[JBoss-dev] [Design of JCA on JBoss] - Re: Impossible to subclass IdleRemover and InternalManagedCo

2005-02-09 Thread [EMAIL PROTECTED]
IdleRemover is a static instance that runs a single remover thread for all pools. To override the implementation would require an IdleRemoveFactory and a global configuration somewhere in jbossjca-service.xml Why do you need to change it? All it does it callback to the registered pool on a timeo

[JBoss-dev] [Design of JCA on JBoss] - Re: Impossible to subclass IdleRemover and InternalManagedCo

2005-02-09 Thread [EMAIL PROTECTED]
I would be prepared to accept a modification that allows the internal pool classname to be specified on the ManagedConnectionPool. In fact there are three configurations that are hardwired and need exposing in -ds.xml for more advanced use cases: | Defines the ManagedConnectionPool implementat

[JBoss-dev] [Design of JCA on JBoss] - Re: Impossible to subclass IdleRemover and InternalManagedCo

2005-02-09 Thread [EMAIL PROTECTED]
On your immediate problem, it seems to that what is really required is the ability define a plugin for the remover logic in InternalManagedConnectionPool. Something like: | com.acme.RemoverPolicy | | - if (cl.isTimedOut(timeout)) | + if (removerPolicy.isTimedOut(cl,

[JBoss-dev] [Design of JCA on JBoss] - Re: Impossible to subclass IdleRemover and InternalManagedCo

2005-02-09 Thread [EMAIL PROTECTED]
In anycase. What feature are you adding? If it is generically useful, this could be added to the base implementation rather than needing all this factory/plugins complication. View the original post : http://www.jboss.org/index.html?module=bb&op=viewtopic&p=3865913#3865913 Reply to the post : h

[JBoss-dev] [Design of JCA on JBoss] - Re: Impossible to subclass IdleRemover and InternalManagedCo

2005-02-09 Thread andreit
"[EMAIL PROTECTED]" wrote : In anycase. What feature are you adding? | If it is generically useful, this could be added to the base implementation | rather than needing all this factory/plugins complication. I belive that it is generically useful and would be appreciated by community. Basicly,

[JBoss-dev] [Design of JCA on JBoss] - Re: Impossible to subclass IdleRemover and InternalManagedCo

2005-02-09 Thread [EMAIL PROTECTED]
I still don't see why you need to change the IdleRemover. If the aged timeout is less than the idle timeout, you pass that as the interval. It just means it is going to make more frequent calls to removedTimeOut. There is no loss here since it just checks one connection (the oldest unused) to se

[JBoss-dev] [Design of JCA on JBoss] - Re: Impossible to subclass IdleRemover and InternalManagedCo

2005-02-10 Thread andreit
"[EMAIL PROTECTED]" wrote : I still don't see why you need to change the IdleRemover. | | If the aged timeout is less than the idle timeout, you pass that as the interval. | It just means it is going to make more frequent calls to removedTimeOut. | Right. But IdleRemover makes addition

[JBoss-dev] [Design of JCA on JBoss] - Re: Impossible to subclass IdleRemover and InternalManagedCo

2005-02-10 Thread [EMAIL PROTECTED]
Do you have a sourceforge id? If you do I will get you cvs access where you can prototype your ideas in jboss-head (the development branch which will become JBoss5). Make sure you read the developers guide on coding standards (they are not the same as Sun's guidelines), cvs branches, etc. Also

[JBoss-dev] [Design of JCA on JBoss] - Re: Impossible to subclass IdleRemover and InternalManagedCo

2005-02-11 Thread andreit
"[EMAIL PROTECTED]" wrote : Do you have a sourceforge id? | | If you do I will get you cvs access where you can prototype your ideas in jboss-head | (the development branch which will become JBoss5). | My id is 'andreitar'. "[EMAIL PROTECTED]" wrote : Make sure you read the developers