Re: national chars - libiconv not used?

[Michal Hajek <hajek@xxxxxxxx>]

Alain Bench wrote:
>  On Monday, June 19, 2006 at 8:04:57 +0200, Michal Hajek wrote:
> > [Latin-2] picture attached.
>     Good news! That's a (mostly) correct Latin-2 table. Which probably
> means that HP iconv functions are at work, and work well. Main problem
> remains with ~A and such for those chars that are converted to 80-9F.
> And the non-break space looks like "a" instead of " ": Font? HP iconv
> specificity? What is the output of printf "'\xFF'" at shell?

It is matter of font, "Terminal" shows "a", "Fixedsys" shows " ".

> > Alain Bench wrote:
> >
> > > ("~A" for char 0x81) and such are a signature of Ncurses
>
>     What exact version of Ncurses 5.5 do you use (tic -V)? I vaguely

ncurses 5.5.20051010 acording to /usr/local/bin/tic (come with ncurses) 
System tic in hpux does not know the switch -V.

> > > > Bad IDN 'mutt.org' in alias 'mutt-dev'
> >
> > This error is not shown anymore. This was matter of "old" libiconv, I
> > believe.
>
>
>     In the context, the exact cause of this error is: Mutt's $charset is
> not known by the libiconv called by libidn. So the solution was perhaps
> the libiconv recompilation with --enable-extra-encodings (necessary for
> CP-852).

Yes, that's my opinion too.
>

> > checklocale returns no error
>
>
>     Does it print for 80-9F: Two lines full of "#", or real chars?
Real chars. Picture attached.

Thanks, Michal