Bug 300102 – [EventTiming] keydown PerformanceEntry duration is always too long (waits for keyup event)

• Status: RESOLVED
• Resolution: FIXED
• Component: DOM

Created attachment 476950 [details]
Screenshot of Event Timing entries in Safari TP for keydown keyup

Safari Technology Preview
Release 228 (WebKit 20623.1.7.19.1)

I saw the announcement for Event Timing support in Safari Tech preview (https://webkit.org/blog/17447/release-notes-for-safari-technology-preview-229/).  Nice work!  I gave it a quick test locally.

I noticed when playing with this simple INP demo page (https://chrome.dev/inp-demo/) that the reported duration value for `keydown` captures the amount of time my finger is pressing the key until keyup.  Looking at the PerformanceEntry details, it seems the startTime, processing times, are correctly assigned, but the total duration value is too large-- suggesting that that the end time waits for the keyup event (potentially waiting for interactionID assignment?).

---

https://www.w3.org/TR/event-timing/#sec-performance-event-timing

duration
The duration attribute’s getter returns the difference between the next time the update the rendering steps are completed for the associated event’s Document after the associated event has been dispatched, and the startTime, rounded to the nearest 8ms.

---

https://www.w3.org/TR/event-timing/#sec-dispatch-pending

Step 6. For each pendingKeyDownEntry in the values from window’s pending key downs:

Set event timing entry duration passing pendingKeyDownEntry, window, and renderingTimestamp.

[Radar WebKit Bug Importer]

<rdar://problem/161911473>

[Franco Vieira de Souza]

Thank you for the bug report!

> suggesting that that the end time waits for the keyup event

This is spot on; duration of keydowns is currently being assigned when the entries are queued.

[Franco Vieira de Souza]

Pull request: https://github.com/WebKit/WebKit/pull/51788

[EWS]

Committed 302107@main (1337f64ad120): <https://commits.webkit.org/302107@main>

Reviewed commits have been landed. Closing PR #51788 and removing active labels.