I haven't used meta-programming operators in R for a while, and I came across this today. What is the recommended way to add a new column?

# Using embrace operator
add_col1 <- function(data, col_name){
  out <- data |> 
    dplyr::mutate({{ col_name }} := "new")
  
  return(out)
}

# Use 
add_col2 <- function(data, col_name){
  out <- data |> 
    dplyr::mutate(!!col_name := "new")
  
  return(out)
}

x1 <- add_col1(cars, "new_col")
x2 <- add_col2(cars, "new_col")

all.equal(x1, x2)

add_col1(cars, new_col)

# Fails due to missing enquo() call 
add_col2(cars, new_col)

What is the recommended way to write such dplyr-type function? From reading some of the docs, I found that using enquo() for normal usage is no longer recommended, which is why add_col2 fails without quotes here.

The only semantic advantage I can see to using add_col2 is that it enforces the use of quotes with col_name, which kinda suggests that the column does not yet exist. However, this is a very weak argument.