Le 8 juin 07 à 22:41, Arnaud Spiwack a écrit :

> 1/ Do the reader of this list encounter the need of existential  
> type often?

Sometimes it can be useful to express an abstract type + operations  
without going through functors which only provide static existential  
types. This allows to store in the same datastructure such types +  
operations with different implementations. For example imagine a  
spatial datastructure with both float32 bigarrrays and regular caml  
float arrays.

> 2/ How would the addition of existential types impact the typing  
> algorithm of OCaml? (because I must confess that I have absolutely  
> no clue, would there still be a principal type to every expression?  
> would that increase complexity?)

I don't know.

But if you really need them they can be encoded in the current type  
system, see the discussion here [1].

Best,

Daniel

[1] http://caml.inria.fr/pub/ml-archives/caml-list/ 
2004/01/52732867110697f55650778d883ae5e9.fr.html