J3/99-251 Date: 1999/11/12 To: J3 From: Matthijs van Waveren Subject: Issues 33 and 129 (ASCII and ISO10646) A. Introduction This discussion paper addresses some questions raised in unresolved issues 33 and 129. These are related to the ASCII and ISO/IEC 10646 character sets. All standard references are to J3/99-007R2. A.1 ASCII The J3 internal note in section 4.4.4.1 on page 40 (unresolved issue 129) mentions that the description of SELECTED_CHAR_KIND (section 13.16.97) would imply that an ASCII character kind must be available in a processor. However, I am not sure whether this requirement has ever been explicitly stated. If it has not yet explicitly been stated, then, to my impression, J3 needs a straw vote: Straw vote no. 1: Is it a requirement for a Fortran 2000 processor that the ASCII character kind is supported ? Yes No Undecided Depending upon the outcome of this straw vote, edits for section 4.4.4.1 and/or section 13.16.97 will need to be written. A.2 ISO/IEC 10646 The J3 internal note in section 4.4.4.1 on page 40 (unresolved issue 129) also mentions that a reference to ISO/IEC 10646 is needed in the normative references. This standard is referred to in section 13.16.97. The edit for this reference is listed in section B. The first J3 internal note in section 13.16.97 on page 349 (unresolved issue 32) states "ISO10646 refers to the 32 bit character set". Because it is listed as an unresolved issue, it seems that some discussion might be needed. ISO/IEC 10646 provides two alternative forms of coded representations of characters: the two-octet BMP form, which may be implemented by a 16 bit integer, and the four-octet canonical form, which may be implemented by a 32 bit integer. This paper proposes the following straw vote. Straw vote no. 2: Does J3 want to allow separate values of the kind type parameter in order to distinguish between the two-octet BMP form and the four-octet canonical form ? Yes No Undecided The second J3 internal note in section 13.16.97 on page 349 (unresolved issue 33) mentions that the value ISO10646 for NAME in SELECTED_CHAR_KIND is error prone. So we might need a straw vote to decide for an alternative value for NAME. The wording of this straw vote will depend upon the outcome of straw vote no. 2. B. Edits [7:6] Add the following reference: ISO/IEC 10646-1:1993 Information technology -- Universal Multiple-Octet Coded Character Set (UCS) -- Part 1: Architecture and Basic Multilingual Plane.