On 27/07/07, Taral <taralx@gmail.com> wrote:
On 7/27/07, Andrzej Jaworski <himself@poczta.nom.pl> wrote:
It offers first hand information that the usefulness of Haskell as programming language amounts exactly to 0. This is a big surprise for me. On the contrary, I think Haskell has been getting too expressive at the expense of fundamental issues left neglected like relatively weak module system.
I think you very much misunderstood what was being said. What Simon appears to say (in my opinion, admittedly) is that Haskell pre-IO was not useful. By adding selective unsafeness (IO, unsafe* functions), Haskell is made into a significantly more useful programming language.
Exactly. As I understand it, he's saying that the C style of languages and Haskell are trying to approach the same goal from two different, somewhat orthogonal, directions. Haskell by starting pure and doing what they can to make it pragmatic without destroying any nice properties. The imperative languages are doing it by starting unsafe/unproductive and "useful" and trying to make it less unsafe/unproductive (often by borrowing from languages like Haskell, but general non-FP-specific features like garbage collection certainly comes to mind)... So I think he's saying that Haskell, with monads for IO, ST, STM, and concurrency, and FFI etc. *is* very useful and securely located in the realm of practical feasibility for most projects, but that it arrived at this position from a different starting point. -- Sebastian Sylvan +44(0)7857-300802 UIN: 44640862