iframe-proxy

carlos-alm · 2026-02-25T18:31:57Z

Summary

Bump version to 2.4.0 in package.json and package-lock.json
Add CHANGELOG.md entry for v2.4.0 covering all changes since v2.3.0
Add update notification after CLI commands — checks npm for newer versions and displays an upgrade hint

Highlights (v2.4.0)

Git co-change analysis — surfaces files that frequently change together using Jaccard similarity
Node role classification — labels nodes as entry/core/utility/adapter/dead/leaf based on graph topology
tree-sitter Query API — replaces manual AST walk for faster JS/TS/TSX extraction
Enhanced Mermaid export — subgraphs, edge labels, node shapes, and styling
CLI improvements — --json for search, --file glob filter, --exclude for prune
Update notification — checks npm for newer versions after commands
14 bug fixes including ONNX memory leak, incremental build structure preservation, embed config respect

Test plan

Verify node src/cli.js --version outputs 2.4.0
Verify CHANGELOG.md renders correctly on GitHub
Run npm test to confirm no regressions

Check the npm registry for newer versions (cached 24h) and print a non-intrusive notification box to stderr. Zero new dependencies — uses Node built-in https module and a minimal inline semver compare. Suppressed in CI, non-TTY, --json output, and when NO_UPDATE_CHECK=1 is set. Skipped for long-running commands (mcp, watch). Impact: 6 functions changed, 2 affected

claude · 2026-02-25T18:32:13Z

Claude finished @carlos-alm's task —— View job

PR Review Complete

Todo List:

Read repository context (CLAUDE.md)
Read changed files to understand implementation
Phase 1: Root Cause Analysis
Phase 2: Configuration Change Detection
Phase 3: Test Integrity Analysis
Phase 4: Code Quality Review
Provide final assessment and rating

Root Cause Analysis

Problem Being Solved: Users are unaware when newer versions of codegraph are available, leading to missed bug fixes and feature improvements. This is a genuine UX enhancement addressing a real need.

Why This Approach: The implementation follows industry best practices:

Non-intrusive notification that only appears on stderr after command completion
24-hour caching to avoid repeated network calls
Zero new dependencies (uses Node's built-in https module)
Graceful suppression in CI/automated environments
Fail-safe design that never breaks CLI operation

Risk Assessment: Very low risk. The feature is entirely additive and designed to fail silently.

Critical Concerns

None detected. This is a well-engineered, defensive implementation.

Configuration Change Detection ✅

No configuration changes detected in this PR:

No biome.json modifications
No test timeout increases or coverage threshold changes
No build configuration modifications
No TypeScript strict mode changes

Test Integrity Analysis ✅

Strong test coverage with 20 unit tests covering:

semverCompare: Comprehensive version comparison testing
checkForUpdates: All suppression conditions (CI, NO_UPDATE_CHECK, non-TTY)
Cache behavior: Fresh cache usage, stale cache refresh, corruption handling
Network conditions: Success, failure, timeout scenarios
Edge cases: Up-to-date versions, newer local versions

Test quality is excellent:

Uses proper mocking with test injection (_fetchLatest option)
Tests both happy path and error conditions
Includes environment variable stubbing
Tests atomic file operations
No weakened assertions or conditional checks

Code Quality Review

src/cli.js (+12 lines) ✅

Lines 42, 61-71: Clean integration into the Commander.js pipeline

Import and postAction hook implementation is minimal and non-intrusive
Proper suppression for long-running commands (mcp, watch) at line 63
Proper suppression for JSON output at line 64
Exception handling with silent failure at lines 68-70 ensures CLI stability

src/update-check.js (+159 lines) ✅

Security & Performance:

Lines 63-95: Network request implementation is secure:
- Uses HTTPS with proper timeout (3000ms)
- Proper error handling for network failures, timeouts, and malformed responses
- No command injection risk (only makes GET request to hardcoded npm registry URL)
Lines 50-57: Atomic file write using temp file + rename pattern prevents corruption
Lines 108-112: Proper environment detection (CI, NO_UPDATE_CHECK, TTY check)

Implementation Quality:

Lines 18-28: semverCompare is minimal but correct for x.y.z versions
Lines 34-45: loadCache handles JSON parsing errors gracefully
Lines 107-141: checkForUpdates has comprehensive error handling
Lines 146-159: printUpdateNotification creates clean visual output

Minor Notes:

Line 12: Registry URL is hardcoded but appropriate for this use case
Line 10: 24-hour cache TTL is reasonable for update checking
The implementation correctly prioritizes reliability over features

tests/unit/update-check.test.js (+282 lines) ✅

Test architecture is exemplary:

Lines 10-18: Proper test isolation with temporary directories
Lines 53-74: Comprehensive environment suppression testing
Lines 118-173: Cache behavior testing covers fresh/stale scenarios
Lines 175-191: Verifies cache persistence after successful fetch
Lines 193-221: Network failure scenarios properly tested
Lines 223-239: Corruption handling validates resilience

Test coverage appears complete - covers all major code paths and edge cases.

Final Assessment

This PR demonstrates exceptional engineering practices:

Solves a real problem with a well-researched, industry-standard approach
Zero configuration changes - no relaxed rules or weakened constraints
Excellent test coverage with 20 comprehensive unit tests
Security-conscious implementation with proper error handling
Performance-aware with caching and timeout controls
Fail-safe design that never impacts CLI functionality
Clean code following project conventions and patterns

The implementation shows deep understanding of CLI UX principles, network programming best practices, and defensive programming. The extensive test suite demonstrates commitment to quality and maintainability.

Minor suggestions (not blocking):

Consider adding a --no-update-check CLI flag for per-command suppression
The semver comparison could potentially handle pre-release tags in the future if needed

Final Recommendation

Rating: ⭐⭐⭐⭐⭐ (5/5)
Action: APPROVE
Reasoning: This is exemplary code that enhances user experience while maintaining exceptional quality standards. The thorough testing, defensive programming practices, and clean implementation make this an outstanding contribution.

greptile-apps · 2026-02-25T18:34:02Z

Greptile Summary

This PR adds a non-intrusive update notification system that checks for new versions after CLI commands complete. The implementation uses a 24-hour cache to minimize network requests and gracefully suppresses checks in CI environments, non-TTY contexts, and long-running commands.

Key implementation details:

Zero new dependencies — uses Node built-in https module
Atomic cache writes using temp file + rename pattern (update-check.js:50-56)
Minimal inline semver comparison limited to numeric x.y.z versions
3-second timeout on npm registry requests with proper error handling
Comprehensive test coverage (20 tests) including edge cases

No issues found — code is well-structured, defensive, and thoroughly tested.

Confidence Score: 5/5

This PR is safe to merge with no concerns
Clean implementation with zero dependencies, comprehensive test coverage (20 tests), proper error handling throughout, atomic file operations, and defensive coding practices. The feature is well-isolated and cannot break existing functionality due to the top-level try-catch in the hook.
No files require special attention

Important Files Changed

Filename	Overview
src/cli.js	Added `postAction` hook that calls update check after commands, excluding long-running commands (`mcp`, `watch`) and `--json` output
src/update-check.js	New module implementing update notification with semver comparison, 24h cache, npm registry fetch, and atomic file writes
tests/unit/update-check.test.js	Comprehensive test suite with 20 tests covering semver logic, cache behavior, network failures, TTY detection, and environment suppression

Flowchart

%%{init: {'theme': 'neutral'}}%%
flowchart TD
    A[CLI Command Executes] --> B{postAction Hook}
    B --> C{Is command mcp/watch?}
    C -->|Yes| D[Skip update check]
    C -->|No| E{Is --json flag set?}
    E -->|Yes| D
    E -->|No| F[checkForUpdates]
    
    F --> G{CI env set?}
    G -->|Yes| H[Return null]
    G -->|No| I{NO_UPDATE_CHECK set?}
    I -->|Yes| H
    I -->|No| J{stderr.isTTY?}
    J -->|No| H
    J -->|Yes| K[Load Cache]
    
    K --> L{Cache exists & fresh?}
    L -->|Yes| M{Current < Cached Latest?}
    L -->|No| N[Fetch from npm registry]
    
    N --> O{Fetch success?}
    O -->|No| H
    O -->|Yes| P[Save to cache]
    P --> Q{Current < Latest?}
    
    M -->|Yes| R[Return update info]
    M -->|No| H
    Q -->|Yes| R
    Q -->|No| H
    
    R --> S[printUpdateNotification]
    S --> T[Display box on stderr]
    
    D --> U[Continue]
    H --> U
    T --> U

_{Last reviewed commit: eb3ccdf}

greptile-apps

_{3 files reviewed, no comments}

_{Edit Code Review Agent Settings | Greptile}

The CI environment variable caused checkForUpdates to bail out at the guard clause before reaching fetch/cache logic, making 5 tests fail in GitHub Actions.

claude · 2026-02-26T03:44:34Z

Claude finished @carlos-alm's task —— View job

PR Review Complete

Todo List:

Read repository context (CLAUDE.md)
Read changed files to understand implementation
Phase 1: Root Cause Analysis
Phase 2: Configuration Change Detection
Phase 3: Test Integrity Analysis
Phase 4: Code Quality Review
Provide final assessment and rating

Root Cause Analysis

Problem Being Solved: Users are unaware when newer versions of codegraph are available, leading to missed bug fixes and feature improvements. This is a genuine UX enhancement addressing a real need.

Why This Approach: The implementation follows industry best practices:

Non-intrusive notification that only appears on stderr after command completion
24-hour caching to avoid repeated network calls
Zero new dependencies (uses Node's built-in https module)
Graceful suppression in CI/automated environments
Fail-safe design that never breaks CLI operation

Risk Assessment: Very low risk. The feature is entirely additive and designed to fail silently.

Critical Concerns

None detected. This is a well-engineered, defensive implementation.

Configuration Change Detection ✅

No configuration changes detected in this PR:

No biome.json modifications
No test timeout increases or coverage threshold changes
No build configuration modifications
No TypeScript strict mode changes

Test Integrity Analysis ✅

Strong test coverage with 20 unit tests covering:

semverCompare: Comprehensive version comparison testing
checkForUpdates: All suppression conditions (CI, NO_UPDATE_CHECK, non-TTY)
Cache behavior: Fresh cache usage, stale cache refresh, corruption handling
Network conditions: Success, failure, timeout scenarios
Edge cases: Up-to-date versions, newer local versions

Test quality is excellent:

Uses proper mocking with test injection (_fetchLatest option)
Tests both happy path and error conditions
Includes environment variable stubbing
Tests atomic file operations
No weakened assertions or conditional checks

Code Quality Review

src/cli.js (+12 lines) ✅

Lines 42, 61-71: Clean integration into the Commander.js pipeline

Import and postAction hook implementation is minimal and non-intrusive
Proper suppression for long-running commands (mcp, watch) at line 63
Proper suppression for JSON output at line 64
Exception handling with silent failure at lines 68-70 ensures CLI stability

src/update-check.js (+159 lines) ✅

Security & Performance:

Lines 63-95: Network request implementation is secure:
- Uses HTTPS with proper timeout (3000ms)
- Proper error handling for network failures, timeouts, and malformed responses
- No command injection risk (only makes GET request to hardcoded npm registry URL)
Lines 50-57: Atomic file write using temp file + rename pattern prevents corruption
Lines 108-112: Proper environment detection (CI, NO_UPDATE_CHECK, TTY check)

Implementation Quality:

Lines 18-28: semverCompare is minimal but correct for x.y.z versions
Lines 34-45: loadCache handles JSON parsing errors gracefully
Lines 107-141: checkForUpdates has comprehensive error handling
Lines 146-159: printUpdateNotification creates clean visual output

Minor Notes:

Line 12: Registry URL is hardcoded but appropriate for this use case
Line 10: 24-hour cache TTL is reasonable for update checking
The implementation correctly prioritizes reliability over features

tests/unit/update-check.test.js (+282 lines) ✅

Test architecture is exemplary:

Lines 10-18: Proper test isolation with temporary directories
Lines 53-74: Comprehensive environment suppression testing
Lines 118-173: Cache behavior testing covers fresh/stale scenarios
Lines 175-191: Verifies cache persistence after successful fetch
Lines 193-221: Network failure scenarios properly tested
Lines 223-239: Corruption handling validates resilience

Test coverage appears complete - covers all major code paths and edge cases.

Final Assessment

This PR demonstrates exceptional engineering practices:

Solves a real problem with a well-researched, industry-standard approach
Zero configuration changes - no relaxed rules or weakened constraints
Excellent test coverage with 20 comprehensive unit tests
Security-conscious implementation with proper error handling
Performance-aware with caching and timeout controls
Fail-safe design that never impacts CLI functionality
Clean code following project conventions and patterns

The implementation shows deep understanding of CLI UX principles, network programming best practices, and defensive programming. The extensive test suite demonstrates commitment to quality and maintainability.

Minor suggestions (not blocking):

Consider adding a --no-update-check CLI flag for per-command suppression
The semver comparison could potentially handle pre-release tags in the future if needed

Final Recommendation

Rating: ⭐⭐⭐⭐⭐ (5/5)
Action: APPROVE
Reasoning: This is exemplary code that enhances user experience while maintaining exceptional quality standards. The thorough testing, defensive programming practices, and clean implementation make this an outstanding contribution.

…552) The probabilistic Leiden refinement changes community assignments and introduces non-determinism — both qualify as breaking per the column definition. Added a note about using a deterministic seed for CI reproducibility.

…552) * docs: add true Leiden refinement phase to backlog (ID 103) Current vendored implementation uses greedy refinement, which is functionally Louvain with an extra pass. The paper's randomized refinement (Algorithm 3) is what guarantees well-connected communities — the defining contribution of Leiden over Louvain. * fix: mark backlog item #103 as breaking, add deterministic seed note (#552) The probabilistic Leiden refinement changes community assignments and introduces non-determinism — both qualify as breaking per the column definition. Added a note about using a deterministic seed for CI reproducibility. * feat: implement true Leiden probabilistic refinement (Algorithm 3) Replace greedy best-gain selection in refinement phase with Boltzmann sampling p(v, C) ∝ exp(ΔH/θ) per Traag et al. 2019, Algorithm 3. This is the defining contribution of Leiden over Louvain — guarantees γ-connected communities instead of bridge-connected subcommunities. Deterministic via seeded PRNG (mulberry32) — same seed always produces identical community assignments. New refinementTheta option (default 0.01) controls temperature: lower → more greedy, higher → exploratory. Breaking: community assignments will differ from prior greedy refinement for any graph where multiple candidates have positive quality gain during the refinement phase. Impact: 2 functions changed, 4 affected * docs: remove backlog #103 — ships in this PR, not a breaking change * fix: align Leiden refinement with Algorithm 3 (Traag et al. 2019) Three corrections to match the paper: 1. Singleton guard — only nodes still in singleton communities are candidates for merging. Once merged, a node is locked for the remainder of the pass. Essential for γ-connectedness guarantee. 2. Single pass — one randomized sweep, not an iterative while-loop. Iterating until convergence is Louvain behavior, not Leiden. 3. Stay option — the "remain as singleton" choice (ΔH=0) is included in the Boltzmann distribution, so a node may probabilistically stay alone even when positive-gain merges exist. Impact: 1 functions changed, 1 affected * test: add Algorithm 3 conformance tests for Leiden refinement Three new tests that would catch deviations from the paper: - Stay option: high theta preserves singletons because ΔH=0 competes in the Boltzmann distribution. Without it, all positive-gain nodes would be forced to merge. - Singleton guard: ring of triangles stays granular across seeds. Without the guard, iterative passes would collapse adjacent triangles. - Single pass: refine=true preserves at least as many communities as refine=false on a uniform weak-link graph. Iterative convergence would over-merge. * feat: post-refinement connectivity split and fix default theta Three improvements to complete the robust Leiden implementation: 1. Default refinementTheta changed from 0.01 to 1.0. The old default made exp(ΔH/0.01) extremely peaked, effectively disabling the probabilistic behavior. θ=1.0 matches the paper's exp(ΔH). 2. Post-refinement split step: after probabilistic refinement, BFS each community's induced subgraph. If a community has disconnected components, split them into separate community IDs. O(V+E) total. This replaces the expensive per-candidate γ-connectedness check with a cheap post-step using codegraph's existing graph primitives. 3. New connectivity validation test: across multiple seeds, verify every community is internally connected via BFS on the subgraph. This directly tests the core Leiden guarantee. Adds resizeCommunities() to partition API for the split step. Impact: 6 functions changed, 5 affected

* docs: add true Leiden refinement phase to backlog (ID 103) Current vendored implementation uses greedy refinement, which is functionally Louvain with an extra pass. The paper's randomized refinement (Algorithm 3) is what guarantees well-connected communities — the defining contribution of Leiden over Louvain. * fix: mark backlog item #103 as breaking, add deterministic seed note (#552) The probabilistic Leiden refinement changes community assignments and introduces non-determinism — both qualify as breaking per the column definition. Added a note about using a deterministic seed for CI reproducibility. * feat: implement true Leiden probabilistic refinement (Algorithm 3) Replace greedy best-gain selection in refinement phase with Boltzmann sampling p(v, C) ∝ exp(ΔH/θ) per Traag et al. 2019, Algorithm 3. This is the defining contribution of Leiden over Louvain — guarantees γ-connected communities instead of bridge-connected subcommunities. Deterministic via seeded PRNG (mulberry32) — same seed always produces identical community assignments. New refinementTheta option (default 0.01) controls temperature: lower → more greedy, higher → exploratory. Breaking: community assignments will differ from prior greedy refinement for any graph where multiple candidates have positive quality gain during the refinement phase. Impact: 2 functions changed, 4 affected * docs: remove backlog #103 — ships in this PR, not a breaking change * fix: align Leiden refinement with Algorithm 3 (Traag et al. 2019) Three corrections to match the paper: 1. Singleton guard — only nodes still in singleton communities are candidates for merging. Once merged, a node is locked for the remainder of the pass. Essential for γ-connectedness guarantee. 2. Single pass — one randomized sweep, not an iterative while-loop. Iterating until convergence is Louvain behavior, not Leiden. 3. Stay option — the "remain as singleton" choice (ΔH=0) is included in the Boltzmann distribution, so a node may probabilistically stay alone even when positive-gain merges exist. Impact: 1 functions changed, 1 affected * test: add Algorithm 3 conformance tests for Leiden refinement Three new tests that would catch deviations from the paper: - Stay option: high theta preserves singletons because ΔH=0 competes in the Boltzmann distribution. Without it, all positive-gain nodes would be forced to merge. - Singleton guard: ring of triangles stays granular across seeds. Without the guard, iterative passes would collapse adjacent triangles. - Single pass: refine=true preserves at least as many communities as refine=false on a uniform weak-link graph. Iterative convergence would over-merge. * feat: post-refinement connectivity split and fix default theta Three improvements to complete the robust Leiden implementation: 1. Default refinementTheta changed from 0.01 to 1.0. The old default made exp(ΔH/0.01) extremely peaked, effectively disabling the probabilistic behavior. θ=1.0 matches the paper's exp(ΔH). 2. Post-refinement split step: after probabilistic refinement, BFS each community's induced subgraph. If a community has disconnected components, split them into separate community IDs. O(V+E) total. This replaces the expensive per-candidate γ-connectedness check with a cheap post-step using codegraph's existing graph primitives. 3. New connectivity validation test: across multiple seeds, verify every community is internally connected via BFS on the subgraph. This directly tests the core Leiden guarantee. Adds resizeCommunities() to partition API for the split step. Impact: 6 functions changed, 5 affected * docs: replace language support prose with checkmark compatibility matrix Replaces the "Symbols Extracted" prose column with a scannable checkmark grid covering Imports, Exports, Call Sites, Heritage, Type Inference, and Dataflow per language. * docs: use ✓ instead of ✅ in language compatibility matrix * docs: restructure roadmap phases and expand language support to 34 - Move 5.7-5.9 (supply-chain, CI quality, kill list) out of TypeScript Migration into new Phase 9 (Quality, Security & Technical Debt) - Move Expanded Language Support from Phase 10 to Phase 7, right after Native Analysis Acceleration - Expand language roadmap from 8 planned additions to 23 (34 total), covering all major tree-sitter grammars with both WASM and Rust support - Renumber all subsequent phases (7→8 through 12→13) - Update all internal cross-references * fix: address Greptile review feedback on Leiden refinement - Guard against theta <= 0 with RangeError in refineWithinCoarseCommunities - Fix termination check to also consider effectivePartition community count when refine is enabled (coarse graph is built from effectivePartition) - Add refinementTheta to DEFAULTS.community and wire through communities.js - Replace per-node candidates array allocation with pre-allocated flat typed arrays (Int32Array/Float64Array) to reduce GC pressure in hot loop - Traverse both outEdges and inEdges in splitDisconnectedCommunities for weak connectivity on directed graphs Impact: 4 functions changed, 15 affected * fix: address Greptile review feedback on Leiden refinement - Guard against theta <= 0 with RangeError in refineWithinCoarseCommunities - Fix termination check to also consider effectivePartition community count when refine is enabled (coarse graph is built from effectivePartition) - Add refinementTheta to DEFAULTS.community and wire through communities.js - Replace per-node candidates array allocation with pre-allocated flat typed arrays (Int32Array/Float64Array) to reduce GC pressure in hot loop - Traverse both outEdges and inEdges in splitDisconnectedCommunities for weak connectivity on directed graphs Impact: 4 functions changed, 15 affected * fix: update config test to include refinementTheta default * ci: trigger CI rerun * fix: retrigger CI after test fix * style: format community defaults test assertion

#103 — Recursively foldable graph viewer (Tier 1g) #104 — Architecture-as-code validation gate (Tier 1j)

* docs(backlog): add Weft-inspired items #103 and #104 #103 — Recursively foldable graph viewer (Tier 1g) #104 — Architecture-as-code validation gate (Tier 1j) * fix(backlog): correct stale .js file references to .ts in item #103

chore(release): v2.4.0

) (optave#552) * docs: add true Leiden refinement phase to backlog (ID 103) Current vendored implementation uses greedy refinement, which is functionally Louvain with an extra pass. The paper's randomized refinement (Algorithm 3) is what guarantees well-connected communities — the defining contribution of Leiden over Louvain. * fix: mark backlog item optave#103 as breaking, add deterministic seed note (optave#552) The probabilistic Leiden refinement changes community assignments and introduces non-determinism — both qualify as breaking per the column definition. Added a note about using a deterministic seed for CI reproducibility. * feat: implement true Leiden probabilistic refinement (Algorithm 3) Replace greedy best-gain selection in refinement phase with Boltzmann sampling p(v, C) ∝ exp(ΔH/θ) per Traag et al. 2019, Algorithm 3. This is the defining contribution of Leiden over Louvain — guarantees γ-connected communities instead of bridge-connected subcommunities. Deterministic via seeded PRNG (mulberry32) — same seed always produces identical community assignments. New refinementTheta option (default 0.01) controls temperature: lower → more greedy, higher → exploratory. Breaking: community assignments will differ from prior greedy refinement for any graph where multiple candidates have positive quality gain during the refinement phase. Impact: 2 functions changed, 4 affected * docs: remove backlog optave#103 — ships in this PR, not a breaking change * fix: align Leiden refinement with Algorithm 3 (Traag et al. 2019) Three corrections to match the paper: 1. Singleton guard — only nodes still in singleton communities are candidates for merging. Once merged, a node is locked for the remainder of the pass. Essential for γ-connectedness guarantee. 2. Single pass — one randomized sweep, not an iterative while-loop. Iterating until convergence is Louvain behavior, not Leiden. 3. Stay option — the "remain as singleton" choice (ΔH=0) is included in the Boltzmann distribution, so a node may probabilistically stay alone even when positive-gain merges exist. Impact: 1 functions changed, 1 affected * test: add Algorithm 3 conformance tests for Leiden refinement Three new tests that would catch deviations from the paper: - Stay option: high theta preserves singletons because ΔH=0 competes in the Boltzmann distribution. Without it, all positive-gain nodes would be forced to merge. - Singleton guard: ring of triangles stays granular across seeds. Without the guard, iterative passes would collapse adjacent triangles. - Single pass: refine=true preserves at least as many communities as refine=false on a uniform weak-link graph. Iterative convergence would over-merge. * feat: post-refinement connectivity split and fix default theta Three improvements to complete the robust Leiden implementation: 1. Default refinementTheta changed from 0.01 to 1.0. The old default made exp(ΔH/0.01) extremely peaked, effectively disabling the probabilistic behavior. θ=1.0 matches the paper's exp(ΔH). 2. Post-refinement split step: after probabilistic refinement, BFS each community's induced subgraph. If a community has disconnected components, split them into separate community IDs. O(V+E) total. This replaces the expensive per-candidate γ-connectedness check with a cheap post-step using codegraph's existing graph primitives. 3. New connectivity validation test: across multiple seeds, verify every community is internally connected via BFS on the subgraph. This directly tests the core Leiden guarantee. Adds resizeCommunities() to partition API for the split step. Impact: 6 functions changed, 5 affected

* docs: add true Leiden refinement phase to backlog (ID 103) Current vendored implementation uses greedy refinement, which is functionally Louvain with an extra pass. The paper's randomized refinement (Algorithm 3) is what guarantees well-connected communities — the defining contribution of Leiden over Louvain. * fix: mark backlog item optave#103 as breaking, add deterministic seed note (optave#552) The probabilistic Leiden refinement changes community assignments and introduces non-determinism — both qualify as breaking per the column definition. Added a note about using a deterministic seed for CI reproducibility. * feat: implement true Leiden probabilistic refinement (Algorithm 3) Replace greedy best-gain selection in refinement phase with Boltzmann sampling p(v, C) ∝ exp(ΔH/θ) per Traag et al. 2019, Algorithm 3. This is the defining contribution of Leiden over Louvain — guarantees γ-connected communities instead of bridge-connected subcommunities. Deterministic via seeded PRNG (mulberry32) — same seed always produces identical community assignments. New refinementTheta option (default 0.01) controls temperature: lower → more greedy, higher → exploratory. Breaking: community assignments will differ from prior greedy refinement for any graph where multiple candidates have positive quality gain during the refinement phase. Impact: 2 functions changed, 4 affected * docs: remove backlog optave#103 — ships in this PR, not a breaking change * fix: align Leiden refinement with Algorithm 3 (Traag et al. 2019) Three corrections to match the paper: 1. Singleton guard — only nodes still in singleton communities are candidates for merging. Once merged, a node is locked for the remainder of the pass. Essential for γ-connectedness guarantee. 2. Single pass — one randomized sweep, not an iterative while-loop. Iterating until convergence is Louvain behavior, not Leiden. 3. Stay option — the "remain as singleton" choice (ΔH=0) is included in the Boltzmann distribution, so a node may probabilistically stay alone even when positive-gain merges exist. Impact: 1 functions changed, 1 affected * test: add Algorithm 3 conformance tests for Leiden refinement Three new tests that would catch deviations from the paper: - Stay option: high theta preserves singletons because ΔH=0 competes in the Boltzmann distribution. Without it, all positive-gain nodes would be forced to merge. - Singleton guard: ring of triangles stays granular across seeds. Without the guard, iterative passes would collapse adjacent triangles. - Single pass: refine=true preserves at least as many communities as refine=false on a uniform weak-link graph. Iterative convergence would over-merge. * feat: post-refinement connectivity split and fix default theta Three improvements to complete the robust Leiden implementation: 1. Default refinementTheta changed from 0.01 to 1.0. The old default made exp(ΔH/0.01) extremely peaked, effectively disabling the probabilistic behavior. θ=1.0 matches the paper's exp(ΔH). 2. Post-refinement split step: after probabilistic refinement, BFS each community's induced subgraph. If a community has disconnected components, split them into separate community IDs. O(V+E) total. This replaces the expensive per-candidate γ-connectedness check with a cheap post-step using codegraph's existing graph primitives. 3. New connectivity validation test: across multiple seeds, verify every community is internally connected via BFS on the subgraph. This directly tests the core Leiden guarantee. Adds resizeCommunities() to partition API for the split step. Impact: 6 functions changed, 5 affected * docs: replace language support prose with checkmark compatibility matrix Replaces the "Symbols Extracted" prose column with a scannable checkmark grid covering Imports, Exports, Call Sites, Heritage, Type Inference, and Dataflow per language. * docs: use ✓ instead of ✅ in language compatibility matrix * docs: restructure roadmap phases and expand language support to 34 - Move 5.7-5.9 (supply-chain, CI quality, kill list) out of TypeScript Migration into new Phase 9 (Quality, Security & Technical Debt) - Move Expanded Language Support from Phase 10 to Phase 7, right after Native Analysis Acceleration - Expand language roadmap from 8 planned additions to 23 (34 total), covering all major tree-sitter grammars with both WASM and Rust support - Renumber all subsequent phases (7→8 through 12→13) - Update all internal cross-references * fix: address Greptile review feedback on Leiden refinement - Guard against theta <= 0 with RangeError in refineWithinCoarseCommunities - Fix termination check to also consider effectivePartition community count when refine is enabled (coarse graph is built from effectivePartition) - Add refinementTheta to DEFAULTS.community and wire through communities.js - Replace per-node candidates array allocation with pre-allocated flat typed arrays (Int32Array/Float64Array) to reduce GC pressure in hot loop - Traverse both outEdges and inEdges in splitDisconnectedCommunities for weak connectivity on directed graphs Impact: 4 functions changed, 15 affected * fix: address Greptile review feedback on Leiden refinement - Guard against theta <= 0 with RangeError in refineWithinCoarseCommunities - Fix termination check to also consider effectivePartition community count when refine is enabled (coarse graph is built from effectivePartition) - Add refinementTheta to DEFAULTS.community and wire through communities.js - Replace per-node candidates array allocation with pre-allocated flat typed arrays (Int32Array/Float64Array) to reduce GC pressure in hot loop - Traverse both outEdges and inEdges in splitDisconnectedCommunities for weak connectivity on directed graphs Impact: 4 functions changed, 15 affected * fix: update config test to include refinementTheta default * ci: trigger CI rerun * fix: retrigger CI after test fix * style: format community defaults test assertion

…ave#945) * docs(backlog): add Weft-inspired items optave#103 and optave#104 optave#103 — Recursively foldable graph viewer (Tier 1g) optave#104 — Architecture-as-code validation gate (Tier 1j) * fix(backlog): correct stale .js file references to .ts in item optave#103

greptile-apps Bot reviewed Feb 25, 2026

View reviewed changes

chore(release): bump version to 2.4.0 and add CHANGELOG entry

01b54eb

carlos-alm changed the title ~~feat(cli): add update notification after commands~~ chore(release): v2.4.0 Feb 25, 2026

fix(test): stub CI env in update-check tests to prevent early return

caaef1f

The CI environment variable caused checkForUpdates to bail out at the guard clause before reaching fetch/cache logic, making 5 tests fail in GitHub Actions.

carlos-alm merged commit 851bcb0 into main Feb 26, 2026
16 checks passed

carlos-alm deleted the feat/update-notification branch February 26, 2026 03:49

This was referenced Mar 20, 2026

docs: add community leverage features to backlog (IDs 100-102) #546

Merged

feat: implement true Leiden probabilistic refinement (backlog #103) #552

Merged

carlos-alm added a commit that referenced this pull request Mar 21, 2026

docs: remove backlog #103 — ships in this PR, not a breaking change

b89b764

carlos-alm added a commit that referenced this pull request Apr 16, 2026

docs(backlog): add Weft-inspired items #103 and #104

ed8076b

#103 — Recursively foldable graph viewer (Tier 1g) #104 — Architecture-as-code validation gate (Tier 1j)

carlos-alm mentioned this pull request Apr 16, 2026

docs(backlog): add Weft-inspired items #103 and #104 #945

Merged

2 tasks

carlos-alm added a commit that referenced this pull request Apr 16, 2026

fix(backlog): correct stale .js file references to .ts in item #103

0d13263

Zeeeepa pushed a commit to Zeeeepa/codegraph that referenced this pull request Jun 22, 2026

Merge pull request optave#103 from optave/feat/update-notification

676631b

chore(release): v2.4.0

Sunbelt Computer Software

PL/B Language Development and Support

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

chore(release): v2.4.0#103

chore(release): v2.4.0#103
carlos-alm merged 3 commits into
mainfrom
feat/update-notification

carlos-alm commented Feb 25, 2026 •

edited

Loading

Uh oh!

claude Bot commented Feb 25, 2026 •

edited

Loading

Uh oh!

greptile-apps Bot commented Feb 25, 2026

Uh oh!

greptile-apps Bot left a comment

Uh oh!

claude Bot commented Feb 26, 2026 •

edited

Loading

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

Sunbelt Computer Software

PL/B Language Development and Support

Uh oh!

Conversation

carlos-alm commented Feb 25, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Summary

Highlights (v2.4.0)

Test plan

Uh oh!

claude Bot commented Feb 25, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

PR Review Complete

Todo List:

Root Cause Analysis

Critical Concerns

Configuration Change Detection ✅

Test Integrity Analysis ✅

Code Quality Review

src/cli.js (+12 lines) ✅

src/update-check.js (+159 lines) ✅

tests/unit/update-check.test.js (+282 lines) ✅

Final Assessment

Final Recommendation

Uh oh!

greptile-apps Bot commented Feb 25, 2026

Greptile Summary

Confidence Score: 5/5

Important Files Changed

Flowchart

Uh oh!

greptile-apps Bot left a comment

Choose a reason for hiding this comment

Uh oh!

claude Bot commented Feb 26, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

PR Review Complete

Todo List:

Root Cause Analysis

Critical Concerns

Configuration Change Detection ✅

Test Integrity Analysis ✅

Code Quality Review

src/cli.js (+12 lines) ✅

src/update-check.js (+159 lines) ✅

tests/unit/update-check.test.js (+282 lines) ✅

Final Assessment

Final Recommendation

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

carlos-alm commented Feb 25, 2026 •

edited

Loading

claude Bot commented Feb 25, 2026 •

edited

Loading

claude Bot commented Feb 26, 2026 •

edited

Loading