delete-char and composed characters

delete-char and composed characters

[Eli Zaretskii]

It looks like C-f and C-b move by composed character positions on the
screen, but DEL and C-d delete individual characters (i.e. a single
Unicode codepoint).

For example, after typing "C-h H", go to the parenthesized text in the
Bengali line and hit C-d.  You will see that you need to type C-d 5
times to delete the entire text in parentheses, but C-f traverses it
in only 2 keystrokes.

Do we have a single-codepoint variant of C-f and C-b?  If not, should
we?  I think we should, because without them, there's no simple way of
deleting or editing just one codepoint in the middle of a composition.

Or did I miss something?

Re: delete-char and composed characters

[Stefan Monnier]

> Do we have a single-codepoint variant of C-f and C-b?  If not, should we?

Where/how would we display the cursor in order for the user to know on
which codepoint he is?


        Stefan

Re: delete-char and composed characters

[Eli Zaretskii]

> From: Stefan Monnier <monnier@IRO.UMontreal.CA>
> Cc: emacs-devel@gnu.org
> Date: Fri, 24 Oct 2008 14:47:57 -0400
> 
> > Do we have a single-codepoint variant of C-f and C-b?  If not, should we?
> 
> Where/how would we display the cursor in order for the user to know on
> which codepoint he is?

We should display the corresponding portion of the composite character
in reverse video, IMO.

For example, in this example from etc/HELLO:

           Здра́вствуйте

moving one notch forward from `р' (the 3rd letter) puts the cursor on
`а' and should display in reverse video only the lower part occupied
by `а', moving one more notch should display the `́' accent in reverse
video, thus showing something reminiscent of a small cursor up on the
accent.

Re: delete-char and composed characters

[Kenichi Handa]

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

> We should display the corresponding portion of the composite character
> in reverse video, IMO.

That doesn't work in general because, typically in Indic
case, multiple characters are displayed by a signle font
glyph.

---
Kenichi Handa
handa@ni.aist.go.jp

Re: delete-char and composed characters

[Eli Zaretskii]

> From: Kenichi Handa <handa@m17n.org>
> CC: monnier@IRO.UMontreal.CA, emacs-devel@gnu.org
> Date: Sat, 25 Oct 2008 09:49:57 +0900
> 
> In article <uljwdydux.fsf@gnu.org>, Eli Zaretskii <eliz@gnu.org> writes:
> 
> > We should display the corresponding portion of the composite character
> > in reverse video, IMO.
> 
> That doesn't work in general because, typically in Indic
> case, multiple characters are displayed by a signle font
> glyph.

What is the "prior art" in this area? how do other editors and word
processors handle this in Indic scripts?

In any case, the fact that this won't work in general does not mean we
shouldn't do it where it does work.

Re: delete-char and composed characters

[Kenichi Handa]

In article <jwvy70dzv8i.fsf-monnier+emacs@gnu.org>, Stefan Monnier <monnier@IRO.UMontreal.CA> writes:

> > Do we have a single-codepoint variant of C-f and C-b?  If not, should we?
> Where/how would we display the cursor in order for the user to know on
> which codepoint he is?

I'm not sure if we need such a command or not, but when you
run such a command, it is possible disable the automatic
composition temporarily for that part.

---
Kenichi Handa
handa@ni.aist.go.jp

Re: delete-char and composed characters

[Stefan Monnier]

>> > Do we have a single-codepoint variant of C-f and C-b?  If not, should we?
>> Where/how would we display the cursor in order for the user to know on
>> which codepoint he is?
> I'm not sure if we need such a command or not, but when you
> run such a command, it is possible disable the automatic
> composition temporarily for that part.

Indeed, that's how it currently works with the `composition' text
property.  E.g. I typically place a λ composition over the `lambda'
symbols in Lisp and if I somehow force point to land in the middle of
`lambda', then the display is temporarily changed to show "lambda".


        Stefan

Re: delete-char and composed characters

[Eli Zaretskii]

> From: Stefan Monnier <monnier@iro.umontreal.ca>
> Date: Sat, 25 Oct 2008 14:19:14 -0400
> Cc: eliz@gnu.org, emacs-devel@gnu.org
> 
> >> > Do we have a single-codepoint variant of C-f and C-b?  If not, should we?
> >> Where/how would we display the cursor in order for the user to know on
> >> which codepoint he is?
> > I'm not sure if we need such a command or not, but when you
> > run such a command, it is possible disable the automatic
> > composition temporarily for that part.
> 
> Indeed, that's how it currently works with the `composition' text
> property.

This is IMO the least we should do for Emacs 23.1, but it should be a
user-level command, or better yet, some variation of normal motion
commands, because we don't really want the user to mess with
composition directly.