Heh... yeah... When I started my post, your post wasn't there yet... I got interupted and came back to it.  When I finally did post (forum hint: I always copy...

Kinda... if you don't consider any type of rounding, the FLOAT 17 digit's in the example I gave came up with the wrong answer...

But, after re-reading your posts, I...

>> For the above if there is no backup with in last 24 hrs then it will show null.

Does that mean the query did want you wanted or not? 

Yeah... didn't even look at that part, Serqiy... just knew the outer table was the wrong one and quit there... good catch.

Heh... the "golden plate" was correct... somebody forgot to wash the glass though JacekO is correct... the method being suggested doesn't use...

Oh boy... you sure about that?

FLOAT is based on BINARY math which is also why it's so fast.  The problem with Binary math is that you sometimes need more bits...

<insert sound of applause here>

Heh... I absolutely agree on these types of posts and I used to get mad as hell at some of these folks... then it dawned on...

Nicely done, Ed...

What I'd really like to know, though, in the original post, Janet said "I would like to call the function with my own datepart."...

My questions would be......

Heh... must be because SQL Server couldn't easily read your code, either.

You need the outer join to be on the database table, not...

The same @Message will be sent every time, it won't change depending on the record.

Ok... dunno about...

It's always a golden plate... and a lot of it isn't available on the web until Serqiy puts it there...

I don't have any websites I can direct you to, but here's a couple of my own...

1. Unless perfectly written parameterized code is used, there is no chance of a cached...

I'm pretty sure that when OSQL is used, any procs called by it are given the same spid.  Of course, that's why testing is important.

Thanks for the heads-up...

Than is impressive.  Thanks, Paul.

Heh... sorry Lynn... wasn't directed specifically at you... was a general comment.  I agree with the idea of them allowing folks to "build comples queries incrementally". 

The only thing I...