`rowwise()` + `mutate()` doesn't respect rowwise-ness when copying a list-col

[DavisVaughan]

I don't think this is expected. I would have expected the first to return an x column containing c(1, 2, 3, 4) as a double vector.

library(dplyr)

df <- tibble(x = list(1, 2, 3, 4))
df <- rowwise(df)

# `x` isn't accessed an element at a time
mutate(df, y = x)
#> # A tibble: 4 × 2
#> # Rowwise: 
#>   x         y        
#>   <list>    <list>   
#> 1 <dbl [1]> <dbl [1]>
#> 2 <dbl [1]> <dbl [1]>
#> 3 <dbl [1]> <dbl [1]>
#> 4 <dbl [1]> <dbl [1]>

# `x` is accessed an element at a time
mutate(df, y = x + 1)
#> # A tibble: 4 × 2
#> # Rowwise: 
#>   x             y
#>   <list>    <dbl>
#> 1 <dbl [1]>     2
#> 2 <dbl [1]>     3
#> 3 <dbl [1]>     4
#> 4 <dbl [1]>     5

I think something is off with this section. We can't just use the .data[[name]] result here, we need to unchop it (checking that all elements are size 1 first)

dplyr/R/mutate.R

Lines 323 to 336 in 55dfc1c

	# column from the original data
	result <- .data[[name]]
	chunks <- mask$resolve(name)
	}
	if (inherits(.data, "rowwise_df") && vec_is_list(result)) {
	sizes <- list_sizes(result)
	wrong <- which(sizes != 1)
	if (length(wrong)) {
	# same error as would have been generated by mask$eval_all_mutate()
	group <- wrong[1L]
	mask$set_current_group(group)
	abort(

[DavisVaughan]

Here is another example that follows a slightly different code path (because the column doesn't come from the original data) but it also has the same problem

library(dplyr)

df <- tibble(x = list(1, 2, 3, 4))
df <- rowwise(df)

# `y` isn't accessed an element at a time
mutate(df, y = list(1), z = y)
#> # A tibble: 4 × 3
#> # Rowwise: 
#>   x         y         z        
#>   <list>    <list>    <list>   
#> 1 <dbl [1]> <dbl [1]> <dbl [1]>
#> 2 <dbl [1]> <dbl [1]> <dbl [1]>
#> 3 <dbl [1]> <dbl [1]> <dbl [1]>
#> 4 <dbl [1]> <dbl [1]> <dbl [1]>

# `y` is accessed an element at a time
mutate(df, y = list(1), z = y + 1)
#> # A tibble: 4 × 3
#> # Rowwise: 
#>   x         y             z
#>   <list>    <list>    <dbl>
#> 1 <dbl [1]> <dbl [1]>     2
#> 2 <dbl [1]> <dbl [1]>     2
#> 3 <dbl [1]> <dbl [1]>     2
#> 4 <dbl [1]> <dbl [1]>     2

This one goes through here, which is the other branch from above

dplyr/R/mutate.R

Lines 319 to 321 in 55dfc1c

	# already have result and chunks
	result <- new_columns[[name]]
	chunks <- mask$resolve(name)