It looks like a bug or at least something to be improved.
Your analysis is correct and reverting the change will correct the problem.
I don't know why Rodney committed this change but we better make sure we don't break anything if we undo the change.


JUnit tests are needed to show the error and prove the correctness of the solution.

Anyway please create a bugzilla issue to track the problem.

-- Dirk

Simon Matic Langford wrote:

Hi,

I've recently upgraded to the latest version of Tomcat (4.1.29), which
includes DBCP 1.1, over 1.0 in previous versions.

On starting my application and attempting to logon to it, it falls over
with a NullPointerException, included here:
java.lang.NullPointerException
        at
oracle.jdbc.driver.ScrollableResultSet.close(ScrollableResultSet.java:14
9)
        at
org.apache.commons.dbcp.DelegatingResultSet.close(DelegatingResultSet.ja
va:193)
        at
org.apache.commons.dbcp.DelegatingResultSet.close(DelegatingResultSet.ja
va:193)
        at
org.apache.commons.dbcp.DelegatingPreparedStatement.passivate(Delegating
PreparedStatement.java:298)
        at
org.apache.commons.dbcp.DelegatingPreparedStatement.close(DelegatingPrep
aredStatement.java:185)
        at
com.pcmsgroup.v21.star.framework.persistence.BaseDbDAO.releaseResources(
BaseDbDAO.java:205)
        at
com.pcmsgroup.v21.star.persistence.logon.LogonDbDAO.fetchByLogonName(Log
onDbDAO.java:149)
        at
com.pcmsgroup.v21.star.domain.logon.Logon.logon(Logon.java:95)
        at
com.pcmsgroup.v21.star.service.logon.LogonService.logon(LogonService.jav
a:52)
        at
com.pcmsgroup.v21.star.application.logon.LogonCtrl.process(LogonCtrl.jav
a:87)
        at
com.pcmsgroup.v21.star.framework.application.BaseCtrl.execute(BaseCtrl.j
ava:180)
        at
org.apache.struts.action.RequestProcessor.processActionPerform(RequestPr
ocessor.java:484)
        at
org.apache.struts.action.RequestProcessor.process(RequestProcessor.java:
274)
        at
org.apache.struts.action.ActionServlet.process(ActionServlet.java:1482)
        at
org.apache.struts.action.ActionServlet.doPost(ActionServlet.java:525)
        at javax.servlet.http.HttpServlet.service(HttpServlet.java:760)
        at javax.servlet.http.HttpServlet.service(HttpServlet.java:853)
        at
org.apache.catalina.core.ApplicationFilterChain.internalDoFilter(Applica
tionFilterChain.java:247)
        at
org.apache.catalina.core.ApplicationFilterChain.doFilter(ApplicationFilt
erChain.java:193)
        at
com.pcmsgroup.v21.star.application.logon.LogonFilter.doFilter(LogonFilte
r.java:168)
        at
org.apache.catalina.core.ApplicationFilterChain.internalDoFilter(Applica
tionFilterChain.java:213)
        ...

I've looked around the web, and in your CVS, and think that this is a
bug.

The code in
com.pcmsgroup.v21.star.framework.persistence.BaseDbDAO.releaseResources
is as follows:

  public final void releaseResources(Connection connection,
PreparedStatement ps, ResultSet rs) throws PersistenceException
  {
    try
    {
      if (rs != null)
      {
        rs.close();
      }
      if (ps != null)
      {
        ps.close();
      }
      if (connection != null)
      {
        // get the Factory
        ConnectionFactory factory = ConnectionFactory.getInstance();
        // turn the Connection back to pool
        factory.releaseConnection(connection);
      }
    }
    catch (SQLException sqle)
    {
      thisLog.error("Failed to release resources", sqle);
    }
    catch (Exception e)
    {
      thisLog.error("Failed to release resources", e);
      throw new PersistenceException(e);
    }
  }

As you can see, we close ResultSet, PreparedStatement and Connection in
the reverse order of that which we
got them in. On DBCP 1.0, this used to work fine. According to the JDK
API docs: "When a Statement object is closed, its current ResultSet object, if one exists, is also closed."
Thus closing the statement after closing the
result set is safe.


However, for DBCP 1.1, a change was made to DelegatingPreparedStatement
(version 1.6 to 1.7), to "ensure PreparedStatment can only be closed once", which entailed changing:


     public void close() throws SQLException {
         passivate();
         _stmt.close();
     }

to:

     public void close() throws SQLException {
         _stmt.close();
         passivate();
     }

I believe this has introduced my problem and is a bug. As now, following
the logic:

- The statement is closed, and in accordance with the JDK docs, the
result set is closed.
- The passivate method is then called, which calls ResultSet.close on
the underlying result set,
  regardless of whether it has been closed already.
- This causes the underlying result set, which is already closed to try
to close again, thus
  the NullPointer.

I couldn't spot this in the bug reports, so could someone answer if this
is a bug or not, or is there
something I am doing wrong, or should this be directed at the developers
list? I don't believe I should quietly catch Exception's in the close() methods as per your examples,
as I'd really like to know if there are problems!


Thanks

Simon




---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]



Reply via email to