Untill you get a deadlock in the system tables and that the system is screwed for all the users...

Yup, it can work, but it takes a derired table to finish the job. Performance is almost the same. But his approach takes lots more typing

Thanx for clarifying. I don't know either if there's a performance difference... but on further analysis I'm not even sure that this query can work the way he's written...

Yup... or someone's a slow typer .

Let us stop guessing, post the actual code and some sample data so that we can see what's really going on.

In what circumstance??? The answer is right, it ALWAYS DEPENDS.

Surely, a scan is bad but n scans is n times worse...

Didn't read the whole script posted, so there might have some more optimizations to do on it.

That too, but I preffer doing the sum(case) approach... but maybe it's not important.

it would look something like this :

dateadd(D, 0, Datediff(D, 0, DateCol)) = DateCol

Can you provide all the informations then??

If you post the original query then I might be able to provide more info.

If you say so .

That could continue for a while... so I'll stop after this one .

Think this can be usefull??

Same query, usable for any amount of cities :

select prod_id

from dbo.prod_cities

where city_id in (207,208)

GROUP by prod_id

having count(*) = 2 -- number of cities in the in list

But where do...

That's almost right .