I'm creating my own UI handling logic that runs on a single thread. Basically, what I'm doing is
void HandlerMain()
{
while (true)
{
Task[] events = PumpEvents();
Task.WaitAll(events);
}
}
where one example task that PumpEvents() returns is
async Task ButtonClick()
{
var httpClient = new HttpClient();
string s = await httpClient.GetStringAsync("http://microsoft.com");
Console.WriteLine(s);
}
The problem of my code is, if one of events takes a long time, it's stuck at Task.WaitAll(), so it can't pump new events making the UI not responsive. Is there any other method than WaitAll() something like
Task[] remainingEvents = PumpEvents();
while (true)
{
remainingEvents = WaitUntilEveryTasksAwait(remainingEvents);
remainingEvents.Append(PumpEvents());
}
Maybe I'm on wrong track. I'd appreciate your advice!
#ScottChamberlain No. The built-in UI processor, say WPF, correctly handles async events so whenever the async events do "await", it skips the current context and handles the next events. I want to duplicate this behavior
Based on your comment to me I now understand what your problem is. You need more logic than your simple while(true) loop to be able to process the await messages. The entire system is built up upon the class SynchronizationContext, what you will need to do is derive your own class from SynchronizationContext and override it's methods to queue up your work to be done inside your while loop.
See this article from Stephen Cleary to give you more information on how a Synchronization Context works and posibly some ideas on where to start writing your own.
If I'm understanding, you don't actually need to wait until those tasks are complete to continue. Just remove the call to Task.WaitAll.
For the record, Task.WaitAll is synchronous -- it blocks. You're better off using Task.WhenAll, which returns a task that completes when all of the provided tasks are complete. That lets you await it just like you'd await any of the individual tasks. Just doing that would solve the problem and keep the UI responsive.
E.g.:
async Task HandlerMain()
{
while (true)
{
Task[] events = PumpEvents();
await Task.WhenAll(events);
}
}
Of course, in this case, depending on what was calling HandlerMain, you'd still have the problem where PumpEvents was hung up waiting on the long-running task.
This starts to lead down the path of questions like, "Why are you writing your own custom UI message pump when WPF and WinForms have that problem already solved?"
Related
I have WinForms app where button click calls some async method of external library.
private async void button1_Click(object sender, EventArgs e)
{
await CallLibraryAsync();
}
private static async Task CallLibraryAsync()
{
var library = new Library();
await library.DoSomethingAsync();
}
The library looks like this:
public class Library
{
public async Task DoSomethingAsync()
{
Thread.Sleep(2000);
await Task.Delay(1000).ConfigureAwait(false);
// some other code
}
}
Before any asynchronous code there is some calculation simulated by Thread.Sleep call. In that case this call will block UI thread for 2 seconds. I have no option to change the code in DoSomethingAsync.
If I want to solve blocking problem, I could call the library in Task.Run like this:
private static async Task CallLibraryAsync()
{
var library = new Library();
// added Task.Run
await Task.Run(() => library.DoSomethingAsync());
}
It solves the problem, UI is not blocke anymore, but I've consumed one thread from ThreadPool. It is not good solution.
If I want to solve this problem without another thread, I can do something like this:
private static async Task CallLibraryAsync()
{
var library = new Library();
// added
await YieldOnlyAsync().ConfigureAwait(false);
await library.DoSomethingAsync();
}
// added
private static async Task YieldOnlyAsync()
{
await Task.Yield();
}
This solution works. Task.Yield() causes that method YieldOnlyAsync() always runs asynchronously and ConfigureAwait(false) causes that next code (await library.DoSomethingAsync();) runs on some ThreadPool thread, not UI thread.
But it is quite complicated solution. Is there any simpler?
Edit:
If the library method looks like this
public class Library
{
public async Task DoSomethingAsync()
{
await Task.Delay(1000).ConfigureAwait(false);
Thread.Sleep(2000);
await Task.Delay(1000);
// some other code
}
}
UI thread would not be blocked and I do not need to do anything. But that's the problem that it is some implementation detail I do not see directly because that could be in some nuget package. When I see that the UI freezes in some situations, I may find this problem (mean CPU-bound calculation before any await in async method) just after some investigation. There is no Wait() or Result, that would be easy to find, this is more problematic.
What I would like is to be prepared for that situation if possible in some simpler way. And that's why I do not want to use Task.Run whenewer I call some third-party library.
If I want to solve blocking problem, I could call the library in Task.Run like this:
It solves the problem, UI is not blocke anymore, but I've consumed one thread from ThreadPool. It is not good solution.
This is exactly what you want to do in a WinForms app. CPU-intensive code should be moved to a separate thread to free up the UI thread. There isn't any downside to consuming a new thread in WinForms.
Use Task.Run to move it to a different thread, and wait asynchronously from the UI thread for it to complete.
To quote the Asynchronous programming article from Microsoft:
If the work you have is CPU-bound and you care about responsiveness, use async and await, but spawn off the work on another thread with Task.Run.
I have no option to change the code
People say that, but you might not actually be hamstrung thus..
Here's a simple app with the same problem you face:
It's definitely pretty sleepy:
So let's whack it into ILSpy with the Reflexil plugin loaded:
We can perhaps shorten that timeout a bit.. Right click, Edit..
Make it 1ms, Right click the assembly and Save As..
That's a bit quicker!
Have a play, NOP it out etc..
You wrote:
If I want to solve blocking problem, I could call the library in Task.Run like this:
private static async Task CallLibraryAsync()
{
var library = new Library();
// added Task.Run
await Task.Run(() => library.DoSomethingAsync());
}
It solves the problem, UI is not blocked anymore, but I've consumed one thread from ThreadPool. It is not good solution.
(emphasis added)
...and then you proceed with inventing a convoluted hack that does the same thing: offloads the invocation of the DoSomethingAsync method to the ThreadPool. So you either want to:
Invoke the DoSomethingAsync method without using any thread at all, or
Invoke the DoSomethingAsync method on a non-ThreadPool thread.
The first is impossible. You can't invoke a method without using a thread. Code runs on CPUs, not on thin air. The second can be done in many ways, with the easiest being to use the Task.Factory.StartNew method, in combination with the LongRunning flag:
await Task.Factory.StartNew(() => library.DoSomethingAsync(), default,
TaskCreationOptions.LongRunning, TaskScheduler.Default).Unwrap();
This way you will invoke the DoSomethingAsync on a newly created thread, which will be destroyed immediately after the invocation of the method has completed. To be clear, the thread will be destroyed when the invocation completes, not when the asynchronous operation completes. Based on the DoSomethingAsync implementation that you have included in the question (the first one), the invocation will complete immediately after creating the Task.Delay(1000) task, and initiating the await of this task. There will be nothing for the thread to do after this point, so it will be recycled.
Side notes:
The CallLibraryAsync method violates the guideline for not exposing asynchronous wrappers for synchronous methods. Since the DoSomethingAsync method is implemented as partially synchronous and partially asynchronous, the guideline still applies IMHO.
If you like the idea of controlling imperatively the current context, instead of controlling it with wrappers like the Task.Run method, you could check out this question: Why was SwitchTo removed from Async CTP / Release? There are (not very many) people who like it as well, and there are libraries available that make it possible (SwitchTo - Microsoft.VisualStudio.Threading).
When you use async/await for I/O or CPU-bound operations, your UI thread will not blocked. In your example, you use Thread.Sleep(2000);command for simulating your CPU-bound operations but this will block your thread-pool thread not UI thread. You can use Task.Delay(2000); for simulating your I/O operations without blocking thread-pool thread.
I am pretty sure, this was answered already and I read many related stuff but somehow I am not getting it to work in my code. Here is the exact code base.
I have this async method in my library, that returns a string upon doing some DB entries:
public class MyLibrary
{
public async Task<string> DoSomethingAsync()
{
return await DoAsync();
}
// some private method, with multiple parameters
private Task<string> DoAsync()
{
return Task.Run(() => Do());
}
}
Now on UI, below resulted a frozen state or deadlock:
var myTask = MyLibraryobject.DoSomethingAsync();
console.Write(myTask.Result);
Since my call is awaitable, I think, UI thread waits for my call to finish its business and populate the result, doesn't it? Or since the call is running on another thread, the task might not yet completed when my cursor hits the line 2. So what now? the cursor will wait till the task gets completed or the line 2 gets executed whenever the task gets completed? sounds synchronous to me..
Also if at all, I want to 'exlicitly' wait till the task gets finished, how do i enforce that? some posts, suggested in doing like below, which gave me the result, but created some more confusion:
var myTask = Task.Run(async () => await MyLibraryobject.DoSomethingAsync());
myTask.Wait();
console.Write(myTask.Result);
What is happening above? why should i create another task and set it to wait? can't I wait for the thread, used by the async method? Again, i am clearly missing some basic stuff here.
Lastly, this is an async call, but the UI thread is waiting for it to be completed thus gives me an impression that it's synchronous. Am I thinking it right? Then what is the main purpose of async methods, which return something, and UI waits for it to complete?
Also what's the difference between a fire-and-forget call being non-async and async? is there any benefit of returning Task, instead of void, for such calls?
I am pretty sure, some broken link that connects all these pieces together in my brain. Can someone explain that to me, please??
You should probably start with my async intro and follow up with my article on async best practices. They answer almost all your questions.
For the details...
how and when to fetch return from awaitable(async) method
You should use await to get the result from an async task.
I have this async method in my library, that returns a string upon doing some DB entries
Then it should not be using Task.Run. It should be using naturally-asynchronous APIs, e.g., Entity Framework's FirstAsync.
Since my call is awaitable, I think, UI thread waits for my call to finish its business and populate the result, doesn't it?
No. The UI is blocking because your code is calling Result. Result is a blocking call.
sounds synchronous to me..
That's because you're making it synchronous by using Result. If you use await, the code will execute serially, but asynchronously.
Also if at all, I want to 'exlicitly' wait till the task gets finished, how do i enforce that?
Your code is already doing that. It's explicitly blocking until the task is complete. If you change it to properly use await, it will be explicitly waiting (but not blocking) until the task is complete.
some posts, suggested in doing like below
No. Don't use Task.Run unnecessarily.
can't I wait for the thread, used by the async method?
Pure async methods don't have threads that they use.
Then what is the main purpose of async methods, which return something, and UI waits for it to complete?
Async makes sense when it's consumed with await.
Now on UI, below resulted a frozen state or deadlock:
var myTask = MyLibraryobject.DoSomethingAsync();
This is blocking the UI because it is not awaited. You simply need to await an async method
string res = await MyLibraryobject.DoSomethingAsync();
and the above should be placed in the UI inside an async context too.
Edit - reply to comment do "you mean another wrapper async method, which returns Task<string>..."? No,I don't. When I wrote "inside an async context" I meant to add the async keyword to a subscribed event, like a button click, if you are in the code behind or to your delegate command implementation, if you are in a more advanced MVVM implementation of the GUI.
Furthermore, your class library is supposed to do something truly asynchronous, not just starting a task and wrapping it inside a formally async method, like it is your DoAsync.
Edit - in reply to comment "how should avoid" it? If you can't do async all the way, just keep the library and the API sync and unblock the UI by starting a separate thread to call the sync api.
I need somehow to bypass Thread.Sleep() method and don't get my UI Thread blocked, but I don't have to delete the method.
I need to solve the problem without deleting the Sleep method. The Sleep method simulates a delay(unresponsive application). I need to handle that.
An application is considered non-responsive when it doesn't pump its message queue. The message queue in Winforms is pumped on the GUI thread. Therefore, to make your application "responsive", you need to make sure the GUI thread has opportunities to pump the message queue - in other words, it must not run your code.
You mentioned that the Thread.Sleep simulates a "delay" in some operation you're making. However, you need to consider two main causes of such "delays":
An I/O request waiting for completion (reading a file, querying a database, sending an HTTP request...)
CPU work
The two have different solutions. If you're dealing with I/O, the best way would usually be to switch over to using asynchronous I/O. This is a breeze with await:
var response = await new HttpClient().GetAsync("http://www.google.com/");
This ensures that your GUI thread can do its job while your request is pending, and your code will restore back on the UI thread after the response gets back.
The second one is mainly solved with multi-threading. You should be extra careful when using multi-threading, because it adds in many complexities you don't get in a single-threaded model. The simplest way of treating multi-threading properly is by ensuring that you're not accessing any shared state - that's where synchronization becomes necessary. Again, with await, this is a breeze:
var someData = "Very important data";
var result = await Task.Run(() => RunComplexComputation(someData));
Again, the computation will run outside of your UI thread, but as soon as its completed and the GUI thread is idle again, your code execution will resume back on the UI thread, with the proper result.
something like that maybe ?
public async void Sleep(int milliseconds)
{
// your code
await Task.Delay(milliseconds); // non-blocking sleep
// your code
}
And if, for reasons that escape me, you HAVE to use Thread.Sleep, you can handle it like that :
public async void YourMethod()
{
// your code
await Task.Run(() => Thread.Sleep(1000)); // non-blocking sleep using Thread.Sleep
// your code
}
Use MultiThreading.
Use a different thread for sleep rather than the main GUI thread. This way it will not interfere with your Main application
I have such question. For example I create dynamically custom userControl using mvvm pattern. So I've sended a command to create an userControl. So creation looks like
private async Task<bool> OnAddUserControl1(List<ALV_VM_ADWERT> control)
{
try
{
_cancellationTokenSource = new CancellationTokenSource();
var userControl = _userControlsContainer.CreateUserControl1(control);
var task1 = Task.Factory.StartNew(() =>
{
userControl.ViewModel.InOperationEvent += OnUsercontrolInOperationChanged;
userControl.ViewModel.ValueTypeChangedEvent += OnValueTypeChanged;
userControl.ViewModel.SetExpandableName += OnSetExpandableName;
}, _cancellationTokenSource.Token, TaskCreationOptions.AttachedToParent, TaskScheduler.FromCurrentSynchronizationContext());
var task2 = Task.Factory.StartNew(() => FinalCreatingStep(userControl, control[0].RAUMNAME.Trim()), _cancellationTokenSource.Token, TaskCreationOptions.AttachedToParent, TaskScheduler.FromCurrentSynchronizationContext());
await Task.WhenAll(task1, task2);
return true;
}
catch (Exception)
{
return false;
}
}
and my question is - does it make sence to create child tasks, or it is better to have code without child tasks? And if the answer is yes, then should I make all methods async? If not, what methods I should not make async?
Do those event subscriptions really need to be asynchronous? You may be trying too hard to use asynchronous code.
The user control constructor is usually the most time consuming part and must be done on the UI thread. Asynchronous operations are generally only required when some form of IO or processing is involed;
Reading Files
Writing Files
Processing large data sets
Crossing the process boundary to talk to a server or connect device
In short, an asynchronous task is probably overkill here.
Does it make sence to create child tasks, or it is better to have code without child tasks?
It is depending on your requirements. If your UI will blocked (Freeze) for long time you have to create a child task, otherwise not!
And if the answer is yes, then should I make all methods async? If not, what methods I should not make async?
Here it is also depending on your requirements and your .Net Version. If you are using .NET 4.5 the easiest way to do that with Async await. If you are using .Net 3.5 than just use Task. If .Net 2 use BackgorundWorker else use Thread class. ONLY Asynchrony methods must get the word async. Other methods you do not have to change them. In other words only the method which blocked the UI.
Your current code doesn't make any sense.
The point of async code in a UI application is responsiveness - i.e., moving long-running operations off the UI thread. As #Gusdor pointed out, most of the use cases for async are I/O-based (or event-based) operations, where you don't want to block the UI thread just to wait for some result. Another use case is when you have some CPU-bound work to do, but you don't want to tie up the UI thread; in this case, you can use Task.Run.
But in your code, you're calling StartNew with TaskScheduler.FromCurrentSynchronizationContext, which means your "child" tasks are going to execute on the UI thread. So your OnAddUserControl1 is just starting tasks which will run on the same thread and asynchronously waiting for them to complete. This is a very complex way of doing nothing.
While we're on the subject of StartNew, there are a number of other problems:
The code is passing a CancellationToken without ever observing it in the delegate.
The code is specifying AttachedToParent which is incorrect for await-compatible tasks.
As mentioned above, the code is passing a TaskScheduler which will run the delegate right back on the UI thread.
If you need to use background (thread pool) tasks, you should use Task.Run instead of Task.Factory.StartNew; I go into more detail on my blog.
So for this example, it doesn't make sense to use async or await at all.
The best way to start using async is to identify the I/O-bound (or event-driven) parts first (e.g., HTTP requests, database calls), make them async, and then work your way up the call stack.
I am writing a simple metro app. However, the API blocks when accessing files. By blocking, I mean the programs waits forever. Creating/opening a file or folder should take at most a few seconds. In this case, it takes forever.
When I run the program, it never comes back from OnTest. Is it what you get.
I understand .Wait will wait for the creation of files and folders to finishe. Maybe that's not great design. However, that's not the point.
My question is:
Do you get the same behavior (blocks the program forever)
Is it what's supposed to happen or is it a bug in WinRT? (I am using the consumer preview)
If that's the expected behavior, why does it take forever?
Here is the XAML code:
<Button Click="OnTest">Test</Button>
Here is the C# code:
private async void OnTest(object sender, RoutedEventArgs e)
{
var t = new Cache("test1");
t = new Cache("test2");
t = new Cache("test3");
}
class Cache
{
public Cache(string name)
{
TestRetrieve(name).Wait();
}
public static async Task TestRetrieve(string name)
{
StorageFolder rootFolder = ApplicationData.Current.LocalFolder;
var _folder = await rootFolder.CreateFolderAsync(name, CreationCollisionOption.OpenIfExists);
var file = await _folder.CreateFileAsync("test.xml", CreationCollisionOption.OpenIfExists);
}
}
It blocks on the second call to new Cache("test2");
I have not attempted to run your program or reproduce your problem, but I can make an educated guess as to what is going on.
Suppose you wrote yourself the following to-do list:
Put a letter to mom in the mailbox.
Set the alarm to wake me up as soon as I've read her reply.
Go to sleep.
Check the mailbox for the reply.
Read the reply.
Now do everything on that list strictly in order from top to bottom. What happens?
The problem is not with the post office or with mom; they are picking up the letter you put in the mailbox, sending it to mom, mom is writing her reply and the post office is bringing it back to you. The problem is that you never get to the fourth step because you only can start the fourth step after you complete the fifth step and the alarm wakes you up. You'll sleep forever because you are essentially waiting for your future self to wake your present self up.
Eric, Thank you for the explanation.
You're welcome.
However, I am still confused as to why my code does not work.
OK, let's break it down. What does your program really do? Let's simplify:
void M()
{
Task tx = GetATask();
tx.Wait();
}
async Task GetATask()
{
Task ty = DoFileSystemThingAsync();
await ty;
DoSomethingElse();
}
First off: what is a task? A task is an object that represents (1) a job to be done, and (2) a delegate to the continuation of the task: the thing that needs to happen after the task is done.
So you call GetATask. What does it do? Well, the first thing it does is it makes a Task and stores it in ty. That task represents the job "start some operation on the disk, and notify the I/O completion thread when it is done".
What is the continuation of that task? What has to happen after that task is done? DoSomethingElse needs to be called. So the compiler transforms the await into a bunch of code that tells the task to ensure that DoSomethingElse is called when the task is done.
The moment that the continuation of the I/O task has been set, the method GetATask returns a task to the caller. What task is that? This is a different task than the task that got stored into ty. The task that is returned is the task that represents the job do everything that the method GetATask needs to do.
What is the continuation of that task? We don't know! That is up to the caller of GetATask to decide.
OK, so let's review. We have two task objects. One represents the task "go do this thing on the file system". It will be done when the file system does its work. It's continuation is "call DoSomething". We have a second task object that represents the job "do everything in the body of GetATask". It will be done after the call to DoSomethingElse returns.
Again: the first task will be complete when the file I/O succeeds. When that happens, the file I/O completion thread will send a message to the main thread saying "hey, that file I/O you were waiting for is done. I am telling you this because it is now time for you to call DoSomethingElse".
But the main thread is not examining its message queue. Why not? Because you told it to synchronously wait until everything in GetATask, including DoSomethingElse, is complete. But the message that is telling you to run DoSomethingElse now cannot be processed because you are waiting for DoSomethingElse to be complete.
Now is it clear? You are telling your thread to wait until your thread is done running DoSomethingElse before you check to see if "please call DoSomethingElse" is in the queue of work to be performed on this thread! You are waiting until you have read the letter from mom, but the fact that you are waiting synchronously means that you are not checking your mailbox to see if the letter has arrived.
Calling Wait is obviously wrong in this case because you are waiting for yourself to do something in the future, and that's not going to work. But more generally, calling Wait completely negates the entire point of being async in the first place. Just don't do that; it doesn't make any sense to say both "I want to be asynchronous" and "but I want to synchronously wait". Those are opposites.
You're using Wait() in the constructor of the Cache class. That's going to block until whatever is currently executing asynchronously has finished.
This is not the way to design this. Constructors and async do not make sense. Perhaps a factory method approach like this would work better:
public class Cache
{
private string cacheName;
private Cache(string cacheName)
{
this.cacheName = cacheName;
}
public static async Cache GetCacheAsync(string cacheName)
{
Cache cache = new Cache(cacheName);
await cache.Initialize();
return cache;
}
private async void Initialize()
{
StorageFolder rootFolder = ApplicationData.Current.LocalFolder;
var _folder = await rootFolder.CreateFolderAsync(this.cacheName, CreationCollisionOption.OpenIfExists);
var file = await _folder.CreateFileAsync("test.xml", CreationCollisionOption.OpenIfExists);
}
}
And then you use it like this:
await Task.WhenAll(Cache.GetCacheAsync("cache1"), Cache.GetCacheAsync("cache2"), Cache.GetCacheAsync("cache3"));
TestRetrieve(name).Wait();
You're telling it to block specifically by using the .Wait() call.
Remove the .Wait() and it shouldn't block anymore.
The existing answers provide very thorough explanations of why it blocks and code examples of how to make it not block, but these many be 'more information' than some users understand. Here is a simpler 'mechanics oriented' explanation..
The way the async/await pattern works, each time you await an async method, you are "attaching" that method's async context to the current method's async context. Imagine await as passing in a magic hidden paramater "context". This context-paramater is what allows nested await calls to attach to the existing async context. (this is just an analogy...the details are more complicated than this)
If you are inside an async method, and you call a method synchronously, that synchronous method doesn't get that magic hidden async context paramater, so it can't attach anything to it. It's then an invalid operation to create a new async context inside that method using 'wait', because the new context does not get attached to your thread's existing top-level async context (because you don't have it!).
Described in terms of the code-example, the TestRetrieve(name).Wait(); is not doing what you think it's doing. It's actually telling the current thread to re-enter the async-activity-wait-loop at the top. In this example, this is the UI-thread, which is called your OnTest handler. The following picture may help:
UI-thread context looks like this...
UI-Thread ->
OnTest
Since you didn't have a connected chain of await/async calls all the way down, you never "attached' the TestRetrieve async context to the above UI-Thread async chain. Effectively, that new context you made is just dangling off in nowhere land. So when you "Wait" the UIThread, it just goes right back to the top.
For async to work, you need to keep a connected async/await chain from the top-level synchronous thread (in this case it's the UI-thread doing this) through all async actions you need to do. You can't make a constructor async, so you can't chain an async context into a constructor. You need to construct the object synchronously and then await TestRetrieve from outside. Remove the 'Wait' line from your constructor and do this...
await (new Cache("test1")).TestRetrieve("test1");
When you do this, the 'TestRetrieve' async context is properly attached, so the chain looks like this:
UI-Thread ->
OnTest ->
TestRetrieve
Now the UI-thread async-loop-handler can properly resume TestRetrieve during async completions and your code will work as expected.
If you want to make an 'async constructor', you need to do something like Drew's GetCacheAsync pattern, where you make a static Async method which constructs the object synchronously and then awaits the async method. This creates the proper async chain by awaiting and 'attaching' the context all the way down.