DSXchange

Execute each routine via a Routine activity in a job sequence and use the activity variable $ReturnValue from each to supply the value to the job parameters.

Exactly as I posted earlier. Use stage variables for check for "changed" (Col1 and Col2) and "larger" (Col3) in data sorted by Col1, Col2 and Col3.
Derive your output columns accordingly.

Open the job sequence. Open its job properties. Go to the Dependencies tab. All the jobs used in the job sequence should be listed there. As to which ones are actually being called, based on the logic in the job sequence, the only place from which you can obtain this information is the log of the jo...

There's nothing out of the box as a function in parallel jobs; instead there are three specific stage types for comparing rows to determine whether two rows are different (changed). Try not thinking like a server job developer. Otherwise you can write your own version of the CRC32() function in C++.

Once you have upgraded to version 8 you will have a resource estimation tool at your fingertips. The main things that you need to estimate on the technical side are to do with resource consumption: CPU, memory, disk space and so on. I suspect you are not looking for this at all, but rather how to es...

You use the two fields Data Source Type and Data Source Name as the two components of the Category name in a table definition.

(This changes in version 8.)

I don't think there's a quick fix - you'll have to settle for revenge.

Transformer stage with a complex set of If..Then..Else tests.

Database query (if available) using COALESCE function.

NLS_LANG is an environment variable used for database (e.g. Oracle). Although DataStage does not use it, DataStage checks that it is set.

Not me. Have you tried to determine any pattern to the behaviour? Are any warnings logged? Your description is a little too broad to afford effective diagnosis.

We do not diagnose messages from APT_CombinedOperatorController.

Disable operator combination to learn which stage (operator) is generating the message. This may shed some light on what is happening.

What is the data type of the source column?

That's true, but the question was about "change the query".

If the grouping columns are indexed that is another argument for performing the summary in the database; it can leverage the indexes to establish the groupings.

Change the value of APT_STRING_PADCHAR to something other than 0x00.