r - 如果未提供变量，为什么用 data.table 插槽初始化 R S4 类不起作用

Question

我正在尝试创建一个包含 data.table 的对象，如果在 new 语句中没有提供任何内容，我希望将其初始化为一个空的 data.table。但是，如果我将其遗漏，则会收到一条错误消息。

我尝试使用 data.frame 代替，这很好，但我想了解下面最后一个新声明中出了什么问题，因为前 2 个新声明是好的？

library(data.table)

setClass("TmpClass", slots = c(v1 = "character", v2 = "data.frame"))
new("TmpClass", v1 = "I work")
# An object of class "TmpClass"
# Slot "v1":
#   [1] "I work"
# 
# Slot "v2":
#   data frame with 0 columns and 0 rows

setClass("JunkClass", slots = c(v1 = "character", v2 = "data.table"))
new("JunkClass", v1 = "I work as well", v2 = data.table(a = 1, b = 2))
# An object of class "JunkClass"
# Slot "v1":
#   [1] "I work as well"
# 
# Slot "v2":
#    a b
# 1: 1 2

new("JunkClass", v1 = "hmm, i get an error")
# Error in validObject(.Object) : 
#   invalid class “JunkClass” object: invalid object for slot "v2" in class 
# "JunkClass": got class "list", should be or extend class "data.table"

Answer 1

我认为这里一个好的解决方案是正式承认 @v2 插槽可以是 NULL。

library(data.table)
setClassUnion("nullOrDatatable", c("NULL", "data.table"))
setClass("JunkClass", slots = c(v1 = "character", v2 = "nullOrDatatable"))
new("JunkClass", v1 = "No more error!")
# An object of class "JunkClass"
# Slot "v1":
#   [1] "No more error!"
# 
# Slot "v2":
#   NULL