ashb commented on issue #7508: [AIRFLOW-6886] Cleanup the SageMakerTrainingOperator result URL: https://github.com/apache/airflow/pull/7508#issuecomment-590295099 The reason is for consistency with the rest of the suite of Sagemaker operators: > I guess the CR author is only using SageMaker training operator. So this ‘Training’ field is weird and unnecessary to him. But if you look at other SageMaker operators, for example the batch transform one (https://github.com/apache/airflow/blob/1a9d218611502118132a93b0c12c1ad511cbdd6e/airflow/providers/amazon/aws/operators/sagemaker_transform.py#L117-L123) , it has two description results. So we definitely need a ‘Model’ field and a ‘Transform’ field here. And to make all SageMaker operators being consistent in the result format, I just had that ‘Training’ field although there’s only one description result. > > In this case I personally value consistency over cleanness (otherwise I won’t write it that way). So I don’t recommend to remove this ‘Training’ key although it’s useless. However if you guys think it’s better to remove it according to Airflow culture. I totally respect it. Just let Yuan Zha [Ed: member of the Sagemaker team at AWS, I have his email] knows this. He has more information on the impact of this change (whether customer’s workflow will be broken) and can stand for Amazon better than I can. What do you think @BasPH? I think I'm leaning to a -0 after this clarification.
---------------------------------------------------------------- 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. For queries about this service, please contact Infrastructure at: [email protected] With regards, Apache Git Services
