[XeTeX] Math italics with unicode-math.

[Khaled Hosny]

On Sat, Oct 09, 2010 at 04:27:28PM -0400, Cole Leahy wrote:
> I'm using Tex Live 2010. Here's the example I want to discuss.
> 
> \documentclass{article}  
> \RequirePackage{amsmath}  
> \RequirePackage{unicode-math}  
> \setmainfont{Linux Libertine O}   
> \setmathfont{xits-math.otf}  
> \setmathfont[range=\mathit/{latin, Latin, greek, Greek}]{Linux Libertine O}  
> \begin{document}  
> Transfinite induction will reveal that $\kappa$ is $\alpha$-Mahlo for each $\
> alpha < \kappa$. We have proved elsewhere that $\kappa$ is 1-Mahlo, and hence
> 0-Mahlo. If $\kappa$ is $\alpha$-Mahlo for all $\alpha$ below some limit
> ordinal $\lambda < \kappa$, then $\kappa$ is $\lambda$-Mahlo by definition.  
> \end{document}  
> 
> My problem is that the latin, Latin, greek, and Greek characters are set in
> XITS, not Libertine. Now fontspec issues the warning,

For some reason, the spaces breaks parsing of the range option, this
example works for me (but I only run it with luatex, xetex don't see my
texlive fonts and I'm lazy to configure it):

\documentclass{article}
\usepackage{amsmath}
\usepackage{unicode-math}
\setmainfont{Linux Libertine O}
\setmathfont{xits-math.otf}
\setmathfont[range=\mathit/{latin,Latin,greek,Greek}]{Linux Libertine O Italic}

\begin{document}
Transfinite induction will reveal that $\kappa$ is $\alpha$-Mahlo for each
$\alpha < \kappa$. We have proved elsewhere that $\kappa$ is 1-Mahlo, and
hence 0-Mahlo. If $\kappa$ is $\alpha$-Mahlo for all $\alpha$ below some
limit ordinal $\lambda < \kappa$, then $\kappa$ is $\lambda$-Mahlo by
definition.
\end{document}

Besides removing the spacing, you should be using "Linux Libertine O
Italic" unless you want upright math of course. Apart from that, you
should be using \usepackage not \RequirePackage (the later is supposed
to be used in styles and classes AFAIK).

> >fontspec Warning: Font 'Linux Libertine O' does not contain script 'Math'.
> 
> I guess Libertine isn't a math font.
Right, and this warning should be harmful as regard of this context (it
basically means you want get proper radicals or large operators and
stuff like that, but you are using XITS Math for that anyway).

>                                       Nevertheless, similar markup seemed to
> yield the desired result in June at <a href="http://www.charlietanksley.net/
> philtex/the-unicode-math-package-for-xelatex-and-the-stix-fonts/">this blog</
> a>. Has something changed dramatically since then?

Their example does not have spaces, so I'm not sure of this is a new
regression or it was like that from the start.

> I'm unwilling to abandon unicode-math. But I desperately want to use, within
> math mode, Latin and Greek glyphs that were designed for text mode. Since (as
> far as I know) mathspec is no longer compatible with unicode-math, I'm not sure
> what to do. Any suggestions?

Hope this helps.

Regards,
 Khaled


-- 
 Khaled Hosny
 Arabic localiser and member of Arabeyes.org team
 Free font developer