Heh... damned straight it's slow... you've written it as you would in Oracle... you've used a "Correlated Subquery" in the SET clause and that means that the query must be...

Unless the worksheet is protected against mistakes like adding an unexpected column or hitting the space-bar to "delete" data, you may get some wildly unexpected results and failures using Excel...

You actually expect an Access ODBC driver to operate as fast as an SQL Server OLE DB driver when used against SQL SERVER?   ...

Heh... I have to laugh at what they call "best practices" for most performance issues... no where do they say "review all code for possible performance issues before it goes...

Can you post the code for your final solution, please?

Thank you for the feedback and I have a very high speed solution in mind... but I need to know what you want done with the data that I posted......

First, no need for a cursor in triggers... ever!  You can simply join to either the INSERTED table, the DELETED table, or both.

Second, I recommend that you do NOT log...

Ya gotta love those that don't even try the suggestions given... guess these "experts" can just look and see "bad code"

I'm not sure...

Last, but not least... security... everybody can make a table in TempDB... might not be true on the "given" DB for whatever user the sproc is logged in as...

AND, TempDB...

No, pretty sure there's nothing built in... you'd have to do a SELECT INTO a temp table and then read the datatype from the columns something like this...

    ...

I agree with Rob and Alex... I'd also like to add that using something that you know "could" fail but hasn't so far (UDF_ID function) is a bit like sitting...

So... if there are, say, 7 dates that are all where the startdate of the next row is 1 greater than the enddate of the previous row like this...

1;AA;20070921;20070922;10

1;AA;20070922;20070923;20

1;AA;20070923;20070924;30

1;AA;20070924;20070925;40

1;AA;20070925;20070926;50

1;AA;20070926;20070927;60

1;AA;20070927;20070928;70

... would the...

The problem with selecting either the MAX or the TOP 1 DESC, even in the presence of the read locks that Andy included in the code, is that you only...

Ok, Cees... I was just making sure that you had everything that you needed.  And Lowell's snippet above will take care of the zeros...

Wow... let's triple the work the server does... some vendors just don't get it...

Thanks for the feedback, Jim...