The plugin design pattern explained (as described by Martin Fowler)

Question

I am trying to understand and exercise the plugin pattern, as explained by Martin Fowler.

I can understand in which way it makes use of the separated interface pattern, and that it requires a factory to provide the right implementation of the interface, based on the currently used environment (test, prod, dev, etc). But:

• How exactly does the factory read the environment values and decide which object (implementing the IdGenerator interface) to create?
• Is the factory a dependency of the domain object (DomainObject)?

Thank you very much.

Answer 1

You will want to get a full PFD of the "Patterns of Enterprise Application Architecture". What is visible on Fowler's site is basically first half-page of any chapter :)
What is being describes is basically the expanded version of idea behind polymorphism.

I don't think "plugin" can actually be described as a "pattern". It's more like result of other design choices.

What you have are .. emm ... "packages", where the main class in each of them implements a third party interface. Each of those packages also have their internal dependencies (other classes or even other libraries), which are used for some specific task. Each package has it's of configuration (which might be added through DIC config) ans each of them get "registered" in your main application.

The mentioning of a factory is almost a red herring, because these days that functionality would be applied using DIC.