GitLab

Before, the engine worker would pop a task and block on send to the
bitswap worker even if the bitswap worker wasn't to receive. Since the
task could have been invalidated during this blocking send, a small
number of stale (already acquired) blocks would be send to partners.

Now, tasks are only popped off of the queue when bitswap is ready to
send them over the wire. This is accomplished by removing the
outboxChanBuffer and implementing a two-phase communication sequence.

	refactor: peerRequestQueue

	it's a mistake to make one queue to fit all. Go's lack of algebraic
	types turns a generalized queue into a monstrosity of type
	checking/casting. Better to have individual queues for individual
	purposes.

	Conflicts:
		exchange/bitswap/decision/bench_test.go
		exchange/bitswap/decision/tasks/task_queue.go

	fix(bitswap.decision.PRQ): if peers match, always return result of pri comparison

	fix(bitswap.decision.Engine): push to the queue before notifying

	TOCTOU bug

	1. client notifies
	2. worker checks (finds nil)
	3. worker sleeps
	3. client pushes (worker missed the update)

	test(PQ): improve documentation and add test

	test(bitswap.decision.Engine): handling received messages

	License: MIT
Signed-off-by: Brian Tiger Chow <brian@perfmode.com>

I think it's time to move a lot of the peer-to-peer networking
but-not-ipfs-specific things into its own package: p2p.
This could in the future be split off into its own library.
The first thing to go is the peer.

this is a major refactor of the entire codebase
it changes the monolithic peer.Peer into using
a peer.ID and a peer.Peerstore.

Other changes:
- removed handshake3.
-	testutil vastly simplified peer
-	secio bugfix + debugging logs
-	testutil: RandKeyPair
-	backpressure bugfix: w.o.w.
-	peer: added hex enc/dec
-	peer: added a PeerInfo struct
  PeerInfo is a small struct used to pass around a peer with
 	a set of addresses and keys. This is not meant to be a
 	complete view of the system, but rather to model updates to
 	the peerstore. It is used by things like the routing system.
-	updated peer/queue + peerset
-	latency metrics
-	testutil: use crand for PeerID gen
 	RandPeerID generates random "valid" peer IDs. it does not
 	NEED to generate keys because it is as if we lost the key
 	right away. fine to read some randomness and hash it. to
 	generate proper keys and an ID, use:
 	  sk, pk, _ := testutil.RandKeyPair()
 	  id, _ := peer.IDFromPublicKey(pk)
 	Also added RandPeerIDFatal helper
- removed old spipe
- updated seccat
- core: cleanup initIdentity
- removed old getFromPeerList

License: MIT
Signed-off-by: Brian Tiger Chow <brian@perfmode.com>

License: MIT
Signed-off-by: Brian Tiger Chow <brian@perfmode.com>

License: MIT
Signed-off-by: Brian Tiger Chow <brian@perfmode.com>

(merely by convention)

License: MIT
Signed-off-by: Brian Tiger Chow <brian@perfmode.com>

License: MIT
Signed-off-by: Brian Tiger Chow <brian@perfmode.com>

performing CR in the form of a PR. Let me know what you think.

License: MIT
Signed-off-by: Brian Tiger Chow <brian@perfmode.com>

make vendor

License: MIT
Signed-off-by: Brian Tiger Chow <brian@perfmode.com>

for safety!

use mockpeer.WithID methods to create peers in tests

License: MIT
Signed-off-by: Brian Tiger Chow <brian@perfmode.com>

to emphasize idempotence

implementation will be patched to ensure bitswap messages cannot contain
duplicate blocks or keys

![](http://m.memegen.com/77n7dk.jpg)

true -> always send to peer

false -> use ledger-based strategy described in IPFS paper draft 3

> Why expose num bytes sent and received?

    Makes it easy to test consistency of the ledgers

> Got a better reason?

    Makes it possible to expose metrics to the people-facing API

didn't run tests after the refactor. apologies.

Move go-ipfs/bitswap package to go-ipfs/exchange/bitswap

* Delineates the difference between the generic exchange interface and
  implementations (eg. BitSwap protocol)

  Thus, the bitswap protocol can be refined without having to overthink
  how future exchanges will work. Aspects common to BitSwap and other
  exchanges can be extracted out to the exchange package in piecemeal.

  Future exchange implementations can be placed in sibling packages next
  to exchange/bitswap. (eg. exchange/multilateral)