[Edu-sig] Re: The Great Computer Language Shootout

[Brent Burley]

Brent Burley wrote:
> Not to knock ocaml, but this is one of the worst cases of cheating in
> these benchmarks....

OK, before I get flamed to death (or maybe I already have and don't know
since I only get the daily digest ;-) ...

I was confusing currying and memoizing - Monday morning brain vacation. 
In Bagley's discussion of Ackermann's function, he says that languages
that perform automatic tail-recursion elimination (which ocaml does)
will perform significantly better.  He also mentions that some
functional languages implement a feature similar to automatic
memoization (though he doesn't mention ocaml explicitly in this regard).

Ocaml is clearly an impressive language, and I withdraw my accusation of
cheating on the benchmark - there's no fault on ocaml or the ocaml
programs.  However, if a language can bypass a function call on a test
that's meant to measure function call overhead, then the test isn't
doing its job (IMHO).  This adds more to the argument that it's hard to
make a meaningful benchmark.


	Brent