Re: prelude extension proposal

From: Michael Hanus <mh_at_informatik.uni-kiel.de>
Date: Wed, 12 Oct 2005 17:19:30 +0200

Sebastian Fischer wrote:
> > The second one serves the same purpose as the prelude function
> > until, but for constrains instead of booleans.
> >
> > cuntil f x = case allValuesOf (f x) of
> > [] -> x
> > (r:_) -> cuntil f r
>
> this would be a nice extension, although I think the name is a bit
> misleading: "until" stops when the given boolean predicate holds, and
> "cuntil" continues while the condition holds. Maybe we should call it
> cwhile then?

It seems to me that the operation is even more general since
it applies the transformation as long as the result is a defined value.
Thus, "whileDefined" could be a better name. For the reasons
already explained by Bernd, I'd prefer to put it into a
module like AllSolutions
(http://www.informatik.uni-kiel.de/~pakcs/lib/CDOC/AllSolutions.html)
as an I/O action. Using the operation getFirstSolution already
defined there, one can implement whileDefined as follows:

whileDefined :: (a -> a) -> a -> IO a
whileDefined f x =
  getFirstSolution (=:=(f x)) >>= maybe (return x) (whileDefined f)

However, the analogon to "until" would be an operation "untilValid"
which could be defined as follows:

untilValid :: (a -> Success) -> (a -> a) -> a -> IO a
untilValid c f x =
  getFirstSolution (=:=(c x)) >>=
  maybe (untilValid c f (f x)) (const (return x))


Of course, if such operations are often used, it seems reasonable
to provide them also in the prelude. At least the function
"getAllValues" is a good candidate.

Best regards,

Michael

_______________________________________________
curry mailing list
curry_at_lists.RWTH-Aachen.DE
http://MailMan.RWTH-Aachen.DE/mailman/listinfo/curry
Received on Wed Oct 12 2005 - 17:50:04 CEST

This archive was generated by hypermail 2.3.0 : Mon Nov 18 2019 - 07:15:05 CET