Marion Gunn <mgunn at egt dot ie> wrote:
> To recap: dot above is a traditional diacritic in Irish, reserved for
> use with certain consonants (its function being served, in Roman
> script, by placing the 'letter' h after those same consonants). I
> suppose (with thanks to Antoine for reading my msg so carefully) I
> should add that dotting an i, even in Romanized text, was unusual in
> Irish handwriting until recently, presumably influenced by its
> prevalence in type.
All of this is absolutely true. You're still not getting the main
point, though:
Unicode has NOTHING to do with how font makers design their fonts.
If some type designer has chosen to create an Uncial font, genuinely
intended for Irish use, with a dot over the lower-case "i", then that
font designer has failed in creating authentic-looking glyphs. This is
just as bad as creating a Fraktur font, genuinely intended for authentic
Germanic use, with an incorrect lower-case "k". (Font designers and
readers of pre-WWII German will know what I mean.)
However, the solution to this is NOT to use an inappropriate character,
in this case U+0131.
Earlier in this thread, it was stated that the character U+0069 should
be used to write the "i" in, say, "Marion" when using Antiqua-type
fonts. If that is the case, then U+0069 is also the character to use
when using Uncial, or any other Gaelic-style font, or any other font
with Basic Latin glyphs. Is this clear? You do NOT change to a
different code point when changing fonts. That's what we had to do with
font hacks of the '80s and early '90s (and, sadly, still today with
Nunacom-7).
This is the basic premise of the Character-Glyph Model: characters
retain their identity even when they are displayed with different
glyphs, or appear in different contextual forms (as in Arabic, or in
ligatures). There is no separate code point for Uncial undotted "i"
because it is the SAME CHARACTER as Antiqua-style dotted "i". The
Turkish and Azeri use of both dotted and undotted "i" is totally
irrelevant to this discussion, because Celtic fonts are not designed for
use with Turkish text, nor vice-versa.
WG2 has created, and revised many times, a "Principles and Procedures"
document that clearly explains the difference between character and
glyph. This document is available at:
http://www.dkuug.dk/JTC1/SC2/WG2/docs/n2652r.pdf
I sincerely hope that if you have the experience and expertise in
Unicode that you claim, you can come away with somewhat better
understanding of the principles of Unicode than you are demonstrating
here. Unicode is NOT a glyph registry.
> So, my question still is (having scanned dozens of Unicode responses
> to my msg this wk) our perennial, modest request of how to guarantee
> continuance, in the specific context of Irish text computing, of the
> traditional restriction of the Irish diacritic dot (having only one
> single function in Irish) to the consonants to which it belongs?
Help to educate the designers of Gaelic-style fonts. Attack the problem
from the glyph perspective, where it belongs, not the character-encoding
perspective, where it doesn't.
-Doug Ewell
Fullerton, California
http://users.adelphia.net/~dewell/
