Reputation: 3446
I use following statement to insert or update current document versions our customers have signed:
MERGE INTO schema.table ccv
USING (select :customer_id as customer_id, :doc_type as doc_type, :version as Version FROM DUAL) n
ON (ccv.customer_id = n.customer_id and ccv.doc_type = n.doc_type)
WHEN MATCHED
THEN UPDATE
set ccv.version = n.version
DELETE WHERE ccv.version is null
WHEN NOT MATCHED
THEN INSERT
( customer_id, doc_type, version)
VALUES
(:customer_id,:doc_type,:version)
Basically I want to avoid inserting on the same condition when I am deleting with DELETE WHERE statement.
The thing is, that there is three different document types (doc_type), but only one or two might be simulatenously signed.
If a client signed a document, then I want to store it's version, if not, then I do not want a record with that document in database.
So, when the new :version is null I delete the existing row. That works great. The problem is, when there were no documents of that customer stored, then oracle actually inserts a record with version = NULL.
How can I avoid inserting records when :version is null?
Is splitting the merge to separate delete, update and insert statement the only way to do that?
Upvotes: 3
Views: 14512
Reputation: 21522
If you're using 10g, you may use conditions for the insert as well:
http://www.oracle-developer.net/display.php?id=310
Rgds.
Upvotes: 3