mutate(var = NULL) results in error

[rossellhayes]

Within regular dplyr, setting a variable to NULL removes that column:

library(dplyr, warn.conflicts = FALSE)

tibble(a = 1:3, b = 4:6) %>%
  mutate(a = NULL)
#> # A tibble: 3 x 1
#>       b
#>   <int>
#> 1     4
#> 2     5
#> 3     6

Following these same steps within a lazy_dt with dtplyr produces an error:

library(dtplyr)
library(dplyr, warn.conflicts = FALSE)

tibble(a = 1:3, b = 4:6) %>%
  lazy_dt() %>%
  mutate(a = NULL) %>%
  as_tibble()
#> Error: Invalid input

[MichaelChirico]

I think we're supposed to end up in the is_symbol branch of dt_squash? But rlang:::is_symbol(NULL) returns FALSE despite appearing like it's intended to return TRUE?

is_symbol
function (x, name = NULL) 
{
    # typeof(NULL) is "NULL", but is_null is returning
    #   x == R_NilValue from C, so I'm not sure how we'd
    #   get to the latter branch without returning from this one
    if (typeof(x) != "symbol") {
        return(FALSE)
    }
    if (is_null(name)) {
        return(TRUE)
    }
    as_string(x) %in% name
}