Thanks Knut,
Figured out some of this about 30 minutes after posting the question but
your explanation clears thing up a lot.
regards
Tony Seebregts
Subject:
Re: Re. PooledConnection being closed
From:
Knut Anders Hatlen <[EMAIL PROTECTED]>
Date:
Mon, 15 Aug 2005 18:44:41 +0200
To:
Derby Discussion <derby-user@db.apache.org>
To:
Derby Discussion <derby-user@db.apache.org>
Tony Seebregts <[EMAIL PROTECTED]> writes:
Ok, thanks Knut - my misunderstanding.
I understood the ConnectionPoolDataSource as merely facilitating
connection pooling. If it manages the actual pooling though, is
there any point in handling the ConnectionEvents ? Other than for
monitoring/interest ...
The ConnectionPoolDataSource doesn't manage the connection pool. It is
just an instrument which is used by a connection pool manager. A
ConnectionPoolDataSource can create PooledConnection objects, but
since each PooledConnection can have only one open Connection object
at a time, you'll probably need more than one PooledConnection object,
and the need for a pool manager emerges.
A typical connection pool manager has a getConnection() method which
checks whether there is an unused PooledConnection in the pool. If
there is one, PooledConnection.getConnection() is called, otherwise a
new PooledConnection is created and its getConnection() is called. In
addition the manager has to make sure that the pool doesn't become too
big, that invalid PooledConnection objects are removed from the pool,
etc. That's where the ConnectionEvent handlers come into use. The
connectionClosed() method marks a PooledConnection as unused, and
removes it from the pool if the pool has become too big. The
connectionErrorOccurred() removes an invalid PooledConnection from the
pool.
