Revert allowing data frames in list_transpose()

NEWS.md

@@ -1,6 +1,7 @@
 # purrr (development version)
 
-* `list_transpose()` now works with data.frames (@KimLopezGuell, #1109).
+* Added a test to assert that `list_transpose()` does not work on data frames
+  (@KimLopezGuell, #1141, #1149).
 * Added `imap_vec()` (#1084)
 * `list_transpose()` inspects all elements to determine the correct
   template if it's not provided by the user  (#1128, @krlmlr).

R/list-transpose.R

@@ -69,8 +69,7 @@ list_transpose <- function(x,
                            simplify = NA,
                            ptype = NULL,
                            default = NULL) {
-
-  check_list(x)
+  obj_check_list(x)
   check_dots_empty()
 
   if (length(x) == 0) {

tests/testthat/_snaps/list-transpose.md

@@ -1,3 +1,11 @@
+# can't transpose data frames
+
+    Code
+      list_transpose(df)
+    Condition
+      Error in `list_transpose()`:
+      ! `x` must be a list, not a <data.frame> object.
+
 # integer template requires exact length of list() simplify etc
 
     Code
@@ -57,7 +65,7 @@
       list_transpose(10)
     Condition
       Error in `list_transpose()`:
-      ! `x` must be a list, not a number.
+      ! `x` must be a list, not the number 10.
     Code
       list_transpose(list(1), template = mean)
     Condition

tests/testthat/test-list-transpose.R

@@ -4,10 +4,13 @@ test_that("can transpose homogenous list", {
   expect_equal(out, list(a = c(x = 1, y = 3), b = c(x = 2, y = 4)))
 })
 
-test_that("can transpose data frames", {
+test_that("can't transpose data frames", {
   df <- data.frame(x = 1:2, y = 4:5)
-  out <- list_transpose(df)
-  expect_equal(out, list(c(x = 1, y = 4), c(x = 2, y = 5)))
+
+  # i.e. be consistent with other `list_*()` functions from purrr/vctrs
+  expect_snapshot(error = TRUE, {
+    list_transpose(df)
+  })
 })
 
 test_that("transposing empty list returns empty list", {