Versions Compared

Key

  • This line was added.
  • This line was removed.
  • Formatting was changed.
Comment:

Revise comparison to contrast SAF1 and SAF2

Feature

SAF 1

SAF 2

Action classes

SAF 1 requires Action classes to extend an abstract base class. A common problem in SAF 1 is programming to abstract classes instead of interfaces.

An SAF 2 Action may optionally implement an Action interface, along with other interfaces to enable optional and custom services. SAF 2 provides a base ActionSupport class to implement commonly used interfaces. Albeit, the Action interface is not required. Any POJO object with a execute signature can be used as an SAF 2 Action object.

Threading Model

SAF 1 Actions are singletons and must be thread-safe because since there will only be one instance of a class to handle all requests for that Action. The singleton strategy places restrictions on what can be done with SAF 1 Actions and requires extra care to develop. Action resources must be thread-safe or synchronized.

SAF 2 Action objects are instantiated for each request, so there are no thread-safety issues. (In practice, servlet containers generate many throw-away objects per request, and one more object does not impose a performance penality or impact garbage collection.)

Servlet Dependency

SAF 1 Actions have dependencies on the servlet API since the HttpServletRequest and HttpServletResponse is passed to the execute method when an Action is invoked.

SAF 2 Actions are not tied coupled to a container. Most often the servlet contexts are represented as simple Maps, allowing the Action Actions to be tested in isolation. SAF 2 Actions can still access to the original request and response, when needed. Other if required. However, other architectural elements reduce or eliminate the need to access the HttpServerRequest HttpServetRequest or HttpServletResponse directly.

Testability

A major hurdle to testing SAF 1 Actions is that the execute method exposes the Servlet API. A third-party extension, Struts TestCase, offers a set of mock object for SAF 1.

SAF 2 Actions can be tested by instantiating the Action, setting properties, and invoking methods. Dependency Injection support can make also makes testing even simpler.

FormBeans Harvesting Input

SAF 1 uses an ActionForm object to capture input, which can create a lot of extra classes. DynaBeans are often . Like Actions, all ActionForms must extend a base class. Since  other JavaBeans cannot be used as ActionForms, developers often create redundant classes to capture input. DynaBeans can used as an alternative to creating conventional ActionForm classes, but, here too, developers may be redescribing existing JavaBeans.

SAF 2 allows you to have all of your properties directly accessible on your Action as regular JavaBean properties, including uses Action properties as input properties, eliminating the need for a second input object. Input properties may be rich object types which can may have their own properties which . The Action properties can can be accessed from the web page via the taglibs. SAF 2 also supports the ActionForm pattern, as well as POJO form objects and POJO Actions. which allow you to have a rich Object type or domain object as your form bean, with its properties directly accessible to the web page, rather than accessing them as sub-properties of a property of the ActionRich object types, including business or domain objects, can be used as input/output objects. The ModelDriven feature simplifies taglb references to POJO input objects.

Expression Language

SAF 1 integrates with JSTL, so it uses the JSTL EL. The EL has basic object graph traversal, but relatively weak collection and indexed property support.

SAF 2 uses OGNL which is a powerful can use JSTL, but the framework also supports a more powerfule and flexible expression language SAF 2 can also use JSTLcalled "Object Graph Notation Language" (OGNL).

Binding values into views

SAF 1 uses the standard JSP mechanism for binding objects into the page context for access.

SAF 2 sets up uses a "ValueStack which " technology so that the taglibs can access to dynamically find values without tightly coupling your view to the types object type it is rendering. This The ValueStack strategy allows reuse of views across a range of types which may have the same propertiesproperty name but different property types.

Type Conversion

SAF 1 ActionForm properties are usually all Strings. SAF 1 uses Commons-Beanutils for type conversion. Converters are per-class, and not configurable per instance.

SAF 2 uses OGNL for type conversion with added converters provided for all basic types. The framework includes converters for basic and common object types and primitives.

Validation

SAF 1 supports manual validation via a validate method on the ActionForm, or through an extension to the Commons Validator. Classes can have different validation contexts for the same class, but cannot chain to validations on sub-objects.

SAF 2 supports manual validation via the validate method and the XWork Validation framework. The Xwork Validation Framework supports chaining validation into sub-properties using the validations defined for the properties class type and the validation context.

Control Of Action Execution

SAF 1 supports separate Request Processors (lifecycles) for each module, but all the Actions in the module must share the same lifecycle.

SAF 2 supports creating different lifecycles on a per Action basis via Interceptor Stacks. Various Custom stacks can be created and used with various different Actions, as needed.