DSXchange

The metadata are stored in shared tables in the combined repository (XMETA). Tools read and write it there. On that basis, the concept of IA pushing metadata to MWB, or even of metadata being "in" MWB, are wide of the mark. No matter which tools import the metadata, it ends up (or can end ...

It is accurate. It is still accurate. Of course, all DataStage parallel jobs run osh. Information Analyzer bypasses the graphical design interface and instead generates appropriate osh scripts directly. Because all required frequencies are - once column analysis has been completed - in IADB, the oth...

The question is about before/after subroutines.

Yes but, before we do, first explain why these characters are "special" and whether you have the permission of the owner of the data to make such a change.

Prove that it's a "problem", rather than "the way it works".

Before/after subroutines are, however, written in DataStage BASIC.

The reference manual is DataStage BASIC.

A training class Programming with DataStage BASIC, which includes a module on before/after subroutines, is available on DVD from DSXchange Learning Center

You must take control of what happens. Partitioning on the sort key will not destroy sorted order (think about it).

It's your first choice, with the added benefit that it can process only one key value at a time from the driver input, and therefore can (because both inputs are sorted) free that memory once all the rows from the probe input have been collected and joined. And THEY are all adjacent because the data...

A Search of DSXchange will suggest that what you want to do is not easily possible, though it IS possible to pass the entire SQL statement as a parameter and to use that as user-defined SQL.

I'd still advise to use Director. As soon as you want to delete jobs as well as categories there is an awful lot of work that has to be done on the server. Deleting an empty category is easy; deleting a category in which jobs exist not so.

Would the Num() function be adequate for your purpose?

More likely to be a problem in a row whose row number is more than 700K.
What's the data type of the key? If VarChar, are they identically trimmed? If not, are the data types identical on stream and reference inputs?

No you don't have to turn it off, but the importer will always assume Unicode while it is on. You can edit the imported table definitions if you prefer.

The message about data being truncated is coming from the database server (ODBC driver), not from the Transformer stage. Verify that the table metadata correspond to what's in your job. Establish a trace to determine which rows are generating the fault, and isolate the cause by that means.

Exception Handler is only triggered by a failure, not by a warning.

If you want to manage warnings, you need explicit triggers.