@feigi/fleet-ctl 0.0.0
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/.claude-plugin/plugin.json +9 -0
- package/LICENSE +202 -0
- package/NOTICE +18 -0
- package/agents/fleet-finisher.agent.md +7 -0
- package/agents/fleet-implementer-alt.agent.md +210 -0
- package/agents/fleet-implementer.agent.md +210 -0
- package/agents/fleet-merge-bot.agent.md +7 -0
- package/agents/fleet-review-comments.agent.md +40 -0
- package/agents/fleet-review-correctness.agent.md +43 -0
- package/agents/fleet-review-runner.agent.md +73 -0
- package/agents/fleet-review-silent-failure.agent.md +49 -0
- package/agents/fleet-review-simplify.agent.md +46 -0
- package/agents/fleet-review-snapshot.agent.md +24 -0
- package/agents/fleet-review-tests.agent.md +46 -0
- package/agents/fleet-review-types.agent.md +41 -0
- package/agents/fleet-review-verifier.agent.md +29 -0
- package/commands/review-and-fix.md +98 -0
- package/commands/run-merge-bot.md +357 -0
- package/commands/run-team-help.md +17 -0
- package/package.json +23 -0
- package/scripts/agent-test-dir-prose.test.mjs +59 -0
- package/scripts/agent-test.test.mjs +1636 -0
- package/scripts/ambient-git-vars-mjs-prose.test.mjs +354 -0
- package/scripts/ambient-git-vars-prose.test.mjs +192 -0
- package/scripts/ancestry-check-position-prose.test.mjs +175 -0
- package/scripts/apply-ruleset.test.mjs +420 -0
- package/scripts/arg-header-probes-prose.test.mjs +393 -0
- package/scripts/arg.mjs +600 -0
- package/scripts/arg.test.mjs +1354 -0
- package/scripts/board-cli.test.mjs +694 -0
- package/scripts/board-identity.test.mjs +429 -0
- package/scripts/board-liveness-banner.test.mjs +119 -0
- package/scripts/board-prev-shape.test.mjs +229 -0
- package/scripts/board-tryparse-null.test.mjs +119 -0
- package/scripts/board.html +471 -0
- package/scripts/board.mjs +2037 -0
- package/scripts/board.test.mjs +3086 -0
- package/scripts/brief-after-cut-prose.test.mjs +53 -0
- package/scripts/candidates-exit3-prose.test.mjs +302 -0
- package/scripts/candidates.mjs +674 -0
- package/scripts/candidates.test.mjs +2101 -0
- package/scripts/caveat-unification-prose.test.mjs +98 -0
- package/scripts/ci-comment-rot-prose.test.mjs +2025 -0
- package/scripts/ci-completes-premise-prose.test.mjs +257 -0
- package/scripts/ci-state-prose.test.mjs +113 -0
- package/scripts/ci-state.mjs +761 -0
- package/scripts/ci-state.test.mjs +1640 -0
- package/scripts/ci-vacuous-green.test.mjs +368 -0
- package/scripts/citation-sweep-prose.test.mjs +1572 -0
- package/scripts/claim-lifecycle.test.mjs +185 -0
- package/scripts/claim-ticket.sh +417 -0
- package/scripts/claim-ticket.test.mjs +1121 -0
- package/scripts/closing-keyword-prose.test.mjs +193 -0
- package/scripts/cockpit-launch-idempotent-prose.test.mjs +165 -0
- package/scripts/cockpit-tick-nonblocking.test.mjs +210 -0
- package/scripts/compute-board.mjs +579 -0
- package/scripts/compute-board.test.mjs +1060 -0
- package/scripts/compute-spend.mjs +329 -0
- package/scripts/compute-spend.test.mjs +451 -0
- package/scripts/cross-repo-citation-prose.test.mjs +156 -0
- package/scripts/cwd-isolation-pins.mjs +83 -0
- package/scripts/dedupe-guard-prose.test.mjs +243 -0
- package/scripts/derive-testcmd.sh +239 -0
- package/scripts/derive-testcmd.test.mjs +380 -0
- package/scripts/diff-stats.mjs +231 -0
- package/scripts/diff-stats.test.mjs +408 -0
- package/scripts/dispatch-block-golden-prose.test.mjs +1024 -0
- package/scripts/dispatch-block-pins-prose.test.mjs +615 -0
- package/scripts/drop-merged-label.sh +173 -0
- package/scripts/drop-merged-label.test.mjs +480 -0
- package/scripts/expected-jobs-prose.test.mjs +80 -0
- package/scripts/filing-bar-seat-prose.test.mjs +242 -0
- package/scripts/filing-claim-bar-prose.test.mjs +166 -0
- package/scripts/finisher-dispatch-premise-prose.test.mjs +312 -0
- package/scripts/finisher-halt-prose.test.mjs +106 -0
- package/scripts/finisher-name-prose.test.mjs +134 -0
- package/scripts/finisher-own-tree-prose.test.mjs +285 -0
- package/scripts/finisher-pin-race-prose.test.mjs +339 -0
- package/scripts/fix-applier-correction-rules-prose.test.mjs +201 -0
- package/scripts/fix-applier-result-file-prose.test.mjs +100 -0
- package/scripts/fleet-bootstrap +155 -0
- package/scripts/fleet-heartbeat.mjs +272 -0
- package/scripts/fleet-heartbeat.test.mjs +450 -0
- package/scripts/fleet-provenance +207 -0
- package/scripts/fleet-provenance.test.mjs +64 -0
- package/scripts/fleet-run +148 -0
- package/scripts/fleet-run.test.mjs +90 -0
- package/scripts/fleet-state.mjs +400 -0
- package/scripts/fleet-state.test.mjs +373 -0
- package/scripts/fleet-tick-prose.test.mjs +299 -0
- package/scripts/fleet-tick.mjs +832 -0
- package/scripts/fleet-tick.test.mjs +1332 -0
- package/scripts/force-with-lease-reason-prose.test.mjs +184 -0
- package/scripts/frontmatter-allowlist.json +64 -0
- package/scripts/frontmatter-check.mjs +256 -0
- package/scripts/frontmatter-check.test.mjs +436 -0
- package/scripts/git-env.mjs +124 -0
- package/scripts/git-env.test.mjs +149 -0
- package/scripts/git-status-untracked-mode-sweep.test.mjs +216 -0
- package/scripts/gone-walk.test.mjs +213 -0
- package/scripts/gp-sep-invariant-prose.test.mjs +188 -0
- package/scripts/head-moved-after-label-prose.test.mjs +233 -0
- package/scripts/immutable-body-claim-prose.test.mjs +413 -0
- package/scripts/implementer-model-tier.test.mjs +643 -0
- package/scripts/inflight-citation-prose.test.mjs +144 -0
- package/scripts/inflight-exit2-prose.test.mjs +81 -0
- package/scripts/inflight-probe-guard-prose.test.mjs +109 -0
- package/scripts/inflight-scan-scope-prose.test.mjs +57 -0
- package/scripts/inflight.sh +1106 -0
- package/scripts/inflight.test.mjs +2984 -0
- package/scripts/injection-control-prose.test.mjs +404 -0
- package/scripts/install-root-audit.test.mjs +464 -0
- package/scripts/instrument-check-prose.test.mjs +235 -0
- package/scripts/instruments.sh +394 -0
- package/scripts/instruments.test.mjs +787 -0
- package/scripts/issue-tracker-prose.test.mjs +71 -0
- package/scripts/json.sh +248 -0
- package/scripts/json.test.mjs +364 -0
- package/scripts/ledger-dispatch.test.mjs +516 -0
- package/scripts/ledger-grammar.mjs +85 -0
- package/scripts/ledger-grammar.test.mjs +142 -0
- package/scripts/ledger-read-require-file.test.mjs +197 -0
- package/scripts/ledger.mjs +1753 -0
- package/scripts/ledger.test.mjs +3554 -0
- package/scripts/line-distance-prose.test.mjs +245 -0
- package/scripts/liveness-rationale-prose.test.mjs +83 -0
- package/scripts/locale-pin-prose.test.mjs +189 -0
- package/scripts/member-outcome-prose.test.mjs +167 -0
- package/scripts/member-outcomes-header.test.mjs +148 -0
- package/scripts/member-outcomes-prose.test.mjs +63 -0
- package/scripts/member-outcomes.mjs +231 -0
- package/scripts/member-outcomes.test.mjs +507 -0
- package/scripts/member-prompt-prose.test.mjs +161 -0
- package/scripts/member-record.mjs +605 -0
- package/scripts/member-record.test.mjs +516 -0
- package/scripts/member-write-guard.mjs +279 -0
- package/scripts/member-write-guard.test.mjs +211 -0
- package/scripts/merge-commit-probe-prose.test.mjs +122 -0
- package/scripts/merge-gate.mjs +262 -0
- package/scripts/merge-gate.test.mjs +613 -0
- package/scripts/muted-git-guard-sweep.test.mjs +493 -0
- package/scripts/net-ssh-precedence-prose.test.mjs +91 -0
- package/scripts/net-watchdog-prose.test.mjs +79 -0
- package/scripts/net.sh +409 -0
- package/scripts/net.test.mjs +211 -0
- package/scripts/no-ci-gate-prose.test.mjs +108 -0
- package/scripts/no-undo-audit.sh +1134 -0
- package/scripts/no-undo-audit.test.mjs +3965 -0
- package/scripts/no-wave.test.mjs +138 -0
- package/scripts/node-floor-sweep.test.mjs +611 -0
- package/scripts/pointer-target-prose.test.mjs +186 -0
- package/scripts/pr-overlap.mjs +417 -0
- package/scripts/pr-overlap.test.mjs +695 -0
- package/scripts/printf-die-sweep.test.mjs +384 -0
- package/scripts/prompt-renderer.mjs +39 -0
- package/scripts/prompt-renderer.test.mjs +68 -0
- package/scripts/prose-pin.mjs +808 -0
- package/scripts/prose-pin.test.mjs +1034 -0
- package/scripts/prove-merge.sh +228 -0
- package/scripts/prove-merge.test.mjs +995 -0
- package/scripts/pull-table-relabel-prose.test.mjs +60 -0
- package/scripts/quiet-payload-prose.test.mjs +222 -0
- package/scripts/reap.sh +1768 -0
- package/scripts/reap.test.mjs +4395 -0
- package/scripts/reaping-prose.test.mjs +159 -0
- package/scripts/refuter-scratch-prose.test.mjs +135 -0
- package/scripts/release-label-required-prose.test.mjs +71 -0
- package/scripts/release-label-split-prose.test.mjs +169 -0
- package/scripts/release-ticket.sh +1716 -0
- package/scripts/release-ticket.test.mjs +5106 -0
- package/scripts/renovate-release-contract.test.mjs +379 -0
- package/scripts/repo-root.mjs +544 -0
- package/scripts/repo-root.test.mjs +873 -0
- package/scripts/reporter-output-rule-prose.test.mjs +208 -0
- package/scripts/review-agent-report-only.test.mjs +49 -0
- package/scripts/review-core-cwd-isolation.test.mjs +374 -0
- package/scripts/review-core-reads.test.mjs +478 -0
- package/scripts/review-core-refuter-scratch.test.mjs +148 -0
- package/scripts/review-core-snapshot-path.test.mjs +777 -0
- package/scripts/review-core-specialist-scratch.test.mjs +189 -0
- package/scripts/review-core-testcmd.test.mjs +288 -0
- package/scripts/review-core-unrun.test.mjs +530 -0
- package/scripts/review-core-unverified-discriminant.test.mjs +208 -0
- package/scripts/review-core.mjs +1028 -0
- package/scripts/review-eval-resolver-path.test.mjs +55 -0
- package/scripts/review-eval.mjs +197 -0
- package/scripts/review-host-fixture.mjs +73 -0
- package/scripts/review-in-run-retry.test.mjs +187 -0
- package/scripts/review-no-vendor-dispatch.test.mjs +45 -0
- package/scripts/review-path-default.test.mjs +796 -0
- package/scripts/review-runner-spawns.test.mjs +93 -0
- package/scripts/review-runner.test.mjs +155 -0
- package/scripts/review-tier-audit.test.mjs +140 -0
- package/scripts/run-artifact-pr-prose.test.mjs +341 -0
- package/scripts/run-merge-bot-prose.test.mjs +1391 -0
- package/scripts/scratch-partition-invariant-prose.test.mjs +74 -0
- package/scripts/scripts-path-citation-sweep.test.mjs +73 -0
- package/scripts/section-between-guard-prose.test.mjs +378 -0
- package/scripts/select-dimensions.test.mjs +674 -0
- package/scripts/shared-refusal.test.mjs +388 -0
- package/scripts/shell-traps-prose.test.mjs +507 -0
- package/scripts/shortlist.mjs +355 -0
- package/scripts/shortlist.test.mjs +366 -0
- package/scripts/slow-transport.mjs +72 -0
- package/scripts/snapshot-repo.test.mjs +772 -0
- package/scripts/snapshot-runner-audience-prose.test.mjs +129 -0
- package/scripts/spend-view.test.mjs +312 -0
- package/scripts/staleness-qualifier-prose.test.mjs +251 -0
- package/scripts/staleness-verdict-prose.test.mjs +142 -0
- package/scripts/staleness.mjs +414 -0
- package/scripts/staleness.test.mjs +893 -0
- package/scripts/stalled-member-liveness-prose.test.mjs +219 -0
- package/scripts/stash-prohibition-prose.test.mjs +313 -0
- package/scripts/strip-comments.mjs +44 -0
- package/scripts/tier-check.mjs +468 -0
- package/scripts/tier-check.test.mjs +726 -0
- package/scripts/tier-guard-gate-prose.test.mjs +216 -0
- package/scripts/tier-outcomes-header.test.mjs +586 -0
- package/scripts/tier-outcomes.mjs +441 -0
- package/scripts/tier-outcomes.test.mjs +342 -0
- package/scripts/tier-roles.mjs +250 -0
- package/scripts/tier-roles.test.mjs +316 -0
- package/scripts/tracker-block-copy-prose.test.mjs +288 -0
- package/scripts/triage-labels-gloss-prose.test.mjs +97 -0
- package/scripts/unattended-git-sweep.test.mjs +142 -0
- package/scripts/verify-sha-exit2-prose.test.mjs +269 -0
- package/scripts/verify-sha.sh +117 -0
- package/scripts/verify-sha.test.mjs +693 -0
- package/scripts/watcher-degraded-block.test.mjs +372 -0
- package/scripts/watcher-key-prose.test.mjs +106 -0
- package/scripts/within-run-pair-prose.test.mjs +208 -0
- package/scripts/worktree-audit.sh +482 -0
- package/scripts/worktree-audit.test.mjs +1407 -0
- package/scripts/worktree-head-prose.test.mjs +276 -0
- package/scripts/worktree-listing-sweep.test.mjs +217 -0
- package/scripts/worktree-model-prose.test.mjs +216 -0
- package/scripts/worktree.sh +470 -0
- package/skills/next-ticket/SKILL.md +138 -0
- package/skills/run-team/SKILL.md +3603 -0
- package/skills/run-team/references/ci-and-staleness.md +75 -0
- package/skills/run-team/references/correction-tickets.md +37 -0
- package/skills/run-team/references/isolation.md +23 -0
- package/skills/run-team/references/member-lifecycle.md +73 -0
- package/skills/run-team/references/reaping.md +73 -0
- package/skills/sizing-a-ticket/SKILL.md +39 -0
|
@@ -0,0 +1,674 @@
|
|
|
1
|
+
import { test } from "node:test";
|
|
2
|
+
import assert from "node:assert/strict";
|
|
3
|
+
import { readFileSync } from "node:fs";
|
|
4
|
+
import { join } from "node:path";
|
|
5
|
+
import { computeStats } from "./diff-stats.mjs";
|
|
6
|
+
import { stripComments } from "./strip-comments.mjs";
|
|
7
|
+
import { DEFAULT_DIMENSIONS, selectDimensions, resolveDimensions, verifiersFor } from "./review-core.mjs";
|
|
8
|
+
|
|
9
|
+
// The reason #118 existed: every count claim about `selectDimensions` had to
|
|
10
|
+
// be hand-derived, and two hand-derived comments were wrong. Every value
|
|
11
|
+
// under test here is the REAL export now, never a re-derived or lifted copy.
|
|
12
|
+
const REPO = join(import.meta.dirname, "..");
|
|
13
|
+
const SOURCE = readFileSync(join(REPO, "scripts", "review-core.mjs"), "utf8");
|
|
14
|
+
|
|
15
|
+
// Drive the matrix from REAL file lists through the real classifier, not from
|
|
16
|
+
// hand-written profile strings. A `diff-stats` classifier change that silently
|
|
17
|
+
// shifts a profile then fails here instead of silently retuning the fan-out.
|
|
18
|
+
const dimensionKeys = (files) => selectDimensions(DEFAULT_DIMENSIONS, computeStats(files)).map((d) => d.key);
|
|
19
|
+
const f = (path, additions = 5, deletions = 5) => ({ path, additions, deletions });
|
|
20
|
+
|
|
21
|
+
test("the imported values have the shape the rest of this file assumes", () => {
|
|
22
|
+
assert.equal(DEFAULT_DIMENSIONS.length, 6);
|
|
23
|
+
assert.deepEqual(
|
|
24
|
+
DEFAULT_DIMENSIONS.map((d) => d.key),
|
|
25
|
+
["correctness", "silent-failure", "tests", "comments", "types", "simplify"],
|
|
26
|
+
);
|
|
27
|
+
});
|
|
28
|
+
|
|
29
|
+
test("unknown, unparseable or empty stats widen to the full set", () => {
|
|
30
|
+
assert.equal(selectDimensions(DEFAULT_DIMENSIONS, null).length, 6);
|
|
31
|
+
assert.equal(selectDimensions(DEFAULT_DIMENSIONS, undefined).length, 6);
|
|
32
|
+
// An empty `files` array is NOT a signal to trim: `gh` can report no files for
|
|
33
|
+
// a real PR, and treating that as "nothing to review" would drop FOUR
|
|
34
|
+
// dimensions on production code.
|
|
35
|
+
assert.equal(dimensionKeys([]).length, 6);
|
|
36
|
+
});
|
|
37
|
+
|
|
38
|
+
test("a docs-only diff runs correctness + comments", () => {
|
|
39
|
+
assert.deepEqual(dimensionKeys([f("README.md")]), ["correctness", "comments"]);
|
|
40
|
+
});
|
|
41
|
+
|
|
42
|
+
test("a tests-only diff drops types, silent-failure and simplify", () => {
|
|
43
|
+
assert.deepEqual(dimensionKeys([f("scripts/foo.test.mjs")]), ["correctness", "tests", "comments"]);
|
|
44
|
+
});
|
|
45
|
+
|
|
46
|
+
test("a production diff runs everything, less `tests` when the diff has none", () => {
|
|
47
|
+
const withTests = [f("scripts/review-core.mjs", 100, 50), f("scripts/a.test.mjs", 30, 10)];
|
|
48
|
+
assert.equal(dimensionKeys(withTests).length, 6);
|
|
49
|
+
|
|
50
|
+
const noTests = [f("scripts/review-core.mjs", 100, 50), f("scripts/b.mjs", 40, 20)];
|
|
51
|
+
const keys = dimensionKeys(noTests);
|
|
52
|
+
assert.equal(keys.length, 5);
|
|
53
|
+
assert.ok(!keys.includes("tests"));
|
|
54
|
+
});
|
|
55
|
+
|
|
56
|
+
// --- Size tier. `profile` is assigned through an else-if chain in computeStats,
|
|
57
|
+
// --- so these values are mutually exclusive and cannot race `docsOnly`.
|
|
58
|
+
// #218: `comments` is part of the size-tier floor, so this row carries it. This
|
|
59
|
+
// single `.js` file whose whole substance is prose scores `docs: 0`, because
|
|
60
|
+
// `classify()` returns `src` for any code extension before it checks `isDocs`
|
|
61
|
+
// — one shape of the diffs #218 measured in production (four of them, below).
|
|
62
|
+
test("a single-file source diff trims to correctness + silent-failure + comments", () => {
|
|
63
|
+
assert.deepEqual(dimensionKeys([f("scripts/review-core.mjs", 3, 2)]), [
|
|
64
|
+
"correctness",
|
|
65
|
+
"silent-failure",
|
|
66
|
+
"comments",
|
|
67
|
+
]);
|
|
68
|
+
});
|
|
69
|
+
|
|
70
|
+
// The measured instances, as file lists, driven through the real classifier: a
|
|
71
|
+
// comment-only `.mjs` (#682), a comment-only `.mjs` (#710), a `.sh` + its test
|
|
72
|
+
// (#1091), and a majority-comment `.sh` + test at `small` (#1172). Every one
|
|
73
|
+
// scores `docs: 0` and every one now keeps `comments`. Under the old
|
|
74
|
+
// `stats.kinds?.docs !== 0` carve-out all four returned without it.
|
|
75
|
+
test("the diffs #218 measured in production all keep comments — none of them has a docs file", () => {
|
|
76
|
+
for (const files of [
|
|
77
|
+
[f("scripts/board.mjs", 1, 1)],
|
|
78
|
+
[f("scripts/arg.mjs", 2, 1)],
|
|
79
|
+
[f("scripts/no-undo-audit.sh", 1, 1), f("scripts/no-undo-audit.test.mjs", 1, 1)],
|
|
80
|
+
[f("scripts/claim-ticket.sh", 7, 0), f("scripts/claim-ticket.test.mjs", 20, 2)],
|
|
81
|
+
]) {
|
|
82
|
+
const stats = computeStats(files);
|
|
83
|
+
assert.equal(stats.kinds.docs, 0, `${files.map((x) => x.path)} unexpectedly scores a docs file`);
|
|
84
|
+
assert.ok(
|
|
85
|
+
dimensionKeys(files).includes("comments"),
|
|
86
|
+
`${files.map((x) => x.path)} lost the comments dimension`,
|
|
87
|
+
);
|
|
88
|
+
}
|
|
89
|
+
});
|
|
90
|
+
|
|
91
|
+
// The size-tier floor OUTRANKS the hasSrc guard for `silent-failure` (#236). A
|
|
92
|
+
// single `.github/workflows/ci.yml` change is profile "single-file" with hasSrc
|
|
93
|
+
// false, so the guard had already removed silent-failure before the tier ran and
|
|
94
|
+
// the documented floor did not exist: PR #226 reviewed CI's own gating logic with
|
|
95
|
+
// `dimensionsRun: ["correctness"]`. A CI workflow diff is mostly shell, which is
|
|
96
|
+
// what the silent-failure hunter is for, and it is where this fleet got bitten.
|
|
97
|
+
test("a single-file config diff keeps the silent-failure floor", () => {
|
|
98
|
+
assert.deepEqual(dimensionKeys([f(".github/workflows/ci.yml", 2, 1)]), [
|
|
99
|
+
"correctness",
|
|
100
|
+
"silent-failure",
|
|
101
|
+
"comments",
|
|
102
|
+
]);
|
|
103
|
+
// Shell under `.github/` is the ticket's other named case, and it classifies
|
|
104
|
+
// `config` for the same reason. A `.sh` ANYWHERE ELSE is classify()'s `src`
|
|
105
|
+
// residue, so it always had silent-failure and is not part of this class.
|
|
106
|
+
assert.deepEqual(dimensionKeys([f(".github/scripts/release.sh", 4, 2)]), [
|
|
107
|
+
"correctness",
|
|
108
|
+
"silent-failure",
|
|
109
|
+
"comments",
|
|
110
|
+
]);
|
|
111
|
+
});
|
|
112
|
+
|
|
113
|
+
// AC-2: `small` is the other size-tier profile and gets the same floor. Two
|
|
114
|
+
// config files at 6 loc — `single-file` needs files === 1, so this row can only
|
|
115
|
+
// reach the tier through `small`, and it went in as [correctness] alone too.
|
|
116
|
+
test("a small config-only diff keeps the silent-failure floor", () => {
|
|
117
|
+
assert.deepEqual(
|
|
118
|
+
dimensionKeys([f(".github/workflows/ci.yml", 2, 1), f(".github/workflows/release.yml", 2, 1)]),
|
|
119
|
+
["correctness", "silent-failure", "comments"],
|
|
120
|
+
);
|
|
121
|
+
});
|
|
122
|
+
|
|
123
|
+
// The floor's own former boundary. OUTSIDE the size tier the hasSrc guard used
|
|
124
|
+
// to drop silent-failure again — five config files at 50 loc profile
|
|
125
|
+
// `production`, and neither #236's floor (SIZE_TIER_PROFILES) nor #739's
|
|
126
|
+
// `tests-only` arm reached it, so growing a config-only diff past the tier
|
|
127
|
+
// LOST the specialist a smaller diff of that SAME KIND kept: non-monotonic on
|
|
128
|
+
// #236's own motivating class, the sibling gap #1285 closed alongside #739.
|
|
129
|
+
// Fixed by a third `keepsSilentFailure` arm gated on `hasTests === false`: a
|
|
130
|
+
// config-only diff keeps silent-failure regardless of size, the same way a
|
|
131
|
+
// config+test diff already does via #739.
|
|
132
|
+
test("a large config-only diff keeps the silent-failure floor even past the size tier (#1285)", () => {
|
|
133
|
+
const files = [
|
|
134
|
+
f(".github/workflows/a.yml", 6, 4),
|
|
135
|
+
f(".github/workflows/b.yml", 6, 4),
|
|
136
|
+
f(".github/workflows/c.yml", 6, 4),
|
|
137
|
+
f("tsconfig.json", 6, 4),
|
|
138
|
+
f("package.json", 6, 4),
|
|
139
|
+
];
|
|
140
|
+
assert.equal(computeStats(files).profile, "production");
|
|
141
|
+
assert.deepEqual(dimensionKeys(files), ["correctness", "silent-failure", "comments"]);
|
|
142
|
+
});
|
|
143
|
+
|
|
144
|
+
// #1285's own reproduction, mirrored on #739's own test shape below: a
|
|
145
|
+
// config-only diff UNDER the size tier keeps silent-failure (`small`), and
|
|
146
|
+
// adding a SECOND config file that pushes `loc` past the tier into
|
|
147
|
+
// `production` must not subtract it. Crossing into a larger, more-scrutinized
|
|
148
|
+
// profile can only ADD scrutiny, never remove a specialist a smaller diff of
|
|
149
|
+
// the same kind already had — the monotonicity property this file's
|
|
150
|
+
// dimension-selection logic exists to guarantee.
|
|
151
|
+
test("a second config file that crosses the size tier into production does not drop silent-failure", () => {
|
|
152
|
+
const underTier = [f(".github/workflows/ci.yml", 3, 2), f("tsconfig.json", 3, 2)];
|
|
153
|
+
assert.equal(computeStats(underTier).profile, "small");
|
|
154
|
+
assert.deepEqual(dimensionKeys(underTier), ["correctness", "silent-failure", "comments"]);
|
|
155
|
+
|
|
156
|
+
const crossesTier = [...underTier, f("package.json", 30, 10)];
|
|
157
|
+
assert.equal(computeStats(crossesTier).profile, "production");
|
|
158
|
+
assert.deepEqual(dimensionKeys(crossesTier), ["correctness", "silent-failure", "comments"]);
|
|
159
|
+
});
|
|
160
|
+
|
|
161
|
+
// #739. `tests-only` is assigned AHEAD of `single-file`/`small` in computeStats'
|
|
162
|
+
// else-if chain, so a config diff that also touches a test file never reaches
|
|
163
|
+
// SIZE_TIER_PROFILES and the #236 floor could not re-admit silent-failure. That
|
|
164
|
+
// made coverage NON-MONOTONIC: `ci.yml` alone kept the hunter and `ci.yml` + a
|
|
165
|
+
// test SUBTRACTED it — adding a file to a diff removed a specialist, on #236's
|
|
166
|
+
// own motivating class (CI gating logic, no src). The gate's other half — a pure
|
|
167
|
+
// test diff has no config, so it must still lose silent-failure — is already
|
|
168
|
+
// pinned above by `a tests-only diff drops types, silent-failure and simplify`.
|
|
169
|
+
test("a tests-only diff keeps the silent-failure floor when it also carries config", () => {
|
|
170
|
+
const configPlusTest = [
|
|
171
|
+
f(".github/workflows/ci.yml", 2, 1),
|
|
172
|
+
f("scripts/ci-vacuous-green.test.mjs", 3, 1),
|
|
173
|
+
];
|
|
174
|
+
assert.equal(computeStats(configPlusTest).profile, "tests-only");
|
|
175
|
+
assert.deepEqual(dimensionKeys(configPlusTest), ["correctness", "silent-failure", "tests", "comments"]);
|
|
176
|
+
});
|
|
177
|
+
|
|
178
|
+
// AC-3: the floor must not reach a diff with nothing for it to do, and the ORDER
|
|
179
|
+
// is the whole guarantee — `docsOnly` returns BEFORE the tier. A file list cannot
|
|
180
|
+
// pin that: computeStats assigns profile `docs` ahead of `files === 1`, so a docs
|
|
181
|
+
// diff never matches SIZE_TIER_PROFILES, both branches agree on [correctness,
|
|
182
|
+
// comments], and `a docs-only diff runs correctness + comments` above already
|
|
183
|
+
// pins that agreement. Pin the DIRECTION instead, on the one blob where the two
|
|
184
|
+
// disagree — the same shape as the missing-`kinds` row below. Measured: leaking
|
|
185
|
+
// the tier into the docsOnly filter, and deleting that branch outright, each
|
|
186
|
+
// passed all 1030 tests without this.
|
|
187
|
+
test("docsOnly returns before the size tier — the floor never reaches a docs diff", () => {
|
|
188
|
+
const docsInTier = { profile: "single-file", docsOnly: true, hasSrc: false, hasTests: false };
|
|
189
|
+
assert.deepEqual(
|
|
190
|
+
selectDimensions(DEFAULT_DIMENSIONS, docsInTier).map((d) => d.key),
|
|
191
|
+
["correctness", "comments"],
|
|
192
|
+
);
|
|
193
|
+
});
|
|
194
|
+
|
|
195
|
+
test("a small multi-file source diff trims to correctness + silent-failure + comments", () => {
|
|
196
|
+
assert.deepEqual(
|
|
197
|
+
dimensionKeys([f("scripts/a.mjs", 5, 5), f("scripts/b.mjs", 5, 4)]),
|
|
198
|
+
["correctness", "silent-failure", "comments"],
|
|
199
|
+
);
|
|
200
|
+
});
|
|
201
|
+
|
|
202
|
+
// Same carve-out shape for `tests`: when the diff's substance IS a test, the
|
|
203
|
+
// mutation-discrimination check is the one it most needs. Without this the row
|
|
204
|
+
// below trims to whatever else the size tier keeps (`SIZE_TIER_DIMS`) — not
|
|
205
|
+
// restated here as a count, which is exactly what went stale last time (#218).
|
|
206
|
+
test("a small diff that adds a test keeps the tests dimension", () => {
|
|
207
|
+
assert.deepEqual(
|
|
208
|
+
dimensionKeys([f("scripts/a.mjs", 5, 5), f("scripts/a.test.mjs", 5, 5)]),
|
|
209
|
+
["correctness", "silent-failure", "tests", "comments"],
|
|
210
|
+
);
|
|
211
|
+
});
|
|
212
|
+
|
|
213
|
+
// Fail DIRECTION, not a matrix row. The blob reaches review-core.mjs relayed by an
|
|
214
|
+
// agent, so a field can go missing without failing JSON.parse, and absence must
|
|
215
|
+
// widen rather than narrow — the `=== true` guards elsewhere in selectDimensions
|
|
216
|
+
// exist for that. `comments` no longer reads any field at all since #218, so what
|
|
217
|
+
// this row pins is the surviving half: `tests` is `=== true`, so a blob without
|
|
218
|
+
// `hasTests` loses it, while `comments` cannot be lost to a missing field.
|
|
219
|
+
test("a size-tier stats blob missing `kinds` and `hasTests` still keeps comments", () => {
|
|
220
|
+
const stats = { profile: "small", docsOnly: false, hasSrc: true };
|
|
221
|
+
assert.deepEqual(
|
|
222
|
+
selectDimensions(DEFAULT_DIMENSIONS, stats).map((d) => d.key),
|
|
223
|
+
["correctness", "silent-failure", "comments"],
|
|
224
|
+
);
|
|
225
|
+
});
|
|
226
|
+
|
|
227
|
+
// #1349 removed the per-call `model` field entirely (per #1303's gap 3): no
|
|
228
|
+
// `agent()` call may carry `model`/`effort`, so DEFAULT_DIMENSIONS carries
|
|
229
|
+
// only `key`/`prompt`/`agentType` now, and tier lives in each fleet-owned
|
|
230
|
+
// `fleet-review-<key>` definition's own frontmatter instead. omp's native
|
|
231
|
+
// `agent()` resolves a bare frontmatter `name:` exactly, so `agentType` is
|
|
232
|
+
// bare here — no namespace prefix to carry (omp's own registry is
|
|
233
|
+
// bare-name/unnamespaced, ADR 0014). This replaces
|
|
234
|
+
// the retired "only the recoverable-miss dimensions carry a model downgrade"
|
|
235
|
+
// test, which pinned a field that no longer exists.
|
|
236
|
+
test("every dimension names a distinct fleet-owned agentType and carries no model field", () => {
|
|
237
|
+
const byKey = Object.fromEntries(DEFAULT_DIMENSIONS.map((d) => [d.key, d]));
|
|
238
|
+
for (const key of ["correctness", "silent-failure", "tests", "comments", "types", "simplify"]) {
|
|
239
|
+
assert.equal(byKey[key].agentType, `fleet-review-${key}`);
|
|
240
|
+
assert.equal(byKey[key].model, undefined, `${key} still carries a model field`);
|
|
241
|
+
}
|
|
242
|
+
assert.equal(new Set(DEFAULT_DIMENSIONS.map((d) => d.agentType)).size, DEFAULT_DIMENSIONS.length);
|
|
243
|
+
});
|
|
244
|
+
|
|
245
|
+
// The test above pins WHICH dimensions carry the downgrade. This pins the fact
|
|
246
|
+
// its comment rests on: `verifiersFor` is keyed on SEVERITY ALONE, so no
|
|
247
|
+
// dimension can draw a different refuter budget and "faces refuters" was never
|
|
248
|
+
// able to separate these six (#221). Without this the corrected rationale is
|
|
249
|
+
// prose with nothing under it — which is how the retired one survived.
|
|
250
|
+
test("the refuter budget is keyed on severity alone, never on a dimension", () => {
|
|
251
|
+
const forSeverity = verifiersFor({ verifiers: 7 });
|
|
252
|
+
// One parameter, and it is the severity. A dimension-aware budget needs a
|
|
253
|
+
// second one, and passing a dimension anyway must not move the answer.
|
|
254
|
+
assert.equal(forSeverity.length, 1);
|
|
255
|
+
for (const d of DEFAULT_DIMENSIONS)
|
|
256
|
+
assert.equal(forSeverity("critical", d.key), 7, `critical/${d.key}`);
|
|
257
|
+
// `silent-failure` and `tests` are dispatched under different fleet-owned
|
|
258
|
+
// agent tiers (frontmatter, not a per-call field) and still draw the same
|
|
259
|
+
// budget — the whole of why the retired rule did not
|
|
260
|
+
// separate them. `suggestion` is the one band that differs, and it is a
|
|
261
|
+
// SEVERITY, not a dimension.
|
|
262
|
+
assert.equal(forSeverity("important"), forSeverity("critical"));
|
|
263
|
+
assert.equal(forSeverity("suggestion"), 0);
|
|
264
|
+
});
|
|
265
|
+
|
|
266
|
+
// --- args.dimensions normalization (#113). The "Specialists" section of
|
|
267
|
+
// `commands/review-and-fix.md` documents the override as
|
|
268
|
+
// accepting "keys or dimension objects"; before this, only objects worked and
|
|
269
|
+
// a key array passed through untouched, dereferencing `d.key`, `d.prompt`,
|
|
270
|
+
// `d.model` and `d.agentType` to `undefined` with no throw and no warning.
|
|
271
|
+
test("no override leaves the size tier in charge", () => {
|
|
272
|
+
assert.equal(resolveDimensions(undefined, DEFAULT_DIMENSIONS), null);
|
|
273
|
+
assert.equal(resolveDimensions(null, DEFAULT_DIMENSIONS), null);
|
|
274
|
+
});
|
|
275
|
+
|
|
276
|
+
// ...but only an ABSENT override does. A falsy-but-present one is a caller
|
|
277
|
+
// error, and falling back to the size tier would silently run a DIFFERENT set
|
|
278
|
+
// than the one that was pinned, with nothing in the log naming the override.
|
|
279
|
+
test("a falsy-but-present override stops the run rather than degrading to the size tier", () => {
|
|
280
|
+
for (const override of ["", 0, false, NaN]) {
|
|
281
|
+
assert.throws(() => resolveDimensions(override, DEFAULT_DIMENSIONS), /must be an array/);
|
|
282
|
+
}
|
|
283
|
+
});
|
|
284
|
+
|
|
285
|
+
test("an object override behaves as it does today — passed through unchanged", () => {
|
|
286
|
+
const objs = [DEFAULT_DIMENSIONS[0], DEFAULT_DIMENSIONS[2]];
|
|
287
|
+
assert.deepEqual(resolveDimensions(objs, DEFAULT_DIMENSIONS), objs);
|
|
288
|
+
});
|
|
289
|
+
|
|
290
|
+
test("a key override resolves against the workflow's own dimension list", () => {
|
|
291
|
+
assert.deepEqual(
|
|
292
|
+
resolveDimensions(["correctness", "comments"], DEFAULT_DIMENSIONS),
|
|
293
|
+
[DEFAULT_DIMENSIONS[0], DEFAULT_DIMENSIONS[3]],
|
|
294
|
+
);
|
|
295
|
+
});
|
|
296
|
+
|
|
297
|
+
test("keys and objects can mix in the same override", () => {
|
|
298
|
+
assert.deepEqual(
|
|
299
|
+
resolveDimensions(["correctness", DEFAULT_DIMENSIONS[2]], DEFAULT_DIMENSIONS),
|
|
300
|
+
[DEFAULT_DIMENSIONS[0], DEFAULT_DIMENSIONS[2]],
|
|
301
|
+
);
|
|
302
|
+
});
|
|
303
|
+
|
|
304
|
+
test("an unknown key stops the run and names the key", () => {
|
|
305
|
+
assert.throws(
|
|
306
|
+
() => resolveDimensions(["not-a-real-dimension"], DEFAULT_DIMENSIONS),
|
|
307
|
+
/unknown key "not-a-real-dimension"/,
|
|
308
|
+
);
|
|
309
|
+
});
|
|
310
|
+
|
|
311
|
+
test("an object override missing a required field stops the run and names the field", () => {
|
|
312
|
+
assert.throws(
|
|
313
|
+
() => resolveDimensions([{ key: "x", agentType: "y" }], DEFAULT_DIMENSIONS),
|
|
314
|
+
/missing required field\(s\): prompt/,
|
|
315
|
+
);
|
|
316
|
+
// All three REQUIRED fields are checked, not just one. `model` is not among
|
|
317
|
+
// them — #1349 retired it entirely (dispatch tier lives only in the agent
|
|
318
|
+
// definition's own frontmatter now), and a caller that still sends one is
|
|
319
|
+
// refused outright — see the dedicated test below, not this one.
|
|
320
|
+
assert.throws(
|
|
321
|
+
() => resolveDimensions([{}], DEFAULT_DIMENSIONS),
|
|
322
|
+
/missing required field\(s\): key, prompt, agentType/,
|
|
323
|
+
);
|
|
324
|
+
assert.deepEqual(
|
|
325
|
+
resolveDimensions([{ key: "x", prompt: "p", agentType: "a" }], DEFAULT_DIMENSIONS),
|
|
326
|
+
[{ key: "x", prompt: "p", agentType: "a" }],
|
|
327
|
+
);
|
|
328
|
+
});
|
|
329
|
+
|
|
330
|
+
// Presence alone let a non-string field (a number, an object, ...) straight
|
|
331
|
+
// through to the specialist dispatch machinery with no throw, contradicting
|
|
332
|
+
// this function's own job of validating the boundary. Mutation-verified: with
|
|
333
|
+
// the `wrongType` check in resolveDimensions loosened back to the
|
|
334
|
+
// presence-only `missing` check, this test reds.
|
|
335
|
+
test("a required field of the wrong type stops the run and names the field and the type it got", () => {
|
|
336
|
+
assert.throws(
|
|
337
|
+
() => resolveDimensions([{ key: 42, prompt: "p", agentType: "a" }], DEFAULT_DIMENSIONS),
|
|
338
|
+
/review-pr: args\.dimensions\[0\] field "key" must be a string, got number$/,
|
|
339
|
+
);
|
|
340
|
+
// `0`, `""`, `null` etc are already caught as MISSING (falsy) above — a
|
|
341
|
+
// truthy non-string, like an object, is the case the presence check lets
|
|
342
|
+
// through unnoticed.
|
|
343
|
+
assert.throws(
|
|
344
|
+
() => resolveDimensions([{ key: "x", prompt: {}, agentType: "a" }], DEFAULT_DIMENSIONS),
|
|
345
|
+
/review-pr: args\.dimensions\[0\] field "prompt" must be a string, got object$/,
|
|
346
|
+
);
|
|
347
|
+
assert.deepEqual(
|
|
348
|
+
resolveDimensions([{ key: "x", prompt: "p", agentType: "a" }], DEFAULT_DIMENSIONS),
|
|
349
|
+
[{ key: "x", prompt: "p", agentType: "a" }],
|
|
350
|
+
);
|
|
351
|
+
});
|
|
352
|
+
|
|
353
|
+
// #1349 (per #1303's gap 3): `model` used to be a fourth, optional field on a
|
|
354
|
+
// dimension override, dereferenced at dispatch to override the tier. Now it
|
|
355
|
+
// is refused OUTRIGHT — any value, not just a wrong-typed one — because tier
|
|
356
|
+
// lives only in the fleet-owned agent definition's frontmatter, never per
|
|
357
|
+
// call. A caller-facing knob that silently worked on one harness and not the
|
|
358
|
+
// other is the exact shape #1349 exists to close, so this is loud rather than
|
|
359
|
+
// a silent no-op.
|
|
360
|
+
test("a dimension override carrying a model field is refused outright, regardless of its type", () => {
|
|
361
|
+
assert.throws(
|
|
362
|
+
() => resolveDimensions([{ key: "x", prompt: "p", agentType: "a", model: "opus" }], DEFAULT_DIMENSIONS),
|
|
363
|
+
/review-pr: args\.dimensions\[0\] field "model" is no longer supported/,
|
|
364
|
+
);
|
|
365
|
+
assert.throws(
|
|
366
|
+
() => resolveDimensions([{ key: "x", prompt: "p", agentType: "a", model: 7 }], DEFAULT_DIMENSIONS),
|
|
367
|
+
/review-pr: args\.dimensions\[0\] field "model" is no longer supported/,
|
|
368
|
+
);
|
|
369
|
+
// A missing `model` is still fine — there is nothing left to demand.
|
|
370
|
+
assert.deepEqual(
|
|
371
|
+
resolveDimensions([{ key: "x", prompt: "p", agentType: "a" }], DEFAULT_DIMENSIONS),
|
|
372
|
+
[{ key: "x", prompt: "p", agentType: "a" }],
|
|
373
|
+
);
|
|
374
|
+
});
|
|
375
|
+
|
|
376
|
+
// #279. These five all used to produce "object is missing required field(s):
|
|
377
|
+
// key, prompt, agentType" — a message asserting the entry IS an object and
|
|
378
|
+
// sending the caller off to add three fields to a `42`. The type is the defect;
|
|
379
|
+
// say so. `null` and `undefined` are the case the old `!entry?.[f]` optional
|
|
380
|
+
// chain covered by accident: it kept the run from dying on an uncaught
|
|
381
|
+
// `TypeError: Cannot read properties of null (reading 'key')`, but it answered
|
|
382
|
+
// with the wrong diagnosis. The explicit type check answers with the right one.
|
|
383
|
+
test("a non-object entry is reported as the wrong TYPE, not as an object missing fields", () => {
|
|
384
|
+
for (const [entry, name] of [
|
|
385
|
+
[42, "number"],
|
|
386
|
+
[null, "null"],
|
|
387
|
+
[undefined, "undefined"],
|
|
388
|
+
[true, "boolean"],
|
|
389
|
+
[["x"], "array"],
|
|
390
|
+
]) {
|
|
391
|
+
// Deliberately NOT anchored on the `[0]` index here — index coverage for
|
|
392
|
+
// THIS type-mismatch branch is a dedicated assertion below, not this one:
|
|
393
|
+
// dropping `[${i}]` from the throw does not fail #281's "identifies which
|
|
394
|
+
// entry" test either, since that test only exercises the missing-field
|
|
395
|
+
// and unknown-key branches. "Reverting one must red one" does not hold
|
|
396
|
+
// for this branch without that dedicated assertion.
|
|
397
|
+
assert.throws(
|
|
398
|
+
() => resolveDimensions([entry], DEFAULT_DIMENSIONS),
|
|
399
|
+
new RegExp(`review-pr: args\\.dimensions\\[0\\] must be a key string or a dimension object, got ${name}$`),
|
|
400
|
+
`a ${name} entry no longer names its own type, and not as an object missing fields`,
|
|
401
|
+
);
|
|
402
|
+
}
|
|
403
|
+
});
|
|
404
|
+
|
|
405
|
+
// The loop above only ever indexes ONE entry ([0]), so it cannot pin the
|
|
406
|
+
// index for a multi-entry override — and separately from #281's "identifies
|
|
407
|
+
// which entry" test above, which never exercises this branch at all. Dropping
|
|
408
|
+
// `[${i}]` from the type-mismatch throw (review-core.mjs) fails no
|
|
409
|
+
// test above. Mutation-verified: reverting the `${i}` in that throw back to a
|
|
410
|
+
// fixed string reds this assertion.
|
|
411
|
+
test("a type-mismatch entry error identifies which entry of a multi-entry override failed", () => {
|
|
412
|
+
assert.throws(
|
|
413
|
+
() => resolveDimensions(["correctness", 42], DEFAULT_DIMENSIONS),
|
|
414
|
+
/review-pr: args\.dimensions\[1\] must be a key string or a dimension object, got number$/,
|
|
415
|
+
);
|
|
416
|
+
});
|
|
417
|
+
|
|
418
|
+
// #281. With a multi-entry override the field names alone do not say WHICH
|
|
419
|
+
// entry is short — and the map callback has the index already. Both throwing
|
|
420
|
+
// branches carry it, so an unknown key in a long list is locatable too.
|
|
421
|
+
test("an entry error identifies which entry of a multi-entry override failed", () => {
|
|
422
|
+
assert.throws(
|
|
423
|
+
() => resolveDimensions(["correctness", "comments", { key: "x", agentType: "y" }], DEFAULT_DIMENSIONS),
|
|
424
|
+
/args\.dimensions\[2\] is missing required field\(s\): prompt/,
|
|
425
|
+
);
|
|
426
|
+
assert.throws(
|
|
427
|
+
() => resolveDimensions(["correctness", "not-a-real-dimension"], DEFAULT_DIMENSIONS),
|
|
428
|
+
/args\.dimensions\[1\] named an unknown key "not-a-real-dimension"/,
|
|
429
|
+
);
|
|
430
|
+
});
|
|
431
|
+
|
|
432
|
+
// #274. A repeated key resolved to two entries pointing at ONE dimension
|
|
433
|
+
// object: the fan-out dispatched the same specialist twice against a single
|
|
434
|
+
// scratch dir, and the coverage record double-counted it — which
|
|
435
|
+
// `run-team/SKILL.md` reads AS coverage. Redundant-but-valid input WHEN THE
|
|
436
|
+
// ENTRIES AGREE, so it dedupes silently in that case; throwing would refuse a
|
|
437
|
+
// request whose meaning is not in doubt. A DIVERGENT collision is a different
|
|
438
|
+
// case — see the dedicated test below.
|
|
439
|
+
test("a repeated key resolves once, and does not throw", () => {
|
|
440
|
+
assert.deepEqual(
|
|
441
|
+
resolveDimensions(["correctness", "comments", "correctness"], DEFAULT_DIMENSIONS),
|
|
442
|
+
[DEFAULT_DIMENSIONS[0], DEFAULT_DIMENSIONS[3]],
|
|
443
|
+
);
|
|
444
|
+
// A key and the object it names are the same dimension; first occurrence wins.
|
|
445
|
+
assert.deepEqual(
|
|
446
|
+
resolveDimensions(["correctness", DEFAULT_DIMENSIONS[0]], DEFAULT_DIMENSIONS),
|
|
447
|
+
[DEFAULT_DIMENSIONS[0]],
|
|
448
|
+
);
|
|
449
|
+
// Dedupe must not empty a set that had entries: an all-duplicate override
|
|
450
|
+
// resolves to one dimension, not to the "resolved to no dimensions" throw.
|
|
451
|
+
assert.equal(resolveDimensions(["types", "types", "types"], DEFAULT_DIMENSIONS).length, 1);
|
|
452
|
+
});
|
|
453
|
+
|
|
454
|
+
// A second entry sharing a key but DIVERGING (different `prompt`/`agentType`)
|
|
455
|
+
// used to dedupe SILENTLY too, discarding a caller's real attempt to
|
|
456
|
+
// override or customize a catalog entry with zero signal — the one silent
|
|
457
|
+
// exception to what this whole function exists to do (#279/#281 turn silent/
|
|
458
|
+
// ambiguous failures into loud, indexed ones). Mutation-verified: comparing
|
|
459
|
+
// only `d.key` in the dedupe loop (the old behaviour) reds this test; the
|
|
460
|
+
// identical-duplicate cases above stay green either way, so this is
|
|
461
|
+
// independent of them, not a replacement. `model` used to be a third
|
|
462
|
+
// divergence field here too, before #1349 retired it outright (see the
|
|
463
|
+
// dedicated "refused regardless of its type" test above) — a `model`-only
|
|
464
|
+
// divergence can no longer reach this comparison at all, since the field is
|
|
465
|
+
// refused before resolveDimensions ever gets to dedupe.
|
|
466
|
+
test("a repeated key with different fields throws — a divergent collision is not silently dropped", () => {
|
|
467
|
+
const dup = { key: "correctness", prompt: "other", agentType: "code-reviewer" };
|
|
468
|
+
assert.throws(
|
|
469
|
+
() => resolveDimensions([DEFAULT_DIMENSIONS[0], dup], DEFAULT_DIMENSIONS),
|
|
470
|
+
/review-pr: args\.dimensions\[1\] repeats key "correctness" with different fields$/,
|
|
471
|
+
);
|
|
472
|
+
});
|
|
473
|
+
|
|
474
|
+
test("an override resolving to nothing stops the run", () => {
|
|
475
|
+
assert.throws(() => resolveDimensions([], DEFAULT_DIMENSIONS), /resolved to no dimensions/);
|
|
476
|
+
});
|
|
477
|
+
|
|
478
|
+
test("a non-array override stops the run rather than crashing on .map", () => {
|
|
479
|
+
assert.throws(() => resolveDimensions({ key: "correctness" }, DEFAULT_DIMENSIONS), /must be an array/);
|
|
480
|
+
});
|
|
481
|
+
|
|
482
|
+
// #278. The header claimed "the three fields the fan-out actually dereferences"
|
|
483
|
+
// and enumerated three; four were dereferenced, back when `model` was a
|
|
484
|
+
// fourth optional field. #1349 retired `model` entirely, so the
|
|
485
|
+
// dereferenced set is three again — `key`, `prompt`, `agentType` — and this
|
|
486
|
+
// pin now confirms the header says exactly that rather than re-asserting the
|
|
487
|
+
// four-field count #278 corrected. Derived from the source rather than
|
|
488
|
+
// hand-listed, because hand-derived count claims about this function have
|
|
489
|
+
// been wrong before (#118) — a fifth dereferenced field reds this and forces
|
|
490
|
+
// the prose to be updated with it.
|
|
491
|
+
test("the resolveDimensions header states the dereferenced-field situation correctly", () => {
|
|
492
|
+
const m = SOURCE.match(/((?:^\/\/.*\n)+)^export function resolveDimensions\(/m);
|
|
493
|
+
assert.ok(m, "resolveDimensions no longer carries a header comment block — update this test");
|
|
494
|
+
const header = m[1];
|
|
495
|
+
// Comments stripped BEFORE deriving the dereferenced set: the header's own
|
|
496
|
+
// prose lists `d.key`, `d.prompt`, `d.model`, `d.agentType` in backticks,
|
|
497
|
+
// which matches the same `\bd\.` pattern used to derive the set. Deriving
|
|
498
|
+
// from raw SOURCE would let the header vote for its own claim — it could
|
|
499
|
+
// catch the header UNDER-counting a field the code dereferences, but never
|
|
500
|
+
// OVER-counting one the code no longer does, since the header's mention
|
|
501
|
+
// would keep padding the derived set to match itself.
|
|
502
|
+
const dereferenced = [...new Set([...stripComments(SOURCE).matchAll(/\bd\.([a-zA-Z]+)/g)].map((x) => x[1]))].sort();
|
|
503
|
+
assert.deepEqual(dereferenced, ["agentType", "key", "prompt"], "the dereferenced field set changed");
|
|
504
|
+
for (const f of dereferenced) {
|
|
505
|
+
assert.match(header, new RegExp("`d\\." + f + "`"), `the header no longer enumerates d.${f}`);
|
|
506
|
+
}
|
|
507
|
+
assert.doesNotMatch(header, /three fields the fan-out actually dereferences/, "the header undercounts again");
|
|
508
|
+
assert.doesNotMatch(header, /absent means inherit/i, "the refuted reword is back — see the models-sent log");
|
|
509
|
+
});
|
|
510
|
+
|
|
511
|
+
// Everything above tests a LIFTED COPY. Nothing above proves review-core.mjs
|
|
512
|
+
// wires resolveDimensions AND selectDimensions into the same call site:
|
|
513
|
+
// replacing it with `explicitDimensions || DEFAULT_DIMENSIONS` left this file
|
|
514
|
+
// green before (#118) by disconnecting the size tier, and a version that
|
|
515
|
+
// dropped resolveDimensions here would do the same to the override
|
|
516
|
+
// normalization, silently. This is the only assertion that fails on either.
|
|
517
|
+
//
|
|
518
|
+
// The two halves now sit ~570 lines apart (#275 moved normalization up to the
|
|
519
|
+
// required-args guard), so this pin is TWO pins — one per line. Weakening either
|
|
520
|
+
// to a looser match, or dropping one because the other still passes, restores
|
|
521
|
+
// exactly the hole #118 opened.
|
|
522
|
+
test("runReview actually calls resolveDimensions, then selectDimensions, to pick the fan-out", () => {
|
|
523
|
+
assert.match(
|
|
524
|
+
SOURCE,
|
|
525
|
+
/^\s*const explicitDimensions = resolveDimensions\(A\.dimensions, DEFAULT_DIMENSIONS\);$/m,
|
|
526
|
+
"the args.dimensions normalization changed — the override may be passed through unresolved",
|
|
527
|
+
);
|
|
528
|
+
assert.match(
|
|
529
|
+
SOURCE,
|
|
530
|
+
/^\s*const dimensions = explicitDimensions \|\| selectDimensions\(DEFAULT_DIMENSIONS, stats\);$/m,
|
|
531
|
+
"the dimensions call site changed — override normalization and/or size-tier selection may be disconnected",
|
|
532
|
+
);
|
|
533
|
+
});
|
|
534
|
+
|
|
535
|
+
// #275, AC: "refused with no agent dispatched and no snapshot directory
|
|
536
|
+
// created". Both are pinned by ONE ordering fact, because the snapshot
|
|
537
|
+
// directory is `mkdir -p`'d inside the snapshot agent's own bash — the dispatch
|
|
538
|
+
// is the only thing that creates it. Asserting the message alone would stay
|
|
539
|
+
// green with the throw back below the snapshot, which is the whole defect.
|
|
540
|
+
//
|
|
541
|
+
// The workflow cannot be imported to test this by execution: it runs a
|
|
542
|
+
// top-level `await pipeline(...)` and compiles as a function body (#538), so
|
|
543
|
+
// source position is the observable. Same shape as
|
|
544
|
+
// review-pr-snapshot-path.test.mjs's guard-ordering pin.
|
|
545
|
+
test("an unresolvable override is refused before the snapshot agent is dispatched", () => {
|
|
546
|
+
const resolveAt = SOURCE.indexOf("const explicitDimensions = resolveDimensions(A.dimensions, DEFAULT_DIMENSIONS);");
|
|
547
|
+
const guardAt = SOURCE.indexOf('if (!pr || !worktree) throw new Error("review-pr: args.pr and args.worktree are required");');
|
|
548
|
+
const snapshotAt = SOURCE.indexOf("const snap = await agent(");
|
|
549
|
+
const mkdirAt = SOURCE.indexOf('mkdir -p "${runRootParent}"');
|
|
550
|
+
assert.ok(resolveAt !== -1 && guardAt !== -1 && snapshotAt !== -1, "one of the three sites moved — update this test");
|
|
551
|
+
assert.ok(guardAt < resolveAt, "the required-args guard no longer runs first — args.pr errors are now masked");
|
|
552
|
+
assert.ok(
|
|
553
|
+
resolveAt < snapshotAt,
|
|
554
|
+
"args.dimensions is validated after the snapshot agent — a typo'd key pays for a snapshot agent first (#275)",
|
|
555
|
+
);
|
|
556
|
+
// The directory is created by that dispatch, not before it, so "no snapshot
|
|
557
|
+
// directory" follows from "no dispatch" only while the mkdir stays inside.
|
|
558
|
+
assert.ok(
|
|
559
|
+
mkdirAt > snapshotAt,
|
|
560
|
+
"the run-root mkdir moved out of the snapshot agent's bash — 'no directory created' no longer follows from 'no agent dispatched'",
|
|
561
|
+
);
|
|
562
|
+
});
|
|
563
|
+
|
|
564
|
+
// Slice, then match. A whole-file `assert.match` for this proved nothing: it
|
|
565
|
+
// survived commenting the line out, planting the phrase in a comment elsewhere,
|
|
566
|
+
// AND moving the `model` option off the review dispatch onto the refuter
|
|
567
|
+
// dispatch. Comment lines are stripped so a commented-out option cannot satisfy
|
|
568
|
+
// it, and the slice is the review dispatch's options object alone.
|
|
569
|
+
function reviewDispatchOptions() {
|
|
570
|
+
const m = SOURCE.match(/\{ label: `review:\$\{d\.key\}`[^\n]*\},/);
|
|
571
|
+
assert.ok(m, "review-core.mjs no longer passes an options object labelled review:${d.key} — update this test");
|
|
572
|
+
return m[0]
|
|
573
|
+
.split("\n")
|
|
574
|
+
.filter((l) => !/^\s*\/\//.test(l))
|
|
575
|
+
.join("\n");
|
|
576
|
+
}
|
|
577
|
+
|
|
578
|
+
// #1349 (per #1303's gap 3) retired the whole per-call override this test
|
|
579
|
+
// used to pin: `args.specialistModel`/per-dimension `model` are both gone,
|
|
580
|
+
// and no `agent()` call anywhere in review-core.mjs may carry a `model` option
|
|
581
|
+
// at all — review-tier-audit.test.mjs is the general-purpose guard for that;
|
|
582
|
+
// this pin is the SPECIFIC regression check that the review dispatch's own
|
|
583
|
+
// options object never grows one back.
|
|
584
|
+
test("the review dispatch never carries a model option — tier lives only in agentType's definition", () => {
|
|
585
|
+
assert.doesNotMatch(
|
|
586
|
+
reviewDispatchOptions(),
|
|
587
|
+
/\bmodel\s*:/,
|
|
588
|
+
"the review dispatch carries a `model` option again — #1349 ruled that out entirely",
|
|
589
|
+
);
|
|
590
|
+
assert.doesNotMatch(
|
|
591
|
+
stripComments(SOURCE),
|
|
592
|
+
/specialistModel/,
|
|
593
|
+
"args.specialistModel is back in CODE — #1349 retired this per-call override entirely (a comment explaining the removal is fine and expected)",
|
|
594
|
+
);
|
|
595
|
+
});
|
|
596
|
+
|
|
597
|
+
// The log is the only runtime evidence of WHICH fleet-owned definition each
|
|
598
|
+
// dimension dispatched to — the tier itself is no longer runtime-visible
|
|
599
|
+
// here at all (#1349): it lives in that definition's own frontmatter, read
|
|
600
|
+
// off the definition file or the dispatch transcript, never off this log.
|
|
601
|
+
test("the dispatch log reports the agentType sent for each dimension, not a model tier", () => {
|
|
602
|
+
const m = SOURCE.match(/`agents dispatched \$\{dimensions[\s\S]*?\n/);
|
|
603
|
+
assert.ok(m, "the `agents dispatched` log line is gone — nothing then reports which definition each dimension used");
|
|
604
|
+
assert.match(m[0], /d\.agentType/, "the log no longer names the dispatched agentType");
|
|
605
|
+
assert.doesNotMatch(m[0], /"frontmatter"|"inherit"/, "the log still reasons about a model tier that no agent() call carries anymore");
|
|
606
|
+
});
|
|
607
|
+
|
|
608
|
+
// `gh pr view --json files` pages at 100 and exits 0, so a truncated list is a
|
|
609
|
+
// short MEASUREMENT, not a small PR: `loc` under-counts and `docsOnly` can be
|
|
610
|
+
// true only because the src files fell off the end. Either one trims dimensions
|
|
611
|
+
// off production code — the docs profile drops four. Widen, the same safe
|
|
612
|
+
// direction as an unparseable blob.
|
|
613
|
+
test("a truncated file list widens to the full set, whatever it profiles as", () => {
|
|
614
|
+
const docsish = computeStats(
|
|
615
|
+
[
|
|
616
|
+
{ path: "docs/a.md", additions: 3, deletions: 0 },
|
|
617
|
+
{ path: "docs/b.md", additions: 3, deletions: 0 },
|
|
618
|
+
],
|
|
619
|
+
124,
|
|
620
|
+
);
|
|
621
|
+
assert.equal(docsish.docsOnly, true, "the short list really does look docs-only — that is the trap");
|
|
622
|
+
assert.equal(docsish.truncated, 124);
|
|
623
|
+
assert.equal(
|
|
624
|
+
selectDimensions(DEFAULT_DIMENSIONS, docsish).length,
|
|
625
|
+
DEFAULT_DIMENSIONS.length,
|
|
626
|
+
"a capped list was sized as a docs PR — four dimensions dropped off code gh never listed",
|
|
627
|
+
);
|
|
628
|
+
// And the widen must not fire when the list is complete, or every review runs
|
|
629
|
+
// the full set forever and the size tier is dead.
|
|
630
|
+
const complete = computeStats(
|
|
631
|
+
[
|
|
632
|
+
{ path: "docs/a.md", additions: 3, deletions: 0 },
|
|
633
|
+
{ path: "docs/b.md", additions: 3, deletions: 0 },
|
|
634
|
+
],
|
|
635
|
+
2,
|
|
636
|
+
);
|
|
637
|
+
assert.equal(complete.truncated, undefined);
|
|
638
|
+
assert.deepEqual(
|
|
639
|
+
selectDimensions(DEFAULT_DIMENSIONS, complete).map((d) => d.key),
|
|
640
|
+
["correctness", "comments"],
|
|
641
|
+
"the docs trim no longer fires on a complete docs-only list",
|
|
642
|
+
);
|
|
643
|
+
});
|
|
644
|
+
|
|
645
|
+
// The invariant that made two `x.length ? x : all` fallbacks dead code, deleted
|
|
646
|
+
// in #669. This is the pin that has to hold for the deletion to stay safe: with
|
|
647
|
+
// `all = DEFAULT_DIMENSIONS` — which is what the ONE call site passes, an
|
|
648
|
+
// `args.dimensions` override having short-circuited via `resolveDimensions` —
|
|
649
|
+
// no combination of stats can empty the set. Sweeping the stats SHAPE rather
|
|
650
|
+
// than a file list on purpose: the fallbacks guarded against a corrupt or
|
|
651
|
+
// unforeseen blob, so the values under test include the non-booleans and
|
|
652
|
+
// unknown profiles a real payload could carry, not only what computeStats emits.
|
|
653
|
+
test("no stats shape empties the dimension set — `correctness` survives every filter", () => {
|
|
654
|
+
// `""` stands in for the whole falsy family (`undefined`, `null`, `""`) plus
|
|
655
|
+
// `"empty"`: all four hit `!stats.profile || stats.profile === "empty"` and
|
|
656
|
+
// return `all` before any other field is read, so they are one outcome, not
|
|
657
|
+
// four — verified by re-reading that guard, not assumed from this trim.
|
|
658
|
+
const profiles = ["", "docs", "tests-only", "single-file", "small", "production", "bogus"];
|
|
659
|
+
const values = [true, false, undefined, null, "yes", 0, 1];
|
|
660
|
+
for (const profile of profiles)
|
|
661
|
+
for (const docsOnly of values)
|
|
662
|
+
for (const hasSrc of values)
|
|
663
|
+
for (const hasTests of values)
|
|
664
|
+
for (const truncated of values) {
|
|
665
|
+
const stats = { profile, docsOnly, hasSrc, hasTests, truncated };
|
|
666
|
+
assert.ok(
|
|
667
|
+
selectDimensions(DEFAULT_DIMENSIONS, stats).some((d) => d.key === "correctness"),
|
|
668
|
+
`correctness was filtered out by ${JSON.stringify(stats)}`,
|
|
669
|
+
);
|
|
670
|
+
}
|
|
671
|
+
// The non-object arms of the same guard.
|
|
672
|
+
for (const stats of [null, undefined, 0, "", false])
|
|
673
|
+
assert.equal(selectDimensions(DEFAULT_DIMENSIONS, stats).length, DEFAULT_DIMENSIONS.length);
|
|
674
|
+
});
|