Intent to Deprecate and Remove: Calling PaymentRequest.show without user activation

[Stephen Mcgruer]

Contact emails

smcg...@chromium.org

Specification

https://www.w3.org/TR/payment-request/#show-method

Summary

Allowing PaymentRequest.show() to be triggered without a user activation could be abused by malicious websites. To protect users, the spec was changed to require user activation, and we are now following through in the Chrome implementation.

Plan is to deprecate in M98 and remove in M99. We may push the M99 date to M100 based on compat risk; see below.

Blink component

Blink>Payments

TAG review

N/A - enforcement of feature from an already-reviewed specification

TAG review status

Pending

Risks

Interoperability and Compatibility

Interoperability: no risk. Firefox has not shipped PaymentRequest at all, whilst Safari's implementation already requires user activation for calling show(). Compatibility: the main risk. If a website is calling PaymentRequest.show() without a user activation today, it will stop working. If that website doesn't have fallback code to use another payments flow, it may lead to a broken purchase experience for the user. Due to this risk, we added a UseCounter, kPaymentRequestShowWithoutGesture, which tracks use of the feature. Although hits on the UseCounter have reduced significantly since 2019*, there is still non-zero usage which is growing slowly over time. We believe the growth to be related to the general increase of web payments, rather than an expanded number of sites. To tackle the remaining usage, we have performed a UKM analysis, and identified the primary remaining site. We are in contact with them, and expect them to roll out a fix in the coming weeks - after which we will revisit the numbers and this thread. * https://chromestatus.com/metrics/feature/timeline/popularity/2398

Gecko: In development (https://bugzilla.mozilla.org/show_bug.cgi?id=1445138)

WebKit: Shipped/Shipping (https://bugs.webkit.org/show_bug.cgi?id=179056)

Web developers: No signals

Other signals:

Debuggability

As we are treating this as a deprecation, we intend to use the issues tab (as per the checklist) to warn developers of the upcoming removal. Once the support is removed, calling show() will throw a SecurityError with a clear error message.

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

Yes - https://wpt.fyi/results/payment-request/show-consume-activation.https.html?label=experimental&label=master&aligned

Requires code in //chrome?

False

Tracking bug

https://crbug.com/825270

Estimated milestones

Deprecate in M98, remove in M99 or M100 (compat risk depending).

Link to entry on the Chrome Platform Status

https://chromestatus.com/feature/5948593429020672

Links to previous Intent discussions

Intent to prototype: https://groups.google.com/u/1/a/chromium.org/g/blink-dev/c/2PhPgk_k9a0/m/alO4yt_HBQAJ
Intent to Experiment: https://groups.google.com/u/1/a/chromium.org/g/blink-dev/c/i6pAWsjU7zg/m/CzqgcGAXAwAJ

• This is a bit of a strange case, where we initially believed that we needed Capability Delegation to support deprecating this feature. However, the partner who needed that ability has instead solved their problem in a different way. As such, we believe it safe to require user activation for show() calls without Capability Delegation being available.

This intent message was generated by Chrome Platform Status and hand edited by smcgruer@.

[Yoav Weiss]

On Wed, Dec 1, 2021 at 4:43 PM Stephen Mcgruer <smcg...@chromium.org> wrote:

Contact emails

smcg...@chromium.org

Specification

https://www.w3.org/TR/payment-request/#show-method

Summary

Allowing PaymentRequest.show() to be triggered without a user activation could be abused by malicious websites. To protect users, the spec was changed to require user activation, and we are now following through in the Chrome implementation.

Plan is to deprecate in M98 and remove in M99. We may push the M99 date to M100 based on compat risk; see below.

Blink component

Blink>Payments

TAG review

N/A - enforcement of feature from an already-reviewed specification

TAG review status

Pending

Risks

Interoperability and Compatibility

Interoperability: no risk. Firefox has not shipped PaymentRequest at all, whilst Safari's implementation already requires user activation for calling show(). Compatibility: the main risk. If a website is calling PaymentRequest.show() without a user activation today, it will stop working. If that website doesn't have fallback code to use another payments flow, it may lead to a broken purchase experience for the user. Due to this risk, we added a UseCounter, kPaymentRequestShowWithoutGesture, which tracks use of the feature. Although hits on the UseCounter have reduced significantly since 2019*, there is still non-zero usage which is growing slowly over time. We believe the growth to be related to the general increase of web payments, rather than an expanded number of sites. To tackle the remaining usage, we have performed a UKM analysis, and identified the primary remaining site. We are in contact with them, and expect them to roll out a fix in the coming weeks - after which we will revisit the numbers and this thread.

Does the primary remaining site have fallback code, or will it be broken?

 

* https://chromestatus.com/metrics/feature/timeline/popularity/2398

Gecko: In development (https://bugzilla.mozilla.org/show_bug.cgi?id=1445138)

WebKit: Shipped/Shipping (https://bugs.webkit.org/show_bug.cgi?id=179056)

Web developers: No signals

Other signals:

Debuggability

As we are treating this as a deprecation, we intend to use the issues tab (as per the checklist) to warn developers of the upcoming removal. Once the support is removed, calling show() will throw a SecurityError with a clear error message.

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

Yes - https://wpt.fyi/results/payment-request/show-consume-activation.https.html?label=experimental&label=master&aligned

Requires code in //chrome?

False

Tracking bug

https://crbug.com/825270

Estimated milestones

Deprecate in M98, remove in M99 or M100 (compat risk depending).

Link to entry on the Chrome Platform Status

https://chromestatus.com/feature/5948593429020672

Links to previous Intent discussions

Intent to prototype: https://groups.google.com/u/1/a/chromium.org/g/blink-dev/c/2PhPgk_k9a0/m/alO4yt_HBQAJ
Intent to Experiment: https://groups.google.com/u/1/a/chromium.org/g/blink-dev/c/i6pAWsjU7zg/m/CzqgcGAXAwAJ

• This is a bit of a strange case, where we initially believed that we needed Capability Delegation to support deprecating this feature. However, the partner who needed that ability has instead solved their problem in a different way. As such, we believe it safe to require user activation for show() calls without Capability Delegation being available.

This intent message was generated by Chrome Platform Status and hand edited by smcgruer@.

--
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/CADY3Mae4RVpVxnjMS8oJ7WE7yOtAiqqa79%3D8v%2ByNf2XhCtHWgg%40mail.gmail.com.

[Stephen Mcgruer]

> Does the primary remaining site have fallback code, or will it be broken?

Yes and no :). It doesn't have automatic fallback for the specific payment method the user has selected (Google Pay), but the user could then select one of the other payment methods that the site supports (either a credit card flow or I think PayPal IIRC).

[Stephen Mcgruer]

To be clear; I think we have a good enough shot of that remaining site fixing their code 'soon' (I expect O(weeks)) that we both:

1. Shouldn't do the removal till they have, and

2. Don't need to provide an alternative in the form of capability delegation.

But the code change to at least start this deprecation would have to land by December 9th (or we punt for 1.5 months), hence why we're filing this ahead of them fixing their site :).

[Yoav Weiss]

LGTM1 to deprecate in M98 and remove in M99, assuming no surprises come up on the usage front.

[Chris Harrelson]

LGTM2

To view this discussion on the web visit https://groups.google.com/a/chromium.org/d/msgid/blink-dev/CAL5BFfU3ebwnoKvHPkXhQeSZ2mSfqgW_i_pXJVqEGaFjPJWWKA%40mail.gmail.com.

[Mike Taylor]

LGTM3

To view this discussion on the web visit https://groups.google.com/a/chromium.org/d/msgid/blink-dev/CAOMQ%2Bw-19DXQBytn%2BUChj%3D5p9JrgrhMZYGxVDYgkv262ttDkoA%40mail.gmail.com.

[Stephen McGruer]

Hey folks,

Following up here - we have determined that the remaining uses do necessitate making Capability Delegation available for web developers (see our Intent to Experiment - unfortunately our partner didn't engage at that time or we would have caught this earlier :(. )

We expect an Intent to Ship to be sent for Capability Delegation 'soon', targeting M100, and so are planning to push this deprecation out to M100 as well to align with that.

Thanks,

Stephen

LGTM2

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

To view this discussion on the web visit https://groups.google.com/a/chromium.org/d/msgid/blink-dev/CADY3Mae4RVpVxnjMS8oJ7WE7yOtAiqqa79%3D8v%2ByNf2XhCtHWgg%40mail.gmail.com.

--
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+unsubscribe@chromium.org.

To view this discussion on the web visit https://groups.google.com/a/chromium.org/d/msgid/blink-dev/CAL5BFfU3ebwnoKvHPkXhQeSZ2mSfqgW_i_pXJVqEGaFjPJWWKA%40mail.gmail.com.

--
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+unsubscribe@chromium.org.

[Stephen McGruer]

Hey all,

Unfortunately we've hit a snag in our deprecation; a partner has been having trouble integrating this change into their system, and though we are engaged in helping them we haven't made much progress yet.

As such, I'm currently requesting that we delay this deprecation until M102, to give us more time to help solve their problem before we require user activation. (I'm not sure how many LGTMs delaying a deprecation requires?)

Thanks,

Stephen

[Mike Taylor]

Hi Stephen,

Is there anything we can learn about their challenges that might apply to the broader ecosystem?

[Stephen Mcgruer]

> Is there anything we can learn about their challenges that might apply to the broader ecosystem?

A little, though largely it appears to be a bug in either their application or in Chrome (we're still trying to figure out which!). Simplifying, the problem is that they seem to be losing the Capability Delegation between click and (in a different iframe) the call to PR.show(), and it's quite tricky to debug this in a large async application. I can think of a few things that might help:

1. Adding capability delegation support to navigator.userActivation would likely be useful, e.g. exposing an array of capabilities currently active. This would make it much easier to quickly debug 'do I have a CD right here'. I hope the Capability Delegation folks might consider adding this! :)

2. Pausing user activation timeout when code execution in devtools is paused would be useful.

3. More generally (and more hand-wavingly), being able to more easily trace flows through async iframes 'somehow'. Devtools has some support for this, and it might just be user error that we and the partner are struggling, but when we're trying to answer questions like "Is it possible that this event flowed through an intermediary iframe that was created and destroyed again before this line of code executed", it can be tricky. 

LGTM2

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/CADY3Mae4RVpVxnjMS8oJ7WE7yOtAiqqa79%3D8v%2ByNf2XhCtHWgg%40mail.gmail.com.

--
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/CAL5BFfU3ebwnoKvHPkXhQeSZ2mSfqgW_i_pXJVqEGaFjPJWWKA%40mail.gmail.com.

--
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.

[Mike Taylor]

Thanks for the thoughtful answers!

LGTM1. I'll trust you to file bugs / feature requests for those 3 items (and yeah, 3 sounds like a useful, but hard problem to solve).

[Mustaq Ahmed]

I think [1] would be useful for developers but I see two blockers here: first we need to land the Capability Delegation patch in HTML  spec as a "reference point" for this idea, then the PR for navigator.userActivation needs to land too.

[Chris Harrelson]

LGTM2 for the extension to 102, but comments below. It would be very good to make progress on landing additional spec pieces.

On Tue, Feb 15, 2022 at 8:09 AM 'Mustaq Ahmed' via blink-dev <blin...@chromium.org> wrote:

I think [1] would be useful for developers but I see two blockers here: first we need to land the Capability Delegation patch in HTML  spec as a "reference point" for this idea, then the PR for navigator.userActivation needs to land too.

Hi Mustaq,

Is there anything blocking integrating the delegation patch into the HTML spec, and landing the PR for userActivation? There seems to be implementer interest from at least Gecko.

Chris

 

To view this discussion on the web visit https://groups.google.com/a/chromium.org/d/msgid/blink-dev/CAB0cuO4_9hPrmzJ2kw26iBzt09dSscvGY%3DsVNOBGeTQQmQ-7Ug%40mail.gmail.com.

[Mustaq Ahmed]

On Tue, Feb 15, 2022 at 11:16 AM Chris Harrelson <chri...@chromium.org> wrote:

LGTM2 for the extension to 102, but comments below. It would be very good to make progress on landing additional spec pieces.

On Tue, Feb 15, 2022 at 8:09 AM 'Mustaq Ahmed' via blink-dev <blin...@chromium.org> wrote:

I think [1] would be useful for developers but I see two blockers here: first we need to land the Capability Delegation patch in HTML  spec as a "reference point" for this idea, then the PR for navigator.userActivation needs to land too.

Hi Mustaq,

Is there anything blocking integrating the delegation patch into the HTML spec, and landing the PR for userActivation? There seems to be implementer interest from at least Gecko.

- For the Capability Delegation patch, yes we are already working with Gecko and will start working on an HTML PR soon (see its intent thread).

- The PR for navigator.userActivation still "needs implementer interest" I think, cc-ing dtapuska@ if I missed something.  (Note that this is separate from the "user activation v2" model which is already spec-ed.)

[Domenic Denicola]

On Tue, Feb 15, 2022 at 11:38 AM 'Mustaq Ahmed' via blink-dev <blin...@chromium.org> wrote:

On Tue, Feb 15, 2022 at 11:16 AM Chris Harrelson <chri...@chromium.org> wrote:

LGTM2 for the extension to 102, but comments below. It would be very good to make progress on landing additional spec pieces.

On Tue, Feb 15, 2022 at 8:09 AM 'Mustaq Ahmed' via blink-dev <blin...@chromium.org> wrote:

I think [1] would be useful for developers but I see two blockers here: first we need to land the Capability Delegation patch in HTML  spec as a "reference point" for this idea, then the PR for navigator.userActivation needs to land too.

Hi Mustaq,

Is there anything blocking integrating the delegation patch into the HTML spec, and landing the PR for userActivation? There seems to be implementer interest from at least Gecko.

- For the Capability Delegation patch, yes we are already working with Gecko and will start working on an HTML PR soon (see its intent thread).

- The PR for navigator.userActivation still "needs implementer interest" I think, cc-ing dtapuska@ if I missed something.  (Note that this is separate from the "user activation v2" model which is already spec-ed.)

At the last HTML triage meeting we got multi-implementer interest and agreement on navigator.userActivation. We still weren't clear on the use cases for the postMessage() parts of that PR. So, if you or someone else are willing to split out the PR into two pieces, we can land the navigator.userActivation piece quickly.

 

To view this discussion on the web visit https://groups.google.com/a/chromium.org/d/msgid/blink-dev/CAB0cuO6LSycEt_gm1VKHP-_VUgo-ri1x3Ux9f9jrzGaZufWr9g%40mail.gmail.com.

[Stephen Mcgruer]

Thanks folks for all the discussion here. We're close to branch point so I wanted to check - do I need, and may have I have, the third LGTM for a delay to M102?

[Daniel Bratell]

LGTM3 to delay.

/Daniel

To view this discussion on the web visit https://groups.google.com/a/chromium.org/d/msgid/blink-dev/CADY3MacxQgXZbh5v5x7wniH4zeW46iis_%2B3CtGG56xE3x5cNtQ%40mail.gmail.com.