From d02ac377a5c8a4d3bc76489168df5ab9bfab8353 Mon Sep 17 00:00:00 2001
From: Neil <neil@stably.ai>
Date: Thu, 10 Sep 2026 06:46:01 -0700
Subject: [PATCH] fix(runtime): let a failed create retry inside its own
 subscription

The returned-failure release freed the module latch but the caller still
latched its closure-local requestedInitialTerminal whenever the dispatch
reported it owned the create, including a create that returned
{ status: 'failed' }. A thrown failure never set it and retried on the
next frame; since every RPC and network failure is reported as a return,
the live path was the one that suppressed the whole subscription. The
dispatch now reports false for a failure whichever way it arrives.
---
 ...ime-initial-terminal-bootstrap-dispatch.ts | 17 +++++++------
 ...bs-sync-initial-terminal-relaunch.test.tsx | 25 +++++++++++++++++++
 2 files changed, 35 insertions(+), 7 deletions(-)

diff --git a/src/renderer/src/runtime/web-runtime-initial-terminal-bootstrap-dispatch.ts b/src/renderer/src/runtime/web-runtime-initial-terminal-bootstrap-dispatch.ts
index a435f9bcb87..d4f097a2a68 100644
--- a/src/renderer/src/runtime/web-runtime-initial-terminal-bootstrap-dispatch.ts
+++ b/src/renderer/src/runtime/web-runtime-initial-terminal-bootstrap-dispatch.ts
@@ -18,8 +18,10 @@ import {
  * is parked as awaiting-mirror and released by the next frame the mirror accepts for the worktree
  * (see web-runtime-initial-terminal-bootstrap.ts).
  *
- * Returns true when this call owned the create, so the caller can keep its closure-local flag in
- * step; false when another closure already held the latch.
+ * Returns true only when this call owned a create that was not reported as failed, because the
+ * caller latches a closure-local flag on it. A failure must report false whichever way it arrives:
+ * releasing the shared latch alone still leaves the subscription that issued the failed create
+ * unable to retry for as long as its closure lives, which is the same suppression one level up.
  */
 export async function dispatchWebRuntimeInitialTerminalBootstrap(
   environmentId: string,
@@ -36,11 +38,12 @@ export async function dispatchWebRuntimeInitialTerminalBootstrap(
     throw error
   }
   // Why check the outcome: the create reports RPC and network failures as `{ status: 'failed' }`
-  // rather than throwing, so the catch above never sees them.
-  if (
-    outcome.status === 'failed' ||
-    Object.hasOwn(useAppStore.getState().tabsByWorktree, worktreeId)
-  ) {
+  // rather than throwing, so the catch above never sees them. Both arms report the same way.
+  if (outcome.status === 'failed') {
+    endWebRuntimeInitialTerminalBootstrap(environmentId, worktreeId)
+    return false
+  }
+  if (Object.hasOwn(useAppStore.getState().tabsByWorktree, worktreeId)) {
     endWebRuntimeInitialTerminalBootstrap(environmentId, worktreeId)
   } else {
     markWebRuntimeInitialTerminalBootstrapAwaitingMirror(environmentId, worktreeId)
diff --git a/src/renderer/src/runtime/web-session-tabs-sync-initial-terminal-relaunch.test.tsx b/src/renderer/src/runtime/web-session-tabs-sync-initial-terminal-relaunch.test.tsx
index 52f10c11856..fb7808e472d 100644
--- a/src/renderer/src/runtime/web-session-tabs-sync-initial-terminal-relaunch.test.tsx
+++ b/src/renderer/src/runtime/web-session-tabs-sync-initial-terminal-relaunch.test.tsx
@@ -255,6 +255,31 @@ describe('useWebSessionTabsSync initial-terminal bootstrap across an effect re-r
     hook.unmount()
   })
 
+  // Third suppression path (review): the returned-failure release above frees the *module* latch,
+  // but the caller sets its closure-local `requestedInitialTerminal` whenever the dispatch reports
+  // it owned the create — including when that create returned `{ status: 'failed' }`. A thrown
+  // failure never sets it and retries on the very next frame; since the create reports every RPC
+  // and network failure as a return rather than a throw, the live path was the one that suppressed
+  // the whole subscription. A failed create must leave the closure free to retry, exactly like a
+  // thrown one.
+  it('retries on the next frame of the same subscription after a create that returned failed', async () => {
+    mocks.createTerminal.mockResolvedValue({ status: 'failed', message: 'host unreachable' })
+
+    const hook = renderHook(() => useWebSessionTabsSync())
+    await act(settle)
+
+    const subscription = findActiveSubscription(0)
+    await publish(subscription, { type: 'snapshot', ...emptyActiveSnapshot(1) })
+    expect(mocks.createTerminal).toHaveBeenCalledTimes(1)
+    expect(useAppStore.getState().tabsByWorktree[WORKTREE]).toBeUndefined()
+
+    // Same closure, next accepted frame: nothing was created, so the workspace is still
+    // never-initialized and the retry is the only thing that will ever give it a terminal.
+    await publish(subscription, { type: 'snapshot', ...emptyActiveSnapshot(2) })
+    expect(mocks.createTerminal).toHaveBeenCalledTimes(2)
+    hook.unmount()
+  })
+
   // Readiness review: the inverse hazard of the test above. A create that succeeds but whose frame
   // never lands (host accepted, the mirror never got a row) must not hold the latch until environment
   // teardown. The frame right after the settle is the mirror's answer and may not seed (pre-mirror