On Wed, 26 Sep 2001, Mattias Waldau wrote:
> I looked thru the documentation, but FGL doesn't seem to be any operations
> like closures and topological sorting in fgl, it mostly seems to be about
> traversing the structure. Have I missed something?

The interfaces don't have, say, topsort as a builtin function, but the
papers describe how to implement it using the provided functions.

> When I look at one of the papers, you get the impression that an imperative
> implementation is one to two magnitudes faster, is that right?
> (Speed is irrelevant right now for me.)

I don't think the papers measured against an imperative implementation,
but rather a functional array based on binary trees versus a functional
array based on version arrays. But I could be wrong and I don't have them
handy at the moment. Anyways, I'd like to do a version array version :-)
too in the future. It would be nice if there was a version array library
around; I've written some quick ones but I don't know how they compare
against some of the better ones. Lots of SML implementations around.

Beware of surmising too much based on some quick benchmarks.

-- Brian


-------------------
Bug reports: http://caml.inria.fr/bin/caml-bugs  FAQ: http://caml.inria.fr/FAQ/
To unsubscribe, mail caml-list-request@inria.fr  Archives: http://caml.inria.fr