guarantee that they don't alter the state of any global variables) is that they can be called by other pure functions. This makes it possible to have strongly pure functions which call all kinds of useful functions which may alter their arguments, because the arguments of the strongly pure function are never altered, and its return value is always the same. - Jonathan M Davis
