added ctxgroup and router

Godeps/Godeps.json

@@ -92,6 +92,10 @@
 			"ImportPath": "github.com/jbenet/go-base58",
 			"Rev": "568a28d73fd97651d3442392036a658b6976eed5"
 		},
+		{
+			"ImportPath": "github.com/jbenet/go-ctxgroup",
+			"Rev": "6b9437e8517175306e30ffec241c523752a38303"
+		},
 		{
 			"ImportPath": "github.com/jbenet/go-datastore",
 			"Rev": "6a1c83bda2a71a9bdc936749fdb507df958ed949"
@@ -126,6 +130,10 @@
 			"ImportPath": "github.com/jbenet/go-random",
 			"Rev": "2e83344e7dc7898f94501665af34edd4aa95a013"
 		},
+		{
+			"ImportPath": "github.com/jbenet/go-router",
+			"Rev": "7a4053217b7bfe3a14cc79541557d47d0c4ad85f"
+		},
 		{
 			"ImportPath": "github.com/kr/binarydist",
 			"Rev": "9955b0ab8708602d411341e55fffd7e0700f86bd"

Godeps/_workspace/src/github.com/jbenet/go-ctxgroup/Godeps/Godeps.json

@@ -0,0 +1,14 @@
+{
+	"ImportPath": "github.com/jbenet/go-ctxgroup",
+	"GoVersion": "go1.3",
+	"Packages": [
+		"./..."
+	],
+	"Deps": [
+		{
+			"ImportPath": "code.google.com/p/go.net/context",
+			"Comment": "null-144",
+			"Rev": "ad01a6fcc8a19d3a4478c836895ffe883bd2ceab"
+		}
+	]
+}

Godeps/_workspace/src/github.com/jbenet/go-ctxgroup/Godeps/Readme

@@ -0,0 +1,5 @@
+This directory tree is generated automatically by godep.
+
+Please do not edit.
+
+See https://github.com/tools/godep for more information.

Godeps/_workspace/src/github.com/jbenet/go-ctxgroup/Makefile

@@ -0,0 +1,16 @@
+all:
+	# no-op
+
+GODEP=$(which godep)
+
+godep: ${GODEP}
+
+${GODEP}:
+	echo ${GODEP}
+	go get github.com/tools/godep
+
+# saves/vendors third-party dependencies to Godeps/_workspace
+# -r flag rewrites import paths to use the vendored path
+# ./... performs operation on all packages in tree
+vendor: godep
+	godep save -r ./...

Godeps/_workspace/src/github.com/jbenet/go-ctxgroup/README.md

@@ -0,0 +1,35 @@
+# ContextGroup
+
+
+- Godoc: https://godoc.org/github.com/jbenet/go-ctxgroup
+
+ContextGroup is an interface for services able to be opened and closed.
+It has a parent Context, and Children. But ContextGroup is not a proper
+"tree" like the Context tree. It is more like a Context-WaitGroup hybrid.
+It models a main object with a few children objects -- and, unlike the
+context -- concerns itself with the parent-child closing semantics:
+
+- Can define an optional TeardownFunc (func() error) to be run at Closetime.
+- Children call Children().Add(1) to be waited upon
+- Children can select on <-Closing() to know when they should shut down.
+- Close() will wait until all children call Children().Done()
+- <-Closed() signals when the service is completely closed.
+
+ContextGroup can be embedded into the main object itself. In that case,
+the teardownFunc (if a member function) has to be set after the struct
+is intialized:
+
+```Go
+type service struct {
+  ContextGroup
+  net.Conn
+}
+func (s *service) close() error {
+  return s.Conn.Close()
+}
+func newService(ctx context.Context, c net.Conn) *service {
+  s := &service{c}
+  s.ContextGroup = NewContextGroup(ctx, s.close)
+  return s
+}
+```

Godeps/_workspace/src/github.com/jbenet/go-ctxgroup/ctxgroup.go

@@ -0,0 +1,265 @@
+// package ctxgroup provides the ContextGroup, a hybrid between the
+// context.Context and sync.WaitGroup, which models process trees.
+package ctxgroup
+
+import (
+	"sync"
+
+	context "github.com/jbenet/go-ipfs/Godeps/_workspace/src/code.google.com/p/go.net/context"
+)
+
+// TeardownFunc is a function used to cleanup state at the end of the
+// lifecycle of a process.
+type TeardownFunc func() error
+
+// ChildFunc is a function to register as a child. It will be automatically
+// tracked.
+type ChildFunc func(parent ContextGroup)
+
+var nilTeardownFunc = func() error { return nil }
+
+// ContextGroup is an interface for services able to be opened and closed.
+// It has a parent Context, and Children. But ContextGroup is not a proper
+// "tree" like the Context tree. It is more like a Context-WaitGroup hybrid.
+// It models a main object with a few children objects -- and, unlike the
+// context -- concerns itself with the parent-child closing semantics:
+//
+// - Can define an optional TeardownFunc (func() error) to be run at Close time.
+// - Children call Children().Add(1) to be waited upon
+// - Children can select on <-Closing() to know when they should shut down.
+// - Close() will wait until all children call Children().Done()
+// - <-Closed() signals when the service is completely closed.
+//
+// ContextGroup can be embedded into the main object itself. In that case,
+// the teardownFunc (if a member function) has to be set after the struct
+// is intialized:
+//
+//  type service struct {
+//  	ContextGroup
+//  	net.Conn
+//  }
+//
+//  func (s *service) close() error {
+//  	return s.Conn.Close()
+//  }
+//
+//  func newService(ctx context.Context, c net.Conn) *service {
+//  	s := &service{c}
+//  	s.ContextGroup = NewContextGroup(ctx, s.close)
+//  	return s
+//  }
+//
+type ContextGroup interface {
+
+	// Context is the context of this ContextGroup. It is "sort of" a parent.
+	Context() context.Context
+
+	// SetTeardown assigns the teardown function.
+	// It is called exactly _once_ when the ContextGroup is Closed.
+	SetTeardown(tf TeardownFunc)
+
+	// Children is a sync.Waitgroup for all children goroutines that should
+	// shut down completely before this service is said to be "closed".
+	// Follows the semantics of WaitGroup:
+	//
+	//  Children().Add(1) // add one more dependent child
+	//  Children().Done() // child signals it is done
+	//
+	// WARNING: this is deprecated and will go away soon.
+	Children() *sync.WaitGroup
+
+	// AddChildGroup registers a dependent ContextGroup child. The child will
+	// be closed when this parent is closed, and waited upon to finish. It is
+	// the functional equivalent of the following:
+	//
+	//	parent.Children().Add(1) // add one more dependent child
+	//  go func(parent, child ContextGroup) {
+	//  	<-parent.Closing()       // wait until parent is closing
+	//  	child.Close()            // signal child to close
+	//  	parent.Children().Done() // child signals it is done
+	//	}(a, b)
+	//
+	AddChildGroup(c ContextGroup)
+
+	// AddChildFunc registers a dependent ChildFund. The child will receive
+	// its parent ContextGroup, and can wait on its signals. Child references
+	// tracked automatically. It equivalent to the following:
+	//
+	//  go func(parent, child ContextGroup) {
+	//
+	//  	<-parent.Closing()       // wait until parent is closing
+	//  	child.Close()            // signal child to close
+	//  	parent.Children().Done() // child signals it is done
+	//	}(a, b)
+	//
+	AddChildFunc(c ChildFunc)
+
+	// Close is a method to call when you wish to stop this ContextGroup
+	Close() error
+
+	// Closing is a signal to wait upon, like Context.Done().
+	// It fires when the object should be closing (but hasn't yet fully closed).
+	// The primary use case is for child goroutines who need to know when
+	// they should shut down. (equivalent to Context().Done())
+	Closing() <-chan struct{}
+
+	// Closed is a method to wait upon, like Context.Done().
+	// It fires when the entire object is fully closed.
+	// The primary use case is for external listeners who need to know when
+	// this object is completly done, and all its children closed.
+	Closed() <-chan struct{}
+}
+
+// contextGroup is a Closer with a cancellable context
+type contextGroup struct {
+	ctx    context.Context
+	cancel context.CancelFunc
+
+	// called to run the teardown logic.
+	teardownFunc TeardownFunc
+
+	// closed is released once the close function is done.
+	closed chan struct{}
+
+	// wait group for child goroutines
+	children sync.WaitGroup
+
+	// sync primitive to ensure the close logic is only called once.
+	closeOnce sync.Once
+
+	// error to return to clients of Close().
+	closeErr error
+}
+
+// newContextGroup constructs and returns a ContextGroup. It will call
+// cf TeardownFunc before its Done() Wait signals fire.
+func newContextGroup(ctx context.Context, cf TeardownFunc) ContextGroup {
+	ctx, cancel := context.WithCancel(ctx)
+	c := &contextGroup{
+		ctx:    ctx,
+		cancel: cancel,
+		closed: make(chan struct{}),
+	}
+	c.SetTeardown(cf)
+
+	c.Children().Add(1) // initialize with 1. calling Close will decrement it.
+	go c.closeOnContextDone()
+	return c
+}
+
+// SetTeardown assigns the teardown function.
+func (c *contextGroup) SetTeardown(cf TeardownFunc) {
+	if cf == nil {
+		cf = nilTeardownFunc
+	}
+	c.teardownFunc = cf
+}
+
+func (c *contextGroup) Context() context.Context {
+	return c.ctx
+}
+
+func (c *contextGroup) Children() *sync.WaitGroup {
+	return &c.children
+}
+
+func (c *contextGroup) AddChildGroup(child ContextGroup) {
+	c.children.Add(1)
+	go func(parent, child ContextGroup) {
+		<-parent.Closing()       // wait until parent is closing
+		child.Close()            // signal child to close
+		parent.Children().Done() // child signals it is done
+	}(c, child)
+}
+
+func (c *contextGroup) AddChildFunc(child ChildFunc) {
+	c.children.Add(1)
+	go func(parent ContextGroup, child ChildFunc) {
+		child(parent)
+		parent.Children().Done() // child signals it is done
+	}(c, child)
+}
+
+// Close is the external close function. it's a wrapper around internalClose
+// that waits on Closed()
+func (c *contextGroup) Close() error {
+	c.internalClose()
+	<-c.Closed() // wait until we're totally done.
+	return c.closeErr
+}
+
+func (c *contextGroup) Closing() <-chan struct{} {
+	return c.Context().Done()
+}
+
+func (c *contextGroup) Closed() <-chan struct{} {
+	return c.closed
+}
+
+func (c *contextGroup) internalClose() {
+	go c.closeOnce.Do(c.closeLogic)
+}
+
+// the _actual_ close process.
+func (c *contextGroup) closeLogic() {
+	// this function should only be called once (hence the sync.Once).
+	// and it will panic at the bottom (on close(c.closed)) otherwise.
+
+	c.cancel()                    // signal that we're shutting down (Closing)
+	c.closeErr = c.teardownFunc() // actually run the close logic
+	c.children.Wait()             // wait till all children are done.
+	close(c.closed)               // signal that we're shut down (Closed)
+}
+
+// if parent context is shut down before we call Close explicitly,
+// we need to go through the Close motions anyway. Hence all the sync
+// stuff all over the place...
+func (c *contextGroup) closeOnContextDone() {
+	<-c.Context().Done() // wait until parent (context) is done.
+	c.internalClose()
+	c.Children().Done()
+}
+
+// WithTeardown constructs and returns a ContextGroup with
+// cf TeardownFunc (and context.Background)
+func WithTeardown(cf TeardownFunc) ContextGroup {
+	if cf == nil {
+		panic("nil TeardownFunc")
+	}
+	return newContextGroup(context.Background(), cf)
+}
+
+// WithContext constructs and returns a ContextGroup with given context
+func WithContext(ctx context.Context) ContextGroup {
+	if ctx == nil {
+		panic("nil Context")
+	}
+	return newContextGroup(ctx, nil)
+}
+
+// WithContextAndTeardown constructs and returns a ContextGroup with
+// cf TeardownFunc (and context.Background)
+func WithContextAndTeardown(ctx context.Context, cf TeardownFunc) ContextGroup {
+	if ctx == nil {
+		panic("nil Context")
+	}
+	if cf == nil {
+		panic("nil TeardownFunc")
+	}
+	return newContextGroup(ctx, cf)
+}
+
+// WithParent constructs and returns a ContextGroup with given parent
+func WithParent(p ContextGroup) ContextGroup {
+	if p == nil {
+		panic("nil ContextGroup")
+	}
+	c := newContextGroup(p.Context(), nil)
+	p.AddChildGroup(c)
+	return c
+}
+
+// WithBackground returns a ContextGroup with context.Background()
+func WithBackground() ContextGroup {
+	return newContextGroup(context.Background(), nil)
+}

Godeps/_workspace/src/github.com/jbenet/go-ctxgroup/ctxgroup_test.go

@@ -0,0 +1,187 @@
+package ctxgroup
+
+import (
+	"testing"
+	"time"
+
+	context "github.com/jbenet/go-ipfs/Godeps/_workspace/src/code.google.com/p/go.net/context"
+)
+
+type tree struct {
+	ContextGroup
+	c []tree
+}
+
+func setupCGHierarchy(ctx context.Context) tree {
+	t := func(n ContextGroup, ts ...tree) tree {
+		return tree{n, ts}
+	}
+
+	if ctx == nil {
+		ctx = context.Background()
+	}
+	a := WithContext(ctx)
+	b1 := WithParent(a)
+	b2 := WithParent(a)
+	c1 := WithParent(b1)
+	c2 := WithParent(b1)
+	c3 := WithParent(b2)
+	c4 := WithParent(b2)
+
+	return t(a, t(b1, t(c1), t(c2)), t(b2, t(c3), t(c4)))
+}
+
+func TestClosingClosed(t *testing.T) {
+
+	a := WithBackground()
+	Q := make(chan string)
+
+	go func() {
+		<-a.Closing()
+		Q <- "closing"
+	}()
+
+	go func() {
+		<-a.Closed()
+		Q <- "closed"
+	}()
+
+	go func() {
+		a.Close()
+		Q <- "closed"
+	}()
+
+	if q := <-Q; q != "closing" {
+		t.Error("order incorrect. closing not first")
+	}
+	if q := <-Q; q != "closed" {
+		t.Error("order incorrect. closing not first")
+	}
+	if q := <-Q; q != "closed" {
+		t.Error("order incorrect. closing not first")
+	}
+}
+
+func TestChildFunc(t *testing.T) {
+	a := WithBackground()
+
+	wait1 := make(chan struct{})
+	wait2 := make(chan struct{})
+	wait3 := make(chan struct{})
+	wait4 := make(chan struct{})
+	go func() {
+		a.Close()
+		wait4 <- struct{}{}
+	}()
+
+	a.AddChildFunc(func(parent ContextGroup) {
+		wait1 <- struct{}{}
+		<-wait2
+		wait3 <- struct{}{}
+	})
+
+	<-wait1
+	select {
+	case <-wait3:
+		t.Error("should not be closed yet")
+	case <-wait4:
+		t.Error("should not be closed yet")
+	case <-a.Closed():
+		t.Error("should not be closed yet")
+	default:
+	}
+
+	wait2 <- struct{}{}
+
+	select {
+	case <-wait3:
+	case <-time.After(time.Second):
+		t.Error("should be closed now")
+	}
+
+	select {
+	case <-wait4:
+	case <-time.After(time.Second):
+		t.Error("should be closed now")
+	}
+}
+
+func TestTeardownCalledOnce(t *testing.T) {
+	a := setupCGHierarchy(nil)
+
+	onlyOnce := func() func() error {
+		count := 0
+		return func() error {
+			count++
+			if count > 1 {
+				t.Error("called", count, "times")
+			}
+			return nil
+		}
+	}
+
+	a.SetTeardown(onlyOnce())
+	a.c[0].SetTeardown(onlyOnce())
+	a.c[0].c[0].SetTeardown(onlyOnce())
+	a.c[0].c[1].SetTeardown(onlyOnce())
+	a.c[1].SetTeardown(onlyOnce())
+	a.c[1].c[0].SetTeardown(onlyOnce())
+	a.c[1].c[1].SetTeardown(onlyOnce())
+
+	a.c[0].c[0].Close()
+	a.c[0].c[0].Close()
+	a.c[0].c[0].Close()
+	a.c[0].c[0].Close()
+	a.c[0].Close()
+	a.c[0].Close()
+	a.c[0].Close()
+	a.c[0].Close()
+	a.Close()
+	a.Close()
+	a.Close()
+	a.Close()
+	a.c[1].Close()
+	a.c[1].Close()
+	a.c[1].Close()
+	a.c[1].Close()
+}
+
+func TestOnClosed(t *testing.T) {
+
+	ctx, cancel := context.WithCancel(context.Background())
+	a := setupCGHierarchy(ctx)
+	Q := make(chan string, 10)
+
+	onClosed := func(s string, c ContextGroup) {
+		<-c.Closed()
+		Q <- s
+	}
+
+	go onClosed("0", a.c[0])
+	go onClosed("10", a.c[1].c[0])
+	go onClosed("", a)
+	go onClosed("00", a.c[0].c[0])
+	go onClosed("1", a.c[1])
+	go onClosed("01", a.c[0].c[1])
+	go onClosed("11", a.c[1].c[1])
+
+	test := func(ss ...string) {
+		s1 := <-Q
+		for _, s2 := range ss {
+			if s1 == s2 {
+				return
+			}
+		}
+		t.Error("context not in group", s1, ss)
+	}
+
+	cancel()
+
+	test("00", "01", "10", "11")
+	test("00", "01", "10", "11")
+	test("00", "01", "10", "11")
+	test("00", "01", "10", "11")
+	test("0", "1")
+	test("0", "1")
+	test("")
+}

Godeps/_workspace/src/github.com/jbenet/go-router/LICENSE

@@ -0,0 +1,21 @@
+The MIT License (MIT)
+
+Copyright (c) 2014 Juan Batiz-Benet
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in
+all copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
+THE SOFTWARE.

Godeps/_workspace/src/github.com/jbenet/go-router/README.md

@@ -0,0 +1,8 @@
+# go-router - networking inspired muxing
+
+This is a networking-inspired model. It's useful for muxing and lots of other
+things. It enables you to construct your abstractions as you would compose
+computer networks (endpoints, switches, routing tables). These could represent
+processing workers, entire subsystems, or even real computers ;).
+
+See https://godoc.org/github.com/jbenet/go-router

Godeps/_workspace/src/github.com/jbenet/go-router/distance.go

@@ -0,0 +1,33 @@
+package router
+
+// HammingDistance is a DistanceFunc that interprets Addresses as strings
+// and uses their Hamming distance.
+// Return -1 if the Addresses are not strings, or strings length don't match.
+func HammingDistance(a1, a2 Address) int {
+	s1, ok := a1.(string)
+	if !ok {
+		return -1
+	}
+
+	s2, ok := a2.(string)
+	if !ok {
+		return -1
+	}
+
+	// runes not code points
+	r1 := []rune(s1)
+	r2 := []rune(s2)
+
+	// hamming distance requires equal length strings
+	if len(r1) != len(r2) {
+		return -1
+	}
+
+	d := 0
+	for i := range r1 {
+		if r1[i] != r2[i] {
+			d++
+		}
+	}
+	return d
+}

Godeps/_workspace/src/github.com/jbenet/go-router/impl.go

@@ -0,0 +1,81 @@
+package router
+
+import (
+	"errors"
+)
+
+type packet struct {
+	a Address
+	p interface{}
+}
+
+// ErrNoRoute signals when there is no Route to a destination
+var ErrNoRoute = errors.New("routing error: no route")
+
+// NewPacket constructs a trivial packet linking a destination Address to
+// an interface{} payload.
+func NewPacket(destination Address, payload interface{}) Packet {
+	return &packet{destination, payload}
+}
+
+func (p *packet) Destination() Address {
+	return p.a
+}
+
+func (p *packet) Payload() interface{} {
+	return p.p
+}
+
+// QueueNode is a trivial node, which accepts packets into a queue
+type QueueNode struct {
+	a Address
+	q chan Packet
+}
+
+// NewQueueNode constructs a node with an internal chan Packet queue
+func NewQueueNode(addr Address, q chan Packet) *QueueNode {
+	return &QueueNode{addr, q}
+}
+
+// Queue returns the chan Packet queue
+func (n *QueueNode) Queue() <-chan Packet {
+	return n.q
+}
+
+// Address returns the QueueNode's Address
+func (n *QueueNode) Address() Address {
+	return n.a
+}
+
+// HandlePacket consumes the incomng packet and adds it to the queue.
+func (n *QueueNode) HandlePacket(p Packet, s Node) error {
+	n.q <- p
+	return nil
+}
+
+type switchh struct {
+	addr   Address
+	router Router
+	nodes  []Node
+}
+
+// NewSwitch constructs a switch with given Router and list of adjacent Nodes.
+func NewSwitch(a Address, r Router, adj []Node) Switch {
+	return &switchh{a, r, adj}
+}
+
+func (s *switchh) Address() Address {
+	return s.addr
+}
+
+func (s *switchh) Router() Router {
+	return s.router
+}
+
+func (s *switchh) HandlePacket(p Packet, n Node) error {
+	next := s.router.Route(p)
+	if next != nil {
+		return next.HandlePacket(p, s)
+	}
+	return ErrNoRoute
+}

Godeps/_workspace/src/github.com/jbenet/go-router/interface.go

@@ -0,0 +1,81 @@
+// Package router is a networking-inspired routing model. It's useful for
+// muxing and lots of other things. It enables you to construct your
+// abstractions as you would compose computer networks (endpoints, switches,
+// routing tables). These could represent processing workers, entire
+// subsystems, or even real computers ;).
+package router
+
+// Address is our way of knowing where we're headed. Traditionally, addresses
+// are things like IP Addresses, or email addresses. But filepaths, or URLs
+// can be seen as addresses too. We really leave it up to you. Our routing
+// is general, and forces you to pick an addressing scheme, and some logic to
+// discriminate addresses that you'll plug into Routers (Address DistanceFunc).
+type Address interface{}
+
+// Packet is the unit of moving things in our network. Anything can be routed
+// in our network, as long as it has a Destination.
+type Packet interface {
+
+	// Destination is the Address of the endpoint this Packet is headed to.
+	// They could use the same Addressing throughout (recommended) like the
+	// internet, or face the pain of translating addresses at inter-network
+	// gateways.
+	Destination() Address
+
+	// Payload is here for completeness. This can be the Packet itself, but this
+	// function encourages the client to think through their packet design.
+	Payload() interface{}
+}
+
+// Node is an object which has interfaces to connect to networks. This
+// is an "endpoint" object.
+type Node interface {
+
+	// Address returns the node's address.
+	Address() Address
+
+	// HandlePacket receives a packet sent by another node.
+	HandlePacket(Packet, Node) error
+}
+
+// Switch is the basic forwarding device. It listens to all its interfaces (it
+// is a Node in our network) and Forwards all Packets received by any interface
+// out another interface, according to its ForwardingTable.
+type Switch interface {
+	Node
+
+	// Router returns the Router used to decide where to forward packets.
+	Router() Router
+}
+
+// Router is an object that decides how a Packet should be Routed. This should
+// be as close to a static table lookup as possible, meaning it would be best
+// to prepare a forwarding table in parallel, instead of blocking.
+//
+// Our Router captures the entire Control Plane, meaning that we can implement:
+// - Static Routing - forwarding table only
+// - Dynamic Routing - routing table computed with an algorithm or protocol
+// - "SDN" Routing - Control Plane separated from Data Plane
+// And even:
+// - URL Routers (like gorilla.Muxer)
+// - Protocol Muxers
+// entirely within different Router implementations.
+//
+// Note that this is a break from traditional networking systems. Instead of
+// having the abstractions of FIB, RIB, Routing/Forwarding Tables, Routers,
+// and Switches, we only have the last two:
+// - Router -- the things that "route" (decide where things go)
+// - Switch -- connecting Nodes, "switch" Packets according to a Router.
+type Router interface {
+
+	// Route decides how to route a Packet.
+	// It returns the next hop Node chosen to send the Packet to.
+	// Route may return nil, if no route is suitable at all (equivalent of drop).
+	Route(Packet) Node
+}
+
+// DistanceFunc returns a measure of distance between two Addresses.
+// Examples:
+// - masked longest prefix match (IP, CIDR)
+// - XOR distance (Kademlia)
+type DistanceFunc func(a, b Address) int

Godeps/_workspace/src/github.com/jbenet/go-router/ip/ip.go

@@ -0,0 +1,6 @@
+// Package ip provides a Table for ip matching
+package ip
+
+// import (
+// 	ipaddr "github.com/mikioh/ipaddr"
+// )

Godeps/_workspace/src/github.com/jbenet/go-router/ip/ip_test.go

@@ -0,0 +1 @@
+package ip

Godeps/_workspace/src/github.com/jbenet/go-router/pkt_test.go

@@ -0,0 +1,77 @@
+package router
+
+import (
+	"testing"
+)
+
+type mockNode struct {
+	addr string
+	pkts []Packet
+}
+
+func (m *mockNode) Address() Address {
+	return m.addr
+}
+
+func (m *mockNode) HandlePacket(p Packet, n Node) error {
+	m.pkts = append(m.pkts, p)
+	return nil
+}
+
+type mockPacket struct {
+	a string
+}
+
+func (m *mockPacket) Destination() Address {
+	return m.a
+}
+
+func (m *mockPacket) Payload() interface{} {
+	return m
+}
+
+func TestAddrs(t *testing.T) {
+
+	ta := func(a, b Address, expect int) {
+		actual := HammingDistance(a, b)
+		if actual != expect {
+			t.Error("address distance error:", a, b, expect, actual)
+		}
+	}
+
+	a := "abc"
+	b := "abc"
+	c := "abd"
+	d := "add"
+	e := "ddd"
+
+	ta(a, a, 0)
+	ta(a, b, 0)
+	ta(a, c, 1)
+	ta(a, d, 2)
+	ta(a, e, 3)
+}
+
+func TestNodes(t *testing.T) {
+
+	a := &mockPacket{"abc"}
+	b := &mockPacket{"abc"}
+	c := &mockPacket{"abd"}
+	d := &mockPacket{"add"}
+	e := &mockPacket{"ddd"}
+
+	n := &mockNode{addr: "abc"}
+	n2 := &mockNode{addr: "ddd"}
+	n.HandlePacket(a, n2)
+	n.HandlePacket(b, n2)
+	n.HandlePacket(c, n2)
+	n.HandlePacket(d, n2)
+	n.HandlePacket(e, n2)
+
+	pkts := []Packet{a, b, c, d, e}
+	for i, p := range pkts {
+		if n.pkts[i] != p {
+			t.Error("pkts not handled in order.")
+		}
+	}
+}

Godeps/_workspace/src/github.com/jbenet/go-router/table.go

@@ -0,0 +1,137 @@
+package router
+
+import (
+	"sync"
+)
+
+// TableEntry is an (Address, Node) pair for a routing Table
+type TableEntry interface {
+	Address() Address
+	NextHop() Node
+}
+
+func NewTableEntry(addr Address, nexthop Node) TableEntry {
+	return &tableEntry{addr, nexthop}
+}
+
+type tableEntry struct {
+	addr Address
+	next Node
+}
+
+func (te *tableEntry) Address() Address {
+	return te.addr
+}
+
+func (te *tableEntry) NextHop() Node {
+	return te.next
+}
+
+// Table is a Router (Routing Table, really) based on a distance criterion.
+//
+// For example:
+//
+//   n1 := NewQueueNode("aaa", make(chan Packet, 10))
+//   n2 := NewQueueNode("aba", make(chan Packet, 10))
+//   n3 := NewQueueNode("abc", make(chan Packet, 10))
+//
+//   var t router.Table
+//   t.Distance = router.HammingDistance
+//   t.AddNodes(n1, n2)
+//
+//   p1 := NewPacket("aaa", "hello1")
+//   p2 := NewPacket("aba", "hello2")
+//   p3 := NewPacket("abc", "hello3")
+//
+//   t.Route(p1) // n1
+//   t.Route(p2) // n2
+//   t.Route(p3) // n2, because we don't have n3 and n2 is closet
+//
+//   t.AddNode(n3)
+//   t.Route(p3) // n3
+type Table interface {
+	Router
+
+	// Entries are the entries in this routing table
+	Entries() []TableEntry
+
+	// Distance returns a measure of distance between two Addresses
+	Distance() DistanceFunc
+}
+
+type SimpleTable struct {
+	entries  []TableEntry
+	distance DistanceFunc
+	sync.RWMutex
+}
+
+// Entries are the entries in this routing table
+func (t *SimpleTable) Entries() []TableEntry {
+	return t.entries
+}
+
+// Distance returns a measure of distance between two Addresses.
+func (t *SimpleTable) Distance() DistanceFunc {
+	return t.distance
+}
+
+// AddEntry adds an (Address, NextHop) entry to the Table
+func (t *SimpleTable) AddEntry(addr Address, nextHop Node) {
+	t.Lock()
+	defer t.Unlock()
+	t.entries = append(t.entries, NewTableEntry(addr, nextHop))
+}
+
+// AddNode calls AddTableEntry for the given Node
+func (t *SimpleTable) AddNode(n Node) {
+	t.AddEntry(n.Address(), n)
+}
+
+// AddNodes calls AddTableEntry for the given Node
+func (t *SimpleTable) AddNodes(ns ...Node) {
+	t.Lock()
+	defer t.Unlock()
+	for _, n := range ns {
+		t.entries = append(t.entries, NewTableEntry(n.Address(), n))
+	}
+}
+
+// Route decides how to route a Packet out of a list of Nodes.
+// It returns the Node chosen to send the Packet to.
+// Route may return nil, if no route is suitable at all (equivalent of drop).
+func (t *SimpleTable) Route(p Packet) Node {
+	if t.entries == nil {
+		return nil
+	}
+
+	t.RLock()
+	defer t.RUnlock()
+
+	dist := t.Distance()
+	if dist == nil {
+		dist = equalDistance
+	}
+
+	var best Node
+	var bestDist int
+	var addr = p.Destination()
+
+	for _, e := range t.entries {
+		d := dist(e.Address(), addr)
+		if d < 0 {
+			continue
+		}
+		if best == nil || d < bestDist {
+			bestDist = d
+			best = e.NextHop()
+		}
+	}
+	return best
+}
+
+func equalDistance(a, b Address) int {
+	if a == b {
+		return 0
+	}
+	return -1
+}

Godeps/_workspace/src/github.com/jbenet/go-router/table_test.go

@@ -0,0 +1,80 @@
+package router
+
+import (
+	"testing"
+)
+
+func TestTable(t *testing.T) {
+
+	na := &mockNode{addr: "abc"}
+	nc := &mockNode{addr: "abd"}
+	nd := &mockNode{addr: "add"}
+	ne := &mockNode{addr: "ddd"}
+
+	pa := &mockPacket{"abc"}
+	pb := &mockPacket{"abc"}
+	pc := &mockPacket{"abd"}
+	pd := &mockPacket{"add"}
+	pe := &mockPacket{"ddd"}
+
+	table := &SimpleTable{
+		entries: []TableEntry{
+			&tableEntry{na.Address(), na},
+			&tableEntry{nc.Address(), nc},
+			&tableEntry{nd.Address(), nd},
+			&tableEntry{ne.Address(), ne},
+		},
+	}
+
+	s := NewSwitch("sss", table, []Node{na, nc, nd, ne})
+	s.HandlePacket(pa, na)
+	s.HandlePacket(pb, na)
+	s.HandlePacket(pc, na)
+	s.HandlePacket(pd, na)
+	s.HandlePacket(pe, na)
+
+	tt := func(n *mockNode, pkts []Packet) {
+		for i, p := range pkts {
+			if len(n.pkts) <= i {
+				t.Error("pkts not handled in order.", n, pkts)
+				return
+			}
+			if n.pkts[i] != p {
+				t.Error("pkts not handled in order.", n, pkts)
+			}
+		}
+	}
+
+	tt(na, []Packet{pa, pb})
+	tt(nc, []Packet{pc})
+	tt(nd, []Packet{pd})
+	tt(ne, []Packet{pe})
+}
+
+func TestTable2(t *testing.T) {
+
+	n1 := NewQueueNode("aaa", make(chan Packet, 1))
+	n2 := NewQueueNode("aba", make(chan Packet, 1))
+	n3 := NewQueueNode("abc", make(chan Packet, 1))
+
+	var tb SimpleTable
+	tb.distance = HammingDistance
+	tb.AddNodes(n1, n2)
+
+	p1 := NewPacket("aaa", "hello1")
+	p2 := NewPacket("aba", "hello2")
+	p3 := NewPacket("abc", "hello3")
+
+	testRoute := func(p Packet, expect Node) {
+		if tb.Route(p) != expect {
+			t.Error(p, "route should be", expect)
+		}
+	}
+
+	testRoute(p1, n1)
+	testRoute(p2, n2)
+	testRoute(p3, n2) // n2 because we don't have n3 and n2 is closet
+
+	tb.AddNode(n3)
+	testRoute(p3, n3)
+}