[m-dev.] for review: tuples [1]

[Simon Taylor]

Peter Schachte wrote:
> Hang on.  If you're going to do that, wouldn't you be a lot better off using
> the defn:
> 
> 	:- type (T1,T2) ---> (T1,T2)
 
> I can only see one disadvantage:  for triples and above, the representation is
> inefficient (and for pairs, the standard pair type is probably fine).  This
> could be fixed by implementing structure inlining of single-constructor types.
> Of course, this would have benefits in many other situations, as well.  Is
> there some reason that this is a more difficult optimization than it would
> seem?

It doesn't work well with polymorphic predicates because
the structures would need to be un-inlined before passing
them to a predicate such as io__write or std_util__deconstruct.
This optimization is in the same category as the type unfolding
discussed earlier in this thread -- it's difficult to work out
when it will improve performance.

Simon.
--------------------------------------------------------------------------
mercury-developers mailing list
Post messages to:       mercury-developers at cs.mu.oz.au
Administrative Queries: owner-mercury-developers at cs.mu.oz.au
Subscriptions:          mercury-developers-request at cs.mu.oz.au
--------------------------------------------------------------------------