% \CheckSum{1849}
%
% \title{The \package{tclldoc} package and class}
% \author{Lars Hellstr\"om^^A
% \thanks{E-mail: \texttt{Lars.Hellstrom@math.umu.se}}
% \emph{et al.}}
% \date{2003/07/19}
% \maketitle
%
% \iffalse
%
%<class|pkg>\NeedsTeXFormat{LaTeX2e}
%<class>\ProvidesClass{tclldoc}
%<compclass>\ProvidesClass{tcldoc}
%<pkg>\ProvidesPackage{tclldoc}
%<comppkg>\ProvidesPackage{tcldoc}
%<class|pkg|compclass|comppkg> [2003/04/05 v2.40
%<class|pkg> Tcl documentation
%<compclass|comppkg> tclldoc compatibility
%<class|compclass> class]
%<pkg|comppkg> package]
%
%<*driver>
\documentclass{ltxdoc}
\usepackage{amsmath}
\IfFileExists{xdoc2.sty}{\usepackage[dolayout]{xdoc2}[2000/11/18]}{}
\makeatletter
\@ifpackageloaded{xdoc2}{}{%
\@ifpackagelater{doc}{2000/05/20}{}{%
\let\XD@fragile@meta=\meta
\def\meta{%
\ifx \protect\@typeset@protect
\expandafter\futurelet \expandafter\@let@token
\expandafter\XD@fragile@meta
\else
\noexpand\meta
\fi
}%
}%
}
\@ifundefined{option}{%
\newenvironment{option}[1]{%
\trivlist
\if@inlabel\else \macro@cnt=\z@ \fi
\item[]\ignorespaces
}{\endtrivlist}%
}{}
\@ifundefined{instance}{%
\newenvironment{instance}[2]{%
\trivlist
\if@inlabel\else \macro@cnt=\z@ \fi
\item[]\ignorespaces
}{\endtrivlist}%
}{}
\makeatother
\DeclareRobustCommand\package[1]{\textsf{#1}}
\DeclareRobustCommand\Tcl{T\kern-0.1em cl}
\DeclareRobustCommand\word[1]{\mbox{$\{$\itshape#1\/$\}$}}
\makeatletter
\newcommand{\swtpc}{%
\sbox\z@{ }%
\expandafter\makebox \expandafter[\the\wd\z@]%
{\textperiodcentered}%
}
\makeatother
\providecommand*\describeoption[1]{\marginpar{\texttt{#1} option}}
\providecommand*\PrintChar[1]{\char#1\relax}
\providecommand*{\DoNotIndexHarmless}[1]{}
\providecommand*{\DoNotIndexBy}[1]{}
\DoNotIndexBy{@}
\DoNotIndexBy{@@}
\DoNotIndexBy{TD@}
\newenvironment{isyntax}{%
\list{}{%
\setlength\parsep{0pt plus 1pt}%
\setlength\listparindent{-1em}%
\setlength\itemindent{\listparindent}%
\addtolength\leftmargin{1em}%
}%
\item\relax \rightskip=1.0\rightskip plus 5cm
}{\endlist}
\setcounter{IndexColumns}{2}
\hfuzz=10pt
\expandafter\def \expandafter\GlossaryParms \expandafter{%
\GlossaryParms \hbadness=10000
}
\CodelineIndex
\EnableCrossrefs
\RecordChanges
\begin{document}
\DocInput{tclldoc.dtx}
\PrintChanges
\PrintIndex
\end{document}
%</driver>
%
% \fi
%
% \begin{abstract}
% The \package{tclldoc} package defines a couple of environments and
% macros for documenting \Tcl\ source code in \texttt{.dtx}-style
% documented source files, much like what the \package{doc}
% package~\cite{doc} does for \TeX\ source code. The \package{tclldoc}
% class is analogous to the \package{ltxdoc} document
% class~\cite{ltxdoc}---it loads the package to gain the basic
% functionality and changes some layout parameters to values that
% are better suited for documented source than those set by the
% standard \package{article} document class.
%
% The \package{tclldoc} package builds on the \package{doc},
% \package{xdoc}~\cite{xdoc}, and \package{docindex}~\cite{docindex}
% packages.
%
% Note: The \package{tclldoc} package and class used to be called
% \package{tcldoc}, but it turned out that there already existed a
% Perl~(!!)\ script \Tcl Doc that has a similar function. Therefore I
% added an extra `\textsf{l}'---which stands for both \LaTeX\ and
% Literate---to avoid confusing the community of \Tcl\ programmers.
% For compatibility, a package and a class named \package{tcldoc} are
% installed with \package{tclldoc}, but they shouldn't be used for
% new documents.
% \end{abstract}
%
% \tableofcontents
%
% \DoNotIndex{\advance,\begingroup,\bgroup,\big,\Big,\bigg,\Bigg,\box}
% \DoNotIndex{\catcode,\char,\chardef,\cleaders,\color@begingroup}
% \DoNotIndex{\color@endgroup,\copy,\csname,\DeclareOption}
% \DoNotIndex{\DeclareRobustCommand,\DeclareTextCommand}
% \DoNotIndex{\DeclareTextSymbol,\def,\discretionary,\divide,\do}
% \DoNotIndex{\edef,\egroup,\else,\@empty,\endcsname,\endgroup}
% \DoNotIndex{\expandafter,\fi,\@firstoftwo,\font,\fontdimen}
% \DoNotIndex{\futurelet,\gdef,\global,\@gobble,\hb@xt@,\hbox}
% \DoNotIndex{\hfill,\if,\ifcase,\ifcat,\ifmmode,\ifnum,\ifvbox}
% \DoNotIndex{\ifvmode,\ifvoid,\ifx,\ignorespaces,\immediate,\indent}
% \DoNotIndex{\input,\@@input,\InputIfFileExists,\IfFileExists}
% \DoNotIndex{\item,\kern,\language,\lastbox,\lastpenalty,\lastskip}
% \DoNotIndex{\lccode,\leavevmode,\let,\llap,\LoadClass,\lowercase}
% \DoNotIndex{\m@ne,\m@th,\@makeother,\mathord,\mkern,\multiply}
% \DoNotIndex{\@ne,\newbox,\newcommand,\newcount,\NewDescribeCommand}
% \DoNotIndex{\newenvironment,\newif,\NewMacroEnvironment,\nfss@text}
% \DoNotIndex{\noexpand,\normalfont,\null,\number,\or,\p@,\@@par}
% \DoNotIndex{\parshape,\PassOptionsToClass,\PassOptionsToPackage}
% \DoNotIndex{\penalty,\@plus,\predisplaypenalty,\prevdepth,\prevgraf}
% \DoNotIndex{\ProcessOptions,\protect,\protected@edef}
% \DoNotIndex{\providecommand,\ProvideTextCommandDefault,\relax}
% \DoNotIndex{\RequirePackage,\RequirePackageWithOptions,\rightarrow}
% \DoNotIndex{\sbox,\selectfont,\setbox,\skewchar,\space,\string}
% \DoNotIndex{\TextSymbolUnavailable,\texttt,\the,\tracingparagraphs}
% \DoNotIndex{\ttfamily,\tw@,\typeout,\unhbox,\unpenalty,\unskip}
% \DoNotIndex{\unrestored@protected@xdef,\unskip,\unvbox,\vadjust}
% \DoNotIndex{\vbox,\vert,\vrule,\vskip,\wd,\write,\xdef,\z@,\z@skip}
%
% \section{Introduction}
% \label{Sec:Introduction}
%
% This introduction is meant to be comprehensible even for readers
% unfamiliar with writing \texttt{.dtx} files and using the \package{doc}
% package. Readers who are experienced with this will probably want to
% skip right to the next section.
%
% A \texttt{.dtx} file has a dual nature. On one hand it is a container
% for some lines of code---it could be a program, a macro package, a
% configuration file for some program, merely a part of any of the
% aforementioned, or even arbitrary combinations of the above---and on
% the other hand it is a \LaTeX\ document which documents this code. One
% important advantage with this arrangement is that one can keep all the
% pieces of a project that has to do with a specific task at one place;
% experience has shown that this greatly furthers really keeping all
% parts of a project up to date with each other.
%
% Slightly simplified, one can say that a \texttt{.dtx} file contains
% three kinds of lines. A \emph{code line} is a line that doesn't begin
% with a `|%|' character; such lines can be extracted (copied) using the
% \package{docstrip} program~\cite{docstrip}. A \emph{guard line} is a
% line that begins with the two characters `|%<|'; guards are used to
% structure the set of code lines so that \package{docstrip} can extract
% different code lines to different generated files. A \emph{comment
% line}, finally, is a line that begins with a `|%|' character that is
% not immediately followed by a `|<|' character. The comment lines are
% ignored by \package{docstrip}, but are part of (and usually make up
% most of) the \LaTeX\ document in the \texttt{.dtx} file.
%
%
% \subsection{Special conventions and basic features in a \texttt{.dtx}
% \LaTeX\ document}
%
% An important difference between normal \LaTeX\ documents and
% \texttt{.dtx} \LaTeX\ documents is that the percent character `|%|'
% doesn't start a comment in the latter; in fact it is usually ignored.
% This allows \LaTeX\ to see and typeset the text in the comment lines
% of a \texttt{.dtx} file. Hence if one wants to include the sentence
% ``Your hovercraft is full of eels!'', which in a normal \LaTeX\
% document could have been written as the line\iffalse
%<*example>
% \fi
%\begin{verbatim}
%Your hovercraft is full of eels!
%\end{verbatim}
% one would instead write the line as
%\begin{verbatim}
%% Your hovercraft is full of eels!
%\end{verbatim}
% in a \texttt{.dtx} document. The space after the |%| is not
% necessary, but most \texttt{.dtx} documents you see include
% it---probably because the ``comment out \TeX\ code'' action of most
% text editors consists of inserting a percent \emph{and} a space at
% the beginning of each line.
%
% The code lines present the opposite problem, as they usually
% shouldn't be treated as normal \LaTeX\ code although the normal \LaTeX\
% reading conventions would make them the entire document. The usual way
% around this is to surround each group of code lines with two comment
% lines that begin and end an environment in which the code lines get
% typeset verbatim. The \package{tclldoc} package provides the \texttt{tcl}
% environment for this purpose, so the code lines
%\begin{verbatim}
%proc factorial {n} {
% set result 1
% for {set i 1} {$i<=$n} {incr i} {
% set result [expr {$result * $i}]
% }
% return $result
%}
%\end{verbatim}
% could be included in a \texttt{.dtx} document as the lines
%\begin{verbatim}
%% \begin{tcl}
%proc factorial {n} {
% set result 1
% for {set i 1} {$i<=$n} {incr i} {
% set result [expr {$result * $i}]
% }
% return $result
%}
%% \end{tcl}
%\end{verbatim}
% When typeset, this will look as
% \begin{macrocode}
proc factorial {n} {
set result 1
for {set i 1} {$i<=$n} {incr i} {
set result [expr {$result * $i}]
}
return $result
}
% \end{macrocode}
% The tiny numbers at the beginning of each line enumerate the code lines.
% Index references to code usually specify such code line numbers, but
% the enumeration can be switched off.
%
% In mathematical papers, the statements of e.g.\ theorems are usually
% made inside a \texttt{theorem} (or whatever) environment which
% provides certain text formatting, a heading, and a position in the
% document that can be referenced from other parts of it. In \texttt{.dtx}
% documents one usually does something similar for each named piece of
% code: macros, environments, templates, etc. In particular, the
% \package{tclldoc} package provides two environments \texttt{proc} (for
% procedures) and \texttt{variable} (for variables).
% Figure~\ref{Fig:procex} contains an example of how \texttt{proc}
% might be used in describing a procedure for computing the greatest
% common divisor of two integers.
%
% \begin{figure}
%
% \begin{trivlist}
% \item[]\leavevmode\llap{\texttt{gcd} (proc)\kern\marginparsep}^^A
% The |gcd| procedure takes two arguments $a$ and $b$ which must be
% integers and returns their greatest common divisor $\gcd(a,b)$,
% which is computed using Euclid's algorithm. As a special case,
% $\gcd(0,0)$ is considered to be $0$, so formally |gcd $a $b|
% computes $\lvert a \rvert \wedge \lvert b \rvert$, where $\wedge$
% denotes the meet operation in the divisor lattice of non-negative
% integers.
% \begin{macrocode}
proc gcd {a b} {
set a [expr {abs($a)}]
set b [expr {abs($b)}]
while {$b>0} {
set r [expr {$a%$b}]
set a $b
set b $r
}
return $a
}
% \end{macrocode}
% \end{trivlist}
%
% \begin{center}
% (a) A typeset procedure with description
% \end{center}
%
%\begin{verbatim}
%% \begin{proc}{gcd}
%% The |gcd| procedure takes two arguments $a$ and $b$ which must be
%% integers and returns their greatest common divisor $\gcd(a,b)$,
%% which is computed using Euclid's algorithm. As a special case,
%% $\gcd(0,0)$ is considered to be $0$, so formally |gcd $a $b|
%% computes $\lvert a \rvert \wedge \lvert b \rvert$, where $\wedge$
%% denotes the meet operation in the divisor lattice of non-negative
%% integers.
%% \begin{tcl}
%proc gcd {a b} {
% set a [expr {abs($a)}]
% set b [expr {abs($b)}]
% while {$b>0} {
% set r [expr {$a%$b}]
% set a $b
% set b $r
% }
% return $a
%}
%% \end{tcl}
%% \end{proc}
%\end{verbatim}
%
% \begin{center}
% (b) The code for the example in (a)
% \end{center}
%
% \caption{An example of the \texttt{proc} environment}
% \label{Fig:procex}
% \end{figure}
%
% What does the \texttt{proc} environment do more precisely? First
% there's the marginal heading which can be seen in Figure~^^A
% \ref{Fig:procex}. Such headings make it easier to find the procedure in
% the typeset form of the document. Then the \texttt{proc} environment
% makes an index entry which tells you where the procedure is defined,
% and finally it stores the procedure name in a variable so that
% subsequent |\changes|\footnote{The \cs{changes} command is defined by
% the \package{doc} package~\cite{doc}. It is used for adding entries
% to a global list of changes for code in the \texttt{.dtx} document.}
% commands know to what the change that they are recording was made.
%
% The \texttt{variable} environment does the same things except that it
% writes ``(var.)'' rather than ``(proc)''. This environment wasn't used
% for describing the three local variables |a|, |b|, and |r| in the
% example; this is since there is no point in referring to these
% variables from elsewhere in the program. Instead the \texttt{variable}
% environment is primarily meant for global variables (although it could
% also be useful for local variables that are meant to be accessed using
% |upvar| or |uplevel|), and as such it can often be of great help, since
% the description of a global variable can otherwise be hard to find,
% especially with languages like \Tcl\ where variables don't have to be
% declared and thus have no natural ``home'' in the code.
%
% Another noteworthy feature in the example is the use of vertical
% bar `\verb+|+' characters to delimit short pieces of verbatim \Tcl\
% code in the comment lines. It is often necessary for the explanation
% to include short examples of code in the documentation, and the
% standard \LaTeX\ |\verb| command is exactly what one would need for
% this. As such code sections are rather frequent however, it has become
% the custom to use a single character for both starting and ending such
% a piece of code. The \package{tclldoc} document class defines \verb+|+
% as a shorthand for \verb+\tclverb|+, where |\tclverb| is a variant of
% |\verb| which has been designed specifically for \Tcl\ code.
%
% The above description was meant to give a basic understanding of how
% \Tcl\ code and documentation thereof can be mixed in a \texttt{.dtx}
% file, it neither explains all the environments and commands that
% the \package{tclldoc} package provides, nor mentions all the
% features of the environments that were described. That information can
% instead be found in Section~\ref{Sec:Manual} of this paper. It should
% also be mentioned that the \package{doc} package~\cite{doc} defines
% several commands and environments that may be of use for describing
% code, and it is well worth getting acquainted with the features of
% that package as well (although parts of its documentation has become
% rather archaic).
%
%
% \subsection{Guards and \package{docstrip} installation scripts}
%
% The central command in a \package{docstrip} installation script is
% |\generate|, since this is the command which actually causes code to
% be extracted. |\generate|'s syntax is
% \begin{quote}
% |\generate|\marg{files}
% \end{quote}
% where \meta{files} consists of one or several |\file| commands, each
% of which has the syntax
% \begin{quote}
% |\file|\marg{output}\marg{sources}
% \end{quote}
% where \meta{output} is a filename and \meta{sources} consists of one
% or several |\from| commands, each of which has the syntax
% \begin{quote}
% |\from|\marg{input}\marg{options}
% \end{quote}
% where, finally, \meta{input} is a filename and \meta{options} is a
% comma-separated list of alphanumerical strings. Thus a |\generate|
% command might look like
%\begin{verbatim}
% \generate{\file{p1.sty}{\from{s1.dtx}{foo,bar}}
% \file{p2.sty}{\from{s2.dtx}{baz}
% \from{s3.dtx}{baz}}
% \file{p3.sty}{\from{s1.dtx}{zip}
% \from{s2.dtx}{zap}}
% }
%\end{verbatim}
% The meaning of this command is
% \begin{quote}
% Generate the three files \texttt{p1.sty}, \texttt{p2.sty}, and
% \texttt{p3.sty}. Extract the code for \texttt{p1.sty} from
% \texttt{s1.dtx} with options \texttt{foo} and \texttt{bar}, extract
% the code for \texttt{p2.sty} from \texttt{s2.dtx} with option
% \texttt{baz} and \texttt{s3.dtx} (the code from \texttt{s2.dtx}
% will be put before the code from \texttt{s3.dtx}) with option
% \texttt{baz}, and finally extract the code for \texttt{p3.sty} from
% \texttt{s1.dtx} with option \texttt{zip} and \texttt{s2.dtx} with
% option \texttt{zap}.
% \end{quote}
% The \emph{options} are used to control which parts of the source files
% should be extracted to which generated file. A source file can
% contain a number of \emph{modules}, and at the beginning of each
% module \package{docstrip} decides, for each output file separately,
% whether the code lines in that module should be extracted to the
% output file. The beginning of a module is marked by a guard line which
% has the syntax
% \begin{quote}
% |%<*|\meta{expression}|>|
% \end{quote}
% and the end by a corresponding
% \begin{quote}
% |%</|\meta{expression}|>|
% \end{quote}
% guard line. In their simplest form, the \meta{expression}s are names
% of options, and in that case the code lines in the module are only
% extracted if that option appears in the \meta{options} for that
% combination of input file and output file. The \meta{expression}s can
% however be arbitrarily complicated boolean expressions;
% see~\cite{docstrip} for more information.
% Modules may nest inside each other, and in that case the code lines in
% an inner module can only be included if all surrounding modules are
% being included. It is checked that matching |*| and |/| guard lines
% contain the same (as strings) \meta{expression}, and case is
% significant in the names of options.
%
% One application of modules which has already been mentioned is to
% bundle code for several different generated files in the same
% \texttt{.dtx} file---one example of this is the file
% \texttt{doc.dtx} (part of the \LaTeX\ base distribution) which
% contains both the \package{doc} package (\texttt{doc.sty}), the
% \package{shortvrb} package (\texttt{shortvrb.sty}), and two
% \package{makeindex} style files (\texttt{gglo.ist} and
% \texttt{gind.ist}). Another application is to keep variant sections
% of code---such as special code for debugging or gathering
% statistics---in the \texttt{.dtx} source file for a program without
% thereby making it a part of the normal form of that program. It is
% quite possible to use \package{docstrip} as a simple pre-processor
% for languages whose compiler\slash interpreter has not got one built
% in.
%
% There are many other commands available in a \package{docstrip}
% installation script beside those listed above, but those are well
% described in the \package{docstrip} manual~\cite{docstrip} and need
% little attention here. Instead I'm going to finish this subsection
% with a quick guide to the particular difficulties one faces when using
% \package{docstrip} to extract \Tcl\ code, and how to overcome them.
%
% The main problem is that \package{docstrip} insert a few comment lines
% at the beginning and end of each file it generates. This is a good
% thing, because a file consisting entirely of extracted code lines
% would normally be completely void of commentary and quite
% unintelligible for the casual user. These few comment lines explain
% that the file was generated by \package{docstrip} from other files
% (which contain the documentation of the code), lists those files,
% and normally also contains a copyright (or more commonly some kind of
% copyleft) notice. The problem lies in that comments look different in
% different languages, and as the default is to write \TeX\ style
% comments, one must tell \package{docstrip} to write \Tcl\ style
% comments. This can be done through the command
%\begin{verbatim}
%\edef\MetaPrefix{\string#}
%\end{verbatim}
% which tells \package{docstrip} to begin each inserted comment line
% with the character `|#|'.
%
% The comment lines inserted at the beginning of a generated file are
% called the \emph{preamble} and those at the end the \emph{postamble}.
% To set the preamble, one writes
% \begin{trivlist}
% \item |\preamble|\\
% \meta{preamble lines}\\
% |\endpreamble|
% \end{trivlist}
% and correspondingly to set the postamble
% \begin{trivlist}
% \item |\postamble|\\
% \meta{postamble lines}\\
% |\endpostamble|
% \end{trivlist}
% The \meta{preamble lines} and \meta{postamble lines} can be any number
% of lines (including zero). Unlike the text in source files, the text in
% these preamble and postamble lines is not read verbatim, so things in
% these lines which have special meaning to \TeX\ (such as control
% sequences) will be treated as such; the only exception is that spaces
% and newlines are preserved (instead of concatenated to single spaces
% as they normally would). It is important that the preamble and
% postamble \emph{are} set after |\MetaPrefix| is changed, because each
% line specified between |\preamble| and |\endpreamble| or |\postamble|
% and |\endpostamble| respectively will be prefixed by the current value
% of |\MetaPrefix|.
%
% Finally, some programs (such as the UNIX core) assign special meaning
% to the first line of a file, so one might want to control what gets
% put there. Merely using |\preamble| doesn't achieve this, because the
% \meta{preamble lines} specified that way are put after the lines
% saying ``this is a generated file \textellipsis''. You can however
% add things to the preamble by explicitly setting the macro
% |\defaultpreamble|, which is where \package{docstrip} stores the
% preamble. To make the first line a comment which simply contains the
% text `|-*-Tcl-*-|', you could give the command
%\begin{verbatim}
%\edef\defaultpreamble{\MetaPrefix\space -*-Tcl-*-^^J\defaultpreamble}
%\end{verbatim}
% Similarly to begin the file by the three standard lines
%\begin{verbatim}
%#! /bin/sh
%#\
%exec tclsh "$0" ${1+"$@"}
%\end{verbatim}
% (for an explanation see~\cite{execMagic})---which on UNIX allow the file
% to function both as a \Tcl\ script and a shell script which terminates
% the shell and runs \texttt{tclsh} on the script instead---you can use
% the command
%\begin{verbatim}
%\edef\defaultpreamble{%
% \MetaPrefix! /bin/sh^^J%
% \MetaPrefix\string\^^J%
% exec tclsh "$0" ${1+"$@"}^^J%
% \defaultpreamble
%}
%\end{verbatim}
% The full explanation of these commands is however far beyond this
% introduction.\footnote{Those who want to fully understand them should
% read \emph{The \TeX book}~\cite{TeXbook}, in particular Chapter~8.}
%
% In summary, a \package{docstrip} installation script for extracting a
% file \texttt{foo.tcl} from \texttt{foo.dtx}, using \Tcl\ style
% comments, inserting a BSD-style license notice in the preamble, and
% beginning with the line |# -*-Tcl-*-| could look as follows:
%\begin{verbatim}
%\input docstrip.tex
%
%\edef\MetaPrefix{\string#}
%
%\preamble
%
%Copyright (c) <YEAR>, <OWNER>
%All rights reserved.
%
%Redistribution and use in source and binary forms, with or without
%modification, are permitted provided that the following conditions
%are met:
%* Redistributions of source code must retain the above copyright
%notice, this list of conditions and the following disclaimer.
%* Redistributions in binary form must reproduce the above
%copyright notice, this list of conditions and the following
%disclaimer in the documentation and/or other materials provided
%with the distribution.
%
%THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
%"AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
%LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS
%FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE
%COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT,
%INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING,
%BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
%LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
%CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
%LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN
%ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
%POSSIBILITY OF SUCH DAMAGE.
%
%\endpreamble
%
%\postamble
%\endpostamble
%
%\edef\defaultpreamble{\MetaPrefix\space -*-Tcl-*-^^J\defaultpreamble}
%
%\generate{\file{foo.tcl}{\from{foo.dtx}{bar}}}
%
%\end
%\end{verbatim}
%
% The generated file \texttt{foo.tcl} will contain
%\begin{verbatim}
%# -*-Tcl-*-
%#
%# This is file `foo.tcl',
%# generated with the docstrip utility.
%#
%# The original source files were:
%#
%# foo.dtx (with options: `bar')
%#
%# Copyright (c) <YEAR>, <OWNER>
%# All rights reserved.
%#
%# Redistribution and use in source and binary forms, with or without
%# modification, are permitted provided that the following conditions
%# are met:
%# * Redistributions of source code must retain the above copyright
%# notice, this list of conditions and the following disclaimer.
%# * Redistributions in binary form must reproduce the above
%# copyright notice, this list of conditions and the following
%# disclaimer in the documentation and/or other materials provided
%# with the distribution.
%#
%# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
%# "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
%# LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS
%# FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE
%# COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT,
%# INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING,
%# BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
%# LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
%# CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
%# LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN
%# ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
%# POSSIBILITY OF SUCH DAMAGE.
%#
%\end{verbatim}
% \meta{lines extracted from \texttt{foo.dtx}}
%\begin{verbatim}
%#
%#
%# End of file `foo.tcl'.
%\end{verbatim}
%
%
%
% \subsection{The structure of the \LaTeX\ document}
%
% All this has been about the local appearance of a \texttt{.dtx} file,
% but what about the overall structure? There are several points to
% raise about that as well.
%
% The first is that \LaTeX\ to begin with treat \texttt{.dtx} documents
% just like any other document---a `|%|' starts a comment and only lines
% \emph{not} beginning with a `|%|' contain anything that \LaTeX\ can
% see. Somehow \LaTeX\ must be instructed to start applying the special
% reading rules that were described above. This is the job of the
% so-called \emph{driver}, which (for a file \texttt{myfile.dtx}) in its
% simplest form can look like
%\begin{verbatim}
%\documentclass{tclldoc}
%\begin{document}
%\DocInput{myfile.dtx}
%\end{document}
%\end{verbatim}
% The important command here is |\DocInput|, because that is what tells
% \LaTeX\ to apply the special \texttt{.dtx} reading rules. More
% precisely it means ``Start ignoring `|%|' characters in the text you
% read, input the file \texttt{myfile.dtx}, and when you're done return
% to treating `|%|' characters as before.''
%
% The driver is usually put in the very first stretch of code lines in
% the \texttt{.dtx} file. This means that \LaTeX, when ordered to typeset
% the \texttt{.dtx} file, will start to read along, possibly ignoring
% hundreds of lines beginning with `|%|' because they are comments. Then
% it encounters the driver, and after the |\documentclass| and
% |\begin{document}| commands it executes the |\DocInput|. This will
% cause it to not ignore lines beginning with `|%|', so when it starts
% reading the file again it will see all the lines it skipped the first
% time through. The file will be read to end, after which \LaTeX\
% returns to the command after the |\DocInput|. As that command happens
% to be |\end{document}|, it finishes the typeset document and stops.
% This stop prevents it from seeing and interpreting as \LaTeX\ commands
% the remaining code lines in the file.
%
% The second time through the driver shouldn't be interpreted as
% \LaTeX\ commands, since for example the |\documentclass| command may
% only be used once in a \LaTeX\ document. One way of achieving this is
% to put an |\iffalse| command right before the driver and a |\fi|
% command right after it. This says to \LaTeX\ that the driver code is
% conditional material, and since the condition evaluates to false
% (|\iffalse| always evaluates to false), this conditional material
% should be skipped. Thus the first few lines of \texttt{myfile.dtx}
% typically might be
%\begin{verbatim}
%% \iffalse
%%<*driver>
%\documentclass{tclldoc}
%\begin{document}
%\DocInput{myfile.dtx}
%\end{document}
%%</driver>
%% \fi
%\end{verbatim}
% The \texttt{driver} guard lines are there to stop \package{docstrip}
% from including the driver code in the generated files.
%
% After the driver comes the actual \LaTeX\ document. The document
% usually consists of two parts, where the first part is a manual for
% the \emph{usage} of the code defined in the file, and the second part
% contains the actual \emph{implementation} (documented code). The idea
% is that most people are (at least the first time) quite content with
% learning how to use something, so one should make it simple for them
% to find that information.\footnote{One needn't take this as an absolute
% rule---I for one haven't written all my packages that way---but
% structuring the document like this generally makes it more accessible.}
% To further this approach one puts the command |\StopEventually| at the
% start of the implementation part and the command |\Finale| at the end
% of it. Normally |\StopEventually| doesn't make itself felt, but if one
% previously has given the command |\OnlyDescription| then rest of the
% file will not be read; this can be used to produce a ``manual only''
% version of the documentation. |\StopEventually| takes one argument and
% the code in this argument is executed at the |\Finale| (if the
% implementation part is being included) or immediately (if the
% implementation part isn't being included). Thus this argument is the
% place to put things that should appear at the very end of the
% document.
%
% The \texttt{tcl}, \texttt{proc}, and \texttt{variable} environments
% described above all typically appear in the implementation part of the
% document.
%
%
% \section{Usage of commands and environments}
% \label{Sec:Manual}
%
%
% \subsection{The actual source code}
%
% The \DescribeEnv{tcl}\texttt{tcl} environment is used for wrapping up
% a group of code lines which contain \Tcl\ code. Lines inside this
% environment which begin with a percent character are called
% \emph{command} lines and can contain \LaTeX\ commands which get
% executed, whereas lines that do not begin with a percent character are
% called \emph{normal} lines and get typeset verbatim (or nearly
% verbatim). Lines that begin with |%<| (\package{docstrip} guard lines)
% do however constitute a special case, as the guard expression will get
% typeset as in \package{doc}'s \texttt{macrocode} environment and the
% remainder of the line will get processed in command mode if it is a
% |*| or |/| guard, but in normal mode if the guard line was of any other
% type.
%
% The \texttt{tcl} environment uses the same general formatting
% parameters as \package{doc}'s \texttt{macrocode} environment. In
% particular this means that the text on a normal line is typeset in
% |\MacroFont| (by default the same thing as \cs{normalfont}\penalty0
% \cs{ttfamily}\penalty0\cs{small}) or |\AltMacroFont| (by default the same
% thing as \cs{normalfont}\penalty0\cs{ttfamily}\penalty0\cs{itshape}^^A
% \penalty0\cs{small}) depending on the current \package{docstrip} module
% nesting level. The \package{tclldoc} class sets the
% \texttt{StandardModuleDepth} counter to 1, which means that the
% |\AltMacroFont| is used when the modules are nested two levels deep or
% more.
%
% If a normal line is too long to fit on one line then the \texttt{tcl}
% environment will try to break it. Legal breakpoints are spaces which
% could be replaced by a backslash and a newline without changing the
% meaning of the command; thus most spaces are legal breakpoints. When a
% line is broken at a space like this, the space is replaced by a
% backslash so that the line is still ``syntactically correct''. The
% opposite happens to lines which actually end with an escaped newline;
% such lines are concatenated with the following line and are treated as
% one long line. This is so that a sequence of breakpoints can be chosen
% which is optimal for the actual line width of the document (as opposed
% to the line width used in the text file, which can be something quite
% different). For example
%\begin{verbatim}
%% \begin{tcl}
%lsearch -exact\
% [concat $a [lrange $b $first end] c d e f]\
% [lindex $g $h]
%% \end{tcl}
%\end{verbatim}
% could be typeset as any of the code examples in
% Figure~\ref{Fig:Radbrytning}, depending on how wide a line is.
% \begin{figure}
% \begin{center}
% \small
% \(\left\lvert
% \begin{minipage}{350pt}
% \begin{flushleft}
% |lsearch -exact [concat $a [lrange $b $first end] c d e f]|^^A
% | [lindex $g $h]|
% \end{flushleft}
% \end{minipage}
% \right\rvert\)
%
% \smallskip
%
% \(\left\lvert
% \begin{minipage}{298pt}
% \begin{flushright}
% \noindent |lsearch -exact [concat $a [lrange $b $first|^^A
% | end] c d e f]|\textbackslash\hfill\vadjust{}\\
% |[lindex $g $h]|
% \end{flushright}
% \end{minipage}
% \right\rvert\)
%
% \smallskip
%
% \(\left\lvert
% \begin{minipage}{241pt}
% \begin{flushright}
% \noindent |lsearch -exact|\textbackslash\hfill\vadjust{}\\
% |[concat $a [lrange $b $first end] c d e f]|\textbackslash\\
% |[lindex $g $h]|
% \end{flushright}
% \end{minipage}
% \right\rvert\)
%
% \smallskip
%
% \(\left\lvert
% \begin{minipage}{184pt}
% \begin{flushright}
% \noindent |lsearch -exact [concat $a|\textbackslash\hfill
% \vadjust{}\\
% |[lrange $b $first end] c d e f]|\textbackslash\\
% |[lindex $g $h]|
% \end{flushright}
% \end{minipage}
% \right\rvert\)
% \end{center}
%
% \caption{The same \Tcl\ code, set in different linewidths}
% \label{Fig:Radbrytning}
% \end{figure}
% The environment tries to put the linebreaks at the lowest possible
% nesting (of braces and brackets) level; I believe this gives the best
% readability.\footnote{When I first started programming in \Tcl\ I
% used a completely different method for breaking long lines in the
% code---I mainly implemented the current method because it was
% simple to program---but I was quite surprised by how readable it made
% the code.} There is however a way to override this automatic choice
% of breakpoints: if a normal line which ends with an escaped newline is
% followed by a command line (even a command line that doesn't contain
% any commands) then it will not be concatenated with the next normal
% line.
%
% There are a couple of restrictions on the code in command lines. First
% of all it is not allowed to start a new paragraph (there will be an
% error message). Secondly a command may not be broken across several
% lines---all the arguments must appear on the same line as the
% control sequence. Thirdly some characters have other catcodes than
% in normal \LaTeX, so it is not certain that all commands work. Some
% commands that do work and may be useful are:
% \begin{itemize}
% \item Vertical space commands (|\smallskip|, |\medskip|, etc.)
% The command line `|% \medskip|' is more to type than a blank
% normal line, but it looks slightly better.
% \item Indexing commands (|\index|, |\IndexEntry|,\footnote{This
% command is defined by the \package{xdoc} package~\cite{xdoc}.}
% etc.)
% \item The \cs{TclInput} and |\settabsize| commands (see below).
% \end{itemize}
% And of course the |\end{tcl}| command works in a command line, since
% that is how one ends a \texttt{tcl} environment.
%
% Besides the \texttt{tcl} environment there is also a
% \DescribeEnv{tcl*}\texttt{tcl*} environment which is different from
% \texttt{tcl} only in that spaces and tabs are typeset as special
% visible space `\textvisiblespace' and visible tab
% `\makebox[6\fontdimen2\font]{\( - \mkern-7mu
% \cleaders\hbox{$\mkern-2mu - \mkern-2mu$}\hfill \mkern-7mu
% \mathord\rightarrow \mkern-1mu \vrule \mkern1mu\)}' characters. This
% can be useful for pieces of code where the exact number of spaces in
% a sequence is significant, such as code for writing tables that align.
%
% For shorter pieces of \Tcl\ code, e.g.\ examples, there's the
% \DescribeMacro{\tclverb}|\tclverb| command. |\tclverb| is very similar
% to the standard \LaTeX\ command |\verb|, but there are two differences.
% The first is that text typeset by |\tclverb| can contain breakpoints
% at whitespace; these behave just as in the \texttt{tcl} environment.
% The second is that the verbatim text that follows |\tclverb| may contain
% newlines, provided that these newlines are escaped by a backslash. Like
% a \Tcl\ interpreter, |\tclverb| ignores whitespace following an escaped
% newline. Unlike a \Tcl\ interpreter, |\tclverb| also ignores one percent
% character before the ignored whitespace, if it is the first character
% on the following line. Thus
%\begin{verbatim}
%% \tclverb|append a $b| is much more efficient than \tclverb|set a\
%% $a$b| if \tclverb|$a| is long.
%\end{verbatim}
% is perfectly legal, and the escaped newline between |a| and |$a$b| is
% treated just like the space between |set| and |a|. Like |\verb|, the
% |\tclverb| command has a starred form \DescribeMacro{\tclverb*}^^A
% |\tclverb*| which also typesets spaces and tabs as visible characters.
%
%
% The \DescribeMacro{\MakeShortTclverb}|\MakeShortTclverb| command works
% just like the \package{doc}\slash\package{shortvrb} command
% |\MakeShortVerb|, except that it makes the active character a shorthand
% reference for |\tclverb|\textellipsis\ instead of |\verb|\textellipsis.
% Use |\DeleteShortVerb| to undo the effect of a |\MakeShortTclverb|. The
% \package{tclldoc} class executes the command
% \begin{quote}
% \verb"\MakeShortTclverb{\|}"
% \end{quote}
% which makes \verb"|" a shorthand for \verb"\tclverb|".
%
% Since there is no universally accepted standard for the size
% (equivalent number of spaces) of a tab, there is a command
% \DescribeMacro{\settabsize}|\settabsize| for changing this.
% |\settabsize| takes as its only argument the new tab size, which must
% be an integer in the range 2--255. The default value is 8.
% |\settabsize| makes a local assignment to the tab size. The tab size
% can be changed inside a \texttt{tcl} (or \texttt{tcl*}) environment.
%
% There is also a command \DescribeMacro{\TclInput}|\TclInput| which
% is used for typesetting ``raw'' (not in \texttt{.dtx} format) \Tcl\ code
% files. |\TclInput| is meant to be used on a \emph{command} line of a
% \texttt{tcl} or \texttt{tcl*} environment, and it efficiently makes
% things look as if the |\TclInput| command had been replaced by the
% inputted file in its entirety (preceded by a newline, and followed
% by a percent and a space). |\TclInput| takes as its only argument the
% name of the file to input.
%
% To typeset the file \texttt{myscript.tcl} one would write
%\begin{verbatim}
%% \begin{tcl}
%% \TclInput{myscript.tcl}
%% \end{tcl}
%\end{verbatim} \iffalse
%</example>
% \fi
% or even
%\begin{verbatim}
% \begin{tcl}\TclInput{myscript.tcl}\end{tcl}
%\end{verbatim}
% anywhere in a \package{tclldoc} document. This works since the
% \texttt{tcl} environment is in command mode right after the initial
% |\begin{tcl}|, and the |\end{tcl}| needs not be the first command on a
% command mode line.
%
%
% \subsection{Markup of named things}
%
% The two environments \DescribeEnv{proc}\texttt{proc} and
% \DescribeEnv{variable}\texttt{variable}, which are analogues of
% \package{doc}'s \texttt{macro} environment, for procedures and
% variables respectively have already been mentioned in
% Section~\ref{Sec:Introduction}. In addition to those there are two
% environments \DescribeEnv{arrayentry}\texttt{arrayentry} and
% \DescribeEnv{arrayvar}\texttt{arrayvar} which are meant for entries
% in array variables and array variables as a whole. The complete
% syntaxes of these environments are
% \begin{quote}
% |\begin{proc}|\oarg{namespace}\marg{proc name}\\
% \vadjust{}\quad$\vdots$\\
% |\end{proc}|
%
% |\begin{variable}|\oarg{namespace}\marg{variable name}\\
% \vadjust{}\quad$\vdots$\\
% \quad|\end{variable}|
%
% |\begin{arrayentry}|\oarg{namespace}\marg{array name}^^A
% \marg{entry name}\\
% \vadjust{}\quad$\vdots$\\
% |\end{arrayentry}|
%
% |\begin{arrayvar}|\oarg{namespace}\marg{array name}^^A
% \oarg{index-des}\\
% \vadjust{}\quad$\vdots$\\
% |\end{arrayvar}|
% \end{quote}
% The \meta{proc name}, \meta{variable name}, and \meta{array name}
% arguments are quite evidently the names of the procedure, variable,
% and array respectively. The \meta{namespace} argument can be to
% specify the namespace part of a qualified name; having the name split
% like this makes it easier to treat the namespace differently from the
% rest of the qualified name. The command \DescribeMacro{\buildname}
% |\buildname| is used by the commands and environments described here to
% construct a qualified name from a namespace and a name.
% If there is no \meta{namespace} argument then the namespace used will
% be the default namespace. The default namespace is set using the
% \DescribeMacro{\setnamespace}|\setnamespace| command, which takes the
% namespace name as its only argument. The default namespace at the
% beginning of the document is the global namespace, whose name is the
% empty string.
%
% The \texttt{arrayentry} environment is intended for certain
% distinguished entries in an array, such as entries inserted to
% make the boundary cases of an algorithm work correctly and entries
% which have a special meaning to the program. Not all arrays contain
% such special entries, but when they do it is a good practice
% to explain them explicitly. The \meta{index-des} argument of the
% \texttt{arrayvar} environment can be used to specify what is used as
% index into the array; the text in this argument will appear both in
% the margin and in the index, but note that \meta{index-des} is a
% moving argument. There is little difference between the
% \texttt{variable} and \texttt{arrayvar} environments when the
% \meta{index-des} argument of the latter isn't used, but the index
% entries they make behave differently with respect to
% \texttt{arrayentry} index entries. An \texttt{arrayentry} index entry
% will be put as a subentry of the corresponding \texttt{arrayvar}
% entry, whereas a \texttt{variable} entry would appear separately.
%
% The above environments usually only appear in the implementation part
% of a \texttt{.dtx} file. For the usage part there is a command
% \DescribeMacro{\describestring}|\describestring| which produces
% marginal headings and index entries. The syntax of |\describestring|
% is
% \begin{quote}
% |\describestring|\oarg{type}\oarg{namespace}\marg{text}
% \end{quote}
% The \meta{text} is the string for which a heading and index entry will
% be made, whereas the \meta{type} (if given) is put after the text.
% If the \meta{namespace} is given then the thing described is supposed
% to be the name of something namespace-relative (like a procedure or
% global variable) and in this case the complete name is formed by
% passing \meta{namespace} and \meta{text} to |\buildname|. If
% \meta{type} is \texttt{proc}, \texttt{var.}, or \texttt{array} and a
% namespace is given then the index entry made will fit that made by
% a corresponding \texttt{proc}, \texttt{variable}, or
% \texttt{arrayvar} respectively environment. The \meta{type} argument
% is, in \LaTeX\ terminology, moving.
%
% The \meta{text} and \meta{namespace} arguments can contain arbitrary
% characters and most characters can be entered verbatim. Amongst the
% exceptions are `|%|', `|\|', `|{|', and `|}|', which instead can be
% entered as
% |\PrintChar|\discretionary{}{}{}|{`\%}|,
% |\PrintChar|\discretionary{}{}{}|{`\\}|,
% |\PrintChar|\discretionary{}{}{}|{`\{}|, and
% |\PrintChar|\discretionary{}{}{}|{`\}}| respectively. See the
% \package{xdoc} package~\cite{xdoc} documentation for an explanation of
% the |\PrintChar| command. The \meta{text} argument can also contain
% ``variant'' parts made using the \DescribeMacro{\meta}|\meta| command.
% As an example,
% \begin{quote}
% |\describestring[array]{\meta{mode}modeVars}|
% \end{quote}
% puts the text
% \begin{quote}
% \meta{mode}|modeVars| (array)
% \end{quote}
% in the margin and index. The arguments of such |\meta| commands are
% moving.
%
% A case which deserves special treatment is that of options of
% commands and for that there is the \DescribeMacro{\describeopt}
% |\describeopt| command. The syntax of this command is
% \begin{quote}
% |\describeopt|*\oarg{namespace}\marg{command}\oarg{type}^^A
% \marg{option}
% \end{quote}
% where \meta{command} is the command whose option is being described
% and \meta{option} is the name of the option. \meta{namespace} is the
% namespace of the \meta{command} and defaults to the global namespace.
% \meta{type} is the type of the command and defaults to |proc| for
% procedure. The \meta{namespace}, \meta{command}, and \meta{type} are
% (currently) only used in the index entry that is generated. Options
% for procedures will be put as subentries of the main procedure entry.
% For built-in commands it might be more appropriate to use |command| as
% \meta{type}, e.g.
% \begin{quote}
% |\describeopt{lsort}[command]{-real}|
% \end{quote}
% will put ``\texttt{-real} option'' in the margin and in the index as
% a subentry of ``\texttt{lsort} (command), global namespace''. The |*|
% is optional---including it will supress the marginal note normally
% generated by this command.
%
% Since \Tcl\ is often used together with C, it might be useful to
% also have something similar to the \texttt{proc} environment for C
% things. This is what the \DescribeEnv{Cfunction}\texttt{Cfunction},
% \DescribeEnv{Cvariable}\texttt{Cvariable}, and \DescribeEnv{Ctype}
% \texttt{Ctype} environments are for. These take as their only
% argument the name of the identifier that is defined, e.g.
%\begin{verbatim}
% \begin{Cfunction}{main}
%\end{verbatim}
% Since there seems to be several schools on how C code should be
% formatted when typeset, the formatting of identifiers passed to these
% environments is configurable. The three commands
% \DescribeMacro{\Cfunctionidentifier}|\Cfunctionidentifier|,
% \DescribeMacro{\Cvariableidentifier}|\Cvariableidentifier|, and
% \DescribeMacro{\Ctypeidentifier}|\Ctypeidentifier| handle all
% typesetting of identifiers; each takes as its only argument the
% identifier (as a harmless string) to typeset. The default is to set
% the argument in italic; this is what CWEB does.
%
% If you are using the \texttt{C}\dots\ environments for identifiers
% whose names contain underscores (\_), you may want to pass the
% \describeoption{notrawchar}\texttt{notrawchar} to \package{tclldoc}
% (it is really an option of the \package{xdoc} package and will be
% passed on to that automatically). This option addresses a problem
% with \texttt{OT1}-encoded fonts that may cause underscores to display
% as a quite different character (the \texttt{cmtt} typewriter fonts are
% however not affected by this problem).
%
%
%
% \subsection{Describing command syntaxes}
%
% One important part of documentation is to describe the syntaxes of
% commands. The previous subsection contains examples of the conventions
% for this that has been developed for \LaTeX\
% commands---mandatory arguments are denoted as `\marg{argument}' and
% optional arguments are denoted as `\oarg{argument}'. These two classes
% suffice rather well for \LaTeX\ commands, but the syntaxes of \Tcl\
% commands are not seldom much more complex. Therefore a more powerful
% form of syntax specification is called for, and one which is close at
% hand is that used in regular expressions since it is already part of
% the \Tcl\ language anyway.
%
% \DescribeMacro\regopt
% \DescribeMacro\regstar
% \DescribeMacro\regplus
% The simplest commands available are the modifiers |\regopt|,
% |\regstar|, and |\regplus|, which correspond to the |?|,
% |*|, and |+| metacharacters in a regular expression; using |\regopt|
% after a term says that it is optional, |\regstar| says that the term
% can be repeated an arbitrary number of times (including zero), and
% |\regplus| says that the term occurs at least once. The typeset
% results of these commands are $^?$, $^*$, and $^+$ respectively
% (recall that exponents are sometimes used to denote repetition).
%
% The terminals in the expression are best made using |\tclverb| (for
% ``fixed'' material, e.g.\ procedure names) and
% \DescribeMacro{\word}|\word| (for variable material, e.g.\
% arguments). The syntax of |\word| is
% \begin{quote}
% |\word|\marg{text}
% \end{quote}
% and e.g.\ |\word{script}| gets typeset as
% \begin{quote}
% \word{script}
% \end{quote}
% Using these, one can for example specify the syntaxes of the \Tcl\
% commands |append| and |catch| through
% \begin{quote}
% \verb"|append| \word{var-name} \word{value}\regplus"\\
% \verb"|catch| \word{script} \word{var-name}\regopt"
% \end{quote}
% (recall that `\verb"|"' is a shorthand for `\verb"\tclverb|"'). These
% get typeset as
% \begin{quote}
% |append| \word{var-name} \word{value}\(^+\)\\
% |catch| \word{script} \word{var-name}\(^?\)
% \end{quote}
%
% Terms in regular expressions can also consist of parenthesised
% subexpressions, which are made using the \DescribeEnv{regblock}^^A
% \texttt{regblock} environment. The syntax of this environment is
% \begin{quote}
% |\begin{regblock}|\oarg{modifier}\quad\dots\quad |\end{regblock}|
% \end{quote}
% If \texttt{regblock} environments are nested then the parentheses of
% the outer environment will grow to be larger than those of the inner
% environment. A side-effect of this is that the \texttt{regblock}
% environment wants to know if a modifier will be applied to it, since
% the amount by which the modifier should be raised in this case depends
% on the size of the parenthesis before it, and this is what the
% \meta{modifier} optional argument is for. \LaTeX\ does not provide
% for arguments at the |\end| of an environment, so it has to be placed
% at the |\begin|. Using these elements, the syntax of |if| can be
% specified through
% \begin{quote}
% \verb"|if| \word{expression} |then|\regopt\ \word{script}"\\
% \verb"\begin{regblock}[\regstar]|elseif| \word{expression}"\\
% \verb"|then|\regopt\ \word{script}\end{regblock}"\\
% \verb"\begin{regblock}[\regopt]|else| \word{script}\end{regblock}"
% \end{quote}
% which typesets as
% \begin{isyntax}
% |if| \word{expression} |then|\(^?\) \word{script} \(\bigl(\)^^A
% |elseif| \word{expression} |then|\(^?\) \word{script}\(\bigr)^*\)
% \(\bigl(\)|else|~\word{script}\(\bigr)^?\)
% \end{isyntax}
% In versions of \package{tclldoc} before 2.40, the \texttt{regblock}
% environment used to be called \DescribeEnv{regexp}\texttt{regexp}.
% That other name is still supported, but should be avoided in new
% documents.
%
% The final regular expression construction that is supported is that
% of branches of a regular expression. A \texttt{regblock} environment
% consist of one or several branches that are separated by
% \DescribeMacro{\regalt}|\regalt| commands. Visually the |\regalt|
% command gets typeset as a vertical bar that has the same size as the
% parentheses of the surrounding \texttt{regblock} environment. The
% |\regalt| command may only be used inside a \texttt{regblock}
% environment.
% An example of the use of |\regalt| is the following specification of
% the syntax of \Tcl's |regexp| command:
% \begin{quote}
% \verb"|regexp| \begin{regblock}[\regstar]|-nocase|\regalt"\\
% \verb"|-indices|\end{regblock} |--|\regopt \word{regular expression}"\\
% \verb"\word{string} \word{var-name}\regstar"
% \end{quote}
% which typesets as
% \begin{isyntax}
% |regexp| \((\)|-nocase|~\(\mid\) |-indices|\()^*\) |--|\(^?\)
% \word{regular expression} \word{string} \word{var-name}\(^*\)
% \end{isyntax}
%
% Finally a note about the relationship between the |\word| command and
% \package{doc}'s |\meta| command. Whereas the argument of |\word| is
% encapsulated in braces (and thus ought to be a separate word for a
% \Tcl\ interpreter), the argument of |\meta| is encapsulated in angle
% brackets. The idea is that |\word| should be used for things which are
% separate words to \Tcl, whereas |\meta| should be used for things which
% corresponds to parts of words or to several words. Thus in the command
% |set b Z${a}Y|, the second word |b| would be a `\word{var-name}' and
% the third word |Z${a}Y| would be a `|Z|\meta{string}|Y|'. In the
% command |label .a -text "Hello there!"|, the last two arguments could
% be summarised as an \meta{option}, but not as an \word{option}.
%
%
% \subsection{Non-ASCII characters}
%
% One problem, which is only going to be more common in the future, is
% how to deal with non-ASCII characters in scripts. The main problem
% here lies not on the output side, as \LaTeX\ is actually pretty good
% at producing a requested character as long as it is available in some
% font, but on the input side. \LaTeX\ can handle input in most 8-bit
% encodings, but in order for that to work the file must contain an
% |\inputencoding| command which tells \LaTeX\ which encoding is being
% used. As transporting a file from one platform to another most likely
% changes the encoding, but not the argument of |\inputencoding|, this
% method is rather fragile. Certainly there is room for improvements
% but the world of 8-bit encodings is generally such a mess anyway that
% it probably isn't worth the effort.
%
% A more progressive approach is to decide that all source code is in
% Unicode (more precisely in UTF-8). The main arguments for this are:
% (i)~\Tcl\ uses Unicode internally, (ii)~it is equally foreign on
% all platforms and can be treated as binary data rather than ``extended
% ASCII'' text, and (iii)~since it isn't converted, there is no loss of
% data. Interestingly enough, \emph{it is possible to use UTF-8 ``out of
% the box'' today!} Using the \package{ucs} package~\cite{ucs-package}
% allows \LaTeX\ to interpret UTF-8 input and this works just as well
% for the \Tcl\ code in a \texttt{tcl} environment as for the normal
% \LaTeX\ text outside it. If \package{docstrip} is run on a \LaTeX\
% format\footnote{Or, in some implementations, the \TeX\ program gets a
% suitable option.} that preserves characters whose most significant
% bit is set\footnote{Rather than converting them to
% \texttt{\^{}\^{}}-sequences, which is the default.} then the non-ASCII
% characters are simply copied verbatim and it makes no difference that
% they may occupy more than one byte of data. Alternatively one can run
% \package{docstrip} on Omega and (with a little extra work) get the
% ability to have the \Tcl\ code translated to some other encoding as
% the files are being generated!\footnote{At least in theory; I have to
% admit I haven't actually tested the \package{docstrip} part of it.}
%
% But although the above paragraph describes the way to go in the long
% run, there are some matters which make this approach slightly
% unfeasible in the near future. This is of course my own subjective
% opinion, but I find that two good reasons not to start using Unicode
% throughout quite yet are that (i)~my favourite text editor doesn't
% support Unicode (yet) and (ii)~even if I do start using it, there
% wouldn't be that much people around who could make sense of such files
% if I were to send it to them. Therefore I \emph{intend} to implement,
% but as yet haven't, a kind of intermediate format where non-ASCII
% Unicode characters are encoded using only ASCII characters plus an
% extra escape character. The basic idea is simply that any string
% `\meta{escape}\meta{hex digits}\meta{escape}' should be interpreted as
% the Unicode character \texttt{U+}\meta{hex digits}, so that arbitrary
% Unicode characters can be encoded using a character set that only
% comprises ASCII plus one extra \meta{escape} character. Supposing that
% this \meta{escape} character is the centered dot `\textperiodcentered',
% I could then encode my name as
% \begin{quote}
% \texttt{Lars Hellstr\swtpc 00f6\swtpc m}
% \end{quote}
% whereas the centered dot itself would be \texttt{\swtpc 00b7\swtpc}.
% The idea is that the file itself should contain the declaration of
% which character is used as this Unicode escape, so that a change due
% to translation from one 8-bit encoding to another will identically
% alter both declaration and use of the escape character, thereby
% preserving the internal logic of the file.
%
% The weak point with this scheme is that \package{docstrip} would have
% to translate the escape sequences to proper characters when it
% generates files. Implementing that under \TeX\ is highly non-trivial.
% It can be done with a reasonable effort under Omega, but it still
% requires hacking \package{docstrip}. The really interesting approach
% would however be to implement it in a port of \package{docstrip} to
% \Tcl, as that would remove the need to have \TeX\ to install the files.
% Porting \package{docstrip} to \Tcl\ is by the way a project of mine
% which I unfortunately haven't spent much time on, but if it is to be
% of any use to have the Unicode escape format described above
% implemented in \package{tclldoc} then I will have to make some
% progress with it.
%
% \medskip
%
% One rather recent advancement in this direction is the code in
% \texttt{sourcedtx.tcl}, which can be generated from the file
% \texttt{sourcedtx.dtx} that is distributed with \package{tclldoc} as
% an example. This implements a \Tcl\ command |dtx::source| that makes
% it possible to source \Tcl\ code in a \texttt{.dtx} file without
% docstripping it to a file first. This code does currently not bother
% about encodings, but that is easy enough to add.
%
% \medskip
%
% Finally, a few notes on the old mechanism for non-ASCII characters
% that is included in the \package{tcldoc} compatibility class. It
% cooperates with the \package{rtkinenc} package~\cite{rtkinenc}, when
% that has been loaded, in order to detect when an input character isn't
% avaiable in any active font encoding. Rather than raising an error and
% printing nothing it these cases, missing characters are written as the
% corresponding |\x|\meta{hh} backslash sequence in a slightly different
% font than the rest of the text. A problem here is however that most
% input encodings contain a few characters which are interpreted as
% \emph{math} character by \LaTeX. When such a character appears on a
% code line it makes \TeX\ switch to math mode and things generally get
% quite messy afterwards.
%
% The cure for this is to redeclare these input characters to \LaTeX\
% so that they work as intended in text mode, but that does take some
% lines of code. The \package{tcldoc} class does contain the
% declarations needed for the \texttt{applemac} input encoding; passing
% it the \describeoption{macinputenc}\texttt{macinputenc} option will
% load the \package{rtkinenc} package, the \texttt{applemac} (macRoman)
% input encoding, the \texttt{TS1} output encoding, and make the
% necessary redeclarations to allow all input characters to work in
% text. As nothing is provided for any other input encoding however,
% this solution never was a good solution. The \texttt{macinputenc}
% should be considered as unsupported as of \package{tclldoc} v\,2.30.
%
%
%
% \subsection{Options and customisation}
%
% The \package{tclldoc} package does not have any options of its own, but
% all options passed to it are passed on to the \package{xdoc} package.
% The \package{tclldoc} class accepts all options that the standard
% \LaTeX\ document class \package{article} accepts, with the exception
% of \texttt{a5paper}.
%
% Like the \package{ltxdoc} class, the \package{tclldoc} class will look
% for a special configuration file \texttt{tclldoc.cfg} and input that
% file if it exists. This file can be used to declare extra options for
% the class, have certain options always given, etc. Section~2 of
% \texttt{ltxdoc.dtx}~\cite{ltxdoc} is a good introduction to how such
% configuration files can be used with \texttt{.dtx} sources in general.
%
% When you use a \texttt{tclldoc.cfg} file to customise the
% \package{tclldoc} document class, you affect how all documents using
% that class will be typeset in your particular \TeX\ installation. It
% is \emph{not} something you have to do, but it can
% make \package{tclldoc} documents work better with the printers, paper
% formats, fonts, etc.\ that are available in your installation. It
% will usually cause line and page breaks to occur at other places than
% they would do if typeset using an uncustomised \package{tclldoc}
% class, so the typographical quality of the document can be decreased,
% but it is uncommon to find an \texttt{.dtx} document whose author
% have given these matters much attention anyway. Hence the typographic
% arguments against customisation are weak.
%
% A common form of customisation is to use additional packages, since
% various kinds of document-wide font selection is often done by
% packages. Due to that the code in \texttt{tclldoc.cfg} is executed when
% the \package{tclldoc} class does its option processing, at which time
% \LaTeX\ does not allow loading packages, such customisation is not
% straightforward. There is a way around that however; to load e.g.\ the
% \package{times} package, use the command
%\begin{verbatim}
%\AtEndOfClass{\usepackage{times}}
%\end{verbatim}
% Using |\AtEndOfClass| like this delays the command until it may be
% executed.
%
%
% \subsection{Miscellanea}
%
% For writing ``\Tcl'', the \package{tclldoc} package defines the command
% \DescribeMacro{\Tcllogo}|\Tcllogo|, which for most fonts look slightly
% better than simply typing |Tcl|. (|\Tcllogo| becomes \Tcl, whereas
% |Tcl| becomes Tcl.)
%
% Between the namespace and the tail part of a qualified name, the
% \package{tclldoc} package commands naturally put the name\-space
% separator `|::|'. This text is stored in the macro
% \DescribeMacro{\namespaceseparator}|\namespaceseparator|, which can be
% redefined using the |\Declare|\-|Robust|\-|Command| command. This is
% mainly useful for modifying how this separator behaves with respect to
% line breaking; the default behaviour is that a line break can occur
% between the colons.
%
% Another configurable piece of text is stored in the
% \DescribeMacro{\namespacephrase}|\namespacephrase| macro. This
% contains word `namespace' as that appears in index entries, e.g.~in
% the last word of
% \begin{quote}
% \texttt{platform} (var.), \texttt{alpha} name\-space
% \end{quote}
% It is often convenient to replace this by something shorter. The
% redefinition
% \begin{quote}
% |\renewcommand{\namespacephrase}{\textsc{ns}}|
% \end{quote}
% turns the above into
% \begin{quote}
% \texttt{platform} (var.), \texttt{alpha} \textsc{ns}
% \end{quote}
% Note however that either |\namespacephrase| itself or its expansion
% must be robust.
%
%
% \section{Acknowledgements}
%
% The \package{tclldoc} document class and \LaTeX\ package were
% constructed starting from three other sources: (i)~the \package{ltxdoc}
% document class~\cite{ltxdoc} by David Carlisle,
% (ii)~the \package{doc} package~\cite{doc} by Frank Mittelbach,
% B.~Hamilton Kelly, Andrew Mills, Dave Love, and Joachim Schrod, and
% (iii)~my own \package{pasdoc} document class. Hence the `et al.'\ in
% the author field above. This complicated heritage in the code is
% mirrored by the documented source---there are paragraphs below that
% are rather about one of (i)--(iii), than about \package{tclldoc}.
%
%
%
% \StopEventually{
% \begin{thebibliography}{9}
% \bibitem{ltoutenc}
% Johannes Braams, David Carlisle, Alan Jeffrey, Frank
% Mittelbach, Chris Rowley, and Rainer Sch\"opf:
% \textit{ltoutenc.dtx}, The \LaTeX3 Project;
% \textsc{ctan}:\discretionary{}{}{\thinspace}\texttt{macros}\slash
% \texttt{latex}\slash \texttt{base}\slash \texttt{ltoutenc.dtx}.
% \bibitem{ltxdoc}
% David Carlisle:
% \textit{The file \texttt{ltxdoc.dtx} for use with \LaTeXe},
% The \LaTeX3 Project;
% \textsc{ctan}:\discretionary{}{}{\thinspace}\texttt{macros}\slash
% \texttt{latex}\slash \texttt{base}\slash \texttt{ltxdoc.dtx}.
% \bibitem{rtkinenc}
% Lars Hellstr\"om:
% \textit{The \package{rtkinenc} package}, 2000;
% \textsc{ctan}:\discretionary{}{}{\thinspace}\texttt{macros}\slash
% \texttt{latex}\slash \texttt{contrib}\slash
% \texttt{supported}\slash \texttt{rtkinenc}\slash
% \texttt{rtkinenc.dtx}.
% \bibitem{xdoc}
% Lars Hellstr\"om:
% \textit{The \package{xdoc} package --- experimental
% reimplementations of features from \package{doc},
% second~prototype}, 2000, 2001;
% \textsc{ctan}:\discretionary{}{}{\thinspace}\texttt{macros}\slash
% \texttt{latex}\slash \texttt{exptl}\slash \texttt{xdoc}\slash
% \texttt{xdoc2.dtx}.
% \bibitem{docindex}
% Lars Hellstr\"om:
% \textit{The \package{docindex} package}, 2001;
% \textsc{ctan}:\discretionary{}{}{\thinspace}\texttt{macros}\slash
% \texttt{latex}\slash \texttt{exptl}\slash \texttt{xdoc}\slash
% \texttt{docindex.dtx}.
% \bibitem{TeXbook}
% Donald E.\ Knuth, Duane Bibby (illustrations):
% \textit{The \TeX book}, Ad\-di\-son--Wes\-ley, 1984;
% ISBN~0-201-13448-9 and~0-201-13447-0.
% \bibitem{docstrip}
% Frank Mittelbach, Denys Duchier, Johannes Braams, Marcin
% Woli\'nski, and Mark Wooding: \textit{The \textsf{DocStrip}
% program}, The \LaTeX3 Project;
% \textsc{ctan}:\discretionary{}{}{\thinspace}\texttt{macros}\slash
% \texttt{latex}\slash \texttt{base}\slash \texttt{docstrip.dtx}.
% \bibitem{doc}
% Frank Mittelbach, B.~Hamilton Kelly, Andrew Mills, Dave Love, and
% Joachim \mbox{Schrod}: \textit{The \package{doc} and
% \package{shortvrb} Packages}, The \LaTeX3 Project;
% \textsc{ctan}:\discretionary{}{}{\thinspace}\texttt{macros}\slash
% \texttt{latex}\slash \texttt{base}\slash \texttt{doc.dtx}.
% \bibitem{execMagic}
% ``Tom'', Donal Fellows, Larry Virden, Richard Suchenwirth:
% \textit{exec magic}, The \Tcl'ers Wiki page \textbf{812};
% \textsc{http}:/\slash \texttt{mini.net}\slash \texttt{tcl}\slash
% \texttt{812.html}.
% \bibitem{ucs-package}
% Dominique~P.~G.~Unruh:
% \textit{\texttt{ucs.sty} - Unicode Support}, 2000;
% \textsc{ctan}:\discretionary{}{}{\thinspace}\texttt{macros}\slash
% \texttt{latex}\slash \texttt{contrib}\slash
% \texttt{supported}\slash \texttt{unicode}/.
% \end{thebibliography}
% \begin{flushleft}\footnotesize
% The ``\textsc{ctan}:'' above is short for ``any of the servers
% in the Comprehensive \TeX\ Archive Network (or mirror thereof)''.
% You get a working URL if you replace this by e.g.\
% ``\texttt{ftp://ftp.tex.ac.uk/tex-archive/}''.%
% \end{flushleft}
% }
%
%
% \changes{v\,1.90}{2000/01/02}{Took a copy of \texttt{pasdoc.dtx} and
% started modifying it into \texttt{tcldoc.dtx}. (LH)}
%
% \section{Initial stuff in the \package{tclldoc} package}
%
% The |\NeedsTeXFormat| and |\ProvidesPackage| commands for the
% \package{tclldoc} package appear at the top of \texttt{tclldoc.dtx} (not
% shown in the typeset document). Apart from that the only initial
% action required is to load the \package{xdoc} package (and thereby
% the \package{doc} package) and the \package{docindex} (or rather the
% \LaTeXe\ version \package{docidx2e}) package.
% \SortIndex{docindex package}{\package{docindex} package}
%
% \changes{v\,2.23}{2001/07/21}{Added message about using
% docindex.ist. (LH)}
% \begin{macrocode}
%<*pkg>
\RequirePackageWithOptions{xdoc2}[2001/11/03]
\RequirePackage{docidx2e}
\AtEndDocument{%
\typeout{********************************}%
\typeout{* Use docindex.ist when\@spaces\@spaces*}%
\typeout{* sorting .idx and .glo files. *}%
\typeout{********************************}%
}%
% \end{macrocode}
%
%
% \section{Verbatim typesetting of \Tcl\ code}
%
% \changes{v\,2.10}{2000/10/03}{Major overhaul of the \texttt{tcl}
% environment, to make it possible to assign different penalties to
% different breakpoints. (LH)}
%
% The main feature in the \package{tclldoc} package is the \texttt{tcl}
% environment, which typesets embedded code in a way more adapted to
% \Tcl\ code than what the \texttt{macrocode} environment does.
%
%
% \subsection{Beginning of line processing}
%
% The organization of the text in \emph{lines} is crucial for its
% interpretation by the \texttt{tcl} environment. In particular the
% following features rely on being at the beginning of a line:
% \begin{itemize}
% \item indentation of \Tcl\ code,
% \item invoking command mode,
% \item recognizing \package{docstrip} guards.
% \end{itemize}
%
% \begin{macro}{\TD@percent@token}
% A control sequence |\let| to a |%|$_{12}$, mainly for use in |\ifx|
% comparisons.
% \DoNotIndex{\%}
% \begin{macrocode}
\begingroup
\catcode`\%=12
\global\let\TD@percent@token=%
\endgroup
% \end{macrocode}
% \end{macro}
%
% \begin{macro}{\TD@start@n@line}
% \begin{macro}{\TD@start@n@line@}
% \begin{macro}{\TD@start@n@line@i}
% The macro |\TD@start@n@line| is used at the beginning of every line.
% The macro first clears the |\TD@line@indent| and |\TD@nesting@level|
% registers and the |\TD@nesting@stack| stack. Then it checks if the
% next character is a percent. If it is then this is a command line.
% If it isn't then it is a normal line and it is time to start a
% codeline paragraph and count the character indentation.
%
% |\TD@|\-|start@|\-|n@|\-|line@| is the macro which actually does
% these things; |\TD@|\-|start@|\-|n@|\-|line| is usually |\let| to
% |\TD@|\-|start@|\-|n@|\-|line@|. When a file is being |\TclInput|ed
% however, |\TD@|\-|start@|\-|n@|\-|line| is |\let| to
% |\TD@|\-|start@|\-|n@|\-|line@i| which peeks at the next token
% before calling |\TD@|\-|start@|\-|n@|\-|line@|. If it didn't,
% |\TD@|\-|start@|\-|n@|\-|line@| would have a problem at the end of
% the file, since the end of a file counts as being |\outer|.
% \begin{macrocode}
\def\TD@start@n@line@#1{%
\global\TD@line@indent=\z@
\TD@nesting@level=\z@
\def\TD@nesting@stack{\TD@nesting@level}%
\ifx #1\TD@percent@token
\expandafter\TD@module
\else
\global\advance \c@codelineno \@ne
\TD@begin@tclpar
\expandafter\TD@count@indent \expandafter#1%
\fi
}
% \end{macrocode}
% \begin{macrocode}
\let\TD@start@n@line=\TD@start@n@line@
\def\TD@start@n@line@i{\futurelet\next\TD@start@n@line@}
% \end{macrocode}
% \end{macro}\end{macro}\end{macro}
%
%
% \begin{macro}{\TD@count@indent}
% The purpose of the |\TD@count@indent| macro is to compute the
% indentation of the current line. It grabs the next character, and
% as long as that is a space or tab it will just increment
% |\TD@line@indent| accordingly and continue with the next character.
% |!| and |"| are used to make space and tab respectively with
% suitable catcode.
% \DoNotIndex{\!,\",\ ,\^}
% \begin{macrocode}
\begingroup
\catcode`\!=\active
\catcode`\"=\active
\lccode`\!=`\ %
\lccode`\"=`\^^I%
\lowercase{%
\endgroup
\def\TD@count@indent#1{%
\ifx !#1%
\global\advance \TD@line@indent \@ne
\expandafter\TD@count@indent
\else \ifx "#1%
\global\divide \TD@line@indent \TD@tab@size
\global\advance \TD@line@indent \@ne
\global\multiply \TD@line@indent \TD@tab@size
\expandafter\expandafter \expandafter\TD@count@indent
\else
\TD@setup@parshape
\expandafter\expandafter \expandafter#1%
\fi\fi
}%
}
% \end{macrocode}
% \end{macro}
%
%
% \begin{macro}{\TD@module}
% When command mode has been properly established it is time to
% check if the next character is |<|, in which case the following
% \package{docstrip} guard must be processed. |\TD@command| must
% appear before the |\futurelet| so that the following character
% won't get tokenized with incorrect catcodes.
% \changes{v\,2.13}{2000/12/20}{Always check whether the current line
% is a guard line, since the \texttt{\PrintChar{\number`\<}} in
% the guard would otherwise start a new paragraph, which is an
% error. (LH)}
% \begin{macrocode}
\def\TD@module{%
\TD@command
\futurelet\next
\TD@ch@angle
}
% \end{macrocode}
% \end{macro}
%
%
% \begin{macro}{\TD@ch@angle}
% The module checking in \package{tclldoc} takes advantage of the
% macros for this in \package{doc}---|\ch@plus@etc| is a \package{doc}
% macro. |<| is active because it is in the nolig list
% (|\verbatim@|\-|nolig@|\-|list|).
% \changes{v\,2.20}{2001/03/12}{Moved \cs{TD@normal} previously in
% \cs{TD@pm@module} to here. This means \package{docstrip} guard are
% now read in normal mode rather than command mode. This change
% was made because of grouping problems that occured on non-block
% guard lines. Also added local redefinition of \PrintChar{13} to
% prevent that guard lines are concatenated with the following
% line. (LH)}
% \DoNotIndex{\^}
% \begin{macrocode}
\begingroup
\catcode`<=\active
\catcode`\^^I=\catcode`\^^M \endlinechar=`\^^I \relax
\catcode`\^^M=\active
\gdef\TD@ch@angle{%
\ifx <\next
\TD@normal
\global\advance \c@codelineno \@ne
\TD@begin@tclpar
\def^^M{\TD@active@CR}%
\expandafter\futurelet \expandafter\next
\expandafter\ch@plus@etc
\fi
}
\endgroup
% \end{macrocode}
% \end{macro}
%
% \begin{macro}{\TD@pm@module}
% The |\TD@pm@module| macro is used instead of \package{doc}'s
% |\pm@module| when in the \texttt{tcl} environment. This is necessary
% since the paragraph man\oe uvres of the \texttt{tcl} environment would
% not work right with \package{doc}'s |\pm@module|.
%
% If we're not dealing with a block directive (|*| or |/|), i.e.,
% it's a single special line, we set everything up to the next |>|
% appropriately, return to normal mode, possibly change font, and
% start counting indentation spaces.
% \DoNotIndex{\>}
% \begin{macrocode}
\begingroup
\catcode`\>=\active
\gdef\TD@pm@module#1>{%
\Module{#1}%
\ifnum \guard@level<\c@StandardModuleDepth \else
\AltMacroFont
\fi
\TD@count@indent
}
\endgroup
% \end{macrocode}
% \end{macro}
%
%
% \begin{macro}{\TD@star@module}
% \changes{v\,2.20}{2001/03/12}{Removed \cs{TD@end@tclpar}. (LH)}
% \changes{v\,2.22}{2001/06/13}{Reinserted \cs{TD@end@tclpar} and
% added \cs{TD@command}; this solves a problem with grouping that
% prevented a font change from getting effect. (LH)}
% \begin{macro}{\TD@slash@module}
% \changes{v\,2.20}{2001/03/12}{Removed \cs{TD@end@tclpar}. (LH)}
% \changes{v\,2.22}{2001/06/13}{Same change as in
% \cs{TD@star@module}. (LH)}
% The |\TD@star@module| and |\TD@slash@module| macros are, like
% |\TD@pm@module|, used instead of \package{doc}'s |\star@module| and
% |\slash@module| respectively when in the \texttt{tcl} environment.
% \DoNotIndex{\>}
% \begin{macrocode}
\begingroup
\catcode`\>=\active
\gdef\TD@star@module#1>{%
\Module{#1}%
\TD@end@tclpar
\TD@command
\global\advance \guard@level \@ne
\ifnum \c@StandardModuleDepth<\guard@level
\global\let\macro@font=\AltMacroFont
\macro@font
\fi
}
\gdef\TD@slash@module#1>{%
\Module{#1}%
\TD@end@tclpar
\TD@command
\global\advance \guard@level \m@ne
\ifnum \guard@level=\c@StandardModuleDepth
\global\let\macro@font\MacroFont
\macro@font
\fi
}
\endgroup
% \end{macrocode}
% \end{macro}\end{macro}
%
%
% \begin{macro}{\TD@gobble@whitespace}
% This macro gobbles all normal mode spaces and tabs following it.
% \begin{macrocode}
\def\TD@gobble@whitespace#1{%
\if \ifx #1\TD@active@space
0%
\else
\ifx #1\TD@active@tab 0\else 1\fi
\fi 0%
\expandafter\TD@gobble@whitespace
\else
\expandafter#1%
\fi
}
% \end{macrocode}
% \end{macro}
%
%
%
% \subsection{Typographical elements}
%
% This subsection contains definitions of the typographical
% elements---special symbols and the like--that are used in the
% \texttt{tcl} and \texttt{tcl*} environments.
%
%
% \begin{macro}{\TD@typography}
% \begin{macro}{\TD@nontcl@font}
% The |\TD@typography| macro does all the typographic set-up that is
% common to the \texttt{tcl} environment, the \texttt{tcl*}
% environment, and the |\tclverb| command.
%
% The |\TD@nontcl@font| macro selects an ``escape'' font for \Tcl\
% material. This font is used for various kinds of escape codes that
% are typeset instead of some character---either because the
% character is not available, or because the line needs to be broken.
% The escape font will have the same font family (and same encoding)
% as the text surrounding the \texttt{tcl} environment or |\tclverb|
% command. Therefore |\TD@nontcl@font| must be defined before the
% \Tcl\ material font is selected. I don't think it is necessary to
% use |\protected@edef| below, but it can't hurt either.
% \begin{macrocode}
\def\TD@typography{%
\protected@edef\TD@nontcl@font{%
\noexpand\fontencoding{\cf@encoding}%
\noexpand\fontfamily{\f@family}\noexpand\selectfont
}%
% \end{macrocode}
% The most frequent escape symbol is the single backslash used for
% escaping the end of lines, and this is stored in the
% |\TD@backslash@box| box.
%
% The following code have to do with how characters outside
% visible ASCII are typeset when they appear in the \texttt{tcl} or
% \texttt{tcl*} environment. The commands are defined by the
% \package{rtkinenc} package.
% \begin{macrocode}
\InputModeCode
\SetUnavailableAction{\leavevmode{%
\TD@nontcl@font\textbackslash x\TypesetHexNumber{##1}%
}}%
\DeclareInputMath{0}{\RIE@undefined{0}}%
\DeclareInputMath{12}{\RIE@undefined{12}}%
}
% \end{macrocode}
% \end{macro}\end{macro}
%
% \begin{macro}{\TD@typeset@space}
% This macro typesets a space, as it should be in the \texttt{tcl}
% environment (and other places too). It generates a kern, not glue, so
% that it can be used in a |\discretionary|.
% \begin{macrocode}
\def\TD@typeset@space{\kern\fontdimen\tw@\font}
% \end{macrocode}
% \end{macro}
%
% \begin{macro}{\TD@discretionary@space}
% This is the macro which offers all the breakpoints in |\tclverb|
% code. The name comes from the fact that all the breakpoints are
% discretionaries which appear to be spaces when not broken. When
% they are broken, they are backslashes. To give some visual hint
% that these backslashes need not be backslashes in the source, they
% are typeset in the same font family (and same encoding) as the
% text surrounding the \texttt{tcl} environment.
% \begin{macrocode}
\def\TD@discretionary@space{%
\discretionary{\copy\TD@backslash@box}{}{\TD@typeset@space}%
}%
% \end{macrocode}
% \end{macro}
%
% \begin{macro}{\TD@breakable@space}
% The valid breakpoints inside \texttt{tcl} and \texttt{tcl*}
% environments are made in the |\TD@|\-|breakable@|\-|space| macro.
% This macro contributes an |\hbox| (which contains a
% |\TD@typeset@space|), a penalty, and an empty |\vadjust| to the
% current horizontal list. If a line break is made at the penalty,
% the |\hbox| will later be removed and replaced by a non-macro font
% backslash (from |\TD@backslash@box|). The |\vadjust| is there to
% prevent that any discardable items disappears after the breakpoint.
% \begin{macrocode}
\def\TD@breakable@space{%
\hbox{\TD@typeset@space}%
\TD@nesting@penalty
\vadjust{}%
}
% \end{macrocode}
% \end{macro}
%
% \begin{macro}{\TD@typeset@tab}
% This is an unbreakable space whose width equals that of
% |\TD@tab@size| spaces.
% \begin{macrocode}
\def\TD@typeset@tab{\kern\TD@tab@size\fontdimen\tw@\font}
% \end{macrocode}
% \end{macro}
%
% \begin{macro}{\TD@visible@whitespace}
% The |\TD@visible@whitespace| macro locally redefines the macros
% which typesets spaces and tabs to make visible symbols. The visible
% tab is roughly a |\rightarrowfill| and a |\vrule|, but a difference
% is that the minuses that form the extensible ``tail'' to the
% |\rightarrow| are not smashed. Not doing that gives the |\vrule| a
% suitable height, but I had rather expected the nominal height of
% the |\rightarrow| to be larger \textellipsis\ It's yet another
% mysterious math font feature, I suppose.
%
% \begin{macrocode}
\def\TD@visible@whitespace{%
\def\TD@typeset@space{\char32 }%
\def\TD@typeset@tab{%
\hb@xt@\TD@tab@size\fontdimen\tw@\font{%
$\m@th-\mkern-7mu%
\cleaders\hbox{$\mkern-2mu-\mkern-2mu$}\hfill
\mkern-7mu\mathord\rightarrow\mkern-1mu\vrule\mkern1mu$%
}%
}%
}
% \end{macrocode}
% \end{macro}
%
%
% \subsection{Catcodes and active characters}
%
% Like any \texttt{verbatim}-like environment, the \texttt{tcl}
% environment does some rather extensive alterations of catcodes. This
% is handled by the three macros |\TD@general|, |\TD@normal|, and
% |\TD@command|. Set-up that is common for both modes is made by
% |\TD@general| when the \texttt{tcl} environment is entered.
% Mode-dependent set-up is made by |\TD@normal| and |\TD@command|
% whenever a mode switch occurs.
%
% \begin{macro}{\TD@let@active}
% One technicality here is that all of these routines need to set the
% definition of some active character, and this character is often not
% active with normal \LaTeX\ catcodes. To overcome this difficulty,
% there is a macro |\TD@let@active| which takes two arguments: a
% control sequence with a single character name, and an arbitrary
% control sequence. The active character with the same name as the
% first control sequence is |\let| locally to the second control
% sequence.
% \begin{macrocode}
\begingroup
\catcode\z@=\active
\gdef\TD@let@active#1{%
\begingroup
\lccode\z@=`#1%
\lowercase{%
\endgroup
\let^^@%
}%
}%
\endgroup
% \end{macrocode}
% \end{macro}
%
%
% \begin{macro}{\TD@general}
% The |\TD@general| macro does the initial set-ups that are common for
% both modes of the \texttt{tcl} environment and for the |\tclverb|
% command. These include definitions of various active characters and
% |\catcode| assignments.
%
% \DoNotIndex{\^,\ ,\\,\[,\]}
% \DoNotIndexHarmless{\PrintChar{`\{}}
% \DoNotIndexHarmless{\PrintChar{`\}}}
% \begin{macrocode}
\def\TD@general{%
\let\do\do@noligs
\verbatim@nolig@list
\let\do\@makeother
\dospecials
\catcode`\^^M=\active
\TD@let@active\^^I\TD@active@tab
\TD@let@active\^^M\TD@active@CR
\TD@let@active\ \TD@active@space
\TD@let@active\\\TD@active@backslash
\TD@let@active\{\TD@active@braceleft
\TD@let@active\}\TD@active@braceright
\TD@let@active\[\TD@active@bracketleft
\TD@let@active\]\TD@active@bracketright
}
% \end{macrocode}
% \end{macro}
%
%
% \begin{macro}{\TD@normal}
% This macro sets up the normal mode, changing everything that is
% different in any of the other modes. Note that the catcodes in
% normal mode are such that no character is a space. This makes it
% possible for macros used in normal mode to look ahead by simply
% grabbing the next character in an undelimited argument.
% \DoNotIndex{\^,\ ,\\,\[,\]}
% \DoNotIndexHarmless{\PrintChar{`\{}}
% \DoNotIndexHarmless{\PrintChar{`\}}}
% \begin{macrocode}
\def\TD@normal{%
\catcode`\^^I=\active
\catcode`\ =\active
\catcode`\\=\active
\catcode`\{=\active
\catcode`\}=\active
\catcode`\[=\active
\catcode`\]=\active
}
% \end{macrocode}
% \end{macro}
%
% \begin{macro}{\TD@command}
% This macro switches from normal mode to command mode.
% \DoNotIndex{\^,\ ,\\,\[,\]}
% \DoNotIndexHarmless{\PrintChar{`\{}}
% \DoNotIndexHarmless{\PrintChar{`\}}}
% \begin{macrocode}
\def\TD@command{%
\catcode`\^^I=10%
\catcode`\ =10%
\catcode`\[=12%
\catcode`\]=12%
\catcode`\\=\z@
\catcode`\{=\@ne
\catcode`\}=\tw@
}
% \end{macrocode}
% \end{macro}
%
%
% \begin{macro}{\TD@active@space}
% \begin{macro}{\TD@active@space@}
% This macro is used for spaces in normal mode. Its main problem is
% to determine whether it is followed by more whitespace or not. If it
% is, then it is an unbreakable space. If it isn't, then it is a
% breakable space.
% \begin{macrocode}
\def\TD@active@space#1{%
\ifx #1\TD@active@backslash
\expandafter\TD@active@space@
\else
\ifx #1\TD@active@space
\TD@typeset@space
\else\ifx #1\TD@active@tab
\TD@typeset@space
\else
\TD@breakable@space
\fi\fi
\expandafter#1%
\fi
}
% \end{macrocode}
% The special case to look out for is that the next character is a
% backslash which escapes a newline.
% \begin{macrocode}
\def\TD@active@space@#1{%
\ifx #1\TD@active@CR
\TD@typeset@space
\else
\TD@breakable@space
\fi
\TD@active@backslash #1%
}
% \end{macrocode}
% \end{macro}\end{macro}
%
% \begin{macro}{\TD@active@tab}
% The |\TD@active@tab| macro is what the active character tab is
% |\let| to inside the \texttt{tcl} and \texttt{tcl*} environments.
% \begin{macrocode}
\def\TD@active@tab{\TD@typeset@tab}
% \end{macrocode}
% \end{macro}
%
% \begin{macro}{\TD@active@CR}
% This macro is called at the end of a line. The |\catcode| tests the
% current mode.
% \DoNotIndex{\\}
% \begin{macrocode}
\def\TD@active@CR{%
\ifnum \catcode`\\=\z@
\ifvmode\else
\PackageError{tclldoc}{Horizontal material on command line}\@ehc
\@@par
\fi
\TD@normal
\else
\TD@end@tclpar
\fi
\TD@start@n@line
}
% \end{macrocode}
% \end{macro}
%
% \begin{macro}{\TD@active@CRv}
% This is what |\tclverb| defines the |^^M| active character as
% being. Cf.\ the standard \LaTeX\ macro |\verb@eol@error|.
% \begin{macrocode}
\def\TD@active@CRv{%
\verb@egroup
\PackageError{tclldoc}{\protect\tclverb\space ended by end of line}%
\@ehc
}
% \end{macrocode}
% \end{macro}
%
%
% \begin{macro}{\TD@active@backslash}
% \changes{v\,2.10}{2000/10/08}{First checking if next character is
% active, and only expands the macro containing the detailed tests
% if it is. (LH)}
% \begin{macro}{\TD@active@backslash@}
% \begin{macro}{\TD@active@backslash@v}
% |\TD@active@backslash| is used for backslashes in normal mode. Its
% main problem is to determine whether the backslash is escaping
% something (which would normally be treated specially by the
% environment) or not. The primary test for this is whether the
% argument grabbed is an active character, and if it is then
% processing continues (in |\TD@active@backslash@|) to determine what
% should be done.
% \begin{macrocode}
\def\TD@active@backslash#1{%
\ifcat \noexpand#1\noexpand~%
\expandafter\TD@active@backslash@
\else
\@backslashchar
\fi
#1%
}
% \end{macrocode}
%
% In order to have guard lines processed correctly, the first test
% below really must be for |\TD@active@CR| rather than for an
% active |^^M|.
% \changes{v\,2.11}{2000/11/06}{Escaped spaces are no longer breakable.
% (LH)}
% \begin{macrocode}
\def\TD@active@backslash@#1{%
\ifcase
\ifx #1\TD@active@CR 0%
\else\ifx #1\TD@active@backslash 1%
\else\ifx #1\TD@active@braceleft 2%
\else\ifx #1\TD@active@braceright 2%
\else\ifx #1\TD@active@bracketleft 2%
\else\ifx #1\TD@active@bracketright 2%
\else\ifx #1\TD@active@space 3%
\else 4\fi\fi\fi\fi\fi\fi\fi
\space
\expandafter\TD@active@backslash@i
\or
\@backslashchar\@backslashchar
\or
\@backslashchar \string#1%
\or
\@backslashchar \TD@typeset@space
\else
\@backslashchar
\expandafter#1%
\fi
}
% \end{macrocode}
% Escaped newlines that are followed by command mode lines are not
% converted to discretionary spaces.
% \begin{macrocode}
\def\TD@active@backslash@i#1{%
\ifx #1\TD@percent@token
\copy\TD@backslash@box
\expandafter\TD@active@CR
\else
\TD@breakable@space
\global\advance \c@codelineno \@ne
\expandafter\TD@gobble@whitespace
\fi
#1%
}
% \end{macrocode}
% The |\tclverb| command uses |\TD@active|\-|@backslash@v| instead of
% |\TD|\-|@active|\-|@backslash@i|, since it is intended for use on
% lines that may begin with |%|.
% \begin{macrocode}
\def\TD@active@backslash@v#1{%
\TD@discretionary@space
\ifx #1\TD@percent@token
\expandafter\@firstoftwo
\fi
\TD@gobble@whitespace #1%
}
% \end{macrocode}
% \end{macro}\end{macro}\end{macro}
%
%
% \begin{macro}{\TD@active@bracketleft}
% \begin{macro}{\TD@active@bracketright}
% These macros mainly typeset the corresponding characters, but they
% also increase or decrease the |\TD@nesting@level| count by one.
% \begin{macrocode}
\def\TD@active@bracketleft{[\advance\TD@nesting@level\@ne}
\def\TD@active@bracketright{]\advance\TD@nesting@level\m@ne}
% \end{macrocode}
% \end{macro}\end{macro}
%
% \begin{macro}{\TD@active@braceleft}
% \begin{macro}{\TD@active@braceright}
% \begin{macro}{\TD@nesting@stack}
% \changes{v\,2.14}{2001/01/19}{Nesting stack added to cope with code
% (brace-delimited strings) where brackets aren't properly balanced.
% (LH)}
% The |\TD@active@braceleft| and |\TD@active@braceright| work very
% much like their \texttt{bracket} counterparts, but they also push
% and pop the current nesting level on and off the |\TD@nesting@stack|
% macro. The reason for doing this is that brackets aren't always
% properly balanced in \Tcl\ code, which may result in a
% |\TD@nesting@level| value which does not mirror how the code really
% will be interpreted. Braces must however be properly balanced, and
% therefore it is always right to restore the value of
% |\TD@nesting@level| that was in force before a brace group began
% when it is ended.
%
% The |\TD@nesting@stack| macro is a \meta{nesting stack}, which is
% one of the following two things
% \begin{quote}
% |\TD@nesting@level|
% \meta{number}|\def\TD@nesting@stack{|\meta{nesting stack}|}|
% \end{quote}
% This means a value is popped off the stack by doing
% \begin{quote}
% |\TD@nesting@level=\TD@nesting@stack|
% \end{quote}
% and |\TD@nesting@level| is unchanged if the stack is empty.
% \DoNotIndex{\[,\]}
% \DoNotIndexHarmless{\PrintChar{`\{}}
% \DoNotIndexHarmless{\PrintChar{`\}}}
% \begin{macrocode}
\begingroup
\catcode`\{=12 \catcode`\}=12
\catcode`\[=1 \catcode`\]=2
\gdef\TD@active@braceleft[{%
\expandafter\def \expandafter\TD@nesting@stack \expandafter[%
\the\expandafter\TD@nesting@level
\expandafter\def \expandafter\TD@nesting@stack
\expandafter[\TD@nesting@stack]%
]%
\advance\TD@nesting@level\@ne
]
\gdef\TD@active@braceright[}%
\advance\TD@nesting@level\m@ne
\TD@nesting@level=\TD@nesting@stack
]
\endgroup
\def\TD@nesting@stack{\TD@nesting@level}
% \end{macrocode}
% \end{macro}\end{macro}\end{macro}
%
%
%
% \subsection{Paragraph formatting}
%
% This subsection contains definitions of control sequences that are
% mainly related to the formatting of paragraphs.
%
% The new implementation of the \texttt{tcl} and \texttt{tcl*}
% environments is based on inserting the ``discretionary'' backslashes
% at the linebreaks \emph{after} the paragraph has been broken.
% Doing this involves reboxing every line in the paragraph, and in this
% process the line adjustments are changed as well.
%
% \begin{macro}{\TD@backslash@box}
% The |\TD@backslash@box| box register is set to a backslash in the
% special |\TD@nontcl@font| at the beginning of each \texttt{tcl} and
% \texttt{tcl*} environment. These backslashes are used to denote
% ``backslash escaping newline''. The primary reason for keeping this
% glyph in a box is not that this might be slightly faster to
% typeset, but to have the \emph{width} of it easily accessible.
% \begin{macrocode}
\newbox\TD@backslash@box
% \end{macrocode}
% \end{macro}
%
% \begin{macro}{\TD@nesting@level}
% \begin{macro}{\TD@nesting@penalty}
% |\TD@nesting@level| is a count register which keeps track of how
% many brace\slash bracket groups surrounds the current position in
% the line. It is cleared to zero at the beginning of each paragraph.
% \begin{macrocode}
\newcount\TD@nesting@level
% \end{macrocode}
% |\TD@nesting@penalty| makes a |\penalty| whose value should depend
% on the nesting level (the deeper the nest, the larger the penalty).
% It may also adjust the value of |\linepenalty| to counter that
% the nesting level has become negative. This is typically needed for
% lines which say things like ``|} else {|''. The default definition is
% to make the nesting penalty $100$ times the nesting level and to
% make the |\linepenalty| equal to
% \(10 - \min\{0,\mbox{least penalty}\}\).
% \begin{macrocode}
\def\TD@nesting@penalty{%
\penalty \the\TD@nesting@level 00\relax
\ifnum \lastpenalty<-\linepenalty
\linepenalty=10%
\advance \linepenalty -\lastpenalty
\fi
}
% \end{macrocode}
% I doubt that this is necessarily the best algorithm for computing
% these penalties, so if you are really thorough about your
% typography you may well want to experiment with other definitions of
% |\TD@nesting@penalty|. In case you should then find something which
% works even better then I'm interested to learn about it.
% \end{macro}\end{macro}
%
% \begin{macro}{\TD@reformat@lines}
% The |\TD@reformat@lines| macro calls itself recursively to reformat
% all lines on the current vertical list. The first line will remain
% flush left, but all other lines will be reset flush right. The
% visible material on the last line will be left as it is, but the
% last box in all other lines will be replaced by a non-macro font
% backslash.
%
% It is very important that the current vertical list is not the main
% vertical list.
% It is assumed that the current vertical list consists of a sequence
% of \meta{box}, \meta{penalty}, \meta{glue}, with an extra glue item
% at the top of the list. It is OK if some penalty or glue item is
% missing. In case the list contains other material as well the line
% reformatting may be stopped prematurely, but there is a trick that
% allows one to put arbitrary material between the lines of the
% reformatted paragraph: rather than doing e.g.
% \begin{quote}
% |\mark|\marg{text}
% \end{quote}
% in the paragraph, do
% \begin{quote}
% |\vadjust{\vbox{\mark{|\meta{text}|}}}|
% \end{quote}
% The |\vbox| will be recognised by the paragraph reformatting
% mechanism as a container for vertical mode material that appears
% between the lines of the paragraph, so it will simply be unboxed.
% \changes{v\,2.12}{2000/11/11}{\cs{vbox} containers for vertical
% material are allowed between the lines of a reformatted
% paragraph. (LH)}
%
% Each line's horizontal list ends with
% \begin{itemize}
% \item a box (which contains the space that is to be replaced by a
% backslash),
% \item a penalty (at which the paragraph was broken), and
% \item a glue item (the |\rightskip|).
% \end{itemize}
%
% A tricky feature in the implementation is that the
% |\bgroup|--|\egroup| nesting will be off by one. The |\bgroup| at
% the beginning of a |\TD@reformat@line| will be matched by the
% |\egroup| at the end of the |\TD@reformat@line| that the first one
% calls!
% \begin{macrocode}
\def\TD@reformat@lines{%
\bgroup
\unskip
\count@=\lastpenalty \unpenalty
\setbox\z@=\lastbox
\ifvoid\z@
% \end{macrocode}
% The recursion had already descended down to the last line of the
% paragraph, and it is now time to reformat it.
% \begin{macrocode}
\egroup
\prevdepth=\TD@prevdepth
\hbox{%
\unhbox\z@
\unskip \unpenalty
\setbox\z@=\lastbox
\copy\TD@backslash@box
}%
\else
% \end{macrocode}
% Else there may be another line, and the |\TD@reformat@lines| recursion
% must continue to descend. Upon return the box currently in box
% register zero must be reformatted as a non-first line (flush
% right) and it cannot be the last line in the paragraph, so it is
% always correct to replace the last box by a backslash.
% \begin{macrocode}
\TD@reformat@lines
\ifvbox\z@ \unvbox\z@ \else
\hb@xt@\dimen@{%
\hfill
\unhbox\z@
\unskip \unpenalty
\setbox\z@=\lastbox
\copy\TD@backslash@box
}%
\fi
\fi
\ifnum \count@=\z@ \else \penalty\count@ \fi
\egroup
}
% \end{macrocode}
% \end{macro}
%
% \begin{macro}{\TD@reformat@par}
% The |\TD@reformat@par| macro reformats all lines (they're supposed
% to constitute a paragraph, but that isn't so important) in the
% current vertical list. The restrictions of |\TD@reformat@lines| on
% what may appear in the list apply and there must be at least two
% lines in the list. |\dimen@| is used to hold the desired width of
% reformatted paragraphs.
%
% More precisely |\TD@reformat@par| takes care of the last line of
% the paragraph and the possible |\vbox| containers for vertical
% material that may follow it. Everything in the paragraph that comes
% before the last line is handled by |\TD@reformat@lines|.
% \changes{v\,2.12}{2000/11/11}{\cs{vbox} containers for vertical
% material are allowed after the last line of a reformatted
% paragraph. (LH)}
% \begin{macrocode}
\def\TD@reformat@par{%
\unskip
\count@=\lastpenalty \unpenalty
\setbox\z@=\lastbox
\ifvbox\z@
\bgroup
\TD@reformat@par
\egroup
\unvbox\z@
\else\ifnum \prevgraf>\@ne
\dimen@=\@totalleftmargin
\advance \dimen@ \linewidth
\bgroup
\unskip
\count@=\lastpenalty \unpenalty
\setbox\z@=\lastbox
\TD@reformat@lines
\hb@xt@\dimen@{\hfill \unhbox\z@ \unskip}%
\else
\unskip
\prevdepth=\TD@prevdepth
\box\z@
\fi\fi
\ifnum \count@=\z@ \else \penalty\count@ \fi
}
% \end{macrocode}
% \end{macro}
%
% \begin{macro}{\TD@prevdepth}
% |\TD@prevdepth| is a macro which is used for storing the value of
% |\prevdepth| at times where \TeX\ modifies the primitive in unwanted
% ways. It should always be set globally.
% \end{macro}
%
%
% \begin{macro}{\TD@begin@tclpar}
% The |\TD@begin@tclpar| macro is called when a paragraph in a
% \texttt{tcl} or \texttt{tcl*} environment is about to start. It
% takes care of setting up things so that the paragraph can later be
% reformatted using |\TD@reformat@par|, but it also has to make sure
% that this reformatting doesn't affect the way the paragraph blends
% in with vertical material before and after it.
%
% Reformatting requires that the paragraph is first built in
% restricted vertical mode, i.e., it has to be built in an explicit
% |\vbox|. A problem with this is however that it changes the value of
% |\prevdepth|, which must therefore be explicitly restored.
% \begin{macrocode}
\def\TD@begin@tclpar{%
\xdef\TD@prevdepth{\the\prevdepth}%
\setbox\z@=\vbox\bgroup
\color@begingroup
\prevdepth=\TD@prevdepth
\indent
}
% \end{macrocode}
% \end{macro}
%
% \begin{macro}{\TD@end@tclpar}
% The |\TD@end@tclpar| macro ends a paragraph begun by
% |\TD@begin@tclpar|, reformats it (|\TD@reformat@par|), and
% contributes it to the surrounding vertical list. The |\begingroup|
% and |\endgroup| are there to sort things out in case the recursion
% in |\TD@reformat@par| fails to match as intended. The second
% |\@@par| sees to that the page builder is exercised (without it,
% several pages may go onto the main vertical list without anything
% being shipped out).
% \begin{macrocode}
\def\TD@end@tclpar{%
\@@par
\begingroup
\skip@=\lastskip
\TD@reformat@par
\vskip\skip@
\endgroup
\xdef\TD@prevdepth{\the\prevdepth}%
\color@endgroup
\egroup
\unvbox\z@
\prevdepth=\TD@prevdepth
\@@par
}
% \end{macrocode}
% \end{macro}
%
%
% \begin{macro}{\TD@line@indent}
% |\TD@line@indent| is a |\count| register holding the indentation of
% the current line, in number of spaces. |\TD@line@indent| should
% always be assigned globally.
% \begin{macrocode}
\newcount\TD@line@indent
% \end{macrocode}
% \end{macro}
%
% \begin{macro}{\TD@par@setup}
% This macro sets up various paragraph formatting parameters for the
% normal mode. It mainly consists of code from |\macro@code| of
% \package{doc}.
% \begin{macrocode}
\def\TD@par@setup{%
% \end{macrocode}
%
% In theory a \texttt{tcl} environment should consist of a
% \texttt{trivlist} environment, but the empty space before and after
% the environment should not be too large.
% \begin{macrocode}
\topsep\MacrocodeTopsep
% \end{macrocode}
%
% The next parameter we set is |\@beginparpenalty|, in order
% to prevent a page break before such an environment.
% \begin{macrocode}
\@beginparpenalty\predisplaypenalty
% \end{macrocode}
%
% We then start a |\trivlist|, set |\parskip| back to
% zero and start an empty |\item|.
% \changes{ \package{doc} v\,1.9b}{1993/12/03}{Forcing any label
% from macro env.}
% \begin{macrocode}
\if@inlabel\leavevmode\fi
\trivlist
\parskip\z@skip
\item[]%
% \end{macrocode}
% However, the \texttt{trivlist} environment is only started to make
% standard \LaTeX\ happy (and to have the various list-related
% measurements properly calculated). Everything below will instead
% by done with \TeX\ primitives.
% \changes{ \package{pasdoc} v\,1.31}{1999/09/02}{\cs{parshape}
% assignment added. (LH)}
% \changes{v\,2.10}{2000/10/08}{\cs{parshape} assignment removed (now
% done in \cs{TD@begin@tclpar}). (LH)}
%
% It is important that the change to a special font does not take
% place before the above |\item|, otherwise a change to
% |\baselineskip| will affect the paragraph above.
% \begin{macrocode}
\macro@font
\frenchspacing
% \end{macrocode}
%
% \leavevmode
% \changes{v\,2.12}{2000/11/11}{Assignment to \cs{TD@backslash@box}
% moved from \cs{TD@typography} to here, so that the size will be
% right. (LH)}^^A
% One escape symbol that is particularly common is a single
% backslash, and this is stored in the |\TD@backslash@box| box. As
% some encodings (e.g.\ \texttt{OT1}) does not contain a backslash,
% these escape backslashes are made with |\textbackslash|, rather
% than an explicit character.
% \begin{macrocode}
\sbox\TD@backslash@box{\TD@nontcl@font\textbackslash}%
% \end{macrocode}
%
% The |\rightskip| is given a generous stretchability of |\linewidth|
% so that it doesn't matter too much if the lengths of lines in the
% paragraph varies. Decreasing this value would make the nesting in
% the code less important.
% \begin{macrocode}
\parindent=\@totalleftmargin
\advance \parindent \MacroIndent
\leftskip=\z@skip
\rightskip=\z@ \@plus \linewidth\relax
% \end{macrocode}
% The next two lines are from the definition of the
% \texttt{macrocode} environment, and I (LH) have no idea what they are
% good for, but I suppose I can keep them, at least for now.
% \begin{macrocode}
\global\@newlistfalse
\global\@minipagefalse
% \end{macrocode}
% \changes{ \package{pasdoc} v\,1.31}{1999/09/02}{Made list
% environments indent line numbers. (LH)}
% \changes{v\,2.33}{2001/12/12}{Made list environments not indent
% line numbers per default, but providing a way for the user to
% change this. (LH)}
% Line numbers are inserted using |\everypar|.
% \begin{macrocode}
\ifcodeline@index
\everypar={\llap{%
\PrintCodelineNo\ \hskip\codelineindentfactor\@totalleftmargin
}}%
\else
\everypar={}%
\fi
% \end{macrocode}
%
% These commands replace \package{doc}'s final module processing
% macros with macros that work in the \texttt{tcl} environment.
% \begin{macrocode}
\let\pm@module=\TD@pm@module
\let\star@module=\TD@star@module
\let\slash@module=\TD@slash@module
}
% \end{macrocode}
% \end{macro}
%
% \begin{macro}{\codelineindentfactor}
% The |\codelineindentfactor| macro specifies by which fraction of
% |\@totalleftmargin| that the list environment indentation of
% codeline numbers will be countered. |0| means that the numbers will
% be indented, |1| or empty (the default) that they will not be
% indented. |0.5| would mean that the effective indentation is half
% that of other material.
% \begin{macrocode}
\let\codelineindentfactor\@empty
% \end{macrocode}
% \end{macro}
%
% \begin{macro}{\TD@setup@parshape}
% This macro sets the shape of the current paragraph, primarily based
% on the value in |\TD@line@indent|, and makes the final indentation
% of the first line. By the time it gets called, there usually already
% is some material (|\indent|, codeline number, and possibly a
% \package{docstrip} guard) in the paragraph horizontal list.
%
% The basic area in which the paragraph material is to be put has a
% left margin at |\@totalleftmargin|${}+{}$|\MacroIndent| and a right
% margin at |\@totalleftmargin|${}+{}$|\linewidth| (in both cases the
% position is that much to the right of the left edge of the galley).
% The printed codeline number (if there is one) will be put to the
% left of the left margin, but all other material is indented at least
% the width of another |\TD@line@indent| spaces. In the first line
% that will be done through an explicit kern, but in subsequent lines
% it will instead be done through a |\parshape| assignment. On these
% subsequent lines, there will be an additional indentation of\/
% $1\,\mathrm{em}$ to mark the continuation.
%
% When the paragraph is broken, the right margin will be moved to the
% right by the width of a space minus the width of the
% |\TD@backslash@box|. The reason for this is that the paragraph will
% be broken with spaces (from |\TD@breakable@space|) at the end of
% lines, but that these spaces will later be replaced by the
% backslash in |\TD@backslash@box|. As that doesn't happen in the
% very last line of the paragraph however, something special must be
% done there. The option chosen is to give the |\parfillskip| a
% nonzero natural width.
%
% The length |\@totalleftmargin|${}+{}$|\MacroIndent| is put in
% |\parindent| at the beginning of the environment.
% \begin{macrocode}
\def\TD@setup@parshape{%
\parfillskip=\fontdimen\tw@\font \@plus 1fil%
\advance \parfillskip -\wd\TD@backslash@box
\dimen@=\MacroIndent
\advance \dimen@ \TD@line@indent\fontdimen\tw@\font
\advance \dimen@ 1em%
\dimen@ii=\linewidth
\advance \dimen@ii -\dimen@
\dimen4=\linewidth
\advance \dimen@ \@totalleftmargin
\advance \dimen4 \@totalleftmargin
\advance \dimen@ii -\parfillskip
\advance \dimen4 -\parfillskip
\parshape \tw@ \z@ \dimen4 \dimen@ \dimen@ii
% \end{macrocode}
% Then one just needs one more kern to indent the first line
% properly.
% \begin{macrocode}
\kern \TD@line@indent\fontdimen\tw@\font
}
% \end{macrocode}
% \end{macro}
%
% \begin{macro}{\TD@trace@pardimen}
% This macro is for tracing dimens that are of interest for paragraph
% breaking. Its call syntax is
% \begin{quote}
% |\TD@trace@pardimen|\marg{text}\marg{dimen}
% \end{quote}
% where \meta{text} will be written to the log file to identify the
% dimension, and \meta{dimen} is the dimen register in question (it
% can be anything that can appear after |\the|).
%
% To reduce the amount of text written, this macro only does anything
% if the \TeX\ parameter |\tracingparagraphs| is positive.
% \begin{macrocode}
%<*partrace>
\def\TD@trace@pardimen#1#2{%
\ifnum \tracingparagraphs>\z@
\immediate\write\m@ne{#1 \the#2}%
\fi
}
%</partrace>
% \end{macrocode}
% \changes{ \package{pasdoc} v\,1.21}{1999/08/15}{Added macro and
% some calls of it. All of those are however surrounded by
% \texttt{partrace} guards. (LH)}
% \end{macro}
%
%
%
% \subsection{User interface}
%
% This subsection defines the actual environment and the
% commands and parameters intended to be user-definable.
%
%
% \begin{macro}{\settabsize}
% \changes{v\,2.12}{2000/11/09}{New command for setting the tab size.
% (LH)}
% \begin{macro}{\TD@tab@size}
% \changes{v\,2.12}{2000/11/09}{New name for \cs{c@tabSize}; no longer a
% count register. (LH)}
% The |\TD@tab@size| control sequence stores (as a |\chardef| token)
% the number of spaces that are equivalent to one tab; the default
% value is 8. To set this number, one uses the command |\settabsize|,
% which has the syntax
% \begin{quote}
% |\settabsize|\marg{new size}
% \end{quote}
% The effect of |\settabsize| is local.
% \begin{macrocode}
\newcommand*\settabsize[1]{\chardef\TD@tab@size=#1\relax}
\settabsize{8}
% \end{macrocode}
% \end{macro}\end{macro}
%
%
% \begin{environment}{tcl}
% \begin{environment}{tcl*}
% This is some kind of a definition. There are aspects of the
% interaction with \texttt{trivlist} that I (LH) don't understand,
% but it appears to work.
% \begin{macrocode}
\newenvironment{tcl}{%
\TD@general
\TD@typography
\TD@par@setup
\TD@command
}{%
\global\@inlabelfalse
\endtrivlist
}
% \end{macrocode}
% \begin{macrocode}
\newenvironment{tcl*}{%
\TD@visible@whitespace
\TD@general
\TD@typography
\TD@par@setup
\TD@command
}{%
\global\@inlabelfalse
\endtrivlist
}
% \end{macrocode}
% \end{environment}\end{environment}
%
% \begin{macro}{\tclverb}
% \begin{macro}{\TD@verb}
% The |\tclverb| macro is for typesetting short passages of \Tcl\ code
% which one does not want to have a special paragraph for. It is
% similar to the standard \LaTeX\ command |\verb|, but it has a couple
% of extra features. One is that spaces that are not followed by more
% whitespace are discretionary breakpoints, just as in the \texttt{tcl}
% environment. Another is that a newline can be escaped by putting a
% backslash at the end of an input line. Such an escaped newline will
% count as a space, and whitespace at the beginning of the line
% following it will be ignored. A percent character immediately
% following an escaped newline will also be ignored.
%
% |\tclverb| has a |*| form |\tclverb*|, which is analogous to the
% standard \LaTeX\ |\verb*|, i.e., spaces are typeset using ``visible
% space'' symbols.
% \begin{macrocode}
\newcommand\tclverb{%
\relax\ifmmode\hbox\else\leavevmode\null\fi
\bgroup
\@ifstar{\TD@visible@whitespace\TD@verb}\TD@verb
}
% \end{macrocode}
% \DoNotIndex{\ ,\^,\\}
% \begin{macrocode}
\def\TD@verb{%
\let\TD@active@backslash@i=\TD@active@backslash@v
\let\TD@active@CR=\TD@active@CRv
\let\TD@breakable@space=\TD@discretionary@space
\TD@general
\catcode`\^^I=\active
\catcode`\ =\active
\catcode`\\=\active
\TD@typography
\verbatim@font
\sbox\TD@backslash@box{\TD@nontcl@font\textbackslash}%
\@sverb
}
% \end{macrocode}
% \end{macro}\end{macro}
%
%
% \begin{macro}{\TclInput}
% The |\TclInput| macro is very much like the standard \LaTeX\
% |\input|, but it is intended to be used in command mode, for
% inputting \Tcl\ files that have not been marked up. |\input| would
% work in this case, but there would be errors at the first and last
% lines of the |\input|ted file.
% \begin{macrocode}
\newcommand\TclInput[1]{%
\IfFileExists{#1}{%
\@addtofilelist{#1}%
\begingroup
\TD@normal
\let\TD@start@n@line=\TD@start@n@line@i
\expandafter\TD@start@n@line
\@@input\@filef@und\TD@percent@token
\endgroup
}{\PackageError{tclldoc}{No file #1}\@eha}%
}
% \end{macrocode}
% \end{macro}
%
% \begin{macro}{\MakeShortTclverb}
% This definition follows \package{doc}'s |\MakeShortVerb| pretty
% close.
% \DoNotIndex{\~}
% \begin{macrocode}
\newcommand\MakeShortTclverb[1]{%
\expandafter\ifx \csname cc\string#1\endcsname\relax
\PackageInfo{tclldoc}{%
Made \expandafter\@gobble\string#1 a short %
reference for \string\tclverb}%
\add@special{#1}%
\expandafter\xdef \csname cc\string#1\endcsname{\the\catcode`#1}%
\begingroup
\lccode`\~=`#1%
\lowercase{%
\global\expandafter\let
\csname ac\string#1\endcsname ~%
\gdef~{\tclverb~}}%
\endgroup
\global\catcode`#1\active
\else
\PackageInfo{tclldoc}{%
\expandafter\@gobble\string#1 already a short verb %
reference}%
\fi
}
% \end{macrocode}
% \end{macro}
%
%
% \section{Miscellaneous markup features}
%
% \subsection{Namespaces}
%
% \changes{v\,2.10}{2000/10/13}{Added namespace support. (LH)}
%
% \begin{macro}{\setnamespace}
% \begin{macro}{\TD@convert@colons}
% Since it is common that many identifiers in the same namespace are
% defined in sequence, one can specify a default namespace to use for
% all commands where no explicit namespace is given. This is done
% using the |\setnamespace| command, whose syntax is
% \begin{quote}
% |\setnamespace|\marg{namespace}
% \end{quote}
% This converts the \meta{namespace} to a harmless character string
% and locally assigns it to the |\TD@namespace| macro, which stores
% the current default namespace. |\TD@convert@colons| replaces all
% |::| substrings in the namespace by |\namespaceseparator|s.
% \changes{v\,2.33}{2001/12/12}{Converting \texttt{::} in namespace
% names. (LH)}
% \begin{macrocode}
\newcommand\setnamespace[1]{%
\MakeHarmless\TD@namespace{#1}%
\protected@edef\TD@namespace{%
\expandafter\TD@convert@colons \TD@namespace ::\relax
}
}
% \end{macrocode}
% \begin{macrocode}
\def\TD@convert@colons#1::#2{%
#1%
\ifx \relax#2\else
\noexpand\namespaceseparator
\TD@convert@colons #2%
\fi
}
% \end{macrocode}
% \end{macro}\end{macro}
%
% \begin{macro}{\TD@namespace}
% At the beginning of a document, the default namespace is the global
% namespace, which is represented by the empty string.
% \begin{macrocode}
\let\TD@namespace\@empty
% \end{macrocode}
% \end{macro}
%
% \begin{macro}{\buildname}
% \changes{v\,2.13}{2000/12/16}{Added \cs{@empty} after test for
% \cs{NoValue}, so that it works for empty arguments. (LH)}
% \begin{macro}{\namespaceseparator}
% \changes{v\,2.12}{2000/11/25}{Command added. (LH)}
% The |\buildname| macro builds a qualified name from a namespace and
% a name tail. It has the syntax
% \begin{quote}
% |\buildname|\marg{namespace}\marg{name}
% \end{quote}
% where \meta{name} is a harmless character sequence and
% \meta{namespace} is a harmless character sequence or the token
% |\NoValue|. If \meta{namespace} is |\NoValue| then the default
% namespace is used. |\buildname| does all processing at expand-time.
%
% If the namespace is the empty string (the global namespace) then
% \meta{name} is returned without a `|::|' prefix; this is for the
% sake of compatibility with pre-v\,8 \Tcl s.
% \begin{macrocode}
\newcommand\buildname[2]{%
\ifx \NoValue#1\@empty
\ifx \@empty\TD@namespace \else
\TD@namespace\namespaceseparator
\fi
\else
\ifx $#1$\else #1\namespaceseparator \fi
\fi
#2%
}
% \end{macrocode}
% The |\namespaceseparator| command typesets a namespace separator. By
% default it is two colons separated by a penalty, but the user may
% redefine it.
% \begin{macrocode}
\DeclareRobustCommand\namespaceseparator{:\penalty\hyphenpenalty:}
% \end{macrocode}
% \end{macro}\end{macro}
%
% \begin{macro}{\namespacephrase}
% The |\namespacephrase| macro stores the phrase that will be used
% for `namespace' in indices and the like. It is meant to be
% redefinable. The contents must be robust.
% \changes{v\,2.25}{2001/08/14}{Macro added. (LH)}
% \begin{macrocode}
\providecommand*\namespacephrase{name\-space}
% \end{macrocode}
% \end{macro}
%
% \begin{instance}{docindex}{index}
% \changes{v\,2.21}{2001/03/26}{Modified as per the v\,1.00 change to
% \package{docindex}. (LH)}
% The \texttt{index} instance of template type \texttt{docindex}
% (defined by the \package{docindex} package) is redefined to allow
% name and namespace levels to be joined.
% \begin{macrocode}
\@ifpackagewith{docidx2e}{usedocindexps}{%
\@namedef{TP@I{}{docindex}{index}}#1#2{%
\begingroup
\@letinstance\DI@indexitem@i{indexitem}{fixed-r1a}%
\@letinstance\DI@indexitem@ii{indexitem}{aloneaccept2}%
\@letinstance\DI@indexitem@iii{indexitem}{fixed3}%
\columnsep=10pt%
\parskip=0pt plus 1pt%
\def\DI@letter@skip{10pt plus 2pt minus 3pt}%
\def\DI@letter@format##1{%
\par
\hb@xt@\hsize{\hfil\textbf{##1}\hfil}%
\nopagebreak
}%
\def\+{-}%
\IfFileExists{\jobname.ind}{%
\ifnum \c@IndexColumns>\@ne
\begin{multicols}{\c@IndexColumns}[#1][\IndexMin]%
\else
\enough@room{\IndexMin}%
#1\par
\addvspace\multicolsep
\fi
\pagestyle{docindex}%
\small
\@nobreakfalse
\DI@ind@setup
\input{\jobname.ind}%
\DI@item@nojoin
\thispagestyle{docindex}
\ifnum \c@IndexColumns>\@ne
\end{multicols}%
\else
\enough@room\postmulticols
\addvspace\multicolsep
\fi
\normalsize #2\par
}{\typeout{No file \jobname.ind.}}%
\endgroup
}
}{%
\@namedef{TP@I{}{docindex}{index}}#1#2{%
\begingroup
\@letinstance\DI@indexitem@i{indexitem}{fixed-r1a}%
\@letinstance\DI@indexitem@ii{indexitem}{aloneaccept2}%
\@letinstance\DI@indexitem@iii{indexitem}{fixed3}%
\columnsep=10pt%
\parskip=0pt plus 1pt%
\def\DI@letter@skip{10pt plus 2pt minus 3pt}%
\def\DI@letter@format##1{%
\par
\hb@xt@\hsize{\hfil\textbf{##1}\hfil}%
\nopagebreak
}%
\def\+{-}%
\IfFileExists{\jobname.ind}{%
\ifnum \c@IndexColumns>\@ne
\begin{multicols}{\c@IndexColumns}[#1][\IndexMin]%
\else
\enough@room{\IndexMin}%
#1\par
\addvspace\multicolsep
\fi
\small
\@nobreakfalse
\DI@ind@setup
\input{\jobname.ind}%
\DI@item@nojoin
\ifnum \c@IndexColumns>\@ne
\end{multicols}%
\else
\enough@room\postmulticols
\addvspace\multicolsep
\fi
\normalsize #2\par
}{\typeout{No file \jobname.ind.}}%
\endgroup
}
}
% \end{macrocode}
% \end{instance}
%
%
%
% \subsection{\texttt{macro}-like environments}
%
% This subsection contains the definitions of a couple of environments
% which, like \package{doc}'s \texttt{macro}, are for marking up the
% document (what is defined and where). These environments are defined
% using the |\New|\-|Macro|\-|Environment| command of the \package{xdoc2}
% package.
% \changes{ \package{pasdoc} v\,1.40}{1999/12/05}{\texttt{macro}-like
% environments redefined. (LH)}
% \changes{v\,2.00}{2000/07/17}{\texttt{macro}-like environments are
% defined using the \cs{NewMacroEnvironment} command. The
% \cs{TD@identifier}, \cs{SpecialMainTclIndex}, and
% \cs{PrintTclName} macros were removed. (LH)}
% \changes{v\,2.12}{2000/11/26}{Using starred form of
% \cs{NewMacroEnvironment} for the \texttt{variable}, \texttt{proc},
% \texttt{arrayvar}, and \texttt{arrayentry} environments, so that
% the marginal headings can be broken. (LH)}
%
% \begin{macrocode}
\MetaNormalfont
% \end{macrocode}
% \changes{v\,2.32}{2001/11/30}{Added \cs{MetaNormalfont} as a global
% declaration and removed it from the commands that did this locally.
% There were simply so many places where it should have been
% otherwise that the resulting code would have looked rudiculous.
% (LH)}
%
% \begin{environment}{variable}
% The \texttt{variable} environment has the syntax
% \begin{quote}
% |\begin{variable}|\oarg{namespace}\marg{variable}
% \end{quote}
% It makes a marginal heading, index entry, etc.\@ for the
% \meta{variable} variable in the \meta{namespace} namespace
% (if specified, otherwise in the default namespace).
% \begin{macrocode}
\NewMacroEnvironment*{variable}%
{\XD@grab@harmless@oarg\XD@grab@harmless\relax}{2}%
{\XDParToMargin{\MacroFont\buildname{#1}{#2} \normalfont (var.)}}%
{\TD@main@index{#1}{#2}{var.}{}}%
{{\buildname{#1}{#2}}{\texttt{\buildname{#1}{#2}} variable}}
{}%
% \end{macrocode}
% \end{environment}
%
% \begin{environment}{proc}
% The \texttt{proc} environment has the syntax
% \begin{quote}
% |\begin{proc}|\oarg{namespace}\marg{proc}
% \end{quote}
% It makes a marginal heading, index entry, etc.\@ for the \meta{proc}
% procedure in the \meta{namespace} namespace (if specified, otherwise
% in the default namespace).
% \begin{macrocode}
\NewMacroEnvironment*{proc}%
{\XD@grab@harmless@oarg\XD@grab@harmless\relax}{2}%
{\XDParToMargin{\MacroFont\buildname{#1}{#2} \normalfont (proc)}}%
{\TD@main@index{#1}{#2}{proc}{}}%
{{\buildname{#1}{#2}}{\texttt{\buildname{#1}{#2}} proc}}
{}%
% \end{macrocode}
% \end{environment}
%
% \begin{environment}{arrayvar}
% \changes{v\,2.00}{2000/07/17}{Environment added. (LH)}
% \changes{v\,2.12}{2000/11/09}{Made last argument optional. (LH)}
% \changes{v\,2.13}{2000/12/15}{Not using \cs{meta} for the
% \meta{index-des} argument. (LH)}
% \begin{environment}{arrayentry}
% \changes{v\,2.02}{2000/08/21}{Environment added. (LH)}
% The \texttt{arrayvar} and \texttt{arrayentry} environments are for
% array variables; \texttt{arrayvar} is for the array as a whole,
% whereas \texttt{arrayentry} is for individual entries in the array
% that (for some reason) are important enough to warrant special
% attention. The respective syntaxes are
% \begin{quote}
% |\begin{arrayvar}|\oarg{namespace}\marg{array}\oarg{index-des}\\
% |\begin{arrayentry}|\oarg{namespace}\marg{array}\marg{entry}
% \end{quote}
% Here \meta{array} is the name of an array variable, \meta{entry} is
% the name of an entry, and \meta{index-des} is a short piece of text
% (seldom more than one word) which describes what is used as indices
% into that array.
% \changes{v\,2.14}{2001/01/12}{Corrected grabbing of the third
% argument of the \texttt{arrayvar} environment: it should be an
% oarg, not a harmless oarg. (LH)}
% \begin{macrocode}
\NewMacroEnvironment*{arrayvar}{%
\XD@grab@harmless@oarg\XD@grab@harmless\relax\XD@grab@oarg
}{3}%
{\XDParToMargin{\MacroFont \buildname{#1}{#2}%
\ifx \NoValue#3%
\space\normalfont (array)%
\else
\penalty\hyphenpenalty(\mbox{\meta@font@select#3})%
\fi
}}%
{\TD@main@index{#1}{#2}{array}{%
\ifx \NoValue#3\@empty\else
\LevelSorted{#3}{\protect\mbox{\textit{#3}} entries}%
\fi
}}%
{{\buildname{#1}{#2}}{\texttt{\buildname{#1}{#2}} array}}
{}%
% \end{macrocode}
% \begin{macrocode}
\NewMacroEnvironment*{arrayentry}{%
\XD@grab@harmless@oarg\XD@grab@harmless\relax\XD@grab@harmless\relax
}{3}%
{\XDParToMargin{%
\MacroFont \buildname{#1}{#2}%
\penalty\hyphenpenalty(#3)%
}}%
{\TD@main@index{#1}{#2}{array}{\LevelSorted{#3}{\texttt{#3}}}}%
{{\buildname{#1}{#2}(#3)}{\texttt{\buildname{#1}{#2}(#3)}}}
{}%
% \end{macrocode}
% \end{environment}\end{environment}
%
% \begin{macro}{\describestring}
% \changes{v\,2.01}{2000/08/12}{Command added. (LH)}
% \changes{v\,2.12}{2000/11/21}{Optional argument added. (LH)}
% \changes{v\,2.13}{2000/12/16}{Second optional argument added. (LH)}
% \changes{v\,2.20}{2001/03/03}{Putting namespace on the second
% level. (LH)}
% The |\describestring| command has the syntax
% \begin{quote}
% |\describestring|\oarg{type}\oarg{namespace}\marg{string}
% \end{quote}
% It is similar to \package{xdoc}'s own |\describe|\-|cs|\-|family|
% command, but it has a couple of extra features which are connected
% to the optional arguments and it doesn't prepend a backslash to
% the name of the thing being described. When the \meta{type} argument
% is present, it is taken as a ``type declaration'' of the
% \meta{string} being described and `(\meta{type})' will be put as a
% suffix to the \meta{string}. When the \oarg{namespace} argument is
% present it means that the string being described is really the name
% of something (e.g.\ a procedure or variable) which exists in a
% namespace, and the complete name is formed by passing both
% \meta{namespace} and \meta{string} as arguments to |\buildname|.
% Note that an empty \meta{namespace} argument is not quite the
% same thing as no \meta{namespace} argument.
% \begin{macrocode}
\NewDescribeCommand{\describestring}{%
\XD@grab@oarg\XD@grab@harmless@oarg\XD@grab@harmless{}%
}{3}{%
\GenericDescribePrint{%
\MacroFont
\ifx \NoValue#2\@empty
#3%
\else
\buildname{#2}{#3}%
\fi
\ifx \NoValue#1\@empty \else\ \normalfont(#1)\fi
}%
\begingroup
\def\meta##1{(##1)}%
\unrestored@protected@xdef\@gtempa{#3}%
\endgroup
\IndexEntry{%
\ifx \NoValue#2\@empty
\LevelSorted{\@gtempa}{%
\texttt{#3}%
\ifx \NoValue#1\@empty \else\space(#1)\fi
}%
\else
\LevelSorted{\@gtempa}{%
\texttt{#3}\ifx \NoValue#1\@empty \else\space(#1)\fi
}%
\ifx $#2$%
\LevelSorted{ }{global \namespacephrase}%
\else
\LevelSorted{#2}{\texttt{#2} \namespacephrase}%
\fi
\fi
}{usage}{\thepage}%
}
% \end{macrocode}
% \end{macro}
%
% \begin{macro}{\describeopt}
% \changes{v\,2.24}{2001/08/11}{Command added. (LH)}
% \changes{v\,2.31}{2001/11/03}{Star-form added. (LH)}
% The |\describeopt| macro is a |\describe|\dots\ command for options
% of commands. The syntax is
% \begin{quote}
% |\describeopt|*\oarg{namespace}\marg{command}\oarg{type}^^A
% \marg{option}
% \end{quote}
% where \meta{command} is the name of the command and \meta{option} is
% the name of the option. \meta{namespace} is the namespace of the
% \meta{command}, if that is not the global namespace. \meta{type} is
% the type description for the command, if different from `proc'.
% The |*| form of this command does not print a marginal note: it
% only makes an index entry.
% \begin{macrocode}
\NewDescribeCommand{\describeopt}{%
\XD@grab@sarg{*}\XD@grab@harmless@oarg\XD@grab@harmless{}%
\XD@grab@oarg\XD@grab@harmless{}%
}{5}{%
\ifx \BooleanFalse#1%
\GenericDescribePrint{%
\MacroFont #5\ \normalfont option%
}%
\fi
\begingroup
\def\meta##1{(##1)}%
\unrestored@protected@xdef\@gtempa{#3}%
\endgroup
\IndexEntry{%
\LevelSorted{\@gtempa}{%
\texttt{#3} (\ifx \NoValue#4\@empty proc\else #4\fi)%
}%
\ifx \NoValue#2\@empty
\LevelSorted{ }{global \namespacephrase}%
\else
\LevelSorted{#2}{\texttt{#2} \namespacephrase}%
\fi
\LevelSorted{#5}{\texttt{#5} option}%
}{usage}{\thepage}%
}
% \end{macrocode}
% \end{macro}
%
% \begin{macro}{\TD@main@index}
% \changes{v\,2.10}{2000/10/14}{Generalized macro. (LH)}
% \changes{v\,2.12}{2000/11/25}{Sorting by base name first, not by
% namespace first. (LH)}
% \changes{v\,2.20}{2001/03/03}{Putting namespace on the second
% level. (LH)}
% The |\TD@main@index| macro has the syntax
% \begin{quote}
% |\TD@main@index|\marg{namespace}\marg{name}\marg{type}^^A
% \marg{extra}
% \end{quote}
% It makes a \texttt{main} index entry. The top level of this entry
% is like an entry for the \meta{name} object of type \meta{type} in
% the \meta{namespace} namespace. The entry may have one or more
% sublevels\footnote{But no more than one extra level is recommended,
% as \package{makeindex} can handle at most three levels.} specified
% in the \meta{extra}; these are then made using explicit
% |\LevelSame| or |\LevelSorted| commands. \meta{name} and
% \meta{namespace} are as for the |\buildname| macro, whereas
% \meta{type} is a description, e.g. `\texttt{proc}', that gets
% appended to the printed text of the entry's top level.
% \begin{macrocode}
\def\TD@main@index#1#2#3#4{%
\XDMainIndex{%
\LevelSorted{#2}{\texttt{#2} (#3)}%
\ifx \NoValue#1\@empty
\ifx \@empty\TD@namespace
\LevelSorted{ }{global \namespacephrase}%
\else
\LevelSorted{\TD@namespace}%
{\texttt{\TD@namespace} \namespacephrase}%
\fi
\else
\ifx $#1$%
\LevelSorted{ }{global \namespacephrase}%
\else
\LevelSorted{#1}{\texttt{#1} \namespacephrase}%
\fi
\fi
#4%
}%
}
% \end{macrocode}
% \end{macro}
%
% \begin{environment}{Cfunction}
% \changes{v\,2.40}{2003/04/05}{Environment added. (LH)}
% \begin{environment}{Cvariable}
% \changes{v\,2.40}{2003/04/05}{Environment added. (LH)}
% \begin{environment}{Ctype}
% \changes{v\,2.40}{2003/04/05}{Environment added. (LH)}
% The \texttt{Cfunction}, \texttt{Cvariable}, and \texttt{Ctype}
% environments are for making a marginal heading, index entry, etc.\@
% for a C function, variable, or type respectively. They all take as
% their single argument the name of the thing that will be defined.
% \begin{macrocode}
\NewMacroEnvironment*{Cfunction}{\XD@grab@harmless\relax}{1}%
{\XDParToMargin{\small\Cfunctionidentifier{#1} (C~function)}}%
{\XDMainIndex{%
\LevelSorted{#1}{\Cfunctionidentifier{#1} (C~function)}%
}}%
{{#1}{\Cfunctionidentifier{#1} C~function}}
{}%
\NewMacroEnvironment*{Cvariable}{\XD@grab@harmless\relax}{1}%
{\XDParToMargin{\small\Cvariableidentifier{#1} (C~variable)}}%
{\XDMainIndex{%
\LevelSorted{#1}{\Cvariableidentifier{#1} (C~variable)}%
}}%
{{#1}{\Cvariableidentifier{#1} C~variable}}
{}%
\NewMacroEnvironment*{Ctype}{\XD@grab@harmless\relax}{1}%
{\XDParToMargin{\small\Ctypeidentifier{#1} (C~type)}}%
{\XDMainIndex{%
\LevelSorted{#1}{\Ctypeidentifier{#1} (C~type)}%
}}%
{{#1}{\Ctypeidentifier{#1} C~type}}
{}%
% \end{macrocode}
% \end{environment}\end{environment}\end{environment}
%
% \begin{macro}{\Cfunctionidentifier}
% \changes{v\,2.40}{2003/04/05}{Macro added. (LH)}
% \begin{macro}{\Cvariableidentifier}
% \changes{v\,2.40}{2003/04/05}{Macro added. (LH)}
% \begin{macro}{\Ctypeidentifier}
% \changes{v\,2.40}{2003/04/05}{Macro added. (LH)}
% These macros are used by \texttt{Cfunction}, \texttt{Cvariable}, and
% \texttt{Ctype} to format the identifiers of C functions, variables,
% and types respectively; the syntax is e.g.
% \begin{quote}
% |\Cfunctionidentifier|\marg{function name}
% \end{quote}
% They default to typeset them in an italic font---this is what CWEB
% uses---but it is meant to be configurable.
% \begin{macrocode}
\newcommand*{\Cfunctionidentifier}[1]{\textit{#1}}
\newcommand*{\Cvariableidentifier}[1]{\textit{#1}}
\newcommand*{\Ctypeidentifier}[1]{\textit{#1}}
% \end{macrocode}
% \end{macro}\end{macro}\end{macro}
%
%
% \subsection{Regular expressions and friends}
%
% I have noticed that when one is describing the syntax of some \Tcl\
% command\slash procedure, one often ends up writing argument lists with
% lots of optional, repeated, or variant parts. In general one can of
% course specify these using BNF syntax diagrams, but whereas occurrences
% of the three syntactic constructions mentioned above are plenty,
% there are much fewer occurrences of for example nesting. Since
% furthermore even the simplest optional or repeated element
% requires an extra syntactic term to be defined, it becomes interesting
% to search for an alternative way of specifying these constructions.
% One that is close at hand is that of regular expressions, as that is
% anyway a part of the \Tcl\ language.
%
% \begin{macro}{\regstar}
% \begin{macro}{\regplus}
% \begin{macro}{\regopt}
% These commands are used to modify the meaning of another syntactic
% element. |\regstar| puts a star $^*$ on something, meaning it is
% repeated zero or more times. |\regplus| puts a plus $^+$ on
% something, meaning it is repeated one or more times. |\regopt| puts
% a question mark $^?$ on something, meaning it is optional. Normally
% these commands are simply put right after what they should modify,
% but if the ``something'' is a regular expression block, i.e., a
% \texttt{regblock} environment, then these things should be put as
% the \meta{modifier} in
% \begin{quote}
% |\begin{regblock}|\oarg{modifier}\\
% | |\textellipsis\\
% |\end{regblock}|
% \end{quote}
% In particular, they should \emph{not} be put after the
% |\end{regblock}|, as that may impair the placement of the exponent.
% \begin{macrocode}
\newcommand\regstar{\ensuremath{^*}}
\newcommand\regplus{\ensuremath{^+}}
\newcommand\regopt{\ensuremath{^?}}
% \end{macrocode}
% \end{macro}\end{macro}\end{macro}
%
% \begin{macro}{\TD@delim@size}
% \begin{macro}{\TD@delim@size@G}
% These control sequences are used in determining what size the
% delimiters (parentheses) around a regular expression block should
% have. The values specify what should be the least size of a
% delimiter that bracket a piece of text, and are interpreted as
% follows: \(0={}\)normal size, \(1={}\)|\big| size, \(2={}\)|\Big|
% size, \(3={}\)|\bigg| size, and $4$ and above is |\Bigg| size.
%
% |\TD@delim@size@G| is a macro which is always set globally and
% which keeps track of the delimiter size at the innermost level.
% |\TD@delim@size| is a |\count| register that is assigned locally
% and is used for keeping track of delimiter sizes at all other
% levels. Commands like |\word|, which contribute a delimiter to some
% piece of text, should set |\TD@delim@size@G|.
% \begin{macrocode}
\newcount\TD@delim@size
\gdef\TD@delim@size@G{0}
% \end{macrocode}
% \end{macro}\end{macro}
%
% \begin{environment}{regblock}
% The \texttt{regblock} environment embarks upon a rather tricky scheme
% of boxing and unboxing to determine the proper delimiter size. The
% basic idea is that the contents of the environment are first typeset
% in an \texttt{lrbox} environment, and |\TD@delim@size@G| is set to
% its proper value. Then the left delimiter is set at this size, the
% saved box is unboxed, and finally the right delimiter is set. All
% through the process, the value of |\TD@delim@size@G| at the
% beginning of the \texttt{regblock} environment has been saved in
% |\TD@delim@size| and at the end |\TD@delim@size@G| is set to the
% maximum of its old value and one plus its value inside the
% \texttt{regblock} environment. This makes it possible to nest
% \texttt{regblock} environments.
%
% It is not quite that easy, though. A regular expression can consist
% of several branches, which must be separated by vertical bars.
% These bars should be the same size as the delimiters around the
% regular expression, and hence the bars cannot be typeset before
% every branch has been typeset. Therefore the \texttt{regblock}
% environment starts an extra horizontal list (the |\hbox|) for the
% sole purpose of storing the branches that have already been typeset
% (and \emph{only} these). Each branch is put on the list as a box.
% When all branches have been typeset, they are first removed from the
% list (by |\TD@join@branches|), and then they are inserted again, but
% this time they are unboxed and separated by suitable delimiters.
% Finally the |\hbox| is ended and almost immediately unboxed.
%
% \changes{v\,2.40}{2003/04/05}{Renamed the \texttt{regexp}
% environment to \texttt{regblock}. A \texttt{regexp} environment
% is provided for compatibility. (LH)}
% \begin{macrocode}
\newenvironment{regblock}[1][]{%
\leavevmode
\def\TD@modifier{#1}%
\TD@delim@size=\TD@delim@size@G
\let\regalt=\TD@regalt
\setbox\z@=\hbox\bgroup
\TD@delim@size=\z@
\begin{lrbox}{\z@}%
\gdef\TD@delim@size@G{0}%
\ignorespaces
}{%
\end{lrbox}%
\ifnum \TD@delim@size@G>\TD@delim@size
\TD@delim@size=\TD@delim@size@G\relax
\fi
{\TD@join@branches}%
\unhbox\z@
$\TD@size@delimiter)\TD@modifier$%
\advance \TD@delim@size \@ne
\xdef\TD@delim@size@G{\the\TD@delim@size}%
\egroup
\ifnum \TD@delim@size@G<\TD@delim@size
\xdef\TD@delim@size@G{\the\TD@delim@size}%
\fi
\unhbox\z@
}
% \end{macrocode}
% \end{environment}
%
% \begin{environment}{regexp}
% This is an alias to the \texttt{regblock} environment, provided for
% compatibility.
% \begin{macrocode}
\newenvironment{regexp}{\regblock}{\endregblock}
% \end{macrocode}
% \end{environment}
%
% \begin{macro}{\TD@join@branches}
% \changes{v\,2.12}{2000/11/13}{Made penalty for linebreak after
% \cs{regalt} depend on the delimiter size. (LH)}
% The |\TD@join@branches| macro uses |\lastbox| to retrieve all boxes
% in the current horizontal list, and unboxes them. Between the
% unboxed boxes it puts a |\vert| delimiter, and before the first it
% puts a left parenthesis. Both will be the current delimiter size.
% \begin{macrocode}
\def\TD@join@branches{%
\setbox\z@=\lastbox
\ifvoid\z@
$\TD@size@delimiter($%
\else
{\TD@join@branches}%
\unhbox\z@
~$\TD@size@delimiter\vert$%
\count@=\TD@delim@size
\advance \count@ \@ne
\multiply \count@ -\@lowpenalty
\penalty\count@\ %
\fi
}
% \end{macrocode}
% \end{macro}
%
% \begin{macro}{\regalt}
% \begin{macro}{\TD@regalt}
% The |\regalt| command is used in a \texttt{regblock} environment
% to separate two branches. It ends an \texttt{lrbox} environment,
% puts the box collected in the surrounding horizontal list (which
% was begun by the |\hbox| in |\lrbox|), and begins a new
% \texttt{lrbox} environment. As this would not make sense outside a
% \texttt{regblock} environment however, the |\regalt| command does by
% default only produce an error message. The ``real'' definition is
% kept in |\TD@regalt| and the \texttt{regblock} environment locally
% redefines |\regalt|.
% \begin{macrocode}
\newcommand\regalt{%
\PackageError{tclldoc}{%
Lonely \protect\regalt--perhaps a missing regblock environment%
}\@eha
}
\def\TD@regalt{%
\end{lrbox}%
\box\z@
\ifnum \TD@delim@size@G>\TD@delim@size
\TD@delim@size=\TD@delim@size@G\relax
\fi
\begin{lrbox}{\z@}%
\gdef\TD@delim@size@G{0}%
}
% \end{macrocode}
% \end{macro}\end{macro}
%
% \begin{macro}{\TD@size@delimiter}
% This macro expands to nothing, |\big|, |\Big|, |\bigg|, or |\Bigg|
% depending on the value of |\TD@delim@size|.
% \begin{macrocode}
\def\TD@size@delimiter{%
\ifcase\TD@delim@size
\or \expandafter\big \or \expandafter\Big \or
\expandafter\bigg \else \expandafter\Bigg
\fi
}
% \end{macrocode}
% \end{macro}
%
% \begin{macro}{\word}
% \changes{v\,2.10}{2000/10/13}{Reimplemented like the new \cs{meta}.
% (LH)}
% \changes{v\,2.12}{2000/11/21}{Using braces instead of angle brackets.
% (LH)}
% This command works like |\meta|, but the argument is enclosed in
% braces rather than angle brackets and it communicates with the
% \texttt{regblock} environment (the braces count as a smallest-size
% delimiter). Something marked out using |\word| rather than |\meta|
% is meant to always denote a separate \Tcl\ word.
% \begin{macrocode}
\DeclareRobustCommand\word[1]{%
\leavevmode
\ifmmode \expandafter \nfss@text \fi
{%
$\{$%
\meta@font@select
\edef\meta@hyphen@restore
{\hyphenchar\the\font\the\hyphenchar\font}%
\hyphenchar\font\m@ne
\language\l@nohyphenation
#1\/%
\meta@hyphen@restore
$\}$%
}%
\ifnum \TD@delim@size@G<\@ne \gdef\TD@delim@size@G{1}\fi
}
% \end{macrocode}
% \end{macro}
%
%
% \subsection{Logos}
%
% \begin{macro}{\Tcllogo}
% \changes{v\,2.03}{2000/09/15}{Macro added. (LH)}
% The |\Tcllogo| command makes a \Tcl\ ``logo'', which simply consists
% of the three letters `T', `c', and `l' in the current font. The only
% special thing about it is that there is a kern between the `T' and
% the `c', since most fonts doesn't have that kern. No work has been
% put into tuning the kern, but the size is about right. As a
% comparison the kern between `T' and `o' in \texttt{cmr10} is
% $-0.083334\,\mathrm{em}$.
% \begin{macrocode}
\DeclareRobustCommand\Tcllogo{T\kern-0.1em cl}
%</pkg>
% \end{macrocode}
% \end{macro}
%
% \changes{v\,2.03}{2000/09/15}{\cs{DescribeResource} command removed.
% Just in case anyone is interested in using it, I have included it as
% a code example in the source for the \package{xdoc} package. (LH)}
%
%
% \section{The \package{tclldoc} class}
%
% \subsection{Option declarations and processing}
%
% \begin{option}{a5paper}
% \begin{option}{olddocinclude}
% The \texttt{a5paper} (to the \package{article} document class) and
% the \texttt{olddocinclude} (to the \package{xdoc} package) are not
% supported for the \package{tclldoc} class.
% \begin{macrocode}
%<*class>
\DeclareOption{a5paper}{%
\ClassError{tclldoc}{The a5paper option is not supported}\@eha
}
\DeclareOption{olddocinclude}{%
\ClassError{tclldoc}{The olddocinclude option is not supported}\@eha
}
% \end{macrocode}
% \end{option}\end{option}
%
% \begin{option}{notrawchar}
% The \texttt{notrawchar} option is passed on to the
% \package{tclldoc} package (which will pass it on to the
% \package{xdoc} package).
% \changes{v\,2.40}{2003/04/05}{Added processing of this option; it
% wouldn't take if it was simply given as a global option. (LH)}
% \begin{macrocode}
\DeclareOption{notrawchar}{%
\PassOptionsToPackage{notrawchar}{tclldoc}%
}
% \end{macrocode}
% \end{option}
%
%
% \begin{macrocode}
\DeclareOption*{%
\PassOptionsToClass{\CurrentOption}{article}}
% \end{macrocode}
%
%
%
% Input a local configuration file, if it exists.
% \begin{macrocode}
\InputIfFileExists{tclldoc.cfg}
{\typeout{*************************************^^J%
* Local config file tclldoc.cfg used^^J%
*************************************}}
{}
% \end{macrocode}
%
%
%
% \begin{macrocode}
\ProcessOptions\relax
% \end{macrocode}
%
% \subsection{Loading the base class}
%
% \begin{macrocode}
\LoadClass{article}
% \end{macrocode}
%
%
% \subsection{The layout}
%
% Increase the text width slightly so that with the standard fonts
% 72 columns of code may appear in a |macrocode| environment.
% \changes{ \package{ltxdoc} v\,2.0c}{1994/03/15}{Set \cs{textwidth}.}
% \begin{macrocode}
\setlength{\textwidth}{355pt}
% \end{macrocode}
%
% Increase the marginpar width slightly, for long command names.
% And increase the left margin by a similar amount
% \changes{ \package{ltxdoc} v\,2.0l}
% {1994/05/25}{Increase \cs{marginparwidth}}
% \changes{ \package{ltxdoc} v\,2.0q}{1995/11/28}
% {Increase \cs{marginparwidth} and page margin.}
% \begin{macrocode}
\addtolength\oddsidemargin{20pt}
\addtolength\evensidemargin{20pt}
% \end{macrocode}
%
%
% Change some defaults for list formatting. In particular, continue to
% indent paragraphs in |\list| environments and don't put extra space
% between them.
% \changes{ \package{pasdoc} v\,1.30}{1999/08/31}{Modifying the list
% formatting. (LH)}
%
% \begin{macro}{\TD@list}
% \begin{macro}{\@listi}
% \begin{macro}{\@listI}
% \begin{macro}{\@listii}
% \begin{macro}{\@listiii}
% \begin{macro}{\@listiv}
% \begin{macro}{\@listv}
% \begin{macro}{\@listvi}
% \begin{macrocode}
\def\@tempa#1{%
\expandafter\def \expandafter#1\expandafter{#1\TD@list}%
}
\def\TD@list{%
\advance \itemsep \parsep
\parsep=\z@\@plus\p@\relax
\advance \itemsep -\parsep
\listparindent=1em\relax
}
\@tempa\@listi
\let\@listI\@listi
\@tempa\@listii
\@tempa\@listiii
\@tempa\@listiv
\@tempa\@listv
\@tempa\@listvi
% \end{macrocode}
% \end{macro}\end{macro}\end{macro}\end{macro}\end{macro}\end{macro}
% \end{macro}\end{macro}
%
%
% \subsection{Loading the \package{tclldoc} package}
%
% \begin{macrocode}
\RequirePackage[dolayout]{tclldoc}
% \end{macrocode}
%
% \begingroup
% Make \verb+|+ be a `short verb' character, but not in the document
% preamble, where an active character may interfere with packages that
% are loaded.
% \DoNotIndex{\|}
% \begin{macrocode}
\AtBeginDocument{\MakeShortTclverb{\|}}
% \end{macrocode}
% \endgroup
%
% As \texttt{.dtx} documents tend to have a lot of monospaced material,
% set up some \texttt{cmtt} substitutions to occur silently.
% \changes{ \package{ltxdoc} v\,2.0p}{1995/11/02}{Add font substitutions}
% \begin{macrocode}
\DeclareFontShape{OT1}{cmtt}{bx}{n}{<-> ssub * cmtt/m/n}{}
\DeclareFontFamily{OMS}{cmtt}{\skewchar\font'60}
\DeclareFontShape{OMS}{cmtt}{m}{n}{<-> ssub * cmsy/m/n}{}
\DeclareFontShape{OMS}{cmtt}{bx}{n}{<-> ssub * cmsy/b/n}{}
% \end{macrocode}
% This substitution is in the standard fd file, but not silent.
% \begin{macrocode}
\DeclareFontShape{OT1}{cmss}{m}{it}{<->ssub*cmss/m/sl}{}
% \end{macrocode}
%
%
% \subsection{More layout}
%
% \changes{v\,2.12}{2000/11/26}{Setting \cs{marginparwidth} so that the
% outer edge of the marginpar is $1\,\mathrm{cm}$ from the paper
% edge. (LH)}
% \changes{v\,2.12}{2000/11/27}{Rearranged code so that
% the \cs{marginparwidth} is set after \package{tclldoc} has been
% loaded; otherwise \package{doc} might overwrite it! (LH)}
% \begin{macrocode}
\setlength\marginparwidth{\evensidemargin}
\addtolength\marginparwidth{1in}
\addtolength\marginparwidth{-\marginparsep}
\addtolength\marginparwidth{-1cm}
% \end{macrocode}
%
%
% \begin{macrocode}
\CodelineNumbered
\DisableCrossrefs
% \end{macrocode}
%
% \begin{macrocode}
\setcounter{StandardModuleDepth}{1}
%</class>
% \end{macrocode}
%
%
% \section{The \package{tcldoc} compatibility package and class}
%
% \changes{v\,2.30}{2001/09/02}{Renamed the package and class to
% \package{tclldoc} to prevent confusion with the \Tcl Doc Perl
% script. Added files \texttt{tcldoc.sty} and \texttt{tcldoc.cls}
% for compatibility. Moved code related to the \texttt{macinputenc}
% option to \texttt{tcldoc.cls}; this option is now unsupported.
% (LH)}
%
% \subsection{Pre-options definitions}
%
% \begin{macro}{\if@rtkinenc@}
% Switch. True iff the \package{rtkinenc} package should be loaded.
% It starts out false.
% \begin{macrocode}
%<*compclass>
\newif\if@rtkinenc@
% \end{macrocode}
% \end{macro}
%
%
% \subsection{Option declarations and processing}
%
% \begin{option}{macinputenc}
% The \texttt{macinputenc} option enables some special code for
% supporting the typesetting of \Tcl\ code written for the Macintosh
% input encoding (cf.\ the \texttt{applemac} option to the
% \package{inputenc} package).
% \begin{macrocode}
\DeclareOption{macinputenc}{%
\PassOptionsToPackage{applemac}{rtkinenc}%
\@rtkinenc@true
}
% \end{macrocode}
% \end{option}
%
%
% \begin{macrocode}
\DeclareOption*{%
\PassOptionsToClass{\CurrentOption}{tclldoc}}
% \end{macrocode}
%
%
%
% Input a local configuration file, if it exists.
% \begin{macrocode}
\InputIfFileExists{tcldoc.cfg}
{\typeout{*************************************^^J%
* Local config file tcldoc.cfg used^^J%
*************************************}}
{}
% \end{macrocode}
%
%
%
% \begin{macrocode}
\ProcessOptions\relax
% \end{macrocode}
%
% \subsection{Loading the base package or class}
%
% \begin{macrocode}
\LoadClass{tclldoc}
%</compclass>
%<comppkg>\LoadPackageWithOptions{tclldoc}
% \end{macrocode}
%
%
% \subsection{Input encoding and font representations}
%
% There are some rather tricky issues concerning interpretation of the
% input in \Tcl\ code. If the |\MacroFont| font contains a character that
% correctly represents a character in the input, then that character is
% obviously what one would like to appear in the typeset output. If no
% font in the entire \TeX\ system contains a character that looks like
% the one in the input, then just as obviously must one examine whether
% there exists some other method of specifying that character. In the
% case of \Tcl\ code, a suitable method seems to be to typeset the
% equivalent |\x|\meta{dd} escape sequence instead. It is for the
% purpose of being able to detect these cases that the
% \package{tcldoc} class (if the user so requests) loads the
% \package{rtkinenc} package, rather than the standard \LaTeX\
% \package{inputenc} package.
%
% The problem of determining for each input character which method of
% representation that should be used---single character or escape
% sequence---does not have a unique solution. Therefore the default
% set-up is to not define any special fakes at all, which means the
% escape sequence method will be chosen in the doubtful cases. There are
% also a couple of options for the \package{tcldoc} class that define
% all (or nearly all) possible fakes---the difference between them lie
% in which fonts are assumed to be present in the system. A user who
% does not like any of these predefined options also has the option of
% putting the corresponding |\Declare|\-|Text|\-|Default|\textellipsis
% \ commands directly into the \texttt{tcldoc.cfg} file.
%
% \begin{macro}{\InputModeCode}
% \begin{macro}{\SetUnavailableAction}
% \begin{macro}{\DeclareInputMath}
% \begin{macro}{\RIE@undefined}
% The \package{tclldoc} package calls some commands defined by the
% \package{rtkinenc} package, but since \package{rtkinenc} is not
% required by the \package{tclldoc} package, the \package{tclldoc}
% package must provide its own definitions, even though these can be
% no-ops. Definition of these commands is delayed until
% |\begin|\nolinebreak[1]|{document}| to avoid problems if the user
% loads the \package{rtkinenc} package after the \package{tclldoc}
% package.
% \begin{macrocode}
%<*pkg>
\AtBeginDocument{%
\providecommand*\InputModeCode{}%
\providecommand*\SetUnavailableAction[1]{}%
\providecommand*\DeclareInputMath[2]{}%
\providecommand*\RIE@undefined[1]{\@inpenc@undefined}%
}
%</pkg>
% \end{macrocode}
% \end{macro}\end{macro}\end{macro}\end{macro}
%
% \begin{macrocode}
%<*compclass>
\if@rtkinenc@
\RequirePackage{rtkinenc}
% \end{macrocode}
%
% \begin{macro}{\.}
% \begin{macro}{\H}
% \begin{macro}{\textemdash}
% \begin{macro}{\textendash}
% \begin{macro}{\textexclamdown}
% \begin{macro}{\textquestiondown}
% \begin{macro}{\textquotedblleft}
% The following text commands are defined by \texttt{OT1enc.def}, but
% their definitions will not work for \texttt{OT1}-encoded typewriter
% fonts. Therefore they are redefined, either so that they work
% anyway (|\textexclamdown| and |\textquestiondown|) or so that they
% call |\TextSymbolUnavailable| when \package{rtkinenc} is in code
% mode. This should make things work for the best in most
% reasonable situations.
%
% It could be noted here that the definitions below for |\.| and |\H|
% does not quite behave as if the commands had not been defined at
% all (I'm assuming that \package{rtkinenc} is in code mode here); an
% undefined command would leave its argument to be typeset, but the
% definitions below will swallow it in those cases.
% \begin{macrocode}
\DeclareTextCommand{\.}{OT1}[1]{%
\IfInputModeCode{\TextSymbolUnavailable\.}%
{\add@accent{95}{#1}}%
}
\DeclareTextCommand{\H}{OT1}[1]{%
\IfInputModeCode{\TextSymbolUnavailable\H}%
{\add@accent{125}{#1}}%
}
\DeclareTextCommand{\textemdash}{OT1}{%
\IfInputModeCode{\TextSymbolUnavailable\textemdash}{\char 124 }%
}
\DeclareTextCommand{\textendash}{OT1}{%
\IfInputModeCode{\TextSymbolUnavailable\textendash}{\char 123 }%
}
\DeclareTextCommand{\textexclamdown}{OT1}{!`}
\DeclareTextCommand{\textquestiondown}{OT1}{?`}
\DeclareTextCommand{\textquotedblleft}{OT1}{%
\IfInputModeCode{\TextSymbolUnavailable\textquotedblleft}%
{\char 92 }%
}
% \end{macrocode}
% \end{macro}\end{macro}\end{macro}\end{macro}\end{macro}\end{macro}
% \end{macro}
%
% Several of the definitions in \texttt{applemac.def} can be improved.
% Most of it is about seeing to that all characters have a proper text
% mode definition, so that their unavailability can be trapped.
%
% \begin{macrocode}
\def\@tempa{applemac}
\ifx \@tempa\RIE@encoding
\DeclareInputBoth{173}{\textneq}{\neq}
\DeclareInputBoth{176}{\textinfty}{\infty}
\DeclareInputBoth{177}{\textpm}{\pm}
\DeclareInputBoth{178}{\textleq}{\leq}
\DeclareInputBoth{179}{\textgeq}{\geq}
\DeclareInputBoth{181}{\textmu}{\mu}
\DeclareInputBoth{182}{\textpartial}{\partial}
\DeclareInputBoth{183}{\textSigma}{\Sigma}
\DeclareInputBoth{184}{\textPi}{\Pi}
\DeclareInputBoth{185}{\textpi}{\pi}
\DeclareInputBoth{186}{\textint}{\int}
\DeclareInputBoth{189}{\textohm}{\Omega}
\DeclareInputBoth{194}{\textlnot}{\lnot}
\DeclareInputBoth{195}{\textsurd}{\surd}
\DeclareInputBoth{197}{\textapprox}{\approx}
\DeclareInputBoth{198}{\textDelta}{\Delta}
\DeclareInputBoth{214}{\textdiv}{\div}
\DeclareInputBoth{215}{\textdiamond}{\diamond}
\DeclareInputBoth{218}{\textfractionsolidus}{/}
% \end{macrocode}
%
% Several of the above text commands are not defined in the \LaTeX\
% kernel. Therefore one must either define them in some encoding, or
% define a default for them, to make them proper text commands. If they
% are not then their unavailability cannot be trapped. |\textapplelogo|
% appears in \texttt{applemac.def}.
% \begin{macrocode}
\ProvideTextCommandDefault{\textneq}{%
\TextSymbolUnavailable\textneq
}
\ProvideTextCommandDefault{\textapplelogo}{%
\TextSymbolUnavailable\textapplelogo
}
\DeclareTextSymbol{\textinfty}{OMS}{49}
\DeclareTextSymbol{\textpm}{OMS}{6}
\DeclareTextSymbol{\textleq}{OMS}{20}
\DeclareTextSymbol{\textgeq}{OMS}{21}
% \DeclareTextSymbol{\textmu}{TS1}{181}
\DeclareTextSymbol{\textpartial}{OML}{64}
\DeclareTextSymbol{\textSigma}{OT1}{6}
\DeclareTextSymbol{\textPi}{OT1}{5}
\DeclareTextSymbol{\textpi}{OML}{25}
\DeclareTextSymbol{\textint}{OMS}{115}
\DeclareTextSymbol{\textohm}{OT1}{10} % Kind of \textOmega
% \DeclareTextSymbol{\textlnot}{TS1}{172}
% \DeclareTextSymbol{\textsurd}{TS1}{187}
\DeclareTextSymbol{\textapprox}{OMS}{25}
\DeclareTextSymbol{\textDelta}{OT1}{1}
% \DeclareTextSymbol{\textdiv}{TS1}{246}
\DeclareTextSymbol{\textdiamond}{OMS}{5}
% \DeclareTextSymbol{\textfractionsolidus}{TS1}{47}
\input{TS1enc.def}
\fi
\fi
% \end{macrocode}
% \begin{macrocode}
\let\if@rtkinenc@=\@undefined
\let\@rtkinenc@true=\@undefined
\let\@rtkinenc@false=\@undefined
% \end{macrocode}
% The |@rtkinenc@| switch isn't needed any more, so it might just as
% well be undefined.
%
%
% \subsection{Useful abbreviations}
% \changes{v\,2.10}{2000/10/11}{\LaTeX\ abbreviation commands from
% \package{ltxdoc} commented out. (LH)}
%
% The commands currently found in this section are inherited from the
% \package{ltxdoc} package, but as they aren't of much use with \Tcl,
% they have been commented out.
%
% |\cmd{\foo}| Prints |\foo| verbatim. It may be used inside moving
% arguments. |\cs{foo}| also prints |\foo|, for those who prefer that
% syntax. (This second form may even be used when |\foo| is |\outer|).
% \begin{macro}{\cmd}
% \begin{macro}{\cs}
% \DoNotIndex{\\}
% \begin{macrocode}
% \def\cmd#1{\cs{\expandafter\cmd@to@cs\string#1}}
% \def\cmd@to@cs#1#2{\char\number`#2\relax}
% \DeclareRobustCommand\cs[1]{\texttt{\char`\\#1}}
% \end{macrocode}
% \end{macro}
% \end{macro}
%
% \begin{macro}{\marg}
% |\marg{text}| prints \marg{text}, `mandatory argument'.
% \DoNotIndexHarmless{\PrintChar{`\{}}
% \DoNotIndexHarmless{\PrintChar{`\}}}
% \begin{macrocode}
% \providecommand\marg[1]{%
% {\ttfamily\char`\{}\meta{#1}{\ttfamily\char`\}}}
% \end{macrocode}
% \end{macro}
%
% \begin{macro}{\oarg}
% |\oarg{text}| prints \oarg{text}, `optional argument'.
% \begin{macrocode}
% \providecommand\oarg[1]{%
% {\ttfamily[}\meta{#1}{\ttfamily]}}
% \end{macrocode}
% \end{macro}
%
% \begin{macro}{\parg}
% |\parg{te,xt}| prints \parg{te,xt}, `picture mode argument'.
% \begin{macrocode}
% \providecommand\parg[1]{%
% {\ttfamily(}\meta{#1}{\ttfamily)}}
%</compclass>
% \end{macrocode}
% \end{macro}
%
%
%
%
% \changes{v\,2.00}{2000/07/17}{Section with \cs{DocInclude} removed,
% since that is defined by \package{xdoc}. (LH)}
%
% \Finale
%
\endinput