QEventLoop proper usage

Question

I have doubts how should I use QEventLoop. I have 2 pieces of code, both of them work for me (get web resource downloaded).

First one:

QNetworkAccessManager *manager = new QNetworkAccessManager( this );
QNetworkRequest request;
request.setUrl(QUrl(url));
request.setRawHeader("User-Agent", "Mozilla Firefox");
connect(manager, SIGNAL(finished(QNetworkReply*)),this,SLOT(replyFinished(QNetworkReply*)));
manager->get( request )  ;

QEventLoop loop;
connect(manager, SIGNAL(finished(QNetworkReply*)),&loop, SLOT(quit()));
loop.exec();

Second one:

QNetworkAccessManager *manager = new QNetworkAccessManager( this );
QNetworkRequest request;
request.setUrl(QUrl(url));
request.setRawHeader("User-Agent", "Mozilla Firefox");
manager->get( request )  ;

QEventLoop loop;
connect(manager, SIGNAL(finished(QNetworkReply*)),this, SLOT(replyFinished(QNetworkReply*)));
loop.exec();

What I want to know is which one should I use. I mean, does the event loop quit in the second one after signal is emmited? Or do I have to call quit() like in the first one? I found the second solution somewhere but it didn't seem proper to me so I modified it into first piece of code.

Answer 1

I agree with @Mher-Didaryan - that the event loop started by following line of code loop.exec(); in the 2nd code snippet - will never exit. This is because the connect() between the SIGNAL and SLOT is being done for a different event loop than the event loop indicated through EventLoop loop; .

In the case of the 1st code snippet, the logic depends on the finished(QNetworkReply*) signal associated with one & same GET request being emitted to two different event loops. But it is quite possible that

    connect(manager, SIGNAL(finished(QNetworkReply*)),&loop, SLOT(quit()));

may well execute after the manager->get( request ) ; has emitted the finished(QNetworkReply*) signal. Maybe it can happen for a GET type HTTP operation involving a very small file or response. In such a scenario the event loop started out by the loop.exec(); in the 1st code snippet will also not exit. I guess this is what @Mher-Didaryan is also querying in his answer.

Maybe you can use the below QEventLoop logic that would handle the following negative execution scenarios too

1. Timing out of the GET request (say due to network connectivity issues)

2. Error type response from server side of network

   QNetworkAccessManager *manager = new QNetworkAccessManager(this);
   QNetworkRequest request;
   QEventLoop loop;
   QTimer getTimer; // let's use a 10 second period for timing out the GET opn
   request.setUrl(QUrl(url));
   request.setRawHeader("User-Agent", "Mozilla Firefox");
   // connect the timeout() signal of getTimer object to quit() slot of event loop
   QTimer::connect(&getTimer,SIGNAL(timeout()),&loop, SLOT(quit()));
   QObject::connect(manager, SIGNAL(finished(QNetworkReply*)),&loop, SLOT(quit()));
   QNetworkReply *resp = manager->get( request );        
   getTimer.start(10000); // 10000 milliSeconds wait period for get() method to work properly
   loop.exec();
   
   if(NULL == resp)
   {
       // Error. we probably timed out i.e SIGNAL(finished()) did not happen
       // this handles above indicated case (1)
       return -1; // or return some timeout related error value
   }
   else if( QNetworkReply::NoError != resp->error() )
   {
       // Error - SIGNAL(finished()) was raised but get() opn failed & returned with error
       // Refer http://doc.qt.io/qt-4.8/qnetworkreply.html#NetworkError-enum
       // This section of code handles above indicated case (2)
   }
   else
   {
       // get() operation was Successful !.
       // read the response available in the 'resp' variable as a QString & parse it. 
       // Obtain the necessary result and etc.
   }
   
   delete resp;
   delete manager;