Daniel Keep wrote: > Actually, I've now come up with a counter-example for the idea of using > pure at all: That's right, lazy evaluation can't be pure. So, the question is is this an important enough case to justify a whole new syntax?