mathdots (12jun14)

git-svn-id: svn://tug.org/texlive/trunk@34301 c570f23f-e606-0410-a88d-b1316a301751

1 files changed, 257 insertions, 122 deletions

diff --git a/Master/texmf-dist/source/generic/mathdots/mathdots.dtx b/Master/texmf-dist/source/generic/mathdots/mathdots.dtx
index 45862da44ff..d5ace76661b 100644
--- a/Master/texmf-dist/source/generic/mathdots/mathdots.dtx
+++ b/Master/texmf-dist/source/generic/mathdots/mathdots.dtx
@@ -4,11 +4,11 @@
 % Copyright 2001--2006, Daniel H. Luecking
 %
 % Mathdots may be distributed and/or modified under the conditions of the
-% LaTeX Project Public License, either version 1.3 of this license or (at
+% LaTeX Project Public License, either version 1.3c of this license or (at
 % your option) any later version. The latest version of this license is in
 %    <http://www.latex-project.org/lppl.txt>
-% and version 1.3 or later is part of all distributions of LaTeX version
-% 2003/12/01 or later.
+% and version 1.3c or later is part of all distributions of LaTeX version
+% 2008/05/04 or later.
 %
 % Mathdots has maintenance status "author-maintained". The Current
 % Maintainer is Daniel H. Luecking. The Base Interpreters associated with
@@ -21,13 +21,14 @@
 %<sty>\ifx\documentclass\UndEfInEd\input mathdots \expandafter\endinput\fi
 %<sty>\ProvidesPackage{mathdots}
 %<*sty>
-[2006/03/16 v0.8 Improve and add various macros for dots in math.]%
+[2014/06/11 v0.9 Improve and add various macros for dots in math.]%
 %</sty>
-%<tex>\def\mathdotsfiledate{2006/03/16}%
-%<tex>\def\mathdotsfileversion{0.8}%
+%<tex>\def\mathdotsfiledate{2014/06/11}%
+%<tex>\def\mathdotsfileversion{0.9}%
 %<*driver>
-\documentclass[draft]{ltxdoc}
-
+\documentclass{ltxdoc}
+\usepackage{amsmath}
+\usepackage{mathdots}
 \addtolength{\textwidth}{.5878pt}
 
 \def\mytt{\upshape\mdseries\ttfamily}
@@ -62,7 +63,7 @@
 % Stupid index!
 \def\usage#1{\textrm{#1}}
 \def\index@prologue{\section*{Index}\markboth{Index}{Index}%
-  Numbers refer to the page where the corresponding entry is described.
+  Numbers refer to the page(s) where the corresponding entry is described.
 }
 \def\IndexParms{%
   \parindent \z@ \columnsep 15pt
@@ -83,14 +84,13 @@
 \DisableCrossrefs
 \CodelineIndex
 \AlsoImplementation
-
 \begin{document}
   \DeleteShortVerb{\|}
   \DocInput{mathdots.dtx}
 \end{document}
 %</driver>
 %\fi
-% \CheckSum{327}
+% \CheckSum{364}
 % \CharacterTable
 %  {Upper-case    \A\B\C\D\E\F\G\H\I\J\K\L\M\N\O\P\Q\R\S\T\U\V\W\X\Y\Z
 %   Lower-case    \a\b\c\d\e\f\g\h\i\j\k\l\m\n\o\p\q\r\s\t\u\v\w\x\y\z
@@ -116,85 +116,182 @@
 % in either direction. It should work equally well in \LaTeX{} or plain
 % \TeX{}. It improves on the default definitions of plain \TeX{} and
 % \LaTeX{}. Similar improvements are provided for the triple and
-% quadruple dot accents of AMS\TeX{}/\LaTeX{}.
+% quadruple dot accents of AMS\TeX{}/AMS\LaTeX{}.
 % \end{abstract}
 %
 % \StopEventually{\PrintIndex}
 %
 % \section{Introduction}
 %
-%  Running \LaTeX{} or \TeX{} on \file{mathdots.ins} generates
+% Running \LaTeX{} or \TeX{} on \file{mathdots.ins} generates
 % \file{mathdots.sty}, a \LaTeX{} package, and \file{mathdots.tex} a plain
 % \TeX{} input file. If one puts \verb"\input mathdots.sty" in a plain
 % \TeX{} file, it will cause \file{mathdots.tex} to be input. This is for
 % backward compatibility.
 %
-% Unlike the commands \cs{vdots} and \cs{ddots} provided by \LaTeX{},
+% Unlike the commands \cs{ddots} and \cs{vdots} provided by \LaTeX{},
 % the versions defined by \MD{} change size with \LaTeX{} size
 % changing commands, as well as in \{sub[sub]\|super[super]\}scripts.
-% After \MD{} is loaded, \verb"\Large$\ddots$" will be larger and
-% \verb"\small$\ddots$" will be smaller than \verb"$\ddots$". Moreover,
-% \verb"$2^{2^\ddots}$" will also produce a smaller symbol. This last
-% example is very unlikely to be needed, but the version with the
-% opposite slant is very common.
+% After \MD{} is loaded, \verb"{\large$\ddots$}" will be larger and
+% \verb"{\small$\ddots$}" will be smaller than \verb"$\ddots$" at normal
+% size. Moreover, \verb"$2^{2^\ddots}$" will also produce a smaller
+% symbol. This last example is very unlikely to be needed, but the version
+% with the opposite slant is very common (and is provided by \MD{}).
+%
+% The \prog{amsmath} accents \cs{dddot} and \cs{ddddot} have similar
+% problems that \MD{} attempts to correct when \prog{amsmath} is loaded.
+%
+% See Table~\ref{examples} for examples of the \MD{} commands at
+% different sizes and in exponents. See Table~\ref{compare} for
+% comparison, showing the results without \MD{}.
+%
+% \def\dott#1{$#1$}
+% \def\dotts#1{$2^{#1}\quad 2^{2^{#1}}$}
+% \begin{table}[htbp]
+% \centering \renewcommand\arraystretch{1.4}
+% \begin{tabular}{c|cccc}
+% \multicolumn{1}{l}{\large With \MD{}:}\\[4pt]
+% \multicolumn{1}{c}{\textbf{Command}}%
+%                  &\textbf{Large}          &\textbf{normal}   &\textbf{scriptsize}
+% &\textbf{in exponents}\\
+% \hline
+% \verb$\ddots$    & \Large\dott{\ddots}    & \dott{\ddots}    & \scriptsize\dott{\ddots}
+% & \dotts{\ddots}\\
+% \verb$\vdots$    & \Large\dott{\vdots}    & \dott{\vdots}    & \scriptsize\dott{\vdots}
+% & \dotts{\vdots}\\
+% \verb$\iddots$   & \Large\dott{\iddots}   & \dott{\iddots}   & \scriptsize\dott{\iddots}
+% & \dotts{\iddots}\\
+% \verb$\dddot{X}$ & \Large\dott{\dddot{X}} & \dott{\dddot{X}} & \scriptsize\dott{\dddot{X}}
+% & \dotts{\dddot{X}}\\
+% \verb$\ddddot{X}$& \Large\dott{\ddddot{X}}& \dott{\ddddot{X}}& \scriptsize\dott{\ddddot{X}}
+% & \dotts{\ddddot{X}}
+% \end{tabular}
+% \caption{Dots at different sizes and in exponents.}
+% \label{examples}
+% \end{table}
+%
+% \begin{table}[htbp]
+% \centering \renewcommand\arraystretch{1.4}
+% \begin{tabular}{c|cccc}
+% \multicolumn{1}{l}{\large Without \MD{}:}\\[4pt]
+% \multicolumn{1}{c}{\textbf{Command}}%
+%                  &\textbf{Large}                      &\textbf{normal}              &\textbf{scriptsize}
+% &\textbf{in exponents}\\
+% \hline
+% \verb$\ddots$    & \Large\dott{\originalddots}     & \dott{\originalddots}    & \scriptsize\dott{\originalddots}
+% & \dotts{\originalddots}\\
+% \verb$\vdots$    & \Large\dott{\originalvdots}     & \dott{\originalvdots}    & \scriptsize\dott{\originalvdots}
+% & \dotts{\originalvdots}\\
+% \verb$\dddot{X}$ & \Large\dott{\originaldddot{X}}  & \dott{\originaldddot{X}} & \scriptsize\dott{\originaldddot{X}}
+% & \dotts{\originaldddot{X}}\\
+% \verb$\ddddot{X}$& \Large\dott{\originalddddot{X}} & \dott{\originalddddot{X}}& \scriptsize\dott{\originalddddot{X}}
+% & \dotts{\originalddddot{X}}
+% \end{tabular}
+% \caption{Dots without \MD{}.}
+% \label{compare}
+% \end{table}
 %
+% \DescribeMacro{\ddots}
 % The default \LaTeX{} (and plain \TeX) definition changes the size of the
-% dots, and the horizontal dimensions, but not the vertical dimensions.
-% Using the default version in 'scripts produces symbols identical to
-% those produced elsewhere. Finally, the default takes the dots from the
-% current text font, whereas the \MD{} package takes them from the
-% current math symbol font, so they should match other dots in math mode.
-% This is only really a concern in \mtt{17pt} sizes or larger where
-% \file{cmr17} is scaled to get text dots, but \file{cmsy12} is scaled to
-% get math dots.
-%
-% \cs{iddots} is similar to \cs{ddots}, except it produces the other
-% diagonal. This command is not provided in basic \TeX{} or \LaTeX{}.
+% dots, and the horizontal dimensions, but not the vertical dimensions,
+% giving them different slants at different sizes. Also, using the default
+% version in 'scripts produces symbols identical to those produced
+% elsewhere. Finally, the default takes the dots from the current text
+% font, whereas the \MD{} package takes them from the current math symbol
+% font, so they should match other dots in math mode. This is only really
+% a concern in \mtt{17pt} sizes or larger where \file{cmr17} is scaled to
+% get text dots, but \file{cmsy12} is scaled to get math dots.
+%
+% \DescribeMacro{\iddots}
+% \MD{} supplies also \cs{iddots}, with behavior similar to \cs{ddots},
+% except it produces the other diagonal. This command is not provided in
+% basic \TeX{} or \LaTeX{}.
 %
+% \DescribeMacro{\vdots}
 % \cs{vdots} produces vertical dots (as in standard \TeX{} or \LaTeX{}),
 % but again, the \MD{} version will change size with size
 % changing commands.
 %
+% \DescribeMacro{\fixedddots}
+% \DescribeMacro{\fixedvdots}
+% \DescribeMacro{\fixediddots}
 % The old behavior is stored in \cs{fixedddots} and \cs{fixedvdots}.
 % If you wish \cs{iddots} to behave similarly then use \cs{fixediddots}.
 %
 % The \cs{fixed...} versions try to maintain the (awful) default behavior,
 % which is to let the horizontal dimensions change size, as well as the
-% dots themselves, but hard code unchanging vertical dimensions.
+% dots themselves, but hard code unchanging vertical dimensions. The
+% `fixed' part of the name does not mean `corrected', but rather
+% `unchanging' or `frozen'.
 %
+% \DescribeMacro{\originalddots}
+% \DescribeMacro{\originalvdots}
+% \DescribeMacro{\originaliddots}
 % On the chance that some other package changes the behavior before \MD{}
 % is loaded, use \cs{originalddots}, \cs{originalvdots}, and
-% \cs{originaliddots} to obtain the versions in force before the \MD{} is
+% \cs{originaliddots} to obtain the versions in force before \MD{} is
 % loaded.
 %
-% If the \LaTeX{} and the \prog{amsmath} package are detected, the
-% commands \cs{dddot} and \cs{ddddot} are modified to get the same
-% size changing behavior in 'scripts. They also correct a positioning bug
-% in the \prog{amsmath} code. The correction requires the command
-% \cs{text}, and so is only applied if \prog{amsmath} has been loaded
-% \emph{before} \MD{}. The old amsmath commands are saved
-% in \cs{originaldddot} and \cs{originalddddot}. Similar comments apply
-% to plain \TeX{} with regard to \file{amstex.tex}.
+% \DescribeMacro{\dddot}
+% \DescribeMacro{\ddddot}
+% If the \prog{amsmath} package is detected in \LaTeX{}, the commands
+% \cs{dddot} and \cs{ddddot} are modified to get the same size changing
+% behavior in 'scripts. They also correct a positioning bug in the
+% \prog{amsmath} code. Compare the old and new versions and the 2-dot accent
+% (from basic \LaTeX{}): $\originaldddot{X},\dddot{X},\ddot{X}$. I have
+% also taken the liberty of changing the spacing between dots to match
+% that in the 2-dot accent.
 %
+% \DescribeMacro{\originaldddot}
+% \DescribeMacro{\originalddddot}
+% Some of the changes to \cs{dddot} and \cs{ddddot} require the command
+% \cs{text}, and so is only applied if \prog{amsmath} has been loaded
+% \emph{before} \MD{}. The old amsmath commands are saved in
+% \cs{originaldddot} and \cs{originalddddot}. Similar comments apply to
+% plain \TeX{} with regard to \file{amstex.tex}.
+%
+% \DescribeMacro{\MDoprekern}
+% \DescribeMacro{\MDodotkern}
+% \DescribeMacro{\MDopostkern}
+% Different math fonts have different metrics for the \cs{dot} and
+% \cs{ddot} accents. If you use a math font family other than the CM
+% fonts, you can adjust the positioning and spacing of the the dots in
+% \cs{dddot} and \cs{ddddot} to better match the spacing in \cs{ddot}
+% with the following length commands: \cs{MDoprekern}, \cs{MDodotkern} and
+% \cs{MDopostkern}. These adjust the space before the first dot, between
+% the dots and after the last dot, respectively. They are in units of
+% \texttt{mu}, which is $1/3$ the size of the space made by \cs{,}.
+% Reducing the first and/or increasing the last one cause the dots to be
+% shifted left. Reducing the middle one causes the dots to be closer
+% together. As there is usually a bit of white on either side of the dot
+% character, these usually have to be given negative values (so ``reduce''
+% means ``make more negative''). The defaults for the CM fonts are
+% accomplished by the commands
+% \begin{verbatim}
+%   \MDoprekern=0mu
+%   \MDodotkern=-1.3mu
+%   \MDopostkern=-1mu \end{verbatim}
+% I have found that the \prog{mathptmx} package (Times fonts in text and
+% math) benefits from changing these via the following commands
+% \begin{verbatim}
+%   \MDoprekern=-1mu
+%   \MDodotkern=-2mu
+%   \MDopostkern=0mu \end{verbatim}
+% \cs{MDoprekern}, \cs{MDodotkern}, and \cs{MDopostkern} were made
+% user-accessible commands in version 0.9.
+%
+% \DescribeMacro{\MDoddots}
+% \DescribeMacro{\MDovdots}
+% \DescribeMacro{\MDoiddots}
+% \DescribeMacro{\MDodddot}
+% \DescribeMacro{\MDoddddot}
 % In \LaTeX{}, all these commands have been made robust as of version 0.6.
+% On the chance that some later loaded package changes things, the \MD{}
+% versions can be obtained with \cs{MDoddots}, \cs{MDovdots},
+% \cs{MDoiddots}, \cs{MDodddot} and \cs{MDoddddot} (as of version 0.9).
 %
 % \section{The implementation}
 %
-% We define the version and date for plain\TeX{}. The file
-% \file{mathdots.sty} immediately switches to \file{mathdots.tex} if
-% \LaTeX{} is not being used.
-% \iffalse
-%<*dtx>
-% \fi
-%    \begin{macrocode}
-%<tex>\def\mathdotsfiledate{2006/02/07}%
-%<tex>\def\mathdotsfileversion{0.7}%
-%<sty>\ifx\documentclass\UndEfInEd\input mathdots \expandafter\endinput\fi
-%    \end{macrocode}
-% \iffalse
-%</dtx>
-% \fi
-%
 % The following are commands for plain \TeX{} that prevent double loading,
 % announce the package name, and change/restore the category of \mtt{@}.
 %    \begin{macrocode}
@@ -213,14 +310,14 @@
 %</tex>
 %    \end{macrocode}
 %
-% Now various shorthands. For example `\cs{MD@us}\cs{MD@t}' becomes
-% `\cs{textstyle}' and `\cs{MD@uf}\cs{MD@ss}' produces
-% `\cs{scriptscriptfont 0}'.
+% Now various shorthands. For example `\cs{MDo@us}\cs{MDo@t}' becomes
+% `\cs{textstyle}' and `\cs{MDo@uf}\cs{MDo@ss}' produces
+% `\cs{scriptscriptfont0}'.
 %    \begin{macrocode}
-\def\MD@us#1{\@nameuse{#1style}}%
-\def\MD@uf#1{\@nameuse{#1font}0}%
-\def\MD@t{text}\def\MD@s{script}\def\MD@ss{scriptscript}%
-\newdimen\MD@unit \MD@unit\p@
+\def\MDo@us#1{\@nameuse{#1style}}%
+\def\MDo@uf#1{\@nameuse{#1font}0}%
+\def\MDo@t{text}\def\MDo@s{script}\def\MDo@ss{scriptscript}%
+\newdimen\MDo@unit \MDo@unit\p@
 %    \end{macrocode}
 %
 % The following are the main utility macros to implement changes of style.
@@ -234,39 +331,39 @@
 %
 % We pass along the current style via \cs{everymath}.
 %
-% In \cs{MD@palette}, \arg1 is a command which we feed the current
-% style name. For example, \cs{MD@palette}\cs{MD@ddots} will be our
-% definition of \cs{ddots}. Then \cs{MD@ddots} will see the current style
+% In \cs{MDo@palette}, \arg1 is a command which we feed the current
+% style name. For example, \cs{MDo@palette}\cs{MDo@ddots} will be our
+% definition of \cs{ddots}. Then \cs{MDo@ddots} will see the current style
 % as its first argument.
 %    \begin{macrocode}
-\def\MD@changestyle#1{\relax\MD@unit0.1\fontdimen6\MD@uf{#1}%
-  \everymath\expandafter{\the\everymath\MD@us{#1}}}%
-\def\MD@dot{$\m@th\ldotp$}%
-\def\MD@palette#1{\mathchoice{#1\MD@t}{#1\MD@t}{#1\MD@s}{#1\MD@ss}}%
+\def\MDo@changestyle#1{\relax\MDo@unit0.1\fontdimen6\MDo@uf{#1}%
+  \everymath\expandafter{\the\everymath\MDo@us{#1}}}%
+\def\MDo@dot{$\m@th\ldotp$}%
+\def\MDo@palette#1{\mathchoice{#1\MDo@t}{#1\MDo@t}{#1\MDo@s}{#1\MDo@ss}}%
 %    \end{macrocode}
 %
 % In the following commands, \arg1 is the current style (supplied via
-% \cs{MD@palette}). Except for the change in the unit used and the dot
+% \cs{MDo@palette}). Except for the change in the unit used and the dot
 % used, the first two definitions are essentially the same as plain's
 % \cs{ddots} and \cs{vdots}, and the third is the same as the first with
-% the oxes reversed.
+% the boxes reversed.
 %    \begin{macrocode}
-\def\MD@ddots#1{{\MD@changestyle{#1}%
-  \mkern1mu\raise7\MD@unit\vbox{\kern7\MD@unit\hbox{\MD@dot}}%
-  \mkern2mu\raise4\MD@unit\hbox{\MD@dot}%
-  \mkern2mu\raise \MD@unit\hbox{\MD@dot}\mkern1mu}}%
-\def\MD@vdots#1{\vbox{\MD@changestyle{#1}%
-  \baselineskip4\MD@unit\lineskiplimit\z@
-  \kern6\MD@unit\hbox{\MD@dot}\hbox{\MD@dot}\hbox{\MD@dot}}}%
-\def\MD@iddots#1{{\MD@changestyle{#1}%
-  \mkern1mu\raise \MD@unit\hbox{\MD@dot}%
-  \mkern2mu\raise4\MD@unit\hbox{\MD@dot}%
-  \mkern2mu\raise7\MD@unit\vbox{\kern7\MD@unit\hbox{\MD@dot}}}}%
+\def\MDo@ddots#1{{\MDo@changestyle{#1}%
+  \mkern1mu\raise7\MDo@unit\vbox{\kern7\MDo@unit\hbox{\MDo@dot}}%
+  \mkern2mu\raise4\MDo@unit\hbox{\MDo@dot}%
+  \mkern2mu\raise \MDo@unit\hbox{\MDo@dot}\mkern1mu}}%
+\def\MDo@vdots#1{\vbox{\MDo@changestyle{#1}%
+  \baselineskip4\MDo@unit\lineskiplimit\z@
+  \kern6\MDo@unit\hbox{\MDo@dot}\hbox{\MDo@dot}\hbox{\MDo@dot}}}%
+\def\MDo@iddots#1{{\MDo@changestyle{#1}%
+  \mkern1mu\raise \MDo@unit\hbox{\MDo@dot}%
+  \mkern2mu\raise4\MDo@unit\hbox{\MDo@dot}%
+  \mkern2mu\raise7\MDo@unit\vbox{\kern7\MDo@unit\hbox{\MDo@dot}}}}%
 %    \end{macrocode}
 %
 % \DescribeMacro{\fixedddots}
 % \DescribeMacro{\fixedvdots}
-% We include current (2005/12/01) \LaTeX{}/plain\TeX{} definitions
+% We include current (as of 2014/06/11) \LaTeX{}/plain\TeX{} definitions
 % so a user can choose them. \LaTeX{} and plain\TeX{} don't have the other
 % diagonal, so we provide a \DescribeMacro{fixediddots}\cs{fixediddots}
 % which is just \cs{fixedddots} with the boxes reversed.
@@ -304,21 +401,33 @@
 % Here are the actual (re)definitions of these three commands. The
 % plain\TeX{} version emits messages similar to that produced by
 % \LaTeX{}'s \cs{DeclareRobustCommand}.
+%
+% Since \cs{ddots} et al.\ are defined in \LaTeX{} with
+% \cs{DeclareRobustCommand}, saving the original definition with
+% \verb$\let\originalddots\ddots$ will not work unless the internal
+% command ``\verb*$\ddots $'' is unchanged. Thus we can't use
+% \verb$\DeclareRobustCommand\ddots$ directly because it changes that
+% internal. Instead we define \cs{MDoddots} and then do
+% \verb$\let\ddots\MDoddots$
 %    \begin{macrocode}
-%<sty>\DeclareRobustCommand\ddots{%
-%<tex>\def\ddots{%
-  \mathinner{\MD@palette\MD@ddots}}%
-%<sty>\DeclareRobustCommand\iddots{%
-%<tex>\def\iddots{%
-  \mathinner{\MD@palette\MD@iddots}}%
-%<sty>\DeclareRobustCommand\vdots{%
-%<tex>\def\vdots{%
-  \mathinner{\MD@palette\MD@vdots}}%
+%<sty>\DeclareRobustCommand\MDoddots{%
+%<tex>\def\MDoddots{%
+  \mathinner{\MDo@palette\MDo@ddots}}%
+%<sty>\DeclareRobustCommand\MDoiddots{%
+%<tex>\def\MDoiddots{%
+  \mathinner{\MDo@palette\MDo@iddots}}%
+%<sty>\DeclareRobustCommand\MDovdots{%
+%<tex>\def\MDovdots{%
+  \mathinner{\MDo@palette\MDo@vdots}}%
 %<tex>\wlog{Mathdots Info: Redefining \string\ddots.}%
 %<tex>\wlog{Mathdots Info: Redefining \string\vdots.}%
+\let\ddots \MDoddots
+\let\iddots\MDoiddots
+\let\vdots \MDovdots
 %    \end{macrocode}
 %
-% Now we try to get AMS \cs{dddot} and \cs{ddddot} accents to behave.
+% Now we try to get AMS \cs{dddot} and \cs{ddddot} accents to behave
+% nicely.
 %
 % \DescribeMacro{\dddot}
 % \DescribeMacro{\ddddot}
@@ -326,49 +435,75 @@
 % \arg1. I don't know if the \cs{text} command is necessary, but it
 % doesn't seem to hurt. I included it in the original version because that
 % one used text dots and needed the text to change size. Now I use a math
-% symbol (the dot accent) and mathpalette to implement the change.
+% symbol (the dot accent) and \cs{mathpalette} to implement the change.
 % Still, I don't really understand how \cs{ex@} works in the AMS packages,
-% so I leave it in. For possible tuning, I have used two macros that
+% so I leave it in.
+% \DescribeMacro{MDoprekern}
+% \DescribeMacro{MDodotkern}
+% \DescribeMacro{MDopostkern}
+% For possible fine tuning, I have used two macros that
 % represent the math spacing placed before and after the sequence of
-% dots (\cs{MD@prekern} and \cs{MD@postkern}). Also, the actual dot is in
-% the definition of \cs{MD@D} where another spacing command follows the
-% dot (\cs{MD@dotkern}). In fact, by redefining \cs{MD@D} and possibly the
-% spacing, one can define other double accent commands.
+% dots (\cs{MDoprekern} and \cs{MDopostkern}). Also, the actual dot is in
+% the definition of \cs{MDo@D} where another spacing command follows the
+% dot (\cs{MDodotkern}). In fact, by redefining \cs{MDo@D} and changing
+% the spacing, one can define other multiple ``dot'' accent commands.
 %
 % \DescribeMacro{\originaldddot}
 % \DescribeMacro{\originalddddot}
 % We also save the original \prog{amsmath} versions.
 %    \begin{macrocode}
-\newbox\MD@dotsbox
-\newmuskip\MD@prekern
-\newmuskip\MD@postkern
-\newmuskip\MD@dotkern
-\MD@prekern 0mu
-\MD@postkern -1mu
-\MD@dotkern -1.3mu
-\def\MD@dotsaccent#1{\mathpalette{\MD@@dotsaccent{#1}}}%
-\def\MD@@dotsaccent#1#2#3{%
-  {\setbox\MD@dotsbox\hbox{$#2\mkern\MD@prekern#1\mkern\MD@postkern$}%
-   \mathop{#3\kern\z@}\limits^{%
-   \text{\vbox to-1.4\ex@{\kern-1.8\ex@\copy\MD@dotsbox\vss}}}}%
-}%
+\newbox\MDo@dotsbox
+\newmuskip\MDoprekern
+\newmuskip\MDopostkern
+\newmuskip\MDodotkern
+\MDoprekern 0mu
+\MDopostkern -1mu
+\MDodotkern -1.3 mu
+\def\MDo@dotsaccent#1{\mathpalette{\MDo@@dotsaccent{#1}}}%
 \let\originaldddot \dddot
 \let\originalddddot\ddddot
 %<*sty>
+\def\MDo@@dotsaccent#1#2#3{%
+  {\setbox\MDo@dotsbox\hbox{$#2\mkern\MDoprekern#1\mkern\MDopostkern$}%
+   \mathop{#3\kern\z@}\limits^{%
+   \text{\vbox to-1.4\ex@{\kern-1.8\ex@\copy\MDo@dotsbox\vss}}}}%
+}%
 \@ifpackageloaded{amsmath}{%
-  \def\MD@D{\mathchar"5F\mkern\MD@dotkern}%
-  \DeclareRobustCommand\dddot {\MD@dotsaccent{\MD@D\MD@D\MD@D}}%
-  \DeclareRobustCommand\ddddot{\MD@dotsaccent{\MD@D\MD@D\MD@D\MD@D}}%
+  \def\MDo@D{\mathchar"5F\mkern\MDodotkern}%
+  \DeclareRobustCommand\MDodddot {%
+    \MDo@dotsaccent{\MDo@D\MDo@D\MDo@D}}%
+  \DeclareRobustCommand\MDoddddot{%
+    \MDo@dotsaccent{\MDo@D\MDo@D\MDo@D\MDo@D}}%
+  \let\dddot \MDodddot
+  \let\ddddot\MDoddddot
 }{}%
 %</sty>
+%    \end{macrocode}
+%
+% For plain tex + amstex: \cs{ex@} is set in \file{amstex.tex} to
+% \texttt{.2326ex}, so its value depends on the value of \texttt{ex} when
+% \prog{amstex} is loaded. Unlike \prog{amsmath}, it doesn't seem to be
+% recalculated with size changes. Thus, in plain TeX we add a reset.
+%    \begin{macrocode}
 %<*tex>
-\def\amstexfmtname{AmS-TeX}%
-\ifx\amstexfmtname\fmtname
-  \def\MD@D{\mathchar"5F\mkern\MD@dotkern}%
-  \wlog{Mathdots Info: Redefining \string\dddot.}%
-  \wlog{Mathdots Info: Redefining \string\ddddot.}%
-  \def\dddot {\MD@dotsaccent{\MD@D\MD@D\MD@D}}%
-  \def\ddddot{\MD@dotsaccent{\MD@D\MD@D\MD@D\MD@D}}%
+\def\MDo@@dotsaccent#1#2#3{%
+{%
+  \setbox\MDo@dotsbox\hbox{$#2\mkern\MDoprekern#1\mkern\MDopostkern$}%
+  \mathop{#3\kern\z@}\limits^{\text{\ex@.2326ex
+    \vbox to-1.4\ex@{\kern-1.8\ex@\copy\MDo@dotsbox\vss}}}%
+}}%
+\ifx\amstexloaded@\relax % true if amstex has been loaded
+  % A backup test in case \amstexloaded@ became \relax by accident:
+  \ifx\ddddot\UndEfInEd
+  \else
+    \def\MDo@D{\mathchar"5F\mkern\MDodotkern}%
+    \wlog{Mathdots Info: Redefining \string\dddot.}%
+    \wlog{Mathdots Info: Redefining \string\ddddot.}%
+    \def\MDodddot {\MDo@dotsaccent{\MDo@D\MDo@D\MDo@D}}%
+    \def\MDoddddot{\MDo@dotsaccent{\MDo@D\MDo@D\MDo@D\MDo@D}}%
+    \let\dddot \MDodddot
+    \let\ddddot\MDoddddot
+  \fi
 \fi
 \csname MathDots\mathdotsfileversion\endcsname
 %</tex>