Hello Everyone,
I am currently experimenting a little with Genode 14.08. When playing around in Qt5, I encountered some problems when trying to change the geometry of a QMainWindow. I have an additional Genode::Thread running, which can trigger a geometry-change of the application's main window by methods provided by the MainWindow.
When a change of the window's geometry is triggered, it seems as if the window-decorations are properly resized, but the QtWindow underneath is kept at its original size. Additionally, the thread which triggers the geometry-change is blocked (this holds true for the Genode::Thread, as well as for a QThread).
Resizing the window with the mouse works properly.
I took a look at the Nitpicker Qt-Plugin, where I noticed two things: 1. - There is some unused code in qnitpickerplatformwindow.cpp in QNitpickerPlatformWindow::handle_events(). Until now I did not dig into the Window_slave_policy, but the inactive code segment looks to me, as if it should change the QtWindow's geometry.
2. - The Processing seems to stop somewhere before returning the framebuffer of the the PlatformWindow when detaching / attaching it from / to the rm_ssession.
Since I'm not quite familiar with all the internal Nitpicker / Qt stuff, it would be great if anyone with a little more experience could provide some help. ;-) - Or am I even trying something which is not intended?
Cheers Marcus
Hello Marcus,
could you please try with the following patch applied?
https://github.com/cproc/genode/commit/d480771f9e9e8ede711c687ed49281bc40386...
The patch is primarily about resizing Qt windows via the window manager, but it modifies some of the places you mentioned.
If it still doesn't work, could you perhaps provide a test case which can be used for debugging?
Christian
On 09/15/2014 04:29 PM, Marcus Nolte wrote:
Hello Everyone,
I am currently experimenting a little with Genode 14.08. When playing around in Qt5, I encountered some problems when trying to change the geometry of a QMainWindow. I have an additional Genode::Thread running, which can trigger a geometry-change of the application's main window by methods provided by the MainWindow.
When a change of the window's geometry is triggered, it seems as if the window-decorations are properly resized, but the QtWindow underneath is kept at its original size. Additionally, the thread which triggers the geometry-change is blocked (this holds true for the Genode::Thread, as well as for a QThread).
Resizing the window with the mouse works properly.
I took a look at the Nitpicker Qt-Plugin, where I noticed two things:
- There is some unused code in qnitpickerplatformwindow.cpp in
QNitpickerPlatformWindow::handle_events(). Until now I did not dig into the Window_slave_policy, but the inactive code segment looks to me, as if it should change the QtWindow's geometry.
- The Processing seems to stop somewhere before returning the
framebuffer of the the PlatformWindow when detaching / attaching it from / to the rm_ssession.
Since I'm not quite familiar with all the internal Nitpicker / Qt stuff, it would be great if anyone with a little more experience could provide some help. ;-)
- Or am I even trying something which is not intended?
Cheers Marcus
Want excitement? Manually upgrade your production database. When you want reliability, choose Perforce Perforce version control. Predictably reliable. http://pubads.g.doubleclick.net/gampad/clk?id=157508191&iu=/4140/ostg.cl... _______________________________________________ genode-main mailing list genode-main@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/genode-main
Hey Christian,
Thanks for your reply.
I built a little test case and played around a little more. When I emit a QtSignal, which is connected to a slot where the resizing is done, everything works properly (with and without your provided patch). Still, with the patch, simply invoking a method which resizes the window fails.
Since I am definitely no expert on Qt, I can't really say if this is a genode-related problem or if I am using Qt improperly here.
If you feel, this might be a genode-related issue, I can, of course, send you the test case.
Cheers Marcus
Am 15.09.2014 18:23, schrieb Christian Prochaska:
Hello Marcus,
could you please try with the following patch applied?
https://github.com/cproc/genode/commit/d480771f9e9e8ede711c687ed49281bc40386...
The patch is primarily about resizing Qt windows via the window manager, but it modifies some of the places you mentioned.
If it still doesn't work, could you perhaps provide a test case which can be used for debugging?
Christian
On 09/15/2014 04:29 PM, Marcus Nolte wrote:
Hello Everyone,
I am currently experimenting a little with Genode 14.08. When playing around in Qt5, I encountered some problems when trying to change the geometry of a QMainWindow. I have an additional Genode::Thread running, which can trigger a geometry-change of the application's main window by methods provided by the MainWindow.
When a change of the window's geometry is triggered, it seems as if the window-decorations are properly resized, but the QtWindow underneath is kept at its original size. Additionally, the thread which triggers the geometry-change is blocked (this holds true for the Genode::Thread, as well as for a QThread).
Resizing the window with the mouse works properly.
I took a look at the Nitpicker Qt-Plugin, where I noticed two things:
- There is some unused code in qnitpickerplatformwindow.cpp in
QNitpickerPlatformWindow::handle_events(). Until now I did not dig into the Window_slave_policy, but the inactive code segment looks to me, as if it should change the QtWindow's geometry.
- The Processing seems to stop somewhere before returning the
framebuffer of the the PlatformWindow when detaching / attaching it from / to the rm_ssession.
Since I'm not quite familiar with all the internal Nitpicker / Qt stuff, it would be great if anyone with a little more experience could provide some help. ;-)
- Or am I even trying something which is not intended?
Cheers Marcus
Want excitement? Manually upgrade your production database. When you want reliability, choose Perforce Perforce version control. Predictably reliable. http://pubads.g.doubleclick.net/gampad/clk?id=157508191&iu=/4140/ostg.cl... _______________________________________________ genode-main mailing list genode-main@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/genode-main
Want excitement? Manually upgrade your production database. When you want reliability, choose Perforce Perforce version control. Predictably reliable. http://pubads.g.doubleclick.net/gampad/clk?id=157508191&iu=/4140/ostg.cl... _______________________________________________ genode-main mailing list genode-main@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/genode-main
Hello Marcus,
On 09/16/2014 10:27 AM, Marcus Nolte wrote:
Hey Christian,
Thanks for your reply.
I built a little test case and played around a little more. When I emit a QtSignal, which is connected to a slot where the resizing is done, everything works properly (with and without your provided patch). Still, with the patch, simply invoking a method which resizes the window fails.
Since I am definitely no expert on Qt, I can't really say if this is a genode-related problem or if I am using Qt improperly here.
I was not sure about that either, but found this interesting document, which states that GUI objects can only be used from the main thread:
http://qt-project.org/doc/qt-5/threads-qobject.html
Christian
Hi,
On Tue, Sep 16, 2014 at 12:43:16PM +0200, Christian Prochaska wrote:
I was not sure about that either, but found this interesting document, which states that GUI objects can only be used from the main thread:
Christian, you're dead right about accessing QWidgets: When programming Qt one should definitely stay away from manipulating GUI related QObjects from non-main threads. The best practice is to connect signals in QObjects belonging to other threads to slots in QWidgets. All attempts to circumvent this approach I know of failed dramatically.
Greets
Hi Christian(s),
Thanks for the link. I stumbled upon that one after some research yesterday, too. I guess that's another chapter in "lessons learned" ;-)
Cheers Marcus
Am 16.09.2014 12:43, schrieb Christian Prochaska:
Hello Marcus,
On 09/16/2014 10:27 AM, Marcus Nolte wrote:
Hey Christian,
Thanks for your reply.
I built a little test case and played around a little more. When I emit a QtSignal, which is connected to a slot where the resizing is done, everything works properly (with and without your provided patch). Still, with the patch, simply invoking a method which resizes the window fails.
Since I am definitely no expert on Qt, I can't really say if this is a genode-related problem or if I am using Qt improperly here.
I was not sure about that either, but found this interesting document, which states that GUI objects can only be used from the main thread:
http://qt-project.org/doc/qt-5/threads-qobject.html
Christian
Want excitement? Manually upgrade your production database. When you want reliability, choose Perforce. Perforce version control. Predictably reliable. http://pubads.g.doubleclick.net/gampad/clk?id=157508191&iu=/4140/ostg.cl... _______________________________________________ genode-main mailing list genode-main@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/genode-main
-
Christian Helmuth -
Christian Prochaska -
Marcus Nolte