R devtools::check() WARNING: "The argument of a replacement function which corresponds to the right hand side must be named 'value'."

I am using devtools::check() in R to check a package and cannot overcome this warning:

The argument of a replacement function which corresponds to the right hand side must be named 'value'.

You can see below in the example code that I do indeed have value as the RHS argument for the function. Other replacement functions in the same package do not throw this warning. What am I doing wrong?

methods::setGeneric(name = "addTable<-", def = function(x, value, ...) standardGeneric("addTable<-"))

#' MyClass
#'
#' @aliases MyClass
#' @rdname MyClass
#' @exportClass MyClass
MyClass <- methods::setClass(
    Class = "MyClass",
    slots = list(
        table = "data.frame"
    ),
    prototype = prototype(
        table = data.frame()
    )
)

#' Attach a data frame to a MyClass object
#'
#' @description `addTable()` adds a table to a `MyClass` object.
#'
#' @param x A `MyClass`.
#' @param value A `data.frame`.
#'
#' @returns A `MyClass` object.
#'
#' @aliases addTable<-
#' @rdname addTable
#' @exportMethod addTable<-
methods::setMethod(
    f = "addTable<-",
    signature = c(x = "MyClass", value = "data.frame"),
    function(x, value) {
    
    x@table <- value
    x
    
    }
)

Upon devtools::check(), this returns the warning:

checking replacement functions ... WARNING 'addTable<-' '\S4method{addTable<-}{MyClass,data.frame}' The argument of a replacement function which corresponds to the right hand side must be named 'value'.

I've tried:

Making a test package that has just the offending method and only the methods package as a dependency and no other classes or methods (i.e., this example)
Using the tag @aliases addTable<-,MyClass,data.frame-method
Renaming x to object
Swapping the position of x and value in the signature and definition part of the method definition (plus setGeneric)... which is different from any other replacement function I've seen, but I gotta try something

Obviously, I'm just casting rocks in an ocean and hoping to hit a fish.

I'm running R 4.4.0 and RTools 4.4.

Solution

S4 Setter Functions

For S4 setters, the "value" argument must be the last argument in the function definition. Most of the time, setters in the wild have two arguments, the object and the new value like below:

## toy class
methods::setClass("Numbers", slots = c(number = "numeric"))

## create a setter generic
methods::setGeneric("number<-", function(x, value) standardGeneric("number<-"))

## create a method for `Numbers` class
methods::setMethod("number<-", "Numbers", function(x, value) {
  x@number <- value
  x
})

## instantiate a `Numbers` object
obj <- new("Numbers", number = 5)
obj

# An object of class "Numbers"
# Slot "number":
# [1] 5

## usage
number(obj) <- 3
obj

# An object of class "Numbers"
# Slot "number":
# [1] 3

## alternate usage - no assignment
`number<-`(obj, 1)

# An object of class "Numbers"
# Slot "number":
# [1] 1

obj

# An object of class "Numbers"
# Slot "number":
# [1] 3

## alternate usage - requires assignment
obj <- `number<-`(obj, 1)
obj

# An object of class "Numbers"
# Slot "number":
# [1] 1

If you want to add additional arguments to the setter, they must be included before the "value" argument either as named arguments or ellipsis like so:

methods::setGeneric("fancy_number<-", function(x, ..., value) standardGeneric("fancy_number<-"))

These additional arguments can be used to modify how the values are assigned.

methods::setMethod("fancy_number<-", "Numbers", function(x, operation = c("add", "replace"), value) {
  if (operation == "add") {
    x@number <- x@number + value 
  } else if (operation == "replace") {
    number(x) <- value
  }
  x
})

## usage - note position of new arguments on left of arrow
fancy_number(obj, "add") <- 100
obj

# An object of class "Numbers"
# Slot "number":
# [1] 101

fancy_number(obj, "replace") <- 7
obj

# An object of class "Numbers"
# Slot "number":
# [1] 7

This is probably why "value" must be the last argument - R needs to know what one value is being assigned.