I'm using Geomerative to load some RShapes from a font and to divide a single shape (in this case representing a single char) into several 'slices' vertically.

I'm using the intersection method from the RG class, thus intersecting different rectangles with the shape to be cut. Here is an example of what I am doing:



From this image you could tell it is fine at this point, but I am finding many inaccuracies in the ending result (those are more visible when exported to pdf). I'm uploading a second image with a coloured background to show them: there is a less than one pixel space between slices and some extremities in the final shapes are badly 'cut'.





I recognize that much can be traced to numerical inaccuracies (we are talking about floats and their precision...) but I guess there is more.
Here are my questions:
1) is there a way to get a more precise intersection result with geomerative (or even an equivalent on processing)?
2) is the pdf render more subject to inaccuracies (and if so any alternatives to export to vector graphics)?
3) am I approaching the problem in a wrong way or expecting too much?

EDIT: sorry for the big post and merry xmas to everyone : D