3.1050 SGML and hypertext (154)

Willard McCarty (MCCARTY@vm.epas.utoronto.ca)
Wed, 14 Feb 90 20:50:45 EST

Humanist Discussion Group, Vol. 3, No. 1050. Wednesday, 14 Feb 1990.

(1) Date: Wed, 14 Feb 90 09:58 (39 lines)
From: Wujastyk (on GEC 4190 Rim-C at UCL) <UCGADKW@EUCLID.UCL.AC.UK>
Subject: electronic editions

(2) Date: 14 February 1990 09:42:00 CST (61 lines)
From: "Michael Sperberg-McQueen 312 996-2477 -2981" <U35395@UICVM>
Subject: SGML and hypertext

(3) Date: Wed, 14 Feb 90 09:47:45 -0800 (30 lines)
From: Malcolm Brown <mbb@jessica.Stanford.EDU>
Subject: status of links and tags

(1) --------------------------------------------------------------------
Date: Wed, 14 Feb 90 09:58
From: Wujastyk (on GEC 4190 Rim-C at UCL) <UCGADKW@EUCLID.UCL.AC.UK>
Subject: electronic editions

Charles Faulhaber (hi!) wonders about the relative merits of
Hypertext and SGML for representing textual editions. I believe that
there is a category error here: SGML is just a way of defining a set
of tags in a rigorous fashion. Once you have decided what tags you
want to define (e.g., <link page = 10> might be defined to mean "at this
point stop reading this document and go and read another at page ten")
then you sprinkle them liberally throughout the document you wish to code up.

Hypertext (about which I know virtually *nothing*) is -- I gather --
a genre of application program. There seems to me no particular problem
in a Hypertext program being designed in such a manner as to be able
to read and comprehend texts tagged up using SGML-defined tags. In
fact, I think that is part of the point of SGML. To quote ISO 8879
(thanks, Lou):

2 Field of Application

The Standard Generalized Markup Language can be
used for documents that are processed by any text
processing or work processing system. It is
particularly applicable to :

) Documents that are interchanged among
systems with differing text processing

b) Documents that are processed in more than
one way, even when the procedures use the
same text processing language.

Documents that exist solely in final formatted form
are not within the field of application of this
International Standard.

(2) --------------------------------------------------------------70----
Date: 14 February 1990 09:42:00 CST
From: "Michael Sperberg-McQueen 312 996-2477 -2981" <U35395@UICVM>
Subject: SGML and hypertext

It's hard to answer Charles Faulhaber's request for enlightenment as to
"the relative merits of coding a particular feature via SGML or linking
a region of text to another node in the system." This sounds to me too
much like an inquiry "which is to be preferred for expression, the Latin
alphabet or the English language?"

SGML is a (means of defining) markup languages. Markup languages are
used to mark characteristics of text for storage or interchange, or to
guide processing by systems which have been told (separately) what to
make of the markup and what to *do* when they see it.

Hypertext systems are processing systems which make it easy to express
and work with various non-linearities in text.

Those non-linearities must be marked, for storage or interchange, with
some codes which we may call markup (since they fit the definition of
that term). SGML-based markup languages, like a vast host of other
markup languages, may be invented to describe these, like any other,
types of textual features. Most hypertext systems will not use SGML
forms for their disk format (let alone for their in-core storage), but
that does not prevent them from using SGML-based languages as export or
import languages. (Some SGML systems don't use exportable SGML for
their disk storage formats, either. Let's face it, non-portable,
proprietary, opaque languages can easily be more compact than portable,
open, legible languages. No surprises there.)

Perhaps I have misunderstood the question. If the question is "which is
more convenient, (a) writing proprietary hypertext-link codes into an
ASCII file with an ASCII editor, (b) writing SGML-based codes into an
ASCII file with an ASCII editor, (c) making hypertext links with a mouse
or cursor in a running hypertext system, or (d) making hypertext links
with a mouse or cursor in a running SGML system?", then my answer would
be that it all depends on the user interface of the postulated hypertext
systems (whether SGML-based or not) and the user interface of the ASCII
editor one is using. (And on how well one knows or cares to know the
programs. Some people never leave Emacs, and may find (a) easier than
(c) because they *know* emacs and not the hypertext system.)

Of the four, I'd lean toward (c) or (d) over (a) or (b), assuming
reasonable user interfaces. Of (a) and (b), I'd probably choose (b)
because it would be easier to make my existing software print out nice
proofs for me to check the markup; on (c) and (d) I'd express no opinion
beyond pointing out that unless the system of choice (c) allowed a clean
ASCII export of all links, I wouldn't put anything into it in the first
place, just because I expect ASCII files to be readable 50 years from
now, but I don't expect any software made by the keystrokes of human
beings to be running 50 years from now, except for the Cobol programs in
banks and oil companies where I hope not to be working.

A great sage of hypertext (none other than our own Steve DeRose) once
put it in a nutshell when he said, "SGML and Hypertext -- two great
tastes that go great together."

Which is preferable? Both.

-Michael Sperberg-McQueen
University of Illinois at Chicago
(3) --------------------------------------------------------------41----
Date: Wed, 14 Feb 90 09:47:45 -0800
From: Malcolm Brown <mbb@jessica.Stanford.EDU>
Subject: status of links and tags

Charles Faulhaber's inquiry brings up an interesting point
concerning the status of links and tags in their respective
computing environments (hypertext and SGML).

To my knowledge, in many (if not most) hypertext systems the
links are not part of the data. Hence you cannot retrieve links in the
way you might retrieve text; it is impossible, in such
systems, to "search for all links to file x". Indeed, some
systems cannot give any overview of the link structure.

On the other hand, it seems important to be able to do just
exactly what Charles suggests. Conceptually, hypertext
Links are indeed part of the
data, just as SGML tags are part of the data in a marked
up file. This is why I've been less than wildly enthusiastic
about many of the hypertext systems I've looked at. Until links
are part of the data, hypertext systems will be useful for
displaying canned sequences (and hence be useful for lower
level instruction). But their usefulness for research will
be limited at best.

Malcolm Brown