[Patch, fortran] PR30746 - 50th Anniversary Bug - Forward reference to contained function

[Paul Richard Thomas]

FX,

> It looks OK to me, and certainly very well explained (I guess without
> the explanation, I wouldn't have understood what difference it really
> made). I'd prefer that you wait 24 hours to give other people time to
> comment or suggest changes, though.

OK on the time delay and thanks for the review.

It offended me that (i) gfortran failed what is presented as a
benchmark for a good compiler and (ii) that there are still failures
of symbol association.  In fact, although I have never come across it,
I can imagine occasions where this structure might arise;
particularly, where separately developed procedures that use associate
a module are then absorbed into it.  Double containment is certainly a
very good way of keeping symbols private!

Cheers

Paul