🧪 Raise server unit-test coverage (81.8% line baseline, tooling included)

[marceld23]

What this is

Help raising the server unit-test coverage. The tooling is already in the repo, the low-hanging targets are known, and CI verifies your work objectively — a great first code contribution.

Current baseline (June 2026, may have drifted slightly)

Line 81.8% / branch 73.8% / method 88.8%, per assembly:

Assembly	Line coverage
Persistence	96.5%
WorldGeneration	93.9%
Shared	85.7%
GameServer	78.7%
Api	68.4%
Networking	59.3%

⚠️ Don't chase the raw numbers blindly: a big share of the low Api/Networking figures is intentionally-untested I/O (LiteNetLib/WebSocket transports, HTML page rendering, console logger). The valuable targets are pure, deterministic logic that currently lacks tests.

Suggested targets (verify names/constants against current source first)

• Primitive types — Vector3i, ChunkCoord, BlockId: equality, hashing, distance
• FrequencyExtensions — weight/probability factors per enum level
• Server presets — lookup case-insensitivity, unknown-name handling
• Localizer — hit / fallback / unknown-key wrap / empty key
• Mission validation — each error case (empty id, no objectives, unknown type/target)
• Noise + deterministic RNG — same seed → same output, output ranges, seamless wrap
• World/structure state — block modifier set/get roundtrips, air-clears-modifiers

How to run

• Fast daily loop: dotnet test -c Release --filter Category!=Slow
• Full coverage report: ./scripts/test-coverage.ps1 -Open (uses coverlet.runsettings; slow — ~30 min — because the worldgen tests are heavy under the collector)

CI rules that will bite you (warnings are errors)

• Async test methods need an Async suffix (VSTHRD200)
• Don't await a TaskCompletionSource in tests (VSTHRD003) — use SemaphoreSlim
• Any non-Slow-tagged test must finish well under 120 s

[salem-wdev]

Hi, I would like to work on this issue if it is still available.

[marceld23]

Hi @salem-wdev, welcome — it's all yours! 🎉 I've assigned you.

A few tips to get started:

• The suggested targets in the issue are still open — none of them have gained tests since. Best starting point is the primitive types (Vector3i, ChunkCoord) since they're small and self-contained.
• The coverage baseline is from June and has drifted a bit with recent releases, but that doesn't change the targets.
• Feel free to split your work into several small PRs rather than one big one — easier to review, faster to merge.
• Note the CI rules in the issue (warnings are errors) — they catch most first-time contributors.

If anything is unclear, just ask here. Looking forward to your PR!

[salem-wdev]

Thanks for assigning this to me @marceld23, I appreciate that, I'll start by looking into the primitive types (Vector3i and ChunkCoord) as you suggested, I'll reach out if I hit any roadblocks.