% $StyleId: rcs.doc,v 2.10 2003/02/02 20:42:38 schrod Exp $ %---------------------------------------------------------------------- % Written by Joachim Schrod . % Copyright conditions see below. % % LaTeX package rcs % to use RCS tag values in one's document, and typeset revision logs % % [LaTeX in MAKEPROG] % (history at end) % If you have received this style file without the user manual (in the % file rcs-user.tex or the respective DVI 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{rcs-doc} \usepackage{path} \usepackage{alltt} \usepackage{fullpage} \RCS $StyleDate: 2003/02/02 20:42:38 $ \begin{document} \title{The \texttt{rcs} Package\\ {\large (Implementation)}% } \author{% % LaTeX does not discard unnecessary glue... Joachim Schrod% \thanks{% \protect\raggedright R\"odermark, Germany. Email: \texttt{jschrod@acm.org}% }% } \date{% Revision \RCSStyleRevision\\ (as of \RCSStyleDate)% } \maketitle % ------------------------------------------------------------ % % subdocument: The user interface of rcs.sty % \input{rcs-user} % % ------------------------------------------------------------ % % subdocument: The internal interface % \input{rcs-conf} % % ------------------------------------------------------------ \chap Implementation. This module reserves the namespace |rcs|. \beginprog \ifx \rcs@loaded\undefined \def\rcs@loaded{$StyleRevision: 2.10 $} \else \PackageWarningNoLine{rcs}% {Some other package already uses namespace `rcs'} \fi \endprog \sect Let's identify this package. Now I'm caught in a catch-22 situation. The code below---part of my standard templates for \LaTeX{} packages---uses principles that are explained later in this document. In particular, the explanation how the information from the following RCS fields are gathered starts at chunk~\ref{chunk:value-split}. There you'll find also the explanation for the funny strings behind the RCS fields. Actually, I would not need them any more, they make the initial document (before the first check-in) work. But they won't be discarded in this package source, as they'll demonstrate another utilization of RCS at work. \beginprog \begingroup \def\RCSPackage#1#2 $#3: #4 #5\endRCS $#6: #7 #8\endRCS{% \def\date{#4}\def\id{v#7}% \ProvidesPackage{#1}[\date\space\id\space #2]% } \RCSPackage{rcs}{typeset info from RCS fields} $StyleDate: 2003/02/02 20:42:38 $: 9999/00/00 \endRCS $StyleRevision: 2.10 $: 0.0 \endRCS \endgroup \endprog \sect This module is supported. Send bug reports, comments, and repairs. The reference version may be retrieved via anonymous ftp from CTAN, directory \path|/tex-archive/macros/latex/contrib/supported/rcs|. \sect This is freely distributable software; you can redistribute it and/or modify it under the terms of the GNU General Public License (version~2) as published by the Free Software Foundation. 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 {\sc 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 option the at sign is a letter anyhow; so we can use the ``private'' Plain and \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. \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 \chardef\CatUsCode=\catcode`\_ \fi \catcode`\_=\CatLetter % top level macro file \endprog \chap Parsing RCS fields. Each RCS field consists of a keyword and an optional value. The value is delimited by `\verb*|: |' from the keyword, this delimiter is missing if no value does exist. At the end of a value there's always a space. I.e., the possible input looks like one of the two following lines: % \begin{quote} \itshape |\RCS $|Keyword\/|$|\\ |\RCS $|Keyword\/\verb*|: |value\/\verb*| $| \end{quote} % where \textit{Keyword} and \textit{value} may be arbitrary strings. In fact, we assume that \textit{Keyword} is a sequence of letters---only then the creation of a cseq |\RCS|\textit{Keyword} is sensible. Our implementation will not produce an error with an obscure keyword, but the result will not be of use for the author. We must not depend on a fixed set of keywords, this style shall be usable with a configurable RCS version. In such a version one can bind arbitrary keywords to the information available in RCS\@. This means we will parse all RCS fields with the same macro, |\rcs_split_field|. This macro will place its results in two macros: |\RCS_keyword| and |\RCS_value|. Afterwards we can bind the expansion of |\RCS_value| to the target cseq. \sect |\RCS_value| will be set to the expansion of |\RcsEmptyValue| if there was no RCS value available, i.e., if the field was not processed by RCS yet. This way the values of non-expanded RCS fields can be configured. The default value is the empty token list. \beginprog \def\RcsEmptyValue{} \endprog \sect For those fields where a special handling is demanded, a cseq |\RcsHandle|\textit{Keyword} must exist. This cseq must handle the parsing of the value. Usually this means that it will create a macro named |\RCS|\textit{Keyword} with a transformed (or filtered) value. If you happen to write a handler you might want to look at the definition of |\RcsHandleDate| starting at section~\ref{sec:date} If a special handler for a field exists, the original value is placed automatically in the macro |\RCSRaw|\textit{Keyword}. This is done before the handler is called. \sect \label{chunk:value-split} The macro |\rcs_split_field| will split the field into |\RCS_keyword| and |\RCS_value|. This split is done by the argument parsing capability of \TeX{}: We can separate parameters by token lists and \TeX{} will use a shortest pattern match to determine the arguments. Let's look what kind of pattern we can use. As an example we use the RCS field `\verb*|$Revision: 2.11 $|', i.e., the unexpanded field is `|$Revision: 2.11 $|'. The macro |\RCS| will use dollars to delimit its argument, we don't need to worry for them. Eventually we'll have to discard the space of the expanded field's value---after all, |\RCSRevision| shall expand to `1.1', not to `1.1~'. First, we can see that the token list $\it (colon, space)$ may be used as the seperator of keyword and value: % \begin{quote} \macroCall{\\rcs_split_field \[Revision\(: \)1.1 \]\\end_value} \end{quote} % ($d_i$ is the separating token list between parameter $i$ and~$i+1$. |#1| denotes the argument of |\RCS|. |\end_value| is a stop token, used to delimit the last parameter; we have to supply this token at the call. In fact, all tokens marked by $s$ and the dotted line---i.e., those after |#1|---are supplied by us.) When we want this token list as a separator, we have to supply it for the unexpanded field, since there is no $\it (colon, space)$ in there: % \begin{quote} \macroCall{\\rcs_split_field \[Revision\]\(: \)\\end_value} \end{quote} % Then the call with the expanded field gets % \begin{quote} \macroCall{\\rcs_split_field \[Revision\(: \)1.1 \]: \\end_value} \end{quote} Well, we want neither the space nor the colon and the other space after the value---we have to get rid of them. We can utilize the fact that no RCS field will ever have the token list $\it (space, colon)$ in it. This means we can use it as a terminator for the RCS value: % \begin{quote} \macroCall{\\rcs_split_field \[Revision\(: \)1.1\( \]:\) \\end_value} \end{quote} % Please note, one token of the delimiter~$d_2$ comes from the |#1|~argument, the other from the tokens we'll add at the call. Let's check our unexpanded field again---we don't have $d_2$ there. We have to add it by the caller: % \begin{quote} \macroCall{\\rcs_split_field \[Revision\]\(: \)\{\}\( :\)\\end_value} \end{quote} % (Without the empty group the two spaces would be combined into one token.) Going back to the expanded field, we notice that these additional tokens do not harm us---we'll ignore everything after $d_2$ anyhow. % \begin{quote} \macroCall{\\rcs_split_field \[Revision\(: \)1.1\( \]:\) \{\} :\\end_value} \end{quote} We can take another look at the whole affair and point out the parameters we get for the split. Remember, that spaces after a control word belong to this word. First, the expanded field: % \begin{quote} \macroCall [p]{\\rcs_split_field \[\(Revision\): \(1.1\) \]:\( \{\} :\)\\end_value} \end{quote} % Then the unexpanded one: % \begin{quote} \macroCall[p]{\\rcs_split_field \[\(Revision\)\]: \(\{\}\) :\(\|\)\\end_value} \end{quote} % ($p_3$, the third parameter will be empty in this case.) \medskip \noindent As explained above: if the value does not exist, a default value is supplied. \beginprog \def\rcs_split_field #1: #2 :#3\end_value{% \def\RCS_keyword{#1}% \def\RCS_value{#2}% \ifx \RCS_value\empty \let\RCS_value\RcsEmptyValue \fi } \endprog \sect |\RCS| triggers the split. After it has occured, we can bind the expansion of |\RCS_value| to a new name. This name is constructed from the prefix |RCS|, the string |Raw| iff a handle for this field exists, and the keyword itself. Afterwards we call the handle, on non-existance this will be a no-op. Gathering the field is a bit more complicated, as we need to pay attention to underscores in file names. \beginprog \def\RCS{% \bgroup \catcode`\_ \CatActive \RCS_get_argument } \def\RCS_get_argument $#1${% \gdef\RCS_argument{#1}% \egroup \RCS_process } \def\RCS_process{% \expandafter\rcs_split_field \RCS_argument: {} :\end_value \expandafter\let \csname RCS\rcs_raw_if_handle\RCS_keyword\endcsname \RCS_value \csname RcsHandle\RCS_keyword\endcsname } \def\rcs_raw_if_handle{% \@ifundefined{RcsHandle\RCS_keyword}{}{Raw}% } \endprog \chap Transforming the value of \Date{} fields. \label{sec:date} When we implement the special handling for the \Date{} field, we take care that we don't rely on the fact that the keyword is |Date|. Certain RCS versions (one of them created at our site) allow the configuration of the used keywords. I.e., there might be another keyword used for this field, but we still want to use the functionality provided here. For example, let's assume the keyword of the \Date{} field is |StyleDate|. Then the assignment % \begin{quote} |\let\RcsHandleStyleDate=\RcsHandleDate| \end{quote} % is all that's needed to turn on the `date special handling' for this field. (In fact, that's done in this style file you're reading.) \sect The special handler for \Date{} fields sets up |\RCS|\textit{Keyword} (|\RCSDate| usually) to produce a textual representation of the date part of the \Date{} value in a |\today|-like format. Thus the constructed tag can be used in many circumstances, e.g., as the argument of the |\date| tag on the titlepage. In addition, |\RCSTime| is defined; it expands to the checkin time. If the value is empty, i.e., if there's no \Date{} value from RCS available, we use the current date. Later we might supply the current time, too---currently |\RCSTime| expands to an empty token list in such a case. Of course, this whole special handling is only done if |\today| is defined in some way; otherwise we just handle the \Date{} field like every other RCS field: The value is stored unfiltered in |\RCS|\textit{Keyword}. \beginprog \def\RcsHandleDate{% \ifx \today\undefined \expandafter\let \csname RCS\RCS_keyword\endcsname \RCS_value \else \ifx \RCS_value\RcsEmptyValue \@namedef{RCS\RCS_keyword}{\today}% \let\RCSTime\empty \else \expandafter\rcs_set_date \RCS_value\end_date \fi \fi } \endprog \sect If the value is not empty we split it using the argument parsing mechanism of \TeX{}. Then we'll define the value using the current definition of |\today|. This way language-specific styles (e.g., from the \textsf{babel} system) may supply their own way to present a date. \beginprog \def\rcs_set_date #1/#2/#3 #4\end_date{% \begingroup \day #3 % <-- space! \month #2 % <-- space! \year #1 % <-- space! \expandafter\xdef \csname RCS\RCS_keyword\endcsname {\today}% \endgroup \def\RCSTime{#4}% } \endprog \chap Typesetting revision logs. If a revision log is to be typeset, we demand a special format. Namely, it is to be enclosed in the |rcslog| environment. This environment has an optional parameter, the configuration tokens. The contents of the environment is the \Log{} field followed by a list of revision entries. The \Log{} field must be on a line of its own. Each revision entry is introduced by a line consisting of the tag |\Revision|, the revision number, the checkin date, the checkin time, and the author's user id. I.e., the input looks like the following example sketch, where italic material is substituted by current values. % \begin{quote} \chardef\\=`\\ % just for the example \chardef\{=`\{ \chardef\}=`\} \begin{alltt}\frenchspacing \\begin\{rcslog\} $Log: rcs.doc,v $ Revision 2.11 2003/02/02 22:30:22 schrod Removed dependeny on \LaTeX2.09 font selection commands. (Though they're still used in the package documentation, as convenience.) \\Revision {\it revision date time uid} \qquad {\it log text\/} \quad \([\,\ldots\) {\it perhaps more entries\/} \(\ldots\,]\) \\end\{rcslog\} \end{alltt} \end{quote} We typeset the revision log similar to a |description| environment. Each revision entry is an item. \sect The tag for the log intro is implemented as a token register. In fact, |\rcsLogIntro| is a hook. \beginprog \newtoks\rcsLogIntro \rcsLogIntro={} \endprog \sect The author names are stored in a table named |rcs_author|. We supply the lookup macro |\RCS_get_author|, which expands to the table value or to the table key itself. \beginprog \def\rcsAuthor#1#2{\@namedef{rcs_author:#1}{#2}} \def\RCS_get_author#1{% \@ifundefined{rcs_author:#1}{#1}{\@nameuse{rcs_author:#1}}% } \endprog \sect Since this style option shall be usable with many styles, we must provide means to configure its behaviour. In particular, it must be possible to configure the layout. What remains fixed, is the layout of the log like a |description| list. I might add this flexibility if there is enough interest (or if somebody sends me tested changes). The first thing to parametrize is the overall layout and the header. This is set up by |\RcsLogStyle|. This cseq \emph{must} define |\RcsLogHeading|, which is the macro to set the header. E.g., here one could insert a |\newpage| to start a new page for the revision log. In addition, we reduce the base size of the log. That's the reason why these two configurations are combined in one macro---if you change one, you also have to consider to change the other. \beginprog \def\RcsLogStyle{% \def\RcsLogHeading{\subsubsection*}% \footnotesize } \endprog \sect The header itself is created by |\rcs_log_hdr|, which is called with the expanded \Log{} value as its argument, terminated by |\end_value|. We use \TeX{}'s parameter parsing mechanism to chop off the tail of the value---if the value exists, it always ends in~`|,v|'. Since only an existing value ends in~`|,v|', we have to supply a fitting empty value. We will use |\rcs_empty_log_value| for that. The empty log value gets a bit more complicated: If we have an empty \Log{} value, we will not have any revision entries. (Well, except in pathological cases. Then we will produce spurious text.) Missing revision entries are missing items; since \LaTeX{} will complain about a missing |\item|, we must provide one to stop it from doing so. But here is no place for such tokens, the list environment hasn't started yet. We will set the flag |@rcs_empty_log@| and check this flag after we invoked the list. \beginprog \def\rcs_log_hdr#1,v\end_value{% \@rcs_empty_log@false \RcsLogHeading{\RcsLogHeadingName #1}% \the\rcsLogIntro } \def\rcs_empty_log_value{% \global\@rcs_empty_log@true % that's most probably in a group! \RcsUnknownFile % text for unknown file name ,v% % assert calling env } \newif\if@rcs_empty_log@ \endprog \sect The user might want to adapt the text output. He can do so by redefining |\RcsLogHeadingName|, |\RcsUnknownFile|, and |\RcsEmptyLog|. \beginprog \def\RcsLogHeadingName{Revision Log for \ttfamily} % file name in monospace \def\RcsUnknownFile{\rmfamily\mdseries $\langle\,$Unknown file name$\,\rangle$} \def\RcsEmptyLog{{\itshape No log entries available.}} \endprog \sect OK, we have fixed the header, let's address the entries. Each entry is typeset by |\RcsLogRevision| which is called with four arguments: (1)~the revision number, (2)~the checkin date, (3)~the checkin time, and (4)~the uid of the author. Such an entry is inserted as an item in the list. The uid is transformed to an author name, if the respective information is available. The checkin timestamp is mapped to a neater appearance by |\RcsLogDate| and |\RcsLogTime|. \beginprog \def\RcsLogRevision#1#2#3#4{% \item [Revision #1]% (created at \RcsLogDate #2\endDate\space \RcsLogTime{#3} % <-- space! by \RCS_get_author{#4})\\\relax } \endprog \sect |\RcsLogDate| sets the date. We can use \TeX{}'s parameter parsing mechanism to separate the date value (passed as the argument) in its sub-values. The code is very similar to |\rcs_set_date|. |\RcsLogTime| sets the time. As explained in the user manual, the time is ignored usually. (I don't think that this info is relevant in a printout.) Here we assume something about the context it's expanded in: Spaces will be before and after the tag. One of them must be discarded. The user option |\settime| is implemented here as a private macro, too. \beginprog \def\RcsLogDate #1/#2/#3\endDate{% \begingroup \day #3 % <-- space! \month #2 % <-- space! \year #1 % <-- space! \today \endgroup } \def\RcsLogTime#1{\ignorespaces} \def\rcs_settime{% \def\RcsLogTime##1{[##1]}% } \endprog \sect Well, that are all possibilities to configure the revision log. Let's put together the pieces. Within the |rcslog| environment the tags |\Revision| and |\settime| are used, they are implemented in the |rcs| namespace. At the start, we use the optional argument of the environment to configure it. |\settime| might be placed here, or font switches, etc. The next entity shall be the \Log{} field. In fact, it shall be on the next line. If an optional environment argument was there, the newline is still in the input and must be discarded before we can use |\RCS| to parse the field. \beginprog \def\rcslog{% \@ifnextchar[% % ] (Emacs) \rcslog_configure {\rcslog_configure[]}% } \def\rcslog_configure[#1]{% \let\Revision\rcslog_revision \let\settime\rcs_settime \RcsLogStyle #1% \afterassignment\rcslog_skipcr \let\next } \endprog \sect There are users---programmer's life would be so nice without them. An empty line (i.e., a |\par|) might have been inserted between |\begin{rcslog}| and the \Log{} field. There might be even other garbage between them. We have to check these possibilities and provide proper responses. \beginprog \def\rcslog_skipcr{% \ifx \next $% \def\next{\rcslog_field $}% \else\ifx \next\par \let\next\rcslog_field \else \PackageError{rcs}{Missing RCS Log field in environment}{% The RCS Log field must be the very first text in the rcslog\MessageBreak environment. I will discard the rest of your input line and try to\MessageBreak recover then. If you think this won't work, exit now.\MessageBreak Otherwise press and cross your fingers!} \let\next\rcslog_discard_line \fi\fi \next } \endprog \sect This is the code for the luser. He provided tokens we didn't wait for---we discard them and check anew if we've got it right now. (Remember, we're waiting for the `|$|' of the \Log{} field.) %$ Emacs \beginprog \def\rcslog_discard_line{% \begingroup \obeylines \rcslog_gobble_line } \begingroup \obeylines \gdef\rcslog_gobble_line #1^^M{% \endgroup% \afterassignment\rcslog_skipcr% \let\next% } \endgroup \endprog \sect Yep, we've reached a dollar; let's hope that it's our \Log{} field. It must be on one line, this line is fetched first and then analyzed. \beginprog \def\rcslog_field{% \begingroup \obeylines \catcode`\_ \CatActive \rcslog_get_field } \begingroup \obeylines \gdef\rcslog_get_field $#1$^^M{% \gdef\RCS_argument{#1}% \endgroup% \rcslog_set_field% } \endgroup \endprog \sect When we analyze the \Log{} field, we must supply the default value. Within the log list there will be no other RCS fields, therefore we can redefine |\RcsEmptyValue| in this environment. The header gets passed the expanded \Log{} value, as explained above. We use |\RCS_value| instead of |\RCSLog| since we don't know which RCS keyword is used. At last, we start the list and check the |@rcs_empty_log@| flag, perhaps we have to supply a dummy |\item|. \beginprog \def\rcslog_set_field{% \let\RcsEmptyValue\rcs_empty_log_value \RCS_process \expandafter\rcs_log_hdr \RCS_value\end_value \list{}\RcsLogListStyle % empty label, layout configurable \if@rcs_empty_log@ \item []\RcsEmptyLog % supply missing item \fi } \endprog \sect The setup of the log list is done by a macro from the protected interface, the user shall be able to change it. In the default supplied, there is no skip between paragraphs, but a quad indentation. |rcslog| is here no list environment in the sense like |itemize| or |enumerate|, therefore a usage of these environments within an entry shall result in top-level item markers. \textbf{Caveat}:\quad Since we globally set |\@listdepth| at the start of the environment, we must not execute |\endlist| that decrements it. Instead we use |\endtrivlist| for the termination of the |rcslog| environment. It might be, that this must be changed if |\RcsLogListStyle| is changed. \beginprog \def\RcsLogListStyle{% \global\@listdepth\z@ \labelwidth\z@ % item label w/o fixed size \itemindent -\leftmargin % move label before start of line \advance\itemindent \labelsep % label was moved too much \def\makelabel##1{\bf ##1}% % how to typeset labels \parsep\z@skip % no space between pars in items \listparindent 1em % <-- space! indent pars in items }% \let\endrcslog\endtrivlist \endprog \sect The arguments of |\Revision| are separated by spaces and terminated by the end of the line. We have to fetch them, then we can pass them to |\RcsLogRevision| which does the real work. \beginprog \def\rcslog_revision{% \begingroup \obeylines \rcslog_get_revinfo } \begingroup \obeylines \gdef\rcslog_get_revinfo #1 #2 #3 #4^^M{% \endgroup% \RcsLogRevision{#1}{#2}{#3}{#4}% } \endgroup \endprog \chap Convenience tags. As a shortcut we support the direct supply of the checkin date for the titlepage material. Here we also take care not to depend on the keyword. \beginprog \def\RCSdate $#1${% \RCS $#1$% \expandafter\date \expandafter{\csname RCS\RCS_keyword\endcsname}% } \endprog \sect |\RCSID| parses the RCS field and redefines the footline. This tag is from Nelson's |rcs| style. In his style he defines that the RCS field is passed as an argument to the macro. To be upward compatible, we have to support this as well: We look at the next token, if it is an opening brace we fetch the param and trigger the `real' invocation by |\rcs_id|. If no brace follows, we call |\rcs_id| immediately. \beginprog \def\RCSID{% \@ifnextchar\bgroup \rcsid_get_field \rcs_id } \def\rcsid_get_field#1{\rcs_id #1} \endprog \sect As a side effect, we'll produce the normal value tags. When we define the footline, we must assert that |\RCS_keyword| and |\RCS_value| are expanded. (They will get new bindings at the next invocation of |\RCS|.) All other macros must not be expanded, they shall be used in the context of the footline, not in our definition context. We utilize |\protect| for that, |\framebox| is the only macro that must be protected. \beginprog \def\rcs_id $#1${% \RCS $#1$% \begingroup \let\protect\noexpand \xdef\@oddfoot{% \reset@font \protect\framebox[\textwidth]{\RCS_keyword: \RCS_value}% }% \endgroup \global\let\@evenfoot\@oddfoot } \endprog \sect |\RCSdef| just outputs its parameter before it's passed to |\RCS|. This tag is from Tom Verhoeff's |rcs| style. In fact, we're not fully compatible: Tom did output `RCS keyword string'. I substitute this by `RCS field', this is consistent with the rest of this style's description. \beginprog \def\RCSdef $#1${% \typeout{RCS field: $#1$}% \RCS $#1$% } \endprog \chap The end. Well, after all we're finished with this style. 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 style. % In particular, XXX provided XXXsubstantial parts of the code. \textsc{Marcus Speh} pointed out that Nelson's style option exists. \textsc{Lee Wittenberg} pointed out that Tom style option exists---Tom posted the style option on |comp.programming.literate| afterwards. I borrowed the idea of appending a second colon to the call of |\rcs_split_field| from Tom's style. \textsc{Harald Fuchs} asked for the possibility to configure the log list layout. \textsc{Dave Love} brought the problem with the |-kv| option of |co| to my attention. It's a pitty that I haven't found a solution yet. (I mean, one that convinces me.) \textsc{Christine Detig} told me where my documentation was uncomprehensible. In particular, you won't have the visual explanation of |\rcs_split_field| without her. barbara beeton proofread the user manual. In addition, she pointed out where a non-programmer does not understand it. After 7.5 years of quiteness, \textsc{Martin Kneißl} emerged and asked for support of underscores in file names. He got it, and I used the opportunity to update contact information. \vskip \PltxPreSectSkip \rcsLogIntro{Of course, we shouldn't stop without the RCS log of this style file.} \begin{rcslog} $StyleLog: rcs.doc,v $ \Revision 2.10 2003/02/02 20:42:38 schrod Support underscores in file names in RCS field values. E.g., in \rcsField{Id} or \Log{}. Update my contact information. \Revision 2.9 1995/08/02 12:09:07 schrod Transformed this style option into a \LaTeXe{} package. Updated to \LaTeXe{}. \Revision 2.8 1993/11/10 12:29:49 schrod Added acknowledgement of barbara. \Revision 2.7 1993/11/08 20:15:15 schrod Added possibility to configure the layout of the log list (|\RcsLogListStyle|). Now the log list is on level~0. Thereby item lists or enumerations in revision entries get first-level labels. This has consequences for the configuration, this implementation strategy must therefore be specified in the internal interface description. \Revision 2.6 1993/11/03 20:04:57 schrod Cleaned up for distribution: Added email address to each document, added copyright info to |rcs.doc|, added acknowledgements. Checked my English and the spacing. Explained the restriction concerning the |-kv| option of |co|. Rewrote the explanation to |\rcs_split_field|. Added the visual clues to explain the (mis?)usage of macro expansion and pattern matching. \Revision 2.5 1993/11/02 20:02:51 schrod Implemented |\RCSdef|. Deleted unnecessary macro that did discard the space of an RCS value. Moved this in the parameter context of |\rcs_split_field|. Added an explanation to |\RCS| where it is called. \Revision 2.4 1993/11/02 18:44:49 schrod Implemented |\RCSID|.\\ Need |\reset@font| now, for the definition of the footline in |\RCSID|. Paragraphs in revision logs are not separated by vertical space any more. They are indented instead. I use the ``classic'' indentation of one quad. \Revision 2.3 1993/11/01 19:49:49 schrod Added subdocument about the internal interface. Cleaned the cseq names: All public names start with `|RCS|' or `|rcs|', protected names start with `|Rcs|', both have only letters. Observer names start with `|RCS|', internal names with `|rcs|', both feature non-letters in the names. \Revision 2.2 1993/11/01 19:20:58 schrod Cleaned up the documentation. The enhancement of the |rcslog| environment was just appended, now it's integrated in the whole document. Improved the explanation of the implementation: the valid input is now explained more clearly. Use |rcs-doc.sty| for documentation of |rcs| style. It does not only load style options, but does also define |\RCSStyleRevision| for access of the style's revision. Added hint that the user manual is urgently needed.\\ The load tag is now in the protected interface.\\ Commented out section with definition of |\reset@font|, that cseq is not needed in the implementation.\\ OALD tells me to use ``heading'' instead of ``header.'' \Revision 2.1 1993/10/29 19:07:38 schrod Reorganized the source structure---the style gets too large to keep everything in one directory. I use the usual setup for my larger styles. Changed the RCS keywords. All keywords in use now start with `|Style|'. Stuff in the text which looks like RCS fields, but where the keywords do not start with `|Style|', are examples! \Revision 1.3 1993/10/29 18:08:19 schrod Each field value supplied by RCS ends with a space; this space is discarded. Supports the typesetting of revision logs. The overall documentation is not yet fully integrated. Of course, our own revision log is typeset at the end, too.\\ But we'll have to use other RCS keywords for this document in the future. E.g., I can't use original RCS keywords currently in the examples without much hassle. An hook for the configuration of this style option is provided. \Revision 1.2 1993/09/08 15:49:00 schrod Use canonical names for category codes. \Revision 1.1 1993/09/03 21:01:29 schrod Re-implemented |rcs| style option. Made it a documented option. \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 End: