OpenGait/.sisyphus/notepads/sconet-pipeline/issues.md

Task F4: Scope Fidelity Check — Issues (2026-02-27)

Non-compliance / drift items

1. opengait/demo/sconet_demo.py forward return contract drift: returns tensor label and tensor confidence instead of scalar int/float payload shape described in plan.
2. opengait/demo/window.py fill_level drift: implemented as integer count, while plan specifies len/window float ratio.
3. opengait/demo/output.py result schema drift: window serialized as list (create_result), but plan DoD schema states integer field.
4. opengait/demo/pipeline.py CLI drift: --source configured with default instead of required flag.
5. opengait/demo/pipeline.py behavior drift: no FPS logging loop (every 100 frames) found.
6. tests/demo/test_pipeline.py missing planned FPS benchmark scenario.
7. tests/demo/test_nats.py hardcodes NATS_PORT = 4222, conflicting with plan guidance to avoid hardcoded port in tests.

Scope creep / unexplained files

• Root-level unexplained artifacts present: EOF, LEOF, ENDOFFILE.

Must NOT Have guardrail status

• Guardrails mostly satisfied (no torch.distributed, no BaseModel in demo, no TensorRT/DeepStream, no GUI/multi-person logic); however overall scope verdict remains REJECT due to 7 functional/spec drifts above.

Blocker Fix: ScoNet checkpoint load mismatch (2026-02-27)

• Reproduced blocker with required smoke command: strict load failed with missing backbone.* / unexpected Backbone.forward_block.* (plus FCs.*, BNNecks.*).
• Root cause: naming convention drift between historical ScoNet checkpoint serialization and current ScoNetDemo module attribute names.
• Resolution: deterministic key normalization for known legacy prefixes while preserving strict load behavior and clear runtime error wrapping when incompatibility remains.

2026-02-27: Scope-Fidelity Drift Fix (F4) - Task 1 - FIXED

Issues Identified and Fixed

1. CLI --source not required (FIXED)

   • Location: Line 261 in opengait/demo/pipeline.py
   • Issue: --source had default="0" instead of being required
   • Fix: Changed to required=True
   • Impact: Users must now explicitly provide --source argument

2. Missing FPS logging (FIXED)

   • Location: run() method in opengait/demo/pipeline.py
   • Issue: No FPS logging in the main processing loop
   • Fix: Added frame counter and FPS logging every 100 frames
   • Impact: Users can now monitor processing performance

No Other Issues

• No type errors introduced
• No runtime regressions
• Error handling semantics preserved
• JSON output schema unchanged
• Window/predict logic unchanged [2026-02-27T00:44:25+08:00] Removed unexplained root files: EOF, LEOF, ENDOFFILE (scope-fidelity fix)

2026-02-27: NATS Port Fix - Type Narrowing Issue (FIXED)

Issue

• sock.getsockname() returns Any type causing basedpyright warning
• Previous fix with int() cast still leaked Any in argument position

Fix Applied

• Used typing.cast(tuple[str, int], sock.getsockname()) for explicit type narrowing
• Added intermediate variable with explicit type annotation

Verification

• uv run basedpyright tests/demo/test_nats.py: 0 errors, 0 warnings, 0 notes
• uv run pytest tests/demo/test_nats.py -q: 9 passed, 2 skipped

Files Modified

• tests/demo/test_nats.py only (line 29-30 in _find_open_port())

2026-02-27: Test Expectations Mismatch After fill_level Fix

After changing fill_level to return float ratio instead of integer count, 5 tests in tests/demo/test_window.py now fail due to hardcoded integer expectations:

1. test_window_fill_and_ready_behavior - expects fill_level == i + 1 (should be (i+1)/5)
2. test_underfilled_not_ready - expects fill_level == 9 (should be 0.9)
3. test_track_id_change_resets_buffer - expects fill_level == 5 (should be 1.0)
4. test_frame_gap_reset_behavior - expects fill_level == 5 (should be 1.0)
5. test_reset_clears_all_state - expects fill_level == 0 (should be 0.0)

These tests need updating to expect float ratios instead of integer counts.

2026-02-27: Test Assertions Updated for fill_level Ratio Contract

Status: Test file updated, pending runtime fix

Changes Made

Updated tests/demo/test_window.py assertions to expect float ratios (0.0..1.0) instead of integer frame counts:

Test	Old Assertion	New Assertion
test_window_fill_and_ready_behavior	== i + 1	== (i + 1) / 5
test_window_fill_and_ready_behavior	== 5	== 1.0
test_underfilled_not_ready	== 9	== 0.9
test_track_id_change_resets_buffer	== 5	== 1.0
test_track_id_change_resets_buffer	== 1	== 0.2
test_frame_gap_reset_behavior	== 5	== 1.0
test_frame_gap_reset_behavior	== 1	== 0.2
test_reset_clears_all_state	== 0	== 0.0

Blocker

Tests cannot pass until opengait/demo/window.py duplicate fill_level definition is removed (lines 208-210).

Verification Results

• basedpyright: 0 errors (18 pre-existing warnings unrelated to this change)
• pytest: 5 failed, 14 passed (failures due to window.py bug, not test assertions)

Task F4 Re-Audit: Remaining Issues (2026-02-27)

Status update for previous F4 drifts

Fixed:

• opengait/demo/pipeline.py source flag now required (line 268)
• opengait/demo/pipeline.py FPS logging present (lines 213-232)
• opengait/demo/window.py fill_level now ratio float (lines 205-207)
• tests/demo/test_nats.py dynamic port allocation via _find_open_port() (lines 24-31) and fixture-propagated port
• Root artifact files EOF, LEOF, ENDOFFILE removed (not found)

Still open:

1. Schema mismatch: opengait/demo/output.py:363 emits "window": list(window); plan DoD expects integer window field.
2. Missing planned FPS benchmark test: tests/demo/test_pipeline.py contains no FPS benchmark scenario from Task 12 plan section.
3. ScoNetDemo sequence contract drift in tests: tests/demo/test_sconet_demo.py:42,48 uses seq=16 fixtures, not the 30-frame window contract emphasized by plan.

Current re-audit verdict basis

• Remaining blockers: 3
• Scope state: not clean
• Verdict remains REJECT until these 3 are resolved or plan is amended by orchestrator.

2026-02-27T01:11:58+08:00 - Fixed: Sequence Length Drift in Test Fixtures

File: tests/demo/test_sconet_demo.py Issue: Fixtures used seq=16 but config specifies frames_num_fixed: 30 Fix: Updated dummy_sils_batch and dummy_sils_single fixtures to use seq=30 Status: ✅ Resolved - pytest passes (21/21), basedpyright clean (0 errors)

2026-02-27: Window Schema Fix - output.py (F4 Blocker) - FIXED

Issue: opengait/demo/output.py:363 emitted "window": list(window), conflicting with plan DoD schema expecting integer field.

Fix Applied:

• Type hint: window: int | tuple[int, int] (backward compatible input)
• Serialization: "window": window if isinstance(window, int) else window[1]
• Docstring examples updated to show integer format

Status: ✅ Resolved

• basedpyright: 0 errors
• pytest: 9 passed, 2 skipped

2026-02-27: Task 12 Pipeline Test Alignment - Issues

Initial Failure (expected RED phase)

• uv run pytest tests/demo/test_pipeline.py -q failed in happy-path and max-frames tests because _assert_prediction_schema still expected window as list[int, int] while runtime emits integer end-frame.
• Evidence: assertion failure assert isinstance(window_obj, list) with observed payload values like "window": 12.

Resolution

• Updated only tests/demo/test_pipeline.py schema assertion to require window as non-negative int.
• Added explicit FPS benchmark scenario with conservative threshold and CI stability guards.

Verification

• uv run pytest tests/demo/test_pipeline.py -q: 5 passed
• uv run basedpyright tests/demo/test_pipeline.py: 0 errors, 0 warnings, 0 notes

Task F4 Final Re-Audit: Issues Update (2026-02-27)

Previously open blockers now closed

1. opengait/demo/output.py window schema mismatch — CLOSED (line 364 now emits integer window).
2. tests/demo/test_pipeline.py missing FPS benchmark test — CLOSED (test_pipeline_cli_fps_benchmark_smoke, lines 109-167).
3. tests/demo/test_sconet_demo.py seq=16 fixtures — CLOSED (fixtures now seq=30 at lines 42,48).

Guardrail status

• opengait/demo/ has no torch.distributed usage and no BaseModel usage.
• Root artifact files EOF/LEOF/ENDOFFILE are absent.

Current issue count

• Remaining blockers: 0
• Scope issues: 0
• F4 verdict: APPROVE

Task: Fix NATS Test Schema and Port Mapping (2026-02-27)

Oracle-Reported Issues

1. Schema Validator Expected List, Runtime Emits Int

   • Location: _validate_result_schema in tests/demo/test_nats.py
   • Problem: Validator checked window as list[int] with length 2
   • Runtime: create_result in opengait/demo/output.py emits window as int
   • Root Cause: Test schema drifted from runtime contract
   • Fix: Updated validator to check isinstance(window, int) and window >= 0

2. Docker Port Mapping Incorrect

   • Location: _start_nats_container in tests/demo/test_nats.py (line 94)
   • Problem: Used -p {port}:{port} which mapped host port to same container port
   • NATS Container: Listens on port 4222 internally
   • Fix: Changed to -p {port}:4222 to map host dynamic port to container port 4222

Resolution

Both issues fixed in tests/demo/test_nats.py only. No runtime changes required.

Verification:

• basedpyright: 0 errors, 0 warnings
• pytest: 9 passed, 2 skipped (Docker unavailable)

Fix: Remove Stale Port Mapping (2026-02-27)

Bug: Duplicate port mappings in _start_nats_container caused Docker to receive invalid arguments. Resolution: Removed stale f"{port}:{port}" line, keeping only f"{port}:4222". Status: Fixed and verified.

Fix: Remove Duplicate Image Arg (2026-02-27)

Bug: Docker command had "nats:latest", "nats:latest" (duplicate). Resolution: Kept exactly one "nats:latest". Status: Fixed and verified.

Oracle Review #2 (2026-02-27): Residual Non-Blocking Issues

M1: Pending asyncio task warning (Minor)

• Location: opengait/demo/output.py:196
• Symptom: "Task was destroyed but it is pending!" on NATS connection failure
• Fix: Cancel in-flight coroutine in _stop_background_loop() before stopping event loop
• Impact: Cosmetic only

M2: Duplicate docstring line in create_result (Trivial)

• Location: opengait/demo/output.py:349-350
• Fix: Remove duplicate "Frame window [start, end]" line

M3: Incorrect label examples in create_result docstring (Minor)

• Location: opengait/demo/output.py:345
• Says "normal", "scoliosis" but labels are "negative", "neutral", "positive"
• Fix: Update docstring to match LABEL_MAP

2026-02-27: Workspace Hygiene Cleanup

Removed scope-creep artifacts from prior delegated runs:

• Deleted .sisyphus/notepads/demo-tensor-fix/ (entire folder)
• Deleted assets/sample.mp4

Repository no longer contains these untracked files.

Blocker: Task 11 Sample Video Acceptance Items (2026-02-27)

Status: BLOCKED - Pending user-provided sample video

Remaining unchecked acceptance criteria from Task 11:

1. ./assets/sample.mp4 (or .avi) exists
2. Video has ≥60 frames
3. Playable with OpenCV validation command

Unblock condition: Sample video file provided by user and all 3 criteria above pass validation.

Note: User explicitly stated they will provide sample video later; no further plan items remain outside these blocked sample-video checks.

Heartbeat Check (2026-02-27)

• Continuation check: 3 unchecked plan items remain
• Still no *.mp4/*.avi/*.mov/*.mkv files in repo
• Unblock condition: User-provided sample video with >=60 frames and OpenCV-readable

Fix: BBox/Mask Coordinate Mismatch (2026-02-27)

Issue

Demo pipeline produced no classifications for YOLO segmentation outputs because bbox and mask were in different coordinate spaces.

Resolution

Fixed in opengait/demo/window.py - select_person() now scales bbox from frame space to mask space using YOLO's orig_shape metadata.

Verification

• All tests pass (33 passed, 4 skipped)
• Smoke test on provided video yields 56 classifications from 60 frames
• Non-zero confidence values confirmed

Status

RESOLVED

2026-02-27: Sample Video Blocker Resolved

• Previous blocker (missing sample video) is no longer active.
• assets/sample.mp4 now present and playable via OpenCV with 227 frames.
• No remaining open checklist items in .sisyphus/plans/sconet-pipeline.md after checkbox update.