[mercury-users] Existential types and list.map

[Ralph Becket]

Peter Hawkins, Friday, 29 July 2005:
> Mark Brown wrote:
> 
> >This is not type safe because there is nothing in the signature of 
> >get_thing
> >that guarantees that all of the elements in list Y have the same type.  To
> >build such a list you would need to have get_thing return a structure that
> >contains an existentially typed value.  For example, the signature for
> >get_thing could be:
> >
> >	:- func get_thing(int) = univ.
> >
> >Then Y would have type list(univ), which would be okay.
> >
> > 
> >
> Yes, but I want to put typeclass constraints on the resulting type, so I 
> can't use univ. As far as I can see, my only option is to define a new 
> list type, which is just hideous.

No, a data constructor can have existentially quantified arguments:

:- type thing ---> some [T] thing(T) => thingable(T).

:- func get_thing(int) = thing.

get_thing(X) = 'new thing'(X).

And now you can have lists of things.

-- Ralph
--------------------------------------------------------------------------
mercury-users mailing list
post:  mercury-users at cs.mu.oz.au
administrative address: owner-mercury-users at cs.mu.oz.au
unsubscribe: Address: mercury-users-request at cs.mu.oz.au Message: unsubscribe
subscribe:   Address: mercury-users-request at cs.mu.oz.au Message: subscribe
--------------------------------------------------------------------------