Asset 목록이 `assetKey` 를 앞세우고 있었다. key 는 본문이 가리키는 불변 식별자이지
사람이 기억하는 이름이 아니다 — 스무 줄이면 해시 같은 문자열 스무 개에서 찾던 도식을
눈으로 골라야 했다. 파일 이름을 앞에 두고 형식·치수·크기를 함께 적는다. key 는 아래에
작게 남긴다(복사해 쓰는 값이라 지우지 않는다).
「사용 0건」과 「사용 3건」이 같은 회색으로 나란했다. 쓰이지 않는 자료는 지워도 되는
것이고 쓰이는 자료는 지우면 화면이 깨지므로, 그 차이를 말로 가른다. 상세에서는 어느
기록이 쓰는지를 링크로 연다 — 바꾸거나 지우기 전에 무엇이 깨질지 숫자만으로는 알 수
없다. 관리 상태도 `ARCHIVED` 대신 「보관됨」으로 적는다.
검증 결과는 오류와 경고가 한 목록에 섞여 있었다. 게시 전에 작성자가 알아야 하는 것은
"게시를 막는 것이 몇 개인가" 하나인데, 그 답을 얻으려면 줄마다 배지를 읽어야 했다.
둘을 갈라 각각의 수를 제목에 적는다.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_015c8cBFUaVr7KLVUx6ujz7D
B1. 검증·미리보기 화면이 CSS 없이 배포됐다. `workflow.module.css` 를 import
하는 파일이 하나도 없었고, Vite 의 CSS Module 은 import 되어야 번들에 들어간다.
제목이 62px 가 아니라 16px 로 나왔고 미리보기의 링크 두 개가 공백 없이 붙었다.
규칙이 전부 `:global()` 이었으므로 모듈일 이유가 없다 — studio.css 로 옮겼다.
B2. Asset 라이브러리에 스타일이 없고 진입 경로도 없었다. 네 클래스에 규칙이
하나도 없어 key·상태·사용 수가 공백 없이 이어 붙었다. 스무 개짜리 목록에
썸네일도 없어 무엇을 고르는지 눌러 봐야 알았다. 내비게이션에도 없어서 URL 을
직접 쳐야 닿았다 — "정리하는 화면" 에 정리하러 갈 길이 없었다.
B3. 삭제에 확인이 없었다. 문서·주제·프로젝트·릴리즈·Asset 전부 한 번에 영구
삭제였고, 그 「삭제」가 행에서 시각적으로 가장 강한 요소였다(16px/600 빨강).
정작 그 행에서 해야 할 일인 「검증하기」는 13px/400 이었다. 확인 창을 하나
만들어 다섯 곳에 붙이고, 행의 우선순위를 되돌렸다.
B4(일부). robots.txt 가 광고하던 sitemap 이 404 였다. 무엇이 게시됐는지는
서버만 알므로 정적 파일이 아니라 조회로 만든다(백엔드 커밋 참조). robots.txt
가 그 주소를 가리키게 했다.
B5. 라우트 프리픽스 밖 URL 이 nginx 의 맨 `Not Found` 로 끝났다. 404 상태는
그대로 두고 — 크롤러에게 링크가 깨졌다고 말하는 것이 그것이다 — 본문만 앱을
주도록 `error_page 404 @not_found` 를 뒀다.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01PTnWufiY9ATv57egkKKVAV
`node:assert` builds its `AssertionError` eagerly, running `util.inspect`
over both operands with `depth: 1000`, `getters: true` and
`maxArrayLength: Infinity`. A React-rendered DOM element carries
`__reactFiber$*` / `__reactProps$*` as own enumerable properties, and that
fiber graph re-expands once per traversal path, so inspecting a single
rendered element allocates without bound. Measured on the Asset Library
heading: depth 6 = 1.3MB, 8 = 7.8MB, 10 = 36MB, 12 = 135MB -- at Node's
depth 1000 the worker dies before any `AssertionError` exists.
The damage is not the crash, it is the disguise. Equality assertions only
inspect their operands on failure, so these sites stayed invisible while
green and detonated exactly when the behaviour they guard regressed --
reporting as `worker exited unexpectedly` with a truncated count
(`8 passed (13)`) and no failing test named. Breaking the delete-path focus
restoration in `asset-library.tsx` reproduced it: 29.5GB anon-rss and the
system OOM killer, or a V8 heap abort in 3s under a 512MB cap. The same
regression now fails in 1.15s with `expect(element).toHaveFocus()` naming
both the expected heading and the `<body>` that took focus instead.
`expect` is not affected -- vitest prints and diffs DOM nodes through
pretty-format's DOM plugin, which reads tag/attributes/children and never
touches the fiber -- so every unsafe site converts to a matcher:
`toHaveFocus()` for the three focus comparisons, `not.toBeInTheDocument()`
for the sixteen `assert.equal(queryBy..., null)` absence checks, which are
equally lethal (proved separately: element-vs-null inspects the element).
Three layers so this cannot come back:
- the 20 live sites in asset-library/asset-picker now use matchers;
- `test-assertion-boundary/no-element-operand-equality` fails `pnpm lint`
when a DOM-element expression reaches `node:assert` equality, resolving
local bindings and exempting the forms that cannot fail with an element
in hand (`assert.notEqual(el, null)`, `el.textContent`);
- a 2048MB worker old-space ceiling in `vitest.config.ts` bounds any future
runaway to a legible `Reached heap limit` abort in seconds instead of an
OOM-killed machine (heaviest suite peaks near 1.3GB RSS).
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
`StudioAssetGateway.listAssets` has always accepted `q`/`cursor` and
returned `nextCursor`, and this screen's own heading promises search
("업로드한 Asset을 검색하고..."). The component asked for `{ limit: 50 }`
and dropped `nextCursor`, so past the 51st asset older assets were
unmanageable with nothing on screen admitting anything was left out.
Follows `document-list.tsx`'s established shape: a `searchDraft`/`q`
pair so only a submitted query is an effect dependency (exactly one
request per deliberate search, never a trailing one), and a
`nextCursor`-driven control that appends rather than replaces.
Two things the pattern did not already cover:
- `RequestOptions.signal` is advisory, so aborting is not enough to stop
an abandoned response from painting over a newer one. An `active` flag
that flips synchronously on dependency change closes that race.
- "더 보기" unmounts exactly when the last page arrives, which would
strand focus on `<body>`. Focus moves to the first appended row
unconditionally, falling back to the page heading -- the same anchor
the delete path already uses.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Fix round 1 for the Task 11 asset library review. I1 (Important): a
successful delete removed the row from state, so remove()'s reuse of
closeDetail() queued .focus() on a now-detached button -- a silent
no-op that stranded focus on <body>. Splits closeDetail() (cancel/close,
row still on screen, restores focus to the trigger) from a new
closeDetailAfterRemoval() (post-delete, focuses the page heading, the
one anchor guaranteed to survive any list change) so the two paths
stop sharing a helper that only one of them can safely use.
Also closes four Minors from the same review round:
- route-contract.test.ts's "27-route inventory" test title corrected
to 28.
- canHardDelete's usageCount clause gets its own isolating assertion
(every prior case used usageCount: 0, so that clause was never
independently falsified).
- Dropped a duplicate role="status" announcement on a listAssets
load failure; the existing role="alert" paragraph is now the sole
announcement.
- Added success-path tests for archive() and remove() via a new
recordingGateway() test helper that pins the exact gateway call
shape (expectedVersion, managementStatus: ARCHIVED, idempotency
keys), not just the resulting UI text; the delete-success test also
pins the I1 focus fix so a regression back to the removed trigger
fails loudly.
See task-11-report.md's "Fix round 1" section for the RED/GREEN
evidence (the pre-fix code reliably crashes the test worker rather
than failing the assertion cleanly -- explained there) and the
correction to this task's original claim about matching
publication-list.tsx's focus pattern.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Adds TECH_LOG_STUDIO_ASSETS (/studio/assets) as a route reachable but
excluded from primary Studio navigation (navigationLabel/navigationOrder
null), plus the AssetLibrary screen that lists assets, shows usage, and
lets an operator archive or hard-delete one. canHardDelete() is a pure
gate mirroring the server's ASSET_IN_USE rule so the screen never offers
an action the server would refuse.
Adding a 28th route also required updating the route-scoped CI
accessibility-evidence gate (FE-GATE-009 in config/ci/gates.json, plus
its authority-baseline counts and shape digest in
scripts/contracts/ci-gates.ts) and the Vite route-to-chunk map that
scripts/generate-build-manifest.ts depends on, or test:unit and the
production build both fail. See task-11-report.md for the full
breakdown.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>