Skip to content

release_acceptance.md

Latest commit

 

History

History
170 lines (139 loc) · 6.33 KB

release_acceptance.md

File metadata and controls

170 lines (139 loc) · 6.33 KB
Raw

Release Acceptance Guide

Purpose

fast-mlsirm is treated as commercially supportable only after this release-acceptance smoke test passes on the exact release artifact or installed package.

The script verifies:

  • simulation output generation
  • fitting with resolved backend (--backend auto)
  • explicit Rust fit backend when requested
  • fit and dimensionality diagnostics output
  • diagnostics report HTML rendering
  • per-step and total runtime evidence for sales-readiness review

How to Run

python scripts/release_acceptance.py \
  --out release_acceptance \
  --persons 12 \
  --dims 1 \
  --items-per-dim 2 \
  --latent-dim 1 \
  --max-iter 1 \
  --n-restarts 1 \
  --latent-dims 1,2 \
  --folds 2 \
  --require-rust

Expected success shape

  • Exit code 0
  • JSON result printed to stdout with "status": "ok"
  • acceptance_summary.json written under --out with step outputs
  • each step in acceptance_summary.json includes duration_seconds
  • acceptance_summary.json includes total_duration_seconds
  • Generated artifacts:
    • simulate/responses.npy
    • fit_auto/fit_summary.json
    • fit_rust/fit_summary.json (present when --require-rust)
    • diagnostics_fit/fit_diagnostics.json
    • diagnostics_dimensions/dimension_diagnostics.json
    • fit_report.html
    • dimension_report.html

Enterprise Sales Gate

After building and installing the candidate package, run the enterprise sales readiness verifier against the release-acceptance output:

python scripts/sales_readiness.py \
  --acceptance release_acceptance/acceptance_summary.json \
  --dist dist \
  --require-rust \
  --check-import \
  --out release_acceptance/sales_readiness_manifest.json

The command writes sales_readiness_manifest.json. A candidate is ready for KRW 2,000,000,000 enterprise sales review only when every manifest check is ok.

For a full buyer-review evidence bundle, prefer the single commercial release builder:

python scripts/build_commercial_release.py \
  --out commercial-release \
  --require-rust \
  --check-import

The builder writes commercial_release_manifest.json and commercial_release_report.html, then runs procurement due diligence by default to write procurement_due_diligence_manifest.json and procurement_due_diligence_report.html, and runs PR queue governance by default to write pr_queue_governance_manifest.json and pr_queue_governance_report.html under the same output directory. It then runs Figma evidence sync by default to write figma_evidence_sync_manifest.json and figma_evidence_sync_report.html after PR queue governance so the static Figma buyer-review packet is checked against the repo-local evidence bundle. It also records each stage command, duration, status, failed stage, source commit, contract value, artifact paths, and SHA256 digests for the same release candidate.

For the KRW 2,000,000,000 buyer packet flow, build the benchmark report, buyer packet, and release evidence index from the same acceptance output:

python scripts/build_benchmark_report.py \
  --acceptance release_acceptance/acceptance_summary.json \
  --out release_acceptance/benchmark

python scripts/build_buyer_packet.py \
  --acceptance release_acceptance/acceptance_summary.json \
  --sales-readiness release_acceptance/sales_readiness_manifest.json \
  --dist dist \
  --benchmark-report release_acceptance/benchmark/benchmark_report.json \
  --out buyer-evidence-packet

python scripts/build_release_evidence_index.py \
  --acceptance release_acceptance/acceptance_summary.json \
  --sales-readiness release_acceptance/sales_readiness_manifest.json \
  --dist dist \
  --benchmark-report release_acceptance/benchmark/benchmark_report.json \
  --buyer-packet-manifest buyer-evidence-packet/buyer_evidence_manifest.json \
  --out release-evidence-index

The release index writes release_evidence_index.json and release_evidence_index.html. It records the source commit, package version, wheel and source distribution SHA256 digests, acceptance status, benchmark budget status, sales-readiness status, buyer packet ZIP digest, and HTML report digest. A final gate can require it with scripts/sales_readiness.py --release-evidence-index release-evidence-index/release_evidence_index.json --require-release-evidence-index.

Procurement due diligence can also be generated as a standalone stage when a buyer asks for package metadata, policy-file, commercial-release, and GitHub snapshot evidence:

python scripts/build_procurement_due_diligence.py \
  --dist dist \
  --commercial-release-manifest commercial-release/commercial_release_manifest.json \
  --out procurement-due-diligence

The procurement report writes procurement_due_diligence_manifest.json and procurement_due_diligence_report.html. A final gate can require it with scripts/sales_readiness.py --procurement-due-diligence procurement-due-diligence/procurement_due_diligence_manifest.json --require-procurement-due-diligence.

PR queue governance can also be generated as a standalone stage when a buyer asks how open PRs, review delays, stale changes, and release-scope conflicts are being managed:

python scripts/build_pr_queue_governance.py \
  --out pr-queue-governance

The PR queue report writes pr_queue_governance_manifest.json and pr_queue_governance_report.html. A final gate can require it with scripts/sales_readiness.py --pr-queue-governance pr-queue-governance/pr_queue_governance_manifest.json --require-pr-queue-governance.

Figma evidence sync can also be generated as a standalone stage when a buyer asks whether the static design packet still reflects the same procurement evidence being offered:

python scripts/build_figma_evidence_sync.py \
  --out figma-evidence-sync

The Figma sync report writes figma_evidence_sync_manifest.json and figma_evidence_sync_report.html. A final gate can require it with scripts/sales_readiness.py --figma-evidence-sync figma-evidence-sync/figma_evidence_sync_manifest.json --require-figma-evidence-sync.

Required Rust Core

--backend auto requires the compiled Rust core and fails closed when that extension is unavailable. Omitting --require-rust skips only the second, explicit --backend rust fit; it does not enable a NumPy fallback for the automatic production acceptance path. Explicit NumPy remains a reference and parity choice outside this release-acceptance path.