pdp11-g77

[Paul Koning]

I'm starting to do some pdp11 cleanup...

In the process, I ran into a g77 assumption (apparently required by
the fortran standard) that "float" and "long" are the same size.  Or
something along those lines...

The pdp11 back end by default has float the same size as double: 64
bits.  That's selectable, the alternative is float = 32 bits.

I'm not sure why this option is there much less why the default is
that way.  In any case, I can see some ways to fix this:

1. Change the default to -mfloat32.
2. Take out the selectable size and make float 32 bits, which seems
   like it would better match expectations anyway.
3. Have -mfloat32 supplied automatically by fortran compiles.

For (3) I assume that's one of those magic xxx_SPEC strings... but I
don't see an xxx_SPEC for Fortran in the docs, only for C, C++, Link,
etc. 

Suggestions?  Right now, (2) is tempting as the least-effort fix...

      paul