Re: Fun with History: “Why wasn't Ada83 object oriented?”

[Simon Wright <simon@pushface.org>]

"Yannick Duchêne (Hibou57)" <yannick_duchene@yahoo.fr> writes:

> By the way, I do too believe if one want to implement in C or
> assembly, that's indeed profitable to first design in Ada (if really
> Ada is not an option at implementation stage for some weird
> reasons). But then the design have to be checked, so I guess it is
> compiled in some way, so why not use the compiled product instead?
> Due to a not available platform?  What platform then? Or else there
> may be other reasons, then could you tell more about these reasons?

The target processor was a Ferranti F2420[1], a 3-address word-addressed
24-bit-word machine (with an assembler looking strikingly like the
notation used by Ada to discuss the Analytical Engine[2]).

The requirement was to produce a Mascot[3] kernel, a light-weight RTS.

I used Ada to make sure the design was internally consistent, but there
was a point at which I had to say 'at this point a miracle happens'[4],
ie switching between one activity (thread, task) and another. The
compiler was an early Telelogic compiler, I think, and host/target was
VAX.

[1] http://en.wikipedia.org/wiki/Ferranti#Computers - end of section
[2] http://www.fourmilab.ch/babbage/sketch.html - eg Note C
[3]
http://en.wikipedia.org/wiki/Modular_Approach_to_Software_Construction_Operation_and_Test
[4] http://www.flickr.com/photos/skepticalist/4372728626/