[noise] minor suggestion: pre-message pattern instead of pre-message patternS ?

Trevor Perrin trevp at trevp.net
Wed Nov 1 17:33:35 PDT 2017


On Wed, Nov 1, 2017 at 5:55 PM, David Wong <davidwong.crypto at gmail.com> wrote:
> Sorry for the spam today :)
>
> a handshake pattern is:
>
> * a pre-message pattern for the initiator
> * a pre-message pattern for the responder
> * a sequence of message patterns for the actual handshake messages
> (called a handshake pattern?)

I believe the "sequence of message patterns" is just called that
("sequence of message patterns"), in the spec.

The term "handshake pattern" refers to the whole thing: A sequence of
pre-message patterns followed by a sequence of message patterns.

Maybe you think this is confusing because the Overview says "Message
patterns are arranged into handshake patterns"?  That's admittedly a
loose explanation, but that seems OK for the Overview, since
pre-message patterns haven't been introduced yet.

The actual section on "Handshake patterns" (Section 7) is more precise.


> Just two minor suggestions (btw, are the issues on github a better
> place for that?):

I think the list is fine for spec discussions (typo fixes or nits
could be git PRs).


> 1. the two pre-message patterns could be merged into one pre-message
> pattern where the responder tokens are represented as "rs" and "re". I
> know this would make the diagrams less nice, but this would make the
> code nicer.

That makes pre-message patterns less consistent with message patterns,
so I think I'd be opposed to that.


> 2. a handshake pattern has a handshake pattern inside of it. Maybe
> renaming one of the handshake pattern would be a good idea?

I don't quite agree that we're abusing terminology in that way, see if
my explanation above makes sense.

Trevor


More information about the Noise mailing list