sample_n.sf can't handle grouped samples when sample size refers to another column

[daranzolin]

Consider the following pattern whereby I indicate an ideal sample size per group, but if the group size is not large enough, it defaults to the group size:

suppressPackageStartupMessages(library(dplyr))
#> Warning: package 'dplyr' was built under R version 4.0.2
new_iris <- sample_n(iris, 30) %>% 
  mutate(Species = "New Species") %>% 
  bind_rows(iris)

new_iris %>% 
  add_count(Species, name = "obs") %>% 
  group_by(Species) %>% 
  mutate(sample_size = ifelse(obs < 40, obs, 40)) %>% 
  sample_n(sample_size, replace = FALSE) %>% 
  count(Species)
#> # A tibble: 4 x 2
#> # Groups:   Species [4]
#>   Species         n
#>   <chr>       <int>
#> 1 New Species    30
#> 2 setosa         40
#> 3 versicolor     40
#> 4 virginica      40

^{Created on 2020-12-06 by the reprex package (v0.3.0)}

This same pattern fails with objects of class 'sf':

library(spData)
library(sf)
#> Linking to GEOS 3.7.2, GDAL 2.4.2, PROJ 5.2.0
suppressPackageStartupMessages(library(dplyr))

world %>% 
  add_count(continent, name = "obs") %>% 
  group_by(continent) %>% 
  mutate(sample_size = ifelse(obs < 10, obs, 10)) %>% 
  sample_n(sample_size, replace = FALSE)
#> Error in ~size: object 'sample_size' not found

^{Created on 2020-12-06 by the reprex package (v0.3.0)}

Is this a bug or is there another method for obtaining these results?