The GitHub Actions job "Tests AMD" on airflow.git/serialization/op-defaults has 
failed.
Run started by GitHub user kaxil (triggered by kaxil).

Head commit for run:
966a2797d9bc975d23c3dbd480f7f315ef3f1e03 / Kaxil Naik <[email protected]>
Optimize `MappedOperator.operator_class` memory usage

Store only essential type information (task_type, _operator_name) in
MappedOperator.operator_class instead of full serialized operator data.

This addresses the TODO comment about storing unnecessary data and
provides significant memory savings for DAGs with many mapped operators:
- Before: ~60+ operator fields per mapped operator
- After: Only 2 essential strings per mapped operator
- Memory reduction: ~90-95% for this field

The optimization maintains all existing functionality:
- Operator extra links continue to work via task_type lookup
- Plugin matching works correctly
- All serialization/deserialization flows preserved

This completes the cleanup that should have happened when unmap()
functionality was removed - we no longer need full operator data
for scheduler-side mapped operators.

Report URL: https://github.com/apache/airflow/actions/runs/17145944595

With regards,
GitHub Actions via GitBox


---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to