morluto/rea

Fork 0

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

[PR #3849] [CLOSED] fix(javascript): stop classifying replaced globals as built-in effects #5945

New issue

Closed

opened 2026-10-31 01:10:38 +00:00 by bsakaguchi · 0 comments

bsakaguchi commented

2026-10-31 01:10:38 +00:00

Owner

📋 Pull Request Information

Original PR: https://github.com/morluto/rea/pull/3849
Author: @kaoru0822-kitauji
Created: 10/15/2026
Status: ❌ Closed

Base: main ← Head: fix/3367-replaced-global-effects

📝 Commits (1)

3be4511 fix(javascript): stop classifying replaced globals as built-in effects

📊 Changes

4 files changed (+221 additions, -63 deletions)

View changed files

📝 src/domain/javascript/javascriptSemanticAmbientWrites.ts (+89 -18)
➕ src/domain/javascript/javascriptSemanticAnalysis.replaced_globals.test.ts (+57 -0)
📝 src/domain/javascript/javascriptSemanticAsyncEffects.ts (+38 -18)
📝 src/domain/javascript/javascriptSemanticDataEffects.ts (+37 -27)

📄 Description

Summary

Fixes #3367. Semantic timer, request, coercion and JSON.parse effects recognized unbound globals without consulting visible source-local writes. As a result, setTimeout = f, self.fetch = f or JSON.parse = f still produced complete Node timer, fetch and JSON parse operations.

Problem and expected behavior

Smallest trigger: setTimeout = f; setTimeout(work, 1) reports a complete setTimeout schedule. The violated invariant is that a built-in effect requires the callee to still denote the built-in. A visible exact overwrite breaks that identity. A write under an unknown key, such as globalThis[k] = v, may or may not break it, so the result must stay uncertain rather than becoming absent or complete.

Change and scope

Owner of the classification: javascriptSemanticAmbientWrites.ts, which owns ambient write facts, now classifies a global reference as one of:
- not global;
- unmodified;
- possibly replaced: the coverage index matches, but only through an unknown-key or widened write;
- replaced: an exact, all-literal written path equals the reference or member path.
Index: exact written paths are indexed once per ambient write list. The index is keyed on the list identity and its length.
Consumers: the timer, fetch/WebSocket, coercion and JSON.parse consumers use semanticBuiltinGlobalResolution:
- replaced → no built-in effect;
- possibly replaced → the effect is kept with partial resolution;
- unmodified → unchanged.
Removed: isUnmodifiedGlobalMember is superseded and removed. isUnmodifiedGlobalReference keeps its exact prior meaning (unmodified only) for its URL, Worker and undefined callers.
Not included: flow-sensitive ordering. A call before the overwrite is classified the same way as one after it, matching how the existing ambient-write coverage already behaves. Write paths the resolver widens, such as Object.defineProperties(globalThis, {...}), remain "possibly replaced". Making them precise is resolver work outside this fix.

Contract and boundary impact

Semantic owner and earliest changed stage: JavaScript semantic analysis, ambient global identity (javascriptSemanticAmbientWrites.ts).
Contribution scope fit and linked maintainer decision: an ordinary fix within existing boundaries, so not applicable.
CLI and MCP/tool-catalog contract: no schema change. resolution values stay within the existing complete | partial | unresolved enums.
Provider, bridge, target-format, or platform compatibility: none.
Evidence, artifact, provenance, or reconstruction contract: IR operations behind a definite overwrite are no longer emitted. Operations behind an uncertain write become partial. Function fingerprints whose effect components include removed requests change accordingly.
Target selection, authority, upstream behavior, and cleanup details: not applicable.
Generated metadata, package, or installation impact: none.

Expected effects:

Read existing files or captured data.

Evidence and regression coverage

Tests added: src/domain/javascript/javascriptSemanticAnalysis.replaced_globals.test.ts covers the replaced, possibly replaced (globalThis[Symbol.for(...)] = …) and unmodified cases. The replaced and possibly replaced cases fail on the base revision, which reports all three effects as complete.
Base reproduction (executed; base = main build at f09f454f4 vs. this branch, with setTimeout(work, 1); fetch(...); JSON.parse("1") after each prefix):

Prefix	main	this branch
none	setTimeout:complete, fetch:complete, JSON.parse:complete	unchanged
`setTimeout = f; fetch = f; JSON.parse = f`	all three complete	`parse:partial` only (generic parse)
`globalThis.setTimeout = f`	all three complete	fetch:complete, JSON.parse:complete
`delete globalThis.setTimeout`	all three complete	fetch:complete, JSON.parse:complete
`globalThis[k] = v`	all three complete	all three partial

Real-app semantic IR comparison (executed): full analyzeParsedJavaScriptSemantics output for every JS file, main build vs. this change.

Corpus	Files	Identical	Differing	Semantic ms (main → branch)
Discord	1	1	0	35 → 33
Notion	96	96	0	8,347 → 8,206
Signal	572	569	3	22,251 → 21,881
Claude	421	418	3	38,379 → 37,574
Tabby	2,183	2,183	0	11,154 → 10,854
VS Code `out/`	27	25	2	19,194 → 19,155

In the 8 differing files, every difference was classified per operation:

1,152 changed from complete to partial: 859 coercion/JSON.parse boundary operations and 293 timer operations. Each sits behind an unknown-key global write: globalThis[e] = …, a Symbol-keyed registry write re[k] ??= new Map(), or Object.defineProperties(globalThis, …).
11 fetch requests and their 13 json/text response consumers were removed, all in VS Code. They sit behind genuine overwrites: globalThis.fetch = async function… in extensionHostProcess.js and self.fetch = function… in extensionHostWorkerMain.js.
10 function fingerprints changed in the same VS Code file. Their effect components follow the removed requests.
No other IR field changed.
Output size: removed operations shrink the IR slightly. Resolution changes are equal-length string swaps, apart from complete → partial being 1 byte shorter.
Remaining proof gaps: no CLI/MCP workflow output was captured separately. The IR comparison exercises the producer that both adapters consume. Timing figures are single warm runs on one machine.

For evidence-bearing changes:

Observed, derived, and inferred claims remain distinguishable.
Artifact identity, source provenance, and failed attempts remain preserved.
Unsupported, incomplete, unavailable, or uncertain outcomes remain visible.

Validation performed

npm run test:local -- src/domain/javascript/javascriptSemanticAnalysis.replaced_globals.test.ts: 3 passed.
161 JavaScript/domain test files via npm run test:local: 2,985 passed (after rebasing onto 788964d52).
npm run check:fast: typecheck and lint pass.
The semantic IR comparison above on Discord, Notion, Signal, Claude, Tabby and VS Code bundles.

Compatibility, safety, and release

Breaking changes or migration steps: none. Consumers already handle partial resolutions.
Real Hopper/Ghidra, browser, or OS coverage: not applicable (static JavaScript analysis).
Package or release metadata impact: none.
Security, privacy, process, or containment review: no target interaction. Private app bundles are referenced only by name and path.

Author readiness

Human owner:

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.
Owning docs, contracts, and generated metadata are updated where needed.
User-visible CLI/MCP changes include representative output.
I checked the final diff for secrets, unrelated cleanup, and unsupported claims.

🤖 Generated with Claude Code

_{🔄 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/3849 **Author:** [@kaoru0822-kitauji](https://github.com/kaoru0822-kitauji) **Created:** 10/15/2026 **Status:** ❌ Closed **Base:** `main` ← **Head:** `fix/3367-replaced-global-effects` --- ### 📝 Commits (1) - [`3be4511`](https://github.com/morluto/rea/commit/3be45116ab262dd2edde504975ef4fb48107f2ef) fix(javascript): stop classifying replaced globals as built-in effects ### 📊 Changes **4 files changed** (+221 additions, -63 deletions) <details> <summary>View changed files</summary> 📝 `src/domain/javascript/javascriptSemanticAmbientWrites.ts` (+89 -18) ➕ `src/domain/javascript/javascriptSemanticAnalysis.replaced_globals.test.ts` (+57 -0) 📝 `src/domain/javascript/javascriptSemanticAsyncEffects.ts` (+38 -18) 📝 `src/domain/javascript/javascriptSemanticDataEffects.ts` (+37 -27) </details> ### 📄 Description ## Summary Fixes #3367. Semantic timer, request, coercion and `JSON.parse` effects recognized unbound globals without consulting visible source-local writes. As a result, `setTimeout = f`, `self.fetch = f` or `JSON.parse = f` still produced complete Node timer, fetch and JSON parse operations. ## Problem and expected behavior Smallest trigger: `setTimeout = f; setTimeout(work, 1)` reports a complete `setTimeout` schedule. The violated invariant is that a built-in effect requires the callee to still denote the built-in. A visible exact overwrite breaks that identity. A write under an unknown key, such as `globalThis[k] = v`, may or may not break it, so the result must stay uncertain rather than becoming absent or complete. ## Change and scope - **Owner of the classification:** `javascriptSemanticAmbientWrites.ts`, which owns ambient write facts, now classifies a global reference as one of: - not global; - unmodified; - possibly replaced: the coverage index matches, but only through an unknown-key or widened write; - replaced: an exact, all-literal written path equals the reference or member path. - **Index:** exact written paths are indexed once per ambient write list. The index is keyed on the list identity and its length. - **Consumers:** the timer, fetch/WebSocket, coercion and `JSON.parse` consumers use `semanticBuiltinGlobalResolution`: - replaced → no built-in effect; - possibly replaced → the effect is kept with `partial` resolution; - unmodified → unchanged. - **Removed:** `isUnmodifiedGlobalMember` is superseded and removed. `isUnmodifiedGlobalReference` keeps its exact prior meaning (`unmodified` only) for its URL, Worker and `undefined` callers. - **Not included:** flow-sensitive ordering. A call before the overwrite is classified the same way as one after it, matching how the existing ambient-write coverage already behaves. Write paths the resolver widens, such as `Object.defineProperties(globalThis, {...})`, remain "possibly replaced". Making them precise is resolver work outside this fix. ## Contract and boundary impact - Semantic owner and earliest changed stage: JavaScript semantic analysis, ambient global identity (`javascriptSemanticAmbientWrites.ts`). - Contribution scope fit and linked maintainer decision: an ordinary fix within existing boundaries, so not applicable. - CLI and MCP/tool-catalog contract: no schema change. `resolution` values stay within the existing `complete | partial | unresolved` enums. - Provider, bridge, target-format, or platform compatibility: none. - Evidence, artifact, provenance, or reconstruction contract: IR operations behind a definite overwrite are no longer emitted. Operations behind an uncertain write become `partial`. Function fingerprints whose effect components include removed requests change accordingly. - Target selection, authority, upstream behavior, and cleanup details: not applicable. - Generated metadata, package, or installation impact: none. Expected effects: - [x] Read existing files or captured data. ## Evidence and regression coverage - Tests added: `src/domain/javascript/javascriptSemanticAnalysis.replaced_globals.test.ts` covers the replaced, possibly replaced (`globalThis[Symbol.for(...)] = …`) and unmodified cases. The replaced and possibly replaced cases fail on the base revision, which reports all three effects as complete. - Base reproduction (executed; base = main build at f09f454f4 vs. this branch, with `setTimeout(work, 1); fetch(...); JSON.parse("1")` after each prefix): | Prefix | main | this branch | |---|---|---| | none | setTimeout:complete, fetch:complete, JSON.parse:complete | unchanged | | `setTimeout = f; fetch = f; JSON.parse = f` | all three complete | `parse:partial` only (generic parse) | | `globalThis.setTimeout = f` | all three complete | fetch:complete, JSON.parse:complete | | `delete globalThis.setTimeout` | all three complete | fetch:complete, JSON.parse:complete | | `globalThis[k] = v` | all three complete | all three partial | - Real-app semantic IR comparison (executed): full `analyzeParsedJavaScriptSemantics` output for every JS file, main build vs. this change. | Corpus | Files | Identical | Differing | Semantic ms (main → branch) | |---|---:|---:|---:|---:| | Discord | 1 | 1 | 0 | 35 → 33 | | Notion | 96 | 96 | 0 | 8,347 → 8,206 | | Signal | 572 | 569 | 3 | 22,251 → 21,881 | | Claude | 421 | 418 | 3 | 38,379 → 37,574 | | Tabby | 2,183 | 2,183 | 0 | 11,154 → 10,854 | | VS Code `out/` | 27 | 25 | 2 | 19,194 → 19,155 | In the 8 differing files, every difference was classified per operation: - 1,152 changed from `complete` to `partial`: 859 coercion/`JSON.parse` boundary operations and 293 timer operations. Each sits behind an unknown-key global write: `globalThis[e] = …`, a Symbol-keyed registry write `re[k] ??= new Map()`, or `Object.defineProperties(globalThis, …)`. - 11 `fetch` requests and their 13 `json`/`text` response consumers were removed, all in VS Code. They sit behind genuine overwrites: `globalThis.fetch = async function…` in `extensionHostProcess.js` and `self.fetch = function…` in `extensionHostWorkerMain.js`. - 10 function fingerprints changed in the same VS Code file. Their effect components follow the removed requests. - No other IR field changed. - Output size: removed operations shrink the IR slightly. Resolution changes are equal-length string swaps, apart from `complete` → `partial` being 1 byte shorter. - Remaining proof gaps: no CLI/MCP workflow output was captured separately. The IR comparison exercises the producer that both adapters consume. Timing figures are single warm runs on one machine. For evidence-bearing changes: - [x] Observed, derived, and inferred claims remain distinguishable. - [x] Artifact identity, source provenance, and failed attempts remain preserved. - [x] Unsupported, incomplete, unavailable, or uncertain outcomes remain visible. ## Validation performed - `npm run test:local -- src/domain/javascript/javascriptSemanticAnalysis.replaced_globals.test.ts`: 3 passed. - 161 JavaScript/domain test files via `npm run test:local`: 2,985 passed (after rebasing onto 788964d52). - `npm run check:fast`: typecheck and lint pass. - The semantic IR comparison above on Discord, Notion, Signal, Claude, Tabby and VS Code bundles. ## Compatibility, safety, and release - Breaking changes or migration steps: none. Consumers already handle `partial` resolutions. - Real Hopper/Ghidra, browser, or OS coverage: not applicable (static JavaScript analysis). - Package or release metadata impact: none. - Security, privacy, process, or containment review: no target interaction. Private app bundles are referenced only by name and path. ## Author readiness Human owner:  - [ ] 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. - [ ] Owning docs, contracts, and generated metadata are updated where needed. - [ ] User-visible CLI/MCP changes include representative output. - [ ] I checked the final diff for secrets, unrelated cleanup, and unsupported claims. 🤖 Generated with [Claude Code](https://claude.com/claude-code) --- <sub>🔄 This issue represents a GitHub Pull Request. It cannot be merged through Gitea due to API limitations.</sub>

bsakaguchi

2026-10-31 01:10:38 +00:00

closed this issue
added the
pull-request
label

bsakaguchi referenced this issue

2026-11-01 21:45:48 +00:00

[PR #6141] [CLOSED] fix(snapshot): keep mismatch, output-loss, and declaration require diagnostics precise #7157

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#5945

No description provided.

Rows
Columns