-
Notifications
You must be signed in to change notification settings - Fork 103
Testing Data Transformation
Keith Babo edited this page May 5, 2015
·
1 revision
A list of things I normally test to validate a Data Transformation build:
- Start with the starter project and build it out to be identical to xml-to-json. This exercises all mapping types and also requires you to create a new transformation test using the Transformation Test wizard.
- Now test some other changes on the completed starter project:
- Remove mappings
- Add new empty mappings to recreated the ones you deleted. Drag to source/target and also set the
field using the detail view. - Add extra variables, change variable values, change variable mappings, remove variables.
- Use filters on model viewers
- Search for a field
- Set some different expression languages for a mapping
- Close the transformation editor and then retest all of the above bullets to make sure that there are no bugs with existing mapping configuration.
- Test validation logic:
- No mapping types or root nodes
- No mapping from scalar to vectors and vice versa
- Create a new transformation using the Blueprint config
- Create a new transformation test targeting the Blueprint config
- Try all combinations of JSON and XML schema/instance documents for types
- Create a new Other to Other mapping using the csv-to-xml project. You can reuse the existing model classes and data formats defined within that project.
- Test back and cancel within wizards and make sure there are no side effects.
- Test using a schema with multiple root elements to confirm element selection in new transformation wizard.