[Hibernate-issues] [Hibernate-JIRA] Closed: (HB-185) flush() does not cleanup after a SQLException

[<leg...@at...>]

Message:

   The following issue has been closed.

   Resolver: Gavin King
       Date: Fri, 11 Jul 2003 11:39 PM

One of The Rules is that you can't keep working with a Session after it throws an exception. There are some very pathalogical things that can happen and there is basically no way that we can ensure that the internal session state is synchronized with the database state after an exception.
---------------------------------------------------------------------
View the issue:

  http://opensource.atlassian.com/projects/hibernate/secure/ViewIssue.jspa?key=HB-185


Here is an overview of the issue:
---------------------------------------------------------------------
        Key: HB-185
    Summary: flush() does not cleanup after a SQLException
       Type: Bug

     Status: Closed
   Priority: Major
 Resolution: WON'T FIX

    Project: Hibernate2
 Components: 
             core
   Versions:
             2.0.1

   Assignee: 
   Reporter: Dave White

    Created: Fri, 11 Jul 2003 2:07 PM
    Updated: Fri, 11 Jul 2003 11:39 PM

Description:
I have a object mapped with a composite-id.  In the database, there is a unique composite index against the columns that make up the composite-id.  I've found that when I do the following:

try {
  session.save(foo); // where foo is the object with the composite-id
  session.flush();
  // Do some other work with session
}
finally {
  session.flush();
  session.close();
}

that if a SQLException is thrown in the first session.flush (because the insert statement violated the unique constraint of the index), the second flush throws the same exception.  The means that the flush() call doesn't cleanup the "things to be done" in the case that an exception is thrown during the flushing process.  Specifically, I found the following in SessionImpl.

private void executeAll(Iterator iter) throws SQLException, HibernateException {
    while ( iter.hasNext() ) {
	Executable e = (Executable) iter.next();
	executions.add(e);	
        e.execute();
        iter.remove();        		
	if ( batcher!=null ) batcher.executeBatch();
    }
}

I'm wondering if perhaps it would be better if this was:

private void executeAll(Iterator iter) throws SQLException, HibernateException {
    while ( iter.hasNext() ) {
	Executable e = (Executable) iter.next();
	executions.add(e);
        try {	
            e.execute();
        }
        finally {
            iter.remove();        		
        }
	if ( batcher!=null ) batcher.executeBatch();
    }
}

After making this change in my local source tree, the problem I descrived above was fixed.


---------------------------------------------------------------------
JIRA INFORMATION:
This message is automatically generated by JIRA.

If you think it was sent incorrectly contact one of the administrators:
   http://opensource.atlassian.com/projects/hibernate/secure/Administrators.jspa

If you want more information on JIRA, or have a bug to report see:
   http://www.atlassian.com/software/jira