Web Performance Working Group Teleconference

12 Feb 2014

See also: IRC log


Plh, JatinderMann, TobinTitus, rbarnes, Alois, AaronHeady, Ilya


Resource Priorities

Jatinder: Tobin will be making spec updates to Resource Priorities based on our TPAC action items. We should have updates in the next two week.


Jatinder: Looks like there are a few remaining action items on the Beacon spec, but otherwise we should be ready to take Beacon to last call.

Rbarnes: Looks like we're almost done on the remaining issues.

Jatinder: How about we fix the remaining issues within this week and aim to go to LAst Call in two weeks.

Rbarnes: I'm okay with going to last call.

<rbarnes> https://bugzilla.mozilla.org/show_bug.cgi?id=936340

Plh: Works for me.

Jatinder: Tobin and I will follow up with the open feedback this week.

Alois: We were looking at the beacon API, and looks like some of the error codes may need clarification. E.g., if you get a false, you'll have to strip out beacons or send multiple beacons.

Richard: We don't currently have a limit. We should look into that.

Jatinder: My expectations was that we would have a very large limit (yet to be determined), and the API would return false. So I don't expect most analytics to ever hit the limit.

Plh: I've heard feedback that some folks plan to send beacons as a background upload technique. There's another feedback to use this as a replacement for ping.

<igrigorik> mozilla bug: https://bugzilla.mozilla.org/show_bug.cgi?id=822480

Jatinder: I recommend we table this discussion until after we have telemetry if to show what the right limit should be. Let's track this as an open issue in the last call period.

Plh: Agreed

High Resolution Time Level 2

Plh:Should we move HRT L2 to LC?

Jatinder:There is only one piece of open feedback. There was feedback on specifying that Event should use DOMHighResTimeStamp. The DOM4 editor suggested this should be specified in the HRT L2 spec. I would expect that HRT L2 defines the DOMHighResTimeStamp, but others should use it. I am open to making the change if that’s what we want.

Plh:Let me review the change and follow up.

Extensive Specs

Jatinder:Thoughts on Anne’s feedback on our extension specs?

Plh:I read it as we should add hooks in the main spec that we’re trying to extend.

Jatinder:Like how HTML has a processing model hook that says run Page Visibility steps and points to the Page Visibility spec?


Jatinder:I’d love to get the recommendation on how we should do extension specs.

Plh:I’ll follow up.

Network Error Logging

Jatinder:I think we need to close on whether providing just the generics errors, like DNS error, is useful enough on its own? If so, we can at least spec that behavior pending a security review where we consider if we can do more specific error codes.

Aaron:From a Bing point of view, we’re happy with just the generic data. It would be more actionable to get the actual error code, but just learning that an error occurs is already useful information that normally may take longer to obtain.

Jatinder:Let’s talk to Arvind and see if we can make the change to move to generic error codes for now.

Plh:I talked to a privacy group within the W3C that are interesting in reviewing the error codes. Who wants to present to them?

Jatinder:Arvind and I would probably be interested.

Plh:I’ll setup the call.

Plh:Also, what do we think of Anne’s feedback to specify Resource Error Logging in Resource Timing L2?

Jatinder:All of the timing specs just add new objects that are placed on the same timeline. I guess the real question is do developers want to look at all resources at the same time, including successful and failed, or at the two lists separately?

Alois:We’d like to look at both.

Jatinder:I think its fine to add error data to Resource Timing. Today we only get successful Resources, adding the error Resources shouldn’t be a problem. We should run this by Arvind as well, see where he stands.

Older Specs

Plh:What is the status for some of our older specs. Namely, what are we doing on Resource Timing?

Jatinder:Spec is at CR, we still need to review the test cases. But I worry that spec won’t be able to advance until both IE and Chrome implement the new event we added to the spec. Do we know if Firefox is starting an implementation?

RBarnes:I believe there is a bug, but not sure about the status.

Jatinder:I think we need a renewed push to complete the test cases and complete our Resource Timing implementation. On the IE side, Tobin will be helping put some effort here.

Networking Hints

Ilya:I wanted to understand if anyone has feedback on the networking hints spec? I believe Mozilla has started to give some feedback.

Jatinder:I’d love to get the feedback on the mailing list. Tobin has started to look at this proposal and will give some feedback. I think that in the next few weeks if we feel that this is going to be something we want to pursue, we should move it to a spec and raise issues on the spec.

Jatinder:I recall that Arvind last year felt that prerender algorithm specifics should be left to the user agent and not be specified. Your proposal seems to change that position for prerender. Would love to see what Arvind thinks about these changes.


Plh:Head’s up that this is still on my list to determine. I have thread with Souders on the dates.

$Date: 2014-02-13 22:33:46 $