[XeTeX] "new-babel", was: Ancient Greek hyphenation

[Will Robertson]

On 4/24/07, Jonathan Kew <jonathan_kew at sil.org> wrote:

> Suppose we make \selectlanguage
> {hindi} automatically switch to a Devanagari font... then the user
> chooses a *different* Devanagari font... then will \selectlanguage
> {nepali} revert to *its* default Devanagari font, rather than keeping
> the user's chosen one? This doesn't feel right to me.

Right...I hadn't thought of that.
You could have fonts defined by script, though, and setting a font for
a "language" actually sets it for the set of languages that uses that
script. (Override-able to a specific language, of course.)

> Or if the language-switching package provides appropriate "hooks", I
> can add my chosen font-switches to it on the fly. But I don't think
> we should have language changes causing font changes by default (as
> some Babel languages currently do, IIRC).

This is exactly how I imagined it would be set up. For a hypothetical
minimal document

\documentclass{article}
\usepackage{postbabel}
\begin{document}
man and \FrenchText{femme}.
\end{document}

the French text would only change fonts if some language setup were called

\postbabel{French}{%
  hyphenation=french,% implicit
  thisfont=Didot,% as opposed to:
%  scriptfont=Bodoni,
  fontoptions={Language=French},% implicit
}

Unless there are good examples of one language--many scripts, that
can't be serviced by this sort of arrangement (which there probably
are, knowing me), I think it adds a bit too much complexity to have a
"script" syntax as well as a "language" syntax.

> Maybe it would be clearer if \selectlanguage were renamed
> \selecthyphenation or something like that. "Select language" has too
> many possible interpretations.

That's a good idea, but if you need backwards compatibility, and we're
building a wrapper around it anyway...

Will