🚀 go-pugleaf

"Sarr J. Blumson" <sa...@tetris.gpcc.itd.umich.edu> wrote in message
news:FD2Ea.8$_K.659@news.itd.umich.edu...

> My understanding is that language subsystems for the B1x00s actually
> played with the microcode, so that the architecture was different
> if you were running a FORTRAN program than if you were running a COBOL
> program.

All from memory:

More than that.  Each language (including the language that the operating
system was written in) had an interpreter.  FORTRAN was word-oriented
(32-bit?) and looked a whole lot like the Burroughs Large System in
miniature; the operating system language was a combination of 24-bit
word-oriented and *bit* addressible memory; and COBOL and RPG were
"string-oriented" and looked a whole lot like the Burroughs Medium System
hardware.  I seem to recall also that there were IBM 1401 and Burroughs B500
interpreters available (which, of course, are language-independent) but I
didn't know much about them.

Anyway, the basic premise of the B1000 system as I see it was that the
compilers produced object code for what amounted to an imaginary machine
designed to run the language, and the interpreters "implemented" that
machine.  In the case of the 1401 and the B500, the machine actually
existed; in the case of Fortran and COBOL, it was a scaled-down version of
existing Burroughs hardware; and in the case of the operating system
(written in SDL), it was a hybrid of the hardware design and an ideal
machine.  And as I perceived it the operating system didn't run the machine,
an interpreter called GISMO (General Input-Output Service Module) actually
did that; the SDL object code MCP ran when GISMO decided to run it.  There
was also a MicroMCP to speed some functions up.

    -Chuck Stevens

    -Chuck Stevens
> -- 
> --------
> Sarr Blumson                    sa...@umich.edu
> voice: +1 734 998 9932          home: +1 734 665 9591
> JSTOR, University of Michigan   http://www-personal.umich.edu/~sarr/