Re: texinfo and UTF-8

[Sergey Poznyakoff]

Eli Zaretskii <address@hidden> ha escrit:

> But you already know the @encoding that was used to produce the
> document, so why do you request that LC_ALL be set?

Technically speaking, it's not me who's requesting it, it's mbrtowc
and its keen who do :) Well, to be strict enough, they require LC_CTYPE 
to be set.

By the way, regexp search (in its current state) also requires LC_CTYPE to
reflect the document language. 

One way to lift this limitation is to provide our own character tables
for all existing encodings, but I leave this approach for the future.

> That depends on the editor.  For example, with Emacs, this can be done
> even in the text-mode ("emacs -nw") session, assuming, of course, that
> the document does not use any character outside the ISO-8859-2
> repertoire (if it does, the foreign characters will be displayed as
> `?').

Yes, but this assumption is very limiting. (As a side note, I was not
able to edit Polish documentation files written in UTF-8 using Emacs
22.1.1 and a terminal set to ISO-8859-2 - every Polish letter was
represented as `?' even though the terminal charset is able to render
it. But maybe I did something wrong...)
 
> > In my opinion, it is user's responsibility to ensure his terminal is
> > set appropriately.
> 
> I don't mind requiring the user to set up her terminal, but
> unfortunately, this is not all she needs to do.  The user will also
> need to install a version of terminal (fonts, etc.) that fit whatever
> @encoding was used in the document.  And that is not as easy as
> setting LC_ALL.

Sure, that's more difficult. But on the other hand, it will be
impossible to view the document without these fonts anyway. I mean,
if, for example, you are going to read a Japanese info file, you have to
have the necessary fonts, there's no other way out. 

Regards,
Sergey