>IMO, it's the lack of competing implementations. I beg to differ in this point. There are other implementations, but they are not called "python" and they are not a 100% python in syntax and features. For example, Boo is 95% python syntax wise, but statically typed. This fundamental difference makes it as fast as C# or any other .NET (or mono) language. Being statically typed doesn't mean that you have to declare types everywhere, like in C. It uses type inference, so you can declare a variable x= 5 and the compiler will know that x is an integer of value 5. Pyrex is statically typed too, but it's used mainly as an extension language for Cpython.
Now talking specifically about python, there are projects aimed at speeding it up substantially: Pypy is a project that relies heavily in type inference (for translation to lower level code) and dynamic optimization. It's based mainly on psyco, which has already probed that it can massively speed up python code. Shed-Skin: it's a pyton-to-c++ compiler. It won't support the most dynamic features of python, and requires the programmer to restric a little bit his coding style in order to allow static compilation, but so far it looks great (only one developer though..). -- http://mail.python.org/mailman/listinfo/python-list
