I agree... in SQL Server 2005, you could probably do it with a recursive CTE... But not in 2K.  Might be able to do it using a recursive UDF, but...

You bet... thanks for the feedback, Matt.

Ya do pretty darned well yourself, Lowell! You made a very useful tool.  I'm in the process of tweaking it for my own use and thought...

Nicely done, Lowell... you might want to check the ##Tmp table, though... the numeric datatypes always end up having a MaxActualLength of "0" and I think that's probably not right....

This will not produce the results you expect if there is a tied date for any give Job/Status combo... nor is it very good on really large tables.  But it...

Sure... can be done in any language... I just got bogged down at work.

A woman's sister dies of supposedly natural causes. The woman goes to her sister's funeral. During the funeral, she sits next to a man who consoles her quite nicely...

"One measurement is worth a thousand speculations"

Mathew,

Index Scans are NOT always a bad thing... here's proof (and a couple o' tricks)...

First, let's make a million row test table...

--=====...

Recommend rewriting the following using a derived table with a proper join instead of the correlated sub-query that you have...

and sc.CONF_id =

(select max(sc2.CONF_id)

from sep_CONF sc2 , main_score es2...

Thanks, Gail

Ah... I see what you're getting at...

Yes, in that case, if the inserted row size is greater that than free size remaining on the target page due to a clustered...

Nah... this one...   One individual had a devil of a good time leading others on just for the sake of a good bar room brawl.  I rather enjoyed...

Cool... thanks, Michael. 

Yeah, clearing the cashe didn't seem to matter in my testing so I didn't include it in my code.  And, I very much like the idea of using...