Re: Errors in a Microsoft Propaganda Book

["Michael J. Mahon" <mjmahon@aol.com>]

Charlie Springer wrote:
> On Sun, 22 Apr 2007 00:38:25 -0700, Michael J. Mahon wrote
> (in article <E_udnfwlLtSbjLbbnZ2dnUVZ_gSdnZ2d@comcast.com>):

> > Like Sutherland with the TX-2, PARC had chosen to use relatively fast
> > (home-made) minicomputers to support interactive graphics, which gave
> > them enough horsepower to support a practical GUI.  In fact, didn't
> > PARC "invent" BITBLT, the fundamental raster GUI operation (Sketchpad
> > was vector graphics-based)?
>
>
> I used Vector General displays at Boeing with VAX's in 1971. I loved VG 
> displays at the time for their 4096x4096 resolution. Talk about a great 
> platform for Battlezone! Too bad it hadn't been created yet.
>
> When I was back for a contract in about 198? the graphics room had an Alto 
> and a Lisa side by side.

Vector graphics required expensive displays for precision (magnetic
deflection is hard to make both fast and precise), so when memory
began to get cheaper, bit-mapped raster graphics pretty much took
over everything.  I think "Asteroids" was the last vector game to
vanish...appropriate, since it was also the first to arrive.  ;-)

> > I'd like to see Jef's thesis to see what he was thinking about.
> > He must have been familiar with Sutherland's work--I certainly
> > was in 1964, and I had just started writing interactive graphics
> > programs (vector-based, at the time).  (I probably would have done
> > *my* thesis on interactive graphics if my advisor hadn't had other
> > obsessions.  ;-)
>
>
> I don't know where to get a copy.  They may have it in his archive at 
> Stanford.

I'll give that some thought...

> > In 1965 I had written two interactive graphics programs for analyzing
> > nuclear physics data, using a light pen and special keyboard.  Its
> > primary value was in using the light pen to select data on the display,
> > and it was obvious that operations on the data could be selected by
> > menus on the display, too, as well as for entering numbers (point at
> > the upper half of a digit and it counts up, lower half, it counts down).
> > The keyboard became vestigial.
> >
> > In those days, the display was refreshed by the computer, not by a
> > frame buffer, so it was not only "natural" to make the display respond
> > to the light pen, so that things changed when the pen was pointed at
> > them, it was almost hard *not* to use the display to provide visual
> > feedback!  Menus that "flew out" (or "dropped down") were, in fact,
> > the *easiest* way to do things, since screen real estate had to be
> > conserved.
>
>
> The VG had a light pen but it wasn't used much. Everything was assembly or 
> FORTRAN. I wrote what amounted to a BASIC interpreter in FORTRAN in order to 
> do interactive debugging and testing of a huge program.

Most of industry was non-interactive, since using a machine in an
interactive mode rendered it a very expensive "personal computer."

At Caltech, I had the advantage of using an older machine (a Burroughs
220, the last commercial vacuum tube computer!) that was then a machine
you "signed-up" for, in blocks of an hour.  It was my second personal
computer (the first was an IBM 1620, after quitting time ;-).

I was already a big interactive user of the 220 (it had a great "organ
console" front panel--the original Batman had a 220 front panel in the
Bat Cave, where the original neon lights had been replaced by brighter
incandescents for TV.  Sitting in front of this panel, it was easy to
have a program stop, displaying a distinctive pattern in the lights,
so you could enter data directly into the registers, then resume
execution.  Interactivity was both simple and effective.

When a 1024x1024 vector point display was installed on the machine to
support some research project, it was quite easy to move from the
machine's console to the display, and we quickly hooked up a light pen
(to interrupt the display list) and a keyboard (also interrupt-driven).
This required modifying the machine's instruction decoding, so that a
special variant of CAD (Clear and ADd) would start a sequential upward
scan of memory addresses, starting with its operand address, and ending
when either a word fetched from memory had a particular bit on or when
the light pen "saw" a dot displayed.  Since the 220 was a decimal
machine, the display actually was 1000x1000!

One of the first things I did was write a character generator routine
that converted alphanumeric text into a display list of dots.  There
were lots of unused bits in the display list words, so there was plenty
of room to "tag" words so that software could do appropriate things
when the light pen caused the CAD to complete on the sensed display list
word.

Once you had seen the displayed data respond in real time to gestures
with the light pen, every observer was "hooked" on interactive graphics!

A few months later, the display was moved to a conference room one floor
and a few hundred feet away from the 220 console, but still driven by
the 220.  By that time, I hardly needed to use the 220 console, since
the suite of programs was controlled entirely from the display.  But it
was a little disconcerting not to be able to instantly see if the 220
was working, stopped, or in a tight loop, so I brought a cheap AM radio
in and put it by the display.  Since the display was driven from the
memory bus of the 220, all the usual noises were audible, and the
charactistic beeps and warbles of program execution provided instant
feedback on how things were going even when the display wasn't changing.

I've considered writing a 220 + graphics display emulator just to run
some of these programs that changed my career--from nuclear physics
to computer science.  ;-)

-michael

NadaNet networking for Apple II computers!
Home page:  http://members.aol.com/MJMahon/

"The wastebasket is our most important design
tool--and it's seriously underused."