My part II solution is based on the J2EE blueprint petstore example. My question comes in two parts. 1. Has anybody used the following guide below to complete their component diagram?
<<Hi! I saw in a book for UML something like that: If an object implements some interfaces and can be changed with other implementation and not affect the system than this object can be considered a component (This is the case for all DAO's). In this case the EJBAction. HTMLAction and Event are all components? Some architects say that the component diagram represents a class diagram but with only dependencies>>
2. Can a Java client link to a EJBController to access EJB-tier? <SFSB> <Client>-><EJBController>-><servicelocator>-><XXEJB>
Please a brief explanation. I am not clear on the right way to go with this.
Originally posted by Kayman: If an object implements some interfaces and can be changed with other implementation and not affect the system than this object can be considered a component ...
A component in UML is a - pluggable / replaceable - physical - deployable thing like a - .jar, .jsp, .exe, ..., db table, ... - that in case of a .jar realizes the interfaces that it provides. - These "provided interfaces" in UML are drawn outside the component - but are also packaged into the .jar because needed there as well.
Originally posted by Kayman: 2. Can a Java client link to a EJBController to access EJB-tier?
Surely a Java client working in a client-side Application Container (i.e. specified by J2EE spec and provided by the EJB container vendor) can access any EJB via its remote interface (even Entity EJBs, but don't!). Usually it accesses a Session EJB being a SessionFacade.
Additionally some Java Standalone Applications (not running in an Application Container, i.e. not compliant to J2EE) can access EJBs via a proprietary protocol.
Thomas [ April 17, 2006: Message edited by: Thomas Taeger ]