generalize geometrical objects

[stefanobortoli]

Hi, I found out that throughout the code, you implement first level shapes as subclass of serializable. It would be useful to have a generic superclass (e.g. Shape, or Geometry) so that one could define generic interfaces and then implement them accordingly. What do you think? This is just a suggestion, of course.

[lessthanoptimal]

that wouldn't be too hard to add. Would you sub divide it into even more categories?

[stefanobortoli]

I guess it depends whether there are implications related to inheritance. Perhaps it could be useful to distinguish between "cartesian aligned objects" vs "arbitrary aligned objects". My limited understanding is that if we assume geometries to be aligned on the axis is it easier to compute projections and therefore other relational functions. Whereas if they can be rotate arbitrarily, then things can become nasty. I am thinking of IoT or IoV scenarios where it is useful to consider 3D but rather limited with respect to "abstract" mathematical properties. So, perhaps it would be easier to extend the pool of relational functions making some assumptions for some types of geometries. What do you think?