set nocount on

create table #temp

(product_id int not null,

prod_desc char(1) null)

insert into #temp

select 1, 'A'

insert into #temp

select 1, 'B'

insert into #temp

select 1, 'C'

insert into #temp

select 2, 'D'

insert into #temp

select 2,...

SELECT @Output = @Output + CASE @Output

WHEN '' THEN prod_desc

ELSE @Output + ', ' + prod_desc

END

FROM #temp

WHERE Product_id = 1

ORDER BY...

That's the one thing that makes me like EM... you can read up the code it generates and learn new stuff.

He meant that something that worked before might not work after this modification. But since you're adding a column instead of removing one I don't see many likely problems....

CREATE UNIQUE INDEX [IX_TypesObjAccess_TypeAccess_TypeName_Unique] ON [dbo].[TypesObjAccess] ([TypeAccess], [TypeName], [PkTypeObjAccess])

WITH

DROP_EXISTING

ON [PRIMARY]

Learning a db software is never a waste of time.. it's just a learning curve.

HTH

that's one way to go...

Try searching these forums for parameter sniffing... this might also be what's causing the headache here.

The foreign key has to refer to a column of same type and length enforced by a unique constraint (index)... it doesn't have to be a primary key nor the...

Can you convert the function to a stored proc while you debug this?

can you compare both execution plan to find where it is different?

Good luck... we'll be waiting.

Have you tried recompiling the udf b4 rerunning the code??

Might be a case of parameter sniffing...

bug testing (checking to see if the subject will have 8 apostrophes this time)..

Strange is there a little bug in the post script??

I see what you're doing, nice trick. But the problem is that it returns only customers with missing entries. Noeld's solution also provides which months/years are missing. ...