I am interested in porting the protovis javascript visualization library to python for use in scientific computing. I have a general question and a specific one.
General: I have never ported a whole library before. What are some good strategies? Should I first just implement the user-facing API and then fill it in in a pythonic fashion? Or is it better to try to port over the internals/infrastructure first, and then rebuild the API on top of it?
Specific: This library appears to make heavy use of javascript's prototype-based inheritance, which is somewhat different from the python model. I found someone who made a pretty simple method to emulate prototypal-inheritance in python. However, from the perspective of porting a library, I'd rather not arm-twist python to be more like javascript. Any feedback on this issue would be greatly appreciated. Thanks!
Uri