문제: SW를 ACTIVE로 켜면 build-frontend 6단계의 SW Vite 빌드가 Vite
기본값대로 public/을 dist에 다시 복사한다. 4단계가 프로필로 만든
dist/config.json이 public/config.json(local, demo 인증,
http://localhost:8080/, 루트 release manifest)으로 덮어써진다. 빌드와
verify-release는 통과하고 release admission만 거절한다. 그래서 SW를 켠
제품은 local이 아닌 릴리스를 만들 수 없다. VITE_ROUTER_BASE_PATH를
쓰면 RELEASE_MANIFEST_URL도 base를 잃는다.
고른 방법: vite.service-worker.config.ts에 publicDir: false를 넣는다.
dist에는 3단계 앱 빌드가 이미 public/을 복사해 두었고, 워커 번들은
public/ 파일을 쓰지 않는다. 같은 repo의 다른 두 번째 Vite 빌드
(scripts/lib/optional-recipe-bundle.ts)도 같은 설정이다.
build-frontend.ts 6단계 주석에 이 조건을 한 줄 적었다.
버린 대안:
- build.copyPublicDir: false. 복사만 끄고 public 디렉터리는 계속
설정에 남는다. 워커 빌드는 public/을 쓸 일이 없어 publicDir: false가
의도를 바로 말한다.
- 앱 빌드 plugin의 closeBundle에서 워커를 만들기, 또는 Vite build()를
configFile: false로 부르는 스크립트. 둘 다 리뷰 5.2절의 대안이다.
two-pass 구조를 바꾸거나 파일이 늘어 이 결함에 비해 범위가 크다.
테스트: tests/unit/service-worker-build-output.test.ts는 임시 복사본에서
SW ACTIVE + APP_PROFILE=production + VITE_ROUTER_BASE_PATH=/app/로 실제
build-frontend.ts를 돌린다. 그리고 dist/config.json이
generateRuntimeConfig("production", env) 결과와 같은지 본다(리뷰의 완료
기준). 설정 객체에서 publicDir === false를 보는 테스트는 수정을 다시
적을 뿐이고, 설정이 5단계 생성물 없이는 로드되지 않아 버렸다. 빌드
1회가 약 2초라 unit 계층에 둔다(build-environment-bundle.test.ts와 같은
방식, docs/testing/taxonomy.md의 짧은 child process 규칙). 복사와 빌드는
e2e 도우미 prepareFixtureBuilds를 재사용하고, 빌드별 env 옵션만
추가했다(기존 호출은 그대로). 이 테스트로 0925/06-10(SW Vite 설정의
dist/config.json 읽기 테스트 없음)도 닫힌다.
위험·한계: 하위 경로에서 사전 캐시 URL이 base를 잃는 문제(0929/06-02)와
워커의 runtime config URL 기본값은 그대로다. 이 테스트는 /app/ 빌드의
설정 문서만 보고 워커 설치는 보지 않는다. Vite 8.1.5 동작 기준이다.
check:architecture는 pnpm 스크립트로만 돌아서 실행하지 않았다(src는
바꾸지 않았다).
검증:
- RED(수정 전): 새 테스트 실패. APP_ENV local(기대 production),
AUTH_MODE demo(기대 external), API_BASE_URL http://localhost:8080/,
RELEASE_MANIFEST_URL /release-manifest.json(기대 /app/...)
- GREEN: 새 테스트 통과(1.9초)
- 수정 전후 production + /app/ 빌드 비교: dist 파일 34개로 같음. 수정 후
RELEASE_TARGET=production check-release-admission ADMITTED, 수정 전 rc 1
- node scripts/run-vitest.ts run tests/unit(pnpm을 띄우는 2파일 제외):
164파일 2343건 통과
- tsc tsconfig.test.json, tsconfig.node.json 통과. 바꾼 4파일 eslint 통과
Refs: docs/reviews/2026-11-01-maturity-review/06-adapters-service-worker-push.md 2절, 13-build-release-supply-chain.md 실험 E-B
The worker build read public/config.json, the local profile, even in a
production build (F-8). It now reads dist/config.json and fails when that
document's APP_ENV is not the build's APP_PROFILE.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A worker that stops the page from booting had no exit: every off switch was
page code. A REMOVE_REGISTRATION release now writes a service-worker.js that
skips waiting and unregisters itself (D-5), checked end to end in the browser.
The runbook and rollback doc now name only paths that exist in code.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
README said runtime settings live in public/config.json, which the build
replaces with config/runtime/<profile>.json, and named none of APP_PROFILE,
RUNTIME_*, VITE_BUILD_ID, RELEASE_ID or RELEASE_TARGET. A unit test now
requires README to name every variable the generator and admission read.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The validated JSON writer creates files 0600, so dist/config.json,
release-manifest.json and runtime-config.schema.json shipped unreadable to a
web server running as another user. The build now adds read bits to dist files;
evidence under artifacts/ stays private.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The provider sandbox never ran. bubblewrap 0.9.0 stops parsing an `--args`
file at the first non-option and never hands the remainder back, so the
command written into that file was silently dropped: bwrap printed its usage
text, exited 1, and the provider produced no evidence at all. The options
still travel in the args file — that is what keeps host paths and credentials
out of `/proc/<pid>/cmdline` — but the command now rides on real argv, and
`encodeProviderBwrapInput` refuses a `--` so the drop cannot come back.
The scope wrapper then could not exit. It read the supervisor's liveness pipe
through `fs`, which runs a blocking `read(2)` on a threadpool thread; the
supervisor holds that pipe open for the scope's whole life, so the read never
returned and closing the descriptor did not interrupt it. Once bubblewrap
finished the wrapper deadlocked in `process.exit`, the scope outlived the
provider, and a completed run was reported as a timeout kill. The channel is
now read through the event loop, so teardown is observable and terminal.
Creation modes were left to the ambient umask. `mkdir(mode)` and `open(mode)`
are requests the kernel subtracts the umask from, so a runner exporting a
restrictive umask produced directories it could not enter and handed `tar` a
file it could not re-open. Private modes are pinned instead of inherited.
Promotion cleanup deleted before it checked. Removals run through a pinned
descriptor, so a leaf substituted after validation had this promotion's exact
five destroyed first and the substitution reported afterwards, leaving a
half-emptied directory a retry could not tell from a completed one. The name
is re-bound to the inode before anything is removed, so the failure is total.
Separately, release coherence proved the artifacts agreed with each other but
never that they belonged where they were going: a build whose runtime document
said `APP_ENV: local`, `AUTH_MODE: demo` and a loopback API is coherent with
itself and passed every gate. `public/` is copied verbatim into `dist/`, so
that local document shipped with every build regardless of what the build was
for. Runtime configuration now comes from a declared profile, and FE-GATE-027
refuses to admit an artifact to an environment it does not match — including
refusing an undeclared destination, so nothing is admitted by omission.
`REQUEST_TIMEOUT_MS` and `VITE_ROUTER_BASE_PATH` were validated and then
dropped: the V3 executor ran every operation on its contract's own deadline,
and Vite emitted root-absolute assets for a sub-path deployment. The timeout is
now a ceiling that may tighten a contract but never loosen one, and one base
path feeds the router, the Service Worker scope and the asset base together.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>