Anurag-S1ngh opened a new pull request, #41332: URL: https://github.com/apache/beam/pull/41332
## Summary `_BeamRowsToArrowTable` (used by `WriteToParquet` on a schema'd PCollection without an explicit Arrow schema) only implements `process_batch()`, but `BundleBasedDirectRunner` invoked `process()` once per element, so the base `DoFn.process()` raised `NotImplementedError` even though the in-process `FnApiRunner` executes the same pipeline cleanly. The direct runner now buffers bundle elements per (timestamp, windows, pane_info) and invokes `process_batch()` on homogeneous batches at the end of the bundle, which is how batched runners execute batched-only DoFns. Includes a `CHANGES.md` entry. Fixes #41078 ## Testing - New: `sdks/python/apache_beam/io/parquetio_test.py::TestParquet::test_schema_write_without_schema_bundle_based_direct_runner` (fails before, passes after) - New: `sdks/python/apache_beam/runners/direct/direct_runner_test.py::BundleBasedRunnerTest::test_process_batch_only_dofn` - `pytest sdks/python/apache_beam/io/parquetio_test.py -k "write or sink or schema"` (20 passed) - `pytest sdks/python/apache_beam/runners/direct/direct_runner_test.py` (8 passed) - `pytest sdks/python/apache_beam/io/parquetio_test.py::TestParquet::test_schema_read_write` (passed) -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
