Re: Date and McGoveran comments on view updating 'problem'

Bob Badour wrote:
...
> A more important question is: How would one write a constraint that
> deletes from SP ^ S affect SP but not S?
> ...

Using a typical relational algebra, my guess is that it would be extremely lengthy, much longer than enumerating all the domains involved. Eg., a disjunction with N-1 <OR> operations where N is number of possible base headings multiplied by the number of possible combinations of tuples in those relations, multiplied again by at least three (since there are nominally three relations involved)! So, I don't think it is a practical question as far as the RM as I know it is concerned.

Some language notation could be a shorthand for this, eg., S' = S might mean that an S relvar can never be deleted from nor inserted to. One might resort to a kind of catalog expression to establish some kind of prototypical delete form where S' NOTEQUAL S MINUS X and X stands for any possible relation, ie., any possible MINUS operand. But I'd say the question is outside the realm of the algebra and therefore out of the realm of the calculus (even though I know very little of the calculus).   Maybe if the RM admitted recursive structures, some algebra could handle it and therefore some kind of terse algebraic constraint might be possible.

Just guesses ... Received on Wed Dec 17 2008 - 17:29:46 CET