[mercury-users] Mutual Exclusivity & Exhaustiveness

[Lars Yencken]

On 03-Jan-2002, Simon Taylor <stayl at cs.mu.OZ.AU> wrote:
> On 03-Jan-2002, Lars Yencken <lljy at students.cs.mu.oz.au> wrote:
> > > We definitely want to disallow examples like:
> > > 
> > > :- promise_exclusive(X::ground, Y::ground) is det :-
> > > 		( foo(X)
> > > 		; bar(X, Y)
> > > 		; baz(Y)
> > > 		).
> > 
> > If there was a case where this was useful, why would we definitely want to 
> > disallow it, aside from simplicity's sake?
> 
> There are two cases where it would make sense for foo(X) and baz(Y)
> to be mutually exclusive.
> 
> 1. One of them can never succeed (it has determinism `failure' or `erroneous').
>    This case does not need a `promise_exclusive' declaration.
> 
> 2. They are impure, and the mutual exclusivity is as a result
>    of some hidden state.
> 
> For example: 
> 
> :- pragma c_code("int state = 1;").
> 
> :- semipure pred state is semidet.
> :- pragma c_code(state, "SUCCESS_INDICATOR = state;").
> 
> :- semipure pred not_state is semidet.
> :- pragma c_code(not_state), "SUCCESS_INDICATOR = !state;").
> 
> :- promise_exclusive is det where ( state ; not_state ).
> 
> This case should be rare enough that it can be ignored.
> Even when it does occur, I think it is so non-obvious
> that an explicit if-then-else should be required.

Ok, agreed. Thanks ;)
--------------------------------------------------------------------------
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
--------------------------------------------------------------------------