Files
gstack/lib/pregate/context.ts
T
garrytanandcapy-ai[bot] d6f4ea9f12 fix(pregate): test stage never hangs silently; small machines record the remote gate (Lane C, GBRA-95)
C1 diagnosis (gbrain 92f0ea87, a core edit selecting 3,683 local files, 358
serial; Bun 1.4.2 and 1.4.3 identical): --timeout was applied per runner
spawn (unit group plus one per serial file), so the check was unbounded, and
spawnSync buffered all output until it returned: 360 s with 0 bytes on stdout
and stderr. A backgrounded child holding stdout made spawnSync wait the full
timeout and then report exit 0 (a silent wait, then green). Not the Bun 1.4.2
spawnSync hang.

C2/C-b: test-tier commands (lanes runner, guards, the --remote gate) run via
lib/pregate/supervise.ts: own process group (setsid), stdout/stderr appended
to a log file (no pipes), async supervision; CheckDef.run may return a
promise and runPregate/withFingerprint await it (a rejection is incomplete).
Leftover group members are killed after exit; SIGINT/SIGTERM to pregate
SIGKILLs running groups, writes no pregate.json, exits 130/143. Windows:
taskkill /T /F (reparented or detached descendants can survive; documented).

C3/C-a: --timeout bounds the whole check; groups past the deadline are not
started. --stall-timeout <s> (default stall_timeout_s, else 300; 0 disables)
kills a group whose log stops growing and records PREGATE_STALLED naming the
last file started. Live stderr: PREGATE_TEST per file (bun test adapter; any
runner for one-file runs), PREGATE_HEARTBEAT after 60 s without progress,
PREGATE: tests <verdict>. --json stdout unchanged.

C4/C-c: under 8 cores, a base-branch remote_gate or a non-local test_backend
records the local tier as a requires-remote obligation with its command
(pregate.json/remote_gate or test_backend/<backend>). --remote runs the
declared remote_gate now (needs --allow-repo-commands, never --policy-from
head) under --spend-cap/GSTACK_SPEND_CAP_USD; it clears only on
PREGATE_REMOTE: tree=<tested tree> tests>0 failed=0 with exit 0, and reports
spent/reserved/unknown.

G1: new code PREGATE_STALLED has its result-codes row, troubleshooting anchor
and the installer-form error line (CODE; url#anchor).

Files (G2): lib/pregate/supervise.ts (new), lib/pregate/types.ts,
lib/pregate/context.ts, lib/pregate/index.ts, lib/pregate/checks/lanes.ts,
lib/pregate/checks/guards.ts, lib/pregate/config.ts, lib/headless-artifacts.ts,
lib/result-codes.ts, bin/gstack-pregate, bin/gstack-doctor-components.sh
(generated), docs/pregate.md, docs/troubleshooting.md,
test/pregate-supervise.test.ts (new), test/pregate-remote.test.ts (new).

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>
2026-10-15 00:23:34 +00:00

123 lines
8.4 KiB
TypeScript

/**
* pregate/context — what every check reads (plan C8): the base, the touched
* files (committed since the merge-base plus the working tree, renames and
* deletions kept), the reviewed `.gstack/pregate.json` and
* `.gstack/generated.json`, the workflow lanes and the starting fingerprint.
* Fingerprinting follows bin/gstack-evidence: the working-tree content hash
* (bin/gstack-wtree) before a check runs and again after; a tree that moved
* during a check makes its result `incomplete`, never `pass`.
*/
import { spawnSync } from 'node:child_process';
import { createHash } from 'node:crypto';
import * as os from 'node:os';
import * as path from 'node:path';
import { wtreeFingerprint, isRegistryUnavailable, loadRegistryFromBase, loadRegistryFromHead, type Registry } from '../regen';
import { isConfigUnavailable, loadConfigFromBase, loadConfigFromHead, type ConfigError, type PregateConfig } from './config';
import type { CheckResult, PregateContext, TouchedFile } from './types';
import { discoverLanes } from './workflows';
import type { RegistryError } from '../regen';
export interface BuildContextOptions {
repoRoot: string; base: string; policyFrom: 'base' | 'head'; allowRepoCommands: boolean;
timeoutMs?: number; env?: NodeJS.ProcessEnv; platform?: NodeJS.Platform; knownChecks?: readonly string[];
/** `--stall-timeout`; default the config's `stall_timeout_s`, else 300 s. */
stallTimeoutMs?: number; progress?: (line: string) => void; cores?: number; testBackend?: string; remote?: { spendCapUsd: number } | null;
}
export const DEFAULT_STALL_TIMEOUT_S = 300;
const GSTACK_CONFIG = path.resolve(import.meta.dir, '..', '..', 'bin', 'gstack-config');
/** `gstack-config get test_backend` (local | ubicloud); `local` when unreadable. */
export function readTestBackend(env: NodeJS.ProcessEnv): string {
const r = spawnSync(GSTACK_CONFIG, ['get', 'test_backend'], { encoding: 'utf8', timeout: 15_000, env, stdio: ['ignore', 'pipe', 'ignore'] });
return (r.status === 0 && (r.stdout ?? '').trim()) || 'local';
}
export type ContextLoad = { ctx: PregateContext } | { refused: { code: 'POLICY_SOURCE_UNAVAILABLE' | 'PREGATE_CONFIG_INVALID' | 'REGEN_REGISTRY_INVALID'; message: string; errors?: Array<ConfigError | RegistryError> } };
function git(cwd: string, args: string[], env?: NodeJS.ProcessEnv): { ok: boolean; out: string; err: string } {
const r = spawnSync('git', args, { cwd, encoding: 'utf8', timeout: 60_000, maxBuffer: 64 * 1024 * 1024, env });
return { ok: r.status === 0, out: (r.stdout ?? '').replace(/\n$/, ''), err: (r.stderr ?? '').trim() };
}
/** Committed changes since the merge-base (renames detected) plus every working-tree change, untracked files included. */
export function touchedFiles(repoRoot: string, mergeBase: string): TouchedFile[] {
const out = new Map<string, TouchedFile>();
const committed = git(repoRoot, ['diff', '--name-status', '-M', '-z', mergeBase, 'HEAD']).out.split('\0').filter(Boolean);
for (let i = 0; i < committed.length; i++) {
const status = committed[i]![0] as string;
if (status === 'R' || status === 'C') { const from = committed[++i]!; const to = committed[++i]!; out.set(to, { path: to, status: 'R', renamedFrom: from, uncommitted: false }); continue; }
const file = committed[++i]!;
out.set(file, { path: file, status: status === 'D' ? 'D' : status === 'A' ? 'A' : 'M', uncommitted: false });
}
const porcelain = git(repoRoot, ['status', '--porcelain', '-z', '--untracked-files=all']).out.split('\0').filter(Boolean);
for (let i = 0; i < porcelain.length; i++) {
const entry = porcelain[i]!;
const xy = entry.slice(0, 2);
const file = entry.slice(3);
if (xy.includes('R')) { const from = porcelain[++i]!; out.set(file, { path: file, status: 'R', renamedFrom: from, uncommitted: true }); continue; }
if (xy === '!!') continue;
const prior = out.get(file);
const status: TouchedFile['status'] = xy.includes('D') ? 'D' : xy === '??' || xy.includes('A') ? (prior?.status === 'M' ? 'M' : 'A') : (prior?.status ?? 'M');
out.set(file, { path: file, status, renamedFrom: prior?.renamedFrom, uncommitted: true });
}
return [...out.values()].filter(t => t.path !== 'node_modules' && !t.path.startsWith('node_modules/')).sort((a, b) => a.path.localeCompare(b.path));
}
export function buildContext(o: BuildContextOptions): ContextLoad {
const base = o.base.replace(/^origin\//, '');
const baseRef = `origin/${base}`;
const baseSha = git(o.repoRoot, ['rev-parse', '--verify', '--quiet', `${baseRef}^{commit}`]);
if (!baseSha.ok || !baseSha.out) return { refused: { code: 'POLICY_SOURCE_UNAVAILABLE', message: `${baseRef} is not a resolvable commit` } };
const head = git(o.repoRoot, ['rev-parse', 'HEAD']).out;
const mergeBase = git(o.repoRoot, ['merge-base', baseSha.out, 'HEAD']).out || baseSha.out;
const configLoad = o.policyFrom === 'head' ? loadConfigFromHead(o.repoRoot, o.knownChecks) : loadConfigFromBase(o.repoRoot, base, o.knownChecks);
if (isConfigUnavailable(configLoad)) return { refused: { code: 'POLICY_SOURCE_UNAVAILABLE', message: configLoad.detail } };
if (configLoad.errors.length) return { refused: { code: 'PREGATE_CONFIG_INVALID', message: `${configLoad.label}: ${configLoad.errors.map(e => `${e.message} at ${e.path}`).join('; ')}`, errors: configLoad.errors } };
const registryLoad = o.policyFrom === 'head' ? loadRegistryFromHead(o.repoRoot) : loadRegistryFromBase(o.repoRoot, base);
if (isRegistryUnavailable(registryLoad)) return { refused: { code: 'POLICY_SOURCE_UNAVAILABLE', message: registryLoad.detail } };
if (registryLoad.errors.length) return { refused: { code: 'REGEN_REGISTRY_INVALID', message: `${registryLoad.source.label}: ${registryLoad.errors.map(e => `${e.message} at ${e.path}`).join('; ')}`, errors: registryLoad.errors } };
const wtree = wtreeFingerprint(o.repoRoot);
if (!wtree) return { refused: { code: 'POLICY_SOURCE_UNAVAILABLE', message: 'cannot fingerprint the working tree (gstack-wtree failed)' } };
const platform = o.platform ?? process.platform;
const env = o.env ?? process.env;
const config = configLoad.config as PregateConfig;
return {
ctx: {
repoRoot: o.repoRoot, base, baseRef, baseSha: baseSha.out, mergeBase, head, wtree, touched: touchedFiles(o.repoRoot, mergeBase),
config, configLabel: configLoad.label, registry: registryLoad.registry as Registry, registryLabel: registryLoad.source.label,
lanes: discoverLanes(o.repoRoot, platform), platform, allowRepoCommands: o.allowRepoCommands, timeoutMs: o.timeoutMs ?? 600_000,
stallTimeoutMs: o.stallTimeoutMs ?? (config.stall_timeout_s ?? DEFAULT_STALL_TIMEOUT_S) * 1000,
progress: o.progress ?? (line => process.stderr.write(`${line}\n`)),
cores: o.cores ?? os.availableParallelism(), testBackend: o.testBackend ?? readTestBackend(env), policyFrom: o.policyFrom, remote: o.remote ?? null,
env, explain: [],
},
};
}
export function sha256(text: string): string { return createHash('sha256').update(text).digest('hex'); }
/** The common input identity every check records; checks add their own keys. */
export function baseInputs(ctx: PregateContext, extra: Record<string, string> = {}): Record<string, string> {
return { tree: ctx.wtree, head: ctx.head, base: `${ctx.baseRef}@${ctx.baseSha.slice(0, 12)}`, touched: sha256(ctx.touched.map(t => `${t.status} ${t.path}`).join('\n')), ...extra };
}
/**
* Run one check under the before/after fingerprint (bin/gstack-evidence:394):
* a tree that changed while the check ran, a thrown error or rejected promise
* or a timeout is `incomplete`. Wall time is measured here so every row
* carries `ms`.
*/
export async function withFingerprint(ctx: PregateContext, id: string, stage: CheckResult['stage'], fn: () => CheckResult | Promise<CheckResult>): Promise<CheckResult> {
const started = Date.now();
let result: CheckResult;
try { result = await fn(); }
catch (e: any) {
result = { id, stage, status: 'incomplete', detail: `threw: ${e?.message ?? e}`, code: 'PREGATE_INCOMPLETE', fix: 'read the error and rerun the check', inputs: baseInputs(ctx) };
}
const after = wtreeFingerprint(ctx.repoRoot);
if (after !== ctx.wtree) {
result = { ...result, status: 'incomplete', detail: `working tree changed during the check (${ctx.wtree.slice(0, 12)} → ${(after ?? 'unknown').slice(0, 12)}); result not certified`, code: 'PREGATE_INCOMPLETE', fix: 'keep the tree still during the run, then rerun gstack-pregate' };
}
return { ...result, ms: Date.now() - started };
}