Designing for Developers
Regardless of whether we choose the composite approach—constructing a new UI element by assembling together several other controls—or we decide to write a complete control from scratch, there are some design issues that we must consider when designing a new type of control. The problem that faces all controls is that they must be a servant to two masters: the software developer who will reuse the control and the end user who will interact with the control.
Non-visual classes don’t suffer from this problem. They have a single interface—their public programming interface. Their internal workings are their own business. Likewise, visual components not designed for reuse, such as most forms, only present one public face—their user interface. The majority of forms have no public programming interface at all, and those that do usually have a very simple one (such as properties for setting or retrieving fields on a form designed to present or collect data).
But a reusable visual component must consider both types of user—it has both a user interface and a programming interface. It is important not to confuse the two when designing your control; we have already seen how tempting it can be to misuse inheritance when we want to implement one control using the user interface of another. As a rule of thumb, the way the user interface does its job is an implementation detail, and should therefore not be visible to client code.
Because your class derives (either directly ...