Flex SDK 3 Vs. 4 – Part 2 – Introducing Components
While the Flex 4 SDK includes the MX component architecture provided with previous releases of the Flex SDK (mainly for backwards compatibility), it introduces a new set of components that utilize the new SDK features, named Spark.
In most cases, the changes between Spark and MX components are not visible, but rather involve the component’s interaction with the Flex 4 frameworks (i.e. layouts, skins, states etc).
In Spark architecture, Visual components are "split" up to several classes. For example, the Button class in Flex 3 is built up of Button class and in addition the ButtonSkin class in Flex 4. Components are typically contained in the spark.* package, while skin classes are typically contained in the spark.skins.* package.
The Spark component architecture is probably the most significant new feature introduced with Flex 4. The main reason was to provide a well-defined separation between the component's functional logic and visualization logic. As mentioned above, Spark components are built up of two classes to support this separation of logic: an AS class for the component functional logic and an MXML skin class for the visualization logic.
This allows us to either override the component functional logic and keep the visualization, override the visualization logic and keep it's functionality, or override both. Also, being able to define skin in MXML is a great improvement over Flex 3.
The fact that the Spark components are build on MX base classes means that the same component lifecycle applies and we can mix MX and Spark components in the same application (see attached code samples).
Spark components should work within MX containers and MX components should work within Spark containers.
A Spark container cannot be the direct child of a MX navigator. To use a Spark container as the child of a navigator, it must be wrapped in either a MX container or a Spark NavigatorContent container.
All Spark components require FP 10, so if your application makes use of Spark components, your users will have to use FP 10.