1 | // Copyright (C) 2016 The Qt Company Ltd. |
2 | // SPDX-License-Identifier: LicenseRef-Qt-Commercial OR LGPL-3.0-only OR GPL-2.0-only OR GPL-3.0-only |
3 | |
4 | /*! |
5 | \page qtconcurrentrun.html |
6 | \title Concurrent Run |
7 | \brief A simple way to run a task in a separate thread. |
8 | \ingroup thread |
9 | |
10 | The QtConcurrent::run() function runs a function in a separate thread. |
11 | The return value of the function is made available through the QFuture API. |
12 | |
13 | QtConcurrent::run() is an overloaded method. You can think of these overloads as slightly |
14 | different \e modes. |
15 | In \l {Concurrent Run (basic mode)} {basic mode}, the function passed to QtConcurrent::run() |
16 | is able to report merely a single computation result to its caller. |
17 | In \l {Concurrent Run With Promise} {run with promise mode}, the function passed to |
18 | QtConcurrent::run() can make use of the additional |
19 | QPromise API, which enables multiple result reporting, progress reporting, |
20 | suspending the computation when requested by the caller, or stopping |
21 | the computation on the caller's demand. |
22 | |
23 | This function is a part of the Qt Concurrent framework. |
24 | |
25 | \section1 Concurrent Run (basic mode) |
26 | |
27 | The function passed to QtConcurrent::run() may report the result |
28 | through its return value. |
29 | |
30 | \section2 Running a Function in a Separate Thread |
31 | |
32 | To run a function in another thread, use QtConcurrent::run(): |
33 | |
34 | \snippet code/src_concurrent_qtconcurrentrun.cpp 0 |
35 | |
36 | This will run \c aFunction in a separate thread obtained from the default |
37 | QThreadPool. You can use the QFuture and QFutureWatcher classes to monitor |
38 | the status of the function. |
39 | |
40 | To use a dedicated thread pool, you can pass the QThreadPool as |
41 | the first argument: |
42 | |
43 | \snippet code/src_concurrent_qtconcurrentrun.cpp explicit-pool-0 |
44 | |
45 | \section2 Passing Arguments to the Function |
46 | |
47 | Passing arguments to the function is done by adding them to the |
48 | QtConcurrent::run() call immediately after the function name. For example: |
49 | |
50 | \snippet code/src_concurrent_qtconcurrentrun.cpp 1 |
51 | |
52 | A copy of each argument is made at the point where QtConcurrent::run() is |
53 | called, and these values are passed to the thread when it begins executing |
54 | the function. Changes made to the arguments after calling |
55 | QtConcurrent::run() are \e not visible to the thread. |
56 | |
57 | Note that QtConcurrent::run does not support calling overloaded functions |
58 | directly. For example, the code below won't compile: |
59 | |
60 | //! [run-with-overload-calls] |
61 | \snippet code/src_concurrent_qtconcurrentrun.cpp 15 |
62 | |
63 | The easiest workaround is to call the overloaded function through lambda: |
64 | |
65 | \snippet code/src_concurrent_qtconcurrentrun.cpp 16 |
66 | |
67 | Or you can tell the compiler which overload to choose by using a |
68 | \c static_cast: |
69 | |
70 | \snippet code/src_concurrent_qtconcurrentrun.cpp 17 |
71 | |
72 | Or qOverload: |
73 | |
74 | \snippet code/src_concurrent_qtconcurrentrun.cpp 18 |
75 | //! [run-with-overload-calls] |
76 | |
77 | \section2 Returning Values from the Function |
78 | |
79 | Any return value from the function is available via QFuture: |
80 | |
81 | \snippet code/src_concurrent_qtconcurrentrun.cpp 2 |
82 | |
83 | If you don't need the result (for example, because the function returns |
84 | \c{void}), using the QThreadPool::start() overload taking a function object |
85 | is more efficient. |
86 | |
87 | As documented above, passing arguments is done like this: |
88 | |
89 | \snippet code/src_concurrent_qtconcurrentrun.cpp 3 |
90 | |
91 | Note that the QFuture::result() function blocks and waits for the result |
92 | to become available. Use QFutureWatcher to get notification when the |
93 | function has finished execution and the result is available. |
94 | |
95 | \section2 Additional API Features |
96 | |
97 | \section3 Using Member Functions |
98 | |
99 | QtConcurrent::run() also accepts pointers to member functions. The first |
100 | argument must be either a const reference or a pointer to an instance of |
101 | the class. Passing by const reference is useful when calling const member |
102 | functions; passing by pointer is useful for calling non-const member |
103 | functions that modify the instance. |
104 | |
105 | For example, calling QByteArray::split() (a const member function) in a |
106 | separate thread is done like this: |
107 | |
108 | \snippet code/src_concurrent_qtconcurrentrun.cpp 4 |
109 | |
110 | Calling a non-const member function is done like this: |
111 | |
112 | \snippet code/src_concurrent_qtconcurrentrun.cpp 5 |
113 | |
114 | \section3 Using Lambda Functions |
115 | |
116 | Calling a lambda function is done like this: |
117 | |
118 | \snippet code/src_concurrent_qtconcurrentrun.cpp 6 |
119 | |
120 | Calling a function modifies an object passed by reference is done like this: |
121 | |
122 | \snippet code/src_concurrent_qtconcurrentrun.cpp 7 |
123 | |
124 | Using callable object is done like this: |
125 | |
126 | \snippet code/src_concurrent_qtconcurrentrun.cpp 8 |
127 | |
128 | \section1 Concurrent Run With Promise |
129 | |
130 | The \e {Run With Promise} mode enables more control for the running |
131 | task compared to \e basic mode of QtConcurrent::run(). |
132 | It allows progress reporting of the running task, |
133 | reporting multiple results, suspending the execution |
134 | if it was requested, or canceling the task on caller's |
135 | demand. |
136 | |
137 | \section2 The mandatory QPromise argument |
138 | |
139 | The function passed to QtConcurrent::run() in \e {Run With Promise} mode |
140 | is expected to have an additional argument of \c {QPromise<T> &} type, where |
141 | \c T is the type of the computation result (it should match the type \c T |
142 | of QFuture<T> returned by QtConcurrent::run()), like e.g.: |
143 | |
144 | \snippet code/src_concurrent_qtconcurrentrun.cpp 9 |
145 | |
146 | The \c promise argument is instantiated inside the QtConcurrent::run() |
147 | function, and its reference is passed to the invoked \c aFunction, so the |
148 | user doesn't need to instantiate it, nor pass it explicitly |
149 | when calling QtConcurrent::run() in this mode. |
150 | |
151 | The additional argument of QPromise type always needs to appear |
152 | as a first argument on function's arguments list, like: |
153 | |
154 | \snippet code/src_concurrent_qtconcurrentrun.cpp 10 |
155 | |
156 | \section2 Reporting results |
157 | |
158 | In contrast to \e basic mode of QtConcurrent::run(), the function passed to |
159 | QtConcurrent::run() in \e {Run With Promise} mode is expected to always return void type. |
160 | Result reporting is done through the additional argument of QPromise type. |
161 | It also enables multiple result reporting, like: |
162 | |
163 | \snippet code/src_concurrent_qtconcurrentrun.cpp 11 |
164 | |
165 | \note There's no need to call QPromise::start() and QPromise::finish() to |
166 | indicate the beginning and the end of computation (like you would normally do when |
167 | using QPromise). QtConcurrent::run() will always call them before starting and |
168 | after finishing the execution. |
169 | |
170 | \section2 Suspending and canceling the execution |
171 | |
172 | The QPromise API also enables suspending and canceling the computation, if requested: |
173 | |
174 | \snippet code/src_concurrent_qtconcurrentrun.cpp 12 |
175 | |
176 | The call to \c future.suspend() requests the running task to |
177 | hold its execution. After calling this method, the running task |
178 | will suspend after the next call to \c promise.suspendIfRequested() |
179 | in its iteration loop. In this case the running task will |
180 | block on a call to \c promise.suspendIfRequested(). The blocked |
181 | call will unblock after the \c future.resume() is called. |
182 | Note, that internally suspendIfRequested() uses wait condition |
183 | in order to unblock, so the running thread goes into an idle state |
184 | instead of wasting its resources when blocked in order to periodically |
185 | check if the resume request came from the caller's thread. |
186 | |
187 | The call to \c future.cancel() from the last line causes that the next |
188 | call to \c promise.isCanceled() will return \c true and |
189 | \c aFunction will return immediately without any further result reporting. |
190 | |
191 | \note There's no need to call QPromise::finish() to stop the computation |
192 | after the cancellation (like you would normally do when using QPromise). |
193 | QtConcurrent::run() will always call it after finishing the execution. |
194 | |
195 | \section2 Progress reporting |
196 | |
197 | It's also possible to report the progress of a task |
198 | independently of result reporting, like: |
199 | |
200 | \snippet code/src_concurrent_qtconcurrentrun.cpp 13 |
201 | |
202 | The caller installs the \c QFutureWatcher for the \c QFuture |
203 | returned by QtConcurrent::run() in order to |
204 | connect to its \c progressValueChanged() signal and update |
205 | e.g. the graphical user interface accordingly. |
206 | |
207 | \section2 Invoking functions with overloaded operator()() |
208 | |
209 | By default, QtConcurrent::run() doesn't support functors with |
210 | overloaded operator()() in \e {Run With Promise} mode. In case of overloaded |
211 | functors the user needs to explicitly specify the result type |
212 | as a template parameter passed to QtConcurrent::run(), like: |
213 | |
214 | \snippet code/src_concurrent_qtconcurrentrun.cpp 14 |
215 | */ |
216 | |
217 | /*! |
218 | \typedef Function |
219 | \internal |
220 | |
221 | This typedef is a dummy required to make the \c Function |
222 | type name known so that clang doesn't reject functions |
223 | that use it. |
224 | */ |
225 | |
226 | /*! |
227 | \fn QFuture<T> QtConcurrent::run(Function function, ...); |
228 | |
229 | Equivalent to |
230 | \code |
231 | QtConcurrent::run(QThreadPool::globalInstance(), function, ...); |
232 | \endcode |
233 | |
234 | Runs \a function in a separate thread. The thread is taken from the global |
235 | QThreadPool. Note that \a function may not run immediately; \a function |
236 | will only be run once a thread becomes available. |
237 | |
238 | //! [run-description] |
239 | In \l {Concurrent Run (basic mode)} {basic mode} T is the same type as the return value |
240 | of \a function. Non-void return values can be accessed via the QFuture::result() function. |
241 | |
242 | In \l {Concurrent Run (basic mode)} {basic mode} the QFuture returned can only be used to |
243 | query for the running/finished status and the return value of the function. In particular, |
244 | canceling or pausing can be issued only if the computations behind the future |
245 | has not been started. |
246 | |
247 | In \l {Concurrent Run With Promise} {run with promise mode}, the \a function is expected |
248 | to return void and must take an additional argument of \c {QPromise<T> &} type, |
249 | placed as a first argument in function's argument list. T is the result type |
250 | and it is the same for the returned \c QFuture<T>. |
251 | |
252 | In \l {Concurrent Run With Promise} {run with promise mode}, similar to \e basic mode, the |
253 | QFuture returned can be used to query for the running/finished status and the value reported |
254 | by the function. In addition, it may be used for suspending or canceling the |
255 | running task, fetching multiple results from the called \a function or |
256 | monitoring progress reported by the \a function. |
257 | |
258 | \sa {Concurrent Run (basic mode)}, {Concurrent Run With Promise}, QThreadPool::start() |
259 | //! [run-description] |
260 | */ |
261 | |
262 | /*! |
263 | \since 5.4 |
264 | \fn QFuture<T> QtConcurrent::run(QThreadPool *pool, Function function, ...); |
265 | |
266 | Schedules \a function on \a pool. Note that \a function may not run |
267 | immediately; \a function will only be run once a thread becomes available. |
268 | |
269 | \include qtconcurrentrun.cpp run-description |
270 | */ |
271 | |
272 | |