% $Id: genops.tex,v 1.6 2020/01/01 23:12:25 karl Exp $
% This is part of the book TeX for the Impatient.
% Copyright (C) 2003-2020 Paul W. Abrahams, Kathryn A. Hargreaves, Karl Berry.
% See file fdl.tex for copying conditions.
\input macros
\chapter {Commands for \linebreak general operations}
\chapterdef{general}
This section covers
\TeX's ^{programming features} and
everything else that doesn't fit into the categories
of commands in the previous chapters.
For an explanation of the conventions used in this section,
see \headcit{Descriptions of the commands}{cmddesc}.
\begindescriptions
%==========================================================================
\section {Naming and modifying fonts}
\begindesc
\bix^^{fonts//naming and modifying}
\cts font {}
\aux\cts font {\<control sequence> = \<fontname>}
\aux\cts font {\<control sequence> = \<fontname> {\bt scaled} \<number>}
\aux\cts font {\<control sequence> = \<fontname> {\bt at} \<dimen>}
\explain
Used alone, the |\font| control sequence designates the current font.
|\font| isn't a true command when it's used alone,
since it then can appear only as an argument to another command.
For the other three forms of |\font|,
\<font\-name> names a set of files that define a font.
These forms of |\font| are commands. Each of these forms has two effects:
{\tighten
\olist
\li It defines \<control sequence> as a name that selects
the font \<font\-name>, possibly magnified (see below).
\li It causes \TeX\ to load the font ^{metrics file}
(^{\tfmfile}) for \<fontname>.
\endolist
}% end tighten
\noindent
The name of a font file usually indicates its design size.
For example, |cmr10| indicates Computer Modern roman with a
design size of $10$ points.
The design size of a font is recorded in its metrics file.
If neither |scaled| \<number> nor |at| \<dimen>
is present, the font is used
at its design size---the size at which it usually looks best.
Otherwise, a magnified version of the font is loaded:
\ulist
\li If |scaled| \<number>
is present, the font is magnified by a factor of $\hbox{\<number>}/1000$.
\li If |at| \<dimen> is present, the font is scaled to \<dimen> by magnifying
it by $\hbox{\<dimen>}/ds$, where $ds$ is the design size of
\<fontname>.
\<dimen> and $ds$ are nearly always given in points.
\endulist
\noindent
Magnifications of less than $1$ are possible; they reduce the size.
You usually need to provide a shape file (\xref{shape}) for each
magnification of a font that you load.
However, some ^{device drivers} can utilize fonts that are resident
in a printer. ^^{resident fonts}
Such fonts don't need shape files.
See \conceptcit{font} and
\conceptcit{magnification} for further information.
\example
\font\tentt = cmtt10
\font\bigttfont = cmtt10 scaled \magstep2
\font\eleventtfont = cmtt10 at 11pt
First we use {\tentt regular CM typewriter}.
Then we use {\eleventtfont eleven-point CM typewriter}.
Finally we use {\bigttfont big CM typewriter}.
|
\produces
\font\regttfont = cmtt10
\font\bigttfont = cmtt10 scaled \magstep 2
\font\eleventtfont = cmtt10 at 11pt
First we use {\regttfont regular CM typewriter}.
Then we use {\eleventtfont eleven-point CM typewriter}.
Finally we use {\bigttfont big CM typewriter}.
\endexample
\enddesc
\begindesc
\cts fontdimen {\<number> \<font>\param{dimen}}
\explain
^^{fonts//parameters of}
These parameters specify various dimensions associated with
the font named by the control sequence \<font>
(as distinguished from the \<font\-name> that names the font files).
Values of these parameters are specified in the metrics
file for \<font>, but you can
retrieve or change their values during a \TeX\ run.
The numbers and meanings of the parameters are:
\display{\halign{\hfil#\hfil\quad&#\hfil\cr
\it Number&\it Meaning\cr
\noalign{\vskip 1\jot}%
1&slant per point\cr
2&interword space\cr
3&interword stretch\cr
4&interword shrink\cr
5&x-height (size of |1ex|)\cr
6&quad width (size of |1em|)\cr
7&extra space\cr}}
\noindent
\TeX\ uses the slant per point for positioning accents.
It uses the interword parameters for producing interword spaces
(see |\spaceskip|, \xref\spaceskip) and the extra space parameter
for the additional space after a period (see |\xspaceskip|,
\xref\xspaceskip).
The values of these parameters for the
\plainTeX\ fonts are enumerated on \knuth{page~433}.
Math symbol fonts have $15$ additional parameters, which we won't discuss here.
Beware:
assignments to these parameters are \emph{not} undone at the end
of a group.
If you want to change these parameters locally, you'll need to
save and restore their original settings explicitly.
\example
Here's a line printed normally.\par
\dimen0=\fontdimen2\font
\fontdimen2\font=3\fontdimen2\font % triple word spacing
\noindent Here's a really spaced-out line.
\fontdimen2\font=\dimen0
|
\produces
Here's a line printed normally.\par
\dimen0=\fontdimen2\font
\fontdimen2\font=3\fontdimen2\font % triple word spacing
\noindent Here's a really spaced-out line.
\fontdimen2\font=\dimen0
\endexample
\enddesc
\begindesc
\cts magnification {{\bt =} \<number>}
\cts mag {\param{number}}
\explain
\margin{{\tt\\mag} and {\tt\\magnification} combined.}
An assignment to |\magnification| establishes
the ``^{scale factor}'' $f$ that determines
the \minref{magnification} ratio of your document \seeconcept{magnification}.
The assignment to |\magni!-fication| must occur before the first page
of your document has been shipped out.
The assignment sets $f$ to \<number> and also
sets |\hsize| and |\vsize|
^^|\hsize//set by {\tt\\magnification}|
^^|\vsize//set by {\tt\\magnification}|
respectively to |6.5true in| and |8.9true in|,
the values appropriate for an $8 \frac1/2$-%
by-$11$-inch page.
$f$ must be between $0$ and $32768$.
The \minref{magnification} ratio of the
document is $f/1000$.
A scale factor
of $1000$ provides unit magnification, i.e., it leaves the size of your
document
unchanged. It's customary to use powers of $1.2$ as scale factors, and
most libraries of fonts are based on such factors. You can use the
^|\magstep| and ^|\magstephalf| commands to specify magnifications by
these factors.
|\magnification| is not a parameter. You can't use it
to \emph{retrieve} the scale factor. If you write something like
|\dimen0 = \mag!-nifi!-cation|, \TeX\ will complain about it.
The |\mag| parameter contains the scale factor.
Changing the value of |\mag| rescales the page dimensions, which is not
usually what you want.
Therefore it's usually better to change the magnification by
assigning to |\magnification| rather than to |\mag|.
\example
\magnification = \magstep2
% magnify fonts by 1.44 (=1.2x1.2)
|
\endexample
\enddesc
\begindesc
\cts magstep {\<number>}
\explain
This command expands to the \minref{magnification} ratio needed to
magnify everything in your document
(other than |true| dimensions)
by $1.2^r$, where $r$ is
the value of \<number>. \<number> must be between $0$ and $5$.
\example
\magnification = \magstep1 % Magnify by ratio of 1.2.
|
\endexample
\enddesc
\begindesc
\cts magstephalf {}
\explain
This command expands to the \minref{magnification} ratio needed to
magnify everything in your document
(other than |true| dimensions)
by $\sqrt{1.2}$,
i.e., halfway between $1$ and $1.2$.
\example
\magnification = \magstephalf
|
\endexample
\eix^^{fonts//naming and modifying}
\enddesc
%==========================================================================
\section {Converting information to tokens}
\subsection {Numbers}
\begindesc
\xrdef{convert}
\bix^^{numbers//converting to characters}
%
\cts number {\<number>}
\explain
This command produces the representation of a \minref{number}
as a sequence of
character \minref{token}s. The number can be either an explicit integer,
a \<number> parameter, or a \<number> register.
\example
\number 24 \quad \count13 = -10000 \number\count13
|
\produces
\number 24 \quad \count13 = -10000 \number\count13
\endexample
\enddesc
\begindesc
^^{Roman numerals}
\easy\cts romannumeral {\<number>}
\explain
This command produces the roman numeral representation of a \minref{number}
as a sequence of
character \minref{token}s. The number can be either an explicit integer,
a \<number> parameter, or a \<number> register.
If the number is zero or negative, |\romannumeral| produces
no tokens.
\example
\romannumeral 24 \quad (\romannumeral -16)\quad
\count13 = 6000 \romannumeral\count13
|
\produces
\romannumeral 24 \quad (\romannumeral -16)\quad
\count13 = 6000 \romannumeral\count13
\endexample
\eix^^{numbers//converting to characters}
\enddesc
%==========================================================================
\subsection {Environmental information}
\begindesc
^^{time of day}
\cts time {\param{number}}
\explain
\TeX\ sets this parameter to the
number of minutes that have elapsed since midnight (of the current day).
At noon, for instance, |\time| is $720$.
This command and the next three make use of the time and date as
recorded in your computer.
\TeX\ retrieves them just once, at the beginning of your run, so |\time|
at the end of the run always has the same value as |\time| at the
beginning of the run (unless you've explicitly changed it).
\enddesc
\bix^^{date}
\begindesc
\cts day {\param{number}}
\explain
\TeX\ sets this parameter to the current day of the month. It is
a number between $1$ and $31$.
|\day| is set at the beginning of your run (see the comments on
|\time| above).
\enddesc
\begindesc
\cts month {\param{number}}
\explain
\TeX\ sets this parameter to the current month. It is
a number between $1$ and $12$.
|\month| is set at the beginning of your run (see the comments on
|\time| above).
\enddesc
\begindesc
\cts year {\param{number}}
\explain
\TeX\ sets this parameter to the current year ({\sc A.D.}).
It is a number such as $1991$.
|\year| is set at the beginning of your run (see the comments on
|\time| above).
\eix^^{date}
\enddesc
\begindesc
^^{version number}
\cts fmtname {}
\cts fmtversion {}
\explain
These commands produce the name and version number
of the \TeX\ format,
e.g., \minref{\plainTeX} or ^{\LaTeX}, that you're using.
The |\fmtversion| string contains a long list of supported languages,
so is omitted here.
\example
This book was produced with the \fmtname\ format.
|
\produces
This book was produced with the \fmtname\ format.
\endexample
\enddesc
\begindesc
\cts jobname {}
\explain
This command produces the base
name of the file with which \TeX\ was invoked.
For example, if your main input file is |hatter.tex|,
|\jobname|
{\parfillskip=0pt\par\eject\noindent}
will expand to |hatter|.
|\jobname| is most useful when you're
creating an auxiliary file to be associated with a document.
^^{auxiliary files}
\example
\newwrite\indexfile \openout\indexfile = \jobname.idx
% For input file `hatter.tex', open index file `hatter.idx'.
|
\endexample\enddesc
%==========================================================================
\subsection {Values of variables}
\begindesc
\cts meaning {\<token>}
\explain
^^{tokens//showing the meaning of}
This command produces
the meaning of \<token>. It is useful for diagnostic output.
You can use the ^|\the| command (\xref\the) in a similar way
to get information about the values of \minref{register}s and other
\TeX\ entities.
\example
[{\tt \meaning\eject}] [\meaning\tenrm] [\meaning Y]
|
\produces
[{\tt \meaning\eject}] [\meaning\tenrm] [\meaning Y]
\endexample\enddesc
\begindesc
\cts string {\<control sequence>}
\explain
^^{control sequences//converting to strings}
This command produces
the characters that form the name of \<control sequence>,
including the \minref{escape character}.
The escape character is represented by the current value of
^|\escapechar|.
^^{escape character//represented by \b\tt\\escapechar\e}
\TeX\ gives the characters in the list a category code of $12$ (other).
You can perform the reverse operation with
the ^|\csname| command (\xref \csname),
which turns a string into a control sequence.
\example
the control sequence {\tt \string\bigbreak}
|
\produces
the control sequence {\tt \string\bigbreak}
\endexample\enddesc
\begindesc
\cts escapechar {\param{number}}
\explain
This parameter specifies the \ascii\ code \minrefs{\ascii} of the
character that \TeX\ uses to represent the \minref{escape character}
^^{escape character//represented by \b\tt\\escapechar\e}
when it's
converting a control sequence name to a sequence of character tokens.
This conversion occurs when you use the |\string| command and also when
\TeX\ is producing diagnostic messages. The default value of the escape
character is $92$, the {\ascii} character code for a ^{backslash}.
If |\escapechar| is not in the range $0$--$255$,
\TeX\ does not include an escape character in the result of the conversion.
\example
\escapechar = `!!
the control sequence {\tt \string\bigbreak}
|
\produces
\escapechar = `!
the control sequence {\tt \string\bigbreak}
\endexample
\enddesc
\begindesc
\cts fontname {\<font>}
\explain
^^{fonts//names of}
This command produces the filename
for \<font>. The filename is the \<font\-name> that was used to define
\<font>.
\example
\font\myfive=cmr5 [\fontname\myfive]
|
\produces
\font\myfive=cmr5 [\fontname\myfive]
\endexample
\enddesc
%==========================================================================
\section {Grouping}
\begindesc
\bix^^{groups}
%
\cts begingroup {}
\cts endgroup {}
\explain
These two commands begin and end a \minref{group}.
A |\begingroup| does not match up with a right brace,
nor an |\endgroup| with a left brace.
\TeX\ treats |\begingroup| and |\endgroup| like any other
\minref{control sequence} when it's scanning its input. In particular,
you can define a \minref{macro} that contains a |\begingroup|
but not an |\endgroup|, and conversely.
^^{macros//using \b\tt\\begingroup\e\ and \b\tt\\endgroup\e\ in}
This technique is often useful
when you're defining paired macros, one of which establishes
an environment and the other of which terminates that environment.
You can't, however, use |\begingroup| and |\endgroup| as substitutes for
braces other than the ones that surround a group.
\example
\def\a{One \begingroup \it two }
\def\enda{\endgroup four}
\a three \enda
|
\produces
\def\a{One \begingroup \it two }
\def\enda{\endgroup four}
\a three \enda
\endexample
\enddesc
\begindesc
\makecolumns 4/2:
\easy%
\ctsact { \xrdef{@lbrace}
\cts bgroup {}
\ctsact } \xrdef{@rbrace}
\cts egroup {}
\explain
The left and right braces are commands that begin and end a
\minref{group}.
The |\bgroup| and |\egroup| \minref{control sequence}s are equivalent
to `|{|' and `|}|', except that
\TeX\ treats |\bgroup| and |\egroup| like any other
\minref{control sequence} when it's scanning its input.
|\bgroup| and |\egroup| can be useful when you're
defining paired macros, one of which
starts a brace-delimited construct (not necessarily a group)
and the other one of which ends that construct.
^^{macros//using \b\tt\\bgroup\e\ and \b\tt\\egroup\e\ in}
You can't define such macros using ordinary braces---if you try,
your macro definitions will contain unmatched braces
and will therefore be unacceptable to \TeX.
Usually you should use these commands only when you can't use
ordinary braces.
\example
Braces define the {\it boundaries\/} of a group.
|
\produces
Braces define the {\it boundaries\/} of a group.
\nextexample
\def\a{One \vbox\bgroup}
% You couldn't use { instead of \bgroup here because
% TeX would not recognize the end of the macro
\def\enda#1{{#1\egroup} two}
% This one is a little tricky, since the \egroup actually
% matches a left brace and the following right brace
% matches the \bgroup. But it works!!
\a \enda{\hrule width 1in}
|
\produces
\def\a{One \vbox\bgroup}
% You couldn't use { instead of \bgroup here because
% TeX would not recognize the end of the macro
\def\enda#1{{#1\egroup} two}
% This one is a little tricky, since the \egroup actually
% matches a left brace and the following right brace
% matches the \bgroup. But it works!
\a \enda{\hrule width 1in}
\endexample
\enddesc
\begindesc
\cts global {}
\explain
This command makes the following definition
or \minref{assignment} \minref{global} \seeconcept{global} so that it
becomes effective independent of \minref{group} boundaries.
You can apply a |\global| prefix to any kind of definition
or \minref{assignment},
including a \minref{macro} definition or a \minref{register} assignment.
\example
{\global\let\la = \leftarrow}
$a \la b$
|
\produces
% for safety's sake we fake this one!
\let\la = \leftarrow
$a \la b$
\endexample
\enddesc
\begindesc
\cts globaldefs {\param{number}}
\explain
This parameter controls whether or not \TeX\ takes definitions
and other assignments to be
\minref{global}:
\ulist
\li If |\globaldefs| is zero (as it is by default), a definition is global
if and only if it is preceded by |\global| either explicitly or implicitly.
(The ^|\gdef| and ^|\xdef| commands (\xref\gdef) have an implicit
|\global| prefix).
\li If |\globaldefs| is greater than zero, all assignments and
definitions are implicitly prefixed by ^|\global|.
\li If |\globaldefs| is less than zero, all ^|\global| prefixes are ignored.
\endulist
\enddesc
\begindesc
\margin{Order of {\tt\\aftergroup} and {\tt\\afterassignment} changed.}
\cts aftergroup {\<token>}
\explain
When \TeX\ encounters this command during input,
it saves \<token>.
After the end of the current \minref{group},
it inserts \<token> back into the input and expands it.
If a group contains several |\aftergroup|s, the corresponding tokens
are \emph{all} inserted following the end of the group, in the order
in which they originally appeared.
The example that follows shows how you can use |\aftergroup| to postpone
processing a token that you generate within a \minref{conditional test}.
\example
\def\neg{negative} \def\pos{positive}
% These definitions are needed because \aftergroup applies
% to a single token, not to a sequence of tokens or even
% to a brace-delimited text.
\def\arith#1{Is $#1>0$? \begingroup
\ifnum #1>-1 Yes\aftergroup\pos
\else No\aftergroup\neg\fi
, it's \endgroup. }
\arith 2
\arith {-1}
|
\produces
\def\neg{negative} \def\pos{positive}
% These definitions are needed because \aftergroup applies
% to a single token, not a sequence of tokens or even
% a group.
\def\arith#1{Is $#1>0$? \begingroup
\ifnum #1>-1 Yes\aftergroup\pos
\else No\aftergroup\neg\fi
, it's \endgroup. }
\arith 2
\arith {-1}
\endexample
\eix^^{groups}
\enddesc
\begindesc
\cts afterassignment {\<token>}
\explain
When \TeX\ encounters this command it saves \<token> in a special
place. After it next performs an \minref{assignment}, it inserts
\<token> into the input and expands it. If you call |\afterassignment|
more than once before an assignment, only the last call has any effect.
One use of |\afterassignment|
is in writing \minref{macro}s for commands intended to be written
in the
form of assignments, as in the example below.
See \knuth{page~279} for a precise description
of the behavior of |\afterassignment|.
\example
\def\setme{\afterassignment\setmeA\count255}
\def\setmeA{$\number\count255\advance\count255 by 10
+10=\number\count255$}
Some arithmetic: \setme = 27
% After expanding \setme, TeX sets \count255 to 27 and
% then calls \setmeA.
|
\produces
\def\setme{\afterassignment\setmeA\count255}
\def\setmeA{$\number\count255\advance\count255 by 10
+10=\number\count255$}
Some arithmetic: \setme = 27
% After expanding \setme, TeX sets \count255 to 27 and
% then calls \setmeA.
\endexample
\enddesc
%==========================================================================
\section {Macros}
%==========================================================================
\subsection {Defining macros}
\begindesc
\bix^^{macros}
\bix^^{macros//defining}
\xrdef{mac1}% begin the section on macros
%
\cts def {\<control sequence> \<parameter text> \rqbraces{\<replacement text>}}
\explain
This command defines \<control sequence> as a \minref{macro} with the
specified \<parameter text> and \<replacement text>.
See \xrefpg{macro} for a full explanation of how to write a macro
definition.
\example
\def\add#1+#2=?{#1+#2&=
\count255=#1 \advance\count255 by #2 \number\count255\cr}
$$\eqalign{
\add 27+9=?
\add -5+-8=?}$$
|
\dproduces
\def\add#1+#2=?{#1+#2&=
\count255=#1 \advance\count255 by #2 \number\count255\cr}
$$\eqalign{
\add 27+9=?
\add -5+-8=?}$$
\endexample
\enddesc
\begindesc
\cts edef {\<control sequence> \<parameter text> \rqbraces{\<replacement text>}}
\explain
This command defines a macro in the same general way as |\def|.
The difference is that \TeX\ expands the \<replacement text>
of an |\edef| immediately (but still without executing anything).
Thus any definitions within the \<replacement text> are expanded, but
assignments and commands that produce things such as boxes and glue
are left as is. For example, an |\hbox| command within
the \<replacement text> of an |\edef| remains as a command and is not
turned into a box as \TeX\ is processing the definition.
It isn't always obvious what's expanded and what isn't, but you'll
find a complete list of expandable control sequences on
\knuth{pages~212--215}.
You can inhibit the expansion of a control sequence that would otherwise
be expanded by using |\no!-expand| (\xref\noexpand). ^^|\noexpand|
You can postpone the expansion of a control sequence by using
^|\expandafter| (\xref\expandafter).
The |\write|, |\message|, |\errmessage|, |\wlog|, and |\csname|
commands expand their
token lists using the same rules that |\edef| uses to expand its
replacement text.
^^|\write//expanded by {\tt\\edef} rules|
^^|\message//expanded by {\tt\\edef} rules|
^^|\errmessage//expanded by {\tt\\edef} rules|
^^|\wlog//expanded by {\tt\\edef} rules|
^^|\csname//expanded by {\tt\\edef} rules|
\example
\def\aa{xy} \count255 = 1
\edef\bb{w\ifnum \count255 > 0\aa\fi z}
% equivalent to \def\bb{wxyz}
\def\aa{} \count255 = 0 % leaves \bb unaffected
\bb
|
\produces
\def\aa{xy} \count255 = 1
\edef\bb{w\ifnum \count255 > 0\aa\fi z}
% equivalent to \def\bb{wxyz}
\def\aa{} \count255 = 0 % leaves \bb unaffected
\bb
\endexample
\enddesc
\begindesc
\cts gdef {\<control sequence> \<parameter text> \rqbraces{\<replacement text>}}
\explain
This command is equivalent to |\global\def|.
\enddesc
\begindesc
\cts xdef {\<control sequence> \<parameter text> \rqbraces{\<replacement text>}}
\explain
This command is equivalent to |\global\edef|.
\enddesc
\begindesc
\cts long {}
\explain
This command is used as a prefix to a \minref{macro} definition.
It tells \TeX\ that the arguments to the macro are permitted to include
|\par| tokens (\xref{\@par}), which normally indicate the end of a paragraph.
^^|\par//in macro arguments|
If \TeX\
tries to expand a macro defined without |\long| and any of
the macro's arguments include a |\par| token,
\TeX\ will complain about a runaway argument. The purpose
of this behavior is to provide you with some protection against unterminated
macro arguments.
|\long| gives you a way of bypassing the protection.
\example
\long\def\aa#1{\par\hrule\smallskip#1\par\smallskip\hrule}
\aa{This is the first line.\par
This is the second line.}
% without \long, TeX would complain
|
\produces
\medskip
\long\def\aa#1{\par\hrule\smallskip#1\par\smallskip\hrule}
\aa{This is the first line.\par
This is the second line.}
% without \long, TeX would complain
\endexample
\enddesc
\begindesc
\cts outer {}
\explain
\null ^^{outer}
This command is used as a prefix to a \minref{macro} definition.
It tells \TeX\ that the macro is outer (\xref{outer})
and cannot be used in certain contexts.
If the macro is used in a forbidden context, \TeX\ will complain.
\example
\outer\def\chapterhead#1{%
\eject\topglue 2in \centerline{\bf #1}\bigskip}
% Using \chapterhead in a forbidden context causes an
% error message.
|
\endexample
\enddesc
\begindesc
\cts chardef {\<control sequence>=\<charcode>}
\explain
^^{characters//defined by \b\tt\\chardef\e}
This command defines \<control sequence>
to be \<charcode>.
Although |\chardef| is most often used to define characters, you can also
use it to give a name to a number in the range $0$--$255$ even when you
aren't using that number as a character code.
\example
\chardef\percent = `\% 21\percent, {\it 19\percent}
% Get the percent character in roman and in italic
|
\produces
\chardef\percent = `\%
21\percent, {\it 19\percent}
% You'll get the percent character in roman and in italic
\endexample
\enddesc
\begindesc
^^{math characters}
^^{mathcodes}
\cts mathchardef {\<control sequence>=\<mathcode>}
\explain
This command defines \<control sequence> as a math character
with the given \minrefs{mathcode}\<mathcode>.
The control sequence will only be legal in math mode.
\example
\mathchardef\alphachar = "010B % As in plain TeX.
$\alphachar$
|
\produces
\mathchardef\alphachar = "010B % As in plain TeX.
$\alphachar$
\endexample
\eix^^{macros//defining}
\enddesc
%==========================================================================
\subsection {Other definitions}
\begindesc
\cts let {\<control sequence> = \<token>}
\explain
^^{control sequences//defining with \b\tt\\let\e}
\minrefs{token}
This command causes
\<control sequence> to acquire the current meaning of \<token>.
Even if you redefine \<token> later, the meaning of \<control sequence>
will not change. Although \<token> is most commonly a control sequence,
it can also be a \minref{character} token.
\enddesc
\begindesc
\cts futurelet {\<control sequence> \<token$_1$> \<token$_2$>}
\explain
This command tells \TeX\ to make \<token$_2$> the meaning of
\<control sequence> (as would be done with |\let|), and then to
process \<token$_1$> and \<token$_2$> normally.
|\futurelet| is useful at the end of macro definitions
because it gives you a way of looking beyond the token that \TeX\
is about to process before it processes it.
\example
\def\predict#1{\toks0={#1}\futurelet\next\printer}
% \next will acquire the punctuation mark after the
% argument to \predict
\def\printer#1{A \punc\ lies ahead for \the\toks0. }
\def\punc{%
\ifx\next;semicolon\else
\ifx\next,comma\else
``\next''\fi\fi}
\predict{March}; \predict{April}, \predict{July}/
|
\produces
\def\predict#1{\toks0={#1}\futurelet\next\printer}
\def\printer#1{A \punc\ lies ahead for \the\toks0. }
\def\punc{%
\ifx\next;semicolon\else
\ifx\next,comma\else
``\next''\fi\fi
}
\predict{March};
\predict{April},
\predict{July}/
\endexample
\enddesc
\begindesc
\cts csname {\<token list> {\bt \\endcsname}}
\xrdef{\endcsname}
\explain
This command produces a control sequence from \<token list>.
It provides a way of synthesizing control sequences,
including ones that you can't normally write.
\<token list> can itself include control sequences; it is expanded
in the same way as the replacement text of an |\edef| definition (\xref\edef).
If the final expansion
yields anything that isn't a character, \TeX\ will complain.
|\csname| goes from a list of tokens to a control sequence;
you can go the other way with ^|\string| \ctsref\string.
\example
\def\capTe{Te}
This book purports to be about \csname\capTe X\endcsname.
|
\produces
\def\capTe{Te}
This book purports to be about \csname\capTe X\endcsname.
\endexample
\enddesc
%==========================================================================
\subsection {Controlling expansion}
\begindesc
\bix^^{macros//controlling expansion of}
\cts expandafter {\<token$_1$> \<token$_2$>}
\explain
This command tells \TeX\ to expand \<token$_1$> according to its rules
for \minref{macro} expansion \emph{after} it has expanded \<token$_2$>
by one level. It's useful when \<token$_1$> is something like `|{|'
^^|{//with {\tt\\expandafter}|
or ^|\string| that inhibits expansion of \<token$_2$>,
but you want to expand \<token$_2$> nevertheless.
\example
\def\aa{xyz}
\tt % Use this font so `\' prints that way.
[\string\aa] [\expandafter\string\aa]
[\expandafter\string\csname TeX\endcsname]
|
\produces
\def\aa{xyz}
\tt
[\string\aa] [\expandafter\string\aa]
[\expandafter\string\csname TeX\endcsname]
\endexample
\enddesc
\begindesc
\cts noexpand {\<token>}
\explain
This command tells \TeX\ to
suppress expansion of \<token> if \<token> is a
\minref{control sequence} that can be expanded.
If \<token> can't be expanded, e.g., it's a letter,
\TeX\ acts as though the |\noexpand| wasn't there
and processes \<token> normally.
In other words the expansion of `|\noexpand|\<token>'
is simply \<token> no matter what \<token> happens to be.
\example
\def\bunny{rabbit}
\edef\magic{Pull the \noexpand\bunny\ out of the hat!! }
% Without \noexpand, \bunny would always be replaced
% by `rabbit'
\let\oldbunny=\bunny \def\bunny{lagomorph} \magic
\let\bunny=\oldbunny \magic
|
\produces
\def\bunny{rabbit}
\edef\magic{Pull the \noexpand\bunny\ out of the hat! }
% Without \noexpand, \bunny would always be replaced
% by `rabbit'
\let\oldbunny=\bunny \def\bunny{lagomorph} \magic
\let\bunny=\oldbunny \magic
\endexample
\enddesc
\begindesc
\cts the {\<token>}
\explain
This command generally expands to a list of \minref{character} tokens
that represents \<token>. \<token> can be any of the following:
\ulist\compact
\li a \TeX\ \minref{parameter}, e.g., |\parindent| or |\deadcycles|
^^{parameters//using \b\tt\\the\e\ with}
\li a \minref{register}, e.g., |\count0|
^^{registers//with \b\tt\\the\e}
\margin{Item for special registers removed}
\li a code associated with an input character, e.g., |\catcode`(|
\li a font parameter, e.g., |\fontdimen3\sevenbf|
\li the ^|\hyphenchar| or ^|\skewchar| of a font, e.g.,
|\skewchar\teni|
\li ^|\lastpenalty|, ^|\lastskip|, or ^|\lastkern| (values derived from
the last item on the current horizontal \minrefs{horizontal list}
or \minref{vertical list})
\li a control sequence defined by ^|\chardef| or
^|\mathchardef|
\endulist
\noindent
In addition, |\the| can expand to noncharacter tokens in the following two
cases:
\ulist\compact
\li |\the| \<font>, which expands to the most recently defined
control sequence that selects
the same font as the control sequence \<font>
\li |\the| \<token variable>, which expands to a copy of the value of the
variable, e.g., |\the\everypar|
\endulist
See \knuth{pages~214--215} for a more detailed description of what
|\the| does in various cases.
\example
The vertical size is currently \the\vsize.
The category code of `(' is \the\catcode `(.
|
\produces
The vertical size is currently \the\vsize.
The category code of `(' is \the\catcode `(.
\endexample
\enddesc
{\tighten
\see \headcit{Converting information to tokens}{convert},
|\showthe| (\xref\showthe).
\par}
\eix^^{macros//controlling expansion of}
%==========================================================================
\subsection {Conditional tests}
\begindesc
\xrdef{conds}
\bix^^{conditional tests}
%
\ctspecial if {\<token$_1$> \<token$_2$>}\ctsxrdef{@if}
\explain
{\emergencystretch=1em
This command tests if \<token$_1$> and \<token$_2$>
have the same \minref{character} code, independent of their
\minref{category code}s.
Before performing the test, \TeX\ expands tokens following the |\if|
until it obtains two tokens that can't be expanded further.
These two tokens become \<token$_1$> and \<token$_2$>.
The expansion
includes replacing a control sequence |\let| equal to a character token
by that character token.
A \minref{control sequence} that can't be further expanded is
considered to have character code $256$.\par}
\example
\def\first{abc}
\if\first true\else false\fi;
% ``c'' is left over from the expansion of \first.
% It lands in the unexecuted ``true'' part.
\if a\first\ true\else false\fi;
% Here ``bc'' is left over from the expansion of \first
\if \hbox\relax true\else false\fi
% Unexpandable control sequences test equal with ``if''
|
\produces
\def\first{abc}
\if\first true\else false\fi;
% ``c'' is left over from the expansion of \first.
% It lands in the unexecuted ``true'' part.
\if a\first\ true\else false\fi;
% Here ``bc'' is left over from the expansion of \first
\if \hbox\relax true\else false\fi
% Unexpandable control sequences test equal with ``if''
\endexample
\enddesc
\begindesc
\ctspecial ifcat {\<token$_1$> \<token$_2$>}\ctsxrdef{@ifcat}
\explain
^^{category codes//testing}
This command tests if \<token$_1$> and \<token$_2$>
have the same \minref{category code}.
Before performing the test, \TeX\ expands tokens following the |\ifcat|
until it obtains two tokens that can't be expanded further.
These two tokens become \<token$_1$> and \<token$_2$>.
The expansion
includes replacing a control sequence |\let| equal to a character token
by that character token.
A \minref{control sequence} that can't be further expanded is
considered to have category code $16$.
\example
\ifcat axtrue\else false\fi;
\ifcat ]}true\else false\fi;
\ifcat \hbox\day true\else false\fi;
\def\first{12345}
\ifcat (\first true\else false\fi
% ``2345'' lands in the true branch of the test
|
\produces
\ifcat axtrue\else false\fi;
\ifcat ]}true\else false\fi;
\ifcat \hbox\day true\else false\fi;
\def\first{12345}
\ifcat (\first true\else false\fi
% ``2345'' lands in the true branch of the test
\endexample
\enddesc
\begindesc
\ctspecial ifx {\<token$_1$> \<token$_2$>}\ctsxrdef{@ifx}
\explain
This command tests if \<token$_1$> and \<token$_2$> agree.
Unlike |\if| and |\ifcat|, |\ifx| does \emph{not} expand the tokens
following |\ifx|, so \<token$_1$> and \<token$_2$> are the two
tokens immediately after |\ifx|.
There are three cases:
\olist
\li If one token is a \minref{macro} and the other one isn't,
the tokens don't agree.
\li If neither token is a macro, the tokens agree if:
\olist
\li both tokens are characters (or control sequences denoting characters) and
their \minref{character} codes and \minref{category code}s agree, or
\li both tokens refer to the same \TeX\ command,
font, etc.
\endolist
\li If both tokens are macros, the tokens agree if:
\olist\compact
\li their ``first level'' expansions, i.e.,
their replacement texts, are identical, and
\li they have the same status with respect to ^|\long| (\xref\long)
and ^|\outer| (\xref\outer).
\endolist
Note in particular that \emph{any two undefined control
sequences agree}.
\endolist
\noindent
This test is generally more useful than |\if|.
\example
\ifx\alice\rabbit true\else false\fi;
% true since neither \rabbit nor \alice is defined
\def\a{a}%
\ifx a\a true\else false\fi;
% false since one token is a macro and the other isn't
\def\first{\a}\def\second{\aa}\def\aa{a}%
\ifx \first\second true\else false\fi;
% false since top level expansions aren't the same
\def\third#1:{(#1)}\def\fourth#1?{(#1)}%
\ifx\third\fourth true\else false\fi
% false since parameter texts differ
|
\produces
\ifx\alice\rabbit true\else false\fi;
% true since neither \rabbit nor \alice is defined
\def\a{a}%
\ifx a\a true\else false\fi;
% false since one token is a macro and the other isn't
\def\first{\a}\def\second{\aa}\def\aa{a}%
\ifx \first\second true\else false\fi;
% false since top level expansions aren't the same
\def\third#1:{(#1)}\def\fourth#1?{(#1)}%
\ifx\third\fourth true\else false\fi
% false since parameter texts differ
\endexample
\enddesc
\begindesc
\ctspecial ifnum {\<number$_1$> \<relation> \<number$_2$>}\ctsxrdef{@ifnum}
\explain
^^{numbers//comparing}
This command tests if \<number$_1$> and \<number$_2$>
satisfy \<relation>, which must be either `|<|', `|=|', or `|>|'.
The numbers can be constants such as |127|, count registers such as
|\pageno| or |\count22|, or numerical parameters such as |\hbadness|.
Before performing the test, \TeX\ expands tokens following the |\ifnum|
until it obtains a sequence of tokens having
the form \<number$_1$> \<relation> \<number$_2$>, followed by a token
that can't be part of \<number$_2$>.
\example
\count255 = 19 \ifnum \count255 > 12 true\else false\fi
|
\produces
\count255 = 19 \ifnum \count255 > 12 true\else false\fi
\endexample
\enddesc
\begindesc
\ctspecial ifodd {\<number>}\ctsxrdef{@ifodd}
\explain
^^{numbers//testing for odd/even}
This command tests if \<number> is odd.
Before performing the test, \TeX\ expands tokens following the |\ifodd|
until it obtains a sequence of tokens having the form \<number>,
followed by a token that can't be part of \<number>.
\example
\count255 = 19
\ifodd 5 true\else false\fi
|
\produces
\ifodd 5 true\else false\fi
\endexample
\enddesc
\begindesc
\ctspecial ifdim {\<dimen$_1$> \<relation> \<dimen$_2$>}\ctsxrdef{@ifdim}
\explain
^^{dimensions//comparing}
This command tests if \<dimen$_1$> and \<dimen$_2$>
satisfy \<relation>, which must be either `|<|', `|=|', or `|>|'.
The dimensions can be constants such as |1in|, dimension registers
such as |\dimen6|, or dimension parameters such as |\parindent|.
Before performing the test, \TeX\ expands tokens following the |\ifdim|
until it obtains a sequence of tokens having
the form \<dimen$_1$> \<relation> \<dimen$_2$>, followed by a token
that can't be part of \<dimen$_2$>.
\example
\dimen0 = 1000pt \ifdim \dimen0 > 3in true\else false\fi
|
\produces
\dimen0 = 1000pt \ifdim \dimen0 > 3in true\else false\fi
\endexample
\enddesc
\begindesc
\ctspecial ifhmode {}\ctsxrdef{@ifhmode}
\ctspecial ifvmode {}\ctsxrdef{@ifvmode}
\ctspecial ifmmode {}\ctsxrdef{@ifmmode}
\ctspecial ifinner {}\ctsxrdef{@ifinner}
\explain
^^{horizontal mode//testing for}
^^{vertical mode//testing for}
^^{math mode//testing for}
^^{internal mode//testing for}
These commands test what \minref{mode} \TeX\ is in:
\ulist
\li |\ifhmode| is true if \TeX\ is in ordinary or restricted horizontal mode.
\li |\ifvmode| is true if \TeX\ is in ordinary or internal vertical mode.
\li |\ifmmode| is true if \TeX\ is in text math or display math mode.
\li |\ifinner| is true if \TeX\ is in an ``internal'' mode:
restricted horizontal, internal vertical, or text math.
\endulist
\example
\def\modes{{\bf
\ifhmode
\ifinner IH\else H\fi
\else\ifvmode
\ifinner \hbox{IV}\else \hbox{V}\fi
\else\ifmmode \hbox{M}\else
error\fi\fi\fi}}
Formula $\modes$; then \modes,
\hbox{next \modes\ and \vbox{\modes}}.
\par\modes
|
\produces
\def\modes{{\bf
\ifhmode
\ifinner IH\else H\fi
\else\ifvmode
\ifinner \hbox{IV}\fi
\else\ifmmode \hbox{M}\else
error\fi\fi\fi}}
Formula $\modes$; then \modes,
\hbox{next \modes\ and \vbox{\modes}}.
\par\noindent{\bf V} % sorry folks, we have to fake this one
\endexample
\enddesc
\begindesc
\ctspecial ifhbox {\<register>}\ctsxrdef{@ifhbox}
\ctspecial ifvbox {\<register>}\ctsxrdef{@ifvbox}
\ctspecial ifvoid {\<register>}\ctsxrdef{@ifvoid}
\explain
^^{hboxes//testing for}
^^{vboxes//testing for}
^^{boxes//testing if void}
These commands test the contents of
the box register numbered \<reg\-ister>.
Let \<register> be $n$. Then:
\ulist
\li |\ifhbox| is true if |\box|$\,n$ is an \minref{hbox}.
\li |\ifvbox| is true if |\box|$\,n$ is an \minref{vbox}.
\li |\ifvoid| is true if |\box|$\,n$ is void, i.e, doesn't have
a box in it.
\endulist
\example
\setbox0 = \vbox{} % empty but not void
\setbox1 = \hbox{a}
\setbox2 = \box1 % makes box1 void
\ifvbox0 true\else false\fi;
\ifhbox2 true\else false\fi;
\ifvoid1 true\else false\fi
|
\produces
\setbox0 = \vbox{}
\setbox1 = \hbox{a}
\setbox2 = \box1 % empties box1
\ifvbox0 true\else false\fi;
\ifhbox2 true\else false\fi;
\ifvoid1 true\else false\fi
\endexample
\enddesc
\begindesc
\ctspecial ifeof {\<number>}\ctsxrdef{@ifeof}
\explain
^^{end of file, testing for}
\minrefs{file}
This command tests an input stream for end of file.
It is true if input stream \<number> has not been opened,
or has been opened and the associated file has been entirely read in
(or doesn't exist).
\enddesc
\begindesc
\ctspecial ifcase
{\<number>\<case$_0$ text> {\bt \\or }\<case$_1$ text> {\bt \\or}
$\ldots$ {\bt \\or} \<case$_n$ text>\hfil\break
\hglue 3pc{\bt \\else} \<otherwise text> {\bt \\fi}}
\ctsxrdef{@ifcase}
\ctsxrdef{@or}
\explain
^^{case testing}
This command introduces a test with numbered multiple cases.
If \<num\-ber> has the value $k$, \TeX\ will expand \<case$_k$ text> if
it exists, and \<other\-wise text> if it doesn't. You can omit the |\else|---%
in this case, \TeX\ won't expand anything if none of the cases are satisfied.
\example
\def\whichday#1{\ifcase #1<day 0>\or Sunday\or Monday%
\or Tuesday\or Wednesday\or Thursday\or Friday%
\or Saturday\else Nonday\fi
\ is day \##1. }
\whichday2 \whichday3 \whichday9
|
\produces
\def\whichday#1{\ifcase #1<day 0>\or Sunday\or Monday%
\or Tuesday\or Wednesday\or Thursday\or Friday%
\or Saturday\else Nonday\fi
\ is day \##1. }
\whichday2 \whichday3 \whichday9
\endexample
\enddesc
\begindesc
\ctspecial iftrue {}\ctsxrdef{@iftrue}
\ctspecial iffalse {}\ctsxrdef{@iffalse}
\explain
These commands are equivalent to tests that are always true or always
false. The main use of these commands is in defining macros that keep
track of the result of a test.
\example
\def\isbigger{\let\bigger=\iftrue}
\def\isnotbigger{\let\bigger=\iffalse}
% These \let's MUST be buried in macros!! If they aren't,
% TeX erroneously tries to match them with \fi.
\def\test#1#2{\ifnum #1>#2 \isbigger\else\isnotbigger\fi}
\test{3}{6}
\bigger$3>6$\else$3\le6$\fi
|
\produces
\def\isbigger{\let\bigger=\iftrue}
\def\isnotbigger{\let\bigger=\iffalse}
% These \let's MUST be buried in macros!
% If they aren't, TeX erroneously tries to match them with \fi
\def\test#1#2{\ifnum #1>#2 \isbigger\else\isnotbigger\fi}
\test{3}{6}
\bigger$3>6$\else$3\le6$\fi
\endexample
\enddesc
\begindesc
\ctspecial else {} \ctsxrdef{@else}
\explain
This command introduces the ``false'' alternative of a conditional test.
\enddesc
\begindesc
\ctspecial fi {} \ctsxrdef{@fi}
\explain
This command ends the text of a conditional test.
\enddesc
\begindesc
\ctspecial newif {{\bt \\if}\<test name>}\ctsxrdef{@newif}
\explain
This command names a trio of control sequences with names |\alpha!-true|,
|\alphafalse|,
and |\ifalpha|, where |alpha| is \<test name>.
You can use them to define your own tests by
creating a logical variable that records
true\slash false information:
\ulist\compact
\li |\alphatrue| sets the logical variable |alpha| true.
\li |\alphafalse| sets the logical variable |alpha| false
\li |\ifalpha| is a conditional test that is true if the logical
variable |alpha| is true and false otherwise.
\endulist
The logical variable |alpha| doesn't really exist, but \TeX\ behaves as
though it did. After |\newif\ifalpha|, the logical variable is initially
false.
|\newif| is an outer command, so you can't use it inside a macro
definition.
\example
\newif\iflong \longtrue
\iflong Rabbits have long ears.
\else Rabbits don't have long ears.\fi
|
\produces
\newif\iflong
\longtrue
\iflong Rabbits have long ears.\else Rabbits don't have long ears.\fi
\endexample
\eix^^{conditional tests}
\enddesc
%==========================================================================
\subsection {Repeated actions}
{\def\test{{\bt \\if}$\Omega$}%
\begindesc
\bix^^{repeated actions}
\bix^^{loops}
\cts loop {$\alpha$ {\test} $\beta$ {\bt \\repeat}}
\ctspecial repeat {}\ctsxrdef{@repeat}
\explain
These commands provide a looping construct for \TeX.
Here $\alpha$ and $\beta$ are arbitrary sequences of commands
and \test\ is any of the conditional tests described in
\headcit{Conditional tests}{conds}.
The |\repeat| replaces the |\fi| corresponding to the test,
so you must not write an explicit |\fi| to terminate the test.
Nor, unfortunately, can you associate an |\else| with the test.
If you want to use the test in the opposite sense, you need to
rearrange the test or
define an auxiliary test with |\newif| (see above) and use that
test in the sense you want (see the second example below).
\TeX\ expands |\loop| as follows:
\olist
\li $\alpha$ is expanded.
\li {\test} is performed. If the result is false, the loop is terminated.
\li $\beta$ is expanded.
\li The cycle is repeated.
\endolist
\example
\count255 = 6
\loop
\number\count255\
\ifnum\count255 > 0
\advance\count255 by -1
\repeat
|
\produces
\count255 = 6
\loop
\number\count255\
\ifnum\count255 > 0
\advance\count255 by -1
\repeat
\nextexample
\newif\ifnotdone % \newif uses \count255 in its definition
\count255=6
\loop
\number\count255\
\ifnum\count255 < 1 \notdonefalse\else\notdonetrue\fi
\ifnotdone
\advance\count255 by -1
\repeat
|
\produces
\newif\ifnotdone
\count255=6
\loop
\number\count255\
\ifnum\count255 < 1 \notdonefalse\else\notdonetrue\fi
\ifnotdone
\advance\count255 by -1
\repeat
%
\eix^^{repeated actions}
\eix^^{loops}
%
\endexample
\enddesc
} % end scope of definition of \test
%==========================================================================
\subsection {Doing nothing}
\begindesc
\cts relax {}
\explain
This command tells \TeX\ to do nothing. It's useful in a context where
you need to provide a command but there's nothing that you want \TeX\ to do.
\example
\def\medspace{\hskip 12pt\relax}
% The \relax guards against the possibility that
% The next tokens are `plus' or `minus'.
|
\endexample
\enddesc
\begindesc
\cts empty {}
\explain
This command expands to no tokens at all.
It differs from |\relax| in that it disappears after macro expansion.
%
\xrdef{mac2}% end the section on macros
\eix^^{macros}
\enddesc
%==========================================================================
\section {Registers}
%==========================================================================
\subsection {Using registers}
\begindesc
\bix^^{registers}
\makecolumns 11/2:
\cts count {\<register> {\bt =} \<number>}
\cts dimen {\<register> {\bt =} \<dimen>}
\cts skip {\<register> {\bt =} \<glue>}
\cts muskip {\<register> {\bt =} \<muglue>}
\cts toks {\<register> {\bt =} \<token variable>}
\aux\cts toks {\<register> {\bt =} \rqbraces{\<token list>}}
\aux\cts count {\<register>}
\aux\cts dimen {\<register>}
\aux\cts skip {\<register>}
\aux\cts muskip {\<register>}
\aux\cts toks {\<register>}
\explain
^^{assignments//of registers}
The first six commands listed here assign something to a register.
The |=|'s in the assignments are optional.
The remaining five control sequences are not true commands
because they can only appear as part of an argument.
They yield the contents of the specified register.
Although you can't use these control sequences by themselves as commands
in text, you can use ^|\the| to convert them to text so that
you can typeset their values.
You can name and reserve registers
with the ^|\newcount| command and its relatives
(\xref{\@newcount}).
Using these commands is a safe way to obtain registers that
are known not to have any conflicting usage.
^^{count registers}
A |\count| register contains an integer, which can be either positive or
negative.
Integers can be as large as you're ever likely to need them to be.\footnote
{Here's the only exercise in this book: find out what's the largest
integer that \TeX\ will accept.}
\TeX\ uses count registers $0$--$9$ to keep track of the
page number (see \knuth{page~119}).
|\count255| is the only count register available for use
without a reservation.
\example
\count255 = 17 \number\count255
|
\produces
\count255 = 17 \number\count255
\endexample
\medskip\noindent
^^{dimension registers}
A |\dimen| register contains a dimension.
Registers |\dimen0| through |\dimen9| and |\dimen255| are available
for scratch use.
\example
\dimen0 = 2.5in
\hbox to \dimen0{$\Leftarrow$\hfil$\Rightarrow$}
|
\produces
\dimen0 = 2.5in
\hbox to \dimen0{$\Leftarrow$\hfil$\Rightarrow$}
\doruler{\8\8\8}3{in}
\endexample
\medskip\noindent
^^{skip registers}
A |\skip| register contains the dimensions of glue.
Unlike a |\dimen| register, it
records an amount of shrink and stretch as well as a natural size.
Registers |\skip0| through |\skip9| and |\skip255| are available
for use without a reservation.
\example
\skip2 = 2in
$\Rightarrow$\hskip \skip2 $\Leftarrow$
|
\produces
\skip2 = 2in
$\Rightarrow$\hskip \skip2 $\Leftarrow$\par
\noindent\hphantom{$\Rightarrow$}\ruler{\8\8}2{in}
\endexample
\medskip\noindent
^^{muskip registers}
A |\muskip| register is like a |\skip| register,
but the glue in it is always measured in ^|mu|
\seeconcept{mathematical unit}.
The size of a |mu| depends on the current font.
For example, it's usually a little
smaller in a subscript than in ordinary text.
Registers |\muskip0| through |\muskip9| and |\muskip255| are available
for use without a reservation.
\example
\muskip0 = 24mu % An em and a half, no stretch or shrink.
$\mathop{a \mskip\muskip0 b}\limits^{a \mskip\muskip0 b}$
% Note the difference in spacing.
|
\produces
\muskip0 = 24mu % an em and a half
$\mathop{a \mskip\muskip0 b}\limits^{a \mskip\muskip0 b}$
% Note the difference in spacing
\endexample
\medskip\noindent
^^{token registers}
You can assign either a token variable
(a register or a parameter) or a token list
to a |\toks| register.
When you assign a token list to a token register,
the tokens in the token list are \emph{not} expanded.
Once the tokens in a token list have been inserted into text
using ^|\the|, they are
expanded just like tokens that were read in directly.
They have the category codes that they received when \TeX\ first
saw them in the~\hbox{input}.
\example
\toks0 = {the \oystereaters\ were at the seashore}
% This assignment doesn't expand \oystereaters.
\def\oystereaters{Walrus and Carpenter}
\toks1 = \toks0
% the same tokens are now in \toks0 and \toks1
Alice inquired as to whether \the\toks1.
|
\produces
\toks0 = {the \oystereaters\ were at the seashore}
% This assignment doesn't expand \oystereaters
\def\oystereaters{Walrus and Carpenter}
\toks1 = \toks0
% the same tokens are now in \toks0 and \toks1
Alice inquired as to whether \the\toks1.
\endexample
\enddesc
\begindesc
\cts maxdimen {}
\explain
^^{dimensions//maximum}
This control sequence yields a \<dimen> that is the
largest dimension acceptable to \TeX\ (nearly 18 feet).
It is not a true command because it can only appear as part of an argument
to another command.
\example
\maxdepth = \maxdimen % Remove restrictions on \maxdepth.
|
\endexample
\enddesc
\see |\advance| (\xref\advance), |\multiply|,
|\divide| (\xref\divide), |\set!-box|, |\box| (\xref\box).
%==========================================================================
\subsection {Naming and reserving registers, etc.}
\begindesc
\bix^^{registers//reserving}
\makecolumns 11/2:
\ctspecial newcount \ctsxrdef{@newcount}
\ctspecial newdimen \ctsxrdef{@newdimen}
\ctspecial newskip \ctsxrdef{@newskip}
\ctspecial newmuskip \ctsxrdef{@newmuskip}
\ctspecial newtoks \ctsxrdef{@newtoks}
\ctspecial newbox \ctsxrdef{@newbox}
\ctspecial newread \ctsxrdef{@newread}
\ctspecial newwrite \ctsxrdef{@newwrite}
\ctspecial newfam \ctsxrdef{@newfam}
\ctspecial newinsert \ctsxrdef{@newinsert}
\ctspecial newlanguage \ctsxrdef{@newlanguage}
\explain
These commands
reserve and name an entity of the indicated type:
\ulist
\li |\newcount|, |\newdimen|, |\newskip|, |\newmuskip|, |\newtoks|,
|\newbox| each reserve a \minref{register} of the indicated type.
^^{count registers//reserved by \b\tt\\newcount\e}
^^{dimension registers//reserved by \b\tt\\newdimen\e}
^^{skip registers//reserved by \b\tt\\newskip\e}
^^{muskip registers//reserved by \b\tt\\newmuskip\e}
^^{token registers//reserved by \b\tt\\newtoks\e}
^^{box registers//reserved by \b\tt\\newbox\e}
\li |\newread| and |\newwrite| reserve an input stream and
an output stream \minrefs{input stream}\minrefs{output stream}
respectively.
^^{input streams//reserved by \b\tt\\newread\e}
^^{output streams//reserved by \b\tt\\newwrite\e}
\li |\newfam| reserves a \minref{family} of math fonts.
^^{family//reserved by \b\tt\\newfam\e}
\li |\newinsert| reserves an insertion type.
(Reserving an insertion type involves reserving several different registers.)
^^{insertions//numbers reserved by \b\tt\\newinsert\e}
\li |\newlanguage| reserves a set of hyphenation patterns.
\endulist
You should use these commands whenever you need one of these entities,
other than in a very local region,
in order to avoid numbering conflicts.
There's an important difference among these commands:
\ulist
\li The control sequences defined by
|\newcount|, |\newdimen|, |\newskip|, |\newmuskip|, and |\newtoks|
each designate an entity of the appropriate type.
For instance, after the command:
\csdisplay
\newdimen\listdimen
|
the control sequence |\listdimen| can be used as a dimension.
\li The control sequences defined by
|\newbox|, |\newread|, |\newwrite|, |\newfam|, |\newinsert|,
and |\newlanguage| each
evaluate to the \emph{number} of an entity of the appropriate type.
For instance, after the command:
\csdisplay
\newbox\figbox
|
the control sequence |\figbox| must be used in conjunction with
a |\box|-like command, e.g.:
\csdisplay
\setbox\figbox = \vbox{!dots}
|
\endulist
\enddesc
\begindesc
\cts countdef {\<control sequence> {\bt =} \<register>}
\cts dimendef {\<control sequence> {\bt =} \<register>}
\cts skipdef {\<control sequence> {\bt =} \<register>}
\cts muskipdef {\<control sequence> {\bt =} \<register>}
\cts toksdef {\<control sequence> {\bt =} \<register>}
\explain
These commands define \<control sequence> to refer to the
\minref{register} of the indicated category whose number is \<register>.
Normally you should use the commands in the previous group
(|\newcount|, etc.) in preference to these commands in order to avoid
numbering conflicts. The commands in the previous group are
defined in terms of the commands \hbox{in this group}.
\example
\countdef\hatters = 19 % \hatters now refers to \count19
\toksdef\hares = 200 % \hares now refers to \toks200
|
\endexample
\enddesc
\see |\newif| (\xref{\@newif}), |\newhelp| (\xref{\@newhelp}).
\eix^^{registers//reserving}
%==========================================================================
\subsection {Doing arithmetic in registers}
\begindesc
\bix^^{arithmetic}
\bix^^{registers//arithmetic in}
%
\cts advance {\<count register> {\bt by} \<number>}
\aux\cts advance {\<dimen register> {\bt by} \<dimen>}
\aux\cts advance {\<skip register> {\bt by} \<glue>}
\aux\cts advance {\<muskip register> {\bt by} \<muglue>}
\explain
This command adds a compatible quantity to a register. For \<glue>
or \<muglue> all three components (natural value, stretch, and shrink)
\minrefs{glue} are added.
Any of the quantities can be negative. For purposes of these calculations
(and other assignments as well), \<glue> can be converted to a
\<dimen> by dropping the stretch and shrink, and a \<dimen> can be converted
to a \<number> by taking its value in scaled points
\seeconcept{dimension}.
You can omit the word |by| in these commands---\TeX\ will understand them
anyway.
\example
\count0 = 18 \advance\count0 by -1 \number\count0\par
\skip0 = .5in \advance\skip0 by 0in plus 1in % add stretch
\hbox to 2in{a\hskip\skip0 b}
|
\produces
\count0 = 18 \advance\count0 by -1 \number\count0\par
\skip0 = .5in \advance\skip0 by 0in plus 1in % add stretch
\hbox to 2in{a\hskip\skip0 b}
\doruler{\8\8}2{in}
\endexample
\enddesc
\begindesc
\cts multiply {\<register> {\bt by} \<number>}
\cts divide {\<register> {\bt by} \<number>}
\explain
These commands multiply and divide the value in \<register>
by \<number> (which can be negative).
The register can be a ^|\count|, ^|\dimen|, ^|\skip|, or ^|\muskip|
register.
For a ^|\skip| or ^|\muskip| register (\xref\skip),
all three components of the \minref{glue} in the register are modified.
You can omit the word |by| in these commands---\TeX\ will understand them
anyway.
You can also obtain a multiple of a \<dimen> by preceding it by a \<number>
\minrefs{number}
or decimal constant, e.g.,
|-2.5\dimen2|.
You can also use this notation for \<glue>, but watch out---the result
is a \<dimen>, not \<glue>.
Thus |2\baselineskip| yields a \<dimen> that is twice the natural size
of |\baselineskip|, with no stretch or shrink.
\example
\count0 = 9\multiply \count0 by 8 \number\count0 ;
\divide \count0 by 12 \number\count0 \par
\skip0 = 20pt plus 2pt minus 3pt \multiply \skip0 by 3
Multiplied value of skip0 is \the\skip0.\par
\dimen0 = .5in \multiply\dimen0 by 6
\hbox to \dimen0{a\hfil b}
|
\produces
\count0 = 9\multiply \count0 by 8 \number\count0 ;
\divide \count0 by 12 \number\count0 \par
\skip0 = 20pt plus 2pt minus 3pt \multiply \skip0 by 3
Multiplied value of skip0 is \the\skip0.\par
\dimen0 = .5in \multiply\dimen0 by 6
\hbox to \dimen0{a\hfil b}
\doruler{\8\8\8}3{in}
\endexample
\eix^^{arithmetic}
\eix^^{registers//arithmetic in}
\eix^^{registers}
\enddesc
%==========================================================================
\section {Ending the job}
\begindesc
^^{ending the job}
\easy\ctspecial bye \ctsxrdef{@bye}
\explain
This command tells \TeX\ to fill out and produce the last page, print
any held-over \minref{insertion}s, and end the job.
It is the usual way to end your input file.
\enddesc
\begindesc
\cts end {}
\explain
This command tells \TeX\ to produce the last page and end the job.
It does not fill out the page, however,
so it's usually better to use |\bye| rather than |\end|.
\enddesc
%==========================================================================
\section {Input and output}
%==========================================================================
\subsection {Operations on input files}
\begindesc
\bix^^{files}
\bix^^{input files}
\easy\cts input {\<filename>}
\explain
\minrefs{file}\minrefs{file name}
This command tells \TeX\ to read its input from file \<filename>.
When that file is exhausted, \TeX\ returns to reading from its previous
input source. You can nest input files to any level you like
(within reason).
When you're typesetting a large document, it's usually a good idea to
structure your main file as a sequence of |\input| commands that refer
to the subsidiary parts of the document. That way you can process the
individual parts easily as you're working on drafts. It's also a good
practice to put all of your \minref{macro} definitions into a separate file and
summon that file with an |\input| command as the first action in your
main file.
\TeX\ uses different rules for scanning file names than it does for scanning
\minref{token}s in general (see \xref{file name}).
If your implementation expects file names to have extensions (usually
indicated by a preceding dot), then \TeX\ provides a default extension
of |.tex|.
\example
\input macros.tex
\input chap1 % equivalent to chap1.tex
|
\endexample
\enddesc
\begindesc
\cts endinput {}
\explain
This command tells \TeX\ to stop reading input from the current file when it
next reaches the end of a line.
\enddesc
\begindesc
\cts inputlineno {}
\explain
This command yields a number (not a string) giving the line number of the
current line, defined to be the number that would appear in an error message
if an error occurred at this point.
\enddesc
\begindesc
\cts openin {\<number> {\bt =} \<filename>}
\explain
This command tells \TeX\ to open the file named \<filename>
and make it available for reading via the input stream
designated by \<number>.
^^{input streams//opening}
\<number> must be between $0$ and $15$.
Once you've opened a file and connected it to an input stream,
you can read from the file using the |\read| command
with the input stream's number.
You can associate more than one input stream with the same
file. You can then read from several different positions within
the file, one for each input stream.
You should allocate |\openin| stream numbers with
|\newread| (\xref{\@newread}).
\example
\newread\auxfile \openin\auxfile = addenda.aux
% \auxfile now denotes the number of this opening
% of addenda.aux.
|
\endexample
\enddesc
\begindesc\secondprinting{\vglue-.5\baselineskip\vskip0pt}
\cts closein {\<number>}
\explain
This command tells \TeX\ to close the \minref{input stream} numbered
\<number>, i.e.,
end the association between the input stream and its file.
The input stream with this number then becomes available for use with a
different file.
You should close an input stream once you're finished using its file.
\example
\closein\auxfile
|
\endexample
\enddesc
\begindesc\secondprinting{\vglue-.5\baselineskip\vskip0pt}
\cts read {\<number> {\bt to} \<control sequence>}
\explain
^^{input streams//reading with \b\tt\\read\e}
^^{reading a file}
This command tells \TeX\ to read a line from the file
associated with the \minref{input stream}
designated by \<number> and assign the tokens on that line to
\<control sequence>. The \minref{control sequence} then becomes a
parameterless \minref{macro}. No macro expansion takes place
during the reading operation. If the line contains any unmatched
left braces, \TeX\ will read additional lines until the braces are
all matched. If \TeX\ reaches the end of the file without matching all the
braces, it will complain.
If \<number> is greater than $15$ or hasn't been associated with a file
using ^|\openin|, \TeX\ prompts you with `\<control sequence> |=|'
on your terminal and waits for you to type a line of input.
It then assigns the input line to \<control sequence>.
If \<number> is less than zero, it reads a line of input from your
terminal but omits the prompt.
\example
\read\auxfile to \holder
% Expanding \holder will produce the line just read.
|
\endexample
\eix^^{input files}
\enddesc
\secondprinting{\vfill\eject}
%==========================================================================
\subsection {Operations on output files}
\bix^^{output files}
\begindesc
\cts openout {\<number> {\bt =} \<filename>}
\explain
^^{output streams//opening}
This command tells \TeX\ to open the file named \<filename>
and make it available for writing via the \minref{output stream}
designated by \<number>.
\<number> must be between $0$ and $15$.
Once you've opened a file and connected it to an output stream,
you can write to the file using the |\write| command
with the output stream's number.
An |\openout| generates a whatsit that becomes part of a box.
The |\openout| does not take effect until \TeX\ ships out that box
to the \dvifile,
unless you've preceded the |\openout| with ^|\immediate|.
\TeX\ won't complain if you associate more than one output stream with the
same file, but you'll get garbage in the file if you try it!
You should allocate stream numbers for |\openout| using
|\newwrite| (\xref{\@newwrite}).
\example
\newwrite\auxfile \openout\auxfile = addenda.aux
% \auxfile now denotes the number of this opening
% of addenda.aux.
|
\endexample
\enddesc
\begindesc
\cts closeout {\<number>}
\explain
^^{output streams//closing}
This command tells \TeX\ to close the \minref{output stream} numbered
\<number>. i.e.,
end the association between the output stream and its file.
The output stream with this number then becomes available for use with a
different file.
You should close an output stream once you're finished using its file.
A |\closeout| generates a whatsit that becomes part of a box.
The |\closeout| does not take effect until \TeX\ ships out that box
to the \dvifile,
unless you've preceded the |\closeout| with ^|\immediate|.
\example
\closeout\auxfile
|
\endexample
\enddesc
\begindesc
\cts write {\<number> \rqbraces{\<token list>}}
\explain
^^{output streams//writing}
^^{writing a file}
This command tells \TeX\ to write \<token list> to the file
associated with the \minref{output stream}
designated by \<number>.
It generates a whatsit that becomes part of a box.
The actual writing does not take place until \TeX\ ships out that box
to the \dvifile,
unless you've preceded the |\write| with ^|\immediate|.
For a |\write| that is not immediate, \TeX\ does not expand macros in
\<token list> until the token list is actually written to the file.
The macro expansions follow the same rules as |\edef| (\xref\edef).
In particular, any control sequence that is not
the name of a macro is written as
^|\escapechar| followed by the control sequence name
and a space. Any `|#|' tokens
in \<token list> are doubled, i.e., written as `|##|'.
If \<number> is not in the range from $0$ to $15$, \TeX\ writes
\<token list> to the log file.
^^{log file//written by \b\tt\\write\e}
If \<number> is greater than $15$ or isn't associated with an output
stream, \TeX\ also writes \<token list> to the terminal.
\example
\def\aa{a a}
\write\auxfile{\hbox{$x#y$} \aa}
% Writes the string `\hbox {$x##y$} a a' to \auxfile.
|
\endexample
\enddesc
\begindesc
\cts immediate {}
\explain
This command should precede an |\openout|, |\closeout|, or |\write|.
^^|\write//with {\tt\\immediate}|
^^|\openout//with {\tt\\immediate}|
^^|\closeout//with {\tt\\immediate}|
It tells \TeX\ to perform the specified file operation without delay.
\example
\immediate\write 16{I'm stuck!!}
% has the same effect as \message
|
\endexample\enddesc
\begindesc
\cts special {\rqbraces{\<token list>}}
\explain
This command tells \TeX\ to
write \<token list> directly to the \dvifile\ when it next
ships out a page.
A typical use of |\special| would be to tell the device driver to incorporate
the contents of a named graphics file into the output page.
^^{device drivers//instructions from \b\tt\\special\e}
The |\special| command produces a whatsit that associates
\<token list> with a particular position on the page, namely,
the position that a zero-size box would have had if such a box
had appeared instead of the |\special| command.
Any use you might make of |\special| depends strictly on
the ^{device drivers} that you have available.
\example
\special{graphic expic}
% Display the graphics file `expic' here.
|
\endexample
\enddesc
\begindesc
\cts newlinechar {\param{number}}
\explain
This parameter contains a character that indicates a new line on
output. When \TeX\ encounters this character
while reading the argument of
a |\write|, |\message|, or
|\errmessage| command, it starts a new line.
If |\newlinechar| is not in the range $0$--$255$,
there is no character that indicates
a new line on output.
\PlainTeX\ sets |\newlinechar| to $-1$.
\example
\newlinechar = `\^^J
\message{This message appears^^Jon two lines.}
|
\logproduces
This message appears
on two lines.
|
\endexample
\enddesc
\see |\newread|, |\newwrite| (\xref{\@newwrite}).
\eix^^{files}
\eix^^{output files}
%==========================================================================
\subsection {Interpreting input characters}
\begindesc
\cts catcode {\<charcode> \tblentry{number}}
\explain
^^{category codes//in \b\tt\\catcode\e\ table}
This table entry contains the \minref{category code} of the character
whose \ascii\ code is \<charcode>.
The category codes are listed on \xrefpg{catcodes}.
By changing the category code of a character you can get \TeX\ to treat
that character differently.
\example
\catcode `\[ = 1 \catcode `\] = 2
% Make [ and ] act like left and right braces.
|
\endexample
\enddesc
\begindesc
\cts active {}
\explain
This command contains the
category code for an active character, namely, the number $13$.
\example
\catcode `\@ = \active % Make @ an active character.
|
\endexample
\enddesc
\begindesc
\cts mathcode {\<charcode> \tblentry{number}}
\explain
This table entry contains the \minref{mathcode} of the character
whose \ascii\ code is \<charcode> \seeconcept{mathcode}.
The mathcode specifies that character's interpretation in math mode.
\example
\mathcode\> = "313E % as in plain TeX
% The > character has class 3 (relation), family 1 (math
% italic), and character code "3E
|
\endexample
\enddesc
\begindesc
\margin{{\tt\\delcode} was explained in two places. This explanation
combines them. (The other place was in the math section.)}
\cts delcode {\<charcode>\tblentry{number}}
\explain
^^{delimiter codes}
This table entry specifies the \minref{delimiter} code for the input character
whose \ascii\ code is \<charcode>.
The delimiter code tells \TeX\ how to find the best output character to use
for typesetting the indicated input character as a delimiter.
\<number> is normally written in hexadecimal notation.
Suppose that \<number> is the hexadecimal number $s_1s_2s_3\,
l_1l_2l_3$. Then when the character is used as a delimiter,
\TeX\ takes the character to have small variant
$s_1s_2s_3$ and large variant $l_1l_2l_3$. Here $s_1s_2s_3$ indicates
the math character found in position $s_2s_3$ of family $s_1$, and
similarly for $l_1l_2l_3$. This is the same convention as the one
used for ^|\mathcode| (\xref \mathcode),
except that |\mathcode| also specifies a class.
\example
\delcode `( = "028300 % As in plain TeX.
|
\endexample
\enddesc
\begindesc
\cts endlinechar {\param{number}}
\explain
This parameter
contains the character code for the character that
\TeX\ appends to the end of each input line.
^^{input lines}
A value not in the range $0$--$255$
indicates that no character should be appended.
\PlainTeX\ leaves |\end!-line!-char| at
|`\^^M| (the {\ascii} code for \asciichar{return}).
\enddesc
\begindesc
\cts ignorespaces {}
\explain
This command tells \TeX\ to read and expand tokens until it finds one that
is not a space \minref{token}, ignoring any space tokens
that it finds on the way.
|\ignorespaces| is often useful at the end of a \minref{macro} as a way
of making the macro insensitive to any spaces or ends of line
that might follow calls on it.
(An empty line after |\ignorespaces| still produces a |\par| token,
however.)
\example
\def\aa#1{yes #1\ignorespaces}
\aa{may}
be
|
\produces
\def\aa#1{yes #1\ignorespaces}
\aa{may}
be
\endexample
\enddesc
%==========================================================================
\section {Controlling interaction with \TeX}
\begindesc
\bix^^{controlling \TeX}
\bix^^{running \TeX}
\cts errorstopmode {}
\explain
This command tells \TeX\
to stop for interaction whenever it finds an error.
This is the normal mode of operation.
\enddesc
\begindesc
\cts scrollmode {}
\explain
This command tells \TeX\
not to stop for most errors,
but to continue displaying the error messages on your terminal.
Typing `|S|' or `|s|' in response to an error message puts you
into scroll mode.
\enddesc
\begindesc
\cts nonstopmode {}
\explain
This command tells \TeX\
not to stop for errors, even those pertaining to files that
it can't find, but to continue displaying the error messages on your terminal.
Typing `|R|' or `|r|' in response to an error message puts you
into nonstop mode.
\enddesc
\begindesc
\cts batchmode {}
\explain
This command tells \TeX\
not to stop for errors and to suppress all further output to your terminal.
Typing `|Q|' or `|q|' in response to an error message puts you
into batch mode.
\enddesc
\begindesc
\cts pausing {\param{number}}
\explain
If this parameter is greater than zero, \TeX\ will pause
at each line of input to give you an opportunity to
replace it with a different line. If you type in a replacement,
\TeX\ will use that line instead of the original one; if you respond
with \asciichar{return}, \TeX\ will use the original line.
Setting |\pausing| to $1$ can be useful as a way of patching a document as
\TeX\ is processing it. For example,
you can use this facility to insert ^|\show| commands (see below).
\eix^^{running \TeX}
\eix^^{controlling \TeX}
\enddesc
%==========================================================================
\section {Diagnostic aids}
\subsection{Displaying internal data}
\begindesc
\bix^^{tracing}
\bix^^{debugging}
\bix^^{diagnostic aids}
\cts show {\<token>}
\cts showthe {\<argument>}
\cts showbox {\<number>}
\cts showlists {}
\explain
These commands record information in the log of your \TeX\ run:
\ulist
\li |\show| records the meaning of \<token>.\minrefs{token}
^^{tokens//displayed by \b\tt\\show\e}
\li |\showthe| records
whatever tokens would be produced by
|\the| \<arg\-u\-ment> (see \xref \the).
\li |\showbox| records the contents of the \minref{box}
\minref{register} numbered \<num\-ber>.
The number of
leading dots in the log indicates the number of
levels of nesting of inner boxes.
\li |\showlists| records
the contents of each of the lists that \TeX\ is currently constructing.
(These lists are nested one within another.)
See \knuth{pages~88--89} for further information about interpreting
the output of |\showlists|.
\endulist
For |\show| and |\showthe|, \TeX\ also displays the information at your
^{terminal}.
For |\showbox| and |\showlists|, \TeX\ displays the information at your
terminal only
if ^|\tracingonline| (\xref \tracingonline) is greater than zero;
if ^|\tracingonline| is zero or less (the default case),
the information is not displayed.
Whenever \TeX\ encounters a |\show|-type command it
stops for interaction. The request for interaction does \emph{not}
indicate an error, but it does give you an opportunity to ask \TeX\ to
show you something else. If you don't want to see anything else, just
press \asciichar{return}.
You can control the amount of output produced by |\showbox| by setting
|\show!-box!-breadth| and |\show!-box!-depth| (\xref\showboxbreadth).
^^|\showboxbreadth| ^^|\showboxdepth|
These parameters respectively have default values of $5$
and $3$, which is why
just five items appear for each box described
in the log output below. (The `|..etc.|' indicates additional items
within the boxes that aren't displayed.)
\example
\show a
\show \hbox
\show \medskip
\show &
|
\logproduces
> the letter a.
> \hbox=\hbox.
> \medskip=macro:
->\vskip \medskipamount .
> alignment tab character &.
|
\nextexample
\showthe\medskipamount
\toks27={\hbox{Joe's\quad\ Diner}}
\showthe\toks27
|
\logproduces
> 6.0pt plus 2.0pt minus 2.0pt.
> \hbox {Joe's\quad \ Diner}.
|
\nextexample
\setbox 3=\vbox{\hbox{A red dog.}\hrule A black cat.}
\showbox 3
|
\logproduces
> \box3=
\vbox(16.23332+0.0)x53.05565
.\hbox(6.94444+1.94444)x46.41675
..\tenrm A
..\glue 3.33333 plus 1.66498 minus 1.11221
..\tenrm r
..\tenrm e
..\tenrm d
..etc.
.\rule(0.4+0.0)x*
.\hbox(6.94444+0.0)x53.05565
..\tenrm A
..\glue 3.33333 plus 1.66498 minus 1.11221
..\tenrm b
..\tenrm l
..\tenrm a
..etc.
|
\endexample
\vfil\eject
\example
\vbox{A \hbox
{formula
$x \over y\showlists$}}
|
\logproduces
### math mode entered at line 3
\mathord
.\fam1 y
this will be denominator of:
\fraction, thickness = default
\\mathord
\.\fam1 x
### restricted horizontal mode entered at line 2
\tenrm f
\tenrm o
\tenrm r
\tenrm m
\kern-0.27779
\tenrm u
\tenrm l
\tenrm a
\glue 3.33333 plus 1.66666 minus 1.11111
spacefactor 1000
### horizontal mode entered at line 1
\hbox(0.0+0.0)x20.0
\tenrm A
\glue 3.33333 plus 1.66498 minus 1.11221
spacefactor 999
### internal vertical mode entered at line 1
prevdepth ignored
### vertical mode entered at line 0
prevdepth ignored
|
\endexample
\enddesc
\see |\showboxbreadth|, |\showboxdepth| \ctsref\showboxbreadth.
\subsection{Specifying what is traced}
\begindesc
\cts tracingonline {\param{number}}
\explain
If this parameter is greater than zero,
\TeX\ will display the results of tracing
(including ^|\showbox| and ^|\showlists|)
at your terminal in addition to recording them in the log file.
\enddesc
\begindesc
\cts tracingcommands {\param{number}}
\explain
If this parameter is $1$ or greater,
\TeX\ will record in the log file most commands that it executes.
If ^|\tracingonline| is greater than zero, this information will also appear
at your terminal.
Typesetting the first character of a word counts as a command,
but (for the purposes of the trace only)
the actions of typesetting the subsequent characters
and any punctuation following them
do not count as commands.
If |\tracingcommands| is $2$ or greater,
\TeX\ will also record commands that are expanded
rather than executed, e.g., conditional tests and their outcomes.
\example
\tracingcommands = 1 If $x+y>0$ we quit.\par
On the other hand, \tracingcommands = 0
|
\logproduces
{vertical mode: the letter I}
{horizontal mode: the letter I}
{blank space }
{math shift character $}
{math mode: the letter x}
{the character +}
{the letter y}
{the character >}
{the character 0}
{math shift character $}
{horizontal mode: blank space }
{the letter w}
{blank space }
{the letter q}
{blank space }
{\par}
{vertical mode: the letter O}
{horizontal mode: the letter O}
{blank space }
{the letter t}
{blank space }
{the letter o}
{blank space }
{the letter h}
{blank space }
{\tracingcommands}
|
\endexample
\enddesc
\begindesc
\cts tracinglostchars {\param{number}}
\explain
If this parameter is greater than zero,
\TeX\ will record an indication in the log file of each time
that it drops an output character because that character does not exist
in the current font.
If ^|\tracingonline| is greater than zero, this information will also appear
at your terminal.
\PlainTeX\ defaults it to $1$ (unlike the others).
\example
\tracinglostchars = 1
A {\nullfont few} characters.
|
\logproduces
Missing character: There is no f in font nullfont!!
Missing character: There is no e in font nullfont!!
Missing character: There is no w in font nullfont!!
|
\endexample
\enddesc
\begindesc
\cts tracingmacros {\param{number}}
\explain
If this parameter is $1$ or greater,
\TeX\ will record in the log file the expansion and arguments
of every macro that it executes.
^^{macros//tracing}
If |\tracingmacros| is $2$ or greater,
\TeX\ will record, in addition,
every expansion of a \minref{token} list such as
|\output| or |\everycr|.
If ^|\tracingonline| is greater than zero, this information will also appear
at your terminal.
\example
\def\a{first \b, then \c}
\def\b{b} \def\c{c}
\tracingmacros = 2
Call \a once.
|
\logproduces
\a ->first \b , then \c
\b ->b
\c ->c
|
\endexample
\enddesc
\begindesc
\cts tracingoutput {\param{number}}
\explain
If this parameter is greater than zero,
\TeX\ will record in the log file the contents of every box that
it sends to the \dvifile.
^^{\dvifile//boxes recorded in log file}
If ^|\tracingonline| is greater than zero, this information will also appear
at your terminal.
The number of leading dots in each line of the trace output indicates
the nesting level of the box at that line.
You can control the amount of tracing by setting
^|\showboxbreadth| and ^|\showboxdepth| (\xref\showboxbreadth).
Setting |\tracingoutput| to $1$ can be particularly helpful when you're trying
to determine why you've gotten ^{extra space} on a page.
\example
% This is the entire file.
\tracingoutput = 1 \nopagenumbers
One-line page. \bye
|
\logproduces
Completed box being shipped out [1]
\vbox(667.20255+0.0)x469.75499
.\vbox(0.0+0.0)x469.75499, glue set 13.99998fil
..\glue -22.5
..\hbox(8.5+0.0)x469.75499, glue set 469.75499fil
...\vbox(8.5+0.0)x0.0
...\glue 0.0 plus 1.0fil
..\glue 0.0 plus 1.0fil minus 1.0fil
.\vbox(643.20255+0.0)x469.75499, glue set 631.2581fill
..\glue(\topskip) 3.05556
..\hbox(6.94444+1.94444)x469.75499, glue set 386.9771fil
...\hbox(0.0+0.0)x20.0
...\tenrm O
...\tenrm n
...\tenrm e
...\tenrm -
...etc.
..\glue 0.0 plus 1.0fil
..\glue 0.0 plus 1.0fill
.\glue(\baselineskip) 24.0
.\hbox(0.0+0.0)x469.75499, glue set 469.75499fil
..\glue 0.0 plus 1.0fil
|
\endexample
\enddesc
\begindesc
\cts tracingpages {\param{number}}
\explain
If this parameter is greater than zero,
\TeX\ will record in the log file its calculations of the cost of
various page breaks that it tries.
^^{page breaks//tracing}
If |\tracing!-online| ^^|\tracingonline|
is greater than zero, this information will also appear
at your terminal.
\TeX\ produces a line of this output
whenever it first places a box or \minref{insertion}
on the current page list, and also whenever it processes a potential
break point for the page.
Examining this output can be helpful when you're trying to determine
the cause of a bad page break.
See \knuth{pages~112--114} for an illustration and explanation of
this output.
Some production forms of \TeX\ ignore the value of |\tracingpages|
so that they can run faster.
If you need to use this parameter, be sure to use a form that
responds to it.
\enddesc
\begindesc
\cts tracingparagraphs {\param{number}}
\explain
If this parameter is greater than zero,
\TeX\ will record in the log file its calculations of the cost of
various line breaks that it tries.
^^{line breaking//tracing}
If ^|\tracingonline| is greater than zero, this information will also appear
at your terminal.
\TeX\ produces this output when it reaches the end of each paragraph.
See \knuth{pages~98--99} for an illustration and explanation of
this output.
Some production forms of \TeX\ ignore the value of |\tracing!-para!-graphs|
so that they can run faster.
If you need to use this parameter, be sure to use a form that
responds to it.
\enddesc
\begindesc
\cts tracingrestores {\param{number}}
\explain
If this parameter is greater than zero,
\TeX\ will record in the log file
the values that it restores when it encounters the end of a \minref{group}.
If ^|\tracingonline| is greater than zero, this information will also appear
at your terminal.
Some production forms of \TeX\ ignore the value of |\tracing!-restores|
so that they can run faster.
If you need to use this parameter, be sure to use a form that
responds to it.
\enddesc
\begindesc
\cts tracingstats {\param{number}}
\explain
If this parameter is $1$ or greater, \TeX\ will
include a report on the resources that it used to run your job
(see \knuth{page~300} for a list and explanation of these resources).
Moreover, if |\tracingstats| is $2$ or greater,
\TeX\ will report on its memory usage whenever it does a
^|\shipout| (\xref \shipout) for a page.
The report appears at the end of the log file.
^^{log file//tracing statistics in}
If ^|\tracingonline| is greater than zero, the information will also appear
at your terminal.
If you're having trouble with \TeX\ exceeding one of its
capacities, the information provided by |\tracingstats| may help you
pinpoint the cause of the difficulty.
Some production forms of \TeX\ ignore the value of |\tracingstats|
so that they can run faster.
If you need to use this parameter, be sure to use a form that
responds to it.
The following example shows a sample of
the tracing output you'd get on one implementation
of \TeX. It may be different on other implementations.
{\codefuzz = 1in
\example
\tracingstats=1
|
\logproduces
Here is how much of TeX's memory you used:
4 strings out of 5540
60 string characters out of 72328
5956 words of memory out of 262141
921 multiletter control sequences out of 9500
14794 words of font info for 50 fonts, out of 72000 for 255
14 hyphenation exceptions out of 607
7i,4n,1p,68b,22s stack positions out of 300i,40n,60p,3000b,4000s
|
\endexample
}% end scope of codefuzz
\enddesc
\begindesc
\cts tracingall {}
\explain
This command tells \TeX\ to
turn on every available form of tracing.
It also sets ^|\tracingonline| to $1$ so that the trace output will appear
at your terminal.
\enddesc
\begindesc
\cts showboxbreadth {\param{number}}
\explain
This parameter specifies the maximum number of list items
that \TeX\ displays for one level of one box when it is producing
the output for ^|\showbox| or ^|\tracingoutput|.
\PlainTeX\ sets |\showboxbreadth| to $5$.
\enddesc
\begindesc
\cts showboxdepth {\param{number}}
\explain
This parameter specifies the
level of the deepest list that \TeX\ displays when
it is producing the output for ^|\showbox| or ^|\showlists|.
\PlainTeX\ sets |\showboxdepth| is $3$.
\eix^^{tracing}
\eix^^{debugging}
\eix^^{diagnostic aids}
\enddesc
%==========================================================================
\subsection {Sending messages}
\begindesc
\bix^^{messages, sending}
\bix^^{error messages}
\cts message {\rqbraces{\<token list>}}
\cts errmessage {\rqbraces{\<token list>}}
\explain
These commands display the message given by \<token list> on your
terminal and also enter it into the log. Any \minref{macro}s in the
message are expanded, but no commands are executed. This is the same rule
that \TeX\ uses for |\edef| (\xref \edef).
For |\errmessage|, \TeX\ pauses
in the same way that it does for one of its own error messages
and displays the |\errhelp| tokens if you ask for help.
You can generate multiline messages by using the ^|\newlinechar|
character (\xref \newlinechar).
\example
\message{Starting a new section.}
|
\endexample
\enddesc
\begindesc
\cts wlog {\rqbraces{\<token list>}}
\explain
This command writes \<token list> on the log file.
^^{log file//written by \b\tt\\wlog\e}
\minrefs{log file}
\TeX\ expands \<token list> according to the same rules that it uses
for |\edef| (\xref\edef).
\example
\wlog{Take two aspirins and call me in the morning.}
|
\logproduces
Take two aspirins and call me in the morning.
|
\endexample
\enddesc
\begindesc
\cts errhelp {\param{token list}}
\explain
This parameter contains the token list that \TeX\ displays
when you ask for help in response to an |\errmessage| command.
We recommend that when
you're generating an error message with |\errmessage|, you
set |\errhelp| to a string that describes the nature of the
error and use |\newhelp| to produce that string.
You can use the ^|\newlinechar| character to produce multiline messages.
\enddesc
\begindesc
\ctspecial newhelp \ctsxrdef{@newhelp} {\<control sequence>
\rqbraces{\<help text>}}
\explain
This command assigns the ^{help message} given by \<help text> to
\<control sequence>. It provides an efficient way of defining
the ^{help text} that further explains an error message.
Before issuing the error message with the |\errmessage| command,
you should assign \<control sequence> to ^|\errhelp|. The help text
will then appear if the user types `|H|'
or `|h|' in response to the error message.
\example
\newhelp\pain{Your input includes a token that I find^^J
to be offensive. Don't bother me again with this^^J
document until you've removed it.}
\errhelp = \pain \newlinechar = `\^^J
% ^^J will start a new line
\errmessage{I do not appreciate receiving this token}
|
\logproduces
!! I do not appreciate receiving this token.
l.8 ...t appreciate receiving this token.}
? H
\Your input includes a token that I find
to be offensive. Don't bother me again with this
document until you've removed it.
|
\endexample
\enddesc
\begindesc
\cts errorcontextlines {\param{number}}
\explain
This parameter determines the number of pairs of context lines,
not counting the top and bottom pairs, that \TeX\ prints when it
encounters an error. By setting it to $0$ you can get rid of long
error messages.
You can still force out the full context by typing something like:
\csdisplay
I\errorcontextlines=100\oops
|
in response to an error,
since the undefined control sequence |\oops| will cause another error.
\PlainTeX\ sets |\error!-context!-lines| to $5$.
\enddesc
\see |\write| (\xref \write), |\escapechar| (\xref \escapechar).
\eix^^{messages, sending}
\eix^^{error messages}
%==========================================================================
\section {Initializing \TeX}
\begindesc
\cts dump {}
\explain
This command, which must not appear inside a group, dumps the
contents of \TeX's memory
to a ^{format file} (\xref{format file}).
By using ^|virtex|, a special ``virgin'' form of \TeX,
you can then reload the format file at high speed and
continue in the same state that \TeX\ was in at the time of the dump.
|\dump| also ends the run. Since |\dump| can only be used
in ^|initex|, not in production forms of \TeX, it is only useful
to people who are installing \TeX.
\enddesc
\begindesc
\cts everyjob {\param{token list}}
\explain
This parameter contains a \minref{token} list that \TeX\ expands at the
start of every job. Because an assignment to |\everyjob| cannot affect
the current run (by the time you've done the assignment it's already too
late), it is only useful to people who are preparing format files.
\enddesc
\enddescriptions \endchapter \byebye