Python and COBOL

[Tim Roberts]

"John W. Baxter" <jwbnews at scandaroon.com> wrote:
>
>(The CDC Fortran was wonderous to behold.  In one test, a decent Fortran 
>programmer and a top-drawer assember programmer were assigned the same 
>non-trivial program to write.  The Fortran was done in under a month.  
>The assember was done in two months...then the assembly-language guy 
>spent about 4 more months tuning until he managed to get the performance 
>to very slightly exceed the performance of the Fortran version, which 
>hadn't been touched.)

We're travelling rather far afield from Python here, but you're taking me
on quite a journey back into the past.  I spent ten years working for
Control Data during their heyday, back when we were able to timeshare 64
users on a 1 MHz machine with 98k of memory, which required water cooling,
occupied 100 square feet of floor space, and cost $4 million.

Yes, the CDC Fortran compiler was a wondrous thing.  Many were the
assembler programmers who were humbled by the code its optimizer generated.
One of the advantages of having 24 registers is that you can salt many
useful things away for later, without resorting to expensive core memory
operations.

If only Seymour had chosen 8-bit bytes instead of 6-bit, the architecture
might have lasted even longer.
--
- Tim Roberts, timr at probo.com
  Providenza & Boekelheide, Inc.