John Cowan wrote:
> "Modifier letter" is a rather loosely defined category: what's in it
> is kind of random.
You are right. A quick browse in UnicodeData.txt shows that the category
includes members such as Arabic tatweel or CJK iteration marks.
Nevertheless, at least in the Latin (IPA), Greek, and Armenian scripts, the
category seems quite consistently used for spacing diacritic signs derived
from letters.
For this reason, I assume that Peter Constable would insist to keep his
ye-to-be-proposed new phonetic characters in the same category "Lm".
But U+00AA and U+00BA (the two existing characters used as singular ordinal
indicators in Italian) have category "Ll", which is also the category of
U+2071 (the Unicode 3.2 character currently considered for unification with
the masculine plural ordinal indicator).
So, I assume that also the fourth character in the group (the feminine
plural ordinal indicator) should be in the same category, in order to avoid
that applications handle it differently than the other three ordinal
indicators.
> Names don't have to make sense.
I understand that some Unicode names are misnomers, and that it is not
possible to change them because of Unicode and ISO policies. But I thought
that one thing is accepting to live with historical mistakes, and a totally
different thing is generating new ones by purpose.
Anyway, I must agree that a cute name is too weak an argument to decide
whether to unify or disunify two characters.
> Smart fonts are unavoidable, and if you are using IPA, you
> often have to have one anyhow.
Let's assume that they are unavoidable; I don't want to go through that now.
The point is that the character that I am considering to propose, as well as
the other three "ordinal indicators", is mainly needed in plain text,
because of the impossibility to attribute the "superscript" property to a
*run* of text.
The reason for this impossibility is that plain text, by definition, cannot
contain markup that tells "here starts superscript" and "here ends
superscript".
My assumption is that also manually switching of smart-font features for
*runs* of text cannot be done in plain text so, if the characters were
unified, font feature would not be a viable method for distinguishing
different usages of it.
_ Marco
