Re: etc/HELLO markup etc.

[Eli Zaretskii]

> From: Yuri Khan <address@hidden>
> Date: Sun, 23 Dec 2018 14:47:39 +0700
> Cc: Paul Eggert <address@hidden>, address@hidden, 
>       Stefan Monnier <address@hidden>, Emacs developers <address@hidden>
> 
> There is at least one more situation where different glyphs
> could/should be selected for the same Unicode code points, which
> charset markup does not solve.
> 
> I’m talking about italic shapes of Cyrillic letters. For some of them,
> Russian and Bulgarian use one shape but Serbian and Macedonian use
> another shape[1]. There are no examples of Bulgarian, Serbian, or
> Macedonian in HELLO, but Russian, Ukrainian and Mongolian examples are
> all marked up as “cyrillic-iso8859-5”, which is an encoding that does
> not carry language information.
> 
> So: charset markup is not the right solution to the problem of
> rendering the same Unicode code point with different glyphs.

You mean, it's not a perfect solution, right?  Because in the "good"
department, it's "good enough" to solve at least part of the problem.
No one says we need to reject a solution because it is only partial.

I would also like to point out that, as far as the 'charset' property
is considered, HELLO is just an example of what _can_ be done, it
doesn't pretend to show _everything_ that you could do.  E.g., if it's
important to be able to display Ukrainian in a font different from
that used for Russian, we could use the koi8-u charset for the
Ukrainian greeting, and tweak our default fontset to use special fonts
for that.  We could even invent additional charsets (see
define-charset) and then use them for some greetings.  Of course, this
machinery works best when a charset is unequivocally determined by the
prevalent encoding used for text that uses that charset, and that
isn't always the case.  But still, the feature is there, and it can be
extended if needed.

Finally, regarding the special handling of italics in Serbian: is
there _any_ application out there that solves this problem
satisfactorily in multilingual environment?  I'm not sure how you
could go about that, since fonts generally cover scripts, and there's
no special Serbian Cyrillic script, there's just Cyrl to cover them
all.