👷 ci: scope mutation testing to production code

[gaborbernat]

The nightly mutation run covers 18,902 sites, and 1,063 of them sit in code that answers to nobody outside this repository. A surviving mutant marks a gap where the behaviour is a promise to someone. Production code promises behaviour to the people running peryx, so a survivor there is a missing test. Benchmark workloads, the shared harness and fixture binaries make no such promise, so a survivor in them is noise that each run reports again. The triage in #1893 spent 19 of its 48 sites on those paths.

.cargo/mutants.toml takes three paths out of scope, with a reason beside each glob. A benchmark workload under crates/*/src/bench/ builds load for CodSpeed to measure, so changing how much it builds moves the measurement rather than any claim about correctness. Fixture binaries under crates/*/tests/, such as the S3 multipart fixture, hold up assertions that live elsewhere, and mutating one fails the suite that drives it. 🔬 The harness in crates/peryx-test-support/ is the one worth arguing over, since a fault injector that stopped injecting would leave every test using it green while testing nothing. That failure shows up red instead. A test that arms a fault asserts the faulted outcome, so it fails when the fault does not arrive, and -D dead_code covers harness behaviour no test reaches. Both checks land before mutation asks the question, and the inventory agrees, with no mutant in that crate surviving a run. No carve-out earns its place today, and adding a path back means saying in the config why.

Shard length holds rather than growing, because mutation-count feeds mutation-shard-count from the same plan the run mutates, so 148 shards become 140 with each still targeting 128 mutants. 📉 A contract in lint-automation pins the three globs and checks that the count still comes from that plan, so a wider count cannot size the matrix for mutants nothing examines.

Closes #2164

[codspeed-hq]

Merging this PR will not alter performance

✅ 29 untouched benchmarks
⏩ 133 skipped benchmarks¹

---

_{Comparing fix/mutation-scope-2164 (aecb241) with main (68b3d3c)}

Footnotes

1. 133 benchmarks were skipped, so the baseline results were used instead. If they were deleted from the codebase, click here and archive them to remove them from the performance reports. ↩

[read-the-docs-community]

Documentation build overview

📚 peryx | 🛠️ Build #34361705 | 📁 Comparing aecb241 against latest (68b3d3c)

  🔍 Preview build  

2 files changed

± contributing/ci/index.html
± core/repositories/artifact-source/index.html