# See the License for the specific language governing permissions and
# limitations under the License.
# ==============================================================================
#
# Re-direct all links in $1 that are relative to be canonical

BASE="$1"
find "${BASE}" -type l | \
  while read l ; do
    if [[ "$(readlink "$l")" == \.\./* ]]; then
      CANONICAL="$(readlink "$l")";
      rm "$l";
      ln -s "${CANONICAL}" "$l"

  // appropriate.
  virtual tensorflow::FullTypeDef FullType() const = 0;

  // The default debug string includes a shape, dtype and FullType.
  // Implementations are free to override it with something more informative.
  virtual std::string DebugString() const;

  AbstractTensorHandleKind getKind() const { return kind_; }

 private:
  const AbstractTensorHandleKind kind_;
};

# Checks TFCI_BAZEL_COMMON_ARGS for "gpu" or "cuda", implying that the test is
# relevant. All of the GPU test machines have CUDA installed via other means,
# so I am not sure how to verify that the dependencies themselves are valid for
# the moment.
if [[ "$TFCI_BAZEL_COMMON_ARGS" =~ gpu|cuda ]]; then
  echo "Checking to make sure tensorflow[and-cuda] is installable..."

// Implementation detail for the unified execution APIs for Eager and tracing
// backends (graph/MLIR).
//
// This defines a set of abstract classes that are intended to provide the
// functionality of the opaque C types exposed in the public APIs defined in the
// `c_api_unified_experimental.h` header.
// =============================================================================

    Run(inputs, {{output, 0}}, {expected_result});
  }

  // Run the host graph, which now contains a function and check that
  // outputs are as expected.
  // 'T' stands for 'tensor' since the outputs are tensors, not scalars.
  void RunT(const std::vector<std::pair<TF_Operation*, TF_Tensor*>>& inputs,
            std::initializer_list<TF_Output> outputs,

        ":linux_armhf",
    ],
)

config_setting(
    name = "freebsd",
    values = {"cpu": "freebsd"},
    visibility = ["//visibility:public"],
)

# Features that are default ON are handled differently below.
#
config_setting(
    name = "no_gcp_support",
    define_values = {"no_gcp_support": "true"},
    visibility = ["//visibility:public"],
)

config_setting(

    const char* fn_name = "test_fn";
    core::RefCountPtr<AbstractFunction> scoped_func;
    // Returning null tensors from a tf.function is not supported, so we keep
    // track of indices in the model's outputs are nullptr in this set.
    // The FunctionDef only outputs the non-null tensors. We later pad the
    // function op outputs to have nullptrs at the `null_indices`.
    absl::flat_hash_set<int> null_indices;
    {

  // Consumes the internal state of the tape (so cannot be called more than
  // once unless the tape is persistent) and produces the gradient of the target
  // tensors with respect to the source tensors. The output gradients are used
  // if not empty and not null. The result is populated with one tensor per
  // target element.
  Status ComputeGradient(
      AbstractContext* ctx, absl::Span<AbstractTensorHandle* const> targets,

    if (gcs_file->stat_cache->Lookup(path, &stat)) {
      if (offset + read < stat.base.length) {
        TF_SetStatus(status, TF_INTERNAL,
                     absl::StrCat("File contents are inconsistent for file: ",
                                  path, " @ ", offset)
                         .c_str());
      }
      TF_VLog(2, "Successful integrity check for: %s @ %u", path.c_str(),
              offset);

    TFE_OpAddInput(matmul, has_packed_input ? packed_handle : h1_task2, status);
    ASSERT_EQ(TF_GetCode(status), TF_OK) << TF_Message(status);
  } else {
    // Handles are on task0 (local), and task2, but op is on task1.
    matmul = MatMulOp(ctx, h0_task0, h1_task2);
  }
  if (remote) {
    TFE_OpSetDevice(matmul, task1_name, status);