Should the Java – domain model mapper be static?

In many projects I participate in, we often have many classes to map content from one domain model to another For example, from WSDL - generated models to project - specific models

for example

public class FooBarContentMapper {
    public static Foo fromWsToDomain(FooType fooType) {
        ...
    }
}

This can also be a non static method. The service layer can have a mapper object field instead of calling a static method:

public class FooBarContentMapper {
    public Foo fromWsToDomain(FooType fooType) {
        ...
    }
}

I found that both methods used a lot, but:

>One of the solutions to improve efficiency in any way? > Is any solution considered a best practice?

Solution

Define "valid" If, by "efficiency", you mean CPU time and memory requirements, then the "instance" method will never be more effective than the "static" method; At best, it can be as efficient as static methods, and the difference depends on the frequency of object instantiation. Read: the number of times you want to instantiate the "instance" method

no The "best practice" here is to match your design with your requirements

>If the mapping operation needs to maintain state, such as relying on other services / mappers / and so on, it makes more sense to use the "instance" method One problem you don't want to get into is that your application design consists of interdependent singletons Using the "instance" method, it is best to use automatic routing tools, such as spring framework or CDI. > If the mapping operation does not require a state and you have high confidence that you will never need a state in the future, use the "static" method - unless you already have an automatic routing device, in which case you can select the "instance" method of automatic routing and ensure that if the mapping operation requires a state in the future, you will not have to change the design