Operation on empty rowwise list column causes "subscript out of bounds" error

[mlell]

If I perform an operation on an empty rowwise data frame, I would expect the result to be an empty data frame as well. Its similar to a for loop being skipped when the vector of indices is empty.

If I do this on a rowwise tibble with zero rows, it behaves as expected:

tibble(a = character(0)) %>%
  rowwise() %>%
  mutate(n = nchar(a)) %>%
  {stopifnot(all.equal(., 
    rowwise(tibble(a = character(0), n = integer(0))))}

This does not work for list columns

tibble(a = list()) %>%
  rowwise() %>% 
  mutate(n = lengths(a))

Error: Problem with `mutate()` input `n`.
x subscript out of bounds
ℹ Input `n` is `lengths(a)`.

I would instead expect a return value like

rowwise(tibble(a = list(), n = integer()))

I tested this issue with dplyr, commit d56404c