SGML

[Anatole Wilson <awilson -at- VNET -dot- IBM -dot- COM>]

I think the greatest effect SGML will have on writers and the information
we produce is that (to badly wreck an old cliche) "Good design is in the
eye of the beholder."

Since SGML tags define only the relationhip of
the different parts of text to each other, and not the way they will appear,
Customer X may view (and print out) the document with eye-pleasing fonts and
stylized boldface, underlines, etc., that make the document more easy to
read, while Customer Y may choose to view the same document in red, 6-point
Courier, with small headings and boldfaced text that obscures the organization
of the document. It will all depend on the style sheet the customer has for
the SGML viewer.

Is this a bad thing? Yes and no. On one hand, it take design worries away from
the writers.  On the other hand, it puts some designers out of work. On the
other hand, it increases the potential for customers with no style sense to
obscure our writing with sixty fonts on a page, and increases the chance
that they'll miss key information (for which we'll get blamed.) On the
other hand, putting desktop publishing in the hands of the graphically-
challenged didn't stop the world--it just made it a little less pleasant to
look at.

I assume that in many cases, companies will provide specific style sheets
with the SGML documents to minimize the potential "misuse." Of course,
the style sheets would probably be proprietary, which eliminates the
original purpose of SGML, which was that it would be a universal standard.

As far as implementation, I've worked with one text-based SGML editor, and
one WYSIWYG editor, Softquad's "Author/Editor."

I thought the text-based editor was almost exactly the same as any other
tag language, with the same pitfalls (similarity to a programming language,
necessity of making sure ALL tags are used in pairs, difficulty in
editing around the tags...).  It really seemed a step backward, usability-
wise.

"Author/Editor" solved some problems by creating beginning and end tags
together automatically, but I still think the screen became a clunky mess
very quickly, with all the tags totallly obscuring text. And I do find myself
questioning the value of WYSIWYG editors when we can't be sure how the
customer will see it.

We haven't really implemented SGML here on a wide scale, so I can't tell you
if the scenario I presented is playing itself out or not.  I suspect that
as SGML comes out of its infancy, some of these problems will be solved.

What do others think?  Am I totally off base? I'm not necessarily
pessimistic about SGML, but I do think there are some challenges ahead
for SGML before it fulfills its promise of becoming a universal
standard.



================================
Anatole Wilson                       "An invasion of armies can
Sr. Assoc. Information Developer     be resisted, but not an idea
IBM, Santa Teresa Labs               whose time has come."
awilson -at- vnet -dot- ibm -dot- com                               --Victor Hugo
all company disclaimers apply
================================