I have a UI program in which i used QFuture to run a function in a different class that was heavy on processor etc. and caused my UI to crash, after using QFuture to thread this out it works fine, however I have another function that will need to be called right after however this one is within the same class as where I'm calling it from.
I have used the same code to do the second one but it just shows errors about Timers which I'm not even using, also I don't know if its right that I'm using this in the arguments:
future = new QFuture<void>;
*future = QtConcurrent::run(this, &GUI::test);
test being a local function, is it something to do with the fact that I can't have two concurrent::run calls or something, I've tried looking at the documentation which hasn't helped unfortunately
basically when this is called the UI locks up and I get some strange warnings, and I thought the point of threading it was that the UI wouldn't crash
Here's my best guess on this.
I'm guessing that you have the following situation. (Note: understanding the situation you're in is very useful when trying to come up with a solution for a problem. Often it turns out that the OP is going about something in a fundamentally wrong way and should be doing something entirely different.)
You have a GUI. You have an operation that takes a long time, which you want to run in a worker thread. And at the end of the long operation, you need to call a function on the GUI thread, probably to update the GUI or something similar.
So you launch the background task using QtConcurrent::run(&someObject, &SomeObject::task). Then you need some way to be notified, on the GUI thread, when the task is completed.
The correct way to do this is to attach a QtFutureWatcher to the future returned from run() and connect a slot on your GUI class to its finished signal. Qt's signal mechanism will ensure that the slot is called on the right thread.
Related
We are trying to write a portable shared library that makes use of some Qt classes for convenience (mainly QTimer and QTcpSocket); no GUI stuff, though. The according signal/slot connections appear to require some Qt event loop, so we "prime" a QCoreApplication as outlined in this answer. Accordingly, we set up a worker object that does the heavy lifting and move it to a QThread.
The problem we run into now is that the queued connections between the QThread's owner object (within the main thread) and the worker object within the QThread seem to never get handled on Linux systems, at least as long as the program that implements our library does not provide any further Qt event loop of its own in the main thread. This is not very helpful, since the data passed from the worker to the main thread should be passed further using some callback functions, which now never get called, though.
My question is thus: is there a way to get an event loop to work in the library main thread without locking it or the host program up (which seems to be the case when just putting a QCoreApplication::exec() or similar there)? Or will we have to set up a different inter-thread communication scheme (independent from Qt) in order to deal with these data transfers?
Since we do not know if the host software is going to run on a QApplication or not, ideally I'd also have a check for that before setting up a main thread event loop. Is a simple if(qApp != nullptr) enough for that?
P.S.: A few things I tried but which did not work for me, either:
Settings up a QEventLoop in a std::thread launched from the main thread (probably not working because still not in the main thread)
Setting up a QEventLoop in the main thread class and triggering its processEvents() function periodically using a QTimer (probably not working due to the missing event loop for the QTimer::timeout signal in the main function)
Starting the QCoreApplication in a std::thread (gives a run-time warning on Windows that QCoreApplication should be started in the main thread)
In Qt parlance, a callback is called Qt::DirectConnection. But of course those callbacks will run on your worker thread. But that’d be the case with any other library that uses callbacks, so Qt is not a problem here, and neither is your code: the basic idea has this property.
If the host application is not using an event loop (any event loop, not necessarily Qt’s), then there’s nothing you can do other than polling – see below.
If the host application runs an X11 event loop, then you need to ensure that your copy of Qt is using the same underlying event loop as the host application. Usually, this would be the glib’s event loop, and then it should work automagically. Otherwise, you’ll need to pass to the user the file descriptor of the synchronization primitive used by Qt’s event loop, and the user will need to integrate it into their event loop. You’ll face the same problem whether you use Qt or not: rolling your own communication method won’t fix it, since you still need a waitable primitive that will interoperate with whatever event loop the user is using.
The user can of course poll for callbacks whenever they feel like it: expose a mainPoll() method that forwards to QCoreApplication::processEvents().
Despite accepting another answer (which I deem more correct), I'd still like to mention a workaround that worked surprisingly well: We actually managed to get around the event loop/thread problems on most systems by connecting the worker thread signals with lambda functions in constructor of the class that sets up the worker.
Now, I doubt that this behaviour is properly thread-safe, and having relatively lengthy lambda functions declared in connect function calls is certainly not good style. But in case anyone else ends up struggling with this issue, this may be a short-term solution or (temporary) workaround.
When should moveToThread be preferred over subclassing QThread?
This link shows that both methods work. On what basis should I decide what to use from those two?
I would focus on the differences between the two methods. There isn't a general answer that fits all use cases, so it's good to understand exactly what they are to choose the best that fits your case.
Using moveToThread()
moveToThread() is used to control the object's thread affinity, which basically means setting the thread (or better the Qt event loop) from which the object will emit signals and its slots will be executed.
As shown in the documentation you linked, this can be used to run code on a different thread, basically creating a dummy worker, writing the code to run in a public slot (in the example the doWork() slot) and then using moveToThread to move it to a different event loop.
Then, a signal connected to that slot is fired. Since the object that emits the signal (the Controller in the example) lives in a different thread, and the signal is connected to our doWork method with a queued connection, the doWork method will be executed in the worker thread.
The key here is that you are creating a new event loop, run by the worker thread. Hence, once the doWork slot has started, the whole event loop will be busy until it exits, and this means that incoming signals will be queued.
Subclassing QThread()
The other method described in Qt's documentation is subclassing QThread. In this case, one overrides the default implementation of the QThread::run() method, which creates an event loop, to run something else.
There's nothing wrong with this approach itself, although there are several catches.
First of all, it is very easy to write unsafe code, because the run() method is the only one in that class that will be actually run on another thread.
If as an example, you have a member variable that you initialize in the constructor and then use in the run() method, your member is initialized in the thread of the caller and then used in the new thread.
Same story for any public method that could be called either from the caller or inside run().
Also slots would be executed from the caller's thread, (unless you do something really weird as moveToThread(this)) leading to extra confusion.
So, it is possible, but you really are on your own with this approach and you must pay extra attention.
Other approaches
There are of course alternatives to both approaches, depending on what you need. If you just need to run some code in background while your GUI thread is running you may consider using QtConcurrent::run().
However, keep in mind that QtConcurrent will use the global QThreadPool. If the whole pool is busy (meaning there aren't available threads in the pool), your code will not run immediately.
Another alternative, if you are at the least on C++11, is to use a lower level API such as std::thread.
As a starting point: use neither. In most cases, you have a unit of work that you wish to run asynchronously. Use QtConcurrent::run for that.
If you have an object that reacts to events and/or uses timers, it's a QObject that should be non-blocking and go in a thread, perhaps shared with other objects.
Such an object can also wrap blocking APIs.
Subclassing QThread is never necessary in practice. It's like subclassing QFile. QThread is a thread handle. It wraps a system resource. Overloading it is a bit silly.
Simple answer is ALWAYS.
When you move object to thread:
it is easy to write test for code
it is easy to refactor code (you can use thread but you don't have to).
you do not mix functionality of thread with business logic
there is no problem with object lifetime
When you subclass QThread
it is harder to write test
object clean up process can get very confusing leading to strange errors.
There is full description of the problem from Qt blog: You’re doing it wrong….
QtConcurrent::run is also very handy.
Please remember that by default slots are trying to jump between treads when signal is send from other thread object is assigned to. For details see documentation of Qt::ConnectionType.
QThread is low level thread abstraction, first look at high level API QtConcurrent module and QRunnable
If nothing of these is suitable for you, then read this old article, it tells how you should use QThread. Think about thread and task performed in this thread as a separate objects, don't mix them together.
So, if you need to write come custom, specific or extended thread wrapper then you should subclass QThread.
If you have QObject derived class with signals and slots, then use moveToThread on it.
In other cases use QtConcurrent, QRunnable and QThreadPoll.
Im using QFuture to add threads to my ui application, i call a method from one thread and have a watcher and signal set so that when the first function is finished a second one is called
both use the same code to thread the function off
first use of QFuture
future = QtConcurrent::run(&Test, &test::getAll);
watcher->setFuture(*future);
second use of QFuture
*future2 = QtConcurrent::run(this, &GUI::paintAll);
watcher2->setFuture(*future2);
when this second one is called i get this warning
QObject::startTimer: timers cannot be started from another thread
and then it sometimes will crash sometimes it wont but i just cant help but feel its to do with this, iv asked so many questions about this now without answers that work, could someone please tell me why im being warned about timers when im not even using them, also why im warned only the second time i call a qfuture function never on that first one??
In Qt, GUI classes must only be accessed by the main thread (it's also called the "GUI thread"). You cannot run paint functions in another thread. See http://qt-project.org/doc/qt-5.0/qtcore/thread-basics.html#gui-thread-and-worker-thread
I'm guessing that GUI::paintAll() indirectly calls Qt GUI classes, which use timers internally.
The usual scenario, there is an MFC/Win32/WTL/wxWidgets/Qt application that does something useful. It was designed to be single threaded, and there is some logic that handles errors/questions within processing blocks.
So, somewhere deep inside some class, a dialog can be fired that asks the user "Are you sure you want to complete the action?"/"Error with document layout" or something like that.
The problem is the dialog is fired from computationally heavy/strightforward code. Like FFT/image sharpening/file system de-fragmentation function, or something along the lines. Which could be launched in a worker thread easily, if not for the GUI. And would suit there better, as it would avoid GUI stalls that are so annoying for the user.
However, GUI cannot work in a worker thread, and dependency injection is pretty much impossible to do, because it would go down several layers of computational code. In a very unclean way from class interface standpoint, like someclass instance(data_in, data_out, param1, param2, GUI_class_ref) : m_GUI(GUI_class_ref), ... 3 or more levels deep.
Is there a pattern/checklist for such scenarios that can be used to marshall GUI prompts back to main thread and return the result back into the core of the computational code, if the code is split in multiple threads?
You can create synchronization context. It is a queue of commands to be executed by main thread. Worker thread adds command into this queue (which must be locked for single-thread access) and waits. Main thread processes this queue periodically, executes commands (for example, "Cancel operation" dialogs) and notifies worker threads about results.
In C#, this was done with delegates and arguments to call them. In C++, you can go with enum-coded messages to be processed in a switch (like messages in Windows programs.) Or create something with pointers to member functions + object pointer to call them from + arguments to call with.
You are at one classical old code refactoring crossroad. Proper isolation and dependency injection is infeasible, so you are left with making the GUI context globally accessible. That is creating a Singleton. It doesn't necessarily need to be the GUI context directly, so at least some isolation is achieved. It can be some kind of manager which has the GUI context and accepts just specific one purpose calls from the computation code. You could make the GUI thread class a friend of this manager and make the GUI callbacks (upon closing the dialog) private.
I could give more specific ideas what to write as i went through exactly the same challenge (threadization of existing heavy app). But i am confused whether you want only the GUI thread to be running freely, or the background computation as well. The example dialog prompt you gave is confusing as it suggests a decision which needs to be answered to know whether continue at all (which would mean that computation is on hold).
I would like invoke a function call in a one shot manner. What's the best way to do this in Boost / C++?
I would like to pass it two parameters and do not need a result.
Well you can spin up a thread and then never join it again. Something like:
boost::thread some_thread(&SomeFunction, param1, param2);
Once the some_thread variable goes out of scope, the thread of execution will be detached and will not be able to be joined again. It's a waste to spin up a thread unless the processing time of this function is significant though.
I haven't used boost::thread in awhile but I see a quick example on the documentation page for the class:
void find_the_question(int the_answer);
boost::thread deep_thought_2(find_the_question,42);
I believe as soon as it finishes the function, the thread will exit. This may not be what you want in that once the thread goes out of scope, it will be destroyed. If that's not going to work, you probably need to create a long running thread pool and then pass your functors as boost::bind compositions.
Depending on how often you are doing this, you might be best off creating a pool of threads, along with a work queue. Creating a thread can create a lot of overhead if you are trying to do it dozens of times a second. If you don't care about the return value, that makes it really easy.
Spin up a thread or two (or ten); have a thread-safe queue of functors to call (bind the parameters to the function and put that on the queue); the threads wait on the queue for something to show up, the first thread to wake up gets to process the work. When a thread is done running a job, it waits on the queue again.
Take a look at this project for an idea of one way to do it.
Of course if you are only making asynchonous calls every couple of seconds to improve a UI's responsiveness, it'd be easier to just start up a new thread every time.
Perhaps you want to emit a signal?
I really liked Qt's signals and slots functionality, and I know Boost has signals/slots as well. I've never used signals/slots in Boost, though.