fix(linux): Hyprland cursor telemetry + anchor recording start to the first real frame#159
Open
eduardoaugustolb wants to merge 130 commits into
Open
Conversation
…s and snapping - feat(ai-edition): multi-session chat history + multi-clip timeline with drag-drop - feat(ai-edition): timeline interaction P0 — clip resize/delete/duplicate, region drag, Edit Clip dialog - feat(ai-edition): timeline viewport spine — pan/zoom/reorder (T01-T09) - feat(ai-edition): timeline lanes, navigator strip, Place-skip mode (T10-T18) - feat(ai-edition): timeline polish — scrub preview, resequence, snap guides (T19-T25)
…r pipeline into the preview
… and manual compaction
…nd move the transport into the timeline
…nment, then CTranslate2
…e thinking on/off
…atch the v4 design - feat(ai-edition): refactor editor to OpenScreen Editor v4 (Media/Edit/Rec) - style(ai-edition): flatten timeline region pills to v4 design - style(ai-edition): match v4 timeline navigator to design - style(ai-edition): align v4 timeline clip cards + ruler to design - fix(ai-edition): hide timeline lane watermark when lane has regions - style(ai-edition): float clip label as v4 blur chip over waveform - docs(ai-edition): add v4 design-parity + import audit notes
…ounded in the editor
…d add LLM effect tools
…orms and theme readability
…et, HUD drag and Wayland - fix(wgc): stop holding the frame mutex across blocking WriteSample calls - fix(recording): round window-capture dimensions up to even for WGC - fix(recording): bound the video-writer frame wait as a hang safety net - fix(overlay): stop the HUD drag from drifting away from the cursor - test(launch): add regression coverage for HUD drag measurement suppression - fix(recording): disable Vulkan on Wayland so PipeWire capture can import DMA-BUF frames
…ing the native export
…ew limits - fix(zoom): stop a manual zoom drifting during its ease-in, and smooth auto follow - feat(timeline): bring back the global Auto-Focus toggle - fix(editor): say what the preview actually does above 16x - fix(editor): lock the per-zoom focus control under the global toggle, drop the 16x note - docs(engineering): retract the webcam reactive-zoom gap, record the test pass
…ws, masks, images, text - feat(annotations): send annotations to the native compositor - feat(annotations): draw figure arrows in the compositor, and make font size proportional - feat(annotations): render blur and mosaic masks in the compositor - feat(annotations): render image annotations in the compositor - feat(annotations): render text with DirectWrite, closing the preview/render gap
|
No actionable comments were generated in the recent review. 🎉 ℹ️ Recent review info⚙️ Run configurationConfiguration used: defaults Review profile: CHILL Plan: Pro Plus Run ID: 📒 Files selected for processing (5)
🚧 Files skipped from review as they are similar to previous changes (5)
📝 WalkthroughWalkthroughHyprland cursor positions are captured through compositor IPC, normalized into cursor recording samples, and selected by the session factory. Linux screen/audio capture handling is updated, and recording timeline state now starts at the first produced video frame. ChangesCursor recording pipeline
Estimated code review effort: 4 (Complex) | ~45 minutes Sequence Diagram(s)sequenceDiagram
participant ScreenRecorder
participant VideoTrack
participant ElectronAPI
ScreenRecorder->>VideoTrack: waitForFirstVideoFrame
VideoTrack-->>ScreenRecorder: first-frame timestamp
ScreenRecorder->>ElectronAPI: setRecordingState with first-frame timestamp
sequenceDiagram
participant RecordingFactory
participant HyprlandCursorRecordingSession
participant hyprlandCursorIpc
participant HyprlandSocket
RecordingFactory->>HyprlandCursorRecordingSession: create Hyprland session
HyprlandCursorRecordingSession->>hyprlandCursorIpc: queryHyprlandCursorPos
hyprlandCursorIpc->>HyprlandSocket: send j/cursorpos
HyprlandSocket-->>hyprlandCursorIpc: cursor coordinates
hyprlandCursorIpc-->>HyprlandCursorRecordingSession: validated position
Possibly related PRs
Suggested reviewers: 🚥 Pre-merge checks | ✅ 3 | ❌ 2❌ Failed checks (2 warnings)
✅ Passed checks (3 passed)
✨ Finishing Touches🧪 Generate unit tests (beta)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
Actionable comments posted: 4
🧹 Nitpick comments (3)
electron/native-bridge/cursor/recording/hyprlandCursorIpc.test.ts (1)
47-117: 📐 Maintainability & Code Quality | 🔵 Trivial | 💤 Low valueConsider covering the timeout path.
Good coverage overall; the one gap is
SOCKET_TIMEOUT_MS-triggered resolution (server accepts but never responds), which currently isn't exercised.🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the rest with a brief reason, keep changes minimal, and validate. In `@electron/native-bridge/cursor/recording/hyprlandCursorIpc.test.ts` around lines 47 - 117, Add a test in the queryHyprlandCursorPos suite that starts a fake server which accepts the connection but never sends a response, then assert queryHyprlandCursorPos resolves to null after the SOCKET_TIMEOUT_MS timeout path. Ensure the test cleanup still closes the server and removes its temporary socket directory.electron/native-bridge/cursor/recording/hyprlandCursorRecordingSession.ts (2)
27-29: 🩺 Stability & Availability | 🔵 Trivial | ⚡ Quick winSilent no-op when the Hyprland socket path can't be resolved.
If
HYPRLAND_INSTANCE_SIGNATUREis set butXDG_RUNTIME_DIRisn't (or resolution otherwise fails),socketPathisnullandcaptureSample()quietly no-ops on every tick (line 56-58) — the recording ends up with zero cursor samples and no trace of why, which is exactly the diagnosability gap behind the original missing-cursor bug (#99).💡 Suggested diagnostic warning
constructor(private readonly options: HyprlandCursorRecordingSessionOptions) { this.socketPath = options.socketPath ?? resolveHyprlandSocketPath(); + if (!this.socketPath) { + console.warn( + "HyprlandCursorRecordingSession: could not resolve the Hyprland IPC socket path; cursor telemetry will be empty.", + ); + } }Also applies to: 55-58
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the rest with a brief reason, keep changes minimal, and validate. In `@electron/native-bridge/cursor/recording/hyprlandCursorRecordingSession.ts` around lines 27 - 29, Update the HyprlandCursorRecordingSession constructor and captureSample flow to detect when socketPath cannot be resolved and emit a diagnostic warning instead of silently no-oping. Preserve normal capture behavior when a valid path exists, and ensure the warning clearly identifies the missing Hyprland socket path and explains that cursor samples cannot be captured.
19-91: 📐 Maintainability & Code Quality | 🔵 Trivial | ⚡ Quick winAdd a test for
HyprlandCursorRecordingSession.The Hyprland cursor IPC layer has
hyprlandCursorIpc.test.ts, but the new recording session class has no local test. Cover the interval-driven sampling, clamping/normalization,maxSamplescapping, and stop/in-flight-capture behavior, with injectablegetDisplayBounds/socketPath.🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the rest with a brief reason, keep changes minimal, and validate. In `@electron/native-bridge/cursor/recording/hyprlandCursorRecordingSession.ts` around lines 19 - 91, The HyprlandCursorRecordingSession class lacks coverage for its recording lifecycle. Add local tests using injectable getDisplayBounds and socketPath to verify immediate and interval-driven capture, coordinate normalization and clamping, maxSamples eviction, and stop behavior including an in-flight capture without leaving timers or sampling state active; mock the Hyprland cursor query and timer behavior as needed.Source: Coding guidelines
🤖 Prompt for all review comments with AI agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
Inline comments:
In `@electron/native-bridge/cursor/recording/hyprlandCursorRecordingSession.ts`:
- Around line 41-90: Update stop() and captureSample() in the cursor recording
session to track any in-flight capture and await its completion before
returning. Return a snapshot copy of this.samples rather than the mutable array,
ensuring no later capture can alter the returned CursorRecordingData.
In `@src/hooks/useScreenRecorder.ts`:
- Around line 41-94: Add same-package tests covering waitForFirstVideoFrame’s
requestVideoFrameCallback success path, timeout fallback, and unavailable-API
fallback; verify Linux system versus editable cursor branching, audio fallback
cleanup, and that cursor recording receives the anchored first-frame timestamp.
Exercise the affected useScreenRecorder flows rather than adding unrelated
refactors.
- Around line 1273-1308: Update the display/audio capture flow around the
Promise.all call to retain the getUserMedia audio promise and, when
getDisplayMedia rejects, await its result and stop all resolved audio tracks
before propagating the display-capture failure. Preserve the existing video-only
fallback for audio-request failures and the combined stream behavior on success.
- Around line 1265-1266: Update the cursor selection logic in useScreenRecorder
so Linux system mode requests "always" and only cursorCaptureMode ===
"editable-overlay" requests "never". Remove the platform === "linux" condition
while preserving the existing "always" behavior for other non-overlay modes.
---
Nitpick comments:
In `@electron/native-bridge/cursor/recording/hyprlandCursorIpc.test.ts`:
- Around line 47-117: Add a test in the queryHyprlandCursorPos suite that starts
a fake server which accepts the connection but never sends a response, then
assert queryHyprlandCursorPos resolves to null after the SOCKET_TIMEOUT_MS
timeout path. Ensure the test cleanup still closes the server and removes its
temporary socket directory.
In `@electron/native-bridge/cursor/recording/hyprlandCursorRecordingSession.ts`:
- Around line 27-29: Update the HyprlandCursorRecordingSession constructor and
captureSample flow to detect when socketPath cannot be resolved and emit a
diagnostic warning instead of silently no-oping. Preserve normal capture
behavior when a valid path exists, and ensure the warning clearly identifies the
missing Hyprland socket path and explains that cursor samples cannot be
captured.
- Around line 19-91: The HyprlandCursorRecordingSession class lacks coverage for
its recording lifecycle. Add local tests using injectable getDisplayBounds and
socketPath to verify immediate and interval-driven capture, coordinate
normalization and clamping, maxSamples eviction, and stop behavior including an
in-flight capture without leaving timers or sampling state active; mock the
Hyprland cursor query and timer behavior as needed.
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: defaults
Review profile: CHILL
Plan: Pro Plus
Run ID: 2e4590be-7e17-4c8d-80b2-6c24b24ab3e2
📒 Files selected for processing (6)
electron/native-bridge/cursor/recording/factory.tselectron/native-bridge/cursor/recording/hyprlandCursorIpc.test.tselectron/native-bridge/cursor/recording/hyprlandCursorIpc.tselectron/native-bridge/cursor/recording/hyprlandCursorRecordingSession.tssrc/components/video-editor/VideoEditor.tsxsrc/hooks/useScreenRecorder.ts
Comment on lines
+41
to
+94
|
|
||
| const FIRST_FRAME_TIMEOUT_MS = 3000; | ||
|
|
||
| /** | ||
| * Resolve once `track` has produced a real, decodable frame -- not merely once | ||
| * `track.muted` reads false. On Linux/PipeWire the desktop-capture portal can still be | ||
| * negotiating (DMA-BUF modifier renegotiation, `no_hardware_cursors`, etc.) well after | ||
| * the track reports unmuted, so anchoring recording start to "now" makes MediaRecorder's | ||
| * own internal frame-0 arrive silently late: the declared recording duration (wall time | ||
| * from button-press to stop) ends up longer than the video's real content span (playback | ||
| * jumps to the end early), and cursor telemetry -- anchored to the same "now" -- drifts | ||
| * out of sync with the picture. Waiting for an actual rendered frame here, before | ||
| * `recorder.start()` is ever called, keeps duration accounting, cursor telemetry, and | ||
| * MediaRecorder's timeline all pointing at the same wall-clock instant. | ||
| * | ||
| * Falls back to `Date.now()` after `FIRST_FRAME_TIMEOUT_MS` if no frame arrives (e.g. a | ||
| * genuinely stalled capture), so recording start is never blocked indefinitely. | ||
| */ | ||
| function waitForFirstVideoFrame(track: MediaStreamTrack): Promise<number> { | ||
| return new Promise((resolve) => { | ||
| const video = document.createElement("video"); | ||
| video.muted = true; | ||
| video.playsInline = true; | ||
| video.srcObject = new MediaStream([track]); | ||
|
|
||
| let settled = false; | ||
| const cleanup = () => { | ||
| video.srcObject = null; | ||
| video.remove(); | ||
| }; | ||
| const finish = (timeMs: number) => { | ||
| if (settled) return; | ||
| settled = true; | ||
| clearTimeout(timeoutId); | ||
| cleanup(); | ||
| resolve(timeMs); | ||
| }; | ||
|
|
||
| const timeoutId = setTimeout(() => finish(Date.now()), FIRST_FRAME_TIMEOUT_MS); | ||
|
|
||
| if (typeof video.requestVideoFrameCallback !== "function") { | ||
| // API unavailable (older Chromium); fall back to a short settle delay rather | ||
| // than blocking on the full timeout every time. | ||
| finish(Date.now()); | ||
| return; | ||
| } | ||
|
|
||
| video.requestVideoFrameCallback(() => finish(Date.now())); | ||
| void video.play().catch(() => { | ||
| // Autoplay/play() rejection still lets requestVideoFrameCallback or the | ||
| // timeout resolve this promise; nothing else to do here. | ||
| }); | ||
| }); | ||
| } |
There was a problem hiding this comment.
📐 Maintainability & Code Quality | 🟠 Major | 🏗️ Heavy lift
Add tests for the new capture and timing behaviors.
This change adds first-frame waiting, Linux cursor/audio branching, and timeline anchoring without included same-package test coverage. Cover the frame callback/timeout paths, Linux system versus editable cursor modes, audio fallback cleanup, and the timestamp passed to cursor recording.
As per coding guidelines, “Add a test for every new behavior in the same package as the code under test.”
Also applies to: 1250-1318, 1473-1542
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
In `@src/hooks/useScreenRecorder.ts` around lines 41 - 94, Add same-package tests
covering waitForFirstVideoFrame’s requestVideoFrameCallback success path,
timeout fallback, and unavailable-API fallback; verify Linux system versus
editable cursor branching, audio fallback cleanup, and that cursor recording
receives the anchored first-frame timestamp. Exercise the affected
useScreenRecorder flows rather than adding unrelated refactors.
Source: Coding guidelines
…switch, a pane that scrolls
…nd no more pre-v4 shell - fix(editor): wire the New project starting points to the Media/Rec tabs - feat(i18n): translate the right-rail settings panes and editor chrome - refactor(ai-edition): drop the pre-v4 editor shell and its dnd-timeline dep - fix(export): give the live preview back its pause instead of resuming it
… unify the zoom scale - fix(zoom, annotations): the tilt was a quadratic formula collapsing, not a clipping rect - fix(zoom): the tilt was never clipped — one edge was simply parallel to the frame - fix(zoom): round the tilted shadow's corners like the plane's own - fix(zoom): one zoom scale, so the focus can actually reach the recording's edge
main's 24 commits were already reimplemented on this branch, so the resolution keeps the branch's code verbatim: not one file under src/, electron/, tests/ or scripts/ differs from feat/ai-edition. Only release infrastructure is inherited from main: - package.json: version 1.7.0 and the build:win:store (MSIX) script - electron-builder.json5: the appx target - .github/workflows: build.yml release pins, prerelease.yml, promote.yml - .gitignore, and docs/testing/rc-e2e-checklist.md Dropped as superseded by this branch's own implementations: the pre-v4 editor shell (VideoEditor, VideoPlayback, TimelineEditor, videoEventHandlers, rafCoalescer) and main's duplicate HUD-drag regression test. build.yml and .gitignore are normalised back to LF, main's convention, so they land on main without a whole-file rewrite.
Author
|
Addressed in c7adb1b:
Not doing:
|
There was a problem hiding this comment.
Actionable comments posted: 2
🤖 Prompt for all review comments with AI agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
Inline comments:
In
`@electron/native-bridge/cursor/recording/hyprlandCursorRecordingSession.test.ts`:
- Around line 111-120: Update the recording-session test to advance timers by
another 100ms before calling stop(), while preserving the assertion that
stopPromise remains unresolved during the pending capture. In
hyprlandCursorRecordingSession.ts, ensure a skipped tick caused by isSampling
does not replace the existing captureInFlight promise, so stop() continues
waiting for the original IPC query.
In `@src/hooks/waitForFirstVideoFrame.test.ts`:
- Around line 61-64: Update the test case “resolves immediately when
requestVideoFrameCallback is unavailable” to remove the explicit any cast and
clear HTMLVideoElement.prototype.requestVideoFrameCallback via
Object.defineProperty, preserving strict typing and the simulated API gap.
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: defaults
Review profile: CHILL
Plan: Pro Plus
Run ID: a4707dc5-f650-490e-833f-4470902362a7
📒 Files selected for processing (6)
electron/native-bridge/cursor/recording/hyprlandCursorIpc.test.tselectron/native-bridge/cursor/recording/hyprlandCursorRecordingSession.test.tselectron/native-bridge/cursor/recording/hyprlandCursorRecordingSession.tssrc/hooks/useScreenRecorder.tssrc/hooks/waitForFirstVideoFrame.test.tssrc/hooks/waitForFirstVideoFrame.ts
🚧 Files skipped from review as they are similar to previous changes (3)
- electron/native-bridge/cursor/recording/hyprlandCursorIpc.test.ts
- electron/native-bridge/cursor/recording/hyprlandCursorRecordingSession.ts
- src/hooks/useScreenRecorder.ts
Comment on lines
+111
to
+120
| await session.start(); | ||
| await vi.advanceTimersByTimeAsync(100); // kicks off the second, still-pending capture | ||
|
|
||
| const stopPromise = session.stop(); | ||
| let stopped = false; | ||
| void stopPromise.then(() => { | ||
| stopped = true; | ||
| }); | ||
| await Promise.resolve(); | ||
| expect(stopped).toBe(false); // must not resolve while the in-flight capture is pending |
There was a problem hiding this comment.
🎯 Functional Correctness | 🟠 Major | ⚡ Quick win
Cover the repeated-tick in-flight race.
The test stops after the first pending interval tick. A second tick hits isSampling, returns immediately, and replaces captureInFlight; stop() can then resolve before the real IPC query completes. Advance another 100ms before stop() and keep the non-settlement assertion; also prevent skipped ticks from replacing the in-flight promise in hyprlandCursorRecordingSession.ts.
Suggested regression step
await session.start();
await vi.advanceTimersByTimeAsync(100);
+await vi.advanceTimersByTimeAsync(100);
const stopPromise = session.stop();📝 Committable suggestion
‼️ IMPORTANT
Carefully review the code before committing. Ensure that it accurately replaces the highlighted code, contains no missing lines, and has no issues with indentation. Thoroughly test & benchmark the code to ensure it meets the requirements.
Suggested change
| await session.start(); | |
| await vi.advanceTimersByTimeAsync(100); // kicks off the second, still-pending capture | |
| const stopPromise = session.stop(); | |
| let stopped = false; | |
| void stopPromise.then(() => { | |
| stopped = true; | |
| }); | |
| await Promise.resolve(); | |
| expect(stopped).toBe(false); // must not resolve while the in-flight capture is pending | |
| await session.start(); | |
| await vi.advanceTimersByTimeAsync(100); // kicks off the second, still-pending capture | |
| await vi.advanceTimersByTimeAsync(100); | |
| const stopPromise = session.stop(); | |
| let stopped = false; | |
| void stopPromise.then(() => { | |
| stopped = true; | |
| }); | |
| await Promise.resolve(); | |
| expect(stopped).toBe(false); // must not resolve while the in-flight capture is pending |
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
In
`@electron/native-bridge/cursor/recording/hyprlandCursorRecordingSession.test.ts`
around lines 111 - 120, Update the recording-session test to advance timers by
another 100ms before calling stop(), while preserving the assertion that
stopPromise remains unresolved during the pending capture. In
hyprlandCursorRecordingSession.ts, ensure a skipped tick caused by isSampling
does not replace the existing captureInFlight promise, so stop() continues
waiting for the original IPC query.
Electron's screen.getCursorScreenPoint() is frozen at (0,0) under Hyprland's xdg-desktop-portal, which is why cursor recordings there have always shown a static cursor (issue getopenscreen#99). Hyprland exposes its own IPC socket (`hyprctl`'s underlying protocol) that reports the real cursor position at any time. When HYPRLAND_INSTANCE_SIGNATURE is set, route Linux cursor recording through a new HyprlandCursorRecordingSession that samples position from that socket instead of the broken Electron API; other Wayland compositors (no equivalent IPC channel today) keep using the existing TelemetryRecordingSession. Samples are timestamped before the IPC round-trip completes (not after), and the session's stop() awaits any capture still mid-flight and returns a snapshot of its samples, so a race between the sampling interval and stop() can't return a still-mutable array or drop a sample that lands late.
The Linux capture path used the legacy chromeMediaSource constraints, which have no cursor toggle at all. Switch Linux to getDisplayMedia + setDisplayMediaRequestHandler (main.ts already supplies the pre-selected source for any platform), mirroring the Windows path: editable-overlay mode requests cursor:"never" so the editor's overlay (and Hyprland telemetry) is the only cursor source and doesn't double up with one baked into the frame; system mode requests "always". System audio on Linux still goes through the legacy desktop-audio constraints (Electron's getDisplayMedia loopback audio only supports win32/macOS) and gets merged with the getDisplayMedia video track; if that video capture rejects, the merge now awaits and stops the audio-only stream instead of leaving it running with no consumer (mic indicator stuck on).
MediaRecorder's internal frame-0 timeline starts whenever the track delivers its first real frame, not when recorder.start() is called. On Linux/PipeWire the desktop-capture portal can still be negotiating (DMA-BUF modifier renegotiation) for several seconds after the track reports unmuted, so anchoring duration accounting and cursor telemetry to Date.now() at button-press time let both drift out of sync with the actual video content: declared duration exceeded the real content span (playback jumped to the end early) and the cursor overlay lagged the picture. waitForFirstVideoFrame() now blocks on an actual rendered frame (via requestVideoFrameCallback, capped at 3s) before recorder.start() is ever called, so recordingId, segmentStartedAt, and the cursor-telemetry start time all anchor to the same verified wall-clock instant.
eduardoaugustolb
force-pushed
the
feature/hyprland-cursor-telemetry
branch
from
July 26, 2026 15:23
40aa05c to
24231b6
Compare
Author
|
Rebased onto current `main` (24231b6) — the branch had fallen behind the v3/v4 editor rewrite that landed while this PR was open. What that rewrite changed and how this branch adapted:
Verified: |
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
2 participants
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.
Summary
cursor:neverfrom the portal and relying on the editor's editable overlay on Linux.recordingId, and cursor telemetry start time) to the first real captured frame instead of toDate.now()at button-press time. On Linux/PipeWire, the desktop-capture portal can still be negotiating (DMA-BUF modifier renegotiation) for several seconds after the track reports unmuted; anchoring to wall-clock "button pressed" time let the declared recording duration exceed the video's real content span (playback would jump to the end early) and let the cursor overlay drift out of sync with the picture.waitForFirstVideoFrame()now blocks on an actual rendered frame (viarequestVideoFrameCallback, capped at 3s) beforerecorder.start()is called, so everything anchors to the same verified instant.Fixes #99 (comment)
Test plan
tsc --noEmitpasses cleanSummary by CodeRabbit
New Features
Bug Fixes