[Development] QWindowSystemInterface::handleMouseEvent() relative or absolute?

Girish Ramakrishnan girish at forwardbias.in
Fri May 18 17:22:35 CEST 2012


Hi Laszlo,

On Tue, May 15, 2012 at 10:45 PM, Laszlo Agocs <laszlo.p.agocs at nokia.com> wrote:
>
> Hi,
>
> Because windowing systems will usually provide absolute coordinates in
> their pointer events, hiding all the cursor management, raw event
> translation, etc. hassle from the clients. Also the Qt events contain
> absolute coordinates and in the end a handleMouseEvent() call is not
> much more then generating a QMouseEvent.
>
> In your case there is no windowing system so it is up to the platform
> and generic plug-ins to implement all the needed logic. Like you
> noticed, if your system becomes complicated, having the loosely coupled
> 'platform plug-in plus a bunch of generic plug-ins' combination might
> not be feasible anymore because you are effectively implementing your
> own windowing system and thus need to synchronize and translate the
> incoming raw events in a centralized manner instead of just blindly
> pumping out events (via QWindowSystemInterface calls) independently from
> each generic plug-in.
>

Related: eglfs currently is just a "graphical" plugin and has no input
support out of the box. eglfs (from it's background as a reference for
qpa) has however become the gold standard for practically all devices
using Qt5 in the absence of wayland support. As a result, practically
everyone is annoyed by having to pass -plugin for input support :-) I
have proposed https://codereview.qt-project.org/#change,26414 to have
built-in input support (evdev based) in eglfs.

Girish



More information about the Development mailing list