Lack of Feature Parity with Performace.measure() in Level 2

[c3-ThomasHuo]

From the documentation of Performance.measure(), it seems like the startMark and endMark still depend on PerformanceTiming, which is clearly being depreciated. Is there a way to have it now default to PerformaceNavigationTiming properties instead?

[npm1]

Interesting observation! I think that we could change user timing to use PerformanceNavigationTiming instead of PerformanceTiming. It looks like the names used in one or the other are almost the same, or exactly the same? The exception would be startTime and duration, which are not names we'd want to special case and hence we'd need to exclude those (and perhaps any other attributes that are not applicable for Nav Timing).

[yoavweiss]

Looking at the list of names, it seems to be comprised of:

• navigationStart
• unloadEventStart
• unloadEventEnd
• redirectStart
• redirectEnd
• fetchStart
• domainLookupStart
• domainLookupEnd
• connectStart
• connectEnd
• secureConnectionStart
• requestStart
• responseStart
• responseEnd
• domLoading
• domInteractive
• domContentLoadedEventStart
• domContentLoadedEventEnd
• domComplete
• loadEventStart
• loadEventEnd

It seems like the only one that we cannot define by relying on the non-obsoleted parts of the spec is domLoading.

@c3-ThomasHuo - is your issue that this attribute would be missing? Or that we currently rely on parts of the spec we say are obsoleted? Or something else?

[nicjansma]

One of the discussions we've been having in the WebPerf WG lately has been around possibly having more than a singular NavigationTiming entry. For example, from BFCache or SPA Soft navigations.

If we do update UserTiming to be based on PerformaceNavigationTiming, we will want to consider how it would behave if there were more than one entry. (i.e. based on the attributes from the most recent entry?)

[yoavweiss]

Options to resolve this:

• Adapting the spec to point at the new definitions for most of these, other than domLoading
• Gather telemetry and see if this user-timing naming conventions is something we can remove without anyone noticing.

[yoavweiss]

On the second point, I filed https://bugs.chromium.org/p/chromium/issues/detail?id=1318445