1
0
mirror of https://github.com/golang/go synced 2024-11-21 14:34:41 -07:00

An experimental implemenation of Ticker using two goroutines for all tickers.

Feel free to suggest other approaches.

R=rsc
CC=cw, golang-dev
https://golang.org/cl/193070
This commit is contained in:
Rob Pike 2010-01-29 08:34:40 +11:00
parent 8fedbb8c3d
commit d53b426fa0

View File

@ -4,81 +4,23 @@
package time
// TODO(rsc): This implementation of Tick is a
// simple placeholder. Eventually, there will need to be
// a single central time server no matter how many tickers
// are active.
//
// Also, if timeouts become part of the select statement,
// perhaps the Ticker is just:
//
// func Ticker(ns int64, c chan int64) {
// for {
// select { timeout ns: }
// nsec, err := Nanoseconds();
// c <- nsec;
// }
import (
"once"
)
// A Ticker holds a synchronous channel that delivers `ticks' of a clock
// at intervals.
type Ticker struct {
C <-chan int64 // The channel on which the ticks are delivered.
done chan bool
c chan<- int64 // The same channel, but the end we use.
ns int64
shutdown bool
nextTick int64
next *Ticker
}
// Stop turns off a ticker. After Stop, no more ticks will be sent.
func (t *Ticker) Stop() {
t.shutdown = true
go t.drain()
}
func (t *Ticker) drain() {
for {
select {
case <-t.C:
case <-t.done:
return
}
}
}
func (t *Ticker) ticker(c chan<- int64) {
now := Nanoseconds()
when := now
for !t.shutdown {
when += t.ns // next alarm
// if c <- now took too long, skip ahead
if when < now {
// one big step
when += (now - when) / t.ns * t.ns
}
for when <= now {
// little steps until when > now
when += t.ns
}
for !t.shutdown && when > now {
// limit individual sleeps so that stopped
// long-term tickers don't pile up.
const maxSleep = 1e9
if when-now > maxSleep {
Sleep(maxSleep)
} else {
Sleep(when - now)
}
now = Nanoseconds()
}
if t.shutdown {
break
}
c <- now
}
t.done <- true
}
func (t *Ticker) Stop() { t.shutdown = true }
// Tick is a convenience wrapper for NewTicker providing access to the ticking
// channel only. Useful for clients that have no need to shut down the ticker.
@ -89,15 +31,132 @@ func Tick(ns int64) <-chan int64 {
return NewTicker(ns).C
}
// Ticker returns a new Ticker containing a synchronous channel that will
type alarmer struct {
wakeUp chan bool // wakeup signals sent/received here
wakeMeAt chan int64
wakeTime int64
}
// Set alarm to go off at time ns, if not already set earlier.
func (a *alarmer) set(ns int64) {
// If there's no wakeLoop or the next tick we expect is too late, start a new wakeLoop
if a.wakeMeAt == nil || a.wakeTime > ns {
// Stop previous wakeLoop.
if a.wakeMeAt != nil {
a.wakeMeAt <- -1
}
a.wakeMeAt = make(chan int64, 10)
go wakeLoop(a.wakeMeAt, a.wakeUp)
a.wakeMeAt <- ns
}
}
// Channel to notify tickerLoop of new Tickers being created.
var newTicker chan *Ticker
func startTickerLoop() {
newTicker = make(chan *Ticker)
go tickerLoop()
}
// wakeLoop delivers ticks at scheduled times, sleeping until the right moment.
// If another, earlier Ticker is created while it sleeps, tickerLoop() will start a new
// wakeLoop but they will share the wakeUp channel and signal that this one
// is done by giving it a negative time request.
func wakeLoop(wakeMeAt chan int64, wakeUp chan bool) {
for {
wakeAt := <-wakeMeAt
if wakeAt < 0 { // tickerLoop has started another wakeLoop
return
}
now := Nanoseconds()
if wakeAt > now {
Sleep(wakeAt - now)
now = Nanoseconds()
}
wakeUp <- true
}
}
// A single tickerLoop serves all ticks to Tickers. It waits for two events:
// either the creation of a new Ticker or a tick from the alarm,
// signalling a time to wake up one or more Tickers.
func tickerLoop() {
// Represents the next alarm to be delivered.
var alarm alarmer
// All wakeLoops deliver wakeups to this channel.
alarm.wakeUp = make(chan bool, 10)
var now, prevTime, wakeTime int64
var tickers *Ticker
for {
select {
case t := <-newTicker:
// Add Ticker to list
t.next = tickers
tickers = t
// Arrange for a new alarm if this one precedes the existing one.
alarm.set(t.nextTick)
case <-alarm.wakeUp:
now = Nanoseconds()
// Ignore an old time due to a dying wakeLoop
if now < prevTime {
continue
}
wakeTime = now + 1e15 // very long in the future
var prev *Ticker = nil
// Scan list of tickers, delivering updates to those
// that need it and determining the next wake time.
// TODO(r): list should be sorted in time order.
for t := tickers; t != nil; t = t.next {
if t.shutdown {
// Ticker is done; remove it from list.
if prev == nil {
tickers = t.next
} else {
prev.next = t.next
}
continue
}
if tickers.nextTick <= now {
if len(t.c) == 0 {
// Only send if there's room. We must not block.
// The channel is allocated with a one-element
// buffer, which is sufficient: if he hasn't picked
// up the last tick, no point in sending more.
t.c <- now
}
t.nextTick += t.ns
if t.nextTick <= now {
// Still behind; advance in one big step.
t.nextTick += (now - t.nextTick + t.ns) / t.ns * t.ns
}
if t.nextTick > now && t.nextTick < wakeTime {
wakeTime = t.nextTick
}
}
prev = t
}
if tickers != nil {
// Please send wakeup at earliest required time.
// If there are no tickers, don't bother.
alarm.wakeMeAt <- wakeTime
}
}
prevTime = now
}
}
// Ticker returns a new Ticker containing a channel that will
// send the time, in nanoseconds, every ns nanoseconds. It adjusts the
// intervals to make up for pauses in delivery of the ticks.
func NewTicker(ns int64) *Ticker {
if ns <= 0 {
return nil
}
c := make(chan int64)
t := &Ticker{c, make(chan bool), ns, false}
go t.ticker(c)
c := make(chan int64, 1) // See comment on send in tickerLoop
t := &Ticker{c, c, ns, false, Nanoseconds() + ns, nil}
once.Do(startTickerLoop)
// must be run in background so global Tickers can be created
go func() { newTicker <- t }()
return t
}