Why not just... don't even need a UDF this way...

DATEADD(mm,DATEDIFF(mm,0,somedate)+1,0)-1

Wow... I made a mess of that one :blush: It's only fast if it works right, huh 😀

Folks, don't use method 3... Tomm's right... I've got a nasty bug...

Thanks for the heads up, Tomm... I'll check it out...

Are you reporting on all 3 years at the same time?

Heh... and if they change the way Order By works, we'd all be screwed, too!...

...Wait, they did that 😀

dhay1999 (10/4/2007)

---

Jeff,

If you are still around I'd like to see a skeleton of what you accomplished. I'm trying to do the same thing with over 150 million rows. ...

noeld (10/3/2007)

---

Ok that's the statement. Where does it says that it will update the varibles following a certain order ?

We know that, so far, we have been able to force...

Nothing undocumented... you need to look in BOL for the UPDATE... I've bolded the part of the update I used....

UPDATE

{

...

Tomm Carr (10/3/2007)

---

Sorry, Jeff. You got the third method to operate a little faster by removing the setup just before entering the loop. This is a crucial step. Without it,...

Peter,

Although the double correlated subquery will work, I'm also pretty sure it will kill batch performance because of the triangular join in it...

http://www.sqlservercentral.com/Forums/Topic359124-338-1.aspx#bm360151

Vishal,

Just a thought... Is this a permanent nonreporting you're modifiying? The reason I ask is if you're actually doing this to some "real" data (other than a reporting table),...

I still don't know what "undocumented behaviour" Noel is talking about...

The maximum length of a row can be 8060 bytes. In this case, some of the...

Actually, all that stuff worked... for about the first two days...

Heh... already are... hardcore DBA just wouldn't allow this.