Re: [AD] Using system mouse cursor

[Chris <kcat@xxxxxxxxxx>]

Evert Glebbeek wrote:
> > I've used the mickeys function directly and I've tried just 
> > repositioning the mouse and getting the difference myself. They both 
> > have stuttering problems, though the mickeys function moreso.
>
> Weird... I don't understand this problem at all really... I'll see if I can 
> reproduce it when I use some of my own programs a bit more (not just the 
> tests provided by Allegro).

Well.. one think I suppose I should've mentioned, I'm polling a Vorbis 
stream in a timer, too. :x Don't totally freak out though, I'm making 
sure that Allegro's multithreading before even attempting that. AFAIK, 
X's input thread is seperate from the timer thread, so I don't see that 
really being a problem anyway.

> What I'd propose is the following: disable the mouse warping hack in X11 
> windowed mode (so that it behaves as Windows does), but have a default 
> warping mode in fullscreen mode. In this case, Allegro is responsible for 
> drawing the mouse cursor, otherwise we can use the system cursor.
> Add a function that allows the user to override this behavior explicitly 
> say, 
> int set_mouse_mode(int mode)
> where mode can be any of (say) MOUSE_WARP, MOUSE_NORMAL, MOUSE_AUTO. 
> MOUSE_WARP enables the mouse warping, MOUSE_NORMAL disables it, MOUSE_AUTO 
> switches to whatever is the default (non-warped for X11 Windowed, warped 
> for X11 fullscreen and non-warped for Windows in any case. Not sure how 
> the MacOS X port handles these things). Returns 0 on succes, non-zero on 
> failure (in Windows, it wouldn't do anything with the current design, or 
> rather, it wouldn't be nescessary).
> We can add a flag to gfx_capabilities to tell the user wether or not mouse 
> mickeys requires warping the cursor (although this may be abuse of the 
> gfx_capabilities variable).

Sounds quite complex. I'm trying to get away from any kind of mouse 
warping though.. I would like the OS's cursor position to be the same as 
Allegro's cursor position as much as possible. AFAIK, this is 100% 
possible.. set_mouse_position calls XWarpMouse (or whatever it is) and 
offsets Allegro's internal position trackers so the next mouse movement 
event that re-positions the mouse is ignored (I thought I read the 
XWarpMouse causes X to send a mouse movement event), but otherwise the 
internal variables are updated as given by X.

Now, this obviously breaks Allegro programs.. get_mouse_mickeys is 
gauranteed (by default) unlimited mouse movement. So what I propose is 
this. Take a function like I've described, but extend it like so:

int enable_soft_mouse(int enable);
   Enables/disables the software mouse cursor. When enabled, this 
causes Allegro to always trap the OS mouse position to the center of the 
program area, but still update mouse_x/mouse_y to cover the mouse range 
and provide unlimited mouse mickeys while using the timer to draw the 
mouse cursor. Turning it off will disable get_mouse_mickeys, but untrap 
the mouse and possibly allow Allegro to have the OS handle drawing the 
cursor. The mouse_* vars will still be updated as expected. Default is 
/enabled/ to maintain compatibility with existing programs.

So, this is sort of like a mouse warp hack, but it's not really a hack 
anymore since it would be a clean implemention at API-level.

- Kitty Cat