Re: W3C Web Crypto WG - Is our deliverable doomed ?

On 09/18/2012 11:20 AM, GALINDO Virginie wrote:
>
> Dear all,
>
> Highlighting a link that Harry shared with us during the conf call 
> yesterday. I think we should read that and try to see where we have 
> been 'better than expected' for providing security for webapp. This 
> would definitely help to better describe our value proposition, I 
> think -- and clearly know our limits ;-)
>
> http://www.matasano.com/articles/javascript-cryptography/
>

I think requiring CSP plus having the API work natively on the client 
addresses rather than be downloaded his issues re hijacking and 
malleability of the Javascript runtime, although the latter point 
requires serious deliberation on our part. In particular, we will 
clearly address "the lack of systems programming primitives needed to 
implement crypto" and issues he brings up like "random number 
generation." Also, he assumes TLS is working just fine and should just 
be used rather than any JS crypto, which it may be if you are using 
pinning+cert transparency, but that is not always the case - and many of 
the things that we need this API for, like signatures, cannot be easily 
done within WebApps as TLS does not let these functions be accessed in 
an API.

It seems to me the rest of the arguments here can be repurposed against 
using cryptography any programming language, not just a Javascript 
run-time environment. Which basically would mean cryptography is in 
general doomed because its hard, yet given its usage and utility, this 
is self-evidently not the case. I'd like to hear the opinion of others

   cheers,
       harry


> Regards,
>
> Virginie
>