[antlr-interest] Separate compilation units for grammars (C++)

[Bryan Ewbank]

All,

I've been exploring how to split a ANTLR-based compiler (v2.x; C++;
one parser, N tree walkers) into different libraries.  It looks as if
the input token set will be a problem for most systems, since it
appears to be required in the same directory as the *.g processed by
ANTLR.

Has anyone made this kind of front-end/back-end split?  Any pointers?

Oh.  I'm living in ANTLR v2.

Thanks,
- Bryan Ewbank