[Monotone-devel] Re: netsync transport encryption?

[Cem Karan]

> Message: 7
> Date: Wed, 25 Oct 2006 13:41:12 +0100
> From: Bruce Stephens <address@hidden>
> Subject: [Monotone-devel] Re: netsync transport encryption?
> To: address@hidden
> Message-ID: <address@hidden>
> Content-Type: text/plain; charset=us-ascii
>
> Dirk Hillbrecht <address@hidden> writes:
>
> [...]
>
> > The question is what you want: Encryption or authentication.
>
> Ah, that's a good point.  The first time I pull from a repository I
> don't have the server's public key.  And we want to permit anonymous
> connections (as now) in which case the server doesn't know which of my
> public keys to use (and quite likely doesn't have them anyway).
>
> I guess in those cases perhaps encryption isn't really useful?

Well, it won't be useful against eavesdropping, but my question is,  
how is this any different from the non-encrypted case?  As near as I  
can tell, you have three cases:

1) Neither side has the other's public key.  The anonymous access  
algorithm kicks in, and temporary session keys can still be passed  
back and forth.  The only difference is that these keys aren't  
authenticated.

2) One side has the other's public key, but not vice-versa.  This  
might be the case with a distribution of Monotone that public keys  
for the server sent out, but where the new, anonymous user's keys are  
unknown.  This case is similar to 1, except that there is only  
authentication in one direction.  This might be in a case like where  
I'm in, where I want to know that I'm getting sources certified as  
being original Monotone sources, but where I'm not going to upload  
anything.

3) Both sides have each other's keys.  This becomes the case I  
originally outlined.

In short, you would lose authentication and guarantees of privacy if  
you don't have each other's public keys, but it shouldn't affect the  
connection in any way, even for anonymous access.

The other solution is to generate an 'anonymous' set of keys that are  
always distributed with the Monotone sources.  Then you are always  
authenticated as anonymous, and the rest of the algorithm goes on as  
before.  I don't know enough about Montone (I just downloaded it  
yesterday!), but from what I can see, this might be the best  
solution; you can use the authentication and trust methods that are  
already available to determine what you want to do with an anonymous  
peer.  The default in the distribution should be to refuse it  
completely.  That will protect newbies like me from shooting  
ourselves in the foot!

Thanks,
Cem Karan