A 04:28 99-07-05 -0700, Michael Everson a écrit :
>Ar 12:44 -0700 1999-07-04, scríobh Curtis Clark:
>><sarcasm type="off">
>>Okay, a fifth way. An "expert set" font with a dotless j, so that combined
>>diacriticals won't look dorky. *Now* am I getting it?
>></sarcasm>
[Michael]
>It isn't funny. Linguists have a real need to combine diacritics.
>Bounding-box specifications in Unicode stack glyphs really nicely, and the
>user or process can choose dotless i (which is present in the UCS) but
>can't choose dotless j unless some phonetically-minded font guy had the
>foresight to put it in the font.
[Alain] What troubles me is that, in a way similar to what Michael says,
some will be tempted (because of bad rendering systems) to use DOTLESS I in
order to compose an I CIRCUMFLEX or an I DIAERESIS, while canonically these
are not equivalent to Îî or Ïï... This will create searching problems in
texts. I would also have prefered that all the combinations of DOTLESS I be
considered euqivalent to all combinations of LETTER I but this is not the
case and when I tried to express this many times, but perhaps too shyly
years ago, I was told that it was not negociable, that the dogma of
canonical equivelence was cast in concrete... (;
Problem is that ISO/IEC 10646 does not preculde to use any letter with any
combianing character... Unicode is more prescriptive, although I guess it
is not forbidden either to do like in the ISO standard (it is just that
canonical equivalence is not guaranteed then, which puts the problem back
to square one).
The same problem could occur if we were to encode a DOTLESS J. I hear what
Michael says though, and it makes sense to me. Solving it will also create
a problem that exists with DOTLESS I used outside of Turkish.
Alain LaBonté
Québec
This archive was generated by hypermail 2.1.2 : Tue Jul 10 2001 - 17:20:48 EDT