Quick Update: I have finally tried to build xetex from source on my system. That version does _not_ segfault unlike the version from TeX Live.
That may suggest that the segfault problem may be partly due to library incompatibilities between different Linux distributions. (Ubuntu has both 64bit and 32bit versions, but uses the 64 bit versions by default.) Nevertheless, my self-compiled version still shows incorect ouput, e.g. big delimiters not growing and radicals printed without a rule.Essentially, I am back to the exactly the problem I reported to the XeTeX list on 06/06/2010 using a xetex binary from Context minimals.
Regards, Ulrik On 07/14/2010 11:59 PM, Ulrik Vieth wrote:
On 07/12/2010 06:45 PM, Jonathan Kew wrote:I have just checked-in a patch to the xetex and texlive source trees to fix the xetex segfault that was occurring with \XeTeXdelcode on some systems. This brings the xetex version number to 0.9997.4.Anyone who has been experiencing this problem is welcome to build and test the new version to confirm that the crash no longer occurs; obviously, I have not been able to test it on all possible systems and configurations.I have updated the latest binaries from TL Pretest and tested it on Linux x86_64. It is getting better, but it is not perfect yet. It has fixed some problem, but I am still getting segfaults and malfunctions, depending on which formulas I typeset. 1) I am getting segfaults for formulas involving \left and \right, i.e. delimiters. See the attached test file xe-test-delim.tex. The \tracingall output indicates that the segfault occurs at the end of display math, i.e. when the accumulated math list is converted to a horizontal and vertical list. The problem only occurs when the formula contains \left and \right delimters, The fraction in display math is typeset correctly when it appears by itself. Unfortunately, I don't have a xetex binary with debug symbols, so I can't provide a meaningful backtrace from the core dump. 2) I am getting incorrect results (presumably caused by incorrectly reading the font metric data) for formulas involving radicals. See the attached test file xe-test-sqrt.tex and corresponding PDF. This seems to indicate that RadicalRuleThickness and or related stuff is not intialized correctly. Regards, Ulrik http://tug.org/mailman/listinfo/xetex
xe-test-delim.pdf
Description: Adobe PDF document
\documentclass{article}
\usepackage{fontspec}
\usepackage{unicode-math}
\setromanfont{XITS}
\setmathfont{XITS Math}
\begin{document}
$$
\left( \frac{x^2}{y^2} \right)
$$
$$
\biggl( \frac{x^2}{y^2} \biggr)
$$
\end{document}
xe-test-sqrt.pdf
Description: Adobe PDF document
\documentclass{article}
\usepackage{fontspec}
\usepackage{unicode-math}
\setromanfont{XITS}
\setmathfont{XITS Math}
\begin{document}
$$
\sqrt{x^2 + y^2}
$$
$$
\frac{1}{\sqrt{x^2 + y^2}}
$$
\end{document}
-------------------------------------------------- Subscriptions, Archive, and List information, etc.: http://tug.org/mailman/listinfo/xetex
