I don't know if anybody else has noticed this, but the implementation of SQL
Server 2000 seems from my limited experience to be absolutely appalling.
Let's forget about the speed comparison between 2000 and 7, let's forget about
any anti-Microsoft sentiments, lets just look at the functionality.
This afternoon I added four columns to a table in a customer testing database
(on request). Shortly after I received complaints that a view of that table was
returning rubbish. I looked into it. The view was returning four blank columns
in the middle, and the data in every column after that was pushed four columns
along. This meant that the column labels had no relationship to the data which
was actually in them!
I copied the SQL that defined the view, and ran it. Sure enough, it returned
different results to the view. It seems almost definite to me that the view is
recording column numbers, and not updating these numbers if the table
definitions are changed. In fact, how this numbering system works boggles the
mind, because according to both Enterprise Manager and intuition, the new
columns should be at the end, and safely out of harm's way!
If you wish to try to replicate this yourself, create a view containing a select
* from at least one table (plus a few other columns from joined tables), then
add new columns to the table and re-run the query afterward. I will be
interested to see if anybody has the same results.
David Cummins
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Structure your ColdFusion code with Fusebox. Get the official book at
http://www.fusionauthority.com/bkinfo.cfm
Archives: http://www.mail-archive.com/[email protected]/
Unsubscribe: http://www.houseoffusion.com/index.cfm?sidebar=lists
