(j3.2006) binding labels and global identifiers

[Van Snyder]

On Wed, 2011-06-22 at 12:11 -0700, Craig Dedo wrote:
> This experience leads me to wonder if the habitual Unix convention really needs to
> be continued.  How difficult and costly would it be for the Unix processors and linkers to
> change their behavior so that they do not change global identifiers?

There is no problem for linkers, because they just take what they get.
They don't change global identifiers.

For global entities with BIND(C) and without name=, the standard
specifies that the binding label is the same as the entity's Fortran
name, in lower case.  Nothing there about appending some number of
underscores before and/or after the name.  So processors already have to
be able to do this.

Stu Feldman is a bright guy, but his pronouncements don't deserve the
status of those from, say, Moses.