On Dec 9, 2008, at 14:39, Steve Ratcliffe wrote:
On Tue, Dec 09, 2008 at 02:11:55PM +0100, Robert Vollmert wrote:
Another possibility: MapLine.setPoints occasionally gets passed consecutive identical points, I assume from rounding to the same coordinates. I think they're being written as zero deltas, which I could well imagine throwing MapSource off. How going through GPSMapEdit would fix this, no idea.
There is supposed to be code to skip zero deltas in LinePreparer. This is because even if its ok at high zooms at lower ones, zero deltas will start to appear (althogh there are supposed to be filtered out earlier on too).
Right, found it now. There's still possible problems: 1. The effective number of points could drop below two, even though there's a check for at least two points initially. Though from what Alex said about one-point segments, we may want to allow that anyway. 2. If coordinates in the .mp-file are rounded to the same point in map units, this might still confuse the routing graph, since we'd create different routing nodes at the same coordinate. There's then also the possibility of arcs with length zero. We should probably check for this in polish.RoadHelper, though it'd be nice if osm2mp took care of that before. Also with the Coord/CoordNode thing, this isn't quite straightforward to implement. Cheers Robert