Re: www in URL's??

Harold A. Driscoll (harold@driscoll.chi.il.us)
Thu, 28 Dec 1995 17:39:21 +0000


Message-Id: <199512281728.LAA06835@thymaster.interaccess.com>
Date: Thu, 28 Dec 1995 17:39:21 +0000
To: Barry Kopulos <bkopulos@wininfonet.mb.ca>
From: "Harold A. Driscoll" <harold@driscoll.chi.il.us>
Subject: Re: www in URL's??
Cc: www-html@w3.org

At 23:56 27/12/95 -0600, you wrote:
>here is something I would like to know:
>
>Why are some addresses in URL's with a preface of www and others do not.?
>Why all of a sudden are we going to not usingt he www int he addresses.?

The host computer name in the URL is just that, the name of the computer
hosting the service. One should tread very carefully regarding the way
people name computers--you might as well be criticizing the names they gave
their children or pets--maybe even as strong of a reaction as arguing
politics or religion. Why, there is even an RFC (RFC 1178) in the FYI series
(FYI 5) on this very subject, http://ds0.internic.net/fyi/fyi5.html .

The use of "www" as the main or "front door" Web server is a social
convenience. Very commonly it is an alias (a CNAME in DNS lingo) to the
"real" name of the computer. This makes sense, since quite likely the
computer is running other servers as well (and might have several CNAME
records). This allows system administration details to change, without you
needing to know or care.

Another situation is a very busy server, such as Netscape. They have more
than a half dozen Web servers (the last time I looked). So when you
reference www.netscape.com you'll somewhat arbitrarily be assigned to one of
them (www3.*, www4.*, ...). You'll also find some services will hand off
various queries to different hosts. Although you might see the host name
change if you look at the URL, otherwise you've no reason to care.

Some services are known in their own right, and combine the use of the Web
with other servers (such as gopher). An example of this is the U.S. Library
of Congress "Thomas" service. As such, it makes good sense to have it
identified as http://thomas.loc.gov/ , and they've not even bothered
defining a CNAME for http://www.thomas.loc.gov/ .

My experience is that the _main_ server for an organization is increasingly
being known as www.*, so things are getting more convenient. At the same
time, as multiple Web servers within an organization become common, some do
things like www2.* or www-students.*, while others just use a host name
which makes sense within the local context.

>Finally does anyone have a shortcut name when using www as it appears ina
>lmost every URl I have come across. Aha-how about w cubed whenever I refer
>to 3 w's in front of a URL.

I curl my tongue when I say it. If you'd rather cube yours, fine by me. :-)

>When are we going to get some final standards on this??

This is a social convention, in a sense a courtesy. When an architect
designs a building, we expect to be able to find the front entrance. There
are visual devices used which help make it obvious. And usually, it is at
the front of the building, but sometimes by the parking lot. Usually we have
no problem, yet occasionally our thought of the architect might be less then
fond. Yet, we manage.

While this response is pushing the limits of HTML discussions, I'm hopeful
that my answer can provide useful background for some readers, and that we
can move on to other issues.

I'd certainly recommend that anybody interested in this topic read the paper
"The User Interface of URLs" by Hoffman, Paul E. (phoffman@proper.com), in
the  INET'95 Hypermedia Proceedings,
http://info.isoc.org:80/HMP/PAPER/016/abst.html

/Harold

ps. We do, for example, have "final standards" for English usage. And we
also allow some leeway for "creative expression" such as "are we going to
not usingt he www int he addresses.?" Surely, recognizing "non-standard"
host names should hardly present a challenge any more difficult than parsing
and understanding such a string of characters.

pps. I don't mean to flame you personally, but I've often noted a trend,
that the folks calling for "final standards" rather than "guidelines" very
often are themselves unwilling to follow "standard English" in their
postings. It seems a paradox. 

Personally, I think the Internet Robustness Principle (Be liberal in what
you accept, conservative in what you provide) is a good philosophy. In
addition to those issues which _need_ to be standardized, we should have
"guidelines" and hopefully have lots of working examples of good and
outstanding practice upon which others can base their work. 
-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
Harold A. Driscoll                       mailto:harold@driscoll.chi.il.us
#include <std/disclaimer>      http://homepage.interaccess.com/~driscoll/