Next guess...

set nocount on

use northwind

select

 t1.CustomerID

 , t1.OrderDate

from

 orders t1

where

 t1.OrderDate in

(

 select top 2 --with ties

  t2.OrderDate

 from

  orders t2

 where

  t2.CustomerID = t1.CustomerID

 order by

  t2.OrderDate desc

)

order by

 t1.CustomerID

 , t1.OrderDate desc

set nocount...

Yes, when you're explicite in your conversion it works. The commented line fails because SQL Server is trying to bring 999999 down to a DECIMAL(2,0) which obviously won't work. This...

I might be wrong, but I think this is an Indian measure.

Yes, ddl_admin (or owner of the proc should be sufficient)

What happens when the user fires directly a DROP PROCEDURE statement?

db_owner should be fine. What error do you get?

I'll join the club here.

Since you have no control what is written to which log file, having multiple files will yield you nothing.

...except even more trouble in case...

No, not yet

Btw, ...

Anyoen esle with a similar behaviour?

Already trying to pronounce "ASSUG" or "AUSSQLUG"?

ISNUMERIC behaves very funny at times. SQL Server MVP Steve Kass has posted the following in the public MS newsgroups. Enjoy!

SELECT

 ISNUMERIC('$') AS Money_1

 ,ISNUMERIC('2d3') AS Float_1

 , ISNUMERIC('$+,') AS Money_2

Money_1     Float_1    ...

Yes, your

SELECT

CASE WHEN ScheduleTime1 > ScheduleTime2 THEN

CASE WHEN ScheduleTime1 > ScheduleTime3 THEN ScheduleTime1

ELSE ScheduleTime3

END

ELSE

CASE WHEN ScheduleTime2 > ScheduleTime3 THEN ScheduleTime2

ELSE ScheduleTime3

END

END

FROM ...

and the one I posted

SELECT CASE

 WHEN...

Amazing. Now I must find out what has changed in SP3a that caused this error

Thank you both!

I also work in insurance business and the best thing you can do, is to properly design your schema right from the start. One single table should work just fine...

I agree that the *best* advise here is to properly design your schema. However, if you are stuck with this design, what about one of these?

SET NOCOUNT ON

IF OBJECT_ID('max_t') >...

Well, this is a good guess and usually correct, although not guaranteed to.

...but the OP wants to include a WHERE clause and rowcnt will not give you the information how...

SQL Server bases its guesses on its internal statistical informations. I doubt that you can query them so that they suit your needs. I don't know of a way to...

use tempdb

create table #test

(

 c1 varchar(10)

)

insert into #test values('charcoal')

insert into #test values('c arco l')

select * from #test where c1 like '%[ ]arco[ ]%'

drop table #test

c1        

----------

c arco l

(1 row(s) affected)