Re: g77 legacy features in gfortran

[Tim Prince <tprince at computer dot org>]

Jack Howarth wrote:
>     We have had a request to update the g77 package to
> provide MacIntel support in the fink packages on Mac OS X.
> I am told that the following issues require the continued
> existence of a g77 package in fink...
>
> 1) support for multiple entry points in subprocedures (sad but necessary
> for legacy code);
> 2) EQUIVALENCE in COMMON;
> 3) stabs or dwarf debugging information for COMMON and EQUIVALENCE
> symbols;
> 4) source input compatibility (tab handling and free-form continuation).
>
> Are this all still accurate for gfortran in gcc trunk? I assume the
> third may be true, but am wondering about the others.
>       Thanks in advance for any clarifications.
>                  Jack
>   
2) In my experience, EQUIVALENCE in COMMON prevents gfortran compilation 
only under -fopenmp (it conflicts with OpenMP standard).  g77 didn't 
support OpenMP at all.
4) I use every combination I can think of, and never saw any 
incompatibility between g77 and gfortran.  Both have strange behaviors 
with respect to fixed form input with source extending beyond column 72 
or wrapped character strings, so I don't count on anything which 
violates normal practice there.  It's certainly conceivable that source 
which depends on non-standard behavior of some particular version of g77 
would break with another version.