TODO

*  Finish the drawing interface ---  currently draw workds with a variety of objects 
   and complete functionality is available.  The drawing interface tries to tame the raw 
   power of the drawing commands, by separating the graphic context attribute setting from 
   the drawing commands using the drawinfo object.  The file mindraw.py contains a list of the  
   methods that this drawinfo object needs.  These methods just take parameters and produce 
   attach a string to the prim method of the object (in C).  This string will be fed to the 
   ImageMagick engine when the drawinfo object is delivered to the draw command. 

*  Write a generalized slicing interface to Images (the methods are there, but is there
   a sensible way to think about slicing images?)  This may be more effort than it's worth.

DONE:   Add Annotate functionality aside from general drawing (making use of the special format characters)
       
*  How do clip-path's work? 
   It looks like clip-path's can all be done within the drawing primitives.
   Or, we could Set an Image Attribute with the name of a clipping path and then use 
   DrawClipPath with that name.
   
   The easiest thing to do may be to use the drawing info context object to build up a string
   of primitives and then have an image method called clip_path that takes a drawinfo object
   as input, and uses it to set the image attribute and then calls clip_path. 

   The clip_mask then gets set. 
     
started:  Write image-magick backend for kiva --- to generate an mimage.
     Not sure how fast this will be, but it can be complete and relatively 
     similar to the display postscript model.   
     This could be used to support formats that ImageMagick supports that the PIL doesn't. 
     It will also help to clarify further the kiva core.