Jonathan S. Shapiro wrote: > On Wed, 2008-03-12 at 18:17 -0400, Swaroop Sridhar wrote: >> So, are you suggesting that if we write (by-ref 'a), where 'a cannot >> unify with a mutable type? This cannot be the case since the type >> variable must be able to unify with any type. >> >> Prof. Shapiro suggested introducing a notion of (immutable t) as a type. >> So, we can write (fn ((by-ref (immutable 'a))) ()) to represent a >> function that only accepts a immutable by-reference argument. > > A type system that has all of IMMUTABLE, MUTABLE, and parametric types > runs into convergence problems. What we *could* do (Swaroop and I > discussed this offline today) is introduce (IMMUTABLE 'a) where the > argument to immutable can only be a type variable, with the meaning "'a > cannot unify with a mutable type".
When we eventually get to introducing the immutable construct, I think it is better to introduce it in the form of a constraint ex: (forall (Immutable 'a) 'a) rather than as a constructor (immutable 'a). From an implementation point of view, either version is not a problem, but I think introducing immutable in the form of a constructor (as analogous to the mutable constructor) is confusing to the programmer. Isn't it is really constraining the type variable 'a, and not constructing a new type? Swaroop. _______________________________________________ bitc-dev mailing list bitc-dev@coyotos.org http://www.coyotos.org/mailman/listinfo/bitc-dev
