# Is there a more-robust FiniteMap?

**Johannes Waldmann
**
joe@isun.informatik.uni-leipzig.de

*Mon, 25 Feb 2002 09:38:53 +0100 (MET)*

>* Using FiniteMap, I often run into robustness problems.
*
same (?) here! a student of mine recently came across a similar problem
(with a hand-made search tree): he expected to run
listToFM $ [(0, ()) | i <- [0..100000]]
in constant space, but instead it took all the stack and/or heap.
we made the data constructors strict, and used strict fold for insertion.
this helped with ghc (but not wih hugs).
are there cases where non-strict behaviour of a finite-map implementation
is useful? I can't think of any (analogy: there is no lazy sorting either)
- so perhaps the above suggestions should be take as defaults?
perhaps ghc already does this (by specializing) when it uses FiniteMap.lhs.
regards -
--
-- Johannes Waldmann ---- http://www.informatik.uni-leipzig.de/~joe/ --
-- joe@informatik.uni-leipzig.de -- phone/fax (+49) 341 9732 204/252 --