Re: can gnuplot do UTF-8 at all? [was: pm3d for emf terminal and point types]

[Mojca M. <moj...@gm...>]

On 12/10/06, Ethan A Merritt wrote:
> On Sunday 10 December 2006 04:56, Dr. Johannes Zellner wrote:
> > On Thu, Dec 07, 2006 at 09:26:17AM -0800, Ethan Merritt wrote:
> > > Another item on my emf wishlist is allowing UTF-8 character sets.
> >
> > btw.: Can any terminals do UTF-8 at all?
>
> - svg and wxt use UTF-8 by default
> - libgd (used for png/jpeg/gif) uses UTF-8 by default if it is present
>   in the current font.
>         set term png font "arialuni"
> - x11 uses UTF-8 so long as you put it in multi-byte mode and specify
>   a suitable font:
>         set term x11 font "mbfont:sazanami mincho,vera,20"
> - The PostScript language itself is, so far as I know, not capable
>   of handling multi-byte encodings. So we're out of luck there.
>   However, I think Harald Harders has a patch for post.trm that
>   reencodes the single code page corresponding to the UTF-8 values
>   for latin1 characters. That would at least get you the umlauts
>   and accents.
>
> I am not sure about other terminal types such as pdf, win and aquaterm.

TeX-based pseudo-terminals support Unicode "without any problem".

I just tried the behaviour on windows (I never really bothered since I
only use it for viewing functions, for the final result I always use
ConTeXt, where unicode works "out-of-the-box"). Results:

a) when typing non-latin characters directly into "console window",
they are displayed wrong in the terminal window, but work OK on the
plot. However, these seem to be cp1250-encoded strings.

b) I created a utf-8 encoded file. On the plot the characters are
displayed as two characters, but to be honest: gnuplot doesn't know
that the file is utf-8 encoded and interprets it as cp1250 encoded. I
have no idea how to tell gnuplot that I would like to use utf-8
encoded files.

Does anyone still develop the windows terminal? It seems to me as if
the development would have been pretty dead there.


But one more question, which might be independent from the terminal in
use. How does gnuplot calculate the label width when utf-8 encoding is
used? (For TeX-based terminals this is not that relevant since label
size is usually calculated wrongly anyway because of the used control
sequences.)

Thanks,
    Mojca