One small note on your first point ...
I agree that doing a count on a numeric column can lead to
misunderstanding, but take the case of doing a count on a text column
or date column, then it becomes very clear. My point is that
count(column) does have something to do with the column value, it will
only count non null value... and I have come across real business case
where we would do both count(column), count(*) in the same select
statement (but we didn't care about distinct values).
Question on your second option:
Do you know if there is a "significant" cost incured with the inline
view. If people had a million rows and since we don't do any row
triming, would there be a sub table built with 2 columns and a million
rows ? (I still think it's good, I am just curioius)
Received on Wed Jan 12 2005 - 14:02:36 CST
