Path: utzoo!news-server.csri.toronto.edu!cs.utexas.edu!uunet!zaphod.mps.ohio-state.edu!sol.ctr.columbia.edu!lll-winken!taco!cepmax.ncsu.edu!jwb
From: jwb@cepmax.ncsu.edu (John W. Baugh Jr.)
Newsgroups: comp.specification
Subject: Database correctness
Message-ID: <1991Mar4.155552.27494@ncsu.edu>
Date: 4 Mar 91 15:55:52 GMT
Sender: news@ncsu.edu (USENET News System)
Reply-To: jwb@cepmax.ncsu.edu
Organization: North Carolina State University
Lines: 12

I'd like some references on what it means for a database (e.g., the
relational kind) to be "correct", the role of "integrity constraints",
and so forth.  Obviously, if one treats a dbms as just an
implementation of a data type then one can define an abstraction
function and do Hoare-style verification.  Otherwise, general dbms
queries may be thought to compromise abstraction by directly accessing
the internal representation (i.e., the database itself).  So what does
one do, just enforce a representation invariant?  Is this what some
call "integrity constraints"?

John Baugh
jwb@cepmax.ncsu.edu