Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss

Chrome's Interventions Quarterly Newsletter: 2016 Q2-Q3

97 views
Skip to first unread message

Chris Peterson

unread,
Aug 17, 2016, 1:21:29 PM8/17/16
to
Here is a status report from the Chrome team about their current and
future "interventions":

https://docs.google.com/document/d/1vqM_Dbi-V7LtdOwb2IGjgor9Fvl5k_FvO2DJGr4CaSQ/

An intervention is when a user agent decides to deviate slightly from a
standardized behavior in order to provide a greatly enhanced user
experience. An important part of every intervention is closing the
feedback loop and educating developers about the new behavior, so that
they can respond appropriately.

- Rendering pipeline throttling for offscreen cross-origin frames
- Disallow pop-ups from cross-origin frames on non-tap touch events
- Unintervention: re-instate autoplay for muted/silent videos
- Bail out on web fonts with an effectively slow connection
- Throttle the JS timers of offscreen frames
- Blocking perf-taxing document.write for users on 2G
- Scroll anchoring
- Disable the Vibrate API for cross origin frames
- Only allow navigations/popups from cross-origin frames on click
- History back that works!
- Ignore clicks on iframes that have moved in the last XXms
- Project OldSpice: taming annoyances with JS dialogs
- Make touchstart during a fling uncancelable
- Lower loading priority / or not loading offscreen cross origin iframes

Karl Dubost

unread,
Aug 17, 2016, 8:03:56 PM8/17/16
to Chris Peterson, dev-pl...@lists.mozilla.org
Chris,

Le 18 août 2016 à 02:21, Chris Peterson <cpet...@mozilla.com> a écrit :
> Here is a status report from the Chrome team about their current and future "interventions":
> https://docs.google.com/document/d/1vqM_Dbi-V7LtdOwb2IGjgor9Fvl5k_FvO2DJGr4CaSQ/
> An intervention is when a user agent decides to deviate slightly from a standardized behavior in order to provide a greatly enhanced user experience.

Very cool information. Two thoughts.

1. What is the status report about Firefox "interventions"
2. What are the impact of the Chrome "interventions" on Firefox Web compatibility (aka if they change the behavior and developers adjust to the new black, will it break in Firefox).

--
Karl Dubost, Mozilla
http://www.la-grange.net/karl/moz

Chris Peterson

unread,
Aug 17, 2016, 8:26:37 PM8/17/16
to Karl Dubost, dev-pl...@lists.mozilla.org


On 8/17/2016 5:03 PM, Karl Dubost wrote:
> Chris,
>
> Le 18 août 2016 à 02:21, Chris Peterson <cpet...@mozilla.com> a écrit :
>> Here is a status report from the Chrome team about their current and future "interventions":
>> https://docs.google.com/document/d/1vqM_Dbi-V7LtdOwb2IGjgor9Fvl5k_FvO2DJGr4CaSQ/
>> An intervention is when a user agent decides to deviate slightly from a standardized behavior in order to provide a greatly enhanced user experience.
> Very cool information. Two thoughts.
>
> 1. What is the status report about Firefox "interventions"

AFAIK, we don't have a coordinated plan for interventions. I know we
have already shipped some iframe and timer throttling and we are now
testing blocking of non-essential plugin content in Beta 49:

https://blog.mozilla.org/futurereleases/2016/07/20/reducing-adobe-flash-usage-in-firefox/


> 2. What are the impact of the Chrome "interventions" on Firefox Web compatibility (aka if they change the behavior and developers adjust to the new black, will it break in Firefox).

The following presentation from the Chrome team's BlinkOn 6 talks about
some of the behavior changes and some of their decision-making process
when designing interventions. One unfortunate example, advertisers
worked around Chrome for Android's "do not autoplay mobile videos"
intervention by deploying pseudo-video ads using canvas-based video
codecs, which was worse for bandwidth and power usage.

https://youtu.be/wQGa_6CRc9I


kenji...@chromium.org

unread,
Aug 17, 2016, 9:10:53 PM8/17/16
to
Hi!


On Thursday, August 18, 2016 at 9:26:37 AM UTC+9, Chris Peterson wrote:
> On 8/17/2016 5:03 PM, Karl Dubost wrote:
> > Chris,
> >
> > Le 18 août 2016 à 02:21, Chris Peterson a écrit :
> >> Here is a status report from the Chrome team about their current and future "interventions":
> >> https://docs.google.com/document/d/1vqM_Dbi-V7LtdOwb2IGjgor9Fvl5k_FvO2DJGr4CaSQ/
> >> An intervention is when a user agent decides to deviate slightly from a standardized behavior in order to provide a greatly enhanced user experience.
> > Very cool information. Two thoughts.
> >
> > 1. What is the status report about Firefox "interventions"
>
> AFAIK, we don't have a coordinated plan for interventions. I know we
> have already shipped some iframe and timer throttling and we are now
> testing blocking of non-essential plugin content in Beta 49:
>
> https://blog.mozilla.org/futurereleases/2016/07/20/reducing-adobe-flash-usage-in-firefox/
>
>
> > 2. What are the impact of the Chrome "interventions" on Firefox Web compatibility (aka if they change the behavior and developers adjust to the new black, will it break in Firefox).

We would consider this to be a bad intervention. If you can come up with such a scenario, please let us know. Probably, the WICG interventions github [1] would be the right place to discuss these.

Ultimately, we are hoping that if developers adjust then it would be for a better outcome both from a user experience viewpoint and for the web in general. As we've learned from the fallout after disabling autoplay on mobile (as you noted below), we need to proceed with great care.

[1]: https://github.com/WICG/interventions/issues

fun...@gmail.com

unread,
Aug 30, 2016, 1:12:59 PM8/30/16
to
> - Bail out on web fonts with an effectively slow connection

This is terrible and error prone for sure. Hopefully Mozilla won't follow suit.

0 new messages