I was wondering how long it would take for this discussion to roll around
to the question it begs -- what is the Web? It has thus rolled and I agree
that the Web is much more than HTTP, since the clients and servers
integrate other tools. The Web wouldn't be so interesting were it not for
that capability. The definition above suggests that the Web is
standardized by URLs. That makes the URL standard verrrry important.
This is a very important debate because it makes implications about whether
or not there is a default navigational paradigm. If the Web were defined
by HTTP, then the default would implicitly be hypertext... except that with
its foray into forms, HTTP has expanded beyond "straight" hypertext into
the world of fields, menus and buttons and thus into GUIs, QBE, etc. This
seems to have been done largely because the servers have been integrated
with databases, for which the default paradigm is fields, buttons, menus,
etc. The browsers integrate. Gopher, ftp and news, which have a default
paradigm of hierarchical browsing. Actually, however, I think this may
reflect that HTTP has grown beyond its name, not that it's taking a wrong
direction.
I don't want to get into a taxonomy of navigational design paradigms here.
I'd like to suggest that one of the great strengths of the Web is that
through a single user interface (yet at the same time, with many choices
for that single one), a variety of navigational paradigms can be
integrated. This is a truly wonderful thing, I suspect. For a variety of
users and a variety of tasks, a variety of paradigms are essential.
Without really much work or programming talent, I've been able to wire
together a server that incorporates hierarchical, hypertext, full-text
search, and indexed navigational paradigms.
I could have done this before, in fact I did, but each paradigm required a
different software tool. I used a database, AppleScript, HyperCard, a
full-text engine, word processors, page layout tools and the Macintosh file
structure. It was nothing that anyone else could really understand except
in pieces. The Web thus is not only lovely from a production and
maintenance standpoint (now that I've built tools for those purposes), it's
a million times easier for users to grasp and use, cross-platform.
And I know that I can add graphical, time-line and other navigational
paradigms to the server, again without having to spend a lot of time and
without really learning C, which I resist at every turn, though it's harder
every day.
I think of the Web as a standard way to "package" distributed information
from hetereogeneous sources for navigability. What are the implications
for HTTP if we look at HTML as just one of a number of information
navigation/packaging standards of equal importance? Perhaps we're defining
a compound document architecture, but that's what a Web page resembles,
anyway.
Here's the question that's really nagging me. Do we really intend for the
HTTP to become the means by which information from heterogeneous sources
becomes integrated? For example, When browsers speak SQL, will it make
sense to talk to a database through the Web and a CGI application? Or will
the URL architecture encourage browser developers to bypass HTTP?
I'm interested in practical responses. For example, although browsers
speak NNTP, there are some compelling reasons to use HTTP to deliver news
(primarily integration reasons), even though it's theoretically less
efficient to add the extra layer of processing.
Nick
Multimedia Computing Corp.
Campbell, California
----------------------------------------------------------
"We are surrounded by insurmountable opportunity." -- Pogo