Re: Last call: Intro, SGML, MIME sections

Daniel Shields (pepprboy@inforamp.net)
Fri, 5 May 95 02:54:02 EDT

Alex Hopmann writes:
>:) Sounds like alot of work to me so feel free to disregard the above
>suggestion. But whoever is collecting the list of ideas for REL/REV values,
>make sure "GLOSSARY" is in there, or possibly Definition. There is probably
>10 other things you could call it also.. Yet another example of why
>standardized names for these things would be useful. I think it would be
>great if my browser knew that glossary definitions should be hilighted
>differently than normal hot links (and thus make documents like the HTML one
>alot less busy). But this will almost be impossible unless we have a list of
>standard definitions for these things.

The way I see it, the whole reason to classify the relationships implied by
links is similar to why you want to classify data-types in a high-level
programming language. Recall the advantages of supplying programmers with
floating point numbers and signed integers, rather than forcing them to deal
with 32-bit doublewords (the default for many CPUs in use today) so that
they have to invent their own math libraries. In publishing on the Web the
simple link is just not meaningful enough for doing complex publishing tasks
requiring the automatic generation of indices, tables of contents, and
automatic link-maintenance systems.

There are two main sources of Web content that can be addressed at least
partially by clssifying links:
1. current SGML and word-processor encoded texts should be downward
renderable in an agreeable fashion for publishing on the Web;
2. new documents authored electronically should be managed on the
Web such that subdocument relationships are preserved in large
works-in-progress. This is especially true of evolving online works like
Internet Drafts, where we'd like to be able to peruse updates to particular
sections during a debate, rather than waiting for the next ftp-upload of the
full document.

The issues of (1) can be addressed by defining link relationships which are
a superset of applicable industry-standard DTDs. I think that by anchoring
text and classifying the link for inclusion in, say: an index, a table of
contents, a a glossary, a footnote, and a list of figures, covers the main
desired capabilities of the textbook-writing industry. Perhaps someone who
has direct experience with applicable industry-standard DTD's can help us
form a common superset, then we can discuss applying standard names for the
relationships.

I feel (2) is fully addressed if we are adaquate in defining subdocument
relationships of a larger document concept. This is important both for a
browser's navigation of a document, and, for authors. Authors who edit a
piece of a large multidocument document might say, "okay server, I am
publishing chapter 3, please do the server/client-side parsing and link
maintenance to insert this chapter into the superdocument without causing
any broken links. (and why don't you update the table of contents, index,
glossary, etc., while you are at it)"

--
Daniel Shields          Air mobile, son.  I'll pick up your boat and drop it in
(416)597-8239                 that river anywhere you want it, my young captain