Seal up the mouth of outrage for a while,
Till we can clear these ambiguities
As mentioned at several points in earlier chapters, David McGoveran has a proposal for resolving the ambiguities that can arise in connection with view updating (in the absence of information equivalence, that is). In this final chapter, I want to say something about that proposal. Before I do, however, there are some points I must make absolutely clear:
To say only that his proposal resolves the ambiguities doesn’t begin to do justice to David’s scheme overall. Rather, that scheme is (of course) a scheme for addressing the database updating issue in its entirety. It begins by recognizing that, as I explained in Chapter 2, the only true variable—the only thing that can really be updated, logically speaking—is the entire database (or dbvar, rather, though it doesn’t use that term). The database, and the relvars within that database, are characterized by predicates; however, those predicates, unlike the ones I’ve been talking about in this book, are completely formal in nature. In particular, they completely subsume what in Chapter 2 I called the total database constraint. Update requests too are expressed in terms of such predicates (they might refer to relvars by name for user convenience, but such references serve merely as shorthand for the pertinent relvar predicates). Implementing an update request then consists in combining the database predicates ...