Philippe Fremy wrote:

>
>     Hi,
>
> I would like to implement something similar to the type inference of 
> ocaml for the python language. I have always found it very impressive 
> (although I have only used caml light).
>
> I have no experience with the topic, it is just a project that seems 
> cool to me :-)

Yeah me too :)

Checkout "Demand-Driven Type Inference with Subgoal Pruning: Trading 
Precision for Scalability." by Alexander Spoon. Google can finds the 
paper if you type in the title. This paper gives a nice description of 
why type inference is hard thing to do in a dynamicly typed OO language 
like Python (aka smalltalk) and gives pointers to some work in the area 
including Cecile and Squeak. There was a later paper published by the 
same author in 04 but its not on the web yet :(

I spent a little time considering adding interfaces to Ruby 
[http://kvo.itee.uq.edu.au/twiki/bin/view/Main/RjBlog39] which is pretty 
similar to Python, I guess. The main motivation for me, in adding 
interfaces to a language like Ruby, is to make the software structure 
explicit, and to give coders a place to document contracts that are 
shared by a number of classes. So I don't mind if the type inference 
isn't complete in the sense that every variable gets assigned some sort 
of minimal type.

I'm sure one of the Ocaml guys can give you a list of references to the 
ML type system. Is there a canonical reference of the Ocaml type system?

regards,

Richard.