On 3/6/07, Stefano Lanzavecchia <[EMAIL PROTECTED]> wrote:
> "Type safe" languages make it difficult (but not impossible) to
> express certain mathematical concepts. This tends to translate
> into bloat, and extra conditionals, in cases where you need those
> concepts.
Given that most functional programming languages are typesafe and are
renowned to be able to express even non-trivial mathematical concepts in a
very concise way, I find the quoted statement a bit extreme. Type-safety
doesn't necessarily mean that the types must be indicated by the developer.
Hindley-Milner type inference (and its extensions) has gone a long way since
it was first introduced in the early 70s...
http://en.wikipedia.org/wiki/Type_inference
I don't see the conflict that I think you are trying to express.
I agree that type safety -- a prohibition against certain classes of
puns, or mappings which would otherwise be expressible in
a notation -- is not locked to any single notation.
That said, any typing issue which is not detected in a single
run-time evaluation of some sample code is potentially an
example of a mapping that could be prohibited. I don't see
why you have a problem with this idea.
--
Raul
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
