[bittorrent] super-seeding and HAVE messages

Petter Nilsen pettern at opera.com
Fri Apr 28 09:46:14 EDT 2006


On Fri, 28 Apr 2006 15:08:04 +0200, Arvid Norberg <arvid at cs.umu.se> wrote:

> My point is, that since the super seed is waiting for another peer in  
> the swarm to have the piece, the peer which just
> downloaded it doesn't have to send a HAVE back to the super-seed, since  
> the seed won't care about it.

Exactly my point too, but atleast Az will not send another HAVE until it  
gets a (redundant) HAVE from the peer that just downloaded the piece.  If  
B doesn't send HAVE to A, A will never send another HAVE.
So it basically boils down to whether one should try to detect if a peer  
is a super-seeder (is it possible?), always send HAVE or optimize HAVE and  
hope it'll work.

-- 
Petter Nilsen, Software Engineer, Opera Software
Phone no: +4791684569 / mailto:pettern at opera.com




More information about the BitTorrent mailing list