[minutes] 2012-03-14 Web Performance WG Teleconference #65

Meeting Summary:



1.    Navigation Timing

The working group expects Navigation Timing to go to PR at the end of the month as WebIDL appears to be heading towards CR during that time period.



2.    Resource Timing

Aside from updating the Resource Timing diagram, the editors have made all Last Call feedback changes. The editors will update the diagram this week. Seeing that there is no additional changes to make to this spec, the working group would like to move this spec to CR.



3.    User Timing

Per discussion, the working group would like add the ability to share User Timing data to third party scripts via the timing-allow-origin http response header. Seeing that this change is significant enough to merit feedback from the wider security audience, this spec will go to Last Call after the editors make this change.



4.    Performance Timeline

There is no open Last Call feedback issues on this spec. The working group would like to move this spec to CR.



5.    High Resolution Time

This specification has moved into the Last Call phase. The working group will monitor all feedback to this spec.



6.    Page Visibility

The spec editors have been getting feedback suggesting the proposed firing of the visibilitychange event during unload is redundant with the unload event and may impact navigation performance by making more script run during the navigation away period. The editors will discuss this topic on the mailing list.



7.    requestAnimationFrame

There was a discussion on including the animationStartTime attribute to ensure better synchronization of multiple animations. A mailing list thread is to come.


Detailed Notes:



Web Perf Teleconference #65 3/14/2012



IRC log: http://www.w3.org/2012/03/14-webperf-irc


Meeting Minutes: http://www.w3.org/2012/03/14-webperf-minutes.html



Attendees

Present for Navigation Timing, Resource Timing and User Timing (4-5PM EST/1-2PM PST)

Jatinder Mann, Philippe Le Hegaret, Arvind Jain, Tony Gentilcore, Zhiheng Wang, Joseph Scheuhammer


Present for Page Visibility, Efficient Script Yielding, Display Paint Notifications (4-5PM EST/2-3PM PST)

Meeting cancelled.



Scribe

Jatinder Mann



Contents

Agenda

1.     Review feedback on all specifications

--------------------------------------------------------------------------------
Resource Timing
Jatinder: Zhihang and I have responded to all of the last call feedback on the mailing list.
Zhiheng: Jatinder, can you check in the resource timing diagram? I need the diagram to make changes to it.
Jatinder: Will do.
plh: I will review the Resource Timing LC feedback and make sure nothing was missed.
Jatinder: Once we have made the diagram change, I will email you to let you know that we are clear to move forward to CR.
User Timing
Jatinder: As I was responding to the User Timing feedback to reword the privacy section, I noticed that the current privacy section suggests that the getMarks and getMeasures functions will return empty Arrays when invoked from scripts originating from different origins and the PerformanceMark and PerformanceMeasure interfaces will zero out data for the startTime and duration attributes when invoked from scripts originating from different origins. As th
... Do we want to allow sites to be able to allow cross-origin scripts from accessing this information with a CORS-like HTTP header, like we did for Resource Timing?
Tony: Yes, that seems reasonable. Without providing the ability to make this information available cross-origin, the benefit of the feature might be limited.
Jatinder: Especially in scenarios where a site has many sub-domains.
... I will update the spec to include the CORS like header. Do we want to use the timing-allow-origin header for both resource timing and user timing or give them unique headers, so a person can choose to offer one or the other?
Tony: I don't think that degree of granularity will be necessary. If you trust a site to share one metric, you should be able to share the other metric.
Jatinder: I think its reasonable to have one header.
plh: If we make such a change, we will want security folks to review it. I recommend we move the spec back to last call to make sure this change gets reviewed.
Performance Timeline
Jatinder: We expect that Performance Timeline and Resource Timing will move towards CR, considering there are no open items remaining. User Timing and High Resolution Time are currently in LC, and if there is limited feedback, these should move along to CR shortly as well.
Page Visibility
Jatinder: I had been chatted with developers regarding the Page Visibility spec, and I had recieved feedback on the firing of visibilitychange during the unload. Most developers were worried that this code is just redudant with unload and that we will just be introducing more code that will run on the unload slowing down page navigations. If the feedback was to tie this in closer with session history, we may do that.
Arvind: Yes, I was reluctant with firing the event during unload as it is redundant. I am not as familar with session history. We should ask Boris of his motivation for this change.
Jatinder: I will start a mail thread.
requestAnimationFrame
Jatinder: I had some feedback regarding the animationStartTime attribute mentioned in the editor's note in the spec. I will send mail on it to make sure the editor's see it.