Set resizable window on Windows

[dariosortino]

At the moment is impossible to resize the visualizer window in Windows OS.

With @S-Dafarra in https://github.com/robotology/human-dynamics-estimation/pull/341/files , we added the possibility to set the height and width of the window in a configuration file as a workaround to this, but it's not convenient if we keep switching between different screens with different resolutions.

cc @traversaro

[S-Dafarra]

I have tried adding setResizable(true) but as soon as the visualizer starts, the window shrinks and the visualization is messed up.

Investigating

[traversaro]

I have tried adding setResizable(true) but as soon as the visualizer starts, the window shrinks and the visualization is messed up.

Investigating

For testing you could try also to temporary revert #976 (by commenting the line https://github.com/robotology/idyntree/blob/v8.1.0/src/visualization/src/Visualizer.cpp#L283) to see if without using SDL the resize on Windows works. If that works, we can think about exposing that so that we can opt-out of using SDL on Windows, as anyhow the problem reported in #975 is not a problem if HumanStateVisualizer is a process just devoted to visualization .

[S-Dafarra]

I tried it, but it is still not usable. I think it may be a bug on the Windows implementation of irrlicht. I also tried following a bit https://irrlicht.sourceforge.io/forum/viewtopic.php?t=43316&sid=34c5d329ed5118c383d18231a046b4e9, but I had no luck. I am not able to change the size of the window from code.

[S-Dafarra]

I managed to make it working. Instead of removing that line, I had to change it to

irrDevParams.DeviceType = irr::EIDT_WIN32;

[S-Dafarra]

I tried it, but it is still not usable. I think it may be a bug on the Windows implementation of irrlicht. I also tried following a bit https://irrlicht.sourceforge.io/forum/viewtopic.php?t=43316&sid=34c5d329ed5118c383d18231a046b4e9, but I had no luck. I am not able to change the size of the window from code.

Actually, I did a mistake before, and I commented out the wrong lines. Indeed, the suggestion of @traversaro was correct. Now we can decide if we want to resize or close the window 😂

[traversaro]

Cool! While we thing how to do this properly @dariosortino I think you can do the local modification suggested in #1070 (comment) and tested by @S-Dafarra in your system/the system used for the demo, so that for future them you can already resize the window, right?

[S-Dafarra]

While we thing how to do this properly

What about creating a window on our own and have irrlicht printing on it? In this way we might be able to deal with the closure more easily. I have been using glfw in the past and it might also be a quite portable solution. @traversaro what do you think?

[S-Dafarra]

While we thing how to do this properly

What about creating a window on our own and have irrlicht printing on it? In this way we might be able to deal with the closure more easily. I have been using glfw in the past and it might also be a quite portable solution. @traversaro what do you think?

I have quickly tried, but no inputs are received by the engine. I probably need to pass custom events via postEventFromUser

[S-Dafarra]

I have started having some results with the custom window, without the mouse inputs. Here the commit https://github.com/S-Dafarra/idyntree/tree/test_visualization

Funny story. If I run the test where I use two visualizer, the entire graphical system of my PC crashes 😅

[S-Dafarra]

Funny story. If I run the test where I use two visualizer, the entire graphical system of my PC crashes

Actually, I was testing an old version in my PC cause I forgot to install 😓 It seems to be working now after removing this line https://github.com/S-Dafarra/idyntree/blob/212fdb017a6922404b7af3362c1e81f375530ae1/src/visualization/src/Visualizer.cpp#L830

[S-Dafarra]

With S-Dafarra@2ed1e97 I managed to have mouse inputs working. Hence, I should have recovered all the features. I need to test it on Windows