我有一个设想。(Windows Forms, c#， .NET)
There is a main form which hosts some user control.
The user control does some heavy data operation, such that if I directly call the UserControl_Load method the UI become nonresponsive for the duration for load method execution.
To overcome this I load data on different thread (trying to change existing code as little as I can)
I used a background worker thread which will be loading the data and when done will notify the application that it has done its work.
Now came a real problem. All the UI (main form and its child usercontrols) was created on the primary main thread. In the LOAD method of the usercontrol I'm fetching data based on the values of some control (like textbox) on userControl.
伪代码看起来像这样:
代码1
UserContrl1_LoadDataMethod()
{
if (textbox1.text == "MyName") // This gives exception
{
//Load data corresponding to "MyName".
//Populate a globale variable List<string> which will be binded to grid at some later stage.
}
}
它给出的例外是
跨线程操作无效:从创建该控件的线程以外的线程访问的控件。
为了了解更多关于这一点，我做了一些谷歌搜索，并提出了一个建议，如使用以下代码
代码2
UserContrl1_LoadDataMethod()
{
if (InvokeRequired) // Line #1
{
this.Invoke(new MethodInvoker(UserContrl1_LoadDataMethod));
return;
}
if (textbox1.text == "MyName") // Now it won't give an exception
{
//Load data correspondin to "MyName"
//Populate a globale variable List<string> which will be binded to grid at some later stage
}
}
但我似乎还是回到了原点。再次申请
变得反应迟钝。这似乎是由于第1行if条件的执行。加载任务再次由父线程完成，而不是由我生成的第三个线程完成。
我不知道我认为这是对还是错。
我如何解决这个问题，以及执行第1行if块的效果是什么?
情况是这样的:我想根据控件的值将数据加载到全局变量中。我不想改变子线程中控件的值。我不会从子线程中做。
因此只有访问值，才能从数据库中获取相应的数据。
根据Prerak K的更新评论(已删除):
我想我没有恰当地提出这个问题。
情况是这样的:我想根据控件的值将数据加载到全局变量中。我不想改变子线程中控件的值。我不会从子线程中做。
因此只能访问值，以便从数据库中获取相应的数据。
你想要的解决方案应该是这样的:
UserContrl1_LOadDataMethod()
{
string name = "";
if(textbox1.InvokeRequired)
{
textbox1.Invoke(new MethodInvoker(delegate { name = textbox1.text; }));
}
if(name == "MyName")
{
// do whatever
}
}
在尝试切换回控件的线程之前，在单独的线程中进行认真的处理。例如:
UserContrl1_LOadDataMethod()
{
if(textbox1.text=="MyName") //<<======Now it wont give exception**
{
//Load data correspondin to "MyName"
//Populate a globale variable List<string> which will be
//bound to grid at some later stage
if(InvokeRequired)
{
// after we've done all the processing,
this.Invoke(new MethodInvoker(delegate {
// load the control with the appropriate data
}));
return;
}
}
}
这里有一种替代方法，如果您正在处理的对象没有
(InvokeRequired)
如果您使用的是类中的主窗体，而不是主窗体中的对象，但该对象在主窗体中，但没有invokerrequired，那么这是非常有用的
delegate void updateMainFormObject(FormObjectType objectWithoutInvoke, string text);
private void updateFormObjectType(FormObjectType objectWithoutInvoke, string text)
{
MainForm.Invoke(new updateMainFormObject(UpdateObject), objectWithoutInvoke, text);
}
public void UpdateObject(ToolStripStatusLabel objectWithoutInvoke, string text)
{
objectWithoutInvoke.Text = text;
}
它的工作原理与上面的相同，但是如果你没有一个带有invokerequired的对象，但是可以访问MainForm，那么它是一种不同的方法
当我在xamarin工作室外的一个visual studio winforms原型项目中编程iOS-Phone单触摸应用程序控制器时，我发现了这个需求。比起xamarin studio，我更倾向于在VS中编程，我希望控制器能够与手机框架完全分离。通过这种方式，在Android和Windows Phone等其他框架上实现这个功能将更容易用于未来的使用。
我想要一种解决方案，使GUI可以响应事件，而无需处理每次单击按钮背后的跨线程切换代码。基本上让类控制器来处理，以保持客户端代码简单。你可能在GUI上有很多事件，如果你可以在类的一个地方处理它，会更干净。我不是一个多头专家，让我知道，如果这是有缺陷的。
public partial class Form1 : Form
{
private ExampleController.MyController controller;
public Form1()
{
InitializeComponent();
controller = new ExampleController.MyController((ISynchronizeInvoke) this);
controller.Finished += controller_Finished;
}
void controller_Finished(string returnValue)
{
label1.Text = returnValue;
}
private void button1_Click(object sender, EventArgs e)
{
controller.SubmitTask("Do It");
}
}
GUI表单不知道控制器正在运行异步任务。
public delegate void FinishedTasksHandler(string returnValue);
public class MyController
{
private ISynchronizeInvoke _syn;
public MyController(ISynchronizeInvoke syn) { _syn = syn; }
public event FinishedTasksHandler Finished;
public void SubmitTask(string someValue)
{
System.Threading.ThreadPool.QueueUserWorkItem(state => submitTask(someValue));
}
private void submitTask(string someValue)
{
someValue = someValue + " " + DateTime.Now.ToString();
System.Threading.Thread.Sleep(5000);
//Finished(someValue); This causes cross threading error if called like this.
if (Finished != null)
{
if (_syn.InvokeRequired)
{
_syn.Invoke(Finished, new object[] { someValue });
}
else
{
Finished(someValue);
}
}
}
}
UI中的线程模型
为了理解基本概念，请阅读UI应用程序中的线程模型(旧的VB链接在这里)。该链接导航到描述WPF线程模型的页面。但是，Windows窗体利用了相同的思想。
UI线程
只有一个线程(UI线程)被允许访问System.Windows.Forms.Control及其子类成员。
试图从不同的线程访问System.Windows.Forms.Control的成员会导致跨线程异常。
因为只有一个线程，所以所有的UI操作都作为工作项进入该线程:
如果UI线程没有工作，那么就存在空闲间隙，可以由与UI无关的计算使用。
为了使用上述间隙，请使用System.Windows.Forms.Control.Invoke或System.Windows.Forms.Control.BeginInvoke方法:
BeginInvoke和Invoke方法
The computing overhead of method being invoked should be small as well as computing overhead of event handler methods because the UI thread is used there - the same that is responsible for handling user input. Regardless if this is System.Windows.Forms.Control.Invoke or System.Windows.Forms.Control.BeginInvoke.
To perform computing expensive operation always use separate thread. Since .NET 2.0 BackgroundWorker is dedicated to performing computing expensive operations in Windows Forms. However in new solutions you should use the async-await pattern as described here.
Use System.Windows.Forms.Control.Invoke or System.Windows.Forms.Control.BeginInvoke methods only to update a user interface. If you use them for heavy computations, your application will block:
调用
System.Windows.Forms.Control.Invoke导致单独的线程等待，直到调用的方法完成:
BeginInvoke
System.Windows.Forms.Control.BeginInvoke不会导致单独的线程等待被调用的方法完成:
代码的解决方案
阅读问题的答案如何更新GUI从另一个线程在c# ?
对于c# 5.0和。net 4.5，推荐的解决方案在这里。
和之前的答案一样，
但是一个非常简短的添加，允许使用所有的Control属性而没有跨线程调用异常。
辅助方法
/// <summary>
/// Helper method to determin if invoke required, if so will rerun method on correct thread.
/// if not do nothing.
/// </summary>
/// <param name="c">Control that might require invoking</param>
/// <param name="a">action to preform on control thread if so.</param>
/// <returns>true if invoke required</returns>
public bool ControlInvokeRequired(Control c, Action a)
{
if (c.InvokeRequired) c.Invoke(new MethodInvoker(delegate
{
a();
}));
else return false;
return true;
}
示例使用
// usage on textbox
public void UpdateTextBox1(String text)
{
//Check if invoke requied if so return - as i will be recalled in correct thread
if (ControlInvokeRequired(textBox1, () => UpdateTextBox1(text))) return;
textBox1.Text = ellapsed;
}
//Or any control
public void UpdateControl(Color c, String s)
{
//Check if invoke requied if so return - as i will be recalled in correct thread
if (ControlInvokeRequired(myControl, () => UpdateControl(c, s))) return;
myControl.Text = s;
myControl.BackColor = c;
}
使用Async/Await和回调的新外观。如果在项目中保留扩展方法，则只需要一行代码。
/// <summary>
/// A new way to use Tasks for Asynchronous calls
/// </summary>
public class Example
{
/// <summary>
/// No more delegates, background workers etc. just one line of code as shown below
/// Note it is dependent on the XTask class shown next.
/// </summary>
public async void ExampleMethod()
{
//Still on GUI/Original Thread here
//Do your updates before the next line of code
await XTask.RunAsync(() =>
{
//Running an asynchronous task here
//Cannot update GUI Thread here, but can do lots of work
});
//Can update GUI/Original thread on this line
}
}
/// <summary>
/// A class containing extension methods for the Task class
/// Put this file in folder named Extensions
/// Use prefix of X for the class it Extends
/// </summary>
public static class XTask
{
/// <summary>
/// RunAsync is an extension method that encapsulates the Task.Run using a callback
/// </summary>
/// <param name="Code">The caller is called back on the new Task (on a different thread)</param>
/// <returns></returns>
public async static Task RunAsync(Action Code)
{
await Task.Run(() =>
{
Code();
});
return;
}
}
你可以在Extension方法中添加其他东西，比如用Try/Catch语句包装它，允许调用者告诉它在完成后返回什么类型，一个异常回调给调用者:
添加Try Catch，自动异常记录和回调
/// <summary>
/// Run Async
/// </summary>
/// <typeparam name="T">The type to return</typeparam>
/// <param name="Code">The callback to the code</param>
/// <param name="Error">The handled and logged exception if one occurs</param>
/// <returns>The type expected as a competed task</returns>
public async static Task<T> RunAsync<T>(Func<string,T> Code, Action<Exception> Error)
{
var done = await Task<T>.Run(() =>
{
T result = default(T);
try
{
result = Code("Code Here");
}
catch (Exception ex)
{
Console.WriteLine("Unhandled Exception: " + ex.Message);
Console.WriteLine(ex.StackTrace);
Error(ex);
}
return result;
});
return done;
}
public async void HowToUse()
{
//We now inject the type we want the async routine to return!
var result = await RunAsync<bool>((code) => {
//write code here, all exceptions are logged via the wrapped try catch.
//return what is needed
return someBoolValue;
},
error => {
//exceptions are already handled but are sent back here for further processing
});
if (result)
{
//we can now process the result because the code above awaited for the completion before
//moving to this statement
}
}
