2

I am trying to use mhchem inside the short caption for a number of figures and tables, and it is interfering with the thesis class I'm required to use (pittetd). It only happens when forming the list of tables or figures; if this doesn't need to be generated, there is no problem. More specifically, it only occurs when \bond{....} is present; \ce{} is not the issue.

Here is a MWE:

\let\oldyear\year
\documentclass{pittetd}
\let\setyear\year
\let\year\oldyear

\usepackage{booktabs}

\usepackage[version=4]{mhchem}
\newcommand{\arlidimert}{argon\textemdash{}lithium cation dimer}
\newcommand{\arlidimer}{\ce{Ar\bond{....}Li+}}

\begin{document}
\listoftables

\begin{table}
  \caption[Percentage of ALMO-EDA charge transfer in \arlidimer{} due to higher-order effects]{Percentage of ALMO-EDA charge transfer in \arlidimer{} due to higher-order effects}
  \label{tab:almo-eda-higher-order-pct-of-ct}
  \begin{tabular}{ccc}
    \toprule
                    & \multicolumn{2}{c}{basis set} \\
    BSSE corrected? & def2-SVP    & def2-SVPD   \\
    \midrule
    no              & 24.2        & 17.1        \\
    yes             & 25.0        & 19.4        \\
    \bottomrule
  \end{tabular}
\end{table}

\end{document}

If article is used instead, there's no issue.

The error output from latexmk -pdf mwe.tex:

(/usr/local/texlive/2017/texmf-dist/tex/latex/graphics/color.sty
(/usr/local/texlive/2017/texmf-dist/tex/latex/graphics-cfg/color.cfg))
(/usr/local/texlive/2017/texmf-dist/tex/latex/hyperref/nameref.sty
(/usr/local/texlive/2017/texmf-dist/tex/generic/oberdiek/gettitlestring.sty))
(./mwe_mhchem_in_short_captions.out (./mwe_mhchem_in_short_captions.lot
! Undefined control sequence.
<argument> Ar\bond 
                   {....}Li+
l.1 ...ue to higher-order effects}}{ii}{table.0.1}

When inspecting the *.lot files, the difference is clear. For the article (working) example:

\contentsline {table}{\numberline {1}{\ignorespaces Percentage of ALMO-EDA charge transfer in \ce {Ar\bond {....}Li+}{} due to higher-order effects}}{1}

but for pittetd (not working):

\contentsline {table}{\numberline {1}{\ignorespaces Percentage of ALMO-EDA charge transfer in \ce {Ar\bond {....}Li+}{} due to higher-order effects}}{ii}{table.0.1}

Since it seems to be triggered by \bond, I can't tell what the red herring is, or if something else is going on. Assume I don't understand how the class file works. What is the appropriate fix?

pentavalentcarbon
  • 197
  • 1
    Try \usepackage{etoolbox}\robustify\arlidimer. – Ulrike Fischer Jan 30 '18 at 22:42
  • It works. I see that it preserves the command as it passes through the *.lot. Is it just changing when the macro expansion occurs? Why would the environment be different between the two classes? – pentavalentcarbon Jan 30 '18 at 23:14
  • Nevermind, I found https://tex.stackexchange.com/questions/4736/what-is-the-difference-between-fragile-and-robust-commands and expect it'll probably answer those questions for me. – pentavalentcarbon Jan 30 '18 at 23:16
  • pittetd.cls heavily redefines \contentsline to work with hyperref which likely accounts for the problem. – Alan Munn Jan 31 '18 at 00:29
  • Ok, I'm having trouble getting this working in my non-MWE with a similar error. I'll probably read that question first. – pentavalentcarbon Jan 31 '18 at 01:48

1 Answers1

1

Converting the comment into an answer to get it from the unanswered list:

Make the command robust with

\usepackage{etoolbox}
\robustify\arlidimer
Ulrike Fischer
  • 327,261