[minutes] 20160504 Web Performance

Available at

               Web Performance Working Group Teleconference

04 May 2016

    See also: [2]IRC log

       [2] http://www.w3.org/2016/05/04-webperf-irc


           Yoav, Ilya, Plh, Todd, wesleyhales





      * [3]Topics
          1. [4]Charter
          2. [5]Resource Timing
          3. [6]"local" vs "remote" performance API's
          4. [7]PV 2
          5. [8]rIC
          6. [9]Resource Hints
          7. [10]nav timing
          8. [11]polaris
      * [12]Summary of Action Items
      * [13]Summary of Resolutions

Charter updates

    * keep Preload, Resource Hints, and rIC as part of the webperf

    * move Server Timing, Frame Timing, and Reporting + NEL to WICG

    --> [14]https://w3c.github.io/charter-webperf/ draft charter

      [14] https://w3c.github.io/charter-webperf/

    Ilya: we have some implementations for the first set, and none
    for the second, thus the proposal

    Yoav: makes sense to me
    ... Frame Timing is a bit more advance

    Ilya: yes and no, but we keep changing subtantially and no
    update on the implementation side. could be convinced either

    Todd: we should ensure that we publish current work as Notes
    before we drop from the charter
    ... we could also allow folks from the WICG to contribute

    Yoav: only thing is that the work on server timing wouldn't
    move forward

    Ilya: so, transfer ownership of the repo of WICG and publish as
    Notes, and then readopt them in the future if needed

    plh: ok, so server timing, reporting, NEL. what about frame

    todd: don't foresee us implementing unless there is incredible

    ilya: ok, then I think we'll keep working on the use cases and
    current design isn't the best
    ... so it's reasonable to move to WICG

    <scribe> ACTION: plh to publish Frame Timing, Server Timing,
    Reporting, NEL [recorded in

      [15] http://www.w3.org/2016/05/04-webperf-irc]

    <trackbot> Created ACTION-169 - Publish frame timing, server
    timing, reporting, nel [on Philippe Le Hégaret - due

    (as Notes)

    <igrigorik> notes:


    Ilya: we added some estimates as well

Resource Timing L1

    [17]V1 draft
     1. Make sure we looked at
        [18]https://w3ctag.github.io/security-questionnaire/ and
        have answers in the spec for all of the relevant questions.
     2. Get a review or an ack from webappsec
     3. Transition to CR

      [17] https://rawgit.com/w3c/resource-timing/V1/index.html
      [18] https://w3ctag.github.io/security-questionnaire/

    igrigorik: Make sure to address L1 comments:


    <scribe> ACTION: plh to make sure the draft is up-to-date
    [recorded in [20]http://www.w3.org/2016/05/04-webperf-irc]

      [20] http://www.w3.org/2016/05/04-webperf-irc]

    <trackbot> Created ACTION-170 - Make the draft is up-to-date
    [on Philippe Le Hégaret - due 2016-05-11].




      [22] https://github.com/w3c/web-platform-tests/pull/402

    <igrigorik> deep link:


    plh: will keep cranking the wheel on that one

    Yoav: I'd like to use resource timing to test the preload
    ... implementation is flaky at the moment
    ... if I get to a stable implementation, I might try to have it
    ship in webkit

    Ilya: for L2, the issues around resources that don't return
    responses, I summarized the status on the chrome bug and
    started a thread with security and privacy team at Google

"local" vs "remote" performance API's

    Ilya: got some good feedback
    ... should we publish as a Note

    Todd: we should
    ... only downside is that we'll get feedback :)

    Ilya: we're having those discussions anyway

    RESOLUTION: let's publish those best practices as a Note

    <scribe> ACTION: plh to create a repo for best practices
    document [recorded in

      [24] http://www.w3.org/2016/05/04-webperf-irc]

    <trackbot> Created ACTION-171 - Create a repo for best
    practices document [on Philippe Le Hégaret - due 2016-05-11].

    Ilya: I'll transform into respec

PV 2

    Ilya: having a change section would be good

    plh: I'll make a PR for that
    ... and keep into iterating on the tests


    <igrigorik> riC test:

      [25] https://github.com/w3c/web-platform-tests/pull/2163

    Ilya: we may have to add some additional tests

    plh: i'll add that to my TODO list

    Ilya: I need to add a privacy section
    ... we kind of have one but need to pull it out

    Todd: I touched based with Justin on that front
    ... one concern is that Chrome pipeline becoming the pipeline
    for developers
    ... the spec is overly specific sometimes. I need to check on

    Ilya: concept of idle time is user agent defiend

    Todd: main problem comes from layout
    ... so ordering of events and when layout happens
    ... trick is that guidance shouldn't be to a specific engine
    ... so I need to figure this out
    ... will look over it next week or two

    next steps:
     1. update security/privacy, wait for Todd within 2 weeks
     2. then wide review from security/privacy
     3. then move to CR

Resource Hints

    <igrigorik> html updates:

      [26] https://github.com/whatwg/html/pull/1141

    Yoav: Domenic is doing some cleanups related to HTML
    ... and adding the link relations to HTML itself
    ... and ask if the relations apply only to link or also a and
    ... I see a weak use cases for a and area
    ... I don't think they're implemented anywhere else, except for

    Ilya: context is that prefetch was supported for a and area and
    domenic was asking for others
    ... I don't see a huge value so I'm ok to remove those
    ... but not even sure on prefetch for a and area

    Yoav: ok, so next step is testing to see who supports what

    Todd: don't know for Edge

    Ilya: we can write a quick test and see what happens
    ... I found that prerender has some issues in Edge for example
    ... in all scenarios
    ... it's tricky to get some tests since we get no clear signals
    ... looking at TCP dumps to verify that it's working

    Todd: e.g. dns-prefetch

    plh: would be good to send feedback to WebDriver folks btw

nav timing

    Ilya: L1 defines its own time origin because we didn't have the
    current def
    ... we're duplicating for historical reasons


      [27] https://www.w3.org/TR/navigation-timing-2/#historical

    plh: maybe change the styling of that section?

    Ilya: or remove it and link to L1

    <igrigorik> sorry guys, need to drop off, dialing back in..

    "All time values defined in this appendix are measured in
    milliseconds since midnight of January 1, 1970 (UTC)."



      [28] http://web.mit.edu/ravinet/www/polaris_nsdi16.pdf

    plh: I'll invite them for a presentation.

    Next call is in 2 weeks.

Summary of Action Items

    [NEW] ACTION: plh to create a repo for best practices document
    [recorded in [29]http://www.w3.org/2016/05/04-webperf-irc]
    [NEW] ACTION: plh to make sure the RT L1 draft is up-to-date
    [recorded in [30]http://www.w3.org/2016/05/04-webperf-irc]
    [NEW] ACTION: plh to publish Frame Timing, Server Timing,
    Reporting, NEL as Notes [recorded in

      [29] http://www.w3.org/2016/05/04-webperf-irc
      [30] http://www.w3.org/2016/05/04-webperf-irc
      [31] http://www.w3.org/2016/05/04-webperf-irc

Summary of Resolutions

     1. [32]let's publish the best practices as a Note

    [End of minutes]

     Minutes formatted by David Booth's [33]scribe.perl version
     1.147 ([34]CVS log)
     $Date: 2016/05/05 01:14:34 $

      [33] http://dev.w3.org/cvsweb/~checkout~/2002/scribe/scribedoc.htm
      [34] http://dev.w3.org/cvsweb/2002/scribe/

Received on Thursday, 5 May 2016 01:16:03 UTC