fixes default arg overwriting config value in load of Pipeline

dlt/helpers/airflow_helper.py

@@ -81,8 +81,7 @@ def __init__(
         The `data_folder` is available in certain Airflow deployments. In case of Composer, it is a location on the gcs bucket. `use_data_folder` is disabled and should be
         enabled only when needed. The operations on bucket are non-atomic and way slower than on local storage and should be avoided.
 
-        `abort_task_if_any_job_failed` will abort the other dlt loading jobs and fail the Airflow task in any of the jobs failed. This may put your warehouse in
-        inconsistent state. See https://dlthub.com/docs/running-in-production/running#handle-exceptions-failed-jobs-and-retry-the-pipeline.
+        `abort_task_if_any_job_failed` will abort the other dlt loading jobs and fail the Airflow task in any of the jobs failed. See https://dlthub.com/docs/running-in-production/running#handle-exceptions-failed-jobs-and-retry-the-pipeline.
 
         The load info and trace info can be optionally saved to the destination. See https://dlthub.com/docs/running-in-production/running#inspect-and-save-the-load-info-and-trace
 

dlt/pipeline/pipeline.py

@@ -550,7 +550,7 @@ def load(
         credentials: Any = None,
         *,
         workers: int = 20,
-        raise_on_failed_jobs: bool = False,
+        raise_on_failed_jobs: bool = ConfigValue,
     ) -> LoadInfo:
         """Loads the packages prepared by `normalize` method into the `dataset_name` at `destination`, optionally using provided `credentials`"""
         # set destination and default dataset if provided (this is the reason we have state sync here)

tests/pipeline/test_pipeline.py

@@ -832,7 +832,6 @@ def test_run_with_table_name_exceeding_path_length() -> None:
 
 def test_raise_on_failed_job() -> None:
     os.environ["FAIL_PROB"] = "1.0"
-    os.environ["RAISE_ON_FAILED_JOBS"] = "true"  # TODO: why is this necessary?
     pipeline_name = "pipe_" + uniq_id()
     p = dlt.pipeline(pipeline_name=pipeline_name, destination="dummy")
     with pytest.raises(PipelineStepFailed) as py_ex:
@@ -951,7 +950,6 @@ def fail_extract():
     assert py_ex.value.step == "extract"
 
     os.environ["COMPLETED_PROB"] = "0.0"
-    os.environ["RAISE_ON_FAILED_JOBS"] = "true"  # TODO: why is this necessary?
     os.environ["FAIL_PROB"] = "1.0"
     with pytest.raises(PipelineStepFailed) as py_ex:
         for attempt in Retrying(