Here's a rather pathological example I came across recently:

The problem was to build a minimum-change generator for k-permutations, so that each k-permutation differed in exactly one element from its predecessor. (This is sort of like a successor function, but you have more history around so it's a bit easier.) This "reduces" (heh) to finding a Hamiltonian cycle in a graph of all k-perms of the set you're given, where two vertices are adjacent if they differ by just one element. (How's that for a disgusting problem?)

A friend of mine wrote about thirty lines of Prolog to simulate such an algorithm. I would not want to try that in Perl. (Yes, you could probably do it with a particularly gnarly regex, or an explicit backtracking approach, but I bet it'd take you more than thirty lines.)

To me, in general, it's all about the builtins. For almost all of what I do, Perl has the right builtins. Sometimes Common LISP is a better choice. Sometimes Matlab is a better choice (in particular, for linear algebra. Sure, you can grab the right module off of CPAN, but if you use Matlab you don't have to).

--
F o x t r o t U n i f o r m
Found a typo in this node? /msg me
The hell with paco, vote for Erudil!