On standardization

Jim Whitehead (ejw@ics.uci.edu)
Thu, 30 May 1996 16:57:20 -0700


Dan Connolly wrote, first quoting Larry Masinter and Christopher Seiwald:

>In message <96May29.162909pdt.2733@golden.parc.xerox.com>, Larry Masinter
>writes:
>>Wait, I didn't see that anyone was proposing to "standardize the
>>syntax of version identifiers across all servers".
>
>I did:
>
>=============
>Date: Tue, 28 May 1996 22:08:19 -0700
>From: Christopher Seiwald <seiwald@p3.com>
>Message-Id: <199605290508.WAA05273@spice.p3.com>
>To: ejw@ics.uci.edu
>Subject: Re: Version identifier in URL
>Cc: www-vers-wg@ics.uci.edu
>
>[...]
>
>We need to stick to our guns: make a simple version extention to URLs
>and make them standard.
>
>Christopher
>----
>Christopher Seiwald     P3 Software             http://www.p3.com
>seiwald@p3.com          f-f-f-fast SCM          1-510-865-8720
>=============

There are at least two meanings to how people generally use the word
"standard."  In the descriptive sense, a standard describes what is
currently occurring, as in "standard practice."  In the normative sense, a
standard describes how things should occur.  The HTTP/1.0 specification
exhibits both of these meanings, both describing (mostly) standard HTTP
server practice, and also giving meaning to what an HTTP/1.0 server should
be.  Of course, the problem with a normative-style standard is that it may
never become widely adopted, and hence is not really a standard.

It strikes me that we are trying to develop a normative standard for
describing a particular version of an entity using URLs.  Since any
proposal we develop will be subject to discussion in front of a broader
audience (e.g., IETF HTTP Working Group), and since no system (with the
potential exception of the MKS Integrity Engine) will actually implement
the proposal once we develop it (at least right away), we are certainly
incapable of developing a standard in either the "normative" or the
"standard practice" sense of the word just yet.

However, since one goal or our work is to ensure that version-aware
browsers and distributed editors are *interoperable*, it makes sense to
develop a standard (in the normative sense) convention for naming
individual versions of an entity.   I feel confident that given the broad
spectrum of viewpoints and interests represented in this working group, any
naming proposal we develop has a reasonable chance of being accepted by a
larger working group, eventually becoming widely implemented, and hence
standard practice as well.

- Jim Whitehead <ejw@ics.uci.edu>