swarm/network: simplified neighbourhood depth calc #1013
Conversation
swarm/network/kademlia.go
Outdated
| @@ -451,11 +451,14 @@ func (k *Kademlia) neighbourhoodDepth() (depth int) { | |||
| var lastPo int | |||
|
|
|||
| f := func(v pot.Val, i int) bool { | |||
| if bytes.Equal(pot.ToBytes(v), pivotAddr) { | |||
There was a problem hiding this comment.
Under what condition should we be getting ourselves in these iterations. Isn't that actually an anomaly?
There was a problem hiding this comment.
maybe going back to po == 256 is simpler
There was a problem hiding this comment.
@justelad the question is about the iterator, not the value itself. Should the iterator ever return ourselves?
There was a problem hiding this comment.
good question. i guess not since the method name is EachNeighbour and self is not a neighbour. i'll make the change. good point 👍
There was a problem hiding this comment.
@nolash, come to think of it, that method name could be actually changed to EachPeer, since it refers to actual peers connected, not just neighbours.
@zelig, I noted to @nolash that some of the terminology in the kademlia is somewhat misleading (as I already told you about minProxBinSize), EachNeighbour is the same case - we iterate over peers not neighbours, the iterator in this case should not have a notion of what is neighbouring and what is not, and that is up to the iteratee to assert. (my 2 cents)
nolash
force-pushed
the
kademlia-neighbourhooddepth
branch
from
6a2eaf1
to
12f1e68
Compare
|
@justelad Tests fail; I pushed a minor amend in comments, hope it's ok. |
acud
force-pushed
the
kad-depth-calc
branch
2 times, most recently
from
8d2609c
to
fa8e0f1
Compare
…nabled waitTillHealty
this PR simplifies the neighbourhood depth calculation for the kademlia and reuses the same logic in the peer pot map creation