Hi, I attached a picture of the tmap where you would set the id.
<edit> having trouble attaching picture ... to populate the auto-id out field named myId in a tmap, the following expression was entered: Numeric.sequence("seqMyId",1,1) <edit>
So the input is coming from somewhere, and the output is going to the database with an additional field myId, which is being auto incremented.
This is sequencing at it's most basic, but does not have a lot of error control. For instance, what if the table being populated was loaded from multiple sources (some outside of Talend)? You could potentially end up with a duplicate key error, or multiple records with the same myId value. Yikes.
Example 2.
To further refine this, you could grab the max(myId) from the table and use that as the start for the sequence.
tOracleInput "select max(myId) as maxId from table1"
Now, you would bring the output from table1 into your tmap and the sequence function would be: Numeric.sequence("seqMyId", table1.maxId, 1)
This is better, but you have to hope that all other processes loading the table are following a similar methodology. And this is why the auto-increment logic is often set in the database and not in the ETL tool.
Here is a link for how to setup an auto-id field in Oracle:
http://earlruby.org/2009/01/creating-auto-increment-columns-in-oracle/
So unless you know that Talend will now and forever be the only tool populating the table, or that the other tools using the methodology in example 2, it is probably best to set your sequencing in Oracle, bearing in mind that there could be a performance hit in doing so (although that can be mitigated too).
Thanks,
Ben
