\documentstyle[11pt,reduce]{article}
\begin{document}
\title{Guide to Writing \REDUCE Reference Documents}
\date{}
\author{H. Melenk \& R. Sch\"opf}
\maketitle
\section{Introduction}
Since Version 3.5 of \REDUCE, the documentation of a package consists of two
parts, a manual part that is used to introduce the user to the background
and the functionality of the package in a textbook style, and a reference
part to be used more like an encyclopedia for retrieving isolated items or
relations\footnote{%
At present most of the packages lack a special reference document and a
plain text version of the manuals are browsed instead. One purpose of this
text is to encourage package suppliers to add a better-suited reference
material to their product}. Both parts need different formats. While the
\REDUCE manual style was fixed some time ago, the \REDUCE reference style
was developed from scratch for \REDUCE 3.5 and essentially unmodified for
\REDUCE 3.6. This paper is intended as a guide to writing a document in
the reference style.
{\bf Note}: The case of the keywords used in this document is relevant. E.g.
you must write {\bf Examples}, {\bf Operator} on one hand, but {\bf verbatim}
on the other.
\section{Media}
A reference document source is a text in \LaTeX{} language following a
specific set of rules. This source text is converted for two different
target media:
\begin{itemize}
\item printed reference manual for the package and a printed extension of
the \REDUCE reference manual
\item on-line help for the package as an extension of the \REDUCE on-line
help system.
\end{itemize}
The printed text is formatted by \LaTeX{} using the style specific file {\bf %
redref.sty}. For the on-line help system a special translation mechanism is
used that generates data structures for different target systems. A
description of the transformation process is given below. At present the
target systems are
\begin{itemize}
\item Microsoft Windows help format for MS WINHELP;
\item GNU Info format, used by GNU Info, xinfo and the \REDUCE X interface
program XR;
\item WWW (world wide web)\footnote{%
formatter contributed by A. Strotmann, Cologne}.
\end{itemize}
The reference format has been designed to serve both purposes with one
unique source. However, the restrictions are much more rigid than with
ordinary \TeX{} styles. These restrictions must be followed carefully,
otherwise the translation to help files will fail. Please note that a
successful translation by \LaTeX{} is necessary but not sufficient as a
test, since not all restrictions are controlled in the environment
established by the style file.
\section{General Structure}
The overall structure is given by the traditional \LaTeX{} hierarchy
\begin{verbatim}
\begin{document}{...}
\chapter{ }
\section{ }
\subsection{ }
\subsubsection{ }
\end{document}
\end{verbatim}
where -- of course -- some or all of the inner levels may be omitted. The
structure is automatically converted into a directory hierarchy using the
given titles. These {\bf must} be unique in the whole document (not only in
the local section).
Informal guideline: the titles should be as short as possible.
The bottom level information must be cut explicitly into pieces that are
named $nodes$ throughout this paper. There must not be any text outside a
node - any such text may either get lost or disturb the translation process.
Typical nodes are: an operator, a variable, a switch etc. In the manual form
a node is a graphically separated part of a page with an emphasized heading.
In the help form a node is a separate page. In both forms the text for a
node can (and should) contain references to other nodes. These are encoded
either as printed references (manual) or as hypertext links (help) that you
can follow by clicking the word with the mouse. In addition, there are
directories and indexes that alternatively lead the user to the desired
information, and which also use the nodes as reference points.
Informal guideline: hierarchy entries and nodes may be mixed on each level.
On the other hand such a mixture may be unpleasant for online use. Also, the
structure should not be too deep, otherwise the user must click lots of
items until he reaches the desired item.
\section{Nodes}
\subsection{Node Context}
In the source language a node is encoded as \LaTeX{} environment
\begin{verbatim}
\begin{<node-type>}[<options>]{<node-name>}
<line-1>
. . . % the node body
<line-n>
\end{<node-type>}
\end{verbatim}
The third part of the header line \verb|[<options>]| is omitted except for
very special cases.
The leading \verb|<node-type>| describes the type of information that is
given in the node. Specific for the type are the layout and especially the
heading and directory enty. \verb|<node-type>| is one of the following
\begin{itemize}
\item {\bf Introduction}: typically a piece of text used at the beginning
of a section to give general information for the whole section.
\verb|<node-name>| is a free heading, which, however, should be as short as
possible.
\item {\bf Operator}: description of an operator. \verb|<node-name>| is the
operator name. When the operator is represented by a special character
symbol, \verb|<options>| is the (internal) alphanumeric name and \{%
\verb|<node-name>|\} contains the special character equivalent. E.g.
\begin{verbatim}
\begin{Operator}[replace]{=>}
\end{verbatim}
References should then be directed towards the alphanumeric name.
\item {\bf Function}: similar to Operator.
\item {\bf Statement}: description of a syntax extension. \verb|<node-name>|
is the leading keyword or the statement.
\item {\bf Command}: description of a command type syntax extension.
\verb|<node-name>| is the leading keyword or the command.
\item {\bf Declaration}: description of a declaration type syntax
extension. \verb|<node-name>| is the leading keyword or the declaration
statement.
\item {\bf Switch}: description of a switch. \verb|<node-name>| is the
switch name. Note that the body of a Switch node must give information of
the default setting.
\item {\bf Variable}: description of a global (share) variable.
\verb|<node-name>| is the variable name.
\item {\bf Constant}: description of a Constant. \verb|<node-name>| is the
constant name.
\item {\bf Concept}: used to introduce a description of a term that does
not fit into one of the other classes, E.g. ``kernel''. \verb|<node-name>|
is the target keyword.
\end{itemize}
In all cases \verb|<node-name>| must be written exactly as used in all
references (correct case, no additional blanks etc.). For \REDUCE 3.5 and
later versions, this is lower case for most items. Special characters
other than underscore should be avoided whenever possible because these
are often not accepted by the target systems.
\subsection{Node Structure}
Inside a node the default state is ``normal text". Special environments (%
\verb|\begin{..}| -- \verb|\end{..}| contexts) are supported that allow you
to format text. These are
\begin{itemize}
\item {\bf Syntax}: description of the syntax for an operator, statement or
command. This is translated into a fixed font. For special advice see below.
\item {\bf Examples}: description of prototypical applications, including
input and output. Here special formatting is performed - see below.
\item {\bf verbatim}: text is printed unchanged in fixed font.
\item {\bf Tex}: a text that allows full \LaTeX{} functionality; such a
piece of information is ignored during compilation for help.
\item {\bf Info}: a text which is processed when compiling for help and
which is ignored during \LaTeX{} processing.
\end{itemize}
Typically \verb|Tex| and \verb|Info| are used one after the other in order
to describe the same context in an advanced (Tex) and a simple (Info) style.
\subsubsection{Normal Text}
This is text translated to a proportional font (where applicable). An empty
line causes a new paragraph to be started. Only a very limited \TeX{}
compatibility is given: Use the backslash for protecting special characters
like \{ and \}. Additionally formatting elements are available:
\begin{itemize}
\item \verb|\|verb
\item \verb|\|ldots
\item \verb|\|cdots
\item \verb|\|pi
\item \verb|\|em (may be without effect for some targets)
\item \verb|\|meta\{x\}: write $x$ as meta symbol for a syntax
description. In general this will lead to an output similar to $x$.
\item \verb|\|name\{x\}: write $x$ as a \REDUCE symbol. This form should be
used for all keywords, operator names etc.
\item \verb|\|nameref\{x\}: write $x$ as a \REDUCE symbol and generate a
reference to a node with the name $x$. This features is used to establish
the cross links in the information structure.
\item \verb|\|index\{x\}: generate an index entry. $x$ does not appear in
the target text.
\end{itemize}
Additionally there is a new command
\begin{quotation}
\verb|\IFTEX{|\TeX{} text\verb|}{|info text\verb|}|
\end{quotation}
that allows you to write parts of the text in two forms, one fancy form for
\LaTeX{} processing and a simpler form for the help environment. For
example, a mathematical formula like $\exp (x)$ must be written as follows:
\begin{verbatim}
\IFTEX{$\exp(x)$}{exp(x)}
\end{verbatim}
Note: the \verb|*| variants of \verb|\verb| and \verb|verbatim| cannot be
used. Informal guidelines:
\begin{itemize}
\item The text should have a nice structure. When the text has many lines
insert paragraph breaks such that the eye of the reader finds fixing points.
\item All \verb|\|index entries should follow immediately the header line
of the node.
\item Ensure that the spelling of index entries is unified, e.g. use only
singular and lower case (except for proper names).
\item Use \verb|\|nameref for as many names as possible. The
interrelations help the user to navigate through your document. Mention
every related node at least once in a \verb|\|nameref; if a related node
does not have a ``natural'' place in the text add a ``See also'' paragraph
at the end of the node.
\item Don't repeat \verb|\|nameref for the same object in one node unless
the distance between the citation points is rather big (nore than half a
page) - use \verb|\|name for repeated occurrences.
\end{itemize}
\subsubsection{Syntax Context}
This section contains the formal pattern of the introduced operator or
statement. The following items must be preceded by a backslash: {\bf round
brackets}, {\bf curly brackets}, {\bf underscore}, {\bf dollar}. Use
\verb|\meta| in the description for the variable items and explain these
using the same forms in the following plain text. The fixed items should be
encoded using \verb|\name|.
Examples:
\begin{verbatim}
\begin{Syntax}
\meta{logical\_expression} \name{and} \meta{logical\_expression}
\end{Syntax}
\begin{Syntax}
\name{<<}\meta{statement}\{; \meta{statement} \name{or}
\$\meta{statement}\}\optional \name{>>}
\end{Syntax}
\end{verbatim}
\subsubsection{Examples Context}
This section gives prototypical application examples. The text here has to
follow a special format where the line breaks in the input are unimportant
(except in the multiline option):
\begin{enumerate}
\item Each example must be terminated by a double backslash. This signals
that the next example starts (which should begin on the next line) or that
the last example is complete.
\item An example starts with the input line. This will be printed in
verbatim style with fixed font.
\item If the example has an associated output, this is started by an
ampersand (\verb|&|) character. The output must be either coded as \TeX{}
math line with restricted math facilities, using only
\begin{itemize}
\item \verb|^| for raised exponents and
\item \verb|\rfrac| for fractions (similar to \verb|\frac| in \TeX{}),
\end{itemize}
or the output may be a sequence of several lines that are then printed in
verbatim style. In the second case the lines must be enclosed in a pair
\verb|\begin{multilineoutput}| \verb|\end{multilineoutput}|.
\end{enumerate}
Informal guidelines:
Write short examples. Take into account that the space is very restricted
when using an online help system. In particular, the lines are rather short.
Examples:
\begin{verbatim}
\begin{Examples}
alist := \{1,2,\{a,b\}\}; & ALIST := \{1,2,\{A,B\}\} \\
blist := \{3,4,5,sin(y)\}; & BLIST := \{3,4,5,SIN(Y)\} \\
append(alist,blist); & \{1,2,\{A,B\},3,4,5,SIN(Y)\} \\
append(alist,\{\}); & \{1,2,\{A,B\}\} \\
append(list z,blist); & \{Z,3,4,5,SIN(Y)\}
\end{Examples}
\begin{Examples}
solve(log(sin(x+3)),x); &
\begin{multilineoutput}{6cm}
4*arbint(1)*pi + pi - 6 4*arbint(1)*pi + pi - 6
\{x=-------------------------,x=-------------------------\}
2 2
\end{multilineoutput}
\end{Examples}
\end{verbatim}
\section{\LaTeX{} Translation}
This should not cause any difficulties as long as the style files accessible.
\section{Help System Transformation}
In the \REDUCE library a set of utilities is available that allow you to
transform a reference style input file into an input to a target system.
These are source files in \REDUCE syntax, and you need a \REDUCE 3.5 or later
executable to use these. At present the source files are
\begin{itemize}
\item comphelp.red: driver and input analysis,
\item helpwin.red: output for MS Windows syntax
\item helpunx.tex: output for GNU Info format
\item minitex.red: formula formatting
\end{itemize}
First you must compile these using the script file $compile$. The result
will be binaries $cmphelpu.b$ and $cmphelpw.b$. For running these use
scripts $mkhelpu$ or $mkhelpw$ that take the name of the input file as
argument. This file may contain \verb|\include| or \verb|\input| statements
for secondary sources. During processing a list of sections and nodes is
printed to the standard output. The conversion process stops if a context
error is found - then the actual context hierarchy is printed. When the
analysis phase has been successful, in a second phase the target file is
generated. This is either a file $*.x$ (Info), to be further processed by
the GNU makeinfo, or a file $*.rtf$ that is input for the Microsoft help
compiler HC.
\end{document}