Message-Id: <199411291733.JAA21594@rock>
From: Terry Allen <terry@ora.com>
Date: Tue, 29 Nov 1994 09:33:05 PST
In-Reply-To: Peter Deutsch <peterd@bunyip.com>
To: Peter Deutsch <peterd@bunyip.com>, Terry Allen <terry@ora.com>,
"Karen R. Sollins" <sollins@lcs.mit.edu>, uri@bunyip.com
Subject: Re: going too far? (longish diatribe)
| } >We should *NEVER* be putting URLs into documents.
| }
| } If the URL is to a document that expires within a year, and we
| } think the site where it resides will keep it right where it is
| } for the foreseeable future, why not? . . .
| Well, one reason touches upon a user inteface issue. If we
| publish both URLs and URNs we would then have two sets of
This discussion is set in the present, when URNs are not yet useful.
| Another reason is an engineering one. Even if we _think_
| something will be around for only a year, it seems like
| poor engineering to build in such assumptions, since the
| Internet is so inherently flakely and the written word is
In which case even in the future you won't be able to count on
resolving URNs properly and might prefer to have URLs too.
[ ... ]
| } . . However, we should also
| } provide non-URL references to those documents pointed to by URLs.
|
| I think Karen's point was that we should not encourage
| poor practice, even if we must tolerate it for now. If we
| can get the URN architecture in place, then it is in
| everyone's interest to do so ASAP. If we build it right,
| the URLs then become redundent and should be omitted. I'm
But proper bibliographic references do not become redundant, and
should continue to count as complete references. URL vs. URN is
another matter entirely. I should always be able to discover a URN,
perhaps even the top-level URN, for a document given a proper
bibliographic reference (author, title, date, to start with).
| the first to admit we shouldn't omit URLs yet, but I agree
| with the idea that we should move away from them, and as
| soon as possible.
I agree that when and if URNs become useful it will become good
practice to supply them along with the usual natural language
citation; only if URNs achieve unqualified success will it be
sensible to establish doctrine about URLs.
Today, when you write a paper and cite a book in your university's
library, you don't supply the call number of the book, because that's
not portable---in another university library the same book will
probably have a different call number, which can be obtained by use
of that library's catalogue along with the bibliographic info in
your citation.
But when you cite a document on the Net, the URL is portable in a
way the call number is not, and requires no lookup to use. URNs
and URN lookup will have to become as easy, cheap, and fast to use
(perhaps because you can find a nearer, cheaper copy of what you
want, perhaps because you tend to read stale documents) as URLs before
it will be bad form to supply URLs---until then, as Roy(?) pointed
out, they're useful info to supply to the reader.
-- Terry Allen (terry@ora.com) O'Reilly & Associates, Inc. Editor, Digital Media Group 103A Morris St. Sebastopol, Calif., 95472 A Davenport Group sponsor. For information on the Davenport Group see ftp://ftp.ora.com/pub/davenport/README.html or http://www.ora.com/davenport/README.html