Re: Using iconv in stand-alone info

[Eli Zaretskii]

> Date: Mon, 21 Dec 2015 19:51:56 +0000
> From: Gavin Smith <address@hidden>
> Cc: Texinfo <address@hidden>
> 
> >   http://lists.gnu.org/archive/html/bug-wget/2015-12/msg00111.html
> >   http://lists.gnu.org/archive/html/bug-wget/2015-12/msg00113.html
> >
> > So I suggest we fix this in Info by adding a call to 'iconv' when the
> > conversion returns with success.
> 
> Sounds good, I'm not opposed to such a change. Will you write the
> change?

OK.

> Out of curiosity, what is the state shift character in CP1255?

No, cp1255 doesn't have it, if I understand what you are asking.  The
problem is that some cp1255 characters are combining characters, so
libiconv composes them with the preceding character into a single
codepoint of the corresponding precomposed character, and only after
that produces the UTF-8 sequence of that codepoint.  So it needs to be
told there will be no next character for it to output the last UTF-8
sequence.

> In theory there's a problem with the interpretation of shift
> characters, especially locking shifts in the Info format, because you
> have to ask what the shift state should be for things like the "File:"
> keyword. Probably not a real problem though - Info files in encodings
> like ISO-2022-JP probably don't exist.

We also convert when writing to the screen, if the locale's codeset
requires that.