(j3.2006) TKR for C Interop

[Michael Ingrassia]

>shall I predict that all compilers that want to run on Linux better have gcc 
>as one of the companions

You can certainly predict that.  That doesn't necessarily make it likely
though.

By way of thought experiment, given that internal subroutines will be
interoperable:  a Fortran compiler might choose to have as companions
only those C processors which can be made to configure function pointers
as closures (internal contents contain pointer to code and pointer to
host environment).  I don't know if gcc does that.  This is a thought
experiment only, not a description of any known compiler past or future.

Yes, it's potentially nicer if you can first fix the pool of desirable
companion C processors and then settle on Fortran compilers which fish in 
that pool.  But that's not how we have written the standard.  A compiler
supporting BIND(C) doesn't promise to bind to your favorite C processor, but
rather to its favorite.

	--Michael I.