DSXchange

Use stage variables to do group change detection. For the first row in the group, capture the sequence number and for all subsequent rows, decrement it by 1.

I did almost exactly that before posting, worked just fine for me.

Edited to add: just did that exact thing, again worked just fine.

What do you mean by 'binary' in this context? And you don't need to convert anything there, simply send it EBCDIC data to write.

Sure, as long as you include the appropriate ".h" file and add a DEFFUN statement. Search for DEFFUN in the forums for discussions on and examples of this.

Sainath.Srinivasan wrote:Does not matter. DS will still parse and give the error.

I'm sorry, but if you're replying to me about the commented out lines in the sql, you are incorrect. Comments are completely ignored.

"How can i find whether the OCI stg is using that index at run time." It go faster? :wink: As noted, use your tool of choice - Toad, SQl*Plus, etc - to check the explain plan of the DML if you're worried about it. Or have your DBA check for you. Ya Its going faster. I have a test table an...

qutesanju wrote:the same sql is running in sql server so i think datastage must accept it

Not if it's too long.

Sainath.Srinivasan wrote:You appear to be using a temporary table (#temp). But DS might interpret that as a parameter with missing #.

Those parts are commented out.

Yes, there can be and again it centers around just how much C you need to put into the LOB.

Yes, exactly that - an expression that equates to false so no rows go through.

Oh, and off the top of my head:

c) DSPrepareJob() in your own job control code.

You want to reset before the next run, not immediately after the failed run - you lose diagnostic information if you do the latter.

No, you don't. You need to Reset the job, not recompile it.

That's too bad, it can be very helpful when it shows up.

Wow, I'm not sure I'd even consider using the FTP stage for volumes like that.