From: Paul Brannan <pbrannan@...>
Date: 2002-09-17T22:36:36+09:00
Subject: Re: Dependency "trees" - suggestions?

On Tue, Sep 17, 2002 at 10:30:30AM +0900, Massimiliano Mirra wrote:
> Yes.  This means that for each node I add, the entire graph has to be
> traversed again looking checking for cycles.

OIC, you aren't interested in protecting against a corrupted database;
you just want to prevent the user from specifying a cyclic dependency.

> I'm wondering whether a less expensive method exists (and less
> convoluted than the one I'm using now) maybe involving some tricks
> with adiancency or incidence matrices.

Probably.  I don't know enough about graphs to say.

Paul