Re: relational algrabra division?
Date: Sun, 21 Dec 2014 16:53:13 -0500
Message-Id: <20141221165313.1a74ca80.jklowden_at_speakeasy.net>
On Sun, 21 Dec 2014 00:25:14 -0800 (PST) compdb_at_hotmail.com wrote:
> the whole idea of relational division is a bit dubious and it is much
> better to approach queries of the kind that it is used for plus
> queries of a more general kind via relation inequalities and "image
> relations".
How "dubious", exactly? The only references to "image relation" I've been able to find come from CJ Date. The search term " (Abstract:image and Abstract:relation and Abstract:algebra)" in the ACM digital library returns 41 entries, only one of which -- Database Explorations: Essays on The Third Manifesto and related topics -- is germane.
So far as I understand them, image relations are *defined* as relational dividends. (Date doesn't use conventional relational algebra notation to define them, so it's hard to be sure.) Then, having more or less assumed their existence, he explains how nice they are are to use as syntactic elements. That's lovely, but hardly a refutation of division as a relational operator. If anything, it's more a case for set-equality as a language feature.
--jkl Received on Sun Dec 21 2014 - 22:53:13 CET