KINDs

[Paul Brook]

> > As a second question, are you guys going to abandon the f2c calling
> > convention? See the g77 info pages for a description of -fno-f2c to
> > see what I mean. We decided to follow the AMD64 ABI and use a flag for
> > back-compatibility for individual functions. It's unfortunate that
> > this ancient ABI is used on even new platforms with g77.
>
> At one time gfortran was going to have a compiler flag to support the
> 77 calling convention.  Paul Brook, I believe, removed all the
> brokeness in http://gcc.gnu.org/ml/fortran/2004-04/msg00043.html.

Well, that flag never worked properly anyway. That particular patch didn't 
have any effect on compiler operation, it just removed some user-visible 
remnants.

I guess we'll probably want to add f2c compatibility in some form.

> One thing I hate about the CVS commit messages in the GCC tree
> is that the messages actually don't tell you why a change was
> made and the message is often too terse.

This is GCC policy. The idea is that the cvs log contais a brief summary of 
what changes were made. You then cross reference the mailing list to find out 
why.

Paul