On Monday, July 14, 2014, Matt Rice <[email protected]> wrote: > The quotation from the spec referring to "interface compilation unit", > is just the paragraph from the spec where this particular behavior of > headers/types implicitly generating a constructor seems to be defined, > as I observed it in the behavior of the last iteration of the bitc > compiler. >
So I now understand this to be a question about type and value constructors in interface UoCs. In BitC, the root of the problem is that there is no real distinction between a type declaration and a type definition. If the definition appears in an interface UoC, then it defines the type constructor and the value constructor as well. It simply never occurred to me that construction and deconstruction should be distinguished. It's an interesting point to consider. We do currently have opaque declarations, but that's not quite the same. Have I mentioned how much it bugs me that you keep asking really good questions? :-))) Thanks, Matt!
_______________________________________________ bitc-dev mailing list [email protected] http://www.coyotos.org/mailman/listinfo/bitc-dev
