If security is a big factor. then you need to get to the bottom of this...

Work with everone, *NIX, cluster, firewall.. well, all that may be involved. Find out what...

Well, Steve has got a lot better head for numbers than I do

This trimming business seems to be trickier than it first seemed....

Great - looks better now.

Have you read what BOL has on the subject?

/Kenneth

Unfortunately, if the choice is between 'fast but not correct' and 'slow but correct', and at the same time 'slow' is deemed as 'not working', you're between a rock and...

Are you absolutely sure about this?

I tried your example - just copy & paste into QA and got:

Table '#temp1________ ...... Scan count 5, logical reads 5, physical...

Well, does the kb article apply to you? Are you on the latest sp? Do you have parallellism set to ON? If so, tried turning it off?

/Kenneth

The general recommendation (from me and some others) since 7.x, is to default to set parallellism OFF, and only turn it on with individual querys proven to benefit from it and also...

Another method (that also allows you to not use dynamic SQL) is to take the in parameter, and instead of tokenize it, insert into a temptable or tempvariable, then join...

osql uses ODBC, perhaps you have conflicting conversion settings there?

Have you tried loading with bcp? bcp offers some switches regarding collation and format.

/Kenneth

Read it as carefully as possible - no pics in my browser, so I have no idea what the table or data looks like. (it's preferrable to post DDL and data...

(since col1 is int, I also return an int)

Here's my take on it - assuming a 20% trim on the overall data.

January 11, 2005 at 9:13 am