# When we specify multiple JDKs, the final one becomes the default, which is used to execute Maven itself.
      # Our Maven configuration then specifies different JDKs to use for some of the steps:
      # - 11 (sometimes) to *download* to support anyone who runs JDiff or our Gradle integration tests (including our doc snapshots and our Java 11 CI test run) but not to use directly

         * only way to get a TypeVariable instance for the resolved types is to create our own. The
         * created TypeVariable will not interoperate with any JDK TypeVariable. But this is OK: We
         * don't _want_ our new TypeVariable to be equal to the JDK TypeVariable because it has
         * _different bounds_ than the JDK TypeVariable. And it wouldn't make sense for our new

         * only way to get a TypeVariable instance for the resolved types is to create our own. The
         * created TypeVariable will not interoperate with any JDK TypeVariable. But this is OK: We
         * don't _want_ our new TypeVariable to be equal to the JDK TypeVariable because it has
         * _different bounds_ than the JDK TypeVariable. And it wouldn't make sense for our new

            CollectionSize.ANY)
        .suppressing(suppressForArrayList())
        .createTestSuite();
  }

  // We are testing LinkedList / testing our tests on LinkedList.
  @SuppressWarnings("JdkObsolete")
  public Test testsForLinkedList() {
    return ListTestSuiteBuilder.using(
            new TestStringListGenerator() {
              @Override

   * {@link TypeVariableImpl#equals(Object)} is changed to no longer be equal to custom TypeVariable
   * implementations. As a result, we need to make sure our TypeVariable implementation respects
   * symmetry. Moreover, we don't want to reconstruct a native type variable {@code <A>} using our
   * implementation unless some of its bounds have changed in resolution. This avoids creating

              <!-- TODO(cpovirk): Enable NullArgumentForNonNullParameter for
                   prod code. It's disabled automatically for "test code"
                   (which is good: our tests have intentional violations), but
                   Error Prone doesn't know it's building test code unless we
                   pass -XepCompilingTestOnlyCode, and that argument needs to

              <!-- TODO(cpovirk): Enable NullArgumentForNonNullParameter for
                   prod code. It's disabled automatically for "test code"
                   (which is good: our tests have intentional violations), but
                   Error Prone doesn't know it's building test code unless we
                   pass -XepCompilingTestOnlyCode, and that argument needs to

   * {@link TypeVariableImpl#equals(Object)} is changed to no longer be equal to custom TypeVariable
   * implementations. As a result, we need to make sure our TypeVariable implementation respects
   * symmetry. Moreover, we don't want to reconstruct a native type variable {@code <A>} using our
   * implementation unless some of its bounds have changed in resolution. This avoids creating

    <dependency>
      <groupId>com.google.guava</groupId>
      <artifactId>guava</artifactId>
      <version>${project.version}</version>
    </dependency>
    <!-- We redeclare the j2objc-annotations dependency from `guava`: Our Gradle
         Module Metadata hides the dependency declared in `guava` from runtime
         configurations downstream, and GWT uses the runtime configuration for