does every S4 needs to be generic

Question

Suppose we have the following dummy class

Foo <- setClass(Class = "Foo",slots = c(foo = "numeric"), 
    prototype = list(foo = numeric())

I thought, generics are used to overload different functions. So assume we want to implement an accessor:

setMethod(f = "getFoo", signature = "Foo", 
    definition = function(Foo)
    {
        return(Foo@foo)
    }
)

Is this valid? Or do I have to define a generic first:

setGeneric(name="getFoo",
    def=function(Foo)
    {
        standardGeneric("getFoo")
    }
)

If there is just one particular "instance" of this function type, there is no reason to define a generic, correct?

Answer 1

In order to define an S4 method, there must be an existing S4 generic (either from base, imported from another package, or defined yourself). My understand of this design is to provide the flexibility to add on additional methods in the future, even if you can't even conceive of another one ATM.

That said, if you are just trying to be more concise you could just provide the default function directly to the generic function.

setClass(Class = "Foo",slots = c(foo = "numeric"), 
                prototype = list(foo = numeric()))

setGeneric(name="getFoo",
           def=function(Foo)
           {
               standardGeneric("getFoo")
           }, useAsDefault=function(Foo){return(Foo@foo)}
)

# test the function
testFoo <- new("Foo", foo=3)

getFoo(testFoo)
[1] 3

So, now you have your generic including the only functionality you really wanted anyway. You also have the option to add on the the generic in the future depending upon how your application develops.