Stylus/tablet click (ButtonPress) not registering in desktop (CEF) build when device uses absolute positioning — works in Relative mode, in browser, and with physical mouse
維護者通常 1 天內回覆
還沒有人認領這個 Issue。
評估
研究方向
首先使用 Graphite desktop Flatpak 重現此問題，並以調查細節和 #2535 為背景，比較 CEF-based 輸入處理對絕對模式和相對模式平板事件的處理方式。僅在必要時檢查 preferences.ron，以使 disable_ui_acceleration 與此 bug 分開；當絕對模式下筆的 ButtonPress 和 ButtonRelease 事件在 UI 和 viewport 中正常運作，且不會使滑鼠輸入或相對模式輸入發生回歸時，即視為完成。
由索引模型根據 Issue 內容生成。
描述
Describe the Bug
On Linux (Flatpak build), a graphics tablet's pen click does not register in the desktop app's CEF-rendered UI or viewport when OpenTabletDriver is configured in Absolute mode (normal tablet-to-screen 1:1 mapping), even though:
- The exact same tablet/pen works correctly in the web version at https://editor.graphite.art/ (in Chromium), also in Absolute mode
- A physical USB mouse clicks normally in the desktop app
xdotool click 1(synthetic XTEST click) also fails to register in the desktop app- Switching OpenTabletDriver to Relative mode makes clicking work correctly in the desktop app — this isolates the bug to something specific about how absolute-positioning tablet devices are handled, since relative mode is otherwise unusable for real drawing (no 1:1 tablet-to-screen mapping)
Cursor motion from the tablet works perfectly in Absolute mode in the desktop app — only button press/release events fail to have any effect while in that mode.
Environment
- OS: Void Linux, X11 (picom compositor, XFCE)
- Graphite: desktop Flatpak build (
art.graphite.Graphite), downloaded ~July 2026 - Tablet driver: OpenTabletDriver 0.6.7 (Flatpak)
- Tablet: Huion H1161
- GPU: AMD (RADV/Mesa)
Steps To Reproduce
- Install OpenTabletDriver via Flatpak, configure tablet in Absolute mode ("Artist Mode" preset) with tip bound to Mouse Button: Left
- Install Graphite via Flatpak
- Launch Graphite desktop app
- Attempt to click/draw with the stylus tip on the canvas or any UI button
- Motion works; the click has no effect
- Switch OpenTabletDriver to Relative mode, click Apply, retest — clicking now works correctly
Investigation so far
xinput test-xi2confirms the pen'sButtonPress/ButtonReleasearrive as genuine X11 core-protocol events from the same device (OpenTabletDriver Virtual Tablet, XI2 id 19) that generates the working motion events — not routed through the XTEST virtual pointer (id 4)udevadm infoconfirms the OTD virtual device is taggedID_INPUT_MOUSE=1libinput list-devicesshows the OTD virtual device'sCapabilities: pointer, but itsDeviceChangedXI2 announcement in Absolute mode includes absolute X/Y valuators plus an XIScrollClass (relative horizontal/vertical scroll valuators) — attributes a plain mouse does not have- A real physical USB mouse (no absolute valuators, no scroll-class) clicks Graphite's UI/canvas without any issue in either mode
xdotool click 1(XTEST-injected click) also fails silently in the desktop app while in Absolute mode- Confirmed: switching the same physical tablet/driver to Relative mode (removing absolute valuators from the device's XI2 announcement) makes clicks work correctly again
This strongly suggests Graphite's CEF-embedded input handling (see #2535, the CEF-based desktop architecture) filters or mis-routes button events specifically from devices that report absolute positioning valuators, rather than a general click-handling bug — since motion, physical mouse clicks, and relative-mode tablet clicks all work fine, and only absolute-mode tablet clicks fail.
Additional Details
- Toggling Preferences → Render GUI on CPU (
disable_ui_accelerationinpreferences.ron) fixed an unrelated issue where the GUI failed to render at all, but did not affect this click bug. - Happy to provide further logs/testing (e.g.
libinput debug-events, CEF verbose logging flags) if useful for triage.
- 主要語言
- Rust
- 星號
- 27.8k
- 分支
- 1.3k
- 平均合併
- 10 小時 53 分鐘
- 30 天內合併 PR
- 195
環境準備
在瀏覽器裡用你自己的 GitHub 帳號啟動這個專案的開發容器。
- 沒有 Dockerfile 或 Docker Compose 檔案
- 有 Pull Request 範本
- 沒有貢獻指南
從這裡開始
- 先讀完整個 Issue，再讀專案的貢獻指南。
- 在 Issue 下留言說明你要接手 —— 這能避免兩個人做同樣的事。
- Fork 儲存庫，在一個分支上完成修改。
- 送出 Pull Request，並在描述裡引用這個 Issue 編號。
GraphiteEditor/Graphite 的其他 Issue
-
`third-party-licenses` fails on Windows when the parent process is PowerShell可能已有人在做 @sahilshingate01 於 12 天前認領。 未關閉
難度 2/5 1-3 小時 新手友好度 76/100
GraphiteEditor/Graphite#4745 · 1 則留言 ·
維護者通常 1 天內回覆
-
Resizing a text box by one edge handle incorrectly locks both Max Width and Max Height, silently clipping text可能重新可做 關聯的 PR 已關閉且未合併。 未關閉
難度 2/5 1-3 小時 新手友好度 78/100
GraphiteEditor/Graphite#4396 · 2 則留言 ·
維護者通常 1 天內回覆
-
Type hint is selected in node graph view可能重新可做 關聯的 PR 已關閉且未合併。 未關閉
難度 2/5 1-3 小時 新手友好度 76/100
GraphiteEditor/Graphite#4311 ·
維護者通常 1 天內回覆
-
`cargo run` fails on WebAssembly target due to `CARGO_BUILD_TARGET` leaking into `rustc_codegen_spirv` build可能重新可做 關聯的 PR 已關閉且未合併。 未關閉
難度 2/5 1-3 小時 新手友好度 68/100
GraphiteEditor/Graphite#3939 · 2 則留言 · 1 個 reaction ·
維護者通常 1 天內回覆
-
難度 3/5 1-2 天 新手友好度 50/100
GraphiteEditor/Graphite#4901 ·
維護者通常 1 天內回覆
查看 GraphiteEditor/Graphite 的全部 Issue
相似的 Issue
-
A-linter
難度 2/5 1-3 小時 新手友好度 73/100
oxc-project/oxc#28783 ·
維護者通常 1 天內回覆
-
Substrait consumer doesn't recognise `shift_left` / `shift_right` from functions_arithmetic.yaml未關閉bug
難度 2/5 1-3 小時 新手友好度 76/100
apache/datafusion#27059 ·
維護者通常 1 天內回覆
-
Bump calamine to 0.38: RUSTSEC-2026-0436 (CFB/OLE2 panic and unbounded allocation) fails cargo audit未關閉
難度 2/5 1-3 小時 新手友好度 70/100
docling-project/docling.rs#979 ·
維護者通常 1 天內回覆
-
llm translation
難度 2/5 1-3 小時 新手友好度 72/100
維護者通常 1 天內回覆
-
難度 2/5 1-3 小時 新手友好度 74/100
維護者通常 1 天內回覆