fcolumn (28sep15)

git-svn-id: svn://tug.org/texlive/trunk@38489 c570f23f-e606-0410-a88d-b1316a301751

1 files changed, 478 insertions, 310 deletions

diff --git a/Master/texmf-dist/source/latex/fcolumn/fcolumn.dtx b/Master/texmf-dist/source/latex/fcolumn/fcolumn.dtx
index 3e83646dfba..182e45670c0 100644
--- a/Master/texmf-dist/source/latex/fcolumn/fcolumn.dtx
+++ b/Master/texmf-dist/source/latex/fcolumn/fcolumn.dtx
@@ -1,60 +1,22 @@
 % \iffalse meta-comment
-%
 % Copyright (C) 2015 by Edgar Olthof
 %
 % This file may be distributed and/or modified under the
 % conditions of the LaTeX Project Public License, either
-% version 1.2 of this license or (at your option) any later
+% version 1.3 of this license or (at your option) any later
 % version.  The latest version of this license is in
 %
 %    http://www.latex-project.org/lppl.txt
 %
-% and version 1.2 or later is part of all distributions of
+% and version 1.3 or later is part of all distributions of
 % LaTeX version 1999/12/01 or later.
 %
 % \fi
 % \iffalse
 %<package>\NeedsTeXFormat{LaTeX2e}
 %<package>\ProvidesPackage{fcolumn}
-%<package>          [2015/03/07 v1.0 automatic currency package]
+%<package>          [2015/09/20 v1.1.1 Automatic Currency Package (EO)]
 %<package>\RequirePackage{array}
-%<*batchfile>
-\begingroup
-%%
-%% Copyright (C) 2015 by Edgar Olthof
-%%
-%% This file may be distributed and/or modified under the
-%% conditions of the LaTeX Project Public License, either
-%% version 1.2 of this license or (at your option) any later
-%% version.  The latest version of this license is in
-%%
-%%    http://www.latex-project.org/lppl.txt
-%%
-%% and version 1.2 or later is part of all distributions of
-%% LaTeX version 1999/12/01 or later.
-%%
-\input docstrip.tex
-\keepsilent
-\preamble
-
-This is a generated file.
-
-Copyright (C) 2015 by Edgar Olthof
-
-This file may be distributed and/or modified under the
-conditions of the LaTeX Project Public License, either
-version 1.2 of this license or (at your option) any later
-version.  The latest version of this license is in
-   
-   http://www.latex-project.org/lppl.txt
-
-and version 1.2 or later is part of all distributions of
-LaTeX version 1999/12/01 or later.
-\endpreamble
-\askforoverwritefalse
-\generate{\file{fcolumn.sty}{\from{fcolumn.dtx}{package}}}
-\endgroup
-%</batchfile>
 %<*driver>
 \documentclass{ltxdoc}
 \usepackage{fcolumn}
@@ -68,9 +30,7 @@ LaTeX version 1999/12/01 or later.
 \end{document}
 %</driver>
 % \fi
-%
-% \CheckSum{0}
-%
+% \CheckSum{618}
 % \CharacterTable
 %  {Upper-case    \A\B\C\D\E\F\G\H\I\J\K\L\M\N\O\P\Q\R\S\T\U\V\W\X\Y\Z
 %   Lower-case    \a\b\c\d\e\f\g\h\i\j\k\l\m\n\o\p\q\r\s\t\u\v\w\x\y\z
@@ -89,95 +49,233 @@ LaTeX version 1999/12/01 or later.
 %
 % \changes{v0.1}{2015/02/06}{First working version}
 % \changes{v1.0}{2015/03/07}{Three-argument version is working properly}
+% \changes{v1.1}{2015/09/14}{Automatic checking of column balance performed when number of
+%   F-columns is even (behaviour can be overridden).  Empty entries are now recognised and 
+%   correctly treated as such, except for the one ended by the double backslash. Not serious;
+%   workaround possible. Furthermore optimisation of code: minimised the number of private 
+%   counts and resetting of column counter done in a nicer way.}
+% \changes{v1.1.1}{2015/09/27}{Installation procedure changed from .ins-in-.dtx to separate
+% .ins and .dtx after discussion with Karl Berry as well as some minor code improvements.}
 % \GetFileInfo{fcolumn.sty}
-% \DoNotIndex{\fi,\let,\@ne}
+% \DoNotIndex{\@arstrut,\@arstrutbox,\@depth,\@firstampfalse,\@firstamptrue,\@gobble,\@halignto}
+% \DoNotIndex{\@height,\@mkpream,\@ne,\@nextchar,\@preamble,\@sharp,\@tempdima,\@tempswafalse}
+% \DoNotIndex{\@tempswatrue,\@\temptokena,\@tfor,\@whilesw,\@width,\advance,\arraystretch,\baselineskip}
+% \DoNotIndex{\begingroup,\bgroup,\count,\count@,\cr,\csname,\def,\divide,\do,\dp,\edef,\egroup}
+% \DoNotIndex{\else,\endcsname,\endgroup,\expandafter,\extrarowheight,\fi,\gdef,\halignto,\hbox}
+% \DoNotIndex{\hfil,\ht,\if,\if@firstamp,\if@tempswa,\ifcase,\lineskip,\loop,\m@ne,\m@th,\message}
+% \DoNotIndex{\multiply,\or,\par,\prepnext@tok,\protect,\repeat,\setbox,\space,\string,\strutbox}
+% \DoNotIndex{\tabskip,\tabularnewline,\the,\the@toks,\thr@@,\tw@,\unhbox,\vbox,\vcenter,\vrule}
+% \DoNotIndex{\vskip,\vtop,\wd,\xdef,\z@,\z@skip}
 % \title{The \textsf{fcolumn} package\thanks{This file has version
 % number \fileversion, last revised \filedate.}}
-% \author{Edgar Olthof}
+% \author{Edgar Olthof\\
+% \texttt{edgar <dot> olthof <at> inter <dot> nl <dot> net}}
 % \date{Printed \today}
 % \maketitle
 % \begin{abstract}
 % In financial reports, text and currency amounts are regularly put in
 % one table, e.g., a year balance or a profit-and-loss overview.
-% This package provides the settings for automatically typesetting
+% This package provides the settings for automatically typesetting and checking
 % such columns, including the sum line (preceded by a rule of the correct
 % width), using the specifier |f|.
+% \end{abstract}
 %
-% The column specifier |f| itself is rather simple.  It is the predefined version
-% of a generic column |F|.  The generic version expects three arguments:
-% |#1| is the separator, |#2| is the decimal mark, and |#3| the coding used for
+% \section{Introduction}\label{intro}
+% The package |fcolumn| provides the macros for an extra tabular specifier that makes creating 
+% financial tables easy.  The column specifier |f| itself is rather simple.  It is the
+% predefined version of a generic column |F|.  The generic version expects three arguments:
+% |#1| is the group separator, |#2| is the decimal mark, and |#3| the coding used for
 % grouping digits of the integer part and decimal part.
 %
-% The f-column in the current version of the package is defined as
-% |\newcolumntype{f}{F{.}{,}{3,2}}| (which is the continental European standard, implemented here)
-% and people in the Anglo-saxon world would rather code |\newcolumntype{f}{F{,}{.}{3,2}}|.
+% The f-column in the current version of the package is defined for the continental European
+% standard: |\newcolumntype{f}{F{.}{,}{3,2}}|.  This means that a number like 1234567 will be 
+% typeset as $12{.}345{,}67$.  People in the Anglo-saxon world would rather
+% code |\newcolumntype{f}{F{,}{.}{3,2}}|, yielding $12{,}345{.}67$ for the number given above.
 % The default value for |#3| is 3,2, indicating that grouping of the integer part
 % is by three digits and that the decimal part consists of two digits. 
 % If however, in your country or company grouping is done with a thinspace every 
 % four digits and there are three digits after the decimal mark---that happens to 
 % be a |\cdot|---, then simply specify |\newcolumntype{f}{F{\,}{\cdot}{4,3}}| in 
-% that case.  By default two digits are used for the decimal part, so if you really
+% that case.  
+%
+% By default two digits are used for the decimal part, so if you really
 % want no decimal digits (in that case of course also skipping the decimal mark) 
 % you have to explicitly specify |x,0|.
-% \end{abstract}
-%
-% \section{Introduction}
-% To show where and how the f-column is used, let's look at a typical financial table.
 %
+% This package requires and loads the |array| package~\cite{array}.
+% To show where and how the F-column is used, let's look at some typical financial information 
+% as shown in Table~\ref{tab:ex1}
 % \begin{table}[htb]
+% \caption{Example Table}
+% \label{tab:ex1}
 % \begin{tabular}{@{}lflf@{}}
 % \multicolumn4c{\bfseries Balance sheet}\\
-% \multicolumn1{@{}l}{properties}&\multicolumn1r{31 dec 2014}
-% &\multicolumn1l{debts}&\multicolumn1{l@{}}{31 dec 2014}\\
-% \hline
+% \toprule
+% properties&\multicolumn1r{31 dec 2014}&
+% debts&\multicolumn1{l@{}}{31 dec 2014}\\
+% \midrule
 % house &     20000000 & equity capital&  5000000\\
 % bank account& -60323 & mortgage &      15000000\\
-% savings&     2800000 &         & \leeg         \\
+% savings&     2800000 \\
 % cash&          14585 & profit  &        2754262\\
 % \sumline
-% \check12
-% \hline
+% \bottomrule
 % \end{tabular}
 % \end{table}
-%
-% This table was entered as:
+% and how this is entered in \LaTeX\ (Table~\ref{tab:ex2}).
+% \begin{table}[htb]
+% \caption{Verbatim version of Example Table}
+% \label{tab:ex2}
 % \begin{verbatim}
 % \begin{table}[htb]
+% \caption{Example Table}
+% \label{tab:ex1}
 % \begin{tabular}{@{}lflf@{}}
 % \multicolumn4c{\bfseries Balance sheet}\\
-% \multicolumn1{@{}l}{properties}&\multicolumn1r{31 dec 2014}
-% &\multicolumn1l{debts}&\multicolumn1{l@{}}{31 dec 2014}\\
-% \hline
+% \toprule
+% properties&\multicolumn1r{31 dec 2014}&
+% debts&\multicolumn1{l@{}}{31 dec 2014}\\
+% \midrule
 % house &     20000000 & equity capital&  5000000\\
 % bank account& -60323 & mortgage &      15000000\\
-% savings&     2800000 &         & \leeg         \\
+% savings&     2800000 \\
 % cash&          14585 & profit  &        2754262\\
 % \sumline
-% \check12
-% \hline
+% \bottomrule
 % \end{tabular}
 % \end{table}
 % \end{verbatim}
+% \end{table}
 % All the work was done by the column specifier ``f'' (for ``finance'').  In this case
-% it constructs the |\sumline|, typesets the numbers, calculates the totals,
-% and checks whether the two columns are in balance; if not, the user is warned
-% via a |\message|.  Of course for nice settings one should use the
-% |booktabs| package, but that is not the point here.
+% it constructs the |\sumline|, typesets the numbers, calculates the totals, determines
+% the widths of the sumrules, and checks whether the two columns are in balance; if not,
+% the user is warned via a |\message|.  Of course for nice settings the
+% |booktabs| package~\cite{booktabs} was used, but that is not the point here.
 %
-% This package is heavily inspired by the |dcolumn| package by David Carlisle,
+% This package is heavily inspired by the |dcolumn| package by David Carlisle~\cite{dcolumn},
 % some constructions are more or less copied from that package.
 %
-% There is at least one things to improve in this version:
-% An empty entry in an f-column should result in an empty spot in the table.
-% Now an empty entry is not correct and is repaired by \LaTeX\ to be equal to
-% 0, printing a number in the table.  This can be prevented by explicitly specifying 
-% |\leeg|, like in the table above, but that should be automated.
-%
+% \section{Commands}
+% The user only needs to know six commands or constructions.  These six are given here.
+% \begin{macro}{F}
+% In the tabular the column specifier |F| can be given with arguments, or the predefined
+% version |f|, where the three arguments of |F| are |{.}|, |{,}|, and |3,2|.  If you
+% want |g| to be your own definition like the curious one given in Section~\ref{intro}, then
+% specify |\newcolumntype{g}{F{\,}{\cdot}{4,3}}| prior to using |g| in a tabular.
+% 
+% Entries in an F-column are from that moment on, treated as (integer) numbers unless
+% explicitly escaped by |\leeg|, see below.  The numbers are typeset according to the template
+% the user gives with his/her |F|-column.
+% \end{macro}
+% \begin{macro}{\sumline}
+% The numbers in an |F|-column are typeset as a financial amount, but the real benefit comes
+% with the |\sumline|. It does three things:
+% \begin{compactitem}
+% \item It calculates the total of the column so far and the maximum width encountered 
+% so far, including the width of the total;
+% \item It generates a rule with width calculated in the first item;
+% \item It checks the columns that are supposed to balance whether or not they actually do.
+% If so, nothing happens.  If not, a |\message| is given that column $i$ and $j$ do not balance,
+% where $i$ and $j$ are the relevant columns.  This is only done if the total number of 
+% |F|-columns is even, e.g., if there are six |F|-columns, then~1 is checked against~4, 2 against~5, 
+% and 3 against~6.  If the number of |F|-columns is odd then anything could be possible in that table
+% and nothing is assumed about structure within the table.  This behaviour can be overridden, see below.
+% \end{compactitem}
+% \end{macro}
+% \begin{macro}{\resetsumline}
+% Suppose you want to typeset one tabular with the profit-and-loss of many projects individually.
+% The layout of those tabulars is the same and it were nice if all columns were aligned.  This can 
+% be done by making it one big tabular with a fresh start for each project.  The macro |\resetsumlines|
+% is used for that: it resets all totals and all column widths, see for example Table~\ref{tab:ex3}.
+% \begin{table}[htb]
+% \caption{Example: multiple projects}
+% \label{tab:ex3}
+% \begin{tabular}{@{}lfflff@{}}
+% \multicolumn6c{\bfseries Project~1}\\
+% \toprule
+% expense&\multicolumn1r{actual}&\multicolumn1r{budget}&
+% income&\multicolumn1r{actual}&\multicolumn1{r@{}}{budget}\\
+% \midrule
+% food &         45000 & 50000 & tickets&  100000 &  100000\\
+% drinks &       35000 & 40000 \\
+% music &        18000 & 10000 \\
+% profit &        2000 \\
+% \sumline
+% \resetsumline
+% \multicolumn6c{\bfseries Project~2}\\
+% \toprule
+% expense&\multicolumn1r{actual}&\multicolumn1r{budget}&
+% income&\multicolumn1{r}{actual}&\multicolumn1{r@{}}{budget}\\
+% \midrule
+% food &         25000 & 30000 & tickets&  40000 &  45000\\
+% drinks &       10000 &  8000 \\
+% music &         8000 &  7000 & loss &     3000 \\
+% \sumline
+% \bottomrule
+% \end{tabular}
+% \end{table}
+% Note that the rules in the first and third F-columns of project~1 cover $1{.}000{,}00$ whereas
+% in project~2 those rules are shorter since they only cover $430{,}00$; still the columns are 
+% aligned.  The verbatim way of setting up Table~\ref{tab:ex3} is given in Table~\ref{tab:ex4}.
+% \begin{table}[htb]
+% \caption{Verbatim version of Table~\ref{tab:ex3}}
+% \label{tab:ex4}
+% \begin{verbatim}
+% \begin{table}[htb]
+% \caption{Example: multiple projects}
+% \label{tab:ex3}
+% \begin{tabular}{@{}lfflff@{}}
+% \multicolumn6c{\bfseries Project~1}\\
+% \toprule
+% expense&\multicolumn1r{actual}&\multicolumn1r{budget}&
+% income&\multicolumn1r{actual}&\multicolumn1{r@{}}{budget}\\
+% \midrule
+% food &         45000 & 50000 & tickets&  100000 &  100000\\
+% drinks &       35000 & 40000 \\
+% music &        18000 & 10000 \\
+% profit &        2000 \\
+% \sumline
+% \resetsumline
+% \multicolumn6c{\bfseries Project~2}\\
+% \toprule
+% expense&\multicolumn1r{actual}&\multicolumn1r{budget}&
+% income&\multicolumn1r{actual}&\multicolumn1{r@{}}{budget}\\
+% \midrule
+% food &         25000 & 30000 & tickets&  40000 &  45000\\
+% drinks &       10000 &  8000 \\
+% music &         8000 &  7000 & loss &     3000 \\
+% \sumline
+% \bottomrule
+% \end{tabular}
+% \end{table}
+% \end{verbatim}
+% \end{table}
+% \end{macro}
+% \begin{macro}{\leeg}
+% If an |F|-column should be empty then simply leave it empty.  If however it should not be empty
+% but the entry should be treated as text---even it is a number---this can be done with |\leeg|.
+% It expects an argument and this argument is typeset in the column, affecting the maximum column
+% width so far.  The common case is where |p.m.| (\textit{pro memoria}) is entered. An empty |F|-column
+% followed by |\\| does not work: either remove the |&| or specify |\leeg{}|.
+% \end{macro}
+% \begin{macro}{\checkfcolumns}
+% The automatic column balance check can also be done manually.  If |F|-columns~1 and~4 should balance
+% and you want them to be checked, then simply say |\checkfcolumns14|.  With more than nine |F|-columns
+% you may be forced to say something like |\checkfcolumns{10}{12}|.
+% \end{macro}
+% \begin{macro}{\strictaccounting}
+% In the rare occasion that a negative number occurs in a financial table, the sign of that number
+% can be an explicit minus sign ($-$) or the number is coloured red, or it is typeset between parentheses,
+% and there may be even other ways.  By default (for aesthetic reasons) |fcolumn| typesets it with 
+% a minus~sign, but strict accounting prescibes that the number should be put between parentheses.
+% The latter can be accomplished by setting |\strictaccountingtrue|. 
+% \end{macro}
+% \StopEventually{\PrintChanges\PrintIndex}
 % \section{The macros}
 % \begin{macro}{column F}
 % \begin{macro}{column f}
-%
-% Note that the definition of the column type |f| does not use private macros 
+% The column specifier |F| is the generic one, and |f| is the default (European) one
+% for easy use. Note that the definition of the column type |f| does not use private macros 
 % (no |@|), so overriding its definition is easy for a user.
 %    \begin{macrocode}
 \newcolumntype{F}[3]{>{\b@fi{#1}{#2}{#3}}r<{\e@fi}}
@@ -186,18 +284,20 @@ LaTeX version 1999/12/01 or later.
 % \end{macro}
 % \end{macro}
 % \begin{macro}{\FCsc@l}
+% \begin{macro}{\FCtc@l}
 % Two \meta{count}s are defined, that both start at zero: the \meta{count} 
-% |\FCsc@l|, that keeps track at which f-column the tabular is working on and 
-% the \meta{count} |\FCtc@l|, that records the number of f-columns that were 
+% |\FCsc@l|, that keeps track at which F-column the tabular is working on and 
+% the \meta{count} |\FCtc@l|, that records the number of F-columns that were 
 % encountered so far.  Later in the package the code can be found for generating
-% a new \meta{count} and a new \meta{dimen} if the number of requested f-columns
-% is larger than currently available.  This is of course the case when an f-column
+% a new \meta{count} and a new \meta{dimen} if the number of requested F-columns
+% is larger than currently available.  This is of course the case when an F-column
 % is used for the first time.
 %    \begin{macrocode}
-\newcount\FCsc@l \FCsc@l0
-\newcount\FCtc@l \FCtc@l0
+\newcount\FCsc@l \FCsc@l=0
+\newcount\FCtc@l \FCtc@l=0
 %    \end{macrocode}
 % \end{macro}
+% \end{macro}
 % \begin{macro}{\geldm@cro}
 % The macro |\geldm@cro| takes a number and by default interprets this as
 % an amount expressed in cents (dollar cents, euro cents, centen, 
@@ -207,68 +307,68 @@ LaTeX version 1999/12/01 or later.
 % As explained, this can be changed. It uses a private boolean |\withs@p|
 % and an accessable, i.e., without |@| boolean: |\strictaccounting|.  The latter  
 % is used to typeset negative numbers between parentheses.  By default it doesn't
-% do this: a minus sign is used.  It also uses two private \meta{count}s. I'm still
-% figuring out whether this is needed; everything is done local within a column
-% of an |\ialign|, so using |\count0|, |\count1|, etc.\ might also work: it would
-% safe two \meta{count}s.
+% do this: a minus sign is used.
 %    \begin{macrocode}
-\newcount\g@lda
-\newcount\g@ldb
 \newif\ifwiths@p
 \newif\ifstrictaccounting \strictaccountingfalse
 %    \end{macrocode}
 % Actually |\geldm@cro| is only a wrapper around |\g@ldm@cro|.
 %    \begin{macrocode}
 \def\geldm@cro#1#2{\withs@pfalse 
- \afterassignment\g@ldm@cro\count@#2\relax{#1}}
+\afterassignment\g@ldm@cro\count@#2\relax{#1}}
 %    \end{macrocode}
 % \end{macro}
 % \begin{macro}{\g@ldm@cro}
 % This macro starts by looking at the sign of |#2|: if it is negative, it prints the correct
-% indicator (a parenthesis or a minus sign), assigns the absolute value of |#2| to |\g@lda| 
-% and goes on.  Note that |\geldm@cro| and therefore |\g@ldm@cro| are used within |$|s, so 
-% it is really a minus sign that is printed, not a hyphen.
+% indicator (a parenthesis or a minus sign), assigns the absolute value of |#2| to |\count2| 
+% and goes on.  Note that |\geldm@cro| and therefore |\g@ldm@cro| are always used within |$|s, so 
+% it is really a minus sign that is printed, not a hyphen.  All calculations are done with
+% |\count0|, |\count1|, etc.\ i.e., without fcolumn-specific \meta{count}s because it is all
+% done locally. Leaving the tabular environment will restore their values.
 %    \begin{macrocode}
 \def\g@ldm@cro#1\relax#2{%
- \ifnum#2<0 \ifstrictaccounting(\else-\fi\g@lda=-#2 \else\g@lda=#2 \fi
+\ifnum#2<0 \ifstrictaccounting(\else-\fi\count2=-#2 \else\count2=#2 \fi
 %    \end{macrocode}
 % Calculate the entire currency units: this is the result of $x/a$ as integer
 % division, with $a=10^n$ and $n$ the part of |#1| after the separator (if any).
 % Here the first character of |#1| is discarded, so the separator in |#1| is not strict:
 % you could also specify |3.2| instead of |3,2| (or even |3p2|).
 %   \begin{macrocode}
- \ifx\relax#1\relax\g@ldb=2 \else \g@ldb\@gobble#1\relax\fi
- \loop
-  \ifnum\g@ldb>0 \divide\g@lda by 10 \advance\g@ldb by \m@ne
- \repeat
+\count3=\ifx\relax#1\relax2 \else \@gobble#1\relax\fi
+\count4=\count3
+\loop
+ \ifnum\count3>0 \divide\count2 by 10 \advance\count3 by \m@ne
+\repeat
 %    \end{macrocode}
-% The value in |\g@lda| is then output by |\g@ldens| using the separation given.
+% The value in |\count2| is then output by |\g@ldens| using the separation given.
 %    \begin{macrocode}
- \g@ldens{\the\count@}%
+\g@ldens{\the\count@}%
 %    \end{macrocode}
 % If there is a decimal part\dots
 %    \begin{macrocode}
- \ifx\relax#1\relax\g@ldb=2 \else \g@ldb\@gobble#1\relax\fi
- \ifnum\g@ldb>0\decim@lmark
+\ifnum\count4>0\decim@lmark
 %    \end{macrocode}
 % Next the decimal part is dealt with. Now $x\bmod a$ is calculated in the 
-% usual way: $x-(x/a)*a$.  The minus sign necessary for this calculation is 
-% introduced in the next line by changing the comparison to |>| instead of |<|.
-%    \begin{macrocode}
-  \ifnum#2>0 \g@lda=-#2\else\g@lda=#2 \fi
-  \loop
-   \ifnum\g@ldb>0 \divide\g@lda by 10 \advance\g@ldb by \m@ne
-  \repeat
-  \ifx\relax#1\relax\g@ldb=2 \else \g@ldb\@gobble#1\relax\fi
-  \loop
-   \ifnum\g@ldb>0 \multiply\g@lda by 10 \advance\g@ldb by \m@ne
-  \repeat
-  \ifnum#2>0 \advance\g@lda by #2
-  \else \advance\g@lda by -#2
-  \fi
-  \ifx\relax#1\relax\g@ldb=2 \else \g@ldb\@gobble#1\relax\fi
-  \zerop@d{\number\g@ldb}{\number\g@lda}%
+% usual way: $x-(x/a)*a$ with integer division.  The minus sign necessary 
+% for this calculation is introduced in the next line by changing the 
+% comparison from |<| to |>|.
+%    \begin{macrocode}
+ \ifnum#2>0 \count2=-#2\else\count2=#2 \fi
+ \count3=\count4
+ \loop
+  \ifnum\count3>0 \divide\count2 by 10 \advance\count3 by \m@ne
+ \repeat
+%    \end{macrocode}
+% The value of |\count3| is~0 now. Now counting up again, this saved an assignment.
+%    \begin{macrocode}
+ \loop
+  \ifnum\count3<\count4 \multiply\count2 by 10 \advance\count3 by \@ne
+ \repeat
+ \ifnum#2>0 \advance\count2 by #2
+ \else \advance\count2 by -#2
  \fi
+ \zerop@d{\number\count3}{\number\count2}%
+\fi
 %    \end{macrocode}
 % If the negative number is indicated by putting it between parentheses, then the
 % closing parenthesis should stick out of the column, otherwise the alignment of 
@@ -277,53 +377,52 @@ LaTeX version 1999/12/01 or later.
 % may even stick out of the table.  I don't like this, therefore I chose to put 
 % |\strictaccountingfalse|.  Change if you like.
 %    \begin{macrocode}
- \ifnum#2<0 \ifstrictaccounting\rlap{)}\fi\fi}
+\ifnum#2<0 \ifstrictaccounting\rlap{)}\fi\fi}
 %    \end{macrocode}
 % \end{macro}
 % \begin{macro}{\g@ldens}
 % Here the whole currency units are dealt with.  The macro |\g@ldens| is used
 % recursively, therefore the double braces; this allows to use |\count0| locally.
 % Tail recursion is not possible here (at least I don't know of a way), but that 
-% is not very important, as the largest number ($2^{31}-1$) will only cause a 
+% is not very important, as the largest number (which is $2^{31}-1$) will only cause a 
 % threefold recursion.  This also implies that the largest amount this package
 % can deal with is 2.147.483.647 (using |x.0|). For most people this is probably
 % more than enough if the currency is euros or dollars.  If not, then it is 
 % more likely that you are spending your time on the beach than studying this package.
 % And otherwise make clear that you use a currency unit of k\$.
 %    \begin{macrocode}
-\def\g@ldens#1{{\g@ldb=\g@lda \count0=#1
+\def\g@ldens#1{{\count3=\count2 \count0=#1
 %    \end{macrocode}
 % First divide by $10^n$, where $n$ is |#1|.
 %    \begin{macrocode}
- \loop
-  \ifnum\count0>0
-  \divide\g@lda by 10
-  \advance\count0 by \m@ne
- \repeat
+\loop
+ \ifnum\count0>0
+ \divide\count2 by 10
+ \advance\count0 by \m@ne
+\repeat
 %    \end{macrocode}
 % Here is the recursive part,
 %    \begin{macrocode}
- \ifnum\g@lda>0 \g@ldens{#1}\fi
+\ifnum\count2>0 \g@ldens{#1}\fi
 %    \end{macrocode}
 % and then reconstruct the rest of the number.
 %    \begin{macrocode}
- \count0=#1
- \loop
-  \ifnum\count0>0
-  \multiply\g@lda by 10
-  \advance\count0 by \m@ne
- \repeat
- \g@lda=-\g@lda
- \advance\g@lda by \g@ldb \du@zendprint{#1}}}
+\count0=#1
+\loop
+ \ifnum\count0>0
+ \multiply\count2 by 10
+ \advance\count0 by\m@ne
+\repeat
+\count2=-\count2
+\advance\count2 by \count3 \du@zendprint{#1}}}
 %    \end{macrocode}
 % \end{macro}
 % \begin{macro}{\du@zendprint}
 % The macro |\du@zendprint| takes care for correctly printing the separator
 % and possible trailing zeros.
 %    \begin{macrocode}
-\def\du@zendprint#1{\ifwiths@p\sep@rator\zerop@d{#1}{\number\g@lda}%
- \else\zerop@d{1}{\number\g@lda}\fi
- \global\withs@ptrue}
+\def\du@zendprint#1{\ifwiths@p\sep@rator\zerop@d{#1}{\number\count2}%
+ \else\zerop@d{1}{\number\count2}\fi \global\withs@ptrue}
 %    \end{macrocode}
 % \end{macro}
 % \begin{macro}{\zerop@d}
@@ -343,44 +442,31 @@ LaTeX version 1999/12/01 or later.
 % First determine the number of digits of |#2| (expressed in the decimal
 % system).  This number is in |\count0| and is at least~1.
 %    \begin{macrocode}
- \loop
-  \divide \count1 by 10
-  \ifnum\count1>0 \advance\count0 by 1
- \repeat
+\loop
+ \divide \count1 by 10
+ \ifnum\count1>0 \advance\count0 by\@ne
+\repeat
 %    \end{macrocode}
 % The number of zeros to be padded is $\max(0,|#1-\count0|)$ (the second
 % argument can be negative), so a simple loop suffices.
 %    \begin{macrocode}
- \loop
-  \ifnum\count0<#1 0\advance\count0 by 1
- \repeat 
- \number#2}}
+\loop
+ \ifnum\count0<#1\relax 0\advance\count0 by\@ne
+\repeat 
+\number#2}}
 %    \end{macrocode}
 % \end{macro}
 % \begin{macro}{\zetg@ld}
 % This macro takes care for several things: it increases the total for a given
-% f-column, it records the largest width of the entries in that column and it
+% F-column, it records the largest width of the entries in that column and it
 % typesets |#1| via |\geldm@cro|.
 %    \begin{macrocode}
 \def\zetg@ld#1#2{\global\advance\csname 
- FCtot@\romannumeral\FCsc@l\endcsname by #1
- \setbox0=\hbox{$\geldm@cro{#1}{#2}$}%
- \ifdim\wd0>\csname FCwd@\romannumeral\FCsc@l\endcsname
-  \global\csname FCwd@\romannumeral\FCsc@l\endcsname=\wd0
- \fi\unhbox0}
-%    \end{macrocode}
-% \end{macro}
-% \begin{macro}{\check}
-% The macro |\check| provides a way to the user to check that the  appropriate 
-% columns are balanced (as it should in a balance).  Arguments |#1| and |#2| are
-% the column numbers to compare.  It is the responsibility of the user to provide
-% the correct numbers here, otherwise bogus output is generated.  But the output is
-% only to screen and the transcript file; it doesn't change the appearance of your
-% document, so you can safely ignore the result then, repair and go on.
-%    \begin{macrocode}
-\def\check#1#2{\noalign{\ifnum\csname FCtot@\romannumeral#1\endcsname=
- \csname FCtot@\romannumeral#2\endcsname\else
- \message{Columns #1 and #2 do not balance!}\fi}}
+FCtot@\romannumeral\FCsc@l\endcsname by #1
+\setbox0=\hbox{$\geldm@cro{#1}{#2}$}%
+\ifdim\wd0>\csname FCwd@\romannumeral\FCsc@l\endcsname
+ \global\csname FCwd@\romannumeral\FCsc@l\endcsname=\wd0
+\fi\unhbox0}
 %    \end{macrocode}
 % \end{macro}
 % \begin{macro}{\b@fi}
@@ -392,84 +478,107 @@ LaTeX version 1999/12/01 or later.
 % The plain \LaTeX\ command |\@empty| is used for that.  The separator and decimal mark
 % are within a math environment, so you can indeed specify |\,| instead of |\thinspace|,
 % but there is an extra brace around, so it doesn't affect
-% the spacing between the digits (trick copied from |dcolumn|).
+% the spacing between the digits (trick copied from |dcolumn|, Ref.~\cite{dcolumn}).
 %    \begin{macrocode}
 \newcount\bedr@g
 \def\b@fi#1#2#3{\def\sep@rator{{#1}}\def\decim@lmark{{#2}}%
- \let\ignorespaces=\@empty \def\sp@l{#3}\global\advance\FCsc@l by \@ne 
+\def\sp@l{#3}\let\ignorespaces=\@empty \let\unskip=\@empty
+\global\advance\FCsc@l by \@ne
 %    \end{macrocode}
-% The value specified by the user is then captured by |\bedr@g|.
+% The value specified by the user is then captured by |\bedr@g| and this is done in a
+% special way: |\bedr@g| is assigned globally within |\box0|.  Why? To later check
+% the width of this box, see |\e@fi|.
 %    \begin{macrocode}
- \bedr@g=}
+\setbox0=\hbox\bgroup\global\bedr@g=}
 %    \end{macrocode}
 % \end{macro}
 % \begin{macro}{\e@fi}
-% Once captured, the rest is handled by |\e@fi|.  The number is then 
-% fed to |\zetg@ld| for the current column.
-%    \begin{macrocode}
-\def\e@fi{\relax \zetg@ld{\number\bedr@g}{\sp@l}}
+% If the user enters a number, say 10, \LaTeX\ will read
+% |\setbox| |0| |=| |\hbox| |\bgroup| |\global| |\bedr@g| |=| |10| |123| |\relax| |\egroup|, which
+% is perfectly valid.  The count |\bedr@g| is~10 now and |\box0| is non-empty, in fact containing
+% the number 123 (although any number would be good: it is discarded anyway).  If however nothing 
+% was entered by the user, \LaTeX\ will read |\setbox| |0| |=| |\hbox| |\bgroup| |\global| |\bedr@g|
+% |=| |123| |\relax| |\egroup|, which is also perfectly valid, but now |\bedr@g| is~123 and |\box0| 
+% is empty.  So the width of |\box0| can be used to discriminate without actually using its contents
+% in the horizontal list.  Later, with the code for |\leeg| a related trick will be used.  The space
+% at the beginning of |\e@fi| is important, otherwise the 123 might be concatenated with the user 
+% entry, leading to wrong numbers.
+%    \begin{macrocode}
+\def\e@fi{ 123\relax\egroup\ifdim\wd0>\z@
+ \zetg@ld{\number\bedr@g}{\sp@l}%
+\fi}
 %    \end{macrocode}
+% Please note that this trick does not work for an empty F-column that is ended by |\\|.  The reason 
+% is that the first thing |\\| does---even before providing the |\cr| ---is to issue a |\relax|.  
+% This stops the assignment to |\bedr@g| and causes a \TeX\ error, complaining about a missing number.
+% This is then repaired by \TeX\ by inserting a zero, so it doesn't affect the total of that column, 
+% but in particular this zero is then typeset, which it is not supposed to.  I decided to leave it 
+% this way because in practice this is never a problem: simply discard the last |&| and the column
+% before the offending column will now be ended by |\\|.  Iterate for this line until a filled F-column
+% or a non-F-column is encountered.  This is assuming that you did not specify \verb+|+s in the tabular
+% description: for them to work for the whole table you should have all |&|s in place (this is true in 
+% general, not only for |fcolumn|).  On the other hand you should never, ever use vertical rules in 
+% tables, according to Ref.~\cite{booktabs}.  And if you really must, then fill the empty F-columns 
+% with |\leeg{}|.
 % \end{macro}
 %
 % Here are adaptations to existing macros.
-% \begin{macro}{\@xarraycr}
-% Among all its tasks, now |\\| also resets the column number.  This 
-% is a transparant change, i.e., loading this package without using 
-% the ``f'' column specifier does
-% not change the effect of package |array.sty| in any way.
-% This resetting has to be done after the |\cr|, so it is stored
-% in a |\noalign|.
-%    \begin{macrocode}
-\def\@xarraycr{\@ifnextchar [%
- \@argarraycr {\ifnum 0=`{}\fi \cr\noalign{\global\FCsc@l=0 }}}
-%    \end{macrocode}
-% \end{macro}
 % \begin{macro}{\@array}
 % The definition of |\@array| had to be extended slightly because
 % it should also include |\@mksumline| (acting on the same |#2|
-% as |\@mkpream| gets).  Also this change is transparant: it only adds 
+% as |\@mkpream| gets).  This change is transparant: it only adds 
 % functionality and if you don't use that, you won't notice the difference.
+% It starts by just copying the original definition from the |array| package~\cite{array}.
 %    \begin{macrocode}
 \def\@array[#1]#2{%
- \@tempdima \ht \strutbox
- \advance \@tempdima by\extrarowheight
- \setbox \@arstrutbox \hbox{\vrule
-            \@height \arraystretch \@tempdima
-            \@depth \arraystretch \dp \strutbox
-            \@width \z@}%
- \begingroup
- \@mkpream{#2}%
- \xdef\@preamble{\noexpand \ialign \@halignto
-                 \bgroup \@arstrut \@preamble
-                         \tabskip \z@ \cr}%
+\@tempdima \ht \strutbox
+\advance \@tempdima by\extrarowheight
+\setbox \@arstrutbox \hbox{\vrule
+           \@height \arraystretch \@tempdima
+           \@depth \arraystretch \dp \strutbox
+           \@width \z@}%
+%    \end{macrocode}
+% Here comes the first change: after each |\\| (or |\cr| for that matter) the \meta{count}
+% |\FCsc@l| should be reset.  This is easiest done with |\everycr|, but |\everycr| is
+% put to |{}| by |\ialign|, so that definition should change.  The resetting should 
+% be done globally.
+%    \begin{macrocode}
+\def\ialign{\everycr{\noalign{\global\FCsc@l=0 }}%
+ \tabskip\z@skip\halign}
+%    \end{macrocode}
+% Then the definition is picked up again.
+%    \begin{macrocode}
+\begingroup
+\@mkpream{#2}%
+\xdef\@preamble{\noexpand \ialign \@halignto
+\bgroup\@arstrut\@preamble\tabskip\z@\cr}%
+\endgroup 
 %    \end{macrocode}
 % The combination |\endgroup| followed by |\begingroup| seems redundant, but that is
 % not the case: the |\endgroup| restores everything that was not |\global|.  With the
 % following |\begingroup| it is ensured that |\@mksumline| experiences the same
 % settings as |\@mkpream| did.
 %    \begin{macrocode}
- \endgroup 
- \begingroup
- \@mksumline{#2}%
- \endgroup
+\begingroup
+\@mksumline{#2}%
+\endgroup
 %    \end{macrocode}
 % As a side product of |\@mksumline| also the \meta{count}s for the totals and 
 % \meta{dimen}s for the widths of the colums are created.  The columns should start
-% fresh, i.e., totals are 0 and widths are
-% 0\thinspace pt.
+% fresh, i.e., totals are~0 and widths are 0\thinspace pt.
 %    \begin{macrocode}
- \res@tsumline
+\res@tsumline
 %    \end{macrocode}
 % From here on it is just the old definition of |array.sty|.
 %    \begin{macrocode}
- \@arrayleft
- \if #1t\vtop \else \if#1b\vbox \else \vcenter \fi \fi
- \bgroup
- \let \@sharp ##\let \protect \relax
- \lineskip \z@
- \baselineskip \z@
- \m@th
- \let\\\@arraycr \let\tabularnewline\\\let\par\@empty \@preamble} 
+\@arrayleft
+\if #1t\vtop \else \if#1b\vbox \else \vcenter \fi \fi
+\bgroup
+\let \@sharp ##\let \protect \relax
+\lineskip \z@
+\baselineskip \z@
+\m@th
+\let\\\@arraycr \let\tabularnewline\\\let\par\@empty \@preamble} 
 %    \end{macrocode}
 % Because |\@array| was changed here and it is this version that should be used,
 % |\@@array| should be |\let| equal to |\@array| again.
@@ -484,20 +593,22 @@ LaTeX version 1999/12/01 or later.
 % Furthermore most entries will simply add nothing to |\sumline|, e.g., |@|, |!|,
 % and \verb+|+ can be fully ignored. Ampersands are only inserted by |c|, |l|, |r|,  
 % |p|, |m|, and |b|.  So a specifier like |@{}lflf@{}| will yield the sumline
-% |&\a&&\a\\|, where |\a| is a macro that prints the desired result of the column.
-% Had the specifier been \verb+l|f||@{   }l|f+, then the same sumline must
+% |&\a&&\a\\|, (where |\a| is a macro that prints the desired result of the column, 
+% see later).  Had the specifier been \verb+l|f||@{   }l|f+, then the same sumline must
 % be constructed: all difficulties are already picked up and solved in the creation
 % of the preamble.
 %
 % In reality the sumline must be constructed from the expanded form of the specifier,
-% so |@{}lf@{}| will expand as |@{}l>{\b@fi#1#2#2}r<{\e@fi}@{}|. The
+% so |@{}lf@{}| will expand as |@{}l>{\b@fi{.}{,}{3,2}}r<{\e@fi}@{}|. The
 % rules for constructing the sumline are now very simple:
 % \begin{compactitem}
 % \item add an ampersand when |c|, |l|, |r|, |p|, |m|, or |b| is found, unless it is
 % the first one (this is the same as in the preamble);
 % \item add a |\a| when |<{\e@fi}| is found;
-% \item ignore everything else.
+% \item ignore everything else;
+% \item close with a |\\|.
 % \end{compactitem}
+% (In reality also the column check is inserted just before the |\\|, see |\aut@check|.)
 % To discriminate, a special version of |\testpach| could be written, but that is 
 % not necessary: |\testpach| can do all the work, although much of it will be discarded.
 % Here speed is sacrificed for space and this can be afforded because the creation
@@ -505,41 +616,43 @@ LaTeX version 1999/12/01 or later.
 % 
 % The start is copied from |\@mkpream|.
 %    \begin{macrocode}
-\def\@mksumline#1{\gdef\sumline{}
- \@lastchclass 4 \@firstamptrue
+\def\@mksumline#1{\gdef\sumline{}\@lastchclass 4 \@firstamptrue
 %    \end{macrocode}
 % At first the column number is reset and the actual code for what was called |\a| 
 % above is made inactive.
 %    \begin{macrocode}
- \global\FCsc@l=\z@
- \let\prr@sult=\relax
+\global\FCsc@l=0
+\let\prr@sult=\relax
 %    \end{macrocode}
 % Then |\@mkpream| is picked up again.
 %    \begin{macrocode}
- \@temptokena{#1}
- \@tempswatrue
- \@whilesw\if@tempswa\fi{\@tempswafalse\the\NC@list}%
- \count@\m@ne
- \let\the@toks\relax
- \prepnext@tok
+\@temptokena{#1}
+\@tempswatrue
+\@whilesw\if@tempswa\fi{\@tempswafalse\the\NC@list}%
+\count0\m@ne
+\let\the@toks\relax
+\prepnext@tok
 %    \end{macrocode}
 % Next is the loop over all tokens in the expanded form of the specifier.  The change
 % with respect to |\@mkpream| is that the body of the loop is now only dealing
-% with classes 0, 2, and 10.  What to do in those cases is of course different from
+% with F-classes~0, 2, and 10.  What to do in those cases is of course different from
 % what to do when constructing the preamble, so special definitions are created, see
 % below.
 %    \begin{macrocode}
- \expandafter \@tfor \expandafter \@nextchar
-  \expandafter :\expandafter =\the\@temptokena \do
-  {\@testpach
-  \ifcase \@chclass \@classfz
+\expandafter \@tfor \expandafter \@nextchar
+ \expandafter :\expandafter =\the\@temptokena \do
+ {\@testpach
+ \ifcase \@chclass \@classfz
   \or \or \@classfii \or
   \or \or \or \or \or \or \or \@classfx \fi
-  \@lastchclass\@chclass}%
+ \@lastchclass\@chclass}%
 %    \end{macrocode}
-% And the macro is finished by appending the |\\| to the sumline.
+% And the macro is finished by applying the |\aut@check| and appending the |\\| to the sumline.
+% Note that the |\aut@check| is performed {\itshape in\/} the last column, but since it does not
+% put anything in the horizontal list---it only writes to screen and transcript file---, this 
+% is harmless.
 %    \begin{macrocode}
- \xdef\sumline{\sumline\noexpand\\}}
+\xdef\sumline{\sumline\noexpand\aut@check\noexpand\\}}
 %    \end{macrocode}
 % \end{macro}
 % \begin{macro}{\@addtosumline}
@@ -550,26 +663,26 @@ LaTeX version 1999/12/01 or later.
 %    \end{macrocode}
 % \end{macro}
 % \begin{macro}{\@classfx}
-% Class~10 for the sumline creation is a stripped down version of |\@classx|: add an ampersand
+% Class~f10 for the sumline creation is a stripped down version of |\@classx|: add an ampersand
 % unless it is the first.  It deals with the specifiers |b|, |m|, |p|, |c|, |l|, and |r|.
 %    \begin{macrocode}
 \def\@classfx{\if@firstamp \@firstampfalse \else \@addtosumline &\fi} 
 %    \end{macrocode}
 % \end{macro}
 % \begin{macro}{\@classfz}
-% Class~0 is applicable for specifiers |c|, |l|, and |r| and if the arguments of |p|, |m|,
+% Class~f0 is applicable for specifiers |c|, |l|, and |r|, and if the arguments of |p|, |m|,
 % or |b| are given.  The latter three cases, with |\@chnum| is 0, 1, or~2 should be ignored
-% and the first three cases are now similar to class~10.
+% and the first three cases are now similar to class~f10.
 %    \begin{macrocode}
 \def\@classfz{\ifnum\@chnum<\thr@@ \@classfx\fi}
 %    \end{macrocode}
 % \end{macro}
 % \begin{macro}{\@classfii}
-% Here comes the nice and nasty part. Class~2 is applicable if a |<| is specified.  This is
+% Here comes the nice and nasty part. Class~f2 is applicable if a~|<| is specified.  This is
 % tested by checking |\@lastchclass|, which should be equal to~8.  Then it is checked that
 % the argument to |<| is indeed |\e@fi|.  This check is rather clumsy but this was the
 % first way, after many attempts, that worked.  It is necessary because the usage of |<|
-% is not restricted to |\e@fi|: the user may have specified other \TeX-code using |<|.
+% is not restricted to |\e@fi|: the user may have specified other \LaTeX-code using~|<|.
 %    \begin{macrocode}
 \def\@classfii{\ifnum\@lastchclass=8 
  \edef\t@stm{\expandafter\string\@nextchar}
@@ -580,26 +693,32 @@ LaTeX version 1999/12/01 or later.
 %    \begin{macrocode}
   \@addtosumline{\prr@sult}
 %    \end{macrocode}
-% But we're not done yet: in the following lines of code the appropriate \meta{count}s 
-% and \meta{dimen}s are created, if necessary.
+% But we're not done yet: in the following lines of code the appropriate \meta{count}s and
+% \meta{dimen}s are created, if necessary.  Note that |\FCsc@l| was set to~0 in the beginning
+% of |\@mksumline|, so it is well-defined when |\@classfii| is used.
 %    \begin{macrocode}
-  \global\advance\FCsc@l by \@ne
+  \global\advance\FCsc@l by \@ne 
   \ifnum\FCsc@l>\FCtc@l
 %    \end{macrocode}
-% Apparently the number of requested columns is larger than the currently available number,
-% so a new one should be created.  What is checked here is its existence: if it exists, it is not
-% created by |fcolumn| so it may be something other than a \meta{count}, in which things go haywire.
-% Therefore it is assumed that only |fcolumn| is allowed to generate a \meta{count} called 
-% |\FCtot@<some romannumeral>|.
+% Apparently the number of requested columns is larger than the currently available number of
+% relevant \meta{count}s and \meta{dimen}s, so new ones should be created.  What is checked 
+% here is merely the existence of |\FCtot@<some romannumeral>|.  If it already exists---although
+% it may not even be a \meta{count}; that is not checked---it is not created by |fcolumn| and a 
+% warning is given.  In case it is a \meta{count} you're just lucky, although any change to this
+% \meta{count} is global anyway, so things will be overwritten.  In the case it is not a 
+% \meta{count}, things will go haywire and you'll soon find out.  The remedy then is to rename 
+% your \meta{count} prior to |fcolumn| to avoid this name clash.
 %    \begin{macrocode}
    \expandafter\ifx\csname FCtot@\romannumeral\FCsc@l\endcsname\relax
     \expandafter\newcount\csname FCtot@\romannumeral\FCsc@l\endcsname
    \else
-    \message{\noexpand\FCtot@\romannumeral\FCsc@l already exists.}
+    \message{^^JWarning: FCtot@\romannumeral\FCsc@l \space is already 
+    defined and it may not even be a <count>. I'll proceed,
+    but with fingers crossed. }
    \fi
 %    \end{macrocode}
-% And the same is applicable for the \meta{dimen}: it is assumed that only |fcolumn| is allowed to
-% generate a \meta{dimen} called |\FCwd@<some romannumeral>|.
+% And the same is applicable for the \meta{dimen}: in case of a name clash you have to 
+% rename your \meta{dimen} prior to |fcolumn|.
 %    \begin{macrocode}
    \expandafter\ifx\csname FCwd@\romannumeral\FCsc@l\endcsname\relax
     \expandafter\newdimen\csname FCwd@\romannumeral\FCsc@l\endcsname
@@ -608,65 +727,115 @@ LaTeX version 1999/12/01 or later.
 %    \begin{macrocode}
     \global\FCtc@l=\FCsc@l
    \else
-    \message{\noexpand\FCwd@\romannumeral\FCsc@l already exists.}
+    \message{^^JWarning: FCwd@\romannumeral\FCsc@l \space is already 
+    defined and it may not even be a <dimen>. I'll proceed,
+    but with fingers crossed. }
    \fi
   \fi
  \fi
 \fi}
 %    \end{macrocode}
-% Once created they are not initialised here because that is done later in one go.
+% Once created it is not necessary to initialise them here because that is done later in one go.
 % \end{macro}
 % \begin{macro}{\leeg}
-% The macro |\prr@sult| (see below) actually puts the information together.
-% It starts with switching off the default behaviour of the f-column by 
-% providing the number 0 to comfort the assignment in |\b@fi|, although
-% any number would be good: it
-% is discarded anyway.  This is achieved by |\let|ting |\e@fi| to be |\relax|.
-% This |\let| is only local to the current column.  Since the user may from time to
-% time also need an empty entry in the table, this definition is without at-sign.
-%    \begin{macrocode}
-\def\leeg{0\relax \let\e@fi=\relax}
-%    \end{macrocode}
-% A better solution were to check for a number.  If present then do the 
-% default action, otherwise generate an empty entry.
+% This macro is used to overrule the default behaviour of the pair |\b@fi| and |\e@fi|.  It
+% starts with ending the assignment to |\bedr@g| in the same way that |\e@fi| would normally
+% do.  Then the effect of |\e@fi| (that is still in the preamble) is annihilated by |\let|ting 
+% it to be |\relax|.  This |\let| is only local to the current column.  Then the argument to 
+% |\leeg| is treated in a similar way as |\e@fi| would do with a typeset number.
+%
+% Since the user may from time to time also need a column entry other than a number in the 
+% table, e.g., |\leeg{p.m.}|, this definition is without at-sign.
+%    \begin{macrocode}
+\def\leeg#1{ 234\relax\egroup \let\e@fi=\relax \setbox0=\hbox{#1}%
+\ifdim\wd0>\csname FCwd@\romannumeral\FCsc@l\endcsname
+ \global\csname FCwd@\romannumeral\FCsc@l\endcsname=\wd0
+\fi\unhbox0}
+%    \end{macrocode}
+% Note that anything may be given as argument to |\leeg|, so in principle it can also be used
+% to cheat: |\leeg{0,03}| will insert |0,03| in the table but it doesn't increase the totals
+% of that column by~3 (assuming |3,2| coding for the separations).  But you won't cheat, won't you?
+% It may affect the width, so be careful: don't insert the unabridged version of Romeo and Julia here.
 % \end{macro}
 % \begin{macro}{\prr@sult}
-% And here it is: the end result.  First annihilate the effect of |\e@fi|.
+% The macro |\prr@sult| actually puts the information together.  It starts like |\leeg|.
 %    \begin{macrocode}
-\def\prr@sult{\leeg
+\def\prr@sult{ 345\relax\egroup \let\e@fi=\relax
 %    \end{macrocode}
 % Then the information for the last line is computed.  It is not sufficient to calculate
 % the width of the result (in points) to use that as the width of the rule separating 
-% the individual entries and the result.  It may be that the sum is larger
-% (in points) than any of the entries, e.g., when the result of $6+6$ is typeset.  The
-% width of the rule should be equal to the width of |\hbox{$12$}| then.  On the other
-% hand the width of the rule when summing $24$ and $-24$ should be that of |\hbox{$-24$}|
-% (or |\hbox{$(24$}|, see above), not the width of the result |\hbox{$0$}|.  Therefore the 
+% the individual entries and the result.  It may be that the sum is larger (in points) than
+% any of the entries, e.g., when the result of $600+600$ is typeset.  The width of the rule
+% should be equal to the width of |\hbox{$12{,}00$}| then (using specifier |3,2|).  On the other
+% hand the width of the rule when summing $2400$ and $-2400$ should be that of |\hbox{$-24{,}00$}|
+% (or |\hbox{$(24{,}00$}|, see above), not the width of the result |\hbox{$0{,}00$}|.  Therefore the 
 % maximum of all entry widths, including the result, was calculated.
 %    \begin{macrocode}
- \setbox0=\hbox{$\geldm@cro{\number\csname
- FCtot@\romannumeral\FCsc@l\endcsname}{\sp@l}$}%
- \ifdim\wd0>\csname FCwd@\romannumeral\FCsc@l\endcsname
-   \global\csname FCwd@\romannumeral\FCsc@l\endcsname=\wd0
- \fi
- \vbox{\hbox to \csname
-  FCwd@\romannumeral\FCsc@l\endcsname{\hrulefill}\vskip2pt
-  \hbox to \csname 
-  FCwd@\romannumeral\FCsc@l\endcsname{\hfil\unhbox0}}}
+\setbox0=\hbox{$\geldm@cro{\number\csname
+FCtot@\romannumeral\FCsc@l\endcsname}{\sp@l}$}%
+\ifdim\wd0>\csname FCwd@\romannumeral\FCsc@l\endcsname
+ \global\csname FCwd@\romannumeral\FCsc@l\endcsname=\wd0
+\fi
+\vbox{\hrule width \csname FCwd@\romannumeral\FCsc@l\endcsname
+\vskip2pt
+\hbox to \csname FCwd@\romannumeral\FCsc@l\endcsname{\hfil\unhbox0}}}
+%    \end{macrocode}
+% \end{macro}
+% \begin{macro}{\aut@check}
+% If the number of F-columns is even, it is assumed that they are part of two sets of columns of
+% which each column of the first set should balance the appropriate column of the second set.  If
+% on the other hand the number of columns is odd, then at least one column has nothing to balance
+% against and no checking occurs.  It is correct to check for oddness of |\FCsc@l| since this
+% |\aut@check| is only performed in the last column of the tabular: the value of |\FCsc@l| now 
+% equals the number of columns used in the current tabular (and may differ from |\FCtc@l|).
+%
+% The output is only to screen and the transcript file; it doesn't change the appearance of your
+% document, so in case the assumption is wrong you can safely ignore the result and go on.
+% The \meta{count}s 0 and~1 are used here and this can be done because any content of those 
+% \meta{count}s from previous calculations has become irrelevant at this moment.
+% \begin{macrocode}
+\def\aut@check{\ifodd\FCsc@l\else \count0=\@ne \count1=\FCsc@l 
+ \divide\count1 by \tw@ \advance\count1 by \@ne
+ \loop 
+  \ifnum\csname FCtot@\romannumeral\count0\endcsname=
+   \csname FCtot@\romannumeral\count1\endcsname\else
+   \message{^^JWarning: F-columns \number\count0 \space
+    and \number\count1 \space do not balance! }%
+  \fi
+  \ifnum\count1=\FCsc@l\else
+   \advance\count0 by\@ne \advance\count1 by\@ne
+ \repeat
+\fi }
+%    \end{macrocode}
+% \end{macro}
+% \begin{macro}{\checkfcolumns}
+% But the assumptions for |\aut@check| may be wrong, therefore manual
+% control on this checking is also made possible here. The macro |\checkfcolumns| 
+% provides a way to the user to check that the appropriate 
+% columns are balanced (as it should in a balance).  Arguments |#1| and |#2| are
+% the column numbers to compare.  It is the responsibility of the user to provide
+% the correct numbers here, otherwise bogus output is generated. 
+%    \begin{macrocode}
+\def\checkfcolumns#1#2{\noalign{\ifnum\csname FCtot@\romannumeral#1
+\endcsname=\csname FCtot@\romannumeral#2\endcsname\else
+ \message{^^JWarning: F-columns #1 and #2 do not balance! }%
+\fi}}
 %    \end{macrocode}
 % \end{macro}
 % \begin{macro}{\res@tsumline}
 % Since all changes to the totals and widths of the columns are global,
 % they have to be reset actively at the start of a tabular or array.
 % That is an action by itself, but it may occur more often, on request
-% of the user, therefore a special macro is defined.  The count 
-% |\FCsc@l| can be used as a temporary register because its actual
-% value is not important anymore.
+% of the user, therefore a special macro is defined.  A side effect of
+% this macro is that |\FCsc@l| is reset to~0. This is an advantage:
+% it should be zero at the beginning of a line in the table (for other
+% lines this is done by the |\\|).
 %    \begin{macrocode}
-\def\res@tsumline{\FCsc@l\FCtc@l \loop\ifnum\FCsc@l>0
-   \global\csname FCtot@\romannumeral\FCsc@l\endcsname=0
-   \global\csname FCwd@\romannumeral\FCsc@l\endcsname=0pt
-   \advance\FCsc@l by \m@ne \repeat}
+\def\res@tsumline{\FCsc@l=\FCtc@l\loop\ifnum\FCsc@l>0
+ \global\csname FCtot@\romannumeral\FCsc@l\endcsname=0
+ \global\csname FCwd@\romannumeral\FCsc@l\endcsname=\z@
+ \advance\FCsc@l by \m@ne
+\repeat}
 %    \end{macrocode}
 % \end{macro}
 % \begin{macro}{\resetsumline}
@@ -676,19 +845,18 @@ LaTeX version 1999/12/01 or later.
 %    \end{macrocode}   
 % \end{macro}
 % That's it!
-% \StopEventually{\PrintIndex}
-% \Finale
-%\typeout{****************************************************}
-%\typeout{*}
-%\typeout{* To finish the installation you have to move the}
-%\typeout{* following file into a directory searched by TeX:}
-%\typeout{*}
-%\typeout{*\space\space\space fcolumn.sty}
-%\typeout{*}
-%\typeout{* To produce the documentation run the file}
-%\typeout{* fcolumn.dtx through LaTeX.}
-%\typeout{*}
-%\typeout{* Happy TeXing!}
-%\typeout{*}
-%\typeout{****************************************************}
+%\section*{Acknowledgement}
+% Thanks to Karl Berry for valuable comments regarding the consistency of
+% the installation procedure of this version.
+%\bibliographystyle{plain}
+%\begin{thebibliography}{9}
+%\itemsep0pt
+%\parsep0pt
+%\parskip0pt
+%\bibitem{array} Frank Mittelbach and David Carlisle. 
+%    A new implementation of \LaTeX's |tabular| and |array| environment.
+%\bibitem{booktabs} Simon Fear. Publication quality tables in \LaTeX.
+%\bibitem{dcolumn} David Carlisle.  The |dcolumn| package.
+%\end{thebibliography}
+%\Finale
 \endinput