[REQ] [Qt5] Emit a generic signal anytime an api call is completed

The current callback in the API interface emits signals specific to the API, for example

    if (worker->error_type == QNetworkReply::NoError) {
        emit createUsersWithArrayInputSignal();
        emit createUsersWithArrayInputSignalFull(worker);
    } else {
        emit createUsersWithArrayInputSignalE(error_type, error_str);
        emit createUsersWithArrayInputSignalEFull(worker, error_type, error_str);
    }

From the outside we can then connect to these signals

The API has different signals per function call.

I would like to be able to know if the API did “any work” not just some “specific work”

For example I would add a signal called void workDone()

    if (worker->error_type == QNetworkReply::NoError) {
        emit createUsersWithArrayInputSignal();
        emit createUsersWithArrayInputSignalFull(worker);
    } else {
        emit createUsersWithArrayInputSignalE(error_type, error_str);
        emit createUsersWithArrayInputSignalEFull(worker, error_type, error_str);
    }

    emit workDone();

    if (worker->error_type == QNetworkReply::NoError) {
        emit createUsersWithListInputSignal();
        emit createUsersWithListInputSignalFull(worker);
    } else {
        emit createUsersWithListInputSignalE(error_type, error_str);
        emit createUsersWithListInputSignalEFull(worker, error_type, error_str);
    }

    emit workDone();

and this would be added to ALL of the callback functions. From the outside then we could connect to know if the API class has done anything yet without needing to connect to a whole bunch of individual signals

This would be trivial for me to add so I wanted to open it up for discussion before just adding a PR

[etherealjoy]

We don’t use the same worker for all calls to a single api class necessarily though? We create one dynamically in each function, so the find children would only apply to that one worker

That is right, so whenever any worker is destroyed we check in the lambda if this which is the API object has any worker remaining. When all workers are destroyed (triggered by abort, timeout and network responses) the current API object will have no more workers and therefore no more active request.

[etherealjoy]

OK thank you for elaborating the use case. What I was trying to say is that, using your example

void doThing1Somewhere() {
  PFXUserApi* api = new PFXUserApi();
  connect(api, &PFXUserApi::addPetSignal, this, function to process doing this);
  connect(api, &PFXUserApi::addPetSignalE, this, function to process an error when doing this);
  api.addPet(pet1ToAdd);
  api.deletePet(pet2ToDelete);
  m_manager->addTask(api);
}

In the above case, It is unknown whether signal callbackExecutedSignal() comes from deletePet() or addPet(). Maybe not an issue in this case since you only care if the call was completed. But if we were to make a feature we should handle the case where we have to emit the signal when all pending request from that object have been done or timed out.