Capture Document Picture-in-Picture windows in Replay

[sentry-junior]

We have a use case where a modal is shown in Document Picture-in-Picture mode. That opens a separate window where the modal UI is rendered. With Sentry we can capture network requests, logs, errors, and related telemetry from that PiP window, but Session Replay does not capture the PiP window itself, so the modal UI is missing from the replay.

Current behavior

• Replay/rrweb attaches to the opener document.
• A Document PiP window is a separate same-origin Window/document created via documentPictureInPicture.requestWindow() (MDN).
• The SDK does not hook documentPictureInPicture / requestWindow, and only one replayIntegration() instance is supported per page.
• Manual start() / stop() / flush() and canvas snapshot() / addWindow do not record PiP DOM into Session Replay playback.
• Related multi-document work is iframe-scoped only (e.g. _experiments.recordCrossOriginIframes, Support for iframes in Session Replay #7279, Capture DOM events (e.g. clicks) for same-origin iframe content #14008, [Replay] Expose rrweb recordCrossOriginIframes under _experiment #14809, recordCrossOriginIframes option not recording cross origin iframes #16173, discussion Support recording content from cross-origin iframes #11858).

Gap

• Product UI hosted in Document PiP is missing from replays, even when other Sentry signals from that window are present.
• Native <video> PiP is browser chrome (not app DOM) and is a different surface; media is also typically blocked by default privacy settings.

Impact

• Debugging and support workflows break when the relevant UI only exists in the PiP window.
• Customers cannot recover that visual context with current Replay APIs or config.

Requested by david.quintas.

--

View Junior Session [Sentry]

[linear-code]

JS-3502