Rationale
Traditional MVC frameworks work in terms of whole requests and whole pages. In each request cycle, the incoming request is mapped to a method on a controller object, which then generates the outgoing response in its entirety, usually by pulling data out of a model to populate a view written in specialised template markup. This keeps the application's flow-of-control simple and clear, but can make code reuse in the controller difficult.
Design
Wicket, on the other hand, is closely patterned after stateful GUI frameworks such as Swing. Wicket applications are trees of components, which use listener delegates to react to HTTP requests against links and forms in the same way that Swing components react to mouse and keystroke events.
Wicket uses plain XHTML for templating (which enforces a clear separation of presentation and business logic and allows templates to be edited with conventional WYSIWYG design tools). Each component is bound to a named element in the XHTML and becomes responsible for rendering that element in the final output. The page is simply the top-level containing component and is paired with exactly one XHTML template. Reuseable parts of pages may be abstracted into components called panels, which can then be pulled whole into pages or other panels with a special tag.
Each component is backed by its own model, which represents the state of the component. The framework does not have knowledge of how components interact with their models, which are treated as opaque objects automatically serialized and persisted between requests. More complex models, however, may be made detachable and provide hooks to arrange their own storage and restoration at the beginning and end of each request cycle. Wicket does not mandate any particular object-persistence or ORM layer, so applications often use some combination of Hibernate objects, EJB beans or POJOs as models.
Source: Apache Wicket & Wikipedia
