[mercury-users] mode system limitation

[Ian MacLarty]

Hello,

I have the following piece of code:

:- type t1
	--->	a
	;	b.

:- inst a
	--->	a.

:- type t2
	--->	t2(f1 :: t1).

:- inst t2(I)
	--->	t2(I).

:- pred p(t2::in, t1::out(a)) is det.

p(T2, X) :-
	(
		T2 ^ f1 = b,
		X = a
	;
		T2 ^ f1 = a,
		q(T2, X)
	).

:- pred q(t2::in(t2(a)), t1::out(a)) is det.

q(t2(a), a).

When I compile this code I get the following error:

test.m:035: In clause for `p(in, out((test.a)))':
test.m:035:   in argument 1 of call to predicate `test.q/2':
test.m:035:   mode error: variable `T2' has instantiatedness
test.m:035:   `bound(test.t2(ground))',
test.m:035:   expected instantiatedness was `bound(test.t2(bound((test.a))))'

If I then change the call to q/2 in the body of p/2 to be:

		q(T2 ^ f1 := a, X)

it compiles with no errors.

I have the following questions:

1. Is this a limitation of the current mode system or a bug?

2. If it's a limitation would it be easy to extend the mode system to
   handle this case?

3. Is there an alternative modification I can make to the code that
   doesn't involve a redundant update of T2 and avoids this limitation
   of the mode system?

Ian.
--------------------------------------------------------------------------
mercury-users mailing list
Post messages to:       mercury-users at csse.unimelb.edu.au
Administrative Queries: owner-mercury-users at csse.unimelb.edu.au
Subscriptions:          mercury-users-request at csse.unimelb.edu.au
--------------------------------------------------------------------------