Re: Many-to-many with both FK cascading deletes?

From: Erwin <e.smout_at_myonline.be>
Date: Wed, 23 Jun 2010 13:31:23 -0700 (PDT)
Message-ID: <f56e1ba9-29f1-4f84-a40b-829899554666_at_e5g2000yqn.googlegroups.com>


On 23 jun, 16:38, -CELKO- <jcelko..._at_earthlink.net> wrote:
> Some of these cycle problems are known to be ambigous or NP-complete.
> Add in deferred constraints and it is a real mess. The safety rule is
> to keep the cascades paths as trees -- no cycles.

Sorry, but I'm mystified.

Where did the OP mention "cycle problems" ?

The only thing I had the impression the OP was talking of was a junction table J between R1 and R2, with a cascade delete rule from R1 to J, as well as from R2 to J.

You seem to have gotten a different impression.

And then, second.

While of course I recognise that cycles hold the danger of infinite recursion when processing them, isn't your assertion too strong ?

Personally, I would say that it is sufficient to keep the cascades paths as DAGs. No ? Received on Wed Jun 23 2010 - 22:31:23 CEST

Original text of this message