[Hibernate-issues] [Hibernate-JIRA] Commented: (HB-319) PreparedStatementCache should check for NULL on equals()

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

The following comment has been added to this issue:

     Author: Gavin King
    Created: Tue, 9 Sep 2003 4:36 AM
       Body:
I kinda knew this problem existed, and I pored over the code twice looking for the cause.

The trouble is its just not effectively reproduceable.


I suspect the HotSpot bug might indeed be the cause.

Anyway, about three days ago I gave up and removed the prepared statement cache from Hibernate. People should be using another connection pool for this _anyway_.


:-(
---------------------------------------------------------------------
View the issue:

  http://opensource.atlassian.com/projects/hibernate/secure/ViewIssue.jspa?key=HB-319


Here is an overview of the issue:
---------------------------------------------------------------------
        Key: HB-319
    Summary: PreparedStatementCache should check for NULL on equals()
       Type: Patch

     Status: Unassigned
   Priority: Trivial

    Project: Hibernate2
 Components: 
             core

   Assignee: 
   Reporter: Jonas Van Poucke

    Created: Tue, 9 Sep 2003 3:47 AM
    Updated: Tue, 9 Sep 2003 3:47 AM
Environment: JDK 1.4.1 and 1.4.2

Description:
I found PrepeparedStatementCache to throw NullPointerExceptions. The bas news: I could not exactly trace the location of the bug. 
The problem lies in the fact that java.util.LinkedList has some entries where the element is NULL (in our case the element is a PrepeparedStatementCache$Entry). Also, the line
    entries.remove(entryMap.get(ps));
is dangerous: the entryMap.get(ps) sometimes return NULL as well. I could not find the point where the key-value pair is removed.

The good news: I fixed this bug by modifying the PrepeparedStatementCache in two places:

--- Fix 1 ---
The equals() method of the Entry inner class should be:
        public boolean equals(Object other) {
            Entry oe = (Entry) other;
            return oe != null && oe.connection == connection &&
                oe.scrollable == scrollable && oe.sql.equals(sql);
        }
That is, I added oe != null

--- Fix 2 ---
The close(Collection statements) method should also check for NULL elemements:
    private void close(Collection statements) {
        PreparedStatement ps;
        Iterator iter = statements.iterator();
        while (iter.hasNext()) {
            try {
                ps = (PreparedStatement) iter.next();
                if (ps != null) {
                    ps.close();
                }
            } catch (SQLException sqle) {
                log.warn("could not close statement", sqle);
            }
        }
    }

I hope that this patch actually solves a problem in the right place, i.e. that the bug is not created outside of PreparedStatementCache or in the JDK

Background:
By looking on the net, I found there might be a


---------------------------------------------------------------------
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