Process completed with exit code 1.

tests/saved/test_duration_saved.py::test_the_control_a_folder_with_no_duration_still_needs_one took 28.3 s (running 91%: un-jitted loop, heavy compute, or a subprocess). Mark it @pytest.mark.slow, make it faster, or add it to tests/duration_allowlist.txt with the reason it must run on every push.

Process completed with exit code 1.

shard 9 of 10: its job's status is 'failure'

Process completed with exit code 1.

default tests (Python 3.12, shard 9 of 10): its latest attempt concluded 'failure'

tests/oracles/test_mrs_family.py::test_the_mrs_rung_conforms_to_the_hydrodynamic_family_in_free_space_and_swaps_with_the_stokes_sphere[0.3.0] took 36.8 s (running 74%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/differential/test_a0_dynamic.py::test_a0_on_its_dynamic_rungs_matches_its_reference_with_exactly_the_second_declarations_differences[3hz] took 38.6 s (compiling 49%: 154 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_experiment_states.py::test_the_experiment_api_keeps_its_invariants_over_random_scripts took 50.5 s (running 75%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_dynamic_arm.py::test_control_the_arm_with_its_gravity_dropped_fails_the_balance_on_every_draw took 54.2 s (compiling 61%: 9 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_dynamic_arm.py::test_the_energy_balance_closes_to_the_integrators_order took 54.3 s (compiling 64%: 26 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_mrs_family.py::test_the_mrs_rung_conforms_to_the_hydrodynamic_family_in_free_space_and_swaps_with_the_stokes_sphere[0.4.0] took 1.4 min (running 83%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_dynamic_arm.py::test_control_a_stop_that_gives_energy_fails_the_balance_on_every_draw took 1.4 min (compiling 49%: 11 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/differential/test_a0_swap.py::test_a_step_of_lag_removed_or_added_inside_the_compiled_step_fails_the_swap[latency-doubled] took 1.4 min (running 69%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/differential/test_a0_swap.py::test_a0s_controller_runs_alike_inside_the_compiled_step_and_as_a_callback[one-step] took 1.5 min (running 70%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/differential/test_a0_swap.py::test_a0s_controller_runs_alike_inside_the_compiled_step_and_as_a_callback[no-latency] took 1.5 min (running 58%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/differential/test_a0_dynamic_controls.py::test_a_fault_planted_in_each_new_part_fails_the_comparison[the-joint-loops-gravity-term-minus] took 31.4 s (compiling 44%: 1 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/differential/test_a0_dynamic_controls.py::test_a_fault_planted_in_each_new_part_fails_the_comparison[the-joint-loops-gain-minus] took 32.8 s (compiling 47%: 1 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/differential/test_a0_dynamic_controls.py::test_a_fault_planted_in_each_new_part_fails_the_comparison[the-rotor-loops-gain-plus] took 34.7 s (compiling 46%: 1 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_node_walk.py::test_every_node_an_entry_builds_passes_maddenings_verification[observation toy.region_probe 0.1.0] took 34.8 s (running 70%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/differential/test_a0_dynamic_controls.py::test_a_fault_planted_in_each_new_part_fails_the_comparison[the-joint-loops-gain-plus] took 36.1 s (compiling 49%: 44 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/differential/test_a0_armature.py::test_at_the_stand_in_only_a_controller_that_sees_the_armature_holds_a_kick_past_the_limit took 40.9 s (compiling 42%: 1 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_tube_kind.py::test_the_confined_matrix_stays_conditioned_and_its_radius_derivative_is_right_on_the_current_sampling[sphere] took 1.1 min (compiling 87%: 272 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_tube_kind.py::test_the_confined_matrix_stays_conditioned_and_its_radius_derivative_is_right_on_the_current_sampling[capsule] took 1.2 min (compiling 88%: 343 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/differential/test_a0_armature.py::test_a0s_controller_holds_a_geared_wrist_inside_its_limit_and_the_bare_wrist_passes_it took 1.2 min (compiling 61%: 100 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_node_walk.py::test_every_node_an_entry_builds_passes_maddenings_verification[dynamics arm 0.1.0] took 1.4 min (running 44%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/scene/test_rotated_world.py::test_the_body_side_in_a_turned_world_runs_as_the_first_world_turned took 25.4 s (tracing/lowering 51%: large or repeatedly rebuilt program; no cache removes this). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/differential/test_a0_dynamic_b_controls.py::test_the_rotors_j_four_percent_large_fails_the_kick[3hz] took 26.0 s (compiling 58%: 1 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/differential/test_a0_dynamic_b_controls.py::test_the_rotor_loops_ki_fails_its_integral[3hz-minus] took 28.7 s (compiling 38%: 1 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_dynamic_motor.py::test_the_rotor_spins_up_to_its_closed_form_first_order_and_passes_maddenings_battery took 30.2 s (running 67%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/differential/test_a0_dynamic_b_controls.py::test_the_motors_angle_offset_at_its_start_fails_the_initial_angle[10hz] took 32.7 s (compiling 45%: 1 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/differential/test_a0_dynamic_b_controls.py::test_the_rotor_loops_ki_fails_its_integral[3hz-plus] took 34.6 s (compiling 43%: 1 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/differential/test_a0_dynamic_b_controls.py::test_the_motors_angle_offset_at_its_start_fails_the_initial_angle[3hz] took 38.2 s (compiling 50%: 215 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/live/test_a0_live.py::test_a0_through_mime_serve_is_never_ahead_of_real_time_and_each_frame_is_the_batch_runs_state took 51.9 s (running 46%: work in a subprocess (not measured here)). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_hull_contact_screws.py::test_an_anisotropic_screw_turns_broadside_and_is_held_at_the_bores_limit took 55.4 s (compiling 72%: 131 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/results/test_reduction_oracle.py::test_a_speed_that_is_not_linear_in_time_has_its_mean_as_its_series_mean_and_not_as_its_ends_midpoint took 1.3 min (compiling 56%: 147 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_registry_walk.py::test_every_cases_parameter_leaves_reach_its_compiled_step[dynamic_rpm] took 24.8 s (running 40%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_registry_walk.py::test_every_case_round_trips_through_a_fresh_process[tracked_rpm] took 25.4 s (running 37%: work in a subprocess (not measured here)). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_registry_walk.py::test_every_cases_parameter_leaves_reach_its_compiled_step[massive_rpm] took 26.6 s (compiling 46%: 1 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_registry_walk.py::test_every_case_round_trips_through_a_fresh_process[dynamic_rpm] took 44.4 s (compiling 52%: 12 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_registry_walk.py::test_every_cases_parameter_leaves_reach_its_compiled_step[hull_contact] took 45.2 s (compiling 79%: 1 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/scene/test_screw.py::test_the_screws_distance_is_the_brute_force_minimums took 45.4 s (compiling 57%: 208 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_registry_walk.py::test_every_case_round_trips_through_a_fresh_process[massive_rpm] took 49.9 s (compiling 60%: 2 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_registry_walk.py::test_every_case_round_trips_through_a_fresh_process[bound_robot_with_parts] took 1.1 min (compiling 51%: 4 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_registry_walk.py::test_every_kind_that_samples_its_surface_passes_the_kind_conformance_harness took 1.3 min (running 78%: work in a subprocess (not measured here)). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_registry_walk.py::test_every_case_round_trips_through_a_fresh_process[hull_contact] took 2.0 min (compiling 59%: 56 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_armature.py::test_the_checks_ratios_predict_each_joints_kick_over_six_armatures_and_its_controls_fail took 27.5 s (tracing/lowering 50%: large or repeatedly rebuilt program; no cache removes this). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_mrs_near_wall.py::test_the_integrated_image_is_its_sum_over_each_cell_written_again[every cell] took 27.5 s (compiling 57%: 201 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/scene/test_root_prims.py::test_the_stage_writer_emits_only_world_as_a_root_prim[script A0] took 35.9 s (compiling 46%: 2 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/differential/test_a0_controls.py::test_a_fault_planted_in_each_part_fails_the_comparison[the-gradient-force-minus] took 36.1 s (tracing/lowering 36%: large or repeatedly rebuilt program; no cache removes this). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/differential/test_a0_controls.py::test_a_fault_planted_in_each_part_fails_the_comparison[the-gradient-force-plus] took 39.4 s (compiling 49%: 1 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_golden_outputs.py::test_each_name_and_version_reproduces_its_golden_output[interaction contact 0.4.0] took 47.7 s (compiling 76%: 2 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/docs/test_reproducible.py::test_w8_mime_docs_check_fails_on_a_planted_hash_order_dependence_and_passes_without_it took 51.3 s (running 100%: work in a subprocess (not measured here)). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_mrs_near_wall.py::test_control_the_stop_removed_lets_the_indefinite_resistance_through_and_the_drag_puts_energy_in took 1.1 min (compiling 64%: 2 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_mrs_near_wall.py::test_script_as_floored_screw_builds_and_its_run_stops_at_the_first_indefinite_step_before_the_body_takes_it took 1.2 min (compiling 71%: 205 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_helix_reference_band.py::test_quasi_uniform_sampling_brings_the_frozen_point_inside_the_reference_band_and_the_quadrature_does_not took 3.1 min (running 95%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_mrs_near_wall.py::test_control_from_the_axis_near_wall_is_raised_only_where_the_body_comes_within_a_force_spacing took 30.5 s (tracing/lowering 56%: large or repeatedly rebuilt program; no cache removes this). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_armature.py::test_the_checks_ratios_predict_each_joints_kick_over_six_armatures_and_its_controls_fail took 32.9 s (tracing/lowering 61%: large or repeatedly rebuilt program; no cache removes this). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/scene/test_root_prims.py::test_the_stage_writer_emits_only_world_as_a_root_prim[script A0] took 34.1 s (compiling 42%: 2 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/differential/test_a0_controls.py::test_a_fault_planted_in_each_part_fails_the_comparison[the-gradient-force-minus] took 36.2 s (tracing/lowering 38%: large or repeatedly rebuilt program; no cache removes this). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/differential/test_a0_controls.py::test_a_fault_planted_in_each_part_fails_the_comparison[the-gradient-force-plus] took 40.4 s (compiling 42%: 1 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_golden_outputs.py::test_each_name_and_version_reproduces_its_golden_output[interaction contact 0.4.0] took 48.5 s (compiling 67%: 2 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/docs/test_reproducible.py::test_w8_mime_docs_check_fails_on_a_planted_hash_order_dependence_and_passes_without_it took 53.9 s (running 100%: work in a subprocess (not measured here)). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_mrs_near_wall.py::test_control_the_stop_removed_lets_the_indefinite_resistance_through_and_the_drag_puts_energy_in took 1.2 min (compiling 62%: 2 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_mrs_near_wall.py::test_script_as_floored_screw_builds_and_its_run_stops_at_the_first_indefinite_step_before_the_body_takes_it took 1.2 min (compiling 65%: 205 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_helix_reference_band.py::test_quasi_uniform_sampling_brings_the_frozen_point_inside_the_reference_band_and_the_quadrature_does_not took 2.9 min (running 95%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/differential/test_a0.py::test_the_emulation_reproduces_the_reference_and_is_the_declared_prediction[NEW] took 21.4 s (running 100%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/live/test_live_defaults.py::test_every_controllers_declared_defaults_reproduce_the_batch_run took 21.8 s (compiling 51%: 7 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/control/test_in_graph.py::test_a0s_rig_runs_alike_with_its_controller_inside_the_compiled_step_and_as_a_callback[feedback-off] took 22.9 s (compiling 53%: 2 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/differential/test_a0.py::test_a0_matches_its_reference_with_exactly_the_declared_differences[3hz] took 24.6 s (running 42%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/control/test_in_graph.py::test_a0s_rig_runs_alike_with_its_controller_inside_the_compiled_step_and_as_a_callback[feedback-on] took 27.0 s (compiling 55%: 2 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/cli/test_serve_terminal.py::test_a_scripted_session_on_a_terminal_shows_each_event_once_above_a_strip_that_follows_the_status[100] took 33.2 s (running 60%: work in a subprocess (not measured here)). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/control/test_in_graph.py::test_a_controller_inside_the_compiled_step_is_saved_loaded_and_verified took 33.9 s (tracing/lowering 48%: large or repeatedly rebuilt program; no cache removes this). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/families/test_family_walk.py::test_every_interaction_in_a_family_passes_its_suite[interaction drag 0.4.0] took 34.8 s (running 86%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/families/test_family_walk.py::test_every_interaction_in_a_family_passes_its_suite[interaction drag 0.3.0] took 42.3 s (running 74%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/differential/test_a0.py::test_the_emulation_reproduces_the_reference_and_is_the_declared_prediction[TAG] took 50.1 s (running 83%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/differential/test_a0_dynamic.py::test_a0_on_its_dynamic_rungs_matches_its_reference_with_exactly_the_second_declarations_differences[3hz] took 48.5 s (running 45%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_dynamic_arm.py::test_the_energy_balance_closes_to_the_integrators_order took 1.1 min (tracing/lowering 47%: large or repeatedly rebuilt program; no cache removes this). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_dynamic_arm.py::test_control_the_arm_with_its_gravity_dropped_fails_the_balance_on_every_draw took 1.2 min (compiling 44%: 9 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_experiment_states.py::test_the_experiment_api_keeps_its_invariants_over_random_scripts took 1.7 min (running 68%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_mrs_family.py::test_the_mrs_rung_conforms_to_the_hydrodynamic_family_in_free_space_and_swaps_with_the_stokes_sphere[0.4.0] took 1.7 min (running 84%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/differential/test_a0_swap.py::test_a0s_controller_runs_alike_inside_the_compiled_step_and_as_a_callback[one-step] took 1.9 min (running 71%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/differential/test_a0_swap.py::test_a0s_controller_runs_alike_inside_the_compiled_step_and_as_a_callback[no-latency] took 1.9 min (running 65%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/differential/test_a0_swap.py::test_a_step_of_lag_removed_or_added_inside_the_compiled_step_fails_the_swap[latency-doubled] took 2.0 min (running 70%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_dynamic_arm.py::test_control_a_stop_that_gives_energy_fails_the_balance_on_every_draw took 2.5 min (tracing/lowering 44%: large or repeatedly rebuilt program; no cache removes this). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/sharing/test_notice_card.py::test_every_file_the_folder_names_is_in_the_notice took 21.4 s here and 4.0 s on a re-run alone (limit 20 s): one runner's outlier, not a slow test. If it repeats, look at the runner and at what the test shares with its neighbours.

tests/test_collection_guard.py::test_this_suites_own_run_is_guarded_the_hazard_order_over_tests_saved_is_refused_and_the_same_files_in_order_are_not took 6.4 s (running 100%: work in a subprocess (not measured here)); the budget for the default lane is 5 s. Mark it slow or make it faster.

tests/oracles/test_arm_dynamics_battery.py::test_the_dynamic_arm_passes_maddenings_check[finite] took 26.2 s (compiling 46%: 78 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_arm_dynamics_battery.py::test_the_dynamic_arm_passes_maddenings_check[deterministic] took 29.9 s (compiling 36%: 1 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_arm_dynamics_battery.py::test_the_dynamic_arm_is_first_order_in_time took 46.0 s (compiling 64%: 123 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_arm_dynamics_battery.py::test_the_dynamic_arm_passes_maddenings_check[gradient_finite] took 3.4 min (running 53%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_arm_dynamics_battery.py::test_the_dynamic_arm_passes_maddenings_check[params_gradient_finite] took 3.5 min (running 50%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_arm_dynamics_battery.py::test_the_dynamic_arm_passes_maddenings_check[params_effective] took 7.3 min (tracing/lowering 38%: large or repeatedly rebuilt program; no cache removes this). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_registry_walk.py::test_every_cases_parameter_leaves_reach_its_compiled_step[dynamic_rpm] took 30.0 s (running 44%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_registry_walk.py::test_every_case_round_trips_through_a_fresh_process[tracked_rpm] took 31.1 s (running 38%: work in a subprocess (not measured here)). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_registry_walk.py::test_every_cases_parameter_leaves_reach_its_compiled_step[massive_rpm] took 31.3 s (compiling 39%: 1 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/scene/test_screw.py::test_the_screws_distance_is_the_brute_force_minimums took 51.5 s (compiling 47%: 210 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_registry_walk.py::test_every_cases_parameter_leaves_reach_its_compiled_step[hull_contact] took 53.8 s (compiling 63%: 1 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_registry_walk.py::test_every_case_round_trips_through_a_fresh_process[dynamic_rpm] took 54.0 s (compiling 39%: 12 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_registry_walk.py::test_every_case_round_trips_through_a_fresh_process[massive_rpm] took 59.3 s (compiling 46%: 2 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_registry_walk.py::test_every_case_round_trips_through_a_fresh_process[bound_robot_with_parts] took 1.2 min (compiling 42%: 4 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_registry_walk.py::test_every_kind_that_samples_its_surface_passes_the_kind_conformance_harness took 1.6 min (running 76%: work in a subprocess (not measured here)). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_registry_walk.py::test_every_case_round_trips_through_a_fresh_process[hull_contact] took 2.3 min (compiling 51%: 56 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_spinner_tree.py::test_on_a_joint_along_the_axis_the_spinner_converges_to_the_jointed_rotor_and_its_planted_faults_do_not took 36.2 s (tracing/lowering 40%: large or repeatedly rebuilt program; no cache removes this). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_hull_contact_support.py::test_control_a_search_that_ends_shallower_than_its_seed_raises_support_short took 39.1 s (compiling 71%: 1 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_spinner.py::test_control_the_forward_euler_coupling_grows_the_energy_on_every_draw took 41.6 s (running 93%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_spinner.py::test_the_gyroscopic_coupling_does_no_work_over_ten_thousand_steps_at_every_step_size took 43.1 s (running 84%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_spinner_tree.py::test_on_a_gimbal_the_spinner_converges_to_the_jointed_rotor_and_its_gyroscopic_sign_flipped_does_not took 43.8 s (compiling 48%: 9 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_hull_contact_support.py::test_a_body_at_rest_on_the_floor_neither_gains_energy_nor_wanders[fl9] took 47.7 s (compiling 55%: 56 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_hull_contact_support.py::test_the_support_never_leaves_a_held_point_shallower_than_its_seed took 56.2 s (compiling 59%: 105 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_hull_contact_support.py::test_a_healthy_run_stays_under_the_flag took 1.1 min (compiling 52%: 2 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_hull_contact_support.py::test_the_projection_after_the_step_leaves_the_kinetic_energy_as_the_solve_left_it took 2.1 min (compiling 72%: 94 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_hull_contact_support.py::test_the_cut_candidates_choose_the_step_all_the_points_choose_to_the_last_bit took 3.8 min (compiling 71%: 9 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/differential/test_a0_dynamic.py::test_a0_on_its_dynamic_rungs_matches_its_reference_with_exactly_the_second_declarations_differences[3hz] took 54.6 s (compiling 40%: 158 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_mrs_family.py::test_the_mrs_rung_conforms_to_the_hydrodynamic_family_in_free_space_and_swaps_with_the_stokes_sphere[0.3.0] took 57.4 s (running 79%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_dynamic_arm.py::test_the_energy_balance_closes_to_the_integrators_order took 1.5 min (tracing/lowering 53%: large or repeatedly rebuilt program; no cache removes this). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_dynamic_arm.py::test_control_the_arm_with_its_gravity_dropped_fails_the_balance_on_every_draw took 1.6 min (tracing/lowering 46%: large or repeatedly rebuilt program; no cache removes this). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_mrs_family.py::test_the_mrs_rung_conforms_to_the_hydrodynamic_family_in_free_space_and_swaps_with_the_stokes_sphere[0.4.0] took 2.1 min (running 85%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/differential/test_a0_swap.py::test_a0s_controller_runs_alike_inside_the_compiled_step_and_as_a_callback[one-step] took 2.5 min (running 71%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/differential/test_a0_swap.py::test_a0s_controller_runs_alike_inside_the_compiled_step_and_as_a_callback[no-latency] took 2.5 min (running 64%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/differential/test_a0_swap.py::test_a_step_of_lag_removed_or_added_inside_the_compiled_step_fails_the_swap[latency-doubled] took 2.9 min (running 72%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_dynamic_arm.py::test_control_a_stop_that_gives_energy_fails_the_balance_on_every_draw took 3.3 min (tracing/lowering 48%: large or repeatedly rebuilt program; no cache removes this). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/sharing/test_notice_card.py::test_every_file_the_folder_names_is_in_the_notice took 30.8 s here and 4.5 s on a re-run alone (limit 20 s): one runner's outlier, not a slow test. If it repeats, look at the runner and at what the test shares with its neighbours.

tests/test_collection_guard.py::test_this_suites_own_run_is_guarded_the_hazard_order_over_tests_saved_is_refused_and_the_same_files_in_order_are_not took 7.3 s (running 100%: work in a subprocess (not measured here)); the budget for the default lane is 5 s. Mark it slow or make it faster.

tests/oracles/test_arm_dynamics_battery.py::test_the_dynamic_arm_passes_maddenings_check[jit_consistent] took 20.3 s (compiling 44%: 1 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_arm_dynamics_battery.py::test_the_dynamic_arm_passes_maddenings_check[finite] took 30.0 s (compiling 44%: 70 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_arm_dynamics_battery.py::test_the_dynamic_arm_is_first_order_in_time took 53.7 s (compiling 62%: 109 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_arm_dynamics_battery.py::test_the_dynamic_arm_passes_maddenings_check[gradient_finite] took 3.7 min (compiling 52%: 28 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_arm_dynamics_battery.py::test_the_dynamic_arm_passes_maddenings_check[params_gradient_finite] took 4.1 min (compiling 52%: 17 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_arm_dynamics_battery.py::test_the_dynamic_arm_passes_maddenings_check[params_effective] took 8.5 min (tracing/lowering 38%: large or repeatedly rebuilt program; no cache removes this). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/cli/test_serve_terminal.py::test_a_terminal_nobody_reads_does_not_keep_serve_from_writing_the_run_folder_and_exiting[max-wall] took 23.3 s (running 100%: work in a subprocess (not measured here)). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/families/test_family_walk.py::test_every_interaction_in_a_family_passes_its_suite[interaction magnetic.force 0.1.0] took 24.4 s (running 94%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/differential/test_a0.py::test_a0_matches_its_reference_with_exactly_the_declared_differences[3hz] took 29.1 s (running 42%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/control/test_in_graph.py::test_a0s_rig_runs_alike_with_its_controller_inside_the_compiled_step_and_as_a_callback[feedback-off] took 32.4 s (compiling 50%: 2 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/control/test_in_graph.py::test_a0s_rig_runs_alike_with_its_controller_inside_the_compiled_step_and_as_a_callback[feedback-on] took 33.1 s (compiling 50%: 2 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/cli/test_serve_terminal.py::test_a_scripted_session_on_a_terminal_shows_each_event_once_above_a_strip_that_follows_the_status[100] took 41.6 s (running 58%: work in a subprocess (not measured here)). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/control/test_in_graph.py::test_a_controller_inside_the_compiled_step_is_saved_loaded_and_verified took 44.2 s (tracing/lowering 50%: large or repeatedly rebuilt program; no cache removes this). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/families/test_family_walk.py::test_every_interaction_in_a_family_passes_its_suite[interaction drag 0.4.0] took 51.3 s (running 95%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/differential/test_a0.py::test_the_emulation_reproduces_the_reference_and_is_the_declared_prediction[TAG] took 55.5 s (running 85%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/families/test_family_walk.py::test_every_interaction_in_a_family_passes_its_suite[interaction drag 0.3.0] took 1.0 min (running 82%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/differential/test_a0_dynamic_controls.py::test_a_fault_planted_in_each_new_part_fails_the_comparison[the-rotor-loops-gain-minus] took 48.0 s (compiling 43%: 1 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/differential/test_a0_dynamic_controls.py::test_a_fault_planted_in_each_new_part_fails_the_comparison[the-joint-loops-gain-minus] took 49.0 s (compiling 44%: 1 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/differential/test_a0_dynamic_controls.py::test_a_fault_planted_in_each_new_part_fails_the_comparison[the-joint-loops-gravity-term-minus] took 49.3 s (compiling 43%: 1 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/differential/test_a0_dynamic_controls.py::test_a_fault_planted_in_each_new_part_fails_the_comparison[the-rotor-loops-gain-plus] took 50.1 s (compiling 44%: 1 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_node_walk.py::test_every_node_an_entry_builds_passes_maddenings_verification[observation toy.region_probe 0.1.0] took 56.7 s (running 77%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/differential/test_a0_armature.py::test_at_the_stand_in_only_a_controller_that_sees_the_armature_holds_a_kick_past_the_limit took 57.0 s (tracing/lowering 43%: large or repeatedly rebuilt program; no cache removes this). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_tube_kind.py::test_the_confined_matrix_stays_conditioned_and_its_radius_derivative_is_right_on_the_current_sampling[capsule] took 1.4 min (compiling 88%: 343 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/differential/test_a0_armature.py::test_a0s_controller_holds_a_geared_wrist_inside_its_limit_and_the_bare_wrist_passes_it took 1.4 min (compiling 55%: 100 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_tube_kind.py::test_the_confined_matrix_stays_conditioned_and_its_radius_derivative_is_right_on_the_current_sampling[sphere] took 1.5 min (compiling 84%: 272 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_node_walk.py::test_every_node_an_entry_builds_passes_maddenings_verification[dynamics arm 0.1.0] took 1.9 min (running 46%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_hull_contact_support.py::test_control_a_flag_set_below_that_run_rises_on_it took 42.9 s (running 61%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_spinner_tree.py::test_on_a_gimbal_the_spinner_converges_to_the_jointed_rotor_and_its_gyroscopic_sign_flipped_does_not took 43.7 s (compiling 57%: 9 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_hull_contact_support.py::test_control_a_search_that_ends_shallower_than_its_seed_raises_support_short took 45.4 s (compiling 83%: 1 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_hull_contact_support.py::test_a_body_at_rest_on_the_floor_neither_gains_energy_nor_wanders[fl9] took 51.1 s (compiling 63%: 56 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_spinner.py::test_the_gyroscopic_coupling_does_no_work_over_ten_thousand_steps_at_every_step_size took 52.9 s (running 92%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_spinner.py::test_control_the_forward_euler_coupling_grows_the_energy_on_every_draw took 56.0 s (running 95%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_hull_contact_support.py::test_the_support_never_leaves_a_held_point_shallower_than_its_seed took 1.0 min (compiling 65%: 105 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_hull_contact_support.py::test_a_healthy_run_stays_under_the_flag took 1.2 min (compiling 51%: 2 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_hull_contact_support.py::test_the_projection_after_the_step_leaves_the_kinetic_energy_as_the_solve_left_it took 2.2 min (compiling 82%: 49 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_hull_contact_support.py::test_the_cut_candidates_choose_the_step_all_the_points_choose_to_the_last_bit took 4.3 min (compiling 85%: 54 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_mrs_family.py::test_the_mrs_rung_conforms_to_the_hydrodynamic_family_in_free_space_and_swaps_with_the_stokes_sphere[0.3.0] took 1.1 min (running 78%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_dynamic_arm.py::test_the_energy_balance_closes_to_the_integrators_order took 1.8 min (tracing/lowering 49%: large or repeatedly rebuilt program; no cache removes this). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_dynamic_arm.py::test_control_the_arm_with_its_gravity_dropped_fails_the_balance_on_every_draw took 1.9 min (tracing/lowering 46%: large or repeatedly rebuilt program; no cache removes this). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_experiment_states.py::test_the_experiment_api_keeps_its_invariants_over_random_scripts took 2.3 min (running 72%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_mrs_family.py::test_the_mrs_rung_conforms_to_the_hydrodynamic_family_in_free_space_and_swaps_with_the_stokes_sphere[0.4.0] took 2.5 min (running 86%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/differential/test_a0_swap.py::test_a0s_controller_runs_alike_inside_the_compiled_step_and_as_a_callback[no-latency] took 2.8 min (running 64%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/differential/test_a0_swap.py::test_a_step_of_lag_removed_or_added_inside_the_compiled_step_fails_the_swap[latency-doubled] took 2.9 min (running 70%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/differential/test_a0_swap.py::test_a0s_controller_runs_alike_inside_the_compiled_step_and_as_a_callback[one-step] took 3.0 min (running 71%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_dynamic_arm.py::test_control_a_stop_that_gives_energy_fails_the_balance_on_every_draw took 3.2 min (tracing/lowering 48%: large or repeatedly rebuilt program; no cache removes this). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/sharing/test_notice_card.py::test_every_file_the_folder_names_is_in_the_notice took 36.3 s here and 5.7 s on a re-run alone (limit 20 s): one runner's outlier, not a slow test. If it repeats, look at the runner and at what the test shares with its neighbours.

tests/oracles/test_hull_contact.py::test_the_held_points_derivatives_are_the_implicit_functions[screw] took 26.8 s (running 40%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_hull_contact.py::test_the_couples_forces_move_with_the_torque_as_its_closed_form took 30.3 s (compiling 85%: 1 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_hull_contact.py::test_control_from_its_nearest_candidate_alone_the_counterexample_passes_the_margin took 31.4 s (compiling 84%: 1 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_hull_contact.py::test_where_no_point_reaches_the_wall_the_step_is_the_free_bodys_to_the_last_bit took 32.9 s (cache read 47%: reading compiled programs back; a warm run pays this). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_hull_contact.py::test_control_its_search_left_at_five_steps_and_held_where_they_end_turns_the_step_otherwise took 36.3 s (compiling 68%: 2 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_hull_contact.py::test_control_a_hull_held_by_its_samples_lets_its_body_past_the_margin took 37.9 s (running 84%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_hull_contact.py::test_control_without_its_rotation_coupled_in_the_couples_forces_do_not took 46.5 s (compiling 82%: 22 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_hull_contact.py::test_the_contact_never_leaves_the_step_more_energy_than_the_free_step took 57.4 s (compiling 81%: 4 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_hull_contact.py::test_no_hull_point_ends_past_the_wall_and_the_margin_is_kept_to_the_sweeps_convergence took 2.3 min (compiling 85%: 371 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_hull_contact.py::test_turning_the_tube_and_the_body_together_turns_the_step took 3.3 min (compiling 67%: 8 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/differential/test_a0_dynamic_b_controls.py::test_the_rotor_loops_ki_fails_its_integral[10hz-minus] took 1.1 min (compiling 47%: 1 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/differential/test_a0_dynamic_b_controls.py::test_the_rotor_loops_ki_fails_its_integral[10hz-plus] took 1.1 min (compiling 40%: 1 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/differential/test_a0_dynamic_b_controls.py::test_the_rotor_loops_ki_fails_its_integral[3hz-minus] took 1.1 min (compiling 41%: 1 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/differential/test_a0_dynamic_b_controls.py::test_the_rotor_loops_ki_fails_its_integral[3hz-plus] took 1.1 min (running 41%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_dynamic_motor.py::test_the_rotor_spins_up_to_its_closed_form_first_order_and_passes_maddenings_battery took 1.2 min (running 75%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/differential/test_a0_dynamic_b_controls.py::test_the_motors_angle_offset_at_its_start_fails_the_initial_angle[10hz] took 1.2 min (running 42%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/differential/test_a0_dynamic_b_controls.py::test_the_motors_angle_offset_at_its_start_fails_the_initial_angle[3hz] took 1.3 min (compiling 38%: 209 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_hull_contact_screws.py::test_an_anisotropic_screw_turns_broadside_and_is_held_at_the_bores_limit took 1.5 min (compiling 58%: 131 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/live/test_a0_live.py::test_a0_through_mime_serve_is_never_ahead_of_real_time_and_each_frame_is_the_batch_runs_state took 1.7 min (running 48%: work in a subprocess (not measured here)). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/results/test_reduction_oracle.py::test_a_speed_that_is_not_linear_in_time_has_its_mean_as_its_series_mean_and_not_as_its_ends_midpoint took 2.6 min (compiling 52%: 152 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_link_loads.py::test_loads_on_the_frames_enter_inverse_dynamics_as_the_jacobian_transpose_maps_them took 34.7 s (compiling 60%: 12 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_sampling_walk.py::test_every_kind_that_samples_spreads_its_samples_evenly[cylinder-2000] took 39.6 s (running 81%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/live/test_serve.py::test_every_frame_is_delivered_in_order_and_is_the_batch_runs_state took 49.4 s (running 54%: work in a subprocess (not measured here)). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_profile_record.py::test_t1_bare_saves_a_folder_that_loads_pends_and_verifies took 54.4 s (running 84%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_sampling_walk.py::test_control_every_kinds_own_quadrature_fails_the_spacing_check took 1.1 min (compiling 73%: 1470 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_switch_harness.py::test_every_guard_the_scan_lists_is_reached_false_and_every_nonzero_norm_away_from_zero took 1.7 min (tracing/lowering 55%: large or repeatedly rebuilt program; no cache removes this). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_switch_scan.py::test_the_tree_passes_the_scan_with_its_list_and_its_owners took 1.8 min (tracing/lowering 59%: large or repeatedly rebuilt program; no cache removes this). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_switch_scan.py::test_every_owner_passed_is_one_a_collector_records_in_a_fresh_trace_of_the_steps took 1.8 min (tracing/lowering 67%: large or repeatedly rebuilt program; no cache removes this). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_switch_harness.py::test_replacing_every_fallback_changes_nothing_where_its_condition_is_false[the hull contact on FL-9 and an ellipsoid at the wall] took 2.3 min (running 64%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_switch_scan.py::test_an_owner_passed_by_position_to_a_helper_fails took 2.4 min (tracing/lowering 46%: large or repeatedly rebuilt program; no cache removes this). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_link_loads.py::test_loads_on_the_frames_enter_inverse_dynamics_as_the_jacobian_transpose_maps_them took 36.1 s (compiling 60%: 12 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_sampling_walk.py::test_every_kind_that_samples_spreads_its_samples_evenly[cylinder-2000] took 40.2 s (running 80%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/live/test_serve.py::test_every_frame_is_delivered_in_order_and_is_the_batch_runs_state took 51.8 s (running 54%: work in a subprocess (not measured here)). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_profile_record.py::test_t1_bare_saves_a_folder_that_loads_pends_and_verifies took 54.2 s (running 88%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_switch_scan.py::test_every_owner_passed_is_one_a_collector_records_in_a_fresh_trace_of_the_steps took 1.0 min (tracing/lowering 75%: large or repeatedly rebuilt program; no cache removes this). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_sampling_walk.py::test_control_every_kinds_own_quadrature_fails_the_spacing_check took 1.2 min (compiling 73%: 1470 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_switch_harness.py::test_every_guard_the_scan_lists_is_reached_false_and_every_nonzero_norm_away_from_zero took 1.7 min (tracing/lowering 54%: large or repeatedly rebuilt program; no cache removes this). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_switch_scan.py::test_the_tree_passes_the_scan_with_its_list_and_its_owners took 1.9 min (tracing/lowering 56%: large or repeatedly rebuilt program; no cache removes this). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_switch_harness.py::test_replacing_every_fallback_changes_nothing_where_its_condition_is_false[the hull contact on FL-9 and an ellipsoid at the wall] took 2.3 min (running 65%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_switch_scan.py::test_an_owner_passed_by_position_to_a_helper_fails took 2.5 min (tracing/lowering 46%: large or repeatedly rebuilt program; no cache removes this). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_sampling_walk.py::test_every_kind_that_samples_spreads_its_samples_evenly[box-2000] took 23.5 s (running 69%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_sampling_walk.py::test_every_kind_that_samples_spreads_its_samples_evenly[capsule-2000] took 24.7 s (running 64%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_sampling_walk.py::test_every_kind_that_samples_spreads_its_samples_evenly[cylinder-2000] took 26.5 s (running 80%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_profile_record.py::test_t1_bare_saves_a_folder_that_loads_pends_and_verifies took 30.1 s (running 90%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/live/test_serve.py::test_every_frame_is_delivered_in_order_and_is_the_batch_runs_state took 31.5 s (running 56%: work in a subprocess (not measured here)). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_switch_scan.py::test_every_owner_passed_is_one_a_collector_records_in_a_fresh_trace_of_the_steps took 36.4 s (tracing/lowering 66%: large or repeatedly rebuilt program; no cache removes this). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_sampling_walk.py::test_control_every_kinds_own_quadrature_fails_the_spacing_check took 48.2 s (compiling 74%: 1429 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_switch_harness.py::test_every_guard_the_scan_lists_is_reached_false_and_every_nonzero_norm_away_from_zero took 1.0 min (tracing/lowering 45%: large or repeatedly rebuilt program; no cache removes this). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_switch_scan.py::test_an_owner_passed_by_position_to_a_helper_fails took 1.4 min (tracing/lowering 36%: large or repeatedly rebuilt program; no cache removes this). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_switch_harness.py::test_replacing_every_fallback_changes_nothing_where_its_condition_is_false[the hull contact on FL-9 and an ellipsoid at the wall] took 1.4 min (running 62%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/differential/test_a0_dynamic_b_controls.py::test_the_rotors_j_four_percent_large_fails_the_kick[3hz] took 57.5 s (compiling 45%: 1 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/differential/test_a0_dynamic_b_controls.py::test_a_downstream_drift_fails_the_fields_turning_against_the_angle[3hz] took 57.8 s (running 55%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/scene/test_rotated_world.py::test_the_body_side_in_a_turned_world_runs_as_the_first_world_turned took 58.2 s (tracing/lowering 48%: large or repeatedly rebuilt program; no cache removes this). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/differential/test_a0_dynamic_b_controls.py::test_the_rotor_loops_ki_fails_its_integral[3hz-plus] took 1.1 min (running 39%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/differential/test_a0_dynamic_b_controls.py::test_the_motors_angle_offset_at_its_start_fails_the_initial_angle[10hz] took 1.2 min (running 36%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_dynamic_motor.py::test_the_rotor_spins_up_to_its_closed_form_first_order_and_passes_maddenings_battery took 1.2 min (running 73%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/differential/test_a0_dynamic_b_controls.py::test_the_motors_angle_offset_at_its_start_fails_the_initial_angle[3hz] took 1.3 min (compiling 36%: 202 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_hull_contact_screws.py::test_an_anisotropic_screw_turns_broadside_and_is_held_at_the_bores_limit took 1.6 min (compiling 56%: 131 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/live/test_a0_live.py::test_a0_through_mime_serve_is_never_ahead_of_real_time_and_each_frame_is_the_batch_runs_state took 1.8 min (running 48%: work in a subprocess (not measured here)). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/results/test_reduction_oracle.py::test_a_speed_that_is_not_linear_in_time_has_its_mean_as_its_series_mean_and_not_as_its_ends_midpoint took 2.6 min (compiling 51%: 152 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/scene/test_rotated_world.py::test_the_body_side_in_a_turned_world_runs_as_the_first_world_turned took 56.5 s (tracing/lowering 51%: large or repeatedly rebuilt program; no cache removes this). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/differential/test_a0_dynamic_b_controls.py::test_the_rotor_loops_ki_fails_its_integral[3hz-minus] took 57.9 s (compiling 40%: 1 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/differential/test_a0_dynamic_b_controls.py::test_the_rotors_j_four_percent_large_fails_the_kick[10hz] took 1.0 min (compiling 44%: 1 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/differential/test_a0_dynamic_b_controls.py::test_the_rotor_loops_ki_fails_its_integral[3hz-plus] took 1.1 min (running 39%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_dynamic_motor.py::test_the_rotor_spins_up_to_its_closed_form_first_order_and_passes_maddenings_battery took 1.2 min (running 74%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/differential/test_a0_dynamic_b_controls.py::test_the_motors_angle_offset_at_its_start_fails_the_initial_angle[10hz] took 1.2 min (running 40%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/differential/test_a0_dynamic_b_controls.py::test_the_motors_angle_offset_at_its_start_fails_the_initial_angle[3hz] took 1.3 min (compiling 36%: 196 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_hull_contact_screws.py::test_an_anisotropic_screw_turns_broadside_and_is_held_at_the_bores_limit took 1.6 min (compiling 53%: 131 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/live/test_a0_live.py::test_a0_through_mime_serve_is_never_ahead_of_real_time_and_each_frame_is_the_batch_runs_state took 1.8 min (running 49%: work in a subprocess (not measured here)). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/results/test_reduction_oracle.py::test_a_speed_that_is_not_linear_in_time_has_its_mean_as_its_series_mean_and_not_as_its_ends_midpoint took 2.6 min (compiling 48%: 150 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/differential/test_a0_dynamic_controls.py::test_a_fault_planted_in_each_new_part_fails_the_comparison[the-joint-loops-gravity-term-plus] took 1.1 min (running 39%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/differential/test_a0_dynamic_controls.py::test_a_fault_planted_in_each_new_part_fails_the_comparison[the-joint-loops-gain-minus] took 1.1 min (running 39%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/differential/test_a0_dynamic_controls.py::test_a_fault_planted_in_each_new_part_fails_the_comparison[the-joint-loops-gravity-term-minus] took 1.1 min (running 41%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/differential/test_a0_armature.py::test_at_the_stand_in_only_a_controller_that_sees_the_armature_holds_a_kick_past_the_limit took 1.2 min (tracing/lowering 42%: large or repeatedly rebuilt program; no cache removes this). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/differential/test_a0_armature.py::test_a0s_controller_holds_a_geared_wrist_inside_its_limit_and_the_bare_wrist_passes_it took 1.7 min (compiling 53%: 100 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_tube_kind.py::test_the_confined_matrix_stays_conditioned_and_its_radius_derivative_is_right_on_the_current_sampling[capsule] took 1.8 min (compiling 86%: 343 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_tube_kind.py::test_the_confined_matrix_stays_conditioned_and_its_radius_derivative_is_right_on_the_current_sampling[sphere] took 1.8 min (compiling 82%: 272 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_node_walk.py::test_every_node_an_entry_builds_passes_maddenings_verification[observation toy.region_probe 0.1.0] took 2.2 min (running 69%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_node_walk.py::test_every_node_an_entry_builds_passes_maddenings_verification[dynamics arm 0.1.0] took 2.6 min (running 45%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_node_walk.py::test_every_entry_builds_the_node_classes_its_builds_declares[wall kind free_space 0.1.0] took 20.2 s here and 17.6 s on a re-run alone (limit 20 s): one runner's outlier, not a slow test. If it repeats, look at the runner and at what the test shares with its neighbours.

tests/oracles/test_link_loads.py::test_loads_on_the_frames_enter_inverse_dynamics_as_the_jacobian_transpose_maps_them took 35.9 s (compiling 56%: 12 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_profile_record.py::test_t1_bare_saves_a_folder_that_loads_pends_and_verifies took 40.3 s (running 85%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_sampling_walk.py::test_every_kind_that_samples_spreads_its_samples_evenly[cylinder-2000] took 42.1 s (running 84%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/live/test_serve.py::test_every_frame_is_delivered_in_order_and_is_the_batch_runs_state took 51.6 s (running 52%: work in a subprocess (not measured here)). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_sampling_walk.py::test_control_every_kinds_own_quadrature_fails_the_spacing_check took 1.1 min (compiling 72%: 1470 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_switch_scan.py::test_the_tree_passes_the_scan_with_its_list_and_its_owners took 1.8 min (tracing/lowering 64%: large or repeatedly rebuilt program; no cache removes this). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_switch_harness.py::test_every_guard_the_scan_lists_is_reached_false_and_every_nonzero_norm_away_from_zero took 2.0 min (tracing/lowering 57%: large or repeatedly rebuilt program; no cache removes this). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_switch_scan.py::test_every_owner_passed_is_one_a_collector_records_in_a_fresh_trace_of_the_steps took 2.3 min (tracing/lowering 70%: large or repeatedly rebuilt program; no cache removes this). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_switch_harness.py::test_replacing_every_fallback_changes_nothing_where_its_condition_is_false[the hull contact on FL-9 and an ellipsoid at the wall] took 2.4 min (running 68%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_switch_scan.py::test_an_owner_passed_by_position_to_a_helper_fails took 2.6 min (tracing/lowering 51%: large or repeatedly rebuilt program; no cache removes this). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_mrs_near_wall.py::test_the_integrated_image_is_its_sum_over_each_cell_written_again[every cell] took 49.0 s (compiling 59%: 450 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/scene/test_root_prims.py::test_the_stage_writer_emits_only_world_as_a_root_prim[script A0] took 51.8 s (tracing/lowering 46%: large or repeatedly rebuilt program; no cache removes this). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/differential/test_a0_controls.py::test_a_fault_planted_in_each_part_fails_the_comparison[the-gradient-force-minus] took 58.5 s (compiling 46%: 1 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_armature.py::test_the_checks_ratios_predict_each_joints_kick_over_six_armatures_and_its_controls_fail took 59.4 s (tracing/lowering 58%: large or repeatedly rebuilt program; no cache removes this). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/differential/test_a0_controls.py::test_a_fault_planted_in_each_part_fails_the_comparison[the-gradient-force-plus] took 1.1 min (running 36%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/docs/test_reproducible.py::test_w8_mime_docs_check_fails_on_a_planted_hash_order_dependence_and_passes_without_it took 1.3 min (running 100%: work in a subprocess (not measured here)). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_golden_outputs.py::test_each_name_and_version_reproduces_its_golden_output[interaction contact 0.4.0] took 1.3 min (compiling 69%: 2 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_mrs_near_wall.py::test_control_the_stop_removed_lets_the_indefinite_resistance_through_and_the_drag_puts_energy_in took 1.7 min (compiling 61%: 2 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_mrs_near_wall.py::test_script_as_floored_screw_builds_and_its_run_stops_at_the_first_indefinite_step_before_the_body_takes_it took 1.8 min (compiling 63%: 194 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_helix_reference_band.py::test_quasi_uniform_sampling_brings_the_frozen_point_inside_the_reference_band_and_the_quadrature_does_not took 6.5 min (running 97%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_sampling_walk.py::test_every_kind_that_samples_spreads_its_samples_evenly[capsule-2000] took 33.3 s (running 67%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_sampling_walk.py::test_every_kind_that_samples_spreads_its_samples_evenly[cylinder-2000] took 35.2 s (running 82%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/live/test_serve.py::test_every_frame_is_delivered_in_order_and_is_the_batch_runs_state took 39.4 s (running 54%: work in a subprocess (not measured here)). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_profile_record.py::test_t1_bare_saves_a_folder_that_loads_pends_and_verifies took 41.5 s (running 83%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_sampling_walk.py::test_control_every_kinds_own_quadrature_fails_the_spacing_check took 1.0 min (compiling 74%: 1467 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_switch_scan.py::test_every_owner_passed_is_one_a_collector_records_in_a_fresh_trace_of_the_steps took 1.1 min (tracing/lowering 75%: large or repeatedly rebuilt program; no cache removes this). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_switch_scan.py::test_the_tree_passes_the_scan_with_its_list_and_its_owners took 1.2 min (tracing/lowering 50%: large or repeatedly rebuilt program; no cache removes this). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_switch_harness.py::test_every_guard_the_scan_lists_is_reached_false_and_every_nonzero_norm_away_from_zero took 1.5 min (tracing/lowering 46%: large or repeatedly rebuilt program; no cache removes this). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_switch_harness.py::test_replacing_every_fallback_changes_nothing_where_its_condition_is_false[the hull contact on FL-9 and an ellipsoid at the wall] took 1.8 min (running 61%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_switch_scan.py::test_an_owner_passed_by_position_to_a_helper_fails took 1.8 min (tracing/lowering 38%: large or repeatedly rebuilt program; no cache removes this). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_arm_dynamics_battery.py::test_the_dynamic_arm_is_first_order_in_time took 38.4 s (compiling 68%: 118 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_arm_dynamics_battery.py::test_the_dynamic_arm_passes_maddenings_check[gradient_finite] took 2.6 min (compiling 54%: 15 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_arm_dynamics_battery.py::test_the_dynamic_arm_passes_maddenings_check[params_gradient_finite] took 2.8 min (compiling 53%: 17 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_arm_dynamics_battery.py::test_the_dynamic_arm_passes_maddenings_check[params_effective] took 5.9 min (tracing/lowering 39%: large or repeatedly rebuilt program; no cache removes this). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_registry_walk.py::test_every_cases_parameter_leaves_reach_its_compiled_step[massive_rpm] took 56.1 s (running 46%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_registry_walk.py::test_every_cases_parameter_leaves_reach_its_compiled_step[hull_contact] took 1.3 min (compiling 75%: 1 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/scene/test_screw.py::test_the_screws_distance_is_the_brute_force_minimums took 1.3 min (compiling 54%: 205 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_registry_walk.py::test_every_case_round_trips_through_a_fresh_process[dynamic_rpm] took 1.4 min (compiling 48%: 12 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_registry_walk.py::test_every_case_round_trips_through_a_fresh_process[massive_rpm] took 1.4 min (compiling 54%: 2 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_registry_walk.py::test_every_case_round_trips_through_a_fresh_process[bound_robot_with_parts] took 1.6 min (compiling 50%: 4 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_registry_walk.py::test_every_kind_that_samples_its_surface_passes_the_kind_conformance_harness took 2.4 min (running 78%: work in a subprocess (not measured here)). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_registry_walk.py::test_every_case_round_trips_through_a_fresh_process[hull_contact] took 3.2 min (compiling 56%: 56 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/env/test_user_xla_cache.py::test_a_damaged_cache_entry_costs_a_warning_and_never_a_run_or_a_digest took 21.2 s here and 8.4 s on a re-run alone (limit 20 s): one runner's outlier, not a slow test. If it repeats, look at the runner and at what the test shares with its neighbours.

tests/env/test_user_xla_cache.py::test_a_cold_and_a_warm_cache_give_the_bits_of_no_cache_and_a_changed_program_is_a_miss_not_a_stale_hit took 21.4 s here and 10.1 s on a re-run alone (limit 20 s): one runner's outlier, not a slow test. If it repeats, look at the runner and at what the test shares with its neighbours.

tests/differential/test_a0_dynamic_controls.py::test_a_fault_planted_in_each_new_part_fails_the_comparison[the-joint-loops-gravity-term-minus] took 42.0 s (running 43%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/differential/test_a0_dynamic_controls.py::test_a_fault_planted_in_each_new_part_fails_the_comparison[the-rotor-loops-gain-plus] took 42.2 s (running 43%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/differential/test_a0_dynamic_controls.py::test_a_fault_planted_in_each_new_part_fails_the_comparison[the-joint-loops-gain-plus] took 42.7 s (running 42%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/differential/test_a0_dynamic_controls.py::test_a_fault_planted_in_each_new_part_fails_the_comparison[the-rotor-loops-gain-minus] took 42.7 s (running 43%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_node_walk.py::test_every_node_an_entry_builds_passes_maddenings_verification[observation toy.region_probe 0.1.0] took 45.3 s (running 76%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/differential/test_a0_armature.py::test_at_the_stand_in_only_a_controller_that_sees_the_armature_holds_a_kick_past_the_limit took 49.0 s (tracing/lowering 48%: large or repeatedly rebuilt program; no cache removes this). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/differential/test_a0_armature.py::test_a0s_controller_holds_a_geared_wrist_inside_its_limit_and_the_bare_wrist_passes_it took 1.1 min (compiling 49%: 100 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_tube_kind.py::test_the_confined_matrix_stays_conditioned_and_its_radius_derivative_is_right_on_the_current_sampling[capsule] took 1.3 min (compiling 88%: 343 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_tube_kind.py::test_the_confined_matrix_stays_conditioned_and_its_radius_derivative_is_right_on_the_current_sampling[sphere] took 1.3 min (compiling 82%: 272 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_node_walk.py::test_every_node_an_entry_builds_passes_maddenings_verification[dynamics arm 0.1.0] took 1.6 min (running 45%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/differential/test_a0.py::test_a0_matches_its_reference_with_exactly_the_declared_differences[3hz] took 39.2 s (running 42%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/control/test_in_graph.py::test_a0s_rig_runs_alike_with_its_controller_inside_the_compiled_step_and_as_a_callback[feedback-on] took 42.5 s (compiling 45%: 2 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/cli/test_serve_terminal.py::test_a_scripted_session_on_a_terminal_shows_each_event_once_above_a_strip_that_follows_the_status[100] took 49.9 s (running 59%: work in a subprocess (not measured here)). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/control/test_in_graph.py::test_a_controller_inside_the_compiled_step_is_saved_loaded_and_verified took 58.9 s (tracing/lowering 53%: large or repeatedly rebuilt program; no cache removes this). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/families/test_family_walk.py::test_every_interaction_in_a_family_passes_its_suite[interaction drag 0.4.0] took 1.0 min (running 91%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/families/test_family_walk.py::test_every_interaction_in_a_family_passes_its_suite[interaction drag 0.3.0] took 1.3 min (running 83%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/differential/test_a0.py::test_the_emulation_reproduces_the_reference_and_is_the_declared_prediction[TAG] took 1.5 min (running 88%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/build/test_mounted_refusals.py::test_mass_on_a_kinematic_motors_rotor_on_the_dynamic_arm_is_an_articulated_payload took 20.6 s here and 10.3 s on a re-run alone (limit 20 s): one runner's outlier, not a slow test. If it repeats, look at the runner and at what the test shares with its neighbours.

tests/saved/test_duration_saved.py::test_a_saved_folder_runs_as_long_as_it_declared_with_no_duration_given took 21.3 s here and 11.1 s on a re-run alone (limit 20 s): one runner's outlier, not a slow test. If it repeats, look at the runner and at what the test shares with its neighbours.

tests/saved/test_duration_saved.py::test_the_control_a_folder_with_no_duration_still_needs_one took 22.6 s here and 16.7 s on a re-run alone (limit 20 s): one runner's outlier, not a slow test. If it repeats, look at the runner and at what the test shares with its neighbours.

tests/saved/test_registry_walk.py::test_every_cases_parameter_leaves_reach_its_compiled_step[dynamic_rpm] took 51.2 s (running 48%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_registry_walk.py::test_every_cases_parameter_leaves_reach_its_compiled_step[massive_rpm] took 57.7 s (running 46%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/scene/test_screw.py::test_the_screws_distance_is_the_brute_force_minimums took 1.3 min (compiling 46%: 196 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_registry_walk.py::test_every_cases_parameter_leaves_reach_its_compiled_step[hull_contact] took 1.4 min (compiling 61%: 1 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_registry_walk.py::test_every_case_round_trips_through_a_fresh_process[dynamic_rpm] took 1.5 min (compiling 41%: 12 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_registry_walk.py::test_every_case_round_trips_through_a_fresh_process[massive_rpm] took 1.6 min (compiling 40%: 2 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_registry_walk.py::test_every_case_round_trips_through_a_fresh_process[bound_robot_with_parts] took 1.8 min (compiling 39%: 4 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_registry_walk.py::test_every_kind_that_samples_its_surface_passes_the_kind_conformance_harness took 2.5 min (running 76%: work in a subprocess (not measured here)). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_registry_walk.py::test_every_case_round_trips_through_a_fresh_process[hull_contact] took 3.5 min (compiling 47%: 56 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/env/test_user_xla_cache.py::test_a_damaged_cache_entry_costs_a_warning_and_never_a_run_or_a_digest took 22.2 s here and 9.8 s on a re-run alone (limit 20 s): one runner's outlier, not a slow test. If it repeats, look at the runner and at what the test shares with its neighbours.

tests/control/test_in_graph.py::test_a0s_rig_runs_alike_with_its_controller_inside_the_compiled_step_and_as_a_callback[feedback-on] took 41.8 s (compiling 55%: 2 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/cli/test_serve_terminal.py::test_a_scripted_session_on_a_terminal_shows_each_event_once_above_a_strip_that_follows_the_status[100] took 53.0 s (running 58%: work in a subprocess (not measured here)). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/control/test_in_graph.py::test_a_controller_inside_the_compiled_step_is_saved_loaded_and_verified took 56.4 s (tracing/lowering 50%: large or repeatedly rebuilt program; no cache removes this). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/families/test_family_walk.py::test_every_interaction_in_a_family_passes_its_suite[interaction drag 0.4.0] took 1.1 min (running 93%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/families/test_family_walk.py::test_every_interaction_in_a_family_passes_its_suite[interaction drag 0.3.0] took 1.3 min (running 82%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/differential/test_a0.py::test_the_emulation_reproduces_the_reference_and_is_the_declared_prediction[TAG] took 1.4 min (running 87%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/build/test_mounted_refusals.py::test_mass_on_a_kinematic_motors_rotor_on_the_dynamic_arm_is_an_articulated_payload took 20.5 s here and 10.5 s on a re-run alone (limit 20 s): one runner's outlier, not a slow test. If it repeats, look at the runner and at what the test shares with its neighbours.

tests/saved/test_duration_saved.py::test_an_override_beats_the_saved_length took 21.0 s here and 13.0 s on a re-run alone (limit 20 s): one runner's outlier, not a slow test. If it repeats, look at the runner and at what the test shares with its neighbours.

tests/saved/test_duration_saved.py::test_the_built_experiment_reads_the_length_it_declared took 21.1 s here and 10.8 s on a re-run alone (limit 20 s): one runner's outlier, not a slow test. If it repeats, look at the runner and at what the test shares with its neighbours.

tests/saved/test_duration_saved.py::test_a_saved_folder_runs_as_long_as_it_declared_with_no_duration_given took 28.0 s here and 13.7 s on a re-run alone (limit 20 s): one runner's outlier, not a slow test. If it repeats, look at the runner and at what the test shares with its neighbours.

tests/differential/test_a0_dynamic.py::test_a0_on_its_dynamic_rungs_matches_its_reference_with_exactly_the_second_declarations_differences[3hz] took 47.5 s (compiling 42%: 158 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_dynamic_arm.py::test_the_energy_balance_closes_to_the_integrators_order took 1.3 min (compiling 51%: 26 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_dynamic_arm.py::test_control_the_arm_with_its_gravity_dropped_fails_the_balance_on_every_draw took 1.3 min (compiling 50%: 9 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_experiment_states.py::test_the_experiment_api_keeps_its_invariants_over_random_scripts took 1.8 min (running 71%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_mrs_family.py::test_the_mrs_rung_conforms_to_the_hydrodynamic_family_in_free_space_and_swaps_with_the_stokes_sphere[0.4.0] took 1.9 min (running 85%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_dynamic_arm.py::test_control_a_stop_that_gives_energy_fails_the_balance_on_every_draw took 1.9 min (compiling 41%: 10 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/differential/test_a0_swap.py::test_a_step_of_lag_removed_or_added_inside_the_compiled_step_fails_the_swap[latency-doubled] took 2.0 min (running 70%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/differential/test_a0_swap.py::test_a0s_controller_runs_alike_inside_the_compiled_step_and_as_a_callback[no-latency] took 2.0 min (running 63%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/differential/test_a0_swap.py::test_a0s_controller_runs_alike_inside_the_compiled_step_and_as_a_callback[one-step] took 2.0 min (running 73%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/sharing/test_notice_card.py::test_every_file_the_folder_names_is_in_the_notice took 28.4 s here and 4.3 s on a re-run alone (limit 20 s): one runner's outlier, not a slow test. If it repeats, look at the runner and at what the test shares with its neighbours.

tests/oracles/test_hull_contact_support.py::test_control_a_search_that_ends_shallower_than_its_seed_raises_support_short took 1.0 min (compiling 75%: 105 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_spinner_tree.py::test_with_armature_on_its_joints_the_gimbal_converges_to_the_jointed_rotor_and_without_it_does_not took 1.1 min (compiling 48%: 9 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_hull_contact_support.py::test_the_support_never_leaves_a_held_point_shallower_than_its_seed took 1.1 min (compiling 69%: 1 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_hull_contact_support.py::test_a_body_at_rest_on_the_floor_neither_gains_energy_nor_wanders[fl9] took 1.1 min (compiling 57%: 58 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_spinner.py::test_control_the_forward_euler_coupling_grows_the_energy_on_every_draw took 1.5 min (running 95%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_spinner.py::test_the_gyroscopic_coupling_does_no_work_over_ten_thousand_steps_at_every_step_size took 1.6 min (running 92%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_hull_contact_support.py::test_a_healthy_run_stays_under_the_flag took 1.8 min (compiling 49%: 2 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_hull_contact_support.py::test_the_projection_after_the_step_leaves_the_kinetic_energy_as_the_solve_left_it took 2.9 min (compiling 78%: 49 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_hull_contact_support.py::test_the_cut_candidates_choose_the_step_all_the_points_choose_to_the_last_bit took 6.5 min (compiling 74%: 33 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/sharing/test_push_output.py::test_a_terminal_asked_to_be_plain_gets_exactly_what_the_pipe_gets[NO_COLOR] took 21.6 s here and 6.8 s on a re-run alone (limit 20 s): one runner's outlier, not a slow test. If it repeats, look at the runner and at what the test shares with its neighbours.

tests/test_collection_guard.py::test_this_suites_own_run_is_guarded_the_hazard_order_over_tests_saved_is_refused_and_the_same_files_in_order_are_not took 6.7 s (running 100%: work in a subprocess (not measured here)); the budget for the default lane is 5 s. Mark it slow or make it faster.

tests/oracles/test_arm_dynamics_battery.py::test_the_dynamic_arm_passes_maddenings_check[finite] took 26.9 s (compiling 48%: 63 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_arm_dynamics_battery.py::test_the_dynamic_arm_passes_maddenings_check[params_consistent] took 34.3 s (compiling 41%: 1 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_arm_dynamics_battery.py::test_the_dynamic_arm_is_first_order_in_time took 56.0 s (compiling 62%: 124 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_arm_dynamics_battery.py::test_the_dynamic_arm_passes_maddenings_check[gradient_finite] took 3.5 min (compiling 61%: 12 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_arm_dynamics_battery.py::test_the_dynamic_arm_passes_maddenings_check[params_gradient_finite] took 3.6 min (compiling 58%: 16 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_arm_dynamics_battery.py::test_the_dynamic_arm_passes_maddenings_check[params_effective] took 7.1 min (tracing/lowering 45%: large or repeatedly rebuilt program; no cache removes this). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_registry_walk.py::test_every_case_round_trips_through_a_fresh_process[tracked_rpm] took 26.9 s (compiling 37%: 11 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_registry_walk.py::test_every_cases_parameter_leaves_reach_its_compiled_step[dynamic_rpm] took 29.5 s (running 50%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_registry_walk.py::test_every_cases_parameter_leaves_reach_its_compiled_step[massive_rpm] took 31.8 s (compiling 46%: 1 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_registry_walk.py::test_every_case_round_trips_through_a_fresh_process[dynamic_rpm] took 48.0 s (compiling 43%: 12 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/scene/test_screw.py::test_the_screws_distance_is_the_brute_force_minimums took 49.5 s (compiling 53%: 212 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_registry_walk.py::test_every_cases_parameter_leaves_reach_its_compiled_step[hull_contact] took 52.1 s (compiling 66%: 1 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_registry_walk.py::test_every_case_round_trips_through_a_fresh_process[massive_rpm] took 58.1 s (compiling 49%: 2 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_registry_walk.py::test_every_case_round_trips_through_a_fresh_process[bound_robot_with_parts] took 59.6 s (compiling 42%: 4 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_registry_walk.py::test_every_kind_that_samples_its_surface_passes_the_kind_conformance_harness took 1.4 min (running 75%: work in a subprocess (not measured here)). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_registry_walk.py::test_every_case_round_trips_through_a_fresh_process[hull_contact] took 2.1 min (compiling 51%: 56 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_hull_contact.py::test_control_from_its_nearest_candidate_alone_the_counterexample_passes_the_margin took 43.2 s (compiling 77%: 1 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_hull_contact.py::test_control_without_its_rotation_coupled_in_the_couples_forces_do_not took 44.9 s (compiling 67%: 1 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_hull_contact.py::test_the_couples_forces_move_with_the_torque_as_its_closed_form took 50.1 s (compiling 63%: 8 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_hull_contact.py::test_the_held_points_derivatives_are_the_implicit_functions[screw] took 52.5 s (running 52%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_hull_contact.py::test_control_a_hull_held_by_its_samples_lets_its_body_past_the_margin took 57.8 s (running 78%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_hull_contact.py::test_control_its_search_left_at_five_steps_and_held_where_they_end_turns_the_step_otherwise took 1.1 min (compiling 71%: 2 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_hull_contact.py::test_where_no_point_reaches_the_wall_the_step_is_the_free_bodys_to_the_last_bit took 1.3 min (tracing/lowering 73%: large or repeatedly rebuilt program; no cache removes this). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_hull_contact.py::test_the_contact_never_leaves_the_step_more_energy_than_the_free_step took 1.9 min (compiling 53%: 4 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_hull_contact.py::test_no_hull_point_ends_past_the_wall_and_the_margin_is_kept_to_the_sweeps_convergence took 3.5 min (compiling 69%: 371 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_hull_contact.py::test_turning_the_tube_and_the_body_together_turns_the_step took 4.3 min (tracing/lowering 48%: large or repeatedly rebuilt program; no cache removes this). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_hull_contact.py::test_control_from_its_nearest_candidate_alone_the_counterexample_passes_the_margin took 43.6 s (compiling 76%: 1 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_hull_contact.py::test_control_a_hull_held_by_its_samples_lets_its_body_past_the_margin took 1.0 min (running 83%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_hull_contact.py::test_the_held_points_derivatives_are_the_implicit_functions[screw] took 1.2 min (running 45%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_hull_contact.py::test_control_its_search_left_at_five_steps_and_held_where_they_end_turns_the_step_otherwise took 1.2 min (compiling 73%: 2 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_hull_contact.py::test_the_couples_forces_move_with_the_torque_as_its_closed_form took 1.2 min (compiling 65%: 9 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_hull_contact.py::test_the_stress_runs_unconverged_draw_turns_the_step_with_the_frame took 1.3 min (compiling 52%: 1 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_hull_contact.py::test_where_no_point_reaches_the_wall_the_step_is_the_free_bodys_to_the_last_bit took 1.3 min (tracing/lowering 72%: large or repeatedly rebuilt program; no cache removes this). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_hull_contact.py::test_the_contact_never_leaves_the_step_more_energy_than_the_free_step took 1.8 min (compiling 59%: 4 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_hull_contact.py::test_turning_the_tube_and_the_body_together_turns_the_step took 3.4 min (tracing/lowering 46%: large or repeatedly rebuilt program; no cache removes this). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_hull_contact.py::test_no_hull_point_ends_past_the_wall_and_the_margin_is_kept_to_the_sweeps_convergence took 3.6 min (compiling 69%: 371 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/scene/test_root_prims.py::test_the_stage_writer_emits_only_world_as_a_root_prim[script A0] took 57.8 s (tracing/lowering 43%: large or repeatedly rebuilt program; no cache removes this). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/differential/test_a0_controls.py::test_a_fault_planted_in_each_part_fails_the_comparison[the-gradient-force-minus] took 59.0 s (compiling 51%: 1 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_armature.py::test_the_checks_ratios_predict_each_joints_kick_over_six_armatures_and_its_controls_fail took 1.0 min (tracing/lowering 56%: large or repeatedly rebuilt program; no cache removes this). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/differential/test_a0_controls.py::test_a_fault_planted_in_each_part_fails_the_comparison[the-gradient-force-plus] took 1.1 min (running 41%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/docs/test_reproducible.py::test_w8_mime_docs_check_fails_on_a_planted_hash_order_dependence_and_passes_without_it took 1.3 min (running 100%: work in a subprocess (not measured here)). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_golden_outputs.py::test_each_name_and_version_reproduces_its_golden_output[interaction contact 0.4.0] took 1.6 min (compiling 66%: 2 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_mrs_near_wall.py::test_control_the_stop_removed_lets_the_indefinite_resistance_through_and_the_drag_puts_energy_in took 2.1 min (compiling 60%: 2 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_mrs_near_wall.py::test_script_as_floored_screw_builds_and_its_run_stops_at_the_first_indefinite_step_before_the_body_takes_it took 2.2 min (compiling 64%: 205 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_helix_reference_band.py::test_quasi_uniform_sampling_brings_the_frozen_point_inside_the_reference_band_and_the_quadrature_does_not took 6.4 min (running 97%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_golden_outputs.py::test_each_name_and_version_reproduces_its_golden_output[geometry ellipsoid 0.1.0] took 23.6 s here and 18.1 s on a re-run alone (limit 20 s): one runner's outlier, not a slow test. If it repeats, look at the runner and at what the test shares with its neighbours.

tests/oracles/test_hull_contact.py::test_the_held_points_derivatives_are_the_implicit_functions[screw] took 42.7 s (running 49%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_hull_contact.py::test_the_couples_forces_move_with_the_torque_as_its_closed_form took 44.8 s (compiling 69%: 8 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_hull_contact.py::test_control_from_its_nearest_candidate_alone_the_counterexample_passes_the_margin took 45.0 s (compiling 77%: 1 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_hull_contact.py::test_control_without_its_rotation_coupled_in_the_couples_forces_do_not took 50.4 s (compiling 66%: 1 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_hull_contact.py::test_control_a_hull_held_by_its_samples_lets_its_body_past_the_margin took 54.0 s (running 84%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_hull_contact.py::test_control_its_search_left_at_five_steps_and_held_where_they_end_turns_the_step_otherwise took 1.2 min (compiling 72%: 2 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_hull_contact.py::test_where_no_point_reaches_the_wall_the_step_is_the_free_bodys_to_the_last_bit took 1.4 min (tracing/lowering 75%: large or repeatedly rebuilt program; no cache removes this). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_hull_contact.py::test_the_contact_never_leaves_the_step_more_energy_than_the_free_step took 1.9 min (compiling 53%: 4 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_hull_contact.py::test_no_hull_point_ends_past_the_wall_and_the_margin_is_kept_to_the_sweeps_convergence took 3.6 min (compiling 70%: 371 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_hull_contact.py::test_turning_the_tube_and_the_body_together_turns_the_step took 4.5 min (tracing/lowering 49%: large or repeatedly rebuilt program; no cache removes this). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/docs/test_walk.py::test_each_link_form_to_a_page_that_names_nothing_fails[with-a-title] took 17.1 s (running 100%: work in a subprocess (not measured here)); the budget for the default lane is 5 s. Mark it slow or make it faster.

tests/saved/test_node_walk.py::test_every_entry_builds_the_node_classes_its_builds_declares[wall kind free_space 0.1.0] took 17.1 s (compiling 68%: 411 cache misses); the budget for the default lane is 5 s. Mark it slow or make it faster.

tests/saved/test_duration_saved.py::test_the_control_a_folder_with_no_duration_still_needs_one took 17.8 s (running 89%: un-jitted loop, heavy compute, or a subprocess); the budget for the default lane is 5 s. Mark it slow or make it faster.

tests/validity/test_profile_build.py::test_a_profile_changes_no_value_and_no_graph took 18.9 s (running 46%: un-jitted loop, heavy compute, or a subprocess); the budget for the default lane is 5 s. Mark it slow or make it faster.

tests/oracles/test_mrs_near_wall.py::test_the_build_report_states_the_wall_refinement_taken_and_says_where_it_is_not_the_one_written[as written] took 18.9 s (compiling 45%: 1 cache misses); the budget for the default lane is 5 s. Mark it slow or make it faster.

tests/env/test_user_xla_cache.py::test_a_cold_and_a_warm_cache_give_the_bits_of_no_cache_and_a_changed_program_is_a_miss_not_a_stale_hit took 19.2 s (running 99%: work in a subprocess (not measured here)); the budget for the default lane is 5 s. Mark it slow or make it faster.

tests/oracles/test_armature.py::test_the_line_is_never_a_refusal_and_a_geared_wrist_has_none took 19.5 s (tracing/lowering 75%: large or repeatedly rebuilt program; no cache removes this); the budget for the default lane is 5 s. Mark it slow or make it faster.

tests/env/test_user_xla_cache.py::test_a_damaged_cache_entry_costs_a_warning_and_never_a_run_or_a_digest took 22.2 s (running 100%: work in a subprocess (not measured here)); the budget for the default lane is 5 s. Mark it slow or make it faster.

tests/saved/test_golden_outputs.py::test_each_name_and_version_reproduces_its_golden_output[geometry ellipsoid 0.1.0] took 23.6 s (tracing/lowering 56%: large or repeatedly rebuilt program; no cache removes this); the budget for the default lane is 5 s. Mark it slow or make it faster.

tests/sharing/test_notice_card.py::test_every_file_the_folder_names_is_in_the_notice took 30.8 s (running 95%: un-jitted loop, heavy compute, or a subprocess); the budget for the default lane is 5 s. Mark it slow or make it faster.

tests/oracles/test_hull_contact_support.py::test_the_support_never_leaves_a_held_point_shallower_than_its_seed took 1.3 min (compiling 65%: 105 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_spinner_tree.py::test_on_a_gimbal_the_spinner_converges_to_the_jointed_rotor_and_its_gyroscopic_sign_flipped_does_not took 1.3 min (compiling 44%: 9 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_hull_contact_support.py::test_a_body_at_rest_on_the_floor_neither_gains_energy_nor_wanders[fl9] took 1.3 min (compiling 53%: 57 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_hull_contact_support.py::test_control_a_flag_set_below_that_run_rises_on_it took 1.4 min (running 47%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_spinner.py::test_control_the_forward_euler_coupling_grows_the_energy_on_every_draw took 1.8 min (running 95%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_spinner.py::test_the_gyroscopic_coupling_does_no_work_over_ten_thousand_steps_at_every_step_size took 1.8 min (running 90%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_hull_contact_support.py::test_a_healthy_run_stays_under_the_flag took 2.0 min (compiling 46%: 2 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_hull_contact_support.py::test_the_projection_after_the_step_leaves_the_kinetic_energy_as_the_solve_left_it took 3.3 min (compiling 67%: 94 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_hull_contact_support.py::test_the_cut_candidates_choose_the_step_all_the_points_choose_to_the_last_bit took 6.6 min (compiling 70%: 9 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/sharing/test_push_output.py::test_a_terminal_asked_to_be_plain_gets_exactly_what_the_pipe_gets[--plain] took 22.4 s here and 7.0 s on a re-run alone (limit 20 s): one runner's outlier, not a slow test. If it repeats, look at the runner and at what the test shares with its neighbours.

tests/control/test_in_graph.py::test_a_derivative_of_the_run_passes_through_the_controller_inside_the_compiled_step_and_not_through_a_callback took 15.7 s (compiling 54%: 17 cache misses); the budget for the default lane is 5 s. Mark it slow or make it faster.

tests/saved/test_duration_saved.py::test_an_override_beats_the_saved_length took 15.9 s (running 86%: un-jitted loop, heavy compute, or a subprocess); the budget for the default lane is 5 s. Mark it slow or make it faster.

tests/build/test_mounted_refusals.py::test_a_rotor_unequal_across_its_axis_is_carried_on_the_dynamic_arm took 16.1 s (tracing/lowering 61%: large or repeatedly rebuilt program; no cache removes this); the budget for the default lane is 5 s. Mark it slow or make it faster.

tests/saved/test_duration_saved.py::test_the_built_experiment_reads_the_length_it_declared took 17.0 s (running 91%: un-jitted loop, heavy compute, or a subprocess); the budget for the default lane is 5 s. Mark it slow or make it faster.

tests/saved/test_node_walk.py::test_every_entry_builds_the_node_classes_its_builds_declares[wall kind free_space 0.1.0] took 20.2 s (compiling 72%: 411 cache misses); the budget for the default lane is 5 s. Mark it slow or make it faster.

tests/build/test_mounted_refusals.py::test_mass_on_a_kinematic_motors_rotor_on_the_dynamic_arm_is_an_articulated_payload took 20.6 s (compiling 72%: 2 cache misses); the budget for the default lane is 5 s. Mark it slow or make it faster.

tests/saved/test_duration_saved.py::test_a_saved_folder_runs_as_long_as_it_declared_with_no_duration_given took 21.3 s (running 72%: un-jitted loop, heavy compute, or a subprocess); the budget for the default lane is 5 s. Mark it slow or make it faster.

tests/sharing/test_notice_card.py::test_every_file_the_folder_names_is_in_the_notice took 21.4 s (running 96%: un-jitted loop, heavy compute, or a subprocess); the budget for the default lane is 5 s. Mark it slow or make it faster.

tests/sharing/test_push_output.py::test_a_terminal_asked_to_be_plain_gets_exactly_what_the_pipe_gets[--plain] took 22.4 s (running 96%: un-jitted loop, heavy compute, or a subprocess); the budget for the default lane is 5 s. Mark it slow or make it faster.

tests/saved/test_duration_saved.py::test_the_control_a_folder_with_no_duration_still_needs_one took 22.6 s (running 86%: un-jitted loop, heavy compute, or a subprocess); the budget for the default lane is 5 s. Mark it slow or make it faster.

tests/test_collection_guard.py::test_this_suites_own_run_is_guarded_the_hazard_order_over_tests_saved_is_refused_and_the_same_files_in_order_are_not took 7.7 s (running 100%: work in a subprocess (not measured here)); the budget for the default lane is 5 s. Mark it slow or make it faster.

tests/oracles/test_arm_dynamics_battery.py::test_the_dynamic_arm_passes_maddenings_check[deterministic] took 23.5 s (tracing/lowering 64%: large or repeatedly rebuilt program; no cache removes this). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_arm_dynamics_battery.py::test_the_dynamic_arm_passes_maddenings_check[finite] took 33.6 s (compiling 40%: 63 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_arm_dynamics_battery.py::test_the_dynamic_arm_is_first_order_in_time took 58.4 s (compiling 57%: 124 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_arm_dynamics_battery.py::test_the_dynamic_arm_passes_maddenings_check[gradient_finite] took 4.2 min (running 47%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_arm_dynamics_battery.py::test_the_dynamic_arm_passes_maddenings_check[params_gradient_finite] took 4.4 min (running 50%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_arm_dynamics_battery.py::test_the_dynamic_arm_passes_maddenings_check[params_effective] took 8.6 min (tracing/lowering 44%: large or repeatedly rebuilt program; no cache removes this). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/control/test_in_graph.py::test_a0s_rig_runs_alike_with_its_controller_inside_the_compiled_step_and_as_a_callback[feedback-off] took 35.8 s (compiling 56%: 2 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/control/test_in_graph.py::test_a0s_rig_runs_alike_with_its_controller_inside_the_compiled_step_and_as_a_callback[feedback-on] took 39.8 s (compiling 54%: 2 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/differential/test_a0.py::test_a0_matches_its_reference_with_exactly_the_declared_differences[3hz] took 40.6 s (compiling 42%: 1 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/families/test_family_walk.py::test_every_interaction_in_a_family_passes_its_suite[interaction drag 0.4.0] took 49.6 s (running 92%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/cli/test_serve_terminal.py::test_a_scripted_session_on_a_terminal_shows_each_event_once_above_a_strip_that_follows_the_status[100] took 54.4 s (running 62%: work in a subprocess (not measured here)). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/control/test_in_graph.py::test_a_controller_inside_the_compiled_step_is_saved_loaded_and_verified took 57.1 s (tracing/lowering 40%: large or repeatedly rebuilt program; no cache removes this). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/families/test_family_walk.py::test_every_interaction_in_a_family_passes_its_suite[interaction drag 0.3.0] took 1.1 min (running 79%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/differential/test_a0.py::test_the_emulation_reproduces_the_reference_and_is_the_declared_prediction[TAG] took 1.5 min (running 86%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_duration_saved.py::test_a_saved_folder_runs_as_long_as_it_declared_with_no_duration_given took 20.1 s here and 11.4 s on a re-run alone (limit 20 s): one runner's outlier, not a slow test. If it repeats, look at the runner and at what the test shares with its neighbours.

tests/saved/test_duration_saved.py::test_the_control_a_folder_with_no_duration_still_needs_one took 23.3 s here and 16.9 s on a re-run alone (limit 20 s): one runner's outlier, not a slow test. If it repeats, look at the runner and at what the test shares with its neighbours.

tests/differential/test_a0_controls.py::test_a_fault_planted_in_each_part_fails_the_comparison[the-motors-speed-minus] took 45.5 s (running 43%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_armature.py::test_the_checks_ratios_predict_each_joints_kick_over_six_armatures_and_its_controls_fail took 46.7 s (tracing/lowering 44%: large or repeatedly rebuilt program; no cache removes this). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/differential/test_a0_controls.py::test_a_fault_planted_in_each_part_fails_the_comparison[the-gradient-force-minus] took 50.4 s (running 39%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/scene/test_root_prims.py::test_the_stage_writer_emits_only_world_as_a_root_prim[script A0] took 50.6 s (compiling 51%: 2 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/differential/test_a0_controls.py::test_a_fault_planted_in_each_part_fails_the_comparison[the-gradient-force-plus] took 59.5 s (compiling 39%: 1 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_golden_outputs.py::test_each_name_and_version_reproduces_its_golden_output[interaction contact 0.4.0] took 1.0 min (compiling 71%: 2 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/docs/test_reproducible.py::test_w8_mime_docs_check_fails_on_a_planted_hash_order_dependence_and_passes_without_it took 1.1 min (running 100%: work in a subprocess (not measured here)). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_mrs_near_wall.py::test_control_the_stop_removed_lets_the_indefinite_resistance_through_and_the_drag_puts_energy_in took 1.5 min (compiling 65%: 2 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_mrs_near_wall.py::test_script_as_floored_screw_builds_and_its_run_stops_at_the_first_indefinite_step_before_the_body_takes_it took 1.6 min (compiling 66%: 205 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_helix_reference_band.py::test_quasi_uniform_sampling_brings_the_frozen_point_inside_the_reference_band_and_the_quadrature_does_not took 6.1 min (running 97%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/results/test_reduction_oracle.py::test_control_the_float64_rule_on_a_float32_run_is_a_step_off_at_every_start_where_the_node_counts_the_tie took 51.0 s (compiling 42%: 36 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/differential/test_a0_dynamic_b_controls.py::test_the_rotor_loops_ki_fails_its_integral[10hz-plus] took 55.7 s (compiling 45%: 1 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/scene/test_rotated_world.py::test_the_body_side_in_a_turned_world_runs_as_the_first_world_turned took 56.3 s (tracing/lowering 46%: large or repeatedly rebuilt program; no cache removes this). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/differential/test_a0_dynamic_b_controls.py::test_the_motors_angle_offset_at_its_start_fails_the_initial_angle[10hz] took 59.5 s (running 42%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/differential/test_a0_dynamic_b_controls.py::test_the_rotor_loops_ki_fails_its_integral[3hz-plus] took 1.0 min (compiling 44%: 1 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_dynamic_motor.py::test_the_rotor_spins_up_to_its_closed_form_first_order_and_passes_maddenings_battery took 1.1 min (running 75%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/differential/test_a0_dynamic_b_controls.py::test_the_motors_angle_offset_at_its_start_fails_the_initial_angle[3hz] took 1.1 min (compiling 43%: 214 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_hull_contact_screws.py::test_an_anisotropic_screw_turns_broadside_and_is_held_at_the_bores_limit took 1.3 min (compiling 62%: 131 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/live/test_a0_live.py::test_a0_through_mime_serve_is_never_ahead_of_real_time_and_each_frame_is_the_batch_runs_state took 1.7 min (running 52%: work in a subprocess (not measured here)). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/results/test_reduction_oracle.py::test_a_speed_that_is_not_linear_in_time_has_its_mean_as_its_series_mean_and_not_as_its_ends_midpoint took 2.6 min (compiling 55%: 158 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/differential/test_a0_dynamic_controls.py::test_a_fault_planted_in_each_new_part_fails_the_comparison[the-joint-loops-gain-plus] took 58.4 s (compiling 43%: 44 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/differential/test_a0_dynamic_controls.py::test_a_fault_planted_in_each_new_part_fails_the_comparison[the-rotor-loops-gain-minus] took 59.8 s (compiling 43%: 1 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_tube_kind.py::test_the_confined_matrix_stays_conditioned_and_its_radius_derivative_is_right_on_the_current_sampling[sphere] took 1.0 min (compiling 68%: 272 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_node_walk.py::test_every_node_an_entry_builds_passes_maddenings_verification[observation toy.region_probe 0.1.0] took 1.0 min (running 73%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/differential/test_a0_dynamic_controls.py::test_a_fault_planted_in_each_new_part_fails_the_comparison[the-rotor-loops-gain-plus] took 1.0 min (compiling 44%: 1 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/differential/test_a0_dynamic_controls.py::test_a_fault_planted_in_each_new_part_fails_the_comparison[the-joint-loops-gravity-term-plus] took 1.1 min (compiling 45%: 1 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/differential/test_a0_dynamic_controls.py::test_a_fault_planted_in_each_new_part_fails_the_comparison[the-joint-loops-gain-minus] took 1.1 min (compiling 43%: 1 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/differential/test_a0_armature.py::test_at_the_stand_in_only_a_controller_that_sees_the_armature_holds_a_kick_past_the_limit took 1.1 min (compiling 44%: 1 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/differential/test_a0_armature.py::test_a0s_controller_holds_a_geared_wrist_inside_its_limit_and_the_bare_wrist_passes_it took 1.7 min (compiling 60%: 100 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/saved/test_node_walk.py::test_every_node_an_entry_builds_passes_maddenings_verification[dynamics arm 0.1.0] took 2.0 min (running 43%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_hull_contact.py::test_control_from_its_nearest_candidate_alone_the_counterexample_passes_the_margin took 46.3 s (compiling 78%: 1 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_hull_contact.py::test_control_without_its_rotation_coupled_in_the_couples_forces_do_not took 50.2 s (compiling 78%: 1 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_hull_contact.py::test_the_couples_forces_move_with_the_torque_as_its_closed_form took 51.0 s (compiling 70%: 8 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_hull_contact.py::test_the_held_points_derivatives_are_the_implicit_functions[screw] took 53.8 s (running 51%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_hull_contact.py::test_control_a_hull_held_by_its_samples_lets_its_body_past_the_margin took 57.4 s (running 84%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_hull_contact.py::test_where_no_point_reaches_the_wall_the_step_is_the_free_bodys_to_the_last_bit took 1.2 min (tracing/lowering 59%: large or repeatedly rebuilt program; no cache removes this). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_hull_contact.py::test_control_its_search_left_at_five_steps_and_held_where_they_end_turns_the_step_otherwise took 1.2 min (compiling 72%: 2 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_hull_contact.py::test_the_contact_never_leaves_the_step_more_energy_than_the_free_step took 1.6 min (compiling 63%: 4 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_hull_contact.py::test_no_hull_point_ends_past_the_wall_and_the_margin_is_kept_to_the_sweeps_convergence took 3.4 min (compiling 78%: 371 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_hull_contact.py::test_turning_the_tube_and_the_body_together_turns_the_step took 3.8 min (compiling 53%: 8 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_hull_contact_support.py::test_control_a_search_that_ends_shallower_than_its_seed_raises_support_short took 1.0 min (compiling 69%: 105 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_spinner_tree.py::test_on_a_joint_along_the_axis_the_spinner_converges_to_the_jointed_rotor_and_its_planted_faults_do_not took 1.0 min (tracing/lowering 39%: large or repeatedly rebuilt program; no cache removes this). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_hull_contact_support.py::test_a_body_at_rest_on_the_floor_neither_gains_energy_nor_wanders[fl9] took 1.1 min (compiling 52%: 58 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_hull_contact_support.py::test_the_support_never_leaves_a_held_point_shallower_than_its_seed took 1.1 min (compiling 67%: 1 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_spinner_tree.py::test_on_a_gimbal_the_spinner_converges_to_the_jointed_rotor_and_its_gyroscopic_sign_flipped_does_not took 1.3 min (compiling 43%: 9 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_spinner.py::test_the_gyroscopic_coupling_does_no_work_over_ten_thousand_steps_at_every_step_size took 1.5 min (running 91%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_spinner.py::test_control_the_forward_euler_coupling_grows_the_energy_on_every_draw took 1.5 min (running 91%: un-jitted loop, heavy compute, or a subprocess). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_hull_contact_support.py::test_a_healthy_run_stays_under_the_flag took 1.8 min (compiling 43%: 2 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_hull_contact_support.py::test_the_projection_after_the_step_leaves_the_kinetic_energy_as_the_solve_left_it took 2.2 min (compiling 69%: 48 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/oracles/test_hull_contact_support.py::test_the_cut_candidates_choose_the_step_all_the_points_choose_to_the_last_bit took 6.5 min (compiling 68%: 33 cache misses). It is on tests/duration_allowlist.txt, so the job passes, but the allowlist has no ceiling: check that this is still what keeping it on every push costs.

tests/docs/test_docs_check.py::test_the_command_checks_the_refusal_table_against_the_source took 14.0 s (running 100%: work in a subprocess (not measured here)); the budget for the default lane is 5 s. Mark it slow or make it faster.

tests/saved/test_node_walk.py::test_every_entry_builds_the_node_classes_its_builds_declares[wall kind free_space 0.1.0] took 14.1 s (compiling 74%: 411 cache misses); the budget for the default lane is 5 s. Mark it slow or make it faster.

tests/oracles/test_projection.py::test_a_fault_planted_in_the_node_is_found[no hold] took 14.4 s (running 96%: un-jitted loop, heavy compute, or a subprocess); the budget for the default lane is 5 s. Mark it slow or make it faster.

tests/env/test_user_xla_cache.py::test_a_damaged_cache_entry_costs_a_warning_and_never_a_run_or_a_digest took 15.0 s (running 100%: work in a subprocess (not measured here)); the budget for the default lane is 5 s. Mark it slow or make it faster.

tests/oracles/test_mrs_separated.py::test_a_moved_radius_makes_the_held_value_again_from_the_moved_positions took 15.3 s (running 68%: un-jitted loop, heavy compute, or a subprocess); the budget for the default lane is 5 s. Mark it slow or make it faster.

tests/saved/test_duration_saved.py::test_an_override_beats_the_saved_length took 16.0 s (running 93%: un-jitted loop, heavy compute, or a subprocess); the budget for the default lane is 5 s. Mark it slow or make it faster.

tests/saved/test_duration_saved.py::test_a_saved_folder_runs_as_long_as_it_declared_with_no_duration_given took 17.5 s (running 73%: un-jitted loop, heavy compute, or a subprocess); the budget for the default lane is 5 s. Mark it slow or make it faster.

tests/saved/test_duration_saved.py::test_the_control_a_folder_with_no_duration_still_needs_one took 19.9 s (running 92%: un-jitted loop, heavy compute, or a subprocess); the budget for the default lane is 5 s. Mark it slow or make it faster.

tests/sharing/test_push_output.py::test_a_terminal_asked_to_be_plain_gets_exactly_what_the_pipe_gets[NO_COLOR] took 21.6 s (running 97%: un-jitted loop, heavy compute, or a subprocess); the budget for the default lane is 5 s. Mark it slow or make it faster.

tests/sharing/test_notice_card.py::test_every_file_the_folder_names_is_in_the_notice took 28.4 s (running 97%: un-jitted loop, heavy compute, or a subprocess); the budget for the default lane is 5 s. Mark it slow or make it faster.

tests/oracles/test_projection.py::test_a_fault_planted_in_the_node_is_found[no hold] took 16.8 s (running 94%: un-jitted loop, heavy compute, or a subprocess); the budget for the default lane is 5 s. Mark it slow or make it faster.

tests/oracles/test_mrs.py::test_the_sphere_converges_on_stokes_drag_and_torque took 17.0 s (compiling 62%: 214 cache misses); the budget for the default lane is 5 s. Mark it slow or make it faster.

tests/credentials/test_credentials.py::test_the_new_checks_pass_unplanted_twenty_times took 17.7 s (running 100%: work in a subprocess (not measured here)); the budget for the default lane is 5 s. Mark it slow or make it faster.

tests/sharing/test_push_output.py::test_a_terminal_asked_to_be_plain_gets_exactly_what_the_pipe_gets[--plain] took 19.0 s (running 96%: un-jitted loop, heavy compute, or a subprocess); the budget for the default lane is 5 s. Mark it slow or make it faster.

tests/build/test_mounted_refusals.py::test_mass_on_a_kinematic_motors_rotor_on_the_dynamic_arm_is_an_articulated_payload took 20.5 s (compiling 80%: 2 cache misses); the budget for the default lane is 5 s. Mark it slow or make it faster.

tests/saved/test_duration_saved.py::test_an_override_beats_the_saved_length took 21.0 s (running 90%: un-jitted loop, heavy compute, or a subprocess); the budget for the default lane is 5 s. Mark it slow or make it faster.

tests/saved/test_duration_saved.py::test_the_built_experiment_reads_the_length_it_declared took 21.1 s (running 92%: un-jitted loop, heavy compute, or a subprocess); the budget for the default lane is 5 s. Mark it slow or make it faster.

tests/saved/test_duration_saved.py::test_a_saved_folder_runs_as_long_as_it_declared_with_no_duration_given took 28.0 s (running 70%: un-jitted loop, heavy compute, or a subprocess); the budget for the default lane is 5 s. Mark it slow or make it faster.

tests/saved/test_duration_saved.py::test_the_control_a_folder_with_no_duration_still_needs_one took 28.3 s (running 91%: un-jitted loop, heavy compute, or a subprocess); the budget for the default lane is 5 s. Mark it slow or make it faster.

tests/sharing/test_notice_card.py::test_every_file_the_folder_names_is_in_the_notice took 36.3 s (running 95%: un-jitted loop, heavy compute, or a subprocess); the budget for the default lane is 5 s. Mark it slow or make it faster.

tests/live/test_serve_limits.py::test_a_name_the_frames_do_not_carry_is_refused_with_the_nearest_name_before_a_run_folder_is_opened took 15.7 s (running 73%: work in a subprocess (not measured here)); the budget for the default lane is 5 s. Mark it slow or make it faster.

tests/saved/test_duration_saved.py::test_the_built_experiment_reads_the_length_it_declared took 15.9 s (running 92%: un-jitted loop, heavy compute, or a subprocess); the budget for the default lane is 5 s. Mark it slow or make it faster.

tests/oracles/test_mrs.py::test_the_sphere_converges_on_stokes_drag_and_torque took 16.6 s (compiling 66%: 214 cache misses); the budget for the default lane is 5 s. Mark it slow or make it faster.

tests/build/test_mounted_refusals.py::test_mass_on_a_kinematic_motors_rotor_on_the_dynamic_arm_is_an_articulated_payload took 17.1 s (compiling 79%: 2 cache misses); the budget for the default lane is 5 s. Mark it slow or make it faster.

tests/saved/test_duration_saved.py::test_an_override_beats_the_saved_length took 17.8 s (running 87%: un-jitted loop, heavy compute, or a subprocess); the budget for the default lane is 5 s. Mark it slow or make it faster.

tests/saved/test_node_walk.py::test_every_entry_builds_the_node_classes_its_builds_declares[wall kind free_space 0.1.0] took 19.8 s (compiling 74%: 411 cache misses); the budget for the default lane is 5 s. Mark it slow or make it faster.

tests/saved/test_duration_saved.py::test_a_saved_folder_runs_as_long_as_it_declared_with_no_duration_given took 20.1 s (running 71%: un-jitted loop, heavy compute, or a subprocess); the budget for the default lane is 5 s. Mark it slow or make it faster.

tests/env/test_user_xla_cache.py::test_a_damaged_cache_entry_costs_a_warning_and_never_a_run_or_a_digest took 21.2 s (running 100%: work in a subprocess (not measured here)); the budget for the default lane is 5 s. Mark it slow or make it faster.

tests/env/test_user_xla_cache.py::test_a_cold_and_a_warm_cache_give_the_bits_of_no_cache_and_a_changed_program_is_a_miss_not_a_stale_hit took 21.4 s (running 99%: work in a subprocess (not measured here)); the budget for the default lane is 5 s. Mark it slow or make it faster.

tests/saved/test_duration_saved.py::test_the_control_a_folder_with_no_duration_still_needs_one took 23.3 s (running 88%: un-jitted loop, heavy compute, or a subprocess); the budget for the default lane is 5 s. Mark it slow or make it faster.

"The ubuntu-latest label will migrate to Ubuntu 26 beginning October 19, 2026. For more information, see https://github.com/actions/runner-images/issues/14748"

"The ubuntu-latest label will migrate to Ubuntu 26 beginning October 19, 2026. For more information, see https://github.com/actions/runner-images/issues/14748"

"The ubuntu-latest label will migrate to Ubuntu 26 beginning October 19, 2026. For more information, see https://github.com/actions/runner-images/issues/14748"

"The ubuntu-latest label will migrate to Ubuntu 26 beginning October 19, 2026. For more information, see https://github.com/actions/runner-images/issues/14748"

"The ubuntu-latest label will migrate to Ubuntu 26 beginning October 19, 2026. For more information, see https://github.com/actions/runner-images/issues/14748"

"The ubuntu-latest label will migrate to Ubuntu 26 beginning October 19, 2026. For more information, see https://github.com/actions/runner-images/issues/14748"

"The ubuntu-latest label will migrate to Ubuntu 26 beginning October 19, 2026. For more information, see https://github.com/actions/runner-images/issues/14748"