The tight connection which exists between the fragment of Prolog
now known by the name Datalog [ul88] and the various calculi and
algebras for Relational Database Systems was observed at several
places in the late 70-ies and early 80-ies. The problem was to make
this idea operational and to build a system which implemented it.
Such systems today are known as Deductive Databases.
We describe the history of a hardly known project from the mid 80-ies
where a prototype realizing this goal was produced. We explain
why the Relational
Database system called Business System 12,
developed by IBM in the Netherlands, and which became operational in
1983,
turned out to provide
the right functionality. We also indicate
how this project influenced subsequent projects aimed at enhancing the
degree of declarativeness in interfaces with database
systems. Evidently the actual developments in the fields of
Databases, Logic Programming and Artificial Intelligence go far
beyond the small part which the authors observed and participated in.
The observations and opinions expressed in this paper therefore don't
pretend to present a complete view on history, but rather to reflect
the insights and positions held by the authors during these
developments.
However, it has never been validated whether
this strategy could be scaled for
real life applications. The potential of a fully compilation
based
implementation of Datalog on an RDMS has been established.
Furthermore its
philosophy had its impact on the RL-1 project started in 1985 at
the
IBM San Jose Research Center and continued during the years 1986
- 1996
at the University of Amsterdam and Groningen. The RL-1 project
focussed on an
extension of Datalog, with features for numeric constraint
solving.
CV:
Ghica van Emde Boas - Lubsen receiveed a B.S.
in Mathematics from
the University of Amsterdam in 1968. In 1969 she joined IBM,
stationed
at the Uithoorn Laboratory in the Netherlands. She has been
involved
with various projects including test-tool
developments for early mainframe operating systems. During the
1980-ies
she was a member of the team that developed the Business System
12 RDMS.
Over the past 10 years she has been working in the field of
object
technology, teaching, and tool development.
Peter van Emde Boas obtained a M.S. in
Mathematics at the University of
Amsterdam in 1969, and a ph.d. in theoretical Computer Science in
1974,
both at the University of Amsterdam. Since 1977 he holds the
chair in
Theoretical Computer Science at this University. From 1964 until
1977 he
held various part-time positions at the CWI in Amsterdam.
His research areas include computation and complexity theory,
semantics, and logic related theory on databases and general
information systems.
The projects reported on in this paper resulted
in a cooperation between
IBM the Netherlands and the University of Amsterdam in between
1983 and 1986.
During this period the two authors spent together the first eight
months
of 1985 at the IBM Research Laboratory in San Jose (now known as
the
Almaden Research Center).