From: "Kenneth Whistler" <[EMAIL PROTECTED]> > Please disregard Philippe's misleading blatherings on this > topic.
Thanks for denying all what I say, before finally saying the same thing as me... What I have said is not blathering, when I just say that UTR 20 is good only in the context of text with layout or rendering markup. When I answered it was for a generic use of XML, which does not imply any markup related to the text content. So I have concluded exactly like you, that what was important was the effect of the canonical normalization, excluding other "compatibility" characteristic of any Unicode character that has a non-canonical decomposition. I still maintain that the first accurate list of compatibility exists, and it is not in UTR20, but in the composition exclusion list data file (which has comments for singleton canonical decomposition mappings) used in combination of the decomposition column of the UCD. If you want a good and _normative_ UTR, better look in UAX #15 (Unicode Normalization Forms), rather than UTR20 which is only _informative_ for both the XML and Unicode standards, and that, for me, is nearly useless out of some specific XML schemas like XHTML. As Alexandre had not specified a text markup context in his question, but only a generic XML context, UTR20 is not relevant for him...
