The windows-test-parity ratchet greps test source for fs.rmSync-without-
maxRetries (and six other Windows-portability anti-patterns), failing when an
integer offender COUNT exceeds a frozen baseline (rmSync: 95). A count ratchet
is a Goodhart metric: fixing one offender and adding another keeps the count
constant, so a new defect slips through green. Replace it — and every other
count ratchet in the repo — with a layered, masking-proof design.
Behavioral seam test
- tests/helpers-cleanup.test.cjs proves helpers.cleanup() carries the Windows
EBUSY retry budget. cleanup() delegates retries to Node's fs.rmSync via
maxRetries (it owns no loop), so the test asserts the option contract
(recursive/force/maxRetries>0/retryDelay>0) + real-FS removal + the cwd-guard,
rather than a loop that does not exist. The EBUSY risk is now tested ONCE at
the helper, not approximated textually at every call site.
Write-time ESLint rule (AST-accurate, replaces the grep)
- eslint-rules/no-raw-rmsync-in-tests.cjs (error in tests/**/*.test.cjs) bans
raw fs.rmSync, steering to cleanup(). Catches member, computed (fs['rmSync']),
destructured and aliased forms; escape hatch is inline
`// eslint-disable-next-line local/no-raw-rmsync-in-tests -- <reason>` only.
- Migrated 336 raw fs.rmSync teardown calls across ~116 test files to cleanup().
~18 genuinely load-bearing sites (mid-test SUT/fault-injection removals,
error-swallowing or name-colliding local teardown helpers) keep the raw call
with an inline eslint-disable + reason.
Shared anti-ratchet primitive
- scripts/lib/allowlist-ratchet.cjs:
- assertWithinAllowlist: fails on NOVEL ids (new offender introduced) AND on
STALE ids (a known offender was fixed but not pruned) — identity, not count,
and a ratchet DOWN toward zero.
- assertTightCeiling: a size/length budget whose ceiling must stay within a
grace band of the high-water mark, so budgets may only tighten, never creep.
Ratchets converted onto the primitive
- windows-test-parity-guard.test.cjs: rmSync rule deleted (now ESLint-enforced);
the remaining six patterns moved from integer baselines to named-set
allowlists with ratchet-down.
- scripts/lint-test-file-count.{cjs,allowlist.json}: per-module integer counts →
named filename sets (closes the swap-a-file-keep-the-count blind spot); a
module dropping under cap now FAILS to force pruning its allowlist entry.
- enh-2790 skill-count `<= 63` → named skill allowlist (ratchets toward ~58).
Size budgets hardened (tighten-only)
- agent-size / workflow-size / feat-3039 help-tiered: ceilings lowered to the
current high-water mark and an assertTightCeiling anti-creep check added per
tier. Fixed external-contract limits (description ≤100 chars, agent ≤100 KB)
are intentionally left as-is — they are not grandfathered creeping budgets.
No user-facing behavior change (tests + tooling only); no USER_FACING_PREFIXES
touched, so no changeset fragment is required.
Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>
213 lines
9.4 KiB
JavaScript
213 lines
9.4 KiB
JavaScript
'use strict';
|
|
/**
|
|
* Regression test for bug #444: gsd_run resolver must probe
|
|
* <repo-root>/.claude/get-shit-done/bin/gsd-tools.cjs (the project-local
|
|
* `--claude --local` install location) BEFORE checking $HOME/.claude and PATH.
|
|
*
|
|
* Asserts:
|
|
* (A) The canonical snippet file contains the repo-local .claude/ check.
|
|
* (B) Behavioral: when RUNTIME_DIR/get-shit-done/bin/ misses, but a stub
|
|
* exists ONLY at <repo-root>/.claude/get-shit-done/bin/gsd-tools.cjs,
|
|
* gsd_run resolves to that stub (no PATH stub, no HOME stub).
|
|
* (C) Precedence: repo-local .claude/ wins over $HOME/.claude/ when both exist.
|
|
*/
|
|
|
|
// allow-test-rule: structural/behavioral regression for the repo-local .claude/ install
|
|
// arm in the gsd_run launcher snippet -- asserts literal substring presence and exercises
|
|
// the bash resolution path via execFileSync; there is no typed IR for "snippet contains arm X".
|
|
|
|
const { describe, test } = require('node:test');
|
|
const assert = require('node:assert/strict');
|
|
const fs = require('node:fs');
|
|
const path = require('node:path');
|
|
const os = require('node:os');
|
|
const { execFileSync } = require('node:child_process');
|
|
const { cleanup } = require('./helpers.cjs');
|
|
|
|
const WORKFLOWS_DIR = path.join(__dirname, '..', 'get-shit-done', 'workflows');
|
|
const SNIPPET_FILE = path.join(WORKFLOWS_DIR, '_runtime-launcher.snippet.sh');
|
|
|
|
// The probe string that must appear in the snippet for the new repo-local check.
|
|
// The snippet uses _GSD_RUNTIME_ROOT as the intermediate variable.
|
|
const LOCAL_CLAUDE_PROBE = '_GSD_RUNTIME_ROOT}/.claude/get-shit-done/bin/';
|
|
|
|
/**
|
|
* Build a PATH that strips gsd-tools but keeps node and system binaries.
|
|
* Accepts additional bin dirs to prepend.
|
|
*
|
|
* We cannot simply remove the whole directory that contains gsd-tools because
|
|
* that directory may also contain node (e.g. /opt/homebrew/bin on macOS).
|
|
* Instead, we keep the system PATH as-is and rely on the test's RUNTIME_DIR
|
|
* having no get-shit-done/bin/ sub-path, so the resolver's first two checks
|
|
* (RUNTIME_DIR/get-shit-done/bin/ and RUNTIME_DIR/.claude/get-shit-done/bin/)
|
|
* are the only ones exercised before we hit our stub.
|
|
*
|
|
* The extra extraBefore dirs (e.g. noToolsBin) sit first but have no gsd-tools
|
|
* binary, so command -v gsd-tools still falls back to PATH lookup. However,
|
|
* the snippet's elif arm that uses `command -v gsd-tools` will find the real
|
|
* installed one unless we mask it. To mask it without losing node, we create
|
|
* a noToolsBin dir that shadows gsd-tools with a sentinel that must NOT be
|
|
* called — and we only call makeIsolatedPath for tests where the .claude stub
|
|
* must win before PATH is consulted (i.e. the elif PATH arm is never reached).
|
|
*
|
|
* For B: stub is at RUNTIME_DIR/.claude/... so resolver picks it at elif-1 (before command -v).
|
|
* For C: same — local .claude/ is checked before command -v and before $HOME/.claude.
|
|
*/
|
|
function makeIsolatedPath(extraBefore = []) {
|
|
// Keep full system PATH so node remains accessible.
|
|
// Tests B and C exercise only the RUNTIME_DIR/.claude arm which fires
|
|
// before command -v gsd-tools — so the real gsd-tools on PATH is never reached.
|
|
const systemPaths = (process.env.PATH || '/usr/bin:/bin').split(path.delimiter);
|
|
return [...extraBefore, ...systemPaths].join(path.delimiter);
|
|
}
|
|
|
|
describe('bug-444: resolver finds repo-local .claude install', () => {
|
|
// --- (A) Snippet contains the repo-local .claude arm ----------------------
|
|
test('(A) snippet file contains the repo-local .claude/ check arm before $HOME/.claude/', () => {
|
|
const content = fs.readFileSync(SNIPPET_FILE, 'utf8');
|
|
|
|
// Must contain the repo-local .claude/ check (via _GSD_RUNTIME_ROOT variable)
|
|
const localClaudeIdx = content.indexOf(LOCAL_CLAUDE_PROBE);
|
|
assert.ok(
|
|
localClaudeIdx !== -1,
|
|
`_runtime-launcher.snippet.sh must contain the repo-local .claude check ` +
|
|
`('${LOCAL_CLAUDE_PROBE}'). ` +
|
|
`Found snippet content:\n${content.trim()}`,
|
|
);
|
|
|
|
// Must still contain the $HOME/.claude fallback arm
|
|
const homeClaudeIdx = content.indexOf('$HOME/.claude/get-shit-done/bin/');
|
|
assert.ok(
|
|
homeClaudeIdx !== -1,
|
|
`Snippet must still contain the $HOME/.claude fallback arm.`,
|
|
);
|
|
|
|
// Repo-local check must appear BEFORE $HOME/.claude check (local overrides global)
|
|
assert.ok(
|
|
localClaudeIdx < homeClaudeIdx,
|
|
`Repo-local .claude/ check (idx ${localClaudeIdx}) must appear BEFORE ` +
|
|
`$HOME/.claude/ check (idx ${homeClaudeIdx}) in the snippet (local overrides global).`,
|
|
);
|
|
});
|
|
|
|
// --- (B) Behavioral: repo-local .claude stub resolved when only location ---
|
|
test('(B) gsd_run resolves repo-local .claude/get-shit-done/bin/ stub when no other locations present', () => {
|
|
// Create a fake repo root with a stub ONLY at .claude/get-shit-done/bin/gsd-tools.cjs
|
|
// NO stub at get-shit-done/bin/, NOT on PATH, NOT in $HOME/.claude
|
|
const fakeRoot = fs.mkdtempSync(path.join(os.tmpdir(), 'gsd-444-root-'));
|
|
const fakeHome = fs.mkdtempSync(path.join(os.tmpdir(), 'gsd-444-home-'));
|
|
const noToolsBin = path.join(fakeRoot, 'nobin');
|
|
fs.mkdirSync(noToolsBin, { recursive: true });
|
|
|
|
try {
|
|
// Create the stub at the repo-local .claude path ONLY
|
|
const localClaudeBinDir = path.join(fakeRoot, '.claude', 'get-shit-done', 'bin');
|
|
fs.mkdirSync(localClaudeBinDir, { recursive: true });
|
|
const stubPath = path.join(localClaudeBinDir, 'gsd-tools.cjs');
|
|
fs.writeFileSync(
|
|
stubPath,
|
|
'#!/usr/bin/env node\nconsole.log("LOCAL_CLAUDE_STUB:" + process.argv.slice(2).join(","));\n',
|
|
);
|
|
fs.chmodSync(stubPath, 0o755);
|
|
|
|
const snippet = fs.readFileSync(SNIPPET_FILE, 'utf8');
|
|
// Set RUNTIME_DIR to fakeRoot so the resolver uses it as the repo root.
|
|
const scriptContent =
|
|
`unset GSD_TOOLS\n` +
|
|
`export RUNTIME_DIR=${JSON.stringify(fakeRoot)}\n` +
|
|
`export HOME=${JSON.stringify(fakeHome)}\n` +
|
|
snippet +
|
|
`\nprintf "GSD_TOOLS=%s\\n" "$GSD_TOOLS"\n` +
|
|
`gsd_run ping test\n`;
|
|
|
|
const scriptPath = path.join(fakeRoot, 'test-local-claude.sh');
|
|
fs.writeFileSync(scriptPath, scriptContent);
|
|
|
|
// Keep node in PATH (needed to run the .cjs stub); remove gsd-tools
|
|
const isolatedPath = makeIsolatedPath([noToolsBin]);
|
|
|
|
const stdout = execFileSync('bash', [scriptPath], {
|
|
encoding: 'utf8',
|
|
env: { ...process.env, PATH: isolatedPath, HOME: fakeHome },
|
|
});
|
|
|
|
// Must have resolved to the local .claude stub
|
|
const normStdout = stdout.replace(/\\/g, '/');
|
|
assert.ok(
|
|
normStdout.includes('.claude/get-shit-done/bin/gsd-tools.cjs'),
|
|
`Expected GSD_TOOLS to resolve to .claude/get-shit-done/bin/gsd-tools.cjs, got:\n${stdout.trim()}`,
|
|
);
|
|
// The stub must have been invoked with the correct arguments
|
|
assert.ok(
|
|
stdout.includes('LOCAL_CLAUDE_STUB:ping,test'),
|
|
`Expected stub output "LOCAL_CLAUDE_STUB:ping,test" but got:\n${stdout.trim()}`,
|
|
);
|
|
} finally {
|
|
cleanup(fakeRoot);
|
|
cleanup(fakeHome);
|
|
}
|
|
});
|
|
|
|
// --- (C) Precedence: repo-local .claude/ wins over $HOME/.claude/ ----------
|
|
test('(C) repo-local .claude/ install wins over $HOME/.claude/ when both exist', () => {
|
|
const fakeRoot = fs.mkdtempSync(path.join(os.tmpdir(), 'gsd-444-prec-root-'));
|
|
const fakeHome = fs.mkdtempSync(path.join(os.tmpdir(), 'gsd-444-prec-home-'));
|
|
const noToolsBin = path.join(fakeRoot, 'nobin');
|
|
fs.mkdirSync(noToolsBin, { recursive: true });
|
|
|
|
try {
|
|
// Stub at repo-local .claude/ path (should be picked)
|
|
const localClaudeBinDir = path.join(fakeRoot, '.claude', 'get-shit-done', 'bin');
|
|
fs.mkdirSync(localClaudeBinDir, { recursive: true });
|
|
const localStubPath = path.join(localClaudeBinDir, 'gsd-tools.cjs');
|
|
fs.writeFileSync(
|
|
localStubPath,
|
|
'#!/usr/bin/env node\nconsole.log("LOCAL_WINS:" + process.argv.slice(2).join(","));\n',
|
|
);
|
|
fs.chmodSync(localStubPath, 0o755);
|
|
|
|
// Stub at $HOME/.claude/ path (must NOT be picked)
|
|
const homeClaudeBinDir = path.join(fakeHome, '.claude', 'get-shit-done', 'bin');
|
|
fs.mkdirSync(homeClaudeBinDir, { recursive: true });
|
|
const homeStubPath = path.join(homeClaudeBinDir, 'gsd-tools.cjs');
|
|
fs.writeFileSync(
|
|
homeStubPath,
|
|
'#!/usr/bin/env node\nconsole.log("HOME_WINS:" + process.argv.slice(2).join(","));\n',
|
|
);
|
|
fs.chmodSync(homeStubPath, 0o755);
|
|
|
|
const snippet = fs.readFileSync(SNIPPET_FILE, 'utf8');
|
|
const scriptContent =
|
|
`unset GSD_TOOLS\n` +
|
|
`export RUNTIME_DIR=${JSON.stringify(fakeRoot)}\n` +
|
|
`export HOME=${JSON.stringify(fakeHome)}\n` +
|
|
snippet +
|
|
`\nprintf "GSD_TOOLS=%s\\n" "$GSD_TOOLS"\n` +
|
|
`gsd_run check\n`;
|
|
|
|
const scriptPath = path.join(fakeRoot, 'test-precedence.sh');
|
|
fs.writeFileSync(scriptPath, scriptContent);
|
|
|
|
const isolatedPath = makeIsolatedPath([noToolsBin]);
|
|
|
|
const stdout = execFileSync('bash', [scriptPath], {
|
|
encoding: 'utf8',
|
|
env: { ...process.env, PATH: isolatedPath, HOME: fakeHome },
|
|
});
|
|
|
|
assert.ok(
|
|
stdout.includes('LOCAL_WINS:check'),
|
|
`Expected repo-local .claude stub to be invoked ("LOCAL_WINS:check") ` +
|
|
`but got:\n${stdout.trim()}`,
|
|
);
|
|
assert.ok(
|
|
!stdout.includes('HOME_WINS'),
|
|
`Expected $HOME/.claude stub NOT to be invoked, but got:\n${stdout.trim()}`,
|
|
);
|
|
} finally {
|
|
cleanup(fakeRoot);
|
|
cleanup(fakeHome);
|
|
}
|
|
});
|
|
});
|