Le 10/11/15 11:55, Marc LE BIHAN a écrit : > There is no classes designed to draw the contents of features in Apache SIS > yet, isn't it ?
Right. There is a rendering engine in http://www.geotoolkit.org (the sandbox of the code that we are porting to SIS), but I think that it is using an older Feature model than the one in SIS. The Geotk refactoring from its old Feature model to the SIS one is a work in progress in a branch. Johann Sorel could tell more about that. Geotk provides many Swing widgets. But for Apache SIS, I wonder if we should target JavaFX instead. > Won't polygons that resolves in two distincts polygons in fact cause some > troubles in some calculations, intersections, and other mathematical or > geometrical operations ? The computation will need to be robust to those situations. This is the case for example of Java2D geometry objects (e.g. java.awt.geom.Path2D) when computing whether a point is inside or outside the geometry. I did not experimented the ESRI library yet (Geotk uses JTS, but its license is not compatible with Apache - migration from JTS to ESRI library is part of the work that we need to do), but I would expect it to be sufficiently robust too. > Maybe the users / callers of Feature assume that one feature only depicts > one polygon (and so on : one line, one point...) ? I think that users should not assume that. Features may be quite complex as their intend is to represent real-world phenomenons (network of rivers, etc.). Features may have an arbitrary amount of geometries. However we can help the users by telling them that one particular geometry in a Feature can be used as the "default geometry". Martin
