Diagnose macOS Intel E2E flake (Electron quitting mid-test)

.github/workflows/deploy_tests_on_pull_request.yml

@@ -9,32 +9,9 @@ concurrency:
 
 jobs:
 
-  DevTests:
-    uses: ./.github/workflows/testing_dev.yml
-    secrets:
-      CODECOV_TOKEN: ${{ secrets.CODECOV_TOKEN }}
-
-  LiveServices:
-    uses: ./.github/workflows/testing_dev_with_live_services.yml
-    secrets:
-      CODECOV_TOKEN: ${{ secrets.CODECOV_TOKEN }}
-
+  # NOTE: diagnostic branch only. Trimmed to just the flaky E2E job to iterate quickly. Do not merge.
   E2ELiveServices:
     uses: ./.github/workflows/testing_dev_e2e_with_live_services.yml
     secrets:
       CODECOV_TOKEN: ${{ secrets.CODECOV_TOKEN }}
       DANDI_SANDBOX_API_KEY: ${{ secrets.DANDI_SANDBOX_API_KEY }}
-
-  BuildTests:
-    uses: ./.github/workflows/testing_flask_build_and_dist.yml
-
-  ExampleDataCache:
-    uses: ./.github/workflows/example_data_cache.yml
-    secrets:
-      AWS_ACCESS_KEY_ID: ${{ secrets.AWS_ACCESS_KEY_ID }}
-      AWS_SECRET_ACCESS_KEY: ${{ secrets.AWS_SECRET_ACCESS_KEY }}
-      S3_GIN_BUCKET: ${{ secrets.S3_GIN_BUCKET }}
-
-  ExampleDataTests:
-    needs: ExampleDataCache
-    uses: ./.github/workflows/testing_pipelines.yml

.github/workflows/testing_dev_e2e_with_live_services.yml

@@ -22,18 +22,10 @@ jobs:
       fail-fast: false
       matrix:
         include:
-          - os: ubuntu-latest
-            label: environments/environment-Linux.yml
-
-          - os: macos-latest  # Mac arm64 runner
-            label: environments/environment-MAC-apple-silicon.yml
-
+          # NOTE: diagnostic branch only. Reduced to the flaky x64 runner. Do not merge.
           - os: macos-15-intel  # Mac x64 runner
             label: environments/environment-MAC-intel.yml
 
-          - os: windows-latest
-            label: environments/environment-Windows.yml
-
     steps:
       - uses: actions/checkout@v6
 

src/electron/main/main.ts

@@ -36,6 +36,27 @@ if (runByTestSuite) {
   if (process.env.CI) app.commandLine.appendSwitch('disable-gpu') // Avoid GPU process crashes on CI runners (e.g. macOS Intel)
 }
 
+// TODO(e2e-flake): temporary diagnostics to pin why Electron quits mid-test on the macOS Intel runner. Remove once identified.
+if (runByTestSuite) {
+  // Log which signal (if any) arrives, then re-raise it so the default behavior is unchanged (no masking).
+  ;['SIGINT', 'SIGTERM', 'SIGHUP'].forEach((sig) =>
+    process.once(sig as NodeJS.Signals, () => {
+      console.log(`[quit-diagnostic] process received ${sig}`)
+      process.kill(process.pid, sig as NodeJS.Signals)
+    })
+  )
+  process.on('exit', (code) => console.log(`[quit-diagnostic] main process exit, code ${code}`))
+  app.on('render-process-gone', (_e, _wc, details) =>
+    console.log(`[quit-diagnostic] app render-process-gone: ${JSON.stringify(details)}`)
+  )
+  app.on('child-process-gone', (_e, details) =>
+    console.log(`[quit-diagnostic] app child-process-gone: ${JSON.stringify(details)}`)
+  )
+  app.on('will-quit', () => console.log('[quit-diagnostic] app will-quit'))
+  app.on('quit', () => console.log('[quit-diagnostic] app quit'))
+  app.on('window-all-closed', () => console.log('[quit-diagnostic] app window-all-closed'))
+}
+
 /*************************************************************
  * Python Process
  *************************************************************/
@@ -414,7 +435,18 @@ app.on('activate', () => {
 
 const root = runByTestSuite ? path.join(paths.root, '.test') : paths.root
 
-if (runByTestSuite) onWindowReady(() => console.log('WINDOW READY FOR TESTING'))
+if (runByTestSuite) onWindowReady((win) => {
+  console.log('WINDOW READY FOR TESTING')
+  // TODO(e2e-flake): temporary window/renderer lifecycle diagnostics. Remove once identified.
+  const wc = win.webContents
+  wc.on('render-process-gone', (_e, details) => console.log(`[quit-diagnostic] webContents render-process-gone: ${JSON.stringify(details)}`))
+  wc.on('unresponsive', () => console.log('[quit-diagnostic] webContents unresponsive'))
+  wc.on('responsive', () => console.log('[quit-diagnostic] webContents responsive'))
+  wc.on('destroyed', () => console.log('[quit-diagnostic] webContents destroyed'))
+  wc.on('did-fail-load', (_e, code, desc) => console.log(`[quit-diagnostic] webContents did-fail-load: ${code} ${desc}`))
+  win.on('close', () => console.log('[quit-diagnostic] window close'))
+  win.on('closed', () => console.log('[quit-diagnostic] window closed'))
+})
 
 
 const homeDirectory = app.getPath("home");
@@ -450,6 +482,8 @@ app.on("window-all-closed", () => {
 
 app.on("before-quit", async (ev: Event) => {
 
+  if (runByTestSuite) console.log(`[quit-diagnostic] before-quit fired\n${new Error().stack}`) // TODO(e2e-flake): remove once identified
+
   ev.preventDefault()
     if (!runByTestSuite) {
     const { response } = await dialog