}

    @Override
    public NavigableSet<E> subSet(
        @ParametricNullness E fromElement,
        boolean fromInclusive,
        @ParametricNullness E toElement,
        boolean toInclusive) {
      return forward.subSet(toElement, toInclusive, fromElement, fromInclusive).descendingSet();
    }

    @Override
    public SortedSet<E> subSet(@ParametricNullness E fromElement, @ParametricNullness E toElement) {

    }

    @Override
    public NavigableSet<E> subSet(
        @ParametricNullness E fromElement,
        boolean fromInclusive,
        @ParametricNullness E toElement,
        boolean toInclusive) {
      return forward.subSet(toElement, toInclusive, fromElement, fromInclusive).descendingSet();
    }

    @Override
    public SortedSet<E> subSet(@ParametricNullness E fromElement, @ParametricNullness E toElement) {

    }

    @Override
    public NavigableSet<E> subSet(
        E fromElement, boolean fromInclusive, E toElement, boolean toInclusive) {
      synchronized (mutex) {
        return Synchronized.navigableSet(
            delegate().subSet(fromElement, fromInclusive, toElement, toInclusive), mutex);
      }
    }

    @Override
    public SortedSet<E> subSet(E fromElement, E toElement) {

    }

    @Override
    public NavigableSet<E> subSet(
        E fromElement, boolean fromInclusive, E toElement, boolean toInclusive) {
      synchronized (mutex) {
        return Synchronized.navigableSet(
            delegate().subSet(fromElement, fromInclusive, toElement, toInclusive), mutex);
      }
    }

    @Override
    public SortedSet<E> subSet(E fromElement, E toElement) {

  TF_Operation* func_feed = Placeholder(host_graph_, s_);
  TF_Operation* func_op = Use({func_feed, five});
  Run({{func_feed, Int32Tensor(2)}}, func_op, 2 /*+=*/ + 5 + 1);

  // Verify input, output, and subset of edges in fdef.
  // The subset of edges we verify is a chain between feed1 and output to
  // make sure that the correct output is picked.
  tensorflow::FunctionDef fdef;
  ASSERT_TRUE(GetFunctionDef(func_, &fdef));

// `max_func`. In usual usage, max_func should be set to the result of
// TF_GraphNumFunctions(g). In this case, all the functions registered in
// `g` will be returned. Else, an unspecified subset.
//
// If successful, returns the number of TF_Function* successfully set in
// `funcs` and sets status to OK. The caller takes ownership of
// all the returned TF_Functions. They must be deleted with TF_DeleteFunction.

# causing library conflicts (e.g., those reported in github issue 1924).
# On Linux, tell the linker (-Wl,<option>) to use a version script that
# excludes all but a subset of function names.
# On MacOS, the linker does not support version_script, but has an
# an "-exported_symbols_list" command.  -z defs disallows undefined
# symbols in object files.

pywrap_aware_tf_cc_shared_object(

  }

  private TypeToken<?> resolveSupertype(Type type) {
    TypeToken<?> supertype = of(getCovariantTypeResolver().resolveType(type));
    // super types' type mapping is a subset of type mapping of this type.
    supertype.covariantTypeResolver = covariantTypeResolver;
    supertype.invariantTypeResolver = invariantTypeResolver;
    return supertype;
  }

  /**