Connection.close() should close its statements

[Patric]

Currently, BoneCP requires that all statements must be closed.
Otherwise, this may result in an OutOfMemoryError.

In fact this is violating the jdbc spec:

9.4.4 Closing Connection Objects
An application calls the method Connection.close to indicate that it has finished
using a connection. All Statement objects created from a given Connection object
will be closed when the close method for the object is called.

It should be sufficient to close the connection. It is no error when not closing statements.
(even it's a questionable style).

See also http://jolbox.com/forum/viewtopic.php?f=3&t=253

[Patric (patric)]

Hi Wallace,

please note that this is not an enhancement request/wish, but a bug report with high priority, because BoneCP, which implements parts of the jdbc specification, violates the latter.

We have much "old" software code in production which do not close statements (but the connection), which is a completely correct behavior.
Without fixing this bug it would be complicated to use BoneCP as a replacement for our propritary connection pooling.

As a result, this bug has an high importance to me and may even be a blocker for BoneCP for the use within our projects.

[Wallace Wadge (wwadge)]

Fair enough - bumping priority.

[Wallace Wadge (wwadge)]

Fixed in 0.8.0-SNAPSHOT.

New config options:

/** If true, track statements and close them if application forgot to do so. See also:
  * detectUnclosedStatements. */
 boolean closeOpenStatements;
 /** If true, print out a stack trace of where a statement was opened but not closed before
  * the connection was closed. See also: closeOpenStatements. */
  boolean detectUnclosedStatements;

[Patric (patric)]

Thanks, Wallace.
The unclosed statements will be closed during Connection.close(), not later, right?

[Wallace Wadge (wwadge)]

That's correct - closed upon calling connection.close() -- right away even before releaseHelperThreads come into play.

[everflux (tklaunchpad)]

Wallace, it would be great if you could perform a release of BoneCP: There are so many fixes after the last 0.7.1 release already commited but can not yet be used due to the SNAPSHOT version. (Snapshot dependencies prevent a maven release and self-packaging is obviously a bad option.)

[Wallace Wadge (wwadge)]

Hi everflux,

After a long while*, I've resumed bonecp work; expect a release sooner rather than later now

* newborn kids have that effect :-/

[Kevin (kevinshlee)]

I'm just wondering when we can have the 0.8.0.RELEASE version with this issue fixed.