When we need to define multiple algorithms, it's easier to encapsulate each of them and make them interchangeable for a certain client/scope in which they are used. The strategy makes it very flexible to be used depending on the clients that try to.
- Policy
Given a set of Clients C and a set of Algorithms A, we can have the following situations:
- Clients in C supporting multiple algorithms from A, may become too complex and hard to maintain;
- Many algorithms from A may be needed at different times, and we want to support only those need at certain times;
- It would be difficult to add new Algorithms in A, and/or modify the existing ones if we let them become embedded with clients;
We can avoid these issues by encapsulating each algorithm into separate classes. An algorithm that is encapsulated in this way, is called a strategy.
- When you have many related classes that are different in some parts of their behavior;
- When you want to make use of multiple variations of the same algorithm;
- When you have algorithms using data that the client should not be aware of;
- When you end up with a class having multiple variations, defined in separate conditional statements;
