On (20 Jul 97) Cameron Clark wrote to Jerry Coffin...
 JC> to find it in either the C or C++ standard.  C and C++ are both
 JC> fairly machine oriented, meaning they allow the machine to store
 JC> things in the most convenient form for the machine.
 CC>     Nasty, to say the least. I remember stories of an old rusian
 CC> machine that had n memory. That's right, it stored bits as
 CC> positive,none, and negative. As you can tell, the base 3 machine
 CC> suffered from serious physical limitations and was dumped.
Sounds a bit hard to believe to me - Knuth mentions the idea of a
trinary machine in his books, but implies that such a thing has never
existed.  In any case, the standards _do_ specify binary encoding,
though they allow 1's or 2's complement or sign/magnitude.
 JC> There has been a proposal for a standardized format for binary
 JC> files, and I believe it was formally placed before the C9X
 JC> committee, but I believe the proposal was placed before the
 JC> committee after they'd said they weren't going to accept more
 JC> proposals for extensions, so it died a'borning, so to speak.
 CC>     You mean like sequential,indexed sequential, and so on? If so,
 CC>     bad idea - dos systems and the like only have sequential files 
 CC>     natively.
No - just an encoding so you could do things like write a long out to a
file on one system and read it on another and dependably get the same
value on each.  I'm not sure about many of the details, but it would
have only had to do with the encoding of individual data items, not the
overall organization of a file.
    Later,
    Jerry.
... The Universe is a figment of its own imagination.
--- PPoint 1.90
---------------