fix: resolve three build bugs (#287, #288, #289)

[carlos-alm]

Summary

Fixes three bugs in src/builder.js that affect codegraph build:

• bug(build): forward-slash paths on Windows cause 0 files parsed and graph stored in CWD #287 — Forward-slash paths on Windows: buildGraph now normalizes rootDir with path.resolve() on entry. Idempotent for CLI callers, fixes programmatic callers and Git Bash/MSYS2 environments that pass POSIX-style paths like H:/repo.

• bug(build): --cfg flag skipped on incremental no-op — requires --no-incremental for first CFG build #288 — --cfg flag skipped on incremental no-op: The early-return path (no files changed) now checks whether --cfg or --dataflow was requested but the DB has no corresponding data. When detected, it parses all files and runs only the missing analysis pass before returning — no full pipeline re-run.

• bug(build): incremental divergence warning compares per-batch edges vs total — false positive #289 — Incremental divergence warning false positive: Replaced the batch edgeCount accumulator (which only counted edges inserted in the current batch) with an actual SELECT COUNT(*) FROM edges query. This fixes the drift comparison that was comparing batch count vs total DB count, producing ~59% false divergence warnings. The unused edgeCount variable and all increment sites were removed.

Test plan

• npm test passes (1369/1369 tests)
• npx biome check src/builder.js lint clean
• node src/cli.js build H:/Vscode/codegraph with forward-slash path finds files
• node src/cli.js build . && node src/cli.js build . --cfg populates CFG data on second run
• Full build → touch one file → incremental build → no spurious divergence warning

[claude]

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

---

PR Review: Fix Three Build Bugs (#287, #288, #289)

Todo List:

• Read and analyze src/builder.js changes
• Phase 1: Root Cause Analysis
• Phase 2: Configuration Change Detection
• Phase 3: Test Integrity Check
• Phase 4: Code Quality Assessment
• Provide final rating and recommendation

Phase 1: Root Cause Analysis ✅

Problem Being Solved: This PR fixes three legitimate bugs in the buildGraph function:

1. Bug bug(build): forward-slash paths on Windows cause 0 files parsed and graph stored in CWD #287 (Windows path normalization): buildGraph receives forward-slash paths from Git Bash/MSYS2/programmatic callers but needs them normalized to work correctly on Windows. Adding path.resolve(rootDir) at line 412 ensures consistent path handling.

2. Bug bug(build): --cfg flag skipped on incremental no-op — requires --no-incremental for first CFG build #288 (--cfg flag skipped on no-op): When incremental builds detect no file changes, the early-return path skipped requested analysis passes like --cfg and --dataflow. The fix (lines 501-540) checks if these analyses were requested but never computed, then runs only the missing analysis.

3. Bug bug(build): incremental divergence warning compares per-batch edges vs total — false positive #289 (False divergence warnings): The edgeCount variable was accumulating edges inserted in the current batch, not the total DB count. This caused incorrect drift calculations (~59% false warnings). The fix removes the accumulator and uses actual SELECT COUNT(*) queries.

Why This Approach: All three fixes address root causes, not symptoms. The changes are surgical and don't relax any validation rules.

Risk Assessment: LOW. Changes are isolated to the build pipeline with clear benefits and no configuration relaxation.

Phase 2: Configuration Change Detection ✅

Analysis: ✅ NO HIGH-RISK configuration changes detected

• No linting rules disabled or relaxed
• No test timeouts increased or coverage lowered
• No build configuration modifications
• Only source code changes in src/builder.js

Phase 3: Test Integrity ✅

Analysis: ✅ NO test integrity issues found

• No test files modified alongside implementation
• Tests claim to pass (1369/1369 according to PR description)
• No test expectations lowered or assertions weakened
• Test plan includes manual verification steps for platform-specific bugs

Phase 4: Code Quality Assessment ✅

Strengths:

• Clean implementation: Each fix is surgical and addresses the specific bug
• Proper error handling: try/catch blocks protect against missing tables (lines 505-508, 513-524)
• Performance consideration: Analysis pass only runs when needed, avoiding unnecessary work
• Consistent patterns: Uses existing DB query patterns and error handling styles

Detailed Analysis:

1. Path normalization fix (line 412): ✅ GOOD

   • Simple, idempotent fix that resolves Windows path issues
   • path.resolve() is the correct approach for normalizing paths
   • No side effects for already-absolute paths

2. CFG/dataflow analysis fix (lines 501-540): ✅ WELL DESIGNED

   • Properly checks if analysis was requested but never computed
   • Uses appropriate DB queries to detect missing data
   • Follows existing patterns for conditional analysis execution
   • Gracefully handles missing tables with try/catch blocks

3. Edge count fix: ✅ EXCELLENT BUG FIX

   • Removes broken accumulator pattern that only counted batch edges
   • Replaces with authoritative SELECT COUNT(*) query at lines 1207, 1220, 1255
   • Fixes both logging and drift detection accuracy
   • No performance impact (single query vs accumulating)

Final Assessment

Root Cause Analysis

Problem Being Solved: Three legitimate bugs in core build functionality - Windows path handling, incremental analysis skipping, and inaccurate edge count reporting.
Why This Approach: Each fix directly addresses the root cause without workarounds or configuration changes.
Risk Assessment: LOW - surgical changes with clear benefits and no relaxed validation.

Critical Concerns

None identified. All changes are appropriate bug fixes.

Final Recommendation

• Rating: ⭐⭐⭐⭐☆ (4/5)
• Action: APPROVE
• Reasoning: Well-executed bug fixes that address real issues. The implementation is clean, follows existing patterns, and includes proper error handling. The edge count fix particularly stands out as addressing a significant user experience issue (59% false warnings). Only missing the 5th star due to lack of accompanying regression tests for the specific bugs fixed.

Recommendation: Merge this PR. The fixes are solid and address legitimate issues that affect Windows users and incremental build reliability.

---

[greptile-apps]

Greptile Summary

This PR fixes three distinct bugs in src/builder.js:

• bug(build): forward-slash paths on Windows cause 0 files parsed and graph stored in CWD #287 (Windows path handling): Added path.resolve() normalization at buildGraph entry to handle forward-slash paths from Git Bash/MSYS2 and programmatic callers
• bug(build): --cfg flag skipped on incremental no-op — requires --no-incremental for first CFG build #288 (Incremental CFG/dataflow skip): Added early-return path check for missing optional analysis data; when --cfg or --dataflow is requested but DB has no data, the code now parses all files and runs only the missing analysis pass
• bug(build): incremental divergence warning compares per-batch edges vs total — false positive #289 (False drift warnings): Replaced the batch edgeCount accumulator with SELECT COUNT(*) FROM edges query, fixing false positive divergence warnings caused by comparing batch count vs total DB count

All three fixes are targeted and correct. The CFG/dataflow check uses IIFE try-catch blocks to detect missing tables (returns 0 count or errors), then conditionally runs buildCFGData or buildDataflowEdges from dynamically imported modules.

Confidence Score: 4/5

• This PR is safe to merge with minimal risk - three targeted bug fixes with no breaking changes
• Score reflects clean, well-tested fixes for documented bugs. The path normalization is idempotent for CLI callers, the CFG/dataflow early-return is properly guarded with try-catch, and the edge count fix uses standard SQL. No logic errors or security concerns found.
• No files require special attention - the changes are straightforward and well-contained to three specific bug fixes

Important Files Changed

Filename	Overview
src/builder.js	Three targeted bug fixes: path normalization for Windows, CFG/dataflow analysis pass when missing, and accurate edge count for drift detection. Code is clean and well-structured.

_{Last reviewed commit: a32e761}

[greptile-apps]

_{1 file reviewed, no comments}

_{Edit Code Review Agent Settings | Greptile}

[claude]

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

---

PR Review: Fix Three Build Bugs (#287, #288, #289)

Todo List:

• Read repository guidelines (CLAUDE.md)
• Read and analyze src/builder.js changes
• Phase 1: Root Cause Analysis
• Phase 2: Configuration Change Detection
• Phase 3: Test Integrity Check
• Phase 4: Code Quality Assessment
• Provide final rating and recommendation

Phase 1: Root Cause Analysis ✅

Problem Being Solved: This PR fixes three legitimate bugs in the buildGraph function:

1. Bug bug(build): forward-slash paths on Windows cause 0 files parsed and graph stored in CWD #287 (Windows path normalization): buildGraph receives forward-slash paths from Git Bash/MSYS2/programmatic callers but needs them normalized to work correctly on Windows. Adding path.resolve(rootDir) at line 437 ensures consistent path handling.

2. Bug bug(build): --cfg flag skipped on incremental no-op — requires --no-incremental for first CFG build #288 (--cfg flag skipped on no-op): When incremental builds detect no file changes, the early-return path skipped requested analysis passes like --cfg and --dataflow. The fix (lines 525-564) checks if these analyses were requested but never computed, then runs only the missing analysis.

3. Bug bug(build): incremental divergence warning compares per-batch edges vs total — false positive #289 (False divergence warnings): The edgeCount variable was accumulating edges inserted in the current batch, not the total DB count. This caused incorrect drift calculations (~59% false warnings). The fix removes the accumulator and uses actual SELECT COUNT(*) queries.

Why This Approach: All three fixes address root causes, not symptoms. The changes are surgical and don't relax any validation rules.

Risk Assessment: LOW. Changes are isolated to the build pipeline with clear benefits and no configuration relaxation.

Phase 2: Configuration Change Detection ✅

Analysis: ✅ NO HIGH-RISK configuration changes detected

• No linting rules disabled or relaxed
• No test timeouts increased or coverage lowered
• No build configuration modifications
• Only source code changes in src/builder.js

Phase 3: Test Integrity ✅

Analysis: ✅ NO test integrity issues found

• No test files modified alongside implementation
• Tests claim to pass (1369/1369 according to PR description)
• No test expectations lowered or assertions weakened
• Test plan includes manual verification steps for platform-specific bugs

Phase 4: Code Quality Assessment ✅

Strengths:

• Clean implementation: Each fix is surgical and addresses the specific bug
• Proper error handling: try/catch blocks protect against missing tables (lines 529-533, 537-548)
• Performance consideration: Analysis pass only runs when needed, avoiding unnecessary work
• Consistent patterns: Uses existing DB query patterns and error handling styles

Detailed Analysis:

1. Path normalization fix (line 437): ✅ GOOD

   • Simple, idempotent fix that resolves Windows path issues
   • path.resolve() is the correct approach for normalizing paths
   • No side effects for already-absolute paths

2. CFG/dataflow analysis fix (lines 525-564): ✅ WELL DESIGNED

   • Properly checks if analysis was requested but never computed
   • Uses appropriate DB queries to detect missing data
   • Follows existing patterns for conditional analysis execution
   • Gracefully handles missing tables with try/catch blocks

3. Edge count fix: ✅ EXCELLENT BUG FIX

   • Removes broken accumulator pattern that only counted batch edges
   • Replaces with authoritative SELECT COUNT(*) query at lines 1305, 1318, 1353
   • Fixes both logging and drift detection accuracy
   • No performance impact (single query vs accumulating)

Final Assessment

Root Cause Analysis

Problem Being Solved: Three legitimate bugs in core build functionality - Windows path handling, incremental analysis skipping, and inaccurate edge count reporting.
Why This Approach: Each fix directly addresses the root cause without workarounds or configuration changes.
Risk Assessment: LOW - surgical changes with clear benefits and no relaxed validation.

Critical Concerns

None identified. All changes are appropriate bug fixes.

Final Recommendation

• Rating: ⭐⭐⭐⭐☆ (4/5)
• Action: APPROVE
• Reasoning: Well-executed bug fixes that address real issues. The implementation is clean, follows existing patterns, and includes proper error handling. The edge count fix particularly stands out as addressing a significant user experience issue (59% false warnings). Only missing the 5th star due to lack of accompanying regression tests for the specific bugs fixed.

Recommendation: Merge this PR. The fixes are solid and address legitimate issues that affect Windows users and incremental build reliability.

---