>I have no experience with Python, but my reading of the article suggests compiling an entire program, not single or partial lines of code as VFP does with macro expansion.

Craig,

My point was not whether compilation of large or small units of code was faster than in the C implementation, or vice versa. Rather, the point is that it can be done, and it will have performance that is above "acceptable".

Python, like VFP, is a highly dynamic language, with constructs like eval() etc, and this, in my opinion, goes to show that Microsoft did not thouroughly research the possibility of making a VFP.NET that contains everything we know and love, before dismissing it.

Furthermore, the article in questions also states that "The Whidbey release of .NET is expected to include support for "light-weight code generation" that could offer significant performance boosts to the performance of eval with minimal changes to IronPython's implementation. ", which would also be beneficial for a VFP implementation.

Perhaps creating such a compiler is a task for the UT community? <g>

Eyvind.