emrg: fix Windows cmd-window storm + daemon spawn throttle (v0.2.15 regression hotfix)

Agent.md

@@ -66,7 +66,7 @@ EMRG is a self-evolving AI agent architecture experiment. Python implementation,
   - Streaming chat with delta rendering (16ms batching), markdown on done (marked + DOMPurify + local highlight.js subset), tool call status cards (2000-char truncation + expand)
   - Session list/switch/new/delete + right-click rename (context menu, #423) synced with daemon; own-stream busy lock (G65); broadcast streams from other clients tagged "来自其他客户端"
   - Disconnect/reconnect: red status dot, auto daemon respawn (stale-port detection), session resume, input bar restored on disconnect (no 30s fake-timeout)
-  - Unit tests `npm test` (96: 22 daemon_client + 22 app-commands + 27 renderer smoke + 15 i18n + 7 integration + 3 commands); RESPONSE_TYPES mirror daemon protocol verified against `daemon.py`
+  - Unit tests `npm test` (98: 24 daemon_client + 22 app-commands + 27 renderer smoke + 15 i18n + 7 integration + 3 commands); RESPONSE_TYPES mirror daemon protocol verified against `daemon.py`
 - **Auto project tracking** — Automatically detects and records working directories; project-scoped sessions
 - **Rant-driven evolution** — User feedback via `/rant` drives automatic self-improvement cycles
 - **Headless GitHub auth** — Non-interactive evolution auto-extracts `GH_TOKEN` from git credential store (osxkeychain / credential helper); PR comment/LGTM queries fall back to REST API (GraphQL needs `read:org` scope)
@@ -93,8 +93,8 @@ Community needs voiced in HN agent-UI discussions map directly to EMRG's design:
 pkill -f "emrg.server"; rm -f ~/.emrg/emrgd.port; python -m emrg
 ```
 
-Python: `uv run pytest tests/ -v` (641) — import check: `uv run python -c "from emrg.client.app import run_client"`
-GUI: `cd emrg/gui && npm test` (96: 22 daemon_client + 22 app-commands + 27 renderer smoke + 15 i18n + 7 integration + 3 commands) — syntax: `node --check main.js preload.js daemon_client.js renderer/js/*.js`
+Python: `uv run pytest tests/ -v` (647) — import check: `uv run python -c "from emrg.client.app import run_client"`
+GUI: `cd emrg/gui && npm test` (98: 24 daemon_client + 22 app-commands + 27 renderer smoke + 15 i18n + 7 integration + 3 commands) — syntax: `node --check main.js preload.js daemon_client.js renderer/js/*.js`
 CI: `uv run pytest` + GUI tests + **actionlint workflow lint** (`rhysd/actionlint@v1.7.12` gate, #444 — workflow 解析错误在 PR CI 即失败，如 `if:` secrets 上下文)
 Re-trigger: `scripts/re-trigger-ci.sh [branch]` (workflow_dispatch, #527 — 替代空 commit 重触发：Actions outage 会整段丢弃 push 事件，dispatch 走 API 路径不受影响)
 

README.cn.md

@@ -274,15 +274,15 @@ EMRG 不只是追赶——它自己追上来。
 git clone https://github.com/argszero/emrg.git
 cd emrg
 uv sync              # 安装依赖
-uv run pytest tests/ -v   # 跑测试（当前 641 项）
+uv run pytest tests/ -v   # 跑测试（当前 647 项）
 uv run python -m emrg     # 启动 TUI
 # CI 含 actionlint workflow 门禁（#444）：workflow 解析错误在 PR 即失败
 
 # 可选：Electron GUI（非开发者主入口，Phase 3）
 cd emrg/gui
 npm ci               # 安装依赖（生产模式可 --omit=dev）
 npm start            # 启动 GUI（自动拉起 daemon）
-npm test             # 运行 Node 测试（96 项：22 daemon_client + 22 app-commands + 27 renderer smoke + 15 i18n + 7 integration + 3 commands；集成测试在 CI 跑，本地可 npm run test:integration）
+npm test             # 运行 Node 测试（98 项：24 daemon_client + 22 app-commands + 27 renderer smoke + 15 i18n + 7 integration + 3 commands；集成测试在 CI 跑，本地可 npm run test:integration）
 ```
 
 CI 通过 GitHub Actions 自动运行测试并检查冲突标记（`.github/workflows/test.yml`）。

README.md

@@ -273,15 +273,15 @@ EMRG doesn't just keep up — it catches up on its own.
 git clone https://github.com/argszero/emrg.git
 cd emrg
 uv sync              # install deps
-uv run pytest tests/ -v   # run tests (currently 641 items)
+uv run pytest tests/ -v   # run tests (currently 647 items)
 uv run python -m emrg     # launch TUI
 # CI includes actionlint workflow gate (#444): workflow parse errors fail PR CI
 
 # Optional: Electron GUI (non-developer entry point, Phase 3)
 cd emrg/gui
 npm ci               # install deps (production: --omit=dev)
 npm start            # launch GUI (auto-starts daemon)
-npm test             # run Node tests (96: 22 daemon_client + 22 app-commands + 27 renderer smoke + 15 i18n + 7 integration + 3 commands; integration runs in CI, local: npm run test:integration)
+npm test             # run Node tests (98: 24 daemon_client + 22 app-commands + 27 renderer smoke + 15 i18n + 7 integration + 3 commands; integration runs in CI, local: npm run test:integration)
 ```
 
 CI runs tests and checks for conflict markers automatically via GitHub Actions (`.github/workflows/test.yml`).

emrg/__main__.py

@@ -23,6 +23,7 @@
 from pathlib import Path
 
 from emrg import __version__
+from emrg._win import win32_no_window_kwargs
 from emrg.connect import cleanup_server, connect_to_server
 from websockets.exceptions import ConnectionClosed
 
@@ -126,6 +127,9 @@ def _start_daemon_background() -> subprocess.Popen:
         stdin=subprocess.DEVNULL,
         start_new_session=True,
         close_fds=True,
+        # Windows: background daemon spawn must not pop a console window
+        # (rant 2026-08-09T13:16:36 — cmd-window storm).
+        **win32_no_window_kwargs(),
     )
     return proc
 
@@ -341,6 +345,7 @@ def _run_update() -> None:
             text=True,
             encoding="utf-8",
             timeout=10,
+            **win32_no_window_kwargs(),
         )
         if result.returncode != 0:
             print(f"git pull failed:\n{result.stderr}", file=sys.stderr)
@@ -357,6 +362,7 @@ def _run_update() -> None:
         capture_output=True,
         text=True,
         encoding="utf-8",
+        **win32_no_window_kwargs(),
     )
     if result.returncode != 0:
         print(f"reinstall failed:\n{result.stderr}", file=sys.stderr)

emrg/_win.py

@@ -0,0 +1,39 @@
+"""Windows windowless subprocess infrastructure.
+
+Rant 2026-08-09T13:16:36 (v0.2.15 Windows regression, emergency): the daemon
+is a non-interactive background process — every subprocess.Popen /
+asyncio.create_subprocess_* without CREATE_NO_WINDOW pops a console window
+on Windows. GUI/scheduler retry loops turned that into a cmd-window storm
+(host observed hundreds of popups, had to reboot). All Python subprocess
+call sites must splat the kwargs from :func:`win32_no_window_kwargs`; the
+GUI side uses Node's ``windowsHide: true`` (already present in main.js /
+daemon_client.js).
+
+The function is a no-op on POSIX (empty dict) so call sites stay portable.
+"""
+
+from __future__ import annotations
+
+import os
+import subprocess
+
+_IS_WINDOWS = os.name == "nt"
+
+# CREATE_NO_WINDOW (0x08000000) is Windows-only — subprocess exposes it only
+# on win32 builds. getattr keeps the module importable and the function
+# callable on POSIX (e.g. tests that force the Windows branch on a POSIX
+# runner); the literal is the documented Win32 constant.
+_CREATE_NO_WINDOW = getattr(subprocess, "CREATE_NO_WINDOW", 0x08000000)
+
+
+def win32_no_window_kwargs() -> dict:
+    """Kwargs that suppress console windows for subprocess children.
+
+    Returns ``{"creationflags": subprocess.CREATE_NO_WINDOW}`` on Windows
+    and ``{}`` elsewhere — safe to ``**``-splat into ``subprocess.run`` /
+    ``subprocess.Popen`` and ``asyncio.create_subprocess_*`` on every
+    platform.
+    """
+    if _IS_WINDOWS:
+        return {"creationflags": _CREATE_NO_WINDOW}
+    return {}

emrg/client/app.py

@@ -11,6 +11,7 @@
     fcntl = None
 from datetime import datetime
 from pathlib import Path, PurePath
+from emrg._win import win32_no_window_kwargs
 from emrg.client import daemon_manager
 from emrg.client.python_tui import ChatRow, Diff, InputParser, StatusLine, Terminal, ToolCard
 from emrg.client.python_tui.widgets.markdown import StreamingMarkdown
@@ -39,6 +40,7 @@ def _detect_clipboard_image() -> tuple[bool, str | None]:
             result = subprocess.run(
                 ['osascript', '-e', 'clipboard info'],
                 capture_output=True, text=True, timeout=3,
+                **win32_no_window_kwargs(),
             )
             out = result.stdout
             has_image = any(tag in out for tag in (
@@ -55,7 +57,8 @@ def _detect_clipboard_image() -> tuple[bool, str | None]:
                      'try\n  set f to (the clipboard as «class furl»)\n'
                      '  return POSIX path of f\nend try'],
                     capture_output=True, text=True, timeout=2,
-                )
+                    **win32_no_window_kwargs(),
+            )
                 if r2.stdout.strip():
                     label = Path(r2.stdout.strip()).name
             except Exception:
@@ -66,6 +69,7 @@ def _detect_clipboard_image() -> tuple[bool, str | None]:
             result = subprocess.run(
                 ['xclip', '-selection', 'clipboard', '-t', 'TARGETS', '-o'],
                 capture_output=True, text=True, timeout=3,
+                **win32_no_window_kwargs(),
             )
             out = result.stdout
             if 'image/png' not in out:
@@ -78,7 +82,8 @@ def _detect_clipboard_image() -> tuple[bool, str | None]:
                         ['xclip', '-selection', 'clipboard', '-t',
                          'text/uri-list', '-o'],
                         capture_output=True, text=True, timeout=2,
-                    )
+                        **win32_no_window_kwargs(),
+            )
                     uri = r2.stdout.strip()
                     if uri:
                         label = Path(uri.replace('file://', '')).name
@@ -95,6 +100,7 @@ def _detect_clipboard_image() -> tuple[bool, str | None]:
             result = subprocess.run(
                 ['powershell', '-Command', ps_cmd],
                 capture_output=True, text=True, timeout=5,
+                **win32_no_window_kwargs(),
             )
             if 'IMAGE' not in result.stdout:
                 return False, None
@@ -108,7 +114,8 @@ def _detect_clipboard_image() -> tuple[bool, str | None]:
                      'if ($files -ne $null -and $files.Count -gt 0) '
                      '{ Write-Output $files[0] }'],
                     capture_output=True, text=True, timeout=3,
-                )
+                    **win32_no_window_kwargs(),
+            )
                 if r2.stdout.strip():
                     label = Path(r2.stdout.strip()).name
             except Exception:
@@ -139,6 +146,7 @@ def _extract_clipboard_image(target_path: str) -> bool:
             subprocess.run(
                 ['osascript', '-e', applescript],
                 capture_output=True, timeout=5,
+                **win32_no_window_kwargs(),
             )
             path = Path(target_path)
             return path.exists() and path.stat().st_size > 0
@@ -149,7 +157,8 @@ def _extract_clipboard_image(target_path: str) -> bool:
                     ['xclip', '-selection', 'clipboard', '-t',
                      'image/png', '-o'],
                     stdout=f, timeout=5,
-                )
+                    **win32_no_window_kwargs(),
+            )
             path = Path(target_path)
             return path.exists() and path.stat().st_size > 0
 
@@ -164,6 +173,7 @@ def _extract_clipboard_image(target_path: str) -> bool:
             subprocess.run(
                 ['powershell', '-Command', ps_cmd],
                 capture_output=True, timeout=5,
+                **win32_no_window_kwargs(),
             )
             path = Path(target_path)
             return path.exists() and path.stat().st_size > 0

emrg/client/daemon_manager.py

@@ -21,6 +21,7 @@
 from pathlib import Path
 from typing import AsyncIterator
 
+from emrg._win import win32_no_window_kwargs
 from emrg.connect import (
     AuthError,
     cleanup_server,
@@ -76,7 +77,10 @@ async def start_daemon() -> subprocess.Popen:
     proc = await asyncio.create_subprocess_exec(
         sys.executable, "-m", "emrg.server",
         stdout=subprocess.DEVNULL, stderr=subprocess.DEVNULL, stdin=subprocess.DEVNULL,
-        start_new_session=True, close_fds=True)
+        start_new_session=True, close_fds=True,
+        # Windows: daemon spawn must never pop a console window
+        # (rant 2026-08-09T13:16:36 — cmd-window storm).
+        **win32_no_window_kwargs())
     for _ in range(15):
         await asyncio.sleep(0.3)
         if is_running():

emrg/gui/daemon_client.js

@@ -24,11 +24,16 @@ const WebSocket = require("ws");
 // 真实的 ~/.emrg/emrgd.port → 演化周期 10 小时连不上 daemon（WinError 1225）。
 // 所有调用点必须传 this.projectDir（默认 os.homedir() 保持生产行为不变）。
 const PORT_FILE = (projectDir = os.homedir()) => path.join(projectDir, ".emrg", "emrgd.port");
+const EMRGD_LOG = (projectDir = os.homedir()) => path.join(projectDir, ".emrg", "emrgd.log");
 const MAX_PAYLOAD = 16 * 1024 * 1024; // G62/G105：16MB 双向一致（工具输出上限 200KB）
 const AUTH_TIMEOUT_MS = 10_000;
 const SPAWN_WAIT_MS = 5_000;
 const PENDING_TIMEOUT_MS = 5_000;
 const STREAM_END_TIMEOUT_MS = 30_000; // G94：最后帧后 30s 无 done 强制结束
+// Rant 2026-08-09T13:16:36 ⑤（防风暴总闸）：单个"连接生命周期"内最多 spawn
+// MAX_SPAWN_ATTEMPTS 次 daemon——之后不再拉起，只把真实错误（含 emrgd.log 尾部）
+// 抛给上层，杜绝 GUI 每 5s 反复 spawn（每次 spawn 都是一个新的 cmd 窗口来源）。
+const MAX_SPAWN_ATTEMPTS = 3;
 
 const SESSION_ID_RE = /^s_\d{6}_\d{4}_[0-9a-f]{4,8}$/;
 
@@ -72,6 +77,7 @@ class DaemonClient {
     this._authFailed = false;
     this._reconnectTimer = null;
     this._stopReconnect = false;
+    this._spawnAttempts = 0; // 连接生命周期内 spawn 计数（成功 auth 后归零）
   }
 
   // ── 生命周期 ────────────────────────────────────────────
@@ -91,7 +97,28 @@ class DaemonClient {
     }
   }
 
+  _readLogTail(lines = 15) {
+    // R124 对应（daemon_manager.py）：spawn 超时后读 emrgd.log 尾部，
+    // 让宿主看到真实失败原因（缺 DLL / PATH / 端口冲突），而不是干巴巴的
+    // "failed to start within timeout"（rant 2026-08-09T13:16:36 验收项 ②）。
+    try {
+      const data = fs.readFileSync(EMRGD_LOG(this.projectDir), "utf8");
+      const tail = data.trim().split("\n").slice(-lines).join("\n");
+      return tail ? `\n  emrgd.log tail:\n${tail}` : "";
+    } catch {
+      return "";
+    }
+  }
+
   async startDaemon() {
+    // Rant 2026-08-09T13:16:36 ⑤：spawn 节流——超过上限不再拉起（防窗口/重试风暴）。
+    if (this._spawnAttempts >= MAX_SPAWN_ATTEMPTS) {
+      throw new Error(
+        `daemon failed to start after ${MAX_SPAWN_ATTEMPTS} attempts — ` +
+        `please start it manually ('emrg server') and check emrgd.log${this._readLogTail()}`
+      );
+    }
+    this._spawnAttempts += 1;
     // Phase 4（rant #12 §4）：打包模式直接 spawn 捆绑 emrgd 可执行文件（脚本内部
     // exec python -m emrg.server）；源码模式保持 python -m emrg.server。
     if (this._isPackaged) {
@@ -116,7 +143,7 @@ class DaemonClient {
         if (await this.isRunning(500)) return child;
         await new Promise((r) => setTimeout(r, 300));
       }
-      throw new Error("emrgd failed to start within timeout");
+      throw new Error(`emrgd failed to start within timeout${this._readLogTail()}`);
     }
     // G125：spawn 设 cwd=project_dir（daemon load_skills 用 Path.cwd() 加载项目级 skills）
     const python = this._findPython();
@@ -138,7 +165,7 @@ class DaemonClient {
       if (await this.isRunning(500)) return child;
       await new Promise((r) => setTimeout(r, 300));
     }
-    throw new Error("emrgd failed to start within timeout");
+    throw new Error(`emrgd failed to start within timeout${this._readLogTail()}`);
   }
 
   _findDaemonExecutable() {
@@ -240,6 +267,7 @@ class DaemonClient {
 
     this.connected = true;
     this._authFailed = false;
+    this._spawnAttempts = 0; // 连接生命周期成功 → 重置 spawn 节流计数
 
     // 5. 注册 message/close 监听 → 事件流分发
     this.ws.on("message", (data) => this._onFrame(data));

emrg/gui/main.js

@@ -31,6 +31,14 @@ function main() {
   let ownStream = false; // 自有流运行中（G65：禁止切会话）
   let ownStreamRequestId = null; // 自有流 request_id（广播 done 不清锁）
   let reconnectTimer = null;
+  // Rant 2026-08-09T13:16:36 ③/⑤：重连指数退避（1s→2s→4s→…封顶 60s）。
+  // 之前固定 1s——daemon 缺失时每 5s 一轮 spawn，弹窗/日志风暴。成功连接后复位。
+  let reconnectDelayMs = 1000;
+  const MAX_RECONNECT_DELAY_MS = 60_000;
+  // Rant 2026-08-09T13:16:36 ⑤：daemon_stopped 提示每个连接生命周期只发一次——
+  // 否则退避封顶 60s 后每轮重试都命中节流、渲染层每分钟追加一条重复系统消息
+  // （对称 TUI app.py _throttle_warned，PR #594）。成功连接后复位。
+  let daemonStoppedNotified = false;
   let stopping = false;
 
   // ── 窗口 ────────────────────────────────────────────────
@@ -646,20 +654,31 @@ vision = false
       await client.ensureConnected();
       logger.info("[gui] connected to emrgd");
       cancelReconnect();
+      reconnectDelayMs = 1000; // 退避复位
+      daemonStoppedNotified = false; // 节流提示复位（下个生命周期可再提示）
       sendToRenderer("status", { connected: true });
     } catch (e) {
       if (client._authFailed) {
         // G88：认证失败 → 停止自动重试
         sendToRenderer("status", { connected: false, auth_failed: true, error: e.message });
         return;
       }
+      // Rant 2026-08-09T13:16:36 ⑤：spawn 节流命中 → 告知宿主真实原因
+      // （含 emrgd.log 尾部），不再无限拉起 daemon。只提示一次，防退避重试
+      // 每分钟重复追加系统消息。
+      if (String(e.message).includes("after 3 attempts") && !daemonStoppedNotified) {
+        daemonStoppedNotified = true;
+        sendToRenderer("status", { connected: false, daemon_stopped: true, error: e.message });
+      }
       logger.warn(`[gui] ensureConnected failed: ${e.message}`);
       scheduleReconnect();
     }
   }
 
   function scheduleReconnect() {
     if (stopping || reconnectTimer) return;
+    const delay = reconnectDelayMs;
+    reconnectDelayMs = Math.min(reconnectDelayMs * 2, MAX_RECONNECT_DELAY_MS); // 指数退避
     reconnectTimer = setTimeout(async () => {
       reconnectTimer = null;
       sendToRenderer("status", { connected: false, reconnecting: true });

emrg/gui/renderer/js/app.js

@@ -1069,6 +1069,11 @@ const App = (() => {
     } else if (data.auth_failed) {
       updateConnectionDot("red");
       Chat.addSystemMessage(_t("app.authFailed"));
+    } else if (data.daemon_stopped) {
+      // Rant 2026-08-09T13:16:36 ⑤：spawn 节流命中——显示真实失败原因（含
+      // emrgd.log 尾部），提示宿主手动启动，不再无限重试弹窗。
+      updateConnectionDot("red");
+      Chat.addSystemMessage(_t("app.daemonStopped", { msg: data.error || "" }));
     } else {
       updateConnectionDot("red");
     }

emrg/gui/renderer/js/i18n.js

@@ -302,6 +302,7 @@ const I18N = (() => {
       "app.unknownError": "未知错误",
       "app.unknown": "未知",
       "app.authFailed": "认证失败了，请检查设置里的 API Key。",
+      "app.daemonStopped": "daemon 启动失败（已停止自动重试）。请在终端运行 `emrg server` 排查；\n{msg}",
       "app.versionInfo": "EMRG GUI v{ver} · 实例 {id} · 模型 {model} · 已进化 {n} 次",
     },
 
@@ -592,6 +593,7 @@ const I18N = (() => {
       "app.unknownError": "Unknown error",
       "app.unknown": "unknown",
       "app.authFailed": "Authentication failed — check your API Key in Settings.",
+      "app.daemonStopped": "daemon failed to start (auto-retry stopped). Run `emrg server` in a terminal to debug;\n{msg}",
       "app.versionInfo": "EMRG GUI v{ver} · Instance {id} · Model {model} · Evolved {n} times",
     },
   };