Skip to content

Official Codebase of 3DCodeVerse

Python

scene: the one-world session is the default — confirmed under the fixed judge once it is told its window (D70)

gaoypengf1b5017d94e9
diff --git a/harness/codeverse/config.py b/harness/codeverse/config.py
index 61a13570..99678d42 100644
--- a/harness/codeverse/config.py
+++ b/harness/codeverse/config.py
@@ -179,8 +179,8 @@ SEED_RECIPES_ENV = "CV3D_SEED_RECIPES"
#: run, and whether it earns that is a measurement nobody has made yet.
SCENE_TEXTURES_ENV = "CV3D_SCENE_TEXTURES"
#: ONE session authors every zone of a scene (the whole world, one author) instead of the
-#: two-small-zones-per-session fan-out. OFF: the loop-17 gain did not survive the fixed-judge
-#: confirmation (D70); an A/B knob.
+#: two-small-zones-per-session fan-out. ON since the fixed-judge confirmation of the session
+#: that is told its window and asked for full density (D70); `off` is the fan-out control.
ONE_WORLD_SESSION_ENV = "CV3D_ONE_WORLD_SESSION"
_TRUE_WORDS = frozenset({"1", "on", "true", "yes", "y"})
_FALSE_WORDS = frozenset({"0", "off", "false", "no", "n"})
@@ -233,7 +233,7 @@ def scene_textures_enabled() -> bool:
def one_world_session_enabled() -> bool:
- """Do all of a scene's zones go to ONE session? ``$CV3D_ONE_WORLD_SESSION``, else off.
+ """Do all of a scene's zones go to ONE session? ``$CV3D_ONE_WORLD_SESSION``, else ON.
Measured 2026-09-08 under one fixed judge (gemini-3.1-pro-preview, n=2, the brief's
must_have list), both sides ``codex:gpt-6-astra@low``: a bare one-file scene — one author,
@@ -247,12 +247,15 @@ def one_world_session_enabled() -> bool:
goes to one session that owns every zone file (the env, the assets and the heroes keep
their own stages), with a session window scaled to the zone count. Measured on the
same three briefs (loop 17, in-loop judge): round 0 0.17 → 0.69, 0.51 → 0.58, 0.44 →
- 0.43; best round 0.75 → 0.82 PASS, 0.73 → 0.60, 0.60 → 0.77 PASS. The fixed-judge
- confirmation (cmp_scene4, both arms astra-low) did NOT reproduce it: harness 0.60 / 0.60 /
- 0.54 vs the bare one-shot 0.81 / 0.83 / 0.60, the single session finishing four zones in
- five minutes as a block-out. Off by default; a knob for the next measurement.
+ 0.43; best round 0.75 → 0.82 PASS, 0.73 → 0.60, 0.60 → 0.77 PASS. The first fixed-judge
+ confirmation (cmp_scene4) did NOT reproduce it — 0.60 / 0.60 / 0.54 vs the bare one-shot
+ 0.81 / 0.83 / 0.60 — because that session finished four zones in five minutes as a
+ block-out. With the batch header stating the window and asking for each zone's full
+ density (loop 18: round 0 0.81 / 0.60 / 0.88, two passes), the second confirmation
+ (cmp_scene5, fixed judge) reads harness 0.89 / 0.89 (both PASS at round 0) vs one-shot
+ 0.92 / 0.77 on boat / clockmaker. On by default; ``off`` is the fan-out control.
"""
- return env_flag(ONE_WORLD_SESSION_ENV, False)
+ return env_flag(ONE_WORLD_SESSION_ENV, True)
def seed_recipes_enabled() -> bool:
diff --git a/harness/docs/DECISIONS.md b/harness/docs/DECISIONS.md
index afcd7f84..ac3940d1 100644
--- a/harness/docs/DECISIONS.md
+++ b/harness/docs/DECISIONS.md
@@ -1113,8 +1113,8 @@ written) that were accepted because the code works that way and the tests pin it
grounds anything (L7): harness-owned starter geometry, like `worldShell` (D51), that the
agent edits.
-* **D70 One author for the whole world — `CV3D_ONE_WORLD_SESSION`, an A/B knob, OFF: its
- in-loop gain did not survive the fixed judge (2026-09-08).** Under one fixed judge (gemini-3.1-pro-preview, n=2, the brief's
+* **D70 One author for the whole world — `CV3D_ONE_WORLD_SESSION`, ON by default since the
+ fixed-judge confirmation of the session that is told its window (2026-09-08).** Under one fixed judge (gemini-3.1-pro-preview, n=2, the brief's
must_have list) and one model on both sides (`codex:gpt-6-astra@low`), a bare one-file scene
— one author, no plan, no lib, no assets, no tools — scored 0.894 on the clockmaker's
workshop and 0.82 on the boat workshop; the harness's round 0 of the same briefs, its zones
@@ -1138,11 +1138,14 @@ written) that were accepted because the code works that way and the tests pin it
session finished four zones in five minutes as a block-out ("missing stove", "shelves
missing", "primitive tools", 0.32 at round 0) where loop 17's draw had reached 0.69 — one
author is coherent but thin, the fan-out dense but incoherent, and n=1 per arm per brief is
- noise-sized. The knob stays OFF; the batch header now states the session's window and asks
- for each zone's full density (measured next as loop 18). What the whole evening says
- instead: the bare one-file scene wins round 0 on interiors every time (0.81–0.89), so the
- harness's next experiment is to CONTAIN it — a whole-scene single-shot draft as round 0,
- then its gates, heroes and refine rounds on top (the bench's `oneshot+refine` shape).
+ noise-sized. The batch header then states the session's window and asks for each zone's
+ full density. **Loop 18 (in-loop):** round 0 clockmaker 0.81 PASS, boat 0.60 → 0.77 PASS,
+ NYC 0.88 PASS; the session spends 10–19 min instead of 5. **Second confirmation
+ (`cmp_scene5`, fixed judge, both arms astra-low):** harness 0.89 / 0.89, both PASS at round
+ 0, vs the bare one-shot 0.92 / 0.77 on boat / clockmaker (the re-drawn one-shots move by
+ about ±0.1 between draws). The default is ON; `off` keeps the fan-out and its two batching
+ constants as the control. The bare one-file scene is no longer ahead on interiors; the
+ harness keeps its heroes, gates, library and rounds on top of that.
## Rejected / deferred
diff --git a/harness/tests/orchestrator_tracks/test_one_world_session.py b/harness/tests/orchestrator_tracks/test_one_world_session.py
index 3286410e..5436f36b 100644
--- a/harness/tests/orchestrator_tracks/test_one_world_session.py
+++ b/harness/tests/orchestrator_tracks/test_one_world_session.py
@@ -19,11 +19,11 @@ from tests.orchestrator_tracks.conftest import make_spec
from tests.orchestrator_tracks.fakes import FakeRuntime, FakeServices
-def test_the_switch_is_read_at_call_time_and_off_by_default(monkeypatch):
+def test_the_switch_is_read_at_call_time_and_on_by_default(monkeypatch):
monkeypatch.delenv("CV3D_ONE_WORLD_SESSION", raising=False)
- assert one_world_session_enabled() is False # D70: the loop-17 gain did not survive the fixed judge
- monkeypatch.setenv("CV3D_ONE_WORLD_SESSION", "on")
- assert one_world_session_enabled() is True
+ assert one_world_session_enabled() is True # D70: confirmed under the fixed judge with the density header
+ monkeypatch.setenv("CV3D_ONE_WORLD_SESSION", "off")
+ assert one_world_session_enabled() is False # the fan-out control
monkeypatch.setenv("CV3D_ONE_WORLD_SESSION", "garbage")
assert one_world_session_enabled() is False # a typo in a bench command is the control, never a crash