Re: [AD] Inconsistent behaviour with mouse coordinates in different windowed modes in Windows.

[Andrei Ellman <ae-a-alleg@xxxxxxxxxx>]

aj wrote:
> Andrei Ellman wrote:
> > Victor wrote:
> > > 1. What is the version of your Allegro?
> > 4.2.0 (20051015) (sometime after RC2 was released)

At last, I have upgraded to the final 4.2.0 release, and I can confirm
that this bug is still there.


> > > 2. What is the type of your Windows?
> > Windows 2000
> i hope you have the service packs.

Yes, I do.
I just tested it under Windows 98, and the same bug in DXWN is there too.


> I think you may have found a bug.

It looks like it. However, the mouse_mickey functions work fine on both
DXWN and GDIB when the mouse is outside.


> > In the GDIB mode, they are always the last values they were when the 
> > mouse was in the window. Is there a reason why they aren't in the DXWN 
> > mode, or have I found a bug? What about windowed modes on other 
> > platforms?
>
> not 100% sure.. but i would think the mouse_x and mouse_y co-ords should 
> always be the last (valid) values generated.
> regardless of which driver you use, that is to say all drivers should 
> act the same.

mouse_x and mouse_y being the last values of the mouse when it was in
the window is a good solution, but I think an even better one would be
to clamp the mouse_x and mouse_y values to the screen area. For example,
if the mouse moves off the left-edge, mouse_x will be 0 until it returns
back to the window. While the mouse is off the left-edge, mouse_y will
continue to be updated. This means that if the mouse is 'orbited' around
the window, then mouse_x and mouse_y should cycle through the
coordinates of the edges of the window.

But whatever solution is used, the behaviour of mouse_x and mouse_y when
the mouse is outside the window must be consistent on all platforms.


> > > It would be nice too if we had a
> > > function like "is_mouse_inside_window();".

I agree.

Also, if the values of mouse_x and mouse_y are clamped, there should be
an additional set of variables that give the actual position of the
mouse in the space of the allegro-window (to maintain compatibility,
these will be separate from mouse_x and mouse_y), so if the cursor is
100 pixels off the left of the screen, mouse_x should be 0, but
big_mouse_x (or whatever it will be called) will be -100. Either that,
or there should be a function that returns a vector which is the
difference between the current mouse-position, and {mouse_x, mouse_y}. 
Maybe instead of getting the mouse position relative to the window, 
there should be a platform independent means of getting the absolute 
mouse-position in the desktop.




> As to the   is_mouse_inside_window()   this can be done yourself with a 
> little bit of code.
> I can think of 99.9% of the code already in my head.
> If you really need it, i can assist you in writing it, (it would be 
> about 10 lines of code, and require 1 or 2 win32 functions).

Thanks, but I would prefer it if a platform independent version of
is_mouse_inside_window() was contributed to Allegro. In the meantime, I
can just work around the bug by testing to see if the values of mouse_x
and mouse_y are <0


AE.