Terminology for types of code points

L2/02-178

To:	UTC
Re:	Terminology for types of code points
From:	Ed Committee
Date:	2001-04-26

The editorial committee is seeking feedback from the UTC on a matter of terminology, having to do with types of code points.

There are the following main types of code points that we need to distinguish:

'Normal' characters
Format characters
PUA characters*
Surrogate code points*
Noncharacter code points*
'Open' code points

Notes:

The starred ones are closed sets; the contents won't change.
The word 'open' is chosen to not prejudge the terminology discussion below. It is not meant to be a permanent term.
Open code points are not yet assigned as characters, but could be -- thus migrating to Normal or Format, and (very rarely) Normal could go to Format or vice versa.
Normal, Format and PUA are code points assigned to characters, while Surrogate, Noncharacter and Open are not assigned to characters.
PUA's are a wierd beast, but formally we consider them as associated with a (very) 'abstract character', primarily as an conceptual artifact for mapping other character encodings.
Format characters (which include control characters) are also oddities; to the man on the street, not really a 'character', but as a practical matter included in the standard.
Noncharacters are really internal-use characters. Not to be interchanged, but available as programmatic sentinels, or placeholders.

There are different unions of these sets that are used often, and need their own names. Plus at least 'Open' above needs a good name.

All but surrogates
- Code points that are not surrogates
Surrogate, Noncharacter, Open
- Code points not associated with abstract characters
Normal, Control, PUA
- Code points associated with abstract characters
- negation of B
Open, Noncharacter
- = Cn - Hardwired in the UCD
Open

We then have terminology that we have used imprecisely in the past:

Assigned	In the UCD docs, Cn is equated to this
	Also used for #1-#5 (e.g. non-open)
	Also used for #1-#3 (e.g. code points not assigned to characters)
Unassigned	Inverse of Assigned
Scalar Value	Code point
Scalar Value	Nonsurrogates
Reserved	Surrogate, Noncharacter, Open in 10646, with different adjectives qualifying the different groups.

In coming up with names, we also need to make sure that negations are reasonable: that nonX means all code points that are not X.

The question is, what terms should we choose for A-E. There are a couple of different possible positions:

Term	Position1	Position2
A: all but surrogates	Nonsurrogate code point	Scalar Value code point
B: Surrogate, Noncharacter, Open	Noncharacter code point	Unassigned code point
C: Normal, Format, PUA	Character code point	Assigned code point
D: Open, Noncharacter	??? code point	??? code point
E: Open	Unassigned code point	Nondesignated code point
not Open	Assigned code point	Designated code point
Noncharacter	Internal-Use code point (Infernal-Use ;-)	Noncharacter code point
Surrogates	Surrogate code point	Surrogate code point, Nonscalar value code point

Feedback from the UTC would be most appreciated as to which of these choices would be the most reasonable and least confusing. The committee is not really wonderfully happy with either of these sets of terms; it is very open to different suggestions for a cohesive set!