Intent to Experiment: Reduce Accept-Language Origin Trial

[Victor Tan]

Contact emails

vict...@chromium.org, mike...@chromium.org 

Explainer

https://github.com/Tanych/accept-language

Specification

Variants header: https://datatracker.ietf.org/doc/html/draft-ietf-httpbis-variants-06

Summary

We want to reduce the amount of information the Accept-Language header exposes in HTTP requests and JS interface navigator.languages. Instead of sending all user’s Accept-Language, we only send the user’s most preferred language after language negotiation in the Accept-Language header. navigator.languages returns the same value as navigator.language during this experiment. 

We would like to run an origin trial for sites to opt into the Reduce Accept-Language origin trial to proactively test for breakage. See below for more details. 

Implementation Doc
https://docs.google.com/document/d/1RkPDf7DNtcOj4KXeW8wNCuYfto-drnGYST_NvZe3GoY

Blink component

Privacy>Fingerprinting

Risks

Interoperability and Compatibility

The compatibility risk is low since we're planning to reduce the amount of information in the Accept-Language header and navigator.languages, rather than remove the header or change value format in the header. Most existing Accept-Language detection code should continue to work. 

As for interoperability, no signal for other vendors. For multilingual sites to rely on the Accept-Language header, developers would need to depend on a user's full Accept-Language list for some browsers and a primary user's Accept-Language for others. 

Another signal is that the Chrome incognito model already reduced the Accept-Language header and JS interface navigator.languages to one language. The Accept-Language header can potentially expand to two if the first Accept-Language includes a region code, like en-US, the reduced Accept-Language  header will be en-US,en;q=0.9. 

Experiment Summary

The experiment is going to be a little different from a normal Origin Trial. The goal is enabling developers to test and ensure compatibility with our proposed changes. It’s incredibly important we give developers any chance to test systems at every level since this change represents vast dependencies on the introduced headers. 

As for enabling with the origin trial itself, there will be two components controlled by the same origin trial:

• Reducing the information in navigator.languages if the origin trial enabled.

• The Accept-Language HTTP request header contains the user’s primary preferred language, this can change if we detect a more preferred language during the language negotiation process. 

Because of the experimental nature of reducing Accept-Language, a valid origin token must be sent in the response header by origins which opt-in the origin trial. Also two new headers Variants (indicating sites supporting languages) accept-language and Content-Language need to be sent in the response header in order to make the language negotiation to work correctly.

Please see the design and implementation document for more information. 

Experiment Goals

The goal of this origin trial is to enable developers to test how reducing the Accept-Language request header and the JS getter navigator.languages will affect their systems, especially to understand the user cases on navigator.languages. We hope this can provide sufficient time for developers to test. We can validate our current plans for reducing Accept-Language and safely roll out them to the web based on their feedback.

We will be relying heavily on user and developer feedback to identify where breakage occurs,  or where use cases are not accounted for, especially for multilingual sites depending on the Accept-Language header, and navigator.languages.  We will create a GitHub repository and a public mailing list for gathering feedback. When the origin trial is ready, we plan to publish developer guidance on how to enroll and provide feedback.

Experiment Risks

There are some risks, as many multilingual sites have come to rely on the value in Accept-Language header and JS interfaces navigator.languages to send the right representation pages to the user.  Site breakage can take many forms, both obvious and non-obvious. However, since sites are in control of the Origin-Trial, Variants and Content-Language headers, a site can quickly opt out of the experiment when breakage is encountered.

Will this feature be supported on all six Blink platforms (Windows, Mac, Linux, Chrome OS, Android, and Android WebView)?

No (All but WebView)

Is this feature fully tested by web-platform-tests?

No (We fully test in browser_tests, WPT has limits to cover all the test cases in Accept-Language header).

Flag name

ReduceAcceptLanguageOriginTrial

Tracking bug

https://bugs.chromium.org/p/chromium/issues/detail?id=1306905

Launch bug

https://bugs.chromium.org/p/chromium/issues/detail?id=1307484 

Link to entry on the Chrome Platform Status

https://chromestatus.com/feature/5188040623390720

[Yoav Weiss]

How would the OT work for the Accept-Language values of the very-first request sent to the origin? Or are we expecting this request to send higher entropy, but not to hide potential breakage with later requests sending lower entropy?

--
You received this message because you are subscribed to the Google Groups "blink-dev" group.
To unsubscribe from this group and stop receiving emails from it, send an email to blink-dev+...@chromium.org.
To view this discussion on the web visit https://groups.google.com/a/chromium.org/d/msgid/blink-dev/CAJh4P7EvtPH_NQX_mJevEXu2fbePPQ7aYhfdBd%2BYB1J-5cn74g%40mail.gmail.com.

[Victor Tan]

> How would the OT work for the Accept-Language values of the very-first request sent to the origin? 

As described in the implementation doc, there are some limitations for the current OT architecture, we can't validate the response OT token before we send the request. 

For the very first request, we are still sending the full Accept-Language user's list, after we validate the response, all subsequent requests start to send a reduced Accept-Language header. 

Bests,

Victor

[Yoav Weiss]

That's fair. What is the experiment's timeline?

[Victor Tan]

We are expected to start in M109 Beta until 2023 Q2. We will document more in the web blog post. 

Bests,

Victor

[Yoav Weiss]

When do you expect the experiment to end?

[Victor Tan]

Sorry for the unclear end milestone, we would like to experiment from M109 to M115. 

Bests,

Victor

[Yoav Weiss]

Even though this is not a typical OT, I don't think it should go beyond the current policy of 6 milestone before showing significant progress towards shipping.

Would M109 to M114 (inclusive) be enough?

[Victor Tan]

Yeap, M109 to M104 inclusive is enough for this OT.

Bests,

Victor

[Yoav Weiss]

LGTM to experiment M109-M114

[Brian Birtles]

Hi,

Where should developers give feedback about this proposal?

The changes to the Accept-Language header are ameliorated by the Variants header. Is there any such mechanism proposed for the changes to the JS interface?

Thanks,

Brian

2022年10月28日金曜日 2:57:18 UTC+9 vict...@chromium.org:

[Mike Taylor]

Hi Brian,

Feedback as issues opened against https://github.com/Tanych/accept-language would be most welcome.

later,
Mike

--
You received this message because you are subscribed to the Google Groups "blink-dev" group.
To unsubscribe from this group and stop receiving emails from it, send an email to blink-dev+...@chromium.org.

To view this discussion on the web visit https://groups.google.com/a/chromium.org/d/msgid/blink-dev/bfe74326-8ee2-4b7f-87f9-17d642f5da7bn%40chromium.org.

[Victor Tan]

Hi Brian,

Currently, we don't have such a mechanism, but we can see how the origin trial goes regarding the JS interface.

Free feel to add your feedback to https://github.com/Tanych/accept-language once the origin trial is available. 

[Victor Tan]

Hi Will,

Please feel free to add your feedback and cases to https://github.com/Tanych/accept-language once the origin trial is available. Thanks.

Bests,

Victor

On Mon, Oct 31, 2022 at 10:00 PM Will Dillard <digit...@gmail.com> wrote:

You should keep all the languages in the Navigator even though it takes up more file space it helps people understand where they might want to navigate to and shows origin of speech which is not readily available it adds a nice touch and you should also include the narrator from Microsoft Edge or something like it. That's perspective and makes the software more diverse

--
You received this message because you are subscribed to the Google Groups "blink-dev" group.
To unsubscribe from this group and stop receiving emails from it, send an email to blink-dev+...@chromium.org.

To view this discussion on the web visit https://groups.google.com/a/chromium.org/d/msgid/blink-dev/CAJh4P7EvtPH_NQX_mJevEXu2fbePPQ7aYhfdBd%2BYB1J-5cn74g%40mail.gmail.com.

[Will Dillard]

You should keep all the languages in the Navigator even though it takes up more file space it helps people understand where they might want to navigate to and shows origin of speech which is not readily available it adds a nice touch and you should also include the narrator from Microsoft Edge or something like it. That's perspective and makes the software more diverse

On Thu, Oct 27, 2022, 12:57 PM Victor Tan <vict...@chromium.org> wrote:

[Jeffrey Yasskin]

I see that your Chrome Status entry says "Specification being incubated in a Community Group", but your explainer is not hosted by a CG, and the specification you cite is 1) in a WG and 2) not the specification you'll need to write for this to be a web feature. Could you work on migrating this to a CG? And now that you've started an origin trial, it's also time to start working on your web-side specification, which will probably eventually merge into Fetch. Let me or your spec mentor know if you need help with that.

Thanks,

Jeffrey

[Victor Tan]

Hi Jeffrey,

Thanks for your information. I updated the Chrome Status entry to correctly represent current status. 

I will work on spec and migrate the explainer to CG if needed. 

Bests,

Victor

[Victor Tan]

A blog post just went out for this Origin Trial: https://developer.chrome.com/blog/origin-trial-for-accept-language-reduction/

To unsubscribe from this group and stop receiving emails from it, send an email to blink-dev+unsubscribe@chromium.org.