Evaluate performance / quality of using native text engines for any or all of the following tasks:
    - Glyph generation
    - layout
    - drawing

I tend to think that the first two will make sense (i.e.: we will have as good or better performance / quality by using native support for glyph generation / layout), but that when it comes to actually drawing, we are better off doing it as we do today (all the literature and advice we've had from mobile graphics guys working with PowerVR indicate that this should be the case).