TFE_Context* ctx = TFE_NewContext(opts, status);
  EXPECT_EQ(TF_GetCode(status), TF_OK) << TF_Message(status);
  TFE_ContextSetServerDefWithTimeout(ctx, 0, serialized_server_def.data(),
                                     serialized_server_def.size(),
                                     init_timeout_in_ms, status,
                                     /*clear_existing_contexts=*/false);

  virtual Status AddDevices(std::vector<std::unique_ptr<Device>> devices) = 0;

  // Block until all pending nodes are finished.
  virtual Status AsyncWait() = 0;

  // Add a function (serialized FunctionDef protocol buffer) so that it can
  // be executed as an op. Return error if the function with the same name
  // already exists.
  virtual Status AddFunctionDef(const FunctionDef& fdef) = 0;

  }
  return nullptr;
}

void TFE_ContextAddFunctionDef(TFE_Context* ctx,
                               const char* serialized_function_def, size_t size,
                               TF_Status* status) {
  tensorflow::FunctionDef function_def;
  if (!function_def.ParseFromArray(serialized_function_def, size)) {
    status->status =
        tensorflow::errors::InvalidArgument("Invalid FunctionDef proto");
    return;
  }

  // new worker (different incarnation ID).
  server_def.set_task_index(0);
  string serialized_update = server_def.SerializeAsString();
  TFE_ContextUpdateServerDef(ctx, 0, serialized_update.data(),
                             serialized_update.size(), status);
  ASSERT_EQ(TF_OK, TF_GetCode(status)) << TF_Message(status);

  TFE_Context* ctx = TFE_NewContext(opts, status);
  EXPECT_EQ(TF_OK, TF_GetCode(status)) << TF_Message(status);
  TFE_DeleteContextOptions(opts);

  TFE_ContextSetServerDef(ctx, 0, serialized.data(), serialized.size(), status);
  EXPECT_EQ(TF_OK, TF_GetCode(status)) << TF_Message(status);

  TFE_TensorHandle* h0_task0 = TestMatrixTensorHandle(ctx);
  TFE_TensorHandle* h1_task0 = TestMatrixTensorHandle(ctx);

#include "tensorflow/core/platform/refcount.h"
#include "tensorflow/core/platform/status.h"

namespace tensorflow {

class FunctionRecord;

// A traced function: this hides the complexity of converting the serialized
// representation between various supported formats e.g. FunctionDef and Mlir
// function.
class AbstractFunction : public core::RefCounted {
 protected:
  enum AbstractFunctionKind { kGraph, kMlir };

// Tensorflow's CPU implementation through constant folding.
TF_CAPI_EXPORT unsigned char TF_GetXlaConstantFoldingDisabled();
TF_CAPI_EXPORT void TF_SetXlaConstantFoldingDisabled(
    unsigned char should_enable);

// Create a serialized tensorflow.ConfigProto proto, where:
//
// a) ConfigProto.optimizer_options.global_jit_level is set to ON_1 if
// `enable_xla_compilation` is non-zero, and OFF otherwise.

  TFE_Context* ctx = TFE_NewContext(opts, status);
  EXPECT_EQ(TF_GetCode(status), TF_OK) << TF_Message(status);
  TFE_DeleteContextOptions(opts);

  TFE_ContextSetServerDef(ctx, 0, serialized.data(), serialized.size(), status);
  EXPECT_EQ(TF_GetCode(status), TF_OK) << TF_Message(status);

  const char task0_name[] = "/job:localhost/replica:0/task:0/device:CPU:0";

// Sets `value` to the binary-serialized TensorShapeProto of the value of
// `attr_name` attribute of `oper`.
TF_CAPI_EXPORT extern void TF_OperationGetAttrTensorShapeProto(
    TF_Operation* oper, const char* attr_name, TF_Buffer* value,
    TF_Status* status);

// Fills in `values` with binary-serialized TensorShapeProto values of the

  ASSERT_EQ(TF_OK, TF_GetCode(status));

  const OpDef* expected_op_def;
  TF_ASSERT_OK(OpRegistry::Global()->LookUpOpDef("Add", &expected_op_def));
  string expected_serialized;
  expected_op_def->SerializeToString(&expected_serialized);
  EXPECT_EQ(expected_serialized, actual_string);
  TF_DeleteBuffer(g_buf);
  TF_DeleteBuffer(op_buf);
  TF_DeleteStatus(status);
  TF_DeleteFunctionLibraryDefinition(func);
}