Hi Manfred
It didn't make much of a difference in runtime doing it without toList. I'm not quite sure if it is really more efficient or not. Perhaps I would have to create dictonary with some millions entries in order to see a noticable difference?
Probably, Haskell is not very deterministic for memory / runtime forecasts :-) In theory there should be no difference in complexity, both need one sweep over all entries of the Map. There could perhaps be a constant cost for constructing the list. Nevertheless, laziness should make it possible to run in constant space, as neither the list, nor the transformed Map is used. So differences are probably small. But is possible, take some measurements and report, not just runtime, but also heap space usage. regards Matthias