Files
callstack__agent-device/scripts/mutation/modules.ts
Michał Pierzchała 423927fdd8 chore(mutation): shrink to report-only — drop the ratchet, baseline and graduation (#1457, #1781) (#1828)
* chore(mutation): shrink the lane to report-only (#1457, #1781 wave 2)

The mutation harness's two real catches (#1474, #1475) both came from humans
reading the weekly score report. The ratchet half never operated: the baseline
was committed exactly twice (8cce0ef6b, 60400d04b), both times with
`stableRuns: 0, gating: false`, and was never updated after the very fixes it
triggered — the weekly job computed a new baseline and then `git checkout --`d
it, uploading a proposal nobody applied in 3+ weeks. A gate nobody arms is
harness weight; the report is the part that paid.

Deletes ratchet.ts + ratchet.test.ts, mutation-baselines/, and every
baseline/graduation/gating path in run.ts (`--update`, `mutation:baseline`).
run.ts now exits non-zero only on a harness failure, never on a score. The
report renders the per-kernel table (kernel, score, killed, survived, total,
timeouts) plus the surviving mutants a strengthening PR works from.

Kernel scoping stays: stryker.config.json and KERNEL_MODULES are untouched.

* fix(mutation): restore denominator coverage and publish the table before judging the shard set

Review of #1828:
- `report.test.ts` re-asserts that Ignored/CompileError/RuntimeError leave the
  denominator — the one behaviour `ratchet.test.ts` covered and nothing replaced.
  A `tally()` edit that counted tool noise would have deflated every published
  score with a green `mutation:test`.
- `assertShardsCoverModules` now runs after `emit()`, so an incomplete shard set
  still publishes the kernels that completed instead of only an error string.
  This makes the workflow comments' claim about the job summary true rather than
  re-wording them down.

* chore(mutation): trigger the affected lane on exactly the paths that can select mutants

The PR lane returns an empty matrix unless the diff touches the harness, so the
kernel-source and `**/*.test.ts` triggers only bought a 1-4 min no-op job on
~96% of PRs. `on.pull_request.paths` is now exactly `LANE_TOOLING` plus the
workflow file, asserted in both directions by workflow.test.ts against the
exported constant — a missing path would let a harness change merge unproven,
an extra one starts a job that can only answer `[]`.

Also drops the workflow header's contradictory scope paragraph: it claimed the
lane selects on kernel sources and any test reaching one, which has not been
true since the ratchet went.

* fix(mutation): score and publish a short shard set before failing on the count

The expected-count check ran inside readShardedReports, before anything was
summarized, so on the weekly's real `--expect-shards 10` one dead shard threw
away the nine that had reported — the earlier reorder only moved the
zero-mutants check. The merge now returns the shard count, and both verdicts
run after emit() with the same exit code and `score` stage.

Regression uses the weekly argument shape (`--expect-shards 10`, one shard
present) and asserts the reporting kernel's row reaches stdout while the run
still fails.
2026-08-18 17:47:29 +02:00

185 lines
6.7 KiB
TypeScript

// Enumerated decision kernels the mutation lane measures (issue #1415).
//
// Mutation score is the only mechanical answer to "is this test load-bearing or
// decorative", but a full-suite sweep is unaffordable. This registry is the
// single source of truth for what Stryker mutates: `stryker.config.json`'s
// `mutate` globs are asserted against it, and PR-affected selection maps changed
// files onto modules through it.
//
// Membership rule: pure decision kernels only — a surviving mutant here means a
// silently wrong agent-facing decision. Anything that spawns subprocesses or
// waits real time is out of scope by construction (its mutants would be timeout
// noise, not test-strength signal).
export type ModuleId =
| 'kernel-errors'
| 'daemon-ref-frame'
| 'interaction-settle'
| 'scroll-edge-state'
| 'selectors'
| 'target-annotation-serde'
| 'snapshot-occlusion';
export type KernelModule = {
readonly id: ModuleId;
readonly label: string;
/** Globs handed to Stryker's `mutate`. */
readonly mutate: readonly string[];
/**
* Paths this module owns; a trailing `/` marks a directory prefix. Sources
* only — the tests whose strength the score measures are derived from the
* import graph in `ownership.ts`, never listed here.
*/
readonly owns: readonly string[];
/**
* How many parallel jobs the module's mutants are sliced across. One job per
* module is the default; a module big enough to outrun the lane's 30-minute
* budget declares more (`--shard i/n` picks the slice).
*/
readonly shards?: number;
};
export const KERNEL_MODULES: readonly KernelModule[] = [
{
id: 'kernel-errors',
label: 'Error retriability + hints',
mutate: ['packages/kernel/src/errors.ts'],
owns: ['packages/kernel/src/errors.ts'],
},
{
id: 'daemon-ref-frame',
label: 'Ref-frame admission matrix (ADR 0014)',
mutate: ['src/daemon/ref-frame.ts'],
owns: ['src/daemon/ref-frame.ts'],
},
{
id: 'interaction-settle',
label: 'Interaction settle decisions',
mutate: ['src/commands/interaction/runtime/settle.ts'],
owns: ['src/commands/interaction/runtime/settle.ts'],
},
{
id: 'scroll-edge-state',
label: 'Scroll edge-state detection',
mutate: ['src/utils/scroll-edge-state.ts'],
owns: ['src/utils/scroll-edge-state.ts'],
},
{
id: 'selectors',
label: 'Selector parsing + matching',
mutate: [
'packages/selectors/src/**/*.ts',
'!packages/selectors/src/**/*.test.ts',
'!packages/selectors/src/internal/__tests__/**',
],
// Selector tests live under the owned package source, so the prefix covers them.
owns: ['packages/selectors/src/'],
// ~1,280 mutants at the observed ~3s/mutant on a 2-core runner is ~64
// minutes in one job — past the acceptance budget and its own timeout.
shards: 4,
},
{
id: 'target-annotation-serde',
label: 'Target-annotation comment-line codec (ADR 0012 decision 3)',
mutate: ['packages/ad-script/src/internal/target-annotation-serde.ts'],
owns: ['packages/ad-script/src/internal/target-annotation-serde.ts'],
},
{
id: 'snapshot-occlusion',
label: 'Snapshot occlusion (covered/not-covered) decisions',
mutate: ['src/snapshot/snapshot-occlusion.ts'],
owns: ['src/snapshot/snapshot-occlusion.ts'],
},
];
export const ALL_MODULE_IDS: readonly ModuleId[] = KERNEL_MODULES.map((module) => module.id);
export function moduleById(id: ModuleId): KernelModule {
const found = KERNEL_MODULES.find((module) => module.id === id);
if (!found) throw new Error(`Unknown mutation module: ${id}`);
return found;
}
export function isModuleId(value: string): value is ModuleId {
return ALL_MODULE_IDS.includes(value as ModuleId);
}
/** Mutate globs for a module subset, in registry order. */
export function mutateGlobs(ids: readonly ModuleId[] = ALL_MODULE_IDS): string[] {
return KERNEL_MODULES.filter((module) => ids.includes(module.id)).flatMap((module) => [
...module.mutate,
]);
}
/**
* The module a lane-tooling change proves itself against. `kernel-errors` is the
* cheapest real sweep in the registry (one file, ~183 mutants), so a change to
* the harness or the config runs actual mutants end to end without paying for
* the full sweep.
*/
export const LANE_CANARY: ModuleId = 'kernel-errors';
/** One mutation job: a module, optionally one slice of it. */
export type ShardSpec = { name: string; module: ModuleId; shard?: string };
/**
* The jobs a module set expands into. Both workflows' matrices are this list, so
* a registry module (or a change to its shard count) can never leave the sweep
* without the workflow assertions noticing.
*/
export function shardMatrix(ids: readonly ModuleId[] = ALL_MODULE_IDS): ShardSpec[] {
return KERNEL_MODULES.filter((module) => ids.includes(module.id)).flatMap((module) => {
const count = module.shards ?? 1;
if (count === 1) return [{ name: module.id, module: module.id }];
return Array.from({ length: count }, (_unused, index) => ({
name: `${module.id}-${index + 1}`,
module: module.id,
shard: `${index + 1}/${count}`,
}));
});
}
export function normalizePath(filePath: string): string {
return filePath.replaceAll('\\', '/').replace(/^\.\//, '');
}
/**
* Where the mutation lane can ever find a kernel's owning test: root `src/` or
* a workspace package's `src/` — the same two roots `unit-core`'s `include`
* (`vitest.config.ts`) draws tests from. A kernel whose test lives outside
* both (e.g. `scripts/__tests__`) is unreachable by construction, not silently
* dropped.
*/
const KERNEL_TEST_FILE_RE = /^(?:src\/|packages\/[^/]+\/src\/).*\.test\.ts$/;
export function isKernelTestFile(filePath: string): boolean {
return KERNEL_TEST_FILE_RE.test(normalizePath(filePath));
}
/**
* Which kernel module owns a repository-relative *source* path, if any.
*
* Test-file attribution is derived from the import graph — see
* `derivedAffectedModules` in `ownership.ts`, which is what the PR lane calls.
*/
export function moduleForFile(filePath: string): ModuleId | undefined {
const normalized = normalizePath(filePath);
for (const module of KERNEL_MODULES) {
for (const owned of module.owns) {
const match = owned.endsWith('/') ? normalized.startsWith(owned) : normalized === owned;
if (match) return module.id;
}
}
return undefined;
}
/** Kernel modules whose registry-owned paths a diff touches. */
export function affectedModules(changedFiles: readonly string[]): ModuleId[] {
const ids = new Set<ModuleId>();
for (const file of changedFiles) {
const id = moduleForFile(file);
if (id) ids.add(id);
}
return KERNEL_MODULES.filter((module) => ids.has(module.id)).map((module) => module.id);
}