dplyr::if_else() fails when condition is a labelled logical vector

[salim-b]

Starting with dplyr 1.1.0, dplyr::if_else() fails with a meaningless error message when condition is a labelled logical vector:

d <-
    mtcars |>
    dplyr::mutate(mileage = mpg > 20) |>
    labelled::set_variable_labels(mileage = "mileage dummy")

dplyr::if_else(d$mileage,
               "guzzler",
               "moderate")
#> Error in `dplyr::if_else()`:
#> ! `condition` must be a vector with type <logical>.
#> Instead, it has type <logical>.

#> Backtrace:
#>     ▆
#>  1. └─dplyr::if_else(d$mileage, "guzzler", "moderate")
#>  2.   └─vctrs::vec_assert(x = condition, ptype = logical(), arg = "condition")
#>  3.     └─rlang::abort(...)

^{Created on 2023-02-02 with reprex v2.0.2}

Explicitly providing the correct prototype gives the same result:

dplyr::if_else(d$mileage,
               "guzzler",
               "moderate",
               ptype = vctrs::vec_ptype(d$mileage))
#> Error in `dplyr::if_else()`:
#> ! `condition` must be a vector with type <logical>.
#> Instead, it has type <logical>.

#> Backtrace:
#>     ▆
#>  1. └─dplyr::if_else(d$mileage, "guzzler", "moderate", ptype = vctrs::vec_ptype(d$mileage))
#>  2.   └─vctrs::vec_assert(x = condition, ptype = logical(), arg = "condition")
#>  3.     └─rlang::abort(...)