Re: [Curdle] Some work for the group

[Russ Housley <[e-mail]>]

On Dec 9, 2016, at 12:33 PM, Phillip Hallam-Baker <[e-mail]> wrote:

> 
> 
> On Fri, Dec 9, 2016 at 11:42 AM, Russ Housley <[e-mail]> wrote:
> 
>> The text in draft-ietf-curdle-pkix says CA’s MUST NOT use pre-hash version of signatures.  Does anyone object to this?  There is a mention of the trade-offs in doing that at the end of section 5.  Please respond within a week.
> 
>  	
> draft-ietf-curdle-cms-eddsa-signatures is aligned with this decision.  If it changes in one of the documents, it needs to change in both.
> 
> ​I don't think it is sustainable in CMS so if it has to be the same in both places then we have to reject it.
> 
> In order for a crypto mode to be viable, it has to be possible to implement it in a streaming encoder. Memory is cheap but storage is cheaper. If I have a digital camera that is signing pictures as they are taken, that module has to be something that can be implemented in a hardware blob somewhere close to the point the image is captured.
> 
> Another think I have great difficulty with here is actually working out what the pure vs free hash implementation is from the specification. trying to describe crypto in RFC plaintext isn't a good match at the best of times. ​This is particularly hard to follow.

Please take a look at the document.  The content is hashed, and that hash is placed in the message digest attribute, then the set of attributes is signed with pure EdDSA.

Russ