Intent to Continue Experimenting: Digital Goods API

[Glen Robertson]

Contact emails

mgi...@chromium.org, gle...@chromium.org, rou...@chromium.org

Explainer

https://github.com/WICG/digital-goods/blob/master/explainer.md

Specification

None

Design docs

https://github.com/WICG/digital-goods/blob/master/explainer.md

https://docs.google.com/document/d/1Jbt2Mzt-xg1cWVlFScBQsoX_pE8Kg1gYpulxUSV8FM0/edit

Summary

An API for querying and managing digital products to facilitate in-app purchases from web applications, in conjunction with the Payment Request API (which is used to make the actual purchases). The API would be linked to a digital distribution service connected to via the user agent. In Chromium, this is specifically a web API wrapper around the Android Play Billing API.

Blink component

Blink>Payments

Search tags

payments, billing

TAG review

https://github.com/w3ctag/design-reviews/issues/571

TAG review status

In progress

Risks

Interoperability and Compatibility

Similar to Payment Request: this API is used to talk to specific store backends, and so its usage is tailored to the specific store. The reason it's a proposed web standard is so that the same code (which is specific to one store) is portable across browsers.

Gecko: No signal (https://github.com/mozilla/standards-positions/issues/349). Awaiting feedback.

WebKit: No signal

Web developers: Positive (https://discourse.wicg.io/t/proposal-web-payments-digital-product-management-api/4350)

Ergonomics

Used in tandem with the Payment Request API.

Goals for experimentation

- General API design. Determine whether developers need to access more data that would be exposed through the Play Billing API but is not exposed through our web API.

- Specifically, we would like to know whether the API is suitable for abstracting over other non-Play stores. While running an experiment with the current implementation won't tell us this, it will set up real-world clients and we can then try their sites on other implementations.

Experimental timeline

- M87 (2020-11-17): Experiment begins

- M90 (2021-04-13): Original experiment end date

- M93 (2021-08-31): Extended experiment end date

Reason this experiment is being extended

An origin trial ran from M87 to M90 and found some areas of developer friction and new features needed (see bugs labeled DGAPI). We haven't yet had time to fix all the issues and update the API. We are planning to update the API and run a next phase of the experiment with a v2 API soon. We would like to avoid stopping the experiment in between the phases to avoid unnecessarily disrupting current users of the API while we work on the next iteration.

Ongoing technical constraints

None

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

No, Android and Chrome OS only (the two platforms where we have Play Store integration).

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

No

Flag name

None

Tracking bug

https://crbug.com/1061503

Launch bug

https://crbug.com/1017947

Link to entry on the Chrome Platform Status

https://chromestatus.com/feature/5339955595313152

Links to previous Intent discussions

Intent to prototype: https://groups.google.com/a/chromium.org/g/blink-dev/c/vkS3k30lWNs

Intent to Experiment: https://groups.google.com/a/chromium.org/g/blink-dev/c/syI9_M9dANY/m/3lt-QGMHAgAJ

This intent message was generated by Chrome Platform Status.

[Alex Russell]

Hey all,

Met today at API OWNERS to discuss. I don't think folks are necessarily opposed to extending this, but a few things jump out:

• 8 milestones as at the far end of how long we'd like any OT to run
• Given that, it would be helpful to summarize both a plan for what the new API will look like as well as lessons learned to date

I know that some of that is captured in the DGAPI bugs linked, but it would help if there were a summary document that described the developer feedback to date, the changes you'll be making to the API to compensate, and the plan for (hopefully) validating the new API works as developers expect quickly, then launching or sunsetting.

Thanks

[Matt Giuca]

Hi Alex,

Thanks for the notes from your meeting. I think we can create a summary of the proposed design changes (however, note that it's somewhat undecided at this stage, what specific changes will need to be made). I don't want to block the extension of the origin trial on having a design proposal, since that could jeopardize customers' ability to use the API.

Does the 8-milestone run include the planned future "V2 API" origin trial, (i.e. if we run this for another 3 milestones, we'll only have 2 milestones left to do a V2 origin trial)? Or do you mean 8 milestones without making changes to the API.

I think the latter would be pretty reasonable, but as I'm sure I've said before, I'm not too comfortable having a hard deadline on having to ship a new API or losing it. Having hard deadlines for shipping features, especially one this complex, generally results in rushed and buggy experiences. If we, say, get to M95 and have a working implementation for V2, but discover a bug at the last minute, I want to be able to have no hesitation to punt the release for an additional milestone (I don't want us to feel compelled to push it out the door because of an arbitrary 8-milestone limit).

I'll note that I raised this issue for this specific API in a general discussion with Blink API Owners in November 2020 (email). Relevant quotes from myself from that email:

Until this work is done, we're not comfortable shipping this API. We think backwards-incompatible changes may be required. But we want to get it into the hands of developers before then. An origin trial lets us do that, but it's got a soft time limit (3 milestones), after which time we'll have to apply for an extension. Maybe that's fine, because we have specific questions we'll still want answered. But it seems a little unfit for purpose, because we aren't using the OT to answer those questions, we're just extending it to keep things working while we do design work behind the scenes.
...
If API owners are happy with us ... repeatedly extending the origin trial, without specific experimental questions, until we have done the internal research / spec work required to ship, then I think we can do without origin keys. My proposal was essentially an attempt to formalize that behaviour, rather than having to apply for all of these exemptions and extensions when they come up.

Now I didn't get a response to that email at the time. The context was that we were asking for a formal "origin keys" system, and in that meeting, I was told that it wouldn't be necessary because origin trials are suitable for that purpose. Per my final paragraph there, it may not be that origin trials are suitable for this purpose, if in practice API owners are going to block origin trial extensions.

To be clear, we aren't intending to extend this indefinitely. I am hoping we can have this API shipped some time this year. But I am flagging that we can't guarantee that we'll have the necessary changes made by a specific milestone, and I don't want to be hard-bound to make those changes by a specific date.

Thanks

Matt

[Yoav Weiss]

Hey Matt,

The main goals for the duration restrictions that Alex mentioned is to avoid cases where Origin Trials are being used as a "soft launch" mechanism and to avoid prematurely baking in the feature into the platform.

Significant changes to the API shape (as the ones you allude to with the V2 API) would provide some assurances against bake in, and from my perspective would "restart the clock".

However, that doesn't mean that breaking the API every X milestones would enable y'all to have it in OT forevermore. We also want to see that the OT is being used to provide us with meaningful feedback. I believe that's the reason Alex asked for a summary of lessons learned, as well as a plan to eventually ship this.

On Fri, May 7, 2021 at 9:37 AM Matt Giuca <mgi...@chromium.org> wrote:

Hi Alex,

Thanks for the notes from your meeting. I think we can create a summary of the proposed design changes (however, note that it's somewhat undecided at this stage, what specific changes will need to be made). I don't want to block the extension of the origin trial on having a design proposal, since that could jeopardize customers' ability to use the API.

Does the 8-milestone run include the planned future "V2 API" origin trial, (i.e. if we run this for another 3 milestones, we'll only have 2 milestones left to do a V2 origin trial)? Or do you mean 8 milestones without making changes to the API.

I think the latter would be pretty reasonable, but as I'm sure I've said before, I'm not too comfortable having a hard deadline on having to ship a new API or losing it. Having hard deadlines for shipping features, especially one this complex, generally results in rushed and buggy experiences. If we, say, get to M95 and have a working implementation for V2, but discover a bug at the last minute, I want to be able to have no hesitation to punt the release for an additional milestone (I don't want us to feel compelled to push it out the door because of an arbitrary 8-milestone limit).

We definitely don't want you or other feature teams to rush to ship features that are not ready. What I'd generally recommend is for you to have a rough plan for exiting the experimentation phase, and to come back to the API owners if e.g. last minute extensions are needed.

--
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/CAHqYdcYbiwZYA6atQ%3DsN6nkaT6M-7avGhOBdwjA373PG%3DCq2cg%40mail.gmail.com.

[Glen Robertson]

Sorry, I made a mistake in my original email in this thread: we actually delayed the start of the OT so I had the milestones wrong.

Corrected milestones follow:

Experimental timeline
- M88(Android)/M89(Desktop): First experiment milestone
- M90: Original final experiment milestone

- M93: Extended final experiment milestone

On Fri, 7 May 2021 at 23:19, Aer Berkopec <aerber...@gmail.com> wrote:

To view this discussion on the web visit https://groups.google.com/a/chromium.org/d/msgid/blink-dev/CAL5BFfW657p0WOQSZeJtJdNz%2Bo5EYdy8q7uJDfb%3D05MUXtLLSg%40mail.gmail.com.

[Yoav Weiss]

In that case, we're talking about 6 milestones all in all? That sounds perfectly reasonable.

[Matt Giuca]

Thanks for the explanation, Yoav.

Yes, we are talking about 6 milestones total (note: 5 milestones on Chrome OS, which is where a lot of our customers are targeting, since we missed the first milestone there).

As said above, we are anticipating a V2 origin trial at some point in the future. We think we'll have a pretty solid design by the end of the quarter (that's roughly corresponding to the start of M94) but we simply can't promise that V2 will be implemented by then. When we get to M94, I think it is likely that we'll need to further extend the existing origin trial, but we should have a much better picture of the work required by then. Another possibility is that we ship all or part of the current API to general availability (if it is determined that the new changes are additions, rather than changes, to the existing API).

Sorry if this is vague, but I am trying to keep our options open and make sure that we aren't making promises we can't keep. My point above stands, that I don't want to set hard engineering deadlines which, if we don't meet them, will result in the disappearance of the API for our customers. We are not trying to keep this API in origin trial forever, but we need time to work towards a solution.

Matt

[Yoav Weiss]

On Mon, May 10, 2021 at 10:35 AM Matt Giuca <mgi...@chromium.org> wrote:

Thanks for the explanation, Yoav.

Yes, we are talking about 6 milestones total (note: 5 milestones on Chrome OS, which is where a lot of our customers are targeting, since we missed the first milestone there).

As said above, we are anticipating a V2 origin trial at some point in the future. We think we'll have a pretty solid design by the end of the quarter (that's roughly corresponding to the start of M94) but we simply can't promise that V2 will be implemented by then. When we get to M94, I think it is likely that we'll need to further extend the existing origin trial, but we should have a much better picture of the work required by then. Another possibility is that we ship all or part of the current API to general availability (if it is determined that the new changes are additions, rather than changes, to the existing API).

Ideally, at that point you would have well-documented learnings from the OT that you could use to justify either a decision to ship some parts of the API, or change it towards a second OT.

 

Sorry if this is vague, but I am trying to keep our options open and make sure that we aren't making promises we can't keep. My point above stands, that I don't want to set hard engineering deadlines which, if we don't meet them, will result in the disappearance of the API for our customers. We are not trying to keep this API in origin trial forever, but we need time to work towards a solution.

The above SGTM!

[Alex Russell]

SGTM too; LGTM!

 

Matt

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/CAHqYdcYbiwZYA6atQ%3DsN6nkaT6M-7avGhOBdwjA373PG%3DCq2cg%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.

[Matt Giuca]

Thanks Alex and Yoav.

SGTM too; LGTM!

 

Matt

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/CAHqYdcYbiwZYA6atQ%3DsN6nkaT6M-7avGhOBdwjA373PG%3DCq2cg%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.

[Glen Robertson]

Hi Blink Owners,

We would like to extend the DGAPI OT further. Requesting an extension to M96 (inclusive).

Note: that is 9 milestones total (8 on CrOS), some of which are the new shorter milestones. Also CrOS won't ship M95.

We now have an updated API design in internal review (sorry, Google only for now). I plan to update the public explainer once the design is internally approved, then start on implementation of a revised "v2" API. I expect implementation to take another milestone at least, but like before we can't make promises about getting the engineering part done within a tight timeline. When that is ready, we will want to start an OT on that revised API.

Also M94 (the existing post-end milestone) has already branched - do we need to merge something to ensure DGAPI OT is not turned off for M94?

Thanks

-Glen

[Chris Harrelson]

Hi,

The API owners met to discuss this intent today (Chris, Yoav, Alex, Mike). We're glad to see you are still pursuing the API and responding to feedback with the v2 API.

However, 9 milestones is very long, much longer than we are typically comfortable with, and within the range where there starts to be substantial risk of burn-in. That combined with the v2 design being neither publicly available nor implemented yet, leads us to conclude that we can't support extending the current Origin Trial, since there is also no defined experimental justification for doing so.

You can of course come back and ask for a new Origin Trial once v2 is public and implemented.

We also discussed two other points:

* There is the option of shipping a subset now, but we were not sure if there is a useful and stable subset that could be shipped.

* If v2 were implemented now and you wished to transition the Origin Trial "smoothly" to the new API, that would have been reasonable to consider. Reason being the purpose of an Origin Trial is to experimentally fit an API to purpose, and a change to v2 is a strong signal that there was experimental feedback justifying it. Further, a breaking v2 API would reduce burn-in risk.

Thanks,

Chris, on behalf of the API owners

To view this discussion on the web visit https://groups.google.com/a/chromium.org/d/msgid/blink-dev/CAPV%2BSg_pU3Q59ZcBVDyY-TFDmhUjW_3e6wqPRfWLZDqJER_r4Q%40mail.gmail.com.

[Glen Robertson]

Would it be possible for us to extend by 2 milestones to M95 (inclusive) instead? We believe we can get a new v2 API (which will be a significant breaking change) implemented in time for M96; M94 is already in beta and M95 will not ship on CrOS, so the earliest we can get new code out to developers is in M96 anyway. This would make our total OT timeline M88 to M95 (8 milestones total), which is within the maximum OT time limit of 8 milestones Alex mentioned above (in fact shorter total time due to the changing milestone period). We would very much like to avoid the disruption to developers of having the OT turned off and this functionality being entirely unavailable during the intervening period before the new OT starts.

We understand your concern about an extended OT risking burn-in, but this is a complex API for developers to start using, as they have to create a product and payment flow around it. Usage of the API is still low — a few hundred calls per day total for all methods (excluding getDigitalGoodsService, which is used for feature detection even when the API is otherwise unavailable).

Shipping a subset of the API now wouldn't help because we are proposing breaking changes to enough of the API that it probably isn't useful to ship the rest. Also, we are proposing a breaking change in the behaviour of the feature detection function that must be called before any other calls, so it would be mildly risky to ship that immediately.

I have made a draft version of the proposed v2 API publicly available (though it is still being reviewed). It will be a significant breaking change to the API.

Thanks,

-Glen

[TRI DUNG DINH]

La tôi

Đã gửi từ iPad của tôi

Ngày 27-08-2021, vào lúc 11:56, Glen Robertson <gle...@chromium.org> viết:

To view this discussion on the web visit https://groups.google.com/a/chromium.org/d/msgid/blink-dev/CAPV%2BSg-%2BDqTw0coyVgjHLF4nytoDMDAyTBGv4woppiuyEOwwew%40mail.gmail.com.

[Glen Robertson]

Hi Blink Owners,

Have you had a chance to consider this yet? We are quickly approaching the end date of the current OT (2021-09-14).

[Chris Harrelson]

Hi Glen,

Thanks for your patience while we (collectively - your team, API owners, and others) worked through a response to your request.

The API owners met today and discussed this request to extend the Origin Trial. Here is some new data I got:

* The team showed evidence of recent further engagement with partners and getting as soon as possible to the v2 you mentioned. The Web Payments team also signed up to help accelerate that.

* There is new evidence (to me) of the potential & importance of this API for the web that I hadn't considered. This would really help the web platform be more useful, IMO.

I'm still concerned about the length of the trial and its impact on burn-in and other risks of a long OT. But I'm now ok with approving this request.

Finally, we also concluded that because of the length of the Origin Trial, it's outside the bounds of approval via one LGTM, and requires three. So please wait for two more LGTMs.

LGTM1 to extend this trial to M96.

To view this discussion on the web visit https://groups.google.com/a/chromium.org/d/msgid/blink-dev/CAPV%2BSg9Kyca3JTU9ZoqsCsEMr4Uuojq8cXge8W4eokF9qZ1CCQ%40mail.gmail.com.

[Daniel Bratell]

LGTM2 to the plan and comments outlined by Chris.

/Daniel

To view this discussion on the web visit https://groups.google.com/a/chromium.org/d/msgid/blink-dev/CAOMQ%2Bw_0tMBsaSM__mYY8onUaq899GJk6aFJtzG1MAOCcj-jYg%40mail.gmail.com.

[Mike West]

LGTM3.

-mike

--

-mike

[Glen Robertson]

Hi Blink Owners,

We have now implemented DGAPI v2.0 which is in M96 as a separate OT. The DGAPI v1 code has been removed in M96 so the v1 trial will not be exposed to versions greater than M95.

However, the OT system is based around dates instead of milestones, and the v1 OT is currently set to end on 2021-12-14, two weeks after M96 hits stable. At that point we expect a large percentage of CrOS users to still be on M95 (or lower), with that percentage decreasing asymptotically over the following weeks. As previously discussed, we would strongly like to avoid dropping support for users in between the origin trials and launch.

Would it be possible to extend the end date of this v1 API trial (for M95 and below only) to allow more time for users to update to M96 before the API is cut off? Two months total would allow time for the majority of users to update (2022-01-30).

Web developers wishing to use DGAPI are already required to update their code to support the new version in M96, so there is no risk of burn-in of the v1 API.

Thanks for considering,

-Glen

--

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/CAKXHy%3DcnMZ%2Brc_HuKx8fhs0RVc94fmhUacaneftCcM%2BjxOGk9g%40mail.gmail.com.

[Chris Harrelson]

I think it's fine to extend the date of the v1 API OT, targeted only at M95 and below. LGTM for that!

To view this discussion on the web visit https://groups.google.com/a/chromium.org/d/msgid/blink-dev/CAPV%2BSg-er1aE1pvYnG1OVai6J7_RN1haYXTHN%2BV7TpF2bQ3hsg%40mail.gmail.com.

[Glen Robertson]

Thanks!

To view this discussion on the web visit https://groups.google.com/a/chromium.org/d/msgid/blink-dev/CAOMQ%2Bw_dPo2bg7zVyobKN9TO2hLgpq1KW5AS-yfdW1b9ObU7Xw%40mail.gmail.com.