Hi Tim,

> > since it affects a file included directly by interp.c.  I'll send
> > something to TPR for him to include in VMM that will allow things like
> > this to be detected by the header files when necessary.)
> Perhaps a generated .h file would help here.

The places you need to know about this are mostly local to interp.c, so
generating a .h (while being the most general solution) is total overkill.  
I'll send you a one-method cs that emits a #define at the top of interp.c
to tell included files whether the global struct has been declared and/or
defined.

(The CCGen already takes care of defining accessors for every variable you
need to reference from external code.)

> it might be good to generate a makefile fragment as well/ instead.

I don't understand what you think would be the utility in that.

The compiler/linker couldn't care less how globals are arranged in the
source files; it's things like sqGnu.h that need to know about this, and
they're sucked right into interp.c (so no information about foo/fum
actually needs to escape from this file at all).

Cheers,
Ian