Date: Thu, 16 Sep 93 14:19:33 -0400
From: "Chris Weider" <clw@merit.edu>
Message-Id: <9309161819.AA13710@merit.edu>
To: aboba@internaut.com, uri@bunyip.com
Subject: Re: Versioning?
>From uri-request@mocha.bunyip.com Thu Sep 16 12:01:02 1993
>Received: from mocha.bunyip.com by merit.edu (5.65/1123-1.0)
> id AA03373; Thu, 16 Sep 93 12:00:49 -0400
>Received: by mocha.bunyip.com (5.65a/IDA-1.4.2b/CC-Guru-2b)
> id AA04250 on Thu, 16 Sep 93 11:00:22 -0400
>Date: Thu, 16 Sep 93 07:38:03 -1000
>From: aboba@internaut.com (Bernard D. Aboba)
>Message-Id: <9309161738.AA00977@internaut.com>
>Received: by NeXT.Mailer (1.87.1)
>Received: by NeXT Mailer (1.87.1)
>To: uri@bunyip.com
>Subject: Versioning?
>Status: R
>
>Just as an aside, I am wondering what the consensus is on when revision
>of a document requires a new URN. There are documents that may be
>revised quite frequently (such as the PDIAL listings) for which generating
>a new URN with every rev might be an inconvenience. Also, the user will
>probably just want "the latest," not the March 3, 1993 version of PDIAL.
>
The only consensus was that version number would not appear explicitly
as a semantic component of the URN. Other than that it is up to the naming
authority to decide when to assign a new URN. However, some of the inconvenience
you mention above is alleviated by the fact that the 'intellectual content'
pointed to by a URN need not have any relation to a particular physically
embodied resource. If I as a naming authority wanted to maintain the requisite
mappings from URN to URL, I could assign a URN to the concept 'the most recent
version of X'. Of course, such a URN would need to be explicitly resolved into
URLs every time it was used. However, as we discovered in the negotiations
for the creation of the URN, there are many other types of 'metainformation'
which could (pseudo)-plausibly be added to the URN, but that really complicates
matters. In fact, there's a subtle point here that I think needs to be brought
out. Many concepts that humans use to classify resources they wish to retrieve
such as 'latest' or 'biggest' require a comparison between many resources, all
of which are plausible candidates for selection. These comparative attributes,
I argue, cannot be determined by any persistent naming scheme. This means that
much of the work needs to be done by the resource location system rather than
the naming scheme.
(stuff deleted)
>
>Also, the question has come up on the FAQ Maintainers list about
>use of : within file names. This is an issue since the filenames
>of FAQs stored on rtfm.mit.edu are based on the subject lines,
>which often have : within them. So the question came up as to
>how to escape this and other potentially burdensome characters.
>
The final character sets for the URN have still not been determined. I expect
we'll settle this in Houston.
Chris Weider