Re: Mutable Revisions

Geoffrey M. Clemm (gclemm@tantalum.atria.com)
Sun, 24 Jan 1999 13:45:43 -0500


Date: Sun, 24 Jan 1999 13:45:43 -0500
Message-Id: <9901241845.AA19472@tantalum>
From: "Geoffrey M. Clemm" <gclemm@tantalum.atria.com>
To: jamsden@us.ibm.com
Cc: ietf-dav-versioning@w3.org
In-Reply-To: <85256703.0059B08A.00@d54mta03.raleigh.ibm.com>
Subject: Re: Mutable Revisions

Having CHECKOUT/CHECKOUT-NEW/CHECKIN as the protocol for both mutable
and immutable clients is a major part of what I think is important.

I don't see what value there is though in having a server that supports
immutable revisions to *ever* not create a new immutable revision on
checkin.  (This is not to say there aren't any, just that there aren't
any that I currently can see).

In contrast, the advantages of having an immutable revision
server always create a new immutable revision on checkin include:

- no protocol constraints or complexity imposed on either a mutable client
  or mutable server
- allows existing immutable revision CM systems to successfuly handle
  all immutable client requests.
- allows immutable clients to create reliable configurations and do
  intelligent merging of data created by mutable clients.

Cheers,
Geoff