Re: Rules for Representation of Subtypes

[kilgallen@eisner.decus.org (Larry Kilgallen)]

In article <52oi3v$din@mill.gdls.com>, schwarza@gdls.com (Art Schwarz) writes:
> Nearest that I can remember, byte size varies by machine. It is NOT an
> eight byte quantity. This is probably the motivation to use _octet_
> instead of bytes in formal descriptions.
> 
>   An example of byte sizes:  CDC 1604    6 bits
>                              CDC 6000   12 bits
>                              PC          8 bits
>                              UNIVAC 1100 7, 8, 9, 12 bits
>                              PDP-10      7 bits (and others?)

The PDP-10 could handle any byte size from 1 to 35 (36 also worked
in hardware, but was a degenerate case, since it was the fullword).

Sizes which were popular in software included 6 and 7, providing
6 and 5 bytes per fullword.  In the case of 7 bit bytes, there
was one bit left over per fullword, and there were cases where
that bit was used for some purpose.

Now I am not predicting that the PDP-10 instruction set will make
a big comeback, but having seen the PDP-10 and the Alpha, I am
certainly not going to claim the ability to predict the nature
of computer hardware thirty years hence.

Several programming languages have survived 30 years, so whatever
that future hardware looks like, Ada programmers should be ready
for it.

Will the demand for Unicode support make 16 bits the typical
size of a Storage Unit ?  Or will egalitarian support for
those Korean dialects drive it to 32 bits ?

Larry Kilgallen