[Info-vax] Fortran

[abrsvc]

On Wednesday, December 5, 2018 at 4:32:39 AM UTC-5, Dave Froble wrote:
> On 12/5/2018 3:26 AM, Simon Clubley wrote:
> > On 2018-12-04, Dave Froble <davef at tsoft-inc.com> wrote:
> >> On 12/4/2018 9:25 AM, abrsvc wrote:
> >>>
> >>> The removal of Hollerith constants is ongoing in anticipation of the loss of support for those in future compilers.
> >>>
> >
> > That code must be multiple decades old. I know it's a pain, especially
> > when certification is involved, but sometimes code needs to be
> > updated in light of more modern standards and safety requirements.
> >
> > In some ways, it's no different than needing to update VAX C
> > code to work with modern C compilers. The updates are required
> > for a reason and the code is better as a result.
> >
> >>
> >> Apparently there are those "bemoaning" the loss ....
> >>
> >> I got to ask, why were they removed?  Was it necessary?
> >>
> >
> > They were removed because it is a concept which results in both
> > hard to read and hard to maintain code and which was replaced by
> > a safer option decades ago.
> >
> > Fortran has had ideas which seemed good at the time but which have
> > not stood the test of time. The other one which comes immediately
> > to mind are common blocks whose definitions are repeated in more
> > than one place.
> >
> > Simon.
> >
> 
> The question remains, was it necessary?
> 
> If the compiler could have continued to support the "hard to read and 
> maintain code", and there was no reason to remove the capability, then 
> that should not have been done.
> 
> Causing others to do some work, which would not have been necessary, is 
> not a good thing.  Causing others to re-write usable code, because some 
> don't like it, is arrogance carried to the extreme.
> 
> -- 
> David Froble                       Tel: 724-529-0450
> Dave Froble Enterprises, Inc.      E-Mail: davef at tsoft-inc.com
> DFE Ultralights, Inc.
> 170 Grimplin Road
> Vanderbilt, PA  15486

The one that caused the most pain was the loss of the octal constant support.  While trivial to change, the decades old modules that continue to provide the functions required had to be changed only to specify the value in a different way.  Why?  It seems to me that supporting this was not a major problem.  It is water under the bridge now of course, but many man-hours were spent to make this change.

As far as Hollerith goes:  As much of a pain it is to change this, I agree that these are often confusing.  I know of programs that would fail because of incorrect counts.  Easy to make a mistake here without knowing it.  This is one that I believe is a change for the better.
Dan