I've stayed away from using functions, but the CROSS & OUTER apply statements work incredibly well with set based SELECT statements applied to the rows returned by the rest of...

To do that automatically you can either buy a tool like Embarcadero Rapid SQL that will do it for you, write a program using SMO to walk each table generating...

You're asking for improved methods of Row-By-Agonizing-Row (RBAR) processing. That type of processing is inherently slow because TSQL is set up & optimized for set based processing. You're almost guaranteed...

Sorry that wasn't helpful. I thought that covered most everything. However, for drill down, one place I go is the Query Procesing Team blog. This is the Storage Engine team's...

I've tested in 2008 and 2005. I also change the compatibility level of the database I was testing on and it worked on all of them, even 7.0. I'm not...

Did you mean to post this in the 2008 forum? If so, I tested your script on a 2008 database and it ran just fine. I'm not sure where the...

I found Itzik Ben Gan's course on advanced TSQL very helpful. He covered all the new TSQL functionality in great depth. He's from Solid Quality Mentors.

Have you performance tested both approaches? I think the CTE approach, especially the second one, not mine, should be much more effecient than loading data into a temp table.

Automatic verification of changes to the scripts as you save them is a pretty large advantage. Refactoring object names is great. The Static Code Analysis is getting a lot more...

If you can pass the data in as XML, you can use the XQUERY language in 2005 to pull the data out of it, exactly like a table. It looks...

Most of the time, cursors just flat out run too slow. Rewriting the procedures to avoid the use of cursors is usually the best bet. Other than that, I'm not...

This white paper from Microsoft is one of the best sources for this information.

Itzik is the greatest.

It sounds as if you're doing some sort of data migration. If you can, you should look into, in some way, turning this into a bulk load....

The long answer is: it depends.

Let's assume that the 8 column table is supposed to be retained in it's entirety and that you're supposed to create that, load it, and...

As you've already seen, no they're not both the same. Yes, the number of physical reads is the same, but you'll find that this procedure recompiles, a lot, probably every...