summaryrefslogtreecommitdiff
path: root/Master/texmf-dist/doc/latex/cweb-latex/src/cweb.doc
diff options
context:
space:
mode:
Diffstat (limited to 'Master/texmf-dist/doc/latex/cweb-latex/src/cweb.doc')
-rw-r--r--Master/texmf-dist/doc/latex/cweb-latex/src/cweb.doc1396
1 files changed, 1396 insertions, 0 deletions
diff --git a/Master/texmf-dist/doc/latex/cweb-latex/src/cweb.doc b/Master/texmf-dist/doc/latex/cweb-latex/src/cweb.doc
new file mode 100644
index 00000000000..8fb7a0a5582
--- /dev/null
+++ b/Master/texmf-dist/doc/latex/cweb-latex/src/cweb.doc
@@ -0,0 +1,1396 @@
+% $StyleId: cweb.doc,v 3.6 1995/11/30 15:44:53 schrod Exp $
+%----------------------------------------------------------------------
+% Written by Joachim Schrod <schrod@iti.informatik.th-darmstadt.de>.
+% Copyright conditions see below (GPL).
+
+%
+% LaTeX class cweb
+% support for LaTeX markup in CWEB sources
+%
+% [LaTeX in MAKEPROG]
+% (history at end)
+
+% If you have received this style file without the user manual (in the
+% file cweb-user.dvi or the respective LaTeX file), it's incomplete and
+% near to useless. If it was given to you as something that you shall
+% use as an author -- complain bitterly to your provider. You need the
+% documentation and you have a right on it! (Below you can find info
+% where to get the reference version.)
+
+
+
+
+\documentclass{progltx}
+
+\usepackage{cweb-doc} % document-specific markup
+\usepackage{fullpage}
+
+
+\RCS $StyleRevision: 3.6 $
+\RCS $StyleDate: 1995/11/30 15:44:53 $
+
+
+\begin{document}
+
+
+\title{The \texttt{cweb} Class\\
+ {\large (Implementation)}%
+ }
+\author{% % LaTeX does not discard unnecessary glue...
+ Joachim Schrod%
+ \thanks{%
+ \protect\raggedright
+ TU~Darmstadt, Computer Science Department, WG Systems Programming,
+ Alexanderstr.~10, D-64283~Darmstadt, Germany.
+ Email: \texttt{jschrod@acm.org}%
+ }%
+ }
+\date{%
+ Revision \RCSStyleRevision\\
+ (as of \RCSStyleDate)%
+ }
+
+\maketitle
+
+
+% doesn't work with progltx yet
+%\tableofcontents
+
+
+
+% ------------------------------------------------------------
+%
+% subdocument: The user interface of cweb.cls
+%
+
+\input{cweb-user}
+
+%
+% ------------------------------------------------------------
+%
+% subdocument: The internal interface
+%
+
+\input{cweb-conf}
+
+%
+% ------------------------------------------------------------
+
+
+
+
+\chap Implementation.
+
+This implementation must typeset complete files output by \cweave{}.
+Large parts of this task are available from the module \pkg{cwebbase}.
+It remains mainly to implement document-level markup and front and
+back matter (table of contents, index, etc.) In particular, the back
+matter material is not trivial; its tags are behind |\end{document}|,
+we have to read carefully behind the end of our \cweb{} document to
+detect if we have to typeset them at all.
+
+
+\sect This module reserves the namespace |cweb|.
+
+\beginprog
+\ifx \cweb@loaded\undefined
+ \def\cweb@loaded{$StyleRevision: 3.6 $}
+\else
+ \PackageWarningNoLine{cweb}%
+ {Some other package already uses namespace `cweb'}
+\fi
+\endprog
+
+
+\sect Let's identify this class.
+
+\noindent The code below is explained in the implementation
+documentation of the \pkg{rcs} package.
+
+\beginprog
+\begingroup
+ \def\RCSClass#1#2 $#3: #4 #5\endRCS $#6: #7 #8\endRCS{%
+ \def\date{#4}\def\id{v#7}%
+ \ProvidesClass{#1}[\date\space\id\space #2]%
+ }
+
+ \RCSClass{cweb}{LaTeX markup for CWEB sources}
+ $StyleDate: 1995/11/30 15:44:53 $: 9999/00/00 \endRCS
+ $StyleRevision: 3.6 $: 0.0 \endRCS
+\endgroup
+\endprog
+
+
+\sect This module is from a supported bundle. Send bug reports,
+comments and repairs.
+
+The reference version may be retrieved via anonymous ftp from
+|ftp.th-darmstadt.de| [130.83.47.112], directory
+|pub/programming/literate-programming/c.c++/|. It's
+placed there as a gzipped tar file. (The information on the IP~number
+is dated August~29,~1995.
+It might have changed, although this is very unlikely. Use
+your friendly nameserver.)
+
+
+\sect This is freely distributable software; you can redistribute it
+and/or modify it under the terms of the GNU General Public License as
+published by the Free Software Foundation; either version~2 of the
+License, or (at your option) any later version.
+
+This software is distributed in the hope that it will be useful, but
+\textbf{without any warranty}; without even the implied warranty of
+\textbf{merchantability} or \textbf{fitness for a particular purpose}.
+See the GNU General Public License for more details.
+
+You should have received a copy of the GNU General Public License in
+the file |License| along with this package; if not, write to the Free
+Software Foundation, Inc., 675~Mass Ave, Cambridge, MA~02139,~USA.
+
+
+\sect Before we start we declare some shorthands for category codes.
+By declaring the underscore~`(|_|)' as letter we can use it in our
+macros. (I agree with \textsc{D.~Knuth} that
+|\identifier_several_words_long| is more readable than
+|\IdentifierSeveralWordsLong| and in every case better than |\p@@@s|.)
+
+As this is a \LaTeX{} style file the at sign is a letter anyhow; so
+we can use the ``private'' \LaTeX{} macros; and with the underscore we
+can make our own macros more readable. But as we have to restore this
+category code at the end of this macro file we store its former value
+in the control sequence |\CatUsCode|. This method is better than to
+use a group because not all macros have to be defined global this way.
+
+Since somebody might use more styles from me, this cseqs might be
+defined already.
+
+\beginprog
+\ifx \CatEscape\undefined
+ \chardef\CatEscape=0
+ \chardef\CatOpen=1
+ \chardef\CatClose=2
+ \chardef\CatIgnore=9
+ \chardef\CatLetter=11
+ \chardef\CatOther=12
+ \chardef\CatActive=13 % is defined in Plain already
+
+ \chardef\CatUsCode=\catcode`\_
+\fi
+
+\catcode`\_=\CatLetter % top level macro file
+\endprog
+
+
+\sect Problems over problems with different \LaTeX{} versions. I have
+decided that I will take the same stand as the \LaTeX{} team: I will
+only support the last two official \LaTeX{} versions, folks who run
+older software might need to upgrade.
+
+The implementation will use `star'-forms of \LaTeX{}'s command
+definition macros. They were introduced with \LaTeX{}
+\mbox{$\langle$1994/12/01$\rangle$}.
+
+Currently (as of 07 Nov 95), 1995/06/01 is the most recent version.
+I want to use a features introduced with that version: |\newif| is not
+outer any more. So I check here for the version and supply the new
+definition if necessary.
+
+\beginprog
+\NeedsTeXFormat{LaTeX2e}[1994/12/01]
+\@ifl@t@r \fmtversion{1995/06/01}{}% else
+ {
+ \@namedef{newif}#1{%
+ \count@\escapechar \escapechar\m@ne
+ \let#1\iffalse
+ \@if#1\iftrue
+ \@if#1\iffalse
+ \escapechar\count@
+ }
+ \def\@if#1#2{%
+ \expandafter\def
+ \csname \expandafter\@gobbletwo\string#1%
+ \expandafter\@gobbletwo\string#2\endcsname {%
+ \let#1#2%
+ }%
+ }
+ }
+\endprog
+
+
+
+
+\chap Options.
+
+Our options are of the ``keyword-value'' category. We realize them by
+the \pkg{keyvald} package, an enhancement of the \pkg{keyval} package
+from the Graphics bundle that supports defaulting for unknown keys.
+
+\begin{fixme}
+ Most of the code below should be extracted and moved to a
+ \pkg{keyvalx} package (or a new version of \pkg{keyval}). In
+ particular, support for package/class option processing, enumeration
+ values, and set values for keyword options.
+\end{fixme}
+
+|\ProcessKeyValOptions| may be used instead of |\ProcessOptions|, it
+takes the keyval set identifier as argument. At the end it evaluates
+and resets a hook. That hook may be used at option definition to
+define further dependencies. (E.g., enabling that option implies
+enabling an other option, too.)
+%
+\begin{fixme}
+ Like |\ProcessOptions|, I define |\@curroptions|, since I don't know
+ if that's needed later. The only place where it could been needed is
+ in the check for unprocessed options.
+\end{fixme}
+
+\beginprog
+\RequirePackage{keyvald}
+
+\let\ProcessKeyValOptionsHook\empty
+\providecommand*\ProcessKeyValOptions[1]{%
+ \edef\@curroptions{\@ptionlist{\@currname.\@currext}}%
+ \edef\next{\noexpand\setkeys{#1}{\@curroptions}}%
+ \next
+ \ProcessKeyValOptionsHook
+ \let\ProcessKeyValOptionsHook\empty
+ }
+\endprog
+
+
+\sect The enumeration option |structure| selects hierarchic~(0) or
+flat~(1) structure. The predefined value is `hierarchic', there is no
+default. The option value is bound to |\cweb@structure|.
+
+\beginprog
+%% \define@key@enum{cweb}{structure}{hierarchic,flat}[hierarchic]
+\@namedef{KV@cweb@structure@enum:hierarchic}{0 } % <-- space!
+\@namedef{KV@cweb@structure@enum:flat}{1 } % <-- space!
+\define@key{cweb}{structure}{%
+ \expandafter\let \expandafter\@tempa
+ \csname KV@cweb@structure@enum:#1\endcsname % \relax if undefined
+ \ifx \@tempa\relax
+ \PackageError{cweb}%
+ {%
+ Invalid value `#1' for option structure%
+ }{%
+Possible valid values are `hierarchic' and `flat'.\MessageBreak
+The predefined value is `hierarchic'.\MessageBreak
+There's no default value, you have to specify one.%
+ }%
+ \else
+ \let\cweb@structure\@tempa
+ \fi
+ }
+\setkeys{cweb}{structure=hierarchic}
+\endprog
+
+
+\sect With the set option |suppress| one can select suppression of
+different document parts: |changehints| suppresses output of hints
+that a changefile was involved, |unchanged| output of unchanged
+chunks, |index| and |reflist| output of index and reference list, and
+|format| output of `|@f|' directives.
+
+The predefined value is no suppression. There is no default value.
+%
+\begin{fixme}
+ Shouldn't I supply a default value? `|{index,reflist}|'?
+\end{fixme}
+
+Set options are represented by flags named
+|@cweb@|\<option>|@|\<value>|@|.
+
+\beginprog
+%% \define@key@set{cweb}{suppress}{changehints,unchanged,index,reflist,format}
+\@for \@tempa :=changehints,unchanged,index,reflist,format\do {%
+ \expandafter\newif \csname if@cweb@suppress@\@tempa @\endcsname
+ \csname @cweb@suppress@\@tempa @false\endcsname
+ }%
+\define@key{cweb}{suppress}{%
+ \@for \@tempa :=#1\do {%
+ \expandafter\let \expandafter\@tempb
+ \csname @cweb@suppress@\@tempa @true\endcsname
+ %% \relax if undefined
+ \ifx \@tempb\relax
+ \PackageError{cweb}%
+ {%
+ Invalid value `#1' for set option suppress%
+ }{%
+Possible valid values are `changehints', `unchanged',\MessageBreak
+or a comma-separated list of these words enclosed in braces.\MessageBreak
+Specified values are turned on, you can't turn them off.\MessageBreak
+No value is on initially.\MessageBreak
+There's no default value, you have to specify one.%
+ }%
+ \else
+ \@tempb
+ \fi
+ }%
+ }
+
+\g@addto@macro\ProcessKeyValOptionsHook{%
+ \if@cweb@suppress@unchanged@
+ \@cweb@suppress@changehints@true
+ \fi
+ }
+\endprog
+
+
+\sect The base document class may be specified by the option
+|baseclass|.
+
+\beginprog
+\def\CwebBaseClass{article}
+\define@key{cweb}{baseclass}[report]{\def\CwebBaseClass{#1}}
+\endprog
+
+
+\sect Language-specific adaptions are stored in files
+\texttt{cwbl-{\it language}.sty}. They may be accessed as packages or
+by the |language| option that takes the \textit{language} as value. We
+have to require the respective package at the end of the class, then
+all cseqs with default names are defined and may be overwritten.
+
+For some languages, we provide Babel-compatible option names, to ease
+the usage of global class options taken up by other packages (like
+\pkg{varioref}), too.
+
+\beginprog
+\def\cweb_set_language#1{%
+ \IfFileExists{cwbl-#1.sty}{%
+ %% true
+ \RequirePackage{cwbl-#1}%
+ }{%
+ %% false
+ \ClassError{cweb}{Language `#1' is unknown.}{%
+Language definitions are stored in files `cwbl-<language>.sty'.\MessageBreak
+I could not locate the file `cwbl-#1.sty'. If you didn't make a%
+\MessageBreak
+typo, please check if there is such a file in a directory searched\MessageBreak
+by LaTeX.%
+ }%
+ }%
+ }
+
+\define@key{cweb}{language}{\AtEndOfClass{\cweb_set_language{#1}}}
+
+\define@key{cweb}{german}[]{\setkeys{cweb}{language=german}}
+\define@key{cweb}{french}[]{\setkeys{cweb}{language=french}}
+\define@key{cweb}{italian}[]{\setkeys{cweb}{language=italian}}
+\endprog
+
+
+
+
+\chap Configuration, inheritance, and aggregation.
+
+This chapter sets up access to other modules, therefore we restore the
+original catcodes for the time being.
+
+\beginprog
+\catcode`\_=\CatUsCode
+\endprog
+
+
+\sect As usual, we provide the possibility to configure this class.
+
+\beginprog
+\InputIfFileExists{cweb.cfg}{%
+ \typeout{******************************************^^J%
+ *^^J%
+ * Using local configuration file cweb.cfg^^J%
+ *^^J%
+ ******************************************%
+ }%
+ }{}
+\endprog
+
+
+\sect Inherit article class.
+
+\beginprog
+\ifx \CwebBaseClass\undefined
+ \def\CwebBaseClass{article}
+\fi
+\define@key{cweb}*[]{\PassOptionsToClass{\KV@key}{\CwebBaseClass}}
+\ProcessKeyValOptions{cweb}
+\LoadClass{\CwebBaseClass}
+\endprog
+
+
+\sect OK, now comes our own code; switch on again our lexical
+conventions.
+
+\beginprog
+\catcode`\_=\CatLetter
+\endprog
+
+
+
+
+\chap Typesetting \cweave{} output.
+
+This task has been out-sourced to an other module, \pkg{cwebbase}. As
+mentioned at the start of the implementation, there remain mainly some
+bigger tasks, they will be implemented in further sections below. For
+the most part, the necessary preconditions are actually our class
+option definitions, we don't have to do anything. (Gosh, what a
+coincidence. :--) We need to tell \pkg{cwebbase}, that it should use
+class warnings on problems.
+
+\beginprog
+\let\cweb_warning=\ClassWarningNoLine
+\input{cwebbase}
+\endprog
+
+
+\sect But there remain also some small things to be done right now:
+For flat structure, the counter |secnumdepth| must be set to~11, then
+we want to number chunks in the whole document.
+
+\beginprog
+\ifcase \cweb@structure
+ %% hierarchic
+ \or
+ %% flat
+ \setcounter{secnumdepth}{11}
+\fi
+\endprog
+
+
+\sect \pkg{cwebbase} uses |\part|, |\chapter| (if bound), and
+|\section|. We have no use for any other section division that came
+from the base class, let's discard the standard ones.
+
+\beginprog
+\let\subsection\undefined
+\let\subsubsection\undefined
+\let\paragraph\undefined
+\let\subparagraph\undefined
+\endprog
+
+
+
+
+\chap Table of contents.
+
+The table of contents features only sections, chunks don't have
+titles, after all. Titles are indented proportionally to the rank,
+with an basic indentation of |\CwebTocIndent|. An entry on major
+ranks may be differentiated by layout instead of indentation, as it's
+done in standard \LaTeX{} classes, too.
+
+The counter |\CwebTocIndentMaxLevel| constitutes an upper limit for a
+recognized indent (concerning indentation, that is).
+
+\beginprog
+\newdimen\CwebTocIndent
+ \CwebTocIndent=1.5em
+\newcount\CwebTocIndentMaxLevel
+ \CwebTocIndentMaxLevel=6 % increase indentation if <= max level
+\endprog
+
+
+\sect For hierarchic structure, we can ignore chunk entries
+completely, but have to supply a bunch of definitions for entries of
+type `|section|$l$'. But for flat structure, sections are passed as
+chunks, too. Then the current rank is stored in the title, as
+|\cwbbRank{|$r$|}|. These differences are so large that we'll have
+to make a very large |\ifcase|, over many chunks in this macro source
+file. Let's start with the hierarchic stuff.
+
+\beginprog
+\ifcase \cweb@structure
+ %% hierarchic
+ \def\l@chunk#1#2{}
+\endprog
+
+
+\sect Let's assume that we're subclassed from one of \LaTeX{}'s
+standard classes; otherwise this stuff must be redefined, too. They
+indent |\section| entries always by the same amount (1.5\,em), our
+layout must fit to that.
+
+If we have chapters, they are also indented by 1.5\,em and
+distinguished by layout. The first minor section starts on rank~3 and
+is placed on level~2, it should be indented larger. Let's simply add
+1.5\,em for each rank, that makes 3\,em for sections of type
+|section2|. I.e., we use $f=l$ as a factor for our basic indentation.
+
+If we don't have chapters, the first minor section starts on rank~2
+and is placed also on level~2, it should be indented like sections
+(1.5\,em), as it will be in another layout. I.e., we use $f=l-1$ as
+the indentation factor.
+
+The actual definitions are generated in a loop. `|\l@section|$l$'
+expands to `|\cweb@toc_entry{|$l$|}|'. The indentation will be
+computed within that macro.
+
+\beginprog
+ \count@=2
+ \@whilenum \count@<11 \do {
+ \expandafter\edef \csname l@section\number\count@\endcsname{%
+ \noexpand\cweb@toc_entry{\number\count@}%
+ }
+ \advance \count@ by 1
+ }
+\endprog
+
+
+\sect A toc entry may be produced by |\@dottedtocline|. That cseq
+needs the level (to decide if that entry is produced after all), the
+entry indentation, the inner-entry indentation used for the number,
+entry text, and page number.
+
+As explained above, we use an indentation of $f \cdot
+|\CwebTocIndent|$, if the level is not larger than
+|\CwebTocIndentMaxLevel|. Otherwise we use the indentation of that
+level.
+
+Each entry is kind of an item with the number as the label. We need an
+inner-entry indentation to be used for subsequent lines within this
+entry. If a number is specified, it's tagged by |\numberline| in the
+entry text. With a proper local definition and a pro-forma evaluation
+of the text, we get at the width of the number (with a fitting space
+for separation to text) and can use that as the inner-entry
+indentation.
+
+\beginprog
+ \def\cweb@toc_entry#1#2{% % #3 (page) will follow
+ %% \count@ == factor for basic indentation
+ %% min(#1,TocIndentMaxRank) - correct_level
+ \ifnum #1>\CwebTocIndentMaxLevel
+ \count@\CwebTocIndentMaxLevel
+ \else
+ \count@ #1 % <-- space!
+ \fi
+ \ifx \chapter\undefined
+ \advance \count@\m@ne
+ \fi
+ %% \dimen@ == inner-entry indentation
+ %% width of number & space, if existing
+ \begingroup
+ \global\dimen@ \z@ % assert value
+ \def\numberline##1{\settowidth{\global\dimen@}{##1\enskip}}%
+ \setbox\z@ \hbox{#2}% % evaluates \numberline if specified
+ \endgroup
+ \@dottedtocline
+ {#1}% % level
+ {\count@ \CwebTocIndent}% % basic indent
+ \dimen@ % num width
+ {#2}% % entry; page will follow
+ }
+\endprog
+
+
+\sect We're done with toc configuration for hierarchic structure,
+since entry definitions for |chapter| and |section| types exist
+already. Let's turn over to the flat structure.
+
+\beginprog
+ \or
+ %% flat
+\endprog
+
+
+\sect The section rank determines how they are featured in the table.
+Titles on rank~0 are typeset boldface, other titles are dotted
+contents lines in roman.
+
+The rank is specified within the title, i.e., in the
+first argument to |\l@chunk|, as an argument for the |\cwbbRank| tag.
+With an appropriate definition we might set |\count@| to the
+respective value. If it is not set, an explicite invocation of
+|\section| or |\addcontentsline| is responsible for this entry. Then
+we assume that the rank is~0.
+
+\beginprog
+ \def\l@chunk#1{% % page will be processed later
+ \count@\@ne % default value of group level
+ \begingroup
+ \let\numberline\@gobble % width not known yet
+ \def\cwbbRank##1{\global\count@ ##1\relax}%
+ \setbox\z@ \hbox{#1}%
+ \endgroup
+ \ifnum \count@=\z@
+ \let\next\cweb@bold_toc_line
+ \else
+ \let\next\cweb@normal_toc_line
+ \fi
+ \next{#1}%
+ }
+\endprog
+
+
+\sect A boldface line is typeset similar to the definition of
+|\@dottedtocline|, only with fewer parameters. It takes two
+parameters, the entry and the page number.
+
+\beginprog
+ \def\cweb@bold_toc_line#1#2{%
+ \addpenalty{\@secpenalty}%
+ \addvspace{1em plus\p@}%
+ \begingroup
+ \noindent \bf
+ \hangindent\CwebTocIndent
+ \rightskip\@tocrmarg \parfillskip -\rightskip
+ \interlinepenalty\@M
+ \@tempdima\CwebTocIndent % for \numberline
+ #1\nobreak\hfill \hbox to\@pnumwidth{\hss #2}%
+ \par
+ \endgroup
+ }
+\endprog
+
+
+\sect A normal line uses |\@dottedtocline| like the hierarchic stuff.
+Rank~1 shall still be on the same level as the bold toc line, so we
+have to use $r-1$ as the proportional factor for |\CwebTocIndent|.
+
+\beginprog
+ \def\cweb@normal_toc_line#1{% % page will be processed later
+ \edef\next{\noexpand\@dottedtocline{\the\count@}}% % toc entry level
+ \ifnum \count@>\CwebTocIndentMaxLevel
+ \count@ \CwebTocIndentMaxLevel
+ \fi
+ \advance\count@\m@ne % group level -= 1
+ \next{\count@\CwebTocIndent}% % basic indent
+ \CwebTocIndent % numwidth
+ {#1}% % entry
+ }
+\endprog
+
+
+\sect The first argument of |\@dottedtocline| is used to specify a
+depth of the issued entry. All entries with a depth larger than
+|tocdepth| are discarded. We typeset all entries, our |tocdepth|
+value is the largest possible section level.
+
+\beginprog
+ \setcounter{tocdepth}{10}
+\endprog
+
+
+\sect That's it. Close our distinction between hierarchic and flat
+structure.
+
+\beginprog
+\fi
+\endprog
+
+
+
+
+\chap The end of the document.
+
+The end of a \cweave{} document is a rather complicated thing. It is
+described in detail in the \cweave{} interface specification, we repeat the
+most important points shortly.
+%
+\begin{itemize}
+
+\item The user has issued an |\end{document}| tag in the documentation
+part of the last chunk.
+
+This tag is therefore part of a chunk, the chunk end (i.e., the
+|\fi| cseq) does appear later in the document. Particularly, this
+means that at the accurence of this tag there is still an open |\if|.
+
+\item Between the |\end{document}| and the |\fi| a lot of text may
+come which must be ignored. Within this text no |\fi| is allowed.
+
+\item After the |\fi| the document may be finished by |\end|.
+(Then the |-x| option of \cweave{} was used.)
+
+Or additional information about the whole document is added: An
+optional tag |\ch|, two mandatory tags |\inx\fin|, and another
+optional tag |\con|.
+
+\end{itemize}
+
+Btw, the last chunk is marked as changed if any changefile entry has
+been applied in the document. (In Plain \cweb{} that chunk is
+colloquially used as the introduction to the index, and that shall be
+printed always.) If we suppress unchanged chunks, the last
+chunk will therefore always be processed, independent if it was really
+changed or not. Yet another reason to use a whole chunk just for
+|\end{document}|.
+
+We'll use a the following approach for the implementation of the
+iidentifier index and the refinement list (remember that we do not
+produce a table of contents): We will check if there is any of these
+two lists available. The available ones are typeset. Then we'll
+finish the document. This functionality will be bound to |\fin|.
+|\inx| will be simply a no-op. |\con| will never be reached, so we
+don't have to cope with it.
+
+
+\sect With the introduction of an end-document hook in \LaTeXe{}, one
+might have hoped that we can utilize that hook for our task. But this
+is not possible: We don't want to add some additional actions
+\emph{into} the end-document handling, we want to look ahead and issue
+lots of commands \emph{before} the end-document handling has happened.
+(If you know \TeX{} by heart, you may skip the rest of the explanation
+and go to the next chunk.)
+
+As a macro language, \TeX{}'s cseqs take those tokens as arguments
+that are behind them in the token stream. The token stream is
+initially the document source, evaluation of a macro pushes its
+expansion (a token list) on front of the stream. I.e., the evaluation
+of |\end| will push the expansion on the token stream before
+|\enddocument| is even looked at. And |\enddocument| will also push
+its expansion before the hook is evaluated.
+
+If we would allow |\enddocument| to push its expansion, all these
+tokens have to be evaluated after we have typeset the backmatter --
+after all, we need a proper document end, don't we? One would need to
+skip over them, look at and eventually evaluate tokens that come
+behind (from the initial document stream, the document source), and
+then evaluate the skipped tokens. That's too hairy, we don't know
+exactly how many tokens to skip, this is probable to change in future
+\LaTeX{} revisions, etc.
+
+The pushed tokens from |\end| don't pose so much problems. As
+|\enddocument| will terminate the document processing, they are not
+evaluated anyhow, we can just throw them away. And |\end| has changed
+only once in the last few years, so it seems to be a stable macro --
+necessary as we need to know its expansion for propper skipping.
+
+A very first consequence of the demand to process tokens after
+|\end{document}| is that we have to save the ``standard'' \LaTeX{}
+document termination, since we will execute that later.
+
+\beginprog
+\let\cweb@save@enddocument=\enddocument
+\endprog
+
+
+\sect When we encounter |\end{document}| we have to skip tokens
+until we reach a |\fi|. Then we establish the correct
+environment for the material behind.
+
+The skip itself is actually a bit difficil, so we'll look at it more
+closely. |\enddocument| is executed within |\end|. As mentioned above,
+the original
+definition of |\enddocument| cheats, it terminates \LaTeX{} before
+the rest of |\end| is executed. |\end| first checks if we're in
+the correct current environment, then it establishs the end
+processing for paragraph building environments, closes the
+group---each environment is in its own group, executes the end
+processing established before, and it may ignore following white
+space.
+
+The |document| pseudo-environment is different in that it isn't in a
+group. I.e., we have to open a group which can be closed by |\end|
+afterwards. The environment check will be done in this group. With
+|\aftergroup| we establish a macro which will skip the rest of the
+|\end| processing. This processing is not needed anyhow: |document|
+is no paragraph building environment and the following white space is
+ignored by ourselves. Since |document| is no paragraph building
+environment, the end processing described above is empty, i.e., not
+existant. Then come a few tokens which ignore the white space. One of
+these tokens is a |\fi|, so we have to skip them first separately.
+They are discarded by |\cweb@skip_end|, we can start to look
+for our |\fi| afterwards.
+
+If |\end{document}| is the first (i.e., the only) text in this chunk
+we discard the chunk start marker. This is possible since it isn't set
+until now, it's in the |\everypar| token register. Of course, then we
+have to do the house keeping work stored in |\everypar| ourselves.
+
+\beginprog
+\def\enddocument{%
+ \if@noskipsec
+ \global\everypar{}%
+ \global\@noskipsecfalse
+ \global\@nobreakfalse
+ \global\clubpenalty\@clubpenalty
+ \fi
+ \begingroup
+ \aftergroup\cweb@skip_end
+ }
+\endprog
+
+
+\sect Let's make sure that the expansion of |\end| is the one that we
+expect. Then we can define |\cweb@skip_end| appropriately, we have to
+skip 5~tokens.
+
+\beginprog
+\CheckCommand*\end[1]{%
+ \csname end#1\endcsname\@checkend{#1}%
+ \expandafter\endgroup\if@endpe\@doendpe\fi
+% 1 2 3 4 5
+ \if@ignore\global\@ignorefalse\ignorespaces\fi}
+
+\def\cweb@skip_end#1#2#3#4#5{\cweb@check_fi}
+\endprog
+
+
+\sect |\cweb@check_fi| gets the next token, and invokes
+|\cweb@do_check_fi| who really checks it. If this token is |\fi| we
+can prepare to finish our document. Otherwise we have to check the
+next token.
+
+Of course, first the token we have checked already must be discarded.
+Therefore we insert the followup action by |\afterassignment|. It's
+important that the assignment gets \emph{only} the next token, we
+have to supply both the equals sign and the optional blank which are
+allowed between the both arguments to |\let|.
+
+\beginprog
+\def\cweb@check_fi{%
+ \futurelet\next \cweb@do_check_fi
+ }
+\def\cweb@do_check_fi{%
+ \ifx \next\fi
+ \afterassignment\cweb@finish
+ \else
+ \afterassignment\cweb@check_fi
+ \fi
+ \let\next= % <-- blank!!
+ }
+\endprog
+
+
+\sect When we found the |\fi|, we also discard it. But we're still
+within a chunk, i.e., there is an open |\if|. So we must supply the
+|\fi|.
+
+\TeX{}nical note: |\next| is now equivalent to |\fi|. It must not be
+used within a loop where it denotes the tail recursion's action. I.e.,
+code like
+ %
+\begin{verbatim}
+ \ifx \foo\bar
+ \let\next\relax
+ \else
+ \let\next\continue
+ \fi
+ \next
+\end{verbatim}
+ %
+will fail. This trap is closed by a redefinition of |\next|.
+
+The output of all following lists may be suppressed by some option.
+That is tested at the place where the respective list shall be
+produced.
+
+All following lists are typeset in CR state.
+
+\beginprog
+\def\cweb@finish{%
+ \fi
+ \let\next\relax
+ \let\ch\cweb@changed_list
+ \let\inx\relax
+ \let\fin\cweb@end_document
+ \let\end\cweb@save@enddocument
+ \cwbb@CR
+ }
+\endprog
+
+
+\sect The following lists are started with unnumbered section
+divisions. They are either chapters or sections.
+
+\beginprog
+\ifx \chapter\undefined
+ \let\CwebListSectionDiv=\section
+\else
+ \let\CwebListSectionDiv=\chapter
+\fi
+\def\CwebListSection{\CwebListSectionDiv*}
+\endprog
+
+
+\sect The list of changed chunks is a cross reference list, (nearly)
+like all others at the end of a chunk. The only difference is that
+we do not show changeflags any more---each chunk number in this list
+carries a change flag by definition.
+
+\TeX{}nical note: The redefinition of |\*| is part of the second
+argument of |\CwebCrossRef|. It is \emph{not} a global redefinition.
+
+\beginprog
+\def\CwebCRChanged{%
+ \CwebCrossRef{The following chunks were changed by the change file:}%
+ }
+\def\CwebChangedListName{Changed Chunks}
+
+\if@cweb@suppress@changehints@
+ \def\cweb@changed_list#1.{}
+\else
+ \def\cweb@changed_list{%
+ \CwebListSection{\CwebChangedListName}%
+ \@mkboth{\uppercase{\CwebChangedListName}}%
+ {\uppercase{\CwebChangedListName}}%
+ \message{\CwebChangedListName}%
+ \let\CwebCRSize\relax
+ \CwebCRChanged \let\*\relax
+ }
+\fi
+\endprog
+
+
+\sect As outlined above, we want to check if there are any entries in
+the identifier index or the refinement list. The former is stored in
+the file |\jobname.idx|, the latter in the file |\jobname.scn|.
+
+We write a macro |\cweb@if_file_not_empty| that's like |\IfFileExists|.
+
+\beginprog
+\def\cweb@end_document{%
+ \cweb@if_file_not_empty{\jobname.idx}{%
+ \if@cweb@suppress@index@ \else
+ \CwebIdIndex
+ \fi
+ }{}%
+ \cweb@if_file_not_empty{\jobname.scn}{%
+ \if@cweb@suppress@reflist@ \else
+ \CwebRefList
+ \fi
+ }{}%
+ \cweb@save@enddocument
+ }
+\endprog
+
+
+\sect The paragraph layout for both the index and the refinement list
+is much the same, it is taken from Plain \cweb{}: Each entry is a
+paragraph, nearly no skip between the paragraphs (just a bit to
+prevent underfull vboxes), no paragraph indentation, ragged right.
+Overfull hboxes in the lists don't make sense, so we prevent them. And
+we don't allow hyphenation.
+
+\beginprog
+\def\cweb@list_par_layout{%
+ \parskip \z@ plus .5\p@
+ \parindent\z@
+ \rightskip \z@ plus 2.5em
+ \tolerance\@M \hyphenpenalty\@M
+ }
+\endprog
+
+
+\sect We consider a file as ``not empty'' if (1)~the file exists,
+(2)~is not empty, and (3)~has a non-empty line at the very front. (In
+fact, an empty line at the front means that there should not be
+anything behind it---but we can't test this portably.)
+
+|\cweb@if_file_not_empty| tests the property, it's parametrized by the
+file name and gets passed actions for the true and the false case.
+
+We open the file first. Then we check if it doesn't exist or if it's
+empty, both conditions deliver true on |\ifeof|; in this case we
+pretend that there was an empty line. (The \LaTeX{} kernel already
+once provided a macro for an empty line, |\@defpar|, we repeat that
+definition here since it was discarded there.) Otherwise we read the
+first line. At this state, the emptiness of the first line is
+equivalent to the non-availability of entries, we can easily construct
+an appropriate macro call to set the flag.
+
+\beginprog
+\def\@defpar{\par}
+\def\cweb@if_file_not_empty#1#2#3{%
+ \openin\@inputcheck #1\relax
+ \ifeof \@inputcheck
+ \let\next\@defpar
+ \else
+ \read\@inputcheck to \next
+ \fi
+ \closein\@inputcheck
+ \ifx \next\@defpar
+ \def\reserved@a{#3}%
+ \else
+ \def\reserved@a{#2}%
+ \fi
+ \reserved@a
+ }
+\endprog
+
+
+\sect The identifier index is available in the file |\jobname.idx|.
+The setup for the index is a mixture of the |theindex| environment of
+the \cls{article} class and DEK's index macros. It's typeset in two
+columns; the user may specify an introductionary text for the index
+by |\cwebIndexIntro|. If there is any introductionary text we add a
+medium skip below.
+
+For the index, |\parfillskip| set in such a way that almost empty
+lines are avoided.
+
+Before we read in this file, we have to bind the special cseqs used
+therein.
+
+\begin{fixme}
+ \LaTeX{} feature or bug? The optional argument of |\twocolumn| is
+ typeset as a vbox, therefore the space at top of the section heading
+ is not discarded at the top of a page. But we know that |\twocolumn|
+ will start a page in any case, we must prevent the addition of this
+ space. Looking in the source, we see that setting the flag
+ |@nobreak| will prevent this addition, let's do that as a
+ workaround. File a \LaTeX{} bug report?
+\end{fixme}
+
+\begin{fixme}
+ Surprise: If |@nobreak| is set \emph{and} if there are no index
+ entries, |\twocolumn| will drop its optional argument. (That
+ happened to be noticed when there was an error in
+ |\cweb@if_file_not_empty|.) File a \LaTeX{} bug report?
+\end{fixme}
+
+\begin{fixme}
+ If |\CwebListSection| is a chapter tag of the standard classes, that
+ macro will check if it's in twocolumn mode and will try to set
+ itself at the top of the current page in full width, as it would
+ have been given in an optional argument of |\twocolumn|. As the
+ method that optional argument uses (a float) must not be nested, we
+ must prevent this to happen. We set the |@twocolumn| flag to false,
+ chapter macros will then think they're evaluated in a one-column
+ document. It's debatable if that is a kludge or even \emph{the}
+ solution -- after all, within the optional argument we aren't in
+ twocolumn mode any more. File a \LaTeX{} bug report?
+\end{fixme}
+
+\beginprog
+\newtoks\cwebIndexIntro
+ \cwebIndexIntro={}
+\def\CwebIdIndex{%
+ %% two column format
+ \columnseprule\z@
+ \columnsep 35\p@ % value is from article.sty
+ %% introduction of index
+ \twocolumn[%
+ \cwbb@tex
+ \@nobreaktrue % don't add space in front of section
+ \@twocolumnfalse % \CwebListSection might be \chapter
+ \CwebListSection{\indexname}%
+ \edef\intro{\the\cwebIndexIntro}% % is a local def
+ \ifx \intro\empty
+ \else
+ \the\cwebIndexIntro\unskip
+ \par\medskip
+ \fi
+ ]%
+ \@mkboth{\uppercase{\indexname}}{\uppercase{\indexname}}%
+ \message{\indexname:}% % tell the user what we're doing
+ %% paragraph layout
+ \cweb@list_par_layout
+ \parfillskip \z@ plus .6\hsize
+ %% bindings
+ \let\I\CwebIndexEntry
+ \let\[\CwebIndexDeclared
+ \let\*\cwbb@change_flag
+ %% read the index
+ \CwebReadIndex
+ }
+\def\CwebReadIndex{\@input{\jobname.idx}}
+\endprog
+
+
+\sect An index entry is typeset with the same hanging indentation like
+a cross reference list.
+
+The entry is tagged with |\.|, if it was entered by the \cweb{}
+operator~`|@.|'. Then it shall be typeset as a string. But an indexed
+name may also want to use |\.| as an accent. This is the same
+situation as we had at the refinement names (in module
+\pkg{cwebbase}), where we introduced |\cwbb@check_dot| to handle this
+case. The same minor restriction as there holds here, a refinement
+name may not consist of a single dot-accented expression. (|@:|~helps
+in this singular case.)
+
+\beginprog
+\def\CwebIndexEntry#1, {%
+ \par
+ \hangindent\CwebNumberListHangindent
+ \leavevmode
+ \cwbb@check_dot{#1}:\quad
+ }
+\endprog
+
+
+\sect The chunks where identifiers are declared are noted with
+underlined numbers. We also must not forget the default declaration of
+|\9|, the tag for the user definable index layout.
+
+\beginprog
+\def\CwebIndexDeclared#1]{$\underline{#1}$}
+\def\9#1{}
+\endprog
+
+
+\sect The list of the refinement names is available in the file
+|\jobname.scn|. The layout is taken from the plain version: ragged
+right, each entry is a paragraph, the different cross reference
+categories are separated by a quad.
+
+We must initialize our whole paragraph layout, as the index might have
+been suppressed and no initialization has happened there then. If
+there was an index, it changed |\parfillskip| and |\*|, let's restore it.
+
+\beginprog
+\def\CwebRefListName{List of Refinements}
+
+\def\CwebRefList{%
+ %% paragraph layout: like in index, but
+ \onecolumn
+ \cweb@list_par_layout
+ \parfillskip\@flushglue % must be reset before section heading
+ \CwebListSection{\CwebRefListName}%
+ \@mkboth{\uppercase{\CwebRefListName}}%
+ {\uppercase{\CwebRefListName}}%
+ \message{\CwebRefListName:}%
+ %% different bindings
+ \def\I{\par \hangindent\CwebNumberListHangindent}%
+ \def\CwebCrossRef##1##2.{\quad {\reset@font\footnotesize ##1~##2.}}%
+ \let\Q\CwebXRCite
+ \let\Qs\CwebXRCites
+ \let\U\CwebXRUse
+ \let\Us\CwebXRUses
+ \let\*\cwbb@change_flag
+ %% Read the list, allow configuration first.
+ \csname CwebRefListHook\endcsname
+ \CwebReadPPList
+ }
+\def\CwebReadPPList{\@input{\jobname.scn}}
+\endprog
+
+
+\sect Extra definitions for the sake of internationalization.
+
+\beginprog
+\def\CwebXRCite{\CwebCrossRef{Cited in chunk}}
+\def\CwebXRCites{\CwebCrossRef{Cited in chunks}}
+\def\CwebXRUse{\CwebCrossRef{Used in chunk}}
+\def\CwebXRUses{\CwebCrossRef{Used in chunks}}
+\endprog
+
+
+
+\chap The end.
+
+Well, after all we're finished with this class. We must not forget to
+restore the underscore catcode.
+
+\beginprog
+\catcode`\_=\CatUsCode
+
+\endinput
+\endprog
+
+
+\sect I would like to thank those who helped me to improve this module.
+% In particular, XXX provided XXXsubstantial parts of the code.
+
+\textsc{Michael M\"uller} and \textsc{Zden\v{e}k Wagner} did thorough
+checks that helped me to improve the alpha test version.
+
+
+
+%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+
+\vskip \PltxPreSectSkip
+
+\begin{rcslog}
+$StyleLog: cweb.doc,v $
+\Revision 3.6 1995/11/30 15:44:53 schrod
+Still didn't work with version \mbox{$\langle$1994/12/01$\rangle$},
+the copy of |\newif| was not complete.\\
+Problem reported by Michael Seel \path|<seel@mpi-sb.mpg.de>|.
+
+\Revision 3.5 1995/11/20 22:34:18 schrod
+Add language support.\\
+Triggered by Christian Kumpf \path|<smurf@igd.fhg.de>| and Andreas
+Scherer \path|<scherer@physik.rwth-aachen.de>|.
+
+Use my `official' (ACM) email address.
+
+\Revision 3.4 1995/11/07 17:55:18 schrod
+ \LaTeX{} \cweb{} should work at least with the last two \LaTeX{}
+versions; make it work with the previous-to-last one, version
+\mbox{$\langle$1994/12/01$\rangle$}. For that, one has to install the
+\mbox{$\langle$1995/06/01$\rangle$} (non-outer) definition of
+|\newif|, and |\hb@xt@| must not be used.\\
+Problem reported by Laurent Desnogues
+\path|<laurent.desnogues@aiguemarine.unice.fr>| and somebody else (XXX
+-- add name).
+
+\Revision 3.3 1995/09/17 14:15:36 schrod
+Initialize list paragraph layout in refinement list, too. With the
+introduction of index suppression, the initialization -- located in
+the index code -- might have been skipped.
+
+\Revision 3.2 1995/09/15 10:22:59 schrod
+The \cls{cweb} class needs at least \LaTeX{} version
+\mbox{$\langle$1994/12/01$\rangle$}.\\
+Problem reported by John S. Robinson \path|jsrobin@umiacs.umd.edu|.
+
+\Revision 3.1 1995/09/12 23:00:48 schrod
+Moved all code that does the actual typesetting of \cweave{} tags and
+is therefore also needed for a |cweb| environment. It's now an own
+module named \pkg{cwebbase}.
+
+\Revision 2.8 1995/09/12 11:44:33 schrod
+Standard base classes with chapters (i.e., report \& book) produced an
+error at the start of the index: |\chapter| may not be called in the
+optional argument of |\twocolumn|. Add a workaround to |\twocolumn|.\\
+Problem reported by Christopher Higgins \path|<C.Higgins@cit.gu.edu.au>|.
+
+\Revision 2.7 1995/08/29 17:32:36 schrod
+Add hook for refinement list.
+
+Allow internationalization of refinement list.\\
+(Problem reported by Christian Kumpf \path|<kumpf@igd.fhg.de>|.)
+
+Cweb bundle, version 1.0.
+
+\Revision 2.6 1995/08/29 15:19:50 schrod
+Added (hint to non-existence of) description of the internal interface.
+
+\Revision 2.5 1995/08/29 02:07:26 schrod
+Discard dependencies on 10\,pt fonts.
+
+Support suppression of format directives.
+
+\Revision 2.4 1995/08/27 19:31:43 schrod
+Discard superfluous space at the top of index. Spacing in refinement
+heading was wrong. Repair test on empty auxilliary files. Make
+section heading for index and refinement list divisions (more)
+configurable. Suppression of index and reference list is supported.
+
+Make configuration of change flag easier.
+
+Put section title of changed chunks list in marks, and tell the user
+that it got typeset.
+
+\Revision 2.3 1995/08/27 17:24:46 schrod
+Make usage of baseclass with chapters work.
+
+\Revision 2.2 1995/08/27 13:26:22 schrod
+Add possibility to suppress change hints. Suppression of unchanged
+chunks suppress change hints as well, they are meaningless as all
+printed chunks are changed by definition.
+
+Realize dependencies in options by |\ProcessKeyValOptionsHook|.
+Illegal values for enumeration options don't alter the current value
+any more.
+
+List of changed chunks is an unnumbered section now.
+
+\Revision 2.1 1995/08/25 19:11:18 schrod
+Add keyword-value option style, with new \pkg{keyvald} package.
+
+Hierarchic strucutures are supported now, in addition to the flat
+structure of the beta-test version. One can choose with an option. For
+that step, the terminology was cleaned up, too: Chunks are not named
+sections any more. (That change involved reimplementation of almost
+all the structure and toc stuff.)
+
+The chunk number supplied by \cweave{} is used now, not some computed
+number. Change flags are printed, too.
+
+One can suppress output of unchanged sections.
+
+One can select the baseclass with an option. That may be used to use
+\cls{report} or \cls{book} to get chapter-style layout. Of course,
+using an arbitrary baseclass is dangerous, it must conform to the
+conventions of \LaTeX{} standard classes.
+
+\Revision 1.13 1995/08/08 00:14:29 schrod
+Updated to \LaTeXe{}, the |cweb| style is now a document class. Used
+my standard templates for that, no changes in functionality.
+
+\Revision 1.12 1993/08/10 14:15:43 schrod
+New page on main section only if group level $<$ |\cwebSecNoEject|.
+Default for the latter is 3.
+
+Document that logos will not be defined in this style file.
+
+Copy of plain macros for |\CwebNumber| does not work. Repaired the
+most important one (subscript must be accessed via |\sb|).
+Incompatibility to NFSS will be addressed later.\\
+(Problems reported by Zden\v{e}k Wagner \path|<wagner@csearn.bitnet>|.)
+
+\Revision 1.11 1993/08/10 11:21:07 schrod
+Reference to section number does not render a period after the
+number any more.
+
+\Revision 1.10 1993/08/09 20:08:20 schrod
+|\cweb@cweave_bindings| is now a no-op if \cweave{} bindings are in
+effect already.\\
+(Problem reported by Michael M\"uller \path|<mimu@mpi-sb.mpg.de>|.)
+
+\Revision 1.9 1993/08/09 18:05:28 schrod
+Left shift operator wasn't defined correctly.\\
+(Problem reported by Michael M\"uller \path|<mimu@mpi-sb.mpg.de>|.)
+
+\Revision 1.8 1993/06/15 15:22:30 schrod
+\textbf{Version 0.2:} Moved from |pub/incoming| to official place, in
+Literate Programming Archive.
+
+\Revision 1.7 1993/06/15 13:25:48 schrod
+First attempt to make it a style: |article| is imported. Warns the
+user if he tries to use |cweb| as a style option.
+
+\Revision 1.6 1993/06/15 08:49:23 schrod
+|\cweb@check_dot| must not evaluate its argument in an |\edef|, this
+causes problems if a |\PB| is within. Now I try hard not to evaluate any
+tokens outside of my control.
+
+Can use |\@defpar| for an empty line in |\cweb@has_entries|, don't
+need an own macro.
+
+\Revision 1.5 1993/06/14 17:50:30 schrod
+Handle a missing |\con|: |\inx| and |\con| are now never looked at.
+Instead |\fin| checks for the existence of the lists and typeset them
+if there are entries available.
+
+\Revision 1.4 1993/06/14 15:54:18 schrod
+Made it work with older versions of \LaTeX{}, too. The |\enddocument|
+implementation of cweb.sty depends on the |\end| implementation; check
+if the current implementation is one we know of (currently: 91-01-14
+and 92-03-25).\\
+(Problem reported by Michael M\"uller \path|<mimu@mpi-sb.mpg.de>|.)
+
+Add FSA diagram about processing states. CR state is also switched
+to from \TeX{} state (that happens with |\ch| at the document end).
+
+\Revision 1.3 1993/05/13 17:51:21 schrod
+Refinements may also be filenames (`|@(|'). Then the complete name
+consists of a |\.| macro call, which is handled now.\\
+(Problem reported by Michael M\"uller \path|<mimu@mpi-sb.mpg.de>|.)
+
+Made the detection of `|@.|' index entries more robust.
+
+\Revision 1.2 1993/05/12 18:28:59 schrod
+Adapted to recent changes of \cweave{} (of April 93):
+
+Main sections have a group level, represented in the table of
+contents. This changed the complete implementation of section tags.
+
+New C token cseqs: |\Z| and |\MRL|, implemented as |\CwebLe| and
+|\CwebCombinedOp|.
+
+\Revision 1.1 1993/04/09 15:00:37 schrod
+Initial revision
+
+\end{rcslog}
+
+
+
+\end{document}
+
+
+
+%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+Local Variables:
+mode: LaTeX
+TeX-brace-indent-level: 4
+indent-tabs-mode: t
+TeX-auto-untabify: nil
+TeX-auto-regexp-list: LaTeX-auto-regexp-list
+compile-command: "make cweb.cls"
+End: