Fix pwsh availability cold-start cache (#6313)

src/main/daemon/daemon-entry.ts

@@ -8,6 +8,7 @@
  */
 import { startDaemon, type DaemonHandle } from './daemon-main'
 import { createPtySubprocess } from './pty-subprocess'
+import { warmPwshAvailabilityCache } from '../pwsh'
 
 export function parseArgs(argv: string[]): { socketPath: string; tokenPath: string } {
   let socketPath = ''
@@ -32,6 +33,7 @@ export function parseArgs(argv: string[]): { socketPath: string; tokenPath: stri
 
 async function main(): Promise<void> {
   const { socketPath, tokenPath } = parseArgs(process.argv.slice(2))
+  void warmPwshAvailabilityCache()
 
   // Why: node-pty can throw a C++ Napi::Error that escapes all JS try/catch
   // blocks (e.g. writing to a PTY whose fd was closed between the native

src/main/pwsh.test.ts

@@ -1,10 +1,12 @@
 import { beforeEach, describe, expect, it, vi } from 'vitest'
 
-const { execFileSyncMock } = vi.hoisted(() => ({
+const { execFileMock, execFileSyncMock } = vi.hoisted(() => ({
+  execFileMock: vi.fn(),
   execFileSyncMock: vi.fn()
 }))
 
 vi.mock('child_process', () => ({
+  execFile: execFileMock,
   execFileSync: execFileSyncMock
 }))
 
@@ -26,6 +28,8 @@ function setPlatform(platform: NodeJS.Platform): () => void {
 describe('isPwshAvailable', () => {
   beforeEach(() => {
     vi.resetModules()
+    vi.useRealTimers()
+    execFileMock.mockReset()
     execFileSyncMock.mockReset()
   })
 
@@ -84,4 +88,70 @@ describe('isPwshAvailable', () => {
       restorePlatform()
     }
   })
+
+  it('repro: does not keep a cold-start timeout cached for the daemon lifetime', async () => {
+    const restorePlatform = setPlatform('win32')
+    execFileSyncMock
+      .mockImplementationOnce(() => {
+        const error = Object.assign(new Error('spawnSync pwsh.exe ETIMEDOUT'), {
+          code: 'ETIMEDOUT'
+        })
+        throw error
+      })
+      .mockReturnValue('PowerShell 7.5.0')
+
+    try {
+      const { isPwshAvailable } = await import('./pwsh')
+      expect(isPwshAvailable()).toBe(false)
+      expect(isPwshAvailable()).toBe(true)
+      expect(execFileSyncMock).toHaveBeenCalledTimes(2)
+    } finally {
+      restorePlatform()
+    }
+  })
+
+  it('warms pwsh availability asynchronously with a longer timeout', async () => {
+    const restorePlatform = setPlatform('win32')
+    execFileMock.mockImplementation((_file, _args, _options, callback) => {
+      callback(null, 'PowerShell 7.5.0', '')
+    })
+
+    try {
+      const { isPwshAvailable, warmPwshAvailabilityCache } = await import('./pwsh')
+      await expect(warmPwshAvailabilityCache()).resolves.toBe(true)
+      expect(execFileMock).toHaveBeenCalledWith(
+        'pwsh.exe',
+        ['-Version'],
+        { timeout: 30_000 },
+        expect.any(Function)
+      )
+      expect(isPwshAvailable()).toBe(true)
+      expect(execFileSyncMock).not.toHaveBeenCalled()
+    } finally {
+      restorePlatform()
+    }
+  })
+
+  it('retries non-timeout failures after the negative cache TTL', async () => {
+    vi.useFakeTimers()
+    vi.setSystemTime(1_000)
+    const restorePlatform = setPlatform('win32')
+    execFileSyncMock
+      .mockImplementationOnce(() => {
+        throw new Error('missing pwsh')
+      })
+      .mockReturnValue('PowerShell 7.5.0')
+
+    try {
+      const { isPwshAvailable } = await import('./pwsh')
+      expect(isPwshAvailable()).toBe(false)
+      expect(isPwshAvailable()).toBe(false)
+      vi.setSystemTime(31_001)
+      expect(isPwshAvailable()).toBe(true)
+      expect(execFileSyncMock).toHaveBeenCalledTimes(2)
+    } finally {
+      restorePlatform()
+      vi.useRealTimers()
+    }
+  })
 })

src/main/pwsh.ts

@@ -1,31 +1,92 @@
-import { execFileSync } from 'child_process'
+import { execFile, execFileSync } from 'child_process'
 
-// Cached pwsh availability check — evaluated once per process lifetime
-let pwshAvailableCache: boolean | null = null
+const PWSH_SYNC_PROBE_TIMEOUT_MS = 5000
+const PWSH_WARMUP_PROBE_TIMEOUT_MS = 30_000
+const PWSH_NEGATIVE_CACHE_TTL_MS = 30_000
+
+type PwshAvailabilityCache =
+  | { available: true }
+  | { available: false; cachedAt: number; retryable: boolean }
+
+let pwshAvailableCache: PwshAvailabilityCache | null = null
+let pwshWarmupInFlight: Promise<boolean> | null = null
+
+function isCacheFresh(cache: PwshAvailabilityCache): boolean {
+  return (
+    cache.available || !cache.retryable || Date.now() - cache.cachedAt < PWSH_NEGATIVE_CACHE_TTL_MS
+  )
+}
+
+function isTimeoutError(error: unknown): boolean {
+  return (
+    typeof error === 'object' &&
+    error !== null &&
+    'code' in error &&
+    (error as NodeJS.ErrnoException).code === 'ETIMEDOUT'
+  )
+}
+
+function cachePwshProbeFailure(error: unknown): void {
+  // Why: pwsh.exe cold starts can exceed the sync timeout; do not let one slow
+  // .NET startup disable the user's PowerShell 7 preference for the daemon.
+  if (isTimeoutError(error)) {
+    pwshAvailableCache = null
+    return
+  }
+  pwshAvailableCache = { available: false, cachedAt: Date.now(), retryable: true }
+}
 
 /**
  * Check whether pwsh.exe is available on this Windows machine.
- * Result is cached for the process lifetime.
+ * Positive results are cached for the process lifetime; negative results are
+ * retried so transient cold-start failures cannot outlive the daemon.
  */
 export function isPwshAvailable(): boolean {
-  if (pwshAvailableCache !== null) {
-    return pwshAvailableCache
+  if (pwshAvailableCache && isCacheFresh(pwshAvailableCache)) {
+    return pwshAvailableCache.available
   }
 
   if (process.platform !== 'win32') {
-    pwshAvailableCache = false
+    pwshAvailableCache = { available: false, cachedAt: Date.now(), retryable: false }
     return false
   }
 
   try {
     execFileSync('pwsh.exe', ['-Version'], {
       stdio: ['pipe', 'pipe', 'pipe'],
-      timeout: 5000
+      timeout: PWSH_SYNC_PROBE_TIMEOUT_MS
     })
-    pwshAvailableCache = true
-  } catch {
-    pwshAvailableCache = false
+    pwshAvailableCache = { available: true }
+  } catch (error) {
+    cachePwshProbeFailure(error)
   }
 
-  return pwshAvailableCache
+  return pwshAvailableCache?.available ?? false
+}
+
+export function warmPwshAvailabilityCache(): Promise<boolean> {
+  if (pwshAvailableCache?.available) {
+    return Promise.resolve(true)
+  }
+  if (process.platform !== 'win32') {
+    pwshAvailableCache = { available: false, cachedAt: Date.now(), retryable: false }
+    return Promise.resolve(false)
+  }
+  if (pwshWarmupInFlight) {
+    return pwshWarmupInFlight
+  }
+
+  pwshWarmupInFlight = new Promise((resolve) => {
+    execFile('pwsh.exe', ['-Version'], { timeout: PWSH_WARMUP_PROBE_TIMEOUT_MS }, (error) => {
+      pwshWarmupInFlight = null
+      if (!error) {
+        pwshAvailableCache = { available: true }
+        resolve(true)
+        return
+      }
+      cachePwshProbeFailure(error)
+      resolve(false)
+    })
+  })
+  return pwshWarmupInFlight
 }