Re: Relative URLs, // and ;

Ron Daniel Jr. (
Mon, 27 Jan 1997 16:23:45 -0700

Message-Id: <>
Date: Mon, 27 Jan 1997 16:23:45 -0700
To: Tim Berners-Lee <>
From: "Ron Daniel Jr." <>
Subject: Re: Relative URLs, // and ;

[I'm adding to the Cc list since that is the
list for the IETF's URN-WG.]

Thus spoke Tim Berners-Lee:

>The URN scheme should be
>mapped onto

The proposed URN syntax is laid out in draft-ietf-urn-syntax, but
the short description is:
for example

Some namespaces, such as ISO Formal Public Identifiers, already make
use of '//'. For such namespaces we say they need to %encode occurances
of '/' that do not follow the rules of RFC1639. Those rules were that
'/' denoted hierarchy with the most significant component on the left.

>At 11:29 am 27-01-97 -0600, Daniel LaLiberte wrote:
>>Are you just putting your foot down now  .. finally.
>I've been out of the loop.  URIs are not supposed to be irrevocably linked
>to DNS!
>Unfortunately there have been many demands on my time.

I don't understand your assertion about DNS. No URN namespace is
required to use domain names, as shown by the examples above. Certainly the
namespaces we are considering for use at the lab have NO tie to the DNS.
(They may be resolved through the DNS, but the names themselves don't
use domain names).
Members of the URN-WG have expressed a desire for a namespace that is
easy for people to get started with, and lots of people think that using
the domain name hierarchy (perhaps with some augmentations to overcome
the reassignment problem) is the way to go. Such a namespace might
look like
but I emphasize that this is only one namespace out of many.

[Lots of stuff on "///", "//", "/" deleted.]
At this time we have not defined any relative URN specification.
Nevertheless we do not allow unencoded occurances of '/' in URNs
unless they follow the rules in RFC1630. Similarly, '?', '#', and
a few other characters must be encoded unless they follow the
rules of RFC1630.

One of the URN requirements was to allow old namespaces to be grandfathered
into URNs. Because of this, I am wary about requiring ALL URN namespaces
to use '//'. Some namespaces will not have the notion of a relative identifier,
others (such as path) may not need '//'. What we do require is that
when representing a name from an old namespace as a URN, reserved
characters are encoded unless they meet the semantics established by 1630.

Ron Daniel Jr.                       email:
Advanced Computing Lab,  MS B287     voice: +1 505 665 0597
Los Alamos National Laboratory         fax: +1 505 665 4939
Los Alamos, NM, USA  87545  
 Want to buy: "The Five Laws of Library Science" by S.R. Ranganathan