[Haskell-cafe] Proof that Haskell is RT

[Thomas Davie]

On 12 Nov 2008, at 14:47, Mitchell, Neil wrote:

>
>> It's possible that there's some more direct approach that
>> represents types as some kind of runtime values, but nobody
>> (to my knowledge) has done that.
>
> It don't think its possible - I tried it and failed.
>
> Consider:
>
> show (f [])
>
> Where f has the semantics of id, but has either the return type  
> [Int] or
> [Char] - you get different results. Without computing the types
> everywhere, I don't see how you can determine the precise type of [].

Surely all this means is that part of the semantics of Haskell is the  
semantics of the type system -- isn't this expected?

Bob