Hi
Well, if this doesnt take too much calculation time, please go ahead :-)
OK here is a first attempt.
The parsers are modified to save a flag on the way if it is actually closed. It also aggressively refuses to make anything into a polygon that is not closed.
So there should be errors as a result of splitter. But on the other hand there may be more due to OSM data errors. Will have to investigate this and adjust if necessary.
Attached is the before and after results of my test and the patch.
A precompiled jar is at: http://files.mkgmap.org.uk/download/72/mkgmap.jar
..Steve
Afer a quick code inspection I would say: Looks good! The only thing that must be checked carefully is the new meaning of the isClosed() method. I am quite sure that there are some places at least in the multipolygon code that relies on isClosed() == true => getPoints().get(0).equals(getPoints().get(getPoints().size()-1)) WanMil