// context_id is initialized with a dummy value and is later set when the worker
// is initialized (either locally or remotely). The context_id can change during
// the process lifetime although this should cause the worker to be
// reinitialized (e.g. cleared caches) as well.
TF_CAPI_EXPORT extern uint64_t TFE_GetContextId(TFE_Context* ctx);

// -----------------------------------------------------------------------------

    **Note:** You can add `--test_sharding_strategy=disabled` to the `flags` to
    disable the sharding so that all the test outputs are in one file. However,
    it may slow down the tests for not running in parallel and may cause the
    test to timeout but it could be useful when you need to execute a single
    test or more in general your filtered/selected tests have a very low
    execution time and the sharding

  // in an initialized context.
  for (auto d = devices.begin(); d != devices.end();) {
    if (absl::StrContains(d->get()->name(), "CPU:0")) {
      d = devices.erase(d);
    } else {
      ++d;
    }
  }

  status->status = tensorflow::unwrap(ctx)->AddDevices(std::move(devices));
}

void TFE_InsertConfigKeyValue(TFE_Context* ctx, const char* key,

  it->second--;
  if (it->second != 0) {
    return;
  }
  tensor_usage_.erase(it);
  auto tensor_op_it = tensor_tape_.find(tensor_id);
  if (tensor_op_it == tensor_tape_.end()) {
    return;
  }
  const int64_t op_id = tensor_op_it->second;
  if (op_id == -1) {
    // Do not delete watched tensors.
    return;
  }
  tensor_tape_.erase(tensor_op_it);
  auto op_it = op_tape_.find(op_id);

                        const int in_flight_nodes_limit)
      : status_(TF_NewStatus()),
        // If the context's default exector is set to async, re-using that in
        // each thread would cause collectives to deadlock. For consistency we
        // create a new sync executor for every thread.
        //
        // TODO(allenl): We should have an async API that works with the
        // parallel device.

// on both the main function side and the component function side, an error will
// be thrown in the registered graph optimization pass.
TEST(CAPI, DistributedFunctionGraphPassOnlyOnce) {
  // Register graph pass that will raise error if called more than once.
  tensorflow::optimization_registration::OptimizationPassRegistration
      register_test_pass(tensorflow::OptimizationPassRegistry::PRE_PLACEMENT, 0,