\providecommand{\ie}{\emph{i.e.}}
\providecommand{\ctan}{\textsc{Ctan}}
\providecommand{\cd}{\textsc{CD-rom}}
\providecommand{\unix}{\textsc{Unix}}
\providecommand{\dos}{\textsc{Dos}}
\providecommand{\ie}{\emph{i.e.{}}}
\let\textss\textsf
\title{Malcolm's Gleanings}
\author[Malcolm Clark]{Malcolm Clark\\
Computing Services, University of Warwick\\
Coventry, CV4 7AL\\\texttt{m.clark@warwick.ac.uk}}

\begin{Article}
\section{True colours?}
The boost given to the use of colour in \LaTeXe\ is not an unmixed
blessing.  There are pitfalls and problems lurking round these new
corners.  There is always a huge discrepancy between the colours we
can produce on a monitor screen and the hard copy version that a
printer will produce. It can be very frustrating to spend time and
effort getting the colour balances `just right', and then finding that
the hard copy looks nothing like the original. Especially when you
have to pay for the hard copy. Why are there discrepancies? The
mismatch is in the very nature of the processes involved -- it is not
because something has gone wrong.

Let's start at the beginning, and assume that we are concerned with
the transition from colour on the screen to colour on the page.
Perhaps the first thing we should realise is that no two people see
colour in quite the same way. Intra-species perception is well-known
to be different, but even between individuals (ignoring those with
some physiological perceptual malfunction, like colour blindness),
response to different colours is different -- \ie\ what different
people call `blue' differs -- well, most people agree about blue, but
how many agree about `mauve' or `puce'? Any perceived colour may be
understood as the sum of three primary colours. This gives us a first
clue, since it means that we can represent a colour as the sum of
three stimuli (a tristimulus), and we can think in terms of
characterising any colour by locating a unique point defined by three
(orthogonal) axes. In fact no monitors can represent the full range
(or \emph{gamut}) of colours. Their actual performance falls short.
Even before we start, we cannot hope to represent all possibilities.

The colours we view on a monitor are emitted colours -- they are
generated by the three electron guns of the device (there are other
technologies, but the principle remains much the same). The advantage
here is that each model of monitor can be calibrated, and in fact the
range of possible colours can be determined. From time to time the
colours may drift, and it is possible that adjacent `identical'
machines may appear to have different colour outputs. If you always
work on the same machine, you can guard against this. Recalibrating
every machine is rather time consuming. Another advantage of using the
same machine every time is that the background and lighting is
unlikely to change to much.

By virtue of our evolution, our eyes are well adapted to slow changes
in light level, and also to slow, but moderate changes in the major
illuminant (as daylight varies through the day, for example). Changing
the incident lighting does not change the emitted light on the
monitor, but since the frame of the monitor and everything else in the
room is changed by changing the illumination, you can end up thinking
that it looks different. The illumination in many rooms is a bit
arbitrary: it may be fluorescent tubes, which have a rather limited
spectrum, or it may be normal incandescent lights, which have a
different spectrum, and occasionally it is north light (also known as
\emph{daylight}) illumination.

This last is an interesting diversion. This is the lighting condition
under which most, if not all, standardised colour determinations are
performed. Anyone who is working seriously with colour will use north
light (most artist's studios, for example, will be oriented to be lit
naturally in this way). In fact, this is one of the reasons why the
hard copy will appear incorrect: the lighting conditions are likely to
have changed. And anyway, the hard copy will look different under
different illumination. The hard copy is made visible by reflected
light.  What is reflected is dependent on the light source. The
distortions of sodium street lights are well known, but the same sort
of effect is true of fluorescent or incandescent lights. The north
light works reasonably well as a standard because of our environmental
background, and because it is based on the light source which was
available as our eyes evolved.

There is at least one more factor to be considered: the printer. Just
as a monitor can only produce a proportion of the possibilities, so
too the printer is restricted in its range of possible colours. In
fact it is worse: while the monitor can use the linear additivity of
the primaries, the printer's primaries are anything but linear, and
the mixing is subtractive.  Adding the primaries on a monitor gives
white: on the printer it will give a theoretical black (more of a
muddy purplish darkness). It is therefore rather difficult to
transform from the co-ordinates representing the colour on screen to
some faithful (whatever that means) rendering on the page. There are
lots of dodges which can be adopted. The most comprehensive is to
employ some sort of lookup table. This depends on the measurement of
many samples -- it is a lot of rather tedious work, spread over lots
of subjects (remember the underlying psycho physical variation).

Briefly then:
\begin{enumerate}
\item the colours you see on the monitor are a subset of possible
  colours;
\item the illumination of the room can be important
\item the printer does not have the capability of rendering all the
  colours visible on the monitor;
\item you will probably view the hard copy under a variety of lighting
  conditions.
\end{enumerate}

Is there any hope? Some printers are better than others. The machine
we use here at Warwick, a Tektronix Phaser, is essentially a wax
crayon machine; it is reasonably good, for the price.  The best hard
copy devices are dye sublimation printers and they are very expensive.
If you stick to primary colours, and colour gradation is not a key
issue, the Phaser is excellent.  With a lot of time and effort you
might achieve finer quality work with it. If you are looking for a
photographic quality reproduction you would have to use other
technologies -- and even then, you can fail. In many cases though,
hard copy is irrelevant. The display medium is genuinely the monitor
screen. If we think in this way, we realise that many other
possibilities are introduced, like running video clips, or even pieces
of software, within a document or presentation (yes, \LaTeXe\ does not
yet support such extensions explicitly, but a suitable
\texttt{\char'134special} could). We usually have the feeling that
paper is the objective, but it does not have to be. Sometimes it is
more convenient, but at other times it is restrictive.

Any of the good ideas for this came from presentations by Chris Lilley
of the Computer Graphics Unit at Manchester Computer Centre; the
mistakes are mine.

\section{Trivial pursuit}
For many years I have been toying with the notion of creating a
special \TeX\ edition of \emph{Trivial Pursuit}. There are those who
contend that \TeX\ \emph{is} a trivial pursuit, but I will have
nothing to do with this view. To assist in the creation of this game,
and to tempt the likes of Waddington's, I will start to include
examples of the questions in this column. The first question is in two
parts. The first part is borrowed from the ACM 1994 Computer Bowl
quiz: ``Only one person in all of computing's history has ever won
both the ACM Turing Award for lasting technical achievement and the
Grace Murray Hoppper Award for work done prior to reaching the age of
thirty. Who is that person?''

The second part is this: ``The Computer Bowl quiz was created and
produced by the Computer Museum, Boston. What link does this have to
the answer for the previous question?''

\section{Dutch bearing gifts}
I quickly snapped up my NTS \cd{} when it was offered through the
group.  I was motivated by a curious mixture of support for the
efforts of another group, and a willingness to get tangled up in a
flashily trendy technology, although 
I have to admit that I'm not yet convinced by \cd{} technology. 
%But I have to admit that I'm not yet
%convinced by \cd{}. It seems to me like a suspect technology which is
%only half way there, or one which has been rather over-hyped. While
%the shiny disks can contain a fair amount of data -- 640 Mbytes --
%this seems to me to be pretty inadequate for multimedia applications
%which include still or moving pictures, and the access times for
%random movements around the disk are appalling (almost as bad as World
%Wide Web connections around the InterNet). It is fine for sequential
%text when it is read in a linear fashion. In other words, pretty
%boring stuff. However, as an archiving mechanism it offers some
%potential, ignoring the extreme difficulty with which additional
%material can be added.

One advantage of the NTS \cd{} is the booklet
which comes with
it. Once you have installed the suite, and basically it wants to run
from the \cd{} drive, you find that it is enormously configurable.
Because of the many varieties of printers, screens and editors which
are available in \dos{} systems, you could have many happy hours ahead
of you getting things right. I get the feeling that it is assumed that
you will like fiddling. There is no doubt that great effort went into
this, and equally, that you can eventually install all the relevant
bits and pieces. I wonder how you would manage to install something
useful if you didn't want to use the 4\dos{} shell which is the
default. Clearly the developers thought this an unlikely decision.
They say ``One might object that using 4\dos{} batch files deprives
the old-fashioned \texttt{command.com} users from the benefits of
4\TeX. We happen to think that this would only be a mild punishment
for not recognizing how good 4\dos{} really is.'' Well, pardon me!
Oddly, they fail to mention that continued use of 4\dos{} involves
payment of a shareware fee.

Once you do get it installed, you still appear to have great power to
customise the options the shell offers. However, I am still unclear
how to change the drivers (you are provided with a variety of screen
drivers and I really wanted the one which occupied least memory; this
is not the default). I was also surprised that among the huge range of
choices of formats you are offered, the simple \LaTeXe\ or plain \TeX\ 
is not among them. It is rather like going into Macdonald's and trying
to get a burger without the limp lettuce or the ketchup. In this case
of the formats, the lettuce and ketchup is Babel.

On the whole, I suspect the shell is about as good as you can get with
\dos{}. It appears workmanlike. I'm afraid it does look dated though.
Windows is here. The true \textsc{blu}e \TeX ie may resent it, and see
it as a step towards perdition, but for many of us Windows is the
least unacceptable face of the Intel chip's range of operating
systems.

But compared to PrimeTime's \cd{}, this is a joy. At least the NTG's
offering has the advantage that you can browse it a bit like you
browse \ctan. If there is a useful file there, you can find it and
pull it out. The PrimeTime \cd{} is tarred and feathered (or zipped)
so that you can't actually find anything unless you know where it is.
It contains almost all the contents of \ctan, but nowhere does it
actually give you the date of this snapshot.  The sheer mass is
daunting. And layered on the top of this is an assumption that \unix{}
is the way the truth and the life. It makes a point about distributing
source code: ``Binary-only distribution prevents recipients from
modifying or learning from the internals of software.'' I have to
confess that I'm not terribly interested in computer programs.  I'm
much more interested with what I can do with the program. The days of
immersing yourself in the exciting details of source code should
surely be passing. The terms `nerd' and `propellor-head' or
`techno-weenie' spring to mind, almost unbidden. People you would
prefer to avoid at parties. Like the ones who want to tell you the
latest exciting details of \LaTeXe\ or \textss{dvips}. If I have to
examine the internals to work out why \TeX\ inserts a skip at some
particular point I'll drift silently to Quark Xpress.\footnote{In
  passing, if I had put a fraction of the effort into Quark that I
  have in \TeX\ and \LaTeX, I would be emminently employable at any
  number of publishers. On the other hand, I would probably also have
  a pony tail and an earring.}

What is the basis for my rant? First, virtually all the discussion in
the handbook is in terms of \unix{}. This is partly fair, since
PrimeTime admit that their experience is almost wholly with \unix{},
and does not really spread to other platforms. Until you actually look
at the \cd{} itself (and yes, it is readable under \dos{}, Windows and
the Mac), you don't realise that it might just be useful to you. There
are the tools needed to unzip the many files on platforms other than
\unix{}. I was mildly amused that the Windows unzip application came
as a \texttt{.z} file itself (\ie\ \emph{zipped}), but since there is
also a \dos{} executable, that is hardly insuperable. The Mac version
of \textsf{unzip} came in two versions -- one a BinHex file, and the
other a self extracting archive. The latter is the one that most of us
would prefer. Those of us who enjoy the Mac tend to enjoy it because
we don't have to mess around. We can get straight on with the real
jobs. In order to use this archive you woud have had to dust down your
copy of ResEdit and mess around changing file types: this is close to
binary editing. It is not the sort of thing you do every day. So I
BinHexed it, and there was the archive, which worked wonderfully.
Except. There is always an except. Turning it loose on a part of the
\cd, on a genuine \textsf{zip} file, it just wouldn't work until I
realised that you had to copy the \textsf{zip} file to your hard disk
and then unpack it.  Otherwise you get unintelligible errors. This is
clearly a failing of this particular implementation of \textsf{unzip},
but it would have been `friendly' if the manual had pointed it out.
After all, someone must have tested it all out, mustn't they? This
means you have to waste disc space on the \textsf{zip}ped and unzipped
files together.

This aside, my main complaint is that to find anything you must first
unzip the file you think it might be in.  Since the unzipped files
would amount to two gigabytes or so, you cannot really just unzip the
whole archive and browse. Maybe I'm being uncharitable. After all, if
you want a particular driver, you look for that driver, and not
another one. Unzip it and off you go. No-one really wants to browse
through all the bits of particular drivers, do they? But think,
there's all that source to modify and learn from. I can hardly wait.
%Does it have \LaTeXe? you ask. Clearly you haven't been paying
%attention. It is the contents of \ctan.  Therefore \LaTeXe\ is there,
%and just where you would expect it. It is also the `current' version,
%rather than the pre-release which the chronologically earlier NTG
%\cd{} contained.

The text of the handbook is
largely taken up with technical details, but has a foreword
George Greenwade, the \emph{emminence grise} of \ctan, a
note on TUG by Sebastian `\emph{Il Presidente}' Rahtz (in
uncharacteristically mellow and benevolent mode), and one of Kees van
der Laan's idiosyncratic expositions entitled `What is \TeX\dots'.
There is no doubt of Kees' enthusiasm. He writes at a frenetic pace
which often leaves the tedious details of conventional sentence
structure far behind. You will either find it exhilarating or
debilitating. At times he seems capable of raising himself several
feet off the ground purely by belief and pace alone. The ground looms
up very large and unfriendly towards the end.

I'm a little bemused by it all. In both cases, opportunities seem to
have been missed. Or perhaps worse, \TeX\ is again cast in the light
of the avid techno-whizz. Both of these offerings are just barely
useful, and save themselves from consignment to one of the outer
Hells. What is it that we should be doing? The contents of \ctan\ are
undoubtedly convenient, but the compression of the PrimeTime disc does
erect a barrier which is too high.  Better to trim it mercilessly but
make it more accessible (yes, I know there is a problem with depth of
directories: \ctan\ goes deeper than the \cd\ format will allow). The
NTG \cd, thanks to the shareware component, is just waiting to turn
sour. The scrupulously honest will have difficulty creating an
alternative integrated \TeX\ system. And anyway, it still isn't
Windows! Time to get into the 90s, before the century ends. But at
least it is possible to find useful files before you find them disc
space.

\end{Article}