Set benchmark run job to Default to reduce flakiness

[sanelli]

Summary

BenchmarkDotNet is currently invoked with the Short job (-j Short) in Scripts/RunBenchmarks.ps1. Short runs use fewer iterations/warmups and produce noisier results, which makes benchmark comparisons and charts flaky.

Proposed change

• Change the BenchmarkDotNet job from Short to Default (either -j Default or omit -j so BenchmarkDotNet uses its default job).
• Keep exporters (Csv, Html, GitHub) and filter arguments unchanged unless they must be adjusted for the longer run.

Acceptance criteria

• Scripts/RunBenchmarks.ps1 runs benchmarks with the Default job instead of Short.
• Document any expected increase in wall-clock time if relevant (script comments / README).
• Benchmark run completes successfully with Default job.

Context

Current invocation (approx.):

dotnet run -c Release --project ./Mappa.Benchmark/ -- -j Short -e "Csv" "Html" "GitHub" @filterArgs

[sanelli]

Implementation plan

PR: #332

---

name: 331 benchmark default job
overview: Switch BenchmarkDotNet from Short to Default, stop PR benchmark runs/comments (keep main merge only), bump version, and verify a chart-subset run.
todos:

• id: "1"
  content: 1. Scripts — change -j Short to -j Default in RunBenchmarks.ps1; comment on longer wall-clock time
  status: completed
• id: "2"
  content: 2. Documentation — update development.md and Mappa.Benchmark/README.md (Default job + longer runtime)
  status: completed
• id: "3"
  content: 3. Bump MappaVersion.targets alpha and run .\Scripts\RebuildUponVersionChange.ps1
  status: completed
• id: "4"
  content: 4. Verify .\Scripts\RunBenchmarks.ps1 -ChartBenchmarksOnly completes successfully with Default job
  status: completed
• id: "5"
  content: 5. Remove benchmark run from .github/workflows/pr.yml
  status: completed
• id: "6"
  content: 6. Remove benchmark section/params from CommentPullRequestCoverage.ps1
  status: completed
• id: "7"
  content: 7. Update docs that CI/PR still imply PR benchmarks
  status: completed
• id: "8"
  content: 8. Bump MappaVersion.targets and run RebuildUponVersionChange.ps1
  status: completed
  isProject: false

---

Issue #331 — Set benchmark run job to Default

Context

#331 asks to replace BenchmarkDotNet Short with Default so CI/local benchmark charts are less noisy. Branch 331-set-benchmark-run-job-to-default-to-reduce-flakiness already matches the issue prefix.

Follow-up: remove benchmark invocation and PR-comment benchmark results from the pull-request workflow; keep benchmarks only on merge-main.yml.

Scope: Scripts + Documentation + workflows + version bump only — no C# project changes. Per mappa-development-workflow, omit the coverage baseline and the final coverage non-decrease check.

Approach

Use explicit -j Default. Keep exporters and filter args unchanged. PR workflow runs coverage comment only; benchmarks remain on main merge.

Steps

1. Scripts — Default job in RunBenchmarks.ps1

Change -j Short to -j Default and comment that Default is slower but more stable.

2. Documentation — development.md and Mappa.Benchmark README

Document Default job and longer wall-clock time.

3. Version bump and rebuild

10.2.0-alpha.62 → 10.2.0-alpha.63 + rebuild.

4. Verify ChartBenchmarksOnly run with Default

Run .\Scripts\RunBenchmarks.ps1 -ChartBenchmarksOnly successfully.

5. Remove benchmark run from pr.yml

Delete the Run benchmarks step from .github/workflows/pr.yml.

6. Remove benchmark posting from CommentPullRequestCoverage.ps1

Drop BenchmarkSummaryPath / BenchmarkComparisonPath params and the ## Benchmarks section from Scripts/CommentPullRequestCoverage.ps1.

7. Docs — PRs do not run benchmarks

Clarify in Documentation/development.md and Mappa.Benchmark/README.md that chart-subset Default runs are for main merge CI only.

8. Version bump and rebuild

10.2.0-alpha.63 → 10.2.0-alpha.64 + rebuild.

[sanelli]

Generator tests report

No new or modified generator integration tests for this issue.