Re: [boost] 3-state boolean type (more historical trivia)

From: Victor A. Wagner, Jr. (vawjr_at_[hidden])
Date: 2002-08-02 11:51:17

At Thursday 2002/08/01 15:20, you wrote:
>A historical note - circa 1968 - Computer Design magazine ran a series of 
>articles on three-state hardware logic.
>
>Knuth called the three states flip-flop-flap, IIRC:-)

ternary logic (or base 3 systems) were indeed discussed c1968.  The biggest 
drawback to them from a hardware point, IIRC, was the difficulty of making 
circuitry which would switch between any of the three states without (in 
some cases) "passing through" the 3rd.  There were also all of the 
"off-color" comments about what to call the equivalent of "bit" and such 
nonsense.  The appeal, of course, was an increase in density of computing 
elements (shorter word sizes to hold the same magnitude of numbers).  The 
most common "word size" was 27 ternary-digits which gives numbers in the 
±3,812,798,742,493 range.
My favorite argument was that you could build the system such that 
truncation and rounding were the same operation (use the values -1, 0, 1 
for the digits rather than the  more traditional 0, 1, 2).
I think the thing that really shot it down tho, was trying to figure out 
how to deal with the "boolean" operators.  Instead of having 16 (2^(2^2)) 
of them as we do in binary we'd have 19683 (3^(3^2)).  Where would we come 
up with all the _names_???!!

Even more bizarre, IMO, than the radix 3 (ternary) logic systems were the 
articles (same time frame) about using -2 as the radix.

>--Beman

Victor A. Wagner Jr.      http://rudbek.com
PGP RSA fingerprint = 4D20 EBF6 0101 B069 3817 8DBF C846 E47A
PGP D-H fingerprint = 98BC 65E3 1A19 43EC 3908 65B9 F755 E6F4 63BB 9D93
The five most dangerous words in the English language:
               "There oughta be a law"