Open jsstevenson opened 3 years ago
The above is fine, but I think the big one is restructuring tests of ETL classes to simply confirm that data is correctly being uploaded into the DB. Right now, we check this indirectly by having test cases use the query module to run searches, but that's really just an indirect test (more of an integration). Instead, we should be testing that ETL classes upload data in the correct way, and then, that given a particular test DB, the query class can look it up correctly.
In general, this library's test infrastructure costs a lot of time relative to how informative and helpful those tests are -- I think it would benefit greatly from rethinking how it is tested.
This issue is stale because it has been open 90 days with no activity. This issue will be closed if no further activity occurs in 14 days.
[ ] Move existing tests -> integration, run them consistently
[ ] Write new unit tests for internal functions
[ ] Cover main.py
[ ] Cover cli.py
[ ] Update default ignore options
[ ] Exclude tests in .coveragerc?
[ ] Pull more boilerplate test code out -> conftest.py
[ ] Restructure tests from record-oriented to match type-oriented (more in line with how feature development/breaking works)?