From info-tsql-sender Mon Jul  5 15:45:31 1993
Date: Mon, 5 Jul 1993 15:32:21 MST
From: "Rick Snodgrass" <rts>
To: tsql@cs.arizona.edu
Subject: SQL temporal language extensions
Content-Length: 3588
X-Lines: 81
Status: RO

One major objective of the temporal database infrastructure workshop
(held three weeks ago in Arlington, TX) was to initiate the design of
a consensus temporal extension of SQL.  There were three conflicting
viewpoints on such an extension voiced by the workshop participants
(with associated rationales, not repeated here):

(a) With the addition of an interval data type, there will be
sufficient support in SQL2/3's data model to support applications
using temporal data.  Further temporal support within the data
model should not be added.

(b) SQL2, and the proposed SQL3, require temporal support to be added
to their data model and query language. A two-pronged effort should be
initiated, the first being a short-term effort to define a temporal
extension to SQL2 and the second being a long-term effort to define a
comprehensive extension to SQL3.

(c) Temporal support should be added, but only SQL3 should be
extended.

These three viewpoints are clearly at variance. So, instead of a
single consensual effort, which appears to be unattainable, three
separate efforts have been initiated, each reflecting the approach
espoused by a significant portion of the TDB community, and together
enabling further progress. The following working groups have been
started. If you would like to participate in one of these working
groups, please contact the group coordinator.

* The SQL2/3 working group
	Those agreeing with viewpoint (a) will define an interval data
type and write SQL2 (or SQL3) queries for the benchmark queries, so
that this approach can be compared with other proposals. Nikos
Lorentzos <elio@isosun.ariadne-t.gr> has volunteered to coordinate
this effort.

* The TSQL2 working group
	Those agreeing with viewpoint (b) will define a short-term
temporal extension to SQL2. I have volunteered to coordinate this
effort.

* The TSQL3 working group
	Those agreeing with viewpoint (b) or with viewpoint (c) will
define a long-term temporal extension to SQL3. Gene Wuu
<wuu@ctt.bellcore.com> has volunteered to coordinate this effort.

The initial reports of these working groups, if they are completed by
August 23, will be attached to final report of the workshop.  However,
the activities of these working groups are also relevant apart from
the workshop, and so should not be prescribed by it.  The workshop can
be viewed as an impetus for further consensual infrastructure
activities withing the temporal database community.

The specific charters of the working groups will be determined by the
coordinators. I now turn to discussing the particulars of the TSQL2
working group.

The TSQL2 working group design will proceed in the context of the
following constraints.

* The initial design must be completed by August 23, 1993.

* The design will be based on SQL2. In particular, none of the
	significant extensions envisioned in SQL3 will be assumed
	to be present.

* Initially, the language design will support user-defined time and
	valid time. Support for transaction time may be added later.

* The design will be implementable in the context of conventional
	relational database technology. In particular, implementations
	of the data model and query language constructs will be
	possible without necessitating major changes to conventional
	database management systems.

* Discrete time will be assumed.

* Past and future valid time will be supported.

The activities of this working group will be parallel to those of the
SQL2/3 and TSQL3 working groups. Interaction in the medium term
between these groups will attempt to arrive at consistent approaches.