Closed fdmalone closed 1 month ago
Should we put a time limit so at least PRs greatly increasing the test time delta cause a failure.
Is excluding the slow tests? Running pytest on my machine gives:
=============================================================================================================== slowest 20 durations ================================================================================================================
44.36s call qualtran/bloqs/chemistry/chem_tutorials_test.py::test_resource_estimation_notebook
41.22s call qualtran/bloqs/hubbard_model_test.py::test_hubbard_model_notebook
11.77s call qualtran/cirq_interop/_bloq_to_cirq_test.py::test_notebook
11.47s call qualtran/bloqs/rotations/phase_gradient_test.py::test_add_scaled_val_into_phase_reg[bloq2]
11.35s call qualtran/bloqs/unary_iteration_bloq_test.py::test_notebook
11.19s call qualtran/bloqs/arithmetic/multiplication_test.py::test_multiplication_notebook
10.05s call qualtran/bloqs/generalized_qsp_test.py::test_real_polynomial_has_real_complementary_polynomial[100]
9.96s call qualtran/_infra/composite_bloq_test.py::test_notebook
9.79s call qualtran/bloqs/chemistry/thc/prepare_test.py::test_notebook
9.59s call qualtran/bloqs/chemistry/trotter/trotter_test.py::test_notebook
9.54s call qualtran/bloqs/arithmetic/addition_test.py::test_notebook
9.37s call qualtran/bloqs/chemistry/pbc/first_quantization/select_and_prepare_test.py::test_notebook
9.33s call qualtran/bloqs/swap_network_test.py::test_notebook
9.23s call qualtran/bloqs/chemistry/trotter/trotter_test.py::test_costing_notebook
9.20s call qualtran/_infra/bloq_test.py::test_notebook
8.31s call qualtran/bloqs/reflection_test.py::test_notebook
8.18s call qualtran/bloqs/util_bloqs_test.py::test_notebook
8.17s call qualtran/bloqs/apply_gate_to_lth_target_test.py::test_notebook
8.17s call qualtran/_infra/adjoint_test.py::test_notebook
8.14s call qualtran/bloqs/basic_gates_test.py::test_notebook
============================================================================================= 1473 passed, 267 skipped, 63 warnings in 69.70s (0:01:09) =============================================================================================
For the slow tests, I think I'm the culprit here with some recently added slow tests to verify the assumptions and error propagation in phasing via cost function infrastructure.
Seems to be taking around 17 minutes vs 10 minutes a few days ago.
cc @mpharrigan