Nicely done... still going to take a fair bit more than that if table has more than 1 DateTime column, has no DateTime column, or, as you said, has different...

Very cool... nice job, Antonio... takes only 28 seconds on the million row test...

1.  Don't use anything but SQLCHAR when importing text files.

2.  Don't use anything but "" for coalation on non-character based columns.

Heh... thanks Chris.  And, nice job, by the way

But... try this... one less join = faster performance

Dinendra,

The question remains... what are you doing where you think it might be a good idea to find a way to sort without using ORDER BY?

David,

Here's an attempt to explain the "errors" that creep in... same "error" bit me big time... Lynn pointed it out to me and Serqiy gave me the tool that explained...

C'mon old friend...

Hey folks... I screwed up way back when and I've crossed out my posts about this because they were wrong... what I saw...

Crud... there goes the neighborhood   Just kiddin' guys... but cut it out, please.

Serqiy, ran the query you posted you were absolutely correct on...

With a fair amount of dynamic SQL...

Do you mean you want to do an "auto discovery" of each table having any DateTime column...

Antonio,

Heh... had to do a double take... but no... it's working correctly because of the DISTINCT transfer and the GROUP BY... here's the results with the original data and your data...

Probably, Serqiy.... just I think they did it at a machine language level which would be a lot faster the at the "T-SQL Level". 

Heh... I agree about the OO hints......

Yeah... and it get's worse... check it out... didn't use to work this way...

select str(612.005, 10, 2) UNION ALL

select str(612.015, 10, 2) UNION...