Bug#434056: Inputenc and XeTeX don't work together

[Jonathan Kew <jonathan_kew@sil.org>]

On 31 Aug 2007, at 7:05 pm, Frank Küster wrote:

> Hi Jonathan,
>
> here's a bug report we got in the Debian BTS about using inputenc with
> XeTeX.  The full conversation is at http://bugs.debian.org/434056, but
> the first paragraph cited describes the wish quite well.  The  
> problem is
> described in more detail in the initial messages, but maybe you're
> familiar with it.
>
> What's your view on that? TIA for your answer,

Hi Frank,

Yes, I'm familiar with the issue. I normally tell XeTeX users that  
they should not be using [utf8]{inputenc} at all, as the engine reads  
UTF-8 natively. I've sometimes thought that it would be good for the  
package to recognize when it is loaded under XeTeX, and automatically  
disable itself (perhaps with a warning), as this is a fairly common  
mistake for new users.

I seem to recall discussing this with one of the LaTeX team at a  
conference some time ago (maybe Chris? Morten?), but have not  
followed up on it recently.

A further step would be to also support other input encodings via the  
inputenc package. This would require changing the \XeTeXinputencoding  
setting to map the text to Unicode correctly. Then a legacy-encoded  
file that says
  \usepackage[cp1250]{inputenc}
or
  \usepackage[applemac]{inputenc}
(or whatever) could work correctly with Unicode fonts in XeTeX. But  
the utf8 case is the common one, so it would be nice if at least that  
one worked transparently.

The correct place to address this issue is in the base LaTeX release;  
it's not a Debian (or other distro) bug. But in the absence of an  
upstream fix, you might want to try and come up with a patch -- I  
think it would be helpful to users.

JK


> Frank
>
> Juliusz Chroboczek <Juliusz.Chroboczek@pps.jussieu.fr> wrote:
>
> > > > TeX, pdfTeX, Omega or XeTeX, he should be able to say
> > > >
> > > >   \usepackage[utf8]{inputenc}
> > > >
> > > > and the right thing for the current implementation of TeX should
> > > > magically happen.
> >
> > > I suspect you can think so because you use a language
> > > in which there is little difference between utf8 and
> > > normal encoding, for CJK (, Arabic, Hindi, ?) I'm afraid
> > > things are not going so magically ;-)
> >
> > Ehm... no.  XeTeX uses UTF-8 for input, and so does TeX (or e-TeX, or
> > pdfTeX) with utf8.def.  Legacy encodings are completely irrelevant  
> > for
> > this discussion.
> >
> > The point is that the four languages I use regularly are all covered
> > by the small subset of Unicode that works correctly when you say
> >
> >   \usepackage[utf8]{inputenc}
> >   \usepackage[T1]{fontenc}
> >
> > I realise that people for whom that is the case are a minority  
> > (there's
> > probably not much more than 1.2 billion of us in the world). However,
> > just because it doesn't work for most people doesn't mean it should
> > stop working for us.
> >
> > > \usepackage{ifxetex}
> > > \ifxetex\else
> > > \usepackage[utf8]{inputenc}
> > > \fi
> >
> > Yes, i'm currently doing roughly that (but with TeX primitives rather
> > than the ifxetex package).  However I believe that this should be  
> > handled
> > automatically by inputenc.
> >
> >                                         Juliusz
>
> --
> Frank Küster
> Debian Developer (teTeX/TeXLive)