Ryzom

sfb wrote:

setCapture still needs to be implemented.

I have begun to implement it. I have strange things with the window position. I go on it asap.

kishan wrote:

The setCapture() now works fine. The XGetWindowAttributes does not seem to give the window position correctly I did not understood why. Anyway, I retrieved it from the "Configure" events (move/resize window) to know how to move the mouse when it goes outside of the window.

Why don't you use XGrabPointer() for pointer confinement?

Made a small test program which relies on XGrabPointer (see attachments for set_capture.c, compile with gcc -lX11 set_capture.c). Enter-key changes between grabbed and ungrabbed pointer state, Escape (or click on window close button / "(X)") exits the program). Also note, that the example application doesn't hide the standard cursor nor implements a user defined cursor, which may or may not affect the outcome.

I have to test it on more platforms, but so far XGrabPointer works fine. The used window manager has an effect on the outcome, i.e. with a window size of 640x480, fvwm2 gave coordinates from (0,0) to (639,479), while awesome returned (-1,-1) to (640,480). Tests with other window managers are open, still, but I guess they will give similar results. Tests with multiscreen etc. are left open, too. A non-working XGrabPointer is most likely a window manager issue, and should be reported as such; I don't see any reason to implement the setCapture() method any more complicated than necessary, unless it doesn't work properly with a hidden/"soft-cursor" replacement.