[Tex/LaTex] unicode-math and amsmath environments

amsmathunicode-mathxetex

I'd like to use unicode-math but I've run into some problems. The following mwe, compiles without problems. The aligned environment is used

\documentclass{article}
\usepackage{amsmath}
\usepackage{unicode-math}
\setmathfont{XITS Math}
\usepackage{xltxtra}

\begin{document} 

\begin{equation} 
    α = \beta^2_i + β^2_i 
\end{equation}

\begin{align} 
    a &= b\\
    b &= c
\end{align}

\end{document}

but if I add the following code (using cases environment – defined in amsmath) then I get a segmentation fault

\begin{equation}
\begin{cases}
    a &= b\\ 
    b &= c 
\end{cases}
\end{equation}

What I want to ask is if the two packages can be used together (so this is a bug) or if I am going to encounter more problems like this one.

The output of the compilation is this one and here is the log file.

ps. I know that it is easy to emulate the cases environment.

ps2. Although it must be obvious, I am using xelatex.

UPDATE
I am using TexLive 2010 64bit on Linux

Best Answer

Andrew’s hint is correct: the Linux 64-bit version of XeTeX has a regression, see the bug tracker. You can either:

  1. Wait for the bug fix to propagate to downstream TeX Live (won’t probably happen until TeX Live 2011), or

  2. Install the 32-bit architecture. TeX Live supports multiple architectures on Linux:

    tlmgr platform add i386-linux

    After that you can place a symlink to the 32-bit xetex and xelatex executables in a directory which comes early in your PATH (e.g. ~/bin). This way, you can continue using the other 64-bit binaries.

  3. Or use LuaTeX instead, which doesn’t have that bug.
Related Question