GUI event blocking

JonB

@ambershark
OK, now had a chance to break and examine stack.

Nobody said this code would be easy...

The "long running operation" involves producing HTML, and saving to PDF file (think of it like a report). I do this via QWebEngineView (partly because this is in shared code, at other times the generated HTML is displayed to the user and is editable prior to producing the PDF). In this case the view is never displayed interactively to the user, but it still follows that route.

Because QWebEngineView is asynchronous (its own thread, I believe), and here we need the HTML/PDF synchronously, before we can save the PDF to file we must await QWebEngineView finished loading (else we get empty PDF saved to file). I have code there like:

    def synchronousRenderHtml(self):
        # see synchronousWebViewLoaded() below
        self.rendered = False
        self.webView.loadFinished.connect(self.synchronousWebViewLoaded)
        #
        # see http://doc.qt.io/qt-5/qtwebenginewidgets-qtwebkitportingguide.html
        # because setHtml() is asynchronous, the call to make editable has to be moved into synchronousWebViewLoaded()
        htmlFile = filesystemfunctions.findFileInDataOrCodeDirectory(Paths.s166Html())
        self.webView.setHtml(self.html, QUrl.fromLocalFile(str(htmlFile)))
        #
        # see synchronousWebViewLoaded() below
        if not self.rendered:
            self.renderLoop.exec()

    def synchronousWebViewLoaded(self):
        # This function is called on self.webView.loadFinished() from synchronousRenderHtml() above
        # The code uses this because self.webView.setHtml() is asynchronous
        # so in effect this implements a "blocking" call (with event processing) in synchronousRenderHtml()
        # This is the only place in code where a QEventLoop is used explicitly
        # It's all very complicated, and I did wish this behaviour was not needed
        # but it seems, at minimum in the case where this dialog is used non-interactively, it is :(
        self.qWebEngineView.page().runJavaScript("document.documentElement.contentEditable = true")
        self.rendered = True
        # cause the self.renderLoop.exec() in synchronousRenderHtml() above to exit now
        self.renderLoop.quit()

Turns out, the re-entrant call for the top-level button being pressed a second time is coming from the self.renderLoop.exec() (which is awaiting the self.renderLoop.quit()).

Cutting a long story short, unless someone can suggest a better way than this self.renderLoop.exec() principle(?), I think I have to stick with that. Hence the button re-entrancy, and I'll have to live with the need to using a re-entrancy blocking flag....

kshegunov

@JonB said in GUI event blocking:

Cutting a long story short, unless someone can suggest a better way than this self.renderLoop.exec() principle(?),

Not in this particular case. But I have a suggestion:
Grab the mouse and keyboard in synchronousRenderHtml and discard those events in your event() override while self.rendered is false and you're owning the user input. In synchronousWebViewLoaded release back the mouse and keyboard. Tread carefully though, as grabbing the mouse and keyboard can make your system unusable (i.e. use -nograb with gdb to prevent you from cursing). ;)

JonB

@kshegunov
I did think about that "dirty" way of handling things.

However (apart from the debugging issue, I use Python + PyCharm debugger, not gdb, and I wouldn't hold your breath that it will have a -nograb functionality!), the problem will be which "event() override" you have in mind.

Because, as I said, the QWebEngineView, which produces the HTML/PDF asynchronously and has the synchronousRenderHtml() call, is not displayed to the user in this case, the UI events are being received to the original dialog where a button is pressed to produce the output. This in itself has no connection to the ("invisible") QWebEngineView --- it doesn't even know one is being used, the implementation is invisible to it. I would have to modify its event handling. And since that would break encapsulation, I really don't feel it would be a good idea to implement!

So I will soldier on with the "blocking flag" I currently have to prevent "re-entrancy", which at least works reasonably well in practice. At least I now understand why this whole behaviour is caused.

kshegunov

@JonB said in GUI event blocking:

the problem will be which "event() override" you have in mind.

That would be the event() override of the class (widget) in which you grab the input. When you grab the input events all of them are redirected to the widget that called the mentioned functions, so you'd need to filter them in that same class. Basically, you grab the input in the widget/dialog/w/e and filter the input there. Whenever the operation's done (i.e. you connect the finished signal to a slot in that same dialog you release the mouse/keyboard.

JonB

@kshegunov
Sorry, my friend, but as per the explanation I wrote above, I still don't understand which widget's events you mean.

Here is the architecture of how the code works:

The user starts from a dialog, which has buttons on it, like Run long-running report.
That button on that dialog makes call produceReportAndWaitForFinish().
Behind the scenes, produceReportAndWaitForFinish() goes: create invisible QWebEnginePage and await final HTML/PDF returned, invisible QWebEnginePage calls synchronousRenderHtml(), which in turn has the self.renderLoop.exec() & self.renderLoop.quit() inside itself (the QWebEnginePage), in order to produce the HTML/PDF.
You want me to place the "event filter" there (inside the QWebEnginePage), when I call self.renderLoop.exec().
However, I don't think that (the QWebEnginePage) receives any inputs. The inputs (which would cause re-entrancy) are directed to buttons on the original top-level dialog where the user originally clicked the Run long-running report button, which I don't want him to click again.

If that is correct(?), that is why I don't want to fiddle inside QWebEnginePage with the events directed to the top-level dialog, because of encapsulation/separation of code. The QWebEnginePage does not know if it was invoked from a dialog in the first place, and the dialog has no idea that a QWebEnginePage (which will do threads/events) is involved in the production of the HTML/PDF.

kshegunov

@JonB said in GUI event blocking:

If that is correct

Nope. I mean exactly the dialog with the buttons. See (untested) code below, which I hope will clear this up:

class DialogWithButtons : public QDialog
{
public:
    DialogWithButtons(QWidget * parent);

public slots:
    void produceReportAndWaitForFinish();
    void reportProduced();

protected:
    bool event(QEvent *) override;
private:
    QPushButton startLongRunningOpButton;
    bool inputBlocked;
}

DialogWithButtons::DialogWithButtons(QWidget * parent)
    : QWidget(parent), startLongRunningOpButton(this), inputBlocked(false)
{
    // Just connect the button
    QObject::connect(&startLongRunningOpButton, &QPushButton::clicked, this, &DialogWithButtons::produceReportAndWaitForFinish);
}

void DialogWithButtons::produceReportAndWaitForFinish()
{
    inputBlocked = true;
    grabMouse();
    grabKeyboard();

    // Create/Init the webengine and so on, put to render and all that goodness. 
    QWebEngineView webEngine;

    // Connect the handlers & QEventLoop blocking
    QObject::connect(&webEngine, &QWebEngineView::loadFinished, this, &DialogWithButtons::reportProduced);

    QEventLoop loop;
    QObject::connect(&webEngine, &QWebEngineView::loadFinished, &loop, &QEventLoop::quit);
    loop.exec();   // Wait for processing to finish
}

void DialogWithButtons::reportProduced()
{
     releaseMouse();
     releaseKeyboard();
     inputBlocked = false;
}

bool DialogWithButtons::event(QEvent * e)
{
    if (inputBlocked && dynamic_cast<QInputEvent *>(e))
        return true;   // Filter out input events - we receive all of them if `inputBlocked` is true
 
    return QDialog::event(e);  // Input was not blocked or event was not UI input - delegate to the default implementation
}

JonB

@kshegunov
Yes, OK, I thought so.

The problem is, my produceReportAndWaitForFinish() in the top-level dialog does not have any kind of

    // Create/Init the webengine and so on, put to render and all that goodness. 
    QWebEngineView webEngine;

inside it (and I don't want it to). It has more like:

    //  Call *completely opaque* ReportProducer to generate HTML/PDF
    //  Here we have *no knowledge* of how ReportProducer functions
    //  and to maintain code separation/encapsulation we do not want to know here
    //  We might completely change how ReportProducer works at any time
    //  without affecting any code/behaviour here in this dialog.
    ReportProducer rp;
    rp.doWhateverToProduceHtmlAndPdf()

So I do understand what your approach requires, but I hope you can see why I do not wish to go down that route. But thanks for your suggestion anyway!

kshegunov

@JonB said in GUI event blocking:

The problem is, my produceReportAndWaitForFinish() in the top-level dialog does not have any kind of
[snip]
inside it (and I don't want it to). It has more like:

This is of no consequence, I've put the event loop/QWebEngineView here only for completeness. You can connect the grab/release slots directly from the button/custom class that generates the pdf respectively. Basically, the same as what you have now, but including 1 event override and putting in the grabMouse and releaseMouse at the appropriate places.

JonB

@kshegunov
Hmm, so you mean in my dialog button:

blockInputs()
ReportProducer rp;
rp.doWhateverToProduceHtmlAndPdf()
releaseInputs()

Right?

kshegunov

Yep, that should work.

JonB

@kshegunov
So, to be pedantic, and hoping I'm not annoying you, I want precisely that code/behaviour wrapped around every single button handler I have anywhere in my code, just in case the code the button invokes happens to create a thread or process events which could cause re-entrance. Which was the reason I posted this thread in the very first place.

There is supposed to be nothing special about my doWhateverToProduceHtmlAndPdf(), it's a complete accident/coincidence that it happens to use something (a QWebEngineView) which itself has to spin an event loop to achieve its work. I might change that tomorrow. Which is why I wanted to isolate anything about its event behaviour down inside the class which creates the QWebEngineView and not up at the caller dialog level. This approach will work in this particular case, but not in other random places in code where a button's actions might happen to spin an event loop, and that bothers me....

Ah well, such are the dirty practicalities of coding...!

EDIT: Bear with me, I'm going to post a new thread to ask a particular question which would satisfy me, just to know if it's possible.... See https://forum.qt.io/topic/88481/is-it-possible-to-intercept-a-modal-dialog-s-event-loop

kshegunov

@JonB said in GUI event blocking:

This approach will work in this particular case, but not in other random places in code where a button's actions might happen to spin an event loop, and that bothers me....

You could in principle adapt it to the general case. Something along:

class UiGuard : public QObject
{
public:
    UiGuard(QWidget * ui)
        : widget(ui)
    {
        widget->grabMouse();
        widget->grabKeyboard();
        widget->installEventFilter(this);
    }

    UiGuard::~UiGuard() override
    {
        widget->releaseKeyboard();
        widget->releaseMouse();
    }

    bool eventFilter(QObject *, QEvent * event) override
    {
        return dynamic_cast<QInputEvent *>(event);  // Eat up the input events
    }

private:
    QWidget * widget;
};

And then you use in the places you have threads and/or event loops spinning, by just creating an instance:

// ...
{
    UiGuard guard(this);  // Prevent input events while in this method
    ReportProducer rp;
    rp.doWhateverToProduceHtmlAndPdf()
}

JonB

@kshegunov
Yes indeed, but as we said:

And then you use in the places you have threads and/or event loops spinning, by just creating an instance:

In principle I have no idea where code might have "threads and/or event loops". (Seriously, it took me days & debugging to even discover this particular button invoked code which did that, I had no idea it did! Which is my point.) So in effect I'd like this around every single UI slot handler (or probably 99% of them but not 1%), and I'm not prepared to edit thousands of places to do that! :)

kshegunov

@JonB said in GUI event blocking:

So in effect I'd like this around every single UI slot handler, and I'm not prepared to edit thousands of places to do that!

That might be rather bad idea, as event filters do hurt performance, even though in this case it's only a single dynamic_cast. You're trying to fight against the way UIs work - i.e. event driven. You could install that kind of class directly on the QCoreApplication object (with the proper modification), where you'd filter the events globally, but seems rather dubious decision.

JonB

@kshegunov

You could install that kind of class directly on the QCoreApplication object (with the proper modification), where you'd filter the events globally, but seems rather dubious decision.

That indeed sounds more like it, because I could do that in my own object which tests a flag to cause this behaviour, and where the flag would be set by the called code which knows it needs it and not the calling code which does not know that. This could be precisely what I am looking for....

You might look to look at my new https://forum.qt.io/topic/88481/is-it-possible-to-intercept-a-modal-dialog-s-event-loop now, where I'm asking just that about application versus dialog event loops....

kshegunov

@JonB said in GUI event blocking:

You might look to look at my new

Will do, but later, as I have some work to do now. :)

JonB

@kshegunov
Of course you must, and get back to fixing those nuclear reactors/missiles you work on :) Thanks so much for your comments!

ambershark

@JonB Another thing you could try is to disable all the controls you don't want being activated during processing. Then enable them when processing is done.

This would be better handled via a modal dialog with a progress bar (even an indefinite one) and a cancel button.

But the other option is just setting QWidget::setEnabled(false). People using the app may be confused why things were disabled all of a sudden though so I still think the modal progress dialog would be the best way.

But you could keep a list of QActions and/or QWidgets you want disabled during that call and just iterate through the list and setEnabled(false) until complete then iterate through again and enable them.

Just another suggestion that may work for you. You have lots of choices on how to handle this and I don't have any great advice on the "best" way so just throwing things out as I think of them. :)

JonB

@ambershark
Hi amber,

The application has 100 "top-level" dialogs, with an average of 3 buttons each (without even including other input controls which cause code to run). That's 300 potential "callers", which might lead to "re-entrancy".

Meanwhile, it has 3 "bottom-level" functions which cause a thread/event loop to be spun, which, if hit, will allow fatal "re-entrancy" in the ultimate, top-level dialog (i.e. allow another a button to be pressed while still servicing the original button press).

To use setEnabled(false), or keep a stack of callers, or any solution aimed at the caller, requires me to put code in 100, or 300, places. And furthermore to maintain that as further dialogs/buttons are added in future. To deal with it by causing a block, somehow, in the offending callee requires code in 3 places.

Which would you rather do?

That's why I started new thread https://forum.qt.io/topic/88481/is-it-possible-to-intercept-a-modal-dialog-s-event-loop ...

ambershark

@JonB Lol yea that sounds nasty... It sounds like it needs a massive rewrite of it's GUI. These things are not normal issues for normal guis.

Having 100s of dialogs is already pretty suspicious by itself. I've written some very large GUIs (million+ lines of code) and those never even came close to 100s of anything, much less of the rarely used (in modern guis) dialogs.

It sucks you got stuck supporting that. It will not be a fun job, lol.