From: Samantha Atkins (samantha@objectent.com)
Date: Mon Apr 16 2001 - 04:10:18 MDT
KPJ wrote:
>
> |that a friend helped me hack an old iron ferrite memory core to. Then a
> |Trash 80, some early integrated system with the huge old actually floppy
> |floppy drive. Then a job on ancient CDC hardware hacking Fortran.
>
> Control Data Compass, CD-8090, or something else?
Compass. IIRC 7600 (or was it 7800?), 176 series and one of their
supers (Star?).
>
> |Many years before all that I was pulled into this elite kid computer
> |training thing on some antique IBM iron. Input was only punched cards.
> |The sick bastards only taught us Cobol! And the turn-around on a deck
> |averaged 2 days. I didn't want anything to with computers after that
> |until the first micro computer chips became available.
>
> Well it depends.. I wrote a disassembler fos OS/360 in COBOL. The COBOL-68
> had excellent file, buffer, and data control. I wrote the Ackermann function
> in COBOL (had to add a stack in software, since PERFORM stacks could only
> handle a stack depth of 9 addresses), and first used LISP (INTERLISP) through
> punched cards.
>
Not that you can't write interesting stuff in Cobol. Just that it
doesn't help you much in doing so. It was sick becuase the way they
picked us was to look for math whiz kids. You don't teach math people
Cobol. "Add 1 to X"?
Barf city.
As a "technical assistant" (cheap programmer) for an oil company in the
early 80s I created multi-dimensional spreadsheets for evaluating
reservoir simulation runs (as well as various sets of field data) from
their supercomputer. (this was 1983. too bad I didn't know what I had in
these spreadsheets on steroids). The supercomputer results came out in
a long serial mess on tape from which huge printouts were produced. It
was sad seeing the Reservior Engineering folks attempting to manipulated
this data using calculators because writing Fortran programs was too
tedious and error prone against such a data format (as well as too
slow). So I developed this spreadsheet on steroids that allowed them to
query all the field well data by x,y,z and time coordinates. I
invented a fairly natural language for expressing formulas and variables
against the data that would recompute if any dependent variables or
formulas changed, used an inverted file structure to transform the data
from tape to disk and wrote a fairly sophisticated LALR parser for the
language by hand. This was all in Fortran 4, on a 128K fixed memory
machine with 60 bit words.
So yeah, very interesting things can be done even with relatively
primitive hardware. But it isn't very intuitive or time-efficient.
> The high tech of yesteryear become the junk of tomorrow. For some time, I had
> 27 SUN-3's stacked in my living room. Today I mostly have SPARCs, DECs, and
> obsolescent PCs. The Singularity cometh...
<grin> I know the dance. Four tech goddesses in my house. Twenty
computers in use. That many again in various stages of disrepair or
just not being plugged in at the moment. Some old SGI hardware that was
considered tremendously powerful in its day. A hacked up Next Box, a few
Apple machines, a couple of Sparcs, at least one AIX, a top-end laptop
from 1994 (now nearly useless) and even an antique PS/2.
A lot has happened in hardware in the last couple of decades.
Unfortunately, not nearly as much has changed in software.
- samantha
This archive was generated by hypermail 2.1.5 : Sat Nov 02 2002 - 08:06:59 MST