super(experimentRunner, dataReporter, buildContext)
        this.pluginCommitSha = pluginCommitSha
        this.testGroup = "build scan plugin"

    }

    @Override
    CrossBuildPerformanceResults newResult() {
        new CrossBuildPerformanceResults(
            testClass: testClassName,
            testId: testId,
            testProject: testProject,
            testGroup: testGroup,

            throw new IllegalArgumentException("Gradle invocation display name must start with 'Gradle '")
        }
    }

    @Override
    GradleVsMavenBuildPerformanceResults newResult() {
        new GradleVsMavenBuildPerformanceResults(
            testClass: testClassName,
            testId: testId,
            testProject: testProject,
            testGroup: testGroup,

    main_func.setArgAttr(
        i, kTfSavedModelIndexPathAttr,
        ArrayAttr::get(context, {StringAttr::get(context, input_names[i])}));
  }

  const int num_results = main_func.getNumResults();
  for (int i = 0; i < num_results; ++i) {
    main_func.setResultAttr(
        i, kTfSavedModelIndexPathAttr,
        ArrayAttr::get(context, {StringAttr::get(context, output_names[i])}));
  }

  }

  return success();
}

LogicalResult CheckOutputConsumer(
    Operation* call_op, int expected_num_outputs,
    llvm::DenseSet<int> expected_consumer_indices) {
  const int num_results = call_op->getNumResults();
  if (num_results != expected_num_outputs) return failure();

  for (int i = 0; i < expected_num_outputs; ++i) {
    auto it = expected_consumer_indices.find(i);
    if (it == expected_consumer_indices.end()) {

  for (Operation& op : func.getOps()) {
    if (!IsTPUOp(&op)) continue;
    op.setAttr(TF::kReplicationInfoAttr, new_group);
  }
  TF::TPUCompilationResultOp new_result = compilation_op.clone();
  new_result->setAttr(kTpuCompilationStatus, new_group);
  builder.insert(new_result);
}

template <typename OpType>
LogicalResult FindAndExcludeOp(func::FuncOp func,
                               const StringAttr& replication_attr,

    auto av = MakeAvailableAsyncValueRef<FallbackTensor>(input_tensor);
    exec_arguments.push_back(av.release());
  }

  // Space for returned values.
  llvm::SmallVector<RCReference<AsyncValue>> results(compute->num_results());

  compute->Execute(*exec_ctx_, exec_arguments, results);

  // Wait for the function execution to finish, as well as the side-effects.
  host_context_->Await(results);

    const int64_t num_results = graph.getNumResults();
    if (mlir::isa<ControlType>(operand.getType())) {
      if (i != num_results)
        return fetch.emitOpError()
               << "operand #" << i
               << " is a control type, can't be bound to a graph result";
      break;
    }
    if (i >= num_results)
      return fetch.emitOpError()

		t.SetHasShape(true)
	}
	return t
}

func newResults(types []*Type) *Type {
	t := newType(TRESULTS)
	t.extra.(*Results).Types = types
	return t
}

func NewResults(types []*Type) *Type {
	if len(types) == 1 && types[0] == TypeMem {
		return TypeResultMem
	}
	return newResults(types)
}

func newSSA(name string) *Type {
	t := newType(TSSA)
	t.extra = name

	if conf.RawPrevResult != nil {
		resultBytes, err := json.Marshal(conf.RawPrevResult)
		if err != nil {
			return nil, fmt.Errorf("could not serialize prevResult: %v", err)
		}
		res, err := version.NewResult(conf.CNIVersion, resultBytes)
		if err != nil {
			return nil, fmt.Errorf("could not parse prevResult: %v", err)
		}
		conf.RawPrevResult = nil
		conf.PrevResult, err = cniv1.NewResultFromResult(res)

  }
  size_t num_results = ps.result().tuple_shapes_size();
  int readonly_variables = absl::c_count_if(
      config.variable(),
      [](const tf2xla::Variable& var) { return var.readonly(); });
  const int actual_num_results =
      config.fetch_size() + config.variable_size() - readonly_variables;
  if (actual_num_results != num_results) {