If you remove the dashes from your "uniqueidentifer" examples, you will see that they are both 32-character hexadecimal strings.

Nope. It's based on Dark...

You cannot hide forever...

You underestimate the power of the Dark Side!

Greg J (1/27/2009)

---

... I'm understanding the query more fully now. After checking the numbers, I realized I needed a distinct clause.

Here's my final result. Thanks again a...

[h4]You don't know the power of the...

OK, glad it worked out for you.

How did this work out for you?

I should also add, that my previous numbers were incorrect (left a debugging limiter in), the total time is 30 seconds, compared to Jeffs 7+ seconds.

Only 8 seconds are in...

Garadin (1/25/2009)

---

I get it now. Great explanation. So you've also found that it's quicker to dump repeatedly into the slightly bigger varbinary's than just dumping into one big...

The second performance problem that the unoptimized version has is that the second select is much slower than I thought it would be. Because I know that the conversion...

OK, how it works. First the unoptimized (simple) version:

The idea here is, just like an Update pseudocursor, use a Select pseudocursor to accumulate state across rows, to determine the...

Here is the vastly simpler non-optimized version:

DECLARE @PrevType CHAR(2)

Declare @buffer varchar(MAX)

Select @buffer = ''

--====== Load the buffer with all of the output values

Select TOP 10000--00

@buffer =...

OK, I checked my solution code on 2005, and it works there too In fact its even faster, 8.5 seconds.

Actually, I think that it was just a typo from a left over diagnostic column.

Anyway, here is a slightly reqwritten version of the Load script:

Drop Table TestPCursors

go

CREATE TABLE TestPCursors

...