as per jdbc spec 3.0 a transaction for a jdbc driver is a single sql statement. but what if one wants application level transaction, i.e suppose i want:
transaction begin: select a select b update c update d delete e transaction end.
i want to create such transactions which span beyond a single sql statement, i.e are a combination of multiple sql statements. i want to implement such transactions in servlets. i take a connection from a connection pool via DataSource object(on weblogic server). then should i set auto commit to false, carry out the sql statements, and commit. is this the right approach or is it that in weblogic, every single connection from the connection pool forms a single transaction?
What you describe is the correct way to perform a mulit-statement transaction in JDBC. I don't know about Weblogic ... perhaps one for the Weblogic forum?
I expect that the issue of how you go about committing would depend on whether your transaction is distributed, e.g. whether you're using full J2EE with container managed persistence and so on.
I'm no expert on J2EE, but if you're just using good ole J2SE under Weblogic then your approach is fine.
What you have specified is one way of handling transaction. If you don't need distributed transactions, you can manage with con.setAutoCommit(false), con.rollback() & con.commit(). Only if you need distributed transactions you should worry about using JTA.
Software Engineer (SCJP)<br />USTechnology, CA
I agree. Here's the link: http://ej-technologies/jprofiler - if it wasn't for jprofiler, we would need to
run our stuff on 16 servers instead of 3.
0
