Re: Version issues

Jim Whitehead (ejw@ics.uci.edu)
Thu, 1 Apr 1999 17:06:30 -0800 

From: Jim Whitehead <ejw@ics.uci.edu>
To: Versioning <ietf-dav-versioning@w3.org>
Date: Thu, 1 Apr 1999 17:06:30 -0800
Message-ID: <005101be7ca5$0b256240$d115c380@ics.uci.edu>
Subject: Re: Version issues



-----Original Message-----
From: Geoffrey M. Clemm [mailto:gclemm@tantalum.atria.com]
Sent: Monday, March 15, 1999 9:34 PM
To: ckaler@exchange.microsoft.com
Cc: Bradley.Sergeant@merant.com; BCragun.ORM2-1.OREM2@gw.novell.com;
jamsden@us.ibm.com; dgd@cs.bu.edu; ejw@ics.uci.edu
Subject: Re: Version issues


One possibility is that although multiple levels of versioning models
are almost inevitable, by the time we have abstracted out the versioning
models to a protocol, that these distinctions may no longer be visible.

So we may want to pospone discussions about the protocol levels for
when we are discussing the protocol rather than the versioning model.

Cheers,
Geoff

   From: "Chris Kaler (Exchange)" <ckaler@Exchange.Microsoft.com>
   Cc: BCragun.ORM2-1.OREM2@GW.Novell.com, jamsden@us.ibm.com,
dgd@cs.bu.edu,
	   ejw@ics.uci.edu
   Date: Mon, 15 Mar 1999 09:50:14 -0800
   X-Mailer: Internet Mail Service (5.5.2232.9)
   Content-Type: text
   Content-Length: 1363
   X-Lines: 28


   I still vote for two levels.  I believe there is a minimal level that we
   can get people to agree on.  I also believe there is a maximal level that
   we can get people to agree on.  Finding an intermediate level that is
   right for more than a couple of vendors seems unlikely to me, and I'd
hate
   to spend too much time trying to find one.
   [CK] I'm not wild about > 2 levels, but this approach will result in
	limited level 2 support.

   In particular, I find it much more likely that if we carefully design the
   maximal features to be orthogonal, then vendors can pick the subset of
   those features that get them to where they need to be, while still
   maintaining
   a reasonable degree of interoperability based on the shared minimal
feature
   set.
   [CK] I believe that if we do this, then a "de facto" level 1.5 will
	emerge.  Better that we define it up front?

   Another way of phrasing it is that I see the definition of these
   intermediate
   levels to be a prime candidate for the more informal process that JimA
   has described for standard properties.
   [CK] I believe we will do a disservice to versioning and will result
	in non-interoperable servers.  Clients will have to use OPTIONS
	to determine if the server supports there level 1+ features and
	in the end, clients will become too complex (because of the
	degrees of freedom) or interoperable.