Re: [ppsp] Progress on core tracker protocol?

[Arno Bakker <[e-mail]>]

Hi Rui et al

after reading the base protocol -00 I have the following comments:

* General: basic design now nice and simple, description still verbose.

* p. 14: The Terminated state does not make much sense, IMHO and is 
equivalent to INIT.

* p. 24: In a PeerGroup in a CONNECT a peer may provide a list of IP 
addresses. How to you intend to protect against malicious peers 
registering victim IP addresses as part of an amplification attack (see
draft-ietf-ppsp-peer-protocol-02, p. 13.2.3). For IP addresses (other 
than the one from which the request comes from) some form of return 
routability check is needed.

* p. 31: Seeders will need to be able to retrieve peer lists from the 
tracker too, in particular when the seeder is behind a NAT itself. 
Please adjust the spec to allow this.

* p. 35: For the peer protocol's security we need the tracker to return 
a true random sample of the list of registered peers in a swarm (see 
peer draft Sec. 13.2.3). Could you make this the default behaviour, or 
allow a peer to request this explicitly?

Regards,
      Arno