bug#31588: 26.1; format counts some Unicode characters as two

[Eli Zaretskii]

> From: Noam Postavsky <npostavs@gmail.com>
> Date: Thu, 24 May 2018 20:50:45 -0400
> Cc: 31588@debbugs.gnu.org
> 
> > As far as I can tell, something is off about the styled_format and how
> > it counts characters- I suspect something about counting
> > multibyte characters, but it's a little beyond me.
> 
> I think it's intended, ➖ and ❌ are wide characters:
> 
>     (mapcar #'char-width '(?➖ ?❌ ?▶ ?✚ ?▷ ?◼))
>     ;=> (2 2 1 1 1 1)

Yes, that's the reason.

> Although in my current font the width seems to be more like 1.8 or
> something.

Unfortunately, the actual width depends on the font, and what's more,
sometimes varies significantly between different fonts.  Maybe we
should have a variant of char-width that actually looks at the font
(of course, such a variant will need a window or frame as one of its
arguments).

But with the current state of our support for wide characters, this is
not a bug.