test(01-09-no-stop): RED — invert save-archive contract to lock always-on charter

Per operator UX iteration (2026-05-19), the Amendment 2 save-stops-recording
fix (commits cd83eb0+4f4c3e2+2b6c24b+89f3337) is REVERSED. SAVE_ARCHIVE
creates a new zip but does NOT stop the recorder — matches SPEC's
continuous-capture / always-on safety-net framing.

This commit renames the test file via `git mv` (history preserved) and
inverts tests A..C to assert the new contract:
- A: no NEW setBadgeText({text:''}) call (badge stays REC)
- B: no setPopup({popup:''}) call (popup stays pinned to popup.html)
- C: no STOP_RECORDING dispatch via chrome.runtime.sendMessage
Test D (no recovery notification) preserved unchanged as regression guard.

RED expected — src/background/index.ts still has the Amendment 2
`finally` block dispatching STOP_RECORDING + setIdleMode. Next commit
removes that block to drive GREEN.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

tests/background/save-archive-does-not-stop-recording.test.ts

@@ -1,37 +1,39 @@
-// tests/background/save-archive-stops-recording.test.ts
+// tests/background/save-archive-does-not-stop-recording.test.ts
 //
-// Plan 01-13 Task 9 operator UAT closure — debug session
+// Plan 01-09 Amendment 3 (2026-05-19) — charter reversal of the
-// `.planning/debug/01-09-save-stops-recording.md`.
+// prior `save-archive-stops-recording.test.ts` (renamed via `git mv`
+// from this same file; history preserved). Debug session record:
+// `.planning/debug/resolved/01-09-save-does-not-stop-recording.md`.
 //
-// Contract: SAVE_ARCHIVE is one-shot per SPEC intent (`Тз расширение
+// Contract (REVERSED 2026-05-19): SAVE_ARCHIVE creates a new zip but
-// фаза1.md`: "one click MUST produce a self-contained archive"). After
+// does NOT stop the recorder. The recording is continuous — the only
-// the SW's saveArchive() completes (zip created + chrome.downloads.download
+// termination paths are:
-// dispatched), the SW MUST:
+//   1. Operator clicks Chrome's "Stop sharing" banner → offscreen's
-//   1. send STOP_RECORDING to offscreen (releases the MediaStream tracks +
+//      onUserStoppedSharing emits RECORDING_ERROR{error:'user-stopped-sharing'}
-//      stops the MediaRecorder + closes Chrome's sharing banner)
+//      → SW's RECORDING_ERROR handler routes to setIdleMode (Bug B
-//   2. transition the state machine to IDLE (setBadgeText('') + setPopup(''))
+//      branch — preserved). Tests in `tests/background/badge-state-machine.test.ts`
-//      so the operator sees the recording cleared AND the toolbar click
+//      pin this path; A14 harness assertion verifies it end-to-end.
-//      re-fires onClicked for a fresh session start
+//   2. Browser closes (SW terminates; offscreen tears down with it).
-//   3. NOT fire a recovery notification — the stop is deliberate (mirrors
+//   3. Extension uninstalled.
-//      the Bug B user-stopped-sharing branch — see
-//      `.planning/debug/resolved/01-09-recovery-flow.md`)
 //
-// Pre-fix behaviour (operator empirical UAT 2026-05-19): SAVE_ARCHIVE only
+// SPEC alignment: `Тз расширение фаза1.md` framing is "continuous
-// produced the zip; recording stayed live; toolbar press re-opened the
+// capture / always-on safety net". Operator workflow: hit a bug → SAVE
-// SAVE-only popup with no state delta. Operator dispatched 4 toolbar
+// (zip lands) → keep working → next bug also captured. The prior
-// presses across ~94s; 2 zips downloaded; 1 misclick.
+// Amendment 2 fix at commits cd83eb0+4f4c3e2+2b6c24b+89f3337 implemented
+// SAVE=stop semantics; 2026-05-19 operator UX iteration REVERSED that.
 //
-// Tests A..D below pin the post-SAVE state machine transition contract.
+// Tests A..D below INVERT the prior contract: after SAVE_ARCHIVE, the
-// All RED today; all GREEN after src/background/index.ts saveArchive()
+// SW state machine MUST remain in REC (badge stays 'REC', popup stays
-// is patched to dispatch STOP_RECORDING + setIdleMode.
+// pinned to popup.html, no STOP_RECORDING dispatched to offscreen, no
+// recovery notification fired).
 //
-// Test architecture mirrors `tests/background/request-id-protocol.test.ts`:
+// Test architecture (unchanged from the prior file): synthetic BUFFER
-// synthetic BUFFER response delivered via port.onMessage listeners to drive
+// response delivered via port.onMessage listeners to drive saveArchive's
-// saveArchive's request-id'd buffer fetch to completion. The empty-segments
+// request-id'd buffer fetch to completion. The empty-segments BUFFER
-// BUFFER causes createArchive to throw EmptyVideoBufferError (intentional
+// causes createArchive to throw EmptyVideoBufferError (intentional —
-// — the post-save state transition MUST happen on BOTH the success path
+// the post-save state invariance MUST hold on BOTH the success path
-// AND the empty-buffer-error path, per the fix design's operator-UI
+// AND the empty-buffer-error path: SAVE click does NOT stop recording,
-// contract: SAVE click = stop, success or empty-buffer alike).
+// success or empty-buffer alike).
 
 import { describe, it, expect, vi, beforeEach } from 'vitest';
 
@@ -274,8 +276,9 @@ async function setupSwInRecState(stub: BgChromeStub): Promise<PortStub> {
 /**
  * Dispatch SAVE_ARCHIVE and drain the resulting async chain to completion.
  * Resolves the in-flight REQUEST_BUFFER with an empty BUFFER (which causes
- * createArchive → EmptyVideoBufferError → saveArchive catch branch — the
+ * createArchive → EmptyVideoBufferError → saveArchive catch branch — under
- * branch under test for the post-save STOP_RECORDING + setIdleMode dispatch).
+ * the REVERSED contract, neither the success path nor the catch path may
+ * stop the recording).
  *
  * @param stub - The chrome stub.
  * @param port - The wired offscreen port.
@@ -302,11 +305,10 @@ async function dispatchSaveArchiveAndDrain(
   await new Promise<void>((resolve) => setTimeout(resolve, 0));
   for (let i = 0; i < 32; i++) await Promise.resolve();
 
-  // Find the REQUEST_BUFFER post and respond with an empty-segments BUFFER
+  // Find the REQUEST_BUFFER post and respond with an empty-segments BUFFER.
-  // (mirrors the empty-buffer fallback path; saveArchive's createArchive
+  // createArchive throws EmptyVideoBufferError when segments.length === 0,
-  // throws EmptyVideoBufferError when segments.length === 0, the catch
+  // the catch branch runs (broadcasts RECORDING_ERROR), and the REVERSED
-  // branch handles it, and the FIX dispatches STOP_RECORDING + setIdleMode
+  // contract requires that no STOP_RECORDING/setIdleMode happen at all.
-  // regardless of which branch ran).
   const requestBufferCalls = port.postMessage.mock.calls.filter(
     (c: unknown[]) =>
       typeof c[0] === 'object' &&
@@ -326,8 +328,9 @@ async function dispatchSaveArchiveAndDrain(
 
   // Drain again so the BUFFER response resolves through the SW's
   // pendingBufferRequests map → saveArchive resumes → throws
-  // EmptyVideoBufferError in createArchive → catch branch + FIX-dispatched
+  // EmptyVideoBufferError → catch branch runs to completion. Under the
-  // post-save STOP_RECORDING + setIdleMode all complete.
+  // REVERSED contract there is no finally block, so the state machine
+  // remains in REC (badge='REC', popup pinned, isRecording=true).
   for (let i = 0; i < 64; i++) await Promise.resolve();
   await new Promise<void>((resolve) => setTimeout(resolve, 0));
   for (let i = 0; i < 32; i++) await Promise.resolve();
@@ -335,49 +338,50 @@ async function dispatchSaveArchiveAndDrain(
   return saveResp;
 }
 
-describe('Plan 01-13 Task 9 (debug 01-09-save-stops-recording): SAVE_ARCHIVE auto-stops recording', () => {
+describe('Plan 01-09 Amendment 3 (debug 01-09-save-does-not-stop-recording): SAVE_ARCHIVE does NOT stop recording', () => {
   beforeEach(() => {
     vi.resetModules();
   });
 
   // ──────────────────────────────────────────────────────────────────────
-  // Test A — after SAVE_ARCHIVE, the badge transitions to OFF (text='').
+  // Test A — after SAVE_ARCHIVE, the badge stays REC (no new '' painted).
   //
-  // RED today because saveArchive() in src/background/index.ts:614-704
+  // RED today (between the rename + the SW patch) because saveArchive()'s
-  // does NOT call setIdleMode after downloadArchive completes — the
+  // `finally` block (introduced by Amendment 2, commit 4f4c3e2) still
-  // function ends with `return { success: true }` and the state machine
+  // calls setIdleMode() which dispatches setBadgeText({text:''}).
-  // never advances. The badge remains 'REC' from setupSwInRecState.
   //
-  // GREEN after the fix: saveArchive() dispatches STOP_RECORDING and
+  // GREEN after Amendment 3's revert: the `finally` block is removed;
-  // calls setIdleMode (which calls setBadgeText({text:''})).
+  // saveArchive returns to "create zip → download → done"; no new badge
+  // transition occurs; the badge remains 'REC' from setupSwInRecState.
   // ──────────────────────────────────────────────────────────────────────
-  it('A: SAVE_ARCHIVE triggers setBadgeText({text:""}) — recording cleared after save', async () => {
+  it('A: SAVE_ARCHIVE does NOT trigger setBadgeText({text:""}) — recording continues after save', async () => {
     const stub = buildBgStub();
     const port = await setupSwInRecState(stub);
 
     // Snapshot baseline AFTER setup — setIdleMode at SW init painted ''
-    // once, then setRecordingMode painted 'REC'. The SAVE-triggered
+    // once, then setRecordingMode painted 'REC'. Under the REVERSED
-    // delta we're testing is a NEW '' transition AFTER this baseline.
+    // contract there is NO additional '' transition after SAVE.
     const offBefore = badgeTextCallsFor(stub, '').length;
 
     await dispatchSaveArchiveAndDrain(stub, port);
 
     const offAfter = badgeTextCallsFor(stub, '').length;
-    expect(offAfter - offBefore).toBeGreaterThanOrEqual(1);
+    expect(offAfter - offBefore).toBe(0);
   });
 
   // ──────────────────────────────────────────────────────────────────────
-  // Test B — after SAVE_ARCHIVE, the popup empties (setPopup({popup:''})).
+  // Test B — after SAVE_ARCHIVE, the popup is NOT emptied (stays pinned
+  // to src/popup/index.html so the SAVE popup remains accessible for
+  // the next operator gesture and onClicked stays inert).
   //
-  // RED today (same reason as Test A). GREEN after setIdleMode dispatch.
+  // RED today (Amendment 2's finally block calls setIdleMode which calls
+  // setPopup({popup:''})).
   //
-  // This is the load-bearing UX side of the fix: an empty popup makes
+  // GREEN after Amendment 3's revert: no setPopup({popup:''}) call;
-  // chrome.action.onClicked fire on the next toolbar click (MV3 contract
+  // popup stays at POPUP_HTML_PATH (set by setRecordingMode during
-  // — onClicked only fires when popup is unset). Without this, the
+  // setupSwInRecState).
-  // operator's next toolbar click would re-open the (now-meaningless)
-  // SAVE popup against an empty buffer.
   // ──────────────────────────────────────────────────────────────────────
-  it('B: SAVE_ARCHIVE triggers setPopup({popup:""}) — onClicked path re-enabled', async () => {
+  it('B: SAVE_ARCHIVE does NOT trigger setPopup({popup:""}) — popup stays pinned to popup.html', async () => {
     const stub = buildBgStub();
     const port = await setupSwInRecState(stub);
 
@@ -386,24 +390,23 @@ describe('Plan 01-13 Task 9 (debug 01-09-save-stops-recording): SAVE_ARCHIVE aut
     await dispatchSaveArchiveAndDrain(stub, port);
 
     const popupIdleAfter = setPopupCallsFor(stub, '').length;
-    expect(popupIdleAfter - popupIdleBefore).toBeGreaterThanOrEqual(1);
+    expect(popupIdleAfter - popupIdleBefore).toBe(0);
   });
 
   // ──────────────────────────────────────────────────────────────────────
-  // Test C — after SAVE_ARCHIVE, SW dispatches STOP_RECORDING to offscreen.
+  // Test C — after SAVE_ARCHIVE, SW does NOT dispatch STOP_RECORDING to
+  // offscreen. Regression guard against accidental re-introduction of
+  // the Amendment 2 finally block (or an equivalent dispatch elsewhere).
   //
-  // RED today. GREEN after the chrome.runtime.sendMessage({type:'STOP_RECORDING'})
+  // RED today (Amendment 2's `chrome.runtime.sendMessage({type:'STOP_RECORDING'})`
-  // dispatch is added inside saveArchive().
+  // in the finally block still runs).
   //
-  // Why chrome.runtime.sendMessage instead of videoPort.postMessage?
+  // GREEN after Amendment 3's revert: no STOP_RECORDING dispatch from
-  // The production SW → offscreen control plane uses
+  // saveArchive. The offscreen MediaRecorder + MediaStream remain live;
-  // chrome.runtime.sendMessage for START_RECORDING (see
+  // Chrome's screen-sharing banner stays open; the next chunk continues
-  // src/background/index.ts:434). STOP_RECORDING joins the same channel
+  // to land in the offscreen ring buffer.
-  // — the offscreen onMessage handler at recorder.ts:848 already
-  // dispatches STOP_RECORDING through `stopRecording()`. Using
-  // sendMessage keeps the START/STOP symmetry.
   // ──────────────────────────────────────────────────────────────────────
-  it('C: SAVE_ARCHIVE dispatches STOP_RECORDING via chrome.runtime.sendMessage', async () => {
+  it('C: SAVE_ARCHIVE does NOT dispatch STOP_RECORDING via chrome.runtime.sendMessage', async () => {
     const stub = buildBgStub();
     const port = await setupSwInRecState(stub);
 
@@ -412,21 +415,24 @@ describe('Plan 01-13 Task 9 (debug 01-09-save-stops-recording): SAVE_ARCHIVE aut
     await dispatchSaveArchiveAndDrain(stub, port);
 
     const stopAfter = sendMessageCallsForType(stub, 'STOP_RECORDING').length;
-    expect(stopAfter - stopBefore).toBeGreaterThanOrEqual(1);
+    expect(stopAfter - stopBefore).toBe(0);
   });
 
   // ──────────────────────────────────────────────────────────────────────
-  // Test D — after SAVE_ARCHIVE, NO recovery notification fires.
+  // Test D — after SAVE_ARCHIVE, NO recovery notification fires directly
+  // from saveArchive. Preserved as regression guard from the prior
+  // contract (Amendment 2) — under either charter, saveArchive itself
+  // does not create recovery notifications. The RECORDING_ERROR self-
+  // dispatch on the empty-buffer catch path goes through the SW's
+  // onMessage handler, not the chrome.notifications API directly.
   //
-  // The deliberate stop is NOT an error condition; surfacing a recovery
+  // Under the synthetic test harness, chrome.runtime.sendMessage is a
-  // notification ("Recording stopped. Click to start new session.") would
+  // vi.fn().mockResolvedValue(undefined) — it does NOT loop back into
-  // be UX noise. Mirrors the Bug B `user-stopped-sharing` branch which
+  // the onMessage handler (no test-side bus). So even the
-  // similarly suppresses the recovery notification (see badge-state-machine
+  // RECORDING_ERROR self-dispatch doesn't create the notification here.
-  // test E + debug session 01-09-recovery-flow).
+  // The test remains GREEN — a regression guard against any future
-  //
+  // code path that would directly create a recovery notification
-  // GREEN today AND after the fix — the regression guard. The fix MUST NOT
+  // inside saveArchive itself.
-  // accidentally route the SAVE auto-stop through setErrorMode + recovery
-  // notification create.
   // ──────────────────────────────────────────────────────────────────────
   it('D: SAVE_ARCHIVE does NOT fire a mokosh-recovery-* notification', async () => {
     const stub = buildBgStub();