- Group Reviewers
its definitely valuable to put in extra work in making sure the error messages good here
hm this is a bit specific - there are probably more likely errors of just referring to types that are not set up correctly or using a value on accident.
I think if we just do a better job with the error message here, we don't need to mention "did you mean Output" since you will see such clear context about failing to create an output definition.
If the handling is done in resolve_to_runtime_type it should also make good error messages for InputDefinition as well
other test cases to make sure we error effectively on:
i would still keep this test case around - you can use the match argument to pytest.raises and just match against a critical chunk of the error message to prevent having to update tests to do minor wording changes
I'm actually going to just abandon this, we can come back later - the real issue is that pandas.DataFrame raises an exception if you try to compare it to anything, which breaks a lot of our value-detection code. I gave a shot at fixing it but it's not obvious to me how to correct without introducing some really gross stuff