Re: Unicode...

[John C Klensin <KLENSIN@xxxxxxxxxxxxxxx>]

>> eight zeros and ones.  The letter Y, for instance, is represented by
>> the sequence 10111001.
>
>Y, incidentally, is 1011001, in 7 bits.

I've been trying to figure out where the eight-bit form came from, and 
either I just realized or it is strange coincidence.  Back in the dawn 
of EBCDIC history, when the System/360 was just coming out, there was a 
table in the back of the S/360 Principles of Operation manual that 
proported to show a character code that I think was called USASCII-8 -- 
IBM's ASCII approximation for use when one turned on the flag that said 
"run 'ASCII', not EBCDIC".  And, rather than use seven consecutive bits 
and a zero at one end or the other, my vague memory is that they put the 
extra bit In The Middle.  For what they set it to -- 0, 1, or some kind 
of parity notion -- I don't know and would have to dig the manual out of 
my "dead computers" archive.
  But note that stuffing an extra 1-bit into the middle of ASCII-Y 
generates New York Times Y rather efficiently.

  I don't agree completely with all of Eric's remarks.  But I would 
certainly agree that, while 10646 has its problems (many of them, 
possibly enough that it should get another round of review and 
comments), Unicode has more than its own share... this a not a choice 
between "problem" and "solution|" but between "one set of problems" and 
"different set of problems".

  Finally, my impression is that, if there has been a major actor in 
Unicode, it is Microsoft, not IBM.  Not clear that improves anything.
    --john
-------