DSXchange

Add a column generator instead of transformer.

That can use schema.

Are they char or varchar ? What are their lengths ?

Make all varchar and do a trim on A and B.

Is the name hard-coded in the job ?

Why do you want to extract using field() and not define them as seperate fields in the metadata ?

Do your project use a phrase called "archiving" ?

Anyhow, if you want to clear, you can do a

Why from routine?

You can call it from a job using stored procedure stage.

Completely agree with Ray. But one method I think may work...... Get a reject link from your first Oracle insert and funnel it together to pass into second Oracle insert. But the plan will be to include a dummy column generator or modify in the reject link to add all columns you need to insert into ...

Use parameters.

The initial requirement (as updated on 10:19 am) appears to be a simple aggregation followed by ordering by date.

Btw, what did you try ?

Can your OS handle it ? If yes, search for 64 bit hashed files.

Forgot to ask, did you look into the LIKE operator in filter stage ?

Did that help you in any way ?

Will the rows having ids identical to your company's will be discarded completely or loaded in few target but not all ? Also what will be the volume of data processed ? If the volume is small, I will load the patterns in a table and do a spare lookup with pattern matching in the sql. For large volum...

Can you explain more and give some examples.

You have loads of functions apart from index in px.

Wonder why someone will need to do that.

Especially if there is only one column.

The options are
1.) if output is seq file, write with rec delim as none.
2.) if source is seq file, you can remove newline using unix utilities.

You do not have match function in px.

you need to include server transfer and perform it there. will be slow.

What do you plan to do ?