You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Our test suite is independent, each test can work on a empty db assuming tables were created.
It wasn't always so. Before tests were running fast but were unreliable. Now they are reliable but a bit slow.
We could parallelize the test execution with pytest-xdist. The only requirement is that each distributed worker needs it's own database. This means we need to switch database used during tests, so maybe some refactor is needed, see #247
See this comment about how approximately this should work. Test database should be created only once, and worker databases are created from main (empty) test db as a template. This will be fast. Of course after tests databases need to be cleaned up (note: we don't need posix ipc use as in the comment, just the approximate flow of db creation)
The text was updated successfully, but these errors were encountered:
Our test suite is independent, each test can work on a empty db assuming tables were created.
It wasn't always so. Before tests were running fast but were unreliable. Now they are reliable but a bit slow.
We could parallelize the test execution with
pytest-xdist
. The only requirement is that each distributed worker needs it's own database. This means we need to switch database used during tests, so maybe some refactor is needed, see #247See this comment about how approximately this should work. Test database should be created only once, and worker databases are created from main (empty) test db as a template. This will be fast. Of course after tests databases need to be cleaned up (note: we don't need posix ipc use as in the comment, just the approximate flow of db creation)
The text was updated successfully, but these errors were encountered: