From: Peter Constable (firstname.lastname@example.org)
Date: Tue Jul 17 2007 - 14:58:51 CDT
The problem is the multi-component vowel marks: there's no way to assign (e.g.) 09CB BENGALI VOWEL SIGN O to a non-zero class. And because you can't do it for 09CB, you can't do it for the characters that make up its canonical decomposition, 09C7 and 09BE. And since you can't do it for those, it also can't be done for any of the left- or right-positioning vowel signs.
There are plenty of things about which we can say "Such-and-such SHOULD have been done." Reality is that these are all assigned to class 0, and that's not changing. That's what we have to implement against.
> -----Original Message-----
> From: email@example.com [mailto:firstname.lastname@example.org] On
> Behalf Of Kent Karlsson
> Sent: Tuesday, July 17, 2007 2:01 AM
> To: 'John H. Jenkins'; 'Unicode MailingList'
> Subject: RE: Generic base characters
> John H. Jenkins wrote:
> > Or, as a similar question, how does a rendering engine distinguish:
> > <092E, 093E, 093F, 093F> (MA + AA + I + I) from
> > <092E, 093F, 093E, 093F> (MA + I + AA + I) and
> > <092E, 093F, 093F, 093E> (MA + I + I + AA) ?
> And this is a major reason why these SHOULD have been assigned
> combining classes other than 0 (224 for the left reordrant ones
> and 226 for the rightones) and have them stack just like above/below
> diacritics. Then all of your examples here would have been
> canonically equivalent, as they should have been.
> /kent k
This archive was generated by hypermail 2.1.5 : Tue Jul 17 2007 - 15:02:04 CDT