ThreadStaticAttribute Classe

Définition

public ref class ThreadStaticAttribute : Attribute

[System.AttributeUsage(System.AttributeTargets.Field, Inherited=false)]
public class ThreadStaticAttribute : Attribute

[System.AttributeUsage(System.AttributeTargets.Field, Inherited=false)]
[System.Serializable]
public class ThreadStaticAttribute : Attribute

[System.AttributeUsage(System.AttributeTargets.Field, Inherited=false)]
[System.Serializable]
[System.Runtime.InteropServices.ComVisible(true)]
public class ThreadStaticAttribute : Attribute

[<System.AttributeUsage(System.AttributeTargets.Field, Inherited=false)>]
type ThreadStaticAttribute = class
    inherit Attribute

[<System.AttributeUsage(System.AttributeTargets.Field, Inherited=false)>]
[<System.Serializable>]
type ThreadStaticAttribute = class
    inherit Attribute

[<System.AttributeUsage(System.AttributeTargets.Field, Inherited=false)>]
[<System.Serializable>]
[<System.Runtime.InteropServices.ComVisible(true)>]
type ThreadStaticAttribute = class
    inherit Attribute

Public Class ThreadStaticAttribute
Inherits Attribute

Héritage

Object

Attribute

ThreadStaticAttribute

Attributs

AttributeUsageAttribute SerializableAttribute ComVisibleAttribute

Exemples

L’exemple suivant instancie un générateur de nombres aléatoires, crée dix threads en plus du thread principal, puis génère deux millions de nombres aléatoires dans chaque thread. Il utilise l’attribut pour calculer la ThreadStaticAttribute somme et le nombre de nombres aléatoires par thread. Il définit également deux champs supplémentaires par thread et previous abnormal, ce qui lui permet de détecter l’altération du générateur de nombres aléatoires.

using System;
using System.Threading;

public class Example
{
   [ThreadStatic] static double previous = 0.0;
   [ThreadStatic] static double sum = 0.0;
   [ThreadStatic] static int calls = 0;
   [ThreadStatic] static bool abnormal;
   static int totalNumbers = 0;
   static CountdownEvent countdown;
   private static Object lockObj;
   Random rand;
   
   public Example()
   { 
      rand = new Random();
      lockObj = new Object();
      countdown = new CountdownEvent(1);
   } 

   public static void Main()
   {
      Example ex = new Example();
      Thread.CurrentThread.Name = "Main";
      ex.Execute();
      countdown.Wait();
      Console.WriteLine("{0:N0} random numbers were generated.", totalNumbers);
   }

   private void Execute()
   {   
      for (int threads = 1; threads <= 10; threads++)
      {
         Thread newThread = new Thread(new ThreadStart(this.GetRandomNumbers));
         countdown.AddCount();
         newThread.Name = threads.ToString();
         newThread.Start();
      }
      this.GetRandomNumbers();
   }

   private void GetRandomNumbers()
   {
      double result = 0.0;

      for (int ctr = 0; ctr < 2000000; ctr++)
      {
         lock (lockObj) {
            result = rand.NextDouble();
            calls++;
            Interlocked.Increment(ref totalNumbers);
            // We should never get the same random number twice.
            if (result == previous) {
               abnormal = true;
               break;
            }
            else {
               previous = result;
               sum += result;
            }   
         }
      }
      // get last result
      if (abnormal)
         Console.WriteLine("Result is {0} in {1}", previous, Thread.CurrentThread.Name);
       
      Console.WriteLine("Thread {0} finished random number generation.", Thread.CurrentThread.Name);
      Console.WriteLine("Sum = {0:N4}, Mean = {1:N4}, n = {2:N0}\n", sum, sum/calls, calls);        
      countdown.Signal();
   }
}
// The example displays output similar to the following:
//    Thread 1 finished random number generation.
//    Sum = 1,000,556.7483, Mean = 0.5003, n = 2,000,000
//    
//    Thread 6 finished random number generation.
//    Sum = 999,704.3865, Mean = 0.4999, n = 2,000,000
//    
//    Thread 2 finished random number generation.
//    Sum = 999,680.8904, Mean = 0.4998, n = 2,000,000
//    
//    Thread 10 finished random number generation.
//    Sum = 999,437.5132, Mean = 0.4997, n = 2,000,000
//    
//    Thread 8 finished random number generation.
//    Sum = 1,000,663.7789, Mean = 0.5003, n = 2,000,000
//    
//    Thread 4 finished random number generation.
//    Sum = 999,379.5978, Mean = 0.4997, n = 2,000,000
//    
//    Thread 5 finished random number generation.
//    Sum = 1,000,011.0605, Mean = 0.5000, n = 2,000,000
//    
//    Thread 9 finished random number generation.
//    Sum = 1,000,637.4556, Mean = 0.5003, n = 2,000,000
//    
//    Thread Main finished random number generation.
//    Sum = 1,000,676.2381, Mean = 0.5003, n = 2,000,000
//    
//    Thread 3 finished random number generation.
//    Sum = 999,951.1025, Mean = 0.5000, n = 2,000,000
//    
//    Thread 7 finished random number generation.
//    Sum = 1,000,844.5217, Mean = 0.5004, n = 2,000,000
//    
//    22,000,000 random numbers were generated.

open System
open System.Threading

type Example() =
    [<ThreadStatic; DefaultValue>] 
    static val mutable private previous : double

    [<ThreadStatic; DefaultValue>] 
    static val mutable private sum : double
    
    [<ThreadStatic; DefaultValue>] 
    static val mutable private calls : int

    [<ThreadStatic; DefaultValue>] 
    static val mutable private abnormal : bool
   
    static let mutable totalNumbers = 0
    static let countdown = new CountdownEvent(1)
    static let lockObj = obj ()
    let rand = Random()


    member this.Execute() =
        for threads = 1 to 10 do
            let newThread = new Thread(ThreadStart this.GetRandomNumbers)
            countdown.AddCount()
            newThread.Name <- threads.ToString()
            newThread.Start()
        this.GetRandomNumbers()
        countdown.Wait()
        printfn $"{totalNumbers:N0} random numbers were generated."

    member _.GetRandomNumbers() =
        let mutable i = 0
        while i < 2000000 do
            lock lockObj (fun () ->
                let result = rand.NextDouble()
                Example.calls <- Example.calls + 1
                Interlocked.Increment &totalNumbers |> ignore
                // We should never get the same random number twice.
                if result = Example.previous then
                    Example.abnormal <- true
                    i <- 2000001 // break
                else
                    Example.previous <- result
                    Example.sum <- Example.sum + result )
            i <- i + 1
        // get last result
        if Example.abnormal then
            printfn $"Result is {Example.previous} in {Thread.CurrentThread.Name}"
        
        printfn $"Thread {Thread.CurrentThread.Name} finished random number generation."
        printfn $"Sum = {Example.sum:N4}, Mean = {Example.sum / float Example.calls:N4}, n = {Example.calls:N0}\n"
        countdown.Signal() |> ignore

let ex = Example()
Thread.CurrentThread.Name <- "Main"
ex.Execute()

// The example displays output similar to the following:
//    Thread 1 finished random number generation.
//    Sum = 1,000,556.7483, Mean = 0.5003, n = 2,000,000
//    
//    Thread 6 finished random number generation.
//    Sum = 999,704.3865, Mean = 0.4999, n = 2,000,000
//    
//    Thread 2 finished random number generation.
//    Sum = 999,680.8904, Mean = 0.4998, n = 2,000,000
//    
//    Thread 10 finished random number generation.
//    Sum = 999,437.5132, Mean = 0.4997, n = 2,000,000
//    
//    Thread 8 finished random number generation.
//    Sum = 1,000,663.7789, Mean = 0.5003, n = 2,000,000
//    
//    Thread 4 finished random number generation.
//    Sum = 999,379.5978, Mean = 0.4997, n = 2,000,000
//    
//    Thread 5 finished random number generation.
//    Sum = 1,000,011.0605, Mean = 0.5000, n = 2,000,000
//    
//    Thread 9 finished random number generation.
//    Sum = 1,000,637.4556, Mean = 0.5003, n = 2,000,000
//    
//    Thread Main finished random number generation.
//    Sum = 1,000,676.2381, Mean = 0.5003, n = 2,000,000
//    
//    Thread 3 finished random number generation.
//    Sum = 999,951.1025, Mean = 0.5000, n = 2,000,000
//    
//    Thread 7 finished random number generation.
//    Sum = 1,000,844.5217, Mean = 0.5004, n = 2,000,000
//    
//    22,000,000 random numbers were generated.

Imports System.Threading

Public Class Example
   <ThreadStatic> Shared previous As Double = 0.0
   <ThreadStatic> Shared sum As Double = 0.0
   <ThreadStatic> Shared calls As Integer = 0
   <ThreadStatic> Shared abnormal As Boolean
   Shared totalNumbers As Integer = 0
   Shared countdown As CountdownEvent
   Private Shared lockObj As Object
   Dim rand As Random

   Public Sub New()
      rand = New Random()
      lockObj = New Object()
      countdown = New CountdownEvent(1)
   End Sub

   Public Shared Sub Main()
      Dim ex As New Example()
      Thread.CurrentThread.Name = "Main"
      ex.Execute()
      countdown.Wait()
      Console.WriteLine("{0:N0} random numbers were generated.", totalNumbers)
   End Sub

   Private Sub Execute()
      For threads As Integer = 1 To 10
         Dim newThread As New Thread(New ThreadStart(AddressOf GetRandomNumbers))
         countdown.AddCount()
         newThread.Name = threads.ToString()
         newThread.Start()
      Next
      Me.GetRandomNumbers()
   End Sub

   Private Sub GetRandomNumbers()
      Dim result As Double = 0.0
      
       
      For ctr As Integer = 1 To 2000000
         SyncLock lockObj
            result = rand.NextDouble()
            calls += 1
            Interlocked.Increment(totalNumbers)
            ' We should never get the same random number twice.
            If result = previous Then
               abnormal = True
               Exit For
            Else
               previous = result
               sum += result
            End If   
         End SyncLock
      Next
      ' Get last result.
      If abnormal Then
         Console.WriteLine("Result is {0} in {1}", previous, Thread.CurrentThread.Name)
      End If       
      
      Console.WriteLine("Thread {0} finished random number generation.", Thread.CurrentThread.Name)
      Console.WriteLine("Sum = {0:N4}, Mean = {1:N4}, n = {2:N0}", sum, sum/calls, calls)
      Console.WriteLine()        
      countdown.Signal()
   End Sub
End Class
' The example displays output similar to the following:
'    Thread 1 finished random number generation.
'    Sum = 1,000,556.7483, Mean = 0.5003, n = 2,000,000
'    
'    Thread 6 finished random number generation.
'    Sum = 999,704.3865, Mean = 0.4999, n = 2,000,000
'    
'    Thread 2 finished random number generation.
'    Sum = 999,680.8904, Mean = 0.4998, n = 2,000,000
'    
'    Thread 10 finished random number generation.
'    Sum = 999,437.5132, Mean = 0.4997, n = 2,000,000
'    
'    Thread 8 finished random number generation.
'    Sum = 1,000,663.7789, Mean = 0.5003, n = 2,000,000
'    
'    Thread 4 finished random number generation.
'    Sum = 999,379.5978, Mean = 0.4997, n = 2,000,000
'    
'    Thread 5 finished random number generation.
'    Sum = 1,000,011.0605, Mean = 0.5000, n = 2,000,000
'    
'    Thread 9 finished random number generation.
'    Sum = 1,000,637.4556, Mean = 0.5003, n = 2,000,000
'    
'    Thread Main finished random number generation.
'    Sum = 1,000,676.2381, Mean = 0.5003, n = 2,000,000
'    
'    Thread 3 finished random number generation.
'    Sum = 999,951.1025, Mean = 0.5000, n = 2,000,000
'    
'    Thread 7 finished random number generation.
'    Sum = 1,000,844.5217, Mean = 0.5004, n = 2,000,000
'    
'    22,000,000 random numbers were generated.

L’exemple utilise l’instruction lock C#, la lock fonction en F#et la SyncLock construction dans Visual Basic pour synchroniser l’accès au générateur de nombres aléatoires. Cela empêche l’altération du générateur de nombres aléatoires, ce qui entraîne généralement son retour d’une valeur de zéro pour tous les appels suivants.

L’exemple utilise également la CountdownEvent classe pour s’assurer que chaque thread a fini de générer des nombres aléatoires avant d’afficher le nombre total d’appels. Sinon, si le thread principal termine l’exécution avant les threads supplémentaires qu’il génère, il affiche une valeur inexacte pour le nombre total d’appels de méthode.

Remarques

Un static champ marqué avec ThreadStaticAttribute n’est pas partagé entre les threads. Chaque thread en cours d’exécution a une instance distincte du champ et définit et obtient des valeurs pour ce champ. Si le champ est accessible sur un autre thread, il contient une valeur différente.

Notez que, en plus d’appliquer l’attribut ThreadStaticAttribute à un champ, vous devez également le définir comme un static champ (en C# ou F#) ou un Shared champ (dans Visual Basic).

Utilisez cet attribut comme c’est le cas et ne dérivez pas de celui-ci.

Pour plus d’informations sur l’utilisation d’attributs, consultez Attributs.

Constructeurs

Propriétés

Méthodes

Implémentations d’interfaces explicites

S’applique à

Voir aussi

• Attribute
• Thread
• AsyncLocal<T>
• Extension des métadonnées à l'aide des attributs
• Threading managé