Yeah. "Everybody was doing it" with preprocessors, but they're a pain. And
with no consensus: I co-authored a report listing about 30 "Structured
Fortran" preprocessors - all different and mostly freely available, and each
implementor was sure his preprocessor input syntax was the right one. Jet
Propulsion Lab (Chuck Lawson) hosted a Fortran Preprocessor conference in
Pasadena, where more or less the same set of proposals was presented.
This was right in the middle of the "structured programming revolution" and
"GO TO considered harmful" etc. Control structure ideas were changing too
fast for (conservatively interpreted) Language Standards procedures to keep
up.
A bunch of us on X3J3 predicted that pressure on vendors would soon induce
them to put in some kind of "branch" that didn't depend on GO TO or labels,
and we said let's try to get everyone to do it the same way.
As Walt points out, the existence of so many preprocessors convinced some of
the other Committee members that we WERE trying to "standardize existing
practice" - i.e., the preprocessor input languages were in some sense
"valid" Fortran variants, even though most were not produced by the usual
vendors.
= Loren
-----Original Message-----
From: ... Walt Brainerd
Sent: Saturday, January 08, 2005 10:04 AM
..
Subject: Re: Fortran bashing in ACM Queue magazine
Walt Brainerd wrote:
..
I should have added that the point Van makes (with bad history) is correct.
The block IF just barely made it into F77.
It was put in after the public review.
Many members of X3J3 were opposed to it because it was not "standard
practice".
One of the convincing arguments for it was that many, many preprocessors
(such as Ratfor mentioned by Van)
implemented something like it.
|