ProcessingInput takes Dataset Definition, but does not accept raw S3 definition

[ericangelokim]

sagemaker-python-sdk/src/sagemaker/processing.py

Line 991 in 01c6ee3

dataset_definition=None,

The new ProcessingInput implementation now takes dataset_definition as a parameter, but fails when S3 Dataset Definition is used:

ParamValidationError: Parameter validation failed:
Unknown parameter in ProcessingInputs[1].DatasetDefinition: "Typename", must be one of: AthenaDatasetDefinition, RedshiftDatasetDefinition, LocalPath, DataDistributionType, InputMode
Unknown parameter in ProcessingInputs[1].DatasetDefinition: "Datasetsourcetype", must be one of: AthenaDatasetDefinition, RedshiftDatasetDefinition, LocalPath, DataDistributionType, InputMode
Unknown parameter in ProcessingInputs[1].DatasetDefinition: "Name", must be one of: AthenaDatasetDefinition, RedshiftDatasetDefinition, LocalPath, DataDistributionType, InputMode
Unknown parameter in ProcessingInputs[1].DatasetDefinition: "S3Executioncontext", must be one of: AthenaDatasetDefinition, RedshiftDatasetDefinition, LocalPath, DataDistributionType, InputMode

This is inconvenient to use, as it requires different implementation for S3 inputs. Not entirely sure why this is implemented as such.

[marckarp]

If you want to use S3 as input to the Processing Job you should be setting s3_input ( S3Input).
If you are still facing an issue please provide a code snippet to reproduce the issue.