[mercury-users] Hi, and determinism

[Richard A. O'Keefe]

Of course the problem of relative determinacy can be hacked around in
particular cases, but I think the general problem is interesting.  The
notion of having-some-property-relative-to-an-oracloe is common in
theoretical computer science, and it's really very natural in programming:
	x raises no exceptions of its own, it only passes on exceptions
	raises by y
	x does no i/o of its own, but its argument procedure y may
	x will not fail unless a call to y fails

In the case of semitdet, we could imagine
	if [p(in,in,out) were det] then det else semidet
as a mode, meaning that the analysis of the predicate is to take place
twice, once with p(in,in,out) is det assumed and once with the real mode
of p/3; the resulting mode should be det in the first case and semidet
in the second.

In particular, higher-order predicates like list mapping want to inherit
determinacy from their arguments in some such fashion.

But I'd rather waffle about it than design it properly!
--------------------------------------------------------------------------
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
--------------------------------------------------------------------------