*
   * <p>Cancelling the pipeline will not cancel {@code future}, so that the pipeline can access its
   * value in order to close it.
   *
   * @param future the future to create the {@code ClosingFuture} from. For discussion of the
   *     future's result type {@code C}, see {@link DeferredCloser#eventuallyClose(Object,
   *     Executor)}.
   * @param closingExecutor the future's result will be closed on this executor

  //    the initial write (just stronger guarantees if it does).
  //
  // See: http://cs.oswego.edu/pipermail/concurrency-interest/2015-January/013800.html
  // For a (long) discussion about this specific issue and the general futility of life.
  //
  // For the time being we are OK with the problem discussed above since it requires a caller to

common:windows --features=compiler_param_file
common:windows --features=archive_param_file

# Speed Windows compile times. Available in VS 16.4 (we are on 16.11). See
# https://groups.google.com/a/tensorflow.org/d/topic/build/SsW98Eo7l3o/discussion
common:windows --copt=/d2ReducedOptimizeHugeFunctions
common:windows --host_copt=/d2ReducedOptimizeHugeFunctions

# Before VS 2017 15.8, the member "type" would non-conformingly have an

   * usually be handled more naturally using {@code tryFind(iterator, predicate).or(defaultValue)}.
   *
   * @since 7.0
   */
  // For discussion of this signature, see the corresponding overload of *Iterables*.find.
  public static <T extends @Nullable Object> @Nullable T find(
      Iterator<? extends T> iterator, Predicate<? super T> predicate, @Nullable T defaultValue) {

        new TypeToken<StringForFirstTypeArg<Integer>>() {};
    assertTrue(subtype.isSubtypeOf(supertype));

    // TODO(benyu): This should check equality to an expected value, see discussion in cl/98674873
    TypeToken<?> unused = supertype.getSubtype(subtype.getRawType());
  }

  public void testGetSubtype_baseClassWithNoTypeArgs() {
    class SingleGenericExtendsBase<T> extends Base {}