[minutes] 2012-02-22 Web Performance WG Teleconference #62

Meeting Summary:



1.      Resource Timing Updates

Resource Timing processing model was updated to allow including a third party resource in the PerformanceResourceTiming buffer with the entire duration provided but individual timing attributes zero'd out. In particular, fetchStart and responseEnd attributes were not zero'd out.



2.      JavaScript Timing Proposal

The working group discussed this proposal in more detail this week. It appears that the JavaScript timing proposal and the User Timing features are two different models for gathering the same type of data; the former being a post-processing model where you get all information, and the latter being a run time processing model where you choose what data you want in advance. Further, both models require input from third parties origins from which they want additional script timing data; one requires a CORS like http header and the other requires third party to instrument their code. The working group feels that we do not need a new API that solves the same problem as an existing API in a different way. The working group has decided not to pursue this proposal.



3.      Page Visibility to CR

The working group has decided to remove the non-normative text suggesting User Agents can optionally disable the Page Visibility API with a User Agent setting.


Detailed Notes:



Web Perf Teleconference #62 2/22/2012



IRC log: http://www.w3.org/2012/02/22-webperf-irc


Meeting Minutes: http://www.w3.org/2012/02/22-webperf-minutes.html



Attendees

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

Philippe Le Hegaret, Jatinder Mann, Arvind Jain, Alois Reitbauer, James Simonsen, Tony Gentilcore


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

Meeting cancelled.



Scribe

Jatinder Mann



Contents

Agenda

1.       Discuss all status of all specifications



--------------------------------------------------------------------------------


Resource Timing

Jatinder: Zhiheng updated the Resource Timing processing model Step 8

https://dvcs.w3.org/hg/webperf/raw-file/tip/specs/ResourceTiming/Overview.html

Jatinder: To include cross-origin resource duration.

James: Changes look good.

Jatinder: I think that was the only item we had to discuss for Resource Timing.
JavaScript Timing

Jatinder: Did anyone have a chance to read Alois' proposal?

James: Yes, I did read it.

Jatinder: Thoughts on the proposal?

James: I think there is too much information here to expose to the web platform. This seems like something better suited for Developer Tools.

Alois: If we feel that there is too much information, we may want to reduce the amount of information.

James: I always felt that User Timing is the place to measure script timing.

Karen: What about adding a resource timing attribute for downloading and parsing script resources.

James: I think you can still get parse time using User timing.

Karen: You still can't get third party script.

Jatinder: Let's step back, what is the goal of this feature?

Alois: I want to find out how long third party script takes to download and run.
... Also, interested in script execution times...

<plh> I guess I'll have to check if we're ready to move to CR for Resource Timing

Jatinder: How does JIT compiling play into this?
... What about security risks?

James: You can find more information on where a user has been or if a user is logged in based on script execution time.

Jatinder: Seems like User Timing and JavaScript Timing are two different models for getting the same information. JavaScript Timing is post processing data - it gathers all data. User Timing is run time processing.

Alois: What about if I want to get Facebook timing or twitter timing?

James: Facebook can instrument their code with User Timign if they want you to get that timing information.

Jatinder: Even with a JavaScript Timing, third parties would need to give a CORS like header anyway. If they have to do that work anyway, they can just instrument their code with User Timing.
... Seems like these are two different models of the same thing.

Karen: Besides, we will need to make sure this new model is secure if we are to even consider it.

Jatinder: I think User Timing and the new proposal are two different approaches for doing the same thing. I don't think we would add a new API that does the same thing as an existing API in a different way.

James: I agree.
Page Visibility

Jatinder: Regarding Anne's comment on Page Visibility privacy - http://lists.w3.org/Archives/Public/public-web-perf/2012Feb/0042.html
... I think we are in agreement that we can remove the optional clause for a user agent to turning off the Page Visibility API.

Arvind: Agree that it isn't necessary to be spec'd out