[minutes] 2011-12-14 Web Performance WG Teleconference #56

Meeting Summary:



1.      Standardizing HAR Format

The working group discussed standardizing the HAR format. This format is currently defined here: http://www.softwareishard.com/blog/har-12-spec/. Per Jan and Steve, this format is commonly used by developer tools and performance analytics to store web performance data. Standardizing this format will allow a format that can easily be shared between different entities. The HAR format will include Performance Timeline data (Navigation, Resource and User Timing) and will take a dependency on those specs. None of the existing timing specs will take a dependency on HAR.



Jan Gunderson will present the working group with a draft of the new specification. The working group will work on updating the charter to include this new specification.



The working group does not want to expose a JavaScript API to auto-generate a file containing data in the HAR format, as currently the format contains data that can be a security or privacy risk. The goal in standardizing this format is to ensure that this data can be shared and used in an interoperable way. A site developer should be able to use a browser dev tool to get this data - however, this spec should not have a normative requirement for UAs Developer Tools.



2.      Timing Spec Action Items

Zhiheng and Jatinder are continuing to work on their Timing spec action items.



3.      Holiday Conference Call

The working group has decided to cancel the both December 28th conference calls and tentatively book the 10PST December 21st conference call. We shall reconvene our normal schedule in the new year. We will also discuss feedback on Alois' tool API on the January 11th conference call.



Happy Holidays!



Detailed Notes:



Web Perf Teleconference #56 12/14/2011



IRC log: http://www.w3.org/2011/12/14-webperf-irc


Meeting Minutes: http://www.w3.org/2011/12/14-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, Karen Anderson, Alois Reitbauer, James Simonsen, Zhiheng Wang, Steve Souders, Jan Gunderson, Michelle Park


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

Ganesh Rao, Jatinder Mann, Karen Anderson



Scribe

Jatinder Mann



Contents

Agenda

1.       Standardizing HAR Format

2.       Timing Spec Updates

3.       Holiday Conference Call Schedule



--------------------------------------------------------------------------------
HAR Format and Navigation Timing

http://lists.w3.org/Archives/Public/public-web-perf/2011Nov/0049.html

http://www.softwareishard.com/blog/har-12-spec/

Arvind: Does it make sense to standarize the HAR spec in the W3C?
... Another issue is that does it make for the browser to implement this

Jatinder: Can we get some background on when this was designed and how widely adopted is it?

Jan: Three years ago we created HAR format v1 to create a common template to send data up for many uses like analytics and test cases. We currently have HAR v1.2. HAR v1.3 will eventually include Navigation Timing.
... HAR currently has a <timings> tag.

Steve: I think its important to mention where HAR is supported today. Firebug, Chrome Dev Tools, IE9/10.

http://www.softwareishard.com/blog/har-12-spec/

James: Whats the action here?

Arvind: The questions are: Should the HAR spec be included in the W3C spec?

Jatinder: I think generally the HAR format sounds like a good idea. But I think our timing specs shouldn't take any dependency on a particular format.

Arvind: The two are independent. If anything, HAR format would take a dependency on the Timing specs, not the other way around.

<Karen_> I would want to ensure that if there are 2 mthods of getting the data that they can align together and there are no discrepancies

Steve: I would bet that over 50% of web pages storing performance data use the HAR format. The value here is that in can be interoperable between other tools.
... Another point is to bring a JavaScript API to output a HAR format

James: Would this be a W3C spec? I was under the impression that the W3C specs were browser related.

plh: The web is more than just the browser. We can include this in the W3 spec.

Jan will propose the HAR format in the Web Perf spec format.

Arvind: Seems like it doesn't make sense to expose this to JavaScript, as there may be privacy and security concerns. This is in fact mostly designed to be used by developer tools. Do you agree Jan?

Jan: Yes, I think this is for developer tools rather than from script.
Conference Call Schedule

Jatinder: Let's keep next week's meeting as tentative and cancel 12/28 meeting. We can reconvene in the new year.
... We can discuss Alois' topic of tool apis on the Jan 11 meeting.