Re: Row locking and serializability

There is an interesting philosophical question there.

The example simply has:

    transaction A inserts a (child) row     transaction B deletes a (parent) row

Arguably the fact that there is a business rule that state:

    child rows should not exist unless the parent row exists has nothing to do with serialisable transactions.

If we change transaction A to :

    lock the parent row and insert a child row if the parent is locked and transaction B to:

    delete the parent row if no children exist

then we have two transactions which are inherently not serialisable and the discussion becomes pointless.

I believe the point of this particular section of the manual was simply to demonstrate that serialisable transactions do not, BY THEMSELVES, guarantee referential integrity.

--

Jonathan Lewis
Yet another Oracle-related web site: www.jlcomp.demon.co.uk

Michael Ortega-Binderberger wrote in message ...
>I disagree. The fact that as the original mail suggests, you can
>construct and example where it breaks down, means that Oracles
>serializability is not what is commonly assumed.
Received on Wed Jun 09 1999 - 15:23:46 CDT