From: Kent Karlsson (email@example.com)
Date: Sun Jun 18 2006 - 04:42:27 CDT
John Hudson quoted:
> *If* you want such quotation marks, then please
> use U+201B for them!
That's a very big "if". It's very far from forcing anyone to use
201B, and I don't expect many will (except if regularly used
for some language).
> This is a glyph display issue, not a text encoding issue. No
> one should be having to
> change characters in order to get the glyph they want.
201B is not a glyph variant of any other character. The charts
should not imply that it is.
> > In any case, font setting is inherently fragile, and is therefore
> > inherently unrobust.
> And yet Unicode entirely relies upon and presumes font
> support for a huge variety of text
> display issues including national and linguistic glyph
> preference (e.g. Serbian).
For purely aesthetic changes, as well as overall style changes,
I agree. But not for orthographic changes, where a completely
different (sequence of) letter(s) is used in different languages
(or for the same language at different times, using different
orthographies), it is completely inappropriate.
> >>also differences to Sindhi shaping of some Arabic letters,
> In this context, it means that the normal glyph shape for the
> final form of U+0647 in
> Sindhi differs from the glyph shape for the Arabic use of the
> same letter. So if one has a
> font for multilingual use, one needs a mechanism to affect a
> Sindhi-specific subsitution.
I don't know how different these are in this particular case,
but if substituting one shape for the other is seen as a spelling
error (rather than just "not-the-nicest-but-perfectly-good"),
then the font difference approach is not appropriate, and a
separate character should be used.
> This is the sort of issue we deal with in font development
> every day. It is the inevitable
> results of the Unicode character/glyph distinction, and is
> one of the things that OpenType
> was specifically designed to accomodate.
Going too far that way leads astray. I've seen too many suggestions
going much to far in this respect (making orthographic differences
into font differences). So I'm very suspicious about all and any
such suggestions. They are like "hacked fonts" but worse since
the error in application may not be immediately noticable.
Many of the new features in Opentype should be used vary sparingly,
and with a very high degree of restraint. Indeed, I'd rather see the
"language dependence" feature in OT deprecated (and any support
for it turned off by default or, even better, always) rather than
supported, since just about all suggestions I've heard of to use it
have been inappropriate. That includes at least one, probably both,
of the instances for Sindhi you just took up.
This archive was generated by hypermail 2.1.5 : Sun Jun 18 2006 - 04:49:32 CDT