Re: parse tree --> running program/script...

Herwig Huener wrote:
> In your case, I would do both, 1 first. You will have a kind
> of a tree anyway if you generate code for a virtual
> machine, an then, having both implementations, you can
> check them against each other when doing some error
> diagnosis.
>
> The kind of virtual machine and it's design depends on
> what your language is best suited to, I guess.


I've been thinking about this... The things that this language will do
will not make it easy to write any kind of (not too slow) machine
code.


ie: dynamic arrays, 'eval's, dynamic variables (the second one is seen
it is valid)


hopefully i can come up with something. :)


thanks,
Tom Fjellstrom

Return to the comp.compilers page.
Search the comp.compilers archives again.