Files
PaperClipAI/packages/paperclip-runner/scripts/build-inputs.test.mjs
T
Nicky LeachandClaude Opus 5.5 ff8ec41e6a feat(runner): upgrade ACPX to 0.22.0 and move patched behavior into Paperclip (#20770)
## Thinking Path

> - Paperclip is the open source app people use to manage AI agents for
work
> - Paperclip Runner and the shared `adapter-utils` engine run ACP
agents through the embedded ACPX library
> - Paperclip pins ACPX 0.13.1 and keeps a large patch
(`patches/acpx@0.13.1.patch`) on it
> - ACPX 0.22.0 adds host hooks (`onAcpMessage`, `assertActive`,
`processLifecycle`, prompt-response `_meta`, `spawnAgent`-friendly
launch) that make much of that patch unnecessary
> - A large patch is expensive to port and hides behavior that belongs
to Paperclip
> - This pull request upgrades to ACPX 0.22.0 and moves most patched
behavior into Runner and `adapter-utils`
> - The benefit is a patch that is 484 lines instead of 1,334, with the
moved behavior covered by Paperclip tests

## Linked Issues or Issue Description

Refs #20675, #14430, #15075, #17942, #9980

**Proposed enhancement**
Upgrade the embedded ACPX library from 0.13.1 to 0.22.0. Keep only the
patch behavior that stock 0.22 cannot express.

**Motivation**
The 0.13.1 patch held behavior that Paperclip owns, for example usage
receipts, Copilot policy, model checks and the agent environment. Each
ACPX upgrade had to port all of it. ACPX 0.22 gives hooks that let
Paperclip do this work itself.

**Current behavior**
Runner and `adapter-utils` depend on 1,334 lines of patched ACPX 0.13.1
code.

**Expected behavior**
Runner and `adapter-utils` use ACPX 0.22.0. The patch keeps only:
extension methods, Cursor subagent routing, event `_meta` and the
message boundary chunk, the protocol guard hook, `spawnAgent`, and
`requestExtension`.

## What Changed

- ACPX 0.13.1 → 0.22.0 in Runner, `adapter-utils`, the server, the
Daytona runner images and the qualified profiles (`acpxVersion`).
`@agentclientprotocol/sdk` moves to 1.9.0.
- New `patches/acpx@0.22.0.patch` (484 lines). The 0.12.0 and 0.13.1
patches are removed.
- Delivery receipts: Runner confirms a permission reply when
`onAcpMessage` sees the exact outbound reply. Question and extension
replies count as delivered when Runner hands them back to ACPX. New
`interaction-delivery.ts`.
- Copilot policy: the existing policy function runs from `onAcpMessage`.
A violation stops the agent, cancels later permission requests and fails
the next prompt through `assertActive`.
- Model selection: stock 0.22 validation, plus a sidecar gate
(`createAcpxEffectiveModelGate`) that refuses a prompt if the agent
reports another model. Grok binds its model when the session opens.
Antigravity adds its concrete model to the model list ACPX reads.
- Typed session failures: Runner and `adapter-utils` read them from the
turn result `_meta` and fail the turn. They advertise the capability
themselves.
- Pi and Cursor usage receipts: a collector (`prompt-receipts.ts`) binds
them to the turn's own `session/prompt` reply.
- Goal capability: Runner reads it from the `initialize` reply. It keeps
it under `paperclip.goalCapability` in the saved agent capabilities
`_meta` and restores it when a saved session loads.
- Agent environment: Runner and `adapter-utils` set the agent
environment in `spawnAgent`. ACP terminals keep their current
environment. `adapter-utils` blocks inheritance of `XAI_API_KEY` and
refuses to start if the server has an `ACPX_AUTH_*` variable.
- `adapter-utils` spawn hooks (`spawnCwd`, stderr, `onAgentSpawn`): a
new `stockAcpRuntimeOptions` maps them to stock `spawnAgent` and
`processLifecycle`.
- Windows: ACPX 0.22 rejects raw command strings on Windows.
`adapter-utils` now passes an argv array there.
- Grok rules: a Runner guard (`grok-session-rules.ts`) writes
`session/new` `_meta.rules`.
- Turn-dispatch journal (#20675): ACPX 0.22 ignores errors from
`onAcpMessage`. The adapter keeps a failed `dispatched` write and throws
it from the turn's `assertActive`. ACPX calls `assertActive` before the
prompt write, so the write still stops. The real-ACPX test now checks
this route. `docs/durable-recovery.md` is updated.
- Plan events: the sidecar reads 0.22 `status` events tagged `plan`.
- CI: the backup-checkpoint lane now installs through the shared
`setup-ci-dependencies` action, like PR CI, so it resolves a stale
`pnpm-lock.yaml` in place. It runs on any root `package.json` change,
and pull requests never commit the lockfile, so its own frozen install
failed on dependency changes.
- Tests that checked removed patch internals are removed or rewritten:
the response-delivery, Cursor-usage and Pi-receipt package contracts,
and the goal-persistence packaging test. New unit tests cover the moved
behavior.

## Verification

Local runs on macOS (darwin-arm64), Node 26, on the rebased head:

- `pnpm --filter @paperclipai/paperclip-runner exec vitest run
src/drivers/acpx src/cli`: 2,000 passed.
- `tsc -p packages/paperclip-runner/tsconfig.json` and
`tsconfig.test.json`: no errors, except the existing
`@paperclipai/paperclip-eval-kernel` build gap in `workflow-harness.ts`.
- Runner package contracts (`node --test
test/acpx-*-package-contract.test.mjs
test/acpx-sidecar-contract.test.mjs`), `scripts/build-inputs.test.mjs`,
`scripts/contracts/contract-fingerprints.test.mjs`,
`scripts/acpx-patch-packaging.test.mjs`: all pass. `node
scripts/generate-acpx-sidecar-contract.mjs --check` passes.
- `cargo test -p paperclip-runner-core --test acpx_park_preparation
--test acpx_park_readiness --test acpx_warm_attachment --test
native_provider_backend`: 86 passed. `cargo fmt --check` passes.
- `adapter-utils`: `tsc` clean. `vitest run src/acpx-engine
src/mcp-isolation.integration.test.ts`: 457 passed (one Gemini flag test
failed once under load and passes when run alone).
- Adapters: `claude-local` and `codex-local` `acp.quota.test.ts` (23 and
39 passed), `grok-local` (149 passed).
- Server: `scripts/prepare-runner-vendor.test.mjs`,
`native-session-executor.test.ts`, `quarantined-task-retry.test.ts`
(1,416 passed).
- `cargo test -p paperclip-runner-core --lib -- previous_acpx_release
acpx_upgrade unrestorable_saved_sessions`: 3 passed. `node --test
test/acpx-session-carry-forward-package-contract.test.mjs`: passes.
- `patch -p1 --forward` on a clean `acpx-0.22.0.tgz` gives output
identical to the port.
- Under full-suite load, some Runner live and recovery tests time out.
Each one passes when run alone. Full regression is left to CI.

Not run: live provider runs (they need provider credentials), the opt-in
Kimi live peer test, the Gemini filesystem smoke test, and Windows.

## Risks

- Previous-release story: sessions saved by the previous release (ACPX
0.13.1) carry forward. ACPX 0.22 loads the session records 0.13.1 wrote.
Runner moves a saved descriptor that names 0.13.1
(`CARRIED_ACPX_RELEASES`) to 0.22.0, and the sidecar keeps 0.13.1 as the
ACPX value in the session identity, so the saved session key and
identity record still match. The existing launch-profile rebind still
decides which sessions resume: a settled session, or the same run's
interrupted turn through lost-turn recovery. A session saved by any
other ACPX release is still refused as `provider_state_incompatible`.
`rules/acpx-profile-identity.mdc` and `docs/durable-recovery.md` now
describe this.
- Proof: `acpx-session-carry-forward-package-contract.test.mjs` resumes
a session record written by the real ACPX 0.13.1 with its Paperclip
patch. 0.22 loads the same provider session (no `session/new`), runs a
turn and keeps the earlier turn. Rust tests cover a settled 0.13.1
session that the next run attaches to, and a turn interrupted by the
upgrade. A recovery-identity test checks that the session key and
identity record match across the versions.
- `pnpm-lock.yaml` is not committed. CI resolves it for each run, and
the lockfile refresh workflow updates it after merge.
- Behavior change on the server: `adapter-utils` refuses to create an
ACPX runtime if the server environment has an `ACPX_AUTH_*` variable.
Before, ACPX ignored such variables. Stock 0.22 would use them to
authenticate agents with server credentials.
- Delivery confirmation is now Runner-owned. A permission reply is
confirmed by the exact outbound message. Question and extension replies
are confirmed when Runner hands them to ACPX, not when the pipe accepts
them.
- Users can select only models that the agent advertises. Stock 0.22
rejects other model IDs (Claude is exempt).
- ACP terminals keep the sidecar or server environment, as before. This
existing exposure is not changed by this pull request.
- The `adapter-utils` Gemini version probe now runs with the server
environment plus the run variables, not the run variables alone.
- The pack contract digest and profile `acpxVersion` change, so Daytona
needs a new runner image.
- Grok launch flags (`--model`, `--reasoning-effort`) were checked on
1.0.13 but session open needs a login, so Grok keeps the existing
`_meta` path.

> For core feature work, check [`ROADMAP.md`](ROADMAP.md) first and
discuss it in `#dev` before opening the PR. Feature PRs that overlap
with planned core work may need to be redirected — check the roadmap
first. See `CONTRIBUTING.md`.

## Model Used

Claude Opus 5.5 (`claude-opus-5-5`), Anthropic, in Claude Code with tool
use (shell, file editing, subagents). Extended reasoning was used.

## Checklist

- [x] I have included a thinking path that traces from project context
to this change
- [x] I have specified the model used (with version and capability
details)
- [x] I have checked ROADMAP.md and confirmed this PR does not duplicate
planned core work
- [x] I have searched GitHub for duplicate or related PRs and linked
them above
- [x] I have either (a) linked existing issues with `Fixes: #` / `Closes
#` / `Refs #` OR (b) described the issue in-PR following the relevant
issue template
- [x] I have not referenced internal/instance-local Paperclip issues or
links (only public GitHub `#NNN` / `github.com/paperclipai/paperclip`
URLs)
- [x] My branch name describes the change (e.g. `docs/...`, `fix/...`)
and contains no internal Paperclip ticket id or instance-derived details
- [x] I have run tests locally and they pass
- [x] I have added or updated tests where applicable
- [x] I have updated relevant documentation to reflect my changes
- [x] I have considered and documented any risks above
- [x] All Paperclip CI gates are green
- [x] Greptile is 5/5 with no open P2s, recommendations, or follow-ups
- [x] I will address all Greptile and reviewer comments before
requesting merge

🤖 Generated with [Claude Code](https://claude.com/claude-code)

---------

Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
2026-11-11 07:28:00 -08:00

543 lines
27 KiB
JavaScript

import assert from "node:assert/strict";
import { execFile } from "node:child_process";
import { existsSync, readdirSync } from "node:fs";
import { mkdtemp, mkdir, readdir, readFile, rm, stat, symlink, truncate, utimes, writeFile } from "node:fs/promises";
import { tmpdir } from "node:os";
import path from "node:path";
import { test } from "node:test";
import { promisify } from "node:util";
import { fileURLToPath } from "node:url";
import {
BUILD_INPUTS_SCHEMA,
PENDING_RECORD_MAX_AGE_MS,
assessRunnerBuild,
beginRunnerBuild,
buildInputsPendingDirectory,
buildInputsPendingPath,
buildInputsStampPath,
collectBuildInputFiles,
collectBuildOutputFiles,
computeBuildInputsDigest,
currentBuildId,
expectedBuildOutputs,
readBuildInputsStamp,
recordRunnerBuild,
} from "./build-inputs.mjs";
const execFileAsync = promisify(execFile);
const recordScript = fileURLToPath(new URL("./record-build-inputs.mjs", import.meta.url));
/** A child of this process names its build after this process, as `--begin` names it after the shell. */
const childBuildId = String(process.pid);
async function waitFor(condition, label, timeoutMs = 10_000) {
const deadline = Date.now() + timeoutMs;
while (!condition()) {
if (Date.now() > deadline) throw new Error(`timed out waiting for ${label}`);
await new Promise((resolve) => setTimeout(resolve, 20));
}
}
/**
* The whole build chain the way pnpm runs it: one shell, `--begin` first,
* `--finish` last, with a stand-in for the build steps that waits for a
* marker file so two chains can be held open at the same time.
*/
function runChainInShell(shell, packageRoot, marker) {
const quote = (value) => `'${value.replace(/'/g, `'\\''`)}'`;
const wait =
`${quote(process.execPath)} -e ${quote(
"const fs=require('node:fs');const [marker]=process.argv.slice(1);const deadline=Date.now()+10000;" +
"(function poll(){if(fs.existsSync(marker))return;if(Date.now()>deadline)process.exit(3);setTimeout(poll,20)})()",
)} ${quote(marker)}`;
const phase = (flag) => `${quote(process.execPath)} ${quote(recordScript)} ${flag} --package-root ${quote(packageRoot)}`;
return execFileAsync(shell, ["-c", `${phase("--begin")} && ${wait} && ${phase("--finish")}`]);
}
async function writeTree(root, files) {
for (const [relative, content] of Object.entries(files)) {
const absolute = path.join(root, ...relative.split("/"));
await mkdir(path.dirname(absolute), { recursive: true });
await writeFile(absolute, content);
}
}
/**
* A throwaway repo with one Runner source file, one Rust source, two root
* patches, built outputs (the four every consumer reads plus one helper that
* is only opened by path), and the usual noise.
*/
async function makeRepo() {
const repoRoot = await mkdtemp(path.join(tmpdir(), "runner-build-inputs-"));
await writeTree(repoRoot, {
"package.json": "{}\n",
"pnpm-lock.yaml": "lockfileVersion: 9\n",
"pnpm-workspace.yaml": "packages:\n - packages/*\n",
"tsconfig.base.json": "{}\n",
"patches/acpx@0.22.0.patch": "--- a\n+++ b\n",
"patches/pi-acp@0.0.33.patch": "--- c\n+++ d\n",
"packages/paperclip-runner/package.json": '{"name":"@paperclipai/paperclip-runner"}\n',
"packages/paperclip-runner/src/index.ts": "export const one = 1;\n",
"packages/paperclip-runner/runner/src/main.rs": "fn main() {}\n",
"packages/paperclip-runner/runner/target/release/paperclip-runnerd": "binary\n",
"packages/paperclip-runner/dist/index.js": "export const one = 1;\n",
"packages/paperclip-runner/dist/live/index.js": "export {};\n",
"packages/paperclip-runner/dist/testing.js": "export {};\n",
"packages/paperclip-runner/dist/bin/paperclip-runnerd": "binary\n",
"packages/paperclip-runner/dist/cli/acpx-runtime-sidecar.cjs": "module.exports = {};\n",
"packages/paperclip-runner/dist-browser/index.js": "browser\n",
"packages/paperclip-runner/node_modules/dep/index.js": "dep\n",
"packages/paperclip-runner/.vite/cache.json": "{}\n",
"packages/paperclip-eval-kernel/src/index.ts": "export const kernel = 1;\n",
"packages/paperclip-eval-kernel/dist/index.js": "export const kernel = 1;\n",
});
const packageRoot = path.join(repoRoot, "packages", "paperclip-runner");
return { repoRoot, packageRoot };
}
const INPUT_FILE_COUNT = 10;
const OUTPUT_FILE_COUNT = 5;
async function withRepo(action) {
const repo = await makeRepo();
try {
await action(repo);
} finally {
await rm(repo.repoRoot, { recursive: true, force: true });
}
}
/** A complete, undisturbed build: begin, (nothing changes), finish. */
function recordCurrentBuild({ repoRoot, packageRoot, platform, arch, now }) {
beginRunnerBuild({ repoRoot, packageRoot, platform, arch, now });
return recordRunnerBuild({ repoRoot, packageRoot, platform, arch, now });
}
test("the input set covers sources, the eval kernel, root patches and root config, and skips outputs and caches", async () => {
await withRepo(async ({ repoRoot }) => {
const files = collectBuildInputFiles(repoRoot);
assert.deepEqual(files, [
"package.json",
"packages/paperclip-eval-kernel/src/index.ts",
"packages/paperclip-runner/package.json",
"packages/paperclip-runner/runner/src/main.rs",
"packages/paperclip-runner/src/index.ts",
"patches/acpx@0.22.0.patch",
"patches/pi-acp@0.0.33.patch",
"pnpm-lock.yaml",
"pnpm-workspace.yaml",
"tsconfig.base.json",
]);
assert.equal(files.length, INPUT_FILE_COUNT);
});
});
test("the output manifest lists every file under dist with its size, and nothing outside it", async () => {
await withRepo(async ({ packageRoot }) => {
const outputs = collectBuildOutputFiles(packageRoot);
assert.deepEqual(outputs, [
{ path: "dist/bin/paperclip-runnerd", size: 7 },
{ path: "dist/cli/acpx-runtime-sidecar.cjs", size: 21 },
{ path: "dist/index.js", size: 22 },
{ path: "dist/live/index.js", size: 11 },
{ path: "dist/testing.js", size: 11 },
]);
assert.equal(outputs.length, OUTPUT_FILE_COUNT);
});
});
test("the digest is stable, ignores outputs and caches, and tracks every kind of input", async () => {
await withRepo(async ({ repoRoot }) => {
const before = computeBuildInputsDigest(repoRoot);
assert.equal(before.schema, BUILD_INPUTS_SCHEMA);
assert.equal(before.fileCount, INPUT_FILE_COUNT);
assert.equal(computeBuildInputsDigest(repoRoot).digest, before.digest, "two reads disagree");
// Outputs, dependencies, caches and the Rust target directory may change freely.
await writeTree(repoRoot, {
"packages/paperclip-runner/dist/index.js": "export const one = 2;\n",
"packages/paperclip-runner/dist/cli/acpx-runtime-sidecar.cjs": "changed\n",
"packages/paperclip-runner/dist-browser/index.js": "changed\n",
"packages/paperclip-runner/node_modules/dep/index.js": "changed\n",
"packages/paperclip-runner/.vite/cache.json": '{"changed":true}\n',
"packages/paperclip-runner/runner/target/release/paperclip-runnerd": "rebuilt\n",
"packages/paperclip-runner/provider-assets/copilot/copilot": "installed by the build\n",
"packages/paperclip-eval-kernel/dist/index.js": "changed\n",
});
assert.equal(computeBuildInputsDigest(repoRoot).digest, before.digest, "an output changed the digest");
// Every input kind changes it: TypeScript, Rust, the eval kernel, a root
// patch, root config, a new file, a rename.
const seen = new Set([before.digest]);
const expectChange = async (label, files) => {
await writeTree(repoRoot, files);
const next = computeBuildInputsDigest(repoRoot).digest;
assert.ok(!seen.has(next), `${label} did not change the digest`);
seen.add(next);
};
await expectChange("a Runner source", { "packages/paperclip-runner/src/index.ts": "export const one = 2;\n" });
await expectChange("a Rust source", { "packages/paperclip-runner/runner/src/main.rs": "fn main() { }\n" });
await expectChange("an eval-kernel source", { "packages/paperclip-eval-kernel/src/index.ts": "export const kernel = 2;\n" });
await expectChange("the pi-acp patch the build copies into dist", { "patches/pi-acp@0.0.33.patch": "--- c\n+++ d\n+e\n" });
await expectChange("the acpx patch the profile check hashes", { "patches/acpx@0.22.0.patch": "--- a\n+++ b\n+c\n" });
await expectChange("a new patch", { "patches/brace-expansion@5.0.9.patch": "--- x\n+++ y\n" });
await expectChange("the lockfile", { "pnpm-lock.yaml": "lockfileVersion: 9\nchanged: true\n" });
await expectChange("the base tsconfig", { "tsconfig.base.json": '{"compilerOptions":{}}\n' });
await expectChange("a new file", { "packages/paperclip-runner/protocol/new.json": "{}\n" });
await rm(path.join(repoRoot, "packages/paperclip-runner/protocol/new.json"));
await expectChange("a renamed file", { "packages/paperclip-runner/protocol/renamed.json": "{}\n" });
});
});
test("a symlink is hashed as its target path, not followed", async () => {
await withRepo(async ({ repoRoot, packageRoot }) => {
const before = computeBuildInputsDigest(repoRoot);
await symlink("../node_modules/dep", path.join(packageRoot, "src", "linked"), "dir");
const linked = computeBuildInputsDigest(repoRoot);
assert.notEqual(linked.digest, before.digest, "the link itself is an input");
assert.equal(linked.fileCount, before.fileCount + 1, "the link target's files were followed");
await writeFile(path.join(packageRoot, "node_modules/dep/index.js"), "changed\n");
assert.equal(computeBuildInputsDigest(repoRoot).digest, linked.digest, "a change behind the link changed the digest");
});
});
test("a build is current only with a matching stamp and every recorded output intact", async () => {
await withRepo(async ({ repoRoot, packageRoot }) => {
const platform = "linux";
const arch = "x64";
const assess = () => assessRunnerBuild({ repoRoot, packageRoot, platform, arch });
assert.equal(assess().current, false);
assert.equal(assess().code, "no_stamp");
const record = recordCurrentBuild({ repoRoot, packageRoot, platform, arch, now: () => new Date(0) });
assert.equal(record.recorded, true);
assert.equal(record.stampPath, buildInputsStampPath(packageRoot));
const stamp = readBuildInputsStamp(record.stampPath);
assert.equal(stamp.schema, BUILD_INPUTS_SCHEMA);
assert.equal(stamp.digest, record.digest);
assert.equal(stamp.fileCount, INPUT_FILE_COUNT);
assert.equal(stamp.byteCount, record.byteCount);
assert.equal(stamp.platform, platform);
assert.equal(stamp.arch, arch);
assert.equal(stamp.startedAt, "1970-01-01T00:00:00.000Z");
assert.equal(stamp.recordedAt, "1970-01-01T00:00:00.000Z");
assert.equal(stamp.outputCount, OUTPUT_FILE_COUNT);
assert.deepEqual(stamp.outputs, collectBuildOutputFiles(packageRoot));
const current = assess();
assert.equal(current.current, true, current.detail);
assert.equal(current.code, "inputs_unchanged");
assert.match(current.detail, /10 files and 5 outputs checked/);
assert.equal(assessRunnerBuild({ repoRoot, packageRoot, platform: "win32", arch }).code, "platform_mismatch");
assert.equal(assessRunnerBuild({ repoRoot, packageRoot, platform, arch: "arm64" }).code, "platform_mismatch");
// The four outputs every consumer reads are checked by name.
for (const output of expectedBuildOutputs(packageRoot, platform)) {
const content = await readFile(output);
await rm(output);
const missing = assess();
assert.equal(missing.current, false, `${output} missing still counted as current`);
assert.equal(missing.code, "missing_output");
assert.match(missing.detail, new RegExp(path.basename(output)));
await writeFile(output, content);
}
assert.equal(assess().current, true);
// So is every other file the build left under dist: a helper that is only
// opened by path must be there, at the size the build wrote.
const helper = path.join(packageRoot, "dist", "cli", "acpx-runtime-sidecar.cjs");
const helperContent = await readFile(helper);
await rm(helper);
const gone = assess();
assert.equal(gone.current, false, "a missing helper still counted as current");
assert.equal(gone.code, "outputs_changed");
assert.match(gone.detail, /^build output dist\/cli\/acpx-runtime-sidecar\.cjs is missing since the build recorded at /);
await writeFile(helper, helperContent);
assert.equal(assess().current, true);
await truncate(helper, 3);
const truncated = assess();
assert.equal(truncated.current, false, "a truncated helper still counted as current");
assert.equal(truncated.code, "outputs_changed");
assert.match(truncated.detail, /^build output dist\/cli\/acpx-runtime-sidecar\.cjs is changed size since /);
await writeFile(helper, helperContent);
assert.equal(assess().current, true);
// A file the build did not produce may appear without invalidating it.
await writeFile(path.join(packageRoot, "dist", "extra.txt"), "later\n");
assert.equal(assess().current, true);
await writeFile(path.join(packageRoot, "src", "index.ts"), "export const one = 3;\n");
const changed = assess();
assert.equal(changed.current, false);
assert.equal(changed.code, "inputs_changed");
await writeFile(record.stampPath, JSON.stringify({ ...readBuildInputsStamp(record.stampPath), schema: "other" }));
assert.equal(assess().code, "schema_mismatch");
await writeFile(record.stampPath, JSON.stringify({ ...readBuildInputsStamp(record.stampPath), schema: BUILD_INPUTS_SCHEMA, outputs: undefined }));
assert.equal(assess().code, "schema_mismatch", "a stamp without an output manifest was trusted");
await writeFile(record.stampPath, "not json");
assert.equal(assess().code, "no_stamp");
});
});
test("an input edited while the build runs leaves no stamp, so the next consumer rebuilds", async () => {
await withRepo(async ({ repoRoot, packageRoot }) => {
const stampPath = buildInputsStampPath(packageRoot);
const pendingPath = buildInputsPendingPath(packageRoot, currentBuildId());
const assess = () => assessRunnerBuild({ repoRoot, packageRoot });
// A previous good build, then a new build begins: the old stamp is gone at
// once, because the outputs are about to be replaced.
recordCurrentBuild({ repoRoot, packageRoot });
assert.equal(assess().current, true);
const pending = beginRunnerBuild({ repoRoot, packageRoot, now: () => new Date(1000) });
assert.equal(pending.pendingPath, pendingPath);
assert.equal(pending.buildId, String(process.ppid), "a build is named after the shell that runs its chain");
assert.equal(readBuildInputsStamp(stampPath), null, "the old stamp survived the start of a new build");
assert.equal(readBuildInputsStamp(pendingPath).digest, pending.digest);
assert.equal(readBuildInputsStamp(pendingPath).buildId, pending.buildId);
assert.equal(assess().code, "no_stamp");
// The build "compiles" the old source, then an edit lands before it ends.
await writeFile(path.join(packageRoot, "src", "index.ts"), "export const one = 2;\n");
const result = recordRunnerBuild({ repoRoot, packageRoot });
assert.equal(result.recorded, false);
assert.equal(result.code, "inputs_changed_during_build");
assert.match(result.detail, /^inputs changed while the build ran \(digest [0-9a-f]{12} at start, [0-9a-f]{12} now\); the next consumer rebuilds$/);
assert.equal(result.startedAt, "1970-01-01T00:00:01.000Z");
assert.equal(readBuildInputsStamp(stampPath), null, "a stamp was written for a build whose inputs moved");
assert.equal(readBuildInputsStamp(pendingPath), null, "the pending record was left behind");
assert.equal(assess().code, "no_stamp");
// An undisturbed build records normally, and only the pending record's
// start time is kept.
const record = recordCurrentBuild({ repoRoot, packageRoot });
assert.equal(record.recorded, true);
assert.equal(readBuildInputsStamp(pendingPath), null);
assert.equal(assess().current, true);
});
});
test("a build that did not begin, or that left a required output missing, cannot be recorded", async () => {
await withRepo(async ({ repoRoot, packageRoot }) => {
const stampPath = buildInputsStampPath(packageRoot);
recordCurrentBuild({ repoRoot, packageRoot });
assert.throws(
() => recordRunnerBuild({ repoRoot, packageRoot }),
new RegExp(
`no pending build inputs for build ${process.ppid} or ${process.pid} under .*paperclip-runner-build-inputs\\.pending: ` +
"the Runner build must start with `node scripts/record-build-inputs\\.mjs --begin`",
),
);
assert.equal(readBuildInputsStamp(stampPath), null, "the previous stamp outlived a build that did not begin");
// Another build's record is not this build's: a chain that did not begin
// cannot finish on the strength of one that is still running.
beginRunnerBuild({ repoRoot, packageRoot, buildId: "other-shell" });
assert.throws(() => recordRunnerBuild({ repoRoot, packageRoot }), /no pending build inputs for build /);
assert.equal(
readBuildInputsStamp(buildInputsPendingPath(packageRoot, "other-shell")).digest,
computeBuildInputsDigest(repoRoot).digest,
"a build that did not begin consumed another build's record",
);
await rm(buildInputsPendingPath(packageRoot, "other-shell"));
beginRunnerBuild({ repoRoot, packageRoot });
await rm(path.join(packageRoot, "dist", "testing.js"));
assert.throws(
() => recordRunnerBuild({ repoRoot, packageRoot }),
/the Runner build finished without producing dist\/testing\.js/,
);
assert.equal(readBuildInputsStamp(stampPath), null);
assert.equal(assessRunnerBuild({ repoRoot, packageRoot }).code, "no_stamp");
});
});
test("record-build-inputs.mjs brackets the build for the package it is pointed at", async () => {
await withRepo(async ({ repoRoot, packageRoot }) => {
const run = (...args) => execFileAsync(process.execPath, [recordScript, ...args, "--package-root", packageRoot]);
const expected = computeBuildInputsDigest(repoRoot);
const short = expected.digest.slice(0, 12);
// --finish before any --begin is a wiring error, not a silent stamp.
await assert.rejects(run("--finish"), /no pending build inputs/);
assert.equal(readBuildInputsStamp(buildInputsStampPath(packageRoot)), null);
const begin = await run("--begin");
assert.match(begin.stdout, new RegExp(`^${BUILD_INPUTS_SCHEMA} begin: digest=${short} files=${INPUT_FILE_COUNT} bytes=\\d+\n$`));
assert.equal(readBuildInputsStamp(buildInputsPendingPath(packageRoot, childBuildId)).digest, expected.digest);
const finish = await run("--finish");
assert.match(
finish.stdout,
new RegExp(`^${BUILD_INPUTS_SCHEMA} recorded: digest=${short} files=${INPUT_FILE_COUNT} bytes=\\d+ outputs=${OUTPUT_FILE_COUNT} stamp=node_modules/.cache/paperclip-runner-build-inputs.json\n$`),
);
const stamp = readBuildInputsStamp(buildInputsStampPath(packageRoot));
assert.equal(stamp.digest, expected.digest);
assert.equal(stamp.platform, process.platform);
assert.equal(stamp.arch, process.arch);
assert.equal(stamp.outputCount, OUTPUT_FILE_COUNT);
assert.equal(assessRunnerBuild({ repoRoot, packageRoot }).current, true);
assert.equal((await stat(buildInputsStampPath(packageRoot))).isFile(), true);
// An edit between the two phases is reported and leaves nothing behind.
await run("--begin");
await writeFile(path.join(packageRoot, "src", "index.ts"), "export const one = 2;\n");
const moved = await run("--finish");
assert.match(moved.stdout, /^paperclip\.runner-build-inputs not recorded: inputs changed while the build ran /);
assert.equal(readBuildInputsStamp(buildInputsStampPath(packageRoot)), null);
await assert.rejects(run(), /one of --begin or --finish is required/);
await assert.rejects(run("--begin", "--finish"), /--finish cannot be combined with --begin/);
await assert.rejects(
execFileAsync(process.execPath, [recordScript, "--begin", "--bogus"]),
/unknown argument --bogus/,
);
});
});
test("two builds running at once in one checkout each keep their own pending record", async () => {
await withRepo(async ({ repoRoot, packageRoot }) => {
const stampPath = buildInputsStampPath(packageRoot);
const pendingDirectory = buildInputsPendingDirectory(packageRoot);
const first = { repoRoot, packageRoot, buildId: "shell-1" };
const second = { repoRoot, packageRoot, buildId: "shell-2" };
const digestNow = () => computeBuildInputsDigest(repoRoot).digest;
// Nothing changes: both builds record, in either finishing order, and the
// second to finish does not fail for want of a record.
beginRunnerBuild(first);
beginRunnerBuild(second);
assert.deepEqual((await readdir(pendingDirectory)).sort(), ["shell-1.json", "shell-2.json"]);
assert.equal(recordRunnerBuild(first).recorded, true);
assert.deepEqual(await readdir(pendingDirectory), ["shell-2.json"], "the first finish consumed the wrong record");
assert.equal(readBuildInputsStamp(stampPath).digest, digestNow());
assert.equal(recordRunnerBuild(second).recorded, true);
assert.deepEqual(await readdir(pendingDirectory), []);
assert.equal(assessRunnerBuild({ repoRoot, packageRoot }).current, true);
beginRunnerBuild(first);
beginRunnerBuild(second);
assert.equal(recordRunnerBuild(second).recorded, true);
assert.equal(recordRunnerBuild(first).recorded, true);
assert.equal(assessRunnerBuild({ repoRoot, packageRoot }).current, true);
// An edit lands between the two starts. The build that began before it
// must not leave a stamp whichever of the two finishes last: a stamp means
// every build that wrote dist read the inputs it names.
const source = path.join(packageRoot, "src", "index.ts");
beginRunnerBuild(first);
await writeFile(source, "export const one = 2;\n");
beginRunnerBuild(second);
const staleFirst = recordRunnerBuild(first);
assert.equal(staleFirst.recorded, false);
assert.equal(staleFirst.code, "inputs_changed_during_build");
assert.equal(readBuildInputsStamp(stampPath), null);
assert.equal(recordRunnerBuild(second).recorded, true, "the build that read the edited inputs was not recorded");
assert.equal(assessRunnerBuild({ repoRoot, packageRoot }).current, true);
beginRunnerBuild(first);
await writeFile(source, "export const one = 3;\n");
beginRunnerBuild(second);
assert.equal(recordRunnerBuild(second).recorded, true);
assert.equal(readBuildInputsStamp(stampPath).digest, digestNow());
const staleLast = recordRunnerBuild(first);
assert.equal(staleLast.recorded, false);
assert.equal(staleLast.code, "inputs_changed_during_build");
assert.equal(readBuildInputsStamp(stampPath), null, "a build that began before the edit left the other build's stamp standing");
assert.equal(assessRunnerBuild({ repoRoot, packageRoot }).code, "no_stamp");
assert.deepEqual(await readdir(pendingDirectory), []);
});
});
test("a build's start drops the pending records of builds that never finished, and no others", async () => {
await withRepo(async ({ repoRoot, packageRoot }) => {
const pendingDirectory = buildInputsPendingDirectory(packageRoot);
const now = new Date("2026-10-11T22:00:00.000Z");
const age = async (buildId, ageMs) => {
const moment = new Date(now.getTime() - ageMs);
await utimes(buildInputsPendingPath(packageRoot, buildId), moment, moment);
};
// A build that failed a step yesterday, one that failed a minute ago, and
// one still running, all left records; a new build starts.
beginRunnerBuild({ repoRoot, packageRoot, buildId: "crashed-yesterday", now: () => now });
await age("crashed-yesterday", PENDING_RECORD_MAX_AGE_MS + 60_000);
beginRunnerBuild({ repoRoot, packageRoot, buildId: "crashed-just-now", now: () => now });
await age("crashed-just-now", 60_000);
beginRunnerBuild({ repoRoot, packageRoot, buildId: "still-running", now: () => now });
await writeFile(path.join(pendingDirectory, "notes.txt"), "not a record\n");
await age("still-running", 0);
beginRunnerBuild({ repoRoot, packageRoot, buildId: "new-build", now: () => now });
assert.deepEqual(
(await readdir(pendingDirectory)).sort(),
["crashed-just-now.json", "new-build.json", "notes.txt", "still-running.json"],
"pruning removed the wrong records",
);
// The running build and the new one still finish on their own records.
assert.equal(recordRunnerBuild({ repoRoot, packageRoot, buildId: "still-running" }).recorded, true);
assert.equal(recordRunnerBuild({ repoRoot, packageRoot, buildId: "new-build" }).recorded, true);
// A record that fell out of the directory is simply absent.
await rm(pendingDirectory, { recursive: true, force: true });
beginRunnerBuild({ repoRoot, packageRoot, buildId: "after-wipe" });
assert.deepEqual(await readdir(pendingDirectory), ["after-wipe.json"]);
});
});
test("two chains run by separate shells at the same time both record", { skip: process.platform === "win32" }, async () => {
await withRepo(async ({ repoRoot, packageRoot }) => {
const pendingDirectory = buildInputsPendingDirectory(packageRoot);
const marker = path.join(repoRoot, "release-the-chains");
const records = () => (existsSync(pendingDirectory) ? readdirSync(pendingDirectory).filter(name => name.endsWith(".json")).sort() : []);
const staged = path.join(pendingDirectory, "999999.json.999999.tmp");
const chains = [runChainInShell("sh", packageRoot, marker)];
let results;
try {
await waitFor(() => records().length === 1, "the first chain's record");
await writeFile(staged, "{");
chains.push(runChainInShell("sh", packageRoot, marker));
await waitFor(() => records().length === 2, "the second chain's record");
const [firstId, secondId] = records().map((name) => name.replace(/\.json$/, ""));
assert.notEqual(firstId, secondId);
for (const id of [firstId, secondId]) {
assert.match(id, /^\d+$/, "a chain's record is named after its shell's pid");
assert.equal(readBuildInputsStamp(buildInputsPendingPath(packageRoot, id)).buildId, id);
}
} finally {
await rm(staged, { force: true });
await writeFile(marker, "go\n");
results = await Promise.allSettled(chains);
}
for (const result of results) {
assert.equal(result.status, "fulfilled", result.reason?.message);
assert.match(result.value.stdout, new RegExp(`${BUILD_INPUTS_SCHEMA} begin: `));
assert.match(result.value.stdout, new RegExp(`${BUILD_INPUTS_SCHEMA} recorded: `), "a chain finished without recording");
}
assert.deepEqual(await readdir(pendingDirectory), []);
assert.equal(assessRunnerBuild({ repoRoot, packageRoot }).current, true);
// The same chain under bash, which may exec its last command: `--finish`
// is then the shell itself and still finds the record `--begin` wrote.
let bash;
try {
bash = (await execFileAsync("bash", ["-c", "echo ok"])).stdout.trim() === "ok";
} catch {
bash = false;
}
if (bash) {
await rm(marker);
const chain = runChainInShell("bash", packageRoot, marker);
await waitFor(() => records().length === 1, "the bash chain's record");
await writeFile(marker, "go\n");
assert.match((await chain).stdout, new RegExp(`${BUILD_INPUTS_SCHEMA} recorded: `));
assert.deepEqual(records(), []);
}
});
});