>> >> Don't you need to also send KEY_UP messages ? > No, the clients (tested on windows and fedora) respond to the key down event. > The next keyboard event contain modifier flags state(Shift, Ctrl, Alt etc.) in it and the html client will update the modifier state. > Amos. I am concerned about this patch. If this code is used to send ctrl-alt-delete to an environment that ignores it, it seems as though it would leave the receiving end in the wrong state. The server thinks ctrl and alt are pressed, so if I type 'c' on the keyboard, the server is going to think it should act on ctrl-alt-c. And what if I'm ignoring ctrl-alt-del, and I have repeat on key press configured in my window manager; now I'm going to flood an infinity of ctrl-alt-del into the target window manager. For ctrl and alt, we have the current state, so it should be possible to preserve those. I can live with the state of del being lost, but if we're going to 'corrupt' the del key, the end state of del should be up, not down. Cheers, Jeremy _______________________________________________ Spice-devel mailing list Spice-devel@xxxxxxxxxxxxxxxxxxxxx http://lists.freedesktop.org/mailman/listinfo/spice-devel