Ack... my fault... I'm so data-centric that I sometimes forget that SQL Server has to live on a Windows server (except for MSDE, Development Edition, etc).

Nope, I don't know how...

Nice code... just a bit of food for thought, though... correlated sub-queries with triangular joins (a<b) can be as bad or worse than cursors on large tables.  Not advocating cursors...

I'll bite ... what is "E&OE"?

It's not free but you could buy SQL Server Developer's Edition for about 64 bucks if you look around.

Noeld probably hit the nail on the head... the server should be much faster than your desktop machine for multiple reasons...

1.  Memory... chances are your desktop computer has between 512Mb...

Mick,

Yes... you can save pictures in IMAGE datatypes... but it's a really bad thing to do for more reasons than I'd care to post here...

Most folks have seen the light...

TempDb is rebuilt from the ground up on a reboot... this does the trick...

 SELECT CrDate AS LastRebootDate

   FROM Master.dbo.SysDatabases

  WHERE Name = 'TempDB'

Vamsi,

Although there's nothing specific about using character arrays to process the W, o, r, d, s, could you be a bit more specific about what you are trying to do?  There's...

If the suppliers of the data have agreed to provide the data with correct row delimiters,

I guess I don't need to do much.  Too bad... I was "dying" to...

Scott,

Post the first 10 records of data (including the header) and let me take a crack at it tonight for ya...  I think there may be a couple of ways...

Prakash,

Serqiy is, although a bit brusk, still absolutely correct.  Why do you need a cursor at all?  And building one dynamically is like the worst of two worlds.  Why can't...

Thanks, DC and Ieea,

I always thought cursors were a part of SQL Server... didn't know they were actually added as a "feature" at...

Just following up... here's a cool problem that I just ran across that I was able to answer because I'd done something like it before...

Here's the problem...

You have a payment...

Absolutely concur!  I was just demonstrating one method of getting the CPU time expended by a single spid.  I usually get the I/O the same way as well as measuring...

Reet,

You'll undoubtably get several suggestions using a self-joined correlated sub-query which are usually pretty darned slow.  They'll usually take about 29 seconds to do 10,000 records and the duration get's exponentially...