SC22/WG15	N229R
							     November 7, 1991

Title:	Liaison statement from WG15 to WG20 on character class and
	identifier member character issues

Source: JTC1/SC22/WG15 __N231__  (Kista WG15 resolutions)

Project: (9945-2 project)

Status: Liaison statement in response to SC22/WG15RIN N042

Distribution: WG20

Requested action: For review and comment



				ISO/IEC JTC1/SC22/WG15
				November 7, 1991

Title:	Liaison statement from WG15 to WG20 on character class and
	identifier member character issues

Source: JTC1/SC22/WG15 __N231__ (Kista WG15 resolutions)

Pursuant to our previously established liaison,
ISO/IEC JTC1/SC22/WG15 would like to bring the following issues
to the attention of ISO/IEC JTC1/SC22/WG20, and requests that
ISO/IEC JTC1/SC22/WG20 consider further investigation of these
matters.

1.  Character classes

1.1 Background

    JTC1/SC22/WG14, JTC1/SC22/WG15, and JTC1/SC22/WG21 are
    addressing the issue of dividing character sets into
    classes (alphabetic, numeric, whitespace etc.).
    Internationalization considerations require that the
    membership of such classes can be varied according to
    the execution locale.  In its recent work,
    JTC1/SC22/WG15RIN (the rapporteur group on internationalization)
    has determined that there may be a need for a new class,
    that of characters which may appear in programming language
    identifiers (e.g.  variable names).  There may well be further
    such classes identified by other groups.

1.2 Issues for resolution

    a. Bearing in mind the requirements of internationalization,
       what minimum set of character classes is appropriate for
       implementation in programming languages?

    b. Does this minimum set include a class of characters which may
       appear in identifiers, as distinct from alphabetic (see also
       item 2, below)?

    c. Is it desirable that programming languages provide the ability
       to define new classes or change the contents of predefined classes
       at translation or execution time?


2. Extended alphabets in identifiers

2.1 Background

   In locales having larger alphabets than that provided by ISO 646, it
   may be desirable to permit the inclusion of characters from those
   larger alphabets in identifiers.  For example, accented letters from
   such larger alphabets might be included in identifiers.

2.2 Issues for resolution

   a. How may this best be achieved, given that any solution
      should:

      (1) not unnecessarily compromise source program portability or
	  system security,
      (2) be applicable to all types of translation system
	  (for example, both compilers and interpreters),
      (3) be applicable to a wide range of national requirements?
      (4) recognize that the class of nonalphabetic characters acceptable
	  in identifiers vary across programming languages (e.g., the
	  hyphen is acceptable in COBOL identifiers, but not in Fortran)
	  and in various positions within the identifiers (e.g., digits
	  and underscores are generally not acceptable in the first position).

   b. Is the concept of a class of characters (one distinct from class
      "alpha") which may appear in identifiers useful in this connection?

===============================================================================
EOF