2.10 CSV input modes (pairs vs wide)
Pairs (robust). Explicit label,phi pairs; tolerant to order; handles missing labels.
# Header
label,phi,label,phi,label,phi,label,phi
# Example row
day,120.0,week,40.0,lunar,275.0,solar,15.0
- Pros: resilient to column order; easy to extend; partial rows acceptable.
- Cons: slightly larger files; parsing needs pair-walking.
Wide (fast). Fixed columns named by labels; must align with manifest channel order.
# Header
day,week,lunar,solar
# Example row
120.0,40.0,275.0,15.0
- Pros: compact and fastest to parse; ideal for streaming.
- Cons: strict column order; missing labels must be imputed or the row rejected.
Headers required. Always include a header line to avoid parser ambiguity.
ASCII-only. Use UTF-8 or ASCII without BOM; decimal dot . only.
2.11 Packaging and snapshotting (one-click, optional)
Goal. Zip the exact acceptance artifacts for provenance (CSV + logs), with a timestamp.
Windows batch (example)
@echo off
setlocal
cd /d "%~dp0"
if not exist dist mkdir dist
for /f %%A in ('powershell -NoProfile -Command "Get-Date -Format yyyyMMdd_HHmmss"') do set TS=%%A
powershell -NoProfile -Command ^
"Compress-Archive -Force -DestinationPath ('dist\\SSM-Clock_Acceptance_'+$env:TS+'.zip') -Path 'notes\\longbench_summary.csv'"
echo WROTE dist\SSM-Clock_Acceptance_%TS%.zip
endlocal
Provenance. Zip the same CSV cited in the blog/release; optionally add logs and per-seed detail.
Repeatability. Keep timestamped zips under dist/ to track runs over time.
Determinism. Do not modify artifacts after packaging; regenerate if parameters change.
Navigation
Back: SSM-Clock—Formats, Determinism & Performance (2.7–2.9)
Next: SSM-Clock—Acceptance Dataset & Results (3.1–3.2)
SHUNYAYA 