On Monday 27 October 2008 19:26:31 Bob Rogers wrote: > All true. But it's unfortunate that the Parrot type system considers > "int" and "Integer" unrelated. As a result, MMD prevents autoboxing > (when chromatic and I had expected otherwise).
I'm not sure it's the type system as much as it is Parrot's Manhattan distance algorithm. I think I know how to promote primitive registers to their autoboxed PMCs in that function; Parrot's calling conventions should take care of the rest. > Which brings up another interesting (to me) question: Does anyone > know of an HLL in which the type system makes a formal distinction > between boxed vs. unboxed values? I am acquainted with only a few > languages that support boxing/unboxing, and all consider an int an int, > boxed or not. Same here, though I note that Parrot's single-dispatch system considers an int an int. -- c