Fri, 4 Jan 2002 11:45:59 -0000
> What's wrong with it? I would probably adopt a similar approach.
> My own thoughts have been to use "jay" for the Python grammar
> and obtain a parse tree of the Python code to be compiled.
> (All in C# btw). Walk the parse tree and use System.Reflection.Emit
> to create the byte code.
Well, I'm a big big fan of Python, its a wonderful prototyping tool, and
good for a wide range of web development. I am however dubious abouts it
suitability as a language in which to produce a compiler (one concern raised
by ActiveState was the slow compile times).
> I think that the real problems are to do with mapping the
> semantics of Python to .NET;
I dont think this is the case (although I could well be wrong), as there
have been no problems mapping Python semantics to Java (no problems in that
it's been done, I have no idea if it was easy to do or not), as we can see
I'd underline the fact that we have a Java/Python implimentation that works
well and is "adult". I don't see the big difference between C#/.NET and
Java/VM that would suddenly throw the whole thing into the realm of
"problematic" which is at the end of the day where ActiveState placed
Python.NET. To paraphrase "harder than we thought, we were going to have a
crack at it but decided its not worth the resource as its running like a
> this I can't yet comment
> on, I haven't researched it enough. I was wondering about
> doing my research by hacking MonoPython and seeing exactly
> why it won't work :-)
Well if it wont work one has to ask why will Java/Python work and not
.NET/Python? If there's a genuine reason then its an area of concern.