A clarification on the MNEMONIC character set. The last meeting of
the working group crafted a working definition of a character set, which
in part declared that a character set must stand on it's own with no
external information needed. This was explicitly done to avoid the
need to support profiling in the charset usages.
Character sets are used both in the header character set encoded
words, and in the charset = "foo" attributes. Neither of these
constructs support profiling.
Using the working definition of a character set, the working group
defined a usage of the general MNEMONIC proposal to be a character set
so that it could be used anywhere any other charset is used. To do
this, a prefix character was choosen as well as a base character set.
By design, these cannot be switched or changed external to the stream
of NMEMONIC characters!
Now, to the choice of anything other than ASCII as a base character
set. The only 7 bit charsets allowed are 2022-jp NMENONIC, and ASCII.
Only ASCII makes sense in terms of MNEMONIC. If a 8 bit charset is
used, it will need to be encoded to be sent over 7 bit paths, making a
really ugly mess and defeating the entire point of NMENONIC! (Even
worse would be 7 bit NMENONIC encoding of 8 bit NMEMONIC.. Whew.)
Please, do not change the basic assumptions just as the final version
of the document is ready to hit the press!
Greg V.