DSXchange

No. And yes. Because, in version 8, QualityStage and DataStage are the same piece (ignoring licensing) you can use a DataStage Transformer stage to apply IsValid() functions or other techniques to effect data type validation.

Given that you've marked this as a server job:

That's nothing to do with Ereplace(). The problem is that you used "END IF" when you should have used "END". Refer to the DataStage BASIC manual for syntax of the IF statement.

DataStage is not really the right tool for analyzing data. Have you investigated Information Analyzer?

DataStage is not really the right tool for analyzing data. Have you investigated Information Analyzer?

There are no libraries. You do, however, need a correct syntax. Since you have posted neither your specification nor your syntax, it's impossible to comment further.

What happens when you let DataStage generate the SQL based on the table definition you have imported?

in a Transformer stage

Then change the design so that missing columns do not cause aborts.

One approach is to use a server job, where you can build a missing column rule into the Columns grid itself in the Sequential File stage.

Why not use a Sort stage on all three columns and generate a Key Change column?

The ODBC drivers are on your UNIX server, not on the client machine.

See how vitally important accurate specifications are!

Only now could someone work on this implementation without wasting time.

Now, to complete the specification, do you need these in separate rows or in separate columns or in some other format (such as a delimited string)?

If the lock is owned by any other user, you need to execute the UNLOCK command while logged in as a DataStage administrator.

Start the process is easy - use the dsjob command. Capturing the parameters should be fairly easy too, assuming they're stored somewhere sensible (though you do need to consider how to protect sensitive data such as passwords). You can build all of that into a script. There are examples posted on DS...

Do you really want overlapping ranges, most of which are N+1 days, or not? Did you really mean
20091001 and 20091003
20091004 and 20091006
20091007 and 20091009