Web Performance

Testing of User Timing

Jatinder: I recommend we move the tests from Intel to approved
... I'll look at the resource timing tests later
... I'll move the user timing tests

Resource Timing cross origin redirect

Jatinder: I'll respond to Pan on that. Didn't close on what to do yet. I think he has a point on per resource point

Navigation Timing 2

Jatinder: one of the goal is perceive latency and with pre-render, the latency changes
... should we add an attribute preRenderSwitch to NV2?
... today, we get duration. would be interesting to have the delta between loadEvent and preRenderSwitch
... it seems logical to me at least

James: sounds good to me. it ties into page visibility as well

Jatinder: we should make it clear that this is not real time performance but rather data to understand page load performance
... I'll follow up by mail
... we'll have to be careful on how preRenderSwitch will come off

Beacon API

Jatinder: sent mail about this regarding the relationship with XHR
... so we don't have to reinvent everything
... seems like there is interest
... I was hoping we can continue exploring, so let's not cut it from our list
... in a perfect world, I would love to have XHR take it on

Alois: yes, this would require some input from our side. let's try to reuse here
... hopefully it will get into XHR

Jatinder: it seems we'll need to provide guidance to them
... we'll send them a proposal

Prerender

Arvind: I started to write it

Jatinder: I have a bunch of questions and will hold until you come up with something

Arvind: two ways, link tag and when the browser does it (like typing in the address bar)
... I'm focusing with the link tag

Jatinder: yes, I don't think we should define heuristic indeeed
... should we say that "user agents can use heuristics" ?
... should we say something about prerendering https?

Arvind: we waited to avoid side effect

Daniel: do we have data on side effects?

Arvind: not really, I'll follow up

Jatinder: should we leave the spec open or not?
... also a performance impact. Should we put limits?
... it seems logical to one, rather than the whole result page

Arvind: ok, I'll keep working on it

Jatinder: if you can share the documentation you have, that would be nice

other specs

Jatinder: resource priorities, diagnostic

Page Visibility, RAF

Plh: we're on track

Milestones

Plh: are the milestones correct?

Arvind, Jatinder: yes

bandwidth testing issue

Jatinder: the nav timing is post processing, while bandwidth is for real time decision
... nav timing isn't a real time api

Daniel: I've been writing some thoughts and will send them along

Plh: should it go into XHR?

Daniel: up for discussion.

[adjourned]