Error when referring to column names with rlang and sym when column name is identical to passed value

Question

What am I missing here.

I'm trying to do some data filtering based on columns that I pass into the filter function dynamically. This works as long as the column name itself is not identical to the object name in which I pass the column name:

Works:

df_a <- data.frame(x = 1:5)
df_b <- data.frame(x = 3:4)

id <- "x"

df_a |> 
  dplyr::filter(!!rlang::sym(id) %in% df_b[, id])

  x
1 3
2 4

Doesn't work

df_a <- data.frame(id = 1:5)
df_b <- data.frame(id = 3:4)

id <- "id"

df_a |> 
  dplyr::filter(!!rlang::sym(id) %in% df_b[, id])

Error in `dplyr::filter()`:
! Problem while computing `..1 = id %in% df_b[, id]`.
Caused by error in `[.data.frame`:
! undefined columns selected
Run `rlang::last_error()` to see where the error occurred.

What can I do to allow passing a column name that is identical to the object name which I pass?

---

Update:

This one seems to work, I don't understand yet, why:

df_a |> 
  dplyr::filter(!!rlang::sym(id) %in% eval(rlang::sym(!!id), df_b))