From: Hans Aberg (haberg@math.su.se)
Date: Sat Apr 23 2005 - 07:01:47 CST
I would suggest that the Unicode adds a formal definition of the
notion of "character name" somehow along these lines:
Firstly, I do not see there is a need to change the word "name" to
say "identifier", as "name" is frequently used to indicate a form of
identifier. For example, one often speaks about a "file name", not a
"file identifier"; the latter would probably be used when the
identifier is a number or something.
Second, character names only apply to abstract characters. The so
called "surrogate code points" are not to be regarded as abstract
characters. The private characters are clearly not part of any
Unicode standard, but the standard could indicate that it is wise to
not give them names conflicting with Unicode standard names. One
might set up an Internet service, to register private character
names. It is considerably easier, to humans, to register new
character names than code points.
Then a "character name" is defined to be an identifier that logically
uniquely identifies the character. Its intent is to be human
readable, and helping the human to identify the abstract character.
From the formal logical point, one defines a metaset of metasymbols
A-Z plus space. The use of "meta" here indicates that these
characters are not to be confused with any Unicode abstract
character; in the notion of a logical theory, these metacharacters
are primitives. A formal character name is a finite sequence from
this metaset, not starting or ending with space, and not containing
two adjacent spaces. One can note that with this definition, any
Unicode text can be represented using these metacharacters, as
Unicode characters can be represented by Unicode character names, and
separated by a double space, in view of that no character name can
contain a double space.
The efforts will then focus on giving each abstract character a fair
character name. The effort of finding a linguistically correct
descriptions will be deflected elsewhere.
-- Hans Aberg
This archive was generated by hypermail 2.1.5 : Sat Apr 23 2005 - 07:03:21 CST