I fear it may already be too late :laugh:

A cheaper room-mate? :blink:

-=JLK=- (10/11/2010)

---

Paul, Thanks for the example that recreates the problem in miniture and with my own tables. Quite helpful in seeing that the problem lies with the function and...

It's the order of the rows arriving at...

Modified version that writes the result of the function call to an ordinary column:

USE Sandpit;

GO

DROP TABLE

...

For anyone else that would like to look at this and see why T-SQL scalar functions that do data access are such a pig, here's a cut-down repro script:

USE ...

tfifield (10/9/2010)

---

The begin_date in skClassBlock is computed and this is then joined to the calendar table. I would make it a persisted computed column and then create an index...

That appears to be the current plan:

https://connect.microsoft.com/SQLServer/feedback/details/387616/tables-node-does-not-show-all-schemas-in-ssmse-2008

The Microsoft comments refer to it as SQL11 with a planned...

I see. Well, I went back and read it again, just...

From whom was that advice obtained, and what were the circumstances? It's not (necessarily) bad advice, but I'm interested to hear...

No, what makes you think it might have been?

The misconceptions concerning shrinking versus truncating logs might make for a better '2sday' blog...

Not this time 😉

The most useful thing at this stage would probably be an actual execution plan. Run the problem query with 'Actual Execution Plan' on, right-click the resulting graphical plan, save...

Have a read...

Oliiii (10/5/2010)

---

I've ran my test on a DB without any row-versioning, the second query select still goes trough no problem even though there is an exclusive lock on the row...