WebPerf Group Call

30 Nov 2016

See also: IRC log


igrigorik, Todd, Shubhie, plh, NicJansma, yoav, xiaoqian


update on longtasks

shubhie: we want to ship v1 in M57 (early Jan)

_ one change we made is reporting a list of iframe attributes: name, id, source

_ security review: iterating on feedback, still an open discussion

_ attribution: plan is to show long scripts URLs

_ next steps: we'll update the explainer next week, ptal.

_ implementation: partial implementation in Canary behind experimental flag -- give it a try.

yoav: where are we with the spec?

shubhie: yeah, we'll work on expanding on explainer to cover processing / security, etc.
... AI: everyone to review explainer, we'll discuss how and where to expand on next call

Paint Timing API

Todd: there are lots of scenarios where main thread can cause compositor to run slower -- e.g. combination of webgl/canvas/svg bogs down; or an image decode appears to be done on main thread, but image shows up way later

Shubhie: do we have a sense for how common that is?

Todd: this is probably the long tail..
... we should probably think about this entire process as a two phase process: main thread > compositor

_ this is implementation specific, but at the same time, this is a common pattern for most browsers

yoav: what's the point of providing both.. the developer only cares about the longer time?

Todd/Shubhie: having two provides a good hint for where the problem may be

AI: let's document alternative models on GitHub and iterate there

Yoav: we need to start a thread WICG thread and we can transfer the repo

Resource Timing L2



AI: Todd will review


Todd: agree that we can't specify a MUST because we have different implementations; we can add a note in the spec, as in PR

_ there is a separate issue: we should gather some telemetry.. how often is this used? if we find that everyone relies on this, then perhaps we ought go back and change to MUST

igrigorik: we should also write some tests to figure out interop between existing implementations

AI: update note with a disclaimer, and in parallel we should open an issue to investigate current implementations and use in the wild


Todd: if we want to make a difference between first byte vs header complete vs ... those should be different attributes

_ if we want to break apart first byte, we're better of with different measurements

AI: update to align with receipt of first byte by the HTTP parser
... NT2 ...

Navigation Timing L2

Todd: it's helpful to be able to tell the tests apart; having a prefix would be helpful

plh: we're ok with the idea, but we do have tests that apply to both

_ it's ok to have tests that don't have a version number

Shubhie: do we duplicate tests?

Todd: implementations that don't have NT1 would want those


scribe: we'll merge the extended diagram, as we want to optimize for those reading the spec
... and, looks like Todd beat us to the punch :)

thanks all!

Summary of Action Items

Summary of Resolutions

[End of minutes]

Minutes formatted by David Booth's scribe.perl version 1.148 (CVS log)
$Date: 2016/12/02 12:14:12 $

Scribe.perl diagnostic output

[Delete this section before finalizing the minutes.]
This is scribe.perl Revision: 1.148  of Date: 2016/10/11 12:55:14  
Check for newer version at http://dev.w3.org/cvsweb/~checkout~/2002/scribe/

Guessing input format: RRSAgent_Text_Format (score 1.00)

No ScribeNick specified.  Guessing ScribeNick: igrigorik
Inferring Scribes: igrigorik

WARNING: No "Topic:" lines found.

Present: igrigorik Todd Shubhie plh NicJansma yoav xiaoqian

WARNING: No meeting title found!
You should specify the meeting title like this:
<dbooth> Meeting: Weekly Baking Club Meeting

WARNING: No meeting chair found!
You should specify the meeting chair like this:
<dbooth> Chair: dbooth

Got date from IRC log name: 30 Nov 2016
Guessing minutes URL: http://www.w3.org/2016/11/30-webperf-minutes.html
People with action items: 

WARNING: No "Topic: ..." lines found!  
Resulting HTML may have an empty (invalid) <ol>...</ol>.

Explanation: "Topic: ..." lines are used to indicate the start of 
new discussion topics or agenda items, such as:
<dbooth> Topic: Review of Amy's report

[End of scribe.perl diagnostic output]