>From a scalability perspective, this fix has made a huge difference. We have different environments where we use the appserver connection pooling (BEA Weblogic or SunONE) and others that use the commons connection pooling. Without statement caching in the commons pooling, our scalability was much less than what could be achieved with appserver connection pooling. Appserver connection pooling typically already handle statement pooling. But by enabling this feature, scalability is now almost identical. Our app really needs this statement pooling.
I see your point about moving the creation of the delegatingconnection and that makes sense. I will do some more thinking about where to place this. If you find a better location, please advise. We are a month from code freeze and I need to get this stable for our performance lab. Thanks. ToddC -----Original Message----- From: Dirk Verbeeck [mailto:[EMAIL PROTECTED] Sent: Monday, February 02, 2004 1:09 PM To: Jakarta Commons Users List Subject: Re: DO NOT REPLY [Bug 24136] - ClassCastException in DriverAdap terCPDS when setPoolPreparedStatements(true) I have tried your fix (in bugzilla) and it seems to work. The construction of the DelegatingConnection can/should moved up the call stack to the PoolablePreparedStatementStub or even to the PooledConnectionImpl. (one DelegatingConnection for each PooledConnectionImpl) There aren't any preparedstatementpool JUnit tests so I'm going to wait a little bit before commiting this fix. (It's not the simplest code ;-) -- Dirk Todd Carmichael wrote: > With this error in the code, translates to no coverage of this feature > by anyone. Statement pooling is a big part of improving scalability > for our application. I am surprised that no one else takes advantage of it. > > > ToddC > > > -----Original Message----- > From: Todd Carmichael [mailto:[EMAIL PROTECTED] > Sent: Thursday, January 29, 2004 12:51 PM > To: '[EMAIL PROTECTED]' > Subject: RE: DO NOT REPLY [Bug 24136] - ClassCastException in DriverAdap > terCPDS when setPoolPreparedStatements(true) > > > Does anyone have a 'quick' fix for this problem? I have the code and > can make changes on my side. > > Thanks! > > ToddC --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED] --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
