Language change and code breaks

[James Logajan]

Grant Edwards wrote:
> And (at least in the Pascal compiler) identifiers were limited
> to 6 characters so that an entire identifier would fit into a
> single word -- so two identifiers could be compared in a single
> instruction.

I too fondly remember the CDC Cyber series. They didn't have a traditional
stack. As I recall, a blank memory space was left at the beginning of each
subroutine where a single return address could be stored. This meant you
couldn't do recursion using the standard assembly language subroutine call
mechanism. I forget how Pascal got around this. I do recall them switching
from 6 bit characters to an optional 12 bit characters to support lowercase
a few years after I encountered Kronos.

> Yikes.  It was an interesting architecture.  One of my
> assignments in a class once upon a time was to write a 6600
> instruction set simulator in 6600 assembly language.  It had a
> small, regular instruction set and it was surprisingly easy...

It has been soooo long that I may misremember, but wasn't it pretty much
automatic to set the r1 register to 1 at the top of an assembly language
program? And to clear r0? Or was it the other way around?

> Doing the same thing for a Pentium would be pretty horrific.

While I am one of the grunting barbarians who doesn't wish to see progress
roll over a lot of legacy code, I suspect any ugliness of the Pentium is due
to the attempts by Intel to retain some form of backward compatibility at
least as far back as the 8008. But it has been so long since I did any
assembly language programming that that may no longer hold true.