path: root/macros/latex/contrib/fcolumn/fcolumn.dtx

% \iffalse meta-comment
% Copyright (C) 2015--2022 by Edgar Olthof
%
% This file may be distributed and/or modified under the
% conditions of the LaTeX Project Public License, either
% 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 1.3 or later is part of all distributions of
% LaTeX version 1999/12/01 or later.
%
% \fi
% \iffalse
%<package>\ProvidesPackage{fcolumn}
%<package>          [2022/02/01 v1.4.1 Automatic Currency Package (EHTO)]
%<package>\NeedsTeXFormat{LaTeX2e}
%<package>\RequirePackage{array}[v2.4k]
%<*driver>
\documentclass{ltxdoc}
\usepackage{longtable}
\usepackage{dcolumn}
\usepackage{fcolumn}
\usepackage{booktabs}
\usepackage[tableposition=top,aboveskip=0pt]{caption}
\DeclareCaptionStyle{normal}{labelfont={rm,bf,small},textfont={rm,small},%
indention=0pt,labelsep=colon,justification=raggedright}
\captionsetup[table]{style=normal}
\captionsetup{singlelinecheck=false}
\usepackage{paralist}
\usepackage{url}
\usepackage[hyperfootnotes=false,hidelinks]{hyperref}
\usepackage{eurosym}
\usepackage{color}
\EnableCrossrefs
\CodelineIndex
\begin{document}
   \DocInput{fcolumn.dtx}
\end{document}
%</driver>
% \fi
% \RecordChanges
% \CheckSum{1200}
% \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
%   Digits        \0\1\2\3\4\5\6\7\8\9
%   Exclamation   \!     Double quote  \"     Hash (number) \#
%   Dollar        \$     Percent       \%     Ampersand     \&
%   Acute accent  \'     Left paren    \(     Right paren   \)
%   Asterisk      \*     Plus          \+     Comma         \,
%   Minus         \-     Point         \.     Solidus       \/
%   Colon         \:     Semicolon     \;     Less than     \<
%   Equals        \=     Greater than  \>     Question mark \?
%   Commercial at \@     Left bracket  \[     Backslash     \\
%   Right bracket \]     Circumflex    \^     Underscore    \_
%   Grave accent  \`     Left brace    \{     Vertical bar  \|
%   Right brace   \}     Tilde         \~}
%
% \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.}
% \changes{v1.1.2}{2015/09/29}{Some inconsistencies between explanatory text and actual 
%   code removed.}
% \changes{v1.2}{2019/03/06}{Input parsing changed after comment from Frank
%   Mittelbach.  He (Frank) also gave various suggestions for improving robustness
%   or user friendlyness of this package.  This version is only backwards compatible
%   when zero decimal digits were and are specified as modifier.}
% \changes{v1.3}{2021/01/03}{Christian Hoff requested the possibility to provide extra formatting
%   information to a column, e.g., colouring.  That was implemented by having an extra parameter
%   to the generic F-column.  The current solution is not very robust, as font and/or size change
%   in math environment are very tricky, but providing colour information works, see the example
%   in the main text.  This version is now compatible with package longtable.  Version~1.3 is 
%   backwards compatible to~1.2: it only adds functionality.}
% \changes{v1.4}{2021/11/21}{More robust version on the extra formatting information: alternative
%   formatting after the comma (if any), instead of additional information.  Corrected a few typos.}
% \changes{v1.4.1}{2022/01/31}{Definition needed for longtable only if that package is loaded.
%   General update, streamlining code.  Excess digits silently ignored when all zero.}
% \GetFileInfo{fcolumn.sty}
% \DoNotIndex{\@@endpbox,\@@startpbox,\@acol,\@arstrut,\@arstrutbox,\@classiv,\@classz,\@depth}
% \DoNotIndex{\@firstampfalse,\@firstamptrue,\@gobble,\@halignto,\@height,\@mkpream,\@ne,\@nextchar}
% \DoNotIndex{\@preamble,\@sharp,\@tempdima,\@tempswafalse,\@tempswatrue,\@\temptokena,\@tfor}
% \DoNotIndex{\@whilesw,\@width,\advance,\arraystretch,\baselineskip,\begingroup,\bgroup,\count}
% \DoNotIndex{\count@,\cr,\csname,\def,\divide,\do,\dp,\edef,\egroup,\hfil,\hline,\ht,\if,\if@firstamp}
% \DoNotIndex{\if@tempswa,\ifcase,\kill,\lineskip,\LT@@hl,\LT@@tabarray,\LT@bchunk,\LT@caption,\LT@endpbox}
% \DoNotIndex{\LT@hline,\LT@kill,\LT@LL@FM@cr,\LT@make@row,\LT@mcol,\LT@no@pgbk,\LT@nofcols,\LT@rows}
% \DoNotIndex{\LT@setprevdepth,\LT@startpbox,\LT@tabularcr,\LTleft,\LTright,\loop,\m@ne,\m@th,\message}
% \DoNotIndex{\multiply,\or,\par,\prepnext@tok,\protect,\refstepcounter,\repeat,\setbox,\space,\string}
% \DoNotIndex{\strutbox,\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\\
% \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 and checking
% such columns, including the sum line (preceded by a rule of the correct
% width), using the specifier |f|.
% \end{abstract}
%
% \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 four arguments: 1)~grouping 
% character of the integer part on output, 2)~decimal mark used on output,
% 3)~compact additional information on input/output characteristics, and
% 4)~anything, but primarily used for providing formatting information, see below.
%
% 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 12345,67 will be typeset as $12{.}345{,}67$.  
% People in the Anglo-Saxon world would rather code 
% |\newcolumntype{f}{F{,}{.}{3,2}{}}| for the same input, yielding 
% $12{,}345{.}67$ as output for the number given above.  The default
% value for |#3| is |3,2|, indicating that grouping of the integer 
% part is by three digits, that a comma is used in the \TeX-source to
% indicate the decimal mark, and that the decimal part consists of
% two digits.  However, if in your country or company grouping is done 
% with a thinspace every four digits, that the decimal mark in the
% source should be the character |p|, and there are three digits after
% the decimal mark---that happens to be a |\cdot|---, then simply specify
% |\newcolumntype{f}{F{\,}{\cdot}{4p3}{}}| in that case.  The input could
% be 123456p78 then, yielding $12{\,}3456{\cdot}780$ as output.
%
% 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|. If you want no grouping
% character, specify |0,x|.
%
% As the fourth parameter you can insert anything just before the typesetting 
% of an amount in a column takes place.  Its purpose is to add additional 
% formatting information, e.g., |\color{red}| to have the contents of a column
% coloured red, but it can be misused, so use with care. And it can't do all!
%
% 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}\\
% \toprule
% properties & \leeg{31 dec 2014} & debts & \leeg{31 dec 2014}\\
% \midrule
% house &        200000    & equity capital & 50000    \\
% bank account &   -603,23 & mortgage &      150000    \\
% savings &       28000    \\
% cash &            145,85 & profit &         27542,62 \\
% \sumline
% \bottomrule
% \end{tabular}
% \end{table}
% and how this is entered in \LaTeX\ (Table~\ref{tab:ex2}).
% \begin{table}[htb]
% \caption{Verbatim version of Example Table~\ref{tab:ex1}.}
% \label{tab:ex2}
% \begin{verbatim}
%\begin{table}[htb]
%\caption{Example Table.}
%\label{tab:ex1}
%\begin{tabular}{@{}lflf@{}}
%\multicolumn4c{\bfseries Balance sheet}\\
%\toprule
%properties & \leeg{31 dec 2014} & debts & \leeg{31 dec 2014}\\
%\midrule
%house &        200000    & equity capital & 50000    \\
%bank account &   -603,23 & mortgage &      150000    \\
%savings &       28000    \\
%cash &            145,85 & profit &         27542,62 \\
%\sumline
%\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, determines the widths of the sumrules, and checks
% whether the two columns are in balance; if not, the user is warned via a
% |\PackageWarning|.  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~\cite{dcolumn}; some constructions are more or less 
% copied from that package.  From version~1.3 onwards it incorporates the idea
% of Christian Hoff of providing additional (formatting) information per column.
% A rather contrived example is given in Table~\ref{tab:ex3}, combining colour and fonts.
% \newcolumntype{q}[1]{F{.}{,}{3,2}{#1}}
% \begin{table}[htb]
% \caption{Example Table with column formatting.}
% \label{tab:ex3}
% \begin{tabular}{@{}lq{\color{red}\mathsf,\mathbf}lq{\color{green}}@{}}
% \multicolumn4c{\bfseries Balance sheet}\\
% \toprule
% properties & \leeg{31 dec 2014} & debts & \leeg{31 dec 2014}\\
% \midrule
% house &        200000    & equity capital & 50000    \\
% bank account &   -603,23 & mortgage &      150000    \\
% savings &       28000   \\
% cash &            145,85 & profit &         27542,62 \\
% \sumline
% \bottomrule
% \end{tabular}
% \end{table}
% How this is entered in \LaTeX\ is shown in Table~\ref{tab:ex4}.
% \begin{table}[htb]
% \caption{Almost verbatim version of Example Table~\ref{tab:ex3}.}
% \label{tab:ex4}
% \begin{verbatim}
%\newcolumntype{q}[1]{F{.}{,}{3,2}{#1}}
%\begin{table}[htb]
%\caption{Example Table with column formatting.}
%\label{tab:ex3}
%\begin{tabular}{@{}lq{\color{red}\mathsf,\mathbf}lq{\color{green}}@{}}
%\multicolumn4c{\bfseries Balance sheet}\\
%...
%... (Same financial contents as in Table 1.)
%...
%\end{tabular}
%\end{table}
% \end{verbatim}
% \end{table}
% The font changing commands like |\mathsf| and |\mathbf| act on an argument, hence require
% braces, but these are already provided internally for this purpose.  For that reason this
% type of commands must be given last, without braces (and if you don't specify a font 
% changing command, these extra internal braces are just redundant).  The fourth argument to the
% new columntype may consist of two parts, separated by a comma.  In that case, the part to
% the left of the comma is applied to the data entered by the user and the right part
% is applied to the result.  The example in Table~\ref{tab:ex3} shows this: the bold font
% is only used in the |\sumline| and there is no colour specification, so that's back to
% the default (black).  If you want formatting for the whole column, like green colour as
% in the last column of Table~\ref{tab:ex3}, leave out the comma.  
%
% Note that changes in font size, e.g., ``|\huge|'' in ``|\huge\color{red}\mathbf|''
% as parameter to column type |q| are ignored by \LaTeX, since the formatting 
% information is used in math environment, which has its own way of handling this.
% This isn't bad, as size changes in one column, without overall changes to the 
% table look terrible.  If you want something huge, make a |\huge| table.
% 
% This package now also works with |longtable|~\cite{longtable}, provided
% |longtable| is loaded before |fcolumn|; it checks for that.  For a change, 
% the raw formatting of the multipage table is shown first, in Table~\ref{tab:ex5}.
% For more information on how to handle |\endhead| and related commands, see the 
% documentation of |longtable|~\cite{longtable}.
% \begin{table}[htb]
% \caption{Almost verbatim version of Example Table~\ref{tab:ex6}.}
% \label{tab:ex5}
% \begin{verbatim}
%\begin{longtable}[l]{@{}lflf@{}}
%\caption{\label{tab:ex6}Table showing compatibility of fcolumn and 
%  longtable.}\\
%\multicolumn4c{\bfseries Balance sheet}\\
%\toprule
%properties & \leeg{31 dec 2014} & debts & \leeg{31 dec 2014}\\
%\midrule
%\endfirsthead
%\caption[]{\textit{(continued from previous page)\/}}\\
%\toprule
%properties & \leeg{31 dec 2014} & debts & \leeg{31 dec 2014}\\
%\midrule
%\endhead
%\bottomrule
%\multicolumn4{r@{}}{\small\textit{(Table continues on next page)\/}}\\
%\endfoot
%\bottomrule
%\endlastfoot
%house &        200000    & equity capital & 50000    \\
%...
%... (Somewhat altered financial contents compared to
%...  Table 1 to demonstrate the page break.)
%...
%\sumline
%\end{longtable}
% \end{verbatim}
% \end{table}
%
% \noindent And here is the result (it's ugly, but it had to be this long to 
% demonstrate the page break).  Not shown here are the new |fcolumn|
% formatting possibilities (like new fonts and/or colours), but it has been 
% checked they do work in combination with |longtable|.
% \begin{longtable}[l]{@{}lflf@{}}
% \caption{\label{tab:ex6}Table showing compatibility of fcolumn and 
%   longtable.}\\
% \multicolumn4c{\bfseries Balance sheet}\\
% \toprule
% properties & \leeg{31 dec 2014} & debts & \leeg{31 dec 2014}\\
% \midrule
% \endfirsthead
% \caption[]{\textit{(continued from previous page)\/}}\\
% \toprule
% properties & \leeg{31 dec 2014} & debts & \leeg{31 dec 2014}\\
% \midrule
% \endhead
% \bottomrule
% \multicolumn4{r@{}}{\small\textit{(Table continues on next page)\/}}\\
% \endfoot
% \bottomrule
% \endlastfoot
% house &        200000    & equity capital & 50000    \\
% bank account &   -603,23 & mortgage &      150000    \\
% savings 1 &      4000    \\
% savings 2 &      4000    \\
% savings 3 &      4000    \\
% savings 4 &      4000    \\
% savings 5 &      4000    \\
% savings 6 &      4000    \\
% savings 7 &      4000    \\
% cash &            145,85 & profit &         27542,62 \\
% \sumline
% \end{longtable}
% \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 four arguments of |F| are |{.}|, 
% |{,}|, |{3,2}|, and |{}|.  If you want |g| to be your own definition like the 
% curious one given in Section~\ref{intro}, then specify 
% |\newcolumntype{g}{F{\,}{\cdot}{4p3}{}}| prior to using |g| in a tabular.
% 
% Entries in an F-column are, from that moment on, treated as numbers unless
% explicitly escaped by |\leeg|, see below.  The numbers are typeset 
% according to the template the user gives with his/her |F|-column.
% The ``middle'' character of |#3| is an important switch: it does more
% than just setting the input decimal mark.  By default the input grouping 
% character is the dot, except when the dot is specified as input decimal
% mark; in that case the comma is acting as input grouping character.  With 
% this convention continental Europe and the Anglo-Saxon part of the world 
% is served.  And using input grouping markers is optional anyway.
% \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{compactenum}[1)]
% \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 |\PackageWarning| 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{compactenum}
% By default the vertical separation between the rule and the total is 
% 2\thinspace pt, but this can be changed by the optional argument to 
% |\sumline|. Give, e.g., |\sumline[10pt]|, in case you want this spacing
% to be 10\thinspace pt.  And you may even give two options, like in 
% |\sumline[10pt][5pt]|, in which the second option is the extra space below
% the summary row.  In fact that second option is parsed to |\\|, that is
% implicit in |\sumline|.
% \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 
% |\resetsumline| is used for that: it resets all totals and all column 
% widths, see for example Table~\ref{tab:ex7}.
% \begin{table}[htb]
% \caption{Example: multiple projects.}
% \label{tab:ex7}
% \begin{tabular}{@{}lfflff@{}}
% \multicolumn6c{\bfseries Project~1}\\
% \toprule
% expense & \leeg{actual} & \leeg{budget} & 
% income & \leeg{actual} & \leeg{budget} \\
% \midrule
% food &         450,20 & 500    & tickets &  1200    &  1000    \\
% drinks &       547,5  & 400    \\
% music &        180    & 100    \\
% profit &        22,3  \\
% \sumline
% \resetsumline
% \multicolumn6c{\bfseries Project~2}\\
% \toprule
% expense & \leeg{actual} & \leeg{budget} & 
% income & \leeg{actual} & \leeg{budget} \\
% \midrule
% food &         250    & 300    & tickets &  400    &  450    \\
% drinks &       100    &  80    \\
% music &         80    &  70    & loss &      30    \\
% \sumline
% \bottomrule
% \end{tabular}
% \end{table}
% Note that the rules in the first and third |F|-columns of project~1 cover
% $1{.}200{,}00$ whereas in project~2 those rules are narrower since they 
% only cover $430{,}00$; still the columns are aligned.  The verbatim way
% of setting up Table~\ref{tab:ex7} is given in Table~\ref{tab:ex8}.
% \begin{table}[htb]
% \caption{Verbatim version of Table~\ref{tab:ex7}.}
% \label{tab:ex8}
% \begin{verbatim}
%\begin{table}[htb]
%\caption{Example: multiple projects.}
%\label{tab:ex3}
%\begin{tabular}{@{}lfflff@{}}
%\multicolumn6c{\bfseries Project~1}\\
%\toprule
%expense & \leeg{actual} & \leeg{budget} & 
%income & \leeg{actual} & \leeg{budget} \\
%\midrule
%food &         450,2  & 500    & tickets &  1200    &  1000    \\
%drinks &       547,5  & 400    \\
%music &        180    & 100    \\
%profit &        22,3  \\
%\sumline
%\resetsumline
%\multicolumn6c{\bfseries Project~2}\\
%\toprule
%expense & \leeg{actual} & \leeg{budget} & 
%income & \leeg{actual} & \leeg{budget} \\
%\midrule
%food &         250    & 300    & tickets &  400    &  450    \\
%drinks &       100    &  80    \\
%music &         80    &  70    & loss &      30    \\
%\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 
% if it is a number---, this can be done with |\leeg|.  It expects an 
% argument and this argument is typeset in the column.  The common case is
% where |p.m.| (\textit{pro memoria}) is entered.  In contrast to v1.1.2 
% of this package, now even an empty |F|-column followed by |\\| is allowed.
% \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}|.  If
% |\checkfcolumns| is used, the automatic check is disabled.  Multiple
% |\checkfcolumns|s are supported; if |F|-columns~1, 2, and~3 should 
% balance, you specify |\checkfcolumns12| and |\checkfcolumns23|.
% There is no explicit command to disable all checking, but
% |\checkfcolumns11| obviously does the trick.
% \end{macro}
% \begin{macro}{\ifstrict@ccounting}
% 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 |\strict@ccountingtrue|, but since this contains a non-letter,
% it is also possible to invoke |fcolumn| with the option |strict|, i.e.,
% |\usepackage[strict]{fcolumn}|, which sets this flag.
% \end{macro}
% \StopEventually{\PrintChanges\PrintIndex}
% \section{The macros}
% Here follows the actual code.
% \subsection{Option}
% \begin{macro}{option strict}
% There is one option. If set, strict accounting rules are used in display.
%    \begin{macrocode}
\newif\ifstrict@ccounting \strict@ccountingfalse
\DeclareOption{strict}{\strict@ccountingtrue}
\ProcessOptions
%    \end{macrocode}
% \end{macro}
%\subsection{Definitions}
% \begin{macro}{column F}
% \begin{macro}{column f}
% The column specifier |F| is the generic one, and |f| is the default 
% (continental 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}[4]{>{\b@fi{#1}{#2}{#3}{#4}}r<{\e@fi}}
\newcolumntype{f}{F{.}{,}{3,2}{}}
%    \end{macrocode}
% \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 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 is used for the first time.
%    \begin{macrocode}
\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, Pfennige,
% centimes, kopecks, groszy) and typesets it as the amount in entire 
% currency units (dollars, euros, guldens, Marke, francs, rubles, z\l oty)
% with comma as decimal mark and the dot as grouping character
% (thousand separator if the first part of |#1| is~3).
% As explained, this can be changed.  It uses two private booleans: 
% |\withs@p| and |\strict@ccounting|.  The latter is used to typeset
% negative numbers between parentheses.  By default it doesn't do this:
% a minus sign is used.
%    \begin{macrocode}
\newif\ifwiths@p
%    \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@#1\relax{#2}}
%    \end{macrocode}
% \end{macro}
% \begin{macro}{\g@ldm@cro}
% After setting the environment for formatting, 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 |\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 |F|-column-specific
% \meta{count}s because it is all done locally.  Leaving the tabular environment will 
% restore their values.  This is also true for the effect of |\FCform@t|, so that 
% formatting information is local to this column.  The reason for inserting the |{|
% between |\FCform@t| and |\ifnum| (and the accompanying |}| just before finishing 
% this definition) is to facilitate the possible use of |\mathbf| or any other font 
% changing command as the last item in |\FCform@t|.
%    \begin{macrocode}
\def\g@ldm@cro#1\relax#2{\FCform@t{\ifnum#2<0 \ifstrict@ccounting 
  (\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}
\count4=\ifx\relax#1\relax 2 \else \@gobble#1\relax\fi
\count3=0
\loop\ifnum\count3<\count4 
  \divide\count2 by 10 \advance\count3 by \@ne
\repeat
%    \end{macrocode}
% Note that |\count3| now equals |\count4|: this going up-and-down will be used more often,
% it saves several assignments.  The value in |\count2| is then output by |\g@ldens| using 
% the separation given (and stored in |\count@|).
%    \begin{macrocode}
\g@ldens{\the\count@}%
%    \end{macrocode}
% If there is a decimal part\dots
%    \begin{macrocode}
\ifnum\count3>0\decim@lmark
%    \end{macrocode}
% Next the decimal part is dealt with.  Now $x\bmod a$ is calculated in the 
% usual way: $x-(x/a)*a$ 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
  \loop\ifnum\count3>0
    \divide\count2 by 10 \advance\count3 by \m@ne
  \repeat
%    \end{macrocode}
% The value of |\count3| is now~0, so counting up again.
%    \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 this entry in the column is wrong.  This is done by an
% |\rlap| and therefore does not influence the column width.  For the last
% column this means that this parenthesis may even stick out of the table.
% I don't like this, therefore I chose to put |\strict@ccountingfalse|.
% Change if you like, by setting the option |strict|.
% 
% If overflow was detected, an exclamation mark is output to the right of
% the value that caused this.  This of course ruins the appearance of the
% table, but in this case that serves a clear goal: there's something wrong
% and you should know.
%    \begin{macrocode}
\ifx\FCs@gn\m@ne \ifnum#2<0 \ifstrict@ccounting
  \rlap{)~!}\else\rlap{~!}\fi\else\ifstrict@ccounting
  \rlap{\phantom{)}~!}\else\rlap{~!}\fi\fi
\else \ifnum#2<0 \ifstrict@ccounting\rlap{)}\fi\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.  This also implies that tail recursion is not 
% possible here, but that is not very important, as the largest number
% (which is $2^{31}-1$) will only cause a threefold recursion using the
% default |3,2| (ninefold when using~|1,0|, but who does that?).  The 
% largest amount this package can deal with is therefore 2.147.483.647 
% (using |3,0|).  For most people this is probably more than enough if
% the currency is euros or dollars.  And otherwise make clear that you
% use a currency unit of~k\EUR{} (or even M\EUR{} for the very rich).
% The author is thinking of ways to use two counters for each number.
% The maximum then becomes $2^{63}-1$.  Even expressed in cents this 
% would lead to a maximum of slightly more than 92.2\thinspace P\EUR{};
% about 100 times the current world economy~\cite{world_economy}. Yet
% another method is to use Heiko Oberdiek's package |bigintcalc|:
% then only memory restrictions apply.  This, however, would require
% a major rewrite of |fcolumn|.
%
% There is no straightforward interpretation of |#1| being zero or 
% negative, therefore this is used as an indicator that no grouping
% character should be used.
%    \begin{macrocode}
\def\g@ldens#1{{\count3=\count2 \count0=#1
%    \end{macrocode}
% First divide by $10^n$, where $n$ is |#1|.
%    \begin{macrocode}
\ifnum\count0<1 \count0=3 \fi
\loop \ifnum\count0>0 \divide\count2 by 10 \advance\count0 by \m@ne
\repeat
%    \end{macrocode}
% Here is the recursive part,
%    \begin{macrocode}
\ifnum\count2>0 \g@ldens{#1}\fi
%    \end{macrocode}
% and then reconstruct the rest of the number.
%    \begin{macrocode}
\count0=#1
\ifnum\count0<1 \count0=3 \fi
\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. The former, however, is only done if |#1|
% is larger than zero.
%    \begin{macrocode}
\def\du@zendprint#1{\ifwiths@p\ifnum#1>0 \sep@rator\fi
 \zerop@d{#1}{\number\count2}%
\else\zerop@d1{\number\count2}\fi\global\withs@ptrue}
%    \end{macrocode}
% \end{macro}
% \begin{macro}{\zerop@d}
% The macro |\zerop@d| uses at least |#1| digits for printing the 
% number |#2|, padding with zeros when necessary. Note: |#1| being
% zero or negative is a flag that it should be interpreted as 3.
% A bit ugly, but it works, since the related code knows about this.
%
% It is done within an extra pair of braces, so that |\count0| 
% and |\count1| can be used without disturbing their values in other macros.
%    \begin{macrocode}
\def\zerop@d#1#2{{\count0=1 \count1=#2
%    \end{macrocode}
% 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 \@ne
\repeat
%    \end{macrocode}
% If |#1| is positive, the number of zeros to be padded is 
% $\max(0,|#1-\count0|)$ (the second argument can be negative), so a
% simple loop suffices.  If it is zero or negative, this is a signal
% that it should be interpreted as 3 (and no separator will be output).
%    \begin{macrocode}
\ifnum#1>0
 \loop \ifnum\count0<#1\relax 0\advance\count0 by \@ne
 \repeat 
\else
 \advance\count0 by -3
 \loop \ifnum\count0<0 0\advance\count0 by \@ne
 \repeat 
\fi\number#2}}
%    \end{macrocode}
% \end{macro}
% \begin{macro}{\zetg@ld}
% This macro takes care for several things: it increases the subtotal for a
% given F-column, it checks whether or not that subtotal has overflown, 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{\count0=#2\relax \let\FCs@gn=\@ne
%    \end{macrocode}
% First it checks whether there is a risk of overflow in this step.  If
% $A$ and $B$ are two \TeX-registers and $B$ is to be added to $A$, 
% overflow will not occur if one is (or both are) zero or if $A$ and~$B$
% have different signs; otherwise be careful.  Note that \TeX\ does not
% check for overflow when performing an |\advance| (done in section~1238
% of Ref.~\cite{CandTB}), in contrast to |\multiply|, see section~105.
%    \begin{macrocode}
\ifnum\count0<0 
  \ifnum\csname FCtot@\romannumeral\FCsc@l\endcsname<0
    \let\FCs@gn=\m@ne 
  \fi
\fi
\ifnum\count0>0
  \ifnum\csname FCtot@\romannumeral\FCsc@l\endcsname>0
    \let\FCs@gn=\m@ne 
  \fi
\fi
\global\advance\csname FCtot@\romannumeral\FCsc@l\endcsname by \count0
\ifx\FCs@gn\m@ne 
%    \end{macrocode}
% They have the same sign, hence risk of overflow.  Record the sign
% of |\count0| (and of the original total of this column; we just established
% they were the same) in |\FCs@gn|. Table~\ref{tab:ex9} 
% \begin{table}[htb]
% \def\PackageError#1#2#3{\PackageWarning{#1}{#2: #3}}
% \caption{Examples on overflow.}
% \label{tab:ex9}
% \begin{tabular}{@{}lfff@{}}
% \multicolumn4c{\bfseries Projects}\\
% \toprule
% income & \leeg{31 dec 2014} & \leeg{31 dec 2015} & \leeg{31 dec 2016} \\
% \midrule
% item~1 &  20000000 & 20000000 &20000000 \\
% item~2 &  10000000 &  2000000 &-1500000 \\
% item~3 &   5000000 & -1500000 & 2000000 \\
% \checkfcolumns12
% \checkfcolumns23
% \sumline
% \bottomrule
% \end{tabular}
% \end{table}
% shows what can go wrong if the numbers are too large: in the left
% |F|-column the sumline is incorrect and the number that caused 
% the overflow is indicated by an exclamation mark. In the middle 
% |F|-column, overflow occurs twice and because this is once positive,
% once negative here, cancellation of errors occurs and the sumline
% is correct in the end.  Nevertheless, it is advised to swap the two
% items that caused the overflow, as shown in the right |F|-column.
%
% Check that the sign of the updated column total is still correct.
% If so, |\FCs@gn| is reset (to |\@ne|) at the end of this chunk.
%    \begin{macrocode}
  \ifnum\count0>0 \let\FCs@gn\@ne \fi
  \count0=\csname FCtot@\romannumeral\FCsc@l\endcsname
  \ifnum\FCs@gn<0 \count0=-\count0 \fi
  \ifnum\count0<0 
    \let\FCs@gn=\m@ne
    \PackageError{fcolumn}{Register overflow}{Overflow occurred
    in fcolumn \number\FCsc@l.\MessageBreak You can press <enter> 
    now and I'll proceed, but check\MessageBreak your table. The
    offending entry is indicated with an\MessageBreak exclamation
    mark in the output.}%
  \else\let\FCs@gn=\@ne
  \fi
\fi
%    \end{macrocode}
% The value of |\FCs@gn| is used in |\geldm@cro| below.
%    \begin{macrocode}
\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}
%
% The \meta{count}s |\FC@l| and |\FC@r| capture the parts to the left and to the 
% right of the decimal mark, respectively.
%    \begin{macrocode}
\newcount\FC@l \newcount\FC@r
%    \end{macrocode}
% Some auxiliary definitions for capturing compacted information.
%    \begin{macrocode}
\def\setucc@de#1#2\relax{\uccode`\~=`#1 }
\def\assignform@t#1,#2,#3\assignform@t{\def\FCform@t{#1}%
  \def\FCform@tt{#2}\ifx\FCform@tt\@empty \def\FCform@tt{#1}\fi}
%    \end{macrocode}
% \begin{macro}{\m@thcodeswitch}
% As will be shown below, once the first digit, or sign, or decimal mark, or
% grouping character is scanned, the decimal digits should loose their
% activeness.  That is done here for the digits in a rather blunt way, by putting their
% |\mathcode|s to zero if |#1| equals~0, since the actual |\mathcode| is not important---as 
% long as it is not |"8000|---because the digits are not used for typesetting.
% (And even if they were; it's inside |\box0|, whose contents will be discarded.)  
% When the |$| in |\e@fi| is encountered, the digits get back their original
% |\mathcode|s so that the actual typesetting in |\zetg@ld| is correct again.
% With a non-zero |#1| the activeness is switched on.
%    \begin{macrocode}
\def\m@thcodeswitch#1{\count0=10 \loop\ifnum\count0>0 
\advance\count0 by \m@ne\mathcode\expandafter`\the\count0=
\ifnum#1=0 0 \else "8000 \fi\repeat}
%    \end{macrocode}
% \end{macro}
% \def\<#1>{\leavevmode\hbox{$\langle$#1\/$\rangle$}}
% \begin{macro}{\b@fi}
% The macro |\b@fi| provides the beginning of the financial column.  It
% will be inserted in the column to capture the number entered by the user.
% 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|, Ref.~\cite{dcolumn}).
%    \begin{macrocode}
\def\b@fi#1#2#3#4{%
%    \end{macrocode}
% An intermediate macro |\sep@xt| to extract the first character
% of |#1|, which in most cases will be the only character.
%    \begin{macrocode}
\def\sep@xt##1##2\end{\def\sep@rator{{##1}}}\sep@xt#1\end
\def\decim@lmark{{#2}}\def\sp@l{#3}\assignform@t#4,,\assignform@t
\global\advance\FCsc@l by \@ne\global\FC@l=0 \global\FC@r=1
%    \end{macrocode}
% The value specified by the user is then captured by |\FC@l| and 
% this is done in a special way: |\FC@l| is assigned globally within
% |\box0|.  Why?  To use it as scribbling paper to examine what the 
% user entered, without dumping it into the horizontal list.
%
% There are four parts to an |F|-column entry, all parts optional,
% making 16 combinations.  The sequence is
% (in the Backus--Naur notation of Ref.~\cite{CandTA}):
% \<sign> \<integer constant> \<decimal mark>
% \<integer constant>. Here \<sign> is a plus or minus character with 
% category code~12, \<integer constant> is a sequence of zero
% or more (decimal) \<digit>s, and \<decimal mark> is the middel 
% part of |#3|, i.e., the comma in |3,2| or the period in |3.2|.  If
% the \<decimal mark> is absent with no space characters between the
% two \<integer constant> terms, these merge, making four redundant
% entries.  One of the combinations is \<empty>, a sequence of characters
% outside the ones that are made active, e.g., plain text or nothing at all:
% this is the only combination that doesn't put anything in an |F|-column---and
% was the most difficult part to handle.
%
% The minus sign must be captured separately, because in an entry like
% |-0,07| the 7~cents are negative, but this cannot be seen from the
% part to the left of the decimal mark, since $-0$ is~$0$ in \TeX\ (in 
% fact in most computer languages, but not in~MIX~\cite{TAOCP1}), so 
% |\ifnum-0<0| yields |false|. |\FCs@gn| is a general purpose flag.  Its 
% first use is to capture the sign.
%    \begin{macrocode}
\let\FCs@gn=\@ne\relax \setbox0\hbox\bgroup$
%    \end{macrocode}
% Do the scan inside a box and inside math mode.  Start with defining all
% characters that may appear as the first one in an |F|-column as active; 
% digits first. This needs a complicated |\edef| with accompanying |\noexpand|
% because the starting digit for |\FC@l| must be packed inside each definition.
%    \begin{macrocode}
\count@=10 \loop\ifnum\count@>0 \advance\count@ by \m@ne
 \uccode`\~=\expandafter`\the\count@ \uppercase{\edef~}{\noexpand
 \m@thcodeswitch0 \global\FC@l=\the\count@}\repeat
%    \end{macrocode}
% For the input decimal mark something extra is needed: if it is the
% first character in an |F|-column (like in |,07|), it should also
% restore the |\mathcode|s of the digits.  Checking whether or not it
% is the first is easy, since in that case the |\mathcode|s of the 
% decimal digits is still |"8000|.  The assignment to |\FC@r| starts 
% with~1, so that appended digits get captured correctly, even if they
% start with~0; postprocessing of |\FC@r| is done in |\e@fi|.
% The input decimal mark switches itself off as active character, 
% so at most one input decimal mark is allowed (N.B.:
% this makes sense).
%    \begin{macrocode}
\def\deactdecm@rk##1##2\relax{\mathcode`##1=0 }%
\afterassignment\setucc@de\count@#3\relax
\uppercase{\def~}{\ifnum\mathcode`\0="8000 \m@thcodeswitch0 \fi
\afterassignment\deactdecm@rk\count@#3\relax \global\FC@r=1}%
%    \end{macrocode}
% The input grouping character effectively expands to ``nothing, i.e.,
% ignore'' in a complicated way: it ignores the character and resumes 
% scanning the number.  The test prior to that action is needed if
% the grouping character is the first character encountered in the
% |F|-column.  Which part to continue with depends on whether or not an
% input decimal mark was encountered; that can be checked by looking
% at its |\mathcode|.
% 
% The input grouping character is the dot ``|.|'', except when that character
% was already chosen as input decimal mark.  In that case, the grouping 
% character will be the comma.  This is easy to check because the |\uccode|
% of |`\~| is still preserved.  
%    \begin{macrocode}
\def\d@cm##1##2{\count@=\mathcode`##1 }
\ifnum\uccode`\~=`. \uccode`\~=`,\relax\else \uccode`\~=`.\relax\fi
\uppercase{\def~}{\ifnum\mathcode`\0="8000 \m@thcodeswitch0 \fi
\afterassignment\d@cm\count@#3\relax
%    \end{macrocode}
% The |\expandafter| below is necessary because the global assignment
% should act after the |\fi|.
%    \begin{macrocode}
\ifnum\count@=\mathcode`- \expandafter\global\FC@l=\the\FC@l
\else \expandafter\global\FC@r=\the\FC@r\fi}%
%    \end{macrocode}
% The signs are relatively simple: record the sign, restore |\mathcode|s if
% needed (it should be: a minus sign between digits screws up everything),
% and start scanning the number.
%    \begin{macrocode}
\uccode`\~=`+\relax \uppercase{\def~}{\ifnum\mathcode`\0="8000
\m@thcodeswitch0 \fi\global\FC@l=0}
\uccode`\~=`-\relax \uppercase{\def~}{\ifnum\mathcode`\0="8000
\m@thcodeswitch0 \fi\global\let\FCs@gn\m@ne \global\FC@l=0}%
%    \end{macrocode}
% Now actually activate all these codes.
%    \begin{macrocode}
\mathcode`-="8000 \mathcode`+="8000 \mathcode`.="8000
%    \end{macrocode}
% These three remain active until the |$| in |\e@fi| is encountered.
% The following ones will, except in the \<empty> case, have their activeness 
% turned off at some time.
%    \begin{macrocode}
\m@thcodeswitch1 \def\actdecm@rk##1##2\relax{%
\ifx##1.\relax \mathcode`,="8000 \else \mathcode`##1="8000 \fi}%
\afterassignment\actdecm@rk\count@#3\relax}
%    \end{macrocode}
% \end{macro}
% \begin{macro}{\e@fi}
% If the digits are still active then either nothing was entered or only
% characters that did not deactivate the digits were entered.  In 
% both cases the output should be \<empty>.  To flag this outside
% the group that started with the opening |$| of |\b@fi|, |\FC@r| is set
% globally to a negative value.  This doesn't harm, because it didn't 
% contain relevant information anyway.  Outside the group, the sign
% of |\FC@r| can then be tested.  This is a slight misuse of this
% |\count|, but now it's documented.  In effect, |\FC@r| can only be~|-1|, 
% |1|, or at least~|10|, so the comparison |\ifnum\FC@r>0| does not miss~0.
%    \begin{macrocode}
\def\e@fi{\ifnum\mathcode`\0="8000 \global\FC@r=\m@ne\fi$\egroup
\ifnum\FC@r>0
%    \end{macrocode}
% If there was no decimal mark or if there was a decimal mark but no 
% decimal part, |\FC@r| will still be~1, which doesn't parse well with
% |\secd@xt|, so a zero is appended, i.e., yielding~|10|.
%    \begin{macrocode}
 \ifnum\FC@r=1 \FC@r=10 \fi
%    \end{macrocode}
% Next is a loop for bringing the decimal part in the correct way to
% the integer part.  The loop is performed the number of decimal
% digits to be printed (the~2 in |3,2| of the default setting).
%    \begin{macrocode}
 \def\i@ts##1##2{\count0=##2}
 \afterassignment\i@ts\count@\sp@l
 \loop\ifnum\count0>0 \multiply\FC@l by 10
  \expandafter\secd@xt\number\FC@r\secd@xt \advance\count0 by \m@ne
 \repeat
%    \end{macrocode}
% This also means that if you provided more decimal digits than this, the excess 
% digit(s) will not be handled.  This is truncation, not rounding!  If all 
% truncated digits are zero, this truncation is a no-op and they are silently ignored,
% see the example Table~\ref{tab:ex10}, 
% \begin{table}[htb]
% \newcolumntype{d}{D{,}{,}{-1}}
% \caption{Truncating excess digits.}
% \label{tab:ex10}
% \begin{tabular}{@{}ldfl@{}}
% \toprule
% composer & \multicolumn1c{raw entry} & \leeg{debt} & remark \\
% \midrule
% Berg   &      123,450  & 123,450  & ``0'' silently ignored \\
% Eisler &      234,567  & 234,567  & ``7'' ignored with warning \\
% Sch\"onberg & 345,6709 & 345,6709 & ``09'' ignored with warning \\
% Webern &        2,3456 &   2,3456 & ``56'' ignored with warning \\
% \sumline
% \bottomrule
% \end{tabular}
% \end{table}
% that was created with |@{}ldfl@{}| (|d| for centering on the decimal
% mark~\cite{dcolumn}) as tabular key. If, however, at least one of them is 
% not zero, a |\PackageWarning| will be given, showing the excess digit(s). 
%    \begin{macrocode}
 \ifnum\expandafter\@gobble\number\FC@r>0
   \def\tw@l##1##2\relax{##2}
   \PackageWarning{fcolumn}{Excess digit\ifnum\FC@r>19 s\fi\space
   ``\expandafter\tw@l\number\FC@r\relax'' in decimal part
   \MessageBreak ignored near or}
 \fi
%    \end{macrocode}
% Don't forget to correct for the sign (once this is done, |\FCs@gn|
% is free again and can and will be used for other purposes).
% Then output the result.
%    \begin{macrocode}
 \ifx\FCs@gn\m@ne\relax \FC@l=-\FC@l \fi
 \zetg@ld{\sp@l}{\FC@l}%
\fi}
%    \end{macrocode}
% \end{macro}
% \begin{macro}{\secd@xt}
% The second digit from the left is needed from a string of characters
% representing a decimal number. The macro |\secd@xt| extracts that digit,
% provided that the number has at least two digits, but that is guaranteed by
% |\e@fi|.  That second digit is then added to |\FC@l|. 
% A new number is assigned to |\FC@r|, that consists 
% of the digits of |#1#3|.  If |#3| was empty, 10 is assigned.  In this way
% |\FC@r| is prepared for insertion in the next invocation of |\secd@xt|.  
% In iterating: 1234 yields~134, yields~14, yields~10, stays~10, etc. 
%    \begin{macrocode}
\def\secd@xt#1#2#3\secd@xt{\advance\FC@l by #2
\FC@r=#1#3 \ifnum\FC@r=1 \FC@r=10 \fi}
%    \end{macrocode}
% \end{macro}
% \subsection{Adaptations to existing macros}
% \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).
% 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 v2.4k of the |array| package~\cite{array},
% compacted.
%    \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@}%
%    \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| in the previous chunk followed by |\begingroup|
% in the next chunk 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}
\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.
%    \begin{macrocode}
\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} 
%    \end{macrocode}
% Because |\@array| was changed here and it is this version that should
% be used, |\@@array| should be |\let| equal to |\@array| again.
%    \begin{macrocode}
\let\@@array=\@array
%    \end{macrocode}
% \end{macro}
% Much of the techniques here are repeated in |\LT@array|.
% \subsection{The sumline, close to a postamble}
% \begin{macro}{\@mksumline}
% The construction of the sumline is much easier than that of the preamble
% for several reasons.  It may be safely assumed that the preamble specifier
% is grammatically correct because it has already been screened by |\@mkpream|.
% Furthermore, most entries will simply add nothing to |\s@ml@ne|, e.g., |@|, |!|,
% and \verb+|+ can be fully ignored. Ampersands are only inserted by |c|, |l|, |r|,  
% |p|, |m|, and |b|.  So, if |\a| is a macro that prints the desired result of the
% column (see below), then a specifier like |@{}lflf@{}| will yield the sumline 
% |&\a&&\a\\|.  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{.}{,}{3,2}{}}r<{\e@fi}@{}|. The rules for constructing the 
% sumline are now very simple:
% \begin{compactitem}[$\bullet$]
% \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 close with a |\\|.
% \end{compactitem}
% (For completeness' sake it should be mentioned that prior to the |\\| also the column
% check is inserted, see |\aut@check|.)
% To discriminate, a special version of |\@testpach|~\cite{array}
% 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 of the sumline is done only 
% once per |\tabular| or |\longtable|.
% 
% The start is copied from |\@mkpream|.
%    \begin{macrocode}
\def\@mksumline#1{\gdef\s@ml@ne{}\@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=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}\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 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\or\or\@classfii\or
\or\or\or\or\or\or\or\@classfx\fi\@lastchclass\@chclass}%
%    \end{macrocode}
% 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\s@ml@ne{\s@ml@ne\noexpand\aut@check\noexpand\\}}
%    \end{macrocode}
% \end{macro}
% \begin{macro}{\@addtosumline}
% Macro |\@addtosumline|, as its name already suggests, adds something to the 
% sumline, like its counterpart |\@addtopreamble| did to the preamble.
%    \begin{macrocode}
\def\@addtosumline#1{\xdef\s@ml@ne{\s@ml@ne #1}}
%    \end{macrocode}
% \end{macro}
% \begin{macro}{\@classfx}
% 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~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~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~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 
% \LaTeX-code using~|<|.
%    \begin{macrocode}
\def\@classfii{\ifnum\@lastchclass=8 
 \edef\t@stm{\expandafter\string\@nextchar}
 \edef\t@stn{\string\e@fi}
 \ifx\t@stm\t@stn
%    \end{macrocode}
% If both tests yield |true|, i.e., we encountered a |<{\e@fi}| where we expect
% one to find, then add the macro to typeset everything.
%    \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.  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 
  \ifnum\FCsc@l>\FCtc@l
%    \end{macrocode}
% 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 cannot be checked---it is not created by 
% |fcolumn| and an error is given.  In case it is a \meta{count} you're just
% lucky, and you could ignore that error, 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
    \PackageError{fcolumn}{Name clash for <count>}{\expandafter\csname
    FCtot@\romannumeral\FCsc@l\endcsname is already defined and it may
    not even be a <count>. If you're\MessageBreak sure it is a <count>,
    you can press <enter> now and I'll proceed, but things\MessageBreak
    will get overwritten.}%
   \fi
%    \end{macrocode}
% 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
%    \end{macrocode}
% If the creation was successful, the \meta{count} |\FCtc@l| should be 
% increased.
%    \begin{macrocode}
    \global\FCtc@l=\FCsc@l
   \else
    \PackageError{fcolumn}{Name clash for <dimen>}{\expandafter\csname
    FCwd@\romannumeral\FCsc@l\endcsname is already defined and it may 
    not even be a <dimen>. If you're\MessageBreak sure it is a <dimen>,
    you can press <enter> now and I'll proceed, but things\MessageBreak
    will get overwritten.}%
   \fi
  \fi
 \fi
\fi}
%    \end{macrocode}
% Once created, it is not necessary to initialise them here because that
% is done later in one go.
% \end{macro}
% \begin{macro}{\sumline}
% The command for the sumline has one optional argument: the separation
% between the rule and the total. By default this is 2\thinspace pt, but
% the user may specify |\sumline[10pt]| if that separation needs to be
% 10\thinspace pt.  The assignment needs to be global, because it is done
% in the first column of the tabular, but is valid for the whole line.
%    \begin{macrocode}
\newdimen\s@mlinesep
\def\sumline{\@ifnextchar[\s@mline{\s@mline[2pt]}}
\def\s@mline[#1]{\global\s@mlinesep=#1 \s@ml@ne}
%    \end{macrocode}
% In the introduction it was stated that |\sumline| has two options, but in 
% reality that second option is the option to |\\| that is issued by |\s@ml@ne|.
% \end{macro}
% \begin{macro}{\prr@sult}
% The macro |\prr@sult| actually puts the information together.  It starts
% like |\leeg|.
%    \begin{macrocode}
\def\prr@sult{$\egroup \let\e@fi=\relax \let\FCform@t=\FCform@tt
%    \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 happen
% that the sum is wider (in points) than any of the entries, e.g., when the
% result of $6+6$ (using specifier~|3,2|) is typeset. The width of the rule 
% should be equal to the width of |\hbox{$12{,}00$}| then.  On the other
% hand the width of the rule when summing $24$ and $-24$ 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{$\geldm@cro{\sp@l}{\number\csname
FCtot@\romannumeral\FCsc@l\endcsname}$}%
\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
\vskip\s@mlinesep
\hbox to \csname FCwd@\romannumeral\FCsc@l\endcsname{\hfil\unhbox0}}}
%    \end{macrocode}
% \end{macro}
% \subsection{Other checks}
% \begin{macro}{\leeg}
% This macro is used to overrule the default behaviour of the pair |\b@fi| 
% and |\e@fi|.  It starts with ending the groups 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. By defining |\leeg| in this way, instead of |\multicolumn1r{}|
% (which contains |\omit|), the default spacing in the column is retained.
% It doesn't alter the width of the sumrule, but has its normal effect on
% the column width, so be careful: don't insert the unabridged version of
% Romeo and Juliet~\cite{RJ} here.  It is not typeset in math mode, nor 
% does it use the extra (formatting) information of |#4| of the fcolumn,
% so you're completely free here.
%    \begin{macrocode}
\def\leeg#1{$\egroup \let\e@fi=\relax #1}
%    \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 the text~|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?
% \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.  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=\z@
 \advance\FCsc@l by \m@ne\repeat}
%    \end{macrocode}
% \end{macro}
% \begin{macro}{\resetsumline}
% To reset a sumline within a table, it should be done within a |\noalign|.
%    \begin{macrocode}
\def\resetsumline{\noalign{\res@tsumline}}
%    \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.
%
% If the list |\FC@chklist| is empty, the list for the automatic check is
% generated (which will remain empty if |\FCsc@l| is odd).
%    \begin{macrocode}
\def\FC@chklist{}
\def\aut@check{\ifx\@empty\FC@chklist\relax
 \ifodd\FCsc@l\else 
  \count0=\@ne \count1=\FCsc@l 
  \divide\count1 by \tw@
  \loop\ifnum\count1<\FCsc@l
   \advance\count1 by \@ne
   \xdef\FC@chklist{\FC@chklist\number\count0,\number\count1;}%
   \advance\count0 by\@ne
  \repeat
 \fi
\fi
%    \end{macrocode}
% Then this list is peeled off and processed.
%    \begin{macrocode}
\loop 
\ifx\FC@chklist\@empty\let\FCs@gn=\@ne\else\let\FCs@gn=\m@ne\fi
\ifx\FCs@gn\m@ne
 \expandafter\fre@t\FC@chklist\end
 \ifnum\csname FCtot@\romannumeral\count0\endcsname=
  \csname FCtot@\romannumeral\count1\endcsname\else
  \PackageWarning{fcolumn}{F-columns \number\count0 \space and
  \number\count1 \space do not balance due to \string\sumline
  \MessageBreak near or}%
 \fi
\repeat}
%    \end{macrocode}
% When |\aut@check| is finished, |\FC@chklist| is empty again, i.e., well
% prepared for the next time it is used.  This also means that the default
% behaviour kicks in again: if that's not what you want, you should
% specify the appropriate |\checkfcolumns| lines again.
% \end{macro}
% \begin{macro}{\fre@t}
% This function eats the first two numbers off |\FC@chklist|.
%    \begin{macrocode}
\def\fre@t#1,#2;#3\end{\count0=#1 \count1=#2 \xdef\FC@chklist{#3}}
%    \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 |F|-column numbers to compare.  It is the responsibility of 
% the user to provide the correct numbers here, otherwise bogus output 
% is generated.  If this manual check is inserted, the automatic check
% will not be performed.
%    \begin{macrocode}
\def\checkfcolumns#1#2{\noalign{\xdef\FC@chklist{\FC@chklist #1,#2;}}}
%    \end{macrocode}
% \end{macro}
% \subsection{Support for package longtable}
% Package |longtable| is used for tables that may span multiple pages.  The
% packages |fcolumn| and |longtable| work together provided |longtable| is 
% loaded first, so that |fcolumn| can adapt one definition of |longtable|. If
% |longtable| is not loaded, it's a waste of memory having that definition, so
% it is only defined when necessary.  For that reason the user is warned (not 
% loading |longtable| is obviously not an error if you don't use it) if 
% |fcolumn| is loaded without prior loading of |longtable|.
%    \begin{macrocode}
\ifx\longtable\@undefined
 \PackageWarningNoLine{fcolumn}{fcolumn is loaded without longtable.
 That's OK, but\MessageBreak if you want to load longtable as well, 
 make sure \MessageBreak it is done before loading fcolumn}\else
%    \end{macrocode}
% \begin{macro}{\LT@array}
% And here is the only definition of |longtable| that needs to be extended
% to make |fcolumn| work with that package.  The lines are compacted a bit
% w.r.t.\ the original |longtable| code; if you want to study the
% code, have a look at the documentation~\cite{longtable}.
%    \begin{macrocode}
 \def\LT@array[#1]#2{\refstepcounter{table}\stepcounter{LT@tables}\if
  l#1 \LTleft\z@\LTright\fill\else\if r#1 \LTleft\fill\LTright\z@\else
  \if c#1 \LTleft\fill\LTright\fill\fi\fi\fi\let\LT@mcol\multicolumn
  \let\LT@@tabarray\@tabarray\let\LT@@hl\hline\def\@tabarray{\let
  \hline\LT@@hl\LT@@tabarray}\let\\\LT@tabularcr\let\tabularnewline\\
  \def\newpage{\noalign{\break}}\def\pagebreak{\noalign{\ifnum`}=0\fi
  \@testopt{\LT@no@pgbk-}4}\def\nopagebreak{\noalign{\ifnum`}=0\fi
  \@testopt\LT@no@pgbk4}\let\hline\LT@hline\let\kill\LT@kill\let\caption
  \LT@caption\@tempdima\ht\strutbox\let\@endpbox\LT@endpbox\ifx
  \extrarowheight\@undefined\let\@acol\@tabacol\let\@classz\@tabclassz
  \let\@classiv\@tabclassiv\def\@startpbox{\vtop\LT@startpbox}\let
  \@@startpbox\@startpbox\let\@@endpbox\@endpbox\let\LT@LL@FM@cr
  \@tabularcr\else\advance\@tempdima\extrarowheight\col@sep\tabcolsep
  \let\@startpbox\LT@startpbox\let\LT@LL@FM@cr\@arraycr\fi\setbox
  \@arstrutbox\hbox{\vrule\@height\arraystretch\@tempdima\@depth
  \arraystretch\dp\strutbox\@width\z@}\let\@sharp##\let\protect\relax
  \begingroup\@mkpream{#2}\xdef\LT@bchunk{\global\advance\c@LT@chunks
  \@ne\global\LT@rows\z@\setbox\z@\vbox\bgroup\LT@setprevdepth\tabskip
  \LTleft\noexpand\halign to\hsize\bgroup\tabskip\z@\@arstrut\@preamble
  \tabskip\LTright\cr}\endgroup
%    \end{macrocode}
% Until this line it was just the code for |\LT@array| from package~|longtable|.
% The two lines of the next chunk are new to |\LT@array|.  Their 
% purpose is the same as in |\@array| above.
%    \begin{macrocode}
  \begingroup\@mksumline{#2}\endgroup\res@tsumline 
  \everycr{\noalign{\global\FCsc@l=0 }}% 
%    \end{macrocode}
% From here on |\LT@array| is picked up again, ending with the |\fi| that belongs
% to the |\ifx| that started this definition.
%    \begin{macrocode}
  \expandafter\LT@nofcols\LT@bchunk&\LT@nofcols\LT@make@row\m@th\let
  \par\@empty\lineskip\z@\baselineskip\z@\LT@bchunk}
\fi
%    \end{macrocode}
% \end{macro}
% That's it!
% \section*{Acknowledgement}
% Thanks to Karl Berry for valuable comments regarding the consistency of the
% installation procedure of this version.  Frank Mittelbach gave various useful
% suggestions for improving the input parsing as well as hints to make the package
% more \LaTeX-like.  He also challenged me to make |fcolumn| compatible with 
% |longtable|.  Christian Hoff's request on column formatting triggered many
% happy hours of coding.
% \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.  The |booktabs| package.  Publication quality tables in \LaTeX.
% \bibitem{dcolumn} David Carlisle.  The |dcolumn| package.
% \bibitem{longtable} David Carlisle.  The |longtable| package.
% \bibitem{world_economy} According to the IMF \url{www.imf.org}.
% \bibitem{CandTB} Donald Knuth, Computers \& Typesetting/B, \TeX: the program.
% \bibitem{CandTA} Donald Knuth, Computers \& Typesetting/A, The \TeX book.
% \bibitem{TAOCP1} Donald Knuth, The Art of Computers Programming, volume~1.
% \bibitem{RJ} William Shakespeare, Romeo and Juliet, a tragedy (1597).
% \end{thebibliography}
% \Finale
\endinput