v1.94.17.0 ci-carry: guard carry, jobs-mapped lanes, aliases, per-file carry of runner-skipped files, path-skipped jobs (GST-1186, GST-1189, GST-1190) (#3929)

* feat(pregate): a guard that already passed at this tree carries from CI or the ledger (GST-1186)

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* feat(ci-carry): jobs-mapped lanes, aliases, narrow-lane counts, --simulate proof, policy-base check on carried (GST-1189, GST-1190)

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* fix(ci-carry): a file whose every test skipped on CI is not carried; tree.json records skipped_files (GST-1190)

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* feat(ci-carry): path-skipped jobs count only under an unchanged gate; find --policy-from head --simulate (GST-1190)

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* ship: name the pre-push guards in Step 17 and the early push (moved from #3923 per GSTA-258)

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* ci-carry: per-file lane carry for runner-skipped files; review fixes (gate inputs, passed ignores CI rows, simulate consent, pull_request-only skips, guard not-owed)

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* ship: Step 17 names the pre-push guards without a code this branch cannot print

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* pregate: gstack's bun run test lane runs its runner-skipped files with per_file

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* ci-carry: a per_file remainder must run tests here and the carried runs must still stand (l6c-p2)

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* ci-carry: read record-ci from its own last record line; the remainder's count must come from this run's row (l6c-p2 red-team)

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* test(ci-tree-reuse): type the skipped_files fixture for the test typecheck ratchet

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* Revert "pregate: gstack's bun run test lane runs its runner-skipped files with per_file"

This reverts commit fafbae5b9b.

* ci-carry: ship the skipped_files reader inert; gstack's writer and per_file move to the adoption PR; Step 17 keeps its refs/heads marker

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* docs: guard carry, per_file and the too-old message in pregate and troubleshooting (docs audit l6cdocs2)

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* gstack-evidence record-ci: recheck the working tree after the post-remainder run recheck (#3929 second fix-check)

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* windows-curation: exclude test/ship-ci-carry.test.ts like its ship-bin-fixture siblings (the path-scoped Windows check runs it at its first spawn)

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

---------

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>
This commit is contained in:
Garry Tanandcapy-ai[bot] authored and GitHub committed 2026-11-03 14:14:50 +00:00
1 parent 94771d927f
commit d2e63181d7
35 files changed
+1144 -134

No files matched your search

+93 -18
View File
@@ -14,7 +14,9 @@
* max age, the run's own tree.json naming HEAD's tree);
* every run executed tests;
* with an inventory command, the runs' tree.json files cover every file it
* prints (a lane carries only what CI provably ran);
* prints (a lane carries only what CI provably ran); with named jobs instead
* (a lane with no test files, GST-1189), each of those jobs succeeded in the
* run's newest attempt, read by verifyRun like the profile's own jobs;
* every run still stands after the inventory ran (runChanged).
*
* Anything else returns a refusal with a reason code, and the caller runs the
@@ -25,12 +27,18 @@
* bounded by a timeout, and returns green, red with the failing jobs, still
* running, or none.
*/
import { runChanged, runReceiptFields, uncoveredFiles, verifyRun, type ApiRun, type RunDeps, type RunEvidence, type RunProfile, type RunSource } from './run-evidence';
import { normalPath, runChanged, runInventory, runReceiptFields, uncoveredFiles, verifyRun, type ApiRun, type RunDeps, type RunEvidence, type RunProfile, type RunSource } from './run-evidence';
export type CarryRefusal = 'dirty_tree' | 'no_profile' | 'no_run' | 'no_verified_run' | 'tree_mismatch' | 'no_tests'
| 'inventory_failed' | 'inventory_empty' | 'inventory_uncovered' | 'run_changed' | 'tree_changed' | 'api_error';
| 'inventory_failed' | 'inventory_empty' | 'inventory_uncovered' | 'inventory_skipped' | 'run_changed' | 'tree_changed' | 'api_error';
export type CarryRefused = { ok: false; reason: CarryRefusal; detail: string };
export type CarryVerdict = { ok: true; runs: RunEvidence[]; tests: number; receipt: string } | CarryRefused;
/**
* `tests` never claims a whole run's count for a narrow lane: with an inventory it is the number of the lane's inventory files
* the runs' tree.json covers (`unit: 'files'`; tree.json has no per-file test counts yet), with named jobs `n/a`, and only
* without either (the whole run carried) the runs' test count. `remainder`: inventory files every run skipped entirely
* (tree.json `skipped_files`), which the caller must run here with the lane's per_file runner before the lane counts.
*/
export type CarryVerdict = { ok: true; runs: RunEvidence[]; tests: number | 'n/a'; unit?: 'files'; notOwed?: string; remainder?: string[]; receipt: string } | CarryRefused;
/** How many of HEAD's runs per workflow are tried, newest first. */
const CANDIDATES = 10;
@@ -38,12 +46,13 @@ const no = (reason: CarryRefusal, detail: string): CarryRefused => ({ ok: false,
const newestFirst = (a: ApiRun, b: ApiRun) => (b.created_at ?? '').localeCompare(a.created_at ?? '') || b.id - a.id;
/** The newest run of `profile` on `head` that verifies; a refusal lists each candidate's reason, newest first. */
function newestVerified(src: RunSource, name: string, profile: RunProfile, head: string): { ok: true; evidence: RunEvidence } | CarryRefused {
function newestVerified(src: RunSource, name: string, profile: RunProfile, head: string, jobs: string[]): { ok: true; evidence: RunEvidence } | CarryRefused {
if (jobs.length) profile = { ...profile, jobs: [...new Set([...profile.jobs, ...jobs])] };
const listed = src.deps.runsForHead(profile.workflow, head).filter(r => profile.events.includes(r.event)).sort(newestFirst).slice(0, CANDIDATES);
if (!listed.length) return no('no_run', `no ${profile.workflow} run (${profile.events.join(', ')}) on ${head.slice(0, 12)} yet`);
const refusals: string[] = [];
for (const run of listed) {
const v = verifyRun(src.deps, { repo: src.repo, runRef: String(run.id), profileName: name, profile, vetoHeads: [head], now: src.now });
const v = verifyRun(src.deps, { repo: src.repo, runRef: String(run.id), profileName: name, profile, vetoHeads: [head], now: src.now, unchanged: src.unchanged });
if (v.ok) return v;
refusals.push(`run ${run.id}: ${v.reason}`);
}
@@ -52,9 +61,13 @@ function newestVerified(src: RunSource, name: string, profile: RunProfile, head:
/**
* The CI evidence for HEAD's tree under `profiles` (one verified run each), or why there is none. `wtree` reads the
* working-tree fingerprint (before the lookup and after it); `inventory` lists the files the runs must cover.
* working-tree fingerprint (before the lookup and after it); `inventory` lists the files the runs must cover, or `jobs`
* names the jobs that must have succeeded in each run's newest attempt (a lane with no test files; no test count needed).
*/
export function findCarry(src: RunSource, o: { profiles: string[]; head: string; tree: string; wtree: () => string | null; inventory?: () => string[] | { error: string } }): CarryVerdict {
export function findCarry(src: RunSource, o: { profiles: string[]; head: string; tree: string; wtree: () => string | null; inventory?: () => string[] | { error: string }; jobs?: string[]; perFile?: boolean }): CarryVerdict {
const jobs = o.jobs ?? [];
let covered: number | undefined;
let remainder: string[] = [];
const before = o.wtree();
if (before !== o.tree) return no('dirty_tree', `the working tree (${before ? before.slice(0, 12) : 'fingerprint unavailable'}) is not HEAD's committed tree ${o.tree.slice(0, 12)}; only a clean, committed tree carries`);
if (!o.profiles.length) return no('no_profile', 'no run profile is named');
@@ -63,20 +76,24 @@ export function findCarry(src: RunSource, o: { profiles: string[]; head: string;
for (const name of o.profiles) {
const profile = src.profiles[name];
if (!profile) return no('no_profile', `${name} is not a run_profiles name on the base branch`);
const v = newestVerified(src, name, profile, o.head);
const v = newestVerified(src, name, profile, o.head, jobs);
if (!v.ok) return v;
runs.push(v.evidence);
}
const other = runs.find(e => e.tree !== o.tree);
if (other) return no('tree_mismatch', `run ${other.run_id} tested tree ${other.tree.slice(0, 12)}, not ${o.tree.slice(0, 12)}`);
const empty = runs.find(e => !e.files.length || e.tests === 0);
const empty = jobs.length ? undefined : runs.find(e => !e.files.length || e.tests === 0);
if (empty) return no('no_tests', `run ${empty.run_id}'s tree.json ${empty.files.length ? 'records 0 tests' : 'lists no test files'}`);
if (o.inventory) {
const listed = o.inventory();
if ('error' in listed) return no('inventory_failed', listed.error);
if (!listed.length) return no('inventory_empty', 'the inventory command printed no files');
const missing = uncoveredFiles(listed, runs);
if (missing.length) return no('inventory_uncovered', `${missing.length} of ${listed.length} inventory file(s) are in no run's tree.json: ${missing.slice(0, 5).join(', ')}${missing.length > 5 ? ', …' : ''}`);
const absent = uncoveredFiles(listed, runs, 'covered');
const some = (fs: string[]) => `${fs.slice(0, 5).join(', ')}${fs.length > 5 ? ', …' : ''}`;
if (absent.length) return no('inventory_uncovered', `${absent.length} of ${listed.length} inventory file(s) are in no run's tree.json: ${some(absent)}`);
remainder = uncoveredFiles(listed, runs, 'uncovered');
if (remainder.length && !o.perFile) return no('inventory_skipped', `${remainder.length} of ${listed.length} inventory file(s) skipped every test in CI and the lane declares no per_file runner: ${some(remainder)}`);
covered = new Set(listed.map(normalPath)).size - remainder.length;
}
for (const e of runs) {
const changed = runChanged(src.deps, e.run_id, e.run_attempt);
@@ -87,6 +104,13 @@ export function findCarry(src: RunSource, o: { profiles: string[]; head: string;
}
const after = o.wtree();
if (after !== o.tree) return no('tree_changed', `the working tree changed while the runs were verified (now ${after ? after.slice(0, 12) : 'unreadable'})`);
if (jobs.length) {
const gates = [...new Set(o.profiles.map(n => src.profiles[n]?.skippable?.gate).filter(Boolean))];
const allSkipped = runs.every(e => jobs.every(j => e.skipped_jobs?.includes(j)));
const notOwed = allSkipped ? `CI skipped at this tree (gate job ${gates.join(', ')}, workflow unchanged)` : undefined;
return { ok: true, runs, tests: 'n/a', ...(notOwed ? { notOwed } : {}), receipt: `ci-carry tree=${o.tree.slice(0, 12)} jobs=${jobs.join(',')}${notOwed ? ` not_owed=${gates.join(',')}` : ''} ${runReceiptFields(runs, src.now)}` };
}
if (covered !== undefined) return { ok: true, runs, tests: covered, unit: 'files', ...(remainder.length ? { remainder } : {}), receipt: `ci-carry tree=${o.tree.slice(0, 12)} files=${covered}${remainder.length ? ` skipped_here=${remainder.length}` : ''} ${runReceiptFields(runs, src.now)}` };
const tests = runs.reduce((n, e) => n + (e.tests ?? e.files.length), 0);
return { ok: true, runs, tests, receipt: `ci-carry tree=${o.tree.slice(0, 12)} tests=${tests} ${runReceiptFields(runs, src.now)}` };
}
@@ -109,24 +133,75 @@ export async function waitForRun(deps: RunDeps, o: { workflow: string; events: s
}
}
/** `ship_lanes_from` (base branch only): exact lane command → the profiles whose runs carry it and the command listing its test files. */
export interface ShipLaneMapping { profiles: string[]; inventory: string }
/**
* `ship_lanes_from` (base branch only): exact lane command → the profiles whose runs carry it and either the command listing
* its test files (`inventory`) or the CI jobs that run it (`jobs`, for a lane with no test files such as a typecheck; GST-1189).
*/
export interface ShipLaneMapping { profiles: string[]; inventory?: string; jobs?: string[]; aliases?: string[]; per_file?: string }
/**
* The `ship_lanes_from` entry for an exact lane command: the entry it keys, else the entry that declares it in `aliases`
* (GST-1190: the same lane run under another exact spelling, such as an env prefix). Never normalized: an alias is a
* second exact command the base's owner declared, not a pattern.
*/
export function laneMapping(from: Record<string, ShipLaneMapping>, command: string): ShipLaneMapping | undefined {
if (Object.prototype.hasOwnProperty.call(from, command)) return from[command];
return Object.values(from).find(m => m.aliases?.includes(command));
}
/** findCarry's coverage rule for one mapping: its inventory (a pin-file command run in `cwd`, with a per_file runner or not) or its named jobs. */
export const mappingCoverage = (m: ShipLaneMapping, cwd: string): { inventory?: () => string[] | { error: string }; jobs?: string[]; perFile?: boolean } =>
m.jobs ? { jobs: m.jobs } : { inventory: () => runInventory(m.inventory!, cwd), perFile: !!m.per_file };
/**
* What `gstack-evidence record-ci` reported, read from its own last record line onward (the per_file remainder's output comes
* first, so a test printing look-alike lines cannot change the reading): the carried runs, count, unit and remainder, any
* not-owed reason, or the last CARRY_NONE reason.
*/
export function recordCiResult(stderr: string): { runs?: string; tests?: string; files?: boolean; notOwed?: string; here?: string; none?: string } {
const lines = stderr.split('\n');
const at = lines.findLastIndex(x => x.startsWith('gstack-evidence: recorded label=') && x.includes(' log=- source=ci runs='));
const none = lines.findLast(x => x.startsWith('CARRY_NONE: '))?.slice('CARRY_NONE: '.length);
if (at < 0) return none ? { none } : {};
const got = / tests_ran=(\d+|n\/a)( unit=files)? log=- source=ci runs=(\S+)/.exec(lines[at]!);
const here = / skipped_here=(\d+) tests_ran_here=(\S+)$/.exec(lines[at]!);
const notOwed = lines.slice(at + 1).find(x => x.startsWith('gstack-evidence: not owed: '))?.slice('gstack-evidence: not owed: '.length);
return { runs: got?.[3], tests: got?.[1], files: !!got?.[2], ...(notOwed ? { notOwed } : {}), ...(here ? { here: `skipped_here=${here[1]} file(s) tests_ran_here=${here[2]}` } : {}) };
}
/** The per_file runner command for `files`: `{files}` replaced by the paths, each single-quoted for bash. */
export const perFileCommand = (template: string, files: string[]): string =>
template.replaceAll('{files}', files.map(f => `'${f.replace(/'/g, `'\\''`)}'`).join(' '));
/** `<path>: <message>` errors for `ship_lanes_from`; empty when valid. Shared with lib/pregate/config.ts. */
export function shipLanesErrors(raw: unknown, profileNames: readonly string[]): Array<{ path: string; message: string }> {
if (raw === undefined) return [];
if (!raw || typeof raw !== 'object' || Array.isArray(raw)) return [{ path: 'ship_lanes_from', message: 'must be an object of exact lane command → { profiles: [run_profiles names], inventory: command }' }];
if (!raw || typeof raw !== 'object' || Array.isArray(raw)) return [{ path: 'ship_lanes_from', message: 'must be an object of exact lane command → { profiles: [run_profiles names], inventory: command } or { profiles, jobs: [job names] }' }];
const errors: Array<{ path: string; message: string }> = [];
const oneLine = (s: unknown) => typeof s === 'string' && !!s.trim() && s.length <= 512 && !/[\0\r\n]/.test(s);
const declared = new Map<string, string>(Object.keys(raw as object).map(k => [k, 'a key']));
for (const [command, m] of Object.entries(raw as Record<string, unknown>)) {
const at = `ship_lanes_from[${JSON.stringify(command)}]`;
if (!oneLine(command)) errors.push({ path: at, message: 'the key must be the exact lane command: one line under 512 characters' });
if (!m || typeof m !== 'object' || Array.isArray(m)) { errors.push({ path: at, message: 'must be { profiles, inventory }' }); continue; }
if (!m || typeof m !== 'object' || Array.isArray(m)) { errors.push({ path: at, message: 'must be { profiles, inventory } or { profiles, jobs }' }); continue; }
const o = m as Record<string, unknown>;
for (const k of Object.keys(o)) if (k !== 'profiles' && k !== 'inventory') errors.push({ path: `${at}.${k}`, message: 'unknown key; valid keys: profiles, inventory' });
for (const k of Object.keys(o)) if (!['profiles', 'inventory', 'jobs', 'aliases', 'per_file'].includes(k)) errors.push({ path: `${at}.${k}`, message: 'unknown key; valid keys: profiles, inventory, jobs, aliases, per_file' });
if ('per_file' in o && (!oneLine(o.per_file) || !(o.per_file as string).includes('{files}') || !('inventory' in o))) errors.push({ path: `${at}.per_file`, message: 'must be one command line under 512 characters containing {files}, on an entry with an inventory (it runs the inventory files CI skipped entirely)' });
if ('aliases' in o) {
const aliases = o.aliases;
if (!Array.isArray(aliases) || !aliases.length || aliases.some(a => !oneLine(a)) || new Set(aliases).size !== aliases.length) errors.push({ path: `${at}.aliases`, message: 'must be a non-empty array of distinct exact lane commands, each one line under 512 characters' });
else for (const a of aliases as string[]) {
if (declared.has(a)) errors.push({ path: `${at}.aliases`, message: `${JSON.stringify(a)} is already ${declared.get(a)}; each exact command maps to one entry` });
else declared.set(a, `an alias of ${JSON.stringify(command)}`);
}
}
if (!Array.isArray(o.profiles) || !o.profiles.length || o.profiles.some(p => typeof p !== 'string')) errors.push({ path: `${at}.profiles`, message: 'must be a non-empty array of run_profiles names' });
else for (const p of o.profiles as string[]) if (!profileNames.includes(p)) errors.push({ path: `${at}.profiles`, message: `${JSON.stringify(p)} is not a run_profiles name` });
if (!oneLine(o.inventory)) errors.push({ path: `${at}.inventory`, message: 'must be one command line under 512 characters that prints the lane\'s test files, one per line' });
if (('inventory' in o) === ('jobs' in o)) errors.push({ path: at, message: 'declare exactly one of inventory (a command listing the lane\'s test files) or jobs (the CI jobs that run it)' });
else if ('jobs' in o) {
const jobs = o.jobs;
if (!Array.isArray(jobs) || !jobs.length || jobs.some(j => typeof j !== 'string' || !oneLine(j) || j.length > 128) || new Set(jobs).size !== jobs.length) errors.push({ path: `${at}.jobs`, message: 'must be a non-empty array of distinct CI job names, each one line under 128 characters' });
} else if (!oneLine(o.inventory)) errors.push({ path: `${at}.inventory`, message: 'must be one command line under 512 characters that prints the lane\'s test files, one per line' });
}
return errors;
}
+38 -3
View File
@@ -11,9 +11,17 @@
* incident: a gbrain serial test landed without its manifest row and
* passed both the isolation check and pregate; `test/cli-goldens.test.ts`
* spawns the CLI as a subprocess, so no import graph selects it.
*
* GST-1186: a guard whose exact command already passed at this tree is not run
* again. #3901's carry (gstack-evidence record-ci) carries it from a verified
* CI run when the base's ship_lanes_from maps the command, else a counted
* ledger pass of the same command at the same working tree stands
* (gstack-evidence passed). ship_lanes_from is read from origin/<base> only, so
* the PR that adds a mapping cannot carry from it yet.
*/
import * as fs from 'node:fs';
import * as path from 'node:path';
import { laneMapping, recordCiResult } from '../../ci-carry';
import { matchesAny } from '../../ship-policy';
import type { Guard } from '../config';
import { baseInputs, sha256 } from '../context';
@@ -23,8 +31,29 @@ import { checkScratch } from '../run-scratch';
import { readRange, STALL_IDLE, supervise } from '../supervise';
import type { EnvironmentCause } from '../types';
import type { CheckDef, CheckResult, PregateContext } from '../types';
import { runScriptSync } from '../../script-invocation';
interface GuardRun { guard: Guard; exit: number | null; timedOut: boolean; stalled: boolean; tail: string[]; log: string; environment: EnvironmentCause | null }
interface GuardRun { guard: Guard; exit: number | null; timedOut: boolean; stalled: boolean; tail: string[]; log: string; environment: EnvironmentCause | null; prior?: string; carryNone?: string }
const EVIDENCE = path.resolve(import.meta.dir, '..', '..', '..', 'bin', 'gstack-evidence');
/**
* The pass that already covers this guard's exact command at this tree (GST-1186): a CI carry (or a CI skip the unchanged gate
* decided, printed as not owed), else a same-tree local ledger pass under the runtime the guard would launch with.
*/
function priorPass(ctx: PregateContext, guard: Guard, env: NodeJS.ProcessEnv): { prior?: string; carryNone?: string } {
const evidence = (args: string[]) => runScriptSync(EVIDENCE, args, { cwd: ctx.repoRoot, env, encoding: 'utf8', timeout: 600_000 });
let carryNone: string | undefined;
if (laneMapping(ctx.config.ship_lanes_from, guard.command)) {
const ci = evidence(['record-ci', '--label', `guard-${guard.id}`, '--base', ctx.base, '--', guard.command]);
const got = recordCiResult(String(ci.stderr ?? ''));
if (ci.status === 0 && got.runs) return { prior: got.notOwed ? `not owed: ${got.notOwed} runs=${got.runs}` : `carried runs=${got.runs}` };
carryNone = got.none ?? `record-ci exit ${ci.status}`;
}
const local = evidence(['passed', '--', guard.command]);
const hit = /^PASSED: (.+)$/m.exec(String(local.stdout ?? ''))?.[1];
return local.status === 0 && hit ? { prior: `reused ${hit}`, carryNone } : { carryNone };
}
async function runGuard(ctx: PregateContext, guard: Guard, env: NodeJS.ProcessEnv, dir: string): Promise<GuardRun> {
const timeoutMs = guard.timeout_s ? guard.timeout_s * 1000 : ctx.timeoutMs;
@@ -53,9 +82,14 @@ async function run(ctx: PregateContext): Promise<CheckResult> {
if (runtime.refusal) return { ...base, ...runtime.refusal, cause: 'environment', environment: launchCause('runtime', runtime.refusal.detail), lines: [`runtime: ${runtime.label}`, environmentLine(launchCause('runtime', runtime.refusal.detail))] };
const dir = checkScratch(ctx.scratch, 'guards');
const runs: GuardRun[] = [];
for (const g of selected) runs.push(await runGuard(ctx, g, runtime.env, dir));
for (const g of selected) {
const p = priorPass(ctx, g, runtime.env);
runs.push(p.prior ? { guard: g, exit: 0, timedOut: false, stalled: false, tail: [], log: '-', environment: null, ...p } : { ...(await runGuard(ctx, g, runtime.env, dir)), carryNone: p.carryNone });
}
const lines = [`runtime: ${runtime.label}`];
for (const r of runs) {
if (r.carryNone) lines.push(`CI_CARRY: none guard=${r.guard.id} (${r.carryNone})`);
if (r.prior) { lines.push(`GUARD: ${r.guard.id} ${r.prior}`); continue; }
lines.push(r.stalled ? `GUARD: ${r.guard.id} stalled: no output for ${ctx.stallTimeoutMs / 1000} s${STALL_IDLE}; process group killed (log ${r.log})` : r.timedOut ? `GUARD: ${r.guard.id} timed out after ${r.guard.timeout_s ?? ctx.timeoutMs / 1000} s (log ${r.log})` : `GUARD: ${r.guard.id} exit=${r.exit} (log ${r.log})`);
if (r.exit !== 0) lines.push(...r.tail.map(l => ` ${l}`));
if (r.environment) lines.push(` ${r.guard.id} ${environmentLine(r.environment)}`);
@@ -68,7 +102,8 @@ async function run(ctx: PregateContext): Promise<CheckResult> {
if (failed.length) return { ...withRuns, status: 'fail', cause: 'diff', code: 'PREGATE_GUARD_FAILED', detail: `${failed.length} guard(s) failed: ${failed.map(r => r.guard.id).join(', ')}${timedOut.length ? `; timed out: ${timedOut.map(r => r.guard.id).join(', ')}` : ''}${blocked.length ? `; blocked by the environment: ${blocked.map(r => r.guard.id).join(', ')}` : ''}`, fix: fixOf(failed) };
if (blocked.length) return environmentResult(withRuns, blocked[0]!.environment!, `${blocked.length} guard(s) could not run on this machine: ${blocked.map(r => `${r.guard.id} (${r.environment!.kind})`).join(', ')}${timedOut.length ? `; timed out: ${timedOut.map(r => r.guard.id).join(', ')}` : ''}`);
if (timedOut.length) return { ...withRuns, status: 'incomplete', code: timedOut.some(r => r.stalled) ? 'PREGATE_STALLED' : 'PREGATE_INCOMPLETE', detail: `${timedOut.length} guard(s) did not finish: ${timedOut.map(r => `${r.guard.id} (${r.stalled ? 'stalled' : 'timed out'})`).join(', ')}`, fix: 'fix the hang, or raise the guard\'s timeout_s, --timeout or --stall-timeout; an unfinished guard never passes' };
return { ...withRuns, detail: `${runs.length} guard(s) passed: ${runs.map(r => r.guard.id).join(', ')}` };
const prior = runs.filter(r => r.prior).length;
return { ...withRuns, detail: `${runs.length} guard(s) passed: ${runs.map(r => r.guard.id).join(', ')}${prior ? ` (${prior} from an earlier pass at this tree)` : ''}` };
}
export const guardsCheck: CheckDef = { id: 'guards', stage: 'preflight', tier: 2, downgradable: false, needsRepoCommands: true, run };
+4 -3
View File
@@ -40,7 +40,7 @@ import { selectLanes, type Selection } from '../lane-selection';
import { launchRuntime } from '../runtime';
import { cargoModel, type CargoModel } from '../rust';
import { adapterFor, readRange, STALL_IDLE, supervise, type Counter, type Supervised } from '../supervise';
import { verifyFromRun, type RunProfile } from '../../run-evidence';
import { ranFiles, verifyFromRun, type RunProfile } from '../../run-evidence';
import { matchesAny } from '../../ship-policy';
import type { CheckDef, CheckResult, PregateContext, RemoteObligation } from '../types';
@@ -643,9 +643,10 @@ function fromRun(ctx: PregateContext, sel: Selection): { line: string; passed: s
const names = 'error' in check ? [] : fromRunProfiles(ctx, check.profiles);
const v = 'error' in check ? { ok: false as const, reason: 'profiles_unavailable', detail: check.error } : verifyFromRun(check, ref, names, ctx.wtree, ctx.head);
if (!v.ok) return { line: `REUSE: refused reason=${v.reason} — ${v.detail}; every selected test runs here`, passed: [] };
const ran = new Set(v.evidence.files);
const ran = new Set(ranFiles(v.evidence));
const passed = sel.localTests.filter(t => ran.has(t));
return { line: `REUSE: accepted ${passed.length} of ${sel.localTests.length} selected local test file(s) from ${v.receipt}; ${sel.localTests.length - passed.length} still run`, passed, receipt: v.receipt };
const skipped = sel.localTests.filter(t => v.evidence.skipped?.includes(t)).length;
return { line: `REUSE: accepted ${passed.length} of ${sel.localTests.length} selected local test file(s) from ${v.receipt}; ${sel.localTests.length - passed.length} still run${skipped ? ` (${skipped} skipped every test in the run)` : ''}`, passed, receipt: v.receipt };
}
/**
+30 -7
View File
@@ -46,9 +46,13 @@
* tree, the command listing the gate's test files they must
* cover, and the command naming the selected files the gate
* itself never runs (gstack: paid files)
* ship_lanes_from exact /ship lane command → `{ profiles, inventory }`: the
* run profiles whose verified runs at HEAD's tree carry that
* lane, and the command listing its test files (lib/ci-carry.ts)
* ship_lanes_from exact /ship lane command → `{ profiles, inventory }` or
* `{ profiles, jobs }`: the run profiles whose verified runs
* at HEAD's tree carry that lane, and either the command
* listing its test files or the CI jobs that run it (a lane
* with no test files, GST-1189), plus optional `aliases`:
* other exact commands for the same lane (GST-1190)
* (lib/ci-carry.ts)
* allowlist `{ finding, paths?, pattern?, reason }` entries the `secrets`
* check accepts as benign: a gstack-redact finding id or a
* non-secret category, scoped by path globs and/or a regex
@@ -74,7 +78,8 @@ import { shipLanesErrors, type ShipLaneMapping } from '../ci-carry';
import type { ResultCodeName } from '../result-codes';
import { runProfileErrors, type RemoteGateRuns, type RunProfile } from '../run-evidence';
import { sealedErrors } from '../sealed-guard';
import { containedRelative, globToRegExp } from '../ship-policy';
import { containedRelative, globToRegExp, runningGstackVersion, SHIP_POLICY_PATH } from '../ship-policy';
import { cmpVersion, parseVersion } from '../version-source';
export const PREGATE_CONFIG_PATH = '.gstack/pregate.json';
export const PREGATE_CONFIG_SCHEMA = 'https://github.com/garrytan/gstack/docs/pregate.md#pregate-v1';
@@ -291,6 +296,22 @@ function git(cwd: string, args: string[]): { ok: boolean; out: string } {
return { ok: r.status === 0 && !r.error, out: (r.stdout ?? '').trim() };
}
/**
* GST-1190: an unknown key in a policy whose project floor (min_gstack_version in the same tree's ship-policy.json) is
* above this gstack belongs to a newer gstack. It still refuses, but as an upgrade, not as a typo.
*/
function newerGstackErrors(parsed: { config: PregateConfig; errors: ConfigError[]; declaresLanes: boolean }, policyText: string | null): typeof parsed {
const unknown = (e: ConfigError) => e.message.startsWith('unknown key');
if (!policyText || !parsed.errors.some(unknown)) return parsed;
let min: unknown;
try { min = (JSON.parse(policyText) as Record<string, unknown> | null)?.min_gstack_version; } catch { return parsed; }
const running = runningGstackVersion();
const want = typeof min === 'string' ? parseVersion(min) : null;
const have = parseVersion(running);
if (!want || !have || cmpVersion(want, have) <= 0) return parsed;
return { ...parsed, errors: parsed.errors.map(e => unknown(e) ? { ...e, code: 'PREGATE_GSTACK_TOO_OLD' as const, message: `needs gstack >= ${min as string}; upgrade (this gstack is ${running} and does not know this key)` } : e) };
}
export function loadConfigFromBase(repoRoot: string, base: string, knownChecks: readonly string[] = []): ConfigLoad {
const ref = `origin/${base.replace(/^origin\//, '')}`;
const sha = git(repoRoot, ['rev-parse', '--verify', '--quiet', `${ref}^{commit}`]);
@@ -302,13 +323,15 @@ export function loadConfigFromBase(repoRoot: string, base: string, knownChecks:
if (!entry.ok || entry.out) return { unavailable: true, base: base.replace(/^origin\//, ''), detail: `${ref}:${PREGATE_CONFIG_PATH} could not be read and git cannot confirm it is absent` };
return { config: DEFAULT_CONFIG, errors: [], present: false, label: `${ref}@${sha.out.slice(0, 12)} (no ${PREGATE_CONFIG_PATH})`, kind: 'none' };
}
return { ...parseConfig(show.out, knownChecks), present: true, label: `${ref}@${sha.out.slice(0, 12)}:${PREGATE_CONFIG_PATH}`, kind: 'base' };
const policy = git(repoRoot, ['show', `${sha.out}:${SHIP_POLICY_PATH}`]);
return { ...newerGstackErrors(parseConfig(show.out, knownChecks), policy.ok ? policy.out : null), present: true, label: `${ref}@${sha.out.slice(0, 12)}:${PREGATE_CONFIG_PATH}`, kind: 'base' };
}
export function loadConfigFromHead(repoRoot: string, knownChecks: readonly string[] = []): ConfigLoad {
const file = path.join(repoRoot, PREGATE_CONFIG_PATH);
if (!fs.existsSync(file)) return { config: DEFAULT_CONFIG, errors: [], present: false, label: `head (no ${PREGATE_CONFIG_PATH})`, kind: 'none' };
return { ...parseConfig(fs.readFileSync(file, 'utf8'), knownChecks), present: true, label: `head (unreviewed) ${PREGATE_CONFIG_PATH}`, kind: 'head' };
const policy = path.join(repoRoot, SHIP_POLICY_PATH);
return { ...newerGstackErrors(parseConfig(fs.readFileSync(file, 'utf8'), knownChecks), fs.existsSync(policy) ? fs.readFileSync(policy, 'utf8') : null), present: true, label: `head (unreviewed) ${PREGATE_CONFIG_PATH}`, kind: 'head' };
}
/** A symlink to nothing on the way to `file` (a dotfiles link not yet cloned): a policy the user meant to have, not an absent one. */
@@ -365,7 +388,7 @@ export function explainConfig(): string {
'remote_gate: the repo\'s remote test gate command; under 8 cores the test tier becomes a requires-remote obligation naming it, and gstack-pregate --remote runs it (it must print PREGATE_REMOTE: tree=<tree> tests=<n> failed=<n>) (docs/pregate.md#remote-gate)',
'run_profiles: name → { workflow, events, jobs, artifact, max_age_hours?, os? }; the GitHub Actions runs gstack-pregate clear --run verifies as evidence on the identical tree (docs/pregate.md#run-evidence)',
'remote_gate_runs: { profiles, inventory, skipped? }; the run profiles that together clear the remote gate obligation, the command listing the gate\'s test files they must cover, and an optional command printing which selected files (PREGATE_FILES_FILE) the gate itself does not run, which then need no run (docs/pregate.md#run-evidence)',
'ship_lanes_from: exact /ship lane command → { profiles, inventory }; a verified run of each profile at HEAD\'s clean tree whose tree.json covers the inventory carries the lane instead of running it again (docs/pregate.md#ci-carry)',
'ship_lanes_from: exact /ship lane command → { profiles, inventory } or { profiles, jobs }, with optional aliases (other exact commands for the same lane); a verified run of each profile at HEAD\'s clean tree whose tree.json covers the inventory, or whose newest attempt ran every named job to success, carries the lane instead of running it again (docs/pregate.md#ci-carry)',
`allowlist: [{ finding, paths?, pattern?, reason }]: benign findings the secrets check accepts and names as allowlisted. finding is a gstack-redact id (env.kv, google.api_key, pii.phone.e164, …) or a category (${ALLOWLISTABLE_CATEGORIES.join(', ')}); paths are globs; pattern is a regex matched at the finding's position in the added line; one of the two is required, reason always. Never ${NON_ALLOWLISTABLE.join(' or ')} or engine.* (docs/pregate.md#allowlist)`,
'no .gstack/pregate.json on the base: <state root>/projects/<slug>/pregate.json is read instead (state root: GSTACK_STATE_ROOT, else GSTACK_HOME, else GSTACK_STATE_DIR, else a gstack plugin data dir, else ~/.gstack, per docs/state-root.md; run --explain prints the resolved path; outside the repo, so a PR cannot reach it; a path inside the checkout is ignored)',
].join('\n') + '\n';
+1 -1
View File
@@ -267,7 +267,7 @@ function userPolicyChange(file: PregateFile): string | null {
export interface ClearInput { runUrl?: string; bundleLane?: string; receiptPath?: string; runs?: string[]; evidence?: RunCheck; flaky?: FlakyInput }
/** What a run clear needs: the base branch's run profiles, the verifier's dependencies, the clock, and the remote gate's inventory. */
export interface RunCheck { repo: string; deps: RunDeps; profiles: Record<string, RunProfile>; gateRuns: RemoteGateRuns | null; now: Date; inventory: (command: string, tree: string, selected?: string[]) => string[] | { error: string }; failures?: FailureDeps }
export interface RunCheck { repo: string; deps: RunDeps; profiles: Record<string, RunProfile>; gateRuns: RemoteGateRuns | null; now: Date; unchanged?: (p: RunProfile) => string | null; inventory: (command: string, tree: string, selected?: string[]) => string[] | { error: string }; failures?: FailureDeps }
export type ClearDecision = { receipt: string; attested: boolean; substitution?: PregateReceiptSubstitution; exception?: PregateException } | { refused: 'usage' | 'PREGATE_RECEIPT_INVALID'; message: string; fix?: string };
/**
+45 -6
View File
@@ -11,20 +11,58 @@ import { spawnSync } from 'node:child_process';
import * as fs from 'node:fs';
import * as os from 'node:os';
import * as path from 'node:path';
import { ghArtifactDeps, ghRunDeps, runInventory } from '../run-evidence';
import { ghArtifactDeps, ghRunDeps, runInventory, type RunProfile } from '../run-evidence';
import { wtreeFingerprint } from '../script-invocation';
import { resolveBaseBranch } from '../ship-policy';
import { normalizeRemoteUrl } from '../ship-queue-train-record';
import { REMOTE_GATE_LANE } from './checks/lanes';
import { PREGATE_CONFIG_PATH, isConfigUnavailable, loadConfigFromBase } from './config';
import { PREGATE_CONFIG_PATH, isConfigUnavailable, loadConfigFromBase, loadConfigFromHead } from './config';
import type { RunCheck } from './index';
/**
* `profiled` says whether `lane` may have a run profile: an unreadable base counts as profiled, so a lane is
* recorded unverified only once the base shows it has none.
* Null when a skippable profile's workflow, its declared `scripts` and every repo path its gate job names are byte-identical
* at HEAD and origin/<base> (GST-1190); else the first that differs. Named paths are read from the gate job's block in the base
* workflow (after `${{ … }}` prefixes, and also under each working-directory), plus package.json when the job runs a package
* script. Files the gate reaches indirectly (imports, a package script's own files) count only when `scripts` declares them.
*/
export function buildRunCheck(root: string, lane: string | null, allowRepoCommands: boolean, base = resolveBaseBranch(root)): RunCheck | { error: string; profiled?: boolean } {
const loaded = loadConfigFromBase(root, base);
export function gateInputsChanged(root: string, base: string, p: RunProfile): string | null {
const ref = `origin/${base.replace(/^origin\//, '')}`;
const wf = `.github/workflows/${p.workflow}`;
const gate = p.skippable?.gate;
if (!gate) return 'the profile names no gate job';
const blob = (rev: string, f: string) => { const r = spawnSync('git', ['rev-parse', '--verify', '--quiet', `${rev}:${f}`], { cwd: root, encoding: 'utf8', timeout: 30_000 }); return r.status === 0 ? r.stdout.trim() : null; };
const show = spawnSync('git', ['show', `${ref}:${wf}`], { cwd: root, encoding: 'utf8', timeout: 30_000, maxBuffer: 8 * 1024 * 1024 });
if (show.status !== 0) return `${wf} is not on ${ref}`;
const lines = show.stdout.split('\n');
const jobsAt = lines.findIndex(l => /^jobs:\s*$/.test(l));
const blocks: string[][] = [];
for (let i = jobsAt + 1; jobsAt >= 0 && i < lines.length && !/^\S/.test(lines[i]!); i++) {
if (/^ [\w-]+:\s*$/.test(lines[i]!)) blocks.push([lines[i]!]);
else blocks.at(-1)?.push(lines[i]!);
}
const block = blocks.find(b => b[0]!.trim() === `${gate}:` || b.some(l => /^ name:/.test(l) && l.trim() === `name: ${gate}`));
if (!block) return `${wf} on ${ref} has no job ${gate}`;
const text = block.join('\n');
const dirs = ['', ...[...text.matchAll(/working-directory:\s*['"]?([\w.\/-]+)/g)].map(m => `${m[1]!.replace(/^\.?\/+|\/+$/g, '')}/`)];
const tokens = [...text.matchAll(/[\w.\/-]+/g)].map(m => m[0]).filter(t => t.includes('/') || /\.(sh|ts|js|mjs|cjs|py|rb)$/.test(t));
const named = [...new Set(tokens.flatMap(t => dirs.map(d => path.posix.normalize(d + t.replace(/^\.?\/+/, ''))).filter(n => !n.startsWith('..'))))];
const pkg = /\b(bun|npm|pnpm|yarn)\s+(run\s+)?[\w:-]+/.test(text) ? dirs.map(d => `${d}package.json`) : [];
for (const f of [wf, ...(p.skippable!.scripts ?? []), ...named, ...pkg]) {
const a = blob(ref, f);
const b = blob('HEAD', f);
if (a === null && b === null && f !== wf && !p.skippable!.scripts?.includes(f)) continue;
if (a === null || a !== b) return `${f} differs from ${ref}`;
}
return null;
}
/**
* `profiled` says whether `lane` may have a run profile: an unreadable base counts as profiled, so a lane is
* recorded unverified only once the base shows it has none. `policy: 'head'` reads the branch's profiles instead, only
* for gstack-ship-preflight --simulate, a dry run that records and clears nothing.
*/
export function buildRunCheck(root: string, lane: string | null, allowRepoCommands: boolean, base = resolveBaseBranch(root), policy: 'base' | 'head' = 'base'): RunCheck | { error: string; profiled?: boolean } {
const loaded = policy === 'head' ? loadConfigFromHead(root) : loadConfigFromBase(root, base);
if (isConfigUnavailable(loaded)) return { error: `the base branch's ${PREGATE_CONFIG_PATH} is unreadable (${loaded.detail}); run profiles come only from the base`, profiled: true };
if (loaded.errors.length) return { error: `the base branch's ${PREGATE_CONFIG_PATH} is invalid: ${loaded.errors.map(e => `${e.path}: ${e.message}`).join('; ')}`, profiled: true };
const origin = spawnSync('git', ['config', '--get', 'remote.origin.url'], { cwd: root, encoding: 'utf8', timeout: 30_000 });
@@ -32,6 +70,7 @@ export function buildRunCheck(root: string, lane: string | null, allowRepoComman
if (!repo) return { error: 'origin is not a github.com repository; run evidence is read from GitHub Actions', profiled: lane === null ? false : lane === REMOTE_GATE_LANE ? !!loaded.config.remote_gate_runs : !!loaded.config.run_profiles[lane] };
return {
repo, deps: ghRunDeps(repo, root), failures: ghArtifactDeps(repo), profiles: loaded.config.run_profiles, gateRuns: loaded.config.remote_gate_runs, now: new Date(),
unchanged: p => gateInputsChanged(root, base, p),
inventory: (command, tree, selected) => {
if (!allowRepoCommands) return { error: 'the inventory command is a pin-file command; pass --allow-repo-commands' };
const at = () => { const fp = wtreeFingerprint({ cwd: root }); return 'tree' in fp ? fp.tree : null; };
+5
View File
@@ -16,6 +16,11 @@ export const RELEASE_RESULT_CODES = {
summary: 'the ship policy sets min_gstack_version above this gstack, so this gstack refuses to read it',
fix: 'upgrade gstack (bin/gstack-upgrade, or the session install with --for <skill>) to the version named, then rerun',
},
PREGATE_GSTACK_TOO_OLD: {
anchor: 'pregate-gstack-too-old',
summary: 'the pregate policy uses a key this gstack does not know, and the project floor (min_gstack_version in .gstack/ship-policy.json) is above this gstack, so the key belongs to a newer gstack; the policy stays refused',
fix: 'upgrade gstack (bin/gstack-upgrade, or the session install with --for <skill>) to at least the version named, then rerun',
},
QUEUE_NOT_READY: {
anchor: 'queue-not-ready',
summary: 'the lane is not ready to report: the working tree, HEAD or CI state does not match the pushed PR head',
+46 -11
View File
@@ -42,7 +42,11 @@ import * as path from 'node:path';
import { sha256Hex, writeReceipt } from './egress-receipt';
import { runBashSync } from './script-invocation';
export interface RunProfile { workflow: string; events: string[]; jobs: string[]; artifact: string; max_age_hours?: number; os?: string }
/**
* `skippable` (GST-1190): named jobs the workflow may skip by path; a skip counts only in a pull_request run whose `gate` job
* succeeded, with the workflow, the declared `scripts` and any path the gate's steps name matching origin/<base> byte for byte.
*/
export interface RunProfile { workflow: string; events: string[]; jobs: string[]; artifact: string; max_age_hours?: number; os?: string; skippable?: { jobs: string[]; gate: string; scripts: string[] } }
export interface RemoteGateRuns { profiles: string[]; inventory: string; skipped?: string }
export interface ApiRun {
id: number; run_attempt: number; status: string; conclusion: string | null; event: string; head_sha: string;
@@ -50,7 +54,7 @@ export interface ApiRun {
repository?: { full_name: string } | null; head_repository?: { full_name: string } | null;
}
export interface ApiJob { name: string; conclusion: string | null; completed_at?: string | null }
export interface TreeRecord { run_id?: unknown; run_attempt?: unknown; event?: unknown; sha?: unknown; head_sha?: unknown; tree?: unknown; reused?: unknown; reuse?: unknown; os?: unknown; files?: unknown; tests?: unknown }
export interface TreeRecord { run_id?: unknown; run_attempt?: unknown; event?: unknown; sha?: unknown; head_sha?: unknown; tree?: unknown; reused?: unknown; reuse?: unknown; os?: unknown; files?: unknown; tests?: unknown; skipped_files?: unknown }
/** What a red run's reader needs beyond RunDeps (lib/pregate/flaky-exception.ts): every file of one artifact, and a pull request's head. */
export interface RunArtifactDeps {
@@ -69,14 +73,15 @@ export interface RunDeps {
}
export type RunRefusal = 'run_ref_invalid' | 'other_repository' | 'fork_run' | 'workflow_mismatch' | 'event_mismatch' | 'run_in_progress'
| 'run_not_success' | 'job_missing' | 'job_ambiguous' | 'job_not_success' | 'artifact_missing' | 'artifact_mismatch'
| 'run_not_success' | 'job_missing' | 'job_ambiguous' | 'job_skipped' | 'job_not_success' | 'artifact_missing' | 'artifact_mismatch'
| 'artifact_reused' | 'tree_mismatch' | 'newer_run_vetoes' | 'run_changed' | 'run_stale' | 'os_mismatch' | 'api_error';
export interface RunEvidence { run_id: number; run_attempt: number; head_sha: string; tree: string; profile: string; event: string; finished_at: string; files: string[]; tests: number | null }
/** `skipped`: the files whose every test skipped on the runner (tree.json `skipped_files`, GST-1190); undefined when an older tree.json does not say. */
export interface RunEvidence { run_id: number; run_attempt: number; head_sha: string; tree: string; profile: string; event: string; finished_at: string; files: string[]; tests: number | null; skipped?: string[]; skipped_jobs?: string[] }
export type RunVerdict = { ok: true; evidence: RunEvidence } | { ok: false; reason: RunRefusal; detail: string };
const SHA = /^[0-9a-f]{40}$/;
const RUN_EVENTS = ['push', 'workflow_dispatch', 'pull_request'];
const PROFILE_KEYS = ['workflow', 'events', 'jobs', 'artifact', 'max_age_hours', 'os'];
const PROFILE_KEYS = ['workflow', 'events', 'jobs', 'artifact', 'max_age_hours', 'os', 'skippable'];
const WORKFLOW_FILE = /^[A-Za-z0-9_.-]+\.ya?ml$/;
const NAME = /^[A-Za-z0-9_.\/-]{1,128}$/;
export const DEFAULT_MAX_AGE_HOURS = 24;
@@ -113,6 +118,14 @@ export function runProfileErrors(profiles: unknown, gateRuns: unknown): Array<{
if (typeof p.artifact !== 'string' || !NAME.test(p.artifact)) errors.push({ path: `${at}.artifact`, message: 'must be the artifact name holding tree.json' });
if ('max_age_hours' in p && (!Number.isInteger(p.max_age_hours) || (p.max_age_hours as number) < 1 || (p.max_age_hours as number) > 168)) errors.push({ path: `${at}.max_age_hours`, message: 'must be an integer from 1 to 168' });
if ('os' in p && (typeof p.os !== 'string' || !/^(Linux|macOS|Windows)$/.test(p.os))) errors.push({ path: `${at}.os`, message: 'must be Linux, macOS or Windows' });
if ('skippable' in p) {
const s = p.skippable as Record<string, unknown> | null;
const names = (v: unknown) => Array.isArray(v) && v.length > 0 && v.length <= 32 && v.every(x => typeof x === 'string' && !!x.trim() && x.length <= 128);
const repoPath = (x: unknown) => typeof x === 'string' && !!x && !x.startsWith('/') && !x.split('/').includes('..') && !/[\0\r\n]/.test(x);
if (!s || typeof s !== 'object' || Array.isArray(s) || Object.keys(s).some(k => !['jobs', 'gate', 'scripts'].includes(k)) || !names(s.jobs) || typeof s.gate !== 'string' || !s.gate.trim()
|| (s.jobs as string[]).includes(s.gate as string) || !Array.isArray(s.scripts) || !s.scripts.length || !s.scripts.every(repoPath))
errors.push({ path: `${at}.skippable`, message: 'must be { jobs: [job names], gate: <the job that decides the skip>, scripts: [every repo file the gate reads] }, the gate not among the jobs' });
}
}
}
if (gateRuns !== undefined && gateRuns !== null) {
@@ -135,13 +148,13 @@ export function runProfileErrors(profiles: unknown, gateRuns: unknown): Array<{
const fail = (reason: RunRefusal, detail: string): RunVerdict => ({ ok: false, reason, detail });
/** When a run's newest attempt started. A run vetoes the candidate when it started at or after the second the candidate was created: a red re-run of an older run counts, and re-running the candidate does not hide a failure in between. */
const started = (r: ApiRun): string => r.run_started_at ?? r.created_at ?? '';
const normalPath = (f: string): string => f.replace(/^(\.\/)+/, '');
export const normalPath = (f: string): string => f.replace(/^(\.\/)+/, '');
/**
* The verdict for run `runRef` under `profile`. `vetoHeads` are further heads
* whose newer runs of the same workflow can veto (the consumer's own head).
*/
export function verifyRun(deps: RunDeps, o: { repo: string; runRef: string; profileName: string; profile: RunProfile; vetoHeads?: string[]; now: Date }): RunVerdict {
export function verifyRun(deps: RunDeps, o: { repo: string; runRef: string; profileName: string; profile: RunProfile; vetoHeads?: string[]; now: Date; unchanged?: (p: RunProfile) => string | null }): RunVerdict {
const id = parseRunRef(o.runRef, o.repo);
if (id === null) return fail('run_ref_invalid', `${JSON.stringify(o.runRef)} is not a run id or a run URL on ${o.repo}`);
const p = o.profile;
@@ -162,10 +175,22 @@ export function verifyRun(deps: RunDeps, o: { repo: string; runRef: string; prof
const maxAge = p.max_age_hours ?? DEFAULT_MAX_AGE_HOURS;
if (!Number.isFinite(finished)) return fail('api_error', `${at} has no completion time`);
if (o.now.getTime() - finished > maxAge * 3600_000) return fail('run_stale', `${at} finished ${finishedAt}, more than ${maxAge} h before now`);
const skippedJobs: string[] = [];
for (const name of p.jobs) {
const hits = jobs.filter(j => j.name === name);
if (hits.length === 0) return fail('job_missing', `${at} has no job named ${name}`);
if (hits.length > 1) return fail('job_ambiguous', `${at} has ${hits.length} jobs named ${name}`);
if (hits[0]!.conclusion === 'skipped') {
const skip = p.skippable;
if (!skip?.jobs.includes(name)) return fail('job_skipped', `${at}'s ${name} job was skipped`);
if (run.event !== 'pull_request') return fail('job_skipped', `${at}'s ${name} job was skipped in a ${run.event} run; only a pull_request run's gate diffs against the PR base, so only its skip counts`);
const gate = jobs.filter(j => j.name === skip.gate);
if (gate.length !== 1 || gate[0]!.conclusion !== 'success') return fail('job_skipped', `${at}'s ${name} job was skipped and its gate job ${skip.gate} ${gate.length === 1 ? `concluded ${gate[0]!.conclusion}` : gate.length ? 'is ambiguous' : 'is missing'}`);
const changed = o.unchanged ? o.unchanged(p) : 'no base copy of the workflow was compared';
if (changed) return fail('job_skipped', `${at}'s ${name} job was skipped, but ${changed}, so this branch decided the skip`);
skippedJobs.push(name);
continue;
}
if (hits[0]!.conclusion !== 'success') return fail('job_not_success', `${at}'s ${name} job concluded ${hits[0]!.conclusion}`);
}
const heads = [...new Set([run.head_sha, ...(o.vetoHeads ?? []).filter(h => SHA.test(h))])];
@@ -195,11 +220,13 @@ export function verifyRun(deps: RunDeps, o: { repo: string; runRef: string; prof
if (p.os && !oses.includes(p.os)) return fail('os_mismatch', `${at} ran on ${oses.filter(Boolean).map(String).join(', ') || 'an unrecorded OS'}; ${o.profileName} requires ${p.os}`);
if (rec.files !== undefined && (!Array.isArray(rec.files) || rec.files.some(f => typeof f !== 'string'))) return fail('artifact_mismatch', `${at}'s ${p.artifact} files is not a list of paths`);
if (rec.tests !== undefined && (!Number.isInteger(rec.tests) || (rec.tests as number) < 0)) return fail('artifact_mismatch', `${at}'s ${p.artifact} tests is not a non-negative integer`);
if (rec.skipped_files !== undefined && (!Array.isArray(rec.skipped_files) || rec.skipped_files.some(f => typeof f !== 'string'))) return fail('artifact_mismatch', `${at}'s ${p.artifact} skipped_files is not a list of paths`);
const files = ((rec.files ?? []) as string[]).map(normalPath);
const tests = rec.tests === undefined ? null : (rec.tests as number);
const changed = runChanged(deps, id, run.run_attempt);
if (changed) return fail('run_changed', changed);
return { ok: true, evidence: { run_id: id, run_attempt: run.run_attempt, head_sha: run.head_sha, tree: headTree, profile: o.profileName, event: run.event, finished_at: finishedAt!, files, tests } };
const skipped = rec.skipped_files === undefined ? {} : { skipped: (rec.skipped_files as string[]).map(normalPath) };
return { ok: true, evidence: { run_id: id, run_attempt: run.run_attempt, head_sha: run.head_sha, tree: headTree, profile: o.profileName, event: run.event, finished_at: finishedAt!, files, tests, ...skipped, ...(skippedJobs.length ? { skipped_jobs: skippedJobs } : {}) } };
} catch (e: any) {
return fail('api_error', `reading run ${id} failed: ${String(e?.message ?? e).slice(0, 300)}`);
}
@@ -219,13 +246,21 @@ export function runInventory(command: string, cwd: string, env?: Record<string,
}
/** The files in `inventory` that no verified run's tree.json lists. */
export function uncoveredFiles(inventory: string[], evidence: RunEvidence[]): string[] {
const ran = new Set(evidence.flatMap(e => e.files.map(normalPath)));
/** A run's files minus the ones whose every test skipped there (GST-1190): only those ran anything. Older tree.json: all files. */
export const ranFiles = (e: RunEvidence): string[] => { const skipped = new Set(e.skipped ?? []); return e.files.map(normalPath).filter(f => !skipped.has(f)); };
/**
* The inventory files no run covers. `skipped: 'uncovered'` (a carry, which drops tests on a machine where they might run)
* also counts a file every test of which skipped in the run; `'covered'` (the remote gate, a Linux runner like CI) does not.
*/
export function uncoveredFiles(inventory: string[], evidence: RunEvidence[], skipped: 'uncovered' | 'covered' = 'covered'): string[] {
const ran = new Set(evidence.flatMap(e => (skipped === 'uncovered' ? ranFiles(e) : e.files.map(normalPath))));
return inventory.filter(f => !ran.has(normalPath(f)));
}
/** What a run check reads: the repository, the verifier's dependencies, the base branch's profiles and the clock. */
export interface RunSource { repo: string; deps: RunDeps; profiles: Record<string, RunProfile>; now: Date }
/** `unchanged` (findCarry only): null when a profile's workflow and gate inputs match the base, else which file differs. */
export interface RunSource { repo: string; deps: RunDeps; profiles: Record<string, RunProfile>; now: Date; unchanged?: (p: RunProfile) => string | null }
export type FromRunVerdict = { ok: true; evidence: RunEvidence; receipt: string } | { ok: false; reason: string; detail: string };
/**
+12 -10
View File
@@ -10,6 +10,7 @@ import * as fs from 'node:fs';
import * as os from 'node:os';
import * as path from 'node:path';
import { fragmentSlug } from './changelog-fragments';
import { laneMapping, recordCiResult, type ShipLaneMapping } from './ci-carry';
import { isConfigUnavailable, loadConfigFromBase } from './pregate/config';
import { parseUnifiedDiff, versionAtRef } from './restamp';
import { branchSides, stampOnlyLine } from './stamp-only';
@@ -174,15 +175,15 @@ export function laneConcurrency(mode: string | undefined, cpus = os.availablePar
return cpus <= 4 ? 1 : Infinity;
}
export interface LaneResult { label: string; exit: number | null; testsRan: string; log: string; carried?: string; reused?: string; carryNone?: string }
export interface LaneResult { label: string; exit: number | null; testsRan: string; log: string; carried?: string; reused?: string; carryNone?: string; notOwed?: string; here?: string }
/** The base branch's `ship_lanes_from` entry for each lane whose command is a key, byte for byte (the ledger binds its sha256). */
export function mappedLanes(cwd: string, base: string, list: { label: string; command: string }[]): Map<string, { profiles: string[]; inventory: string }> {
/** The base branch's `ship_lanes_from` entry for each lane whose command is a key or a declared alias, byte for byte (the ledger binds its sha256). */
export function mappedLanes(cwd: string, base: string, list: { label: string; command: string }[]): Map<string, ShipLaneMapping> {
const loaded = loadConfigFromBase(cwd, base);
const map = new Map<string, { profiles: string[]; inventory: string }>();
const map = new Map<string, ShipLaneMapping>();
if (isConfigUnavailable(loaded) || loaded.errors.length) return map;
const from = loaded.config.ship_lanes_from;
for (const l of list) if (Object.prototype.hasOwnProperty.call(from, l.command)) map.set(l.label, from[l.command]!);
for (const l of list) { const m = laneMapping(from, l.command); if (m) map.set(l.label, m); }
return map;
}
@@ -192,9 +193,9 @@ export function mappedLanes(cwd: string, base: string, list: { label: string; co
*/
export function carryLane(cwd: string, base: string, l: { label: string; command: string }): LaneResult {
const rec = bin(cwd, 'gstack-evidence', ['record-ci', '--label', l.label, '--base', base, '--', l.command], 600_000);
const got = / tests_ran=(\d+) log=- source=ci runs=(\S+)/.exec(rec.stderr);
if (rec.status === 0 && got) return { label: l.label, exit: 0, testsRan: got[1]!, log: '-', carried: got[2]! };
return { label: l.label, exit: null, testsRan: 'unknown', log: '-', carryNone: /^CARRY_NONE: (.+)$/m.exec(rec.stderr)?.[1] ?? `record_failed: ${firstLine(rec.stderr)}` };
const got = recordCiResult(rec.stderr);
if (rec.status === 0 && got.runs) return { label: l.label, exit: 0, testsRan: got.files ? `${got.tests} (files)` : got.tests!, log: '-', carried: got.runs, ...(got.notOwed ? { notOwed: got.notOwed } : {}), ...(got.here ? { here: got.here } : {}) };
return { label: l.label, exit: null, testsRan: 'unknown', log: '-', carryNone: got.none ?? `record_failed: ${firstLine(rec.stderr)}` };
}
/**
@@ -204,7 +205,7 @@ export function carryLane(cwd: string, base: string, l: { label: string; command
*/
export async function runLanes(cwd: string, list: { label: string; command: string }[], concurrency = Infinity, opts: { base?: string; reuse?: boolean } = {}): Promise<LaneResult[]> {
const out: LaneResult[] = [];
const mapped = opts.base ? mappedLanes(cwd, opts.base, list) : new Map<string, { profiles: string[]; inventory: string }>();
const mapped = opts.base ? mappedLanes(cwd, opts.base, list) : new Map<string, ShipLaneMapping>();
let next = 0;
const worker = async () => {
while (next < list.length) {
@@ -225,7 +226,8 @@ export async function runLanes(cwd: string, list: { label: string; command: stri
/** Step 5's lane verdict: a lane passes on its runner count, never exit 0 alone; a carried lane names its CI runs. */
export function laneLine(l: LaneResult): { line: string; ok: boolean } {
if (l.carried) return { line: `LANE: ${l.label} carried run=${l.carried} tests_ran=${l.testsRan}`, ok: true };
if (l.notOwed) return { line: `LANE: ${l.label} not owed: ${l.notOwed} run=${l.carried}`, ok: true };
if (l.carried) return { line: `LANE: ${l.label} carried run=${l.carried} tests_ran=${l.testsRan}${l.here ? ` ${l.here}` : ''}`, ok: true };
const zero = l.exit === 0 && l.testsRan === '0';
const state = l.exit !== 0 ? 'FAIL' : zero ? 'ZERO-RUN' : l.testsRan === 'unknown' ? 'pass (count unavailable)' : 'pass';
return { line: `LANE: ${l.label} ${state} exit=${l.exit} tests_ran=${l.testsRan} log=${l.log}${l.reused ? ` reused=${l.reused}` : ''}`, ok: l.exit === 0 && !zero };
+44
View File
@@ -8,8 +8,12 @@
import * as fs from 'node:fs';
import * as os from 'node:os';
import * as path from 'node:path';
import { findCarry, laneMapping, mappingCoverage } from './ci-carry';
import { defaultBaseBranch } from './default-base';
import { isConfigUnavailable, loadConfigFromHead } from './pregate/config';
import { buildRunCheck } from './pregate/run-check';
import { classifyDiff } from './review-depth';
import { wtreeFingerprint } from './script-invocation';
import { audit } from './ship-audit';
import { bin, carryNoneLine, firstLine, flagReader, gh, git, hostedOrigin, laneConcurrency, laneLine, lanes, Report, runLanes, shipPolicy, short } from './ship-bin-common';
@@ -20,6 +24,11 @@ export const PREFLIGHT_HELP = `gstack-ship-preflight — /ship Steps 1-3 and Ste
--lanes serial|parallel how Step 5 runs its lanes; unset, one at a time on 4 or fewer cores, all at once above
--policy-from head Step 5's pregate reads the branch's .gstack/pregate.json, which may only add test lanes
(a branch that adds the lane for a file it touches); anything else is PREGATE_HEAD_POLICY_LOOSENS
gstack-ship-preflight --tests-only --base <base> --policy-from head --simulate [--allow-repo-commands] --lane <label>=<command>...
a dry run that proves a new ship_lanes_from mapping: which lanes the branch's own map would carry from a verified
CI run at this tree. Every line starts SIMULATION:; no lane runs, no evidence or ledger row is written and no
obligation clears (a real carry reads origin/<base>'s map only). An inventory-mapped lane runs the branch's
inventory command, so it needs --allow-repo-commands. Exit 0 when every mapped lane would carry, else 1
gstack-ship-preflight --audit --base <base> --out <dir> the single reads and Steps 7-8's deterministic audits
(CONTEXT, TODOS, COVERAGE_MAP, Coverage, PLAN_*, READER: coverage=needed|none plan=needed|none)
@@ -164,6 +173,40 @@ async function tests(cwd: string, base: string, list: { label: string; command:
r.next('Step 6: select eval suites for prompt-related changes with the project\'s selector, then Step 7');
}
/**
* --simulate (GST-1190): HEAD's ship_lanes_from and run_profiles, findCarry against this tree, nothing else. It never calls
* gstack-evidence or gstack-pregate, so it writes no row and clears no obligation: a PR controls both its own map and the
* CI that would satisfy it, so head policy never carries. The branch's inventory commands run only with `allow`
* (--allow-repo-commands), as gstack-ci-carry find --simulate requires.
*/
function simulate(cwd: string, base: string, list: { label: string; command: string }[], allow: boolean): { text: string; exit: number } {
const lines: string[] = [];
const say = (s: string) => lines.push(`SIMULATION: ${s}`);
const done = (exit: number) => ({ text: lines.join('\n') + '\n', exit });
const loaded = loadConfigFromHead(cwd);
say(`policy ${isConfigUnavailable(loaded) ? 'unavailable' : loaded.label}; a dry run: only the branch's inventory commands run, and only with --allow-repo-commands; no evidence or ledger row is written, nothing clears (a carry reads origin/${base}'s map only)`);
if (isConfigUnavailable(loaded) || loaded.errors.length) { say(`refused: ${isConfigUnavailable(loaded) ? loaded.detail : loaded.errors.map(e => `${e.path}: ${e.message}`).join('; ')}`); return done(1); }
const check = buildRunCheck(cwd, null, allow, base, 'head');
if ('error' in check) { say(`refused: ${check.error}`); return done(1); }
const head = git(cwd, ['rev-parse', 'HEAD']).stdout.trim();
const tree = git(cwd, ['rev-parse', 'HEAD^{tree}']).stdout.trim();
const wtree = () => { const fp = wtreeFingerprint({ cwd }); return 'tree' in fp ? fp.tree : null; };
let mapped = 0;
let carried = 0;
for (const l of list) {
const m = laneMapping(loaded.config.ship_lanes_from, l.command);
if (!m) { say(`LANE: ${l.label} not mapped by the branch's ship_lanes_from; /ship would run it`); continue; }
mapped++;
if (m.inventory && !allow) { say(`CI_CARRY: none label=${l.label} (inventory_not_allowed: the branch's inventory command is a pin-file command; pass --allow-repo-commands)`); continue; }
const v = findCarry(check, { profiles: m.profiles, head, tree, wtree, ...mappingCoverage(m, cwd) });
if (!v.ok) { say(`CI_CARRY: none label=${l.label} (${v.reason}: ${v.detail})`); continue; }
carried++;
say(`LANE: ${l.label} would carry run=${v.runs.map(e => `${e.run_id}/${e.run_attempt}`).join(',')} tests_ran=${v.tests}${v.unit ? ` unit=${v.unit}` : ''}${v.remainder ? ` and would run ${v.remainder.length} skipped file(s) here with per_file` : ''}`);
}
say(`${carried} of ${mapped} mapped lane(s) would carry once origin/${base} holds this map`);
return done(mapped > 0 && carried === mapped ? 0 : 1);
}
export async function preflight(argv: readonly string[], cwd: string): Promise<{ text: string; exit: number }> {
const { flag, has } = flagReader(argv);
const base = flag('--base');
@@ -178,6 +221,7 @@ export async function preflight(argv: readonly string[], cwd: string): Promise<{
if (!['base', 'head'].includes(policyFrom)) return { text: 'gstack-ship-preflight: --policy-from needs base|head\n', exit: 2 };
if (!base || typeof list === 'string') return { text: `gstack-ship-preflight: ${typeof list === 'string' ? list : '--base <base> is required'}\n`, exit: 2 };
if (has('--tests-only') && !list.length) return { text: 'gstack-ship-preflight: --tests-only needs at least one --lane <label>=<command>\n', exit: 2 };
if (has('--simulate')) return has('--tests-only') && policyFrom === 'head' && list.length ? simulate(cwd, base, list, has('--allow-repo-commands')) : { text: 'gstack-ship-preflight: --simulate needs --tests-only --policy-from head and at least one --lane <label>=<command>\n', exit: 2 };
const mode = flag('--lanes');
const concurrency = has('--lanes') && mode === undefined ? '--lanes needs serial or parallel' : laneConcurrency(mode);
if (typeof concurrency === 'string') return { text: `gstack-ship-preflight: ${concurrency}\n`, exit: 2 };