🚀 go-pugleaf

jhelling@cs.ruu.nl (Jeroen Hellingman) writes:

>[about having an A-umlaut for each language seperately to help sorting...]

I believe the sorting procedure is related to the user, not to the data.
It is the user which needs data sorted in the way the user
expect it, so s/he can find what s/he looks for.
For that purpose the character d (a-umlaut, a-diaresis, swedish/finnish
ae) should not have different encodings, according to language, as 
the user does not in many cases know what language the string is
in, for instance is Eberg a Danish or Swedish name. It would be
soretd differently, if it should be sorted according to the
original language.

Most national standards for collating (German, Austrian, Canadian,
Swedish, Danish, Norwegian, Finnish come to my mind) does not
provide specifications for collating according to the original
language, but only with respect to the letters. Some standards are
explicit about this, for example the Danish DS 377 standard.

So I would recommend *not* to have different versions of Latin
characters outside the BMP of 10646. 

I know to little of the Han/Kanji/Hanzi character collation to
give an educated opinion here.

Keld Simonsen