fix(web): middle-click pastes in the terminal on Linux - #11018
Merged
maria-rcks merged 2 commits intoSep 10, 2026
Conversation
The terminal canvas ignored the middle button: onPointerDown returned for any button but 0, so the standard Linux middle-click paste did nothing and Chromium started autoscroll instead. Middle-click now pastes through the same bracketed-paste path as every other paste. A browser cannot read the X11/Wayland PRIMARY buffer, so the terminal's own selection is the source when the user highlighted here and the system clipboard covers the rest. Applications that track the mouse still receive the click. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Contributor
Bugbot is paused — on-demand spend limit reachedBugbot uses usage-based billing for this team and has hit its on-demand spend limit. A team admin can raise the spend limit in the Cursor dashboard, or wait for the next billing cycle to continue. |
github-actions
Bot
added
size:M
![macroscopeapp[bot]](https://avatars.githubusercontent.com/in/900172?s=60&v=4){kind=small}
Contributor
ApprovabilityVerdict: Approved at Macroscope's review found this PR approvable — This is a localized Linux/BSD terminal interaction fix with targeted tests and no schema, deployment, security, billing, or configuration impact. The production behavior is confined to middle-click event handling within the existing terminal surface. You can add or adjust custom eligibility rules. Learn more. |
|
No actionable comments were generated in the recent review. 🎉 ℹ️ Recent review info⚙️ Run configurationConfiguration used: Path: .coderabbit.yaml Review profile: CHILL Plan: Advanced Run ID: 📒 Files selected for processing (2)
Limit details: You’ve used all 10 included reviews currently available. 📝 WalkthroughWalkthroughThe Ghostty terminal now supports Linux and BSD middle-click paste from the terminal selection only. It preserves split-pane activation and suppresses browser autoscroll and duplicate paste handling. Tests verify selection paste and no clipboard fallback. ChangesMiddle-click terminal interaction
Estimated code review effort: 3 (Moderate) | ~20 minutes Severity of issue fixed: Medium Merge Risk: ⚪ Minimal · up to Linux and BSD middle-click now pastes only an existing terminal selection while preserving pane activation and preventing duplicate browser handling. The current change is ready to merge with no identified actionable risk. Suggested reviewers: 🚥 Pre-merge checks | ✅ 3 | ❌ 2❌ Failed checks (2 warnings)
✅ Passed checks (3 passed)
Full details: Linked Issues checkExplanation The implementation fixes middle-click event handling and supports pasting the terminal's own selection, but issue
✨ Finishing Touches🧪 Generate unit tests (beta)
Comment |
Middle click no longer falls back to the system clipboard: with nothing selected in the terminal there is no primary-selection-like buffer to read, and CLIPBOARD holds text the user only ever copied. The gesture also stops cancelling pointerdown, which used to drop the compatibility mousedown that activates an inactive split pane, and now cancels the middle mouseup so Chromium cannot paste PRIMARY into the hidden textarea on top of the paste already sent. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Contributor
Bugbot is paused — on-demand spend limit reachedBugbot uses usage-based billing for this team and has hit its on-demand spend limit. A team admin can raise the spend limit in the Cursor dashboard, or wait for the next billing cycle to continue. |
macroscopeapp
Bot
dismissed
their stale review
September 10, 2026 00:49
Dismissing prior approval to re-evaluate d3cf911
aorwall
added a commit
to aorwall/t3code
that referenced
this pull request
Sep 10, 2026
Merges `pingdotgg/t3code` `2a3035353..0f602b3` (16 commits) into the fork. - **Landed:** 283 files (`HEAD^1..HEAD`) against 277 in the upstream range — `merge-stats.mjs` reports an exact 277/277 file match, so nothing in the range was dropped and nothing extra came in. The six over are three typecheck fixes and three fork docs, both listed below. Fork delta 733 files (`HEAD^2..HEAD`). - **Conflicts:** 6 files, all on one upstream feature (pingdotgg#10839, linking several pull requests to a thread). Resolutions in `docs/fork/upstream-merge-log.md`. - **Sweep:** 13 owned-concern hits, all `infra/relay/**` FCM/Android-push files under the decided-out `cloud-relay-connect` concern. Inherited in tree, adopted by nothing. - **Unsupported methods:** 0 ADD, 0 DROP — no `packages/contracts/src/rpc.ts` edit needed. ## What upstream shipped ### Usable as-is against Moatless Pure client work, no backend involvement — these are live the moment this merges. - **pingdotgg#11020** message copy buttons show on touch devices. - **pingdotgg#11018** middle-click pastes in the terminal on Linux. - **pingdotgg#10869** expanded images zoom and pan. - **pingdotgg#11002** the composer uses the available space for model names. - **pingdotgg#10981** duplicate tool-call commands can be expanded independently. - **pingdotgg#10947** provider settings grow a bulk model toggle. - **pingdotgg#10609** the PR list's diff counts return to the top right. - **pingdotgg#11022** remote projects open in Zed (`packages/contracts/src/editor.ts` plus the desktop shell — the fork ships both). - **pingdotgg#10998 / pingdotgg#10983 / pingdotgg#10964** three Android glass/overlap fixes in `apps/mobile`. ### Unsupported in Moatless — needs backend implementation - **pingdotgg#10839 — several pull requests per thread.** This is the substantive decision in the merge. Upstream now carries `thread.pullRequests: ThreadPullRequestLink[]`, `packages/shared/src/threadPullRequests.ts`, and a `ThreadPullRequestBadgeControl` pill with its own `pull-requests` stack tab. That is exactly the equivalent the fork's `task-bound-pull-request` convergence entry said to re-home its `+N` menu onto — but it cannot be re-homed yet: Moatless serves no `pullRequests` array on a thread and does not advertise the new `threadPullRequests` capability, so upstream's badge would resolve to nothing and paint an empty pill over a working one. Taking `theirs` would have silently deleted live fork behaviour. **Resolution:** upstream's implementation landed whole, and the two presentations are switched on `useSupportsMultiplePullRequests` — upstream's badge and stack where the server advertises the capability, the fork's binding-derived pill and `+N` menu where it does not. Additive, no prop threading, and it re-homes itself the day the backend advertises. `docs/fork/inventory.json` and `docs/fork/gaps.md` are updated with the switch and with the exact deletion list for when that happens. **To close it:** serve `thread.pullRequests` on `OrchestrationThread`/`OrchestrationThreadShell` from `task_bindings`, and report `capabilities.threadPullRequests: true`. - **pingdotgg#10870 — find threads by linked pull request.** Search terms come off the same `thread.pullRequests` array, so sidebar and command-palette search by PR number/URL match nothing here until the array is served. Closes with pingdotgg#10839. - **pingdotgg#10875 — navigate, merge and rebase GitHub stacks.** Adds two RPC methods, `pullRequests.stack` and `pullRequests.linkedThreads`, which the Moatless backend does not dispatch. Both are already covered by the shared `PullRequestRpcError` union, so the client decodes the refusal correctly and the stack UI stays inert — no contract change needed. Implementing the two methods is what turns it on. - **pingdotgg#10416 — Android agent notifications and ongoing activity.** Rides FCM through `infra/relay`, which is part of the decided-out `cloud-relay-connect` concern (being removed with Clerk). Inherited in tree, not adopted. ### Backend behaviour worth reproducing in Moatless - **pingdotgg#11007 — recent PR reads survive a server restart.** Upstream added `apps/server/src/pullRequest/PullRequestReadCache.ts`, persisting which pull requests a user has already read so a restart does not re-mark the whole list unread. Moatless owns this surface itself, so nothing in this repository holds it open — recorded so whoever touches the backend's PR read state knows the answer exists upstream. ## Verification `verify.mjs`, seven of eight green: `duplicate-adds`, `tripwires`, `resolution-check`, `unsupported-methods`, `fmt:check`, `lint`, `typecheck`. `test` is red on `@t3tools/desktop` alone — `scripts/browser-secret-native.test.mjs > bundled libsecret helper` fails to compile because `libsecret-1` is not installed in this sandbox. **Pre-existing environment gap, not merge-introduced:** it is already an entry in `docs/fork/gaps.md`, and `git diff --name-only HEAD^1 HEAD | grep browser-secret` is empty. 100 of 102 desktop files pass. Four packages did not finish under `vp run -r test` (`@t3tools/mobile`, `t3`, `@t3tools/web`, `t3code-relay`) and all four pass when run alone, which is parallel load rather than the merge. Three typecheck failures were fixed in the merge commit, all fork-only web code that upstream's widened shared types reached: `sandboxControl.placement.test.tsx` needed the two new `RightPanelTabs` props, and `useSandboxAvailability.ts` / `useSandboxDetail.ts` needed `isSuccess` threaded through now that `EnvironmentQueryView` carries it. Nothing is unresolved. --- Moatless task: https://moatless.soaplabstest.com/tasks/db1b3cbe-4401-441b-bbec-6b0c725c93ce
github-actions Bot
added a commit
to omarcresp/t3code-flake
that referenced
this pull request
Sep 10, 2026
## What's Changed * fix(web): restore pr list diff counts to the top right by @maria-rcks in pingdotgg/t3code#10609 * fix(web): show message copy buttons on touch devices by @maria-rcks in pingdotgg/t3code#11020 * fix(web): middle-click pastes in the terminal on Linux by @maria-rcks in pingdotgg/t3code#11018 * fix(editors): open remote projects in Zed by @maria-rcks in pingdotgg/t3code#11022 * feat: add blue and orange diff color palette by @maria-rcks in pingdotgg/t3code#10671 * fix(server): resolve project identity before legacy pr relinks by @t3-code[bot] in pingdotgg/t3code#11045 **Full Changelog**: pingdotgg/t3code@v0.0.41-nightly.20260910.1473...v0.0.41-nightly.20260910.1486 Upstream release: https://github.com/pingdotgg/t3code/releases/tag/v0.0.41-nightly.20260910.1486
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Middle-clicking the integrated terminal did nothing:
onPointerDowninapps/web/src/terminal/ghostty/surface.tsreturned early for any button but 0, so the standard Linux middle-click paste never ran and Chromium started autoscroll instead.Middle click on Linux/BSD now pastes the terminal's own Ghostty selection through
pasteFromClipboard, the same bracketed-paste path (and paste race token) used byCtrl+Shift+V,Shift+Insert, and the right-click menu. With nothing selected in the terminal it does nothing: a browser cannot read the X11/Wayland PRIMARY buffer, and CLIPBOARD is deliberately not substituted, so this never emits text the user only copied. Applications that track the mouse (vim, less) still receive the middle-button press instead of a paste.This covers in-terminal selections only. External PRIMARY is not wired: reading a highlight made in another Linux app needs Electron
clipboard.readText('selection')(and a matching writer on selection change) plus a desktop IPC channel, which is not part of this PR. Middle-clicking with no terminal selection stays a no-op rather than pasting the wrong buffer.Two details worth reviewing:
pointerdownis intentionally left uncancelled, because cancelling it drops the compatibilitymousedownthatThreadTerminalDraweruses to activate an inactive split pane; andmouseupis cancelled, because Chromium otherwise pastes PRIMARY into the focused hidden textarea and would deliver a second paste throughonPaste.Verified:
vp test run src/terminal/ghostty/surface.test.tsinapps/web— 54 passed, including a new case asserting the selection is pasted and that the clipboard is never read.vp run --filter @t3tools/web typecheck— exit 0, no errors.vp lint apps/web/src/terminal/ghostty/surface.ts apps/web/src/terminal/ghostty/surface.test.ts— exit 0. Not verified on a real X11 or Wayland session.Mobile has no middle button; desktop wraps web and picks this up automatically.
Refs #10921
UI evidence: unverified. The browser preview host was unavailable in this session, so the changed interaction was not exercised in a real client. Scoped tests, typecheck, and lint pass; a reviewer should exercise the interaction locally before merge.
Done by Claude Opus 5 (1M context) in Claude Code.
🤖 Generated with Claude Code
Summary by CodeRabbit