Revert "don't pass on ..."

This reverts commit 5cb62d2.

R/new.R

@@ -3,20 +3,49 @@
 #' Creates a subclass of a tibble.
 #'
 #' @param x A tibble-like object
+#' @param ... Passed on to [structure()]
 #' @param nrow The number of rows, guessed from the data by default
 #' @param subclass Subclasses to assign to the new object, default: none
 #' @export
-new_tibble <- function(x, nrow = NULL, subclass = NULL) {
+new_tibble <- function(x, ..., nrow = NULL, subclass = NULL) {
   stopifnot(is.list(x))
   stopifnot(has_nonnull_names(x))
 
+  x <- update_tibble_attrs(x, ...)
   x <- set_tibble_class(x, subclass = subclass)
+  # Make sure that we override any row names that
+  # may have been there previously, in x or in ...
   if (is.null(nrow)) nrow <- guess_nrow(x)
   attr(x, "row.names") <- .set_row_names(nrow)
   validate_nrow(x)
   x
 }
 
+update_tibble_attrs <- function(x, ...) {
+  # Can't use structure() here because it breaks the row.names attribute
+  attribs <- list(...)
+
+  # reduce2() is not in the purrr compat layer
+  nested_attribs <- map2(names(attribs), attribs, function(name, value) set_names(list(value), name))
+  x <- reduce(
+    .init = x,
+    nested_attribs,
+    function(x, attr) {
+      if (!is.null(attr[[1]])) {
+        attr(x, names(attr)) <- attr[[1]]
+      }
+      x
+    }
+  )
+
+  x
+}
+
+set_tibble_class <- function(x, subclass = NULL) {
+  class(x) <- c(subclass, "tbl_df", "tbl", "data.frame")
+  x
+}
+
 guess_nrow <- function(x) {
   if (!is.null(.row_names_info(x, 0L))) .row_names_info(x, 2L)
   else if (length(x) == 0) 0L
@@ -37,8 +66,3 @@ validate_nrow <- function(x) {
 
   invisible(x)
 }
-
-set_tibble_class <- function(x, subclass = NULL) {
-  class(x) <- c(subclass, "tbl_df", "tbl", "data.frame")
-  x
-}

tests/testthat/test-tbl-df.R

@@ -301,6 +301,8 @@ test_that("is_tibble", {
 test_that("new_tibble", {
   tbl <- new_tibble(
     data.frame(a = 1:3),
+    attr1 = "value1",
+    attr2 = 2,
     subclass = "nt"
   )
 
@@ -310,6 +312,8 @@ test_that("new_tibble", {
     unclass(tbl),
     structure(
       list(a = 1:3),
+      attr1 = "value1",
+      attr2 = 2,
       .Names = "a",
       row.names = .set_row_names(3L)
     )