morluto/rea

Fork 0

mirror of https://github.com/morluto/rea.git synced 2026-11-05 06:45:22 +00:00

[PR #4035] [MERGED] fix(javascript): preserve mapped runtime digest disagreements #6006

New issue

Closed

opened 2026-10-31 01:11:07 +00:00 by bsakaguchi · 0 comments

bsakaguchi commented

2026-10-31 01:11:07 +00:00

Owner

📋 Pull Request Information

Original PR: https://github.com/morluto/rea/pull/4035
Author: @N0zoM1z0
Created: 10/15/2026
Status: ✅ Merged
Merged: 10/15/2026
Merged by: @N0zoM1z0

Base: main ← Head: fix/runtime-reconciliation-path-digest-3316

📝 Commits (2)

34f628e fix(javascript): preserve mapped runtime digest disagreements
9a52df0 docs(javascript): clarify captured-source matching scope

📊 Changes

5 files changed (+316 additions, -84 deletions)

View changed files

📝 docs/javascript-runtime-reconciliation.md (+7 -4)
📝 src/domain/javascript/javascriptRuntimeLoadState.ts (+9 -4)
📝 src/domain/javascript/javascriptRuntimeReconciliationMatching.ts (+69 -53)
➕ tests/boundary/filesystem/javascript-applications/javascriptRuntimeContentPriority.test.ts (+231 -0)
📝 tests/boundary/filesystem/javascript-applications/javascriptRuntimeReconciliation.test.ts (+0 -23)

📄 Description

Summary

Fixes #3316. A caller mapping to a/main.js (value 1) no longer reports a high-confidence match to b/main.js just because the captured bytes contain value 2. It reports the mapped-file content disagreement and retains the selected candidate reference.

Problem and expected behavior

Global digest fallback ran before the mapped-path disagreement check. It could hide a mismatch behind either a unique match elsewhere or unrelated ambiguous matches. A known mapped candidate must own the comparison.

Change and scope

Compare captured source against mapped candidates first; only use the global digest index when no mapped candidate exists. Mapped ambiguity includes only qualified mapped matches.
Keep unavailable mapped digests unknown, and preserve those candidate references through static load-state classification instead of claiming absence.
Reuse parsed candidate facts and the per-call index. This removes the per-entity scan of an unrelated global same-digest bucket; no persistent cache or extra tool call is introduced.

Existing read-only reconciliation scope and Evidence contracts are preserved. Successful correspondence remains inferred and does not prove execution. No new provider or target interaction capability is introduced.

Known affected behavior: [x] read existing files/captured data. Scope follows contribution scope.

Evidence and regression coverage

At 6688ed6, the ten new cases produce six genuine failures and four passing controls. They cover automatic/explicit file mappings, zero/one/multiple alternate matches, qualified ambiguity, missing digests and digest-only fallback when the mapped path has no candidate. Replaces the narrower old mismatch case.
The original public Linux CLI reproduction now returns a+2 → unmatched, b+2 → content-and-location match, a+3 → unmatched. Those captures are synthetic passive Electron Evidence over real static files.
Real Chrome for Testing 156.0.8078.4, owned local HTTP fixture, actual CDP source capture: mapping the captured module to a reports disagreement; mapping to b matches. Public CLI and stdio MCP normalized results are identical. Replaying the same actual browser Evidence on parent 337d2e3 reproduces the incorrect unique-content match. Input bytes remain unchanged; the owned browser/server and MCP transport close successfully.
A distinct missing-digest boundary case additionally proved the downstream false-absence behavior before the load-state fix. It models an unavailable parsed digest; it does not claim a real provider produced that case.

Evidence assertions: [x] observation/derivation/inference remain distinct; [x] identities and provenance retained; [x] incomplete outcomes stay visible.

Validation performed

55 relevant cases across six reconciliation, filesystem/process, CLI and MCP files. All passed before the unrelated main refresh. On runtime commit 34f628e, the combined run passed 54/55 with one existing MCP comprehensive case hitting its 20-second timeout under shared-host load. Its complete file then passed 5/5 in isolation with unchanged limits (the affected case took 8.95 seconds). The failed attempt is retained; no timeout threshold was changed.
npm run check — passed on runtime commit 34f628e: typecheck, lint, formatting, unused-code, metadata, module boundaries and architecture guards.
npm run docs:check — passed on current documentation, including five documentation consistency tests.
Actual browser/CLI/stdio-MCP probe above — passed. Windows/macOS and other real providers were not run for this read-only JavaScript matcher change.

Measured completed CLI task

Parent 337d2e3 versus runtime commit 34f628e, identical actual-browser Evidence and correct b mapping. Every normalized output is equal; full retained JSON is 17,357 cl100k_base tokens in both versions. Median of two alternating samples per mode, baseline / fixed:

Node compile cache	Wall (s)	Reaped child CPU (s)	Sampled peak family RSS (MiB)
cold	1.97 / 1.84	2.28 / 2.18	332.9 / 329.5
warm	1.86 / 1.75	2.18 / 2.07	330.4 / 341.4

Linux x86_64, pinned Node 24.18.0/npm 11.16.0; two CPU cores, at most two test workers, serial heavy tasks and an 8 GiB overall local budget. Cold disables Node compile caching; warm primes a per-version cache. OS caches and shared-host load are uncontrolled. RSS is sampled at 200 ms and can double-count shared pages/miss short children; reported wall time includes up to 200 ms observer delay. These small samples do not establish an isolated speedup. No model consumed these payloads; complete agent-task tokens, input/schema negotiation and billing are unmeasured. All local guards stayed below budget; peak measured validation family RSS was approximately 4.46 GiB.

Compatibility and release

Existing call/configuration/schema contracts remain intact. Observable corrections are explicit mapped mismatches, narrower qualified ambiguity, and unknown load states for missing digests. Source-map authority and complete retained Evidence remain available. No package/release metadata change.

Latest head: 9a52df0 adds only the captured-source qualifier to the matching documentation; runtime code and the measured behavior remain identical to 34f628e. npm run docs:check passed again after that clarification.

Author readiness

Human owner: @N0zoM1z0

The human owner explicitly authorized delegated review, proxy completion of these declarations, ready-for-review publication and normal merge for this batch. The agent reviewed the diff and evidence described above; this records that delegation and does not assert a separate manual review occurred.

I reviewed the final diff and relevant existing code, understand the change and its edge cases, and take responsibility for the complete PR, including AI-assisted work.
I checked the evidence above, confirmed it supports the current change, and recorded meaningful validation gaps.
I declare this PR ready for initial maintainer review.

Before merge

This PR addresses a concrete problem or an agreed enhancement.
Material scope and authority questions are resolved, and any required maintainer scope decision covers the final diff.
The PR has one focused outcome and the title follows type(scope): outcome.
Related issue is linked, or the reason for not linking one is stated above.
Tests cover changed observable behavior and meaningful failure paths.

_{🔄 This issue represents a GitHub Pull Request. It cannot be merged through Gitea due to API limitations.}

## 📋 Pull Request Information **Original PR:** https://github.com/morluto/rea/pull/4035 **Author:** [@N0zoM1z0](https://github.com/N0zoM1z0) **Created:** 10/15/2026 **Status:** ✅ Merged **Merged:** 10/15/2026 **Merged by:** [@N0zoM1z0](https://github.com/N0zoM1z0) **Base:** `main` ← **Head:** `fix/runtime-reconciliation-path-digest-3316` --- ### 📝 Commits (2) - [`34f628e`](https://github.com/morluto/rea/commit/34f628efe075b19b948882c3507d6b9cc250926d) fix(javascript): preserve mapped runtime digest disagreements - [`9a52df0`](https://github.com/morluto/rea/commit/9a52df0a939ccb3233cede21fda0a7aeb13e932d) docs(javascript): clarify captured-source matching scope ### 📊 Changes **5 files changed** (+316 additions, -84 deletions) <details> <summary>View changed files</summary> 📝 `docs/javascript-runtime-reconciliation.md` (+7 -4) 📝 `src/domain/javascript/javascriptRuntimeLoadState.ts` (+9 -4) 📝 `src/domain/javascript/javascriptRuntimeReconciliationMatching.ts` (+69 -53) ➕ `tests/boundary/filesystem/javascript-applications/javascriptRuntimeContentPriority.test.ts` (+231 -0) 📝 `tests/boundary/filesystem/javascript-applications/javascriptRuntimeReconciliation.test.ts` (+0 -23) </details> ### 📄 Description ## Summary Fixes #3316. A caller mapping to `a/main.js` (value 1) no longer reports a high-confidence match to `b/main.js` just because the captured bytes contain value 2. It reports the mapped-file content disagreement and retains the selected candidate reference. ## Problem and expected behavior Global digest fallback ran before the mapped-path disagreement check. It could hide a mismatch behind either a unique match elsewhere or unrelated ambiguous matches. A known mapped candidate must own the comparison. ## Change and scope - Compare captured source against mapped candidates first; only use the global digest index when no mapped candidate exists. Mapped ambiguity includes only qualified mapped matches. - Keep unavailable mapped digests unknown, and preserve those candidate references through static load-state classification instead of claiming absence. - Reuse parsed candidate facts and the per-call index. This removes the per-entity scan of an unrelated global same-digest bucket; no persistent cache or extra tool call is introduced. Existing read-only reconciliation scope and Evidence contracts are preserved. Successful correspondence remains inferred and does not prove execution. No new provider or target interaction capability is introduced. Known affected behavior: [x] read existing files/captured data. Scope follows [contribution scope](https://github.com/morluto/rea/blob/main/CONTRIBUTING.md#contribution-scope). ## Evidence and regression coverage - At `6688ed6`, the ten new cases produce **six genuine failures and four passing controls**. They cover automatic/explicit file mappings, zero/one/multiple alternate matches, qualified ambiguity, missing digests and digest-only fallback when the mapped path has no candidate. Replaces the narrower old mismatch case. - The original public Linux CLI reproduction now returns `a+2 → unmatched`, `b+2 → content-and-location match`, `a+3 → unmatched`. Those captures are synthetic passive Electron Evidence over real static files. - **Real Chrome for Testing 156.0.8078.4**, owned local HTTP fixture, actual CDP source capture: mapping the captured module to `a` reports disagreement; mapping to `b` matches. Public CLI and **stdio MCP** normalized results are identical. Replaying the same actual browser Evidence on parent `337d2e3` reproduces the incorrect unique-content match. Input bytes remain unchanged; the owned browser/server and MCP transport close successfully. - A distinct missing-digest boundary case additionally proved the downstream false-absence behavior before the load-state fix. It models an unavailable parsed digest; it does not claim a real provider produced that case. Evidence assertions: [x] observation/derivation/inference remain distinct; [x] identities and provenance retained; [x] incomplete outcomes stay visible. ## Validation performed - **55 relevant cases across six reconciliation, filesystem/process, CLI and MCP files.** All passed before the unrelated main refresh. On runtime commit `34f628e`, the combined run passed 54/55 with one existing MCP comprehensive case hitting its 20-second timeout under shared-host load. Its complete file then passed **5/5** in isolation with unchanged limits (the affected case took 8.95 seconds). The failed attempt is retained; no timeout threshold was changed. - `npm run check` — passed on runtime commit `34f628e`: typecheck, lint, formatting, unused-code, metadata, module boundaries and architecture guards. - `npm run docs:check` — passed on current documentation, including five documentation consistency tests. - Actual browser/CLI/stdio-MCP probe above — passed. Windows/macOS and other real providers were not run for this read-only JavaScript matcher change. ### Measured completed CLI task Parent `337d2e3` versus runtime commit `34f628e`, identical actual-browser Evidence and correct `b` mapping. Every normalized output is equal; full retained JSON is **17,357 cl100k_base tokens** in both versions. Median of two alternating samples per mode, **baseline / fixed**: | Node compile cache | Wall (s) | Reaped child CPU (s) | Sampled peak family RSS (MiB) | | --- | --- | --- | --- | | cold | 1.97 / 1.84 | 2.28 / 2.18 | 332.9 / 329.5 | | warm | 1.86 / 1.75 | 2.18 / 2.07 | 330.4 / 341.4 | Linux x86_64, pinned Node 24.18.0/npm 11.16.0; two CPU cores, at most two test workers, serial heavy tasks and an 8 GiB overall local budget. Cold disables Node compile caching; warm primes a per-version cache. OS caches and shared-host load are uncontrolled. RSS is sampled at 200 ms and can double-count shared pages/miss short children; reported wall time includes up to 200 ms observer delay. These small samples do not establish an isolated speedup. No model consumed these payloads; complete agent-task tokens, input/schema negotiation and billing are unmeasured. All local guards stayed below budget; peak measured validation family RSS was approximately 4.46 GiB. ## Compatibility and release Existing call/configuration/schema contracts remain intact. Observable corrections are explicit mapped mismatches, narrower qualified ambiguity, and unknown load states for missing digests. Source-map authority and complete retained Evidence remain available. No package/release metadata change. Latest head: `9a52df0` adds only the captured-source qualifier to the matching documentation; runtime code and the measured behavior remain identical to `34f628e`. `npm run docs:check` passed again after that clarification. ## Author readiness Human owner: @N0zoM1z0 The human owner explicitly authorized delegated review, proxy completion of these declarations, ready-for-review publication and normal merge for this batch. The agent reviewed the diff and evidence described above; this records that delegation and does not assert a separate manual review occurred. - [x] I reviewed the final diff and relevant existing code, understand the change and its edge cases, and take responsibility for the complete PR, including AI-assisted work. - [x] I checked the evidence above, confirmed it supports the current change, and recorded meaningful validation gaps. - [x] I declare this PR ready for initial maintainer review. ## Before merge - [x] This PR addresses a concrete problem or an agreed enhancement. - [x] Material scope and authority questions are resolved, and any required maintainer scope decision covers the final diff. - [x] The PR has one focused outcome and the title follows `type(scope): outcome`. - [x] Related issue is linked, or the reason for not linking one is stated above. - [x] Tests cover changed observable behavior and meaningful failure paths. --- <sub>🔄 This issue represents a GitHub Pull Request. It cannot be merged through Gitea due to API limitations.</sub>

bsakaguchi

2026-10-31 01:11:07 +00:00

closed this issue
added the
pull-request
label

bsakaguchi referenced this issue

2026-11-01 21:45:30 +00:00

[PR #6006] [MERGED] fix(process): reject finite trace variants that only reorder unordered members #7116

No milestone

No project

No assignees

1 participant

Notifications

Due date

The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set.

Reference

morluto/rea#6006

No description provided.

Rows
Columns