Incorrect results of `fct_reorder()` with a character vector and duplicates

[DanChaltiel]

Hi,

I encountered an unexpected result with fct_reorder() when .x is not numeric and .f has duplicates:

• If .x is a character, the result is incorrect and there is a warning
• If .x is a factor, there is an unexpected error.

f = c("a", "b", "b")
x = c(3,2,1)

#This is OK
forcats::fct_reorder(f, x)
#> [1] a b b
#> Levels: b a

x = c("z", "x", "y")
forcats::fct_reorder(f, x)
#> Warning in mean.default(sort(x, partial = half + 0L:1L)[half + 0L:1L]):
#> argument is not numeric or logical: returning NA
#> [1] a b b
#> Levels: a b

x = factor(c("z", "x", "y"))
forcats::fct_reorder(f, x)
#> Error in median.default(x, ...): need numeric data

#This is OK
x = as.numeric(factor(c("z", "x", "y")))
forcats::fct_reorder(f, x)
#> [1] a b b
#> Levels: b a

^{Created on 2025-12-02 with reprex v2.1.1}

Reordering a factor by another categorical variable is an important feature that has proven very useful in my work.

Thanks a lot for maintaining this great package.