On Tue, Aug 26, 2014 at 11:42 PM, John Cowan <xxxxxx@mercury.ccil.org> wrote:
Alex Shinn scripsit:

> Specifically I think because of the combinatorial explosion in memory
> needed, such operations are better served by a generating API than
> one which returns a complete data structure, either as an abstract
> generator or as a loop syntax extension.

I agree that such an API would be a Good Thing.  However, I don't agree
that because something might run till the Big Rip (or Big Crunch)
if applied to sufficient data, that it should never be used at all.
When you need exponential algorithms, and your problem is small enough
to use them, you should have them.

Sure, I'm not trying to argue against the utility of these
procedures.  I just wanted to point out that the issues
involved are, as usual, deeper than they initially appear
and suggest this be left to a separate library or SRFI.

Also, as a user of (srfi 113) I would prefer if it didn't
have too much bloat, especially from lesser-used
utilities.  Just my 2¢.

-- 
Alex