Skip to content

Fix impact count totals independent of limits - #5240

Merged
Widthdom merged 4 commits into
mainfrom
fix-issue5226
Aug 31, 2026
Merged

Fix impact count totals independent of limits#5240
Widthdom merged 4 commits into
mainfrom
fix-issue5226

Conversation

@Widthdom

Copy link
Copy Markdown
Owner

Summary

  • Make impact count modes independent of presentation limits and avoid materializing normal result rows.
  • Batch bounded identity-target traversal so multi-hop MCP counts complete within the request deadline while preserving graph-readiness and lower-bound authority signals.
  • Report MCP presentation limits as ignored in count-only mode, and add bilingual documentation, changelog, and regression coverage.

Testing

  • dotnet build CodeIndex.sln --no-restore -p:UseSharedCompilation=false (Debug; 0 warnings, 0 errors)
  • Full Release .NET 8 suite: 12,114 passed, 9 skipped; one unrelated allocation microbenchmark failed once and passed on an immediate isolated rerun (1/1)
  • Full Release .NET 9 suite: 11,630 passed, 431 skipped, 0 failed
  • Impact test suite: 141 passed on each of .NET 8 and .NET 9
  • dotnet format CodeIndex.sln --no-restore --verify-no-changes
  • Changelog fragment validation and git diff origin/main...HEAD --check
  • Codex adversarial review: 2 rounds (workflow maximum); all findings from both rounds were fixed
  • Representative 300-result CLI count reduced from about 54 seconds to 8.38 seconds; the equivalent MCP request completed without timing out

Fixes #5226

@Widthdom
Widthdom merged commit 4a45d5e into main Aug 31, 2026
12 checks passed
@Widthdom
Widthdom deleted the fix-issue5226 branch August 31, 2026 10:54
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

Make impact --count ignore presentation limits and return truthful totals

1 participant