dsh-mobile-flow 0.8.0 → 0.8.1

package/README.md

@@ -30,7 +30,7 @@ On viewports ≤720px (the same breakpoint the official question card uses) the
 5. **No auto-focus on session switch** — the stock UI returns focus to the input box on every mount / session switch (a desktop convenience), which pops the on-screen keyboard over half the screen on phones (and is just as unwelcome on a tablet). Narrow viewports and touch devices swallow that programmatic focus; tapping the input box still focuses it normally.
 6. **Workspace row actions always visible** — the trailing buttons on workspace and session rows (the ⋯ menu with rename/delete and the ＋ for a new session in that workspace) surface on hover only; touch has no hover, so narrow viewports and touch devices show them whenever the sidebar list is rendered.
 7. **Native input takeover** — narrow screens and touch devices are taken over by default (a mouse-driven desktop is left alone); the draft surface is drawn by a native textarea whose metrics match the stock box exactly (geometry is measured from the live composer every frame the card relayouts). Enter still performs the official send gesture (including `/` menu arbitration and the busy-Enter policy), file pastes are forwarded to the official attachment intake, and locked states follow the product's own editability gate.
-8. **The send button follows the field** — as soon as the field holds text, the send button reads as enabled (a live blue circle instead of the disabled grey one) and one tap sends it (queue/steer while a turn streams). While typing that state is painted by pure CSS with **zero DOM writes**, so the "no DOM movement while typing" invariant stands.
+8. **The send button follows the field** — as soon as the field holds text, the send button reads as enabled (a live blue circle instead of the disabled grey one) and one tap sends it (queue/steer while a turn streams), with the tap's own trailing `click` swallowed so the product can never read that same tap as an interrupt (v0.8.1). While typing that state is painted by pure CSS with **zero DOM writes**, so the "no DOM movement while typing" invariant stands.
 9. **The switch is not a phone control** — the tool-row 「输入法✓/✗」 chip renders on every viewport, so a large tablet always has a one-tap say in this; it is also the fallback for a device the default gate fails to recognise as touch.
 
 **A mouse-driven desktop is completely unaffected** (a touch device is taken over however wide it is — that is the point of this change).
@@ -223,6 +223,20 @@ Two fixes:
 - **The look**: the machine draft is stale, but the FIELD knows whether it holds text, and `:placeholder-shown` is the one pure-CSS value probe. While the field holds text and the machine draft is still empty, the takeover paints that button enabled (opacity 1, pointer cursor, pressed feedback) — **zero DOM writes**, the hard rule stands. Once the draft is in sync (e.g. after a commit) the button is left exactly as the product renders it.
 - **The gesture**: the card's capture `pointerdown` now recognises the submit control (the trailing row's last primary button; an interruptible child's extra Stop circle renders before it) and, when a draft is still uncommitted, **takes the gesture**: commit the draft, then run **the product's own send gesture** — the same path Enter takes (slash adjudication, busy policy, steer/queue). One tap, one send, no engine timing involved. Under a running turn the button now wears the send arrow while the field holds text (the Stop square is hidden and the product's own arrow path is painted as a mask), so face and gesture agree.
 
+**v0.8.1 (tablet re-check: "sending a message kills the session")**: the report was that sending while a turn streams
+**terminates that turn** — the message lands in the pending queue instead, and 「插话发送」 (steer) is greyed out
+forever (only a *running* turn can be steered), so the only way out was delete-and-resend. The cause sits in the
+v0.7.3 gesture bridge: canceling `pointerdown` does **not** cancel the tap's own `click` (measured 2026-10-12, Chrome
+152: a trusted `click` on that very button ~100 ms later, when the finger lifts). By then the carried submit has
+emptied the machine draft, so the product reads its own button as the running turn's **Stop** control and calls
+`session/cancel` — the turn the user just sent into dies with `turn/end aborted reason user`, the queued message stays
+behind (`keepInbox`), exactly what the host log showed. The fix records the carried control and **swallows that one
+trailing click** (same button, within 1 s) in the card's capture phase (`preventDefault` + `stopPropagation`), so one
+tap is one send; a click with no carried tap in front of it still reaches the product (tapping the button with an
+empty field remains the stock Stop). Verified by two new jsdom assertions plus the live page at 640×1024 and
+1024×768: only `session/prompt` is sent, no `session/cancel`, the turn keeps streaming; the empty-field tap still
+sends `session/cancel`.
+
 **Escape hatch**: a small tool-row button (**输入法✓ / 输入法✗**, narrow viewports only) swaps back to the stock
 input box and remembers the choice — no device can be left stuck.
 

package/README.zh.md

@@ -30,7 +30,7 @@
 5. **切会话不自动聚焦** —— 官方在每次挂载/切换会话时把焦点还给输入框（桌面便利设计），手机上会弹出输入法占掉半屏，平板上也一样挡事。窄屏与触屏设备上拦截这类程序化聚焦；你主动点输入框时照常聚焦。
 6. **workspace 行操作按钮常显** —— 分组行与会话行尾部的按钮（⋯ 菜单含重命名/删除、＋ 在该 workspace 新建会话）官方仅在悬停时显示；触屏没有悬停，窄屏与触屏设备下侧边栏列表渲染时常显这两个按钮。
 7. **原生输入框接管** —— 窄屏或触屏设备默认接管（鼠标操作的桌面端不动），输入区绘制成原生 textarea，外观、内边距、字号行高与官方完全一致（几何按官方输入区实时测量对齐）。回车仍是官方发送手势（含 `/` 菜单仲裁、busy-Enter 策略），粘贴图片转发给官方附件入口，锁定态跟随官方可编辑性开关。
-8. **发送按钮跟「输入框里有什么」走** —— 只要输入框里有字，右下角发送按钮就是**可点的亮色**（不再是灰的），点一下即发送（运行中的回合则是排队/引导）；打字期间这个状态由纯 CSS 驱动，**零 DOM 写入**，不破坏「打字期零 DOM 变动」这条硬约束。
+8. **发送按钮跟「输入框里有什么」走** —— 只要输入框里有字，右下角发送按钮就是**可点的亮色**（不再是灰的），点一下即发送（运行中的回合则是排队/引导），并且把这一记 tap 自己的尾随 `click` 吞掉，官方就再也不会把同一次点击读成「打断」（v0.8.1）；打字期间这个状态由纯 CSS 驱动，**零 DOM 写入**，不破坏「打字期零 DOM 变动」这条硬约束。
 9. **开关不分屏幕大小** —— 工具行的「输入法✓/✗」在任何视口都显示（不只在手机上），所以「大屏平板要不要用原生输入框」永远有一键可切的地方；它同时也兼容「本插件没识别出这是触屏设备」的情况。
 
 **鼠标操作的桌面端完全不受影响**（触屏设备即使屏幕很宽也会接管——这正是本次改动要的结果）。
@@ -166,6 +166,8 @@ localStorage.removeItem("dsh-mobile-flow:growth");       // 恢复默认
 - **外观**：机器草稿是旧的，但**字段自己知道有没有字**，而 `:placeholder-shown` 是纯 CSS 里唯一能读字段值的探针。于是接管卡片在「字段有字 + 机器草稿仍为空」时把发送按钮画成启用态（opacity 1、手型光标、按下有反馈）——**零 DOM 写入**，硬约束不破；机器草稿已经同步（如已提交）时不动它，保持官方原样。
 - **手势**：卡片捕获阶段的 `pointerdown` 判断「点的是不是提交按钮（trailing 行里最后一颗 primary，可中断子会话额外的 Stop 圆钮在它前面）」且「还有未提交的文字」，是则**拿走这次手势**：先落草稿，再走**官方自己的发送手势**（与回车同一条路：斜杠仲裁、busy 策略、排队/引导），一次点击一条消息，不再看引擎脸色。运行中的回合里，字段有字时按钮**换成发送箭头**（隐藏官方的 Stop 方块，用官方同一份箭头 path 画 mask），点它就是发送——脸和手势一致。
 
+**v0.8.1（平板复验反馈：发一条消息，正在跑的回合被打断）**：现象是「一发送，会话就被终止」——消息进了待发送队列，而「插话发送」永远是灰的（只有**运行中**的回合才能插话），只能删掉那条消息重新发。根因在 v0.7.3 的手势桥接：**取消 `pointerdown` 并不能取消这一记 tap 自己的 `click`**（2026-10-12 在 Chrome 152 上实测：手指抬起后约 100 ms，同一颗按钮上照样派发一次 `isTrusted` 的 click）。那时我们 carry 的那次提交已经把机器草稿清空，官方读到「运行中 + 草稿为空」就把它当成 **Stop 圆钮**，于是发出 `session/cancel` —— 用户刚发进去的那个回合当场被杀（宿主机日志里正是 `turn/end aborted reason user`，且因为 `keepInbox` 排队消息留在队列里），与现象完全对上。修法：carry 成功时记下这颗按钮，紧随其后的那一记 click（同一颗按钮、1 秒内）在卡片捕获阶段 `preventDefault + stopPropagation` 吞掉——一次 tap 就是一次发送；**没有 carry 的 click 照旧放行**（空输入框时点它就是官方的 Stop，不能挡）。验证：jsdom 新增两条断言 + 真实页面探针（竖屏 640×1024、横屏 1024×768 各跑一遍：只发 `session/prompt`、没有 `session/cancel`、回合继续跑；空输入框点击仍发出 `session/cancel`）。
+
 **v0.7.2（手机反馈：新建会话后点输入框没反应）**：新会话刚建出来、**workspace 还没解析出来**时，官方卡片其实是「选择工作区」的**选择器触发器**（`aria-haspopup="menu"`、`contenteditable="false"`、没有绑编辑器），根本不是输入框。旧版接管照样压上去，还把这份「不可编辑」**锁存**成原生 textarea 的 `readOnly` —— 于是：点输入框**不弹键盘、不出现光标**，同时把官方那记「打开选择器」的点击吞掉，用户既不能打字也打不开选择器；而且 workspace 解析好之后它**不会自愈**（只有换会话重新挂载才恢复），正好对上「第一次新建会话没反应、切走再新建就好」的现象（2026-09-19）。
 
 处理：

package/lib/client.js

@@ -1643,10 +1643,14 @@ window.__ModuleLoader__.load({
           var primaries = trailing.querySelectorAll("[class*='_primary']");
           return primaries.length !== 0 && primaries[primaries.length - 1] === button;
         };
+        /** The tap whose gesture this bridge carried, waiting for the engine's
+            own trailing `click` (see onClickCapture). */
+        var carriedClick = null;
         var onPointerDown = function (event) {
           var area = areaRef.current;
           var wrap = wrapRef.current;
           if (area === null || wrap === null || event.target === area) return;
+          carriedClick = null; /* a new tap supersedes any carry still waiting */
           var x = event.clientX;
           var y = event.clientY;
           debug("tap " + Math.round(x) + "," + Math.round(y) + " target=" + (event.target && event.target.tagName));
@@ -1676,11 +1680,37 @@ window.__ModuleLoader__.load({
           if (submitAt(button) === true && pendingRef.current !== null) {
             event.preventDefault();
             mirrorNow();
-            debug("carry tap -> " + (sendGesture() ? "submit" : "nothing"));
+            var carried = sendGesture();
+            debug("carry tap -> " + (carried ? "submit" : "nothing"));
+            /* Canceling `pointerdown` does NOT cancel the tap's own `click`
+               (measured 2026-10-12 on Chrome 152: a trusted click on this very
+               button ~100 ms later, after the finger lifts). By then the
+               carried submit has emptied the machine draft, so the product
+               reads its own button as the running turn's Stop control and
+               cancels the turn the user just sent into — "sending a message
+               kills the session", then the message sits in the queue with
+               steering unavailable (only a running turn can be steered).
+               Swallow that one trailing click. */
+            carriedClick = carried ? { button: button, at: Date.now() } : null;
             return;
           }
+          carriedClick = null;
           mirrorNow();
         };
+        /** The engine's click for a tap this bridge already carried: the same
+            control, moments later, and never a second gesture. */
+        var onClickCapture = function (event) {
+          var carried = carriedClick;
+          carriedClick = null;
+          if (carried === null || Date.now() - carried.at > 1000) return;
+          var target = event.target;
+          var insideButton = carried.button !== null && typeof carried.button.contains === "function"
+            && target !== null && (target === carried.button || carried.button.contains(target) === true);
+          if (insideButton !== true && submitAt(buttonAt(event.clientX, event.clientY)) !== true) return;
+          event.preventDefault();
+          event.stopPropagation();
+          debug("swallow trailing click of a carried tap");
+        };
         /* Engines without a working inert: if anything hands focus back to the
            stock editor, take it straight back to the field. */
         var onFocusIn = function (event) {
@@ -1697,11 +1727,13 @@ window.__ModuleLoader__.load({
           if (document.visibilityState === "hidden") mirrorNow();
         };
         card.addEventListener("pointerdown", onPointerDown, true);
+        card.addEventListener("click", onClickCapture, true);
         document.addEventListener("focusin", onFocusIn, true);
         window.addEventListener("pagehide", onHide);
         document.addEventListener("visibilitychange", onHide);
         return function () {
           card.removeEventListener("pointerdown", onPointerDown, true);
+          card.removeEventListener("click", onClickCapture, true);
           document.removeEventListener("focusin", onFocusIn, true);
           window.removeEventListener("pagehide", onHide);
           document.removeEventListener("visibilitychange", onHide);

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "dsh-mobile-flow",
-  "version": "0.8.0",
+  "version": "0.8.1",
   "description": "Mobile fixes for the DeepSeek Harness Web UI: on narrow (≤720px) screens the input bar and AI confirmation cards scroll with the page instead of pinning to the viewport floor, edges slim down, and touch devices (any width, a tablet included) no longer auto-focus the input or hide the workspace row actions; the draft surface becomes a native textarea on narrow screens and on any touch device (ArkWeb/Android-IME hardened: zero DOM writes while typing, resizing only at commit points, with a tool-row escape hatch and an on-device diagnostics bench) so mobile IMEs stop clearing typed text and stop dropping the keyboard; the send button follows the field's content (live as soon as there is text, one tap sends), not the deliberately stale machine draft.",
   "license": "MIT",
   "type": "module",