Re: VBA : shrinking a multidimensional array to 'nothing' and restarting again...


Gerry Quinn wrote:
> In article <1112354886.910021.195270@xxxxxxxxxxxxxxxxxxxxxxxxxxxx>,
> spinoza1111@xxxxxxxxx says...
> > Gerry Quinn wrote:
>
> > I need to follow up on this. Like I said, the incident happened in
> > 1992. But more than one language has provided integer pow without
> > providing either negative powers or noninteger arguments. By far
the
> > most common use of exponentiation is probably integer and in
> > determining limits given word size.
>
> That language doesn't sound like C. Besides, any such function would

> have been documented (or how would Nash have known of it?). The
> documentation would have stated that it returned an int, not a 64-bit

> integer.

Not necessarily. We're talking about Microsoft.
>
> > > > At the time, I was using C frequently and my favorite compiler
was
> > > > Borland. I probably in reality saw the use of pow(), guessed it
was
> > > > an integer hack, and directed Nash to Borland.
> > > >
> > > > This revision of the reconstructed story preserves the main
point:
> > > > that Nash knew the C standard and the Microsoft developers
either did
> > > > not, or else did not care.
>
> There is no integer pow() in the C standard.

We're talking about what Microsoft does, not the C standard. Microsoft
is famous for ignoring standards.

>
> > > The main point must be considered in extreme doubt, to put it
mildly.
> > > We are told that a mathematician, known to be brilliant but
unstable,
> > > and who had not studied C to any degree, had difficulty with a
> > > certain function not operating as he expected in one compiler.
We are asked
> > > to
>
> > His expectation was correct (if my memory serves). He wanted to
> > calculate 2 to the 31st power and decrement and the C standard
library
> > provides the function as you so helpfully pointed out. Therefore,
IF he
> > used pow() at all (which is now in doubt since I reconstructed the
> > story incorrectly in 1995 when Nasar interviewed me, although not
its
> > overall import) he used it right, the Microsoft compiler betrayed
him,
> > your hero here, experienced in Microsoft betrayal and constant
> > evaluation, steered him to Borland, and end of story.
>
> He didn't use it right, because it returns a floating point value
that
> needs to be explicitly converted to the integer type of his choice.
>
OK, suppose he casted (int)(pow). Then the bug was obvious to me. But I
need to do more research to explain how the Borland compiler worked
without any change to the code.

> > No. Despite my own false reconstruction of the actual notation
used,
> > I'll stand by the overall thrust of my story which was you could
not
> > trust Microsoft at the time to get it right. I need to download if
> > possible a copy of the old compiler and confirm my hunch, that it
> > quietly truncated its arguments to long int (32 bit at the time).
>
> You'll need to show more than that to demonstrate anything other than

> that a guy who either didn't know much about the programming language
he
> was using - or didn't care - made a typical blunder, the details of
> which you can't remember. Nothing mystical about it.
>
I'd rather you flame me for false memory syndrome and creatively
writing a story, complete with my usual profound (and still true)
observations on life, the universe, and everything. I'd rather you NOT
attack Nash's programming skilz.

That way, I could fight one of my celebrated "flame" wars and hammer
you down, again.

That's because I recall in general that Nash's code was great and I am
on record in Sylvia's book as saying so, and also that when he used the
Borland compiler, without to my recollection changing the code, this
code worked. That's because I could see from the code alone (and
confirmed later by reading his Rand papers) that Nash knew all about
programming per se and did not make elementary blunders, as do mere
mortals like me.

Fortunately, and probably because it did not check with Nash's own
recollection, Sylvia Nasar did not print my false memory, nor did I
retail the story in my own book.

I would like to tell you that Nash welcomed me into one of his study
groups and that I suggested a solution to the Riemann Hypothesis. I
would also like to tell ya that I played quarterback for Princeton at
the age of 37 and won the 1987 game against Yale. But even Scottie
Fitzgerald remembered in an alcoholic daze that he did not graduate and
I follow his example.

But in fact, I was only honored to assist him a couple of times, and at
the time in question, I solved a problem he was having. I have always
been careful to qualify my association with Princeton as being an
employee and only a very part time student, and to qualify my
association with Nash.

This is because I am all too well aware that what Walter Benjamin
called the aura attaches to institutions like the Vatican, the British
monarchy and Princeton. This aura today not only attracts a fearsome
first-order crowd of phonies and cranks, it also attracts a global
suspicion that one is being had, which produces epistemological
deadlock.

You have given an excellent demonstration of how mere programming can
break this deadlock by demonstrating that my original story MUST BE
FALSE. In part because I deconstructed nonsense at my own university by
debugging a Fortran compiler in object code, I acknowledge that my
original story MUST BE FALSE.

But lay off Nash.

I fucked up when I told Sylvia the wrong story because in 1995 I was a
VB 3 addict. I remain obliged to you for deconstructing my story but
I'll stand by the grand outline and the moral lessons I've drawn.
> - Gerry Quinn

.