Skip to content

test: regenerate placement solver pins #334

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Draft
SamGu-NRX wants to merge 2 commits into
base: gpt-burst/1022z-bs-port-11
Choose a base branch
from
Draft

test: regenerate placement solver pins #334

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
2cabf47
test: regenerate placement solver pins
SamGu-NRX Oct 23, 2026
5384d3b
test: refresh placement output pins
SamGu-NRX Oct 23, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
28 changes: 14 additions & 14 deletions backend/server/placement/tests/data/capture_cases_v1/manifest.json
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,7 +1,7 @@
{
"schema_version": "placement-case-manifest/v0",
"description": [
"Burst8 supported-case packet. Each positive case pairs a schema-valid synthetic base-scene/v0 source (metres, hand-authored) with an independently authored root-wire scene (feet, hand-derived as metres divided by 0.3048 and rounded to six decimals) — the same wall, on-wall meter anchor, observed wall_rect hazard, level ground, coverage gap, and chain ends, described twice by construction, never produced by running the converter.",
"Burst8 supported-case packet. Each positive case pairs a schema-valid synthetic base-scene/v0 source (metres, hand-authored) with an independently authored root-wire scene (feet, hand-derived as metres divided by 0.3048 and rounded to six decimals) \u2014 the same wall, on-wall meter anchor, observed wall_rect hazard, level ground, coverage gap, and chain ends, described twice by construction, never produced by running the converter.",
"Positive cases run the unchanged top-level solver through POST /v1/placements on the wire scene and on the bridge's conversion of the source scene; the two responses must agree except stats.elapsed_ms (timing) and stats.input_sha256 plus response.input_sha256 (the posted bytes differ: the wire file carries a schema_version envelope field the bridge does not emit). They must also match these expectations, derived from the authored dimensions: the spot is the last failing start pushed to the 3 ft gas-clearance line (gas edge + 3.0), gas_clearance is unsure by margin (measured exactly 3.0 ft against 0.590552 ft of uncertainty), the ground gap makes ground_surface unsure-unobserved, and the unexplored end within reach asks for a walk past it. The merged ground band's six-decimal extent is the buffer union over the authored gaps and the clearance radii of rules.yaml; its span and out_ft are pinned as the serialized geometric consequence of the authored dimensions.",
"Negative cases are source-only: the bridge must refuse them with the declared error code and pointer. multiwall-source-unsupported and generated-object-evidence-unsupported are newly authored to be clean of every other unsupported feature, so the declared refusal is the scene's first one. The other three preserve W7's documented bridge limitations: standoff-meter-anchor-unsupported is a byte-identical copy of the W7 hazard-right-ground-gap source, generated-regions-unsupported a byte-identical copy of the W7 generated-hazard source, and footprint-hazard-unsupported is the W7 right source with only the meter's standoff_m field removed so the footprint hazard is the first unsupported feature left (its hash pins the derived bytes).",
"Every fixture path is relative to this manifest's directory and pinned by sha256; a missing or drifted file fails the suite with its path."
Expand Down Expand Up @@ -53,10 +53,10 @@
"kind": "band",
"band": "ground",
"span_ft": [
-3.307693,
8.495
-3.603325,
8.4951
],
"out_ft": 12.141026
"out_ft": 12.431758
},
{
"kind": "past_end",
Expand All @@ -70,8 +70,8 @@
"sweep_failing_runs": [
{
"start_ft": [
2.846457,
5.725066
3.141733,
5.42979
],
"failing": [
"gas_clearance",
Expand All @@ -80,8 +80,8 @@
},
{
"start_ft": [
5.74147,
6.069554
5.446194,
6.36483
],
"failing": [
"gas_clearance"
Expand All @@ -93,7 +93,7 @@
"right": "unexplored"
}
},
"note": "Geometry facts, not policy: the wall runs s [-1.5, 10.0] ft with a limit left end and an unexplored right end; the gas meter at s [3.0, 4.0] ft pushes the best spot to the 3 ft clearance line (s0 = 4.0 + 3.0 = 7.0, measured 3.0 ft +- 0.590552, unsure by margin, not by observation); every start closer to the meter either overlaps the working space or the meter (estimate-failing) and every start past the clearance line has unsure ground too but a longer route. The wall_rect vent obj-tap-vent (the replay runner's TAP_HAZARD geometry) sits at s [5.249344, 5.905512] ft +- 0.164042: the chosen spot clears it (gap 1.094488 ft), while the pinned sweep runs overlap the gas meter and the vent — the second run ends exactly at the vent's tolerance line 5.905512 + 0.164042 = 6.069554. Ground is observed only at [-1.5, 3.0] and [8.5, 10.0] ft out 13, so the spot's ground is unsure-unobserved and the response must request the ground band and a walk past the right end."
"note": "Geometry facts, not policy: the wall runs s [-1.5, 10.0] ft with a limit left end and an unexplored right end; the gas meter at s [3.0, 4.0] ft pushes the best spot to the 3 ft clearance line (s0 = 4.0 + 3.0 = 7.0, measured 3.0 ft +- 0.590552, unsure by margin, not by observation); every start closer to the meter either overlaps the working space or the meter (estimate-failing) and every start past the clearance line has unsure ground too but a longer route. The wall_rect vent obj-tap-vent (the replay runner's TAP_HAZARD geometry) sits at s [5.249344, 5.905512] ft +- 0.164042: the chosen spot clears it (gap 1.094488 ft), while the pinned sweep runs overlap the gas meter and the vent \u2014 the second run ends exactly at the vent's tolerance line 5.905512 + 0.164042 = 6.069554. Ground is observed only at [-1.5, 3.0] and [8.5, 10.0] ft out 13, so the spot's ground is unsure-unobserved and the response must request the ground band and a walk past the right end."
},
{
"id": "hazard-left-ground-gap",
Expand Down Expand Up @@ -143,9 +143,9 @@
"band": "ground",
"span_ft": [
-10.0,
-1.004999
3.603325
],
"out_ft": 12.141026
"out_ft": 12.431758
},
{
"kind": "past_end",
Expand All @@ -160,7 +160,7 @@
{
"start_ft": [
-8.975557,
-4.458333
-4.441109
],
"failing": [
"gas_clearance"
Expand All @@ -169,7 +169,7 @@
{
"start_ft": [
-0.590552,
0.621391
0.326115
],
"failing": [
"meter_working_space"
Expand Down Expand Up @@ -223,7 +223,7 @@
"code": "placement_scene_unsupported",
"pointer": "/objects/0/geometry/standoff_m"
},
"note": "Awaiting conversion: a byte-identical copy of the W7 hazard-right-ground-gap source, preserved unchanged so the declared refusal is the scene's actual first one — the stand-off meter anchor the bridge does not synthesise, at /objects/0/geometry/standoff_m. The suite never runs the missing bridge, and this case is never reported as a successful conversion."
"note": "Awaiting conversion: a byte-identical copy of the W7 hazard-right-ground-gap source, preserved unchanged so the declared refusal is the scene's actual first one \u2014 the stand-off meter anchor the bridge does not synthesise, at /objects/0/geometry/standoff_m. The suite never runs the missing bridge, and this case is never reported as a successful conversion."
},
{
"id": "footprint-hazard-unsupported",
Expand Down
22 changes: 11 additions & 11 deletions backend/server/placement/tests/data/cases/manifest.json
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -49,10 +49,10 @@
"kind": "band",
"band": "ground",
"span_ft": [
-4.427701,
-4.728063,
20.0
],
"out_ft": 13.761035
"out_ft": 14.056496
},
{
"kind": "past_end",
Expand All @@ -66,8 +66,8 @@
"sweep_failing_runs": [
{
"start_ft": [
2.708333,
5.541667
2.694445,
5.270833
],
"failing": [
"gas_clearance"
Expand Down Expand Up @@ -98,8 +98,8 @@
"wall_id": "side",
"segment": 0,
"span_ft": [
2.708333,
5.291667
2.694445,
5.277779
],
"outcome": "unsure"
},
Expand All @@ -124,10 +124,10 @@
"kind": "band",
"band": "ground",
"span_ft": [
2.004999,
16.451777
2.004899,
16.736026
],
"out_ft": 12.993443
"out_ft": 13.28668
},
{
"kind": "past_end",
Expand All @@ -141,8 +141,8 @@
"sweep_failing_runs": [
{
"start_ft": [
-6.958333,
-5.291667
-8.958333,
-5.277779
],
"failing": [
"gas_clearance"
Expand Down
4 changes: 2 additions & 2 deletions backend/tests/test_capture_placement_cases.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -427,7 +427,7 @@ def _expected_a() -> dict:
"route_length": {"outcome": "pass", "unsure_cause": None},
},
"missing_evidence": [
{"kind": "band", "band": "ground", "span_ft": [-0.614621, 22.321313], "out_ft": 12.009634},
{"kind": "band", "band": "ground", "span_ft": [-0.680316, 22.387009], "out_ft": 12.07043},
{"kind": "past_end", "side": "left", "span_ft": [-9.84252, -9.84252]},
],
"ends": {"left": "unexplored", "right": "limit"},
Expand All @@ -452,7 +452,7 @@ def _expected_b() -> dict:
"route_length": {"outcome": "pass", "unsure_cause": None},
},
"missing_evidence": [
{"kind": "band", "band": "ground", "span_ft": [-23.666497, -0.664867], "out_ft": 12.042482},
{"kind": "band", "band": "ground", "span_ft": [-23.765041, -0.566324], "out_ft": 12.136125},
{"kind": "past_end", "side": "right", "span_ft": [9.84252, 9.84252]},
],
"ends": {"left": "limit", "right": "unexplored"},
Expand Down