path: root/Master/texmf-dist/doc/generic/components-of-TeX/etexkomp.tex

% etexkomp.tex                                  25 Mar 91
%------------------------------------------------------------
% (c) 1990 by J.Schrod (TeXsys).

%
% originally published in Die TeXnische Komoedie (DANTE e.V.) as:
%     Die Komponenten von TeX
%
% ILaTeX with standard styles and
%	style for reports about TeX

% Changes made by Adrian Clark for publication in `Baskerville'
% and editorial remarks by Malcolm Clark
% incorporated.


\documentstyle[texrep]{article}


% terms `output' and `driver' in english, used in picture `figtotal'.
\def\OutputInFigtotal{output}
\def\DriverInFigtotal{\DVI{} driver}



\begin{document}


\title{The Components of \TeX{}}
\author{
   Joachim Schrod\\
   Detig$\,\cdot\,$Schrod \TeX{}sys
   }
  \date{March 1991}

\titlenote{
   \copyright{} Copyright 1990, 1991 by Joachim Schrod.
   All rights reserved.
   Republication and Distribution is allowed under conditions analog
   to the GNU General Public License.
   Previous revisions of this paper appeared in ``Die \TeX{}nische
   Kom\"odie'' and in ``Baskerville.''
   }
\titlenote{
   {\sl Actual address\/}:
   Joachim Schrod, Detig$\,\cdot\,$Schrod \TeX{}sys,
   Kranichweg 1, D-6074 R\"odermark, FR~Germany,
   Email: {\tt xitijsch@ddathd21.bitnet}
   }

\maketitle



 \begin{abstract}
 \TeX{} needs a great amount of supplementary components (files and
programs) of which the meaning and interaction often is unknown. This
paper explains the components of the kernel system \TeX{} that are
visible for the \TeX{} user and their relations.
 \end{abstract}




\section{About this Report}

\TeX{} is a typesetting system which offers authors easy usage of
powerful typesetting features to produce printed matter which is the
state of the art of computer typesetting. This is, however, not done by the
\TeX{} program alone: A significant number of supplementary
programs and files together form the complete typesetting and
authoring system. Along with the programs that belong to \TeX{}
directly, there exist two other major programs which were built by {\sc
Donald Knuth} in connection with \TeX{} and must be included in an
explanation of the full system: \MF{}, for the generation of fonts,
and \WEB{}, a documentation and developing `language' for programming.
\TeX{} and \MF{} are written in \WEB{}.

This text describes this `kernel' \TeX{} from a user's viewpoint: at
the end you should have an overview of the ingredients of the \TeX{}
system, and about the files and support programs that are essential
for you as a user. This will not, however, be an introduction to the
capabilities of \TeX{} or how you may run \TeX{} on your computer.

I will use marginal notes to identify the places where terms are
explained for the first time. Abbrevations for file types -- usually
identified by common suffixes or extensions -- are set in a monospace
type (``{\tt typewriter\/}'') and those abbrevations are put into the
margin, too. Please note that these abbreviations are sometimes not
identical with the file extensions (see also
Table~\ref{tab:filetypes}).

This report is the start of a series that describes the subsystems
mentioned above and their respective components. In that series each
report will focus on one subsystem in a special point of view; it
should not result in gigantic descriptions which tell everything (and
then nothing). In my opinion the following reports will be of
interest:
 %
 \begin{itemize}

\item the structure of a standard installation of \TeX{}
\item \DVI{} drivers and fonts
\item possibilities of graphics inclusion in \TeX{} documents
\item the components of \MF{}
\item the structure of a standard installation of \MF{}
\item \WEB{} systems --- the concept of Literate Programming
\item other (though not yet planned) themes of interest are perhaps
\begin{itemize}
\item differences between \TeX{} and DTP systems
\item the way how \TeX{} works (there exists some good books on this topic!)
\item the limits of \TeX{}
\item \TeX{} as a programming language
\end{itemize}

 \end{itemize}
 %
 The reports will be published in this sequence.




\section{What is \TeX{}?}

 \TeX{} is a typesetting system with great power for the typesetting
of formulae. Its basic principle is that structures in the document
are marked and transformed into typeset output. Providing such
information about the structure of a document is known as {\it
^|markup|}. If the marks describe the look of the document, it is
called {\it ^|optical markup|}, while, if document structures are
marked, it is called {\it ^|logical markup|}. \TeX{}  provides both
forms of  markup, \ie{}, exact control of the layout of parts of the
document and their positioning as well as the markup of the structure
of formulae or document components. The logical markup is mapped to
the optical one by \TeX{} so that layout may serve for the
identification of structures by the reader.%
   \footnote{
      Layout -- and book design in general -- do not represent useless
beauty. A good book design must first support the understanding of
the content to produce readable text. So it is {\it
\ae{}sthetic\/} in its best sense, since it connects form and contents
and builts a new quality.
      }

 The kernel of the \TeX{} typesetting system is the formatting program
\TeX82^^|typesetting machine \TeX82|, which is often simply called
\TeX{}. This usage shall be adopted here whenever the difference
between the complete system and the formatter is unimportant or
obvious. \TeX82 is a big monolithic program which is published in the
book {\it \TeX{}: The Program\/} by {\sc Donald Knuth}. Its features
may be separated into two levels:
 %
 \begin{enumerate}

\item \TeX82 formats text, \ie{}, it breaks it into paragraphs
(including automatic hyphenation) and produces page breaks.

\item It provides the programming language \TeX{} which incorporates a macro
mechanism. This allows new commands to be built to support markup
at a higher level. {\sc Donald Knuth} presents an example in the
\TeX{}book: \Plain{}. A collection of macros which supports a special
task and has (hopefully) a common philosophy of usage is called a
{\it ^|macro package|}.

\end{enumerate}
 %
 High level features for optical markup, as represented by \Plain{},
allow one to build additional levels leading to full logical markup.
At the moment, two macro packages for logical markup are widespread:
\AmSTeX{} and \LaTeX{}. Both systems are built on top of \Plain{} to
greater or lesser extents and the user can use the optical markup of
\Plain{} in addition to logical markup if desired. This results in
the effect that the author can use a mixture of structural
information and explicit layout information -- a situation with a
high potency of features that nevertheless can (and does) lead to a
lot of typographic nonsense!

As \TeX82 was built only for typesetting texts and to allow the
realization of new markup structures, many features are lacking which are
required by authors. To provide features like the production of
an index or a bibliography or the inclusion of graphics, additional
programs^^|supplementary programs| have been written, which use 
information from a \TeX82 formatting run, process them, and provide
them for the next \TeX82 run. Two supplementary programs are
in widespread use and available for many computer/operating system
combinations: \BibTeX{}, for the production of a bibliography from a
reference collection, and \MakeIndex{}, for the production of an index.

A special case of the processing of information provided by a \TeX{}
run is the production of a table of contents or the usage of cross
references in a text. For this only informations about page numbers,
section numbers, etc., are needed. These are
provided by \TeX82 and can be processed by \TeX82 itself, so \TeX82 is
used as its own post processor in this situation.

We have now seen that the \TeX{} typesetting system  is
a collection of tools that consists of the typesetting `engine' \TeX82,
macro packages (maybe several that are based on others) and
supplementary programs, used together with these macro packages.
This relation is illustrated by Figure~\ref{fig:components}.

\begin{figure}
\begin{center}
\input figkomp
\caption{The Components of \TeX{}}
\label{fig:components}
\end{center}
\end{figure}



\section{Formatting}

The formatting process of \TeX{} needs information about the
dimensions of characters used for the paragraph breaking. A set
of characters is grouped in {\it ^|font|s}. (But this is a
simplification as the notion ``font'' should be used for the
realization of a type in a fixed size for a specific output device.)
The dimensions of the characters of a font are called {\it font
metrics}.


The format in which the font metrics are used by \TeX{} was
defined by {\sc Donald Knuth} and is called  \ftype{tfm} format
(``{\sl \TeX{} font metrics\/}''). In this format, every character is
descibed as a {\it box\/} with a height, a depth, and a width.
\TeX{} only needs these measurements, it is not interested in the shape
of the character. It is even possible that the character may extend
outside 
the box, which may result in an overlap with other characters.
The character measures are specified in a device independent
dimension because \TeX{} processes its breaking algorithm
independent of any output device.

During paragraph breaking, \TeX{} hyphenates automatically, which
can be done in an almost language-independent way. For the adaption to
different languages, {\it ^|hyphenation patterns|\/} are needed to
parametrize the hyphenation algorithm.

The result of a \TeX{} formatting run is a \ftype{DVI} document, in which
the type and position on the page are specified for each character to
be output. The resolution that is used is so small that
every possible output device will have a coarser raster, so that the
positioning is effectively device independent. The \DVI{} document
specifies only types, not the fonts themselves, so that the name \DVI{}%
   \footnote{
      This name is a problem because ``DVI'' is a trademark of 
      Intel Corp.\ now, but the name \DVI{} for \TeX{} output files 
      pre-dates this.
      }
 (``{\sl device independent\/}'') is accurate. To make the result of
the formatting run available, the \DVI{} file must be output by a
so-called ^|\DVI{} driver| on the desired output device.

If problems occur during the formatting, error messages or warnings
are output on the terminal. {\it Every\/} message that appears on the
terminal will also be written into a protocol file named \ftype{log}
file. In this \LOG{} file additional information may be placed that
would have been too verbose for the output to the terminal. If this
is the case, \TeX{} will tell the user so at the end of the
formatting run. The messages of \TeX{} are not built in the program,
they are stored in a (string) \ftype{pool} file. These messages must
be read in at the beginning of a run.



\section{Macro Packages}
The basic macro package is ^|\Plain|, developed by {\sc Donald Knuth}
together with \TeX82. It parametrizes the \TeX82 typesetting machine
so that it can typeset English texts with the Computer Modern type
family. Additionally, \Plain{} provides optical markup features.
\Plain{} is available as one source file, {\tt plain.tex}.

All other macro packages known to the author are based on \Plain{},
\ie{}, they contain the source file {\tt plain.tex\/} either
originally or with modifications of less important parts. Next to
\Plain{}, the most important (free) available macro packages are
^|\AmSTeX{}| by {\sc Michael Spivak} and ^|\LaTeX{}| by {\sc Leslie
Lamport}. Other free macro packages are often of only local
importance (\eg{} Blue\TeX{}, TEXT1, or \TeX{}sis) or are used in
very special environments only (\eg{} {\tt texinfo\/} in the
GNU~project or {\tt webmac\/} for \WEB{}). Important commercial macro
packages are ^|Macro\TeX{}| by {\sc Amy Hendrickson} and
^|\LAMSTeX{}|, also written by {\sc Michael Spivak}.

These macro packages usually consist of a kernel that
provides additional markup primitives. With such primitives,
{\it document styles\/}^^|styles| can be built which realize logical
markups by a corresponding layout. This layout can often be varied by
{\it sub-styles\/} or {\it style options\/} which may also provide
additional markups.

The macro packages produce {\it ^|supplementary files|\/} which contain
information about the page breaks or the document markup. This
information may be used by support programs -- \eg{}, the
specification of a reference from a bibliography database or the
specification of an index entry with corresponding page number for the
construction of an index. A special case is the information about
cross references and headings for the building of a table of contents,
as this information can be gathered and reused by \TeX{} directly.

^|\SliTeX{}| is a special component of \LaTeX{} for the preparation
of slides with overlays. In TUGBoat volume~10, no.~3 (1989)
^|\LAMSTeX{}| was announced, which will provide the functionality of
\LaTeX{} within \AmSTeX{}. ^|Macro\TeX{}| is a toolbox of macro
``modules'' which may be used to realize new markups but, as it
became available only short time ago, it is not yet widespread.

For the usage of these (and other) macro packages, one must check
whether they need ^|additional fonts| which do not belong to the Computer
Modern type family. For \LaTeX{}, \eg{}, fonts with additional symbols
and with invisible characters (for the slide overlays) are needed,
while \AmSTeX{} needs several additional font sets with mathematical and
Cyrillic characters.



\section{Support Programs}

Only two support programs will be discussed here: ^|\BibTeX{}| by
{\sc Oren Patashnik} for the preparation of bibliographies and
^|\MakeIndex{}| by {\sc Pehong Chen} and {\sc Michael Harrison} for
the preparation of a sorted index. For both tasks exist other,
funcionally equivalent, support programs. But the abovementioned are
available on many operating systems, and have an ``official'' state
as they are {\sc Leslie Lamport} encourages their usage with
\LaTeX{} in his documentation, and the TUG supports them for general use.

There is no totally portable mechanism for the inclusion of general
graphics in \TeX{} documents, so that there are no machine independent
support programs available.

\BibTeX{} is used to handle references collected in \ftype{bib}
files. \TeX{} produces supplementary files which contain information
about the required references, and \BibTeX{} generates from them a
sorted bibliography in a \ftype{bbl} file which may be subsequently
used by \TeX{}. The kind of sorting and the type of ^|cite keys| are
defined by {\it ^|bibliography styles|\/},  specified in \ftype{bst}
files. The messages of a \BibTeX{} run are written to a \ftype{blg}
logfile.

\MakeIndex{} reads an \ftype{idx} support file that contains the index
entries and the according page numbers, sorts these items, unifies
them and writes them as \TeX{} input in an \ftype{ind} file. The
formatting style may be specified by an {\it ^|index style|}. The
messages of a \MakeIndex{} run are written to a \ftype{ilg} file.



\section{Performance Improvements}

Much of the work that \TeX82 has to do is the same for every
document:
%
 \begin{enumerate}

\item \label{enum:hyphen}
 All text has to be broken into lines. Text pieces in the same
language are hyphenated with the same hyphenation patterns.

\item \label{enum:markup}
 The basic markups of the corresponding macro packages must be
available.

\item The required font metrics are much alike for many documents, as the
font set used usually doesn't differ that much.

\end{enumerate} 
% 
 To improve \TeX{}'s performance, hyphenation, markup, and font
metrics descriptions are converted from an external, for
(\ref{enum:hyphen}) and~(\ref{enum:markup}) textual, representation
into an internal representation which can easily be used by \TeX82.
It is sensible to do this transformation only once, not for every
document. The internal representation is stored in a \ftype{fmt} file.
The storing is done with the \TeX{} command {\tt \bs{}dump}, so that
\FMT{} files often are called ``{\it dumped formats}.'' A \FMT{} file
can be read at the beginning of a \TeX82 run and is thus available
for the processing of the actual text.

As the creation of a \FMT{} file is done infrequently -- usually for
the update of a macro package -- the formatting of texts can be done
with a reduced version of the \TeX82 program that doesn't contain the
storage and the program parts for the transformation of the
hyphenation patterns and for the dumping. The complete version of
\TeX82 is needed in an initialization phase only and therefore called
^|\INITeX{}|. Additional improvements of the performance can be
reached  by the usage of ^|production version|s of \TeX82 from which
parts for statistical analysis and for debugging are stripped. 

\TeX{} versions that have no dumped formats preloaded, have the
ability to load a dumped format (\ie{} a \FMT{} file), and have no
ability to dump a \FMT{} file (\ie{}, they are not \INITeX{}) are
often called ^|Vir\TeX{}|, which stands for {\it virgin \TeX{}}.



\section{Connections Between File Types and Components}

In the above sections, the components of the \TeX{} authoring system 
were described, and the files that are read or written by these
components mentioned. The connections between them all is 
demonstrated graphically in Figure~\ref{fig:total}. In this graphic,
file types are represented by rectangles, and programs by ovals. The
arrows mean ``is read by'' or ``is produced by.'' The abbreviations
of the file types are explained in Table~\ref{tab:filetypes}, which
also lists the file identifications (suffixes or extensions) that
these files usually have (but note that  other file identifications
are also in use).

\begin{figure} 
\begin{center} 
\input figtotal 
\caption{The Connection of Components and File Types} 
\label{fig:total} 
\end{center}
\end{figure}

\begin{table}
\begin{center}

\noindent
\vbox\bgroup			% halign, LaTeX is too inflexible!!
%
\tabskip=1em
\catcode`\,=\active % Roman Komma, also in typewriter text!
\def,{{\rm \char`\,\relax \space}\ignorespaces}

\let\par=\cr \obeylines%
\halign{\tt \hfil #\unskip\hfil & #\unskip\hfil & \tt #\unskip\hfil
%
\sc File Type & \hfil \sc Explanation & \sc File Identification
              &                       & \sc (suffix, extension, etc.)
\noalign{\vskip 3pt}%
%
TEX & Text input & tex, ltx
DVI & \TeX82 output, formatted text & dvi
LOG & \TeX82 log file & log, lis, list
HYP & Hyphenation patterns & tex
TFM & Font metrics & tfm
POOL & String pool & pool, poo, pol
FMT & Format file & fmt
MAC & \TeX{} macro file & tex, doc
STY & \TeX{} style file & sty, tex, st, doc
AUX & Support files & aux, toc, lot, lof,\unskip
    &               & glo, tmp, tex
BIB & Reference collections & bib
BBL & References or bibliographies & bbl
BLG & \BibTeX{} log file & blg
BST & \BibTeX{} style file & bst
IDX & Unsorted index & idx
IND & Sorted index & ind
IST & Index markup specification
ILG & \MakeIndex{} log file & ilg
}%
%
\egroup

\caption{File Types}
\label{tab:filetypes}
\end{center}
\end{table}



\subsubsection*{Acknowledgements}

I would like to thank {\sc Christine Detig} who was so kind to
provide the English translation. {\sc Nelson Beebe} suggested
performing the  translation. {\sc Klaus Guntermann} made valuable
comments on the first (German) version.
{\sc Nico Poppelier} contributed a new version of figure
\ref{fig:total}, better than my original one.



\end{document}