Add DiffPlex benchmarks and reduce builder allocations

[Copilot]

Adds a BenchmarkDotNet harness for measuring DiffPlex core diff APIs and records baseline/final results for comparison. Applies low-risk runtime improvements while keeping the public API unchanged.

• Benchmark coverage
  • Added DiffPlex.Benchmarks with MemoryDiagnoser.
  • Covers core differ APIs, inline builder, and side-by-side builder.
  • Includes small/large inputs across identical, many-small-change, few-large-change, completely different, and long-common-subsequence scenarios.

dotnet run -c Release --project DiffPlex.Benchmarks -- --filter "*" --exporters github csv

• Recorded results

  • Added benchmarks/RESULTS.md with baseline, final, before/after summary, and ROI notes.

• Optimizations

  • Reused stateless line/character chunker singletons.
  • Pre-sized side-by-side diff output lists.
  • Replaced LINQ scans in hot/read paths with indexed loops.
  • Tracked side-by-side change summaries during piece construction instead of rescanning.

• Measured impact

  • Large side-by-side many-small-change case: 1,939.46 μs → 1,712.49 μs.
  • Side-by-side allocations reduced ~5–9% across key large changed-input scenarios.
  • Additional measured candidates were not retained where results were mixed or low ROI.

[coderabbitai]

Important

Review skipped

Bot user detected.

To trigger a single review, invoke the @coderabbitai review command.

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Advanced

Run ID: 5223aabf-02bd-415f-8d7f-b18c87d2d751

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

Use the checkbox below for a quick retry:

• 🔍 Trigger review

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}