Re: functional languages ill-suited for large programs?

[Alan Bawden <address@hidden>]

   Cc: ll1-discuss@ai.mit.edu
   From: Jim Blandy <jimb@redhat.com>
   Date: 01 Nov 2003 00:03:40 -0500

   Alan Bawden <Alan@LCS.MIT.EDU> writes:
   ...
   >   fib = ((map fib' [0 ..]) !!)
   >     where
   >       fib' 0 = 0
   >       fib' 1 = 1
   >       fib' n = fib (n - 1) + fib (n - 2)
   > 
   > is a memoized version that executes quite quickly.  It seems to me that I
   > have just expressed the notion of a memoized function in a purely
   > functional language.

   Yes, you have.  Please check my memory of Haskell: this computes
   (lazily) an infinite list whose n'th element is fib' n, and then
   curries the subscripting operator !! --- is that right?

Yup.  

   But what if I needed to be able to forget old memoized values?

Then you'd be moving the goal posts!  What you -said- was that a functional
language prevented you from expressing memoization.  Now you've changed
your tune!

   If you've got an argument that concealed state *isn't* often useful,
   then that'd be a direct rebuttal.

Well, I wasn't trying to rebut your overall argument nor was I trying to
claim that concealed state isn't useful.  I was just pointing out that your
-specific- claim (that functional languages prevent you from expressing
memoization) isn't true.

I was also indirectly making the point that it can be surprising what -can-
be expressed in a functional language.  I think you're right that state is
indispensable in many situations.  But I'm not as certain as I used to be
about exactly what those situations are, I've been surprised too often...