apache/beam · error · NotImplementedError

Execution of [ ] not implemented in runner .

Error message

Execution of [%s] not implemented in runner %s.

What it means

The DirectRunner's TransformEvaluatorRegistry walks the transform's MRO looking for a registered evaluator class. If none is found, get_evaluator raises NotImplementedError, meaning the DirectRunner has no local implementation for this PTransform type.

Solutions

1. Replace the transform with a composable one built from supported primitives (ParDo, GroupByKey, etc.).
2. Register a _TransformEvaluator for your transform class with the registry (subclass and extend _evaluators).
3. Run the pipeline on a runner that supports the transform.
4. Upgrade apache-beam — a newer DirectRunner may have added the evaluator.

Example fix

# before
output = pcoll | MyCustomPrimitive()
# after
output = pcoll | beam.ParDo(MyDoFn())  # composed of supported primitives

from apache_beam.runners.direct.transform_evaluator import _TransformEvaluatorRegistry
# assert your transform class (or an ancestor) has an evaluator before submitting

def is_direct_runner_supported(t) -> bool:
    from apache_beam.runners.direct import transform_evaluator as te
    return any(cls in te._TransformEvaluatorRegistry._evaluators for cls in type(t).__mro__)

try:
    result = pipeline.run()
except NotImplementedError as e:
    if 'not implemented in runner' in str(e):
        rewrite_transform_with_supported_primitives()

When it happens

Trigger: Applying a PTransform subclass (custom transform, external transform, or cloud-only primitive like a native Pub/Sub/BigQuery sink variant) to a pipeline executed with DirectRunner, where the transform class and all its bases are absent from _evaluators.

Common situations: Custom PTransform not registered with the DirectRunner; using cross-language/external transforms; transforms meant only for distributed runners; stale Beam version where a new primitive has no local evaluator.

Related errors

• DirectRunner: id_label is not supported for PubSub reads
• Expected a PTransform object, got
• Expecting a PCollection argument.
• PCollection not part of a pipeline.
• Root provider for [ ] not implemented in runner

AI-assisted analysis of apache/beam@12126d8942 (2026-09-13). Data as JSON: /api/errors/aea1768fb0610585. Report an issue: GitHub.

Appendix: source

Thrown at sdks/python/apache_beam/runners/direct/transform_evaluator.py:129

        core.PTransform: DefaultRootBundleProvider,
        _TestStream: _TestStreamRootBundleProvider,
    }

  def get_evaluator(
      self, applied_ptransform, input_committed_bundle, side_inputs):
    """Returns a TransformEvaluator suitable for processing given inputs."""
    assert applied_ptransform
    assert bool(applied_ptransform.side_inputs) == bool(side_inputs)

    # Walk up the class hierarchy to find an evaluable type. This is necessary
    # for supporting sub-classes of core transforms.
    for cls in applied_ptransform.transform.__class__.mro():
      evaluator = self._evaluators.get(cls)
      if evaluator:
        break

    if not evaluator:
      raise NotImplementedError(
          'Execution of [%s] not implemented in runner %s.' %
          (type(applied_ptransform.transform), self))
    return evaluator(
        self._evaluation_context,
        applied_ptransform,
        input_committed_bundle,
        side_inputs)

  def get_root_bundle_provider(self, applied_ptransform):
    provider_cls = None
    for cls in applied_ptransform.transform.__class__.mro():
      provider_cls = self._root_bundle_providers.get(cls)
      if provider_cls:
        break
    if not provider_cls:
      raise NotImplementedError(
          'Root provider for [%s] not implemented in runner %s' %
          (type(applied_ptransform.transform), self))

View on GitHub (pinned to 12126d8942)