What causes "append() was removed in Pandas 2.0." in apache/beam?

Replace append with pd.concat: pd.concat([s, to_append]) (works in both deferred and eager pandas).

apache/beam · error · WontImplementError

append() was removed in Pandas 2.0.

Error message

append() was removed in Pandas 2.0.

What it means

DeferredSeries.append is a deprecated wrapper: pandas removed Series.append in 2.0, so Beam's deferred implementation raises WontImplementError whenever the installed pandas is >= 2.0. The operation must be replaced with pandas.concat, which Beam DataFrames supports.

Solutions

Replace append with pd.concat: pd.concat([s, to_append]) (works in both deferred and eager pandas).
Pin pandas<2.0 only as a short-term stopgap.
Update code samples/docs that reference Series.append.

Example fix

// before
combined = s.append(to_append)
// after
combined = pd.concat([s, to_append])

Defensive patterns

Strategy: try-catch

Validate before calling

import pandas as pd
PD_VERSION = tuple(int(p) for p in pd.__version__.split('.')[:2])
if PD_VERSION >= (2, 0) and hasattr(s, 'append'):
    raise DeprecationWarning("Series.append removed in pandas 2.0; use pd.concat")

Try / catch

from apache_beam.dataframe import frame_base
try:
    combined = s.append(to_append)
except frame_base.WontImplementError:
    combined = pd.concat([s, to_append])

Prevention

Replace all Series.append calls with pd.concat before upgrading to pandas 2.x.
Grep the codebase for '.append(' on Series objects during pandas upgrades.
Prefer pd.concat even on pandas 1.x so the code is version-agnostic.

When it happens

Trigger: Calling s.append(other) on a DeferredSeries while pandas >= 2.0 is installed.

Common situations: Upgrading pandas from 1.x to 2.x in a Beam pipeline; following pre-2.0 pandas tutorials; requirements.txt floating to pandas 2.x.

Understand the failure class

Background: "is deprecated and will be removed" — deprecation warnings for old API names, keywords, and options, and how to migrate before the removal release — this error's family across 29 libraries.

Related errors

AI-assisted analysis of apache/beam@12126d8942 (2026-09-13). Data as JSON: /api/errors/8f9abd7c5ed45e16. Report an issue: GitHub.

Appendix: source

Thrown at sdks/python/apache_beam/dataframe/frames.py:1379

  @frame_base.with_docs_from(pd.Series)
  def keys(self):
    return self.index

  # Series.T == transpose. Both are a no-op
  T = frame_base._elementwise_method('T', base=pd.Series)
  transpose = frame_base._elementwise_method('transpose', base=pd.Series)
  shape = property(
      frame_base.wont_implement_method(
          pd.Series, 'shape', reason="non-deferred-result"))

  @frame_base.with_docs_from(pd.Series, removed_method=PD_VERSION >= (2, 0))
  @frame_base.args_to_kwargs(pd.Series, removed_method=PD_VERSION >= (2, 0))
  @frame_base.populate_defaults(pd.Series, removed_method=PD_VERSION >= (2, 0))
  def append(self, to_append, ignore_index, verify_integrity, **kwargs):
    """``ignore_index=True`` is not supported, because it requires generating an
    order-sensitive index."""
    if PD_VERSION >= (2, 0):
      raise frame_base.WontImplementError('append() was removed in Pandas 2.0.')
    if not isinstance(to_append, DeferredSeries):
      raise frame_base.WontImplementError(
          "append() only accepts DeferredSeries instances, received " +
          str(type(to_append)))
    if ignore_index:
      raise frame_base.WontImplementError(
          "append(ignore_index=True) is order sensitive because it requires "
          "generating a new index based on the order of the data.",
          reason="order-sensitive")

    if verify_integrity:
      # We can verify the index is non-unique within index partitioned data.
      requires = partitionings.Index()
    else:
      requires = partitionings.Arbitrary()

    return frame_base.DeferredFrame.wrap(
        expressions.ComputedExpression(

View on GitHub (pinned to 12126d8942)