Doc No:         X3J16/96-120
                                                WG21/N0938
                                Date:           28 May 1996
                                Reply To:       Ichiro Koshida
                                                koshida@cc.teu.ac.jp
 
 
 
        Q and A on Extended Identifiers and Extended Literals
 
          Yuriko Sawatani, Norihiro Kumagai, Ichiro Koshida
 
 
Q: How does it work when a character, which is not defined in the
   execution character set is used as _universal-character-name_ ?
A: For identifiers, it is OK if they are distinguished each other even
   if there are no correspondence characters in the execution character
   set. For literals of strings or characters, a character is translated
   by an implementation-defined encoding. For comments, they are removed
   during translation phase and need not to think about this.
 
Q: Will size of an external identifier be too long if a non basic
   character in an identifier is converted to universal-character-name?
A: In C++, there is no limitation of the size for identifiers.
 
Q: In C, for example, uppercase and lowercase can be treated as the
   same in a linking time. In this situation, will this proposal have a
   problem?
A: In C++, uppercase and lowercase are treated as the different
   characters. So it is OK.
 
Q: I suppose that identifiers have no problem if they can be recognized
   each other, but how about in a debug environment?
A: It is OK to show a character using ??uNNNN, ??UNNNNNNNN if needed.
 
Q: Does it need to prepare all of the mapping of characters in ISO10646?
A: No. You only need to prepare tables from supported source file
   character to universal-character-name in  translation phase 1, and
   tables from universal-character-name to the execution character set
   in translation phase 5.
 
Q: Is a run-time code conversion from universal-character-set to the
   execution character set required in this proposal? If it is true,
   then, will it has an impact for run-time environment, ex. code conversion
   tables and a performance impact for a run-time code conversion?
A: This proposal does not assume a run-time code conversion. In phase 5
   defined in 2.1 Phases of translation, universal-character-name is
   converted to a member of the execution character set.
 
Q: If a compiler supports universal-character-name, i.e. ??uNNNN, ??UNNNNNNNN,
   can it not be used on an OS which does not support ISO/IEC 10646?
A: Yes, it can. universal-character-name is introduced to have a
   portability from one platform to another. When the OS does not
   support ISO/IEC 10646, then universal-character-name is replaced to
   a character defined in translation phase 5.
 
Q: Is it possible to map ??u2323 to a Japanese character x2323 in JIS
   code since I want to develop a compiler for JIS environment?
A: It will not ISO Strictly Conforming Implementation. The character
   designated by the universal-character-name ??uNNNN is that character
   whose encoding in ISO/IEC 10646.