User bio
404 bio not found
Member since Aug 10, 2022

I have also been playing around now with SQL inserts and updates to see how the DB reacts to uncommitted data.

If I insert a record after a tstart and it has not been committed yet, then in another session I set IsolationMode to 1 and do a select on the table for that specific record.

It does not give me an SQL 100 as I would have expected. It is a new record and hasn't been committed yet.
What it did give is -114 (Unable to get Shared Lock), BUT the values were still put into the binded variables I selected into.

With an update the same thing. I got a SQLCODE of -114, but the new values were actually put into the binded variables.

I guess there is no real uncommitted data functionality then.

I do want it in the Transaction. If the commit of the records related to this event being published fails, this event should not be rolled back too.

Problem is that it may have been picked up by the other process already.

I like the idea of using the PPG and merge after the commit. This does however add more complexity to the code in "remembering" to do the merge after the commit.

Open Exchange applications:
Certifications & Credly badges:
Global Masters badges: