> On May 9, 2024, at 8:58 PM, Chuck Guzis via cctalk <cctalk@classiccmp.org>
> wrote:
>
> On 5/9/24 16:30, Michael Thompson wrote:
>> I have a source code tape for Pascal on a CDC 6600 from CDC in France.
>> I am not sure which version it is.
>
> Broadly speaking, there were only three major CDC versions; the 1972
> original, the 1975 rewrite, and the (I think) 1980s version. There were
> intermediate versions, of course.
>
> I think that the 1975 version was widely used as a reference for many
> other implementations.
>
> But now comes the question, "Does one design a machine to a language or
> a language to a machine?" If you take the former course, you have the
> problem of not being able to implement features that the language
> designers didn't imagine. In the latter case, you wind up with a
> language that isn't easily made portable.
Or neither. "Machine to a language" can be seen in the Burroughs mainframes
(ALGOL-60), in the IBM 360 (Fortran and COBOL) and perhaps some others. But a
lot of machines are not built to a particular language, certainly that's the
case for most if not all modern machines. Some machines might have features
to optimize certainly languages while still being quite general; the "display"
support in the Electrologica X8 is an example, but it is just as happy running
Fortran or LISP.
As for "language to the machine" that's pretty much unheard of. While there
certainly are languages that only were seen on one or a few machines or
architectures -- SYMPL, CYBIL, BLISS, TUTOR -- it isn't because that was the
intent of those languages. I suppose you could pose ESPOL as an example of a
language for a machine, though I suspect it could have been generalized, as C
was, if there had been a desire to do so.
paul
