How to design my setter method?

Question

I have a method which I wrote like this:

void updateMyJaxb(final JAXBElement<?> jaxbElement) {

addElementtoJaxb(jaxbElement)
}


void addElementTJaxb(jaxbElement)
{
   //have to cast to myown type 
((JAXBElement<MyType>) jaxbElement).getValue().setSomeValue(somevalue);
}

so in essence I add some staff to JAXBElement. The question is : is it a good design, take a reference and change the content, ? Will it be better to return the reference to a updated object? Otherwise the method may not be straightforward to understand ?

Answer 1

Setters should not have any effects on the parameter they take directly. Personally I would consider that as bad design. If you stil want to do that I would strongly recommend you to document this behavior, because the behavior might be unexpected.

Returning the objects is also not required, since Java passes parameters (objects in general) by reference.

I would use the Adapter pattern, which takes an JAXBElement as a parameter and returns a JAXBElement. Instead of updating the JAXBElement the adapter clones it and changes the clone and returns it. This way no unexpected side-effect are introduced. This pattern is also called defensive-copying.

private JAXBElement clone(JAXBElement o) {
    JAXBElement clone = new JAXBElement();

    clone.setName(o.getName());
    ...

    return clone;
}

This should do the job.