[saket0187]

This PR replaces old tests with pytest , which will allow us to complete #7530 .

[petrasovaa]

@echoix it looks like the fixture runs twice although it's a module level fixture, can we prevent that?

2026-06-15T10:31:05.6217830Z ============================== slowest durations ===============================
2026-06-15T10:31:05.6219290Z 21.74s setup    temporal/t.vect.db.select/tests/t_vect_db_select_test.py::test_t_vect_db_select_interval_columns
2026-06-15T10:31:05.6221320Z 16.23s setup    temporal/t.vect.db.select/tests/t_vect_db_select_test.py::test_t_vect_db_select_interval_default

@saket0187 the setup runs too long, we might not be able to do much about that, but try lowering the number of maps to see if that makes at least little bit of difference.

[echoix]

Is there a fixture with the same name up in a parent contest.py?

From what I remember, a non negligible time is taken only to import the temporal Python modules (the file generated for ctypes), plus the init that is long. It ends up going through almost everything just to "start".

[saket0187]

I've updated the conftest.py fixture, lowering the generated maps from 6 to 3 and the random points from 5 to 3. This cut the total test execution time down to roughly 10–11 seconds on my end.