⟦caf89868d⟧

TextFile

\subsection{DVI File Structure}
A \dvi\ (`DeVice Independent') file
contains a stream of special codes
that instruct conversion programs
in the art of building output pages.
Most of these details are irrelevant to anyone else,
but a few are especially important.
A \dvi\ file has three major sections.
It starts with a {\em preamble\/}
listing particular constants used in the file.
After this comes the main body, as a series of separate pages.
Finally, there is the {\em postamble},
which summarises things about the file overall.
Many \dvi\ file readers look first at the postamble,
since it has most of what they need to know.
If you hand a plain file to a \dvi\ reader,
it should complain about a missing postamble
or missing preamble.

\subsubsection{Magnifications}\label{sec:dvimag}
Each \dvi\ file begins with a {\em magnification},
set with the \TeX\ primitive \verb|\magnification|.
Given a magnification factor $m$,
the document will be printed at $m/1000$ times normal size.
Thus, values larger than 1000 cause the document to expand;
values smaller than this make it shrink,
and the default value is of course 1000.
This is sometimes useful for obtaining extra resolution.
For instance, if you have a copier that reduces to 77\% original size,
a magnification of $1000/.77$, or 1299,
would produce output that, when shrunk,
would appear normal size.
(Note that this magnification
applies to {\em all\/} the dimensions in the \dvi\ file,
and is not like \LaTeX's \verb|\documentstyle| options
for 11 or 12 point output.)
Most \dvi\ drivers
allow a second magnification to be applied
in addition to that in the \dvi\ file.
For instance, by applying another magnification of 770,
you could make a document with a \verb|\magnification| of 1299
appear at normal size ($1.299 \cdot 0.77 \approx 1.0$).

\subsubsection{Fonts}
After the \verb|\magnification|,
a \dvi\ file lists the {\em fonts\/} it needs.
A font is a collection of {\em glyphs\/} (character shapes)
along with details about how to print them on paper
(or on a video screen).
\TeX\ fonts come in several varieties,
but the ones you are likely to encounter are called
`\gf' and `\pk' fonts.
GF and PK fonts are directly interconvertible
and can usually be regarded as identical.
These examples will use \pk\ fonts.

Fonts, too, come in various magnifications.
The standard set of magnifications is:
none (unmagnified),
\verb|\magstephalf| (1.095 times normal size),
\verb|\magstep1| (1.2 times normal size),
\verb|\magstep2| (1.44 times normal size),
and so on,
up through \verb|\magstep6| (2.986 times normal size).
Each \verb|\magstep| is 1.2 times bigger than the previous one.
What this means to you
is that your system must have each font in eight different sizes
for you to use any font at any standard magnification.
Most systems do not have every font in every size.
When you come across one that is not in the size you want---for
instance, were you to ask for \verb|cmssqi8 scaled \magstep3|,
it would probably not be there---you may get an error like this:
\begin{quote}
\begin{verbatim}
Warning: no font for cmssqi8 scaled 1728
(wanted, e.g., /usr/local/lib/tex/fonts/cmssqi8/cmssqi8.518pk)
\end{verbatim}
\end{quote}
If this happens, the \dvi\ driver may quit\footnote
{In this case the word {\tt warning} will be omitted.}
(possibly after noting other missing fonts)
and leave you to solve the problem yourself.
You can often do this by choosing some other font in your \TeX\ file,
or you may be able to get a local guru
to generate the font in your size.
You can also substitute a font that is nowhere near the right size
by creating a file---typically a link or a symbolic link---with the
appropriate name (here {\tt cmssqi8.518pk}),
either in your own directory
or in the system font directory.
The output will, however, tend to look bad.
On the other hand, the \dvi\ driver may add this admonishment:
\begin{quote}
\begin{verbatim}
I will substitute boxes for the missing glyphs
\end{verbatim}
\end{quote}
In this case, you will get little square boxes
like those in Chapter~11 of {\em The \TeX book\/}---or,
if the message said {\tt white space} instead of {\tt boxes},
just blank space---instead of glyphs from your font.
(Which you get is up to the person who does the installation.
For all the gory details,
including how \verb|\magstep3| translates to {\tt .518pk},
see \S\ref{sec:fontdesc} on p.~\pageref{sec:fontdesc}.)

\subsubsection{Count Registers}\label{sec:dvicount}
Each individual \dvi\ page is marked with ten counters,
known as \ccount0 through \ccount9.
These are the numbers \TeX\ prints in square brackets
when it formats your document.
Normally, \ccount0 holds the page number
and the rest of the counters are not set (are zero).
The \dviselect\ program described in \S\ref{sec:dviselect}
on p.~\pageref{sec:dviselect} makes use of these counters.
You can make \LaTeX\ put the current chapter number of a book
or report
into \ccount1 by adding the magic invocation
\begin{quote}
\begin{verbatim}
\makeatletter \countdef\c@chapter=1 \makeatother
\end{verbatim}
\end{quote}
between your \verb|\documentstyle| declaration and your
\verb|\begin{document}| command.\footnote
{Such tricks for use with \dviselect\ should be agreed-upon
and put into standard \LaTeX\ style files,
where the {\tt\string\makeatletter}
and {\tt\string\makeatother} invocations
would be unnecessary (and in fact, would be wrong).
As of yet, this has not happened.}

\subsubsection{Drift}\label{sec:drift}
\TeX\ imagines that it can put any character anywhere on a page,
with a resolution of less than one millionth of an inch,
which is more precise than anyone could measure.
Real printers and displays cannot do this.
To make the output look right,
\dvi\ drivers are required to follow information kept in each font
as to where to put characters,
and this sometimes disagrees with where \TeX\ put the characters.
Normally, the disagreement is negligible.
In some peculiar cases, however, it can build up,
and the characters would `drift away' from where they should be.
Because of this,
the disagreement is always limited to a small number of pixels
(individual dots on the printer or screen).
This limit is called the {\em maximum drift},
often abbreviated to {\em maxdrift}.
The average low-resolution (300 dot per inch) printer driver
will limit the drift to three pixels, or $1/100$ of an inch.
This can sometimes be changed on the command line.
Normally you will never notice, or care.

\subsubsection{Margins}
\TeX\ documents are supposed to be printed with a one inch
top margin and a one inch left margin.\footnote
{Yes, there really {\em is\/} a standard for \dvi\ converters,
and it really {\em does\/} say this,
despite international standards and the metric system.}
That is, the \dvi\ coordinate $(0,0)$
is supposed to correspond to the position one inch away
from the top left corner of the page,
measured horizontally and vertically.
\TeX\ and \LaTeX\ may add more margin space on top of this,
but this amount is built into each \dvi\ driver.
Most of these programs have options
to override or adjust the amount of margin.
Typically you will need this
only if you are printing on special paper.
Be aware, though, that some laser printers
move the $(0,0)$ point away from the corner of the page
to try to hide the fact that they cannot print along the edges,
and you (or the system administrator, during installation)
may have to compensate for this.

Incidentally, many `300 dot per inch' laser printers
are not exactly 300 dots per inch;
a typical low-cost printer
might have 297 dots per inch in one direction,
and 302 dots per inch in the other.
Not only does this vary from one printer to the next,
it may also vary on the same printer
due to changes in temperature, humidity, and the like.
You may be able to tweak a printer into alignment
if you really expect perfection,
but it probably will not stay perfect for long.\footnote
{For real precision,
you need something with more resolution anyway.
At the time of this writing,
the University of Maryland charged \$2 per foot
for printing on a Mergenthaler 1000-line-per-inch phototypesetter,
while laser printer output cost about \$.15 per page,
and previewing was free.
Those typesetting books
were advised to preview and to use laser printers
until they were sure that the book was in its final form.}

\subsubsection{Specials and Graphics}
\TeX\ can be told to pass arbitrary strings
on to the \dvi\ reader.
The construct \verb|\special{hello}|, for instance,
passes the string {\tt hello}.
It is up to the \dvi\ conversion program
to make sense out of these strings.
Unfortunately,
there is no standard for meanings or even for formats of \verb|\special|s,
and the result has been chaos.
Each attempt to set some standard
has foundered,
and it seems unlikely that the situation will improve soon.
What this really means
is that any \verb|\special|s you use
may well tie you to a particular kind of printer.
You may not be able to preview their effects.
Nonetheless, there is no substitute for real graphics
of the sort available only with \verb|\special|.
Just be aware that your pictures will not be as portable
as the rest of your text.
DataMuseum.dk

DKUUG/EUUG Conference tapes

⟦caf89868d⟧ TextFile

Derivation

TextFile
