TFE_DeleteOp(add_op);

  TF_Tensor* t = TFE_TensorHandleResolve(retval, status);
  if (async) {
    if (forward_input) {
      // Since the input was forwarded, we released the input handle right away
      // and hence expect the input to be forwarded to the return tensor.
      EXPECT_EQ(orig_ptr, TF_TensorData(t));
    } else {
      // In async mode we expect forwarding to work without releasing the input

  return schemes;
}

// `INSTANTIATE_TEST_SUITE_P` is called once for every `TEST_P`. However, we
// only want to analyze the user provided schemes and those that are registered
// only once. Hence, this function keeping another static pointer to a vector
// which contains only the schemes under test.
//
// Without this additional step, when there are schemes available but the user

    TFE_TensorHandle* h, TF_Status* status);

// This function will block till the operation that produces `h` has
// completed. The memory returned might alias the internal memory used by
// TensorFlow. Hence, callers should not mutate this memory (for example by
// modifying the memory region pointed to by TF_TensorData() on the returned
// TF_Tensor).
TF_CAPI_EXPORT extern TF_Tensor* TFE_TensorHandleResolve(TFE_TensorHandle* h,

// that provides the filesystem functionality. However, the POSIX filesystem
// needs to be statically registered in some tests and utilities for building
// the API files at the time of creating the pip package. Hence, we need to
// expose this function so that this filesystem can be statically registered
// when needed.
void TF_InitPlugin(TF_FilesystemPluginInfo* info);

    srcs = ["posix_filesystem_helper.cc"],
    hdrs = ["posix_filesystem_helper.h"],
    deps = [":copy_file"],
)

# On Linux, we can copy files faster using `sendfile`. But not elsewhere.
# Hence, this private library to select which implementation to use.
cc_library(
    name = "copy_file",
    srcs = select({
        "//tensorflow:linux_x86_64": ["copy_file_linux.cc"],

# bazel-out directory. Next, it archives the generated source files
# into a srcjar directly under the root. There doesn't appear to be a
# simple way to parameterize this from bazel, hence this helper to
# "normalize" the srcjar layout.
#
# Arguments:
#   src_jar - path to the original srcjar
#   dest_jar - path to the destination
# Returns:
#   None
function cp_normalized_srcjar() {

// values and hence receives 5 gradient values during backprop. However the
// gradient function uses only the first of those values and ignores the rest.
// The entry, "FusedBatchNorm": [1, 2, 3, 4], indicates that only the gradient
// corresponding to index 0 is used, and the gradient values at indices 1-4 are
// ignored (and hence can be None). The backprop algorithm can then leverage

    and
    [tensorflow/python](https://github.com/tensorflow/tensorflow/tree/master/tensorflow/python).
    TensorFlow has passed version 1.0 and hence cannot make
    non-backward-compatible API changes without a major release. Reviewers of
    your pull request will comment on any API compatibility issues

# bazel-out directory. Next, it archives the generated source files
# into a srcjar directly under the root. There doesn't appear to be a
# simple way to parameterize this from bazel, hence this helper to
# "normalize" the srcjar layout.
#
# Arguments:
#   src_jar - path to the original srcjar
#   dest_jar - path to the destination
# Returns:
#   None
function cp_normalized_srcjar() {

// maintains the map from `op_id` to a `OpTapeEntry` which stores the `op_type`,
// inputs and outputs and the gradient function These data structures combined
// allow us to trace the data dependencies between operations and hence compute
// gradients.
//
// `ZerosLike` is not expected to be called and returns a nullptr. The creation
// of default zeros grads is handled by the `DefaultGradientFunction` registered
// for each op.