Re: bounding box calculations

[David Turner]

Hello,

Werner LEMBERG a écrit :
> 
> > Is there a reason why outline bounding boxes are computed by
> > considering the control points of curves, rather than their extents?
> > For example, the curve (0.0,0.0)-(0.5,5.0)-(1.0,0.0) has a bounding
> > box of (0.0,0.0)-(1.0,2.5), but freetype2 looks as if it uses
> > (0.0,0.0)-(1.0,5.0).
> 
> Yes.  Currently, FreeType returns CBoxes only (i.e., a bounding box
> determined by its control points which always guarantees full
> inclusion of the glyph).
> 
> The code for computing exact bounding boxes is missing (cf. ftbbox.h).
> It has been removed during heavy reorganization and rewriting of
> FreeType to become more modular, and it hasn't been added again.
> 
> I can't remember why David has omitted the code...
> 
Because nobody seemed to care about it, I'll try to look at the CVS
Attic to ressurect it (they should work unmodified..)

- David