I am developing a Java-ME Based Mobile Application. Now My Requirements are like whenever I am updating one of my RMS, I want my application to be stay in a Freeze kind of mode; which means no other action like clicking button or anything else should happen. My Method is already "Synchronized".
Kindly guide me regarding this question.
Thanks.
The best way to handle this is to "serialize" your tasks. You can do this with a message queue - a class that maintains a Vector of message objects (tasks) and runs code based on each message. The queue runs on a thread that processes each task (message) in series. You create a simple message class for the different tasks - read RMS etc. A message can be an Integer if you like that wraps a number. The operation of adding and retrieving messages is synchronized but the code than does the tasks is not and runs on a simple switch block. The benefit of serializing your tasks is you don't have to worry about concurrency. Here is some of the essential code from a class I use to do this.
class MessageQueue implements Runnable{
Vector messages;
Thread myThread;
volatile boolean stop;
public void start() {
stop=false;
myThread=new Thread(this);
myThread.start();
}
// add message to queue - this is public
public synchronized void addMessage(Message m) {
messages.addElement(m);
if(stop) {
start();
} else {
// wake the thread
notify();
}
}
// get next message from queue - used by this thread
private synchronized Message nextMessage() {
if(stop) return null;
if(messages.isEmpty()) {
return null;
} else {
Message m=(Message)messages.firstElement();
messages.removeElementAt(0);
return m;
}
}
public void run() {
while (!stop) {
// make thread wait for messages
if (messages.size() == 0) {
synchronized (this) {
try {
wait();
} catch (Exception e) {
}
}
}
if (stop) {
// catch a call to quit
return;
}
processMessage();
}
}
}
// all the tasks are in here
private void processMessage() {
Message m = nextMessage();
switch (m.getType()) {
case Message.TASK1:
// do stuff
break;
case Message.TASK2:
// do other stuff
break;
case Message.TASK3:
// do other other stuff
break;
default: //handle bad message
}
}
}
What you are asking is very code depended. Usually when you want to make some synchronic actions you just write them one after the other. in java it's more complected, since sometimes you "ask" the system to do something (like repaint() method). But since the RMS read/write operations are very quick (few millisecond) i don't see any need in freesing.
Could you please provide some more information about the need (time for RMS to respond)? does your code runs on system thread (main thread) or your own thread?
I want my application to be stay in a Freeze kind of mode; which means no other action like clicking button or anything else should happen.
First of all I would strongly advise against real freezing of UI - this could make a suicidal user experience for your application.
If you ever happened to sit in front of computer frozen because of some programming bug, you may understand why approach like this is strongly discouraged. As they describe it in MIDP threading tutorial, "user interface freezes, the device appears to be dead, and the user becomes frustrated..."
This tutorial by the way also suggests possibly the simplest solution for problems like you describe: displaying a wait screen. If you don't really have reasons to avoid this solution, just do as tutorial suggests.
To be on a safe side, consider serializing tasks as suggested in another answer. This will ensure that when RMS update starts, there are no other tasks pending.
Related
Please help me write Junit for this piece of code using Mockito /Powermock, Finding it difficult due to lamda expression and executor service.
public class myClass {
ExecutorService executorService;
public void testMethod(String a){
Thread thread = new Thread(() -> {
//logic
a= testDAo.getStatus();
while (true) {
if (Thread.interrupted()) {
break;
}
if (a() != "done" || a() != "fail") {
Thread.yield();
try {
Thread.sleep(10000);
} catch (InterruptedException e) {
}
} else {
break;
}
}
}
Future task = executorService.submit(thread);
while (!task.isDone()) {
try {
Thread.sleep(10000);
} catch (InterruptedException e) {
}
}
}
}
Various things here:
first of all: for testing executors and parallel execution, using a same thread executor can be extremely helpful (because it takes out the parallel aspect)
you have difficulties writing a unit test - because your production code is way too complicated.
Thus the real answer is: step back and improve your production code. Why again are you pushing a thread into an executor service?
The executor service is already doing things on a thread pool (at least that is how you normally use them). So you push a thread into a thread pool, and then you have code that waits "two" times (first within that thread, and then outside on the future). That just adds a ton of complexity for small gain.
Long story short:
I would get rid of that "inner thread" - just have the executor task wait until the result becomes available
Then: if lambda's give you trouble - then don't use them. Just create a named small class that implements that code. And then you can write unit tests for that small class. In other words: don't create huge "units" that do 5 different things. The essence of a good unit is to one thing (single responsibility principle!). And as soon as you follow that idea testing becomes much easier, too.
class Driver : Public QObject
{
Q_OBJECT
private:
// method command: sends a command
// signal ok: command executed, sends back a message
MyDevice *device;
public:
Deriver()
{
device = new MyDevice(0);
connect (mydevice,&MyDevice::ok,this,&Driver::onInitOk);
}
public slots:
void init()
{
device->command("init");
//at this point, I want to block this method until the device signals ok with a given msg
}
command()
{
device->command("setmode x");
device->command("cmd");
//at this point, I want to block this method until the device signals ok with a given msg
}
void onInitOk(QString msg)
{
//somehow unblock the actually running command, if the msg matches
}
}
I would like to use the command/init with a QueuedConnection, so they are executed async from the gui thread, and sequentially. (Am I right?)
How can I implement the blocking effectively?
Okay so I've edited based on the clarity of the comments given. The best place to look at would be the Qt Threading Guide. This can give a much better breakdown on the systems used for concurrency.
For your example I've added a QMutex object to your Driver class. It may be worth thinking about if you want to move the thread-based controls into the MyDevice class itself if you have access.
Driver()
{
moveToThread(new QThread());
device = new MyDevice(0);
}
void init()
{
mutex.lock();
const QString& result = device->command("init");
onInitOk(result);
}
void command()
{
mutex.lock();
device->command("setmode x");
const QString& result = device->command("cmd");
onInitOk(result);
}
void onInitOk(QString msg)
{
...[STUFF]
// Even when things go wrong you MUST unlock the mutex at some point.
// You can't keep the thread blocked forever in cases of poor results.
// As such it might be better practice to unlock in
// the same function that locks!
mutex.unlock();
}
QMutex mutex;
Bear in mind I am assuming you are wanting to access the functionality from the slots mechanism. Hence why we use the moveToThead() function. When the object is accessed via slots in the GUI thread it'll now run the function on a different thread.
Likewise the mutex only blocks for all the objects that share that one mutex instance. So depending on your implementation you may have to think about what is right for you in exposing that mutex.
So we're setting up a worker role with windows azure and it's running at a very high cpu utilization. I think it has something to do with this section, but I'm not sure what to do about it. Each individual thread that gets started has its own sleep, but the main thread just runs in a while loop. Shouldn't there be a sleep in there or something?
public class WorkerRole : RoleEntryPoint
{
private List<ProcessBase> backgroundProcesses = new List<ProcessBase>();
public override void Run()
{
// This is a sample worker implementation. Replace with your logic.
Trace.WriteLine("BackgroundProcesses entry point called", "Information");
foreach (ProcessBase process in backgroundProcesses)
{
if (process.Active)
{
Task.Factory.StartNew(process.Run, TaskCreationOptions.LongRunning);
}
}
while (true) { }
}
How about something like this, would this be appropriate?
public override void Run()
{
// This is a sample worker implementation. Replace with your logic.
Trace.WriteLine("BackgroundProcesses entry point called", "Information");
List<Task> TaskList = new List<Task>();
foreach (ProcessBase process in backgroundProcesses)
{
if (process.Active)
{
TaskList.Add(Task.Factory.StartNew(process.Run, TaskCreationOptions.LongRunning));
}
}
Task.WaitAll(TaskList.ToArray());
//while (true) { }
}
Your change looks good to me. Sometimes I use Thread.Sleep(Timeout.Infinite).
Have you tested it? Does it reduce the CPU usage? It could be that the tasks themselves actually consume a lot of CPU. We don't know for sure yet that the while loop is the culprit.
The while loop is probably causing your high CPU. It's basically an infinite busy-wait. Your second code sample should work fine, as long as the Tasks you're waiting on never exit. In my personal opinion the best solution is the one outlined in my answer here. If you don't like that, a simpler solution would be to add a Sleep() inside the loop. eg:
while(true){
Thread.Sleep(10000);
}
while loop with empty body will fully load the CPU core to which the thread is dispatched. That's bad idea - you burn CPU time for no good.
A better solution is to insert a Thread.Sleep() with a period ranging anywhere from 100 milliseconds to infinity - it won't matter much.
while( true ) {
Thread.Sleep( /*anything > 100 */ );
}
Once you've got rid of the empty loop body you're unlikely to do any better than that - whatever you do in your loop the thread will be terminated anyway when the instance is stopped.
Just deployed this to production this morning after testing it last night on staging. Seems to be working great. CPU usage went down to .03% average for the background process down from 99.5% ...
public override void Run()
{
// This is a sample worker implementation. Replace with your logic.
Trace.WriteLine("BackgroundProcesses entry point called", "Information");
List<Task> TaskList = new List<Task>();
foreach (ProcessBase process in backgroundProcesses)
{
if (process.Active)
{
TaskList.Add(Task.Factory.StartNew(process.Run, TaskCreationOptions.LongRunning));
}
}
Task.WaitAll(TaskList.ToArray());
//while (true) { }
}
I'm testing my app with mono in prevision of a Linux port, and I have a threading problem. I initially considered pasting 3000 code lines here, but finally I've devised a small minimal example ;)
You have a form with a button (poetically named Button1, and a label (which bears, without surprise, the name Label1)). The whole lot is living a happy life on a form called Form1. Clicking Button1 launches an infinite loop that increments a local counter and updates Label1 (using Invoke) to reflect its value.
Now in Mono, if you resize the form, the label stops updating, never to restart. This doesn't happen with MS implementation. BeginInvoke doesn't work any better; worse, it makes the UI hang in both cases.
Do you know where this discrepancy comes from? How would you solve it? And finally, why doesn't BeginInvoke work here? I must be making a huge mistake... but which?
EDIT:
Some progress so far:
Calling BeginInvoke does in fact work; only, the UI just doesn't refresh fast enough, so it seems to stop.
On mono, what happens is that the whole thread hangs when you insert a message in the UI queue (eg by resizing the form). In fact, the synchronous Invoke call never returns. I'm trying to understand why.
Of interest: even using BeginInvoke, the asynchronous calls don't get executed before the resizing operation ends. On MS.Net, they keep running while resizing.
The code looks like this (C# version lower):
Public Class Form1
Private Sub Button1_Click(ByVal sender As System.Object, ByVal e As System.EventArgs) Handles Button1.Click
Dim T As New Threading.Thread(AddressOf Increment)
T.Start()
End Sub
Sub UpdateLabel(ByVal Text As String)
Label1.Text = Text
End Sub
Delegate Sub UpdateLabelHandler(ByVal Text As String)
Sub Increment()
Dim i As Long = 0
Dim UpdateLabelDelegate As New UpdateLabelHandler(AddressOf UpdateLabel)
Try
While True
i = (i + 1) Mod (Long.MaxValue - 1)
Me.Invoke(UpdateLabelDelegate, New Object() {i.ToString})
End While
Catch Ex As ObjectDisposedException
End Try
End Sub
End Class
Or, in C#,
public class Form1
{
private void Button1_Click(System.Object sender, System.EventArgs e)
{
System.Threading.Thread T = new System.Threading.Thread(Increment);
T.Start();
}
public void UpdateLabel(string Text)
{
Label1.Text = Text;
}
public delegate void UpdateLabelHandler(string Text);
public void Increment()
{
long i = 0;
UpdateLabelHandler UpdateLabelDelegate = new UpdateLabelHandler(UpdateLabel);
try {
while (true) {
i = (i + 1) % (long.MaxValue - 1);
this.Invoke(UpdateLabelDelegate, new object[] { i.ToString() });
}
} catch (ObjectDisposedException Ex) {
}
}
}
This is a bug in the mono runtime, at least I think it is. The code might not be good practice (I'm not a threading expert), but the thing that suggests a bug is the fact that the behaviour differs on windows and Linux.
On Linux, mono has exactly the same behaviour as MS.Net has on windows. No hanging, continuous updates even while resizing.
On Windows, mono displays all the aforementioned problems. I've posted a bug report at https://bugzilla.novell.com/show_bug.cgi?id=690400 .
Do you know where this discrepancy
comes from? How would you solve it?
I am not sure. I do not see anything obvious in your code that would cause the difference between Mono and .NET. If I had to make a wild guess I would say there is a possibility that you have stumbled upon an obscure bug in Mono. Though, I suppose it is possible that Mono uses a sufficiently different mechanism for handling the WM_PAINT messages that cause the form to get refreshed. The constant pounding of the UI thread from repeated calls to Invoke may be disrupting Mono's ability to get the form refreshed.
And finally, why doesn't BeginInvoke
work here?
Calling Invoke in a tight loop is bad enough, but BeginInvoke will be even worse. The worker thread is flooding the UI message pump. BeginInvoke does not wait until the UI thread has finished executing the delegate. It just posts the requests and returns quickly. That is why it appears to hang. The messages that BeginInvoke is posting to the UI message queue keep building up as the worker thread is likely severely out pacing the UI thread's ability to process them.
Other Comments
I should also mention that the worker thread is nearly useless in the code. The reason is because you have a call to Invoke on every iteration. Invoke blocks until the UI has finished executing the delegate. That means your worker thread and UI thread are essentially in lock-step with each other. In other words, the worker is spending most of its time waiting for the UI and vice versa.
Solution
One possible fix is to slow down the rate at which Invoke is called. Instead of calling it on every loop iteration try doing it every 1000 iterations or the like.
Any even better approach is to not use Invoke or BeginInvoke at all. Personally, I think these mechanisms for updating the UI are way overused. It is almost always better to let the UI thread throttle its own update rate especially when the worker thread is doing continuous processing. This means you will need to place a timer on the form and have it tick at the desired refresh rate. From the Tick event you will probe a shared data structure that the worker thread is updating and use that information to update the controls on the form. This has several advantages.
It breaks the tight coupling between the UI and worker threads that Control.Invoke imposes.
It puts the responsibility of updating the UI thread on the UI thread where it should belong anyway.
The UI thread gets to dictate when and how often the update should take place.
There is no risk of the UI message pump being overrun as would be the case with the marshaling techniques initiated by the worker thread.
The worker thread does not have to wait for an acknowledgement that the update was performed before proceeding with its next steps (ie. you get more throughput on both the UI and worker threads).
First and foremost: clicking on Button1 is asynchronous already, so you don't need to create another thread to increment, just call the increment method Sorry, I was reading your question line by line and by the time I got to the while-loop I forgot about the button:
private void Button1_Click(System.Object sender, System.EventArgs e)
{
Thread t = new Thread(Increment);
t.IsBackground = true;
t.Start();
}
Second: if you do need to use a thread then you should always set your thread to background (i.e. foreground prevents your process from terminating), unless you have a good reason for using a foreground thread.
Third: if you're making updates to the UI, then you should check the InvokeRequired property and call BeginInvoke:
public void UpdateLabel(string Text)
{
if (InvokeRequired)
{
BeginInvoke(new UpdateLabelDelegate(UpdateLabel), Text);
}
else
{
Label1.Text = Text;
}
}
public void Increment()
{
int i = 0;
while(true)
{
i++; // just incrementing i??
UpdateLabel(i.ToString());
Thread.Sleep(1000);// slow down a bit so you can see the updates
}
}
You can also "automate" the Invoke Required "pattern": Automating the InvokeRequired code pattern
And now see if you're still having the same problem.
I tried it on my machine and it works like a charm:
public partial class Form1 : Form
{
private delegate void UpdateLabelDelegate(string text);
public Form1()
{
InitializeComponent();
}
private void button1_Click(object sender, EventArgs e)
{
Thread t = new Thread(Increment);
t.IsBackground = true;
t.Start();
}
private void UpdateLabel(string text)
{
if (label1.InvokeRequired)
{
BeginInvoke(new UpdateLabelDelegate(UpdateLabel), text);
}
else
{
label1.Text = text;
}
}
private void Increment()
{
int i = 0;
while (true)
{
i++;
UpdateLabel(i.ToString());
Thread.Sleep(1000);
}
}
}
As my user changes the CurrentItem of a dataForm, I need to go the server to get addtional data. It's quite likely that the user could scroll through several items before finding the desired one. I would like to sleep for 500ms before going to get the data.
Is there a component already in the SDK or toolkit like a background worker that would assist in getting back to the UI thread to make my WCF async call once the 500ms sleep is done? It seems that if I don't do that, and try instead to call the WCF async method on the sleeper thread then the Completed event fires on the sleeper thread and not the UI thread, which of course is not good.
I think you might be a little off-track in your thinking. I'm not sure why you feel you need to get back to the UI thread in order to make the asych call. Generally you do as much work as you can on a BG thread and only marshal back to the UI thread when you have the results (by way of the Dispatcher).
I typically use a System.Threading.Timer for this purpose:
public class MyViewModel
{
private readonly Timer refreshTimer;
public MyViewModel()
{
this.refreshTimer = new Timer(this.DoRefresh);
}
public object CurrentItem
{
get { ... }
set
{
...
Invalidate();
}
}
// anything that should invalidate the data should wind up calling this, such as when the user selects a different item
private void Invalidate()
{
// 1 second delay
this.refreshTimer.Change(1000, Timeout.Infinite);
}
private void DoRefresh()
{
// make the async call here, with a callback of DoRefreshComplete
}
private void DoRefreshComplete()
{
// update the UI here by way of the Dispatcher
}
}