bumming cycles out of parse_identifier()...

[Joseph S. Myers]

On Mon, 10 Sep 2001, Zack Weinberg wrote:

> Well, (a) I think it's silly to allow \u09B6 and not the literal
> character U+09B6, and (b) even if we just do \u escapes we still have
> to canonicalize, because the scenario I described with
> combined/uncombined accents can still happen with \u escapes.

Do you have any examples (where different identifiers using only
characters allowed from the basic character set or Annex D are equivalent
under NFC)?  In those cases I looked at, the combining accents didn't seem
to be included in Annex D where precomposed forms are available.

-- 
Joseph S. Myers
jsm28@cam.ac.uk