Paul White NZ (4/6/2010)

---

The problem here is not one of being deterministic or not: current versions (including 2008) of SQL Server do not support persisted columns that reference a UDF...

IIRC you need to do some thing like the following in order to let the startup filter work properly:

And (

@bOnlyUnmatched = 0 --...

Interesting, again I learned something new today. Thanks for the reply.

Peter

But that's what the thread was about. The same issue was raised...

Paul made a mistake based on an improper data set. Even...

Still the same mistake. This will match characters like '³' and '²'.

If you are looking for performance I should try Paul's solution as well, with a minor change (replace COUNT(*) with COUNT(DISTINCT name)). Given the small test data set and no...

Well, you were close. I think this...

lmu92 (3/23/2010)

---

Maybe you could make the CASE statement of Lowells nice function a little shorter:

CASE

WHEN (SUBSTRING(@OriginalText,Tally.N,1)) like '[0-Z]'

THEN SUBSTRING(@OriginalText,Tally.N,1)

ELSE '' END

It...

And the modified version returns NULL

Hi Paul, as you just mentioned in another thread, "Never use a UNION where a UNION ALL will do"! I believe the test script contains an error because the comments...

Another way to get rid of the sort is to make all values the same in Tens(n):

Tens (N) AS (SELECT 0 UNION ALL SELECT 0 UNION...

I guess the subtal difference between Wayne's and Paul's solution is the (ORDER BY N) vs. (ORDER BY (SELECT 0)). The last one doesn't require a sort. Check out the...

select

@date = max(obj.date)

from

@Table2 obj

join

@Table1 t1 on t1.date = obj.date and t1.name = 'John'

join

@Table1 t2 on t2.date = obj.date and...

Note that variable names @, @@ or names starting with @@ are deprecated in SQL 2008. See http://msdn.microsoft.com/en-us/library/ms143729.aspx:

Do not use @ or @@ or names that begin with @@ as...