Go

Memory pooling

Introduction#

sync.Pool stores a cache of allocated but unused items for future use, avoiding memory churn for frequently changed collections, and allowing efficient, thread-safe re-use of memory. It is useful to manage a group of temporary items shared between concurrent clients of a package, for example a list of database connections or a list of output buffers.

sync.Pool

Using sync.Pool structure we can pool objects and reuse them.

package main

import (
    "bytes"
    "fmt"
    "sync"
)

var pool = sync.Pool{
    // New creates an object when the pool has nothing available to return.
    // New must return an interface{} to make it flexible. You have to cast
    // your type after getting it.
    New: func() interface{} {
        // Pools often contain things like *bytes.Buffer, which are
        // temporary and re-usable.
        return &bytes.Buffer{}
    },
}

func main() {
    // When getting from a Pool, you need to cast
    s := pool.Get().(*bytes.Buffer)
    // We write to the object
    s.Write([]byte("dirty"))
    // Then put it back
    pool.Put(s)

    // Pools can return dirty results

    // Get 'another' buffer
    s = pool.Get().(*bytes.Buffer)
    // Write to it
    s.Write([]bytes("append"))
    // At this point, if GC ran, this buffer *might* exist already, in
    // which case it will contain the bytes of the string "dirtyappend"
    fmt.Println(s)
    // So use pools wisely, and clean up after yourself
    s.Reset()
    pool.Put(s)

    // When you clean up, your buffer should be empty
    s = pool.Get().(*bytes.Buffer)
    // Defer your Puts to make sure you don't leak!
    defer pool.Put(s)
    s.Write([]byte("reset!"))
    // This prints "reset!", and not "dirtyappendreset!"
    fmt.Println(s)
}

This modified text is an extract of the original Stack Overflow Documentation created by the contributors and released under CC BY-SA 3.0 This website is not affiliated with Stack Overflow