DSXchange

What kind of delimiter - field, record? Where?

A UNIX newline is a CHAR(10), by the way, depending on the 'where' you can use that function or perhaps the value '010' by itself.

And I've never worked with SQL Server, so you'll need to wait for someone who has to know if what you are doing is appropriate for that database.

In the meantime, where exactly are you running this routine? In a Sequence job? Where is the return value used?

It specifically writes a message to the log to help determine what went wrong. Post it.

Read up on NullToZero() and NullToValue() then use whichever is appropriate. Or simply use IsNull() to test and then assign whatever you like. Sam Ting.

I don't know DB2 but from googling around, it seems you don't need the picture clause. Try:

where B.OPEN_DT = DATE('#ps_FM_DT.p_FM_DT#')

Dates have no internal format. Use the DATE() function in your SQL to convert your parameter to internal format.

Perhaps it would help if you posted this exception or any other error/warning that you are seeing. You should also DSLogInfo() to write your final 'Command' out to the job's log so you can know that it is formatted correctly.

All that sounds good, so unsure why you would be gettting that error. I would suspect either a bug or some kind of issue with 11g, since I don't think it is 'officially' supported. First thing I'd check is that you were up-to-date on your patches, and they can be plentiful. I'm also curious if the j...

Guessing someone or something cleaned out your /tmp directory while the job was running.

Right, it just means the job is aborting. The controlling processes see that and then automatically restart it, which can lead to this never-ending cycle until you put an end to it. As noted, usually a 'normal' DS issue, the same kind you would see if you ran the job a single time outside of this en...

Right - get the source file corrected, right now it is invalid.

1|"car|care"|UK

Your best alternative to this is to do the sql join in your source stage, why the reluctance to do that? The only other approach is an 'advanced' approach and requires a multi-key hashed file (to ensure uniqueness), a VOC record, a UV stage over the hashed file with the 'multi-row result set' option...

Regardless, there is no "try it for free" concept, beta or not.

Is the value in each Table2.a column unique? If not, you have destructive overwrite going on in the hashed file (last one in wins) based on that key and thus less records survive in the hashed file than you expect. It would also mean you don't see the cartesian product in the DataStage job's output ...

D'oh!