Re: Trying to input Unicode via GNU Emacs 21.3.1

[David Kastrup]

List account <lists@norvelle.org> writes:

> I am trying to use GNU Emacs 21.3.1 on FreeBSD (5.3) to edit web pages
> (I'm accessing my FreeBSD machine via Terminal.App on a Mac, with
> TERM=xterm-color).  I need to input Unicode characters and have them
> appear properly in web browsers.  Currently, I have gotten Emacs to
> use "Unicode" mode (i.e. the two or three little "u"'s appear at the
> bottom left), and I am able to enter characters that look just fine in
> Emacs, but they display as gibberish in browsers.
>
> For instance, I need to be able to display the typical accented
> Spanish, Italian and French characters.  As an example, I can input
> "Alarcón" in Emacs and it looks fine, but it displays in my browser
> (Camino 0.82 on Mac OS X) as "Alarcón".  The odd thing is that I
> basically copied and modified this text from a page that actually
> works just fine.
>
> I have the following lines in my .emacs:
> (setq locale-coding-system 'utf-8)
> (set-terminal-coding-system 'utf-8)
> (set-keyboard-coding-system 'utf-8)
> (set-selection-coding-system 'utf-8)
> (prefer-coding-system 'utf-8)

It would appear that the browser is of the opinion that the selection
is in latin-1, your system default.  You are explicitly telling Emacs
to ignore the system default.

Also with your other settings you tell Emacs that everything the
locale appears to be is wrong.  The easiest thing probably would be if
you not only told your Emacs that all of your environment is utf-8,
but if you just configured your environment to actually be so, in
which case you would not have to tell all of those lies to Emacs.

It may be that in a Latin-1 locale, Emacs-21.3 does not have a way to
tell the browser "Everything in the selection is utf-8".  I believe
that the development version of Emacs _has_ had some changes, due to
some X conventions that have been introduced or become common-place
only after Emacs 21.3 has been release, so it might fair better with
passing Unicode characters over a selection that it principally
Latin-1, at least when the other program also knows about those
conventions.

-- 
David Kastrup, Kriemhildstr. 15, 44793 Bochum