(j3.2006) Question about DIM
Tue Jan 12 20:25:11 EST 2016
On Tue, 2016-01-12 at 16:58 -0700, Keith Bierman wrote:
> Only Van is likely to appreciate my discovery of the "Athena" Fortran
> compiler, which offered a lot of features beyond the standard. I don't
> recall (but have the manual somewhere) precisely what the benefits
> where (better control structures, a multi-line statement function
Athena Fortran also had
* a powerful macro system, which Dirk Feild used to implement
Fortran 77 control structures without changing the compiler,
* internal subprograms (not a multi-line statement function
* as with the Univac Fortran system, statement functions were
macros instead of functions, so references to them could appear
as the LHS in an assignment statement if their RHS would be
allowed -- one of the few examples of updaters (which are very
powerful, but I've not seen them in other than Univac Fortran
* a method to specify how procedure invocations were done, so you
didn't need to write a wrapper to call the system directly (long
before C interop),
* character concatenation syntax of variables could appear on the
LHS, providing a powerful scatter, ...
and many other facilities that Fortran 2015 won't have -- but no
recursion or derived types. I faked limited forms of derived types
using statement functions; their updater semantics were indispensible.
The mutability of programs written using this uniform syntax is what
inspired me to urge this on X3J3 at the Albuquerque meeting in 1986.
Brian Smith, John Reid, and Rex Page had figured out what I wanted, how
useful it would be, and why I wanted it. The other 30+ members said
"Huh?" One of the members (Kurt?) said "Fortran programmers want to see
what their program is doing." I mentioned that David Parnas, Geschke
and Mitchell, Donald Ross, Yourdon and Constantine, Jensen and Tonies,
and other authors had described quite clearly a decade earlier that this
is exactly what you do NOT want if you want to reduce maintenance cost,
but it didn't carry much weight.
> Years later when working at the Lab I recall running into an ugly
> issue with the TFOR namelist handling and had to look up the author
> (who had retired from the Lab to be a freelance astrologer to the
> stars (and Nancy Regan)). I'd stumbled across an internal error in
> namelist handling .... which started my long road to ruining
> compilation systems by finding flaws ;>
Jack Perrine is still alive, although very frail (or at least none of
our mutual friends has told me he died), living in Pasadena with his pet
gray wolf Djinn. A few years ago I talked him into giving the Athena
Fortran listings to the Computer History Museum. Paul McJones
(paul at mcjones.org) would know exactly where they are.
At JPL, Fred Krogh developed an exceedingly good code to integrate the
initial-value problem for ordinary differential equations, called DIVA.
It can handle second-order systems directly, instead of requiring them
to be converted to first-order systems. This is important for celestial
mechanics. It's available as part of our Math77 library, which Caltech
finally (after 25 years) gave us permission to publish, from netlib.
For some weird reason, Jack converted DIVA to FORTH. BTW, Phil Sharp,
who visited J3 for one meeting, is working very closely with Fred on a
newer ODE code.
Many people worked with Jack at Digitek in the 1950's, including Tom
Lahey, and maybe Don Ryan (of Ryan-MacFarland). Paul McJones worked for
Jack at Athena Computing. Jack doesn't remember any of them.
Jack was once nattering on about astrology and I told him "I don't put
much stock in that, but I'm told that's typical of a Pisces." I knew my
remark had just zoomed right over his head when he said "Oh, yes, that's
very true! Pisces are very skeptical!"
The reason Jack come to work at JPL, instead of continuing Athena
Computing, was that his horoscope had told him to buy silver when Nelson
and Bunker Hunt were trying to corner the market. When the bubble
burst, Jack lost his shirt (and his pants, and his toothbrush, and his
shoelaces, and ...).
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the J3