W3C home > Mailing lists > Public > public-web-perf@w3.org > September 2016

[minutes] Web Performance September 14

From: Philippe Le Hégaret <plh@w3.org>
Date: Wed, 14 Sep 2016 16:05:30 -0400
To: public-web-perf@w3.org
Message-ID: <4efe9988-be53-8fe8-8a62-54a703efaf9d@w3.org>
Available at
   https://www.w3.org/2016/09/14-webperf-minutes.html

Text version:

                             Web Performance

14 Sep 2016

    [2]Agenda

       [2] 
https://lists.w3.org/Archives/Public/public-web-perf/2016Sep/0001.html

Attendees

    Present
           Todd, Ilya, Shubie, Nate, Yoav, Ben, Plh

    Regrets
           Wesley

    Chair
           Ilya

    Scribe
           plh

Contents

      * [3]Topics
          1. [4]RT Level 1 forward?
          2. [5]Long Tasks API and msFirstPaint
          3. [6]getEntries
          4. [7]TPAC
      * [8]Summary of Action Items
      * [9]Summary of Resolutions
      __________________________________________________________

    <wesleyhales> thanks guys

RT Level 1 forward?

    see
    [10]https://lists.w3.org/Archives/Public/public-web-perf/2016Se
    p/0000.html

      [10] 
https://lists.w3.org/Archives/Public/public-web-perf/2016Sep/0000.html

    igrigorik: do we have enough coverage and are we ok moving
    forward?
    ... didn't a chance to dig, but otherwise, it looks good

    Todd: I think the tests are sufficient

    plh: I wasn't able to run the tests on safari

    yoav: there is support behind a flag. I can run them.

    ilya: I'm comfortable moving forward

    RESOLUTION: RT Level 1 to Proposed Recommendation

    ilya: there is an [11]old PR for resource timing

      [11] 
https://github.com/w3c/web-platform-tests/pulls?q=is:open+is:pr+label:resource-timing

    plh: i'll look into
    [12]https://github.com/w3c/web-platform-tests/pull/402

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

Long Tasks API and msFirstPaint

    Ilya: any feedback from MS on this front?

    Todd: we looked at msFirstPaint and its implementation. not the
    best place at the moment.
    ... it's the first we trigger all of the work we need to do the
    first paint. could include layout, some scriptings,
    rasterization, etc.
    ... so we don't think it's the best representation
    ... we haven't yet completed the conversation. next would be
    TPAC.
    ... we're looking at long tasks and are still in favor for it

    Ilya: ok, I'll reserve time at TPAC for the long task
    discussion

    shubhie: we have a proposal for the V2 attribution
    ... I'll send an email about that

    Todd: 50ms constant might be too low for today's Web

    shubhie: it came up before indeed. we can discuss that.

    Todd: so just tweaks at the moment. nothing blocking.
    ... time should be as close to the rasterization as possible
    ... but we all have incentive to cheat on that :)
    ... we haven't settled yet

    Nate: similar concerns around attribution and long task

    shubhie: we can keep iterating on that

    Nate: we want to measure if the engine is taking a really long
    time

    Ben: [...]

    Nate: this API is helping for the developer for debugging
    performance issues.

    shubhie: are folks still on board with firstPaint and
    firstContentPaint?

    Todd: not sure yet
    ... biggest concern is firstContentPaint is significant yet

    Ben: not sure if that's something that I'd be able to expose to
    developers. It's an important idea but maybe through the
    Hero(?) API

    Todd: beyond firstPaint, it seems arbitrary

    Yoav: having an API to allow the developer to say what they
    want to say is important, but deployed content doesn't have
    that. how do we target the content in the wild? those don't
    have to be exclusive.

    Ben: we could come up with a metric that'd be relevant for a
    lot of sites.

    Yoav: third party analytics can't change the content to
    annotate it

    Ilya: I'm hearing agreement for simple heuristic metrics, as
    well as more sophisticated ones

getEntries

    <igrigorik>
    [13]https://github.com/w3c/performance-timeline/issues/57#issue
    comment-246773875

      [13] 
https://github.com/w3c/performance-timeline/issues/57#issuecomment-246773875

    [14]https://www.w3.org/2016/09/pt-naming.html

      [14] https://www.w3.org/2016/09/pt-naming.html

    plh: current results are a little over the map, see results in
    [15]https://www.w3.org/2016/09/pt-naming.html

      [15] https://www.w3.org/2016/09/pt-naming.html

    Yoav: don't have a strong opinion there. this seems related to
    the value of as attribute on preload
    ... which was case sensitive and doesn't seem consistent with
    the rest of the platform

    <yoav> [16]https://github.com/whatwg/html/issues/1665

      [16] https://github.com/whatwg/html/issues/1665

    plh: I don't think we need to be consistent since it's not an
    html attribute. it's problematic that
    getEntriesByType('RESOURCE') would return nothing

    Yoav: we shouldn't make a decision here without a bigger
    context

    Todd: agreed.

    plh: I could look for history
    ... an other solution is to do case-by-case basis ...

    Yoav: nope :)

    plh: I'll provide more info for TPAC

TPAC

    <igrigorik>
    [17]https://docs.google.com/document/d/1U2DfWlToLlDoJyGvDZytM6V
    M2p67qisuCOfgnWIclzw/edit#heading=h.cqwljymoo9m

      [17] 
https://docs.google.com/document/d/1U2DfWlToLlDoJyGvDZytM6VM2p67qisuCOfgnWIclzw/edit#heading=h.cqwljymoo9m

    [Ilya goes through the proposed agenda]

    Ilya: any preferences for time?

    [none heard]

    Todd: memory notification api?

    Ilya: we have some notes on that from the f2f. I'll dig those.

    [adjourned]

Summary of Action Items

Summary of Resolutions

     1. [18]RT Level 1 to Proposed Recommendation

    [End of minutes]
Received on Wednesday, 14 September 2016 20:05:35 UTC

This archive was generated by hypermail 2.3.1 : Wednesday, 14 September 2016 20:05:36 UTC