TF_EXPECT_OK(ReadCache(&cache, "a", 8, 1, &out));
  EXPECT_EQ(calls, 3);
  // Advance the fake timestamp so that "a" becomes stale via its first block.
  env->SetNowSeconds(now + 2);
  // The pruning thread periodically compares env->NowSeconds() with the oldest
  // block's timestamp to see if it should evict any files. At the current fake
  // timestamp of `now` + 2, file "a" is stale because its first block is stale,

Status CreateParamsForInputs(AbstractContext* ctx,
                             absl::Span<AbstractTensorHandle* const> inputs,
                             std::vector<AbstractTensorHandle*>* params);

// A callable that takes tensor inputs and returns zero or more tensor outputs.
using Model = std::function<Status(AbstractContext*,
                                   absl::Span<AbstractTensorHandle* const>,

# installed into a virtual environment, and then that venv is used to run all
# bazel tests with a special flag "--define=no_tensorflow_py_deps=true", which
# drops all the bazel dependencies for each py_test; this makes all the tests
# use the wheel's TensorFlow installation instead of the one made available
# through bazel. This must be done in a different root directory, //bazel_pip/...,

  opts = TF_NewImportGraphDefOptions();
  TF_ImportGraphDefOptionsSetPrefix(opts, "imported");
  TF_ImportGraphDefOptionsAddInputMapping(opts, "scalar", 0, {scalar, 0});
  TF_ImportGraphDefOptionsAddInputMapping(opts, "fake", 0, {scalar, 0});

  tensorflow::GraphDef graph_def_proto;
  ASSERT_TRUE(tensorflow::ParseProtoUnlimited(&graph_def_proto, graph_def->data,
                                              graph_def->length));

      r._elem.remove(testsuite._elem)
  if len(r) > 0:  # pylint: disable=g-explicit-length-test
    result += r

# Insert the number of failures for each test to help identify flakes
# need to clarify for shard
for p in result._elem.xpath(".//error | .//failure"):
  key = re.sub(r"0x\w+", "", p.getparent().get("name", "")) + p.text
  p.text = runfiles_matcher.sub("[testroot]/", p.text)

  std::unique_ptr<SingleMachine> cluster(new SingleMachine(5 * 60, 3, 0));
  TF_ASSERT_OK(cluster->Provision());

  TrivialTestGraphInputYielder fake_input(4, 1, 10, false,
                                          cluster->GetDeviceNames());
  GrapplerItem item;
  CHECK(fake_input.NextItem(&item));

  TF_Status* status = TF_NewStatus();
  TF_GraphProperties* graph_properties =

# "/path/bar". This script uses "gsutil rsync" instead, which acts on directory
# contents. About arguments to gsutil:
# "gsutil -m rsync" runs in parallel.
# "gsutil rsync -r" is recursive and makes directories work.
# "gsutil rsync -d" is "sync and delete files from destination if not present in source"

DOWNLOADS="$(mktemp -d)"
mkdir -p "$DOWNLOADS"
gsutil -m rsync -r "$TFCI_ARTIFACT_STAGING_GCS_URI" "$DOWNLOADS"

//       .Attr("N: int >= 2")
//       .Attr("T: type");
// that defines two inputs, "concat_dim" and "values" (in that order).
// You must use TF_AddInput() for the first input (since it takes a
// single tensor), and TF_AddInputList() for the second input (since
// it takes a list, even if you were to pass a list with a single
// tensor), as in:
//   TF_OperationDescription* desc = TF_NewOperation(graph, "Concat", "c");

//
// The given param (which may be null) is passed to work_func when the thread
// starts. In this way, data may be passed from the thread back to the caller.
//
// Caller takes ownership of the result and must call TF_JoinThread on it
// eventually.
TF_CAPI_EXPORT extern TF_Thread* TF_StartThread(const TF_ThreadOptions* options,
                                                const char* thread_name,

  // The devices operations run on.
  const std::vector<std::string>& underlying_devices() const {
    return underlying_devices_;
  }

  // Takes a description of a single operation being executed on the
  // ParallelDevice, and in turn runs one operation per component device with
  // its corresponding inputs from the input ParallelTensors. Wraps the