Before SDL, the windows behaved differently than with SDL now, everyone is cursing at the new one.
Old behaviour:
The mouse is released from the center of the window when the console is opened, or when the application loses focus, through task switching (win key, alt+tab).
Current behaviour:
Simple task switching is impossible, because the mouse will hardly get unlocked from the window. Even if so, it will get kicked everytime it passes over the window. You can open the console for this, but it is annoying to do so, and it's blocking half of the window. But after getting back into the game, in both cases (using the console, or ALT+TABbing) a mouse movement hitch occurs.
Created attachment 1799[details]
adapt to task switching
With this patch, not only every time the console is opened, but also when the window focus is lost, the mouse appears at the center of the window. During switches, the mouse does not hitch.
I believe this makes the obscure in_nograb obsolete.
The patch should work for OSes other than the Windows XP I've tested this on.
I question the changes of r1341. It does combine a couple of mouse movement events into one, but the sequential UI calls tell it to "do something", and instead of "move here, click, move there, click", it tell it to "click and click, then move there".
Another thing, In Sys_Idle():
// If we have no input focus at all, sleep a bit
if( !( appState & ( SDL_APPMOUSEFOCUS | SDL_APPINPUTFOCUS ) ) )
{
Cvar_SetValue( "com_unfocused", 1 );
sleep += 16;
}
First, we should definitely not sleep if the mouse is outside of the window borders (SDL_APPMOUSEFOCUS), but the window itself has focus. Second, if the window is not minimized, should we even sleep at all? As long as the graphics are drawn, it not sleep. I mean I would like to do something else (like chat on ICQ) while spectating the game.
With this patch, things go very wrong for me.
Basically, mouse is over the window, so it gets focus from my WM, and then ioq3 steals the mouse causing my WM to defocus the window, causing ioq3 to release the mouse, and then the mouse is back over the window so it goes on a nice high speed focus/defocus loop until I open the console.
Going full screen and back makes everything work until I open the console and the mouse is released, and it all starts again.
Removing the lines 719-724 part of the patch makes everything go back to normal for me.
To reproduce you need a WM that's set to have focus follow the mouse, and that removes focus when the mouse is grabbed. I use wmii, which meets both of these criteria.
Created attachment 1799 [details] adapt to task switching With this patch, not only every time the console is opened, but also when the window focus is lost, the mouse appears at the center of the window. During switches, the mouse does not hitch. I believe this makes the obscure in_nograb obsolete. The patch should work for OSes other than the Windows XP I've tested this on.I question the changes of r1341. It does combine a couple of mouse movement events into one, but the sequential UI calls tell it to "do something", and instead of "move here, click, move there, click", it tell it to "click and click, then move there". Another thing, In Sys_Idle(): // If we have no input focus at all, sleep a bit if( !( appState & ( SDL_APPMOUSEFOCUS | SDL_APPINPUTFOCUS ) ) ) { Cvar_SetValue( "com_unfocused", 1 ); sleep += 16; } First, we should definitely not sleep if the mouse is outside of the window borders (SDL_APPMOUSEFOCUS), but the window itself has focus. Second, if the window is not minimized, should we even sleep at all? As long as the graphics are drawn, it not sleep. I mean I would like to do something else (like chat on ICQ) while spectating the game.