Fix Float.parse/1 inconsistent error handling for non-scientific notation overflow

`Float.parse/1` was returning `:error` for overflow in scientific notation (e.g.,
"1.0e400") but raising `ArgumentError` for equivalent overflow in non-scientific
notation (e.g., a 310-digit number). This violated the documented behavior that
states `Float.parse/1` should return `:error` when the float exceeds the maximum
size.

Author: pnezis

lib/elixir/lib/float.ex

@@ -201,13 +201,14 @@ defmodule Float do
   end
 
   defp parse_unsigned(rest, dot?, false = _e?, acc) do
-    float =
-      acc
-      |> add_dot(dot?)
-      |> :lists.reverse()
-      |> :erlang.list_to_float()
-
-    {float, rest}
+    acc
+    |> add_dot(dot?)
+    |> :lists.reverse()
+    |> :erlang.list_to_float()
+  rescue
+    ArgumentError -> :error
+  else
+    float -> {float, rest}
   end
 
   defp add_dot(acc, true), do: acc

lib/elixir/test/elixir/float_test.exs

@@ -57,6 +57,11 @@ defmodule FloatTest do
     assert Float.parse("1.7976931348623159e308") === :error
     assert Float.parse("1.7976931348623159e+308") === :error
     assert Float.parse("9e8363") === :error
+
+    # Non-scientific notation overflow
+    assert Float.parse(String.duplicate("9", 310) <> ".0") === :error
+    assert Float.parse("-" <> String.duplicate("9", 310) <> ".0") === :error
+    assert Float.parse(String.duplicate("9", 310) <> ".0foo") === :error
   end
 
   test "floor/1" do