If you insist on using the sp_OAxx procedures, I suggest that you read up on them in BOL very carefully. If one doesn't know how to use these properly, it's...

Triggers fire once per statement - not once per row affected.

You need to rewrite your trigger so it can deal with more than one row per statement.

/Kenneth

I'll assume that you know how to get your 'raw' data, and you just want to count/sum up the different numbers...

You can do something like this for the different counts;

SELECT...

While doing checks on FK's, don't forget to also check if it's trustworthy or not..

select OBJECTPROPERTY(object_id(myForeignKey), 'CnstIsNotTrusted')

/Kenneth

The numbers in the cost isn't the most significant thing to compare with in this case.

What you want to look at is the scan count, where scan count = 1...

Why go for the hard way, and what's so non-elegant about an outer join..?

To find the actual gaps - ie the numbers that's really missing, an outer join and a...

Seems to be a problem with the language setting.

Easiest method to avoid it is to use ISO format for the stringdate.

Try '20080101' instead and see what happens.

Also, the convert(datetime,...

Ah, I see...

Well, I was mainly thinking about how to avoid the dupe issue in the first place 😉

/Kenneth

Is it something like this you're looking for?

Problem: get key + latest date - ie select key, max(date) as maxDate

select h.col1,

...

Henrik,

it seems you have a bit of a mess regarding collations in db's vs server installations.

Consider that the choice of collation and sortorder is a fundamental piece of a system's...

In the beginning summary they mention that all concepts...

Heh. Well, you need to do your fair share of testing also.

After all, it's your data =;o)

Especially look out for 'weird' chars (from an ascii point of view) that may...

Hmmm... perhaps the old arcane ways are up for revision...

Did a very tiny test in tempdb on a 2005 box, and it indeed looks like

alter table alter column collate ...

If your existing data may contain extended ascii (åäö and such), you'd probably want to store away your data before changing collation.

The 'old and proven' way is something along these...

There is no such thing as a time and date format in SQL Server.

Assuming you have a datetime datatype, how the values from this will be displayed is up to...