[J3] [EXTERNAL] Re: END statement

[Van Snyder]

On Tue, 2022-12-06 at 09:54 -0700, Keith Bierman via J3 wrote:
> 
> 
> On Tue, Dec 6, 2022 at 9:45 AM Clune, Thomas L. (GSFC-6101)
> <thomas.l.clune at nasa.gov> wrote:
> > Ah – thank you for returning the favor and making me feel young!☺
> > 
> 
> 
> HTH. I was disappointed that google wouldn't find an online copy of
> the TFOR manual (the CHM has the citation, so it has one in its
> archive). I happened to have one on a shelf  in my basement ;> I
> didn't have an EFL manual handy, but memory says it allowed ";" as a
> statement separator as well (its source is probably online
> somewhere). 
> 
> One of TFOR's principal authors, Jack Perrine left JPL and became an
> Astrologer, and as I understand it, Nancy Regan was one of his
> clients.

I might still have my TFOR manual. Or maybe I gave it to CHM.

Jack always regulated his life according to his horoscope. It told him
to buy silver at $50/oz, when Nelson and Bunker Hunt were trying to
corner the market. He lost his shirt (and his underwear, and his
toothbrush, and his shoelaces), so his little company, Athena, went
bust and he came to work at JPL. He was in charge of all the compilers,
mostly from Univac. He kept the PI/1 compiler going for Dan Alderson
because the SEQGEN program, that generated command sequences for
Voyager, was in PL/1.

All of his astrology software was in FORTH, so he converted Fred
Krogh's differential-equations solver, that is used in all the
navigation software, to FORTH. He (and the navigation software team)
especially liked that integrator because it handles second-order
equations directly.

He gave me the listings for TFOR, and I gave them to Paul McJones at
CHM. Paul had briefly worked for Jack at Athena, but Jack didn't
remember him. Tom Lahey and Jack had worked for the same company
(Digitek?), but Jack also didn't remember working with Tom. I think one
of the principals at that company founded Ryan-McFarland.

In addition to allowing semicolon as a statement separator, TFOR had
internal subprograms, a macro system, and the at-sign as a comment
introduction. One could also define how procedure calls were compiled,
so that it was possible to write ER (Executive Return) calls to Exec
directly instead of needing to write wrappers in assembler. One of my
colleagues, Dirk Feild, added block-IF and DO constructs using the
macro system. He called it FORCE, for FORtran Control Extended. That
wasn't portable to other machines, so John Flynn wrote SFtran, a
structured-fortran preprocessor. Walt Brainerd told me that the SFtran
control structures influenced Fortran 90, although the principles
involved were pretty obvious, and similar to RATFOR.

Both TFOR and the Univac compilers treated statement functions as
macros, so that a reference to one could appear anywhere that its body
could appear -- in particular in variable-definition contexts. I used
this to fake structures. It's the reason I proposed the C(S) notation
for structure component references at the Albuquerque meeting in 1986,
while there was still controversy over the percent notation.


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mailman.j3-fortran.org/pipermail/j3/attachments/20221206/9b51e54c/attachment.htm>