From 04ea29fad142dcebe50600e99ba4a3f1bf43d7d0 Mon Sep 17 00:00:00 2001
From: jeqcho <jeqin_chooi@college.harvard.edu>
Date: Tue, 14 Jul 2026 10:39:03 -0700
Subject: [PATCH] test: sweep blueprints at runtime seeds; make task versions
 explicit

Post-merge cleanups from the #24 review:

- test_sim_blueprint's sweep paired scenes with derive_seed(0, index, epoch),
  seeds that no longer occur at runtime now that init_seed is crc32-derived.
  Sweep with derive_seed(0, scene.init_seed, epoch) instead. Pinned tests keep
  their fixed seed literals (they check blueprint/realization consistency at a
  seed, any seed); the one comment claiming those literals were real epochs is
  reworded.

- TaskSpec.version's default was bumped to "2" in #24, which would mint future
  new tasks at version 2. Restore the default to "1" and set version="2"
  explicitly on the nine existing tasks (scoop_pasta already pins "3").
  Realized version stamps are unchanged.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
---
 src/kitchenbench/specs.py   | 11 ++++++++++-
 tests/test_sim_blueprint.py |  6 ++++--
 2 files changed, 14 insertions(+), 3 deletions(-)

diff --git a/src/kitchenbench/specs.py b/src/kitchenbench/specs.py
index 2b1cf37..afbacca 100644
--- a/src/kitchenbench/specs.py
+++ b/src/kitchenbench/specs.py
@@ -34,7 +34,7 @@ class TaskSpec:
     bimanual: bool
     max_steps: int
     instances: tuple[TaskInstance, ...]
-    version: str = "2"
+    version: str = "1"
     description: str = ""
 
 
@@ -1341,6 +1341,7 @@ def _sort_fixtures(
         bimanual=False,
         max_steps=60,
         instances=_PLACE_CUTLERY,
+        version="2",
         description="Pick a single piece of cutlery and place it on a target surface.",
     ),
     TaskSpec(
@@ -1350,6 +1351,7 @@ def _sort_fixtures(
         bimanual=False,
         max_steps=80,
         instances=_STACK,
+        version="2",
         description="Stack multiple like items into a single neat stack.",
     ),
     TaskSpec(
@@ -1359,6 +1361,7 @@ def _sort_fixtures(
         bimanual=False,
         max_steps=80,
         instances=_PLACE_IN_RACK,
+        version="2",
         description="Drop a dish into the correct slot of a dish rack.",
     ),
     TaskSpec(
@@ -1368,6 +1371,7 @@ def _sort_fixtures(
         bimanual=True,
         max_steps=100,
         instances=_POUR_PASTA,
+        version="2",
         description="Pour dry pasta into a receiving vessel; one arm steadies, the other pours.",
     ),
     TaskSpec(
@@ -1377,6 +1381,7 @@ def _sort_fixtures(
         bimanual=True,
         max_steps=120,
         instances=_OPEN_CONTAINER,
+        version="2",
         description="Remove or unscrew a lid — one arm braces while the other twists or pries.",
     ),
     TaskSpec(
@@ -1386,6 +1391,7 @@ def _sort_fixtures(
         bimanual=True,
         max_steps=120,
         instances=_FOLD_CLOTH,
+        version="2",
         description="Deformable manipulation: grasp opposite corners and manage slack.",
     ),
     TaskSpec(
@@ -1395,6 +1401,7 @@ def _sort_fixtures(
         bimanual=True,
         max_steps=120,
         instances=_SEAL_CONTAINER,
+        version="2",
         description="Align and press-or-twist a matching lid onto a base while one arm holds it.",
     ),
     TaskSpec(
@@ -1404,6 +1411,7 @@ def _sort_fixtures(
         bimanual=True,
         max_steps=80,
         instances=_HANDOFF,
+        version="2",
         description="A pure handover that a single arm cannot do — the must-use-both-arms anchor.",
     ),
     TaskSpec(
@@ -1413,6 +1421,7 @@ def _sort_fixtures(
         bimanual=False,
         max_steps=200,
         instances=_SORT_CUTLERY,
+        version="2",
         description="Sort a mixed pile into spoon/fork/knife compartments — multi-instance.",
     ),
     TaskSpec(
diff --git a/tests/test_sim_blueprint.py b/tests/test_sim_blueprint.py
index e0f2f59..b1856f5 100644
--- a/tests/test_sim_blueprint.py
+++ b/tests/test_sim_blueprint.py
@@ -44,7 +44,9 @@ def _all_scene_seed_pairs() -> list[tuple[str, object, int]]:
     for spec in SPEC_BY_KEY.values():
         for index, scene in enumerate(build_scenes(spec)):
             for epoch in range(3):
-                pairs.append((f"{spec.key}[{index}]e{epoch}", scene, derive_seed(0, index, epoch)))
+                pairs.append(
+                    (f"{spec.key}[{index}]e{epoch}", scene, derive_seed(0, scene.init_seed, epoch))
+                )
     return pairs
 
 
@@ -208,7 +210,7 @@ def test_pinned_handoff_receiving_arms() -> None:
 
 def test_pinned_size_order_mixed_sizes_both_orders() -> None:
     scene = _scene("stack", 3)  # mixed-sizes: size_order sampled
-    # Epoch 0 samples "largest_first", epoch 1 samples "shuffled" (pinned).
+    # These fixed seeds sample "largest_first" and "shuffled" (pinned).
     seed_first = derive_seed(0, 3, 0)
     values_first = realize_scene(scene, seed_first).values  # type: ignore[arg-type]
     assert values_first["size_order"] == "largest_first"