[noise] NoiseSocket API feedback

Nemanja Mijailovic metalnem at mijailovic.net
Sun Apr 29 01:18:58 PDT 2018


I found one more part of the specification that I think could be a bit clearer. In the section 3 (Negotiation), the part describing the Acceptance case says:

"The responder sends a NoiseSocket handshake message containing the next handshake message in the initial Noise protocol. The negotiation_data field must be empty.”

That is true only for two-way patterns. In one-way patterns, there will be no next handshake message, so the responder can only send empty noise_message field. I think that the spec should probably be explicit about that.

Nemanja

> On Apr 24, 2018, at 10:40 AM, Trevor Perrin <trevp at trevp.net> wrote:
> 
> On Thu, Apr 19, 2018 at 7:08 AM, Nemanja Mijailovic
> <metalnem at mijailovic.net> wrote:
>> I think it may be difficult to define the API that is easy to directly implement in all programming languages. The spec itself is clear enough, so I wouldn’t miss the API section much if we remove it.
> 
> Yeah, I think I agree with you and Christopher [1] that the API isn't
> adding much value, so I'll prepare a NoiseSocket draft that removes it
> (and also clarifies the padding of handshake payloads, as was
> discussed).
> 
> Trevor
> 
> [1] https://moderncrypto.org/mail-archive/noise/2018/001543.html
> 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://moderncrypto.org/mail-archive/noise/attachments/20180429/e8d8c344/attachment.html>


More information about the Noise mailing list