>Len,
>
>>> One or other of us may be missing the point (it may be me) - but even in the
>>> description you gave, there was nothing to contradict & a lot to confirm that
>>> IL is "a proprietry representation of the symbol tables and tokens in a
>>> manner optimised for the back-end executor" - the back-end executor just
>>> happens to be a JIT compiler. There is no real reason why a JIT compiler
>>> cannot be produced for p-code either.
>
>The point you are missing is that IL isn't or doesn't contain the symbol tables you speak of. The compiler (C# or VB) will create symbol tables, parse trees, tokens etc. internally as part of its compilation process, however the output from the backend is meta-data and IL. The previous IL example source code shows the variables being accessed through a stack, pretty much like x86 assembly language.

OK, I give in, I'm being particularly dense today.

(The p-code I was used to was stack based too, so I still don't see where the real difference is, it just seems that it's just a matter of internal representation rather than a fundamental difference, as Anders Hejlsberg said. All of the same information must be there, just in a different form)