Message-Id: <9305212207.AA03782@mocha.bunyip.com>
Date: Fri, 21 May 93 18:03:47 EDT
From: Richard W Wiggins <WIGGINS@msu.edu>
Subject: Granularity of URNs
To: uri@bunyip.com
Coming from the Gopher community, I've longed for a scheme to break
up documents into logical segments:
-- Look at the Project Gutenberg books, where chapters of Moby Dick are
artificially split by hand. (And numbered out of order, since Chapter
100 sorts before Chapter 1, but that's another story.)
-- Look at the Presidential Debate we have at gopher.msu.edu, where each
document is a question and its answers. A 100 minute (50M) single
document would be unwieldy to the point of being useless. Split
up, you get better random access over the Net than you would with
a DAT on your desktop.
In the Gopher world, the solution to a too-large document is for a human
to segment it into labeled smaller documents. The user can't fetch the
entire thing at once. But if you made it one huge document, each
click would grab it all. We get comments from users on per-packet
charged networks that some of our documents are unexpectedly expensive.
For any large document, whether text, audio, or video, we conceive of it
as one logical work, but with lots of parts. With a little agreed-upon
external structure, those parts could be defined for different types of
documents. A book has chapters. An audio file has "tracks" and,
following the CD model, has "index marks". The Bible has books, chapters
and verses. And so forth.
Assuming documents are marked up into their relevant segments -- a lot of
work in some cases, perhaps aided by automated tools in others -- a
reader could retrieve all or part of the document at will. Presumably
the client would let the user choose; maybe by default the first click
on a document would fetch the table of contents, not the document
itself. In cases where the user retrieves the first segment, and wants no
more, this would make for faster response time, lighter load on
networks, and lower packet charges.
Since SGML/HTML and WWW are new to me, there may be all sorts of schemes
for solving this problem. Please forgive if this is well-explored turf.
No doubt the URN drafters would rather not burden the proposal with
this; maybe this should come later.
/Rich Wiggins, Gopher Coordinator, Michigan State U
========================================================================