Fix dates and version numbers in the package files.
[mdwtools] / mdwtab.dtx
CommitLineData
86f6a31e 1% \begin{meta-comment} <general public licence>
2%%
3%% mdwtab package -- another rewrite of the tabular environment, etc.
8bc5bdd2 4%% Copyright (c) 1996, 2002, 2003, 2007, 2016, 2020 Mark Wooding
86f6a31e 5%%
6%% This program is free software; you can redistribute it and/or modify
7%% it under the terms of the GNU General Public License as published by
8%% the Free Software Foundation; either version 2 of the License, or
9%% (at your option) any later version.
10%%
11%% This program is distributed in the hope that it will be useful,
12%% but WITHOUT ANY WARRANTY; without even the implied warranty of
13%% MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
14%% GNU General Public License for more details.
15%%
16%% You should have received a copy of the GNU General Public License
17%% along with this program; if not, write to the Free Software
18%% Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
19%%
20% \end{meta-comment}
21%
22% \begin{meta-comment} <Package preambles>
23%<+mdwtab>\NeedsTeXFormat{LaTeX2e}
24%<+mdwtab>\ProvidesPackage{mdwtab}
af8af7eb 25%<+mdwtab> [2020/09/06 1.14.0 Table typesetting with style]
86f6a31e 26%<+mathenv>\NeedsTeXFormat{LaTeX2e}
27%<+mathenv>\ProvidesPackage{mathenv}
af8af7eb 28%<+mathenv> [2020/09/06 1.14.0 Various maths environments]
7abe48b3 29%<+colour>\NeedsTeXFormat{LaTeX2e}
30%<+colour>\ProvidesPackage{mtcolour}
af8af7eb 31%<+colour> [2020/09/06 1.14.0 Colour support for mdwtab]
7abe48b3 32%<+color>\NeedsTeXFormat{LaTeX2e}
33%<+color>\ProvidesPackage{mtcolor}
af8af7eb 34%<+color> [2020/09/06 1.14.0 Fix for people who can't spell]
86f6a31e 35% \end{meta-comment}
36%
912d2f01 37% \CheckSum{3424}
86f6a31e 38%% \CharacterTable
39%% {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
40%% 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
41%% Digits \0\1\2\3\4\5\6\7\8\9
42%% Exclamation \! Double quote \" Hash (number) \#
43%% Dollar \$ Percent \% Ampersand \&
44%% Acute accent \' Left paren \( Right paren \)
45%% Asterisk \* Plus \+ Comma \,
46%% Minus \- Point \. Solidus \/
47%% Colon \: Semicolon \; Less than \<
48%% Equals \= Greater than \> Question mark \?
49%% Commercial at \@ Left bracket \[ Backslash \\
50%% Right bracket \] Circumflex \^ Underscore \_
51%% Grave accent \` Left brace \{ Vertical bar \|
52%% Right brace \} Tilde \~}
53%%
54%
55% \begin{meta-comment}
56%
57%<*driver>
58\input{mdwtools}
59\describespackage{mdwtab}
7abe48b3 60\describespackage{mtcolour}
86f6a31e 61\describespackage{mathenv}
62\addcontents{lot}{\listoftables}
63\mdwdoc
64%</driver>
65%
66% \end{meta-comment}
67%
68%^^A-------------------------------------------------------------------------
69% \renewcommand{\tabstyle}{\small}
70%
71% \section{User guide}
72%
73%
74% The \package{mdwtab} package contains a reimplementation of the standard
75% \LaTeX\ \env{tabular} and \env{array} environments. This is not just an
76% upgraded version: it's a complete rewrite. It has several advantages over
77% the official \package{array} package (not raw \LaTeX's, which is even less
78% nice), and it's more-or-less compatible. Most of these are rather
79% technical, I'll admit.
80%
81% \begin{itemize}
82%
83% \item The newcolumn system is properly and perfectly integrated into the
84% system. There are now \emph{no} `primitive' column types -- all the
85% standard types are created as user-defined columns.
86%
87% \item You can define entirely different table-like environments using the
88% equipment here. It's still hard work, although less so than before.
89% I'll do an example of this some time.
90%
91% \item Construction of the preamble is generally much tidier. I've used
92% token registers rather than |\edef|, and it's all done very nicely.
93%
94% \item Fine spacing before and after rules (described by DEK as `a mark of
95% quality') is now utterly trivial, since the preamble-generator will
96% store the appropriate information.
97%
98% \item You can use \env{array} in LR and paragraph modes without having
99% to surround it with `|$|' signs.
100%
101% \item Usually you don't want tables in the middle of paragraphs. For these
102% cases, I've provided a simpler way to position the table
103% horizontally.
104%
105% \item Footnotes work properly inside \env{tabular} environments (hoorah!).
106% You can `catch' footnotes using the \env{minipage} environment if
107% you like. (It uses an internal version of the \package{footnote}
108% package to handle footnotes, which doesn't provide extra goodies like
109% the \env{footnote} environment; you'll need to load the full package
110% explicitly to get them.)
111%
112% \item Standard \LaTeX\ tabular environments have a problem with lining up
113% ruled tables. The |\firsthline| command given in the \textit{\LaTeX\
114% Companion} helps a bit, but it's not really good enough, and besides,
115% it doesn't \emph{actually} line the text up right after all. The
116% \package{mdwtab} package does the job properly to begin with, so you
117% don't need to worry.
118%
119% \end{itemize}
120%
121% I've tested the following packages with \package{mdwtab}, and they all
122% work. Some of the contortions required to make them work weren't pleasant,
123% but you don't need to know about them. By a strange coincidence, all the
124% packages were written by David Carlisle. Anyway, here's the list:
125% \begin{itemize}
126% \item The quite nice \package{dcolumn} package.
127% \item The more useful \package{delarray} package.
128% \item The rather spiffy \package{hhline} package.
129% \item The truly wonderful \package{tabularx} package.
130% \item The utterly magnificent \package{longtable} package.
131% \end{itemize}
132%
133% Note that I've looked at \package{supertabular} as well: it won't work, so
134% use \package{longtable} instead, 'cos it's much better.
135%
136%
137% \subsection{The downside}
138%
139% There's no such thing as a free lunch. The \package{mdwtab} environment
140% is not 100\% compatible with the \env{tabular} environment found in
141% \LaTeXe\ or the \package{array} package.
142%
143% The differences between \package{mdwtab} and \LaTeXe's \env{tabular}
144% environment are as follows:
145%
146% \begin{itemize} \synshorts \let\`=\lq
147%
148% \item The vertical spacing in \env{array} environments is different to
149% that in \env{tabular} environments. This produces more attractive
150% results in most mathematical uses of \env{array}s, in the author's
151% opinion. The spacing can be modified by playing with length
152% parameters.
153%
154% \item The presence of horizontal and vertical rules will alter the spacing
155% of the table (so a pair of columns separated by a `|' is wider than
156% a pair with no separation by "\\arrayrulewidth". This does mean that
157% horizontal and vertical rules match up properly -- the usual \LaTeX\
158% environment makes the horizontal rules stop just short of the edge
159% of the table, making an ugly mess (check out the \textit{\LaTeX\
160% book} if you don't believe me -- page~62 provides a good example).
161% The \package{array} package handles rules in the same way as
162% \package{mdwtab}.
163%
164% \setbox0=\hbox{\footnotesize`\\def\\xcs{\\tabskip=\\fill}'}
165% \setbox2=\hbox{\footnotesize`...@{\\span\\xcs}...'}
166% \item In common with the \package{array} package, there are some
167% restrictions on the use of the "\\extracolsep" command in preambles:
168% you may use at most one "\\extracolsep" command in each `@' or `!'
169% expression. Also, you can't say
170% \begin{listing}
171%\newcommand{\xcs}{\extracolsep{\fill}}
172% \end{listing}
173% and then expect something like `...@{\\xcs}...' to actually work --
174% the "\\extracolsep" mustn't be hidden inside any other
175% commands. Because things like `@' expressions aren't expanded at
176% the time, "\\extracolsep" has to be searched and processed
177% \`by hand'.\footnote{^^A
178% All \cs{extracolsep} does is modify the \cs{tabskip} glue, so
179% if you were an evil \TeX\ hacker like me, you could just say
180% \unhbox0\ and put \unhbox2\ in your preamble. That'd work nicely.
181% It also works with the \package{array} package.}
182%
183% \item Control sequences (commands) in a table's preamble aren't expanded
184% before the preamble is read. In fact, commands in the preamble are
185% considered to be column types, and their names are entirely
186% independent of normal \LaTeX\ commands. No column types of this
187% nature have yet been defined\footnote{^^A
188% There used to be an internal \cs{@magic} type used by
189% \env{eqnarray}, but you're not supposed to know about that.
190% Besides, it's not there any more.}
191% but the possibility's always there. Use the "\\newcolumntype" or
192% "\\coldef" commands to define new column types.
193%
194% \item The preamble parsing works in a completely different way. There is
195% a certain amount of compatibility provided, although it's heavily
196% geared towards keeping \package{longtable} happy and probably won't
197% work with other packages.
198%
199% \item Obscure constructs which were allowed by the old preamble parser but
200% violate the syntax shown in the next section (e.g., `|@{}|' to
201% suppress the "\\doublerulesep" space between two vertical rules,
202% described in \textit{The \LaTeX\ Companion} as \`a misuse of the
203% `@{...}' qualifier') are now properly outlawed. You will be given
204% an error message if you attempt to use such a construction.
205%
206% \item The `*' forms (which repeat column types) are now expanded at a
207% different time. Previously, preambles like `c@*{4}{{:}@}{--}c'
208% were considered valid (the example would expand to
209% `c@{:}@{:}@{:}@{:}@{--}c'), because `*'s were expanded before the
210% preamble was actually parsed. In the new system, `*' is treated
211% just like any other preamble character (it just has a rather odd
212% action), and preambles like this will result in an error (and
213% probably a rather confusing one).
214%
7abe48b3 215% \item David Carlisle's \package{colortbl} package entirely fails to work
216% with \package{mdwtab}. However, we now have colour support of our
eafdddad 217% own which is at times similar in style.
7abe48b3 218%
86f6a31e 219% \end{itemize}
220%
221% There are also several incompatibilities between \package{mdwtab} and
222% \package{array}:
223%
224% \begin{itemize} \synshorts \let\`=\lq
225%
226% \item Because of the way "\\newcolumntype" works in the \package{array}
227% package, a horrid construction like
228% \begin{listing}
229%\newcolumntype{x}{{:}}
230%\begin{tabular}{|c!xc|}
231% \end{listing}
232% is considered to be valid, and is interpreted as `|c!{:}c|'. My
233% reading of pages~54 and~55 of the \textit{\LaTeX\ book} tells me
234% that this sort of thing is forbidden in normal \LaTeX\ commands.
235% The \package{mdwtab} preamble parser now treats column type letters
236% much more like commands with the result that the hacking above won't
237% work any more. The construction above would actually be interpreted
238% as `|c!{x}c|' (i.e., the `x' column type wouldn't be expanded to
239% `{:}' because the parser noticed that it was the argument to the
240% `!' modifier\footnote{^^A
241% This is a direct result of the way \TeX\ treats undelimited
242% arguments. See chapters~5 and~20 of \textit{The \TeX book} for
243% more information about how grouping affects argument reading.}).
244%
245% \item Most of the points above, particularly those relating to the
246% handling of the preamble, also apply to the \package{array} package.
247% it's not such an advance over the \LaTeXe\ version as everyone said
248% it was.
249%
250% \end{itemize}
251%
252%
253% \subsection{Syntax}
254%
255% \DescribeEnv{tabular}
256% \DescribeEnv{tabular*}
257% \DescribeEnv{array}
258% So that everyone knows where I stand, here's a complete syntax for my
259% version of the \env{tabular} environment, and friends
260%
261% \begin{grammar}
262%
eafdddad 263% <tabular-env> ::= \[[
86f6a31e 264% "\\begin"
265% \begin{stack}
266% "{tabular}" \\ "{tabular*}" "{" <length> "}" \\
267% "{array}" \\ "{smarray}"
268% \end{stack}
269% \[ "[" <position-arg> "]" \]
270% "{" <preamble> "}" <text>
271% "\\end"
272% \( "{tabular}" \\ "{tabular*}" \\ "{array}" \\ "{smarray}" \)
273% \]]
274%
275% <position-arg> ::= (see below)
276%
eafdddad 277% <preamble> ::= \[[
86f6a31e 278% <first-column>
279% \[ \< <column> \> \]
280% \]]
281%
282% <first-column> ::= \[[ \[ <rule> \] <column> \]]
283%
eafdddad 284% <column> ::= \[[
86f6a31e 285% \[ <spacing> \] \[ \< <user-pre-text> \> \] <column-type>
286% \[ \< <user-post-text> \> \] \[ <spacing> \] \[ <rule> \]
287% \]]
288%
eafdddad 289% <spacing> ::= \[[ "@" "{" <text> "}" \]]
86f6a31e 290%
7abe48b3 291% <user-pre-text> ::= \[[ \[ "?" \] ">" "{" <text> "}" \]]
86f6a31e 292%
eafdddad 293% <column-type> ::= \[[
86f6a31e 294% \begin{stack}
295% \[ "T" \\ "M" \] \( "l" \\ "c" \\ "r" \) \\
296% \( "p" \\ "m" \\ "b" \) "{" <length> "}" \\
297% "#" "{" <raw-pre-text> "}" "{" <raw-post-text> "}"
298% \end{stack}
299% \]]
300%
7abe48b3 301% <user-post-text> ::= \[[ \[ "?" \] "<" "{" <text> "}" \]]
86f6a31e 302%
eafdddad 303% <rule> ::= \[[ \( "|" \\ "!" "{" <text> "}" \) \]]
86f6a31e 304%
305% \end{grammar}
306%
307% If you examine the above very carefully, you'll notice a slight deviation
308% from the original -- an |@|-expression \emph{following} a rule is
309% considered to be part of the \emph{next} column, not the current one. This
310% is, I think, an almost insignificant change, and essential for some of the
311% new features. You'll also notice the new |#| column type form, which
312% allows you to define new real column types instead of just modifying
313% existing ones. It's not intended for direct use in preambles -- it's
314% there mainly for the benefit of people who know what they're doing and
315% insist on using |\newcolumntype| anyway.
316%%
317% The actual column types are shown in table~\ref{tbl:columns}.
318%
319% \begin{table}
eafdddad
MW
320% \begin{tabular}[C]{| >{\synshorts} c | m{3in} |} \hlx{hv[1]}
321%
322% \multicolumn{2}{|c|}{\bf Column types} \\ \hlx{v[1]hv}
323% \bf Name & \bf Meaning \\ \hlx{vhv.}
324% "l" & Left aligned text (\env{tabular}) or
325% equation (\env{array}). \\ \hlx{.}
326% "c" & Centred text (\env{tabular}) or
327% equation (\env{array}). \\ \hlx{.}
328% "r" & Right aligned text (\env{tabular}) or
329% equation (\env{array}). \\ \hlx{vhv.}
86f6a31e 330% "Ml", "Mc" and "Mr" & Left, centre and right aligned
eafdddad 331% equations.* \\ \hlx{.}
86f6a31e 332% "Tl", "Tc" and "Tr" & Left, centre and right aligned
eafdddad 333% text.* \\ \hlx{vhv.}
86f6a31e 334% "p{"<width>"}" & Top aligned paragraph with the given
eafdddad 335% width. \\ \hlx{.}
86f6a31e 336% "m{"<width>"}" & Vertically centred paragraph with
eafdddad 337% the given width. \\ \hlx{.}
86f6a31e 338% "b{"<width>"}" & Bottom aligned paragraph with the
eafdddad 339% given width. \\ \hlx{vhv.}
86f6a31e 340% "#{"<pre>"}{"<post>"}" & User defined column type:
eafdddad
MW
341% \<pre> is inserted before the
342% cell entry, \<post> is inserted
343% afterwards.* \\ \hlx{vhhv[1]}
344%
345% \multicolumn{2}{|c|}{\bf Other modifier characters} \\ \hlx{v[1]hv}
346% \bf Name & \bf Meaning \\ \hlx{vhv.}
347% "|" & Inserts a vertical rule between
348% columns. \\ \hlx{.}
7abe48b3 349% "$*["<params>"]" & Inserts a vertical rule of given
86f6a31e 350% width between columns; "*" selects
7abe48b3 351% "\arraythickrulewidth".* \\ \hlx{.}
eafdddad
MW
352% "!{"<text>"}" & Inserts \<text> between columns,
353% treating it as a vertical rule. \\ \hlx{vhv.}
354% "@{"<text>"}" & Inserts \<text> instead of the
355% usual intercolumn space. \\ \hlx{vhv.}
356% ">{"<text>"}" & Inserts \<text> just before the
357% actual column entry. \\ \hlx{.}
358% "<{"<text>"}" & Inserts \<text> just after the
359% actual column entry. \\ \hlx{.}
c8da92a7 360% "?>{"<text>"}" & Inserts \<text> before the column
7abe48b3 361% entry \emph{and} the rules list.* \\ \hlx{.}
c8da92a7 362% "?<{"<text>"}" & Inserts \<text> after the column
912d2f01
MW
363% entry \emph{and} the rules list.* \\ \hlx{.}
364% "'{"<text>"}" & Inserts \<text> into the rules list
365% only.* \\ \hlx{vhv.}
86f6a31e 366% "*{"<count>"}{"<chars>"}" & Inserts \<count>
eafdddad
MW
367% copies of the \<chars> into the
368% preamble. \\ \hlx{vhs}
86f6a31e 369%
370% \multicolumn{2}{@{}l}{* This column type is a new feature}
371% \end{tabular}
372%
373% \caption{\package{array} and \package{tabular} column types and modifiers}
374% \label{tbl:columns}
375% \end{table}
376%
377% Now that's sorted everything out, there shouldn't be any arguments at all
378% about what a column means.
379%
380% The lowercase \<position-arg>s \lit{t}, \lit{c} and \lit{b} do exactly
381% what they did before: control the vertical positioning of the table. The
382% uppercase ones control the \emph{horizontal} positioning -- this is how you
383% create \emph{unboxed} tables. You can only create unboxed tables in
384% paragraph mode.
385%
386% Note that unboxed tables still can't be broken across pages. Use
387% the \package{longtable} package for this, because it already does an
388% excellent job.
389%
390% \DescribeMacro{\tabpause}
391% One thing you can to with unboxed tables, however, is to `interrupt' them,
392% do some normal typesetting, and then continue. This is achieved by the
393% |\tabpause| command: its argument is written out in paragraph mode, and
394% the table is continued after the argument finishes.
395% Note that it isn't a real argument as far as commands like |\verb| are
396% concerned -- they'll work inside |\tabpause| without any problems.
397%
398% \DescribeMacro{\vline}
86f6a31e 399% The |\vline| command draws a vertical rule the height of the current table
400% cell (unless the current cell is being typeset in paragraph mode -- it
401% only works in the simple LR-mode table cells, or in \lit{@} or \lit{!}
7abe48b3 402% modifiers). It's now been given an optional argument which describes
403% parameters for the line. See section~\ref{sec:ruleparams}.
86f6a31e 404%
405% { \let\tabstyle=\relax
406% \begin{demo}{An example of \cmd\vline}
407%\large
408%\begin{tabular}
409% {| c !{\vline[5pt]} c | c |}
410% \hlx{hv}
411% \bf A & \it B & \sf C \\
412% \hlx[2pt]{vhv}
413% \bf D & \it E & \sf F \\
414% \hlx{vh}
415%\end{tabular}
416% \end{demo}
417% }
418%
419% \DescribeMacro{smarray}
420% You've probably noticed that there's an unfamiliar environment mentioned
421% in the syntax shown above. The \env{smarray} environment produces a
422% `small' array, with script size cells rather than the normal full text
423% size cells. I've seen examples of this sort of construction\footnote{^^A
424% There's a nasty use of \env{smallmatrix} in the |testmath.tex| file which
425% comes with the \package{amslatex} distribution. It's actually there to
426% simulate a `smallcases' environment, which the \package{mathenv} package
7abe48b3 427% includes, based around \env{smarray}.} ^^A
86f6a31e 428% being implemented by totally unsuitable commands. Someone may find it
429% handy.
430%
431%
432% \subsection{An updated \cs{cline} command}
433%
434% \DescribeMacro{\cline}
435% \DescribeMacro{\hline}
436% The standard \LaTeX\ |\cline| command has been updated. As well as just
437% passing a range of columns to draw lines through, you can now pass a comma
438% separated list of column numbers and ranges:
439%
440% \begin{grammar}
eafdddad 441% <cline-cmd> ::= \[[
86f6a31e 442% "\\cline"
7abe48b3 443% \[ "*" \] \\ \[ "[" <rule-params> "]" \]
86f6a31e 444% "{" \< <number> \[ "-" <number> \] \\ "," \> "}"
445% \]]
446% \end{grammar}
447%
7abe48b3 448% The rules printed by |\cline| and |\hline| can be modified by rule
449% parameters: see section~\ref{sec:ruleparams}.
86f6a31e 450%
7abe48b3 451% Note that |\cline| rules are rather bodgy. Other rules now have
452% \emph{thickness}, but |\cline|s don't. Instead, they backspace over
453% previous stuff. If you don't like that, insert an appropriate gap, using
454% |\vgap|. The \lit{z} rune in |\hlx| is perfect for this kind of thing --
455% precede your \lit{c} lines by \lit{z} lines for best results.
86f6a31e 456%
457% {\let\tabstyle\relax
458% \begin{demo}[w]{A \cs{cline} example}
459%\newcommand{\mc}{\multicolumn{1}}
460%\begin{tabular}[C]{|c|c|c|c|} \cline{2,4}
461% \mc{c|}{one} & two & three & four \\ \hline
462% five & six & seven & \mc{c}{eight} \\ \cline{1,3}
463%\end{tabular}
464% \end{demo}
465% }
466%
7abe48b3 467% \subsection{Other stuff}
468%
469% \DescribeMacro\nextrow
470% The \env{tabular} and \env{array} environments maintain a counter
471% \textsf{tabrow}. The counter is reset to zero at the start of each table.
472% It is stepped by one by default; you can provide an optional argument which
473% is the amount to add.
474%
475%
86f6a31e 476% \subsection{Spacing control}
477%
478% One of the most irritating things about \LaTeX's tables is that there isn't
479% enough space around horizontal rules. Donald Knuth, in \textit{The
480% \TeX book}, describes addition of some extra vertical space here as `a mark
481% of quality', and since \TeX\ was designed to produce `beautiful documents'
482% it seems a shame that \LaTeX\ doesn't allow this to be done nicely. Well,
483% it does now.
484%
485% \DescribeMacro{\vgap}
486% The extra vertical space is added using a command |\vgap|, with the
487% following syntax:
488%
489% \begin{grammar}
490%
eafdddad 491% <vgap-cmd> ::= \[[
86f6a31e 492% "\\vgap" \[ "[" <which-cols> "]" \] "{" <length> "}"
493% \]]
494%
eafdddad 495% <which-cols> ::= \[[ \< <number> \[ "-" <number> \] \\ "," \> \]]
86f6a31e 496%
497% \end{grammar}
498%
499% This command must appear either immediately after the beginning of the
500% table or immediately after the |\\| which ends a row. (Actually, there are
501% other commands which also have this requirement -- you can specify a
502% collection of them wherever you're allowed to give any one.) It adds some
503% vertical space (the amount is given by the \<length>) to the table,
504% making sure that the vertical rules of the table are extended correctly.
505%
506% The |\vgap| command relies on information stored while your table preamble
507% is being examined. However, it's possible that you might not want some
508% of the rules drawn (e.g., if you've used |\multicolumn|). The optional
509% \<which-cols> argument allows you to specify which rules are \emph{not}
510% to be drawn. You can specify either single column numbers or ranges. The
511% rule at the very left hand side is given the number~0; the rules at the
512% end of column~$n$ are numbered~$n$. It's easy really.
513%
514% \DescribeMacro{\hlx}
515% Using |\vgap| is all very well, but it's a bit cumbersome, and takes up a
516% lot of typing, especially when combined with |\hline| commands. The |\hlx|
517% command tries to tidy things.
518%
519% The syntax is simple:
520% \begin{grammar}
521%
eafdddad 522% <hlx-cmd> ::= \[[
86f6a31e 523% "\\hlx"
7abe48b3 524% \[ "*" \] \[ "[" <rule-params> "]" \]
86f6a31e 525% "{"
526% \begin{rep}
527% \begin{stack}
528% "h" \\
529% \tok{"v["<which-cols>"]["<length>"]"} \\
7abe48b3 530% \tok{"z["<which-cols>"]["<length>"]"} \\
86f6a31e 531% \tok{"s["<length>"]"} \\
532% \tok{"c{"<which-cols>"}"} \\
533% "b" \\
534% \tok{"/["<number>"]"} \\
7abe48b3 535% \tok{"!{"<rule-params>"}"} \\
536% \tok{"?{"<stuff>"}"} \\
537% \tok{"+["<step>"]"} \\
86f6a31e 538% "."
539% \end{stack}
540% \end{rep}
541% "}"
542% \]]
543%
544% \end{grammar}
7abe48b3 545% The |*| or optional \<rule-params> give rule-drawing parameters for the |h|
546% and |c| subcommands. (Note that you can't pass a |*| or an optional
547% parameters argument to the |h| or |c| subcommands directly.) See
e8e9e5d8 548% section~\ref{sec:ruleparams}.
86f6a31e 549%
550% The argument works a bit like a table preamble, really. Each letter is a
551% command. The following are supported:
552%
553% \begin{description}
554%
555% \item [\lit*{h}] Works just like |\hline|. If you put two adjacent to each
556% other, a gap will be put between them.
557%
558% \item [\lit*{v[}\<which-cols>\lit*{][}\<length>\lit*{]}] Works
559% like \syntax{"\\vgap["<which-cols>"]{"<length>"}"}. If the
560% \<length> is omitted, the value of |\doublerulesep| is used.
561% This usually looks right.
562%
7abe48b3 563% \item [\lit*{z[}\<which-cols>\lit*{][}\<length>\lit*{]}] Like \lit{v},
564% except that the default gap is the current rule width (set by the
eafdddad
MW
565% \<rule-params>) rather than |\doublerulesep|. This is a good thing
566% to insert before a |\cline| row.
7abe48b3 567%
86f6a31e 568% \item [\lit*{s[}\<length>\lit*{]}] Leaves a vertical gap with the
569% given size. If you omit the \<length> then |\doublerulesep| is
570% used. This is usually right.
571%
572% \item [\lit*{c\char`\{}\<which-cols>\lit*{\char`\}}] Works just like
573% |\cline|.
574%
575% \item [\lit*{b}] Inserts a backspace the width of a rule. This is useful
576% when doing \package{longtable}s.
577%
578% \item [\lit*{/[}\<number>\lit*{]}] Allows a page break in a table. Don't
579% use this except in a \env{longtable} environment. The \<number>
580% works exactly the same as it does in the |\pagebreak| command,
581% except that the default is 0, which just permits a break without
582% forcing it.
583%
7abe48b3 584% \item [\lit*{!\char`\{}\<rule-params>\lit*{\char`\}}] Change the rule
585% parameters to be used for subsequent subcommands.
586%
587% \item [\lit*{?\char`\{}\<stuff>\lit*{\char`\}}] Do \<stuff>, which can be
588% any commands which \emph{don't} typeset anything.
589%
590% \item [\lit*{+[\<step>]}] Add \<step> (default is 1) to the value of the
591% \textsf{tabrow} counter.
592%
86f6a31e 593% \item [\lit*{.}] (That's a dot) Starts the next row of the table. No
594% more characters may follow the dot, and no |\hline|, |\hlx|, |\vgap|
595% or |\multicolumn| commands may be used after it. You don't have to
596% include it, and most of the time it's totally useless. It can be
597% handy for some macros, though. I used it in (and in fact added it
598% especially for) the table of column types.
599%
600% \end{description}
601%
602% An example of the use of |\hlx| is given, so you can see what's going on.
603%
604% \begin{figure}
605% \let\tabstyle\relax
606% \begin{demo}[w]{Beautiful table example}
607%\newcommand{\zerowidth}[1]{\hbox to 0pt{\hss#1\hss}}
608%\setlength{\tabcolsep}{1.5em}
609%\begin{tabular}[C]{| r | c | r |} \hlx{hv[1,2]}
610% \multicolumn{3}{|c|}{\bf AT\&T Common Stock} \\ \hlx{v[1,2]hv}
611% \multicolumn{1}{|c|}{\zerowidth{\bf Year}} &
612% \multicolumn{1}{c|}{\zerowidth{\bf Price}} &
613% \multicolumn{1}{c|}{\zerowidth{\bf Dividend}} \\ \hlx{vhv}
614% 1971 & 41--54 & \$2.60 \\
615% 2 & 41--54 & 2.70 \\
616% 3 & 46--55 & 2.87 \\
617% 4 & 40--53 & 3.24 \\
618% 5 & 45--52 & 3.40 \\
619% 6 & 51--59 & .95\rlap{*} \\ \hlx{vhs}
620% \multicolumn{3}{@{}l}{* (first quarter only)}
621%\end{tabular}
622% \end{demo}
623% \end{figure}
624%
625%
626% \subsection{Creating beautiful long tables}
627%
628% You can use the |\vgap| and |\hlx| commands with David Carlisle's
629% stunning \package{longtable} package. However, there are some things you
630% should be away of to ensure that your tables always come out looking
631% lovely.
632%
633% The \package{longtable} package will break a table at an |\hline| command,
634% leaving a rule at the bottom of the page and another at the top of the
635% next page. This means that a constructions like |\hlx{vhv}| will be
636% broken into something like |\hlx{vh}| at the bottom of the page and
637% |\hlx{hv}| at the top of the next. You need to design the table headers
638% and footers with this in mind.
639%
640% However, there appears to be a slight problem:\footnote
641% {You might very well call it a bug. I couldn't possibly comment.}
642% if the footer starts with an |\hline|, and a page is broken at an |\hline|,
643% then you get an extra thick rule at the bottom of the page. This is a bit
644% of a problem, because if the rule isn't there in the footer and you get
645% a break between two rows \emph{without} a rule between them, then the page
e8e9e5d8 646% looks very odd.
86f6a31e 647%
648% If you want to do ruled longtables, I'd recommend that you proceed as
649% follows:
650% \begin{itemize}
651% \item End header sections with an |\hlx{vh}|.
652% \item Begin footer sections with an |\hlx{bh}|.
653% \item Begin the main table with |\hlx{v}|.
654% \item Insert |\hlx{vhv}| commands in the main table body as usual.
655% \end{itemize}
656% If \package{longtable} gets modified appropriately, the use of the \lit{b}
657% command won't be necessary.
658%
659% Here's an example of the sort of thing you'd type.
660%
661% \begin{listinglist} \listingsize
662% \verb"\begin{longtable}[c]{|c|l|} \hlx{hv}" \\
663% \verb"\bf Heading & \bf Also heading \\ \hlx{vh}" \\
664% \verb"\endhead" \\
665% \verb"\hlx{bh}" \\
666% \verb"\endfoot" \\
667% \verb"\hlx{v}" \\
668% \verb"First main & table line \\ \hlx{vhv}" \\
669% \verb"Lots of text & like this \\ \hlx{vhv}" \\
670% \null\quad\vdots \\
671% \verb"Lots of text & like this \\ \hlx{vhv}" \\
672% \verb"Last main & table line \\ \hlx{vh}" \\
673% \verb"\end{longtable}"
674% \end{listinglist}
675%
676%
677% \subsection{Rules and vertical positioning}
678%
679% In the \LaTeXe\ and \package{array.sty} versions of \env{tabular}, you run
680% into problems if you try to use ruled tables together with the \lit{[t]} or
681% \lit{[b]} position specifiers -- the top or bottom rule ends up being
682% nicely lined up with the text baseline, giving you an effect which is
683% nothing like the one you expected. The \textit{\LaTeX\ Companion} gives
684% two commands |\firsthline| and |\lasthline| which are supposed to help with
685% this problem. (These commands have since migrated into the \package{array}
686% package.) Unfortunately, |\firsthline| doesn't do its job properly --
687% it gets the text position wrong by exactly the width of the table rules.
688%
689% The \package{mdwtab} package makes all of this automatic. It gets the
690% baseline positions exactly right, whether or not you use rules. Earlier
691% versions of this package required that you play with a length parameter
692% called |\rulefudge|; this is no longer necessary (or even possible -- the
693% length parameter no longer exists). The package now correctly compensates
694% for all sorts of rules and |\vgap|s at the top and bottom of a table and
695% it gets the positioning right all by itself. You've never had it so good.
696%
697%
7abe48b3 698% \subsection{Rule parameters}
699% \label{sec:ruleparams}
700%
701% The rule-drawing commands |\hline|, |\vline|, |\cline| and |\hlx|, and the
702% |$| column type (which is otherwise a synonym for "|") accept \emph{rule
703% parameters}. If the command is followed by a |*|, then the rules are a bit
704% thicker than usual -- they use |\arraythickrulewidth| rather than
705% |\arrayrulewidth|. However, there's an optional argument which can contain
706% one of:
707%
78cdb9cc 708% \begin{description} \setdescriptionlabel{\ttfamily#1}
7abe48b3 709% \item[thin] Use |\arrayrulewidth| as the line width. This is the default.
710% \item[thick] Use |\arraythickrulewidth| as the line width. This is the
711% same as giving a |*| after the command.
712% \item[width=\<length>] Make the rules exactly \<length> wide.
713% \item[\<length>] The same as \texttt{width=\<length>}, for compatibility.
714% \end{description}
715%
716% More of these keywords will be added later if past experience is anything
717% to go by. Note that the individual |\hlx| subcommands \emph{don't} take
718% rule parameters, but see the |!| subcommand for updating the current
719% parameters.
720%
721% \DescribeMacro\tabsetruleparams
722% If you say \syntax{"\\tabsetruleparams{"<rule-params>"}"} then the
723% \<rule-params> will be prepended to any parameters provided to specific
724% rule-drawing commands (including the \lit{\char`\|} preamble command). For
725% example, |\tabsetruleparams{thick}| makes all rules thick. This is a local
726% declaration.
727%
728%
86f6a31e 729% \subsection{User serviceable parts}
730%
731% There are a lot of parameters which you can modify in order to make arrays
732% and tables look nicer. They are all listed in table~\ref{tbl:config}.
733%
734% \begin{table}
eafdddad
MW
735% \begin{tabular}[C]{| l | m{3in} |} \hlx{hv}
736% \bf Parameter & \bf Meaning \\ \hlx{vhv}
737% |\tabstyle| & A command executed at the beginning of
738% a \env{tabular} or \env{tabular$*$}
739% environment. By default does nothing.
740% Change using |\renewcommand|. \\ \hlx{vhv}
741% |\extrarowheight| & A length added to the height of every
742% row, used to stop table rules
743% overprinting ascenders. Default 0\,pt.
744% Usage is deprecated now: use |\hlx|
745% instead. \\ \hlx{vhv}
746% |\tabextrasep| & Extra space added between rows in a
747% \env{tabular} or \env{tabular$*$}
748% environment (added \emph{before} any
749% following |\hline|). Default 0\,pt. \\
750% |\arrayextrasep| & Analogous to |\tabextrasep|, but for
751% \env{array} environments. Default
752% 1\,jot (3\,pt). \\
753% |\smarrayextrasep| & Analogous to |\tabextrasep|, but for
754% \env{smarray} environments. Default
755% 1\,pt. \\ \hlx{vhv}
756% |\tabcolsep| & Space added by default on each side of
757% a table cell (unless suppressed by an
758% \lit{@}-expression) in \env{tabular}
759% environments. Default is defined by
760% your document class. \\
761% |\arraycolsep| & Analogous to |\tabcolsep|, but for
762% \env{array} environments. Default is
763% defined by your document class. \\
764% |\smarraycolsep| & Analogous to |\tabcolsep|, but for
765% \env{smarray} environments. Default
766% is 3\,pt. \\ \hlx{vhv}
767% |\arrayrulewidth| & The width of horizontal and vertical
768% rules in tables. \\
769% |\arraythickrulewidth|& The width of starred rules in tables. \\
770% |\doublerulesep| & Space added between two adjacent
771% vertical or horizontal rules. Also
772% used by |\hlx{v}|. \\ \hlx{vhv}
773% |\arraystretch| & Command containing a factor to
774% multiply the default row height.
775% Default is defined by your document
776% class (usually 1). \\ \hlx{vh}
86f6a31e 777% \end{tabular}
778%
779% \caption{Parameters for configuring table environments}
780% \label{tbl:config}
781%
782% \end{table}
783%
784%
785% \subsection{Defining column types}
786%
787% \DescribeMacro{\newcolumntype}
788% The easy way to define new column types is using |\newcolumntype|. It
789% works in more or less the same way as |\newcommand|:
790%
791% \begin{grammar}
792%
793% <new-col-type-cmd> ::= \[[
794% "\\newcolumntype"
795% "{" <column-name> "}"
796% \[ "[" <num-args> "]" \]
797% \[ "[" <default-arg> "]" \]
798% "{" <first-column> \[ \< <column> \> \] "}"
799% \]]
800%
801% \end{grammar}
802%
803% (The \env{array.sty} implementation doesn't accept the \<default-arg>
804% argument. I've no idea why not, 'cos it was very easy to implement.)
805%
806% \DescribeMacro{\colset}
807% This implementation allows you to define lots of different sets of columns.
808% You can change the current set using the |\colset| declaration:
809% \begin{grammar}
eafdddad 810% <colset-cmd> ::= \[[ "\\colset" "{" <set-name> "}" \]]
86f6a31e 811% \end{grammar}
812% This leaves a problem, though: at any particular moment, the current
813% column set could be anything, since other macros and packages can change
814% it.
815%
816% \DescribeMacro{\colpush}
817% \DescribeMacro{\colpop}
818% What actually happens is that a stack of column sets is maintained. The
819% |\colset| command just replaces the item at the top of the stack. The
820% command |\colpush| pushes its argument onto the top of the stack, making
821% it the new current set. The corresponding |\colpop| macro (which doesn't
822% take any arguments) removes the top item from the stack, reinstating the
823% previous current column set.
824%
825% \begin{grammar}
eafdddad
MW
826% <colpush-cmd> ::= \[[ "\\colpush" "{" <set-name> "}" \]]
827% <colpop-cmd> ::= \[[ "\\colpop" \]]
86f6a31e 828% \end{grammar}
829%
830% The macros which manipulate the column set stack work \emph{locally}.
831% The contents of the stack are saved when you open a new group.
832%
833% To make sure everyone behaves themselves properly, these are the rules for
834% using the column set stack:
835%
836% \begin{itemize}
837%
838% \item Packages defining column types must ensure that they preserve the
839% current column set. Either they must push their own column type
840% and pop it off when they're finished defining columns, or they must
841% avoid changing the stack at all, and use the optional arguments to
842% |\coldef| and |\collet|.
843%
844% \item Packages must not assume that any particular column set is current
845% unless they have made sure of it themselves.
846%
847% \item Packages must ensure that they pop exactly as much as they push.
848% There isn't much policing of this (perhaps there should be more),
849% so authors are encouraged to behave responsibly.
850%
851% \item Packages must change the current column set (using |\colset|) when
852% they start up their table environment. This will be restored when
853% the environment closes.
854%
855% \end{itemize}
856%
857% \DescribeMacro{\coldef}
858% |\newcolumntype| is probably enough for most purposes. However, Real
859% \TeX nicians, and people writing new table-generating environments, require
860% something lower-level.
861%
862% \begin{grammar}
eafdddad 863% <coldef-cmd> ::= \[[
86f6a31e 864% "\\coldef"
865% \[ "[" <set-name> "]" \]
866% <col-name> <arg-template> "{" <replacement-text> "}"
867% \]]
868% \end{grammar}
869%
870% Note that this defines a column type in the current colset. It works
871% almost exactly the same way as \TeX's primitive |\def|. There is a
872% potential gotcha here: a |\tab@mkpream| token is inserted at the end of
873% your replacement text. If you need to read an optional argument or
874% something, you'll need to gobble this token before you carry on. The
875% |\@firstoftwo| macro could be handy here:
876% \begin{listing}
877%\coldef x{\@firstoftwo{\@ifnextchar[\@xcolumn@i\@xcolumn@ii}}}
878% \end{listing}
879% This isn't a terribly pretty state of affairs, and I ought to do something
880% about it. I've not seen any use for an optional argument yet, though.
881% Note that if you do gobble the |\tab@mkpream|, it's your responsibility to
882% insert another one at the very end of your macro's expansion (so that
883% further preamble characters can be read).
e8e9e5d8 884%
86f6a31e 885% The replacement text is inserted directly. It's normal to insert preamble
886% elements here. There are several to choose from:
887%
888% \begin{description}
889%
890% \item [Column items] provide the main `meat' of a column. You insert a
891% column element by saying
892% \syntax{"\\tabcoltype{"<pre-text>"}{"<post-text>"}"}.
893% The user's text gets inserted between these two. (So do user pre-
894% and post-texts. Bear this in mind.)
895%
896% \item [User pre-text items] work like the \lit{>} preamble command. You
897% use the \syntax{"\\tabuserpretype{"<text>"}"} command to insert it.
898% User pre-texts are written in \emph{reverse} order between the
899% pre-text of the column item and the text from the table cell.
900%
901% \item [User post-text items] work like the \lit{<} preamble command. You
902% use the \syntax{"\\tabuserposttype{"<text>"}"} command to insert it.
903% Like user pre-texts, user post-texts are written in reverse order,
904% between the table cell text and the column item post-text.
905%
906% \item [Space items] work like the \lit{@} preamble command. They're
907% inserted with the \syntax{"\\tabspctype{"<text>"}"} command.
908%
909% \item [Rule items] work like the `\verb"|"' and \lit{!} commands. You
910% insert them with the \syntax{"\\tabruletype{"<text>"}"} command.
911% Note that the text is inserted by |\vgap| too, so it should contain
912% things which adjust their vertical size nicely. If you really need
913% to, you can test |\iftab@vgap| to see if you're in a |\vgap|.
914%
915% \end{description}
916%
917% \DescribeMacro{\collet}
918% As well as defining columns, you can copy definitions (rather like |\let|
919% allows you to copy macros). The syntax is like this:
920%
921% \begin{grammar}
922%
eafdddad 923% <collet-cmd> ::= \[[
86f6a31e 924% \[ "[" <set-name> "]" \] <col-name> \[ "=" \] \[ "[" <set-name> "]" \]
925% <col-name>
926% \]]
927%
928% \end{grammar}
929%
930% (In other words, you can copy definitions from other column sets.)
931%
932%
933% \subsection{Defining new table-generating environments}
934%
935% Quite a few routines are provided specifically to help you to define new
936% environments which do alignment in a nice way.
937%
938% \subsubsection{Reading preambles}
939%
940% The main tricky bit in doing table-like environments is parsing preambles.
941% No longer.
942%
943% \DescribeMacro{\tab@readpreamble}
944% \DescribeMacro{\tab@doreadpream}
945% The main parser routine is called |\tab@doreadpream|. Given a user
946% preamble string as an argument, it will build an |\halign| preamble to
947% return to you. However, the preamble produced won't be complete. This is
948% because you can actually make multiple calls to |\tab@doreadpream| with
949% bits of user preambles. The |\newcolumntype| system uses this mechanism,
950% as does the \lit{*} (repeating) modifier. When there really is no more
951% preamble to read, you need to \emph{commit} the heldover tokens to the
952% output. The |\tab@readpreamble| routine will do this for you -- given a
953% user preamble, it builds a complete output from it.
954%
955% A token register |\tab@preamble| is used to store the generated preamble.
956% Before starting, you must initialise this token list to whatever you want.
957% There's another token register, |\tab@shortline|, which is used to store
958% tokens used by |\vgap|. For each column in the table, the list contains
959% an |\omit| (to override the standard preamble) and an |\hfil| space taking
960% up most of the column. Finally, for each rule item in the user preamble,
961% the shortline list contains an entry of the form:
962% \begin{quote} \synshorts
963% "\\tab@ckr{"<column-number>"}{"<rule-text>"}"
964% \end{quote}
965% This is used to decide whether to print the rule or an empty thing of the
966% same width. You probably ought to know that the very first column does
967% \emph{not} have a leading |\omit| -- this is supplied by |\vgap| so that
968% it can then look for optional arguments.
969%
970% \DescribeMacro{\tab@initread}
971% As well as initialising |\tab@preamble| and emptying |\tab@shortline|,
972% there are several other operations required to initialise a preamble read.
973% These are all performed by the |\tab@initread| macro, although you may want
974% to change some of the values for your specific application. For reference,
975% the actions performed are:
976% \begin{itemize}
977% \item initialising the parser state by setting $|\tab@state| =
978% |\tab@startstate|$;
979% \item clearing the token lists |\tab@preamble| and |\tab@shortlist|;
980% \item initialising the macros |\tab@tabtext|, |\tab@midtext|, and
981% |\tab@multicol| to their default values of `|&|',
982% `|\ignorespaces#\unskip|' and the empty token list respectively.^^A
983% \footnote{^^A
984% These are macros rather than token lists to avoid hogging all
985% the token list registers. Actually, the package only allocates
986% two, although it does use almost all of the temporary registers as
987% well. Also, there's a lie: \cs{unskip} is too hamfisted to remove
988% trailing spaces properly; I really use a macro called
989% \cs{@maybe@unskip}}
990% \item clearing the internal token list registers |\tab@pretext|,
991% |tab@userpretext| and |\tab@posttext|;
992% \item clearing the column counter |\tab@columns| to zero;
993% \item clearing the action performed when a new column is started (by making
994% the |\tab@looped| macro equal to |\relax|; this is used to make
995% |\multicolumn| macro raise an error if you try to do more than one
996% column); and
997% \item setting up some other switches used by the parser (|\iftab@rule|,
998% |\iftab@initrule| and |\iftab@firstcol|, all of which are set to be
999% |true|).
1000% \end{itemize}
1001%
1002% The macro |\tab@multicol| is used by the |\multicolumn| command to insert
1003% any necessary items (e.g., struts) before the actual column text. If you
1004% set this to something non-empty, you should probably consider adding a
1005% call to the macro to the beginning of |\tab@preamble|.
1006%
1007% When parsing is finally done, the count register |\tab@columns| contains
1008% the number of columns in the alignment. Don't corrupt this value, because
1009% it's used for handling |\hline| commands.
1010%
1011% \subsubsection{Starting new lines}
1012%
1013% The other messy bit required by table environments is the newline command
1014% |\\|. There are nasty complications involved with starting new lines, some
1015% of which can be handled by this package, and some on which I can only give
1016% advice.
1017%
1018% \DescribeMacro{\tab@cr}
1019% The optional arguments and star-forms etc. can be read fairly painlessly
1020% using the |\tab@cr| command:
1021%
1022% \begin{grammar}
eafdddad 1023% <tabcr-cmd> ::= \[[
86f6a31e 1024% "\\tab@cr" <command> "{" <non-star-text> "}" "{" <star-text> "}"
1025% \]]
1026% \end{grammar}
1027%
1028% This will call your \<command> with two arguments. The first is the
1029% contents of the optional argument, or `|\z@|' if there wasn't one. The
1030% second is either \<star-text> or \<non-star-text> depending on
1031% whether the user wrote the $*$-form or not.
1032%
1033% Somewhere in your \<command>, you'll have to use the |\cr| primitive to
1034% end the table row. After you've done this, you \emph{must} ensure that you
1035% don't do anything that gets past \TeX's mouth without protecting it --
1036% otherwise |\hline| and co.\ won't work. I usually wrap things up in a
1037% |\noalign| to protect them, although there are other methods. Maybe.
1038%
1039% You might like to have a look at the \env{eqnarray} implementation provided
1040% to see how all this gets put into practice.
1041%
1042%
7abe48b3 1043% \subsection{Colour support}
1044%
1045% I've now added colour support to \package{mdwtab}. That is, you can play
1046% with the colours of table cell backgrounds, rules and text. The support
1047% isn't there by default: you have to either give the \textsf{colour} option
1048% when you load \package{mdwtab}, or include the \package{mtcolour} package
1049% yourself. It's very new, and might break. It's probably not as good as
1050% \package{colortbl}. I prefer English spellings for the commands and
1051% declarations: to reduce confusion, I've provided synonyms with fewer `u's.
1052% If only American package authors were so thoughtful. The examples in this
1053% part of the documentation may not display correctly in some DVI viewers:
1054% for best results, run |dvips| and view the PostScript using (say)
1055% GhostScript.
1056%
1057% \subsubsection{New commands and features}
1058%
1059% \DescribeMacro\cellcolour
1060% The |\cellcolour| command changes the background colour for the current
1061% cell. You can use it directly in a table cell, or in the table preamble.
1062% It doesn't matter whereabouts in the table cell it goes. Note that
1063% unlike the \package{colortbl}, the |\cellcolour| command works on the
1064% \emph{entire} contents of the cell, including the |\tabcolsep| space and
1065% the rules, if any. That means that it's robust even if there are |@{...}|
1066% preamble commands present.
1067%
1068% The actual syntax is like this:
1069%
1070% \begin{grammar}
1071% <cell-colour-cmd> ::= \[[
1072% \( "\\cellcolour" \\ "\\cellcolor" \)
1073% \[ "*" \]
1074% \[ "[" <colour-model> "]" \]
1075% "{" <colour> "}"
e8e9e5d8 1076% \[ "[" <left-overhang> "]"
7abe48b3 1077% \[ "[" <right-overhang> "]" \] \]
1078% \]]
1079% \end{grammar}
1080%
1081% The \lit{*} makes |\cellcolour| override an extant |\rowcolour| command
1082% (see below). The \<colour-model> and \<colour> are as for the |\color|
1083% command. The \<left-overhang> is how much the colour band should stick out
1084% to the left of the cell; and similarly for the \<right-overhang>. If you
1085% don't give a \<right-overhang> then the same value is used for both; if you
1086% give neither then there's no overhang. The reason you might want overhang
1087% is to deal with |\extracolsep| glue. I shouldn't worry about it if I were
1088% you.
1089%
1090% It's very useful to use |\cellcolour| in a preamble, in particular, in the
1091% |?>| preamble command (which was added specifically). (If you use only |>|
1092% then |\vgap| leaves very odd-looking gaps in the table.)
1093%
1094% { \let\tabstyle=\relax
1095% \begin{demo}{A coloured table}
1096%\newcolumntype{\c}[2]{%
1097% >{\color{#1}}%
1098% ?>{\cellcolour{#2}}%
1099%}
1100%\begin{tabular}
1101% {|\c{cyan}{red}c|
1102% \c{magenta}{green}c|
1103% \c{yellow}{blue}c|}
1104% \hlx{hv}
1105% One &Two &Three \\ \hlx{vhv}
1106% Four &Five &Six \\ \hlx{vhv}
1107% Seven&Eight&Nine \\ \hlx{vh}
1108%\end{tabular}
1109% \end{demo}
1110% }
1111%
1112% Obviously, judicious use of |\newcolumntype| would abbreviate the above
1113% considerably.
1114%
1115% \DescribeMacro\rowcolour
1116% \DescribeMacro\rowcolouroff
1117% The |\rowcolour| command changes the background colour in the same way as
1118% |\cellcolour|; however, its effect takes precedence over |\cellcolour| (but
1119% not |\cellcolour*|) if both are active, and isn't automatically turned off
1120% at the start of the next cell. To actually turn it off again, say
1121% |\rowcolouroff|.
1122%
1123% \begin{grammar}
1124% <row-colour-cmd> ::= \[[
1125% \( "\\rowcolour" \\ "\\rowcolor" \)
1126% \[ "[" <colour-model> "]" \]
1127% "{" <colour> "}"
1128% \]]
1129% \end{grammar}
1130%
1131% Note that you don't get to specify overhang parameters here. The ones from
1132% the |\cellcolour| declaration are used, unless there isn't one in which
1133% case there aren't any.
1134%
1135% \DescribeMacro\ifmod
1136% A common thing to do is colour alternate rows of the table differently.
1137% This is a bit tricker for \package{mdwtab} than it would be for, say,
1138% \package{array}, since it's hard to spot where the `rows' actually change.
1139% The solution is to use the \textsf{tabrow} counter, and |\ifmod|. Saying
1140% say \syntax{"\\ifmod{"$x$"}{"$m$"}{"$y$"}{"<yes>"}{"<no>"}"} is the same as
1141% saying \<yes> if $x \bmod m = y$, and \<no> otherwise. This is typically
1142% used as follows.
1143%
1144% % { \let\tabstyle=\relax
1145% \begin{demo}{Alternating row colours}
1146%\begin{tabular}
1147% {|?>{\ifmod
1148% {\value{tabrow}}{2}{1}
1149% {\rowcolour{white}}
1150% {\rowcolour[gray]{0.9}}}
1151% c|c|}
1152% \hlx{h+v}
1153% One & Two \\ \hlx{vh+v}
1154% Three & Four \\ \hlx{vh+v}
1155% Five & Six \\ \hlx{vh+v}
1156% Seven & Eight \\ \hlx{vh+v}
1157% Nine & Ten \\ \hlx{vh+}
1158%\end{tabular}
1159% \end{demo}
1160% }
1161%
1162% There are new rule parameters for colours. You get a colourful rule if you
1163% say \syntax{"colour" "=" <colour>}. You can also say \syntax{"colourmodel"
1164% "=" <colour-model>} to choose unnamed colours.
1165%
1166% When I've thought of what other things need doing, I'll do some of them.
1167% The kit I've provided \emph{can} do most interesting things, but it might
1168% require a certain level of hacking. Ask me if you want something and it's
1169% not obvious how to do it.
1170%
1171% \subsubsection{Dirty tricks}
1172%
1173% The colour support interacts with |\vgap| very badly. The preamble rune
1174% |?>{\cellcolour{...}}| works well if you want to colour a column, and
1175% |\rowcolour| works either in the preamble or as
1176% |\hlx{?{\rowcolour{...}}}|. But what if you want to just colour one table
1177% cell? You can, as suggested above, just say |\cellcolour{...}| in the
1178% table text, but that leaves really nasty-looking gaps above and below if
1179% there are adjacent |\vgap| rows.
1180%
1181% This is what |\hlx{?{...}}| was invented for. Here's a demo.
1182%
1183% \begin{demo}[w]{Colouring just one cell}
1184%\let\hack=\relax
1185%\begin{tabular}[C]{|c|?>{\hack}c|} \hlx{hv}
1186%Uncoloured & cells here \\ \hlx{vhv}
1187%And some & more \\
1188% \hlx{vh?{\gdef\hack{\cellcolour{red}}}v}
1189%Yet more & This one's red! \\
1190% \hlx{vh?{\global\let\hack=\relax}v}
1191%And more & uncoloured cells \\ \hlx{vh}
1192%\end{tabular}
1193% \end{demo}
1194%
1195%
11222df6 1196% \subsection{The \package{mathenv} package alignment environments}
86f6a31e 1197%
11222df6 1198% The \package{mathenv} package provides several environments for aligning
86f6a31e 1199% equations in various ways. They're mainly provided as a demonstration of
1200% the table handling macros in \package{mdwtab}, so don't expect great
1201% things. If you want truly beautiful mathematics, use
1202% \package{amsmath}.\footnote{^^A
1203% Particularly since nice commands like \cmd\over\ are being reactivated
1204% in a later release of \package{amsmath}.}
1205% However, the various environments do nest in an approximately useful way.
1206% I also think that the \env{matrix} and \env{script} environments provided
1207% here give better results than their \package{amsmath} equivalents, and
1208% they are certainly more versatile.
1209%
1210% \subsubsection{The new \env{eqnarray} environment}
1211%
1212% \DescribeEnv{eqnarray}
1213% \DescribeEnv{eqnarray*}
1214% As an example of the new column defining features, and because the original
1215% isn't terribly good, I've included a rewritten version of the
1216% \env{eqnarray} environment. The new implementation closes the gap between
1217% \env{eqnarray} and \AmSTeX\ alignment features. It's in a separate,
1218% package called \package{mathenv}, to avoid wasting your memory.
1219%
1220% \begin{grammar}
1221%
1222% <eqnarray-env> ::= \[[
1223% <begin-eqnarray> \< <row> \\ "\\\\" \> <end-eqnarray>
1224% \]]
1225%
1226% <begin-eqnarray> ::= \[[
1227% "\\begin" \( "{eqnarray}" \\ "{eqnarray*}" \)
1228% \[ "[" \< <eqa-column> \> "]" \]
1229% \]]
1230%
eafdddad 1231% <eqa-column> ::= \[[
86f6a31e 1232% \[ "q" \\ ":" \]
1233% \[ \< ">" "{" <pre-text> "}" \> \]
1234% \begin{stack}
1235% \[ "T" \] \( "r" \\ "c" \\ "l" \) \\
1236% "L" \\
1237% "x"
1238% \end{stack}
1239% \[ \< "<" "{" <post-text> "}" \> \]
1240% \]]
1241%
1242% <end-eqnarray> ::= \[[
1243% "\\end" \begin{stack} "{eqnarray}" \\ "{eqnarray*}" \end{stack}
1244% \]]
1245%
1246% \end{grammar}
1247%
1248% Descriptions of the various column types are given in
1249% table~\ref{tbl:eqnarray}.
1250%
1251% \begin{table}
eafdddad
MW
1252% \begin{tabular}[C]{| >{\synshorts} c | m{3in} |} \hlx{hv[1]}
1253%
1254% \multicolumn{2}{|c|}{\bf Column types} \\ \hlx{v[1]hv}
1255% \bf Name & \bf Meaning \\ \hlx{vhv.}
1256% "l" & Left aligned piece of equation. \\ \hlx{.}
1257% "c" & Centred piece of equation. \\ \hlx{.}
1258% "x" & Centred or flush-left whole equation
1259% (depending on \textsf{fleqn} option). \\ \hlx{.}
1260% "r" & Right aligned piece of equation. \\ \hlx{vhv.}
1261% "L" & Left aligned piece of equation whose
1262% width is considered to be 2\,em. \\ \hlx{vhv.}
86f6a31e 1263% "Tl", "Tc" and "Tr" & Left, centre and right aligned
eafdddad
MW
1264% text. \\ \hlx{vhhv[1]}
1265%
1266% \multicolumn{2}{|c|}{\bf Other modifier characters} \\ \hlx{v[1]hv}
1267% \bf Name & \bf Meaning \\ \hlx{vhv.}
1268% ":" & Leaves a big gap between equations.
1269% By default, the `chunks' separated by
1270% \lit{:}s are equally spaced on the
1271% line. \\ \hlx{.}
1272% "q" & Inserts 1\,em of space \\ \hlx{vhv.}
1273% ">{"<text>"}" & Inserts \<text> just before the
1274% actual column entry. \\ \hlx{.}
1275% "<{"<text>"}" & Inserts \<text> just after the
1276% actual column entry. \\ \hlx{vhv.}
86f6a31e 1277% "*{"<count>"}{"<chars>"}" & Inserts \<count>
eafdddad
MW
1278% copies of the \<chars> into the
1279% preamble. \\ \hlx{vh}
86f6a31e 1280% \end{tabular}
1281%
1282% \caption{\package{eqnarray} column types and modifiers}
1283% \label{tbl:eqnarray}
1284% \end{table}
1285%
1286% The default preamble, if you don't supply one of your own, is \lit{rcl}.
1287% Most of the time, \lit{rl} is sufficient, although compatibility is more
1288% important to me.
1289%
1290% By default, there is no space between columns, which makes formul\ae\ in an
1291% \env{eqnarray} environment look just like formul\ae\ typeset on their own,
1292% except that things get aligned in columns. This is where the default
1293% \env{eqnarray} falls down: it leaves |\arraycolsep| space between each
1294% column making the thing look horrible.
1295%
1296% An example would be good here, I think. This one's from exercise 22.9 of
1297% the \textit{\TeX book}.
1298%
1299% \begin{demo}[w]{Simultaneous equations}
1300%\begin{eqnarray}[*3{rc}rl]
1301% 10w & + & 3x & + & 3y & + & 18z & = 1 \\
1302% 6w & - & 17x & & & - & 5z & = 2
1303%\end{eqnarray}
1304% \end{demo}
1305%
1306% Choosing a more up-to-date example, here's some examples from the
1307% \textit{\LaTeX\ Companion}.
1308%
1309% \begin{demo}[w]{Lots of equations}
1310%\begin{eqnarray}[rl:rl:lq]
eafdddad 1311% V_i &= v_i - q_i v_j, & X_i &= x_i - q_i x_j, &
86f6a31e 1312% U_i = u_i, \qquad \mbox{for $i \ne j$} \\
1313% V_j &= v_j, & X_j &= x_j &
1314% U_j u_j + \sum_{i \ne j} q_i u_i. \label{eq:A}
1315%\end{eqnarray}
1316% \end{demo}
1317%
1318% \begin{figure}
1319% \begin{demo}[w]{Plain text column and \cs{tabpause}}
1320%\begin{eqnarray}[rlqqTl]
1321% x &= y & by (\ref{eq:A}) \\
1322% x' &= y' & by definition \\
1323%\tabpause{and}
1324% x + x' &= y + y' & by Axiom~1
1325%\end{eqnarray}
1326% \end{demo}
1327% \end{figure}
1328%
1329% The new features also mean that you don't need to mess about with
1330% |\lefteqn| any more. This is handled by the \lit{L} column type:
1331%
1332% \begin{demo}{Splitting example}
1333%\begin{eqnarray*}[Ll]
1334% w+x+y+z = \\
7abe48b3 1335% & a+b+c+d+e+{} \\
86f6a31e 1336% & f+g+h+i+j
1337%\end{eqnarray*}
1338% \end{demo}
1339%
1340% Finally, just to prove that the spacing's right at last, here's another one
1341% from the \textit{Companion}.
1342%
1343% \begin{demo}{Spacing demonstration}
1344%\begin{equation}
1345% x^2 + y^2 = z^2
1346%\end{equation}
1347%\begin{eqnarray}[rl]
1348% x^2 + y^2 &= z^2 \\
1349% y^2 &< z^2
1350%\end{eqnarray}
1351% \end{demo}
1352%
1353% Well, that was easy enough. Now on to numbering. As you've noticed, the
1354% equations above are numbered. You can use the \env{eqnarray$*$}
1355% environment to turn off the numbering in the whole environment, or say
1356% |\nonumber| on a line to suppress numbering of that one in particular.
1357%
1358% \DescribeMacro{\eqnumber}
1359% More excitingly, you can say |\eqnumber| to enable numbering for a
1360% particular equation, or \syntax{"\\eqnumber["<text>"]"} to choose what to
1361% show instead of the line number. This works for both starred and unstarred
1362% versions of the environment. Now |\nonumber| becomes merely a synonym for
1363% `|\eqnumber[]|'.
1364%
1365% A note for cheats: you can use the sparkly new \env{eqnarray} for simple
1366% equations by specifying \lit{x} as the column description. Who needs
1367% \AmSTeX?\ |;-)|
1368%
1369% \DescribeEnv{eqlines}
1370% \DescribeEnv{eqlines*}
1371% In fact, there's a separate environment \env{eqlines}, which is equivalent
1372% to \env{eqnarray} with a single \lit{x} column; the result is that you can
1373% insert a collection of displayed equations separated by |\\| commands. If
1374% you don't like numbering, use \env{eqlines$*$} instead.
1375%
1376% \subsubsection{The \env{eqnalign} environment}
1377%
1378% \DescribeEnv{eqnalign}
1379% There's a new environment, \env{eqnalign}, which does almost the same
1380% thing as \env{eqnarray} but not quite. It doesn't do equation numbers,
1381% and it wraps its contents up in a box. The result of this is that:
1382%
1383% \begin{itemize}
1384%
1385% \item You can use \env{eqnalign} for just a part of a formula.
1386% The \env{eqnarray} environment must take up the whole display.
1387%
1388% \item You can use \env{eqnalign} within \env{eqnarray} for extra fine
1389% alignment of subsidiary bits.
1390%
1391% \item You can break off from doing an \env{eqnarray} using the |\tabpause|
1392% command. You can't use |\tabpause| inside
1393% \env{eqnalign}.\footnote{^^A
1394% Well, technically speaking there's nothing to stop you. However,
1395% the results won't be pretty.}
1396%
1397% \end{itemize}
1398%
1399% The \env{eqnalign} environment works like this:
1400%
1401% \begin{grammar}
1402%
1403% <eqnalign-env> ::= \[[
1404% <begin-eqnalign> <contents> <end-eqnalign>
1405% \]]
1406%
1407% <begin-eqnalign> ::= \[[
1408% "\\begin" "{eqnalign}"
1409% \[ "[" \< <eqa-column> \> "]" \]
1410% \[ "[" \( "t" \\ "c" \\ "b" \) "]" \]
1411% \]]
1412%
1413% <end-eqnalign> ::= \[[ "\\end" "{eqnalign}" \]]
1414%
1415% \end{grammar}
1416%
1417% As the syntax suggests, the preamble for the \env{eqnalign} environment
1418% works exactly the same way as for \env{eqnarray}. Example time: another
1419% one from the \textit{\TeX book}.
1420%
1421% \begin{figure}
1422% \begin{demo}[w]{Example of \env{eqnalign}}
1423%\[
1424% \left\{ \begin{eqnalign}[rl]
1425% \alpha &= f(z) \\ \beta &= f(z^2) \\
1426% \gamma &= f(z^3)
1427% \end{eqnalign} \right\}
1428% \qquad
1429% \left\{ \begin{eqnalign}[rl]
1430% x &= \alpha^2 - \beta \\ y &= 2\gamma
1431% \end{eqnalign} \right\}.
1432%\]
1433% \end{demo}
1434% \end{figure}
1435%
1436% \DescribeMacro{\multicolumn}
1437% The |\multicolumn| command works correctly in both the \env{eqnarray} and
1438% \env{eqnalign} environments, although you should bear in mind that you
1439% should give \env{eqnarray} column types, not \env{array} ones.
1440%
1441% \subsubsection{A note on spacing in alignment environments}
1442%
1443% Most of the time, equations in \env{eqnarray} and \env{eqnalign}
1444% environments will be beautiful. However, there are some things you should
1445% bear in mind when you produce beautiful equations.
1446%
1447% The main problem with spacing is making sure that binary relations and
1448% binary operators have the correct amount of space on each side of them.
1449% The alignment environments insert `hidden' objects at the ends of table
1450% cells to assist with the spacing: \lit{l} column types have a hidden object
1451% on the left, \lit{r} types have a hidden object on the right, and \lit{c}
1452% types have a hidden object on \emph{both} ends. These hidden objects add
1453% the correct space when there's a binary operator or relation next to them.
1454% If some other sort of object is lurking there, no space is added. So far,
1455% so good.
1456%
1457% The only problem comes when you have something like this:
1458%
1459% \begin{demo}{How not to do an \env{eqnarray}}
1460%\begin{eqnarray*}[rcl]
1461% x + y & = & 12 \\
1462% 2x - 5y & = & -6
1463%\end{eqnarray*}
1464% \end{demo}
1465%
1466% The `$-$' sign in the second equation has been treated as a binary operator
1467% when really it should be a unary prefix operator, but \TeX\ isn't clever
1468% enough to know the difference. (Can you see the difference in the spacing
1469% between $-6$~and~${}-6$?) There are two possible solutions to the
1470% problem. You could wrap the `|-6|' up in a group (`|{-6}|'), or just the
1471% $-$ sign (`|{-}6|'). A better plan, though, is to get rid of the middle
1472% column altogether:
1473%
1474% \begin{demo}{How to do an \env{eqnarray}}
1475%\begin{eqnarray*}[rl]
1476% x + y & = 12 \\
1477% 2x - 5y & = -6
1478%\end{eqnarray*}
1479% \end{demo}
1480%
1481% Since the things in the middle column were the same width, it's not
1482% actually doing any good. Also, now that \TeX\ can see that the thing on
1483% the left of the `$-$' sign is a relation (the `$=$' sign), it will space
1484% the formula correctly.
1485%
1486% In this case, it might be even better to add some extra columns, and line
1487% up the $x$ and $y$ terms in the left hand side:
1488%
1489% \begin{demo}{Extra beautiful \env{eqnarray}}
1490%\begin{eqnarray*}[rrl]
1491% x + & y & = 12 \\
1492% 2x - & 5y & = -6
1493%\end{eqnarray*}
1494% \end{demo}
1495%
1496% ^^A Some hacking now to display box sizes.
1497%
1498% {
1499% \catcode`p=12 \catcode`t=12
1500% \gdef\magni#1pt{#1}
1501% }
1502%
1503% \newcommand{\widthof}[1]{^^A
1504% \settowidth{\dimen0 }{#1}^^A
1505% \expandafter\magni\the\dimen0\,pt^^A
1506% }
1507%
1508% ^^A The text below makes an assumption which looks correct to me (I asked
1509% ^^A TeX, and it agreed with me), although in case anything changes, I want
1510% ^^A to be informed.
1511%
1512% \sbox0{$+$} \sbox2{$-$} \ifdim\wd0=\wd2\else%
1513% \errmessage{Assertion failed: `+' and `-' are different widths!}
1514% \fi
1515%
1516% There's no need to put the `$+$' and `$-$' operators in their own column
1517% here, because they're both \widthof{$+$} wide, even though they don't
1518% look it.
1519%
1520% \subsubsection{Configuring the alignment environments}
1521%
1522% There are a collection of parameters you can use to make the equation
1523% alignment environments (\env{eqnarray} and \env{eqnalign}) look the way
1524% you like them. These are all shown in table~\ref{tbl:eqnparms}.
1525%
1526% \begin{table}
eafdddad
MW
1527% \begin{tabular}[C]{| l | p{3in} |} \hlx{hv}
1528% \bf Parameter & \bf Use \\ \hlx{vhv}
1529% |\eqaopenskip| & Length put on the left of an
1530% \env{eqnarray} environment. By
1531% default, this is |\@centering| (to
1532% centre the alignment) or |\mathindent|
1533% (to left align) depending on whether
1534% you're using the \textsf{fleqn}
1535% document class option. \\
1536% |\eqacloseskip| & Length put on the right of an
1537% \env{eqnarray} environment. By
1538% default, this is |\@centering|, to
1539% align the environment correctly. \\ \hlx{vhv}
1540% |\eqacolskip| & Space added by the \lit{:} column
1541% modifier. This should be a rubber
1542% length, although it only stretches in
1543% \env{eqnarray}, not in \env{eqnalign}.
1544% The default value is 1\smallf1/2\,em
1545% with 1000\,pt of stretch. \\
1546% |\eqainskip| & Space added at each side of a normal
1547% column. By default this is 0\,pt. \\ \hlx{vhv}
1548% |\eqastyle| & The maths style used in the alignment.
1549% By default, this is |\textstyle|,
1550% and you probably won't want to change
1551% it. \\ \hlx{vh}
86f6a31e 1552% \end{tabular}
1553%
1554% \caption{Parameters for the \env{eqnarray} and \env{eqnalign} environments}
1555% \label{tbl:eqnparms}
1556% \end{table}
1557%
1558%
1559% \subsection{Other multiline equations}
1560%
1561% Sometimes there's no sensible alignment point for splitting equations. The
1562% normal thing to do under these circumstances is to put the first line way
1563% over to the left of the page, and the last line over to the right. (If
1564% there are more lines, I imagine we put them in the middle.)
1565%
1566% \DescribeEnv{spliteqn}
1567% \DescribeEnv{spliteqn*}
1568% The \env{spliteqn} environment allows you to do such splitting of
1569% equations. Rather than tediously describe it, I'll just give an example,
1570% because it's really easy. The $*$-version works the same, except it
1571% doesn't put an equation number in.
1572%
1573% \begin{figure}
1574% \begin{demo}[w]{A split equation}
1575%\begin{spliteqn}
1576% \sum_{1\le j\le n}
1577% \frac {1} { (x_j - x_1) \ldots (x_j - x_{j-1})
1578% (x - x_j) (x_j - x_{j+1}) \ldots (x_j - x_n) }
1579% \\
1580% = \frac {1} { (x - x_1) \ldots (x - x_n) }.
1581%\end{spliteqn}
1582% \end{demo}
1583% \end{figure}
1584%
1585% \DescribeEnv{subsplit}
1586% If you have a very badly behaved equation, you might want to split a part
1587% of it (say, a bit of a fraction), particularly if you're doing things in
1588% narrow columns.
1589%
1590% \begin{figure}
1591% \begin{demo}[w]{A \env{subsplit} environment}
1592%\begin{equation}
1593% \frac{
1594% \begin{subsplit}
1595% q^{\frac{1}{2} n(n+1)}(ea; q^2)_\infty (eq/a; q^2)_\infty \\
1596% (caq/e; q^2)_\infty (cq^2/ae; q^2)_\infty
1597% \end{subsplit}
1598% }{
1599% (e; q)_\infty (cq/e; q)_\infty
1600% }
1601%\end{equation}
1602% \end{demo}
1603% \end{figure}
1604%
1605% \subsection{Matrices}
1606%
1607% Also included in the \package{mathenv} package is a collection of things
1608% for typesetting matrices. The standard \env{array} doesn't (in my opinion)
1609% provide the right sort of spacing for matrices. \PlainTeX\ provides some
1610% quite nice matrix handling macros, but they don't work in the appropriate
1611% \LaTeX\ way.
1612%
1613% \textbf{Warning:} These definitions will make old versions of
1614% \package{plain.sty} unhappy; newer versions correctly restore the
1615% Plain~\TeX\ macros |\matrix| and |\pmatrix|.
1616%
1617% \DescribeEnv{matrix}
1618% The simple way to do matrices is with the \env{matrix} environment.
1619%
1620% \begin{grammar}
1621%
eafdddad 1622% <matrix-env> ::= \[[ <begin-matrix> <contents> <end-matrix> \]]
86f6a31e 1623%
1624% <begin-matrix> ::= \[[ "\\begin{matrix}" \[ "[" <matrix-cols> "]" \] \]]
1625%
eafdddad 1626% <matrix-cols> ::= \[[
86f6a31e 1627% \< \[ "[" \] \[ "T" \] \( "l" \\ "c" \\ "r" \) \>
1628% \]]
1629%
eafdddad 1630% <end-matrix> ::= \[[ "\\end{stack}" \]]
86f6a31e 1631%
1632% \end{grammar}
1633%
1634% The \lit{l}, \lit{c} and \lit{r} columns are fairly obvious -- they align
1635% their contents in the appropriate way. The \lit{[} character is more
1636% complicated. It means `repeat the remaining column types forever', so a
1637% preamble of \lit{cc[lr} means `two centred columns, then alternating left-
1638% and right-aligned columns for as often as needed'. The default preamble,
1639% if you don't specify one, is \lit{[c} -- `any number of centred columns'.
1640%
1641% \DescribeMacro{\multicolumn}
1642% The |\multicolumn| command works correctly in matrices, although you should
1643% bear in mind that you should give \env{matrix} column types, not
1644% \env{array} ones.
1645%
1646% \DescribeEnv{pmatrix}
1647% The standard \env{matrix} environment doesn't put any delimiters around the
1648% matrix. You can use the standard |\left| and |\right| commands, although
1649% this is a bit nasty. The \env{pmatrix} environment will put parentheses
1650% around the matrix it creates; it's otherwise exactly the same as
1651% \env{matrix}.
1652%
1653% \DescribeEnv{dmatrix}
1654% A \env{dmatrix} environment is also provided. It takes two extra
1655% arguments: the left and right delimiter characters (without |\left| or
1656% |\right|).
1657%
1658% \begin{figure}
1659% \begin{demo}[w]{Various \env{matrix} environments}
1660%\[ \begin{matrix} 1 & 0 \\ 0 & -1 \end{matrix} \quad
1661% \begin{pmatrix}
1662% \cos\theta & \sin\theta \\
1663% -\sin\theta & \cos\theta
1664% \end{pmatrix} \quad
1665% \begin{dmatrix}[] 0 & -i \\ i & 0 \end{dmatrix}
1666%\]
1667% \end{demo}
1668% \end{figure}
1669%
1670% \DescribeEnv{smatrix}
1671% Normal matrices always come out the same size; they don't change size
1672% according to the surrounding context (unfortunately). However, it can be
1673% occasionally useful to put matrices in running text, so you can talk about
1674% $A$ being $\bigl( \begin{smatrix} a & b \\ b & c \end{smatrix} \bigr)$
1675% being its own transpose (i.e., $A = A^T$). This is accomplished using the
1676% \env{smatrix} (the `s' stands for `small' -- I thought that `smallmatrix'
1677% was too big to type inline). As well as inline text, the \env{smatrix}
1678% can be useful in displays, if the matrix is deep in a subformula. I can't
1679% think of any examples offhand, though.
1680%
1681% \DescribeEnv{spmatrix}
1682% \DescribeEnv{sdmatrix}
1683% The \env{smatrix} environment doesn't supply any delimiters, like
1684% \env{matrix}. There are \env{spmatrix} and \env{sdmatrix} environments
1685% which do, though. Note that delimiters have a tendency to get too big and
1686% mess up the line spacing -- I had to use explicitly |\big| delimiters
1687% in the above example.
1688%
1689% \DescribeEnv{pmatrix*}
1690% \DescribeEnv{spmatrix*}
1691% \DescribeEnv{sdmatrix*}
1692% All the small matrix environments have starred versions, which are more
1693% suitable for use in displays, since they have more space between the rows.
1694% They're intended for typesetting really big matrices in displays.
1695%
1696% \DescribeMacro{\ddots}
1697% \DescribeMacro{\vdots}
1698% The standard |\vdots| and |\ddots| commands don't produce anything at all
1699% nice in small matrices, so this package redefines them so that they scale
1700% properly to smaller sizes.
1701%
1702% \DescribeEnv{genmatrix}
1703% Actually, all these environments are special cases of one: \env{genmatrix}.
1704% This takes oodles of arguments:
1705% \begin{quote} \synshorts
1706% "\\begin{genmatrix}{"<matrix-style>"}{"<outer-style>"}" \\
1707% \null \qquad "{"<spacing>"}{"<left-delim>"}{"<right-delim>"}" \\
1708% \null \quad\vdots \\
1709% "\\end{genmatrix}"
1710% \end{quote}
1711% The two `style' arguments should be things like |\textstyle| or
1712% |\scriptstyle|; the first, \<matrix-style>, is the style to use for the
1713% matrix elements, and the second, \<outer-style>, is the style to assume
1714% for the surrounding text (this affects the spacing within the matrix; it
1715% should usually be the same as \<matrix-style>). The \<spacing> is inserted
1716% between the matrix and the delimiters, on each side of the matrix. It's
1717% usually `|\,|' in full-size matrices, and blank for small ones. The
1718% delimiters are inserted around the matrices, and sized appropriately.
1719%
1720% \DescribeEnv{newmatrix}
1721% You can create your own matrix environments if you like, using the
1722% |\newmatrix| command. It takes two arguments, although they're a bit
1723% odd. The first is the name of the environment, and the second contains
1724% the arguments to pass to \env{genmatrix}. For example, the \env{pmatrix}
1725% environment was defined by saying
1726%
1727% \begin{listing}
1728%\newmatrix{pmatrix}{{\textstyle}{\textstyle}{\,}{(}{)}}
1729% \end{listing}
1730%
1731% If you don't pass all three arguments, then you end up requiring the
1732% user to specify the remaining ones. This is how \env{dmatrix} works.
1733%
1734% \DescribeEnv{script}
1735% Finally, although it's not really a matrix, stacked super- and subscripts
1736% follow much the same sorts of spacing rules. The \env{script} environment
1737% allows you to do this sort of thing very easily. It essentially provides
1738% a `matrix' with the right sort of spacing. The default preamble string is
1739% \lit{c}, giving you centred scripts, although you can say
1740% |\begin{script}[l]| for left-aligned scripts, which is better if the
1741% script is being placed to the right of its operator. If you're really
1742% odd, you can have more than one column.
1743%
1744% \begin{demo}{Example of \env{script}}
1745%\[ \mathop{{\sum}'}_{x \in A}
1746% f(x)
1747% \stackrel{\mathrm{def}}{=}
1748% \sum_{\begin{script}
1749% x \in A \\ x \ne 0
1750% \end{script}} f(x)
1751%\]
1752% \end{demo}
1753%
1754%
1755% \subsection{Other \package{mathenv} environments}
1756%
1757% The \package{mathenv} package contains some other environments which may
1758% be useful, based on the enhanced \env{tabular} and \env{array}
1759% environments.
1760%
1761% \DescribeEnv{cases}
1762% The \env{cases} environment lets you say things like the following:
1763%
1764% \begin{demo}[w]{Example of \env{cases}}
1765%\[ P_{r-j} = \begin{cases}
1766% 0 & if $r-j$ is odd \\
1767% r!\,(-1)^{(r-j)/2} & if $r-j$ is even
1768% \end{cases}
1769%\]
1770% \end{demo}
1771%
1772% The spacing required for this is a bit messy, so providing an environment
1773% for it is quite handy.
1774%
1775% \DescribeEnv{smcases}
1776% The \env{smcases} environment works the same way as \env{cases}, but with
1777% scriptsize lettering.
1778%
1779% \implementation
1780%
1781%
1782%^^A-------------------------------------------------------------------------
1783% \section{Implementation of table handling}
1784%
1785%
1786% Here we go. It starts horrid and gets worse. However, it does stay nicer
1787% than the original, IMHO.
1788%
1789% \begin{macrocode}
1790%<*mdwtab>
1791% \end{macrocode}
1792%
1793%
1794% \subsection{Registers, switches and things}
1795%
1796% We need lots of these. It's great fun.
1797%
1798% The two count registers are simple enough:
1799%
1800% \begin{description}
1801% \item [\cs{tab@state}] contains the current parser state. Since we
1802% probably won't be parsing preambles recursively, this is a global
1803% variable.
1804% \item [\cs{tab@columns}] contains the number of the current column.
1805% \item [\cs{tab@hlstate}] contains the state required for hline management.
7abe48b3 1806% \item [\textsf{tabrow}] contains the row number in the table. It's a
1807% proper \LaTeX\ counter.
86f6a31e 1808% \end{description}
1809%
1810% \begin{macrocode}
1811\newcount\tab@state
1812\newcount\tab@columns
7abe48b3 1813\newcounter{tabrow}
86f6a31e 1814% \end{macrocode}
1815%
1816% We need \emph{lots} of token registers. Fortunately, most of them are only
1817% used during parsing. We'll use \PlainTeX's scratch tokens for this. Note
1818% that |\toks\tw@| isn't used here. It, and |\toks@|, are free for use by
1819% column commands.
1820%
1821% \begin{macrocode}
1822\newtoks\tab@preamble
1823\newtoks\tab@shortline
1824\toksdef\tab@pretext 4
1825\toksdef\tab@posttext 6
1826\toksdef\tab@userpretext 8
1827% \end{macrocode}
1828%
1829% The dimens are fairly straightforward. The inclusion of |\col@sep| is a
1830% sacrifice to compatibility -- judicious use of |\let| in \package{array}
1831% would have saved a register.
1832%
1833% \begin{macrocode}
1834\newdimen\extrarowheight
1835\newdimen\tabextrasep
1836\newdimen\arrayextrasep
1837\newdimen\smarraycolsep
1838\newdimen\smarrayextrasep
1839\newdimen\tab@width
1840\newdimen\col@sep
1841\newdimen\tab@endheight
1842\newdimen\arraythickrulewidth
1843\newdimen\tab@rulewidth
1844% \end{macrocode}
1845%
1846% Some skip registers too. Phew.
1847%
1848% \begin{macrocode}
1849\newskip\tab@leftskip
1850\newskip\tab@rightskip
1851% \end{macrocode}
1852%
1853% And some switches. The first three are for the parser.
1854%
1855% \begin{macrocode}
1856\newif\iftab@firstcol
1857\newif\iftab@initrule
1858\newif\iftab@rule
1859\newif\iftab@vgap
7abe48b3 1860\newif\iftab@colour
86f6a31e 1861% \end{macrocode}
1862%
1863% Now assign some default values to new dimen parameters. These definitions
1864% are essentially the equivalent of an |\openup 1\jot| in \env{array}, but
1865% not in \env{tabular}. This looks nice, I think.
1866%
1867% \begin{macrocode}
1868\tabextrasep\z@
1869\arrayextrasep\jot
1870\smarraycolsep\thr@@\p@
1871\smarrayextrasep\z@
1872\arraythickrulewidth\p@
1873% \end{macrocode}
1874%
1875% Set some things up for alien table environments.
1876%
1877% \begin{macrocode}
1878\let\tab@extrasep\tabextrasep
1879\let\tab@penalty\relax
1880% \end{macrocode}
1881%
1882%
7abe48b3 1883% \subsection{Options processing}
1884%
1885% Notice options, load package.
1886%
1887% \begin{macrocode}
1888\DeclareOption{colour}{\tab@colourtrue}
1889\DeclareOption{color}{\tab@colourtrue}
1890\ProcessOptions
1891\RequirePackage{mdwkey}
1892% \end{macrocode}
1893%
86f6a31e 1894% \subsection{Some little details}
1895%
1896% \begin{macro}{\@maybe@unskip}
1897%
1898% This macro solves a little problem. In an alignment (and in other places)
1899% it's desirable to suppress trailing space. The usual method, to say
1900% |\unskip|, is a little hamfisted, because it removes perfectly reasonable
1901% aligning spaces like |\hfil|s. While as a package writer I can deal with
1902% this sort of thing by saying |\kern\z@| in appropriate places, it can
1903% annoy users who are trying to use |\hfill| to override alignment in funny
1904% places.
1905%
1906% My current solution seems to be acceptable. I'll remove the natural width
1907% of the last glue item, so that it can still stretch and shrink if
1908% necessary. The implementation makes use of the fact that multiplying
1909% a \<skip> by a \<number> kills off the stretch. (Bug fix: don't do this
1910% when we're in vertical mode.)
1911%
1912% \begin{macrocode}
1913\def\@maybe@unskip{\ifhmode\hskip\m@ne\lastskip\relax\fi}
1914% \end{macrocode}
1915%
1916% \end{macro}
1917%
1918% \begin{macro}{\q@delim}
1919%
1920% Finally, for the sake of niceness, here's a delimiter token I can use
1921% for various things. It's a `quark', for what it's worth (i.e., it expands
1922% to itself) although I'm not really sure why this is a good thing. As far
1923% as I'm concerned, it's important that it has a unique meaning (i.e., that
1924% it won't be |\ifx|-equal to other things, or something undefined) and that
1925% it won't be used where I don't expect it to be used. \TeX\ will loop
1926% horridly if it tries to expand this, so I don't think that quarks are
1927% wonderfully clever thing to use. (Maybe it should really expand to
1928% something like `\syntax{<quark>"."}', which will rapidly fill \TeX's memory
1929% if it gets accidentally expanded. Still, I'll leave it as it is until
1930% such time as I understand the idea more.)
1931%
1932% \begin{macrocode}
1933\def\q@delim{\q@delim}
1934% \end{macrocode}
1935%
1936% \end{macro}
1937%
1938%
1939% \subsection{Parser states}
e8e9e5d8 1940%
86f6a31e 1941% Now we start on the parser. It's really simple, deep down. We progress
1942% from state to state, extracting tokens from the preamble and building
1943% command names from them. Each command calls one of the element-building
1944% routines, which works out which state it should be in. We go through each
1945% of the states in between (see later) doing default things for the ones we
1946% missed out.
e8e9e5d8 1947%
86f6a31e 1948% Anyway, here's some symbolic names for the states. It makes my life
1949% easier.
1950%
1951% \begin{macrocode}
1952\chardef\tab@startstate 0
1953\chardef\tab@loopstate 1
1954\chardef\tab@rulestate 1
1955\chardef\tab@prespcstate 2
1956\chardef\tab@prestate 3
1957\chardef\tab@colstate 4
1958\chardef\tab@poststate 5
1959\chardef\tab@postspcstate 6
1960\chardef\tab@limitstate 7
1961% \end{macrocode}
1962%
1963%
1964% \subsection{Adding things to token lists}
1965%
1966% Define some macros for adding stuff to the beginning and end of token
1967% lists. This is really easy, actually. Here we go.
1968%
1969% \begin{macrocode}
1970\def\tab@append#1#2{#1\expandafter{\the#1#2}}
1971\def\tab@prepend#1#2{%
1972 \toks@{#2}#1\expandafter{\the\expandafter\toks@\the#1}%
1973}
1974% \end{macrocode}%
1975%
1976%
1977% \subsection{Committing a column to the preamble}
1978%
1979% Each time we pass the `rule' state, we `commit' the tokens we've gathered
1980% so far to the main preamble token list. This is how we do it. Note the
1981% icky use of |\expandafter|.
1982%
1983% \begin{macrocode}
1984\def\tab@commit{%
1985% \end{macrocode}
1986%
1987% If this isn't the first column, then we need to put in a column separator.
1988%
1989% \begin{macrocode}
1990 \iftab@firstcol\else%
1991 \expandafter\tab@append\expandafter\tab@preamble%
1992 \expandafter{\tab@tabtext}%
1993 \fi%
1994% \end{macrocode}
1995%
7abe48b3 1996% Now dump in the |\tab@lefttext| material.
1997%
1998% \begin{macrocode}
1999 \expandafter\tab@append\expandafter\tab@preamble%
2000 \expandafter{\tab@lefttext}%
2001% \end{macrocode}
2002%
86f6a31e 2003% Now we spill the token registers into the main list in a funny order (which
2004% is why we're doing it in this strange way in the first place.
2005%
2006% \begin{macrocode}
2007 \toks@\expandafter{\tab@midtext}%
2008 \tab@preamble\expandafter{%
2009 \the\expandafter\tab@preamble%
2010 \the\expandafter\tab@pretext%
2011 \the\expandafter\tab@userpretext%
2012 \the\expandafter\toks@%
7abe48b3 2013 \the\expandafter\tab@posttext%
2014 \tab@righttext%
86f6a31e 2015 }%
2016% \end{macrocode}
2017%
2018% Now reset token lists and things for the next go round.
2019%
2020% \begin{macrocode}
2021 \tab@firstcolfalse%
2022 \tab@pretext{}%
2023 \tab@userpretext{}%
2024 \tab@posttext{}%
2025}
2026% \end{macrocode}
2027%
2028%
2029% \subsection{Playing with parser states}
2030%
2031% \begin{macro}{\tab@setstate}
2032%
2033% This is how we set new states. The algorithm is fairly simple, really.
2034%
2035% ^^A Let's see how good my TeX really is... ;-)
2036% ^^A Actually, it doesn't seem to have worked out too badly. Maybe I should
2037% ^^A write a package to do this automatically. It's rather tricky, though.
2038%
2039% \def\qq{\mbox{\quad}}
2040% \sbox{0}{\itshape\textunderscore}\def\_{\usebox{0}}
2041%
2042% \begin{quote}
2043% {\bf while} $\it tab\_state \ne s$ {\bf do} \\
2044% \qq $\mathit{tab\_state = tab\_state}+1$; \\
2045% \qq {\bf if} $\it tab\_state = tab\_limitState$ {\bf then}
eafdddad 2046% $\it tab\_state=tab\_loopState$; \\
86f6a31e 2047% \qq {\bf if} $\it tab\_state = tab\_preSpcState$ {\bf then} \\
2048% \qq \qq {\bf if} $\it tab\_initRule$ {\bf then} \\
2049% \qq \qq \qq $\it tab\_initRule = {\bf false}$; \\
2050% \qq \qq {\bf else} \\
2051% \qq \qq \qq {\bf if} $\it tab\_inMultiCol$ {\bf then moan}; \\
2052% \qq \qq \qq $\it commit$; \\
2053% \qq \qq \qq $\it append(tab\_shortLine,\hbox{`|&\omit|')}$; \\
2054% \qq \qq {\bf end\,if}; \\
2055% \qq {\bf end\,if}; \\
2056% \qq {\bf if} $\it tab\_state \ne s$ {\bf then}
eafdddad 2057% $\it do\_default(tab\_state)$; \\
86f6a31e 2058% {\bf end\,while};
2059% \end{quote}
2060%
2061% First we decide if there's anything to do. If so, we call another macro to
2062% do it for us.
2063%
2064% \begin{macrocode}
2065\def\tab@setstate#1{%
2066 \ifnum#1=\tab@state\else%
2067 \def\@tempa{\tab@setstate@i{#1}}%
2068 \@tempa%
2069 \fi%
2070}
2071% \end{macrocode}
2072%
2073% This is where the fun is. First we bump the state by one, and loop back
2074% if we fall off the end.
2075%
2076% \begin{macrocode}
2077\def\tab@setstate@i#1{%
2078 \global\advance\tab@state\@ne%
2079 \ifnum\tab@state>\tab@limitstate%
2080 \global\tab@state\tab@loopstate%
2081 \fi%
2082% \end{macrocode}
2083%
2084% Now, if we've just passed the ruleoff state, we commit the current text
2085% \emph{unless} this was the strange initial rule at the very beginning. We
2086% provide a little hook here so that |\multicolumn| can moan if you try and
2087% give more than one column there. We also add another tab/omit pair to the
2088% list we use for |\vgap|.
2089%
2090% \begin{macrocode}
2091 \ifnum\tab@state=\tab@prespcstate%
2092 \iftab@initrule%
2093 \tab@initrulefalse%
2094 \else%
2095 \tab@looped%
2096 \tab@commit%
7abe48b3 2097 \expandafter\tab@append\expandafter\tab@shortline%
2098 \expandafter{\tab@rightruletext}%
86f6a31e 2099 \tab@append\tab@shortline{&\omit}%
7abe48b3 2100 \expandafter\tab@append\expandafter\tab@shortline%
2101 \expandafter{\tab@leftruletext}%
86f6a31e 2102 \fi%
2103 \fi%
2104% \end{macrocode}
2105%
2106% Now we decide whether to go round again. If not, we do the default thing
2107% for this state. This is mainly here so that we can put the |\tabcolsep| or
2108% whatever in if the user didn't give an \lit{@} expression.
2109%
2110% \begin{macrocode}
2111 \ifnum#1=\tab@state%
2112 \let\@tempa\relax%
2113 \else%
2114 \csname tab@default@\number\tab@state\endcsname%
2115 \fi%
2116 \@tempa%
2117}
2118% \end{macrocode}
2119%
2120% \end{macro}
2121%
2122% Now we set up the default actions for the various states.
2123%
2124% In state~2 (pre-space) we add in the default gap if either we didn't have
2125% an \lit{@} expression in the post-space state or there was an explicit
2126% intervening rule.
2127%
2128% \begin{macrocode}
2129\@namedef{tab@default@2}{%
2130 \iftab@rule%
2131 \tab@append\tab@pretext{\hskip\col@sep}%
2132 \fi%
2133}
2134% \end{macrocode}
2135%
2136% If the user omits the column type, we insert an `l'-type column and moan
2137% a lot.
2138%
2139% \begin{macrocode}
2140\@namedef{tab@default@4}{%
2141 \tab@err@misscol%
2142 \tab@append\tab@pretext{\tab@bgroup\relax}%
2143 \tab@append\tab@posttext{\relax\tab@egroup\hfil}%
2144 \tab@append\tab@shortline{\hfil}%
2145 \advance\tab@columns\@ne%
2146}
2147% \end{macrocode}
2148%
2149% Finally we deal with the post-space state. We set a marker so that we
2150% put in the default space in the pre-space state later too.
2151%
2152% \begin{macrocode}
2153\@namedef{tab@default@6}{%
2154 \tab@append\tab@posttext{\hskip\col@sep}%
2155 \tab@ruletrue%
2156}
2157% \end{macrocode}
2158%
2159%
2160% \subsection{Declaring token types}
2161%
2162% \begin{macro}{\tab@extracol}
2163%
2164% Before we start, we need to handle |\extracolsep|. This is a right pain,
2165% because the original version of \env{tabular} worked on total expansion,
2166% which is a Bad Thing. On the other hand, turning |\extracolsep| into a
2167% |\tabskip| is also a major pain.
2168%
2169% \begin{macrocode}
2170\def\tab@extracol#1#2{\tab@extracol@i#1#2\extracolsep{}\extracolsep\end}
2171\def\tab@extracol@i#1#2\extracolsep#3#4\extracolsep#5\end{%
2172 \ifx @#3@%
2173 \def\@tempa{#1{#2}}%
2174 \else%
2175 \def\@tempa{#1{#2\tabskip#3\relax#4}}%
2176 \fi%
2177 \@tempa%
2178}
2179% \end{macrocode}
2180%
2181% \end{macro}
2182%
2183% This is where we do the work for inserting preamble elements.
2184%
2185% \begin{macro}{\tabruletype}
2186%
2187% Inserting rules is interesting, because we have to decide where to put
2188% them. If this is the funny initial rule, it goes in the pre-text list,
2189% otherwise it goes in the post-text list. We work out what to do first
2190% thing:
2191%
2192% \begin{macrocode}
2193\def\tabruletype#1{\tab@extracol\tabruletype@i{#1}}%
2194\def\tabruletype@i#1{%
2195 \iftab@initrule%
2196 \let\tab@tok\tab@pretext%
2197 \else%
2198 \let\tab@tok\tab@posttext%
2199 \fi%
2200% \end{macrocode}
2201%
2202% Now if we're already in the rule state, we must have just done a rule.
2203% This means we must put in the |\doublerulesep| space, both here and in the
2204% shortline list. Otherwise we just stick the rule in.
2205%
2206% This is complicated, because |\vgap| needs to be able to remove some bits
2207% of rule. We pass each one to a macro |\tab@ckr|, together with the column
2208% number, which is carefully bumped at the right times, and this macro will
2209% vet the rules and output the appropriate ones. There's lots of extreme
2210% |\expandafter| nastiness as a result. Amazingly, this actually works.
2211%
2212% \begin{macrocode}
2213 \ifnum\tab@state=\tab@rulestate%
2214 \tab@append\tab@tok{\hskip\doublerulesep\begingroup#1\endgroup}%
2215 \expandafter\tab@append\expandafter\tab@shortline\expandafter{%
2216 \expandafter\hskip\expandafter\doublerulesep%
2217 \expandafter\tab@ckr\expandafter{\the\tab@columns}%
2218 {\begingroup#1\endgroup}%
2219 }%
2220 \else%
2221 \tab@setstate\tab@rulestate%
2222 \tab@append\tab@tok{\begingroup#1\endgroup}%
2223 \expandafter\tab@append\expandafter\tab@shortline\expandafter{%
2224 \expandafter\tab@ckr\expandafter{\the\tab@columns}%
2225 {\begingroup#1\endgroup}%
2226 }%
2227 \fi%
2228% \end{macrocode}
2229%
2230% Finally, we say there was a rule here, so that default space gets put in
2231% after this. Otherwise we lose lots of generality.
2232%
2233% \begin{macrocode}
2234 \tab@ruletrue%
2235}
2236% \end{macrocode}
2237%
2238% \end{macro}
2239%
2240% \begin{macro}{\tabspctype}
2241%
2242% We need to work out which space-state we should be in. Then we just put
2243% the text in. Easy, really.
2244%
2245% \begin{macrocode}
2246\def\tabspctype#1{\tab@extracol\tabspctype@i{#1}}%
2247\def\tabspctype@i#1{%
2248 \tab@rulefalse%
2249 \ifnum\tab@state>\tab@prespcstate%
2250 \tab@setstate\tab@postspcstate%
2251 \let\tab@tok\tab@posttext%
2252 \else%
2253 \tab@setstate\tab@prespcstate%
2254 \let\tab@tok\tab@pretext%
2255 \fi%
2256 \tab@append\tab@tok{\begingroup#1\endgroup}%
2257}
2258% \end{macrocode}
2259%
2260% \end{macro}
2261%
2262% \begin{macro}{\tabcoltype}
2263%
2264% If we're already in the column state, we bump the state and loop round
2265% again, to get all the appropriate default behaviour. We bump the column
2266% counter, and add the bits of text we were given to appropriate token lists.
2267% We also add the |\hfil| glue to the shortline list, to space out the rules
2268% properly.
2269%
2270% \begin{macrocode}
2271\def\tabcoltype#1#2{%
2272 \ifnum\tab@state=\tab@colstate%
2273 \global\advance\tab@state\@ne%
2274 \fi%
2275 \advance\tab@columns\@ne%
2276 \tab@setstate\tab@colstate%
2277 \tab@append\tab@pretext{#1}%
2278 \tab@append\tab@posttext{#2}%
2279 \tab@append\tab@shortline{\hfil}%
2280}
2281% \end{macrocode}
2282%
2283% \end{macro}
2284%
2285% \begin{macro}{\tabuserpretype}
2286% \begin{macro}{\tabuserposttype}
2287%
2288% These are both utterly trivial.
2289%
2290% \begin{macrocode}
2291\def\tabuserpretype#1{%
2292 \tab@setstate\tab@prestate%
2293 \tab@prepend\tab@userpretext{#1}%
2294}
2295% \end{macrocode}
2296%
2297% \begin{macrocode}
2298\def\tabuserposttype#1{%
2299 \tab@setstate\tab@poststate%
2300 \tab@prepend\tab@posttext{#1}%
2301}
2302% \end{macrocode}
2303%
2304% \end{macro}
2305% \end{macro}
2306%
2307%
2308% \subsection{The colset stack}
2309%
2310% Let's start with something fairly easy. We'll keep a stack of column sets
2311% so that users don't get confused by package authors changing the current
2312% column set. This is fairly easy, really.
2313%
2314% \begin{macro}{\tab@push}
2315% \begin{macro}{\tab@pop}
2316% \begin{macro}{\tab@head}
2317%
2318% These are the stack management routines. The only important thing to note
2319% is that |\tab@head| must take place \emph{only} in \TeX's mouth, so we can
2320% use it in |\csname|\dots|\endcsname| constructions.
2321%
2322% \begin{macrocode}
2323\def\tab@push#1#2{%
2324 \toks@{{#2}}%
2325 \expandafter\def\expandafter#1\expandafter{\the\expandafter\toks@#1}%
2326}
2327\def\tab@pop#1{\edef#1{\expandafter\@gobble#1}}
2328\def\tab@head#1{\expandafter\tab@head@i#1\relax}
2329\def\tab@head@i#1#2\relax{#1}
2330% \end{macrocode}
2331%
2332% \end{macro}
2333% \end{macro}
2334% \end{macro}
2335%
2336% \begin{macro}{\colset}
2337% \begin{macro}{\colpush}
2338% \begin{macro}{\colpop}
2339%
2340% Now we can define the user macros.
2341%
2342% \begin{macrocode}
2343\def\tab@colstack{{tabular}}
2344\def\colset{\colpop\colpush}
2345\def\colpush{\tab@push\tab@colstack}
2346\def\colpop{\tab@pop\tab@colstack}
2347% \end{macrocode}
2348%
2349% \end{macro}
2350% \end{macro}
2351% \end{macro}
2352%
2353% \begin{macro}{\tab@colset}
2354%
2355% Now we define a shortcut for reading the top item off the stack.
2356%
2357% \begin{macrocode}
2358\def\tab@colset{\tab@head\tab@colstack}
2359% \end{macrocode}
2360%
2361% \end{macro}
2362%
2363%
2364% \subsection{The main parser routine}
2365%
2366% \begin{macro}{\tab@initread}
2367%
2368% This macro sets up lots of variables to their normal states prior to
7abe48b3 2369% parsing a preamble. Some things may need changing, but not many. This
2370% version just sets the major hooks, and then does a subread. The midtext
2371% macro contains what to put in the very middle of each template --
2372% |\multicolumn| will insert its argument here.
86f6a31e 2373%
2374% \begin{macrocode}
2375\def\tab@initread{%
7abe48b3 2376 \def\tab@lefttext{}%
2377 \def\tab@leftruletext{}%
2378 \def\tab@righttext{}%
2379 \def\tab@rightruletext{}%
2380 \def\tab@tabtext{&}%
2381 \def\tab@midtext{\ignorespaces####\@maybe@unskip}%
2382 \tab@initsubread%
2383}
2384% \end{macrocode}
2385%
2386% \end{macro}
2387%
2388% \begin{macro}{\tab@initsubread}
2389%
2390% This is where most of the activity is. We don't replace the left and right
2391% texts, so that we effectively inherit them rfrom the enclosing
2392% environment.
2393%
2394% \begin{macrocode}
2395\def\tab@initsubread{%
86f6a31e 2396% \end{macrocode}
2397%
2398% First, reset the parser state to the start state.
2399%
2400% \begin{macrocode}
2401 \global\tab@state\tab@startstate%
2402% \end{macrocode}
2403%
7abe48b3 2404% We clear the token lists to sensible values, mostly.
86f6a31e 2405%
2406% \begin{macrocode}
2407 \tab@preamble{}%
2408 \tab@shortline{}%
86f6a31e 2409 \tab@pretext{}%
2410 \tab@userpretext{}%
2411 \tab@posttext{}%
2412 \let\tab@multicol\@empty%
2413 \def\tab@startpause{\penalty\postdisplaypenalty\medskip}%
2414 \def\tab@endpause{\penalty\predisplaypenalty\medskip}%
2415% \end{macrocode}
2416%
2417% Finally, reset the column counter, don't raise errors when we loop, and set
2418% some parser flags to their appropriate values.
2419%
2420% \begin{macrocode}
2421 \tab@columns\z@%
2422 \let\tab@looped\relax%
2423 \tab@ruletrue%
2424 \tab@initruletrue%
2425 \tab@firstcoltrue%
2426}
7abe48b3 2427
86f6a31e 2428% \end{macrocode}
2429%
2430% \end{macro}
2431%
2432% \begin{macro}{\tab@readpreamble}
2433%
2434% This is the main macro for preamble handling. Actually, all it does is
2435% gobble its argument's leading brace and call another macro, but it does it
2436% with style.
2437%
2438% \begin{macrocode}
2439\def\tab@readpreamble#1{%
7abe48b3 2440 \expandafter\tab@append\expandafter\tab@shortline%
2441 \expandafter{\tab@leftruletext}%
86f6a31e 2442 \tab@doreadpream{#1}%
2443 \iftab@initrule\global\tab@state\tab@prespcstate\fi%
2444 \tab@setstate\tab@rulestate%
7abe48b3 2445 \expandafter\tab@append\expandafter\tab@shortline%
2446 \expandafter{\tab@rightruletext}%
86f6a31e 2447 \tab@commit%
2448}
2449% \end{macrocode}
2450%
2451% \end{macro}
2452%
2453% \begin{macro}{\tab@doreadpream}
2454%
2455% The preamble is in an argument. Previous versions used a nasty trick using
2456% |\let| and |\afterassignment|. Now we use an explicit end token, to allow
2457% dodgy column type handlers to scoop up the remaining preamble tokens
2458% and process them. Not that anyone would want to do that, oh no (see
2459% the \lit{[} type in the \env{eqnarray} environment |;-)|).
2460%
2461% \begin{macrocode}
2462\def\tab@doreadpream#1{\tab@mkpreamble#1\q@delim}
2463% \end{macrocode}
2464%
2465% \end{macro}
2466%
2467% \begin{macro}{\tab@mkpreamble}
2468%
2469% This is the main parser routine. It takes each token in turn, scrutinises
2470% it carefully, and does the appropriate thing with it.
2471%
2472% The preamble was given as an argument to |\tab@doreadpream|, and that has
2473% helpfully stripped off the initial |{| character. We need to pick off the
2474% next token (whatever it is) so we can examine it. We'll use |\futurelet|
2475% so we can detect groups and things in funny places.
2476%
2477% \begin{macrocode}
2478\def\tab@mkpreamble{\futurelet\@let@token\tab@mkpreamble@i}
2479% \end{macrocode}
2480%
2481% If we find a space token, we'll go off and do something a bit special,
2482% since spaces are sort of hard to handle. Otherwise we'll do it in the old
2483% fashioned way.
2484%
2485% \begin{macrocode}
2486\def\tab@mkpreamble@i{%
2487 \ifx\@let@token\@sptoken%
2488 \expandafter\tab@mkpreamble@spc%
2489 \else%
2490 \expandafter\tab@mkpreamble@ii%
2491 \fi%
2492}
2493% \end{macrocode}
2494%
2495% If we find a |\@@endpreamble| token, that's it and we're finished. We just
2496% gobble it and return. Otherwise, if it's an open group character, we'll
2497% complain because someone's probably tried to put an argument in the wrong
2498% place. Finally, if none of the other things apply, we'll deal with the
2499% character below.
2500%
2501% \begin{macrocode}
2502\def\tab@mkpreamble@ii{%
2503 \ifx\@let@token\q@delim%
2504 \def\@tempa{\let\@let@token}%
2505 \else%
2506 \ifcat\bgroup\noexpand\@let@token%
2507 \tab@err@oddgroup%
2508 \def\@tempa##1{\tab@mkpreamble}%
2509 \else%
2510 \let\@tempa\tab@mkpreamble@iii%
2511 \fi%
2512 \fi%
2513 \@tempa%
2514}
2515% \end{macrocode}
2516%
2517% Handle a character. This involves checking to see if it's actually
2518% defined, and then doing it. Doing things this way means we won't get
2519% stranded in mid-preamble unless a package author has blown it.
2520%
2521% \begin{macrocode}
2522\def\tab@mkpreamble@iii#1{%
2523 \@ifundefined{\tab@colset!col.\string#1}{%
2524 \tab@err@undef{#1}\tab@mkpreamble%
2525 }{%
2526 \@nameuse{\tab@colset!col.\string#1}%
2527 }%
2528}
2529% \end{macrocode}
2530%
2531% If we get given a space character, we'll look up the command name as
2532% before. If no-one's defined the column type we'll just skip it silently,
2533% which lets users do pretty formatting if they like.
2534%
2535% \begin{macrocode}
2536\@namedef{tab@mkpreamble@spc} {%
2537 \@ifundefined{\tab@colset!col. }{%
2538 \tab@mkpreamble%
2539 }{%
2540 \@nameuse{\tab@colset!col. }%
2541 }%
2542}
2543% \end{macrocode}
2544%
2545% \end{macro}
2546%
2547% \begin{macro}{\coldef}
2548%
2549% Here's how to define column types the nice way. Some dexterity is required
2550% to make everything work right, but it's simple really.
2551%
2552% \begin{macrocode}
7abe48b3 2553\def\coldef{\@testopt\coldef@i\tab@colset}
86f6a31e 2554\def\coldef@i[#1]#2#3#{\coldef@ii[#1]{#2}{#3}}
2555\def\coldef@ii[#1]#2#3#4{%
2556 \expandafter\def\csname#1!col.\string#2\endcsname#3{%
2557 #4\tab@mkpreamble%
2558 }%
2559}
2560% \end{macrocode}
2561%
2562% \end{macro}
2563%
2564% \begin{macro}{\collet}
2565%
2566% We'd like to let people copy column types from other places. This is how
2567% to do it.
2568%
2569% \begin{macrocode}
7abe48b3 2570\def\collet{\@testopt\collet@i\tab@colset}
86f6a31e 2571\def\collet@i[#1]#2{%
2572 \@ifnextchar=%
2573 {\collet@ii[#1]{#2}}%
2574 {\collet@ii[#1]{#2}=}%
2575}
7abe48b3 2576\def\collet@ii[#1]#2={\@testopt{\collet@iii[#1]{#2}}\tab@colset}
86f6a31e 2577\def\collet@iii[#1]#2[#3]#4{%
2578 \expandafter\let\csname#1!col.\string#2\expandafter\endcsname%
2579 \csname#3!col.\string#4\endcsname%
2580}
2581% \end{macrocode}
2582%
2583% \end{macro}
2584%
2585% \begin{macro}{\newcolumntype}
2586%
2587% We just bundle the text off to |\newcommand| and expect it to cope. It
2588% ought to. The column type code inserts the user's tokens directly, rather
2589% than calling |\tab@doreadpream| recursively. The magic control sequence
2590% is the one looked up by the parser.
2591%
2592% There's some additional magic here for compatibility with the obscure way
2593% that \package{array} works.
2594%
2595% \begin{macrocode}
7abe48b3 2596\def\newcolumntype#1{\@testopt{\nct@i{#1}}0}
86f6a31e 2597\def\nct@i#1[#2]{\@ifnextchar[{\nct@ii{#1}[#2]}{\nct@iii{#1}{[#2]}}}
2598\def\nct@ii#1[#2][#3]{\nct@iii{#1}{[#2][#3]}}
2599\def\nct@iii#1#2#3{%
2600 \expandafter\let\csname\tab@colset!col.\string#1\endcsname\relax%
2601 \expandafter\newcommand\csname\tab@colset!col.\string#1\endcsname#2{%
2602 \tab@deepmagic{#1}%
2603 \tab@mkpreamble%
2604 #3%
2605 }%
2606}
2607% \end{macrocode}
2608%
2609% Now for some hacking for compatibility with \package{tabularx}.
2610%
2611% \begin{macrocode}
2612\def\newcol@#1[#2]{\nct@iii{#1}{[#2]}}
2613% \end{macrocode}
2614%
2615% And now some more. This is seriously deep magic. Hence the name.
2616%
2617% \begin{macrocode}
2618\def\tab@deepmagic#1{%
2619 \csname NC@rewrite@\string#1\endcsname\NC@find\tab@@magic@@%
2620}
2621\def\NC@find#1\tab@@magic@@{}
2622% \end{macrocode}
2623%
2624% \end{macro}
2625%
2626%
2627% \subsection{Standard column types}
2628%
2629% First, make sure we're setting up the right columns. This also sets the
2630% default for the user. Other packages must not use the |\colset| command
2631% for defining columns -- they should use the stack operations defined above.
7abe48b3 2632% For colour support, we ensure that the total stretch in a table cell is
2633% 1\,fil.
2634%
2635% \begin{macrocode}
2636\def\tab@halfhfil{\hskip\z@\@plus.5fil\relax}
2637% \end{macrocode}
2638%
2639% And now on with the show.
86f6a31e 2640%
2641% \begin{macrocode}
2642\colset{tabular}
2643% \end{macrocode}
2644%
2645% Now do the simple alignment types. These are fairly simple. The
2646% mysterious kern in the \lit{l} type is to stop the |\col@sep| glue from
2647% vanishing due to the |\unskip| inserted by the standard |\tab@midtext| if
2648% the column contains no text. (Thanks for spotting this bug go to that
2649% nice Mr~Carlisle.)
2650%
2651% \begin{macrocode}
2652\coldef l{\tabcoltype{\kern\z@\tab@bgroup}{\tab@egroup\hfil}}
7abe48b3 2653\coldef c{\tabcoltype{\tab@halfhfil\tab@bgroup}{\tab@egroup\tab@halfhfil}}
86f6a31e 2654\coldef r{\tabcoltype{\hfil\tab@bgroup}{\tab@egroup}}
2655% \end{macrocode}
2656%
2657% Some extensions now. These are explicitly textual or mathematical
2658% columns. Can be useful if you're providing column types for other people.
2659% I've inserted a kern here for exactly the same reason as for the \lit{l}
2660% column type above.
2661%
2662% \begin{macrocode}
2663\coldef T#1{\tab@aligncol{#1}{\tab@btext}{\tab@etext}}
2664\coldef M#1{\tab@aligncol{#1}{\tab@bmaths}{\tab@emaths}}
2665\def\tab@aligncol#1#2#3{%
2666 \if#1l\tabcoltype{\kern\z@#2}{#3\hfil}\fi%
7abe48b3 2667 \if#1c\tabcoltype{\tab@halfhfil#2}{#3\tab@halfhfil}\fi%
86f6a31e 2668 \if#1r\tabcoltype{\hfil#2}{#3}\fi%
2669}
2670% \end{macrocode}
2671%
2672% Now for the default rules.
2673%
2674% \begin{macrocode}
7abe48b3 2675\coldef ${\@firstoftwo{\tab@withrp\tab@vrule}}
2676\coldef |{\@firstoftwo{\tab@withrp\tab@vrule[]}}
2677\def\tab@vrule#1{\tabruletype{#1\vrule\@width\dimen@}\tab@mkpreamble}
86f6a31e 2678\coldef !#1{\tabruletype{#1}}
2679% \end{macrocode}
2680%
2681% Deal with \lit{@} expressions.
2682%
2683% \begin{macrocode}
2684\coldef @#1{\tabspctype{#1}}
2685% \end{macrocode}
2686%
2687% And the paragraph types. I've added things to handle footnotes here.
2688%
2689% \begin{macrocode}
2690\coldef p#1{\tabcoltype%
2691 {\savenotes\vtop\tab@bpar{#1}}%
2692 {\tab@epar\spewnotes\hfil}}
2693\coldef m#1{\tabcoltype%
2694 {\savenotes$\vcenter\tab@bpar{#1}}%
2695 {\tab@epar$\spewnotes\hfil}}
2696\coldef b#1{\tabcoltype%
2697 {\savenotes\vbox\tab@bpar{#1}}%
2698 {\tab@epar\spewnotes\hfil}}
2699% \end{macrocode}
2700%
2701% Phew. Only a few more left now. The user text ones.
2702%
2703% \begin{macrocode}
2704\coldef >#1{\tabuserpretype{#1}}
2705\coldef <#1{\tabuserposttype{#1}}
7abe48b3 2706\coldef ?#1#2{%
2707 \ifx>#1\expandafter\tabuserpretype%
cd62ff5f 2708 \else\expandafter\tabuserposttype\fi%
7abe48b3 2709 {#2}%
2710 \tab@append\tab@shortline{#2}%
2711}
912d2f01
MW
2712\coldef '#1{%
2713 \tab@append\tab@shortline{#1}%
2714}
86f6a31e 2715% \end{macrocode}
2716%
2717% The strange column type.
2718%
2719% \begin{macrocode}
2720\coldef ##1#2{\tabcoltype{#1}{#2}}
2721% \end{macrocode}
2722%
b1506c80
MW
2723% And \lit{*}, which repeats a preamble spec. The tricky part is ensuring
2724% that nested \lit{*} specs work correctly: we must save the loop counter, in
2725% |\count 0|, obviously, but \emph{also} the |\iterate| macro built which
2726% |\loop| modifies internally. Usually you'd use grouping to fix this, but
2727% if we introduce a group level then we won't update the preamble registers
2728% correctly. Instead, queue up tokens to restore the values in \TeX's input
2729% list.
86f6a31e 2730%
2731% \begin{macrocode}
2732\coldef *#1#2{%
b1506c80
MW
2733 \toks\tw@\expandafter{\iterate}%
2734 \toks@{%
2735 \count@#1%
2736 \loop\ifnum\count@>0\relax%
2737 \tab@doreadpream{#2}%
2738 \advance\count@\m@ne%
2739 \repeat%
2740 }%
2741 \edef\@tempa{%
2742 \the\toks@%
2743 \def\noexpand\iterate{\the\toks\tw@}%
2744 \count@\the\count@%
2745 }%
2746 \@tempa%
86f6a31e 2747}
2748% \end{macrocode}
2749%
2750%
2751% \subsection{Paragraph handling}
2752%
2753% First of all, starting new paragraphs: the vbox token is already there, and
2754% we have the width as an argument.
2755%
2756% \begin{macro}{\tab@bpar}
2757%
2758% There are some gymnastics to do here to support lists which form the
2759% complete text of the parbox. One of the odd things I'll do here is to
2760% not insert a strut on the first line: instead, I'll put the text into a
2761% box register so that I can inspect it later. So that I have access to
2762% the height of the first line, I'll use a |\vtop| -- I can get at the
2763% final depth by using |\prevdepth|, so this seems to be the most general
2764% solution.
2765%
2766% \begin{macrocode}
2767\def\tab@bpar#1{%
2768 \bgroup%
2769 \setlength\hsize{#1}%
2770 \@arrayparboxrestore%
2771 \setbox\z@\vtop\bgroup%
2772 \global\@minipagetrue%
2773 \global\@noskipsecfalse%
2774 \everypar\expandafter{\the\everypar%
2775 \global\@minipagefalse%
2776 \everypar{}%
2777 }%
2778}
2779% \end{macrocode}
2780%
2781% \end{macro}
2782%
2783% \begin{macro}{\tab@epar}
2784%
2785% To end the paragraph, close the box. That sounds easy, doesn't it?
2786% I need to space out the top and bottom of the box so that it looks as if
2787% struts have been applied.
2788%
2789% \begin{macrocode}
2790\def\tab@epar{%
2791% \end{macrocode}
2792%
2793% Anyway, I should end the current paragraph if I'm still in horizontal
2794% mode. A simple |\par| will do this nicely. I'll also remove any trailing
2795% vertical glue (which may be left there by a list environment), because
2796% things will look very strange otherwise.
2797%
2798% \begin{macrocode}
2799 \ifhmode\@maybe@unskip\par\fi%
2800 \unskip%
2801% \end{macrocode}
2802%
2803% Now I'll look at the depth of the last box: if it's less deep than my
2804% special strut, I'll cunningly backpedal by a bit, and add a box with the
2805% appropriate depth. Since this will lie on the previous baseline, it won't
2806% alter the effective height of the box. There's a snag here. |\prevdepth|
2807% may be wrong for example if the last thing inserted was a rule, or the
7abe48b3 2808% box is just empty. Check for this specially. (Thanks to Rowland McDonnell
2809% for spotting this.)
86f6a31e 2810%
2811% \begin{macrocode}
2812 \ifdim\prevdepth>-\@m\p@\ifdim\prevdepth<\dp\@arstrutbox%
2813 \kern-\prevdepth%
2814 \nointerlineskip%
2815 \vtop to\dp\@arstrutbox{}%
2816 \fi\fi%
2817% \end{macrocode}
2818%
2819% I've finished the bottom of the box now: I'll close it, and start work on
2820% the top again.
2821%
2822% \begin{macrocode}
2823 \egroup%
2824% \end{macrocode}
2825%
2826% For top-alignment to work, the first item in the box must be another box.
2827% (This is why I couldn't just set |\prevdepth| at the beginning.) If the
2828% box isn't high enough, I'll add a box of the right height and then kern
2829% backwards so that the `real' first box ends up in the right place.
2830%
2831% \begin{macrocode}
2832 \ifdim\ht\z@<\ht\@arstrutbox%
2833 \vbox to\ht\@arstrutbox{}%
2834 \kern-\ht\z@%
2835 \fi%
2836 \unvbox\z@%
2837 \egroup%
2838}
2839% \end{macrocode}
2840%
2841% \end{macro}
2842%
2843%
2844% \subsection{Gentle persuasion}
2845%
2846% To persuade \package{longtable} to work, we emulate some features of
2847% the \package{array} way of doing things. It's a shame, but we have to do
2848% it, because \package{longtable} came first.
2849%
2850% Note the horribleness with the grouping here. In order to get everything
2851% expanded at the right time, |\@preamble| just replaces itself with the (not
2852% expanded!) preamble string, using |\the|. This means that the preamble
2853% string must be visible in the group just above us. Now,
2854% \package{longtable} (and \package{array} for that matter) does
2855% |\@mkpreamble| immediately after opening a new group. So all we need to do
2856% is close that group, do our stuff, and reopen the group again. (Evil
2857% laughter\dots)
2858%
2859% \begin{macrocode}
2860\def\@mkpream#1{%
2861 \endgroup%
2862 \colset{tabular}%
2863 \tab@initread%
2864 \def\tab@multicol{\@arstrut}%
2865 \tab@preamble{\tab@multicol}%
b1143cde
MW
2866 \let\tab@lefttext\tab@lefttexthook%
2867 \let\tab@righttext\tab@righttexthook%
2868 \let\tab@leftruletext\tab@leftruletexthook%
2869 \let\tab@rightruletext\tab@rightruletexthook%
2870 \def\tab@midtext{\tab@setcr\ignorespaces\@sharp\@sharp\@maybe@unskip}%
86f6a31e 2871 \tab@readpreamble{#1}%
2872 \gdef\@preamble{\the\tab@preamble}%
2873 \let\tab@bgroup\begingroup%
2874 \let\tab@egroup\endgroup%
2875 \begingroup%
2876}
2877% \end{macrocode}
2878%
86f6a31e 2879% \subsection{Debugging}
2880%
2881% This macro just parses a preamble and displays it on the terminal. It
2882% means I can see whether the thing's working.
2883%
2884% \begin{macrocode}
2885\def\showpream#1{%
2886 \tab@initread%
2887 \tab@readpreamble{#1}%
2888 \showthe\tab@preamble%
2889 \showthe\tab@shortline%
2890}
2891% \end{macrocode}
2892%
2893% A quick macro for showing column types.
2894%
2895% \begin{macrocode}
2896\def\showcol#1{%
2897 \expandafter\show\csname\tab@colset!col.\string#1\endcsname%
2898}
2899% \end{macrocode}
2900%
2901%
2902% \subsection{The \env{tabular} and \env{array} environments}
2903%
2904% This is where we define the actual environments which users play with.
2905%
2906% \subsubsection{The environment routines}
2907%
2908% The real work is done in the |\@array| macro later. We just set up lots
2909% (and I mean \emph{lots}) of parameters first, and then call |\@array|.
2910%
2911% \begin{macro}{\tab@array}
2912%
2913% The |\tab@array| macro does most of the common array things.
2914%
2915% \begin{macrocode}
2916\def\tab@array{%
2917 \tab@width\z@%
2918 \let\tab@bgroup\tab@bmaths%
2919 \let\tab@egroup\tab@emaths%
2920 \@tabarray%
2921}
2922% \end{macrocode}
2923%
2924% \end{macro}
2925%
2926% \begin{macro}{\tab@btext}
2927% \begin{macro}{\tab@bmaths}
2928% \begin{macro}{\tab@etext}
2929% \begin{macro}{\tab@emaths}
2930%
2931% These macros contain appropriate things to use when typesetting
2932% text or maths macros. They're all trivial. They're here only for
2933% later modification by funny things like the \env{smarray} environment.
2934%
2935% \begin{macrocode}
2936\def\tab@btext{\begingroup}
7abe48b3 2937\def\tab@bmaths{\color@begingroup$}
86f6a31e 2938\def\tab@etext{\endgroup}
7abe48b3 2939\def\tab@emaths{\m@th$\color@endgroup}
86f6a31e 2940% \end{macrocode}
2941%
2942% \end{macro}
2943% \end{macro}
2944% \end{macro}
2945% \end{macro}
2946%
2947% \begin{environment}{array}
2948%
2949% Now for the \env{array} environment. The `|$|' signs act as a group, so we
2950% don't need to do extra grouping this time. Closing the environment is
2951% easy.
2952%
2953% \begin{macrocode}
2954\def\array{%
2955 \col@sep\arraycolsep%
2956 \let\tab@extrasep\arrayextrasep%
2957 \tab@normalstrut%
2958 \tab@array%
2959}
2960\def\endarray{%
2961 \crcr%
2962 \egroup%
2963 \tab@right%
7abe48b3 2964 \endgroup%
86f6a31e 2965 \tab@restorehlstate%
7abe48b3 2966 \global\c@tabrow\count@%
2967 \def\@currentlabel{\p@tabrow\thetabrow}%
2968 \tab@endhook%
86f6a31e 2969}
2970% \end{macrocode}
2971%
2972% \end{environment}
2973%
2974% \begin{environment}{smarray}
2975%
2976% Now for something a little different. The \env{smarray} environment
2977% gives you an array with lots of small text.
2978%
2979% \begin{macrocode}
2980\def\smarray{%
2981 \extrarowheight\z@%
2982 \col@sep\smarraycolsep%
2983 \let\tab@extrasep\smarrayextrasep%
9d809eac 2984 \def\tab@bmaths{\color@begingroup$\scriptstyle}%
86f6a31e 2985 \def\tab@btext{\begingroup\scriptsize}%
2986 \setbox\z@\hbox{\scriptsize\strut}%
2987 \dimen@\ht\z@\dimen@ii\dp\z@\tab@setstrut%
2988 \tab@array%
2989}
2990\let\endsmarray\endarray
2991% \end{macrocode}
2992%
2993% \end{environment}
2994%
2995% \begin{macro}{\tabstyle}
2996%
2997% This is a little hook that document designers can use to modify the
2998% appearance of tables throughout a document. For example, I've set it to
2999% make the text size |\small| in all tables in this document. Macro writers
3000% shouldn't try to use it as a hook for their own evilness, though. I've
3001% used |\providecommand| to avoid nobbling an existing definition.
3002%
3003% \begin{macrocode}
3004\providecommand\tabstyle{}
3005% \end{macrocode}
3006%
3007% \end{macro}
3008%
3009% \begin{macro}{\@tabular}
3010%
3011% The two \env{tabular} environments share lots of common code, so we
3012% separate that out. (This needs to be done better.) All we really do here
3013% is set up the |\tab@bgroup| and |\tab@egroup| to localise things properly,
3014% and then go.
3015%
3016% \begin{macrocode}
3017\def\@tabular#1{%
3018 \tabstyle%
3019 \setlength\tab@width{#1}%
3020 \let\tab@bgroup\tab@btext%
3021 \let\tab@egroup\tab@etext%
3022 \col@sep\tabcolsep%
3023 \let\tab@extrasep\tabextrasep%
3024 \tab@normalstrut%
3025 \@tabarray%
3026}
3027% \end{macrocode}
3028%
3029% \end{macro}
3030%
3031% \begin{environment}{tabular}
3032% \begin{environment}{tabular*}
3033%
3034% These environments just call a macro which does all the common stuff.
3035%
3036% \begin{macrocode}
3037\def\tabular{\@tabular\z@}
3038\expandafter\let\csname tabular*\endcsname\@tabular
3039\let\endtabular\endarray
3040\expandafter\let\csname endtabular*\endcsname\endarray
3041% \end{macrocode}
3042%
3043% \end{environment}
3044% \end{environment}
3045%
3046% \subsubsection{Setting the strut height}
3047%
3048% \begin{macro}{\tab@setstrut}
3049%
3050% We use a magical strut, called |\@arstrut|, which keeps the table from
3051% collapsing around our heads. This is where we set it up.
3052%
3053% It bases the array strut size on the given values of |\dimen@| and
3054% |\dimen@ii|, amended by various appropriate fiddle values added in by
3055% various people.
3056%
3057% \begin{macrocode}
3058\def\tab@setstrut{%
3059 \setbox\@arstrutbox\hbox{%
3060 \vrule%
3061 \@height\arraystretch\dimen@%
3062 \@depth\arraystretch\dimen@ii%
3063 \@width\z@%
3064 }%
3065}
3066% \end{macrocode}
3067%
3068% \end{macro}
3069%
3070% \begin{macro}{\tab@normalstrut}
3071%
3072% This sets the strut the normal way, from the size of |\strutbox|.
3073%
3074% \begin{macrocode}
3075\def\tab@normalstrut{%
3076 \dimen@\ht\strutbox\advance\dimen@\extrarowheight%
3077 \dimen\tw@\dp\strutbox%
3078 \tab@setstrut%
3079}
3080% \end{macrocode}
3081%
3082% \end{macro}
3083%
3084% \subsubsection{Setting up the alignment}
3085%
3086% The following bits are mainly for other packages to hook themselves onto.
3087%
3088% \begin{macrocode}
7abe48b3 3089\let\@arrayleft\relax
3090\let\@arrayright\relax
3091\let\tab@beginhook\@empty
3092\let\tab@lefttexthook\@empty
3093\let\tab@righttexthook\@empty
3094\let\tab@leftruletexthook\@empty
3095\let\tab@rightruletexthook\@empty
3096\let\tab@endhook\@empty
3097% \end{macrocode}
3098%
3099% For setting these hooks, we provide some handy commands.
3100%
3101% \begin{macrocode}
3102\def\tab@addhookbefore#1#2{%
3103 \toks@{#2}\toks@\expandafter{\the\expandafter\toks@#1}%
3104 \edef#1{\the\toks@}%
3105}
3106\def\tab@addhookafter#1#2{%
3107 \toks@\expandafter{#1#2}%
3108 \edef#1{\the\toks@}%
3109}
86f6a31e 3110% \end{macrocode}
3111%
7abe48b3 3112% And now we get on with the real thing.
3113%
86f6a31e 3114% \begin{macrocode}
3115\def\@tabarray{%
3116 \let\@arrayleft\relax%
3117 \let\@arrayright\relax%
7abe48b3 3118 \@testopt\@array c%
86f6a31e 3119}
3120% \end{macrocode}
3121%
3122% \begin{macro}{\@array}
3123%
3124% The |\@array| macro does most of the real work for the environments. The
3125% first job is to set up the row strut, which keeps the table rows at the
3126% right height. We just take the normal strut box, and extend its height by
3127% the |\extrarowheight| length parameter.
3128%
3129% \begin{macrocode}
3130\def\@array[#1]#2{%
3131% \end{macrocode}
3132%
3133% Sort out the hline state variable. We'll store the old value in a
3134% control sequence to avoid wasting any more count registers.
3135%
3136% \begin{macrocode}
7abe48b3 3137 \tab@beginhook%
3138 \count@\c@tabrow%
3139 \global\c@tabrow\z@%
86f6a31e 3140 \edef\tab@restorehlstate{%
3141 \global\tab@endheight\the\tab@endheight%
3142 \gdef\noexpand\tab@hlstate{\tab@hlstate}%
3143 }%
7abe48b3 3144 \begingroup%
86f6a31e 3145 \def\tab@hlstate{n}%
3146% \end{macrocode}
3147%
3148% Now we read the preamble. All the clever things we've already done are
3149% terribly useful here.
3150%
3151% The |\tab@setcr| sets up |\\| to be a newline even if users have changed it
3152% using something like |\raggedright|.
3153%
3154% \begin{macrocode}
3155 \colset{tabular}%
3156 \tab@initread%
7abe48b3 3157 \let\tab@lefttext\tab@lefttexthook%
3158 \let\tab@righttext\tab@righttexthook%
3159 \let\tab@leftruletext\tab@leftruletexthook%
3160 \let\tab@rightruletext\tab@rightruletexthook%
86f6a31e 3161 \def\tab@midtext{\tab@setcr\ignorespaces####\@maybe@unskip}%
3162 \def\tab@multicol{\@arstrut\tab@startrow}%
3163 \tab@preamble{\tab@multicol\tabskip\z@skip}%
3164 \tab@readpreamble{#2}%
3165% \end{macrocode}
3166%
3167% Set up the default tabskip glue. This is easy: there isn't any.
3168%
3169% \begin{macrocode}
3170 \tab@leftskip\z@skip%
3171 \tab@rightskip\z@skip%
3172% \end{macrocode}
3173%
3174% Now set up the positioning of the table. This is put into a separate macro
3175% because it's rather complicated.
3176%
3177% \begin{macrocode}
3178 \tab@setposn{#1}%
3179% \end{macrocode}
3180%
3181% Now work out how to start the alignment.
3182%
3183% \begin{macrocode}
3184 \ifdim\tab@width=\z@%
3185 \def\tab@halign{}%
3186 \else%
3187 \def\tab@halign{to\tab@width}%
3188 \fi%
3189% \end{macrocode}
3190%
3191% Finally, do all the normal things we need to do before an alignment. Note
3192% that we define |\tabularnewline| first, then set |\\| from that (using
3193% |\tab@setcr|). Since |\\| is reset in the |\tab@midtext| of every table
3194% cell, it becomes secondary to |\tabularnewline|. Doing things this way
3195% avoids the problems with declarations like |\raggedright| which redefine
3196% |\\| in their own (usually rather strange) way, so you don't need to mess
3197% about with things like the |\PreserveBackslash| command given in the
3198% \textit{\LaTeX\ Companion}.
3199%
3200% \begin{macrocode}
3201 \lineskip\z@\baselineskip\z@%
3202 \m@th%
3203 \def\tabularnewline{\tab@arraycr\tab@penalty}%
3204 \tab@setcr%
3205 \let\par\@empty%
3206 \everycr{}\tabskip\tab@leftskip%
3207 \tab@left\halign\tab@halign\expandafter\bgroup%
3208 \the\tab@preamble\tabskip\tab@rightskip\cr%
3209}
3210% \end{macrocode}
3211%
3212% \end{macro}
3213%
3214% You've no doubt noticed the |\tab@left| and |\tab@right| macros above.
3215% These are set up here and elsewhere to allow other things to gain control
3216% at various points of the table (they include and take the place of the
3217% |\@arrayleft| and |\@arrayright| hooks in \package{array}, put in for
3218% \package{delarray}'s use.
3219%
3220% \subsubsection{Positioning the table}
3221%
3222% \begin{macro}{\tab@setposn}
3223%
3224% This macro sets everything up for the table's positioning. It's rather
3225% long, but not all that complicated. Honest.
3226%
3227% First, we set up some defaults (for centring). If anything goes wrong, we
3228% just do the centring things.
3229%
3230% \begin{macrocode}
3231\def\tab@setposn#1{%
3232 \def\tab@left{%
3233 \savenotes%
3234 \leavevmode\hbox\bgroup$\@arrayleft\vcenter\bgroup%
3235 }%
3236 \def\tab@right{%
3237 \egroup%
3238 \m@th\@arrayright$\egroup%
3239 \spewnotes%
3240 }%
3241 \global\tab@endheight\z@%
3242% \end{macrocode}
3243%
3244% For the standard positioning things, we just do appropriate boxing things.
3245% Note that the dollar signs are important, since \package{delarray} might
3246% want to put its delimiters in here.
3247%
3248% The |\if@tempswa| switch it used to decide if we're doing an unboxed
3249% tabular. We'll set it if we find an unbox-type position code, and then
3250% check that everything's OK for this.
3251%
3252% \begin{macrocode}
3253 \@tempswafalse%
3254 \let\tab@penalty\relax%
3255 \if#1t%
3256 \def\tab@left{%
3257 \savenotes%
3258 \leavevmode\setbox\z@\hbox\bgroup$\@arrayleft\vtop\bgroup%
3259 }%
3260 \def\tab@right{%
3261 \egroup%
3262 \m@th\@arrayright$\egroup%
3263 \tab@raisebase%
3264 \spewnotes%
3265 }%
3266 \gdef\tab@hlstate{t}%
3267 \global\tab@endheight\ht\@arstrutbox%
3268 \else\if#1b%
3269 \def\tab@left{%
3270 \savenotes%
3271 \leavevmode\setbox\z@\hbox\bgroup$\@arrayleft\vbox\bgroup%
3272 }%
3273 \def\tab@right{%
3274 \egroup%
3275 \m@th\@arrayright$\egroup%
3276 \tab@lowerbase%
3277 \spewnotes%
3278 }%
3279 \gdef\tab@hlstate{b}%
3280 \else%
3281 \if#1L\@tempswatrue\fi%
3282 \if#1C\@tempswatrue\fi%
3283 \if#1R\@tempswatrue\fi%
3284 \fi\fi%
3285% \end{macrocode}
3286%
3287% Now for some tests to make sure we're allowed to do the unboxing. We text
3288% for |\@arrayleft| being defined, because people trying to hook us won't
3289% understand unboxed tabulars.
3290%
3291% \begin{macrocode}
3292 \if@tempswa\ifhmode%
3293 \ifinner\tab@err@unbrh\@tempswafalse\else\par\fi%
3294 \fi\fi%
3295 \if@tempswa\ifmmode\tab@err@unbmm\@tempswafalse\fi\fi%
3296 \if@tempswa\ifx\@arrayleft\relax\else%
3297 \tab@err@unbext\@tempswafalse%
3298 \fi\fi%
3299% \end{macrocode}
3300%
3301% Finally, if we're still doing an unboxed alignment, we need to sort out the
3302% spacing. We know that no-one's tried to hook on to the environment, so we
3303% clear |\tab@left| and |\tab@right|.
3304%
3305% \begin{macrocode}
3306 \if@tempswa%
3307 \def\tab@left{\vskip\parskip\medskip}%
3308 \def\tab@right{\par\@endpetrue\global\@ignoretrue}%
3309% \end{macrocode}
3310%
3311% Now we need to sort out the alignment. The only way we can do this is by
3312% playing with tabskip glue. There are two possibilities:
3313%
3314% \begin{itemize}
3315%
3316% \item If this is a straight \env{tabular} or an \env{array}, we just use
3317% infinite glue. This is reasonable, I think.
3318%
3319% \item If we have a width for the table, we calculate the fixed values of
3320% glue on either side. This is fairly easy, and forces the table to
3321% the required width.
3322%
3323% \end{itemize}
3324%
3325% First, set up the left and right glues to represent the prevailing
3326% margins set up by \env{list} environments. I think this is the right
3327% thing to do.
3328%
3329% \begin{macrocode}
3330 \tab@leftskip\@totalleftmargin%
3331 \tab@rightskip\hsize%
3332 \advance\tab@rightskip-\linewidth%
3333 \advance\tab@rightskip-\@totalleftmargin%
3334% \end{macrocode}
3335%
3336% First of all, deal with the simple case. I'm using 10000\,fill glue here,
3337% in an attempt to suppress |\extracolsep| glue from making the table the
3338% wrong width. It can always use filll glue if it really needs to, though.
3339%
3340% \begin{macrocode}
3341 \ifdim\tab@width=\z@%
3342 \if#1L\else\advance\tab@leftskip\z@\@plus10000fill\fi%
3343 \if#1R\else\advance\tab@rightskip\z@\@plus10000fill\fi%
3344% \end{macrocode}
3345%
3346% Now for the fun bit. This isn't too hard really. The extra space I must
3347% add around the table adds up to $|\linewidth| - |\tab@width|$. I just
3348% need to add this onto the appropriate sides of the table.
3349%
3350% \begin{macrocode}
3351 \else%
3352 \dimen@\linewidth%
3353 \advance\dimen@-\tab@width%
3354 \if#1L\advance\tab@rightskip\dimen@\fi%
3355 \if#1R\advance\tab@leftskip\dimen@\fi%
3356 \if#1C%
3357 \advance\tab@leftskip.5\dimen@%
3358 \advance\tab@rightskip.5\dimen@%
3359 \fi%
3360 \fi%
3361% \end{macrocode}
3362%
3363% Don't allow page breaks. David Carlisle's wonderful \env{longtable}
3364% package does page breaks far better than I could possibly do here, and
3365% we're compatible with it (wahey!).
3366%
3367% \begin{macrocode}
3368 \def\tab@penalty{\penalty\@M}%
3369% \end{macrocode}
3370%
3371% Finally, set the new width of the table, and leave.
3372%
3373% \begin{macrocode}
3374 \tab@width\hsize%
3375 \fi%
3376}
3377% \end{macrocode}
3378%
3379% \end{macro}
3380%
3381% \subsubsection{Handling tops and bottoms}
3382%
3383% This is how the tops and bottoms of tables are made to line up with the
3384% text on the same line, in the presence of arbitrary rules and space. The
3385% old method, based on the way the \package{array} package worked, wasn't
3386% terribly good. This new version copes much better with almost anything
3387% that gets thrown at it.
3388%
3389% I'll keep a state in a macro (|\tab@hlstate|), which tells me what I'm
3390% meant to be doing. The possible values are \lit{n}, which means I don't
3391% have to do anything, \lit{t}, which means that I'm meant to be handling
3392% top-aligned tables, and \lit{b}, which means that I'm meant to be lining
3393% up the bottom. There are several other `substates' which have various
3394% magic meanings.
3395%
3396% \begin{macrocode}
3397\def\tab@hlstate{n}
3398% \end{macrocode}
3399%
3400% When all's said and done, I extract the box containing the table, and
3401% play with the height and depth to try and make it correct.
3402%
3403% \begin{macro}{\tab@addruleheight}
3404%
3405% This macro is called by `inter-row' things to add their height to our
3406% dimen register.
3407%
3408% Only do this if the state indicates that it's sensible.
3409%
3410% \begin{macrocode}
3411\def\tab@addruleheight#1{%
3412 \if\tab@hlstate n\else%
3413 \global\advance\tab@endheight#1\relax%
3414 \fi%
3415}
3416% \end{macrocode}
3417%
3418% \end{macro}
3419%
3420% \begin{macro}{\tab@startrow}
3421%
3422% This is called at the start of a row, from within the array preamble.
3423% Currently, this assumes that the rows aren't bigger than their struts:
3424% this is reasonable, although slightly limiting, and it could be done better
3425% if I was willing to rip the alignment apart and put it back together
3426% again.
3427%
3428% \begin{macrocode}
3429\def\tab@startrow{%
3430 \if\tab@hlstate t%
3431 \gdef\tab@hlstate{n}%
3432 \else\if\tab@hlstate b%
3433 \global\tab@endheight\dp\@arstrutbox%
3434 \fi\fi%
3435}
3436% \end{macrocode}
3437%
3438% \end{macro}
3439%
3440% \begin{macro}{\tab@raisebase}
3441%
3442% This macro is called at the end of it all, to set the height and depth
3443% of the box correctly. It sets the height to |\tab@endheight|, and the
3444% depth to everything else. The box is in |\box|~0 currently.
3445%
3446% \begin{macrocode}
3447\def\tab@raisebase{%
3448 \global\advance\tab@endheight-\ht\z@%
3449 \raise\tab@endheight\box\z@%
3450}
3451% \end{macrocode}
3452%
3453% \end{macro}
3454%
3455% \begin{macro}{\tab@lowerbase}
3456%
3457% And, for symmetry's sake, here's how to set the bottom properly instead.
3458%
3459% \begin{macrocode}
3460\def\tab@lowerbase{%
3461 \global\advance\tab@endheight-\dp\z@%
3462 \lower\tab@endheight\box\z@%
3463}
3464% \end{macrocode}
3465%
3466% \end{macro}
3467%
3468%
3469% \subsection{Breaking tables into bits}
3470%
3471% Unboxed tables have a wonderful advantage over boxed ones: you can stop
3472% halfway through and do something else for a bit. Here's how:
3473%
3474% \begin{macro}{\tabpause}
3475%
3476% I'd like to avoid forbidding catcode changes here. I'll use |\doafter|
3477% now I've got it, to ensure that colour handling and things occur
3478% \emph{inside} the |\noalign| (otherwise they'll mess up the alignment
3479% very seriously). We selectively include lots of stuff from
3480% |\arrayparboxrestore|.
3481%
3482% We have to be careful here to ensure that everything works correctly within
3483% lists. (The \package{amsmath} package had this problem in its
3484% |\intertext| macro, so I'm not alone here.)
3485%
3486% \begin{macrocode}
3487\def\tabpause#{%
3488 \noalign{\ifnum0=`}\fi%
3489 \let\if@nobreak\iffalse
3490 \let\if@noskipsec\iffalse
3491 \let\par\@@par
3492 \let\-\@dischyph
3493 \let\'\@acci\let\`\@accii\let\=\@acciii
3494 \everypar{}%
3495 \lineskip\normallineskip%
3496 \let\\\@normalcr%
7abe48b3 3497 \color@begingroup%
86f6a31e 3498 \tab@startpause%
3499 \vskip-\parskip%
3500 \parshape\@ne\@totalleftmargin\linewidth%
3501 \noindent%
3502 \doafter\tabpause@i%
3503}
3504\def\tabpause@i{%
3505 \nobreak%
3506 \tab@endpause%
7abe48b3 3507 \color@endgroup%
86f6a31e 3508 \ifnum0=`{\fi}%
3509}
3510% \end{macrocode}
3511%
3512% \end{macro}
3513%
3514%
3515% \subsection{The wonderful world of \cmd\multicolumn}
3516%
3517% \begin{macro}{\multicolumn}
3518%
3519% This is actually fantastically easy. Watch and learn. Make sure you
3520% notice the |\long|s here: remember that some table cells can contain
3521% paragraphs, so it seems sensible to allow |\par| into the argument.
3522% (As far as I know, most other |\multicolumn| commands don't do this,
3523% which seems a little silly. Then again, I forgot to do it the first
3524% time around.)
3525%
3526% \begin{macrocode}
3527\long\def\multicolumn#1#2#3{%
3528 \multispan{#1}%
3529 \begingroup%
3530 \tab@multicol%
7abe48b3 3531 \tab@initsubread%
86f6a31e 3532 \long\def\tab@midtext{#3}%
3533 \let\tab@looped\tab@err@multi%
3534 \tab@readpreamble{#2}%
3535 \the\tab@preamble%
3536 \endgroup%
3537 \ignorespaces%
3538}
3539% \end{macrocode}
3540%
3541% \end{macro}
3542%
3543%
3544% \subsection{Interlude: range lists}
3545%
3546% For processing arguments to |\vgap| and |\cline|, we need to be able to
3547% do things with lists of column ranges. To save space, and to make my
3548% fingers do less typing, here's some routines which do range handling.
3549%
3550% \begin{macro}{\ranges}
3551%
3552% Given a macro name and a comma separated list of ranges and simple numbers,
3553% this macro will call the macro giving it each range in the list in turn.
3554% Single numbers~$n$ will be turned into ranges $n$--$n$.
3555%
3556% The first job is to read the macro to do (which may already have some
3557% arguments attached to it). We'll also start a group to make sure that
3558% our changes to temp registers don't affect anyone else.
3559%
3560% There's a space before the delimiting |\q@delim| to stop numbers being
3561% parsed to far and expanding our quark (which will stop \TeX\ dead in its
3562% tracks). Since we use |\@ifnextchar| to look ahead, spaces in range lists
3563% are perfectly all right.
3564%
3565% \begin{macrocode}
3566\def\ranges#1#2{%
3567 \gdef\ranges@temp{#1}%
3568 \begingroup%
3569 \ranges@i#2 \q@delim%
3570}
3571% \end{macrocode}
3572%
3573%
3574% We're at the beginning of the list. We expect either the closing marker
3575% (if this is an empty list) or a number, which we can scoop up into a
3576% scratch register.
3577%
3578% \begin{macrocode}
3579\def\ranges@i{%
3580 \@ifnextchar\q@delim\ranges@done{\afterassignment\ranges@ii\count@}%
3581}
3582% \end{macrocode}
3583%
3584% We've read the first number in the range. If there's another number, we'll
3585% expect a `|-|' sign to be next. If there is no `|-|', call the user's code
3586% with the number duplicated and then do the rest of the list.
3587%
3588% \begin{macrocode}
3589\def\ranges@ii{%
3590 \@ifnextchar-\ranges@iii{\ranges@do\count@\count@\ranges@v}%
3591}
3592% \end{macrocode}
3593%
3594% Now we strip the `|-|' off and read the other number into a temporary
3595% register.
3596%
3597% \begin{macrocode}
3598\def\ranges@iii-{\afterassignment\ranges@iv\@tempcnta}
3599% \end{macrocode}
3600%
3601% We have both ends of the range now, so call the user's code, passing it
3602% both ends of the range.
3603%
3604% \begin{macrocode}
3605\def\ranges@iv{\ranges@do\count@\@tempcnta\ranges@v}
3606% \end{macrocode}
3607%
3608% We've finished doing an item now. If we have a `|,|' next, then start
3609% over with the next item. Otherwise, if we're at the end of the list,
3610% we can end happily. Finally, if we're totally confused, raise an
3611% error.
3612%
3613% \begin{macrocode}
3614\def\ranges@v{%
3615 \@ifnextchar,%
3616 \ranges@vi%
3617 {%
3618 \@ifnextchar\q@delim%
3619 \ranges@done%
3620 {\tab@err@range\ranges@vi,}%
3621 }%
3622}
3623% \end{macrocode}
3624%
3625% We had a comma, so gobble it, read the next number, and go round again.
3626%
3627% \begin{macrocode}
3628\def\ranges@vi,{\afterassignment\ranges@ii\count@}
3629% \end{macrocode}
3630%
3631% Here's how we call the user's code, now. We close the group, so that the
3632% user's code doesn't have to do global things to remember its results, and
3633% we expand the two range ends from their count registers. We also ensure
3634% that the range is the right way round.
3635%
3636% \begin{macrocode}
3637\def\ranges@do#1#2{%
3638 \ifnum#1>#2\else%
3639 \expandafter\endgroup%
3640 \expandafter\ranges@temp%
3641 \expandafter{%
3642 \the\expandafter#1%
3643 \expandafter}%
3644 \expandafter{%
3645 \the#2%
3646 }%
3647 \begingroup%
3648 \fi%
3649}
3650% \end{macrocode}
3651%
3652% And finishing the scan is really easy. We close the group after gobbling
3653% the close token.
3654%
3655% \begin{macrocode}
3656\def\ranges@done\q@delim{\endgroup}
3657% \end{macrocode}
3658%
3659% \end{macro}
3660%
3661% \begin{macro}{\ifinrange}
3662%
3663% Something a little more useful, now. |\ifinrange| takes four arguments:
3664% a number, a range list (as above), and two token lists which I'll call
3665% \emph{then} and \emph{else}. If the number is in the list, I'll do
3666% \emph{then}, otherwise I'll do \emph{else}.
3667%
3668% \begin{macrocode}
3669\def\ifinrange#1#2{%
3670 \@tempswafalse%
3671 \count@#1%
3672 \ranges\ifinrange@i{#2}%
3673 \if@tempswa%
3674 \expandafter\@firstoftwo%
3675 \else%
3676 \expandafter\@secondoftwo%
3677 \fi%
3678}
3679\def\ifinrange@i#1#2{%
3680 \ifnum\count@<#1 \else\ifnum\count@>#2 \else\@tempswatrue\fi\fi%
3681}
3682% \end{macrocode}
3683%
3684% \end{macro}
3685%
3686%
3687% \subsection{Horizontal rules OK}
3688%
3689% This is where all the gubbins for |\vgap| and friends is kept, lest it
3690% contaminate fairly clean bits of code found elsewhere.
3691%
7abe48b3 3692% \subsubsection{Common parsing for rule parameters twiddling}
3693%
3694% \begin{macro}{\tab@ruleparams}
86f6a31e 3695%
7abe48b3 3696% Given a macro name, make a (global) macro |\tab@ruledecls|, which sets
3697% |\dimen0| to be the chosen rule thickness, and sets up colours and whatnot,
3698% and then and calls the macro. We parse a `|*|' to mean
3699% |\arraythickrulewidth|, an optional argument which should be something
3700% |\setlength| can understand, or nothing, which gives the default
3701% |\arrayrulewidth|.
86f6a31e 3702%
7abe48b3 3703% To make this properly hookable, we need to make a list of properties and
3704% gather them together.
86f6a31e 3705%
3706% \begin{macrocode}
7abe48b3 3707\let\tab@rp@inithook\@empty
3708\let\tab@rp@sethook\@empty
3709\let\tab@rp@donehook\@empty
3710\let\tab@rp@default\@empty
3711\def\tab@ruleparams#1{%
3712 {\ifnum0=`}\fi%
3713 \tab@rp@inithook%
3714 \def\tab@rp@next{\ifnum0=`{\fi}#1}%
3715 \expandafter\tab@rp@keys\expandafter{\tab@rp@default}%
3716 \@ifstar\tab@rp@star\tab@rp@what%
86f6a31e 3717}
7abe48b3 3718\def\tab@rp@star{\dimen@\arraythickrulewidth\tab@rp@what}
3719\def\tab@rp@what{\@ifnextchar[\tab@rp@opt\tab@rp@done}
3720\def\tab@rp@opt[#1]{\tab@rp@keys{#1}\tab@rp@done}
3721\def\tab@rp@keys{\mkparse{mdwtab:rule}}
3722\def\tab@rp@done{%
3723 \protected@xdef\tab@rp@{\tab@rp@sethook}%
3724 \tab@rp@donehook%
3725 \tab@rp@next%
86f6a31e 3726}
7abe48b3 3727\def\tab@withrp#1{\tab@ruleparams{\tab@withrp@i{#1}}}
3728\def\tab@withrp@i#1{%
3729 \toks@{#1}%
3730 \toks@\expandafter{\the\expandafter\toks@\expandafter{\tab@rp@}}%
3731 \the\toks@%
3732}
3733% \end{macrocode}
3734%
3735% And now to define the width parameters.
3736%
3737% \begin{macrocode}
3738\tab@addhookafter\tab@rp@inithook{\dimen@\arrayrulewidth}
3739\tab@addhookafter\tab@rp@sethook{\dimen@\the\dimen@}
3740\tab@addhookafter\tab@rp@donehook{\global\tab@rulewidth\dimen@}
3741\mkdef{mdwtab:rule}{width}{\setlength\dimen@{#1}}
3742\mkdef{mdwtab:rule}{thin}*{\dimen@\arrayrulewidth}
3743\mkdef{mdwtab:rule}{thick}*{\dimen@\arraythickrulewidth}
3744\mkdef*{mdwtab:rule}*{\setlength\dimen@{#1}}
3745% \end{macrocode}
3746%
3747% \end{macro}
3748%
3749% \begin{macro}{\tabsetruleparams}
3750%
3751% And the user default-parameter list.
3752%
3753% \begin{macrocode}
3754\def\tabsetruleparams{\def\tab@rp@default}
86f6a31e 3755% \end{macrocode}
3756%
3757% \end{macro}
3758%
3759% \subsubsection{Drawing horizontal rules}
3760%
3761% \begin{macro}{\hline}
3762%
3763% Note the funny use of |\noalign| to allow \TeX\ stomach ops like
3764% |\futurelet| without starting a new table row. This lets us see if there's
3765% another |\hline| coming up, so we can see if we need to insert extra
3766% vertical space.
3767%
3768% \begin{macrocode}
7abe48b3 3769\def\hline{\noalign\tab@ruleparams\hline@prep}
86f6a31e 3770\def\hline@prep{%
3771 \tab@dohline%
3772 \noalign{\ifnum0=`}\fi%
3773 \tab@penalty%
3774 \futurelet\@let@token\hline@i%
3775}
3776% \end{macrocode}
3777%
3778% We check here for another |\hline| command, and insert glue if there is.
3779% This looks terrible, though, and |\hlx{hvh}| is much nicer. Still\dots
3780%
3781% \begin{macrocode}
3782\def\hline@i{%
3783 \@tempswafalse%
3784 \ifx\@let@token\hline\@tempswatrue\fi%
3785 \ifx\@let@token\hline@prep\@tempswatrue\fi%
3786 \if@tempswa%
3787 \vskip\doublerulesep%
3788 \tab@addruleheight\doublerulesep%
3789 \fi%
3790 \ifnum0=`{\fi}%
3791}
3792% \end{macrocode}
3793%
3794% \end{macro}
3795%
3796% \begin{macro}{\tab@dohline}
3797%
e8e9e5d8 3798% This is where hlines actually get drawn.
86f6a31e 3799% Drawing lines is more awkward than it used to be, particularly in unboxed
3800% tables. It used to be a case simply of saying |\noalign{\hrule}|.
3801% However, since unboxed tables are actually much wider than they look, this
3802% would make the rules stretch right across the page and look generally
3803% horrible.
3804%
3805% The solution is simple: we basically do a dirty big |\cline|.
3806%
3807% \begin{macrocode}
3808\def\tab@dohline{%
7abe48b3 3809 \multispan\tab@columns%
3810 \color@begingroup%
3811 \tab@rp@\leaders\hrule\@height\dimen@\hfil%
3812 \tab@addruleheight\dimen@%
3813 \color@endgroup%
86f6a31e 3814 \cr%
3815}
3816% \end{macrocode}
3817%
3818% \end{macro}
3819%
3820% \subsubsection{Vertical rules}
3821%
3822% I couldn't fit these in anywhere else, so they'll have to go here. I'll
3823% provide a new optional argument which specifies the width of the rule; this
3824% gets rid of the problem described in the \emph{Companion}, where to get
3825% an unusually wide vertical rule, you have to play with things like
3826% \syntax{"\\vrule width" <dimen>} which really isn't too nice.
3827%
3828% \begin{macro}{\vline}
3829%
7abe48b3 3830% Now uses the general |\tab@ruleparams| parser. We save and restore the
3831% global |\tab@rulewidth| parameter here.
86f6a31e 3832%
3833% \begin{macrocode}
3834\def\vline{%
3835 \begingroup%
7abe48b3 3836 \@tempdima\tab@rulewidth\let\safe@\tab@rp@%
3837 \tab@ruleparams\tab@vline%
3838}
3839\def\tab@vline{%
3840 \tab@rp@\vrule\@width\dimen@%
3841 \global\tab@rulewidth\@tempdima\global\let\tab@rp@\safe@%
3842 \endgroup%
86f6a31e 3843}
3844% \end{macrocode}
3845%
3846% \end{macro}
3847%
3848% \subsubsection{Drawing bits of lines}
3849%
3850% Just for a bit of fun, here's an extended version of |\cline| which takes
3851% a list of columns to draw lines under, rather than just a single range.
3852%
3853% \begin{macro}{\cline}
3854%
3855% Not a single line of code written yet, and we already have a dilemma on
3856% our hands. Multiple consecutive |\cline| commands are meant to draw
3857% on the same vertical bit of table. But horizontal lines are meant to have
7abe48b3 3858% thickness now. Worse, if the lines have real thickness then we leave gaps
3859% in the vertical rules which aren't covered by our line. But if we
3860% backspace over the line, then we overwrite it with coloured blobs.
3861%
3862% We give up on doing the job properly -- that's just doomed. Backspace over
3863% the previous row, and provide a hack for doing the spacing right elsewhere.
86f6a31e 3864%
3865% Now the problem remains how best to do the job. The way I see it, there
3866% are three possibilities:
3867%
3868% \begin{itemize}
3869%
3870% \item We can start a table row, and then for each column of the table
3871% (as recorded in |\tab@columns|) we look to see if that column is
3872% listed in the range list and if so draw the rule. This requires
3873% lots of scanning of the range list.
3874%
3875% \item We can take each range in the list, and draw rules appropriately,
3876% just like the old |\cline| used to do, and starting a new table row
3877% for each.
3878%
3879% \item We can start a table row, and then for each range remember where we
3880% stopped drawing the last row, move to the start of the new one, and
3881% draw it. If we start moving backwards, we close the current row
3882% and open a new one.
3883%
3884% \end{itemize}
3885%
3886% The last option looks the most efficient, and the most difficult. This
3887% is therefore what I shall do |;-)|.
3888%
3889% The first thing to do is to add in a little negative space, and start a
3890% table row (omitting the first item). Then scan the range list, and finally
3891% close the table row and add some negative space again.
3892%
3893% We need a global count register to keep track of where we are. Mixing
3894% local and global assignments causes all sorts of tragedy, so I shall hijack
3895% |\tab@state|.
3896%
3897% \begin{macrocode}
7abe48b3 3898\def\cline{\noalign\tab@ruleparams\cline@do}
86f6a31e 3899% \end{macrocode}
3900%
3901% Now for the tricky bit. When we're given a range, we look to see if the
3902% first number is less than |\tab@state|. If so, we quickly close the
3903% current row, kern backwards and start again with an |\omit| and reset
7abe48b3 3904% |\tab@state| to 1, and try again. This is hardly perfect, but gets the job
3905% done in many cases. Correct |\vgap| insertion fixes the remaining bugs.
86f6a31e 3906%
3907% \begin{macrocode}
7abe48b3 3908\def\cline@do#1{%
3909 \noalign{\kern-\tab@rulewidth}%
3910 \omit%
3911 \global\tab@state\@ne%
3912 \ranges\cline@do@i{#1}\cr%
3913}
3914\def\cline@do@i#1#2{%
86f6a31e 3915 \ifnum#1<\tab@state\relax%
3916 \tab@@cr%
3917 \noalign{\kern-\tab@rulewidth\tab@penalty}%
3918 \omit%
3919 \global\tab@state\@ne%
3920 \fi%
3921% \end{macrocode}
3922%
3923% We are now either at or in front of the column position required. If
7abe48b3 3924% we're too far back, we must |\hfil&\omit| our way over to the correct
86f6a31e 3925% column.
3926%
3927% \begin{macrocode}
3928 \@whilenum\tab@state<#1\do{%
3929 \hfil\tab@@tab@omit%
3930 \global\advance\tab@state\@ne%
3931 }%
3932% \end{macrocode}
3933%
3934% We've found the start correctly. We must deal with a tiny problem now:
3935% if this is not the first table cell, the left hand vertical rule is in the
3936% column to the left, so our horizontal rule won't match up properly. So
3937% we skip back by a bit to compensate. If there isn't actually a vertical
3938% rule to line up with, no-one will notice, because the rules are so thin.
3939% This adds a little touch of quality to the whole thing, which is after all
3940% the point of this whole exercise.
3941%
3942% \begin{macrocode}
3943 \ifnum\tab@state>\@ne%
3944 \kern-\arrayrulewidth%
3945 \fi%
3946% \end{macrocode}
3947%
3948% Now we must stretch this table cell to the correct width.
3949%
3950% \begin{macrocode}
3951 \@whilenum\tab@state<#2\do{%
3952 \tab@@span@omit%
3953 \global\advance\tab@state\@ne%
3954 }%
3955% \end{macrocode}
3956%
3957% We're ready. Draw the rule. Note that this is |\hfill| glue, just in case
3958% we start putting in |\hfil| glue when we step onto the next cell.
3959%
3960% \begin{macrocode}
7abe48b3 3961 \color@begingroup%
3962 \tab@rp@%
86f6a31e 3963 \leaders\hrule\@height\tab@rulewidth\hfill%
7abe48b3 3964 \color@endgroup%
86f6a31e 3965}
3966% \end{macrocode}
3967%
3968% Some alignment primitives are hidden inside macros so they don't get seen
3969% at the wrong time. This is what they look like:
3970%
3971% \begin{macrocode}
3972\def\tab@@cr{\cr}
3973\def\tab@@tab@omit{&\omit}
3974\def\tab@@span@omit{\span\omit}
3975% \end{macrocode}
3976%
3977% \end{macro}
3978%
3979% \subsubsection{Drawing short table rows}
3980%
3981% Before I start on a description of more code, I think I'll briefly discuss
3982% my reasons for leaving the |\vgap| command in its current state. There's a
3983% reasonable case for introducing an interface between |\vgap| and
3984% |\multicolumn|, to avoid all the tedious messing about with column
3985% ranges. There are good reasons why I'm not going to do this:
3986%
3987% \begin{itemize}
3988%
3989% \item It's very difficult to do: it requires either postprocessing of
3990% the table or delaying processing of each row until I know exactly
3991% what's in it; a |\multicolumn| in a row should be able to affect
3992% a |\vgap| before the row, which gets very nasty. This package is
3993% probably far too large already, and adding more complexity and
3994% running the risk of exhausting \TeX's frustratingly finite capacity
3995% for the sake of relieving the user of a fairly trivial job doesn't
3996% seem worthwhile.
3997%
3998% \item Perhaps more importantly, there are perfectly valid occasions when
3999% it's useful to have the current vgap behaviour. For example, the
4000% \texttt{MIX} word layout diagrams found in \emph{The Art of
4001% Computer Programming} use the little `stub lines' to show where
4002% data items cross byte boundaries:
4003%
eafdddad
MW
4004% ^^A This actually looks terrifyingly similar to the original.
4005% ^^A The leading @{} is there to stop the table looking off-centre,
4006% ^^A because there's no left hand rule telling you where the table
4007% ^^A starts, like there is on the right, just the \tabcolsep glue.
4008%
4009% \begingroup
4010% \newcommand{\wide}[2]{\multicolumn{#1}{c|}{\ttfamily #2}}
4011% \begin{tabular}[C]{@{} r @{\qquad} | Mc | *{5}{c|}} \hlx{c{2-7} v}
4012% empty & - & 1 & 0 & 0 & 0 & 0 \\ \hlx{v c{2-7} v}
4013% occupied & + & \wide{2}{LINK} & \wide{3}{KEY} \\ \hlx{v c{2-7}}
4014% \end{tabular}
4015% \endgroup
86f6a31e 4016%
4017% \end{itemize}
4018%
4019% That's my excuses out of the way; now I'll press on with the actual
4020% programming.
4021%
4022% \begin{macro}{\tab@checkrule}
4023%
4024% We have a range list in |\tab@xcols| and a number as an argument. If we
4025% find the number in the list, we just space out the following group,
4026% otherwise we let it be.
4027%
4028% \begin{macrocode}
4029\def\tab@checkrule#1{%
4030 \count@#1\relax%
4031 \expandafter\ifinrange%
4032 \expandafter\count@%
4033 \expandafter{\tab@xcols}%
4034 {\tab@checkrule@i}%
4035 {}%
4036}
4037\def\tab@checkrule@i#1{\setbox\z@\hbox{#1}\hb@xt@\wd\z@{}}
4038% \end{macrocode}
4039%
4040% \end{macro}
4041%
4042% \begin{macro}{\vgap}
4043%
4044% We must tread carefully here. A single misplaced stomach operation can
4045% cause error messages. We therefore start with an |\omit| so we can search
4046% for optional arguments.
4047%
4048% So that |\hlx| can get control after |\vgap| has finished, we provide a
4049% hook called |\vgap@after| which is expanded after |\vgap| has finished.
4050% Here we make it work like |\@empty|, which expands to nothing. (Note that
4051% |\relax| will start a new table row, so we can't use that.) There are
4052% some penalty items here to stick the |\vgap| row to the text row and
4053% |\hline| that are adjacent to it. The \package{longtable} package will
4054% split an |\hline| in half, so this is the correct thing to do.
4055%
4056% \begin{macrocode}
4057\def\vgap{%
4058 \noalign{\nobreak}%
4059 \omit%
4060 \global\let\vgap@after\@empty%
4061 \iffalse{\fi\ifnum0=`}\fi%
4062 \@ifnextchar[\vgap@i\vgap@simple%
4063}
4064% \end{macrocode}
4065%
4066% We set up two different sorts of |\vgap| -- a simple one which allows all
4067% rules to be passed through, and a specific one which carefully vets each
4068% one (and is therefore slower). We decide which to so based on the presence
4069% of an optional argument.
4070%
4071% The optional argument handler just passes its argument to an interface
4072% routine which is used by |\hlx|.
4073%
4074% \begin{macrocode}
4075\def\vgap@i[#1]{\vgap@spec{#1}}
4076% \end{macrocode}
4077%
4078% Now we handle specified columns. Since we're in an omitted table cell, we
4079% must set things up globally. Assign the column spec to a macro, and set up
4080% vetting by the routine above. Then just go and do the job.
4081%
4082% \begin{macrocode}
4083\def\vgap@spec#1#2{%
4084 \gdef\tab@xcols{#1}%
4085 \global\let\tab@ckr\tab@checkrule%
4086 \vgap@do{#2}%
4087}
4088% \end{macrocode}
4089%
4090% Handle all columns. Just gobble the column number for each rule, and let
4091% the drawing pass unharmed. Easy.
4092%
4093% \begin{macrocode}
4094\def\vgap@simple#1{%
4095 \global\let\tab@ckr\@gobble%
4096 \vgap@do{#1}%
4097}
4098% \end{macrocode}
4099%
4100% This is where stuff actually gets done. We set the |\vgap| flag on while
4101% we do the short row. Then just expand the token list we built while
4102% scanning the preamble.
4103%
4104% Note that the flag is cleared at the end of the last column, to allow other
4105% funny things like |\noalign| and |\omit| before a new row is started.
4106%
4107% \begin{macrocode}
4108\def\vgap@do#1{%
4109 \ifnum0=`{}\fi%
4110 \global\tab@vgaptrue%
4111 \the\tab@shortline%
4112 \vrule\@height#1\@width\z@%
4113 \global\tab@vgapfalse
4114 \tab@addruleheight{#1}%
4115 \cr%
4116 \noalign{\nobreak}%
4117 \vgap@after%
4118}
4119% \end{macrocode}
4120%
4121% \end{macro}
4122%
4123% \subsubsection{Prettifying syntax}
4124%
4125% \begin{macro}{\hlx}
4126%
4127% This is like a poor cousin to the preamble parser. The whole loop is
4128% carefully written to take place \emph{only} in \TeX's mouth, so the
4129% alignment handling bits half way down the gullet don't see any of this.
4130%
4131% First, pass the string to another routine.
4132%
4133% \begin{macrocode}
7abe48b3 4134\def\hlx{\noalign\tab@ruleparams\hlx@prep}
86f6a31e 4135\def\hlx@prep#1{\hlx@loop#1\q@delim}
4136% \end{macrocode}
4137%
4138% Now peel off a token, and dispatch using |\csname|. We handle
4139% undefinedness of the command in a fairly messy way, although it probably
4140% works. Maybe.
4141%
4142% \begin{macrocode}
4143\def\hlx@loop#1{%
4144 \ifx#1\q@delim\else%
4145 \@ifundefined{hlx@cmd@\string#1}{%
4146 \expandafter\hlx@loop%
4147 }{%
4148 \csname hlx@cmd@\string#1\expandafter\endcsname%
4149 }%
4150 \fi%
4151}
4152% \end{macrocode}
4153%
4154% \end{macro}
4155%
4156% \begin{macro}{\hlxdef}
4157%
4158% New |\hlx| commands can be defined using |\hlxdef|. This is a simple
4159% abbreviation.
4160%
4161% \begin{macrocode}
4162\def\hlxdef#1{\@namedef{hlx@cmd@#1}}
4163% \end{macrocode}
4164%
4165% \end{macro}
4166%
4167% \begin{macro}{\hlx h}
4168%
4169% Handle an \lit{h} character. Just do an |\hline| and return to the loop.
4170% We look ahead to see if there's another \lit{h} coming up, and if so
4171% insert two |\hline| commands. This strange (and inefficient) behaviour
4172% keeps packages which redefine |\hline| happy.
4173%
4174% \begin{macrocode}
4175\hlxdef h#1{%
4176 \noalign{%
7abe48b3 4177 \ifx#1h\def\@tempa{\hline@prep\hline@prep\hlx@loop}%
4178 \else\def\@tempa{\hline@prep\hlx@loop#1}%
4179 \fi\expandafter
86f6a31e 4180 }%
4181 \@tempa%
4182}
4183% \end{macrocode}
4184%
4185% \end{macro}
4186%
4187% \begin{macro}{\hlx b}
4188%
4189% The \lit{b} character does a nifty backspace, for \package{longtable}'s
4190% benefit.
4191%
4192% \begin{macrocode}
4193\hlxdef b{\noalign{\kern-\arrayrulewidth}\hlx@loop}
4194% \end{macrocode}
4195%
4196% \end{macro}
4197%
4198% \begin{macro}{\hlx /}
4199%
4200% The `"/"' character allows a page break at the current position.
4201%
4202% \begin{macrocode}
7abe48b3 4203\hlxdef /{\noalign{\ifnum0=`}\fi\@testopt\hlx@cmd@break@i0}
4204\def\hlx@cmd@break@i[#1]{\ifnum0=`{\fi}\pagebreak[#1]\hlx@loop}
86f6a31e 4205% \end{macrocode}
4206%
4207% \end{macro}
4208%
4209% \begin{macro}{\hlx v}
7abe48b3 4210% \begin{macro}{\hlx z}
86f6a31e 4211%
7abe48b3 4212% Handle a \lit{v} or \lit{z} character. This is rather like the |\vgap|
4213% code above, although there are syntactic differences.
86f6a31e 4214%
4215% \begin{macrocode}
7abe48b3 4216\hlxdef v{\hlx@vgap\doublerulesep}
4217\hlxdef z{\hlx@vgap\tab@rulewidth}
4218\def\hlx@vgap#1{%
86f6a31e 4219 \noalign{\nobreak}%
4220 \omit%
4221 \iffalse{\fi\ifnum0=`}\fi%
4222 \global\let\vgap@after\hlx@loop%
7abe48b3 4223 \@ifnextchar[{\hlx@vgap@i{#1}}{\hlx@vgap@ii\vgap@simple{#1}}%
86f6a31e 4224}
7abe48b3 4225\def\hlx@vgap@i#1[#2]{%
4226 \ifx!#2!\def\@tempa{\hlx@vgap@ii\vgap@simple{#1}}%
4227 \else\def\@tempa{\hlx@vgap@ii{\vgap@spec{#2}}{#1}}\fi%
86f6a31e 4228 \@tempa%
4229}
7abe48b3 4230\def\hlx@vgap@ii#1#2{\@testopt{\hlx@vgap@iii{#1}}{#2}}
86f6a31e 4231\def\hlx@vgap@iii#1[#2]{#1{#2}}
4232% \end{macrocode}
4233%
4234% \end{macro}
7abe48b3 4235% \end{macro}
86f6a31e 4236%
4237% \begin{macro}{\hlx s}
4238%
4239% Allow the user to leave a small gap using the \lit{s} command.
4240%
4241% \begin{macrocode}
4242\hlxdef s{%
4243 \noalign{\ifnum0=`}\fi%
4244 \nobreak%
7abe48b3 4245 \@testopt\hlx@space@i\doublerulesep%
86f6a31e 4246}
4247\def\hlx@space@i[#1]{%
4248 \vskip#1%
4249 \tab@addruleheight{#1}%
4250 \ifnum0=`{\fi}%
4251 \hlx@loop%
4252}
4253% \end{macrocode}
4254%
4255% \end{macro}
4256%
4257% \begin{macro}{\hlx c}
4258%
7abe48b3 4259% We might as well allow a \lit{c} command to do a |\cline|. The fix to
4260% |\cline| permeates here.
86f6a31e 4261%
4262% \begin{macrocode}
7abe48b3 4263\hlxdef c#1{\cline@do{#1}\hlx@loop}
4264% \end{macrocode}
4265%
4266% \end{macro}
4267%
4268% \begin{macro}{\hlx ?}
4269%
4270% Do some arbitrary stuff which won't typeset. Put the stuff in a box which
4271% is discarded, just in case.
4272%
4273% \begin{macrocode}
4274\hlxdef ?#1{%
4275 \noalign{\setbox\z@\hbox{\color@begingroup#1\color@endgroup}}\hlx@loop%
4276}
4277% \end{macrocode}
4278%
4279% \end{macro}
4280%
4281% \begin{macro}{\hlx !}
4282%
4283% Change parameters in mid-flow.
4284%
4285% \begin{macrocode}
4286\hlxdef !#1{\noalign\tab@ruleparams\hlx@loop[{#1}]}
86f6a31e 4287% \end{macrocode}
4288%
4289% \end{macro}
4290%
4291% \begin{macro}{\hlx .}
4292%
4293% The \lit{.} character forces a start of the new column. There's a little
4294% problem here. Since the \lit{.} character starts the next column, we need
4295% to gobble any spaces following the |\hlx| command before the cell contents
4296% actually starts. Unfortunately, |\ignorespaces| will start the column for
4297% us, so we can't put it in always. We'll handle it here, then. We'll take
4298% the rest of the `preamble' string, and warn if it's not empty. Then we'll
4299% |\ignorespaces| -- this will start the column for us, so we don't need to
4300% |\relax| any more.
4301%
4302% \begin{macrocode}
4303\hlxdef .#1\q@delim{%
4304 \ifx @#1@\else%
4305 \PackageWarning{mdwtab}{%
4306 Ignoring \protect\hlx\space command characters following a
4307 `.'\MessageBreak command%
4308 }%
4309 \fi%
4310 \ignorespaces%
4311}
4312% \end{macrocode}
4313%
4314% \end{macro}
4315%
7abe48b3 4316% \begin{macro}{\hlx +}
4317% \begin{macro}{\nextrow}
4318%
4319% The \lit{+} subcommand just steps the table-row counter.
4320%
4321% \begin{macrocode}
4322\hlxdef +{\nextrow\hlx@loop}
4323\def\nextrow{\noalign{\ifnum0=`}\fi\@testopt\nextrow@i\@ne}
4324\def\nextrow@i[#1]{\global\advance\c@tabrow#1\ifnum0=`{\fi}}
4325% \end{macrocode}
4326%
4327% \end{macro}
4328% \end{macro}
4329%
86f6a31e 4330%
4331% \subsection{Starting new table rows}
4332%
4333% We take a break from careful mouthery at last, and start playing with
4334% newlines. The standard one allows pagebreaks in unboxed tables, which
4335% isn't really too desirable.
4336%
4337% Anyway, we'll try to make this macro rather more reusable than the standard
4338% one. Here goes.
4339%
4340% \begin{macro}{\@arraycr}
4341%
4342% We pass lots of information to a main parser macro, and expect it to cope.
4343%
4344% \begin{macrocode}
4345\def\@arraycr{\tab@arraycr{}}
4346\def\tab@arraycr#1{\tab@cr{\tab@tabcr{#1}}{}{}}
4347% \end{macrocode}
4348%
4349% Now to actually do the work. |\tab@cr| passes us the skip size, and the
4350% appropriate one of the two arguments given above (both of which are empty)
4351% depending on the presence of the $*$.
4352%
4353% \begin{macrocode}
4354\def\tab@tabcr#1#2{%
4355% \end{macrocode}
4356%
4357% If the total height I need to add between rows (from the optional argument
4358% and the `extrasep' parameter) is greater than zero, I'll handle this by
4359% extending the strut slightly. I'm not actually sure whether this is the
4360% right thing to do, to be honest, although it's easier than trying to
4361% to an automatic |\vgap|, because I need to know which columns to skip.
4362% If the space is less than zero, I'll just insert the vertical space with
4363% in a |\noalign|.
4364%
4365% First, to calculate how much space needs adding.
4366%
4367% \begin{macrocode}
4368 \setlength\dimen@{#2}%
4369 \advance\dimen@\tab@extrasep%
4370% \end{macrocode}
4371%
4372% If the height is greater than zero, I need to play with the strut. I must
4373% bear in mind that the current table cell (which I'm still in, remember)
4374% may be in vertical mode, and I may or may not be in a paragraph.
4375%
4376% If I am in vertical mode, I'll backpedal to the previous box and put the
4377% strut in an hbox superimposed on the previous baseline. Otherwise, I can
4378% just put the strut at the end of the text. (This works in either LR
4379% or paragraph mode as long as I'm not between paragraphs.) Again, Rowland's
4380% empty cell bug strikes. (See |\tab@epar| for details.)
4381%
4382% \begin{macrocode}
4383 \ifdim\dimen@>\z@%
4384 \ifvmode%
4385 \unskip\ifdim\prevdepth>-\@m\p@\kern-\prevdepth\fi%
4386 \nointerlineskip\expandafter\hbox%
4387 \else%
4388 \@maybe@unskip\expandafter\@firstofone%
4389 \fi%
4390 {\advance\dimen@\dp\@arstrutbox\vrule\@depth\dimen@\@width\z@}%
4391 \fi%
4392% \end{macrocode}
4393%
4394% This table cell works as a group (which is annoying here). I'll copy the
4395% interrow gap into a global register so that I can use it in the |\noalign|.
4396%
4397% \begin{macrocode}
4398 \global\dimen\@ne\dimen@%
4399 \cr%
4400 \noalign{%
4401 #1%
4402 \ifdim\dimen\@ne<\z@\vskip\dimen\@ne\relax\fi%
4403 }%
4404 \@gobble%
4405}
4406% \end{macrocode}
4407%
4408% \end{macro}
4409%
4410% \begin{macro}{\tab@setcr}
4411%
4412% To set the |\\| command correctly in each table cell, we make it a part of
4413% the preamble (in |\tab@midtext|) to call this routine. It's easy -- just
4414% saves the preamble from being huge.
4415%
4416% \begin{macrocode}
4417\def\tab@setcr{\let\\\tabularnewline}
4418% \end{macrocode}
4419%
4420% \end{macro}
4421%
4422% \begin{macro}{\tab@cr}
4423%
4424% Now we do the parsing work. This is fun. Note the revenge of the funny
4425% braces here. Nothing to worry about, honest. The tricky bit is to keep
4426% track of which arguments are which. (Thanks to David Carlisle for pointing
4427% out that I'd missed out the |\relax| here.)
4428%
4429% \begin{macrocode}
4430\def\tab@cr#1#2#3{%
4431 \relax%
4432 \iffalse{\fi\ifnum0=`}\fi%
4433 \@ifstar{\tab@cr@i{#1}{#3}}{\tab@cr@i{#1}{#2}}%
4434}
7abe48b3 4435\def\tab@cr@i#1#2{\@testopt{\tab@cr@ii{#1}{#2}}\z@}
4436\def\tab@cr@ii#1#2[#3]{\ifnum0=`{}\fi#1{#3}{#2}}
86f6a31e 4437% \end{macrocode}
4438%
4439% \end{macro}
4440%
4441%
4442% \subsection{Gratuitous grotesquery}
4443%
4444% So far we've had an easy-ish ride (or should that be \emph{queasy}?). Now
4445% for something unexplainably evil. We convince \LaTeX\ that it's loaded the
4446% \package{array} package, so that packages which need it think they've got
4447% it.
4448%
4449% The bogus date is the same as the date for the \package{array} package I've
4450% got here -- this will raise a warning if Frank updates his package which
4451% should filter back to me telling me that there's something I need to
4452% know about.
4453%
4454% The messing with |\xdef| and the funny parsing ought to insert the current
4455% \package{mdwtab} version and date into the fake \package{array} version
4456% string, giving a visible clue to the user that this isn't the real
4457% \package{array} package.
4458%
4459% \begin{macrocode}
4460\begingroup
4461\catcode`.=11
4462\def\@tempa#1 #2 #3\@@{#1 #2}
4463\xdef\ver@array.sty
4464 {1995/11/19 [mdwtab.sty \expandafter\@tempa\ver@mdwtab.sty\@@]}
4465\endgroup
4466% \end{macrocode}
4467%
4468%
4469% \subsection{Error messages}
4470%
4471% I've put all the error messages together, where I can find them, translate
4472% them or whatever.
4473%
4474% First, some token-space saving (which also saves my fingers):
4475%
4476% \begin{macrocode}
4477\def\tab@error{\PackageError{mdwtab}}
4478% \end{macrocode}
4479%
4480% Now do the error messages.
4481%
4482% \begin{macrocode}
4483\def\tab@err@misscol{%
4484 \tab@error{Missing column type}{%
4485 I'm lost. I was expecting something describing^^J%
4486 the type of the current column, but you seem to^^J%
4487 have missed it out. I've inserted a type `l'^^J%
4488 column here in the hope that this makes sense.%
4489 }%
4490}
4491% \end{macrocode}
4492%
4493% \begin{macrocode}
4494\def\tab@err@oddgroup{%
4495 \tab@error{Misplaced group in table preamble}{%
4496 I've found an open brace character in your preamble^^J%
4497 when I was expecting a specifier character. I'm^^J%
4498 going to gobble the whole group and carry on as if^^J%
4499 I'd never seen it.%
4500 }%
4501}
4502% \end{macrocode}
4503%
4504% \begin{macrocode}
4505\def\tab@err@undef#1{%
4506 \tab@error{Unknown `\tab@colset' preamble character `\string#1'}{%
4507 I don't understand what you meant by typing this^^J%
4508 character. Anyway, I'll ignore it this time around.^^J%
4509 Just don't you do it again.%
4510 }%
4511}
4512% \end{macrocode}
4513%
4514% \begin{macrocode}
4515\def\tab@err@unbrh{%
4516 \tab@error{Can't use unboxed tabular in LR mode}{%
4517 You've asked for a tabular or array environment with^^J%
4518 `L', `C' or `R' as the position specifier, but you're^^J%
4519 in LR (restricted horizontal) mode, so it won't work.^^J%
4520 I'll assume you really meant `c' and soldier on.%
4521 }%
4522}
4523% \end{macrocode}
4524%
4525% \begin{macrocode}
4526\def\tab@err@unbmm{%
4527 \tab@error{Can't use unboxed tabular in maths mode}{%
4528 You've asked for a tabular or array environment with^^J%
4529 `L', `C' or `R' as the position specifier, but you're^^J%
4530 in maths mode, so it won't work. I'll pretend that^^J%
4531 you really typed `c', and that this is all a bad dream.%
4532 }%
4533}
4534% \end{macrocode}
4535%
4536% \begin{macrocode}
4537\def\tab@err@unbext{%
4538 \tab@error{Can't extend unboxed tabulars}{%
4539 You're trying to use kludgy extensions (e.g.,^^J%
4540 `delarray') on an array or tabular with `L', `C'^^J%
4541 or `R' as the position specifier. I'll assume you^^J%
4542 subconsciously wanted a `c' type all along.%
4543 }%
4544}
4545% \end{macrocode}
4546%
4547% \begin{macrocode}
4548\def\tab@err@multi{%
4549 \tab@error{More than one column in a \protect\multicolumn}{%
4550 You've put more than one column into a \string\multicolumn^^J%
4551 descriptor. It won't work. I have no idea what^^J%
4552 will happen, although it won't be pleasant. Hold^^J%
4553 on tight now...%
4554 }%
4555}
4556% \end{macrocode}
4557%
4558% \begin{macrocode}
4559\def\tab@err@range{%
4560 \tab@error{Expected `,' or `<end>' in range list}{%
4561 I was expecting either the end of the range list,^^J%
4562 or a comma, followed by another range. I've^^J%
4563 inserted a comma to try and get me back on track.^^J%
4564 Good luck.%
4565 }%
4566}
4567% \end{macrocode}
4568%
7abe48b3 4569%
4570% \subsection{Loading the colour package}
4571%
4572% If requested, we load the \package{mtcolour} package here. This ensures
4573% that it can patch this code if it needs to.
4574%
4575% \begin{macrocode}
4576\iftab@colour
4577 \RequirePackage{mtcolour}
4578\fi
4579% \end{macrocode}
4580%
86f6a31e 4581% That's it. No more. Move along please.
4582%
4583% \begin{macrocode}
4584%</mdwtab>
4585% \end{macrocode}
4586%
4587%
4588%^^A-------------------------------------------------------------------------
7abe48b3 4589% \section{Implementation of \package{mtcolour}}
4590%
4591%
4592% This is in a separate package to avoid dragging in the \package{color}
4593% package if it's unwanted.
4594%
4595% I prefer English spellings. Here's a trivial redirection for Americans.
4596%
4597% \begin{macrocode}
4598%<*color>
4599\DeclareOption*{\PassOptionsToPackage{\CurrentOption}{mtcolour}}
4600\ProcessOptions
4601\RequirePackage{mtcolour}
4602%</color>
4603% \end{macrocode}
4604%
4605% And now we can start the thing properly.
4606%
4607% \begin{macrocode}
4608%<*colour>
4609\RequirePackage{color}
4610% \end{macrocode}
4611%
4612%
4613% \subsection{Cell background colours}
4614%
4615% First, some simple preliminaries. The |\iftab@colour| switch is set if the
4616% current cell is meant to have a colour.
4617%
4618% \begin{macrocode}
4619\newif\iftab@colour
4620\tab@colourfalse
4621% \end{macrocode}
4622%
4623% We shall store the cell colour information in |\tab@cellcolour|, and the
4624% row colour information as |\tab@rowcolour|. Because of the structure of
4625% tables, we need to make global assignments; so we must copy the current
4626% value away at the start of a table and put the value back at the end. In
4627% order to transfer the overhang information reliably, we use a separate
4628% control sequence |\tab@colouroverhangs| for that -- otherwise |\color| can
4629% corrupt it.
4630%
4631% \begin{macrocode}
4632\tab@addhookbefore\tab@beginhook{%
4633 \let\tab@saverowcolour\tab@rowcolour%
4634 \let\tab@savecolouroverhangs\tab@colouroverhangs%
4635 \let\tab@savecellcolour\tab@cellcolour%
4636}
4637\tab@addhookafter\tab@endhook{%
4638 \global\let\tab@rowcolour\tab@saverowcolour%
4639 \global\let\tab@colouroverhangs\tab@savecolouroverhangs%
4640 \global\let\tab@cellcolour\tab@savecellcolour%
4641}
4642% \end{macrocode}
4643%
4644% Initially, there are no colours.
4645%
4646% \begin{macrocode}
4647\let\tab@rowcolour\@empty%
4648\let\tab@cellcolour\@empty%
4649\let\tab@colouroverhangs\@empty%
4650% \end{macrocode}
4651%
4652% \begin{macro}{\@snarfcolour}
4653%
4654% Reading a colour specification is something we'll need to do a few times,
4655% so an abstraction is useful. Its single argument is a continuation to
4656% which we pass a colour-spec acceptable to the |\color| command. (This is
4657% the same code as found in the \package{sverb} package. Remember to keep
4658% them in step.)
4659%
4660% \begin{macrocode}
4661\def\@snarfcolour#1{%
4662 \@ifnextchar[{\@snarfcolour@i{#1}}{\@snarfcolour@ii{#1}{}}%
4663}
4664\def\@snarfcolour@i#1[#2]{\@snarfcolour@ii{#1}{[#2]}}
4665\def\@snarfcolour@ii#1#2#3{#1{#2{#3}}}
4666% \end{macrocode}
4667%
4668% \end{macro}
4669%
4670% \begin{macro}{\cellcolour}
4671%
4672% Setting a cell colour is a matter of stashing the right declarations in
4673% |\tab@cellcolour| and |\tab@colouroverhangs|. Note that the overhangs end
4674% up in |\dimen0| and |\dimen2|.
4675%
4676% \begin{macrocode}
4677\def\cellcolour{%
4678 \@ifstar{\tab@ccol@i{\let\tab@rowcolour\@empty}}{\tab@ccol@i{}}%
4679}
4680\def\tab@ccol@i#1{\@snarfcolour{\tab@ccol@ii{#1}}}
4681\def\tab@ccol@ii#1#2{\@testopt{\tab@ccol@iii{#2#1}}\z@}
4682\def\tab@ccol@iii#1[#2]{\@testopt{\tab@ccol@iv{#1}{#2}}{#2}}
4683\def\tab@ccol@iv#1#2[#3]{%
4684 \gdef\tab@cellcolour{\color#1\tab@colourtrue}%
4685 \gdef\tab@colouroverhangs{%
4686 \setlength\dimen@{#2}%
4687 \setlength{\dimen\tw@}{#3}%
4688 }%
4689}
4690% \end{macrocode}
4691%
4692% \end{macro}
4693%
4694% \begin{macro}{\rowcolour}
4695%
4696% Setting the global row colour is simpler, because we don't mess with
4697% overhangs.
4698%
4699% \begin{macrocode}
4700\def\rowcolour{\@snarfcolour\tab@setrowcolour}
4701\let\rowcolor\rowcolour
4702\def\tab@setrowcolour#1{%
4703 \gdef\tab@rowcolour{\color#1\tab@colourtrue}%
4704}
4705% \end{macrocode}
4706%
4707% \end{macro}
4708%
4709% \begin{macro}{\rowcolouroff}
4710%
4711% And turning the global colouring off is easy.
4712%
4713% \begin{macrocode}
4714\def\rowcolouroff{\global\let\tab@rowcolour\@empty}
4715\let\rowcoloroff\rowcolouroff
4716% \end{macrocode}
4717%
4718% \end{macro}
4719%
4720% \begin{macro}{\tab@colourleft}
4721%
4722% Now we start on the table-cell hooks. The left hook starts a box which
4723% will capture the cell's text and natural width. We add the hook to the
4724% rule list as well, so that we can colour the bits in |\vgap|s correctly.
4725%
4726% \begin{macrocode}
4727\tab@addhookbefore\tab@lefttexthook\tab@colourleft
4728\tab@addhookbefore\tab@leftruletexthook\tab@colourleft
4729\def\tab@colourleft{%
4730 \global\let\tab@cellcolour\@empty%
4731 \global\let\tab@colouroverhangs\@empty%
4732 \setbox\z@\hbox\bgroup\color@begingroup%
4733}
4734% \end{macrocode}
4735%
4736% \end{macro}
4737%
4738% \begin{macro}{\tab@colourright}
4739%
4740% The right hook will insert an appropriate rule behind the cell and
4741% retypeset the cell contents over the top. Note that the stretch in a table
4742% cell is exactly 1\,fil. Because we add (leaders) and subtract (negative
4743% |\hskip|) 1\,fil, we retain this stretch exactly. Don't bother unless
4744% there's actually some colouring.
4745%
4746% \begin{macrocode}
4747\tab@addhookafter\tab@righttexthook\tab@colourright
4748\tab@addhookafter\tab@rightruletexthook\tab@colourright
4749\def\tab@colourright{%
4750 \color@endgroup\egroup%
4751 \color@begingroup%
4752 \global\tab@colourfalse%
4753 \tab@cellcolour\tab@rowcolour%
4754 \dimen@\z@\dimen\tw@\z@\tab@colouroverhangs%
4755 \iftab@colour%
4756 \skip@\wd\z@\advance\skip@\z@\@plus1fil%
4757 \skip\tw@\skip@%
4758 \kern-\dimen@%
4759 \advance\skip\tw@\dimen@%
4760 \advance\skip\tw@\dimen\tw@%
4761 \leaders\vrule\hskip\skip\tw@%
4762 \kern-\dimen\tw@%
4763 \hskip-\skip@%
4764 \fi%
4765 \color@endgroup%
4766 \unhbox\z@%
4767}
4768% \end{macrocode}
4769%
4770% \end{macro}
4771%
4772%
4773% \subsection{Coloured rules}
4774%
4775% We hook ourselves onto the rule-parameters edifice. This is rather
4776% straightforward.
4777%
4778% \begin{macrocode}
4779\tab@addhookafter\tab@rp@inithook{%
4780 \let\tab@rulecolour\@empty%
4781 \let\tab@rulecolourmodel\@empty%
4782}
4783\mkdef{mdwtab:rule}{colour}{\tab@setrulecolour{#1}}
4784\mkdef{mdwtab:rule}{colourmodel}{\tab@setrulecolourmodel{#1}}
4785\mkdef{mdwtab:rule}{color}{\tab@setrulecolour{#1}}
4786\mkdef{mdwtab:rule}{colormodel}{\tab@setrulecolourmodel{#1}}
4787\mkdef{mdwtab:rule}{nocolour}*{\let\tab@rulecolour\@empty}
4788\mkdef{mdwtab:rule}{nocolor}*{\let\tab@rulecolour\@empty}
4789\mkdef{mdwtab:rule}{nocolourmodel}*{\let\tab@rulecolourmodel\@empty}
4790\mkdef{mdwtab:rule}{nocolormodel}*{\let\tab@rulecolourmodel\@empty}
4791\def\tab@setrulecolour#1{%
4792 \def\tab@rulecolour{\color\tab@rulecolourmodel{#1}}%
4793}
4794\def\tab@setrulecolourmodel#1{\def\tab@rulecolourmodel{[#1]}}
4795\tab@addhookafter\tab@rp@sethook{\tab@rulecolour}
4796% \end{macrocode}
4797%
4798%
4799% \subsection{Other stuff}
4800%
4801% \begin{macro}{\ifmod}
4802%
55ed6c02 4803% \syntax{"\\ifmod{"$x$"}{"$m$"}{"$y$"}{"<yes>"}{"<no>"}"} -- if $x \bmod m =
7abe48b3 4804% y$ then do \<yes>; otherwise do \<no>.
4805%
4806% \begin{macrocode}
4807\def\ifmod#1#2#3{%
4808 \begingroup%
4809 \@tempcnta#1%
4810 \@tempcntb#2%
4811 \count@\@tempcnta%
4812 \divide\count@\@tempcntb%
4813 \multiply\count@\@tempcntb%
4814 \advance\@tempcnta-\count@%
4815 \count@#3\relax%
4816 \ifnum\@tempcnta=\count@\endgroup\expandafter\@firstoftwo%
4817 \else\endgroup\expandafter\@secondoftwo\fi%
4818}
4819% \end{macrocode}
4820%
4821% \end{macro}
4822%
4823% Done.
4824%
4825% \begin{macrocode}
4826%</colour>
4827% \end{macrocode}
4828%
4829%^^A-------------------------------------------------------------------------
86f6a31e 4830% \section{Implementation of \package{mathenv}}
4831%
4832%
4833% This is in a separate package, mainly to avoid wasting people's memory.
4834%
4835% \begin{macrocode}
4836%<*mathenv>
4837% \end{macrocode}
4838%
4839%
4840% \subsection{Options handling}
4841%
4842% We need to be able to cope with \textsf{fleqn} and \textsf{leqno} options.
4843% This will adjust our magic modified \env{eqnarray} environment
4844% appropriately.
4845%
4846% \begin{macrocode}
4847\newif\if@fleqn
4848\newif\if@leqno
4849\DeclareOption{fleqn}{\@fleqntrue}
4850\DeclareOption{leqno}{\@leqnotrue}
4851\ProcessOptions
4852% \end{macrocode}
4853%
4854% We use the \package{mdwtab} package for all its nice table handling things.
4855% (Oh, and to inflict it on users who want to do nice equations and don't
4856% care about our tables.)
4857%
4858% \begin{macrocode}
4859\RequirePackage{mdwtab}
4860% \end{macrocode}
4861%
4862%
4863% \subsection{Some useful registers}
4864%
4865% The old \LaTeX\ version puts the equation numbers in by keeping a count of
4866% where it is in the alignment. Since I don't know how may columns there are
4867% going to be, I'll just use a switch in the preamble to tell me to stop
4868% tabbing.
4869%
4870% \begin{macrocode}
4871\newif\if@eqalast
4872% \end{macrocode}
4873%
4874% Now define some useful length parameters. First allocate them:
4875%
4876% \begin{macrocode}
4877\newskip\eqaopenskip
4878\newskip\eqacloseskip
4879\newskip\eqacolskip
4880\newskip\eqainskip
4881\newskip\splitleft
4882\newskip\splitright
4883% \end{macrocode}
4884%
4885% Now assign some default values. Users can play with these if they really
4886% want although I can't see the point myself.
4887%
4888% \begin{macrocode}
4889\AtBeginDocument{%
4890 \eqacloseskip\@centering%
4891 \eqacolskip1.5em\@plus\@m\p@
4892 \eqainskip\z@%
4893 \if@fleqn%
4894 \eqaopenskip\mathindent%
4895 \splitleft\mathindent\relax%
e8e9e5d8 4896 \splitright\mathindent\@minus\mathindent\relax%
86f6a31e 4897 \else%
4898 \eqaopenskip\@centering%
4899 \splitleft2.5em\@minus2.5em%
4900 \splitright\splitleft%
4901 \fi%
4902 \relax%
4903}
4904% \end{macrocode}
4905%
4906%
4907% \subsection{A little display handling}
4908%
4909% I'm probably going a little far here, and invading territory already
4910% claimed by the \package{amsmath} stuff (and done a good deal better than
4911% I can be bothered to do), but just for completeness, this is how we handle
4912% attempts to put displays inside other displays without screwing up the
4913% spacing.
4914%
4915% \begin{macro}{\dsp@startouter}
4916%
4917% This is how we start an outermost display. It's fairly easy really. We
4918% make |\dsp@start| start an inner display, and make |\dsp@end| close the
4919% outer display.
4920%
4921% \begin{macrocode}
4922\def\dsp@startouter{%
4923 \let\dsp@end\dsp@endouter%
4924 $$%
4925}
4926% \end{macrocode}
4927%
4928% \end{macro}
4929%
4930% \begin{macro}{\dsp@endouter}
4931%
4932% Ending the outer display is utterly trivial.
4933%
4934% \begin{macrocode}
4935\def\dsp@endouter{$$}
4936% \end{macrocode}
4937%
4938% \end{macro}
4939%
4940% \begin{macro}{\dsp@startinner}
4941%
4942% Starting inner displays is done in a vbox (actually I choose |\vbox| or
4943% |\vtop| depending on the setting of \textsf{leqno} to put the equation
4944% number the right way round).
4945%
4946% \begin{macrocode}
4947\def\dsp@startinner{%
4948 \let\dsp@end\dsp@endinner%
4949 \if@fleqn\kern-\mathindent\fi%
4950 \if@leqno\vtop\else\vtop\fi\bgroup%
4951}
4952% \end{macrocode}
4953%
4954% \end{macro}
4955%
4956% \begin{macro}{\dsp@endinner}
4957%
4958% Ending an inner display is also really easy.
4959%
4960% \begin{macrocode}
4961\def\dsp@endinner{\egroup}
4962% \end{macrocode}
4963%
4964% \end{macro}
4965%
4966% \begin{macro}{\dsp@start}
4967%
4968% This is what other bits of code uses to start displays. It's one of the
4969% start macros up above, and outer by default.
4970%
4971% \begin{macrocode}
4972\def\dsp@start{%
4973 \ifmmode%
4974 %\ifinner\mth@err@mdsp\fi%
4975 \expandafter\dsp@startinner%
4976 \else%
4977 \ifhmode\ifinner\mth@err@hdsp\fi\fi%
4978 \expandafter\dsp@startouter%
4979 \fi%
4980}
4981% \end{macrocode}
4982%
4983% \end{macro}
4984%
4985% \begin{macro}{\dsp@tabpause}
4986%
4987% This sets up the correct pre- and postambles for the |\tabpause| macro in
4988% maths displays. This is fairly simple stuff.
4989%
4990% \begin{macrocode}
4991\def\dsp@tabpause{%
4992 \def\tab@startpause%
4993 {\penalty\postdisplaypenalty\vskip\belowdisplayskip}%
4994 \def\tab@endpause%
4995 {\penalty\predisplaypenalty\vskip\abovedisplayskip}%
4996}
4997% \end{macrocode}
4998%
4999% \end{macro}
5000%
5001%
5002% \subsection{The \env{eqnarray} environment}
5003%
5004% We allow the user to play with the style if this is really wanted. I dunno
5005% why, really. Maybe someone wants very small alignments.
5006%
5007% \begin{macrocode}
5008\let\eqastyle\displaystyle
5009% \end{macrocode}
5010%
5011% \subsubsection{The main environments}
5012%
5013% \begin{environment}{eqnarray}
5014% \begin{environment}{eqnarray*}
5015%
5016% We define the toplevel commands here. They just add in default arguments
5017% and then call |\@eqnarray| with a preamble string. We handle equation
5018% numbers by setting up a default (|\eqa@defnumber|) which is put into
5019% the final column. At the beginning of each row, we globally |\let|
5020% |\eqa@number| equal to |\eqa@defnumber|. The |\eqnumber| macro just
5021% changes |\eqa@number| as required. Since |\eqa@number| is changed globally
5022% we must save it in this environment.
5023%
5024% First, we must sort out the optional arguments and things. This is really
5025% easy. The only difference between the starred and non-starred environments
5026% is the default definition of |\eqa@defnumber|.
5027%
5028% \begin{macrocode}
5029\def\eqnarray{%
5030 \eqnarray@i\eqa@eqcount%
5031}
5032\@namedef{eqnarray*}{\eqnarray@i{}}
7abe48b3 5033\def\eqnarray@i#1{\@testopt{\eqnarray@ii{#1}}{rcl}}
86f6a31e 5034% \end{macrocode}
5035%
5036% Right. Now for the real work. The first argument is the default numbering
5037% tokens; the second is the preamble string.
5038%
5039% \begin{macrocode}
5040\def\eqnarray@ii#1[#2]{%
5041% \end{macrocode}
5042%
5043% Set up the equation counter and labels correctly.
5044%
5045% \medskip\par\noindent|\begin{rant}|\par
5046% The hacking with |\@currentlabel| is here because (in the author's opinion)
5047% \LaTeX's |\refstepcounter| macro is broken. It's currently defined as
5048% \begin{listing}
5049%\def\refstepcounter#1{%
5050% \stepcounter{#1}%
5051% \protected@edef\@currentlabel%
5052% {\csname p@#1\endcsname\csname the#1\endcsname}%
5053%}
5054% \end{listing}
5055% which means that the current label gets `frozen' as soon as you do the
5056% counter step. By redefining the macro as
5057% \begin{listing}
5058%\def\refstepcounter#1{%
5059% \stepcounter{#1}%
5060% \edef\@currentlabel{%
5061% \expandafter\noexpand\csname p@#1\endcsname%
5062% \expandafter\noexpand\csname the#1\endcsname%
5063% }%
5064%}
5065% \end{listing}
5066% these sorts of problems would be avoided, without any loss of functionality
5067% or compatibility that I can see.
5068% \par\noindent|\end{rant}|\par
5069%
5070% \begin{macrocode}
5071 \stepcounter{equation}%
5072 \def\@currentlabel{\p@equation\theequation}%
5073% \end{macrocode}
5074%
5075% The next step is to set up the numbering. I must save the old numbering
5076% so I can restore it later (once in the alignment, I must assign these
5077% things globally).
5078%
5079% \begin{macrocode}
5080 \let\eqa@oldnumber\eqa@number%
5081 \def\eqa@defnumber{#1}%
5082 \global\let\eqa@number\eqa@defnumber%
5083% \end{macrocode}
5084%
5085% The |\if@eqalastfalse| switch is false everywhere except when we're in the
5086% final column.
5087%
5088% \begin{macrocode}
5089 \@eqalastfalse%
5090% \end{macrocode}
5091%
5092% Remove the |\mathsurround| kerning, since it will look very odd inside
5093% the display. We have our own spacing parameters for configuring these
5094% things, so |\mathsurround| is unnecessary.
5095%
5096% \begin{macrocode}
5097 \m@th%
5098% \end{macrocode}
5099%
5100% Time to parse the preamble string now. I must choose the correct column
5101% set, initialise the preamble parser and set up the various macros. The%
5102% extra `|@{\tabskip\eqacloseskip}|' item sets up the tabskip glue to centre
5103% the alignment properly.
5104%
5105% \begin{macrocode}
5106 \colset{eqnarray}%
5107 \tab@initread%
5108 \def\tab@tabtext{&\tabskip\z@skip}%
5109 \tab@preamble{\tabskip\z@skip}%
5110 \tab@readpreamble{#2@{\tabskip\eqacloseskip}}%
5111 \dsp@tabpause%
5112% \end{macrocode}
5113%
5114% Now for some final setting up. The column separation is set from the
5115% user's parameter, the |\everycr| tokens are cleared, and I set up the
5116% newline command appropriately.
5117%
5118% \begin{macrocode}
5119 \col@sep.5\eqainskip%
5120 \everycr{}%
5121 \let\\\@eqncr%
5122% \end{macrocode}
5123%
5124% Now start a maths display and do the alignment. Set up the left hand
5125% tabskip glue to centre the alignment, and do the actual alignment.
5126% The preamble used is mainly that generated from the user's string, although
5127% the stuff at the end is how we set up the equation number -- it repeats
5128% appropriately so we can always find it.
5129%
5130% \begin{macrocode}
5131 \dsp@start%
5132 \tabskip\eqaopenskip%
5133 \halign to\displaywidth\expandafter\bgroup%
5134 \the\tab@preamble%
5135 &&\eqa@lastcol\hb@xt@\z@{\hss##}\tabskip\z@\cr%
5136}
5137% \end{macrocode}
5138%
5139% Now for the end of the environment. This is really easy. Set the final
5140% equation number, close the |\halign|, tidy up the equation counter (it's
5141% been stepped once too many times) and close the display.
5142%
5143% \begin{macrocode}
5144\def\endeqnarray{%
5145 \eqa@eqnum%
5146 \egroup%
5147 \dsp@end%
5148 \global\let\eqa@number\eqa@oldnumber%
5149 \global\@ignoretrue%
5150 \global\advance\c@equation\m@ne%
5151}
5152\expandafter\let\csname endeqnarray*\endcsname\endeqnarray
5153% \end{macrocode}
5154%
5155% \end{environment}
5156% \end{environment}
5157%
e8e9e5d8 5158% Now we can define the column types.
86f6a31e 5159%
5160% \begin{macrocode}
5161\colpush{eqnarray}
5162% \end{macrocode}
5163%
5164% Note the positioning of ord atoms in the stuff below. This will space out
5165% relations and binops correctly when they occur at the edges of columns, and
5166% won't affect ord atoms at the edges, because ords pack closely.
5167%
5168% First the easy ones. Just stick |\hfil| in the right places and
5169% everything will be all right.
5170%
5171% \begin{macrocode}
5172\coldef r{\tabcoltype{\hfil$\eqastyle}{{}$}}
5173\coldef c{\tabcoltype{\hfil$\eqastyle{}}{{}$\hfil}}
5174\coldef l{\tabcoltype{$\eqastyle{}}{$\hfil}}
5175\coldef x{\tabcoltype{\if@fleqn\else\hfil\fi$\eqastyle}{$\hfil}}
5176% \end{macrocode}
5177%
5178% Now for the textual ones. This is also fairly easy.
5179%
5180% \begin{macrocode}
5181\collet T [tabular]T
5182% \end{macrocode}
5183%
5184% Sort of split types of equations. I mustn't use |\rlap| here, or
5185% everything goes wrong -- |\\| doesn't get noticed by \TeX\ in the same way
5186% as |\cr| does.
5187%
5188% \begin{macrocode}
5189\coldef L{\tabcoltype{\hb@xt@2em\bgroup$\eqastyle}{$\hss\egroup}}
5190% \end{macrocode}
5191%
5192% The \lit{:} column type is fairly simple.
5193%
5194% \begin{macrocode}
5195\coldef :{\tabspctype{\tabskip\eqacolskip}}
5196\coldef q{\tabspctype{\quad}}
5197% \end{macrocode}
5198%
5199% The other column types just insert given text in an appropriate way.
5200%
5201% \begin{macrocode}
5202\collet > [tabular]>
5203\collet < [tabular]<
5204\collet * [tabular]*
5205\collet @ [tabular]@
5206% \end{macrocode}
5207%
5208% Finally, the magical `|\magic|' column type, which sets the equation
5209% number. We set up the |\tabskip| glue properly, tab on, and set the flag
5210% which marks the final column. The |\eqa@lastcol| command is there to
5211% raise an error if the user tabs over to this column. I'll temporarily
5212% redefine it to |\@eqalasttrue| when I enter this column legitimately.
5213% The extra magical bits here will make the final column repeat, so that we
5214% can find it if necessary. Well is this column type named.
5215%
5216% That's it. We can return to normal now.
5217%
5218% \begin{macrocode}
5219\colpop
5220% \end{macrocode}
5221%
5222% \subsubsection{Newline codes}
5223%
5224% Newline sequences (|\\|) get turned into calls of |\@eqncr|. The job is
5225% fairly simple, really.
5226%
5227% \begin{macrocode}
5228\def\@eqncr{\tab@cr\eqacr@i\interdisplaylinepenalty\@M}%
5229\def\eqacr@i#1#2{%
5230 \eqa@eqnum%
5231 \noalign{\penalty#2\vskip\jot\vskip#1}%
5232}
5233% \end{macrocode}
5234%
5235% \subsubsection{Setting equation numbers}
5236%
5237% \begin{macro}{\eqa@eqpos}
5238%
5239% Before we start, we need to generalise the flush-left number handling bits.
5240% The macro |\eqa@eqpos| will put its argument in the right place.
5241%
5242% \begin{macrocode}
7ef5ba2c
MW
5243\def\eqa@eqpos#1{%
5244 \if@leqno%
86f6a31e 5245 \hb@xt@.01\p@{}\rlap{\normalfont\normalcolor\hskip-\displaywidth#1}%
7ef5ba2c
MW
5246 \else%
5247 \normalfont\normalcolor#1%
5248 \fi%
5249}
86f6a31e 5250% \end{macrocode}
5251%
5252% \end{macro}
5253%
5254% \begin{macro}{\eqa@eqnum}
5255%
5256% Here we typeset an equation number in roughly the right place. First I'll
5257% redefine |\eqa@lastcol| so that it tells me I'm in the right place, and
5258% start a loop to find that place.
5259%
5260% \begin{macrocode}
5261\def\eqa@eqnum{%
5262 \global\let\eqa@lastcol\@eqalasttrue%
5263 \eqa@eqnum@i%
5264}
5265% \end{macrocode}
5266%
5267% Now for the loop. The |\relax| here is absolutely vital -- it starts the
5268% table column, inserting useful tokens like `|\eqa@lastcol|' which tell
5269% me where I am in the alignment. Then, if I've reached the end, I can
5270% typeset the equation number; otherwise I go off into another macro and
5271% step on to the next column.
5272%
5273% \begin{macrocode}
5274\def\eqa@eqnum@i{%
5275 \relax%
5276 \if@eqalast%
5277 \expandafter\eqa@eqnum@ii%
5278 \else%
5279 \expandafter\eqa@eqnum@iii%
5280 \fi%
5281}
5282\def\eqa@eqnum@ii{%
5283 \eqa@eqpos\eqa@number%
5284 \global\let\eqa@number\eqa@defnumber%
5285 \global\let\eqa@lastcol\eqa@@lastcol%
5286 \cr%
5287}
5288\def\eqa@eqnum@iii{&\eqa@eqnum@i}
5289% \end{macrocode}
5290%
5291% \end{macro}
5292%
5293% \begin{macro}{\eqa@lastcol}
5294%
5295% This is used as a marker for the final column in an \env{eqnarray}
5296% environment. By default it informs the user that they've been very
5297% silly and swallows the contents of the column. I'll redefine it to
5298% something more useful at appropriate times, and then turn it back again.
5299%
5300% \begin{macrocode}
5301\def\eqa@@lastcol{\mth@err@number\setbox\z@}
5302\let\eqa@lastcol\eqa@@lastcol
5303% \end{macrocode}
5304%
5305% \end{macro}
5306%
5307% \subsubsection{Numbering control}
5308%
5309% \begin{macro}{\eqnumber}
5310%
5311% The |\eqnumber| command sets the equation number on the current equation.
5312% This is really easy, actually.
5313%
5314% \begin{macrocode}
5315\newcommand\eqnumber[1][\eqa@eqcount]{\gdef\eqa@number{#1}}
5316% \end{macrocode}
5317%
5318% \end{macro}
5319%
5320% \begin{macro}{\eqa@eqcount}
5321%
5322% This is how a standard equation number is set, stepping the counter and
5323% all. It's really easy and obvious.
5324%
5325% \begin{macrocode}
5326\def\eqa@eqcount{(\theequation)\global\advance\c@equation\@ne}
5327% \end{macrocode}
5328%
5329% \end{macro}
5330%
5331% \begin{macro}{\nonumber}
5332%
5333% The \LaTeX\ |\nonumber| command could be defined by saying
5334% \begin{listing}
5335%\renewcommand{\nonumber}{\eqnumber[]}
5336% \end{listing}
5337% but I'll be slightly more efficient and redefine |\eqa@number| directly.
5338%
5339% \begin{macrocode}
5340\def\nonumber{\global\let\eqa@number\@empty}
5341% \end{macrocode}
5342%
5343% \end{macro}
5344%
5345% \subsubsection{The \env{eqnalign} environment}
5346%
5347% As a sort of companion to \env{eqnarray}, here's an environment which does
5348% similar things inside a box, rather than taking up the whole display width.
5349% It uses the same column types that we've already created, so there should
5350% be no problems.
5351%
5352% \begin{environment}{eqnalign}
5353%
5354% First, sort out some simple things like optional arguments.
5355%
5356% \begin{macrocode}
7abe48b3 5357\def\eqnalign{\@testopt\eqnalign@i{rcl}}
5358\def\eqnalign@i[#1]{\@testopt{\eqnalign@ii{#1}}c}
86f6a31e 5359% \end{macrocode}
5360%
5361% Now we actually do the environment. This is fairly easy, actually.
5362%
5363% \begin{macrocode}
5364\def\eqnalign@ii#1[#2]{%
5365 \let\\\eqn@cr%
5366 \colset{eqnarray}%
5367 \tab@initread%
5368 \def\tab@tabtext{&\tabskip\z@skip}%
5369 \tabskip\z@skip%
5370 \col@sep.5\eqainskip%
5371 \tab@readpreamble{#1}%
5372 \everycr{}%
5373 \if#2t\vtop\else%
5374 \if#2b\vbox\else%
5375 \vcenter%
5376 \fi%
5377 \fi%
5378 \bgroup%
5379 \halign\expandafter\bgroup\the\tab@preamble\cr%
5380}
5381% \end{macrocode}
5382%
5383% Finishing the environment is even simpler.
5384%
5385% \begin{macrocode}
5386\def\endeqnalign{%
5387 \crcr%
5388 \egroup%
5389 \egroup%
5390}
5391% \end{macrocode}
5392%
5393% \end{environment}
5394%
5395% \begin{macro}{\eqn@cr}
5396%
5397% Newlines are really easy here.
5398%
5399% \begin{macrocode}
5400\def\eqn@cr{\tab@cr\eqn@cr@i{}{}}
5401\def\eqn@cr@i#1{\cr\noalign{\vskip\jot\vskip#1}\@gobble}
5402% \end{macrocode}
5403%
5404% \end{macro}
5405%
5406%
5407% \subsection{Simple multiline equations}
5408%
5409% As a sort of example and abbreviation, here's a multiline display
5410% environment which just centres everything.
5411%
5412% \begin{environment}{eqlines}
5413%
5414% We just get |\eqnarray| to do everything for us. This is really easy.
5415%
5416% \begin{macrocode}
5417\def\eqlines{\eqnarray[x]}
5418\let\endeqlines\endeqnarray
5419% \end{macrocode}
5420%
5421% \end{environment}
5422%
5423% \begin{environment}{eqlines*}
5424%
5425% There's a $*$ version which omits numbers. This is easy too. Lots of
5426% hacking with expansion here to try and reduce the number of tokens being
5427% used. Is it worth it?
5428%
5429% \begin{macrocode}
5430\expandafter\edef\csname eqlines*\endcsname{%
5431 \expandafter\noexpand\csname eqnarray*\endcsname[x]%
5432}
5433\expandafter\let\csname endeqlines*\expandafter\endcsname
5434 \csname endeqnarray*\endcsname
5435% \end{macrocode}
5436%
5437% \end{environment}
5438%
5439%
5440% \subsection{Split equations}
5441%
5442% Based on an idea from \textit{The \TeX book}, we provide some simple
5443% environments for doing split equations. There's plenty of scope for
5444% improvement here, though.
5445%
5446% \begin{environment}{spliteqn}
5447% \begin{environment}{spliteqn*}
5448%
5449% The only difference between these two is that the $*$-version doesn't put
5450% in an equation number by default (although this behaviour can be
5451% changed by |\eqnumber|).
5452%
5453% The fun here mainly concerns putting in the equation number at the right
5454% place -- for |leqno| users, we need to put the number on the first line;
5455% otherwise we put it on the last line.
5456%
5457% The way we handle this is to have two macros, |\\| (which clearly does
5458% all the user line breaks) and |\seq@lastcr| which is used at the end of
5459% the environment to wrap everything up. The |\seq@eqnocr| macro puts an
5460% equation number on the current line and then does a normal |\\|. It also
5461% resets |\\| and |\seq@lastcr| so that they don't try to put another
5462% equation number in. This must be done globally, although anyone who tries
5463% to nest maths displays will get what they deserve.
5464%
5465% For the non-$*$ environment, then, we need to step the equation counter,
5466% and set |\\| to |\seq@cr| or |\seq@eqnocr| as appropriate for the setting
5467% of the |leqno| flag -- |\seq@lastcr| always gets set to put an equation
5468% number in (because it will be reset if the number actually gets done
5469% earlier -- this catches stupid users trying to put a single row into
5470% a split environment).
5471%
5472% \begin{macrocode}
5473\def\spliteqn{%
5474 \let\eqa@oldnumber\eqa@number%
5475 \global\let\eqa@number\eqa@eqcount%
5476 \spliteqn@i%
5477}
5478% \end{macrocode}
5479%
5480% For the $*$ variant, we don't need to bother with equation numbering, so
5481% this is really easy.
5482%
5483% \begin{macrocode}
5484\@namedef{spliteqn*}{%
5485 \let\eqa@oldnumber\eqa@number%
5486 \gdef\eqa@number{}%
5487 \spliteqn@i%
5488}
5489% \end{macrocode}
5490%
5491% Ending the environments is easy. Most of the stuff here will be described
5492% later.
5493%
5494% \begin{macrocode}
5495\def\endspliteqn{%
5496 \hfilneg\seq@lastcr%
5497 \egroup%
5498 \dsp@end%
5499 \global\let\eqa@number\eqa@oldnumber%
5500 \global\advance\c@equation\m@ne%
b57a27e1 5501 \global\@ignoretrue%
86f6a31e 5502}
5503\expandafter\let\csname endspliteqn*\endcsname\endspliteqn
5504% \end{macrocode}
5505%
5506% \end{environment}
5507% \end{environment}
5508%
5509% \begin{macro}{\spliteqn@i}
5510%
5511% Here we handle the full display splits. Start a maths display, and make
5512% each row of the alignment take up the full display width.
5513%
5514% The macro |\seq@dosplit| does most of the real work for us -- setting up
5515% the alignment and so forth. The template column is interesting. There
5516% are two items glue on both sides of the actual text:
5517%
5518% \begin{itemize}
5519%
5520% \item Some glue which can shrink. This keeps the display from the edges
5521% of the page unless we get a really wide item.
5522%
5523% \item An |\hfil| to do the alignment. By default, this centres the
5524% equations. On the first line, however, we put a leading |\hfilneg|
5525% which cancels the first |\hfil|, making the first row left aligned.
5526% Similarly, at the end, we put an |\hfilneg| after the last equation
5527% to right align the last line.
5528%
5529% \end{itemize}
5530%
5531% We pass this information on as an argument. It's easy really.
5532%
5533% \begin{macrocode}
5534\def\spliteqn@i{%
5535% \end{macrocode}
5536%
5537% First, set up equation numbering properly. See my rant about
5538% |\refstepcounter| above.
5539%
5540% \begin{macrocode}
5541 \stepcounter{equation}%
5542 \def\@currentlabel{\p@equation\theequation}%
5543% \end{macrocode}
5544%
5545% Right; now to sort out the numbering and newline handling. If the number's
5546% meant to be on the first line (for \textsf{leqno} users), then it gets
5547% typeset on the first like; otherwise we just do a normal newline on
5548% all lines except the first. Once |\seq@eqnocr| has done its stuff, it
5549% redefines all the newline handling not to insert another number.
5550%
5551% \begin{macrocode}
5552 \if@leqno%
5553 \global\let\seq@docr\seq@eqnocr%
5554 \else%
5555 \global\let\seq@docr\seq@cr%
5556 \fi%
5557 \global\let\seq@lastcr\seq@eqnocr%
5558% \end{macrocode}
5559%
5560% For my next trick, I'll do some display handling -- start a (possibly
5561% nested) maths display, set up the |\tabpause| macro appropriately, and
5562% set the newline command to do the right thing.
5563%
5564% \begin{macrocode}
5565 \dsp@start%
5566 \dsp@tabpause%
5567 \def\\{\seq@docr}%
5568% \end{macrocode}
5569%
5570% Finally, call another macro to do the remaining bits of setting up.
5571%
5572% \begin{macrocode}
5573 \seq@dosplit%
5574 {\hb@xt@\displaywidth{%
5575 \hskip\splitleft\hfil$\displaystyle##$%
5576 \hfil\hskip\splitright}}%
5577 {\hfilneg}%
5578}
5579% \end{macrocode}
5580%
5581% \end{macro}
5582%
5583% \begin{environment}{subsplit}
5584%
5585% For doing splits in the middle of equations, we provide a similar
5586% environment. Here, we make |\\| just start a new line. We also use
5587% a |\vcenter| rather than a full maths display. The glue items are also
5588% a bit different: we use plain double-quads on each side of the item, and
5589% we need to remove them by hand at the extremities of the environment.
5590%
5591% \begin{macrocode}
c728af0b
MW
5592\def\subsplit{\@ifnextchar[\subsplit@i{\subsplit@i[c]}}
5593\def\subsplit@i[#1]{%
5594 \let\@tempa\vcenter%
5595 \if#1t\let\@tempa\vtop\fi%
5596 \if#1b\let\@tempa\vbox\fi%
86f6a31e 5597 \let\\\seq@cr%
c728af0b 5598 \@tempa\bgroup%
86f6a31e 5599 \seq@dosplit{\hfil\qquad$##$\qquad\hfil}{\hfilneg\hskip-2em}%
5600}
5601% \end{macrocode}
5602%
5603% Ending the environment is fairly easy. We remove the final glue item,
5604% and close the alignment and the vbox.
5605%
5606% \begin{macrocode}
5607\def\endsubsplit{%
5608 \hfilneg\hskip-2em\cr%
5609 \egroup\egroup%
5610}
5611% \end{macrocode}
5612%
5613% \end{environment}
5614%
5615% \begin{macro}{\seq@dosplit}
5616%
5617% Here we do most of the real work. Actually, since the preamble is passed
5618% in as an argument, most of the work is already done. The only thing to
5619% really note is the template for subsequent columns. To stop users putting
5620% in extra columns (which is where we put the equation number) we raise an
5621% error and discard the input in a scratch box register. This template is
5622% repeated infinitely so as to allow us to put the equation number in nicely.
5623% However, the final negative glue item won't work properly, so the equation
5624% will look awful.
5625%
5626% \begin{macrocode}
5627\def\seq@dosplit#1#2{%
5628 \halign\bgroup%
5629 #1&&\mth@err@number\setbox\z@\hbox{##}\cr%
5630 #2\relax%
5631}
5632% \end{macrocode}
5633%
5634% \end{macro}
5635%
5636% \begin{macro}{\seq@eqnocr}
5637%
5638% Here's how we set equation numbers. Since the column provided raises
5639% errors as soon as a token finds its way into it, we start with a |&\omit|.
5640% Then we just put the equation number in a zero-width box. Finally, we
5641% reset the newline commands to avoid putting in more than one equation
5642% number, and do normal newline things.
5643%
5644% \begin{macrocode}
5645\def\seq@eqnocr{%
5646 &\omit%
5647 \hb@xt@\z@{\hss\eqa@eqpos\eqa@number}%
5648 \global\let\seq@docr\seq@cr%
5649 \global\let\seq@lastcr\seq@cr%
5650 \seq@cr%
5651}
5652% \end{macrocode}
5653%
5654% \end{macro}
5655%
5656% \begin{macro}{\seq@cr}
5657%
5658% Newlines are very easy. We add a |\jot| of extra space, since this is
5659% a nice thing to do.
5660%
5661% \begin{macrocode}
5662\def\seq@cr{\tab@cr\seq@cr@i\interdisplaylinepenalty\@M}
5663\def\seq@cr@i#1#2{\cr\noalign{\penalty#2\vskip\jot\vskip#1}}
5664% \end{macrocode}
5665%
5666% \end{macro}
5667%
5668%
5669% \subsection{Matrix handling}
5670%
5671% There's been a complete and total overhaul of the spacing calculations
5672% for matrices here. The vertical spacing now bears an uncanny similarity
5673% to the rules \TeX\ uses to space out |\atop|-like fractions, the difference
5674% being that you can have more than one column in a matrix. This has the
5675% interesting side-effect that we get an \package{amsmath}-style
5676% sub/superscript environment almost free of charge with the matrix handling
5677% (it just ends up being a script-size single-column matrix).
5678%
5679% What is rather gratifying is that our \env{matrix} environment looks
5680% rather nicer than \package{amsmath}'s (which is based directly on
5681% \env{array}, giving it nasty restrictions on the numbers of columns and
5682% so on); in particular, the version here gives the `correct' result for
5683% Knuth's exercise~18.42 (which states categorically that a |\smallskip|
5684% should be placed between the rows of the big matrix).
5685%
5686% The reason the interrow space doesn't come out in the AMS version is
5687% that \env{array} inserts extra vertical space by extending the depth of
5688% the final row using a strut: the big matrix already extends deeper than
5689% this, so the strut doesn't make any difference. If the space was added
5690% by |\hlx{s[\smallskipamount]}| instead of the |\\| command, things would
5691% be different.
5692%
5693% \begin{figure}
5694%
5695% ^^A This is essentially what amsmath (version 1.2b) does. The real
5696% ^^A implementation requires a counter MaxMatrixCols, and has fewer braces:
5697% ^^A that's all the difference. Oh, and I turn off \arrayextrasep here,
5698% ^^A since amsmath doesn't expect it to be there (accurate emulation, see?)
5699% ^^A and I've used \hspace instead of \hskip since everything else is
5700% ^^A `proper' LaTeX stuff.
5701%
5702% \newenvironment{ams-pmatrix}{^^A
5703% \setlength{\arrayextrasep}{0pt}^^A
5704% \left(^^A
5705% \hspace{-\arraycolsep}^^A
5706% \begin{array}{*{10}{c}}^^A
5707% }{^^A
5708% \end{array}^^A
5709% \hspace{-\arraycolsep}^^A
5710% \right)^^A
5711% }
5712%
5713% \begin{demo}{Exercise 18.42 from \emph{The \TeX book}}
5714%\newcommand{\domatrix}[1]{
5715% \def\mat##1
5716% {\begin{#1}##1\end{#1}}
5717% \[ \begin{#1}
5718% \mat{a & b \\ c & d} &
5719% \mat{e & f \\ g & h}
5720% \\[\smallskipamount]
5721% 0 &
5722% \mat{i & j \\ k & l}
5723% \end{#1}
5724% \]
5725%}
5726%\domatrix{pmatrix}
5727%\domatrix{ams-pmatrix}
5728% \end{demo}
5729%
5730% \end{figure}
5731%
5732% \begin{environment}{genmatrix}
5733%
5734% The first job is to store my maths style and font away, because I'll be
5735% needing it lots later.
5736%
5737% \begin{macrocode}
5738\def\genmatrix#1#2#3#4#5{%
5739 \let\mat@style#1%
5740 \ifx#2\scriptstyle%
5741 \let\mat@font\scriptfont%
5742 \else\ifx#2\scriptscriptstyle%
5743 \let\mat@font\scriptscriptfont%
5744 \else%
5745 \let\mat@font\textfont%
5746 \fi\fi%
5747% \end{macrocode}
5748%
5749% Now to cope with inserted text. This is easy.
5750%
5751% \begin{macrocode}
5752 \ifx\mat@style\scriptstyle%
5753 \let\mat@textsize\scriptsize%
5754 \else\ifx\mat@style\scriptscriptstyle%
5755 \let\mat@textsize\scriptscriptsize%
5756 \else%
5757 \let\mat@textsize\relax%
5758 \fi\fi%
5759% \end{macrocode}
5760%
5761% Now for some fun. I'll remember how to start and end the matrix in a
5762% couple of macros |\mat@left| and |\mat@right|. I haven't yet worked out
5763% exactly what needs to be in |\mat@right| yet, though, so I'll build that
5764% up in a scratch token list while I'm making my mind up.
5765%
5766% Initially, I want to open a group (to trap the style changes), set the
5767% maths style (to get the right spacing), insert the left delimiter, insert
5768% some spacing around the matrix, and start a centred box. The ending just
5769% closes all the groups and delimiters I opened.
5770%
5771% \begin{macrocode}
5772 \def\mat@left{\bgroup\mat@style\left#4#3\vcenter\bgroup}%
5773 \toks@{\egroup#3\right#5\egroup}%
5774% \end{macrocode}
5775%
5776% Now comes a slightly trickier bit. If the maths style is script or
5777% scriptscript, then I need to raise the box by a little bit to make it look
5778% really good. The right amount is somewhere around \smallf 3/4\,pt, I
5779% think, so that's what I'll use.
5780%
5781% \begin{macrocode}
5782 \@tempswatrue%
5783 \ifx\mat@style\displaystyle\else\ifx\mat@style\textstyle\else%
5784 \@tempswafalse%
5785 \setbox\z@\hbox\bgroup$%
5786 \toks@\expandafter{\the\toks@$\m@th\egroup\raise.75\p@\box\z@}%
5787 \fi\fi%
5788% \end{macrocode}
5789%
5790% If I'm not in maths mode right now, then I should enter maths mode, and
5791% remember to leave it later.
5792%
5793% \begin{macrocode}
5794 \if@tempswa\ifmmode\else%
5795 $\m@th%
5796 \toks@\expandafter{\the\toks@$}%
5797 \fi\fi%
5798% \end{macrocode}
5799%
5800% Now I've sorted out how to end the environment properly, so I can set up
5801% the macro, using |\edef|.
5802%
5803% \begin{macrocode}
5804 \edef\mat@right{\the\toks@}%
5805% \end{macrocode}
5806%
5807% Now see if there's an optional argument. If not, create lots of centred
5808% columns.
5809%
5810% \begin{macrocode}
7abe48b3 5811 \@testopt\genmatrix@i{[c}%
86f6a31e 5812}
5813% \end{macrocode}
5814%
5815% Now to sort out everything else.
5816%
5817% \begin{macrocode}
5818\def\genmatrix@i[#1]{%
5819% \end{macrocode}
5820%
5821% Some initial setting up: choose the correct column set, and set up some
5822% variables for reading the preamble.
5823%
5824% \begin{macrocode}
5825 \colset{matrix}%
5826 \tab@initread%
5827% \end{macrocode}
5828%
5829% Now comes some of the tricky stuff. The space between columns should be
5830% 12\,mu (by trial and error). We put the space in a box so we can measure
5831% it in the correct mathstyle.
5832%
5833% \begin{macrocode}
5834 \setbox\z@\hbox{$\mat@style\mskip12mu$}%
5835 \edef\tab@tabtext{&\kern\the\wd\z@}%
5836 \tab@readpreamble{#1}%
5837% \end{macrocode}
5838%
5839% Now we need to decide how to space out the rows. The code here is based
5840% on the information in appendix~G of \emph{The \TeX book}: I think it'd be
5841% nice if my matrices were spaced out in the same way as normal fractions
5842% (particularly |\choose|y things). The standard |\baselineskip| and
5843% |\lineskip| parameters come in really handy here.
5844%
5845% The parameters vary according to the size of the text, so I need to see
5846% if we have scriptsize or less, or not. The tricky |\if| sorts this out.
5847%
5848% \begin{macrocode}
5849 \if1\ifx\mat@style\scriptstyle1\else%
5850 \ifx\mat@style\scriptscriptstyle1\else0\fi\fi%
5851 \baselineskip\fontdimen10\mat@font\tw@%
5852 \advance\baselineskip\fontdimen12\mat@font\tw@%
5853 \lineskip\thr@@\fontdimen8\mat@font\thr@@%
5854 \else%
5855 \baselineskip\fontdimen8\mat@font\tw@%
5856 \advance\baselineskip\fontdimen11\mat@font\tw@%
5857 \lineskip7\fontdimen8\mat@font\thr@@%
5858 \fi%
5859 \lineskiplimit\lineskip%
5860% \end{macrocode}
5861%
5862% Now actually set up for the alignment. Assign |\\| to the correct value.
5863% Set up the |\tabskip|. Do the appropriate |\mat@left| thing set up above.
5864% And then start the alignment.
5865%
5866% \begin{macrocode}
5867 \let\\\mat@cr%
5868 \tabskip\z@skip%
5869 \col@sep\z@%
5870 \mat@left%
5871 \halign\expandafter\bgroup\the\tab@preamble\tabskip\z@skip\cr%
5872% \end{macrocode}
5873%
5874% Now for a little hack to make the spacing consistent between matrices of
5875% the same height. This comes directly from \PlainTeX. This appears to
5876% make the spacing \emph{exactly} the same as the \TeX\ primitives, oddly
5877% enough.
5878%
5879% \begin{macrocode}
5880 \ifx\mat@font\textfont%
5881 \omit$\mat@style\mathstrut$\cr\noalign{\kern-\baselineskip}%
5882 \fi%
5883}
5884% \end{macrocode}
5885%
5886% Finishing the environment is really easy. We do the spacing hack again
5887% at the bottom, close the alignment and then tidy whatever we started in
5888% |\mat@left|.
5889%
5890% \begin{macrocode}
5891\def\endgenmatrix{%
5892 \crcr%
5893 \ifx\mat@font\textfont%
5894 \omit$\mat@style\mathstrut$\cr\noalign{\kern-\baselineskip}%
5895 \fi%
5896 \egroup%
5897 \mat@right%
5898}
5899% \end{macrocode}
5900%
5901% \end{environment}
5902%
5903% \begin{macro}{\mat@cr}
5904%
5905% Newlines are really easy. The $*$-form means nothing here, so we ignore
5906% it.
5907%
5908% \begin{macrocode}
5909\def\mat@cr{\tab@cr\mat@cr@i{}{}}
5910\def\mat@cr@i#1{\cr\noalign{\vskip#1}\@gobble}
5911% \end{macrocode}
5912%
5913% \end{macro}
5914%
5915% \begin{macro}{\newmatrix}
5916%
5917% This is how we define new matrix environments. It's simple fun with
5918% |\csname| and |\expandafter|.
5919%
5920% \begin{macrocode}
5921\def\newmatrix#1#2{%
5922 \@namedef{#1}{\genmatrix#2}%
5923 \expandafter\let\csname end#1\endcsname\endgenmatrix%
5924}
5925% \end{macrocode}
5926%
5927% \end{macro}
5928%
5929% \begin{environment}{matrix}
5930% \begin{environment}{pmatrix}
5931% \begin{environment}{dmatrix}
5932% \begin{environment}{smatrix}
5933% \begin{environment}{spmatrix}
5934% \begin{environment}{sdmatrix}
5935% \begin{environment}{smatrix*}
5936% \begin{environment}{spmatrix*}
5937% \begin{environment}{sdmatrix*}
5938%
5939% Now we define all the other environments we promised. This is easy.
5940%
5941% \begin{macrocode}
5942\newmatrix{matrix}{{\textstyle}{\textstyle}{\,}{.}{.}}
5943\newmatrix{pmatrix}{{\textstyle}{\textstyle}{\,}{(}{)}}
5944\newmatrix{dmatrix}{{\textstyle}{\textstyle}{\,}}
5945\newmatrix{smatrix}{{\scriptstyle}{\scriptstyle}{}{.}{.}}
5946\newmatrix{spmatrix}{{\scriptstyle}{\scriptstyle}{}{(}{)}}
5947\newmatrix{sdmatrix}{{\scriptstyle}{\scriptstyle}{}}
5948\newmatrix{smatrix*}{{\scriptstyle}{\textstyle}{}{.}{.}}
5949\newmatrix{spmatrix*}{{\scriptstyle}{\textstyle}{}{(}{)}}
5950\newmatrix{sdmatrix*}{{\scriptstyle}{\textstyle}{}}
5951% \end{macrocode}
5952%
5953% \end{environment}
5954% \end{environment}
5955% \end{environment}
5956% \end{environment}
5957% \end{environment}
5958% \end{environment}
5959% \end{environment}
5960% \end{environment}
5961% \end{environment}
5962%
5963% \begin{environment}{script}
5964%
5965% Now for superscripts and subscripts. This is fairly easy, because I
5966% took so much care over the matrix handling.
5967%
5968% \begin{macrocode}
5969\def\script{%
5970 \let\mat@style\scriptstyle%
5971 \def\mat@left{\vcenter\bgroup}%
5972 \def\mat@right{\egroup}%
5973 \let\mat@font\scriptfont%
5974 \let\mat@textsize\scriptsize%
7abe48b3 5975 \@testopt\genmatrix@i c%
86f6a31e 5976}
5977\let\endscript\endgenmatrix
5978% \end{macrocode}
5979%
5980% \end{environment}
5981%
5982% Now define the column types.
5983%
5984% \begin{macrocode}
5985\colpush{matrix}
5986\coldef l{\tabcoltype{\kern\z@$\mat@style}{\m@th$\hfil}}
5987\coldef c{\tabcoltype{\hfil$\mat@style}{\m@th$\hfil}}
5988\coldef r{\tabcoltype{\hfil$\mat@style}{\m@th$}}
5989\coldef T#1{\tab@aligncol{#1}{\begingroup\mat@textsize}{\endgroup}}
5990\collet > [tabular]>
5991\collet < [tabular]<
5992\collet * [tabular]*
5993\collet @ [tabular]@
5994% \end{macrocode}
5995%
5996% The repeating type is more awkward. Things will go wrong if this is
5997% given before the first column, so we must do a whole repeat by hand. We
5998% can tell if we haven't contributed a column yet, since |\tab@column| will
5999% be zero. Otherwise, we fiddle the parser state to start a new column, and
6000% insert the |&| character to make \TeX\ repeat the preamble.
6001%
6002% \begin{macrocode}
6003\coldef {[}{%
6004 \@firstoftwo{%
6005 \ifnum\tab@columns=\z@%
6006 \def\@tempa##1\q@delim{%
6007 \tab@mkpreamble##1[##1\q@delim%
6008 }%
6009 \expandafter\@tempa%
6010 \else%
6011 \tab@setstate\tab@prestate%
6012 \tab@append\tab@preamble{&}%
6013 \expandafter\tab@mkpreamble%
6014 \fi%
6015 }%
6016}
6017% \end{macrocode}
6018%
6019% We're done defining columns now.
6020%
6021% \begin{macrocode}
6022\colpop
6023% \end{macrocode}
6024%
6025%
6026% \subsection{Dots\dots}
6027%
6028% Nothing whatsoever to do with alignments, although vertical and diagonal
6029% dots in small matrices look really silly. The following hacky definitions
6030% work rather better.
6031%
6032% \begin{macro}{\mdw@dots}
6033%
6034% First of all, here's some definitions common to both of the dots macros.
6035% The macro takes as an argument the actual code to draw the dots, passing
6036% it the scaled size of a point in the scratch register |\dimen@|; the
6037% register |\box 0| is set to contain a dot of the appropriate size.
6038%
6039% \begin{macrocode}
6040\def\mdw@dots#1{\ensuremath{\mathpalette\mdw@dots@i{#1}}}
6041\def\mdw@dots@i#1#2{%
6042 \setbox\z@\hbox{$#1\mskip1.8mu$}%
6043 \dimen@\wd\z@%
6044 \setbox\z@\hbox{$#1.$}%
6045 #2%
6046}
6047% \end{macrocode}
6048%
6049% \end{macro}
6050%
6051% \begin{macro}{\vdots}
6052%
6053% I'll start with the easy one. This is a simple translation of the original
6054% implementation.
6055%
6056% \begin{macrocode}
6057\def\vdots{%
6058 \mdw@dots{\vbox{%
6059 \baselineskip4\dimen@%
6060 \lineskiplimit\z@%
6061 \kern6\dimen@%
6062 \copy\z@\copy\z@\box\z@%
6063 }}%
6064}
6065% \end{macrocode}
6066%
6067% \end{macro}
6068%
6069% \begin{macro}{\ddots}
6070%
6071% And I'll end with the other easy one\dots
6072%
6073% \begin{macrocode}
6074\def\ddots{%
6075 \mdw@dots{\mathinner{%
6076 \mkern1mu%
6077 \raise7\dimen@\vbox{\kern7\dimen@\copy\z@}%
6078 \mkern2mu%
6079 \raise4\dimen@\copy\z@%
6080 \mkern2mu%
6081 \raise\dimen@\box\z@%
6082 \mkern1mu%
6083 }}%
6084}
6085% \end{macrocode}
6086%
6087% \end{macro}
6088%
6089%
6090% \subsection{Lucky dip}
6091%
6092% Time to round off with some trivial environments, just to show how easy
6093% this stuff is.
6094%
6095% \begin{environment}{cases}
6096% \begin{environment}{smcases}
6097%
6098% These are totally and utterly trivial.
6099%
6100% \begin{macrocode}
6101\def\cases{\left\{\,\array{@{}lTl@{}}}
6102\def\endcases{\endarray\,\right.}
6103\def\smcases{\left\{\smarray{@{}lTl@{}}}
6104\def\endsmcases{\endsmarray\,\right.}
6105% \end{macrocode}
6106%
6107% \end{environment}
6108% \end{environment}
6109%
6110% \subsection{Error messages}
6111%
6112% Some token saving:
6113%
6114% \begin{macrocode}
6115\def\mth@error{\PackageError{mathenv}}
6116% \end{macrocode}
6117%
6118% Now for the error messages.
6119%
6120% \begin{macrocode}
6121\def\mth@err@number{%
6122 \mth@error{Too many `&' characters found}{%
6123 You've put too many `&' characters in an alignment^^J%
6124 environment (like `eqnarray' or `spliteqn') and wandered^^J%
6125 into trouble. I've gobbled the contents of that column^^J%
6126 and hopefully I can recover fairly easily.%
6127 }%
6128}
6129% \end{macrocode}
6130%
6131% \begin{macrocode}
6132\def\mth@err@mdsp{%
6133 \mth@error{Can't do displays in nondisplay maths mode}{%
6134 You're trying to start a display environment, but you're^^J%
6135 in nondisplay maths mode. The display will appear but^^J%
6136 don't blame me when it looks horrible.%
6137 }%
6138}
6139% \end{macrocode}
6140%
6141% \begin{macrocode}
6142\def\mth@err@hdsp{%
6143 \mth@error{Can't do displays in LR mode}{%
6144 You're trying to start a display environment, but you're^^J%
6145 in LR (restricted horizontal) mode. Everything will go^^J%
6146 totally wrong, so your best bet is to type `X', fix the^^J%
6147 mistake and start again.%
6148 }%
6149}
6150% \end{macrocode}
6151%
6152% \vskip\parskip\vbox{ ^^A The best way I could find of keeping this lot
6153% ^^A together, I'm afraid.
6154% That's all there is. Byebye.
6155%
6156% \begin{macrocode}
6157%</mathenv>
6158% \end{macrocode}
6159% \nopagebreak
6160%
6161% \hfill Mark Wooding, \today
6162% }
6163%
6164% \Finale
6165%
6166\endinput