feat(bench): RFC 0031 §7 — M_L2 unfrozen and asserting per RFC 0033 (run #21)

[jensholdgaard]

What

RFC 0031 §7: M_L2 unfrozen and asserting, per its named condition — the RFC 0033 cached template map landed (#522) and was measured (run #21: every pair warm at 187,904 B acquisition).

Frozen values, derived from the record and set below the weakest measurement (the existing freeze discipline):

• Processed channel (primary): M_L2 = 10. Measured 32.5–39.3× across runs feat(miner): add mask() — RFC 0001 §6.2 step 2 #10–feat(miner): add Drain prefix tree skeleton (RFC 0001 §6.2 step 3) #17 pre-artifact; 37.3–45.1× recomputed on the post-artifact totals.
• Storage-side: 1.1× floor, integer-exact (ourios × 11 ≤ loki_storage × 10). Post-artifact severity total = 2,223,171 B (0 + 2,035,267 + 187,904) against the recorded Loki storage band computes 1.20–1.51×; 1.1 sits below the weakest point with wobble margin. The amendment says plainly this is a parity-plus floor, not a 10× claim — write-side sizing remains the recorded, un-chased lever.

Also asserting from this slice:

• Scenario RFC0031.3 un-stubbed green (frozen-value pins, boundary math both directions, the §9.13 reproduction rows on the post-artifact totals, and the pre-artifact 1.05× row correctly failing the floor — the gate isn't vacuous).
• RFC 0033 §5.6's corpus acquisition gate (warm × 2 ≤ fold): harvested from the per-pair probes; in the all-warm steady state the fold comparand comes from one off-timed-path refold, loudly non-evaluable on failure, zero-bytes invalid-never-pass.
• print_pair_bytes_gates relabels L2 PRIMARY/FROZEN; new bytes_must_win_tenths in lgates.rs with boundary/overflow/zero unit tests.

✅ Measured — run #23 (29353634499), exit 0

First run with every new assertion live, all PASS:

• L2 processed (PRIMARY, frozen 10): 43.97×
• L2 storage-side floor (frozen 11/10): 1.49×
• L1 storage 108.3× / L3 storage 24.9× (frozen 10) — the existing gates held
• Latency floors held; RFC 0033 acquisition gate: warm 187,905 B compressed on every pair (≤ fold/2 asserted)
• The L3 pair measured cleanly (run docs(rfc-0001): amend §6.1 to align record schema with OTLP LogRecord #21's Loki-side flake did not recur)

⚠️ HOLD — measure before merge (discharged)

Comparative dispatch run #23 from this branch must pass with the new assertions live (the #506/run-#19 precedent). Numbers land here before merge.

Checks run

cargo fmt --all --check; workspace cargo clippy --all-targets --all-features -- -D warnings; cargo nextest run -p ourios-bench (149 passed, RFC0031.3 running green); mdbook build clean.

🤖 Generated with Claude Code

https://claude.ai/code/session_01WQY9wfrfRggqSpMLH8Xj3Y

Summary by CodeRabbit

• New Features

  • Added more precise tenths-based margin evaluation for comparative byte checks.
  • L2 evaluations now enforce both a processed-channel margin and a 1.1× storage-side floor.
  • Added template-map acquisition checks for warm and cold artifacts.

• Bug Fixes

  • Invalid or overflowing measurements are now rejected instead of passing evaluation.

• Documentation

  • Documented the finalized L2 thresholds and evaluation requirements.

[coderabbitai]

📝 Walkthrough

Walkthrough

The PR adds tenths-based L2 storage gating, freezes the L2 processed and storage-side thresholds, enables RFC0031.3, and integrates RFC0033.6 template-map acquisition validation into comparative runs.

Changes

Comparative gating

Layer / File(s)	Summary
Tenths-based L2 gate crates/ourios-bench/src/lgates.rs, crates/ourios-bench/src/lib.rs	Adds the tenths storage-floor configuration, integer-only gate evaluation with overflow invalidation, defaults and unit tests, and a public re-export.
Frozen L2 comparative integration crates/ourios-bench/tests/rfc0031_comparative.rs, docs/rfcs/0031-comparative-evaluation-loki.md	Enables RFC0031.3 and applies the frozen processed-channel and 1.1× storage-side L2 gates in comparative checks and output.
Template-map acquisition gate crates/ourios-bench/tests/rfc0031_comparative.rs	Adds RFC0033.6 warm-versus-cold acquisition validation, run-level failure aggregation, and coverage for evaluable, non-evaluable, and invalid measurements.

Estimated code review effort: 4 (Complex) | ~45 minutes

Sequence Diagram(s)

sequenceDiagram
  participant ComparativeRun
  participant FrozenGates
  participant TenthsGate
  participant AcquisitionGate
  participant GateFailures
  ComparativeRun->>FrozenGates: evaluate L2 pair
  FrozenGates->>TenthsGate: check storage-side floor
  TenthsGate-->>FrozenGates: return outcome
  ComparativeRun->>AcquisitionGate: compare warm artifact and cold fold bytes
  FrozenGates->>GateFailures: add L2 failures
  AcquisitionGate->>GateFailures: add acquisition failure when decisive

Loading

Possibly related PRs

• jensholdgaard/ourios#464: Earlier RFC0031 Section 7 freeze work connected to the L2 thresholds documented here.
• jensholdgaard/ourios#479: Adds the Loki storage-side fetched-bytes path used by these comparative gates.

Suggested reviewers: copilot

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Title check	✅ Passed	The title clearly and concisely summarizes the main change: unfreezing and asserting M_L2 for RFC 0031/0033.
Description check	✅ Passed	The description is detailed and covers the required areas, though it uses custom headings instead of the exact template.

✨ Finishing Touches

📝 Generate docstrings

• Create stacked PR
• Commit on current branch

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch rfc0031-ml2-unfreeze

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}

[Copilot]

Pull request overview

This PR updates the RFC 0031 comparative-benchmark harness and documentation to reflect that L2 (M_L2) is no longer deferred: it is now frozen and asserted (processed-channel must-win at M_L2 = 10, plus a storage-side parity-plus floor of 1.1× expressed as m_l2_storage_floor_tenths = 11). It also adds an asserting check for the RFC 0033 §5.6 corpus acquisition gate (warm × 2 ≤ fold) when any pair runs warm.

Changes:

• Un-stubs Scenario RFC0031.3 and pins/asserts the new frozen L2 processed + storage-floor gates.
• Adds bytes_must_win_tenths and the frozen margin m_l2_storage_floor_tenths to support integer-exact sub-integer gate math.
• Adds a dispatch-run assertion for the RFC0033.6 acquisition condition when warm measurements exist.

Reviewed changes

Copilot reviewed 4 out of 4 changed files in this pull request and generated 1 comment.

File	Description
docs/rfcs/0031-comparative-evaluation-loki.md	Documents the M_L2 unfreeze/freeze decision and the new processed + storage-floor frozen values.
crates/ourios-bench/tests/rfc0031_comparative.rs	Un-ignores RFC0031.3, adds L2 dual-channel gating, and asserts the RFC0033.6 acquisition gate in the dispatch run.
crates/ourios-bench/src/lib.rs	Re-exports the new bytes_must_win_tenths helper.
crates/ourios-bench/src/lgates.rs	Introduces m_l2_storage_floor_tenths and bytes_must_win_tenths with unit tests.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

Pull request overview

Copilot reviewed 4 out of 4 changed files in this pull request and generated no new comments.