DSXchange

Welcome aboard. :D

Try surrounding the owner name in double quotes. For example "OPS$OWNER".TABLENAME

Are you sure this is a parallel job? There is no Ereplace() function available in parallel Transformer stage. If it's a server job or a BASIC Transformer stage you can use the LETTERS Transform (which is defined as Oconv(Arg1,"MCA")) - this preserves only alphabetic characters. However, if you have ...

When you use the Package Installer it asks which project for installation. Select all projects. You can not add anything to the Template project; it's not a real project, it's only a template of a project. The reason you can't add things there (unless re-installing the server) is that it does not ha...

The only way is to use a UV stage, and declare the actual value (not the ROWID value) as Key in your metadata. This will be terribly slow since it's a non-key-based lookup. You can improve performance by creating an index on the hashed file. Search the forum for CREATE.INDEX command.

Hegemon wrote:One of the UV files literally has over 10K I-Descriptors defined.

(Staying with the Scottish theme - actually Nac Mac Feegle) Crivens!!

What database type do you specify in the Dynamic RDBMS stage?

If you want to identify which lookup failed you have two choices.

A separate Transformer stage for each lookup.

Separate test conditions on each of the reference input links - either check the link variable NOTFOUND or whether the returned key value is NULL.

There is no IsValid() function (unless you've written one) in server jobs.

Was another job that uses the shared container, or the shared container itself, open in another Designer at the time?

You change the TIME category of your locale. This is probably most easily managed using the NLS.ADMIN menu in the UV account. You will, however, need some familiarity with the ED command line editor in DataStage or with DataStage SQL, since the category definition is a record in a hashed file.

In your new dictionaries prefer I-desciptors (compiled) to correlatives (interpreted), both for reliability and performance.

It doesn't help, except to show that you are determining that some rows are making it into the file by inspecting the file. You must play detective to find out why the file is being cleared? Could it be that whatever you're using to watch it has its file pointer at beginning-of-file and, when you cl...

Convert to string and trim the leading space character. Note that this may produce wrong answer if your numeric value is negative.

I predict that IBM will tell you that you are applying server job thinking to parallel job processing, and leave it at that.

Probably. I can't really say 100% without knowing exactly what the logic is, but if you can build it using the expression editor then it's doable.