summaryrefslogtreecommitdiff
path: root/usergrps/uktug/baskervi/5_2/flynn.tex
blob: 97043407560043f04220aa2ed2f5b7816015d6cc (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
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
% this is from the Companion, p108, but simply doesn't work!
% In the tabular, LaTeX complains that \temp is undefined.
\newcommand{\PreserveBackslash}[1]{\let\temp=\\#1\let\\=\temp}
\let\PBS=\PreserveBackslash
% This ought to be a package, but I don't think it's been unpgraded yet.
\title{HTML \& \TeX: Making them sweat}
\author[Peter Flynn]{Peter Flynn\\University College, Cork}
\begin{Article}
\begin{abstract}

\noindent HTML is often criticised for its presentation-oriented conception.
But it does contain sufficient structural information for many
everyday purposes and this has led to its development into a more
stable form. Future platforms for the World Wide Web may support other
applications of SGML, and the present climate of popularity of the Web
is a suitable opportunity for consolidation of the more stable
features.  \TeX\ is pre-eminently stable and provides an ideal
companion for the process of translating HTML into print.

\end{abstract}

\section{Markup}

HTML, a HyperText Markup Language\cite{html-spec}, is the language used
to structure text files for use in the World Wide Web, an Internet-based
hypertext and multimedia distributed information system. HTML is an
application of SGML, the Standard Generalized Markup Language, ISO
8879\cite{iso-sgml}. Contrary to popular belief, neither SGML nor HTML
is new: SGML gained International Standard status in 1986 and HTML has
been in use since 1989.
\begin{figure*}
\begin{verbatim}
#! /bin/sh

echo Content-type: text/html
echo 

cat <<EOH
<html><head><title>Date and time</title></head><body><p>It is now
EOH
date
cat <<EOT
</p></body></html>
EOT
\end{verbatim}
\caption{Example of a Unix shell script to return the date and time as
a HTML file}
\label{script}
\end{figure*}

SGML is a specification for writing descriptions of text structure. In
itself SGML does not \emph{do} anything, any more than, say, Kernighan
and Ritchie's specification of the C~language\cite{krc} \emph{does}
anything: users and implementors have to do something \emph{with}
it. It has been slow to achieve popularity, partly because writing
effective Document Type Descriptions (DTDs) is a non-trivial task, and
partly because software to make full use of its facilities has
traditionally been expensive. It was therefore seen as a `big business
only' solution to text-handling problems until the popularisation of
HTML owing to increased use of the World Wide Web. Since 1992 the
software position has also improved considerably~--- an extensive list of
tools is maintained by Steve Pfeffer at UIO\cite{sgml-tools}.

\section{The World Wide Web}

WWW (W3 or just `the Web') is a client-server application on the
Internet. Users' clients (`browsers') request files from servers run
by information providers and display them, using the HTML markup
embedded in the text to render the formatting. Some of the markup can
provide filenames for the retrieval of graphics as illustrations, or
act as anchor-points for links to other documents, which can be
further text, or graphics, sound or motion video. This latter
capability gives the Web a hypertext and multimedia dimension, and
allows crosslinking of files almost anywhere on the Internet.

Because the HTML files are plain text with embedded plain text markup,
in traditional SGML manner, they are immediately portable between
arbitrary makes and models of computer or operating system, making the
Web one of the first genuinely portable, multiplatform applications of
its kind.

\subsection{HTML Markup}

An example of simple markup and an appropriate rendering is
illustrated in Figure~\ref{html-example}. The conventions of SGML's
Reference Concrete Syntax\cite{iso-sgml} are used, so markup `tags'
are enclosed in angle brackets (less-than and greater-than signs), in
pairs surrounding the text to which they refer, with the end-tag being
preceded by a slash or solidus immediately after its opening angle
bracket.

\begin{figure*}
\begin{verbatim}
<html>
  <head>
    <title>Fleet Street Eats</title>
  </head>
  <body>
  <h1>Where to eat in Fleet Street</h1>
  <p>There are many restaurants in the City, from 
     fast-food joints to <i>haute cuisine</i>.</p>
  ...
\end{verbatim}
\vspace{1ex}
\hrule
\vspace{1em}
\begin{flushright}
\begin{small}
{\bf Document title:} {\tt Fleet Street Eats}
\end{small}
\end{flushright}
\section*{Where to eat in Fleet Street}
There are many restaurants in the City, from
fast-food joints to {\it haute cuisine\/}.\newline\dots
\caption{Example of HTML markup and possible rendering}
\label{html-example}
\end{figure*}

The rendering is left almost entirely to the user's client program, as
there are almost no facilities within HTML for the expression of
appearance apart from a minimal indication of font change (italics,
boldface and typewriter-type). Indeed, most recent browsers allow the
{\em user} arbitrary control over which fonts, sizes and colours
should be used to instantiate the tagged elements of text. 

\subsection{Implementation}

HTML was devised for the Web by non-SGML-experts who saw it as an
ideal mechanism for implementing plain-text portability while
preserving sufficient structural information for online rendering: one
of the classical reasons for adopting SGML. It is now becoming
standardised by an IETF working group who have produced a draft
specification in the form of a formal DTD\cite{html-spec}. Because of
the need to allow this specification to model existing `legacy'
documents (most of which would be regarded as fragments rather than
document instances), as well as provide for more robust usage, the
current DTD has two modes: a non-rigorous `deprecated' mode for
describing the legacy and a `recommended' mode for creating and
maintaining files in conventional form.

HTML is sufficient for minimal documents, providing the structural and
visual features shown in Figure~\ref{html-tags}. A future version
(3.0) is being developed by the IETF Working Group, which will allow
the description of mathematics, tables and some additional visual- and
content-oriented features.

\begin{figure*}
\begin{tabular}{>{\tt}l>{\PBS\raggedright\hspace{0pt}\baselineskip=10pt}p{1in}|>{\tt}l>{\PBS\raggedright\hspace{0pt}\baselineskip=10pt}p{1in}|>{\tt}l>{\PBS\raggedright\hspace{0pt}\baselineskip=10pt}p{1in}}
\multicolumn{2}{c|}{{\bf Structural}}&\multicolumn{2}{c|}{{\bf Descriptive}}&\multicolumn{2}{c}{{\bf Visual}} \\ [3pt]
html            &document type&
  a             &hypertext link anchor-point&
    b           &bold type \\
\quad head      &document header&
  cite          &citations&
    br          &forced line-break \\
\quad title     &document title&
  code          &computer code&
    hr          &horizontal rule \\
\quad base      &root address for incomplete hypertext references&
  em            &emphasis&
    i           &italics \\
\quad meta      &specification of mapped headers&
  kbd           &keyboard input&
    tt          &typewriter type \\
\quad link      &relationship of document to outside world&
  samp          &sample of input&
    img         &illustrations \\
\quad isindex   &specifies a processable document which can take an argument&
  strong        &strong emphasis&
                &\\
                &&
  var           &program variable&
                &\\
body            &contains all the text&
                &&
                &\\
\quad h1\dots h6&six levels of section heading&
                &&
                &\\
\quad p         &paragraph&
                &&
                &\\
\quad pre       &preformatted text&
                &&
                &\\
\quad blockquote&block quotations&
 {\bf Form-fill}&&
    {\bf Obsolete:}&\\
\quad address   &addresses&
  form          &contains a form&
    listing     &use {\tt pre} \\
\quad ol        &ordered lists&
  textarea      &free-text entry &
    xmp         &use {\tt pre} \\
\quad ul        &unordered lists&
  input         &input field (text, checkbox, radio button, {\it etc}&
    plaintext   &use {\tt pre} \\
\quad menu      &menu lists&
  select        &drop-down menu&
    nextid      &editing control \\
\quad dir       &directory lists&
  \quad option  &menu item&
    dfn         &definition of term\\
\qquad li       &list item&
                &&
                &\\
\quad dl        &definition lists&
                &&
                &\\
\qquad dt       &definition list term&
                &&
                &\\
\qquad dd       &definition list description&
                &&
                &\\
\end{tabular}
\caption{Markup available in HTML 2.0 (indentation implies the item
must occur within the domain of its [non-indented] parent)}
\label{html-tags}
\end{figure*}

Despite the coming improvements, HTML is likely to be joined in the
Web by other DTDs in future. One well-known SGML software house
already has a prototype browser which can handle instances of
arbitrary DTDs, given sufficient formatting information.  This would
make it possible to use the Web for transmission and display of
documents using other SGML applications such as CALS (US Military),
DocBook (O'Reilly/Davenport), the TEI (Text Encoding Initiative) and
corporation-specific DTDs (such as those of Elsevier).\label{otherdtds}

The next version of the DTD, HTML3, contains specifications for
mathematics, tables and some additional elements for
content-descriptive material, as well as a few extra visual keys such
as an {\tt ALIGN} attribute for positional specification. Most of this
work is being implemented on a test basis in the Arena browser (Unix/X
only at the moment) at CERN.

Although Web browsers can reference files by any of several methods
(HTTP, the Web's `native' protocol; FTP; Telnet; Gopher; WAIS; and
others) by using the URL (Universal Resource Locator: a form of file
address on the Internet), the most powerful tool lies at the server
end: the ability of servers to execute scripts, provided their output
is HTML. A trivial example is shown in Figure~\ref{script}, which
returns the date and time.

Such a script can contain arbitrary processing, including the
invocation of command-line programs and the passing of arguments. Data
can be gathered from the user either with the \verb+<isindex>+ tag in
the header, which causes a single-line data-entry field to appear, or
with the more complex \verb+<form>+ element with scrollable text
boxes, checkboxes, radio buttons and menus. In this manner, complete
front-ends can be manufactured to drive data-retrieval engines of any
kind, provided that they operate from the command line, and that the
script returns their output in HTML\@. The user (and the browser) remain
unaware that the result has been generated dynamically.

\subsection{Presentation}

HTML is criticised for being `presentation-oriented', but as can be
seen from Figure~\ref{html-tags}, the overwhelming majority of the
markup is structural or content-descriptive. However, this does not
prevent the na\"\i ve or sophisticated author from using or abusing
the markup in attempts to coerce browsers into displaying a specific
visual instantiation, primarily because none of the browsers (with the
partial exceptions of Arena and {\tt w3-mode} for GNU Emacs) performs
any form of validation parsing, and will thus display any random
assemblage of tags masquerading as HTML. This behaviour has misled
even some eminent authorities to dismiss HTML as `not being SGML'.

There is thus a conflict between the SGML purist on the one side, who
decries any attempt at encoding visual appearance; and the uninformed
author on the other, who has been unintentionally misled into thinking
that HTML and the Web constitute some kind of glorified networked DTP
system. 

The purists are few in number but eloquently vocal: however, in
general, they acknowledge that visual keys can be included if they are
carefully coded. A perceived requirement to allow an author to
recommend the centering of an element is thus achieved in HTML3 by the
\verb+align="center"+ attribute, rather than the unnecessary
\verb+<center>+ element proposed by the authors of Netscape.

The demands of the author are at their most marked in the approach of
publishers and marketing users, who have been accustomed for the last
550 years to exert absolute control over the final appearance of their
text. But the Web is not paper, and the freedoms and constraints of
the Press do not apply: it is as much a new medium as radio or
television. For such an author to insist that she must be able to
control the final display to the same extent as on paper is as
pointless as insisting that a viewer with a black-and-white television
must be able to see the colours in a commercial.

The paradigm has been established that the browser controls the
appearance, using the markup as guidelines. There is indeed no reason
at all why attributes could not be added so that an author could write

{\tt <h1 color=green font=LucidaBrightBoldItalic size=24 shading=50>}

\noindent but the user of Lynx or WWW (two popular text-only browsers
for terminal screens) would still only see the heading in fixed-width
typewriter characters. The habit of insisting that everyone `must'
see a particular typographic instantiation is an unfortunate result of
a misinterpretation of the objective of the Web: to deliver
information in a compact, portable and arbitrarily reprocessable form.

But publishers accustomed to paper, insistent on `keeping control',
have of course an entirely valid point, one with which the present
author has great sympathy. Why should a carefully-prepared document be
made a hames of by a typographically illiterate user who has set
\verb+<h2>+ to display as 44pt Punk Bold in diagonal purple and green
stripes?

The solution probably lies in the implementation of style sheets,
perhaps along the lines of those discussed by the authors of
Arena\cite{arena-style}. They would in any case only be
recommendations: not every user has a CD-ROM of Adobe or Monotype
fonts. In any event, if 100\% control is essential, as in the display
of typographic examples, all graphical browsers can be configured to
spawn a window to display PostScript file, although the download time
may be a strong disincentive.

It is entirely possible that the control of content will
ultimately prove a more attractive option than the control of
appearance.

\section{Publishing with HTML}

Setting aside the unresolved questions of display, there are more
pressing business problems about publishing on the Web.

The authentication of users is being addressed at several levels, from
simple, non-authoritative checks using {\tt identd} to the more complex
username-and-password systems employed on some Web pages. From the
user's end, the authentication of the data being accessed is equally
important. The openness of the Internet in its raw form allows
`spoofing' in both directions, so the emergence of protocols to
provide checks is to be welcomed. 

The security of network-accessible texts from break-ins remains a
concern to anyone providing high-value merchandise, and Web text is
in this sense no different from any other computer data. Normal
precautions must therefore be taken to prevent theft through
other channels (such as remote lo\-gin), as distinct from theft
perpetrated by falsification of Web access.

There is a need for robust solutions to charging and billing for
usage, and the secure transmission of financial data, including credit
card numbers, digital signatures, and perhaps even EFT transactions.
The Secure HTTP (SHTTP) mechanism being marketed by MCom and others is
becoming popular as a way of achieving some of this, but the Internet
must shed some of its image of lax controls and sloppy housekeeping if
it is to achieve sufficient `respectability' to attract the business
of those who are not networking specialists.

The handling of copyright and the intellectual property of electronic
texts remains, as ever, an unsolved problem. While copyright law can
be used to provide a remedy for breach, the difficulty lies in
preventing the breach occurring in the first place. The reason is that
(as with other electronic mat\-er\-ial), copying and reproduction is fast,
cheap and easy, once the material is in the hands of the customer.
While a supplier may use SHTTP to protect the details of the
transaction, once a print file has been sent to someone, the supplier
retains no control whatsoever over its use, reuse and abuse. Copies
could be sent to dozens others, or printed many times, in the space of
minutes. 

\subsection{Printing from HTML}

The demand for printed copies of Web material is surprisingly high.
Although in some cases it is reminiscent of those people who insist on
printing their email, it is undeniable that there is a serious
requirement for good quality print from Web documents.

Existing solutions to printing SGML text are usually
application-specific, being embedded in SGML editors or DTP systems,
but there are also some more generic packages:

\begin{itemize}

\item {\tt Format} by Thomas Gordon (\LaTeX)

\item {\tt HTMLtoPS} by Jan K\aa rrman (PostScript)

\item {\tt SGML2TeX} and {\tt WebSet} by Peter Flynn (\TeX/\LaTeX)

\item {\tt SimSim} by Jonathan Fine (\TeX)

\end{itemize}

\noindent The use of \TeX\ systems for most of these seems to indicate
that the similarity of markup concepts has not gone unnoticed by
practitioners. The author's own contributions are experimental, but
the second of them is planned as an interactive Web service, to be
introduced in the summer of 1995. Emailing a URL to the point of
service will cause it to be retrieved, typeset, and the output
returned to the user by email in PostScript form. As a form of email
browser, the control of appearance may lie in the hands of the
user, but suggestions for how implement this are currently
being sought\cite{ttn41}.

\subsection{Problems}

Implementing a professional level of typesetting from HTML raises some
interesting questions:

\begin{itemize}

\item most HTML files are invalid

\item most HTML authors don't understand SGML 

\item most HTML authors couldn't care less

\item most World Wide Web users couldn't care less

\end{itemize}

The handling of missing, damaged or abused tags in a gracious manner
is not a feature of most SGML parsers. At the best, a
typesetter-browser can only be expected to report to the user that a
file is invalid, and while it may be displayed by browsers which do
not make any claim to typographic quality, an attempt to make a
respectable print job of an invalid file is unlikely to succeed.

\section{Development}

The future of the World Wide Web and HTML is uncertain. While
development continues, and while new users are anxious to start
surfing the net, the existing designs and implementations will
suffice. In the longer term, a coalescing of services is likely to
occur, but for this to happen, a number of changes need to take place:

\begin{itemize}

\item The Web will start to make use of other DTDs, as outlined above.
  Any file containing a \verb+<!doctype...>+ at the beginning could
  cause a browser to retrieve the DTD specified, along with a style
  sheet, and work much as any SGML-conformant DTP system would.

\item Browsers will become pickier, able to offer better services at
the expense of rejecting invalid or badly broken files. Arena already
perfoms a form of consistency check on the HTML code of files, and
displays `Bad HTML' in the top corner when an offender is spotted.

\item Users will become pickier, demanding better response from the
browser, better response from the server, and better facilities from
both. As users become more educated about the use of SGML, developers
will no longer be able to hide the deficiencies of products under the
cover of technical detail.

\item This presupposes more user education, which is inevitable in a
developing technology. 100 years ago, motor cars appeared on the
roads, but few passengers in them understood the use of the levers and
rods which controlled them. With some minor exceptions, it is now
expected that a driver knows that turning the wheel clockwise turns
the car to the right, and {\it vice versa\/}. It will not take us that
long to perceive the innards of HTML, but it can only be done by
training and education.

\item At some stage, investment is always needed. Many companies have
put substantial sums into the development of Internet resources, and
those that have done so with forethought and planning deserve to reap
a rich reward. It is a long-term investment, more akin to a
partnership, but support is always needed by those who undertake the
developments, especially as much of it is done in personal time and at
personal expense.

\end{itemize}

\noindent There is still some way to go before we achieve the ease of
use of the telephone or the radio, but the path is becoming easier
with each new development.

\begin{thebibliography}{9}

\bibitem{html-spec} Berners-Lee~T \& Connolly~D, {\em HyperText Markup
Language Specification~---~2.0}, Internet Draft, IETF Working Group on
HTML, December 1994.

\bibitem{ttn41} Flynn~P, {\em Typographers' Inn}, \TeX\ and TUG NEWS,
{\bf 4}, 1, March 1995.

\bibitem{iso-sgml} Goldfarb~C, {\em The SGML Handbook}, OUP, 1990,
ISBN 0--19--853737--9.

\bibitem{krc} Kernighan~BW \& Ritchie~DM, {\em The C Programming
Language}, Prentice-Hall, 1978.

\bibitem{arena-style} Lie~H {\em et al}, {\em HTML Style sheets},
\URL|http://www.w3.org/hypertext/WWW/Style/| 

\bibitem{sgml-tools} Pepper~S, {\em The Whirlwind Guide: SGML
tools and vendors},
\URL|ftp://ftp.ifi.uio.no/pub/SGML/SGML-Tools/SGML-Tools.txt|

\end{thebibliography}

\end{Article}