From: PL

To: Editor

In the 6th edition of AN INTRODUCTION TO DATABASE SYSTEMS, in the section on client/server application programming you have a Note that cursors (FETCH loops) should be avoided. Since the copy I have was published in 1994, do you still feel that this is the case? Especially in the changes to the implementations that have gone on over the last 9 years.

Chris Date Responds: My position has not changed. Improvements in implementation have nothing to do with that position.

·   Avoiding FETCH loops: Obviously these cannot be avoided 100%. What I meant was: Don’t use nested FETCH loops to replace (e.g.) joins or unions, which ought to be handled by the DBMS, not the application programmer.

·   Avoiding DELETE/UPDATE CURRENT: These operations violate the relational model, which has no tuple=level operations at all. Moreover, they are not fully defined (semantically, that is) even in SQL.

Posted 06/20/03

[ABOUT] [QUOTES] [LINKS]