Don't accidentally name anonymous module/class

b00f280d4b9569e7153365d7e1c522b3d6b3c6cf introduced
an accidental behavior change in that defining a module/class under
`m` gives `m` a name when `m` is anonymous.

`ruby -ve 'Module.new { class self::A; end; p name }'` outputs a name
similar to `Module#inspect` when it should output `nil` like in Ruby
2.6.x.

 * variable.c: Use `make_temporary_path` instead of `save_temporary_path`
   when getting the name of the parent module.

 * variable.c (rb_set_class_path): Delegate to `rb_set_class_path_string`
   instead of duplicating the logic.

[Bug #16097]

1 files changed, 18 insertions, 0 deletions

diff --git a/test/ruby/test_module.rb b/test/ruby/test_module.rb
index b8d6ffb19d..d56e63e4be 100644
--- a/test/ruby/test_module.rb
+++ b/test/ruby/test_module.rb
@@ -565,6 +565,24 @@ class TestModule < Test::Unit::TestCase
     assert_equal("TestModule::User",   User.name)
   end
 
+  def test_accidental_singleton_naming_with_module
+    o = Object.new
+    assert_nil(o.singleton_class.name)
+    class << o
+      module Hi; end
+    end
+    assert_nil(o.singleton_class.name)
+  end
+
+  def test_accidental_singleton_naming_with_class
+    o = Object.new
+    assert_nil(o.singleton_class.name)
+    class << o
+      class Hi; end
+    end
+    assert_nil(o.singleton_class.name)
+  end
+
   def test_classpath
     m = Module.new
     n = Module.new