#include "tensorflow/c/eager/c_api.h"

#ifdef __cplusplus
extern "C" {
#endif

// Resets `op_to_reset` with `op_or_function_name` and `raw_device_name`. This
// is for performance optimization by reusing an exiting unused op rather than
// creating a new op every time. If `raw_device_name` is `NULL` or empty, it
// does not set the device name. If it's not `NULL`, then it attempts to parse

  // blocking collective operations at a time), and avoids some dynamic
  // allocation/scheduling.
  //
  // TODO(allenl): Keep a map from outer thread to list of inner threads rather
  // than a single list of threads so aliased nested parallel devices don't
  // re-use a thread.
  std::vector<std::unique_ptr<DeviceThread>> device_threads_;

TF_CAPI_EXPORT void TF_SetXlaMinClusterSize(int size);

// Gets/Sets TF/XLA flag for whether(true) or not(false) to disable constant
// folding. This is for testing to ensure that XLA is being tested rather than
// Tensorflow's CPU implementation through constant folding.
TF_CAPI_EXPORT unsigned char TF_GetXlaConstantFoldingDisabled();
TF_CAPI_EXPORT void TF_SetXlaConstantFoldingDisabled(
    unsigned char should_enable);

    // Ownership of `aid` transferred to CallBackwardFunction below.
    Gradient* aid;
    if (output_tensor.GetDType() == tensorflow::DT_VARIANT) {
      // Note: Needs to be zeros rather than ones since there's currently no
      // ones_like for variants.
      aid = output_tensor.ZerosLike();
    } else {
      // TODO(allenl): Figure out why using zeros_like everywhere causes issues