# Re: Join types

Date: Fri, 25 Jan 2008 09:03:06 -0800 (PST)

Message-ID: <65073ec5-2125-45b2-a2a3-837777320eaf_at_u10g2000prn.googlegroups.com>

On Jan 25, 3:04 am, Gints Plivna <gints.pli..._at_gmail.com> wrote:

> I'm a bit studying join types and trying to make a visualisation of

*> relations among them. As a result I've created an ER diagram
**> describing relations among them and it can be found herehttp://gplivna.blogspot.com/2008/01/sql-join-types-im-studying-bit-sq...
**> I've tried to find something like that using google however the best I
**> could get was textual description. I'm not mathematician and studied
**> set theory a bit 10 years ago in university and almost all have no
**> forgotten :) so maybe it has some problems from set theory viewpoint.
**> So question is - is it generally ok? If you know anything similar
**> please add link either here or in my blog post. All comments welcome!
**> Thanks!
**> Gints
*

Join is indded the most important operation in the RM. However. SQL classification of joins is quite ad-hock and not particularly illuminating. There are many more interesting joins around.

The "normal" natural join is something that can be described as set intersection join. There are also set containment join (aka relational division) and symmetrized form of set containment join -- set equality join.

The other direction is comparing joins in Binary Relation algebras with joins in RA. In Relation algebras join is not commutative and is not a generalisation of set intersection. Received on Fri Jan 25 2008 - 18:03:06 CET