perf(tests): trim remaining dominant waits from the slowest test files

Author: Gustav-Simonsson

scripts/ci-timings.json

@@ -15,15 +15,15 @@ function processAlive(pid: number): boolean {
 }
 
 // These probes are `bun -e` one-liners, so they never import project modules
-// and finish in seconds. The tick test keeps a ~0.7s stall window: under the
+// and finish in seconds. The tick test keeps a ~0.5s stall window: under the
 // `--parallel` load this runner exists to enable, a delayed output tick can
 // look like a stall against a tighter window, so the window keeps several
 // ticks of scheduling headroom while a broken never-resetting timer still
 // fires mid-run (the tick test's total runtime exceeds the window). The
 // remaining tests drive children that never emit output, so their stall is
 // declared deterministically on the watchdog's first tick; they use a short
 // window so the suite does not pay the tick test's headroom in wall clock.
-const TEST_STALL_MS = 700;
+const TEST_STALL_MS = 500;
 const SILENT_CHILD_STALL_MS = 250;
 
 describe("runWithWatchdog", () => {
@@ -56,13 +56,13 @@ describe("runWithWatchdog", () => {
   });
 
   test("output resets the stall timer", async () => {
-    // Prints every 120ms for ~1s against a 0.7s stall window: a broken
+    // Prints every 80ms for ~0.7s against a 0.5s stall window: a broken
     // timer that never reset would fire mid-run.
     const result = await runWithWatchdog({
       command: process.execPath,
       args: [
         "-e",
-        "for (let i = 0; i < 8; i++) { console.log('tick', i); await new Promise(r => setTimeout(r, 120)); }",
+        "for (let i = 0; i < 8; i++) { console.log('tick', i); await new Promise(r => setTimeout(r, 80)); }",
       ],
       stallMs: TEST_STALL_MS,
     });
@@ -104,7 +104,7 @@ describe("runWithWatchdog", () => {
       command: process.execPath,
       args: [
         "-e",
-        'process.on("SIGTERM", () => {}); setTimeout(() => process.exit(3), 400);',
+        'process.on("SIGTERM", () => {}); setTimeout(() => process.exit(3), 300);',
       ],
       stallMs: SILENT_CHILD_STALL_MS,
       onStall: () => {

scripts/test-parallel.test.ts

@@ -44,7 +44,7 @@ describe("background shell through the agent toolset", () => {
         {
           id: "bg-start",
           name: "run_shell",
-          arguments: { command: "sleep 0.4; echo bg-done", background: true },
+          arguments: { command: "sleep 0.2; echo bg-done", background: true },
         },
         new AbortController().signal,
       );

src/agent/background-shell-tool.test.ts

@@ -57,7 +57,7 @@ describe("runGuardedShell", () => {
     const feed = createShellOutputFeed();
     let finished = false;
     const running = runGuardedShell(
-      { command: "echo first; sleep 0.02; echo second; sleep 0.2" },
+      { command: "echo first; sleep 0.02; echo second; sleep 0.15" },
       neverAbort(),
       undefined,
       undefined,
@@ -85,7 +85,7 @@ describe("runGuardedShell", () => {
   test("omitted timeout does not arm a timer", async () => {
     const start = Date.now();
     const { exitCode, timedOut, output } = await runGuardedShell(
-      { command: "sleep 0.1; echo done" },
+      { command: "sleep 0.05; echo done" },
       neverAbort(),
     );
     expect(timedOut).toBe(false);
@@ -180,7 +180,7 @@ describe("runGuardedShell", () => {
     if (process.platform === "win32") return;
     const token = `ic_guard_orphan_${randomUUID()}`;
     const cmd = `bash -c 'IC_GUARD_TAG=${token} sleep 600 & IC_GUARD_TAG=${token} exec sleep 600'`;
-    await runGuardedShell({ command: cmd, timeout: 250 }, neverAbort());
+    await runGuardedShell({ command: cmd, timeout: 150 }, neverAbort());
     await waitUntilGone(token);
   });
 
@@ -479,12 +479,12 @@ describe("background run_shell (shellGuardPlugin)", () => {
         id: "fg3",
         name: "run_shell",
         arguments: {
-          // Fifteen lines ~10 ms apart: far more chunk arrivals than one
+          // Fifteen lines ~5 ms apart: far more chunk arrivals than one
           // cadence window per 100 ms can allow. Without the Date.now() gate
           // in emitPendingOutput every arrival emits (~16 emissions) and this
           // ceiling fails — the assertion is what pins the cadence.
           command:
-            "i=1; while [ $i -le 15 ]; do echo line$i; sleep 0.01; i=$((i+1)); done",
+            "i=1; while [ $i -le 15 ]; do echo line$i; sleep 0.005; i=$((i+1)); done",
         },
       },
       neverAbort(),
@@ -1153,7 +1153,7 @@ describe("shellGuardPlugin", () => {
 
       await first;
       await Promise.race([queued, new Promise((r) => setTimeout(r, 250))]);
-      await new Promise((r) => setTimeout(r, 120));
+      await new Promise((r) => setTimeout(r, 80));
 
       const leftover1 =
         spawnSync("pgrep", ["-f", token1], {

src/plugins/shell-guard-plugin.test.ts

@@ -25,7 +25,7 @@ describe("background shell registry", () => {
   test("start returns a handle immediately while the process runs", async () => {
     const registry = createBackgroundShellRegistry();
     const started = registry.start({
-      command: "sleep 0.2; echo done",
+      command: "sleep 0.1; echo done",
       cwd: tmpCwd,
     });
     if ("error" in started) throw new Error(started.error);

src/shell/background-shell.test.ts

@@ -267,9 +267,9 @@ describe("mountProductHost", () => {
 
   test("setChrome with running agents paints an agents panel clock", async () => {
     const now = Date.now();
-    // Start 500ms before the minute boundary: the rollover assertion stays
+    // Start 300ms before the minute boundary: the rollover assertion stays
     // identical while the boundary wait (up to a full minute from a 59:00
-    // start) shrinks to at most ~0.5s of wall clock, with plenty of margin
+    // start) shrinks to at most ~0.3s of wall clock, with plenty of margin
     // left for the mount + first capture to still see 0:59.
     const { host, renderOnce, captureCharFrame } = await mountHeadless({
       chrome: {
@@ -279,7 +279,7 @@ describe("mountProductHost", () => {
             currentToolStartedAt: null,
             description: "map callers",
             status: "running",
-            startedAt: now - 59_500,
+            startedAt: now - 59_700,
             lastActivityAt: now,
           },
         ],
@@ -1017,7 +1017,7 @@ describe("flat type-to-filter model picker", () => {
       await harness.renderOnce();
       expect(host.shell.overlayKind).toBe("add_provider");
       harness.pressKey("Escape");
-      await new Promise((r) => setTimeout(r, 60));
+      await new Promise((r) => setTimeout(r, 30));
       await harness.renderOnce();
       expect(host.shell.overlayKind).toBe("model_picker");
       expect(host.shell.overlayItems).toEqual(modelItems);
@@ -1040,7 +1040,7 @@ describe("flat type-to-filter model picker", () => {
       await harness.renderOnce();
       expect(host.shell.overlayKind).toBe("add_provider");
       harness.pressKey("Escape");
-      await new Promise((r) => setTimeout(r, 60));
+      await new Promise((r) => setTimeout(r, 30));
       await harness.renderOnce();
       expect(host.shell.overlayKind).not.toBe("model_picker");
       expect(host.shell.overlayKind).toBeNull();
@@ -1078,7 +1078,7 @@ describe("flat type-to-filter model picker", () => {
       await harness.renderOnce();
       expect(host.shell.overlayKind).toBe("add_provider");
       harness.pressKey("Escape");
-      await new Promise((r) => setTimeout(r, 60));
+      await new Promise((r) => setTimeout(r, 30));
       await harness.renderOnce();
       expect(host.shell.overlayKind).toBeNull();
     } finally {

src/tui/product-host.test.ts

@@ -64,7 +64,7 @@ describe("createGateRequestApproval", () => {
         });
         const pending = requestApproval(request);
         // Longer than the budget — frozen while the prompt is open.
-        await new Promise((r) => setTimeout(r, 120));
+        await new Promise((r) => setTimeout(r, 90));
         expect(getToolApprovalBudget()?.signal.aborted).toBe(false);
         captured?.resolve({ allow: true });
         const outcome = await pending;
@@ -179,7 +179,7 @@ describe("attachApprovalBudget", () => {
         );
         // Longer than the budget — frozen while the question is pending,
         // exactly like the permission gate's budget pause.
-        await new Promise((r) => setTimeout(r, 120));
+        await new Promise((r) => setTimeout(r, 90));
         expect(getToolApprovalBudget()?.signal.aborted).toBe(false);
         finish("answered");
         finish("answered again");

src/tui/request-approval.test.ts

@@ -9,13 +9,18 @@ process.on("SIGINT", () => {
   process.exit(0);
 });
 process.stdin.resume();
+// READY tells the driver the raw-mode handlers are installed, so it can send
+// the keypress without a fixed boot sleep.
+process.stdout.write("READY\n");
 process.stdin.on("data", (chunk: Buffer) => {
-  if (chunk.includes(0x03)) process.stdout.write("GOT_CTRL_C_BYTE\n");
+  if (!chunk.includes(0x03)) return;
+  process.stdout.write("GOT_CTRL_C_BYTE\n");
+  // A SIGINT, if the kernel delivered one, would arrive synchronously with
+  // the keypress, so a short grace counted from the byte still proves none is
+  // delivered without paying a fixed 3s (or an 800ms boot-independent wait)
+  // in wall clock.
+  setTimeout(() => {
+    process.stdout.write("NO_SIGINT_ON_CTRL_C\n");
+    process.exit(0);
+  }, 300);
 });
-// A SIGINT, if the kernel delivered one, would arrive synchronously with the
-// keypress, so a short grace after GOT_CTRL_C_BYTE still proves none is
-// delivered without paying a fixed 3s in wall clock.
-setTimeout(() => {
-  process.stdout.write("NO_SIGINT_ON_CTRL_C\n");
-  process.exit(0);
-}, 800);

tests/fixtures/rawmode-sigint/probe.ts

@@ -16,11 +16,9 @@ def main() -> int:
         os.execvp("bun", ["bun", "run", probe_path])
         os._exit(127)
 
-    time.sleep(0.4)
-    os.write(fd, b"\x03")
-
     out = b""
     deadline = time.time() + 2.5
+    sent_ctrl_c = False
     while time.time() < deadline:
         ready, _, _ = select.select([fd], [], [], 0.5)
         if fd not in ready:
@@ -32,6 +30,11 @@ def main() -> int:
         if not chunk:
             break
         out += chunk
+        # Send the keypress only once raw mode is installed (no fixed boot
+        # sleep); the probe's grace window then counts from the keypress.
+        if not sent_ctrl_c and b"READY" in out:
+            os.write(fd, b"\x03")
+            sent_ctrl_c = True
         if b"NO_SIGINT_ON_CTRL_C" in out or b"GOT_SIGINT" in out:
             break
 

tests/fixtures/rawmode-sigint/pty_probe.py

@@ -172,7 +172,7 @@ test("opting out discards events captured before the toggle instead of sending t
   expect(sends).toBe(0);
 
   handler(false);
-  await new Promise((resolve) => setTimeout(resolve, 150));
+  await new Promise((resolve) => setTimeout(resolve, 60));
   expect(getInstance().enabled).toBe(false);
   expect(sends).toBe(0);
 });