[minutes] 2012-08-22 Web Performance WG Teleconference #79

Meeting Summary:



1.       Resource Timing - setResourceTimingBufferSize

The Resource Timing spec has two contradicting statements: in one section we state that when the setResourceTimingBufferSize is called to set the buffer size to a smaller size, the buffer should be truncated immediately, whereas another section mentions that the truncation does not occur until clearResourceTimings method has been called. There is consensus that we shouldn't truncate the buffer until after the clearResourceTimings method has been called. Jatinder will update the spec.



2.       Navigation Timing 2

Philippe has taken the action item to publish Navigation Timing 2 as a First Public Working Draft this week. The editors are planning on updating the spec with the radioAwakeStart and radioAwakeEnd optional attributes that will define the period of time when the radio is initializing. There has been feedback to the WG that during mobile web browsing, sometimes a significant portion of time may be spent turning the radio back on; without these attributes, the timeline isn't clear why the navigation was so delayed.



3.       Charter Updates

The WG is working on updating the WG Charter to include new specifications and ideas that the WG would like to pursue:



a.       HARS

Arvind has nearly finalized a draft of the HARS specification. As mentioned in an earlier meeting, HARS is the de facto standard for formatting performance metric data. We plan on updating the charter to include this specification and will publish this specification once the charter has been updated.



b.      Background Activity

For consideration in the new charter, Jatinder will share in the next two weeks the problem statement and potential solutions for improving power- and CPU-efficiency by minimizing browser background activity to the mailing list for discussion.



c.       Memory Leak

For consideration in the new charter, Tony will share in the next two weeks the problem statement and potential solutions for reducing memory leaks to the mailing list for discussion.



d.      Network Bandwidth

The WG agreed that real time analysis of network bandwidth was something best suited for other networking WGs to tackles instead. We will not be pursuing a charter update for this area.



e.      Data upload during unload

During TPAC 2011, there was discussion of how some web sites use the inefficient method of putting the browser in a busy loop during the unload in an effort to upload data to a server. For consideration in the new charter, Jatinder will share in the next two weeks the problem statement and potential solutions for solving this issue to the mailing list for discussion.



f.        Media Resource Timing

There had been a previous discussion of expanding video, audio and flash performance entries in Resource Timing from being single entries to more detailed entries for each byte range in the those different media types. We had discussed pursuing this as a part of Resource Timing 2 or a Media Resource Timing spec. Jatinder will send out updated text defining this problem space for inclusion in the charter.



4.       Velocity Europe

Alois will represent the Web Performance WG in a talk describing the WG at Velocity Europe this year.


Detailed Notes:



Web Perf Teleconference #79 8/22/2012



IRC log: http://www.w3.org/2012/08/22-webperf-irc



Meeting Minutes: http://www.w3.org/2012/08/22-webperf-minutes.html



Attendees

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

Jatinder Mann, Philippe Le Hegaret, Alois Reitbauer, Arvind Jain


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 and decide on expected behavior for setResourceTimingBufferSize method

2. Discuss any other additional Timing spec feedback

3. Discuss Charter updates and potential new specifications



--------------------------------------------------------------------------------
Resource Timing
Jatinder: We need to close on whether setResourceTimingBufferSize method should truncate the buffer or not. I'll follow up on the mailing list.
Navigation Timing 2
Arvind: Based on feedback, I have added an attribute for radioAwakeStart and radioAwakeEnd. These will be optional attributes. On mobile devices, the time spent waiting for the radio to be turned on during a navigation can be significant. I will submit the changes and Jatinder please also review.
Charter
Arvind: I have a draft prepared for the HARS spec and will be checking in the draft into the w3c respository soon.
Plh: We should update the charter to include the HARS specification with our next charter update.
... I'm going to take an action to put together a draft for the new charter.
Arvind: Jatinder, can you send out an email with the problem statement and potential solutions for the background activity this week? We can look at this idea and decide whether or not we want to include this in the next charter.
plh: Is Tony still interested in the memory leak API?
Arvind: He is out this week, but we should get him to also send out a problem statement and potential solutions for the memory leak API and the working group can analyze whether they want to take this or not in the next charter.
plh: Alois, what about the bandwidth API?
Arvind: Actually, I believe we discussed that this one should be better tracked in other network WGs, rather than this Web Perf.
plh: I will remove this from our list.
... The other ones is media resource timing. This came up in the discussion with NetFlix.
Arvind: Jatinder, do you recall?
Jatinder: There were two parts, one was whether we should provide additional data for video, audio and flash, like byte ranges or only a single entry. I believe this we are interested in doing in Resource Timing 2.
... The other point that Mark brought up was to use the Timing specs to make real time decisions, and we decided that was out of scope for this WG.
plh: Jatinder can you send out a sentence describing what this should look like in the charter?
Jatinder: Will do.
Alois: There was another issue that we wanted to have covered. That was with some sites using a busy loop on the unload to attempt to send up data. This sounds like a good opportunity for the charter.
Jatinder: Yes, I recall Jason Weber brought up this issue at TPAC. Let me respond with a problem statement and potential solutions to the mailing list for the charter.
plh: Sounds good.
... We also need to discuss testing of the specifications, however, I think that can wait for a bit.
Alois: There was one more issue that we needed to close. For Velocity Europe, we needed to choose a representative that would talk about the Web Performance WG.
... Since I will be at the conference I can present.
Plh: Sounds good.