Re: Improving standalone info usability

[Eli Zaretskii]

> Date: Fri, 31 Jan 2014 00:25:36 +0000
> From: Gavin Smith <address@hidden>
> Cc: address@hidden
> 
> > P.S. Regarding your other work, on UTF-8 vs. ISO, it still seems to me
> > that in principle this should be addressed at the terminal level,
> > instead of every single cmdline application having to deal with the same
> > problem.  Not that I have any expectation of that happening
> 
> I agree that every application shouldn't have to deal with the same
> problem and it should be dealt with in some central place. I looked
> into using the iconv functions in libc but unfortunately they don't
> deal with the case when characters cannot be represented.

It is true that characters which cannot be encoded in the terminal's
encoding should be replaced with something that still leaves the text
legible.  However, many characters _can_ be encoded, and Info should
use libiconv for those.  Moreover, the mere fact that a character
cannot be represented should be taken from libiconv's output, rather
than hard-coded in advance in Info's sources.  That way, we won't need
any changes when/if there are terminals or encodings that don't exist
today.

IMO your UTF-8 related patch would be much more complete if it used
libiconv as described above.

Thanks.