Heh... "Worked up" isn't the point... the real fact is, it shouldn't have happened that way and they shouldn't be in the position where they have to shrink it... of...

Yes like bnordberg said this is from a system generated log file.

Then, the bottom line is, you probably...

High praise coming from the likes of you, Peter!  Thank you and I know which "Script Library" you're talking about... I'll throw it in in the next day or two. ...

Thanks for the great feedback, David,

Couldn't believe the pickle someone put you into... had to do something to get you out of that mess

You'll need to grant DDL Admin role privs those users.

Heh... of course if a "lossy" number was stored to begin with, the function will return that same "lossy" number which is what the requirement was... its a FLOAT, not...

...And, now that we've done all of that... I'll bet the damned thing will import using BCP with one of the "native" datatype identifiers... I'll have to try that out...

All right... here ya go... as I always say (in a "Yogi" Berra fashion), "Once ya figure it out, it's easy"

I haven't tried...

'Zactly where I'm headed...

Oh, hell yes!  Got the spreadsheet doing it right... trying to convert it to an SQL function...

It does use an 11 bit mantisa... got the .15626 example and the even numbers through 10 (thinking thats a hint!) to work...  THIS is a bugger!  It would be...

Well, for sure, SQL Server isn't following the IEEE-754 standard.  If you visit the link David pointed out (nice link, by the way... setup a spreadsheet to do all that),...

Actually, there is an ISNUMERIC function (I think... there was one in 2k... did they get rid of it?)

And, actually, you would never want to use it as an ISALLDIGITS...

Hey, have a Happy Easter.  Hope you and yours are doing fine!

Here's the test code I used John... both tables have a composite clustered index on SomeID and SomeNumber...

CREATE VIEW vBIGTEST AS

SELECT *

  FROM BigTest

UNION ALL

SELECT *

  FROM BigTest2

PRINT REPLICATE('-',78)

SELECT...