Re: [Scheme-reports] Bytevectors should be called u8vectors

[Ray Dillinger <bear@x>]

On 07/02/2012 07:08 AM, Robert Smith wrote:

> I don't think the Scheme standard should equate "byte" with "octet".

Okay, I say this as an old-skool programmer who actually
remembers machines with bytes of other lengths, and even
remembers the original formal definition of byte as "a
unit of memory containing the same number of bits as the
distance in bits between two consecutive memory addresses."
I remember 36-bit words containing four nine-bit bytes,
and all the rest of that, and I used to grind my teeth
too when people used byte to mean 8-bits regardless.

But that's over now.

Nobody, but nobody, is still using that old definition. In
fact, the definition of byte as 8 bits, regardless of
machine addressing conventions, is now firmly established
by ISO/IEC 80000-13. The same standard gives us 'kibi'.
'mebi', 'tebi', etc prefixes for powers of 1024, which
I'm still getting used to but appreciate for specificity.

It was too fundamental to leave as a hardware convention,
so they didn't.  Even the architects of new machines where
the distance between two consecutive memory addresses is
16 bits, refer to it as two bytes.

Interestingly, the rationale section in ISO/IEC 80000-13
specifically cites understanding among technical people who
rightly expect the use of the common technical term 'byte'
with a standard meaning, and any supposed ambiguity of
the term 'byte' casting confusion on the meanings of many
standards already published in which the term byte is used
specifically to refer to 8-bit entities.  'octet', which
was originally proposed for exactly eight bits, never
caught on outside of the telecom industry apparently,
and is no longer recommended.

So for the same reason that a DSL network bridge which does
absolutely no modulation or demodulation is now a 'modem',
a 'byte' is forevermore accepted specifically as a unit
consisting of exactly eight bits.  The world moves on and
language evolves. Even technical language.

				Bear






_______________________________________________
Scheme-reports mailing list
Scheme-reports@x
http://lists.scheme-reports.org/cgi-bin/mailman/listinfo/scheme-reports