// EagerService to maintain consistency between client and worker. The
// 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.

// Return a shape op fetching the shape of `a`.
TFE_Op* ShapeOp(TFE_Context* ctx, TFE_TensorHandle* a);

// Return an allreduce op adding up input tensor `in` from `group_size` workers.
TFE_Op* AllReduceOp(TFE_Context* ctx, TFE_TensorHandle* in, int group_size);

// Return a SendOp op `op_name` with send input tensor `in` and attributes
// `send_device`, `recv_device`, and `send_device_incarnation` set.

      TF_Status* status) const;

  void AsyncWait(TFE_Context* context, TF_Status* status) const;

  // Device strings for component devices that only include a
  // worker/task/replica if any of those differ across components. Useful for
  // printing debug messages.
  std::vector<std::string> SummarizeDeviceNames() const;

 private:

// Counts the number of elements in the device list.
TF_CAPI_EXPORT extern int TF_DeviceListCount(const TF_DeviceList* list);

// Retrieves the full name of the device (e.g. /job:worker/replica:0/...)
// The return value will be a pointer to a null terminated string. The caller
// must not modify or delete the string. It will be deallocated upon a call to
// TF_DeleteDeviceList.
//

TF_CAPI_EXPORT extern void TFE_AbortCollectiveOps(TFE_Context* ctx,
                                                  TF_Status* status);

// Checks the health of collective ops peers. Explicit health check is needed in
// multi worker collective ops to detect failures in the cluster.  If a peer is
// down, collective ops may hang.
TF_CAPI_EXPORT extern void TFE_CollectiveOpsCheckPeerHealth(
    TFE_Context* ctx, const char* task, int64_t timeout_in_ms,

                                      int64_t init_timeout_in_ms, int retries,
                                      bool clear_existing_contexts = false) = 0;

  // Initializes context for the local worker and no contexts will be created
  // for remote workers. Currently this only works for resetting context.
  // TODO(b/289445025): Consider removing this when we find a proper fix.
  virtual Status InitializeLocalOnlyContext(const ServerDef& server_def,

TF_CAPI_EXPORT extern TFE_ContextDevicePlacementPolicy
TFE_ContextGetDevicePlacementPolicy(TFE_Context* ctx);

// A tensorflow.ServerDef specifies remote workers (in addition to the current
// workers name). Operations created in this context can then be executed on
// any of these remote workers by setting an appropriate device.
//
// If the following is set, all servers identified by the
// ServerDef must be up when the context is created.