[Development] [I/O, Core] Defer a emitting of signal. It is possible?

Denis Shienkov denis.shienkov at gmail.com
Mon Feb 10 08:45:31 CET 2014


>That's a wrong conclusion. Our direct implementations of select() (like in
>QProcess and QAbstractSocket) do not use QSocketNotifier. They bypass the
>notifiers and go straight to select.

Yes, they bypass in case of waitForXX() methods.
But I'm talk about QSocketNotifier which also is used in QAbstractSocket as
ReadNotifier/WriteNotifier/ExceptionNotifier and so on to watch the (fd)
events.

> What do you mean by Dispatcher? The classes derived from
> QAbstractEventDispatcher?

I mean a pair: QEventDispatcherUNIX && QSocketNotifier.

> Anyway, what did your profiling reveal? Where are the hotspots in the
> application?

No, still not yet.

> Yes, profile it and let us know where the exact problem is.

Ok, thanks.  I will try do it as will have a free time.

Best regards,
Denis





2014-02-09 23:40 GMT+04:00 Thiago Macieira <thiago.macieira at intel.com>:

> Em dom 09 fev 2014, às 23:26:29, Denis Shienkov escreveu:
> > It seems that select() which implicitly uses QSocketNotifier/Dispatcher
> > is a bottleneck. But it is very strange
> > because the serial port is very slow device on the relation with others.
> > Therefore in the theory the I/O with
> > the serial port shan't influence overall performance and the CPU loading.
>
> That's a wrong conclusion. Our direct implementations of select() (like in
> QProcess and QAbstractSocket) do not use QSocketNotifier. They bypass the
> notifiers and go straight to select.
>
> Unless you meant "used by" instead of "uses".
>
> What do you mean by Dispatcher? The classes derived from
> QAbstractEventDispatcher?
>
> Anyway, what did your profiling reveal? Where are the hotspots in the
> application?
>
> > So, I thought that the possible workaround is to use a "deferred"
> > signals. E.g. the event from an defice
>
> Why are you talking about workarounds? First we investigate the actual
> problem. Only if we are unable to fix the problem should we think of a
> workaround. We're not there yet.
>
> > Or maybe, there are other solutions of a problem? Maybe someone can
> > advise something?
>
> Yes, profile it and let us know where the exact problem is.
>
> --
> Thiago Macieira - thiago.macieira (AT) intel.com
>   Software Architect - Intel Open Source Technology Center
> _______________________________________________
> Development mailing list
> Development at qt-project.org
> http://lists.qt-project.org/mailman/listinfo/development
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.qt-project.org/pipermail/development/attachments/20140210/aab3ee54/attachment.html>


More information about the Development mailing list