Re: rs parameter in encrypted content coding

On 2016-12-08 19:29, Martin Thomson wrote:
> James opened this:
>
> https://github.com/httpwg/http-extensions/issues/274
>
> I'm close to pushing a new version based on what we discussed at the
> meeting.  This would seem to be an open issue.  Here are the choices:
>
> 1. what we have now, rs = record without authentication tag
> 2. what James suggests rs includes auth tag
> 3. something else
>
> My original hope was to avoid having any potential values that were
> invalid, but we already got there with padding (rs=0,1,2 are all
> invalid).
>
> Personally, I don't find the lack of generality to be problematic, but
> I agree that the choice is a little arbitrary.  I'd be interested in
> hearing what other people think.

I don't feel strongly about it, but I think James' argument is good. So 
+1 on 2.

Best regards, Julian

Received on Thursday, 8 December 2016 18:43:09 UTC