Date: Fri, 15 Oct 1993 12:58:20 -0700 (PDT)
From: "Rob Raisch, The Internet Company" <raisch@internet.com>
Subject: Re: The URN: wrapper and URLs...
To: Michael Mealling <ccoprmm@oit.gatech.edu>
In-Reply-To: <199310151925.AA16777@oit.oit.gatech.edu>
Message-Id: <Pine.3.03.9310151220.A12629-b100000@hmmm.internet.com>
Ummmmm, correctly if I am wrong here, but I do not see any difference
between the following scenerios:
1) I get a 'descriptor' which looks like 'ftp://some.host/some/file' --
using this, I know that I must do a number of things in order to retrieve
this property:
connect, via tcp, to the ftp control port and set up the connection.
send the file retrieval request
accept the data connection
receive the file
close the connections
2) I get a 'descriptor' which looks like 'urn://namespace/product' --
using this, I know that I must do a number of things in order to retrieve
this property:
connect, via some protocol, to the URN to URL mapping service.
retrieve the list (perhaps null) of URLs which are listed for
this product
select the proper (value judgement) URL to retrieve and... etc.
What is the difference? I would suggest that, from a completely
functional standpoint, there is no difference between a URL and a URN, as
a description of a mechanism used to retrieve some data. They are both
indirect methods of abstractly naming a resource on the Global Internet
and there are a number of steps involved in turning the reference to data
into actual data. These steps are not explicitly described in the
reference - the implementor must know how to turn the reference into the
thing. The actual set of actions is certainly different, just as they are
between ftp:* and gopher:*, but I see little use in adding overhead (even
"4 lines of code") if it achieves nothing.
Rob "Less is More" Raisch