Some nits on clause 6.12, Conversions (5.12 in recent drafts)

[Michel Hack (1-914-784-7648) <hack@xxxxxxxxxxxxxx>]

(1)  Issues of character codes (ASCII, Unicode, etc.) are language-defined.

      Issues of character codes are not really defined by languages,
      except in the bizarre sense that "language" is used in this
      standard.  For example, the language C runs on both Ebcdic and
      Ascii platforms.  Some languages, in particular Java, do define
      the character set for strings (in this case, Unicode).

(2)  Hex character sequence specification:
      hexSignificand ({hexDigit} * "." {hexDigit}+ | {hexDigit}+ ".")
      hexExponent {hexExpIndicator} {sign}? {digit}+
      hexSequence {sign}? {hexIndicator} {hexSignificand} {hexExponent}?

      This says that the radix point is required, even when there is
      a hexExpIndicator.  I observe however that some implementations
      (AIX, glibc) do not require a radix point when an exponent is
      given (as is the case for decimal floating-point notation).

      Is this intentional, or is it a subtle error in the grammar?

Michel.

Sent: 2007-08-28 16:10:06 UTC