Composed with ... by these characters

Composed with ... by these characters

[Eli Zaretskii]

  
  emacs -Q -nw

After "C-h H", go to the "Bengali" line, type C-f to get to one of the
non-ASCII characters, and type "C-u C-x =".  In the "*Help*" buffer,
you will see the following snippet, among other text:

    Composed with the surrounding characters "?" and "??" by these characters:
     ? (#x2476)
     ? (#x2494)
     ? (#x2434)
     ? (#x2482)

What does it mean "composed with ... characters ... BY THESE
CHARACTERS?  This probably needs to be rephrased, but how?

Also, is it normal that on a tty such composed characters show more ?
characters than glyphs on a GUI display?  For example, in a GUI
session, the word inside parentheses after "Bengali" is displayed with
2 glyphs, while a tty shows no less than 5 question marks.  Is this
expected behavior?

Re: Composed with ... by these characters

[Kenichi Handa]

In article <uk583tv67.fsf@gnu.org>, Eli Zaretskii <eliz@gnu.org> writes:

>   emacs -Q -nw

> After "C-h H", go to the "Bengali" line, type C-f to get to one of the
> non-ASCII characters, and type "C-u C-x =".  In the "*Help*" buffer,
> you will see the following snippet, among other text:

>     Composed with the surrounding characters "?" and "??" by these characters:
>      ? (#x2476)
>      ? (#x2494)
>      ? (#x2434)
>      ? (#x2482)

> What does it mean "composed with ... characters ... BY THESE
> CHARACTERS?

It means that Emacs generated a lgstring (Lispy Glyph
String) as an automatic composition, but as there's no way
for the terminal to show correct glyphs in the lgstring,
Emacs just sends characters corresponding to glyphs to the
terminal.

> This probably needs to be rephrased, but how?

I've just fixed describe-char not to show such trivial
composition informaion.  Here, trivial means that a
character constitutes a graphme cluster by itself without
alternate characters.

> Also, is it normal that on a tty such composed characters show more ?
> characters than glyphs on a GUI display?  For example, in a GUI
> session, the word inside parentheses after "Bengali" is displayed with
> 2 glyphs, while a tty shows no less than 5 question marks.  Is this
> expected behavior?

Yes.  On a graphic terminal, those charaters should be
displayed by two grapheme clusters (one for the first three
characters, the other for the last two characters).  On a
terminal, as we can't show such correct grapheme clusters,
Emacs just tries to show each 5 characters, and if the
terminal coding system doesn't support them, they are shown
by ?s.

---
Kenichi Handa
handa@m17n.org