feat: Separate task processor database #68
Conversation
| @pytest.fixture | ||
| def task_processor_database_tasks( | ||
| mocker: MockerFixture, | ||
| settings: SettingsWrapper, | ||
| ) -> Generator[Callable[..., list[Task]], None, None]: | ||
| """ | ||
| Prepare a fake tasks processor database | ||
| """ |
There was a problem hiding this comment.
Good question. We'd need to have a separate test suite, in order to setup a separate database — or something else. I agree it would provide us an improved functional test, but it comes with some drawbacks. I figured we could get away the way it is because:
- The actual task fetching is tested against the default database.
- The effort of setting up a new database could lead us to test Django stuff.
Still, I'm not 100% convinced myself. WDYT?
There was a problem hiding this comment.
As discussed offline, refer to how analytics database is handled in the container entrypoint and tests:
https://github.com/Flagsmith/flagsmith/blob/7f0035e70b548414661c1230be9ab6057d3c454c/api/scripts/run-docker.sh#L56-L62
https://github.com/Flagsmith/flagsmith/blob/7f0035e70b548414661c1230be9ab6057d3c454c/.github/workflows/api-pull-request.yml#L59-L63
https://github.com/Flagsmith/flagsmith/blob/7f0035e70b548414661c1230be9ab6057d3c454c/api/tests/unit/app_analytics/test_analytics_db_service.py#L48-L53
emyller
force-pushed
the
feat/separate-task-processor-database
branch
from
81192cf to
0607086
Compare
Ref: Flagsmith/flagsmith#5136
WIP