This week's book giveaway is in the OCAJP 8 forum. We're giving away four copies of OCA Java SE 8 Programmer I Study Guide and have Edward Finegan & Robert Liguori on-line! See this thread for details.
I'm very new to JPA and have inherited some code which looks pretty inefficient. It's job is to cancel some cashflows in the database by updating the original cashflow's isLatestVersion flag and endDate and inserting new versions with a status of cancelled. To do this it gets all the cashflows to update and then loops through each one, cloning it, updating fields, and then calling EntityManager.persist() for the new cashflow and EntityManager.merge() for the updated cashflow. The code looks like:
Is there a way in JPA to do the inserts in one go without having to loop through each one?
Any other ways to make this more efficient? I don't think there's any transaction flags set anywhere so would doing the work inside a transaction help speed it up (I'm assuming the default is AutoCommit or something similar)?
You cannot persist anything in JPA without a transaction, so I assume you have one somewhere.
Normally in large batch situations splitting the batch into smaller batches per transaction (say 1000) is more efficient, but this depends on how big the batch is.
Most JPA providers support batch writing, which depending on the database can improve performance dramatically.
You seem to be recording history, instead of using persist, you could just create an update trigger in your database that does this. If you are using EclipseLink, it provides history support that automates the historization and allows historical querying.