Problem
tests/ holds 17 files parked as *.py.disabled. Four of them pass unchanged
against the current code — 48 test cases that are written, working, and simply
switched off.
| File |
Cases |
Covers |
test_pelican_routes.py.disabled |
20 |
Pelican routes |
test_update_kafka_datasource.py.disabled |
14 |
Kafka datasource updates |
test_search_routes_get.py.disabled |
8 |
routes/search_routes/get.py |
test_metrics_organization.py.disabled |
6 |
Organization metrics |
This is not only lost tests, it is lost coverage on code that ships.
api/routes/search_routes/get.py currently sits at 0.00% coverage, and the
search router is mounted unconditionally — it is present in every deployment,
whatever the catalog backend. Its test file exists and passes; it is just not
being collected.
Two other low-coverage modules line up with disabled files the same way:
search_datasource_route at 12% and register_routes/post_general_dataset.py
at 26.5%.
Note on test_pelican_routes
Both tests/test_pelican_routes.py and tests/test_pelican_routes.py.disabled
exist. The two need comparing before the disabled one is restored, so the suite
does not end up with duplicate or conflicting cases.
Proposal
Rename the four files back to .py and let CI run them. No test code needs
writing.
Out of scope
The other 13 disabled files fail and need real work. They should be triaged
separately, grouped by area (Kafka, organizations, search, MinIO). Their
failures are currently hard to read because Pydantic V1-style deprecation
warnings from api/models/service_request_model.py and
api/models/update_service_model.py flood the output.
Acceptance criteria
Problem
tests/holds 17 files parked as*.py.disabled. Four of them pass unchangedagainst the current code — 48 test cases that are written, working, and simply
switched off.
test_pelican_routes.py.disabledtest_update_kafka_datasource.py.disabledtest_search_routes_get.py.disabledroutes/search_routes/get.pytest_metrics_organization.py.disabledThis is not only lost tests, it is lost coverage on code that ships.
api/routes/search_routes/get.pycurrently sits at 0.00% coverage, and thesearch router is mounted unconditionally — it is present in every deployment,
whatever the catalog backend. Its test file exists and passes; it is just not
being collected.
Two other low-coverage modules line up with disabled files the same way:
search_datasource_routeat 12% andregister_routes/post_general_dataset.pyat 26.5%.
Note on
test_pelican_routesBoth
tests/test_pelican_routes.pyandtests/test_pelican_routes.py.disabledexist. The two need comparing before the disabled one is restored, so the suite
does not end up with duplicate or conflicting cases.
Proposal
Rename the four files back to
.pyand let CI run them. No test code needswriting.
Out of scope
The other 13 disabled files fail and need real work. They should be triaged
separately, grouped by area (Kafka, organizations, search, MinIO). Their
failures are currently hard to read because Pydantic V1-style deprecation
warnings from
api/models/service_request_model.pyandapi/models/update_service_model.pyflood the output.Acceptance criteria
api/routes/search_routes/get.pyis no longer at 0% coverage.test_pelican_routessituation is resolved, with only onecopy left.