summaryrefslogtreecommitdiff
path: root/usergrps/uktug/baskervi/5_2/fine.tex
blob: e5505f1e6b746f469fdcb5537cb8251d14cee93d (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
\title{Formatting SGML Manuscripts}
\author{Jonathan Fine}
\def\simsim{{\sc simsim}}

\begin{Article}

This article is about typography, SGML, \TeX, and \simsim, which is a
new \TeX\ macro package.  Close by are copies of several of the OHP
transparencies.  They were typeset {\em directly from an SGML
document instance} using \simsim.

First some words about the title slide.  Documents can be formatted
for several purposes.  They may be typeset for printing, or for
conversion to Adobe PDF format.  They might be formatted for viewing
on a computer monitor, as is done by the WEB browsers for HTML.  They
might be formatted for display and alteration by a visual or WYSIWYG
editor.  Formatting is the process of supplying fonts, dimensions,
line and page breaking rules and so forth, so as to produce a
representation of the document that is (we hope) well adapted to the
display medium and the needs of the user.  Rendering will convert
this formatted document into bitmaps or whatever that can be
displayed or printed.

In my opinion SGML is as important for structured documents as ASCII
is for character sets (and SQL is for databases).  It is the standard
that will allow different machines and different software programs to
share documents.  In the title I use the words `manuscripts' to
emphasise that my focus is on human communication from author to
reader, and not transference of bytes from one machine to another. 
Human beings have special qualities, which can be reflected in the
manuscripts they produce.  More on this later.

Still on the title slide, the subtitle `Much Ado about Nothing' has
two meanings.  The first is that in five to ten years the formatting
of SGML manuscripts will be no big deal, just as today Postscript is
nothing very special.  The second is that success requires taking
pains or `making much ado' over the spaces.  Which brings us on to
the second slide.

\subsection*{Spaces Between Words}
Typography is not the only art where a sound sense of space is vital. 
Architecture and music are others.  The quotation from Schnabel
expresses my view beautifully.  It is one thing to get the fonts and
sizes right (to play the notes on the score) and another to get the
little pauses or spaces right, and also the timing of the line and
page breaks.  In {\em The \TeX{}book}, Knuth quote Jan Tschichold
``Every shape exists only because of the space around it. \ldots\
Hence there is a `right' position for every shape in every situation. 
If we succeed in finding that position, we have done our job.''  Much
of the typographic art involves getting the space right.  Getting the
choice of fonts right is another skill.

Even if we cannot reach the subtle virtues just expressed, we should
strive to avoid gross errors.  I'm sure we have all seen two words on
a page with an extra space between them, as compared to their
neighbors.  Often this happens because the author has for some reason
placed two spaces between the words (this is the sort of things that
humans are good at doing) both of which have been treated as
significant by the subsequent processing.  \TeX's default reading
rules automatically solve most of these problems, but not when braces
for emphasised text and the like are present.

\begin{figure*}
\extrarowheight2pt
\centerline{%
\begin{tabular}{|l|l|l|}
\hline
\includegraphics[width=.3\textwidth]{fpic0.ps}&
\includegraphics[width=.3\textwidth]{fpic1.ps}&
\includegraphics[width=.3\textwidth]{fpic2.ps}\\
\hline
\includegraphics[width=.3\textwidth]{fpic3.ps}&
\includegraphics[width=.3\textwidth]{fpic6.ps}&
\includegraphics[width=.3\textwidth]{fpic7.ps}\\
\hline
\end{tabular}}
\end{figure*}
The writing of this article (in \LaTeX) provided an example of this. 
In an earlier version I had written
\begin{verbatim}
\subsection*{ Who owns what?}  
\end{verbatim}
and the like to begin subsections. This results in an unwanted
space at the start of the title.  Like so:

\subsection*{ Who owns what?}  
The making of books involves lots of co-operation, and the
participants benefit when there are clear boundaries and
responsibilities.  For example, many authors expect their spelling
and punctuation to be corrected during the publishing process, but
object to their words being otherwise changed. Newspaper journalism
necessarily has different rules, as does academic journal
publishing.  But as a general rule the author supplies the words, the
formatter the spaces.  Problems arise if the author has control over
spacing, or fonts for that matter.  During production copy-editing
and other changes will be made to the author's words.  If supplied as
a computer file, the author can reasonably expect to be sent back
another computer file just like the one that was sent in, but
containing the words as actually printed.  This returned file should
not exercise any control over the spaces between words, for neither
did the author's original file.

Punctuation is a great problem.  By and large, the author should
supply the correct punctuation mark or logical structure.  The
formatter must choose the font and the spacing around the punctuation
mark.  This will depend on the rules of style required by the
publisher.  So at least three parties are involved.  Should the
design or rules or style used by the formatter be changed, so too may
the punctuation marks used.  The more that can be programmed into the
software, the less need there is for human action.  There will always
be exceptions.  Production staff will need on occasion to impose
their will on the software's production of the formatted document.

\subsection*{Manuscript Problems}
These will, in an ideal world, never arise.  In an ideal world others
do all they can to prevent or solve your problems.  And we do all
we can to help others.  In reality the author might be preparing the
manuscript using an ordinary text editor, or a word-processor with an
SGML add-on.  There are likely to be stray spaces and carriage
returns scattered across the file.  There might even be space between
the last word of a sentence and the closing period or other
punctuation!  Particularly if an end-tag intervenes.  If not ignored,
if they influence the final printed page, then a few authors will
discover and use this feature.  Others will be distracted from the
writing of words by the need to get the spaces `right'.  But we have
agreed that the spaces belong to the formatter.  Thus, the three
`Hello world' messages should be formatted identically.  To do
otherwise is to allow the author power over spacing.

For most elements it is reasonable to assume that their boundaries do
not divide words.  And also that between words a space should be
supplied.  Thus, each line in the displayed nursery rhyme should be
formatted in the same way.  The formatter should ignore `extra'
spaces and supply those that are `missing'.  More subtle is this. 
What is the natural size of space to provide between a bold word and
a word in the default (say roman) font?  This is a typographic
question, and so has nothing to do with which element (if any) the
space character appears in.  Should it be a bold-sized space, a
roman-sized space, the larger, some average, or some other value. 

(The OHPs have been set, for simplicity, with a space between
characters depending only on current font size, but not font style. 
Where speed is more valuable than typography, as when an author is
writing the words of a manuscript, or when the display device is a
computer monitor incapable of subtle expression, this is the right
choice.  A quality publisher might wish to specify more closely the
interword spacing.)

The thrust of this slide is that the formatting process cannot assume
that the input file is `just so' and correct for the intended
processing.  More likely it is an electronic manuscript, with
electronic analogues to the physical imperfections that paper
manuscripts present.  We do hope, however, that it can be read.

\subsection*{What is \TeX\ the program?}
\TeX\ is the portable program {\em par excellence}.  It also has very
few bugs.  It is stable across time.  It has an ethos different from
commercial software, which often charges maintenance for bug reports
to be responded to. With \TeX\ one is given a modest monetary reward
for finding a bug.

It is worth remembering that \LaTeX\ is not only a macro package but
also an input file syntax.  Because \TeX\ is programmable, no fixed
input syntax is required. Given sufficiently tricky macros, the
mighty lion that is \TeX\ can be made to imitate other beasts, such
as the unforgetting elephant that is SGML.  \simsim\ is just such a
set of macros.

(The usual \TeX\ approach, when confronted with SGML files to
typeset, is to translate into \LaTeX\ or the like before calling on
\TeX\ to do the typesetting.  However, it seems to me that this
approach cannot but fail to give the author control over spacing, and
to mishandle manuscript problems, unless the translation process is
extremely sophisticated.  It will need to know about the typography
intended for each element and also the character data attributes.
Add to this the legendary problems \LaTeX\ has with verbatim in
titles and so forth, and the limitations should become apparent. 
Translation to \LaTeX\ might have been the best there was available,
but it is certainly not the best that is possible.)

\subsection*{What is \simsim?}
This brings us to the final part of the talk, which is a software
announcement.  The OHPs were typeset using a preliminary version of a
\TeX\ macro package \simsim\ that I have been developing for several
years, and which is close to completion.  The English word `sesame'
is already a registered computer software trademark, so I have chosen
to use the Arabic word `simsim'.  Both are descended from an Akkadian
word, current in Mesopotamia at least 4500 years ago.  Simsim is one
of the oldest words known to humanity.  It is also the key in the
classic story of Ali Babar.

There are two sides to \simsim.  Input and output.  Input is SGML and
also style files.  Output is pages formatted by \TeX.  The title
slide of the talk was typeset from:
\begin{verbatim}
<title-page title =
"FORMATTING / &SGML / MANUSCRIPTS /
 - or - /
MUCH ADO / ABOUT /NOTHING"
>

<par> UKTUG and BCS-EPSG meeting </>

<ol>
<li> (c) Copyright 1995 </>
<li> Jonathan Fine </>
<li> 203 Coldhams Lane </>
<li> Cambridge </>
<li> CB1 3HY </>
</ol>

</title-page>
\end{verbatim}

Notice that the title has been entered as an attribute value, with the
line breaks denoted by forward slash `\verb"/"' or solidus
characters.  This is a notation in wide use for displaying line
breaks in verse quoted as flowing text within a paragraph.   Suppose
one were presented with the title slide and were asked to encode as an
SGML element.  This is the sort of thing that the Text Encoding
Initiative Guidelines were developed for.  One would record that it
was a title page, that such and such was the title text, and so
forth.  It is this approach that led me to use the solidus to denote
line breaks in the title text.  This then is the sort of input
manuscript that \simsim\ will be dealing with.  Note that the
formatter has not been misled by the irregular spaces in the title
attribute value.  The \verb"&SGML" is an entity reference.  In the
title it produces itself in the current font, but elsewhere it is
appearing in a smaller font.  This is done using \TeX's macro
capabilities.

\subsection*{The Flavour of \simsim}
The parsing of an SGML manuscript makes the data within it available
to the formatting (or whatever) application.  There is even a
specification (the Element Structure Information Set) of what data is
available and when.  Built into \simsim\ is an SGML parser.  Writing a
\simsim\ style file is a matter of linking \TeX\ actions to SGML
events, such as the parsing of a start tag.  The less technically
minded might like to skim the following description as to how this is
done.

Another part of \simsim\ is an enhanced programming environment for
the writing of \TeX\ macros and \simsim\ style files.  Within a
\simsim\ macro file the characters \verb"(par)" denote a token that
is called at the end of the parsing of a \verb"<par>" start tag.  It
is up to the application or style file to define this token to
perform the required actions.

Start tags can carry attributes.  The characters
\begin{verbatim}
(title-page|title)
\end{verbatim}
in a \simsim\ file represent a control sequence whose expansion is
the text read by the parser as the value of the (character data)
attribute \verb"title" of the \verb"title-page" tag.  It is then up
to the style file to typeset this data, or to write it to a file, or
to otherwise dispose of it.

The other main type of attribute is the name-group.  Loosely, this
corresponds to the `radio buttons' that graphical user interfaces
provided.  Each such attribute has a short finite list of possible
values.  For example, the HTML \verb"IMG" tag has an \verb"ALIGN"
name group attribute, whose values can be \verb"top", \verb"middle",
or \verb"bottom".  Because \simsim\ incorporates an SGML parser, the
style file need not worry about getting this information.  Indeed,
great errors are liable to occur if it attempts to do so.  Rather,
the parser makes this data available for the application to use.

For example, with the HTML \verb"ALIGN" name group attribute the
process goes like this.  Within the \simsim\ programming environment
the characters
\begin{verbatim}
(img|align)
\end{verbatim}
represent a token whose expansion will be set by the parser to be one
of
\begin{verbatim}
(img*top)
(img*middle)
(img*bottom)
\end{verbatim}
according to the option selected by the author of the manuscript. 
The style file should assign appropiate values to the three tokens
above, for example
\begin{verbatim}
let (img*top)    = vtop
let (img*middle) = vbox
let (img*bottom) = vcenter
\end{verbatim}
(these are illustrative values, and are not necessarily sensible)
and then
\begin{verbatim}
(img|align)
{
  //  the image goes here
  ...  ...   ...
}
\end{verbatim}
will cause the image to be processed in accordance with the attribute
value specified in the manuscript.  This is all rather easier to do
than to explain.  Similar mechanisms are provided to link actions to
\verb"SDATA" entitities.

The observant reader may notice that I have played fast and loose
with the case of tag and attribute names.  For the reference concrete
syntax (used by almost all SGML applications) these names are to be
converted to uppercase when read.  (This is controlled by a parameter
in the SGML declaration.)  This is in practice quite important, and
so \simsim\ converts to uppercase when it parses tag and attribute
names, and the same with the programming environment.

\subsection*{Five Important Questions}
This slide is my attempt to anticipate the questions the audience
would like to ask.  (The untechnical should stop skimming.)  To
amplify my answers, I am looking for SGML-aware \TeX\ users who would
like to be early users of \simsim.  Tables and math capabilities
will, I hope, be developed to meet customers' specific needs.  I do
not think it best that I try to anticipate their requirements.  So
much will depend on the SGML DTDs they use, or intend to use.  Please
contact me if you have any specific questions, and particularly if
you are interested in being a test site.

At the meeting I was asked some good questions.  Firstly, it is
possible to have the processing attached to a tag depend on the
context?  The answer is yes.  For example, the bulleted items on
slide two are \verb"<li>" elements, as on the title page, but within
a \verb"<bl>" rather that \verb"<ol>" list.  This is because the
action attached to a tag is held as a \TeX\ control sequence token,
whose meaning can be changed just like any other control sequence. So
the token represented by \verb"(bl)" can change the meaning attached
to \verb"(li)".  (In fact this may not be the best method, there are
other ways.)

Another question was how does it relate to \LaTeX?  So far as I am
concerned there is no relation with \LaTeX, and no means of
converting documents from one form to another.  Or style files for
that matter.  \simsim\ and \LaTeX\ both start with uninitialised
\TeX, but from there proceed in different directions and with
different assumptions.  I don't see any interaction between the
\simsim\ and the \LaTeX\ worlds, and if somebody creates one, that's
not my doing. A related question (motivated by legacy documents
perhaps) is whether, if you have well structured \TeX\ documents, you
can get something like SGML out of it.  My answer is that probably
you can, but that is not the problem I set myself, and not a problem
I have plans to solve.  

Performance was another question.  How long would it take to process
a long document?  This depends on the computer one has, and on the
mix of text and markup in the document.  Preliminary tests indicate
the same order of speed as \LaTeX.  And do I have a manual?  At the
moment it's not developed to such a point that I can offer manuals. 
But I'd like to.  I want it to be a proper product.  At this point it
is in the process of development and I'm looking for clients who'd
like to take some risk with me, or at least make some effort.  I also
want to supply support.  Further to that, I was asked, will I be
offering maintenance costs (the usual commercial practice) or rewards
(Knuth's practice with \TeX)?  After the laughter had died down, I
declined to answer the question, explaining that I did need to earn
money.  This was the last question.

\end{Article}