At 01:57 PM 7/23/01 +0900, Martin Duerst wrote:
>The language here is slightly different, and I have no idea whether
>the intent was exactly the same, but in any case it seems that the
>intents were very close to each other.
IA characters were from the beginning intended for in-process use, in other
words, not intended for interchange. Instead, they were intended as
reserved placeholder codes which are needed to allow particular, but
important types of implementations of interlinear annotations.
Recently, a much better way to address this and similar needs has been
found, by reserving 32 contiguous BMP locations as 'not-a-character' or
non-character codes. These are codes that cannot, by definition, be part of
a 'character' data stream and can therefore be used internally by a process
as marker codes to support particular implementations, without the
potential of conflicting with character data.
It's somewhat unfortunate that this better solution to the problem was not
developed sooner, but, at the time, it appeared that any solution would
have to follow the mold of the 'Object Replacement Character', in other
words, reserving characters with publicly identified purpose.
A./
This archive was generated by hypermail 2.1.2 : Mon Jul 23 2001 - 13:56:59 EDT