I have a small application that I am trying to create for windows. I am running into an issue with mixing a background thread designed to process some data. This background engine someone needs to both update the application gui (A windows Form) and get information from it.
Here is the basic application main.
int main() {
Engine engine;
Gui g;
engine.run(); // creates a new thread for engine logic
g.ShowDialog();
bool running = false;
while(1)
{
// Update gui with information from the engine
g.update(engine.GetState());
// transition to running
if(g.isRunning() && !running)
{
engine.play();
running = true;
}
// transition to stopped
else if(!g.isRunning() && running)
{
engine.stop();
running = false;
}
}
}
My main problem comes from the fact that Gui class is managed. See Class declaration below.
public ref class Gui : public System::Windows::Forms::Form
I am not really able to mix these two things, at first I had wanted to just throw the Engine into the Gui but that doesn't work since it is unmanaged.
You will note that the problem here is calling ShowDialog() as this makes the dialog modal and no code afterwards is executed. However, if I use Show() ... the Gui simply doesn't update or process any inputs.
SOLUTION:
I created a background worker in the Gui class so the engine is contained within the Gui but is ran on another thread.
void InitializeBackgoundWorker()
{
this->backgroundWorker1 = gcnew System::ComponentModel::BackgroundWorker;
backgroundWorker1->DoWork += gcnew DoWorkEventHandler( this, &Gui::backgroundWorker1_DoWork );
backgroundWorker1->RunWorkerAsync( );
}
delegate void UpdateCallback(int hp, int maxhp);
void UpdateGui(int hp, int maxhp)
{
this->playerHealthBar->Value = ((float)(hp)/(float)(maxhp) * 100.0f);
};
void backgroundWorker1_DoWork( Object^ sender, DoWorkEventArgs^ e )
{
aBotEngine engine;
while(true)
{
engine.runnable(NULL);
array<Object^>^args = gcnew array<Object^>(2);
args[0] = engine.getPlayerHp();
args[1] = engine.getPlayerMaxHp();
this->playerHealthBar->Invoke(gcnew UpdateCallback(this, &Gui::UpdateGui), args);
}
};
As far as I can tell this is the proper way to have background thread that updates your windows form. I'm sure its not the only way.
void InitializeBackgoundWorker()
{
this->backgroundWorker1 = gcnew System::ComponentModel::BackgroundWorker;
backgroundWorker1->DoWork += gcnew DoWorkEventHandler( this, &Gui::backgroundWorker1_DoWork );
backgroundWorker1->RunWorkerAsync( );
}
delegate void UpdateCallback(int hp, int maxhp);
void UpdateGui(int hp, int maxhp)
{
this->playerHealthBar->Value = ((float)(hp)/(float)(maxhp) * 100.0f);
};
void backgroundWorker1_DoWork( Object^ sender, DoWorkEventArgs^ e )
{
aBotEngine engine;
while(true)
{
engine.runnable(NULL);
array<Object^>^args = gcnew array<Object^>(2);
args[0] = engine.getPlayerHp();
args[1] = engine.getPlayerMaxHp();
this->playerHealthBar->Invoke(gcnew UpdateCallback(this, &Gui::UpdateGui), args);
}
};
Related
Using GTKMM, I'm extending the DrawingArea widget with the idea that an external process provides it with images. My CameraDrawingArea will then display the images at the right size using Cairo.
Each time an image arrives, I store it and I call the invalidate method, which eventually ends up in a call to on_draw, where I can resize and display the image.
My problem is the following:
The first 10 or 20 images are displayed as I expected.
After a while, the images keep coming from the provider process, I keep calling invalidate
but on_draw is not called any more.
To show it here, I've simplified the code so that there is nothing external to the class, and no link with other libraries. I've replaced the process providing the images by a method with for-loops, and the display of the image by printing a simple text in the middle of the widget area:
In the constructor I launch a new std::thread to call the doCapture method in the same instance. I also set up a font description, to use it later.
The doCapture method is a silly CPU eater, that does nothing except calling from time to time the refreshDrawing method, as long as keepCapturing is not false.
refreshDrawing invalidates the whole window's rectangle via a call to invalidate.
Gtk's magic is suppose to call on_draw and provide a Cairo context to draw whatever. In my case, for tests purposes, I draw a brownish centered integer.
The class destructor stops the thread by set keepCapturing to false, and waits for termination with a join.
#include "camera-drawing-area.hpp"
#include <iostream>
CameraDrawingArea::CameraDrawingArea():
captureThread(nullptr) {
fontDescription.set_family("Monospace");
fontDescription.set_weight(Pango::WEIGHT_BOLD);
fontDescription.set_size(30 * Pango::SCALE);
keepCapturing = true;
captureThread = new std::thread([this] {
doCapture();
});
}
void CameraDrawingArea::doCapture() {
while (keepCapturing) {
float f = 0.0;
for (int n = 0; n < 1000; n++) {
for (int m = 0; m < 1000; m++) {
for (int o = 0; o < 500; o++) {
f += 1.2;
}
}
}
std::cout << "doCapture - " << f << std::endl;
refreshDrawing();
}
}
void CameraDrawingArea::refreshDrawing() {
auto win = get_window();
if (win) {
win->invalidate(false);
std::cout << "refreshDrawing" << std::endl;
}
}
bool CameraDrawingArea::on_draw(const Cairo::RefPtr<Cairo::Context>& cr) {
std::cout << "on_draw" << std::endl;
static char buffer[50];
static int n = 0;
sprintf(buffer, "-%d-", n++);
Gtk::Allocation allocation = get_allocation();
const int width = allocation.get_width();
const int height = allocation.get_height();
auto layout = create_pango_layout(buffer);
layout->set_font_description(fontDescription);
int textWidth, textHeight;
layout->get_pixel_size(textWidth, textHeight);
cr->set_source_rgb(0.5, 0.2, 0.1);
cr->move_to((width - textWidth) / 2, (height - textHeight) / 2);
layout->show_in_cairo_context(cr);
cr->stroke();
return true;
}
CameraDrawingArea::~CameraDrawingArea() {
keepCapturing = false;
captureThread->join();
free(captureThread);
}
And this is my header file:
#ifndef CAMERA_DRAWING_AREA_HPP
#define CAMERA_DRAWING_AREA_HPP
#include <gtkmm.h>
#include <thread>
class CameraDrawingArea : public Gtk::DrawingArea {
public:
CameraDrawingArea();
virtual ~CameraDrawingArea();
protected:
bool on_draw(const Cairo::RefPtr<Cairo::Context>& cr) override;
private:
bool keepCapturing;
void doCapture();
void refreshDrawing();
std::thread* captureThread;
Pango::FontDescription fontDescription;
};
#endif
The problem manifests itself as follows:
When starting the application, it faithfully displays 1, 2, 3...
Between 5th and 20th iteration (it's random, but rarely outside these ranges), it stops refreshing.
Because of the cout, I can see that refreshDrawing is called be sure that invalidate is also called, but on_draw isn't.
Also, if I stop the application before it stops refreshing, then it ends up nicely. But, if I stop the application after it stops refreshing, then I see this message below (the ID value varies):
GLib-CRITICAL **: 10:05:04.716: Source ID 25 was not found when attempting to remove it
I'm quite sure that I do something wrong, but clueless about what. Any help would be appreciated.
I also checked the following questions, but they don't seem to be related with my case:
Draw signal doesn't get fired in GTKMM, when derived class doesn't call a superclass's constructor
You can't use GTK methods from any other thread than the one in which you started the GTK main loop. Probably the win->invalidate() call is causing things to go wrong here.
Instead, use Glib::Dispatcher to communicate with the main thread, or use gdk_threads_add_idle() for a more C-style solution.
Based on the answer form #ptomato, I've rewritten my example code. The golden rule is do not call GUI functions from another thread, but if you do, then acquire some specific GDK locks first. That's the purpose of Glib::Dispatcher :
If a Glib::Dispatcher object is constructed in the main GUI thread (which will therefore be the receiver thread), any worker thread can emit on it and have the connected slots safely execute gtkmm functions.
Based on that, I've added a new private member Glib::Dispatcher refreshDrawingDispatcher that will allow threads to safely the invalidate the windows area:
#ifndef CAMERA_DRAWING_AREA_HPP
#define CAMERA_DRAWING_AREA_HPP
#include <gtkmm.h>
#include <thread>
class CameraDrawingArea :
public Gtk::DrawingArea {
public:
CameraDrawingArea();
virtual ~CameraDrawingArea();
protected:
bool on_draw(const Cairo::RefPtr<Cairo::Context>& cr) override;
private:
bool keepCapturing;
void doCapture();
void refreshDrawing();
Glib::Dispatcher refreshDrawingDispatcher;
std::thread* captureThread;
Pango::FontDescription fontDescription;
};
#endif
Then, I've connected the dispatcher to the refreshDrawing method. I do this in the class constructor, which is called during GUI start up and therefore in the main GUI thread:
CameraDrawingArea::CameraDrawingArea():
refreshDrawingDispatcher(),
captureThread(nullptr) {
fontDescription.set_family("Monospace");
fontDescription.set_weight(Pango::WEIGHT_BOLD);
fontDescription.set_size(30 * Pango::SCALE);
keepCapturing = true;
captureThread = new std::thread([this] {
doCapture();
});
refreshDrawingDispatcher.connect(sigc::mem_fun(*this, &CameraDrawingArea::refreshDrawing));
}
Finally, the thread has to call the dispatcher:
void CameraDrawingArea::doCapture() {
while (keepCapturing) {
float f = 0.0;
for (int n = 0; n < 1000; n++) {
for (int m = 0; m < 1000; m++) {
for (int o = 0; o < 500; o++) {
f += 1.2;
}
}
}
std::cout << "doCapture - " << f << std::endl;
refreshDrawingDispatcher.emit();
}
}
And now, this works without further problems.
I have the following code:
tbb::concurrent_bounded_queue<Image> camera_queue_;
camera_queue_.set_capacity(1);
struct Image
{
int hour_;
int minute_;
int second_;
int msec_;
QImage image_;
Image(){hour_ = -1; minute_ = -1; second_ = -1; msec_ = -1; image_ = QImage();}
Image& operator=(Image const& copy)
{
this->hour_ = copy.hour_;
this->minute_ = copy.minute_;
this->second_ = copy.second_;
this->msec_ = copy.msec_;
this->image_ = copy.image_;
return *this;
}
};
In a Qt Thread :
ThreadA:
tbb::concurrent_bounded_queue<Image> image_queue_;
image_queue_.set_capacity(1);
Image cur_image_;
void Worker::process() {
while(1)
{
if(quit_)
break;
{
camera_queue_.pop(cur_image_);
image_queue_.push(cur_image_);
}
emit imageReady();
}
emit finished();
}
Image Worker::getCurrentImage()
{
Image tmp_image;
image_queue_.pop(tmp_image);
return tmp_image;
}
In Another Thread:
ThreadB:
Producer::Producer(){
work_ = new Worker();
work_->moveToThread(workerThread_);
QObject::connect(workerThread_, &QThread::finished, work_, &QObject::deleteLater);
QObject::connect(this, &Producer::operate, work_, &Worker::process);
QObject::connect(work_, &Worker::imageReady, this, &Producer::displayImage);
QObject::connect(this, &Producer::stopDecode, work_, &Worker::stop);
workerThread_->start();
emit operate();
}
void Producer::process() {
while(1)
{
if(quit_)
break;
{
camera_queue_.push(GetImage());
}
}
}
void Producer::displayImage()
{
Image tmp = std::move(work_->getCurrentImage());
widget_->showImage(tmp.image_);
}
However, In main thread, I have a function that enables user to click a button to get current image:
bool Producer::SaveImage()
{
Image img = std::move(work_->getCurrentImage());
std::string fileName = std::to_string(img.hour_) + "-" + std::to_string(img.minute_) + "-" + std::to_string(img.second_) + "-" + std::to_string(img.msec_/1000) + ".jpg";
std::string outFileName = folder + "/" + fileName;
return img.image_.save(QString::fromStdString(outFileName));
}
The problem is:
When user does not click the button to invoke Producer::SaveImage(), the Image Decoding and Showing runs smoothly. But when user invoke Producer::SaveImage(), the whole program will get stuck (Caton phenomenon ?). The GUI response becomes not that smooth. The more user invokes SaveImage, the slower the GUI response becomes.
Can anyone help to explain why ? Is there a way to solve that ?
Why do you want to use concurrent queue? It looks like there is a syncronization mechanism in place and you rely mostly on it instead of using concurrent_queue for synchronisation and communication as it is supposed for.
The issue is that when you set capacity = 1, both operations of concurrent_bounded_queue will block until there is enough space of items in the queue. E.g. if the queue contains an item already, the push operation will block. And since you control your threads with another notification mechanism, you might catch a deadlock.
In particular, try to swap the operations like below:
camera_queue_.pop(cur_image_);
emit imageReady();
image_queue_.push(cur_image_);
This should prepare the thread which receives images (if I understand this correctly) and it will block on its image_queue_.pop() method, then this thread will put new image and unblocks the recipient. There might be other issues similar to this, so, please rethink all of your synchronization.
I have an application which has the parent frame as the MainFrame and two child frames, ChildA and ChildB. This should be a document oriented GUI such that when user clicks a document, it should be able to open another instance of FrameA.
When I click the executable of GUI MainFrame, ChildA and ChildB loads normally. However when the executable is running and when I click a document I realized that Windows is opening it as another process such that MainFrame, ChildA and ChildB loads again which I dont want cause I only want to open another instance of ChildA to show the contents of the document.
My code is as follows:
bool MainFrameApp::OnInit()
{
m_AnotherInstanceRunning=false;
m_checker = new wxSingleInstanceChecker;
if(m_checker->IsAnotherRunning()) m_AnotherInstanceRunning=true;
//Some other code
if(!m_AnotherInstanceRunning)
{
frame= new MainFrame(0L);
std::cout<<"Frame:"<<frame; //Prints the address
auto m_ChildA=new SomeChildFrame(frame);
frame->Show();
m_FrameA->Show(true);
wxTheApp->SetTopWindow(frame);
}
if(m_AnotherInstanceRunning)
{
frame=dynamic_cast<MainFrame*>(wxTheApp->GetTopWindow());
std::cout<<"Frame:"<<frame; //prints 0
frame->OpenDocument(frame,strDocDirectory,strDocName); //Does not work
return false;
}
My question is, when I click on a document how can I open that document within an already running MainFrame.
EDIT 1:
This is how I achieved it after countless of hours...
Below is the code for MainFrame.h
const int SERVER_ID=wxNewId();
const int SERVERSOCKET_ID=wxNewId();
const int CLIENTSOCKET_ID=wxNewId();
class MainFrame;
class MainFrameApp : public wxApp
{
bool m_AnotherInstanceRunning;
wxSocketServer* m_server;
wxString m_DocDirectoryPath;
wxString m_DocFullName; //including extension
protected:
virtual void OnServerEvent(wxSocketEvent& event);
virtual void OnServerSocketEvent(wxSocketEvent& event);
virtual void OnClientSocketEvent(wxSocketEvent& event);
public:
MainFrame* m_MainFrame;
wxSingleInstanceChecker* m_checker;
};
static const wxCmdLineEntryDesc cmdLineDesc[] =
{
{wxCMD_LINE_PARAM, NULL, NULL, "", wxCMD_LINE_VAL_STRING, wxCMD_LINE_PARAM_MULTIPLE},
{wxCMD_LINE_NONE},
};
Below is the code for MainFrame.cpp
bool ScienceSuitApp::OnInit()
{
m_AnotherInstanceRunning=false;
m_checker = new wxSingleInstanceChecker;
if(m_checker->IsAnotherRunning()) m_AnotherInstanceRunning=true;
wxFileName exePath(wxStandardPaths::Get().GetExecutablePath());
glbStrExeDir=exePath.GetPath()+wxFileName::GetPathSeparator();
wxString strDocFullPath; //This is directory+name
if(argc>1)
{
wxCmdLineParser parser(cmdLineDesc, argc, argv);
parser.Parse();
strDocFullPath=argv[1];
wxFileName docPath(strDocFullPath);
m_DocDirectoryPath=docPath.GetPath()+wxFileName::GetPathSeparator();
m_DocFullName=docPath.GetFullName(); //we needed extension of document as well
}
if(!m_AnotherInstanceRunning)
{
//Some other code
//Here Process #1
m_MainFrame= new MainFrame(0L);
m_MainFrame->Show();
wxIPV4address addr;
addr.Service(3000);
m_server = new wxSocketServer(addr,wxSOCKET_WAITALL);
if (!m_server->Ok())
{
if (m_server->LastError() == wxSOCKET_INVPORT) wxMessageBox(wxT("Port in use!"));
m_server->Destroy();
}
m_server->SetEventHandler(*this, SERVER_ID);
m_server->SetNotify(wxSOCKET_CONNECTION_FLAG);
m_server->Notify(true);
Connect(SERVER_ID,wxID_ANY, wxEVT_SOCKET,wxSocketEventHandler(MainFrameApp::OnServerEvent));
Connect(SERVERSOCKET_ID,wxID_ANY,wxEVT_SOCKET,wxSocketEventHandler(MainFrameApp::OnServerSocketEvent));
}
if(m_AnotherInstanceRunning)
{
//Here is for Process #2
Connect(CLIENTSOCKET_ID,wxID_ANY,wxEVT_SOCKET,wxSocketEventHandler(MainFrameApp::OnClientSocketEvent));
wxIPV4address addr;
addr.Hostname(wxT("localhost"));
addr.Service(3000);
wxSocketClient* Socket = new wxSocketClient();
Socket->SetEventHandler(*this, CLIENTSOCKET_ID);
Socket->SetNotify(wxSOCKET_CONNECTION_FLAG | wxSOCKET_INPUT_FLAG | wxSOCKET_LOST_FLAG);
Socket->Notify(true);
if(!Socket->Connect(addr, true)) wxMessageBox("Failed to connect to server");
return true;
}
return true;
}
void MainFrameApp::OnServerEvent(wxSocketEvent& event) //Process #1
{
wxSocketBase* sock = m_server->Accept(false);
sock->SetEventHandler(*this, SERVERSOCKET_ID);
sock->SetNotify(wxSOCKET_INPUT_FLAG | wxSOCKET_LOST_FLAG);
sock->Notify(true);
}
void MainFrameApp::OnServerSocketEvent(wxSocketEvent& event) //Process #1
{
wxSocketBase *sock = event.GetSocket();
wxCharBuffer buf(256);
switch(event.GetSocketEvent())
{
case wxSOCKET_CONNECTION:
{
break;
}
case wxSOCKET_INPUT:
{
sock->Read(buf.data(), 255);
wxString pathstr=wxString::FromUTF8(buf);
sock->Destroy();
wxFileName docPath(pathstr);
wxString DocDirectoryPath=docPath.GetPath()+wxFileName::GetPathSeparator();
wxString DocFullName=docPath.GetFullName(); //we needed extension of document as well
m_MainFrame->OpenDocument(m_MainFrame,DocDirectoryPath,DocFullName);
break;
}
case wxSOCKET_LOST:
{
sock->Destroy();
break;
}
}
}
void MainFrameApp::OnClientSocketEvent(wxSocketEvent& event) //Process #2
{
wxString str=m_DocDirectoryPath+m_DocFullName;
wxSocketBase* sock = event.GetSocket();
switch(event.GetSocketEvent())
{
case wxSOCKET_CONNECTION:
{
sock->Write(str.mb_str(),str.length());
}
}
//Exit the client process (Process #2) staying in the background, otherwise and no messages are sent
//Also we dont want to spawn background processes
exit(0);
}
If you want to prevent more than one instance of your application running at the same time, you can use wxSingleInstanceChecker to check if another one is running and use the IPC classes to send the name of the document to open to the other instance if it is.
Today I encountered a problem with repaint() function from QT libraries. Long story short, I got a slot where I train my neural network using BP algorithm. I had tested the whole algorithm in console and then wanted to move it into GUI Application. Everything works fine except refreshing. Training of neural networks is a process containing a lot of computations, which are made in bp_alg function (training) and licz_mse function (counting a current error). Variable ilosc_epok can be set up to 1e10. Therefore the whole process may last even several hours. Thats why I wanted to display a current progress after each 100000 epochs (the last if contition). wyniki is an object of QTextEdit class used for displaying the progress. Unfortunately, repaint() doesnt work as intended. At the beginning it refreshes wyniki in GUI, but after some random time it stops working. When the external loop is finished, it refreshes once again showing all changes.
I tried to change frequency of refreshing, but sooner or later it always stops (unless the whole training process stops early enough because of satisfying the break condition). It looks like at some moment of time the application decides to stop refreshing because of too many computations. Imo it shouldnt happen. I was looking for a solution among older questions and managed to solve the problem when I used qApp->processEvents(QEventLoop::ExcludeUserInputEvents); instead of wyniki->repaint();. However, Im still curious why repaint() stops working just like that.
Below I paste a part of the code with the problematic part. Im using QT Creator 2.4.1 and QT Libraries 4.8.1 if it helps.
unsigned long int ile_epok;
double mse_w_epoce;
for (ile_epok=0; ile_epok<ilosc_epok; ile_epok++) { //external loop of training
mse_w_epoce = 0;
for (int i=0; i<zbior_uczacy_rozmiary[0]; i++) { //internal loop of training
alg_bp(zbior_uczacy[i], &zbior_uczacy[i][zbior_uczacy_rozmiary[1]]);
mse_w_epoce += licz_mse(&zbior_uczacy[i][zbior_uczacy_rozmiary[1]]);
}
//checking break condition
if (mse_w_epoce < warunek_stopu) {
wyniki->append("Zakończono uczenie po " + QString::number(ile_epok) + " epokach, osiągając MSE: " + QString::number(mse_w_epoce));
break;
}
//problematic part
if ((ile_epok+1)%(100000) == 0) {
wyniki->append("Uczenie w toku, po " + QString::number(ile_epok+1) + " epokach MSE wynosi: " + QString::number(mse_w_epoce));
wyniki->repaint();
}
}
You're blocking your GUI thread, so repaints will not work, it's just plainly bad design. You're never supposed to block the GUI thread.
If you insist on doing the work in the GUI thread, you must forcibly chop the work into small chunks and return to the main event loop after each chunk. Nested event loops are evil, so don't even think you'd want one. All this has a bad code smell, so stay away.
Alternatively, simply move your computation QObject to a worker thread and do the work there.
The code below demonstrates both techniques. It's easy to notice that the chopping-up-of-work requires to maintain loop state inside of the worker object, not merely locally in the loop. It's messier, the code smells bad, again - avoid it.
The code works under both Qt 4.8 and 5.1.
//main.cpp
#include <QApplication>
#include <QThread>
#include <QWidget>
#include <QBasicTimer>
#include <QElapsedTimer>
#include <QGridLayout>
#include <QPlainTextEdit>
#include <QPushButton>
class Helper : private QThread {
public:
using QThread::usleep;
};
class Trainer : public QObject {
Q_OBJECT
Q_PROPERTY(float stopMSE READ stopMSE WRITE setStopMSE)
float m_stopMSE;
int m_epochCounter;
QBasicTimer m_timer;
void timerEvent(QTimerEvent * ev);
public:
Trainer(QObject *parent = 0) : QObject(parent), m_stopMSE(1.0) {}
Q_SLOT void startTraining() {
m_epochCounter = 0;
m_timer.start(0, this);
}
Q_SLOT void moveToGUIThread() { moveToThread(qApp->thread()); }
Q_SIGNAL void hasNews(const QString &);
float stopMSE() const { return m_stopMSE; }
void setStopMSE(float m) { m_stopMSE = m; }
};
void Trainer::timerEvent(QTimerEvent * ev)
{
const int updateTime = 50; //ms
const int maxEpochs = 5000000;
if (ev->timerId() != m_timer.timerId()) return;
QElapsedTimer t;
t.start();
while (1) {
// do the work here
float currentMSE;
#if 0
for (int i=0; i<zbior_uczacy_rozmiary[0]; i++) { //internal loop of training
alg_bp(zbior_uczacy[i], &zbior_uczacy[i][zbior_uczacy_rozmiary[1]]);
currentMSE += licz_mse(&zbior_uczacy[i][zbior_uczacy_rozmiary[1]]);
}
#else
Helper::usleep(100); // pretend we're busy doing some work
currentMSE = 2E4/m_epochCounter;
#endif
// bail out if we're done
if (currentMSE <= m_stopMSE || m_epochCounter >= maxEpochs) {
QString s = QString::fromUtf8("Zakończono uczenie po %1 epokach, osiągając MSE: %2")
.arg(m_epochCounter).arg(currentMSE);
emit hasNews(s);
m_timer.stop();
break;
}
// send out periodic updates
// Note: QElapsedTimer::elapsed() may be expensive, so we don't call it all the time
if ((m_epochCounter % 128) == 1 && t.elapsed() > updateTime) {
QString s = QString::fromUtf8("Uczenie w toku, po %1 epokach MSE wynosi: %2")
.arg(m_epochCounter).arg(currentMSE);
emit hasNews(s);
// return to the event loop if we're in the GUI thread
if (QThread::currentThread() == qApp->thread()) break; else t.restart();
}
m_epochCounter++;
}
}
class Window : public QWidget {
Q_OBJECT
QPlainTextEdit *m_log;
QThread *m_worker;
Trainer *m_trainer;
Q_SIGNAL void startTraining();
Q_SLOT void showNews(const QString & s) { m_log->appendPlainText(s); }
Q_SLOT void on_startGUI_clicked() {
QMetaObject::invokeMethod(m_trainer, "moveToGUIThread");
emit startTraining();
}
Q_SLOT void on_startWorker_clicked() {
m_trainer->moveToThread(m_worker);
emit startTraining();
}
public:
Window(QWidget *parent = 0, Qt::WindowFlags f = 0) :
QWidget(parent, f), m_log(new QPlainTextEdit), m_worker(new QThread(this)), m_trainer(new Trainer)
{
QGridLayout * l = new QGridLayout(this);
QPushButton * btn;
btn = new QPushButton("Start in GUI Thread");
btn->setObjectName("startGUI");
l->addWidget(btn, 0, 0, 1, 1);
btn = new QPushButton("Start in Worker Thread");
btn->setObjectName("startWorker");
l->addWidget(btn, 0, 1, 1, 1);
l->addWidget(m_log, 1, 0, 1, 2);
connect(m_trainer, SIGNAL(hasNews(QString)), SLOT(showNews(QString)));
m_trainer->connect(this, SIGNAL(startTraining()), SLOT(startTraining()));
m_worker->start();
QMetaObject::connectSlotsByName(this);
}
~Window() {
m_worker->quit();
m_worker->wait();
delete m_trainer;
}
};
int main(int argc, char *argv[])
{
QApplication a(argc, argv);
Window w;
w.show();
return a.exec();
}
#include "main.moc"
I'm having problem (basically, I'm confused.) while trying to create a worker thread for my wxWidgets GUI application that WILL MODIFY one of the GUI property itself. (In this case, wxTextCtrl::AppendText).
So far, I have 2 source files and 2 header files for the wx program itself (excluding my own libs, MySQL lib, etc), say MainDlg.cpp which contains a derived class of wxFrame called 'MainDlg' and MainForm.cpp which contains a derived class of wxApp called 'MainForm'.
MainForm.cpp
#include "MainHeader.h" // contains multiple header files
IMPLEMENT_APP(MainForm)
bool MainForm::OnInit()
{
MainDlg *Server = new MainDlg(wxT("App Server 1.0"), wxDEFAULT_FRAME_STYLE - wxRESIZE_BORDER - wxMAXIMIZE_BOX);
Editor->Show();
return true;
}
MainDlg.cpp:
#include "MainHeader.h"
BEGIN_EVENT_TABLE(MainDlg, wxFrame)
EVT_BUTTON(6, MainDlg::StartServer)
EVT_BUTTON(7, MainDlg::StopServer)
END_EVENT_TABLE()
CNETServerConnection *cnServCon;
std::string ServerIP, DBHost, DBUser, DBName, DBPass;
int UserCapacity, DBPort, ServerPort;
MYSQL *sqlhnd;
MainDlg::MainDlg(const wxString &title, long style) : wxFrame(NULL, wxID_ANY, title, wxDefaultPosition, wxSize(301, 230), style)
{
cnServCon = new CNETServerConnection(100);
this->InitializeComponent();
}
void MainDlg::InitializeComponent()
{
this->SetTitle(wxT("App Server 1.0"));
this->SetSize(396, 260);
this->SetBackgroundColour(wxSystemSettings::GetColour(wxSYS_COLOUR_3DFACE));
this->Centre();
statBox = new wxTextCtrl(this, 4, wxT("Welcome to AppServer 1.0\n\n"), wxPoint(10, 10), wxSize(371, 141), wxTE_MULTILINE | wxTE_READONLY);
//.................................
}
void MainDlg::StartServer(wxCommandEvent &event)
{
this->startBtn->Enable(false);
this->AppendStatus(wxT("\nLoading server configuration... "));
//.................................
this->AppendStatus(wxT("OK\n\nServer ready!\n\n"));
// When the server is ready, I need to run a thread
// that will update the statBox (through AppendStatus func or wxTextCtrl::AppendText directly)
// regularly without interrupting the GUI itself.
// Because the thread will contain a while-loop
// to make the program keep receiving message from clients.
this->startBtn->Hide();
this->stopBtn->Show();
this->cmdBtn->Enable();
this->cmdBox->Enable();
}
void MainDlg::StopServer(wxCommandEvent &event)
{
//...................................
}
void MainDlg::AppendStatus(const wxString &message)
{
statBox->AppendText(message);
}
// Well, here is the function I'd like to run in a new thread
void MainDlg::ListenForMessages()
{
int MsgSender = 0;
while(1)
{
if(!cnServCon->GetNewMessage())
continue;
if(MsgSender = cnServCon->GetJoiningUser())
this->AppendStatus(wxT("Someone connected to the server."));
}
}
I also found an usage example of wxThread from Simple example of threading in C++:
class MessageThread : public wxThread
{
private:
MessageThread(const MessageThread ©);
public:
MessageThread() : wxThread(wxTHREAD_JOINABLE)
{
}
void *Entry(void)
{
// My works goes here
return;
}
};
wxThread *CreateThread()
{
wxThread *_hThread = new MessageThread();
_hThread->Create();
_hThread->Run();
return _hThread;
}
But I don't know how to associate it with my program and make it able to modify my GUI property (statBox).
Any kind of help would be appreciated! :)
Thanks.
The easiest is to create an event type id, and use a wxCommandEvent using the type id, set its string member ("evt.SetText"), and send the event to one of your windows (using AddPendingEvent). In the handler of that event, you can then call AppendText on your control using the text you sent (evt.GetText), because you are in the GUI thread by then.
// in header
DECLARE_EVENT_TYPE(wxEVT_MY_EVENT, -1)
// in cpp file
DEFINE_EVENT_TYPE(wxEVT_MY_EVENT)
// the event macro used in the event table. id is the window id you set when creating
// the `wxCommandEvent`. Either use -1 or the id of some control, for example.
EVT_COMMAND(window-id, event-id, handler-function)
Here is an overview how it works: Custom Events.