Re: Foreign key constraints

All of this information from the respondees to my original posting has been very helpful. Based on what everyone has said I can see that I probably don't need to track down missing indexes because the problem I'm seeing with the application I'm working with isn't due to a problem with locks or lack of locks and that is all I am worried about. Now it will be up to the developers of the app to explore what is going on.

Ken

Jonathan Lewis wrote:

> Yes, of course.
> Sincere apologies to the newsgroup for
> putting out spreading confusion. Please
> ignore (most of) my previous post on
> FK constraints.
> One more silly error and I'm going to
> retire, I think.
> I must have explained this dozens of times -
> clearly the child table has to be locked to
> stop another user from sneaking in an
> orphan after the parent is deleted but
> before the commit takes place. It's
> a consequence of read-consistency and
> readers not blocking writers.
>
> (actually, in parent/child RI processing
> Oracle is doing a sneaky bit of read-uncommitted,
> but only for internal consumption so it's not
> really cheating).
>
Received on Fri Jun 18 1999 - 12:09:11 CDT