Quit worrying about the storage used by various products; they will make
more and chepaer storage for you. The real question is how to use them
in a data model. My heuristics are:

1) Avoid them if you can. Use a dummy value and put it into a DEFAULT
clause, if the domain of your column needs it.

2) If you have a NULL, make usre it has one and only one meaning in that
domain. Example: The ICD codes for disease use 000.000 for
"Undiagnosed" and 999.999 for "Diagnosed, and we don't know what it is"
for two kinds of missing data.

3) Make sure the propagation property of NULLs makes sense in your data
model. Otherwise, use COALESCE() to replace it with another value.

4) Remember that you must use "CAST (NULL AS <datatype>)" in places
where the compiler needs to know the datatype of a column or variable.

--CELKO--
===========================
Please post DDL, so that people do not have to guess what the keys,
constraints, Declarative Referential Integrity, datatypes, etc. in your
schema are.

*** Sent via Developersdex http://www.developersdex.com ***
Don't just participate in USENET...get rewarded for it!