SparkR：orderBy由两列按降序排列

Question

我正在尝试按数据框中的两列来订购SparkDataFrame。

示例如下：

library(magrittr)
library(SparkR)
cars <- cbind(model = rownames(mtcars), mtcars)
carsDF <- createDataFrame(cars)

以下失败：

carsDF %>%
  mutate(rank = over(rank(), orderBy(windowPartitionBy(column("cyl")), desc(column("mpg")), desc(column("disp"))))) %>%
  head()

我觉得这与column()函数的使用有关，原因是它也失败了：

carsDF %>%
  mutate(rank = over(rank(), orderBy(windowPartitionBy(column("cyl")), column("mpg"), column("disp")))) %>%
  head()

以下两个命令（带有和不带有orderBy()函数的column()的单个参数）均成功：

carsDF %>%
  mutate(rank = over(rank(), orderBy(windowPartitionBy("cyl"), "mpg"))) %>%
  head()

carsDF %>%
  mutate(rank = over(rank(), orderBy(windowPartitionBy(column("cyl")), column("mpg")))) %>%
  head()