DSXchange

Hello cnu and welcome to DSXchange. This question has been asked numerous times, you can start by using the search function.
Basically, a DataSet is already partitioned for quick and efficient parallel access while a sequential file generally can only be processed, well, sequentially.

What stage are you using to read this data and what Varchar() length did you specify? If you remove the XML stage and no longer get an error in the program, then it probably isn't an issue with input lengths.

The warning message and the segmentation are not necessarily related. If you change the job default to UT8 for test purposes, what happens?

What is the job that you are viewing reading from, i.e. what is the source stage?

Thanks for posting all that - this type of information is what makes DSXchange (with the Search function) so useful to us all!

This looks like an IBM c++ compiler problem that you can search on the internet. Turn off optimization in the stage-specific compiler options for the transform stage, that should do the trick as well.

I think TrimLeadingTrailing is a macro and not its own function, try using TRIM(InColumn,'B') to see if the complexity is reduced.

No, it isn't mandatory to use MSQUERY, but if you have the same problem with a tool outside of DataStage, then it is unlikely you have a DataStage problem - so it is a good way to cross-check connectivity.

Is runtime column propagation enabled in the output link of the stage before the Oracle write?

Can you try the ODBC connection with MSQUERY or another tool of your choice? Does that work correctly?

Most likely in the XML stage, but without additional information it is impossible to say.

Not really - I've never seen this error before. What sort of SQL do you have?

This is a named-pipe error. But you need to supply a bit more information on the job and error in order to diagnose the real cause. Do you always get this error? What happens if you run in a single node?

A 64Bit file can hold 2**64/2 bytes of data. The details of dynamic hashed files are not documented in DataStage but you can look at the IBM UniVerse documentation for details.

DSAttachJob() is a function, not a subroutine. It won't compile or run with a CALL.
Use "JobHandle = DSAttachJob(Arg1,Arg2)"