The CI relies on the files in tests/test_data to accurately perform its function.
- I could switch to generating this data on the fly always but I think this would introduce other kinds of non-obvious behavior which might be more confusing in the long run. Moreover, relying exclusively on remote data could be problematic for a variety of reasons.
- At a minimum, I think a separate workflow job should validate if the
test_data matches what would be generated by the scripts. This would allow PRs which do not necessarily require that the test_data be updated still run through CI while still highlighting the fact that there are still fields or objects on the API unknown to the SDK.
The CI relies on the files in
tests/test_datato accurately perform its function.test_datamatches what would be generated by the scripts. This would allow PRs which do not necessarily require that thetest_databe updated still run through CI while still highlighting the fact that there are still fields or objects on the API unknown to the SDK.