% \iffalse
%% Style file `arydshln'
%% Copyright (C) 1993,1998-2004
%%       Hiroshi Nakashima <nakasima'AT'tutics'DOT'tut'DOT'ac'DOT'jp>
%%       (Kyoto Univ. / Toyohashi Univ. of Tech.)
%% This program can be redistributed and/or modified under the terms
%% of the LaTeX Project Public License Distributed from CTAN
%% archives in directory macros/latex/base/lppl.txt; either
%% version 1 of the License, or any later version.
% \fi
%
% \CheckSum{3605}
%% \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
%%   Digits        \0\1\2\3\4\5\6\7\8\9
%%   Exclamation   \!     Double quote  \"     Hash (number) \#
%%   Dollar        \$     Percent       \%     Ampersand     \&
%%   Acute accent  \'     Left paren    \(     Right paren   \)
%%   Asterisk      \*     Plus          \+     Comma         \,
%%   Minus         \-     Point         \.     Solidus       \/
%%   Colon         \:     Semicolon     \;     Less than     \<
%%   Equals        \=     Greater than  \>     Question mark \?
%%   Commercial at \@     Left bracket  \[     Backslash     \\
%%   Right bracket \]     Circumflex    \^     Underscore    \_
%%   Grave accent  \`     Left brace    \{     Vertical bar  \|
%%   Right brace   \}     Tilde         \~}
%%
% \iffalse
%%
%<*dtx>
\ProvidesFile{arydshln.dtx}
%</dtx>
%<package>\def\next{LaTeX2e}
%<package>\ifx\fmtname\next\def\next{
%<package>\NeedsTeXFormat{LaTeX2e}[1994/12/01]
%<package>\ProvidesPackage{arydshln}}
%<package>\else\def\next[#1]{}\fi\next
%<driver>\ProvidesFile{arydshln-man.tex}
% \fi
% \ProvidesFile{arydshln.dtx}
[2004/08/31 v1.71 ]
%
% \iffalse
%<*driver>
\documentclass{ltxdoc}
\usepackage{arydshln}
\DisableCrossrefs
\PageIndex
\CodelineNumbered
\RecordChanges
%</driver>
%<driver>\OnlyDescription
%<driver>\def\ONLYDESCRIPTION{}
%<*driver>
\advance\textwidth4em
\begin{document}
\DocInput{arydshln.dtx}
\end{document}
%</driver>
% \fi
% \MakeShortVerb{\|}
% \GetFileInfo{arydshln.dtx}
% \IndexPrologue{\section*{Index}}
% \setcounter{IndexColumns}{2}
% \def\th{^{\it th}}
% \def\LOR{{\ {\lor}\ }}
% \def\LAND{{\ {\land}\ }}
% \let\togets\leftrightarrow
% \let\epsilon\varepsilon
% \def\cs#1{\hbox{\tt\char\escapechar#1}}
%
% \makeatletter
% \def\Meta#1{\penalty\@highpenalty \hbox{{\tt\char`\{}\meta{#1}{\tt\char`\}}}}
%
% \def\eqnarray{\stepcounter{equation}\let\@currentlabel=\theequation
% \global\@eqnswtrue
% \global\@eqcnt\z@\tabskip\@centering\let\\=\@eqncr
% $$\halign to \displaywidth\bgroup\hskip\@centering
%   $\displaystyle\tabskip\z@{##}$\@eqnsel&\global\@eqcnt\@ne
%   \hfil$\;{##}\;$\hfil
%   &\global\@eqcnt\tw@ $\displaystyle\tabskip\z@{##}$\hfil
%    \tabskip\@centering&\llap{##}\tabskip\z@\cr}
%
% \let\latex@addmarginpar\@addmarginpar
% \let\latex@latex@warning@no@line\@latex@warning@no@line
% \def\@addmarginpar{\let\@latex@warning@no@line\@gobble
%	\latex@addmarginpar
%	\let\@latex@warning@no@line\latex@latex@warning@no@line}
% \let\@font@warning\@gobble
%
% \long\def\m@cro@#1#2{\endgroup \topsep\MacroTopsep \trivlist
%    \edef\saved@macroname{\string#2}

%   \def\makelabel##1{\llap{##1\hskip\@totalleftmargin}}

%   \if@inlabel
%     \let\@tempa\@empty \count@\macro@cnt
%     \loop \ifnum\count@>\z@
%       \edef\@tempa{\@tempa\hbox{\strut}}\advance\count@\m@ne \repeat
%     \edef\makelabel##1{\llap{\vtop to\baselineskip
%                                {\@tempa\hbox{##1}\vss}

%			       \hskip\@totalleftmargin}}

%     \advance \macro@cnt \@ne
%   \else  \macro@cnt\@ne  \fi
%   \edef\@tempa{\noexpand\item[

%      #1

%        \noexpand\PrintMacroName
%      \else
%        \noexpand\PrintEnvName
%      \fi
%      {\string#2}]}

%   \@tempa
%   \global\advance\c@CodelineNo\@ne
%    #1

%       \SpecialMainIndex{#2}\nobreak
%       \DoNotIndex{#2}

%    \else
%       \SpecialMainEnvIndex{#2}\nobreak
%    \fi
%   \global\advance\c@CodelineNo\m@ne
%   \ignorespaces}
%
% \def\DescribeOpt{\leavevmode\@bsphack\begingroup\MakePrivateLetters
%   \Describe@Opt}
% \def\Describe@Opt#1{\endgroup \Describe@pt{#1}}
% \def\Describe@pt#1#2{

%               \marginpar{\raggedleft\PrintDescribeMacro{#1}}

% 		\SpecialOptUsageIndex{#1}{#2}}
% \def\SpecialOptUsageIndex#1#2{

% 	\index{#1\actualchar{\protect\ttfamily#1} (#2)\encapchar usage}

%	\index{#2s:\levelchar{\protect\ttfamily#1}\encapchar usage}

%	\@esphack\ignorespaces}
% \def\SpecialOptMainIndex#1#2{

% 	\index{#1\actualchar{\protect\ttfamily#1} (#2)\encapchar main}

%	\index{#2s:\levelchar{\protect\ttfamily#1}\encapchar main}

%	\@esphack\ignorespaces}
% \def\SpecialOptIndex#1#2{

% 	\index{#1\actualchar{\protect\ttfamily#1} (#2)}

%	\index{#2s:\levelchar{\protect\ttfamily#1}}

%	\@esphack\ignorespaces}
% \def\SpecialOptIndexS#1#2{

% 	\index{#1\actualchar{\protect\ttfamily#1} (#2)}

%	\index{#2es:\levelchar{\protect\ttfamily#1}}

%	\@esphack\ignorespaces}
%
% \def\SpecialMainIndex#1{\@bsphack
%	\RemoveAtFromCS{#1}

% 	\protected@edef\@tempa{\noexpand\special@index{\RAFCtempa
%		\actualchar \string\verb
% 		\quotechar*\verbatimchar\string#1\verbatimchar
%		\encapchar main}}\@tempa
%	\@esphack}
% 
% \def\SpecialIndex#1{\@bsphack
%	\RemoveAtFromCS{#1}

% 	\protected@edef\@tempa{\noexpand\special@index{\RAFCtempa
%		\actualchar \string\verb
% 		\quotechar*\verbatimchar\string#1\verbatimchar}}\@tempa
%	\@esphack}
% 
% \def\Macro{\begingroup \@sanitize \iMacro}
% \def\iMacro#1{\endgroup
%	\def\SpecialMainIndex##1{\SWiIndex{##1}{\encapchar main}}

%	\macro{#1}}
% \let\endMacro\endmacro
% 
% \def\SpecialWeirdIndex{\@bsphack \begingroup \@sanitize \SWIndex{}}
% \def\SWIndex#1#2{\endgroup \SWiIndex{#2}{#1}}
% \def\SWiIndex#1#2{%
%	\protected@edef\@tempa{\noexpand\special@index{

%		#1\actualchar \string\verb
%		\quotechar*\verbatimchar#1\verbatimchar#2}}\@tempa
% 	\@esphack}
% 
% \let\RAFCempty\@empty
% \let\RAFCgobble\@gobble
% \makeatother
% \def\RemoveAtFromCS#1{\edef\RAFCtempa{\expandafter\RAFCgobble\string#1}

% 	\edef\RAFCtempa{\expandafter\string\csname\RAFCtempa\endcsname}

%	\expandafter\iRemoveAtFromCS\RAFCtempa@\NIL}
% \def\iRemoveAtFromCS#1#2@#3\NIL{\def\RAFCtempb{#3}

% 	\ifx\RAFCtempb\RAFCempty \def\RAFCtempa{#2}\let\next\relax
%	\else \def\RAFCtempa{#2@}\def\next{\iiRemoveAtFromCS#3\NIL}

% 	\def\RAFCtempb{#2}

%	\ifx\RAFCtempb\RAFCempty
%	\else\ifx\RAFCtempb\CommonCSPrefix
%	\else\ifx\RAFCtempb\CommonIfPrefix
%	\else \def\RAFCtempa{}\def\next{\xRemoveAtFromCS#2@#3\NIL}\fi\fi\fi\fi
%	\next}
% \def\iiRemoveAtFromCS#1{\ifx#1@\edef\RAFCtempa{\RAFCtempa@}

%	\let\next\iiRemoveAtFromCS
%	\else \def\next{\xRemoveAtFromCS#1}\fi
% 	\next}
% \def\xRemoveAtFromCS#1@\NIL{\edef\RAFCtempa{#1\RAFCtempa}}
% \makeatletter
% 
% \begingroup
% \gdef\@sverbcat{\catcode`\@\active\catcode`\#\active}
% \catcode`\@\active \gdef@{\char`\@\penalty\exhyphenpenalty}
% \catcode`\#\active \gdef#{\penalty\exhyphenpenalty}
% \endgroup
% \let\latex@sverb\@sverb
% \def\@sverb{\@sverbcat\latex@sverb}
%
% \newcount\ChangesCounter \ChangesCounter10000
% \def\changes@#1#2#3{\global\advance\ChangesCounter\@ne
% 	\protected@edef\@tempa{\noexpand\glossary{#1\levelchar
%		x\number\ChangesCounter
%		\ifx\saved@macroname\@empty
%			\actualchar\generalname
%		\else
%			\expandafter\@gobble\saved@macroname
%			\actualchar \string\verb
%			\quotechar*\verbatimchar \saved@macroname
%			\verbatimchar
%		\fi
%		:\levelchar #3}}\@tempa\endgroup\@esphack}
% \def\theglossary{\onecolumn \glossary@prologue
%	\GlossaryParms \let\item\@idxitem \ignorespaces}
% \def\endtheglossary{\relax}
% 
% \makeatother
% \errorcontextlines1000
%
% 
% 
% \changes{v1.0}{1993/04/01}
%	{The style was born on a good day \ldots}
% \changes{v1.05}{1993/06/18}
%	{Cope with \texttt{\char`\\\char`\\} with negative optional
%	vertical space.}
% \changes{v1.1}{1993/06/24}
%	{Save and restore the \texttt{\char`\\catcode} for
%	`\texttt{@}'.}
% \changes{v1.2-1}{1998/07/16}
%	{Add this document.}
% \changes{v1.2-2}{1998/07/16}
%	{Cope with \protect\LaTeXe.}
% \changes{v1.2-3}{1998/07/16}
% 	{Allow mixture of vertical solid- and dash-lines.}
% \changes{v1.2-4}{1998/07/16}
% 	{Add the feature of explicit dash/gap specification.}
% \changes{v1.2-5}{1998/07/16}
% 	{Fix some bugs and change codes.}
% \changes{v1.4-1}{1999/06/25}
%	{Make compatible with array package and add new features.}
% 
% \changes{v1.4-2}{1999/06/25}
%	{The following are changes of this document.}
% 
% \changes{v1.4-3-1}{1999/06/25}
% 	{The following are for the general compatibility with \textsf{array}.}
% \changes{v1.4-3-2}{1999/06/25}
% 	{The following are to control the effective width of vertical lines.}
% \changes{v1.4-3-3}{1999/06/25}
% 	{The following are for inactivation of dash-line functions.}
% \changes{v1.4-3-4}{1999/06/25}
% 	{The following are for drawing mode to cope with the bug of
%	\cs{xlearders}.}
% \changes{v1.4-3-5}{1999/06/25}
%	{The following are to implement dashed version of \cs{firsthline}
%	and \cs{lasthline} of \textsf{array}.}
% \changes{v1.4-3-6}{1999/06/25}
%	{The following are to fix the bug by which the depth of
%	\texttt{array}\slash\texttt{tabular} was always zero.}
% \changes{v1.4-3-7}{1999/06/25}
%	{The following are to rename macros for \cs{cdashline}.}
% \changes{v1.4-3-8}{1999/06/25}
%	{The following are to cope with very narrow or negative wide columns.}
% \changes{v1.4-4}{2000/07/04}
%	{Released to CTAN.}
%
% \changes{v1.5-1}{2000/07/12}
%	{The following are for the compatibility with \textsf{colortab}.}
% \changes{v1.5-2}{2002/06/26}
%	{The following are for bug fix of \cs{adl@putlrc}.}
% \changes{v1.5-3}{2003/07/11}
%	{The following are for bug fix of \cs{adl@inactivate}.}
% \changes{v1.54-1}{2003/08/25}
%	{The following are for bug fix of \cs{adl@@vl}.}
% \changes{v1.54-2}{2003/08/25}
%	{The following are for bug fix of activation.}
% \changes{v1.54-3}{2003/08/25}
%	{The following are miscellaneous modifications.}
%
% \changes{v1.6-1}{2003/08/25}
%	{The following are for the compatibility with \textsf{longtable}.}
%
% \changes{v1.7}{2003/05/21}
%	{The following are for the compatibility with \textsf{colortbl}.}
% \changes{v1.7-1}{2004/5/21}
%	{The following are for null-wide horizontal lines.}
% \changes{v1.7-2}{2004/5/21}
%	{The following are to fix the bug of \cs{arrayrulecolor} etc. in
%	\textsf{colortbl}.}
% \changes{v1.7-3}{2004/5/21}
% 	{The following are for vertical line coloring.}
%
% \changes{v1.71-1}{2004/7/31}
% 	{The following are for bug fix for \textsf{array}'s
% 	\texttt{m}-columns.}
% 
%
% \def\EXCLAMATION{\texttt{!}}
% \def\VERTBAR{\texttt{\char`\|}}
% \edef\CommonCSPrefix{\expandafter\RAFCgobble\string\adl}
% \edef\CommonIfPrefix{\expandafter\RAFCgobble\string\ifadl}
% 
% \title{The \textsf{arydshln} package\thanks{

%	This file has version number \fileversion, last revised \filedate.}}
% \author{Hiroshi Nakashima\\(Toyohashi Univ. of Tech.)}
% \date{\filedate}
% \maketitle
%
% \begin{abstract}
% This file gives \LaTeX's |array| and |tabular| environments the capability
% to draw horizontal\slash vertical dash-lines.
% \end{abstract}
%
% \tableofcontents
% \newpage
%
% 
%
% \section{Introduction}
% \label{sec:intro}
%
% In January 1993, Weimin Zhang kindly posted a style \textsf{hvdashln}
% written by the author, which draws horizontal\slash vertical dash-lines in
% \LaTeX's |array| and |tabular| environments, to the news group
% \textsf{comp.text.tex}.  The style, unfortunately, has a known problem
% that vertical lines are broken when an array contains tall rows.
%
% In March of the year, Monty Hayes complained of this problem encouraging
% the author to make a new version \textsf{arydshln} to solve the problem.
% The new style also has new features, such as allowing `|:|' to specify
% vertical dash-line in preamble, and |\cdashline| being a counterpart of
% |\cline|.
%
% \changes{v1.4-2}{1999/06/25}
%	{The history on the compatibility with \textsf{array} package.}
% In March 1999, Sebastian Rahtz kindly invited the style, which had been
% improved following the bug report from Takahiro Kubota, to be included in
% \TeX{} CTAN and also in the online catalogue compiled by Graham Williams.
% This invitation gave the style new users including Peter Ehrbar who wished 
% to use it with \textsf{array} style in Standard \LaTeX{} Tools Bundle and
% had trouble because these styles were incompatible with each other.
% Therefore, the style became compatible with \textsf{array} and got
% additional new features.
% 
% \changes{v1.5-1}{2000/7/12}
%	{The history on the compatibility with \textsf{colortab} package.}
% In February 2000, Zsuzsanna Nagy reported that \textsf{arydshln} is not
% compatible with \textsf{colortab} style to let the author work on the
% compatibility issue again.
% 
% \changes{v1.6-1}{2003/8/25}
%	{The history on the compatibility with \textsf{longtable} package.}
% In Feburary 2001, Craig Leech reported another compatibility problem with
% \textsf{longtable}.  Although the author promised that the problem would
% be attacked some day, the issue had left long time\footnote{

%

% Two years and a half!  Sorry Craig.}
%

% until three other complaints.  Then the author attacked the problem hoping 
% it is the last compatibility issue\footnote{

%

% But his hope was dashed as described below.}.
%
% \changes{v1.7}{2004/5/21}
%	{The history on the compatibility with \textsf{colortbl} package.}
% In May 2004, Klaus Dalinghaus found another incompatibility with
% \textsf{colortbl}. Although he was satisfied by a quick hack for cell
% painting, the author attacked a harder problem for line
% coloring to solve the problem\footnote{

%

% Without dreaming it is the last compatibility issue.}.
%
% 
%
% \section{Usage}
% \label{sec:usage}
% \subsection{Loading Package}
% \label{sec:usage-load}
% 
% \changes{v1.4-2}{1999/06/25}
%	{Explanation of package loading is added.}
% 
% The package is usable to both \LaTeXe{} and \LaTeX-2.09 users with their
% standard package loading declaration.  If you use \LaTeXe, simply do the
% following.
% \begin{quote}
% \SpecialUsageIndex{\usepackage}
% |\usepackage{arydshln}|
% \end{quote}
% If you still love \LaTeX-2.09, the following is what you have to do.
% \begin{quote}
% \SpecialUsageIndex{\documentstyle}
% |\documentstyle[..,arydshln,...]|\Meta{style}
% \end{quote}
%
% \changes{v1.5-1}{2000/7/12}
%	{Caution about loading order of \textsf{colortab} is added.}
% \changes{v1.6-1}{2003/8/25}
%	{Caution about loading order of \textsf{longtable} is added.}
% \changes{v1.7}{2004/5/21}
%	{Caution about loading order of \textsf{colortbl} is added.}
% \SpecialOptIndex{array}{package}
% \SpecialOptIndex{colortab}{package}
% \SpecialOptIndex{longtable}{package}
% \SpecialOptIndex{colortbl}{package}
%

% Only one caution given to users of \textsf{array} (v2.3m or later) and
% \textsf{longtable} (v4.10 or later) packages, included in Standard \LaTeX{}
% Tools Bundle, and \textsf{colortab} and \textsf{colortbl} package is that
% \textsf{arydshln} has to be loaded {\em after} \textsf{array},
% \textsf{longtable}, \textsf{colortab} and/or \textsf{colortbl} done.  That
% is, the following is correct but reversing the order of |\usepackage| will
% cause some mysterious error.
%

% \begin{quote}
% |\usepackage{array}     % and/or|\\
% |\usepackage{longtable} % and/or|\\
% |\usepackage{colortab}  % or|\\
% |\usepackage{colortbl}|\\
% |\usepackage{arydshln}|
% \end{quote}
%
% 
% 
% \subsection{Basic Usage}
% \label{sec:usage-basic}
%
% \DescribeEnv{array}\DescribeEnv{tabular}
% \SpecialIndex{\hline}
% \SpecialIndex{\cline}
% \SpecialIndex{\multicolumn}
% You can simply use |array| or |tabular(*)| environments with standard
% preamble, such as \verb!{r|c|ll}!, and standard commands |\\|, |\hline|,
% |\cline| and |\multicolumn|.
%
% \leavevmode\marginpar{\raggedleft\PrintDescribeMacro{:}}

% \index{:={\protect\ttfamily:}|usage}

% Drawing a vertical dash-line is quite simple.  Use `|:|' in the preamble
% as the separator of columns separated by the dash-line, just like using
% `\verb!|!' to draw a vertical solid-line.  The {\em preamble} means not
% only that of the environment, but also the first argument of
% |\multicolumn|.
%
% \DescribeMacro{\hdashline}\DescribeMacro{\cdashline}
% \changes{v1.4-2}{1999/06/25}
%	{Description of \cs{first}\slash\texttt{lasthdashline} is added.}
% It is also simple to draw a horizontal dash-line.  Use |\hdashline| and
% |\cdashline| as the counterparts of |\hline| and |\cline|.
%
% For example;
%

%\begin{quote}
%\begin{verbatim}
%\begin{tabular}{|l::c:r|}\hline
%A&B&C\\\hdashline
%AAA&BBB&CCC\\\cdashline{1-2}
%\multicolumn{2}{|l:}{AB}&C\\\hdashline\hdashline
%\end{tabular}
%\end{verbatim}
%\end{quote}
%

% will produce the following result.
%

%\begin{quote}
%\catcode`\|12
%\begin{tabular}{|l::c:r|}\hline
%A&B&C\\\hdashline
%AAA&BBB&CCC\\\cdashline{1-2}
%\multicolumn{2}{|l:}{AB}&C\\\hdashline\hdashline
%\end{tabular}
%\end{quote}
%
% \DescribeMacro{\firsthdashline}\DescribeMacro{\lasthdashline}
% If you use \textsf{array}, the dashed version of |\firsthline| and
% |\lasthline| named |\first#hdash#line| and |\lasthdashline| are available.
% 
% 
%
% \subsection{Style Parameters}
% \label{sec:usage-par}
%
% \DescribeMacro{\dashlinedash}\DescribeMacro{\dashlinegap}
% You have two style parameters to control the shape of dash-lines:
% |\dashlinedash| is for the length of each dash segment in a dash line;
% |\dashlinegap| controls the amount of each gap between dash segments.
% Both parameters have a common default value, 4\,pt.
%
% 
%
% \subsection{Fine Tuning}
% \label{sec:usage-fine}
%
% \leavevmode\marginpar{\raggedleft\PrintDescribeMacro{;}}

% \index{;={\protect\ttfamily;}|usage}

% Although you can control the shape of dash-lines in an
% |array|\slash|tabular| environment as described in \S\ref{sec:usage-par},
% you might want to draw a dash-line of a shape different from others.  To
% specify the shape of a vertical dash-line explicitly, you may use;
%

% \begin{quote}
% |;{|\meta{dash}|/|\meta{gap}|}|
% \end{quote}
%

% instead of ordinary `|:|' and will have a dash-line with dash segments of
% \meta{dash} long separated by spaces of \meta{gap}.
%
% \DescribeMacro{\hdashline}\DescribeMacro{\cdashline}
% As for horizontal dash-lines, explicit shape specifications may be given
% through optional arguments of |\hdashline| and |\cdashline| as follows.
%

% \begin{quote}
% |\hdashline[|\meta{dash}|/|\meta{gap}|]|\\
% |\cdashline{|\meta{col1}|-|\meta{col2}|}[|\meta{dash}|/|\meta{gap}|]|
% \end{quote}
%

%
% For example;
%

%\begin{quote}
%\begin{verbatim}
%\begin{tabular}{|l::c;{2pt/2pt}r|}\hline
%A&B&C\\\hdashline[1pt/1pt]
%AAA&BBB&CCC\\\cdashline{1-2}[.4pt/1pt]
%\multicolumn{2}{|l;{2pt/2pt}}{AB}&C\\\hdashline\hdashline
%\end{tabular}
%\end{verbatim}
%\end{quote}
%

% will produce the following result.
%

%\begin{quote}
%\catcode`\|12
%\begin{tabular}{|l::c;{2pt/2pt}r|}\hline
%A&B&C\\\hdashline[1pt/1pt]
%AAA&BBB&CCC\\\cdashline{1-2}[.4pt/1pt]
%\multicolumn{2}{|l;{2pt/2pt}}{AB}&C\\\hdashline\hdashline
%\end{tabular}
%\end{quote}
%
% \changes{v1.4-2}{1999/06/25}
%	{Description of the real width of vertical lines is added.}
% 
% \DescribeMacro{\ADLnullwide}
% \DescribeMacro{\ADLsomewide}
% \SpecialIndex{\arrayrulewidth}
%

% The vertical solid and dashed lines are drawn as if their width is zero, as 
% standard \LaTeX's |array| and |tabular| do, if you don't use
% \textsf{array} package.  Otherwise, they have {\em real} width of
% |\arrayrulewidth| as the authors of \textsf{array} prefers.  However, you
% may explicitly tell \textsf{arydshln} to follow your own preference by
% |\ADLnullwide| if you love \LaTeX{} standard, or |\ADLsomewide| if
% you second the preference of \textsf{array} authors.
% 
% 
%
% \subsection{Finer Tuning}
% \label{sec:usage-finer}
% 
% \changes{v1.4-2}{1999/06/25}
%	{Description of drawing mode is added.}
% To draw dash-lines, we use a powerful primitive of \TeX{} called |\xleaders|.
% It replicates a segment that consist of a dash and gap so that a dash-line
% has as many segments as possible and distributes {\it remainder} space to
% make the spaces between adjacent dash segments (almost) equal to each
% other.  Therefore, you will have dash-lines with consistent steps of gaps
% and spaces as the left and upper lines in Figure~\ref{fig:mode}(1) are.
% 
% \begin{figure}[b]
% \catcode`\|12
% \tabcolsep0pt \setbox\strutbox\hbox{\vrule height14pt depth6pt width0pt}
% \def\mb{\makebox[20pt]}
% \dashlinedash3.01pt \dashlinegap3.01pt
% \def\example#1{\begin{tabular}[b]{|c:c;{2pt/2pt}c|}
%	\hline\noalign{\vskip-\arrayrulewidth}
%	\mb{A}&\mb{A}&\mb{A}\\[-\arrayrulewidth]\hdashline
%	\mb{B}&\mb{B}&\mb{B}\\[-\arrayrulewidth]\hdashline[2pt/2pt]
%	\mb{C}&\mb{C}&\mb{C}\\[-\arrayrulewidth]\hline
% 	\multicolumn3c{(#1)}
%	\end{tabular}}
%
% \centerline{\example1\qquad
%	\ADLdrawingmode{2}\example2\qquad
%	\ADLdrawingmode{3}\example3}
% \caption{Drawing mode controlled by \cs{ADLdrawingmode}}
% \label{fig:mode}
% \end{figure}
%
% However, because of a bug (or buggy feature) of |\xleaders|, there is a
% small possibility that a dash segment near the right\slash bottom end
% drops as right and lower lines in (1) of the figure shows.  To cope with this
% problem, you may change the {\it drawing mode} by
% |\ADLdrawingmode|\Meta{$m$} as follows.
%

% \begin{itemize}
% \item $m=1$
% \DescribeMacro{\ADLdrawingmode}\\
% As shown in Figure~\ref{fig:mode}(1), most beautiful in almost all cases
% as the left/upper lines, but has a small possibility to produce an ugly
% result as right/lower lines.  This is default.
%
% \item $m=2$\\
% As shown in (2) of the figure, beautiful if dash-lines are not so sparse
% as right/lower lines, but dash segments near the both ends may be a little 
% bit too long as left/upper lines.
% 
% \item $m=3$\\
% As shown in (3) of the figure, beautiful if dash-lines are not so sparse
% as right/lower lines, but gaps near the both ends may be considerably too
% large as left/upper lines.
% \end{itemize}
%

% It is recommended to use default mode 1 unless you have an ugly result
% in the final version of your manuscript, because the correctness of mode 1
% is very sensitive to the length of dash-lines.
% 
% 
%
% \subsection{Performance Tuning}
% \label{sec:usage-performance}
% 
% \changes{v1.4-2}{1999/06/25}
%	{Description of (in)activation is added.}
%
% Since drawing dash-lines is a hard job, you have to be patient with the
% fact that the performance of typesetting |array|\slash|tabular| with
% dash-lines is poorer than that of ordinary ones.  In fact, according to
% author's small performance evaluation with a |tabular| having nine
% vertical and ten horizontal dash-lines, typesetting the |tabular| is
% approximately ten times as slow as its ordinary counterpart with solid
% lines.
%
% However, this is not a really bad news, unfortunately.  The real one is
% that loading \textsf{arydshln} makes typesetting |array|\slash|tabular|
% slower even if they only have solid lines which the package treats as
% special ones of dash-lines.  The evaluation result shows the degradation
% factor is about nine.  Therefore, if your document has many
% |array|\slash|tabular| with solid lines, \LaTeX{} will run slowly even 
% with quite few (or no) |array|\slash|tabular| with dash-lines, 
%
% \DescribeMacro{\ADLinactivate}
% To cope with this problem, you may inactivate dash-line functions by the
% command |\ADLinactivate| that replaces dash-lines with solid
% lines drawn by a faster (i.e. ordinary) mechanism.  Although the
% inactivation does not completely solve the performance problem, the
% degradation factor will become much smaller and acceptable, approximately
% 1.5 in the author's evaluation.  For example, the draft version of your
% document will have the command in its preamble, which you will remove from
% your final version.
% 
% \DescribeMacro{\ADLactivate}
% Alternatively, you may do |\ADLinactivate| in the preamble, switch on by
% |\ADLactivate| before you really need dash-lines, and switch off again
% afterword.  A wiser way could be surrounding |array|\slash|tabular| by
% |\begin{ADLactivate}| and |\end{ADLactivate}|.
%
% \DescribeEnv{Array}
% \DescribeEnv{Tabular}
% If you feel it tiresome to type the long command\slash environment name for
% the activation, you may use |Array| and |Tabular(*)| environment in which
% dash-line functions are always active.  Note that, however, since these
% environment names are too natural to keep them from being used by authors
% of other packages or yourself, name conflict could occur.  If |Array|
% and/or |Tabular| have already been defined when \textsf{arydshln} is
% loaded, you will get a warning to show you have to define new
% environments, say |dlarray| and |dltabular|, as follows.
%

% \begin{itemize}\item[]
% |\newenvironment{dlarray}{\ADLactivate\begin{array}}%|\\
% |                        {\end{array}}|\\
% |\newenvironment{dltabular}{\ADLactivate\begin{tabular}}%|\\
% |                          {\end{tabular}}|\\
% |\newenvironment{dltabular*}{\ADLactivate\begin{tabular*}}%|\\
% |                           {\end{tabular*}}|
% \end{itemize}
%
% \begingroup
% \let\PDM\PrintDescribeMacro
% \def\PrintDescribeMacro#1{\leavevmode\llap{\PDM{#1}}}
% \DescribeMacro{\ADLnoshorthanded} 
% On the other hand, if they are defined after \textsf{arydshln} is loaded,
% their definitions are silently replaced or \LaTeX{} complains of multiple
% definitions.  The error in the latter case will be avoided by putting
% |\ADLnoshorthanded| just after |\usepackage{arydshln}|.
% \par
% \endgroup
% 
% 
%
% \subsection{Compatibility with Other Packages}
% \label{sec:usage-compat}
% 
% \changes{v1.5-1}{2000/07/12}
%	{Section 2.7 is added.}
% \changes{v1.4-2}{1999/06/25}
%	{Description of characters and commands of \textsf{array} package is
%	 added.}
% \SpecialOptIndex{array}{package}
% \SpecialIndex{\extrarowheight}
% \SpecialIndex{\firsthline}
% \SpecialIndex{\lasthline}
% \SpecialIndex{\newcolumntype}
%

% Users of \textsf{array} package may use all of newly introduced
% preamble characters, such as `|>|', `|<|', `|m|', `|b|', and all the
% commands such as |\extrarowheight|, |\firsthilne| and |\lasthline|.  The
% preamble characters given by \textsf{arydshln} may be included in the
% second argument of |\newcolumntype|.
%
% \changes{v1.5-1}{2000/07/12}
%	{Description of \textsf{colortab} commands is added.}
% \SpecialOptIndex{colortab}{package}
% \SpecialIndex{\LCC}
% \SpecialIndex{\NAC}
% \SpecialIndex{\ENAC}
% \SpecialIndex{\AC}
% \SpecialIndex{\EAC}
%

% Also users of \textsf{colortab} package may use |\LCC|\slash |\ECC|
% construct to color columns.  A horizontal solid\slash dash line may be
% colored by, e.g.\ |\NAC\hdashline\ENAC|\@.  The pair of |\AC| and |\EAC| may 
% be used to color everything between them {\em but}, unfortunately,
% vertical lines are not.  There are no ways to color vertical lines in a
% table having dash lines.  You may color vertical lines of a ordinary table 
% inactivating dash line functions by |\ADLinactivate|.
% 
% \changes{v1.7}{2004/5/21}
%	{Description of \textsf{colortbl} and related commands is added.}
% \SpecialOptIndex{colortbl}{package}
% \SpecialIndex{\rowcolor}
% \SpecialIndex{\columncolor}
% \SpecialIndex{\cellcolor}
% \SpecialIndex{\arrayrulecolor}
% \SpecialIndex{\doublerulesepcolor}
%

% Another (and more convenient) table coloring tool \textsf{colortbl} may be
% also used simply by loading it before \textsf{arydshln}.  Not only the
% painting commands |\rowcolor|, |\columncolor| and |\cellcolor| work well, 
% but both solid and dash lines are also colored by the command
% |\arrayrulecolor| of \textsf{colortbl}\footnote{

%

% The \textsf{colortbl} manual says \cs{arrayrulecolor} and
% \cs{doublerulesepcolor} may be in \texttt{>\char`\{\ldots\char`\}}
% in a preamble but they cause an error with the original implementation.
% This bug is fixed in \textsf{arydshln} and they are now usable to specify
% the color of the vertical (dash) lines whose specifications occur after
% the commands.}.
%

% One caution is that |\arrayrulecolor| defines the color of the dash-part 
% of dash lines and thus gap-part has no color (i.e. color of the paper on
% which the line drawn).  Therefore, if you have a \textsf{tabular} like;
%

%\begin{quote}
%\begin{verbatim}
%\begin{tabular}{|>{\columncolor{red}}l:>{\colomncolor{green}}r|}
%...
%\end{tabular}
%\end{verbatim}
%\end{quote}
%

% you will find the vertical dash line is a sequence of black (or the color of 
% |\arrayrulecolor|) and white segments.  This problem is partly solved by 
% declaring |\ADLnullwide|\footnote{

%

% Since \textsf{colortbl} automatically loads \textsf{array}, the default is
% \cs{ADLsomewide}}
%

% to conjunct the red and blue columns and to draw the dash line on their
% border.
%
% \DescribeMacro{\ADLnullwidehline}
% \DescribeMacro{\ADLsomewidehline}
% Unfortunately, however, |\ADLnullwide| does not affect the real
% width of horizontal (dash) lines and thus you will still see white gaps
% in |\hdashline| and |\cdashline|.  A solution is to put
% |\ADLnullwidehline| before you start a
% \texttt{array}\slash\texttt{tabular}\footnote{

%

% This command also makes \cs{cline} and \cs{cdashline} visisble even if the
% row below is painted.}.
%

% With this command, a horizontal (dash) line is drawn adjusting its bottom
% edge to that of the row above.  The command |\ADLsomewidehline| turns the
% switch to default and the top edge of a horizontal (dash) line will be
% adjusted to the bottom edge of the row above.
%
% \DescribeMacro{\dashgapcolor}
% \DescribeMacro{\nodashgapcolor}
% \SpecialIndex{\color}
% Another method to avoid white gaps is to give a color to gaps by
% |\dashgapcolor| with arguments same as |\color|.  For example;
%

% \begin{quote}
% |\arrayrulecolor{green}\dashgapcolor[rgb]{1,1,0}|
% \end{quote}
%

% makes colorful dash lines with green dashes and yellow gaps.  The command
% can be placed outside of \texttt{array}\slash\texttt{tabular} for dash
% lines in the environment, in the argument of preamble character |>|
% for vertical dash lines following them, or at the beginning of a row for
% horizontal dash lines following the command.  The commmand
% |\nodashgapcolor| (no arguments) nullifies the effect of |\dashgapcolor|.
% Note that |\nodashgapcolor| is different from |\dashgapcolor{white}|
% because the former makes gaps {\it transparent} while the later whiten them.
%
% \DescribeEnv{longtable}
% \DescribeEnv{Longtable}
% \changes{v1.6-1}{2003/8/25}
%	{Description of \textsf{longtable} is added.}
% \SpecialOptIndex{longtable}{package}
% Usage of \textsf{longtable} with \textsf{arydshln} is quite simple.  Just
% loading \textsf{arydshln} after \textsf{longtable} is enough to make the
% |longtable| environment able to draw dash-lines.  A shorthand activation
% of dash-line functions is also available by |Longtable| environment.  One
% caution to \textsf{longtable} users is that the temporary results before
% the {\em convergence} of the column widths may be different from those
% without \textsf{arydshln}.  For example, the following is the first pass
% result of the example shown in Table~3 of the \textsf{longtable} manual.
%

% \begin{quote}
% \catcode`\|12
% \leavevmode\vtop{\offinterlineskip
%	\hbox{\begin{tabular}{|c|c|c|}\hline 1&2&3\end{tabular}}
%	\hbox{\begin{tabular}{|c|}wide multicolumn spanning 1--3\end{tabular}}
%	\hbox{\begin{tabular}{|c|c|}\omit&
% 		\omit\vrule height0pt width116.66695pt\cr
% 		multicolumn 1--2&3\end{tabular}}
% 	\hbox{\begin{tabular}{|c|c|c|}\omit&
% 		\omit\vrule height0pt width68.05573pt&
% 		\omit\vrule height0pt width116.66695pt\cr
% 		wide 1&2&3\\\hline\end{tabular}}}
% \end{quote}
%

% \SpecialOptIndex{LTchunksize}{counter}
% Since |LTchunksize| is one in the example, columns of each row has their own
% widths and thus has vertical lines drawn at the edges of the columns.  On
% the other hand, you will have the following as the first pass result with
% \textsf{arydshln}.
%

% \begin{quote}
% \catcode`\|12
% \settowidth\unitlength{\begin{tabular}{|c|c|c|}\hline 1&2&3\end{tabular}}
% \begin{tabular}[t]{|l|c|c|}
% 	\multispan3\relax
% 		\leaders\hrule height\arrayrulewidth\hskip\unitlength\cr
% 	\hbox to0pt{1\hskip2\tabcolsep 2\hskip2\tabcolsep 3\hss}&&\\
% 	\multicolumn3{|l|}{wide multicolumn spanning 1--3}\\
% 	\multicolumn2{|l|}{\hbox to0pt{multicolumn 1--2\hskip\tabcolsep
% 		\hbox to116.66695pt{\hss3\hss}\hss}}&\\
% 	wide 1&2&3\\\hline
% 	\omit&
% 		\omit\vrule height0pt width68.05573pt&
% 		\omit\vrule height0pt width116.66695pt\cr
% 	\end{tabular}
% \end{quote}
%

% As you see, the vertical lines are drawn at the column edges of the last
% row\footnote{

%

% More precisely, drawn according to the column widths established by all
% the chunks preceding page output.}
%

% because \textsf{arydshln} draws them when it see the last row.  Anyway,
% you may ignore temporary results and will have a compatible result when
% the column widths are converged like the following.
%

% \begin{quote}
% \catcode`\|12
% \begin{tabular}[t]{|c|c|c|}\hline
%	1&2&3\\
%	\multicolumn{3}{|c|}{wide multicolumn spanning 1--3}\\
%	\multicolumn{2}{|c|}{multicolumn 1--2}&3\\
%	wide 1&2&3\\\hline
%	\end{tabular}
% \end{quote}
% 
% 
%
% \section{Known Problems}
% \label{sec:prob}
%
% There are following known problems.
%

% \begin{enumerate}
% \item
% The new preamble specifiers `|:|' and `|;{|\meta{dash}|/|\meta{gap}|}|'
% cannot be followed or preceded by `|@{|\meta{text}|}|', or you will have
% an ugly result.  More specifically, a specifier to draw a dash-line at the
% left edge of a column cannot be preceded by `|@{|\meta{text}|}|', while
% that to draw at the right edge cannot be followed by `|@{|\meta{text}|}|'.
%
% \item\leavevmode
% \changes{v1.4-2}{1999/06/25}
%	{Description about `\protect\EXCLAMATION' of array package is added.}
%

% If you use \textsf{array} package, the restriction of `|@|' shown above is 
% also applied to `|!|'.
% 
% \item
% In order to make it sure that a dash-line always {\em touches} its both
% end, i.e. a dash-line always begins and ends with a dash segment, the
% amount of a gap will slightly vary depending on the dash-line length.
%
% \item
% \changes{v1.4-2}{1999/06/25}
%	{Reference to the section for drawing mode is added.}
%

% As described in \S\ref{sec:usage-finer}, dash-lines drawn in the default
% mode 1 may lack a dash segment near its right\slash bottom end.
% 
% \item
% \changes{v1.4-2}{1999/06/25}
%	{Description on minimum length is added.}
%

% If a dash-line is too short, you will have an ugly result without overfull
% message.  More specifically, in mode 1 or 3, a line will look to protrude
% beyond its column\slash row borders if it is shorter than a half of
% |\dashlinedash|.  In mode 2, the minimum length to avoid the protrusion is
% $1.5\times|\dashlinedash|+ |\dashlinegap|$.
%
% \item
% \changes{v1.4-2}{1999/06/25}
%	{Reference to the performance tuning section is added.}
%

% As described in \S\ref{sec:usage-performance},
% the processing speed for |array| and |tabular| environment will become
% slower even if dash-lines are not included.
% 
% \item
% \changes{v1.5-1}{2000/07/12}
%	{Caution about \cs{AC}/\cs{EAC} pair for vertical line coloring is
%	 added.}
% \changes{v1.7}{2004/5/21}
%	{Comment on vertical line coloring with \textsf{colortbl} is added.}
%

% 
% As described in \S\ref{sec:usage-compat}, |\AC| and |\EAC| pair of
% \textsf{colortab} such as |\AC&\EAC| cannot color the vertical line at
% |&|.  Use |\ADLinactivate| if you want to have a ordinary table with
% colored vertical lines.  Note that you may color vertical lines with
% \textsf{colortbl} package.
% \end{enumerate}
%
% 
%
% \IndexPrologue{\newpage\section*{Index}
% Italicized number refers to the page where the specification of
% corresponding entry is described.}
% \StopEventually{
% \section*{Acknowledgments}
% 
% \changes{v1.4-2}{1999/06/25}
%	{Thank to more people.}
% \changes{v1.5-3}{2003/07/11}
%	{Thank to Yaxin Liu.}
% \changes{v1.6-1}{2003/08/25}
%	{Thank to people for \textsf{longtable}.}
% \changes{v1.7}{2004/5/21}
%	{Thank to Klaus Dalinghaus and refer orignal \textsf{colortbl}.}
% \changes{v1.71-1}{2004/8/31}
%	{Thank to Morten H\o gholm.}
%
% The author thanks to Monty Hayes who gave the author the opportunity to
% make this style, and Weimin Zhang and Takahiro Kubota who pointed out bugs
% in early versions.  He also thanks to the following people; Sebastian
% Rahtz and Graham Williams who kindly invited the style to \TeX{} CTAN and
% online catalogue compiled by Graham;  Peter Ehrbar who showed the style was
% incompatible with \textsf{array} and kindly accepted the offer to be an
% alpha-user of v1.4 alone;  Zsuzsanna Nagy who reported another
% incompatibility problem with \textsf{colortab};  Ralf Heydenreich who
% reported the bug causing that glues in a column have no effect; Yaxin 
% Liu who reported the incompatibility bug of \textsf{array} and
% \cs{ADLinactivate}; Craig Leech who reported the incompatibility 
% problem with \textsf{longtable}, which was also reported by Uwe Jehmlich,
% Torge Thielemann and 
% Florian Weig, and have waited for two years and a half (!) for the
% solution; and Klaus Dalinghaus who reported yet another incompatibility
% with \textsf{colortbl}; Morten H\o gholm who reported the bug of |m|-type
% columns of \textsf{array} which had not manifested in five (!!) years
% since the author realesed the first \textsf{array}-compatible version.
%
% The base implementation of \texttt{array} and \texttt{tabular}
% environments, part of which the author gives new definitions referring
% original ones, are written by Leslie Lamport as a part of \LaTeX-2.09 and
% \LaTeXe{} (1997/12/01) to which Johannes Braams and other authors also
% contributed.  The author also refers \textsf{array} package (v2.3m)
% written by Frank Mittelbach and David Carlisle; \textsf{colortab} package
% (v0.9) written by Timothy van Zandt; and \textsf{longtable} (v4.10) and
% \textsf{colortbl} (v0.1j) packages
% written by David Carlisle; to make the style compatible with those
% packages.
% 
% \PrintIndex}
%
% 
%
% \newpage
% \section{Implementation}
% \label{sec:imp}
% \subsection{Problems and Solutions}
% \label{sec:imp-ps1}
%
% \changes{v1.4-2}{1999/06/25}
%	{The title of section 4.1 is changed.}
% \changes{v1.4-2}{1999/06/25}
%	{\cs{hfil} is replaced with \cs{hss} taking the possibility of
%	negative wide columns into account.}
% \changes{v1.7}{2004/5/21}
%	{Add notes for dash line coloring.}
%

% We have two different problems to solve; how to draw horizontal dash-lines
% and how to draw vertical dash-lines.  The former problem is relatively
% easy because the technique for drawing |\cline|-s can be used.  That is, if
% we know the number of columns, we can draw a dash-line across the
% |\multispan|-ed columns by |\xleaders| of dash.  Modifying a preamble of
% |array|\slash|tabular| to count the number of columns is not hard.  Since
% |\cdashline| is given beginning and ending columns, its implementation is
% also easy.
%
% \SpecialIndex{\@arstrutbox}
%

% The latter problem, however, is much harder.  Remember that
% |array|\slash|tabular| draws vertical solid lines by |\vrule|-s in each row
% without height\slash depth specification exploiting \TeX's sophisticated
% mechanism of the rule extension in the surrounding box.  Since \TeX{} does
% not have such a mechanism for |\xleaders| unfortunately, we at least have
% to know the height and depth of a row which includes vertical dash-lines.
% Although the height and depth are often same as those of |\@arstrutbox|, we
% will have an exceptionally tall and/or deep row that makes dash-lines {\em
% broken} if we assume every row has the standard height and depth.
%
% Moreover, even if we can measure the height\slash depth of each row (in
% fact we will do as describe later), drawing dash-lines in each row will
% not produce a good result.  Look at the following two examples closely.
%

%\begin{quote}
%\makeatletter
%\begin{tabular}{c:c}\hline
%A&B\\\hline
%\noalign{\xdef\adl@rowsL{\adl@rowsL(\relax/0);}
%         \xdef\adl@rowsR{\adl@rowsR(\relax/0);}}
%{\vrule height2\ht\strutbox depth2\dp\strutbox width\z@\LARGE A}&B\\\hline
%\end{tabular}\qquad
%\begin{tabular}{c:c}\hline
%A&B\\\hline
%{\vrule height2\ht\strutbox depth2\dp\strutbox width\z@\LARGE A}&B\\\hline
%\end{tabular}
%\end{quote}
%

% In the left example, two dash-lines are individually drawn in two rows.
% Since the first row is not so tall and deep (8.4\,pt\slash
% 3.6\,pt) as to contain enough number of default dash segments (4\,pt dash and
% 4\,pt gap) to keep |\xleaders| from inserting a large space, the dash-line
% in the first row is {\em sparse}.  On the other hand, the second row is
% enough tall and deep (16.8\,pt\slash 7.2\,pt) and thus the dash-line in the
% row looks better.  Thus the resulting dash-line is awful because it does
% not have a continuous dash/gap sequence.
%
% The right example, which we wish to produce, is much better than the left.
% In this example, the dash line is draw across two rows keeping continuous
% steps of dashes and gaps.  In order to have this result, we have to draw
% the dash-line {\em after} two rows are built because it is necessary to
% know the total hight and depth of two rows.  In general, if we know the
% total hight and depth of rows and whether a column has a dash-line, we can
% draw dash-lines by adding an extra row containing dash-lines.  For
% example, the result shown above is obtained by the following row.
%

% \begin{quote}
% |\omit\hss|\meta{dash-line of 36\,pt high}|&\omit\cr|
% \end{quote}
%

% Note that \meta{dash-line of 36\,pt high} have to be |\smash|-ed.
%
% In addition to this basic scheme, we have to take the following points
% into account.
%

% \begin{itemize}
% \item
% A dash-line drawn by the preamble character `|;|' will have non-default
% dash\slash gap specification.
%
% \item
% A column may have two or more dash-lines separated by spaces of
% |\doublerulesep|.  Mixed sequence of solid- and dash-lines also have to be
% allowed.
%
% \item
% The first column may have dash-lines at both ends, while those of others
% will appear at right ends only.  An exception of this rule is brought by
% |\multicolumn| that may have leading sequence of solid- and/or dash-line
% specifiers in its preamble.
%
% \item
% A |\multicolumn| may break or add a dash-line, or may change the
% dash\slash gap specification of a dash-line.  A sequence of
% |\h(dash)line|-s also break dash-lines.
% 
% \item
% If \textsf{colortbl} is in use, coloring dash\slash gap by
% |\arrayrulecolor| and |\dashgapcolor| gives another possibility of the
% variation of dash\slash gap specification.
% \end{itemize}
%
% In order to cope with them, the following data structure is constructed
% during rows are built.
%

% \begin{enumerate}
% \item
% The list of row information $R=\langle r_1,r_2,\ldots,r_N\rangle$.
%
% \item
% The $i\th$ element of $R$, $r_i$, is one of the following\footnote{
%

% In the real implementation, the structure of $r_i$ is slightly different.}.
%

% \begin{enumerate}
% \item
% A triple $\langle C^L_i,C^R_i,h_i\rangle$, where $C^L_i$ and $C^R_i$ are
% the lists of solid- or dash-line segments drawn at the left and right edge
% of columns respectively, and $h_i$ is the height plus depth of the $i\th$
% row.
%
% \item
% ${\it connect}(h_i)$ for a |\h(dash)line| of $h_i$ wide meaning that $r_i$
% is an empty pseudo row of $h_i$ high and dash-lines are not broken at the
% row.
%
% \item
% \changes{v1.6-1}{2003/8/25}
%	{Description of \textit{discard} is added.}
% In |longtable| environment, ${\it discard}(h_i)$ for a negative vertical
% space inserted by |\\[|\meta{$h_i$}|]| or |\h(dash)line| meaning $r_i$
% is an empty pseudo row of $h_i$ high and dash-lines are not broken but may 
% be discarded by the page break at the row.
%
% \item
% ${\it disconnect}(h_i)$ for a vertical gap generated by a sequence of
% |\h(dash)line| meaning that $r_i$ is an empty pseudo row of $h_i$ high and
% dash-lines are broken at the row.
% \end{enumerate}
%
% \item
% $C^L_i=\langle e^i_1,e^i_2,\ldots,e^i_m\rangle$ where $e^i_j$ corresponds
% to the $j\th$ (leftmost is first) solid- or dash-line segment.  $C^R_i$ is
% similar but its elements are ordered in reverse, i.e. the rightmost segment
% is the first element.
%
% \item
% \changes{v1.7}{2004/5/21}
% 	{A dash/gap specification $d^i_j/g^i_j$ now has color.}
% The $j\th$ element of $C^L_i$ or $C^R_i$, $e^i_j$, is a triple $\langle
% c^i_j,d^i_j,g^i_j\rangle$ where $c^i_j$ is the column number in which the
% segment appears, and $d^i_j$ and $g^i_j$ are dash\slash gap specification,
% length and color, of the segment.  For a solid line segment, the length
% attributes of both $d^i_j$ and $g^i_j$ are 0.
% \end{enumerate}
%
% Then this data structure is processed to draw solid- and dash-lines at the
% end of the |array|\slash|tabular| as follows.  Let $e^i_j=\langle c^i_j,
% d^i_j, g^i_j\rangle$ be the $j\th$ element of $C^L_i$ of $r_i$.  The {\em
% position} $p^i_j$ of $e^i_j$ in the column $c^i_j$ is defined as follows.
%$$
% p^i_j = \cases{1&		if $j=1 \LOR c^i_j\neq c^i_{j-1}$\cr
%		 p^i_{j-1}+1&	otherwise}.
%$$
% The following defines whether two elements $e^i_j$ and $e^{i'}_{j'}$ are
% {\em connected}, or $e^i_j\sim e^{i'}_{j'}$.
%

% \begin{eqnarray*}
% e^i_j\sim e^{i'}_{j'}&
%	\togets&i<i'\LAND\\
% &&	c^i_j=c^{i'}_{j'}\LAND d^i_j=d^{i'}_{j'}\LAND g^i_j=g^{i'}_{j'}\LAND
%		p^i_j=p^{i'}_{j'}\LAND\\
% &&	^\forall k(i<k<i'\to r_k=connect(h_k)).
% \end{eqnarray*}
%

% With these definitions, we can classify all $e^i_j$ into ordered sets
% $S_1,S_2,\ldots S_n$ as follows.
%

% \def\PH{\phantom{^\forall e^i_j\in S_k,\forall e^{i'}_{j'}\in S_{k'}(}}
% \begin{eqnarray*}
% k\neq k'&\togets&
%	S_k\cap S_{k'}=\emptyset\\
% e^i_j\sim e^{i'}_{j'}&\togets&
%	^\exists k(e^i_j,e^{i'}_{j'}\in S_k\LAND
%		   S_k=\{\ldots,e^i_j,e^{i'}_{j'},\ldots\}\\
% k<k'&\togets&
%	^\forall e^i_j\in S_k,\forall e^{i'}_{j'}\in S_{k'}(
%		(c^i_j<c^{i'}_{j'})\LOR\\
% &&	\PH	(c^i_j=c^{i'}_{j'}\LAND p^i_j<p^{i'}_{j'})\LOR\\
% &&	\PH	(c^i_j=c^{i'}_{j'}\LAND p^i_j=p^{i'}_{j'}\LAND i<i')).
% \end{eqnarray*}
%

% Now we can draw a dash-line $L_k=\langle\gamma_k, \pi_k, \delta_k, \xi_k,
% \tau_k, \beta_k\rangle$ corresponding to $S_k=\{e^i_j,\ldots,e^{i'}_{j'}\}$
% as follows.
%

% \begin{itemize}
% \item
% $L_k$ is $\pi_k\th$ line in the $\gamma_k\th$ column where $\gamma_k=
% c^i_j=\ldots=c^{i'}_{j'}$ and $\pi_k=p^i_j=\ldots=p^{i'}_{j'}$.
%
% \item
% $L_k$ has the dash specification (size and color)
% $\delta_k=d^i_j=\ldots=d^{i'}_{j'}$ and gap specification
% $\xi_k=g^i_j=\ldots=g^{i'}_{j'}$.
%
% \item
% The top and bottom ends of $L_k$ are at $\tau_k$ and $\beta_k$ above the
% bottom of the |array|\slash|tabular|, where;
%$$
% \tau_k=\sum_{l=i}^N h_l,\quad\beta_k=\sum_{l=i'+1}^N h_l.
%$$
% \end{itemize}
%

%  The row to draw $L_1,\ldots,L_n$ is;
%$$
% \sigma_1 L_1 \sigma_2 L_2 \ldots L_{n-1}\sigma_n L_n\sigma_{n+1}|\cr|
%$$
% where;
%

% \def\amp{\hbox{\tt\&}}
% \begin{eqnarray*}
% \sigma_1&=&
%	\cs{omit}[\cs{hss}\amp\cs{omit}]^{\gamma_1-1}\\[1ex]
% \sigma_{1<k\leq n}&=&
%	\cases{	\cs{null}&
%			if $\gamma_{k-1}=\gamma_k\LAND \pi_{k-1}=\pi_k$\cr
%		\cs{hskip}\cs{doublerulesep}&
%			if $\gamma_{k-1}=\gamma_k\LAND \pi_{k-1}\neq\pi_k$\cr
%		[\cs{hss}\amp\cs{omit}]^{\gamma_k-\gamma_{k-1}}&
%			if $\gamma_{k-1}\neq\gamma_k$\cr}\\[1ex]
% \sigma_{n+1}&=&
%	[\cs{hss}\amp\cs{omit}]^{{\mit\Gamma}-\gamma_n-1}\cs{hss}.
% \end{eqnarray*}
%

% Note that $[x]^m$ means $m$-times iteration of $x$, and $\mit\Gamma$ is
% the number of columns specified in the preamble.
%
% Dash-lines at the right edges of columns are similarly drawn by processing
% $C^R_i$ with the following modifications.
%

% \begin{eqnarray*}
% k<k'&\togets&
%	^\forall e^i_j\in S_k,\forall e^{i'}_{j'}\in S_{k'}(
%		(c^i_j<c^{i'}_{j'})\LOR\\
% &&	\PH	(c^i_j=c^{i'}_{j'}\LAND p^i_j>p^{i'}_{j'})\LOR\\
% &&	\PH	(c^i_j=c^{i'}_{j'}\LAND p^i_j=p^{i'}_{j'}\LAND i<i'))\\
% \sigma_1&=&
%	\cs{omit}\cs{hss}[\amp\cs{omit}\cs{hss}]^{\gamma_1-1}\\[1ex]
% \sigma_{k>1}&=&
%	\cases{	\cs{null}&
%			if $\gamma_{k-1}=\gamma_k\LAND \pi_{k-1}=\pi_k$\cr
%		\cs{hskip}\cs{doublerulesep}&
%			if $\gamma_{k-1}=\gamma_k\LAND \pi_{k-1}\neq\pi_k$\cr
%		[\amp\cs{omit}\cs{hss}]^{\gamma_k-\gamma_{k-1}}&
%			if $\gamma_{k-1}\neq\gamma_k$\cr}\\[1ex]
% \sigma_{n+1}&=&
%	[\amp\cs{omit}\cs{hss}]^{{\mit\Gamma}-\gamma_n-1}
% \end{eqnarray*}
%
% 
%
% \subsection{Another Problem and Imperfect Solutions}
% \label{sec:imp-ps2}
% 
% In the default mode 1, we draw a dash line of dash size $d$ and gap size
% $g$ as follows.  Let $W$ be the length of the line plus 10\,|sp|\footnote{

%

% This small amount is added by \cs{xleaders} in order to, according to the 
% comment in \texttt{tex.web}, compensate floating point rounding error.},
%

% which is unknown for us if horizontal but known for \TeX, and assume
% $W\geq d/2$ (or the line protrude to the column\slash row boarder.)  At
% the both ends of the columns, dashes of $d/2$ long are drawn to make the
% dash-line {\em touched} to the ends.  Then $n=\lfloor(W-d-g)/(d+g)\rfloor$
% dashes are equally distributed in the remaining space.  Thus we will have;
% $$
% D_0(d/2)G_0(g+\epsilon')D_1(d)G_1(g+\epsilon)\ldots
%	G_{n-1}(g+\epsilon)D_n(d)G_n(g+\epsilon')D_{n+1}(d/2)
% $$
% where $D_i(l)$ and $G_i(l)$ are dash and gap of $l$ long,
% $\epsilon=(W-(n+1)(d+g))/(n+1)$ (rounded), and
% $\epsilon'=(W-(n+1)(d+g)-(n-1)\epsilon)/2$ to compensate the rounding
% error on the calculation of $\epsilon$.  For a horizontal line, this
% result will be obtained by |\xleaders| as follows where $G^m_i(\epsilon)$
% and $G^m_i(\epsilon')$ are the spaces inserted by |\xleaders|.
%

% \begin{eqnarray*}
% \rlap{$\displaystyle D_0(d/2)G^l_0(g/2)
%		\hbox{\cs{xleaders}\cs{hbox}{\tt\char`\{}}
%			G^r(g/2)D(d)G^l(g/2)\hbox{\tt\char`\}\cs{hss}\ }
%		G^r_n(g)D_{n+1}(d/2)$}\qquad\\
% &=&D_0(d/2)G^l_0(g/2)G^m_0(\epsilon')
%	\left(G^r_0(g/2)D_1(d)G^l_1(g/2)\right)G^m_1(\epsilon)\\
% &&	\phantom{D_0(d/2)G^l_0(g/2)G^m_0(\epsilon)}
%	\left(G^r_1(g/2)D_2(d)G^l_2(g/2)\right)G^m_2(\epsilon)\\
% &&	\phantom{D_0(d/2)G^l_0(g/2)G^m_0(\epsilon)}\ldots\\
% &&	\phantom{D_0(d/2)G^l_0(g/2)G^m_0(\epsilon)}
% 	\llap{$\displaystyle G^m_{n-1}(\epsilon)$}
% 	\left(G^r_{n-1}(g/2)D_n(d)G^l_n(g/2)\right)
%	G^m_n(\epsilon')G^r_n(g/2)D_{n+1}(d/2)\\
% &=&D_0(d/2)G_0(g+\epsilon')D_1(d)G_1(g+\epsilon)\ldots
%	G_{n-1}(g+\epsilon)D_n(d)G_n(g+\epsilon')D_{n+1}(d/2)
% \end{eqnarray*}
%

% The problem is that $\epsilon'$ could be negative and \TeX{} mistakingly
% ignores this possibility.  That is, since \TeX{} does not put |\hbox|
% beyond the right edge of |\xleaders|, the rightmost |\hbox| is omitted if
% $\epsilon'$ is negative as described in \S\ref{sec:usage-finer}.
%
% Since it is (almost) impossible to know the length of a horizontal line,
% we cannot cope with this problem by adding or subtracting its length.
% Thus we introduced {\em drawing mode} to have imperfect solutions.  In the 
% mode 2, we draw a line by the following sequence.
%

% \begin{eqnarray*}
% \rlap{$\displaystyle D_0(d/2)G^l_0(g/2)G^r_{0'}(g/2)D_{1'}(d)G^l_{1'}(g/2)
%		G(-d-g)$}\qquad\\
% &&	\hbox{\cs{xleaders}\cs{hbox}{\tt\char`\{}}
%			G^r(g/2)D(d)G^l(g/2)\hbox{\tt\char`\}\cs{hss}\ }\\
% &&	G(-d-g)G^r_{n'}(g/2)D_{n'}(d)G^l_{n'}(g/2)G^r_n(g)D_{n+1}(d/2)
% \end{eqnarray*}
%

% That is, $n\th$ |\hbox| that could be disappeared is put twice and the
% first one is also overlaid for symmetrization.  Therefore the length of the
% first and $n\th$ dashes is $d+\vert\epsilon'\vert$ and thus could be a
% little bit longer than others.
% 
% On the other hand, we replace |\xleaders| of mode 1 with |\cleaders| for
% the drawing in mode 3.  The result will be;
% $$
% D_0(d/2)G_0(g+R)D_1(d)G_1(g)\ldots
%	G_{n-1}(g)D_n(d)G_n(g+R)D_{n+1}(d/2)
% $$
% where $R=(W-(n+1)(d+g))/2$ to make the first and last gaps considerably wider
% than others.
%
% 
%
% \iffalse
%<*package>
% \fi
%
% \subsection{Register Declaration}
% \label{sec:imp-decl}
%
% Here registers and switches are declared.
%
% \begin{macro}{\dashlinedash}
% \begin{macro}{\dashlinegap}
% \begin{macro}{\hdashlinewidth}
% \begin{macro}{\hdashlinegap}
% First of all, two |\dimen| registers |\dashlinedash| and |\dashlinegap|
% to control the shape of dash-lines are declared, and their default values,
% 4\,pt for both, are assigned to them.  They have aliases,
% |\hdashlinewidth| and |\hdashlinegap| respectively, for the backward
% compatibility.
% \end{macro}\end{macro}\end{macro}\end{macro}
%

%    \begin{macrocode}
%% Register Declaration

\newdimen\dashlinedash \dashlinedash4pt %
\newdimen\dashlinegap \dashlinegap4pt %
\let\hdashlinewidth\dashlinedash
\let\hdashlinegap\dashlinegap

%    \end{macrocode}
%
% Next, the following six switches are declared.
% \begin{itemize}
% \item
% \begin{macro}{\ifadl@leftrule}
% |\ifadl@leftrule| is used in the preamble analysis macro |\@mkpream| and
% is true during it processes leading characters for solid- and dash-lines,
% i.e.\ `\verb!|!', `|:|', and `|;|'.
% \end{macro}
%
% \item
% \begin{macro}{\ifadl@connected}
% |\ifadl@connected| is used to indicate the {\em connection} $e^i_j\sim
% e^{i'}_{j'}$.  When we process $e^{i'}_{j'}$, the switch is true iff
% $^\exists e^i_j(e^i_j\sim e^{i'}_{j'})$.
% \end{macro}
%
% \item
% \begin{macro}{\ifadl@doublerule}
% |\ifadl@doublerule| is used to make $\sigma_k$.  When we are to make
% $\sigma_k L_k$, it is true iff $\gamma_{k-1}=\gamma_k\LAND\pi_{k-1}\neq
% \pi_k$.
% \end{macro}
%
% \item
% \begin{macro}{\ifadl@zwvrule}
% \changes{v1.4-3-2}{1999/06/25}
%	{Introduced to indicate vertical lines have null width.}
% |\ifadl@zwvrule| controls the {\em real} width of vertical lines.  If it
% is true, lines are drawn as if their width is zero following \LaTeX's
% standard.  Otherwise, their width |\arrayrulewidth| contribute to the
% width of columns as \textsf{array} does.
% \end{macro}
%
% \item
% \begin{macro}{\ifadl@zwhrule}
% \changes{v1.7-1}{2004/5/21}
%	{Introduced to indicate horizontal lines have null width.}
% |\ifadl@zwhrule| controls the {\em real} width of horizontal lines.  If it
% is true, a line is drawn as if its width is zero and its bottom edge is
% adjusted to that of the row above by inserting |\vskip-\arrayulewidth|
% before the drawing.  Thus a horizontal dash line is included in the row
% above and its gaps look colored properly if the row is painted.  If it is
% false, the width |\arrayrulewidth| contribute to the height of
% \texttt{array}\slash\texttt{tabular} as usual.
% \end{macro}
%
% \item
% \begin{macro}{\ifadl@usingarypkg}
% \changes{v1.4-3-1}{1999/06/25}
%	{Introduced to know if \textsf{array} is loaded.}
% |\ifadl@usingarypkg| is true iff \textsf{array} has been loaded prior to
% \textsf{arydshln}.  This switch shows us which definitions, by \LaTeX{} or 
% \textsf{array}, we have to modify.  Its value is set by examining if
% |\extrarowheight|, which is introduced by \textsf{array}, is defined.
% \end{macro}
%
% \item
% \begin{macro}{\ifadl@inactive}
% \changes{v1.4-3-3}{1999/06/25}
%	{Introduced to indicate dash-line functions are inactive.}
% |\ifadl@inactive| inactivates dash-line functions if it is true.  Its
% default value is false.
% \end{macro}
% \end{itemize}
%

% \SpecialIndex{\@tempswa}
% We also use a working switch |\@tempswa|.
%

%    \begin{macrocode}
\newif\ifadl@leftrule
\newif\ifadl@connected
\newif\ifadl@doublerule
\newif\ifadl@zwvrule
\newif\ifadl@zwhrule
\newif\ifadl@usingarypkg
\ifx\extrarowheight\undefined \adl@usingarypkgfalse
\else                         \adl@usingarypkgtrue  \fi
\newif\ifadl@inactive \adl@inactivefalse

%    \end{macrocode}
%
% \begin{macro}{\ADLnullwide}
% \changes{v1.7-1}{2004/5/21}
%	{Introduced to make horizontal lines null wide.}
% \begin{macro}{\ADLsomewide}
% \changes{v1.7-1}{2004/5/21}
%	{Introduced to make horizontal lines \cs{arraydashline} wide.}
% The switch |\ifadl@hwvrule| is turned on/off by user interface macros
% |\ADLnullwide| and |\ADLsomewide|.  Its initial value is the complement of
% |\adl@usingarypkg|.
% \end{macro}\end{macro}
%
% \begin{macro}{\ADLnullwidehline}
% \changes{v1.4-3-2}{1999/06/25}
%	{Introduced to make vertical lines null wide.}
% \begin{macro}{\ADLsomewidehline}
% \changes{v1.4-3-2}{1999/06/25}
%	{Introduced to make vertical lines \cs{arraydashline} wide.}
% The switch |\ifadl@zwvrule| is turned on/off by user interface macros
% |\ADLnullwidehline| and |\ADLsomewidehline|.  Its initial value is false.
% \end{macro}\end{macro}
%
% \begin{macro}{\ADLactivate}
% \begin{macro}{\ADLinactivate}
% The switch |\ifadl@inactive| is also turned on/off by user interface macros
% |\ADL#inactivate| and |\ADLactivate|.
% \end{macro}\end{macro}
%

%    \begin{macrocode}
\def\ADLnullwide{\adl@zwvruletrue}
\def\ADLsomewide{\adl@zwvrulefalse}
\ifadl@usingarypkg \ADLsomewide \else \ADLnullwide \fi
\def\ADLnullwidehline{\adl@zwhruletrue}
\def\ADLsomewidehline{\adl@zwhrulefalse}
\ADLsomewidehline

\def\ADLactivate{\adl@inactivefalse}
\def\ADLinactivate{\adl@inactivetrue}

%    \end{macrocode}
%
% The following |\box| register and three |\dimen| registers are used to
% measure the height and depth of a row.
%

% \begin{itemize}
% \item
% \begin{macro}{\adl@box}
% The contents of a column is packed into the |\box| register |\adl@box| to
% measure its height and depth.
% \end{macro}
%
% \item
% \begin{macro}{\adl@height}
% \begin{macro}{\adl@depth}
% The |\dimen| registers |\adl@height| and |\adl@depth| contain the
% height\slash depth of the tallest\slash deepest column in a row.  When a
% column is processed, they are compared to the height and depth of
% |\adl@box| and are updated if they are less.
% \end{macro}\end{macro}
%
% \begin{macro}{\adl@heightsave}
% \begin{macro}{\adl@depthsave}
% Since we have to update these register |\global|-ly to pass their value
% across |&| and we may have a column containing |array|\slash|tabular|,
% they are saved into |\adl@heightsave|\slash|\adl@depthsave| at the
% beginning of the environment and are restored at its end.
% \end{macro}\end{macro}
% 
% \begin{macro}{\adl@finaldepth}
% \changes{v1.4-3-6}{1999/06/25}
%	{Introduced to measure the depth of the last row.}
% The other |\dimen| register |\adl@finaldepth| is set to the depth of the
% last row, or zero if the last vertical item is a horizontal line.  This
% value is used to shift |array|\slash|tabular| down because we add extra 
% two |\smash|-ed rows which make the depth of |array|\slash|tabular| zero.
% \end{macro}
% \end{itemize}
%

% \SpecialIndex{\@tempdima}
% \SpecialIndex{\@tempdimb}
% We also use working |\dimen| registers |\@tempdima| and |\@tempdimb|.
%

%    \begin{macrocode}
\newbox\adl@box
\newdimen\adl@height \newdimen\adl@heightsave
\newdimen\adl@depth \newdimen\adl@depthsave
\newdimen\adl@finaldepth
%    \end{macrocode}
%
% Then the following |\count| registers are declared.  Note that some of
% them contain dimensions measured by the unit |sp|.
%

% \begin{itemize}
% \item
% \begin{macro}{\adl@columns}
% \begin{macro}{\adl@ncol}
% \changes{v1.4-3-1}{1999/06/25}
%	{Introduced for new column counting in preamble construction.}
% |\adl@columns| has the number of columns specified in the preamble of the
% environment.  Because of a complicated reason related to the compatibility 
% with \textsf{array}, we cannot count up |\adl@columns| directly but
% increment |\adl@ncol| when each column of preamble is built and move its
% value to |\adl@columns| after the preamble is constructed.
% \end{macro}\end{macro}
%
% \item
% \begin{macro}{\adl@currentcolumn}
% \begin{macro}{\adl@currentcolumnsave}
% To process |\multicolumn|, we have to know the column number where it
% appears.  Thus we have a column counter |\adl@currentcolumn| which is
% |\global|-ly incremented when each column is built.  Because of the
% |\global| assignment, the counter has to be saved\slash restored
% into\slash from |\adl@currentcolumnsave|.
% \end{macro}\end{macro}
%
% \item
% \begin{macro}{\adl@totalheight}
% In the real implementation, $\tau_k$ and $\beta_k$ are calculated by the
% following equations rather than those shown in \S\ref{sec:imp-ps1}.
%$$
% H=\sum_{l=1}^N h_l,\quad
% \tau_k=H-\sum_{l=1}^{i-1} h_l,\quad
% \beta_k=\tau_k-\sum_{l=i'}^i h_l.
%$$
% |\adl@totalheight| contains $\sum_{l=1}^i h_l$ when the $i\th$ row is
% built and thus its final value is $H$.  Since the data structure $R$ are
% represented by a text, we have to pay attention to the precision of its
% dimensional elements, such as $h_i$.  That is, if we append $h_i$ to $R$
% by expanding |\the\dimen|$n$ which has the height plus depth of $i\th$
% row, $h_i$ will be an approximation of |\dimen|$n$ represented by a
% decimal fraction with |pt|.  Although the error of the approximation is
% quite small and may be negligible, the error must be avoided because it is
% avoidable by simply using |\number\dimen|$n$.  Therefore, $h_i$ is an
% integer and thus |\adl@totalheight| is too.
% \end{macro}
%
% \begin{macro}{\adl@totalheightsave}
% Because of the |\global| assignment to |\adl@totalheight| to pass its value
% across rows, it has to be saved\slash restored into\slash from
% |\adl@totalheightsave|.
% \end{macro}
%
% \item
% \begin{macro}{\adl@dash}
% \begin{macro}{\adl@gap}
% In order to check $e^i_j\sim e^{i'}_{j'}$, the size attributes of $d^i_j$
% and $g^i_j$ are kept in
% the registers |\adl@dash| and |\adl@gap| when we process $e^{i'}_{j'}$.
% As explained above, $d^i_j$ and $g^i_j$ are integers and thus |\adl@dash|
% and |\adl@gap| are |\count| registers.
% \end{macro}\end{macro}
%
% \item
% \begin{macro}{\adl@cla}
% \begin{macro}{\adl@clb}
% \SpecialIndex{\@cla}
% \SpecialIndex{\@clb}
% The coding of |\cdashline| is similar to that of |\cline| in \LaTeX-2.09
% which uses two global |\count| registers |\@cla| and |\@clb|.  These
% registers are omitted from \LaTeXe{} because its |\cline| is completely
% recoded.  We could adopt new coding but it requires some other macro
% definitions that \LaTeX-2.09 does not have.  Thus we simply introduce new
% global counters |\adl@cla| and |\adl@clb| for |\cdashline| in order to
% make |\cdashline| work in both \LaTeX-2.09 and \LaTeXe.
% \end{macro}\end{macro}
% \end{itemize}
%

% \SpecialIndex{\@tempcnta}
% \SpecialIndex{\@tempcnta}
% We also use working |\count| registers |\@tempcnta| and |\@tempcntb|.
%

%    \begin{macrocode}
\newcount\adl@columns \newcount\adl@ncol
\newcount\adl@currentcolumn \newcount\adl@currentcolumnsave
\newcount\adl@totalheight \newcount\adl@totalheightsave
\newcount\adl@dash \newcount\adl@gap
\newcount\adl@cla \newcount\adl@clb
%    \end{macrocode}
%
% \begin{macro}{\adl@everyvbox}
% \changes{v1.4-3-1}{1999/06/25}
%	{Introduced for a tricky modification of \cs{@array}.}
% The last register declaration is for a |\toks| register named
% |\adl@everyvbox|.  In order to minimize the copy-and-modify of the codes
% in \LaTeX{} and \textsf{array}, we need to use |\everyvbox| in our own
% definition of |\@array|.  The register is used to save the contents of
% |\everyvbox|.
% \end{macro}
%

%    \begin{macrocode}
\newtoks\adl@everyvbox

%    \end{macrocode}
%
% \begin{macro}{\adl@org@arrayclassz}
% \changes{v1.4-3-3}{1999/06/25}
%	{Introduced to restore \cs{@arrayclassz}.}
% \begin{macro}{\adl@org@tabclassz}
% \changes{v1.4-3-3}{1999/06/25}
%	{Introduced to restore \cs{@tabclassz}.}
% \begin{macro}{\adl@org@classz}
% \changes{v1.4-3-3}{1999/06/25}
%	{Introduced to restore \cs{@classz}.}
% \begin{macro}{\adl@org@@startpbox}
% \changes{v1.4-3-3}{1999/06/25}
%	{Introduced to restore \cs{@@startpbox}.}
% \begin{macro}{\adl@org@@endpbox}
% \changes{v1.4-3-3}{1999/06/25}
%	{Introduced to restore \cs{@@endpbox}.}
% \begin{macro}{\adl@org@endpbox}
% \changes{v1.4-3-3}{1999/06/25}
%	{Introduced to restore \cs{@endpbox}.}
% \begin{macro}{\adl@org@cline}
% \changes{v1.4-3-3}{1999/06/25}
%	{Introduced to restore \cs{cline}.}
% \changes{v1.4-3-6}{1999/06/25}
%	{Introduced to refer original version in modified \cs{cline}.}
% The other declarative stuffs are the sequence of |\let| to capture the
% original definitions of macros that we will modify afterword.  The main
% purpose of them is to nullify the modification when dash-line functions
% are inactive, while |\adl@org@cline| is also referred in its modified
% version.
% \end{macro}\end{macro}\end{macro}\end{macro}\end{macro}\end{macro}\end{macro}
%

%    \begin{macrocode}
\let\adl@org@arrayclassz\@arrayclassz
\let\adl@org@tabclassz\@tabclassz
\let\adl@org@classz\@classz
\let\adl@org@@startpbox\@@startpbox
\let\adl@org@@endpbox\@@endpbox
\let\adl@org@endpbox\@endpbox
\let\adl@org@cline\cline

%%^L
%    \end{macrocode}
%
% 
%
% \subsection{Initialization}
% \label{sec:imp-init}
%
% \begin{macro}{\adl@array}
% \changes{v1.4-3-1}{1999/06/25}
%	{Introduced to save original definition of \cs{@array}.}
% \begin{macro}{\@array}
% \changes{v1.4-3-1}{1999/06/25}
%	{Drastically modified to avoid copy-and-modify.}
% \begin{macro}{\adl@noalign}
% \changes{v1.7-2}{2004/5/21}
% 	{Introduced to fix a bug of \textsf{colortbl}.}
%

% \LaTeX's macro |\@array| is modified to save and initialize registers and
% data structures which are |\global|-ly updated in order to allow nested
% |array|\slash|tabular|.  This saving and initializing are performed by
% |\adl@arrayinit| as explained below.  The problem in the modification is
% that the code of |\@array| in \textsf{array} is completely different from
% that of \LaTeX{} original.
%
% The main difference is that \LaTeX{} builds |\@preamble| locally, while
% \textsf{array} does globally exploiting the fact that the lifetime of
% |\@preamble| ends before another |array|\slash|tabular| appears in a
% column.  The latter implementation will work well unless the building
% process in |\@mkpream| produces something referred after |\@preamble| is
% thrown into \TeX's {\em stomach}.  In our implementation, unfortunately,
% the number of columns has to be counted in \break|\@mkpream| and will be
% referred by |\hdashline| and the vertical line drawing procedure.
%
% Thus we have to change the column counting mechanism depending on whether
% or not \textsf{array} is in use.  The simplest way could be to copy the codes
% of \LaTeX{} and \textsf{array} and modify them appropriately examining the
% value of |\ifadl@usingarypkg|.  However this solution is vulnerable to the
% modification of the original version and thus we wish to refuse it as far
% as possible.
%
% \SpecialIndex{\everyvbox}
% \SpecialIndex{\adl@everyvbox}
% Therefore, we use a trick with |\everyvbox| in which |\adl@arrayinit| is
% temporarily included to initialize registers and locally set
% |\adl@columns| to the number of columns |\global|-ly counted by
% |\adl@ncol|.  This trick work well so far because;
%

% \begin{itemize}
% \item
% the first |\vbox|, |\vtop| or |\vcenter| made by |\@array| is the vertical 
% box surrounding |\halign|, and;
%
% \item
% in |\@array| of \textsf{array} the box is opened {\em after} the preamble
% is constructed;
% \end{itemize}
%

% and will hopefully work in future.
%
% \changes{v1.5-3}{2003/07/11}
%	{Move \cs{adl@inactivate} to \cs{@array} from \cs{adl@arrayinit}.}
% \changes{v1.54-2}{2003/08/25}
%	{Invoke \cs{adl@activate} if not \cs{ifadl@inactive}.}
% \changes{v1.7-2}{2004/5/21}
% 	{Make \cs{adl@noalign} \cs{let}-equal to \cs{noalign}.}
% \SpecialIndex{\arrayrulecolor}
% \SpecialIndex{\doublerulesepcolor}
% \SpecialIndex{\dashgapcolor}
% Next, if |\ifadl@inactive| is true, |\adl@inactivate| is invoked to
% inactivate dash-line functions.  Otherwise, |\adl@activate| is invoked to
% activate them because an inactivated |array|\slash|tabular|
% may have active children in it.  Finally, |\adl@noalign| is made
% |\let|-equal to |\noalign| so that |\arrayrulecolor|,
% |\doublerulesepcolor| and |\dashgapcolor| are expanded with |\noalign| in
% the environment.
% \end{macro}\end{macro}\end{macro}
%
% \begin{macro}{\@@array}
% \changes{v1.4-3-1}{1999/06/25}
%	{Introduced because \textsf{array} uses it.}
%

% \SpecialIndex{\@tabarray}
% Another stuff for the compatibility with \textsf{array} is to |\let| a
% control sequence |\@@array| be equal to |\@array| because it is referred
% in |\@tabarray| in \textsf{array}.
% \end{macro}
%

%    \begin{macrocode}

%% Initialization

\let\adl@array\@array
\def\@array{\adl@everyvbox\everyvbox
        \everyvbox{\adl@arrayinit \the\adl@everyvbox \everyvbox\adl@everyvbox}%
        \ifadl@inactive \adl@inactivate \else \adl@activate \fi
        \let\adl@noalign\noalign
        \adl@array}
\let\@@array\@array

%    \end{macrocode}
%
% \begin{macro}{\adl@arrayinit}
% \changes{v1.4-3-9}{1999/06/25}
%	{The bug of saving \cs{adl@colsR} is fixed.}
% \changes{v1.5-1}{2000/07/12}
%	{Use new macro \cs{adl@arraysave} to save registers/structures.}
% \begin{macro}{\adl@arraysave}
% \changes{v1.5-1}{2000/07/12}
%	{Introduced to use in modified \cs{CC@} of \textsf{colortab}.}
% \SpecialIndex{\adl@height}
% \SpecialIndex{\adl@depth}
% \SpecialIndex{\adl@currentcolumn}
% \SpecialIndex{\adl@totalheight}
%

% As described in \S\ref{sec:imp-decl}, registers updated |\global|-ly,
% which are |\adl@height|, |\adl@depth|, |\adl@currentcolumn| and
% |\adl@totalheight|, are saved in |\adl@arrayinit| by calling
% |\adl@arraysave|, and also given initial values.  The macro also saves the
% following data structures and initializes them to empty lists.
% \end{macro}
%

% \begin{macro}{\adl@rowsL}
% \begin{macro}{\adl@rowsR}
% \begin{macro}{\adl@rowsLsave}
% \begin{macro}{\adl@rowsRsave}
% \begin{macro}{\adl@colsL}
% \begin{macro}{\adl@colsR}
% \begin{macro}{\adl@colsLsave}
% \begin{macro}{\adl@colsRsave}
% \begin{itemize}
% \item
% In the real implementation, the data structure $R$ is split into two
% lists;
%

% \begin{eqnarray*}
% \cs{adl@rowsL}&=&R^L=\langle\langle C^L_1,h_1\rangle,\ldots\rangle\\
% \cs{adl@rowsR}&=&R^R=\langle\langle C^R_1,h_1\rangle,\ldots\rangle
% \end{eqnarray*}
%

% and they are saved into |\adl@rowsLsave| and |\adl@rowsRsave|.
%
% \item
% When the $i\th$ row is building, $C^L_i$ and $C^R_i$ are constructed in
% the macros |\adl@colsL| and |\adl@colsR|.  They are saved
% into |\adl@colsLsave| and |\adl@colsRsave|.
% \end{itemize}
% \end{macro}\end{macro}\end{macro}\end{macro}
% \end{macro}\end{macro}\end{macro}\end{macro}
%

% \begin{macro}{\adl@connect}
% \begin{macro}{\adl@discard}
% \changes{v1.6-1}{2003/8/25}
%	{Add initializetion of \cs{adl@discard}.}
% \SpecialIndex{\@elt}
% \SpecialOptIndex{longtable}{environment}
%

% In the real implementation, $e^i_j$ is represented by a control sequence
% |\@elt|, and ${\it connect}(i)$ by |\adl@connect|.  They are made
% |\let|-equal to |\relax| to keep them from expansion during $R$ is
% constructed.  In |longtable| environment, ${\it connect}(i)$ for negative
% vertical space inserted by |\\[|\meta{h}|]| or a horizontal line has
% another representation |\adl@discard| to indicate it corresponds to a
% discardable item of page breaking.  Since this representation, however, is
% nonsense in usural |array|\slash|tabular| even if they are included in
% |\longtable|, we define |\adl@discard| as |\adl@connect| so that it
% transforms itself into |\adl@connect| when it is added to
% |\adl@rowsL|\slash|R| by |\xdef|.  Note that |\adl@discard| is made
% |\let|-equal to |\relax| to inhibit the transformation at the beginning of
% |longtable| environment.
% \end{macro}\end{macro}
% 
% \changes{v1.4-3-1}{1999/06/25}
%	{Modified for new column counting in preamble construction.}
%

% Then, we set to |\adl@columns| to the value of |\adl@ncol| {\em locally}.
% As explained above, this has an effect with \textsf{array} because
% |\adl@arrayinit| is called {\em after} the preamble is generated.  Without
% \textsf{array}, on the other hand, this assignment has no effect but safe
% because it is included in a group of |\vbox| etc.
% 
% \changes{v1.4-3-3}{1999/06/25}
%	{Modified to call \cs{adl@inactivate}.}
% \changes{v1.5-3}{2003/07/11}
%	{Move \cs{adl@inactivate} from \cs{adl@arrayinit} to \cs{@array}.}
%

% \end{macro}
% \SpecialIndex{\adl@height}
% \SpecialIndex{\adl@depth}
% \SpecialIndex{\adl@currentcolumn}
% \SpecialIndex{\adl@totalheight}
% \SpecialIndex{\adl@heightsave}
% \SpecialIndex{\adl@depthsave}
% \SpecialIndex{\adl@currentcolumnsave}
% \SpecialIndex{\adl@totalheightsave}
%

%    \begin{macrocode}
\def\adl@arrayinit{%
        \adl@arraysave
        \global\adl@height\z@ \global\adl@depth\z@
        \global\adl@currentcolumn\@ne \global\adl@totalheight\z@
        \gdef\adl@rowsL{}\gdef\adl@rowsR{}\gdef\adl@colsL{}\gdef\adl@colsR{}%
        \let\@elt\relax \let\adl@connect\relax \def\adl@discard{\adl@connect}%
        \adl@columns\adl@ncol}
\def\adl@arraysave{%
        \adl@heightsave\adl@height
        \adl@depthsave\adl@depth
        \adl@currentcolumnsave\adl@currentcolumn
        \adl@totalheightsave\adl@totalheight
        \let\adl@rowsLsave\adl@rowsL
        \let\adl@rowsRsave\adl@rowsR
        \let\adl@colsLsave\adl@colsL
        \let\adl@colsRsave\adl@colsR}

%    \end{macrocode}
%
% \begin{macro}{\adl@inactivate}
% \changes{v1.4-3-3}{1999/06/25}
%	{Introduced to inactivate \cs{@arrayclassz} etc.}
% \changes{v1.5-3}{2003/07/11}
%	{Change \cs{adl@inactivate} caller to \cs{@array}.}
% \changes{v1.7-1}{2004/5/21}
% 	{Remove \cs{cline} because our own version is needed for null-wide.}
% If |\ADLinactivate| has effect and thus |\ifadl@inactive| is true, the
% macro |\adl@inactivate| is called from |\@array|\footnote{

%

% Before v1.53, \cs{adl@inactivate} was called from \cs{adl@arrayinit} and
% thus invokded {\em after} the preamble of \textsf{array} is built.  This
% was incorrect of course and made inactive version of \texttt{p},
% \texttt{m} and \texttt{b} produce nothing.}.
%

% This |\let|-s the following control sequences be equal to their
% counterparts in \LaTeX{} and/or \textsf{array} package.
%

% \changes{v1.54-2}{2003/08/25}
%	{Add \cs{adl@argcr} to inactivation.}
%

% \begin{itemize}\item[]
% \SpecialIndex{\@arrayclassz}
% \SpecialIndex{\@tabclassz}
% \SpecialIndex{\@classz}
% \SpecialIndex{\@@startpbox}
% \SpecialIndex{\@@endpbox}
% \SpecialIndex{\@endpbox}
% \SpecialIndex{\adl@cr}
% \SpecialIndex{\adl@argcr}
% \SpecialIndex{\adl@endarray}
% |\@arrayclassz|\quad
% |\@tabclassz|\quad
% |\@classz|\quad
% |\@@startpbox|\quad
% |\@@endpbox|\\
% |\@endpbox|\quad
% |\adl@cr|\quad
% |\adl@argcr|\quad
% |\adl@endarray|
% \end{itemize}
%

% Note that we have to inactivate both |\@@endpbox| for \LaTeX{} and
% |\@endpbox| for \textsf{array}, while |\@startpbox| for \textsf{array} is
% not necessary because it is unmodified.  Also note that \break|\@classz|
% has to
% be |\let|-equal to |\adl@org@classz| only if \textsf{array} is in use,
% because \LaTeX{} does not define |\@classz| but refers it which is either
% |\@arrayclassz| or |\@tabclassz|.  Yet another remark is that we have to
% conceal |\cr| for |\adl@cr|\slash|\adl@argcr| and |\crcr| for
% |\adl@endarray| by bracing them from \TeX's |\halign| mechanism that
% searches them when an |array|\slash|tabular| has an nested
% |array|\slash|tabular|.  This could be done by a tricky |\let|-assignment
% such as;
%

% \begin{quote}
% |\iffalse{\let\adl@cr\cr \iffalse}\fi|
% \end{quote}
%

% but we simply use |\def| instead of |\let| because of clarity.
%
% We also |\let| the following be {\it no-operation} or their inactive
% versions.
%

% \begin{itemize}\item[]
% \SpecialIndex{\adl@hline}
% \SpecialIndex{\adl@ihdashline}
% \SpecialIndex{\adl@cdline}
% \SpecialIndex{\adl@@vlineL}
% \SpecialIndex{\adl@@vlineR}
% \SpecialIndex{\adl@vlineL}
% \SpecialIndex{\adl@vlineR}
% |\adl@hline|\quad
% |\adl@ihdashline|\quad
% |\adl@cdline|\quad
% |\adl@@vlineL|\quad
% |\adl@@vlineR|\\
% |\adl@vlineL|\quad
% |\adl@vlineR|
% \end{itemize}
% Note that we have to inactivate both |\adl@@vlineL| and |\adl@vlineL|,
% because the latter is referred when \textsf{array} is in use while the
% former is referred otherwise.  Their |R| relatives are also inactivated by 
% the same reason.
% \end{macro}
%

% \SpecialIndex{\@arrayclassz}
% \SpecialIndex{\@tabclassz}
% \SpecialIndex{\@classz}
% \SpecialIndex{\@@startpbox}
% \SpecialIndex{\@@endpbox}
% \SpecialIndex{\@endpbox}
% \SpecialIndex{\adl@cr}
% \SpecialIndex{\adl@argcr}
% \SpecialIndex{\adl@endarray}
% \SpecialIndex{\adl@hline}
% \SpecialIndex{\adl@ihdashline}
% \SpecialIndex{\adl@cdline}
% \SpecialIndex{\adl@@vlineL}
% \SpecialIndex{\adl@@vlineR}
% \SpecialIndex{\adl@vlineL}
% \SpecialIndex{\adl@vlineR}
% \SpecialIndex{\adl@org@arrayclassz}
% \SpecialIndex{\adl@org@tabclassz}
% \SpecialIndex{\adl@org@classz}
% \SpecialIndex{\adl@org@@startpbox}
% \SpecialIndex{\adl@org@@endpbox}
% \SpecialIndex{\adl@org@endpbox}
% \SpecialIndex{\adl@cr}
% \SpecialIndex{\adl@argcr}
% \SpecialIndex{\adl@org@cline}
% \SpecialIndex{\adl@inactivehdl}
% \SpecialIndex{\adl@inactivecdl}
% \SpecialIndex{\adl@inactivevl}
%    \begin{macrocode}
\def\adl@inactivate{%
        \let\@arrayclassz\adl@org@arrayclassz
        \let\@tabclassz\adl@org@tabclassz
        \ifadl@usingarypkg \let\@classz\adl@org@classz \fi
        \let\@@startpbox\adl@org@@startpbox
        \let\@@endpbox\adl@org@@endpbox
        \let\@endpbox\adl@org@endpbox
        \def\adl@cr{\cr}%
        \def\adl@argcr##1{\cr}%
        \def\adl@endarray{\crcr}%
        \let\adl@hline\@gobbletwo
        \let\adl@ihdashline\adl@inactivehdl
        \let\adl@cdline\adl@inactivecdl
        \let\adl@@vlineL\adl@inactivevl
        \let\adl@@vlineR\adl@inactivevl
        \let\adl@vlineL\adl@inactivevl
        \let\adl@vlineR\adl@inactivevl}
%    \end{macrocode}
% 
% \begin{macro}{\adl@activate}
% \changes{v1.54-2}{2003/08/25}
%	{Introduced to activate \cs{@arrayclassz} etc.\ again.}
%

% On the other hand, if |\ifadl@inactive| is false, the macro |\adl@activate| 
% is called from |\@array| to make inactivated macros active again in order
% to cope with the case in which an inactive \textsf{array}\slash
% \textsf{tabular} has active children in it\footnote{

%

% Before v1.54, an active \textsf{array}\slash\textsf{tabular} in an
% inactive parent was not activated.}.
%

% To do that, |\adl@activate| makes |\@arrayclassz| etc.\ |\let|-equal to
% their active version |\adl@act@arrayclassz| etc.\ which will be defined
% (|\let|-equal to) as our own |\@arrayclassz| etc.\ in \S\ref{sec:imp-sh}.
% \end{macro}
%

% \SpecialIndex{\@arrayclassz}
% \SpecialIndex{\@tabclassz}
% \SpecialIndex{\@classz}
% \SpecialIndex{\@@startpbox}
% \SpecialIndex{\@@endpbox}
% \SpecialIndex{\@endpbox}
% \SpecialIndex{\adl@cr}
% \SpecialIndex{\adl@argcr}
% \SpecialIndex{\adl@endarray}
% \SpecialIndex{\adl@hline}
% \SpecialIndex{\adl@ihdashline}
% \SpecialIndex{\adl@cdline}
% \SpecialIndex{\adl@@vlineL}
% \SpecialIndex{\adl@@vlineR}
% \SpecialIndex{\adl@vlineL}
% \SpecialIndex{\adl@vlineR}
% \SpecialIndex{\adl@act@arrayclassz}
% \SpecialIndex{\adl@act@tabclassz}
% \SpecialIndex{\adl@act@classz}
% \SpecialIndex{\adl@act@@startpbox}
% \SpecialIndex{\adl@act@@endpbox}
% \SpecialIndex{\adl@act@endpbox}
% \SpecialIndex{\adl@act@cr}
% \SpecialIndex{\adl@act@argcr}
% \SpecialIndex{\adl@act@cline}
% \SpecialIndex{\adl@act@endarray}
% \SpecialIndex{\adl@act@hline}
% \SpecialIndex{\adl@act@ihdashline}
% \SpecialIndex{\adl@act@cdline}
% \SpecialIndex{\adl@act@@vlineL}
% \SpecialIndex{\adl@act@@vlineR}
%    \begin{macrocode}
\def\adl@activate{%
        \let\@arrayclassz\adl@act@arrayclassz
        \let\@tabclassz\adl@act@tabclassz
        \ifadl@usingarypkg \let\@classz\adl@act@classz \fi
        \let\@@startpbox\adl@act@@startpbox
        \let\@@endpbox\adl@act@@endpbox
        \let\@endpbox\adl@act@endpbox
        \let\adl@cr\adl@act@cr
        \let\adl@argcr\adl@act@argcr
        \let\adl@endarray\adl@act@endarray
        \let\adl@hline\adl@act@hline
        \let\adl@ihdashline\adl@act@ihdashline
        \let\adl@cdline\adl@act@cdline
        \let\adl@@vlineL\adl@act@@vlineL
        \let\adl@@vlineR\adl@act@@vlineR
        \let\adl@vlineL\adl@act@@vlineL
        \let\adl@vlineR\adl@act@@vlineR}

%%^L
%    \end{macrocode}
%
% \changes{v1.6-1}{2003/8/25}
%	{Add a summary of activation/inactivation.}
% The summary of the activation and inactivation is shown in
% Table~\ref{tab:act}.
% 
% 
% 
% \begin{table}
% \catcode`\|12
% \caption{Active and Inactive Operations}\label{tab:act}
% \def\!{\texttt{\char`\\\char`\\}}
% \centerline{\begin{tabular}[t]{|l||l|l|}\hline
% command&active&inactive\\\hline\hline
% \tt l c r&&\\
% \qquad with \textsf{array}&	\cs{adl@act@classz}&
%				\cs{adl@org@classz}\\
% \qquad without \textsf{array}&\cs{adl@act@tabclassz}&
%				\cs{adl@org@tabclassz}\\
% &				\cs{adl@act@arrayclassz}&
%				\cs{adl@org@arrayclassz}\\\hline
% {\tt p m b} (open)&&\\
% \qquad with \textsf{array}&	\cs{adl@act@classz}&
%				\cs{adl@org@classz}\\
% \qquad without \textsf{array}&\cs{adl@act@@startpbox}&
%				\cs{adl@org@@startpbox}\\\hline
% {\tt p m b} (close)&		\cs{adl@act@@endpbox}&
%				\cs{adl@org@@endpbox}\\\hline
% \tt \char`\|/:/;&		\cs{adl@act@@vlineL/R}&
%				\cs{adl@inactivevl}\\\hline
% \!&				$\to$\cs{adl@act@(arg)cr}&
%				$\to$\cs{cr}\\\hline
% \cs{hline}&			$\to$\cs{adl@act@hline}&
%				$\to$\cs{@gobbletwo}\\\hline
% \cs{hdashline}&		$\to$\cs{adl@act@ihdashline}&
%				$\to$\cs{adl@inactivehdl}\\\hline
% \cs{cdashline}&		$\to$\cs{adl@act@cdline}&
%				$\to$\cs{adl@inactivecdl}\\\hline
% \end{tabular}}
% \end{table}
% 
% 
%
% \subsection{Making Preamble}
% \label{sec:imp-pream}
%
% Each preamble character is converted to a part of |\halign|'s preamble as
% follows.
%

% \begin{itemize}
% \item
% \begin{macro}{\adl@colhtdp}
% \changes{v1.5-2}{2002/06/26}
%	{The pseudo-formal description of $\langle\hbox{\it put-lrc}\rangle$
%	 is modified.}
% \SpecialIndex{\adl@box}
% \SpecialIndex{\adl@height}
% \SpecialIndex{\adl@depth}
% `|l|', `|r|' and `|c|' are converted to the following \meta{lrc}.
%

% \begingroup \catcode`\|12
% \begin{eqnarray*}
% \meta{lrc}&::=&
%	[\cs{hfil}]\meta{put-lrc}[\cs{hfil}]\\
% \meta{put-lrc}&::=&
%	\cs{setbox}\cs{adl@box}\cs{hbox}\Meta{lrc-contents}\\
% &&	\cs{adl@colhtdp }\cs{unhbox}\cs{adl@box}\\
% \meta{lrc-contents}&::=&
%	\hbox{\texttt{\$\cs{relax}\#\$}}\ |\\
% &&	\hbox{\texttt{\#\cs{unskip}}}\\
% \end{eqnarray*}
% \endgroup
%

% That is, the content of a column is at first packed into the |\box|
% register |\adl@box|, then its height and depth are compared to
% |\adl@height| and |\adl@depth| by the macro |\adl@colhtdp|, and finally
% the box is put with leading and/or trailing |\hfil|.
% \end{macro}
%
% \item
% \begin{macro}{\adl@vlineL}
% \begin{macro}{\adl@vlineR}
% \def\G{{\mit\Gamma}}
% `\verb!|!', `|:|' and |;{|\meta{dash}|/|\meta{gap}|}| are converted to the
% following \meta{vline}.
%

% \begingroup \catcode`\|12
% \begin{eqnarray*}
% \meta{vline}&::=&
%	[\cs{hskip}\cs{doublerulesep}]\meta{vline-LR}\\
% \meta{vline-LR}&::=&
%	\cs{adl@vlineL}\Meta{$\G_d$}\Meta{$\G_g$}Meta{c}
%		\hbox{{\tt\char`\{}\meta{d}/\meta{g}{\tt\char`\}}}\ |\\
% &&	\cs{adl@vlineR}\Meta{$\G_d$}\Meta{$\G_g$}Meta{c}
%		\hbox{{\tt\char`\{}\meta{d}/\meta{g}{\tt\char`\}}}\\
% \meta{d}&::=&
%	0\ |&\hbox to.3\textwidth{\ldots\ for `\texttt{|}'\hfil}\cr
% &&	\cs{dashlinedash}\ |&
%		\hbox to.3\textwidth{\ldots\ for `\texttt{:}'\hfil}\cr
% &&	\meta{dash}&
%		\hbox to.3\textwidth{\ldots\ for `\texttt{;}'\hfil}\cr
% \meta{g}&::=&
%	0\ |&\hbox to.3\textwidth{\ldots\ for `\texttt{|}'\hfil}\cr
% &&	\cs{dashlinegap}\ |&
%		\hbox to.3\textwidth{\ldots\ for `\texttt{:}'\hfil}\cr
% &&	\meta{gap}&
%		\hbox to.3\textwidth{\ldots\ for `\texttt{;}'\hfil}\cr
% \end{eqnarray*}
% \endgroup
%

% \SpecialIndex{\CT@arc@}
% \SpecialIndex{\adl@dashgapcolor}
% Note that \meta{c} is the column number (leftmost is 1) where the
% character appears, and \meta{$\G_d$} and \meta{$\G_g$} is the
% color of dashs and gaps specified in |\CT@arc@| and |\adl@dashgapcolor|.
% \end{macro}\end{macro}
% \end{itemize}
%

% \SpecialIndex{\adl@currentcolumn}
% Additionally, each column except for the last one has;
%

% \begin{quote}
% |\global\advance\adl@currentcolumn\@ne|
% \end{quote}
%

% just before |&| to increment |\adl@currentcolumn|.  Other features, such
% as inserting spaces of |\arraycolsep|\slash|\tabcolsep|, are as same as
% original scheme.  This means that |@|\Meta{text} and |!|\Meta{text} of
% \textsf{array} are {\em not} handled specially although it could interfere
% with drawing vertical lines.  Therefore, we have the problem 1 shown in
% \S\ref{sec:prob}, which is very hard to solve.  Note that the measurement
% of the column of `|p|' of \LaTeX{} original is done by (modified)
% |\@@startpbox| and |\@@endpbox| and thus the preamble for `|p|' is not
% modified.  In case with \textsf{array}, however, the preambles for `|p|'
% and its relatives `|m|' and `|b|' are modified to set |\adl@box| to the
% box for them.
%
% \begin{macro}{\adl@mkpream}
% \begin{macro}{\@mkpream}
% \changes{v1.4-3-1}{1999/06/25}
%	{Modified for new column counting and control sequence redefinition.}
% \changes{v1.71-1}{2004/08/31}
%	{Modified to nullify \cs{adl@startmbox} and \cs{adl@endmbox} for
%	\textsf{array's} \texttt{m}-columns.}
% \SpecialIndex{\adl@colhtdp}
% \SpecialIndex{\adl@vlineL}
% \SpecialIndex{\adl@vlineR}
% \SpecialIndex{\adl@startmbox}
% \SpecialIndex{\adl@endmbox}
% \SpecialIndex{\adl@preaminit}
% \SpecialIndex{\afterassignment}
%

% To make the preamble shown above, |\@mkpream| is modified to |\let|
% control sequences |\adl@colhtpd|, |\adl@vlineL| and |\adl@vlineR| be
% |\relax| in order to keep them from being expanded by |\edef|\slash
% |\xdef| for the preamble construction.  The control sequences
% |\adl@startmbox| and |\adl@endmbox| for \texttt{m}-columns of
% \textsf{array} are also made |\let|-equal to |\relax|.
%
% Giving them their own definition is done by |\adl@preaminit| that is
% called using |\afterassignment| after |\@preamble| is made by
% |\adl@mkpream|, the original version of |\@mkpream|.  If \textsf{array} is
% not in use, |\@mkpream| is followed by an |\edef| of |\@preamble| to add
% |\ialign| etc.\ and thus |\adl@preaminit| is properly called {\em after}
% this final {\em assignment} to make |\@preamble|.
%
% With \textsf{array}, on the other hand, calling |\adl@preaminit| is safe
% because |\@mkpream| is followed by |\xdef| for |\@preamble| too, but has
% no effect because it is in the group for |\@mkpream|.  This grouping,
% however, gives us an easier way to give those control sequences their own
% definition.  That is, we simply initiate them with the definitions that
% will be regained when the group is closed.
% 
% \SpecialIndex{\adl@ncol}
% \SpecialIndex{\adl@columns}
% \SpecialIndex{\ifadl@leftrule}
% The modified |\@mkpream| also initializes |\adl@ncol| and
% |\ifadl@leftrule|, and set |\adl@columns| to the value of |\adl@ncol|
% locally after the preamble is made.  This has an effect in case without
% \textsf{array} because the body of |array|\slash|tabular| is in the same
% grouping context of |\@mkpream|.  With \textsf{array}, on the other hand,
% this assignment has no effect but safe because it is included in a group
% of |\@mkpream|'s own.
% \end{macro}\end{macro}
%

%    \begin{macrocode}

%% Making Preamble

\let\adl@mkpream\@mkpream
\def\@mkpream#1{\let\adl@colhtdp\relax
        \let\adl@vlineL\relax \let\adl@vlineR\relax
        \let\adl@startmbox\relax \let\adl@endmbox\relax
        \global\adl@ncol\@ne \adl@leftruletrue
        \adl@mkpream{#1}\adl@columns\adl@ncol \afterassignment\adl@preaminit}

%    \end{macrocode}
%
% \begin{macro}{\@addamp}
% \changes{v1.4-3-1}{1999/06/25}
%	{Modified for new column counting in preamble construction.}
% \SpecialIndex{\adl@currentcolumn}
% \SpecialIndex{\adl@ncol}
%

% The macro |\@addamp| is also modified to add the code for incrementing the
% counter |\adl@currentcolumn| to |\@preamble| with |&|.  The counter
% |\adl@ncol| is also incremented by |\@addamp| so that we can refer its
% value as \meta{c} of |\adl@vlineL|\slash|R|\@.  This increment is done
% |\global|-ly in order that we locally set |\adl@columns| to the counting
% result outside of the group for |\@mkpream| of \textsf{array}.  Therefore,
% whether or not \textsf{array} is in use, |\adl@columns| will have a
% correct value and will be correctly referred by |\hdashline| to know how
% many columns are specified in the preamble.  Note that this |\global|
% assignment is safe because the life time of |\adl@ncol| is same as that of
% |\@preamble|.
% \end{macro}
%

%    \begin{macrocode}
\def\@addamp{\if@firstamp\@firstampfalse \else
        \@addtopreamble{\global\advance\adl@currentcolumn\@ne &}%
        \global\advance\adl@ncol\@ne \fi}

%    \end{macrocode}
%
% \SpecialIndex{\ifadl@usingarypkg}
% Since the implementation of |\@testpach| and macros for class-0 characters 
% (i.e. |l|, |r| and |c|) is completely different between \LaTeX{} and
% \textsf{array}, we have to have two versions switched by
% |\adl@usingarypkg|.
% 
% \subsubsection*{With \textsf{array}}
% 
% \begin{macro}{\@testpach}
% \changes{v1.4-3-1}{1999/06/25}
%	{The version for \textsf{array} is introduced.}
% \SpecialIndex{\@arrayrule}
% \SpecialIndex{\adl@arrayrule}
% \SpecialIndex{\adl@arraydashrule}
% \SpecialIndex{\adl@argarraydashrule}
%

% Although we introduced two preamble characters `|:|' and `|;|', we did not
% introduce new character {\em class} because we want to minimize the
% modification of original codes.  Therefore, `|:|' and `|;|' is classified
% into class-1 together with `\verb!|!'\@.  Since these characters obviously
% have their own appropriate operations, |\@testpach| is modified so that
% |\@arrayrule|, which is invoked from |\@mkpream| in the case of class-1
% character, is |\let|-equal to the macro corresponding to each character.
% \end{macro}
%

%    \begin{macrocode}
\ifadl@usingarypkg
\def\@testpach{\@chclass
 \ifnum \@lastchclass=6 \@ne \@chnum \@ne \else
  \ifnum \@lastchclass=7 5 \else
   \ifnum \@lastchclass=8 \tw@ \else
    \ifnum \@lastchclass=9 \thr@@
   \else \z@
   \ifnum \@lastchclass = 10 \else
   \edef\@nextchar{\expandafter\string\@nextchar}%
   \@chnum
   \if \@nextchar c\z@ \else
    \if \@nextchar l\@ne \else
     \if \@nextchar r\tw@ \else
   \z@ \@chclass
   \if\@nextchar |\@ne \let\@arrayrule\adl@arrayrule \else
   \if\@nextchar :\@ne \let\@arrayrule\adl@arraydashrule \else
   \if\@nextchar ;\@ne \let\@arrayrule\adl@argarraydashrule \else
    \if \@nextchar !6 \else
     \if \@nextchar @7 \else
      \if \@nextchar <8 \else
       \if \@nextchar >9 \else
  10
  \@chnum
  \if \@nextchar m\thr@@\else
   \if \@nextchar p4 \else
    \if \@nextchar b5 \else
   \z@ \@chclass \z@ \@preamerr \z@ \fi \fi \fi \fi \fi \fi
   \fi \fi  \fi  \fi  \fi  \fi  \fi \fi \fi \fi \fi \fi}

%    \end{macrocode}
% 
% \begin{macro}{\@classz}
% \changes{v1.4-3-1}{1999/06/25}
%	{Introduced because \textsf{array} uses it.}
% \changes{v1.71-1}{2004/08/31}
%	{Modified to call \cs{adl@startmbox} and \cs{adl@endmbox} for
%	\textsf{array}'s \texttt{m}-columns.}
% \SpecialIndex{\adl@putlrc}
% \SpecialIndex{\adl@box}
% \SpecialIndex{\adl@startmbox}
% \SpecialIndex{\adl@endmbox}
%

% In \textsf{array}, |array| and |tabular| share common macro for class-0
% named |\@classz|, which also generates the preamble for `|p|', `|m|' and
% `|b|'.  Thus we modify it to measure the height and depth of the
% class-0 column by the macro |\adl@putlrc|, and to set |\adl@box| to the
% box for `|p|' and its relatives.  Note that |m|-type preambles
% (|@chnum|${}=3$) have to be generated by |\adl@startmbox| and
% |\adl@endmbox| because a |\vcenter| construct cannot be assigned by
% to |\adl@box| by |\setbox| directly.
% \end{macro}
%

%    \begin{macrocode}
\def\@classz{\@classx
   \@tempcnta \count@
   \prepnext@tok
   \@addtopreamble{\ifcase \@chnum
      \hfil
      \adl@putlrc{\d@llarbegin \insert@column \d@llarend}\hfil \or
      \hskip1sp\adl@putlrc{\d@llarbegin \insert@column \d@llarend}\hfil \or
      \hfil\hskip1sp\adl@putlrc{\d@llarbegin \insert@column \d@llarend}\or
   \setbox\adl@box\hbox \adl@startmbox{\@nextchar}\insert@column
        \adl@endmbox\or
   \setbox\adl@box\vtop \@startpbox{\@nextchar}\insert@column \@endpbox \or
   \setbox\adl@box\vbox \@startpbox{\@nextchar}\insert@column \@endpbox
  \fi}\prepnext@tok}
%    \end{macrocode}
% 
% \begin{macro}{\adl@class@start}
% \changes{v1.4-3-1}{1999/06/25}
%	{Introduced for class number identification.}
% \begin{macro}{\adl@class@iiiorvii}
% \changes{v1.4-3-1}{1999/06/25}
%	{Introduced for class number identification.}
%

% Another stuffs for compatibility are to refer the class number of the
% beginning of preamble which is different between \LaTeX{} and
% \textsf{array}, and that of `|p|' or `|@|' to get the argument of `|;|' as 
% explained later.  In case with \textsf{array}, the former is class-4
% and we use `|@|' (class-7) for the latter.
% \end{macro}\end{macro}
%

%    \begin{macrocode}
\def\adl@class@start{4}
\def\adl@class@iiiorvii{7}

%    \end{macrocode}
% 
% \subsubsection*{Without \textsf{array}}
%
% \begin{macro}{\@testpach}
% \SpecialIndex{\@arrayrule}
% \SpecialIndex{\adl@arrayrule}
% \SpecialIndex{\adl@arraydashrule}
% \SpecialIndex{\adl@argarraydashrule}
%

% The reason why and how we modify |\@testpach| of \LaTeX{} is same as those 
% of \textsf{array}.
% \end{macro}
%

%    \begin{macrocode}
\else
\def\@testpach#1{\@chclass \ifnum \@lastchclass=\tw@ 4\relax \else
        \ifnum \@lastchclass=\thr@@ 5\relax \else
                \z@ \if #1c\@chnum \z@ \else
                    \if #1l\@chnum \@ne \else
                    \if #1r\@chnum \tw@ \else
                \@chclass
                    \if #1|\@ne \let\@arrayrule\adl@arrayrule \else
                    \if #1:\@ne \let\@arrayrule\adl@arraydashrule \else
                    \if #1;\@ne \let\@arrayrule\adl@argarraydashrule \else
                    \if #1@\tw@ \else
                    \if #1p\thr@@ \else \z@ \@preamerr 0\fi
        \fi  \fi  \fi  \fi  \fi  \fi  \fi  \fi  \fi}

%    \end{macrocode}
%
% \begin{macro}{\@arrayclassz}
% \begin{macro}{\@tabclassz}
% \SpecialIndex{\adl@putlrc}
%

% Since \LaTeX{} has two macros for class-0, one for |array| and the other
% for |tabular|, we have to modify both.  Since the box for `|p|' is opened
% by |\@@startpbox|, however, we may not worry about it.
% \end{macro}\end{macro}
%

%    \begin{macrocode}
\def\@arrayclassz{\ifcase \@lastchclass \@acolampacol \or \@ampacol \or
                \or \or \@addamp \or
                \@acolampacol \or \@firstampfalse \@acol \fi
        \edef\@preamble{\@preamble
                \ifcase \@chnum
                    \hfil\adl@putlrc{$\relax\@sharp$}\hfil
                \or \adl@putlrc{$\relax\@sharp$}\hfil
                \or \hfil\adl@putlrc{$\relax\@sharp$}\fi}}
\def\@tabclassz{\ifcase \@lastchclass \@acolampacol \or \@ampacol \or
                \or \or \@addamp \or
                \@acolampacol \or \@firstampfalse \@acol \fi
        \edef\@preamble{\@preamble
        \ifcase \@chnum
                    \hfil\adl@putlrc{\@sharp\unskip}\hfil
                \or \adl@putlrc{\@sharp\unskip}\hfil
                \or \hfil\hskip\z@ \adl@putlrc{\@sharp\unskip}\fi}}
%    \end{macrocode}
%
% \begin{macro}{\adl@class@start}
% \changes{v1.4-3-1}{1999/06/25}
%	{Introduced for class number identification.}
% \begin{macro}{\adl@class@iiiorvii}
% \changes{v1.4-3-1}{1999/06/25}
%	{Introduced for class number identification.}
% In \LaTeX{}, the beginning of preamble is class-6 and we use `|p|'
% (class-3) to get the argument of `|;|'.
% \end{macro}\end{macro}
%

%    \begin{macrocode}
\def\adl@class@start{6}
\def\adl@class@iiiorvii{3}
\fi

%    \end{macrocode}
% 
% Hereafter, codes for \LaTeX{} and \textsf{array} are common again.
%
% \begin{macro}{\adl@putlrc}
% \changes{v1.5-2}{2002/06/26}
%	{\cs{adl@putlrc} must do \cs{unhbox}\cs{adl@box} to make glues
%	 effective.} 
% \SpecialIndex{\adl@box}
% \SpecialIndex{\adl@colhtdp}
%

% The macro |\adl@putlrc| is for class-0 preamble characters to set
% |\adl@box| to the contents of a column, measure its height\slash depth by
% |\adl@colhtdp| and put the box by |\unhbox| (not by |\box|) in order to
% make the glues in the contents effective.
% \end{macro}
%

%    \begin{macrocode}
\def\adl@putlrc#1{\setbox\adl@box\hbox{#1}\adl@colhtdp \unhbox\adl@box}

%    \end{macrocode}
% 
% \begin{macro}{\adl@arrayrule}
% \changes{v1.4-3-1}{1999/06/25}
%	{Modified to replace \cs{adl@columns} with \cs{adl@ncol}.}
% \begin{macro}{\adl@arraydashrule}
% \changes{v1.4-3-1}{1999/06/25}
%	{Modified to replace \cs{adl@columns} with \cs{adl@ncol}.}
% \begin{macro}{\adl@argarraydashrule}
% \changes{v1.4-3-1}{1999/06/25}
%	{Modified to replace \cs{adl@columns} with \cs{adl@ncol}.}
% \changes{v1.4-3-1}{1999/06/25}
%	{Modified to pretend \texttt{p} or \texttt{@} depending on if
%	\textsf{array} is in use.}
% \begin{macro}{\adl@xarraydashrule}
% \changes{v1.4-3-1}{1999/06/25}
%	{Modified to refer \cs{adl@class@start} rather than \LaTeX's 6.}
% \changes{v1.4-3-2}{1999/06/25}
%	{Modified to add invisible rule of \cs{arrayrulewidth} wide if
%	\cs{ADLsome}\penalty\exhyphenpenalty\texttt{wide}.}
% \changes{v1.7-3}{2004/5/21}
%	{Modified to add color arguments to \cs{adl@vline}\texttt{L}\slash
%	\texttt{R}}
% \SpecialIndex{\@lastchclass}
% \SpecialIndex{\adl@class@start}
% \SpecialIndex{\ifadl@leftrule}
% \SpecialIndex{\adl@vlineL}
% \SpecialIndex{\adl@vlineR}
% \SpecialIndex{\adl@ncol}
% \SpecialIndex{\CT@arc@}
% \SpecialIndex{\adl@dashgapcolor}
% \SpecialIndex{\color}
% \def\G{{\mit\Gamma}}
%

% The preamble parts for vertical solid- and dash-lines are constructed by
% the macros |\adl@arrayrule| for `\verb!|!', |\adl@arraydashrule| for
% `|:|', and |\adl@argarraydashrule| for `|;|'.\@  The macro;
%

% \begin{quote}
% |\adl@xarraydashrule|\Meta{$c^L$}\Meta{$c^R$}|{|\meta{d}|/|\meta{g}|}|
% \end{quote}
%

% is invoked by them to perform common operations.  It at first checks the
% preamble character is the first element of the preamble
% (|\@lastchclass|${}=|\adl@class@start|$) or it follows another character
% for vertical line (|\@lastchclass|${}=1$).  If this is not satisfied, the
% vertical line is put at the right edge of a column and thus
% |\ifadl@leftrule| is set to false.  Then it adds
% |\adl@vlineL|\Meta{$\G_d$}\Meta{$\G_g$}

% \Meta{$c^L$}|{|\meta{d}|/|\meta{g}|}| if |\ifadl@leftrule| is
% true indicating the vertical line will appear at the left edge of the
% column \meta{$c^L$}, or |\adl@vlineR|\Meta{$\G_d$}\Meta{$\G_g$}
% \Meta{$c^R$}|{|\meta{d}|/|\meta{g}|}| otherwise.  Note that \meta{$c^L$}
% is always 1 for {\em main} preamble while \meta{$c^R$} is the column
% number given by |\adl@ncol|, but \meta{$c^L$} may not be 1 for the
% preamble of |\multicolumn| as described in \S\ref{sec:imp-mc}.  Also note
% that $\G_d$ and $\G_g$ are |\CT@arc@| and |\adl@dashgapcolor|
% respectively whose bodies are |\color| for dashes and gaps specified by
% |\arrayrulecolor| and |\dashgapcolor|, or |\relax| if they are not colored.
% 
% In addition, an invisible |\vrule| of |\arrayrulewidth| wide is added if
% both |\ADL#some#wide| and |\ADLactivate| are in effect, i.e.\ both
% |\ifadl@zwrule| and |\ifadl@inactive| are false, to keep a space for
% the vertical line having {\em real} width.
% \end{macro}\end{macro}\end{macro}\end{macro}
%
% \begin{macro}{\adl@classv}
% \begin{macro}{\adl@classvfordash}
% \SpecialIndex{\@nextchar}
% \SpecialIndex{\@chclass}
% \SpecialIndex{\@classv}
% \SpecialIndex{\adl@class@iiiorvii}
%

% The argument of `|;|' is not provided by |\adl@argarraydashrule| but is
% directly passed from the preamble text through |\@nextchar|.  This direct
% passing is implemented by the following trick.  The macro
% |\adl@argarraydashrule| set |\@chclass| to |\adl@class@iiiorvii| to 
% pretend it is for `|p|' if \textsf{array} is not in use, or `|@|'
% otherwise.  Then it temporally changes the definition of |\@classv|, which
% is incidentally for the argument of `|p|' and `|@|' in case without/with
% \textsf{array} respectively, to |\adl@classvfordash| to process the
% argument of `|;|' rather than that of `|p|' or `|@|'\@.  Then
% |\adl@classvfordash| is invoked by |\@mkpream| and it adds the argument to
% |\@preamble|.  Finally, it restores the definition of |\@classv| and set
% |\@chclass| to 1 to indicate that the last item is a vertical line
% specification.
% \end{macro}\end{macro}
%

%    \begin{macrocode}
\def\adl@arrayrule{%
        \adl@xarraydashrule
                {\@ne}{\adl@ncol}{{\z@/\z@}}}
\def\adl@arraydashrule{%
        \adl@xarraydashrule
                {\@ne}{\adl@ncol}%
                {{\dashlinedash/\dashlinegap}}}
\def\adl@argarraydashrule{%
        \adl@xarraydashrule
                {\@ne}{\adl@ncol}{}%
        \@chclass\adl@class@iiiorvii \let\@classv\adl@classvfordash}
\def\adl@xarraydashrule#1#2#3{%
        \ifnum\@lastchclass=\adl@class@start\else
        \ifnum\@lastchclass=\@ne\else
                \adl@leftrulefalse \fi\fi
        \ifadl@zwvrule\else \ifadl@inactive\else
                \@addtopreamble{\vrule\@width\arrayrulewidth
                        \@height\z@ \@depth\z@}\fi \fi
        \ifadl@leftrule
                \@addtopreamble{\adl@vlineL{\CT@arc@}{\adl@dashgapcolor}%
                        {\number#1}#3}%
        \else   \@addtopreamble{\adl@vlineR{\CT@arc@}{\adl@dashgapcolor}%
                        {\number#2}#3}\fi}
\let\adl@classv\@classv
\def\adl@classvfordash{\@addtopreamble{{\@nextchar}}\let\@classv\adl@classv
        \@chclass\@ne}

%%^L
%    \end{macrocode}
% 
%
% \subsection{Building Columns}
% \label{sec:imp-col}
%
% \begin{macro}{\adl@preaminit}
% \begin{macro}{\adl@colhtdp}
% \begin{macro}{\adl@vlineL}
% \begin{macro}{\adl@vlineR}
%

% If \textsf{array} is not in use, after the |\@preamble| is completed, the
% control sequences for macros in it should regain their own definition.
% The macro |\adl@preaminit| performs this operation for macros we
% introduced, |\adl@colhtdp|, |\adl@vlineL| and |\adl@vlineR|.  For the case 
% with \textsf{array}, we will call |\adl@preaminit| in \textsf{arydshln} to 
% initiate them with the definitions as described later.
% \end{macro}\end{macro}\end{macro}\end{macro}
%

%    \begin{macrocode}

%% Building Columns

\def\adl@preaminit{\let\adl@colhtdp\adl@@colhtdp
        \let\adl@vlineL\adl@@vlineL \let\adl@vlineR\adl@@vlineR}

%    \end{macrocode}
%
% \begin{macro}{\adl@@colhtdp}
% For the measurement of the height and depth of a row, |\adl@@colhtdp|
% compares |\adl@height| and |\adl@depth| to the height and depth of
% |\adl@box| which contains the main part of the column to be built, and
% |\global|-ly updates the registers if they are less.
% \end{macro}
%

%    \begin{macrocode}
\def\adl@@colhtdp{%
        \ifdim\adl@height<\ht\adl@box \global\adl@height\ht\adl@box \fi
        \ifdim\adl@depth<\dp\adl@box \global\adl@depth\dp\adl@box\fi}

%    \end{macrocode}
%
% \begin{macro}{\adl@@vlineL}
% \changes{v1.7-3}{2004/5/21}
% 	{Color arguments are added.}
% \begin{macro}{\adl@@vlineR}
% \changes{v1.7-3}{2004/5/21}
% 	{Color arguments are added.}
% \begin{macro}{\adl@@ivline}
% \changes{v1.7-3}{2004/5/21}
% 	{Invocations of \cs{adl@setcolor} are added.}
% \begin{macro}{\adl@setcolor}
% \changes{v1.7-3}{2004/5/21}
% 	{Introduced to color vertical lines.}
% \begin{macro}{\adl@nocolor}
% \changes{v1.7-3}{2004/5/21}
% 	{Introduced to examine if coloring is specified.}
% \begin{macro}{\adl@dashcolor}
% \changes{v1.7-3}{2004/5/21}
% 	{Introduced as the temporary variable of color specification of
%	 dashes.}
% \begin{macro}{\adl@gapcolor}
% \changes{v1.7-3}{2004/5/21}
% 	{Introduced as the temporary variable of color specification of
%	 gaps.}
% \SpecialIndex{\color}
% \SpecialIndex{\current@color}
% \def\G{{\mit\Gamma}} \let\g\gamma
% The macro
% |\adl@@vlineL|\meta{$\G_d$}\meta{$\G_g$}\meta{c}|{|\meta{d}|/|\meta{g}|}|
% adds the element
% $e=\langle c,d,g\rangle={}$

% |\@elt|\Meta{c}\Meta{d}\Meta{g}\Meta{$\g_d$}\Meta{$\g_g$} to the tail of
% the list |\adl@colsL| to construct $C^L_i$, where $\g_d$ and $\g_g$ are
% the color specifications given by  |\color| macros in $\G_d$ and $\G_g$.
% The macro |\add@@vlineR| performs similar operation but the element is
% added to the head of |\adl@colsR| for $C^R_i$ because it is processed
% right-to-left manner. The argument \meta{d} and \meta{g} are extracted by
% the macro |\adl@ivline| which converts given dimensional values of them to
% integers. It also set \meta{d} and \meta{g} to 0 (i.e. solid-line) if one
% of given values are not positive, in order to make it sure that one dash
% segment has positive length.  Then it invokes |\adl@setcolor| to define
% |\adl@dashcolor| and |\adl@gapcolor| with the color specification of
% $\G_d$ and $\G_g$.  Since |\adl@setcolor| locally expands |\color| macro
% in $\G_d$ and $\G_g$ to define |\currnet@color| that becomes the body of
% |\adl@dashcolor| ($\g_d$) and |\adl@gapcolor| ($\g_g$) with expansion,
% different |\color| specifications of a color, such as |\color{red}|
% and |\color[rgb]{1,0,0}|, will produce a unified result such as
% |{rgb 1 0 0}|.  If $\G_d$ or $\G_g$ is |\relax| which is the body of
% |\adl@nocolor|, $\g_d$ or $\g_g$ is also |\relax| to indicate dashes are
% colored (or not colored) as done in outer world and gaps are transparent.
% \end{macro}\end{macro}\end{macro}\end{macro}\end{macro}\end{macro}\end{macro}
%

%    \begin{macrocode}
\def\adl@@vlineL#1#2#3#4{\adl@ivline#4\@nil{#1}{#2}%
        \xdef\adl@colsL{\adl@colsL
                \@elt{#3}{\number\@tempcnta}{\number\@tempcntb}%
                        {\adl@dashcolor}{\adl@gapcolor}}}
\def\adl@@vlineR#1#2#3#4{\adl@ivline#4\@nil{#1}{#2}%
        \xdef\adl@colsR{%
                \@elt{#3}{\number\@tempcnta}{\number\@tempcntb}%
                        {\adl@dashcolor}{\adl@gapcolor}%
                \adl@colsR}}
\def\adl@ivline#1/#2\@nil#3#4{%
        \@tempdima#1\relax \@tempcnta\@tempdima
        \@tempdima#2\relax \@tempcntb\@tempdima
        \ifnum\@tempcnta>\z@ \else \@tempcnta\z@ \@tempcntb\z@ \fi
        \ifnum\@tempcntb>\z@ \else \@tempcnta\z@ \@tempcntb\z@ \fi
        \adl@setcolor\adl@dashcolor{#3}\adl@setcolor\adl@gapcolor{#4}}
\def\adl@setcolor#1#2{\def\@tempa{#2}\ifx\@tempa\adl@nocolor \def#1{\relax}%
        \else{#2\xdef#1{\current@color}}\fi}
\def\adl@nocolor{\relax}
%    \end{macrocode}
%
% \begin{macro}{\adl@colhtdp}
% \changes{v1.4-3-1}{1999/06/25}
%	{Initialized by calling \cs{adl@preaminit}.}
% \begin{macro}{\adl@vlineL}
% \changes{v1.4-3-1}{1999/06/25}
%	{Initialized by calling \cs{adl@preaminit}.}
% \begin{macro}{\adl@vlineR}
% \changes{v1.4-3-1}{1999/06/25}
%	{Initialized by calling \cs{adl@preaminit}.}
% \SpecialIndex{\adl@preaminit}
%

% After |\adl@@colhtdp|, |\adl@@vlineL| and |\adl@@vlineR| are defined, 
% we call |\adl@preaminit| to |\let| their single |@| counterparts be equal
% to them.  Therefore, in case with \textsf{array}, |\adl@colhtdp| etc.\ are 
% temporarily |\relax| when |\@preamble| is being generated in the group of
% |\@mkpream|, and regain their own definition outside the group where
% the completed |\@preamble| is referred.
% \end{macro}\end{macro}\end{macro}
%

%    \begin{macrocode}
\adl@preaminit

%    \end{macrocode}
% 
% \begin{macro}{\adl@inactivevl}
% \changes{v1.4-3-3}{1999/06/25}
%	{Introduced to emulate `\texttt{:}' and \texttt{;} by
%	\protect\VERTBAR.} 
% \changes{v1.7-3}{2004/5/21}
% 	{Modified to color the \cs{vline} by the first argument.}
% \SpecialIndex{\ifadl@zwvrule}
%

% If |\ADLinactivate| is in effect, |\adl@vlineL|\slash|R| and
% |\adl@@vlineL|\slash|R| are |\let|-equal
% to |\adl@inactivevl|.  This macro simply put a |\vrule| by |\vline| with
% |\color| (or |\relax|) in its first argument and
% with\slash without negative |\hskip| of a half of |\arrayrulewidth| wide
% depending on |\ifadl@zwvrule|, discarding other arguments.
% \end{macro}
%

%    \begin{macrocode}
\def\adl@inactivevl#1#2#3#4{\ifadl@zwvrule \hskip-.5\arrayrulewidth \fi
        {#1\vline}\ifadl@zwvrule \hskip-.5\arrayrulewidth \fi}

%    \end{macrocode}
% 
% 
% \begin{macro}{\@@startpbox}
% \begin{macro}{\@@endpbox}
% \begin{macro}{\@endpbox}
% \changes{v1.4-3-1}{1999/06/25}
%	{Introduced because \textsf{array} uses it.}
% \begin{macro}{\adl@startmbox}
% \changes{v1.71-1}{2004/08/31}
%	{Introduced to the bug fix of \textsf{array}'s \texttt{m}-columns.}
% \begin{macro}{\adl@endmbox}
% \changes{v1.71-1}{2004/08/31}
%	{Introduced to the bug fix of \textsf{array}'s \texttt{m}-columns.}
% \SpecialIndex{\@finalstrut}
% \SpecialIndex{\@startpbox}
% \SpecialIndex{\adl@box}
% \SpecialIndex{\adl@colhtdp}
%

% The macros to make |\parbox| for `|p|' (and `|b|' of
% \textsf{array}), |\@@startpbox| and \hbox{|\@@endpbox|}, are modified for
% height\slash depth measurement.  The code for |\@@endpbox| is based on
% that of \LaTeXe{} to fix the bug of |\strut|-ing in \LaTeX-2.09, but
% |\@finalstrut| is manually expanded because it is not available in
% \LaTeX-2.09.
% 
% In \textsf{array}, |\@@endpbox| is not used but |\@endpbox| is.
% Therefore, we |\let| them be equal.  As for |\@startpbox|, however, we may 
% not worry about it because we have modified \hbox{|\@classz|} in
% \S\ref{sec:imp-pream} for the measurement.  However, we have to take care
% of |m|-type columns specially because its body |\vcenter| cannot be
% assigned directly to |\adl@box| by |\setbox|\footnote{

%

% The author had forgotten this fact until Morten H\o gholm pointed out it.
% Thanks Morten.}.
%

% Thus we enclose a |$\vcenter{|\ldots|}$| construct in a |\hbox| and assign
% it to |\adl@box|.  The macro |\adl@startmbox| opens the construct with
% \textsf{array}'s |\@startpbox|, while |\adl@endmbox| closes it calling
% |\adl@org@endpbox| which is the unmodified |\@endpbox| of \textsf{array}
% and measures the height and depth of the |\hbox| by |\adl@colhtdp|.
% \end{macro}\end{macro}\end{macro}\end{macro}\end{macro}
%

%    \begin{macrocode}
\def\@@startpbox#1{\setbox\adl@box\vtop\bgroup \hsize#1\@arrayparboxrestore}
\def\@@endpbox{\unskip \ifhmode \nobreak
        \vrule\@width\z@\@height\z@\@depth\dp\@arstrutbox \fi
        \par \egroup \adl@colhtdp \box\adl@box \hfil}
\let\@endpbox\@@endpbox
\def\adl@startmbox{\bgroup $\vcenter\@startpbox}
\def\adl@endmbox{\adl@org@endpbox $\egroup \adl@colhtdp \box\adl@box \hfil}

%%^L
%    \end{macrocode}
%
% 
%
% \subsection{Multi-columns}
% \label{sec:imp-mc}
%
% \begin{macro}{\multicolumn}
% \changes{v1.4-3-1}{1999/06/25}
%	{Modified for several reason.}
% \begin{macro}{\adl@preamble}
% \begin{macro}{\adl@mcaddamp}
% \changes{v1.4-3-1}{1999/06/25}
%	{Introduced for the complaint on multiple columns if with
%	\textsf{array}.}
% \begin{macro}{\adl@activatepbox}
% \changes{v1.4-3-1}{1999/06/25}
%	{Introduced to do nothing if with \textsf{array}.}
%

% The macro |\multicolumn| is modified for the followings.
%

% \begin{itemize}
% \item
% The macros to construct the parts of |\@preamble| for vertical lines,
% |\adl@arrayrule|, |\adl@arraydashrule| and |\adl@argarraydashrule|, have
% to perform operations slightly different from those for main preamble.
% Thus they are |\def|-ined to multi-column version |\adl@mcarrayrule|, etc.
% \changes{v1.3}{1998/10/08}
% 	{\cs{def}-s for \cs{adl@mcarrayrule} etc. are enclosed in a group.}
% These |\def|-initions are enclosed in a group so that they are not
% affected to |array| or |tabular| which may occur in the third argument of
% |\multicolumn|.  In order to make |\@preamble| work well outside of the
% group containing |\@makepream|, |\adl@preamble| is |\global|-ly |\let|-equal
% to |\@preamble| just after |\@makepream| in the group and then
% reverse |\let|-assignment is performed just after the group is closed.
% These global assignment is unnecessary with \textsf{array} because
% |\@preamlbe| is constructed |\global|-ly, but safe.
% 
% Since this grouping nullifies the effect of |\adl@preaminit| called in
% |\@mkpream|, we call |\adl@preaminit| again after the group closing.
%
% \item
% In \textsf{array}, |\@addamp| to make |\@preamble| for |\multicolumn| has
% a different definition from that for main one.  Thus it is |\let|-equal
% to |\adl@mcaddamp| whose definition is switched by |\ifadl@usingarypkg|.
% 
% \item
% If \textsf{array} is in use, |\@preamble| has to be |\xdef|-ed once again
% by |\@addpreamble| with an |\@empty| argument after |\@mkpreamble| to
% expand the contents of |\toks| registers.  This is performed whether or
% not with \textsf{array} because it is safe.
%
% \item
% As done in |\@array|, |\set@typeset@protect| is replaced with direct
% |\let|.
%
% \item
% If without \textsf{array}, |\@startpbox| and |\@endpbox| should be
% |\let|-equal to their |@@| counterparts, while should not with
% \textsf{array}.  Thus we define |\adl@activatepbox| to do or not to do so
% depending on |\ifadl@usingarypkg|.
% 
% \item
% The counter |\adl@currentcolumn| is |\global|-ly incremented by the first
% argument of |\multicolumn| (number of columns to be |\span|-ned).
% \end{itemize}
%

% Note that |\adl@columns| is modified by |\@mkpream|, but it is not
% referred |\adl@mcarrayrule| etc., and its value is restored before
% referred by |\hdashline|, etc.
% \end{macro}\end{macro}\end{macro}\end{macro}
%

% \SpecialIndex{\adl@arrayrule}
% \SpecialIndex{\adl@arraydashrule}
% \SpecialIndex{\adl@argarraydashrule}
% \SpecialIndex{\adl@mcarrayrule}
% \SpecialIndex{\adl@mcarraydashrule}
% \SpecialIndex{\adl@mcargarraydashrule}
% \SpecialIndex{\ifadl@usingarypkg}
% \SpecialIndex{\adl@currentcolumn}
%

%    \begin{macrocode}

%% Multi-Columns

\def\multicolumn#1#2#3{\multispan{#1}\begingroup \begingroup
        \def\adl@arrayrule{\adl@mcarrayrule{#1}}%
        \def\adl@arraydashrule{\adl@mcarraydashrule{#1}}%
        \def\adl@argarraydashrule{\adl@mcargarraydashrule{#1}}%
        \let\@addamp\adl@mcaddamp
        \@mkpream{#2}\@addtopreamble\@empty
        \global\let\adl@preamble\@preamble \endgroup
        \let\@preamble\adl@preamble
        \def\@sharp{#3}\let\protect\relax
        \adl@activatepbox
        \adl@preaminit
        \@arstrut \@preamble\hbox{}\endgroup
        \global\advance\adl@currentcolumn#1\ignorespaces}
\ifadl@usingarypkg
        \def\adl@mcaddamp{\if@firstamp\@firstampfalse \else\@preamerror5\fi}
        \let\adl@activatepbox\relax
\else
        \let\adl@mcaddamp\@addamp
        \def\adl@activatepbox{\let\@startpbox\@@startpbox
                \let\@endpbox\@@endpbox}
\fi

%    \end{macrocode}
%
% \begin{macro}{\adl@mcarrayrule}
% \begin{macro}{\adl@mcarraydashrule}
% \begin{macro}{\adl@mcargarraydashrule}
% \changes{v1.4-3-1}{1999/06/25}
%	{Modified to pretend \texttt{p} or \texttt{@} depending on if
%	\textsf{array} is in use.}
% \SpecialIndex{\adl@arrayrule}
% \SpecialIndex{\adl@arraydashrule}
% \SpecialIndex{\adl@argarraydashrule}
% \SpecialIndex{\adl@currentcolumn}
%

% The preamble parts for vertical lines are constructed by the macros
% |\adl@mcarrayrule|, |\adl@mcarraydashrule| and |\adl@mcargarraydashrule|
% which are passed the first argument \meta{$n$} of |\multicolumn| to know
% the number of columns to be |\span|-ned.  They are similar to their
% relatives for main preamble, |\adl@arrayrule|, etc., but the arguments
% \meta{$c^L$} and \meta{$c^R$} passed to |\adl@xarraydashrule| are;
% $$
% c^L=c,\qquad c^R=c+n-1
% $$
% where $c={}$|\adl@currentcolumn|.  This makes leading vertical lines drawn
% at the left edge of the leftmost |\span|-ned column and trailing ones at
% the right edge of the rightmost column.
% \end{macro}\end{macro}\end{macro}
%

%    \begin{macrocode}
\def\adl@mcarrayrule#1{\@tempcnta#1\advance\@tempcnta\adl@currentcolumn
        \advance\@tempcnta\m@ne
        \adl@xarraydashrule
                {\adl@currentcolumn}{\@tempcnta}{{\z@/\z@}}}
\def\adl@mcarraydashrule#1{\@tempcnta#1\advance\@tempcnta\adl@currentcolumn
        \advance\@tempcnta\m@ne
        \adl@xarraydashrule
                {\adl@currentcolumn}{\@tempcnta}%
                {{\dashlinedash/\dashlinegap}}}
\def\adl@mcargarraydashrule#1{\@tempcnta#1\advance\@tempcnta\adl@currentcolumn
        \advance\@tempcnta\m@ne
        \adl@xarraydashrule
                {\adl@currentcolumn}{\@tempcnta}{}%
        \@chclass\adl@class@iiiorvii \let\@classv\adl@classvfordash}

%%^L
%    \end{macrocode}
%
% 
%
% \subsection{End of Rows}
% \label{sec:imp-eor}
%
% \begin{macro}{\@xarraycr}
% \changes{v1.4-3-1}{1999/06/25}
%	{The version for \textsf{array} is introduced.}
% \begin{macro}{\@xtabularcr}
% \begin{macro}{\@xargarraycr}
% \begin{macro}{\@yargarraycr}
% \SpecialIndex{\ifadl@usingarypkg}
%
 
% At the end of the $i\th$ row, we have to calculate $h_i$ which is the height
% plus depth of the row, and add elements \meta{$C^L_i,h_i$} and
% \meta{$C^R_i,h_i$} to $R^L$ and $R^R$.  To do this, |\cr|-s in the macros
% \hbox{|\@xarraycr|}, |\@xtabularcr|, |\@xargarraycr| are replaced with our
% own |\adl@cr|.  The macro |\@yargarraycr|\meta{dimen} is also modified but
% its |\cr| is replaced with |\adl@argcr|\meta{dimen} to add (negative)
% |\dimen| to $h_i$.  Note that |\@xargarraycr|\meta{dimen} uses ordinary
% |\adl@cr| because the extra vertical space of \meta{dimen} is inserted to
% the last column.
% 
% Note that the implementation of |\@xarraycr| is slightly different between
% \LaTeX{} and \textsf{array}, we have to have two versions and choose one.
% \end{macro}\end{macro}\end{macro}\end{macro}
%

%    \begin{macrocode}

%% End of row

\ifadl@usingarypkg
\def\@xarraycr{\@ifnextchar[{\@argarraycr}{\ifnum0=`{}\fi\adl@cr}}
\else
\def\@xarraycr{\@ifnextchar[{\@argarraycr}{\ifnum0=`{\fi}${}\adl@cr}}
\fi
\def\@xtabularcr{\@ifnextchar[{\@argtabularcr}{\ifnum0=`{\fi}\adl@cr}}
\def\@xargarraycr#1{\@tempdima#1\advance\@tempdima\dp\@arstrutbox
        \vrule\@height\z@\@depth\@tempdima\@width\z@
        \adl@cr}
\def\@yargarraycr#1{\adl@argcr{#1}\noalign{\vskip #1}}

%    \end{macrocode}
%
% \begin{macro}{\adl@cr}
% \begin{macro}{\adl@argcr}
% The macro |\adl@cr| and |\adl@argcr| perform |\cr| and then invoke the
% common macro |\adl@@cr|\meta{x}.  The argument \meta{x} is the extra
% (negative) vertical space for |\adl@argcr|, while it is 0 for |\adl@cr|.
% \end{macro}\end{macro}
%
% \begin{macro}{\adl@@cr}
% \changes{v1.4-3-6}{1999/06/25}
%	{Modified to set \cs{adl@finaldepth}.}
% \changes{v1.6-1}{2003/8/25}
%	{Modified to insert \cs{adl@discard}.}
% \SpecialIndex{\@arstrutbox}
% \SpecialIndex{\adl@totalheight}
% \SpecialIndex{\adl@colsL}
% \SpecialIndex{\adl@colsR}
% \SpecialIndex{\adl@rowsL}
% \SpecialIndex{\adl@rowsR}
% \SpecialIndex{\adl@connect}
% \SpecialIndex{\adl@discard}
% \SpecialIndex{\adl@finaldepth}
% \SpecialIndex{\adl@height}
% \SpecialIndex{\adl@depth}
% \SpecialIndex{\adl@currentcolumn}
%

% The macro |\adl@@cr|\meta{x} at first calculate $h_i$ as follows.  The
% registers |\adl@height|${}=\eta$ and |\adl@depth|${}=\delta$ have the
% maximum height and depth of the columns in the row.  However, they could
% be smaller than the height and/or depth of |\@arstrutbox|, $\eta_s$ and
% $\delta_s$.  If so, the height and/or depth of the row are $\eta_s$ and
% $\delta_s$.  Therefore, $h_i$ is caluclated by;
% $$
% h_i = \max(\eta,\eta_s) + \max(\delta,\delta_s).
% $$
% Additionally, if the extra space \meta{x} is negative, a vertical space of
% $x$ is inserted below the row\footnote{

%

% Before v1.54, negative \meta{x} shrinks the hight of the row by $\vert
% x\vert$.  Although the former result may be more appropriate if the row
% has vertical lines than the current because lines extrude to the next row
% now, new feature is considered compatible with original
% \texttt{array}\slash \texttt{tabular}.}.
%

% Thus the integer value of $h_i+x$ is
% |\global|-ly added to |\adl@totalheight|, and the elements
% \meta{$C^L_i{=}\cs{adl@colsL},h_i$} and
% \meta{$C^R_i{=}\cs{adl@colsR},h_i$} are added to the tail of
% $R^L={}$|\adl@rowsL| and $R^C={}$|\adl@rowsR|\@.  If $x$ is not 0
% (negative), {\it discard}$(x)$ or {\it connect}$(x)$ is also added after
% \meta{$C^{L/R}_i,h_i$} according to the current environment (|longtable|
% or not).  In the real implementation, $R^L$ and $R^C$ has the following
% format of \meta{rows}.
% 
%

% \begin{eqnarray*}
% \meta{rows}&::=&
%	[\meta{row}\hbox{\tt;}]^*\\
% \meta{row}&::=&
%	\hbox{{\tt(}\meta{cols}{\tt/}\meta{$h_i$}{\tt)}}\\
% \meta{cols}&::=&
%	[\hbox{\cs{elt}\Meta{c}\Meta{d}\Meta{g}}]^*\ \vert\qquad\qquad&
%	    \hbox to.3\textwidth{\ldots\ $C^L$ or $C^R$\hss}\cr
% &&	\cs{adl@connect}\ \vert&
%	    \hbox to.3\textwidth{\ldots\ for {\it connect}$(h_i)$\hss}\cr
% &&	\cs{adl@discard}\ \vert&
%	    \hbox to.3\textwidth{\ldots\ for {\it discard}$(h_i)$\hss}\cr
% &&	\cs{relax}&
%	    \hbox to.3\textwidth{\ldots\ for {\it disconnect}$(h_i)$\hss}\cr
% \noalign{\vskip-\baselineskip}
% \end{eqnarray*}
%

% Since |\adl@discard| is |\def|-ined as |\adl@connect| by |\adl@arrayinit|, 
% added |\adl@discard| transforms itself into |\adl@connect| if current
% envrionment is not |longtable|.  Otherwise, as we make |\adl@discard|
% |\let|-equal to |\relax| when a |longtable| environment starts, it keeps
% its own form.
% 
% Then, |\adl@finaldepth| is set to |\adl@depth| if $x$ is zero, or to zero
% otherwise (negative), in order to make the depth |array|\slash|tabular|
% equal to that of the last row.  Finally, |\adl@colsL|, |\adl@colsR|,
% |\adl@currentcolumn|, |\adl@height| and |\adl@depth| are reinitialized to
% process the next row.
% \end{macro}
%

%    \begin{macrocode}
\def\adl@cr{\cr\noalign{\adl@@cr\z@}}
\def\adl@argcr#1{\cr\noalign{\adl@@cr{#1}}}
\def\adl@@cr#1{
        \ifdim\adl@height<\ht\@arstrutbox \adl@height\ht\@arstrutbox\fi
        \ifdim\adl@depth<\dp\@arstrutbox \adl@depth\dp\@arstrutbox\fi
        \advance\adl@height\adl@depth
        \global\advance\adl@totalheight\adl@height
        \@tempdima#1\relax \global\advance\adl@totalheight\@tempdima
        \xdef\adl@rowsL{\adl@rowsL
                (\adl@colsL/\number\adl@height);%
                \ifdim#1=\z@\else (\adl@discard/\number\@tempdima);\fi}%
        \xdef\adl@rowsR{\adl@rowsR
                (\adl@colsR/\number\adl@height);%
                \ifdim#1=\z@\else (\adl@discard/\number\@tempdima);\fi}%
        \gdef\adl@colsL{}\gdef\adl@colsR{}
        \global\adl@currentcolumn\@ne
        \ifdim#1=\z@ \global\adl@finaldepth\adl@depth
        \else        \global\adl@finaldepth\z@\fi
        \global\adl@height\z@ \global\adl@depth\z@}

%%^L
%    \end{macrocode}
%
% 
%
% \subsection{Horizontal Lines}
% \label{sec:imp-hline}
%
% \begin{macro}{\hline}
% \changes{v1.4-3-6}{1999/06/25}
%	{Modified to set \cs{adl@finaldepth} to zero.}
% \changes{v1.7-1}{2004/5/21}
% 	{Modified to shift up if null-wide.}
% \begin{macro}{\cline}
% \changes{v1.4-3-6}{1999/06/25}
%	{Modified to set \cs{adl@finaldepth} to zero.}
% \changes{v1.7-1}{2004/5/21}
% 	{Modified to shift up if null-wide.}
% \SpecialIndex{\adl@hline}
% \SpecialIndex{\adl@xhline}
% \SpecialIndex{\adl@connect}
% \SpecialIndex{\adl@finaldepth}
% \SpecialIndex{\adl@org@cline}
%

% The macro |\hline| is modified to insert |\vskip-\arrayrulewidth| before
% drawing if |\ADLnullwidehline| is in effect, or to add the element ${\it
% connect}(w)={}$|(\adl@connect/|\allowbreak|\number\arrayrulewidth)| to the
% end of $R^L$ and $R^R$ by |\adl@hline| otherwise.  The other modifications 
% are to set |\adl@finaldepth| to zero for the case that the last vertical
% item is |\hline|, and to check if it is followed by not only |\hline| but
% also |\hdashline| by |\adl@xhline|.
% 
% The macro |\cline| is also modified to set |\adl@finaldepth| to zero.  As
% for the feature of |\ADLnullwidehline|, it inserts |\vskip-\arrayrulewidth| 
% to shift the line up befor drawing, and |\vskip\arrayrulewidth| after
% drawing to cancel the negative skip inserted by |\adl@org@cline|.
% \end{macro}\end{macro}
%

%    \begin{macrocode}

%% Horizontal Lines

\def\hline{\noalign{\ifnum0=`}\fi
        \ifadl@zwhrule \vskip-\arrayrulewidth
        \else \adl@hline\adl@connect\arrayrulewidth \fi
        \hrule\@height\arrayrulewidth
        \global\adl@finaldepth\z@
        \futurelet\@tempa\adl@xhline}
\def\cline#1{\noalign{\global\adl@finaldepth\z@
                \ifadl@zwhrule \vskip-\arrayrulewidth\fi}
        \adl@org@cline{#1}%
        \noalign{\ifadl@zwhrule \vskip\arrayrulewidth\fi}}

%    \end{macrocode}
%
% \begin{macro}{\hdashline}
% \changes{v1.4-3-5}{1999/06/25}
%	{Modified to make \cs{adl@hdashline} usable for
%	\cs{first}\slash\texttt{lasthdashline}.}
% \begin{macro}{\adl@hdashline}
% \changes{v1.4-3-5}{1999/06/25}
%	{Modified to be usable for
%	\cs{first}\slash\texttt{lasthdashline}.}
% \changes{v1.7-1}{2004/5/21}
% 	{Modified for null-wide horizontal lines.}
% \begin{macro}{\adl@ihdashline}
% \changes{v1.4-3-5}{1999/06/25}
%	{Introduced as the substitute of old \cs{adl@hdashline}.}
% \changes{v1.7-1}{2004/5/21}
% 	{\cs{adl@hline} is moved to \cs{adl@hdashline} for null-wide lines.}
% \SpecialIndex{\adl@hcline}
% \SpecialIndex{\dashlinedash}
% \SpecialIndex{\dashlinegap}
%

% The macro |\hdashline| calls |\adl@hdashline| to open the |\noalign|
% construct by the well-known trick |{\ifnum0=`}\fi| and then to invoke
% |\adl@ihdashline| checking the existence of its optional argument
% |[|\meta{dash}|/|\meta{gap}|]|.  Before the invocation, it inserts
% |\vskip-\arrayrulewidth| if |\ADLnullwidehline| is in effect, or adds
% ${\it connect}(w)$ to the end of $R^L$ and $R^R$.  Then |\adl@ihdashline|
% closes the |\noalign| by |\ifnum0=`{\fi}| to start the pseudo row for the
% horizontal dash-line.  Before the dash-line is drawn by |\adl@hcline|
% which is also used for |\cdashline|, all the columns are |\span|-ned by
% giving |\adl@columns| to |\multispan|.  Finally, the |\noalign| is opened
% again and |\adl@xhline| is invoked to check whether |\h(dash)line| is
% followed.
% \end{macro}\end{macro}\end{macro}
% 
% \begin{macro}{\adl@inactivehdl}
% \changes{v1.4-3-3}{1999/06/25}
%	{Introduced to emulate \cs{hdashline} by \cs{hline}.}
% \changes{v1.7-1}{2004/5/21}
%	{Modified to shift up if null-wide.}
%

% If |\ADLinactivate| is in effect, |\adl@ihdashline| is |\let|-equal
% to |\adl@inactivehdl|.  This macro simply puts a |\hrule| discarding its
% arguments after inserting |\vskip|\allowbreak|-\arrayrulewidth| if
% |\ADLnullwidehline| is in effect.
% \end{macro}
%

%    \begin{macrocode}
\def\hdashline{\adl@hdashline\adl@ihdashline}
\def\adl@hdashline#1{\noalign{\ifnum0=`}\fi
        \ifadl@zwhrule \vskip-\arrayrulewidth
        \else \adl@hline\adl@connect\arrayrulewidth \fi
        \@ifnextchar[%]
                     {#1}%
                     {#1[\dashlinedash/\dashlinegap]}}
\def\adl@ihdashline[#1/#2]{\ifnum0=`{\fi}%
        \multispan{\adl@columns}\unskip \adl@hcline\z@[#1/#2]%
        \noalign{\ifnum0=`}\fi
        \futurelet\@tempa\adl@xhline}
\def\adl@inactivehdl[#1/#2]{\ifadl@zwhrule \vskip-\arrayrulewidth \fi
        \hrule\@height\arrayrulewidth
        \futurelet\@tempa\adl@xhline}

%    \end{macrocode}
%
% \begin{macro}{\adl@xhline}
% \SpecialIndex{\@xhline}
%

% The macro |\adl@xhline| is the counterpart of the original |\@xhline|.
% This is introduced to check the mixed sequence of |\hline| and
% |\hdashline|, and to add the element ${\it
% disconnect}(s)={}$|(\relax/\doublerulesep)| to the end of $R^L$ and $R^R$
% by |\adl@hline| if a pair of |\h(dash)line| is found.
% \end{macro}
%

%    \begin{macrocode}
\def\adl@xhline{\ifx\@tempa\hline \adl@ixhline\fi
        \ifx\@tempa\hdashline \adl@ixhline\fi
        \ifnum0=`{\fi}}
\def\adl@ixhline{\vskip\doublerulesep \adl@hline\relax\doublerulesep}

%    \end{macrocode}
%
% \begin{macro}{\adl@hline}
% \SpecialIndex{\adl@totalheight}
% \SpecialIndex{\adl@rowsL}
% \SpecialIndex{\adl@rowsR}
%

% The macro |\adl@hline|\meta{cs}\meta{dimen} |\global|-ly adds the integer
% value of \meta{dimen} to |\adl@totalheight| and adds the element
% |(|\meta{cs}|/\number|\meta{dimen}|)| to the tail of $R^L$ and $R^R$.  The
% arguments \meta{cs}\meta{dimen} are |\adl@connect\arrayrulewidth| for
% ${\it connect}(w)$ or |\relax\doublerulesep| for ${\it disconnect}(s)$.
% \end{macro}
%

%    \begin{macrocode}
\def\adl@hline#1#2{\@tempcnta#2
        \global\advance\adl@totalheight\@tempcnta
        \xdef\adl@rowsL{\adl@rowsL
                (#1/\number\@tempcnta);}%
        \xdef\adl@rowsR{\adl@rowsR
                (#1/\number\@tempcnta);}}

%    \end{macrocode}
%
% \begin{macro}{\cdashline}
% \changes{v1.4-3-7}{1999/06/25}
%	{Modified to call renamed \cs{adl@cdline}.}
% \begin{macro}{\adl@cdline}
% \changes{v1.4-3-7}{1999/06/25}
% 	{Renamed and modified to call renamed \cs{adl@cdlinea}\slash
%	\texttt{b}.} 
% \changes{v1.7-1}{2004/5/21}
% 	{Modified to shift up if null-wide.}
% \begin{macro}{\adl@cdlinea}
% \changes{v1.4-3-7}{1999/06/25}
% 	{Renamed.}
% \begin{macro}{\adl@cdlineb}
% \changes{v1.4-3-7}{1999/06/25}
% 	{Renamed.}
% \SpecialIndex{\@cline}
% \SpecialIndex{\adl@cla}
% \SpecialIndex{\adl@clb}
% \SpecialIndex{\adl@org@cline}
% \SpecialIndex{\adl@hcline}
%

% The macro |\cdashline| at first opens |\noalign| and then invokes
% |\adl@cdline| checking the existence of its optional argument
% |[|\meta{dash}|/|\meta{gap}|]|.  The macro |\adl@cdline| first inserts
% |\vskip-\arrayrulewidth| if |\ADLnullwidehline| is in effect.  Then it
% performs column |\span|-ing by the code based on that of |\@cline| in
% \LaTeX-2.09 because \LaTeXe's version will not work with \LaTeX-2.09.  The
% main job is done by |\adl@hcline| after the target columns are |\span|-ned
% by |\adl@cdlinea| or |\adl@cdlineb|.
% \end{macro}\end{macro}\end{macro}\end{macro}
%
% \begin{macro}{\adl@inactivecdl}
% \changes{v1.4-3-3}{1999/06/25}
%	{Introduced to emulate \cs{cdashline} by \cs{cline}.}
% \changes{v1.7-1}{2004/5/21}
% 	{Modified to invoke \cs{cline} rather than \cs{adl@orgcline} for
%	 null-wide.}
%

% If |\ADLinactivate| is in effect, |\adl@cdline| is |\let|-equal
% to |\adl@inactivecdl|.  This macro simply calls our own |\cline|, after
% closing the |\noalign| opened by |\cdashline|.
% \end{macro}
%

%    \begin{macrocode}
\def\cdashline#1{\noalign{\ifnum0=`}\fi
        \@ifnextchar[%]
                     {\adl@cdline[#1]}%
                     {\adl@cdline[#1][\dashl