Element Timing API

[digitarald]

Request for Mozilla Position on an Emerging Web Specification

• Specification Title: Element Timing API
• Specification or proposal URL: https://w3c.github.io/element-timing/
• Mozillians who can provide input (optional): @ecsmyth, @vickychin

Other information

A similar metric approach has been used in Firefox's performance automation https://bugzilla.mozilla.org/show_bug.cgi?id=1418368 .

[dbaron]

There may be some useful info in w3ctag/design-reviews#326 and w3ctag/design-reviews#395.

[martinthomson]

We just learned from Microsoft that this is shipping unprefixed in Edge and Chrome. We'd like to have a position soon.

[sefeng211]

I think we can resolve it as worth prototyping, cc @smaug---- @Bas-moz

[smaug----]

@emilio might have an opinion.
I didn't immediately see anything too harmful.

[emilio]

Yeah I don't see anything particularly problematic. I'm a bit confused about how nodes in shadow trees are intended to be handled.

https://wicg.github.io/element-timing/#get-an-element seems to try not to expose them (but seems like they should still generate an entry). But https://wicg.github.io/element-timing/#sec-element-processing doesn't handle shadow DOM and shouldn't even look at shadow trees. So something there is clearly confused.

[sefeng211]

Elements inside shadow trees are not exposed as the resolution of WICG/webcomponents#816. If the #sec-element-processing step doesn't handle shadow DOM, then no entries will be generated for nodes in shadow trees, right? And this should be the desired behaviour.

[Bas-moz]

Based on the earlier comments it seems the standard looks reasonable.

As to the extend of timing leakage, which was discussed, we can apply the appropriate amount of jitter/clamping as we do with many other timings, and this in itself does not make for a concern with the standard.

As per Sean's comment, the shadow tree issue has also been resolved in the original standard.

I suggest our position here should be positive.

[zcorpan]

Is there an issue for moving this spec into the Web Performance WG (or somewhere else)?

[Bas-moz]

See: https://lists.w3.org/Archives/Public/public-web-perf/2024Dec/0005.html

This now lives here: https://w3c.github.io/element-timing/

[keithamus]

bugzilla: https://bugzilla.mozilla.org/show_bug.cgi?id=1940239