permissions:
  contents: read

jobs:
  build:
    if: github.repository == 'tensorflow/tensorflow' # Don't do this in forks
    runs-on: [self-hosted, linux, ARM64]
    strategy:
      fail-fast: false
      matrix:
        pyver: ['3.9', '3.10', '3.11', '3.12']
    steps:
      - name: Stop old running containers (if any)
        shell: bash
        run: |

# This env is only valid when RBE is available on the selected platform, since
# it enables a derived --config setting.  If RBE is not available (i.e. there
# is no --config setting), bazel would fail and quit. This script does a quick
# check This script checks for such errors early
if ! grep "rbe_$TFCI_BAZEL_TARGET_SELECTING_CONFIG_PREFIX" .bazelrc; then
  cat <<EOF

  else:
    importlib.import_module("keras.src.optimizers")
except (ImportError, AttributeError):
  pass

del importlib

# Delete modules that should be hidden from dir().
# Don't fail if these modules are not available.
# For e.g. this file will be originally placed under tensorflow/_api/v1 which
# does not have "python", "core" directories. Then, it will be copied

      immediate_execution_ctx_.reset(ctx_raw);
    }

    // Computing numerical gradients with TensorFloat-32 is numerically
    // unstable. Some forward pass tests also fail with TensorFloat-32 due to
    // low tolerances
    enable_tensor_float_32_execution(false);
  }

  AbstractContextPtr immediate_execution_ctx_;
  GradientRegistry registry_;
  Status status_;

 public:

  if [[ -s $BATS_TEST_TMPDIR/actual_missing_licenses ]]; then
    echo "Please include the missing licenses for the following packages in $LICENSES_TARGET:"
    cat $BATS_TEST_TMPDIR/actual_missing_licenses
  fi

  # Fail if either of the two "extras" or "missing" lists are present. If so,
  # then the user will see the above error messages.
  [[ ! -s $BATS_TEST_TMPDIR/actual_extra_licenses ]] && [[ ! -s $BATS_TEST_TMPDIR/actual_missing_licenses ]]
}

// LINT.IfChange
// Note: Keep in sync with internal copy of enum in eager/context.h.
typedef enum TFE_ContextDevicePlacementPolicy {
  // Running operations with input tensors on the wrong device will fail.
  TFE_DEVICE_PLACEMENT_EXPLICIT = 0,
  // Copy the tensor to the right device but log a warning.
  TFE_DEVICE_PLACEMENT_WARN = 1,
  // Silently copy the tensor, which has a performance cost since the operation

  TF_AbstractOpSetOpName(placeholder_op, "my_ph", status.get());
  ASSERT_EQ(TF_OK, TF_GetCode(status.get())) << TF_Message(status.get());

  // This should fail.
  TF_AbstractOpSetOpType(placeholder_op, "Placeholder", status.get());
  ASSERT_EQ(TF_FAILED_PRECONDITION, TF_GetCode(status.get()));

  TF_DeleteAbstractOp(placeholder_op);
  TF_DeleteExecutionContext(graph_ctx);
}

If you are experiencing or witnessing conflict, we ask you to use the following escalation strategy to address the conflict:

1.  Address the perceived conflict directly with those involved, preferably in a
    real-time medium.
2.  If this fails, get a third party (e.g. a mutual friend, and/or someone with

    string prefix_cmp;
    const char* child_scope_name;
    if (prefix == nullptr) {
      child_scope_name = "gradients";
    } else {
      prefix_cmp = string(prefix) + "/";
      // The operation should fail if the provided name prefix has already been
      // used in this graph
      for (const auto& pair : g->name_map) {
        const string& name = pair.first;

    EXPECT_TRUE(strstr(TF_Message(status), msg) != nullptr)
        << TF_Message(status);
    // Since error is not cleared, the following copy with correct device will
    // still fail.
    TF_SetStatus(status, TF_OK, "");
    TFE_DeleteTensorHandle(retvals[0]);
    TFE_Executor* executor = TFE_ContextGetExecutorForThread(ctx);
    TFE_ExecutorWaitForAllPendingNodes(executor, status);