Tests for `void`.

pkg/front_end/test/fasta/function_type_recovery_test.dart

Index: pkg/front_end/test/fasta/function_type_recovery_test.dart
diff --git a/pkg/front_end/test/fasta/function_type_recovery_test.dart b/pkg/front_end/test/fasta/function_type_recovery_test.dart
new file mode 100644
index 0000000000000000000000000000000000000000..a7483273807f780b7e1e134496933c5bb843f38c
--- /dev/null
+++ b/pkg/front_end/test/fasta/function_type_recovery_test.dart
@@ -0,0 +1,13 @@
+// Copyright (c) 2017, the Dart project authors.  Please see the AUTHORS file
+// for details. All rights reserved. Use of this source code is governed by a
+// BSD-style license that can be found in the LICENSE file.
+
+// Tests that an inline function type inside a `Function` type isn't a
+// parser error.
+
+typedef F = int Function(int f(String x));

[eernst, 2017/02/21 13:39:11]

Actually, upon request I specified that the `T Function(..)` syntax should _not_
allow for nested occurrences of old-style `functionSignature`s like `int
f(String x)`, cf.
https://gist.github.com/eernstg/ffc7bd281974e9018f10f0cb6cfee4aa. 

So we should not test that this is accepted, we should insist that it is
rejected (ideally as a syntax error, and certainly at compile-time).

The same issue arises at line 12 below.

+
+main() {
+  F f = null;
+  String Function(String g(int y)) g = null;
+}