Using a named vector with dplyr::select(dplyr::all_of(...)) changes the table column names

[andreassoteriadesmoj]

When passing a named vector in dplyr::select(dplyr::all_of(...)), the names of the table fields to select are replaced by the names of the vector elements containing the name of the fields. Is this a bug?

I'm using dbplyr 2.4.0

iris_db <- iris %>% 
  dbplyr::memdb_frame()

# Named vector #
species <- c(sp = 'Species')
# Expecting 'Species', getting 'sp' (= names(species))
iris_db |>
  dplyr::select(
    dplyr::all_of(
      species
    )
  ) |> 
  colnames()

# [1] "sp"

# Not named vector #
species_noname <- 'Species'
# Expecting 'Species', getting 'Species' (!= names(species))
iris_db |>
  dplyr::select(
    dplyr::all_of(
      species_noname
    )
  ) |> 
  colnames()

# [1] "Species"

[hadley]

This is by design of dplyr:

library(dplyr, warn.conflicts = FALSE)


species <- c(sp = 'Species')
iris |> select(all_of(species)) |> colnames()
#> [1] "sp"
iris |> rename(all_of(species)) |> colnames()
#> [1] "Sepal.Length" "Sepal.Width"  "Petal.Length" "Petal.Width"  "sp"

^{Created on 2024-02-14 with reprex v2.0.2.9000}