Re: Drastically cutting primary features [was Re: Last call for public comments on Web Crypto charter]

IMO, the division between simple and difficult goes between a
domain-oriented crypto system and a traditional unrestricted ditto.

In a domain-restricted crypto system everything happens between a
specific user/browser and a specific relying part/issuer application.
Privacy-wise there are no issues and security-wise screw-ups are
limited to exactly these two parties.

Traditional unrestricted crypto systems essentially targets the
Internet which have major implications which (based from the comments
on this list...), few people seem to have realized.

I think it is time for the DomCrypt guys to chime-in and say if
this description is wrong or not.  Phrases like "we must support
smart cards etc in the future" has no room in a charter because
that has to proved with respect to *feasibility*.  Oracle has
smart card support in Java (javax.smartcardio.*) which nobody
uses because it simply doesn't work.

Anders

On 2011-11-24 00:14, Harry Halpin wrote:
> On 11/18/2011 06:34 AM, Stephen Farrell wrote:
>>
>> I've very happy to see how this process has gone and the
>> resulting charter. I have two comments:-
>>
>> I would strongly argue to move TLS key extraction into
>> the list of primary features. I guess that function might
>> not always produce a key, depending on client & server
>> implementations, but I think its important that it be
>> available since if/when it works, it would mean that an
>> awful lot of people would not need to develop their own
>> (and probably broken) key distribution schemes.
> 
> I'm catching up on older comments, and I'd like for people on this 
> mailing list to notice the radical simplifying nature of Stephen's 
> proposal.
> 
> By having key establishment come from TLS as the primary feature, it 
> would get rid of the need for key storage, key agreement, and key 
> generation from primary features. Or at least move those to secondary 
> features!
> 
> What do people think?
> 
> It would be a radically simpler starting point for an API, which appeals 
> to me.
> 
>>
>> I would separately argue that the current list of primary
>> functions (esp without TLS key extraction) is not really
>> a "high-level API," right now, it looks much more like
>> just any old crypto API (e.g. if you have D-H, which
>> many developers might not understand very well). I think
>> requiring the WG to more somehow at a higher level than
>> JCE/JCA might be a way to indicate that.
>>
>> Regards,
>> Stephen.
>>
>> On 11/17/2011 03:17 PM, Harry Halpin wrote:
>>> Everyone,
>>>
>>> On next Tuesday, as said earlier, I plan to take the Web Cryptography
>>> charter [1] from the wiki and put it into HTML as an "official draft
>>> charter" then ask for preliminary feedback from the AC, before going to
>>> real AC review in December (thus launching Working Group in January).
>>>
>>> So, if you have any comments, *now* is the time to send to the mailing
>>> list. Suggested text replacement is most welcome.
>>>
>>>        cheers,
>>>           harry
>>>
>>> [1] http://www.w3.org/wiki/IdentityCharter
>>>
>>>
>>
> 
> 
>

Received on Thursday, 24 November 2011 09:41:59 UTC