Thank you all for showing interest and responding.
Check out http://thyer.name/lambda-animator/. Requires Java.
Wow, this is SUCH a cool tool. Best discovery in a long time! I think I need to brush up on my lambda-calculus, because it took me some time to figure out what settings to use to get something similar to Haskell. Function strategy "substitute by name", Argument strategy "call by need" and Reduce to "weak head normal form" seem to work OK, though. One issue, though. When trying out my infinite-list-of-fibs example, I have an auxiliary function (define nth (lambda (n xs) (if (= n 0)(head xs)(nth (- n 1) (tail xs))))) and this is not behaving the way I want, i.e. like the Haskell function nth n (x:xs) = if n == 0 then x else nth (n-1) xs because it doesn't do pattern matching, i.e. it doesn't force the argument to be evaluated until it fits the x:xs pattern. I can't figure out to simulate this eager pattern in the lisp that the lambda-animator uses. This means that if I e.g. do a (nth 8 fibs) in the animator, I get a long string of tail's before it actually starts expanding the fibs list... Anyway, on the side, I also started working on this myself in SML New Jersey. (Sorry, this will probably make me unpopular here on Haskell-cafe, but the ability to use references was just too tempting, and I'm not too experienced with purely functional data structures). My approach isn't as general, though, I don't have lambda's and apply's in my syntax tree. I just have functions there as nodes, and then the application of them has to be implemented as reduction rules. This approach does make the graph a bit less messy to look at, though. Also, since I implement the reduction rules myself, the pattern matching problem described above isn't a problem.
I think HOPS is what you are looking for http://www.cas.mcmaster.ca/~kahl/HOPS/
Yes, HOPS looks very promising. However, I cannot find the download link either. And Kahl hasn't replied to my email yet. Again, thank you all for replying. Kai