[m-dev.] for review: add support for existential types [4/4]

[David Glen JEFFERY]

On 30-Jun-1998, Fergus Henderson <fjh at cs.mu.OZ.AU> wrote:
> [continued from part 3]
> 
>  		%
>  		% check that the types of the candidate predicate/function
>  		% subsume the actual argument types
> +		% [This is the right thing to do even for calls to
> +		% existentially typed preds, because we're using the
> +		% type variables from the callee's pred decl (obtained
> +		% from the pred_info via pred_info_arg_types) not the types
> +		% inferred from the callee's clauses (and stored in the
> +		% clauses_info and proc_info) -- the latter
> +		% might not subsume the actual argument types.

You need a closing "]".

> +		%
> +		% add the constraints for this functor
> +		% to the current constraint set
> +		% For functors which are data constructors,
> +		% the fact that we don't take the dual
> +		% corresponds to assuming that they
> +		% will be used as constructors rather than as
> +		% deconstructors.
> +		%
> +		add_constraints(OldConstraints, ConstraintsToAdd,
>  			ClassConstraints),
>  		type_assign_set_typeclass_constraints(TypeAssign1,
> -			ClassConstraints, TypeAssign),
> +			ClassConstraints, TypeAssign2),
> +		type_assign_get_head_type_params(TypeAssign2,
> +			HeadTypeParams0),
> +		list__append(ConsExistQVars, HeadTypeParams0,
> +			HeadTypeParams),
> +		type_assign_set_head_type_params(TypeAssign2,
> +			HeadTypeParams, TypeAssign),

If you are assuming that it is being used as a constructor and not a
deconstructor, why do you add the ConsExistQVars to the HeadTypeParams?

> @@ -3270,8 +3619,13 @@
>  	--->	type_assign(
>  			map(var, type),		% var types
>  			tvarset,		% type names
> +			headtypes,		% universally quantified
> +						% type variables
>  			tsubst,			% type bindings
> -			list(class_constraint),	% typeclass constraints
> +			class_constraints,	% typeclass constraints:
> +						% both universal (assumed)
> +						% and existential (the ones
> +						% we need to prove)

I think that comment could be clearer. Maybe:
						% typeclass constraints that 
						% we need to prove. (either 
						% universal constraints from
						% callees or existential
						% constraints from the 
						% caller).





Well... that's the final part of the diff. Phew!

I'll have a look at your replies to my comments RSN.


love and cuddles,
dgj
-- 
David Jeffery (dgj at cs.mu.oz.au) |  Marge: Did you just call everyone "chicken"?
PhD student,                    |  Homer: Noooo.  I swear on this Bible!
Department of Computer Science  |  Marge: That's not a Bible; that's a book of
University of Melbourne         |         carpet samples!
Australia                       |  Homer: Ooooh... Fuzzy.