- From: Anders Rundgren <anders.rundgren.net@gmail.com>
- Date: Tue, 1 Sep 2015 13:52:18 +0200
- To: Adrian Hope-Bailie <adrian@hopebailie.com>
- Cc: Web Payments IG <public-webpayments-ig@w3.org>
On 2015-09-01 12:24, Adrian Hope-Bailie wrote: > <snip> > > > Well, if your target is rather creating a standard supporting > - a defined and documented security model > > > Can you explain why the security model that already exists for the Web is insufficient? I was referring to the security model between web-pages and invoked local applications, an arrangement which according to most Web-gurus isn't a [legitimate] part of the Web. If applications rather are "listening" shouldn't make a difference, it is the same problem. Anyway, according to the lot mentioned above, there's no need to do anything of what you and I are currently discussing which makes this even more "interesting"!!! > - a simple JavaScript API > > > I think the existing patterns for communicating with services from client code via > HTTP and Websockets is very mature and well understood. I'm not sure that a new API > adds any value. What does it offer that can't be achieved using the existing pattern? How does localhost-based solutions transfer web-contexts including TLS-cert to the native application except through a direct call to the remote service? How do you maintain web-sessions and cookies? Well, I have done all of that but the code looks like c**p. In addition, this "handover" from remote to local web has undefined security properties. IMO, it smells like workaround. > > - a bidirectional channel interface using OS/process level security > localhost schemes doesn't really match up. > > > I don't understand what you are saying here. A native application will operate under > the policy that is defined for it by the OS. The fact that it exposes an HTTP interface > for integrations from code running in a browser on the same machine doesn't change that does it? > > > Of course the browser vendors could make a specific "patch" to localhost but it will > be equally hard to standardize as the Web2Native Bridge. Probably even more difficult > since localhost already is a standard http/web feature. > > > Using an HTTP service on local host doesn't strike me as requiring any changes to anything. > In fact the pattern is already widely used. What do you think needs to be standardised? I don't know why Google introduced Native Messaging if it redundant, but it *may* have to do with something of the stuff I'm writing. That some services like www.dropboxlocalhost.com define public DNS names as pointing to 127.0.0.1 indicates that the local web-server concept has flaws. Related: https://code.google.com/p/chromium/issues/detail?id=378566 Anders > > </snip> > > > On 31 August 2015 at 16:02, Anders Rundgren <anders.rundgren.net@gmail.com <mailto:anders.rundgren.net@gmail.com> <mailto:anders.rundgren.net@gmail.com <mailto:anders.rundgren.net@gmail.com>>> wrote: > > Hi Payment-Gurus, > > If you have the possibility installing system software like Java SE on a PC or MAC at your disposal, > you can test an early version [1] of a decentralized [2], wallet-based [3] Web Payment system. I will > later create a short YouTube video that (hopefully) will explain the concept for non-programmers. > > For those of you who are in the process of creating some kind of Web Payment standard, I believe > the fact that the PoC system relies on a single and pretty universal (not "payment flavored"), > method added to the browser [4], should be of some interest. > > Although the scope of this project extends quite a bit beyond security, it may be worth mentioning > that the system uses an enhanced smart card for holding payment credentials (albeit in a software > version to make testing feasible without shipping hardware). > > After installing the Wallet software, you can try out the PoC at: https://test.webpki.org/webpay-merchant > > I'm currently working on a second generation that will contain an equally decentralized scheme for > protecting card data in traditional card networks , which is not based on Tokenization since I have found > out that the EMVCo Tokenization scheme has severe scalability issues. > > The system currently lacks some documentation... > Anyway, I'm always responding quickly to questions if you have some :-) > > Cheers, > Anders > Performing his weekly update > > 1] Getting the PoC software: https://github.com/cyberphone/web2native-bridge#installation > > 2] Decentralized payments: http://webpki.org/papers/decentralized-payments.pdf > > 3] The Wallet: https://github.com/cyberphone/web2native-bridge/blob/release/webpayment.client/src/org/webpki/w2nb/webpayment/client/Wallet.java#L116 > > 4] The added browser method: https://github.com/cyberphone/web2native-bridge#api > > > > > >
Received on Tuesday, 1 September 2015 11:52:57 UTC