Of course, you might also want to only do this if fielda has changed...

add this to the where clause about fielda not being null :

... and not exists ( select...

You need to translate your IFs into WHERE clauses...

update table1

set col1 = (select col1_source from inserted where hashFields = table1.unique_id)

where unique_id in (select hashFields from inserted where fielda is not...

No, compatibility mode won't help you there.

For instance, aside from being able to use non-2000 data types, you can use CTEs, CROSS APPLY etc

And even behaviours that are supposed...

Try this:

select Factor,min(PartNumber) as PartFrom,max(PartNumber) as PartTo

from Factors2Process

group BY Factor,stuff(PartNumber,7,5,'')

order by Factor,PartFrom

And by the way - there is no order by on the query in your cursor definition...you should...

Lowell (2/18/2011)

---

Mr Magoo I'd love to hear the back story on this...might be very educational for us....

what happened that made you start thinking something was flooding the server...after looking at...

Roy, you were spot on..I did a bit of research, found out about the idle timeout testing that C3P0 does and that led me to the configuration file that contained...

In this case, I am using CROSS APPLY just as a way of creating placeholders for the calculations.

It just makes the SELECT cleaner - but should not have any other...

Great, thanks very much Roy, that has given me enough...

I think this is what I meant to write just now...

SELECT

sli.itemid ,

sli.title,

qtysold = SUM(qtys.qtysold),

change = SUM(qtys.qtysold - qtys.qtysoldprevious)

FROM

sale_line_item AS sli

JOIN

transactions AS trn

ON

trn.txid=sli.txid

CROSS APPLY

(

SELECT

qtysold...

I agree with Craig, but also - why not try doing it all in one SELECT...?

edit: sorry my bad code removed

GilaMonster (2/17/2011)

---

It does. It reads the leaf level of the smallest index...

How about this one

Ok, I have had a quick check and can't find any noticable difference between them now, so maybe I just remembered something wrong....never mind - no harm done 😛

edit: checking...

I have noticed in the past that COUNT(*) does not always force a full count of all rows

I remember pushing the FIZZBUZZ problem to a trillion rows and used count...

What happens if you try this on both servers (post results back)?

set statistics io on;

set statistics time on;

go

set showplan_text on;

go

select count_big(all column_name)

from your_table

go

set showplan_text off;

go

set statistics time off;

set statistics io...