Colin Davis wrote: >>> >> I wonder what we could do if a node knew it didn't need hole punching >> or a node knew a particular peer didn't need hole punching. (How the >> node would know is irrelevant at this point in the discussion.) >> > > Exactly ;) > But that only works when a node knows, 100% that it is not behind a NAT. NAT is the most common reason for needing UDP hole punching, so I think we were talking about the same thing. > While it would be best if the node automatically detected, right now, > only a human can say. > > After talking on IRC with toad and Zothar, I've made a patch to allow > a human to set that, and thus set the Max backoff time. It appears to me that you've made the maximum ARK fetching backoff time user settable. I'm not clear on what that has to do with handshake/connection attempts or knowing the node is behind a NAT and/or needs UDP hole punching.
Perhaps you were thinking of changing the time between handshake requests? A separate layer would have to be added to the current logic for that to work. I realized now that my timeX, timeY and timeZ description may have been lacking. timeW would be the time when the node is not sending handshake requests to the peer and would be timeX - timeY long. Perhaps I should pseudo-code it.
