La tôi


Đã gửi từ iPad của tôi

Ngày 27-08-2021, vào lúc 11:56, Glen Robertson <[email protected]> viết:

> 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
> 
>> On Fri, 20 Aug 2021 at 06:33, Chris Harrelson <[email protected]> wrote:
>> 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
>> 
>>> On Mon, Aug 16, 2021 at 2:17 AM Glen Robertson <[email protected]> wrote:
>>> 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
>>> 
>>>> On Fri, 14 May 2021 at 10:19, Matt Giuca <[email protected]> wrote:
>>>> Thanks Alex and Yoav.
>>>> 
>>>>> On Fri, 14 May 2021 at 05:24, Alex Russell <[email protected]> 
>>>>> wrote:
>>>>> SGTM too; LGTM!
>>>>> 
>>>>>> On Monday, May 10, 2021 at 3:54:37 AM UTC-7 Yoav Weiss wrote:
>>>>>>> On Mon, May 10, 2021 at 10:35 AM Matt Giuca <[email protected]> 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!
>>>>>>  
>>>>>>> 
>>>>>>> Matt
>>>>>>> 
>>>>>>>> On Mon, 10 May 2021 at 15:30, Yoav Weiss <[email protected]> 
>>>>>>>> wrote:
>>>>>>>> In that case, we're talking about 6 milestones all in all? That sounds 
>>>>>>>> perfectly reasonable.
>>>>>>>> 
>>>>>>>>> On Mon, May 10, 2021 at 6:43 AM Glen Robertson <[email protected]> 
>>>>>>>>> wrote:
>>>>>>>>> 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 <[email protected]> 
>>>>>>>>>> wrote:
>>>>>>>>>> 
>>>>>>>>>> 
>>>>>>>>>>> On Fri, May 7, 2021, 3:44 AM Yoav Weiss <[email protected]> 
>>>>>>>>>>> wrote:
>>>>>>>>>>> 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 <[email protected]> 
>>>>>>>>>>>> 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.
>>>>>>>>>>> 
>>>>>>>>>>>> 
>>>>>>>>>>>> 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
>>>>>>>>>>>> 
>>>>>>>>>>>> 
>>>>>>>>>>>>> On Fri, 7 May 2021 at 05:34, Alex Russell 
>>>>>>>>>>>>> <[email protected]> wrote:
>>>>>>>>>>>>> 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
>>>>>>>>>>>>> 
>>>>>>>>>>>>>> On Thursday, May 6, 2021 at 8:30:38 AM UTC-7 Glen Robertson 
>>>>>>>>>>>>>> wrote:
>>>>>>>>>>>>>> Contact emails
>>>>>>>>>>>>>> [email protected], [email protected], [email protected]
>>>>>>>>>>>>>> 
>>>>>>>>>>>>>> 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.
>>>>>>>>>>>> 
>>>>>>>>>>>> -- 
>>>>>>>>>>>> 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 [email protected].
>>>>>>>>>>>> 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 [email protected].
>>>>>>>>>>> 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.
>>> 
>>> -- 
>>> 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 [email protected].
>>> 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.
> 
> -- 
> 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 [email protected].
> 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.

-- 
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 [email protected].
To view this discussion on the web visit 
https://groups.google.com/a/chromium.org/d/msgid/blink-dev/EBD0C095-66F8-4381-A05C-BF9B97C4D9E7%40gmail.com.

Reply via email to