Re: W3C Web Crypto Use cases - next step ?

Hi Karen!  Thanks for your careful review.


On Jul 8, 2013, at 3:47 PM, Lu HongQian Karen wrote:

> Hi Arun,
>  
> Thanks again for putting this together. I have a few comments below:
>  
> Banking transaction use case
> >Jae-sang presents the derived public key to GB
> Which derived public key?


It shouldn't say "derived."  Early drafts didn't distinguish sufficiently between key generation and key derivation.  I've fixed this now.


>  
> >GB may then use a key exchange mechanism to exchange keys with the server over TLS
> Guess you mean GB client (running in the web browser) and GB server?
>  


Right.  In every use case, I mean with a "user agent" as an intermediary.  So everytime I say "GB" I mean, an instrument such as a web page, iframe, or script hosted on GB's domain.  In this particular example, ALL transactions occur within GB's domain.


> >The signature is verified, and upon successful verfication, is decrypted.
> verfication -> verification
> is decrypted -> the encrypted message is decrypted.


Fixed.


> BrowserID use case
>  
> >BrowserID
> Change to Persona?


Good nit!  So actually, any reference to the protocol itself is "BrowserID."  Persona, generally considered a product, is in fact an implementation of BrowserID.  Stricly speaking, that protocol spec lives here: https://github.com/mozilla/id-specs/tree/prod/browserid

>  
> > The signed assertion is then combined …
> May change to: The javascript of Persona.org in the iFrame combines the signed assertion…

Sure!

>  
> Does Karen need to login to Persona before Persona can access her data?


Yes!  The flow is exactly as if you did this:

1. Visit http://myfavoritebeer.org/
2. Log in using Persona

What is happening between those websites is what the use case is.


>  
> The verification part is confusing to me. Does PSS verifies on the client side or the server side?


EVERYTHING happens on the client side.  PSS obtains the public key from https://persona.org/.well-known/browserid (which in fact uses a semi-legitimate form of JWK).

-- A*

Received on Friday, 12 July 2013 19:43:06 UTC