[wsif] -- support for factory pattern for Java services?

[Andre Tost <at...@us.ibm.com>]

As far as I can tell, in the case of a service being a JavaBean, the
classname of this bean is defined in the bindings. The code uses this
classname directly to instantiate an instance of this object, using
Class.newInstance(). Is there any chance (or plan) to support a level of
indirection for this? Assume I want to retrieve this instance from another
place before invoking a service request on it. Or I want to let a factory
instantiate it. For EJBs, the fact that they are retrieved through naming
gives me additional flexibility, for JavaBeans it seems pretty hardcoded.

Andre Tost

Re: [wsif] -- support for factory pattern for Java services?

[Nirmal Mukhi <nm...@us.ibm.com>]

Hello Andre,

That's a good point, we'll consider something like this in the future, 
though it won't happen right away for sure.

If you could provide a design doc and/or the reqd source code patches that 
would be excellent, I would think patches would be likely to find their 
way into future distributions as long as it doesn't break existing test 
cases.

Thanks,
Nirmal.




Andre Tost/Rochester/IBM@IBMUS
01/19/2003 10:18 AM
Please respond to axis-user
 
        To:     axis-user@xml.apache.org
        cc: 
        Subject:        [wsif] -- support for factory pattern for Java 
services?






As far as I can tell, in the case of a service being a JavaBean, the
classname of this bean is defined in the bindings. The code uses this
classname directly to instantiate an instance of this object, using
Class.newInstance(). Is there any chance (or plan) to support a level of
indirection for this? Assume I want to retrieve this instance from another
place before invoking a service request on it. Or I want to let a factory
instantiate it. For EJBs, the fact that they are retrieved through naming
gives me additional flexibility, for JavaBeans it seems pretty hardcoded.

Andre Tost