Error in kruskal_test (coin) with Weighted Data: “Invalid Object Class 'IndependenceProblem’”

I am trying to perform a weighted Kruskal-Wallis test for each group in a dataset, but I keep encountering the following error:

Error in kruskal_test: Error in validObject(.Object): invalid class “IndependenceProblem” object: FALSE

Here is the code I am using:

# Load necessary libraries
library(tidyverse)
library(coin)

# Set seed for reproducibility
set.seed(1)

# Create the dataset
punteggi <- tibble(
  Codice = paste0("cod_", 1:200),
  Regione = sample(c("FVG", "Lazio", "Sicilia"), size = 200, replace = TRUE, prob = c(.4, .2, .4)),
  Genere = sample(c("Femminile", "Maschile"), size = 200, replace = TRUE, prob = c(.6, .4)),
  Area = sample(c("Urban", "Suburban"), size = 200, replace = TRUE, prob = c(.6, .4))
)

punteggi <- punteggi |>
  group_by(Regione, Genere, Area) |>
  mutate(
    mu = runif(1, 4, 7),
    sd = runif(1, 1, 2),
    weight = 1/n()
  ) |> 
  ungroup() |>
  rowwise() |>
  mutate(pt_tot = rnorm(1, mu, sd))

print(punteggi)

# Function to perform the weighted Kruskal-Wallis test for each group
kruskal_per_gruppo <- function(data) {
  test_result <- tryCatch({
    data <- data |> mutate(across(weight, ~./sum(.)))
    kruskal_test(pt_tot ~ value, data = data, weights = ~ weight)
  }, error = function(e) {
    message("Error in kruskal_test: ", e)
    return(NULL)
  })
  return(test_result)
}

# Apply the Kruskal-Wallis test for each group
results <- punteggi |> 
  pivot_longer(c(Regione, Genere, Area)) |> 
  group_by(name) |> 
  mutate(
    across(value, as.factor),
    kruskal = kruskal.test(pt_tot ~ value)$p.value
  ) |> 
  nest() |> 
  mutate(kruskal_weighted = map(data, kruskal_per_gruppo))

print(results)

What might be causing this error and how can I resolve it? I have checked that the weights are correctly normalized, but it seems there is an issue with creating the IndependenceProblem object. Any suggestions would be greatly appreciated!

The weights for all the coin package tests must be integers. They are treated as frequency weights -- ie, if you have a weight of 3 it is exactly like having three copies of that observation.