FishingIsGoodДжо Селко - дядька популярный и словоохотливый.
Спору нет. Но к нему есть и претензии. Вот цитата с сайта Паскаля/Дейта :
http://www.dbdebunk.com/page/page/1490837.htm

(NOT) MORE ON CELKO, PLEASE!
with Fabian Pascal

From: JH
To: Editor
Date: 04 Oct 2004

Joe Celko has often hyped using nested sets to represent trees in a SQL DBMS. My introduction to Celko and nested sets was his post to a mailing list on this subject in 2001, and the idea intrigued me. There was just one problem. His examples did not work!

On the chance that I could have been using a worse-than-usual SQL implementation, I tried everything I could get my hands on, yet the results were the same. Celko goofed! Somebody that was an alleged expert had published something that was completely wrong. Not just a single post, but every single article on this subject for at least six years. Duplicated in everything from magazines to his book, SQL FOR SMARTIES.


I set out trying to understand how nested sets work. It took some time, but I eventually figured it out and sent him the message included below.

I've been playing around with the nested set model for a little while, thanks to your articles. However, I think I've found an error that has been duplicated in all instances of the article and the message(s) posted to the Postgres mailing list.

1. An employee and all their Supervisors, no matter how deep the tree.


2. The employee and all subordinates. There is a nice symmetry here.


I've tried this on Postgres, Oracle 9i, and mySQL, and get the same results for the second query:


It should be a select from p1.* and not p2.*, which is also a bit more symmetrical. The results for selecting from p1 instead of p2:


His response was brief. "Shit. You're right."


From: Fabian Pascal
To: JH

Too bad you did not check with me first. There is absolutely nothing that will persuade me to consider anything that has to do with Celko. 100% waste of time.