Re: [Monotone-devel] netsync transport encryption?

[Jeronimo]

On Thu, Oct 26, 2006 at 12:08:22AM +0200, Ulf Ochsenfahrt wrote:
> Jeronimo wrote:
> >On Wed, Oct 25, 2006 at 06:08:04PM -0300, Jeronimo Pellegrini wrote:
> >>That would be O (RSA x users) every time you commit... Is it necessary?
> >>(Or DSA, or whatever other asymetric algorithm)
> >
> >Just so I write it correctly:
> >
> >The time would be O ( R x U ) per commit, where
> >
> >R = time of RSA(single key)
> >U = number of users
> >
> >So I'd rather change the key only when necessary.
> 
> I agree that it's more sensible to only switch revision keys when a 
> change of user trust occurs. Additionally, it might be possible to bring 
> the time requirements down to O(R x log(U)) for a key switch, by using a 
> binary tree of users.

Well, I didn't think of any optimizations yet since it's still in its
early days. :-)

> Is Apso a branch of monotone?

No -- more VCSs could be added to Apso by extending a C++ class.
Monotone was the first one (because it's what I use).

> Or does it work entirely different? The 
> website mentions that you plan to support other systems as well, that's 
> why I ask.

Both VC systems and cryptographic libraries should be plugabble.

> Is this already offtopic?

Don't know, but I suppose it is...

J.