- From: Adrian Hope-Bailie <adrian@ripple.com>
- Date: Sat, 21 Jan 2017 20:30:24 +0200
- To: Payments WG <public-payments-wg@w3.org>
- Cc: Ian Jacobs <ij@w3.org>, Adam Roach <abr@mozilla.com>, Tommy Thorsen <tommyt@opera.com>, Jason Normore <jason.normore@shopify.com>
- Message-ID: <CAN1GQt2h=tM38TuEBUav6ti-1WJxKZj8q+oPh05ugByxd0XN0Q@mail.gmail.com>
This thread, which was started between a few members of the Payment Apps TF has some relevance to an issue now being discussed at length on the lists[1]. As such, I am copying the list to ensure we have it publicly available as part of the discussion. [1] https://github.com/w3c/webpayments-payment-apps-api/issues/48 On Fri, Jan 13, 2017 at 7:28 PM, Jason Normore <jason.normore@shopify.com> wrote: > > Without a manifest file we are forced to get app meta-data by executing > a registration script before registration to see what data it might provide > if we ran the registration. Yuck. > > I strongly agree with this statement and is the main reason I was a > proponent of the separate payment app manifest that defines the extra > information required to display and instantiate or install it (pointer to > service worker js). BUT I think this statement in the proposed changes > takes care of this: "To recommend a payment app, a merchant provides a > payment app identifier (PAI), name, and icon information.". I actually like > this better because it gives more control to the merchant. > > > Jason > > On Fri, Jan 13, 2017 at 10:15 AM Ian Jacobs <ij@w3.org> wrote: > >> >> > On Jan 13, 2017, at 9:07 AM, Adrian Hope-Bailie <adrian@ripple.com> >> wrote: >> > >> > Two concerns with the proposed approach: >> > >> > 1. Browsers index service workers based on the scope URL not the script >> URL so we risk having multiple payment apps that conflict in scope and >> therefor cannot co-exist. >> >> That sounds like a reasonable statement and I don’t know enough about >> service workers: >> https://www.w3.org/TR/service-workers/#model >> >> AdamR made the point that we should talk with service worker experts to >> ensure we get this right. :) >> >> > 2. Without a manifest file we are forced to get app meta-data by >> executing a registration script before registration to see what data it >> might provide if we ran the registration. Yuck. >> >> Why do you need to know “what it might provide”? >> >> What is the difference between getting the manifest data from a file and >> getting the manifest data via a script? >> >> For me the difference is “we already have tools in place to do it via a >> service worker install event” and don’t need to ask the browser to invent >> anything new. >> > >> > I agree that derived URLs are also ugly so perhaps we should be >> consistent with the approach for Payment Request and have the identifier >> URL return the manifest (a JSON file) and also support link headers for >> getting the payment app URL without downloading the manifest. >> > >> > So there are two URLS: >> > >> > 1. The service worker scope and manifest file URL are the same. >> > 2. The service worker script is at another URL which is listed in the >> manifest and also available through link relations headers (consistent with >> PR). >> >> Maybe we should get on a call to work through this. As a last resort I >> will put this on next week’s Payment App TF agenda. >> >> Meanwhile, I’ll ping Jake Archibald. >> >> Ian >> -- >> Ian Jacobs <ij@w3.org> http://www.w3.org/People/Jacobs >> Tel: +1 718 260 9447 <(718)%20260-9447> >> >> >> >>
Received on Saturday, 21 January 2017 18:30:58 UTC