Re: Standalone 'info' should recode into display's encoding

[Per Bothner]

On 01/01/2014 10:04 AM, Eli Zaretskii wrote:
> > Date: Wed, 01 Jan 2014 09:44:29 -0800
> > From: Per Bothner <address@hidden>

> > They are - if you make the info program/mode locale-aware.
> > The correct place for locale-dependent behavior is in the
> > viewing program, not the info-generating program.
>
> Sorry, I'm not following.  Take the Emacs Info reader, for example: it
> is already locale-aware, in that it stores the characters internally
> in a Unicode-based encoding, and then encodes them according to the
> locale when it writes them to the screen.
>
> Is that what you had in mind as "locale-aware"?

Yes.

> If so, then it doesn't help in this case, because characters such as
> u+2018 cannot be encoded in any encoding I know of except UTF,
> certainly not in Latin-N.  What would you have a locale-aware viewer
> do in this case, i.e. when the offending characters cannot be
> represented in the locale's encoding?

It should replace it with some other character, of course.  Info/emacs
should replace both u+2018 and u+2019 with plain single quote.

Similarly, distributed info files should contain accented characters,
in UTF-8 encoding.  It is up to the viewing program to select a replacement.
For example @copyright{} should generate the copyright symbol (C in a
circle), and the viewing program may replace this by (C).

Such translation needs to be mode-dependent: mapping ‘this’ to 'this'
is appropriate for info-mode and w3m-mode, but it's probably not desirable
for C-mode.
--
        --Per Bothner
address@hidden   http://per.bothner.com/