More informative error for case_when() RHS type and class mismatches

[eutwt]

closes #6206

Current

dplyr::case_when(c(TRUE, FALSE) ~ c(3, 4), TRUE ~ c('a', 'b'))
#> Error in names(message) <- `*vtmp*`: 'names' attribute [1] must be the same length as the vector [0]

^{Created on 2022-03-31 by the reprex package (v2.0.1)}

New

dplyr::case_when(c(TRUE, FALSE) ~ c(3, 4), TRUE ~ c('a', 'b'))
#> Error in `dplyr::case_when()`:
#> ! `c("a", "b")` must be a double vector, not a character vector.

^{Created on 2022-03-31 by the reprex package (v2.0.1)}

I didn't update snapshots because the line-wrapping in my error messages happens earlier than in the snapshot, and I'm not sure why.

[ericnewkirk]

I was just looking into this too, sorry for butting in. I think you need fmt_calls(quos_pairs[[i]]$rhs[2]) instead of fmt_calls(pair$rhs[i]) here. pair is just the last formula to be processed in the loop above, not necessarily the one triggering the error.

Try your version like this and you'll see what I mean:

case_when(
  1:10 > 8 ~ "a",
  1:10 > 6 ~ "b",
  1:10 > 4 ~ FALSE,
  1:10 > 2 ~ "c",
  TRUE ~ "d"
)

[eutwt]

Thanks @ericnewkirk, you are correct