Skip to content

Official Codebase of 3DCodeVerse

Python

record: telemetry by_round scores a degraded verdict as unscored — it quoted the raw 0.0 where every other reader uses effective_score (bug B10)

gaoypeng013125add213
diff --git a/harness/codeverse3d/record/telemetry.py b/harness/codeverse3d/record/telemetry.py
index 296c7c75..d8f068d4 100644
--- a/harness/codeverse3d/record/telemetry.py
+++ b/harness/codeverse3d/record/telemetry.py
@@ -35,6 +35,7 @@ from codeverse3d.contracts.run import (
from codeverse3d.cost.ledger import load_ledger, summarise
from codeverse3d.cost.types import STAGE_ORDER, CallCost
from codeverse3d.proc import read_json_or_none, write_json_atomic
+from codeverse3d.record.record import effective_score
from codeverse3d.workspace import Workspace
log = logging.getLogger(__name__)
@@ -175,7 +176,7 @@ def cost_summary(record: RunRecord, rows: list[CallCost]) -> CostSummary:
by_role={k: round(b.cost_usd, 6) for k, b in summ.dimension("role").items()},
by_model={k: round(b.cost_usd, 6) for k, b in summ.dimension("model").items()},
by_round=[{"index": r.index, "kind": r.kind, "cost_usd": round(r.usage.cost_usd, 6),
- "minutes": round(r.minutes, 2), "score": r.score} for r in record.rounds],
+ "minutes": round(r.minutes, 2), "score": effective_score(r)} for r in record.rounds],
)
diff --git a/harness/tests/flywheel_cli/test_run_layout_cli.py b/harness/tests/flywheel_cli/test_run_layout_cli.py
index 33059282..c86aa6ef 100644
--- a/harness/tests/flywheel_cli/test_run_layout_cli.py
+++ b/harness/tests/flywheel_cli/test_run_layout_cli.py
@@ -123,6 +123,13 @@ def test_telemetry_summarises_the_runs_own_ledger(fake_run):
assert not (ws.telemetry / "usage.jsonl").exists(), "one ledger, one name"
+def test_telemetry_by_round_scores_a_degraded_verdict_as_unscored(fake_run):
+ ws, rec = fake_run
+ rec.rounds[1].judgment = rec.rounds[1].judgment.model_copy(update={"summary": "judge_error: outage"})
+ by_round = build_telemetry(ws, rec, write=False).cost.by_round
+ assert [r["score"] for r in by_round] == [rec.rounds[0].judgment.overall, None]
+
+
def test_the_settings_snapshot_records_the_temperature_the_track_planned_at(tmp_path):
"""The planner row reported ``planner.plan()``'s default 0.4 for every run — a default
no track uses: each passes its own ``plan_temperature``, and graphics plans at 0.5. The