Re: [AD] scare_mouse_area

["Vincent Penquerc'h" <lyrian@xxxxxxxxxx>]

On Sun, Dec 19, 1999 at 05:41:57PM +0000, Shawn Hargreaves wrote:
> > However, when I tested it, it sometimes leaves artifacts on the screen
> 
> I think that's because you don't freeze the mouse before checking 
> coordinates, so it can move in between when you decide that it's outside the 
> box, and when you freeze it.

Wow! That's only a few tens cycles at most if some branches predictions get
missed! Didn't think of this one...  Nice shot Shawn ;)

> I've also changed it to use a separate flag stack for remembering the freeze 
> states: it seems a bit hackish to put that in a magic pointer value :-) Also 

Err, yes :)
I wondered a bit, and then decided to add the less things possible ;) Since
NULL was also used.... But it's cleaner to add another stack *grin* :)

> having a dedicated place to store this allows it to respect the existing 
> freeze state, in case the user had already frozen the mouse before calling 
> this routine.

It did already, doesn't it ?? I incremented and decremented the flag, so
this shouldn't discard any user selected frozen state ? Or Did I miss
something ?

> I wonder if it would be worth making all the GUI objects use this? On the 
> one hand that would reduce flicker quite a bit, but on the other hand, not 
> all the standard objects draw entirely inside their rectangle (some have a 
> one pixel shadow, or grab handles, etc), and it seems very likely that other 
> widgets people have written might also be a bit lax about this, so I don't 
> know whether we can risk assuming everyone will stick exactly to their 
> stated widget sizes...

For my custom widgets, I always use subbitmaps, so all is automatically
clipped to the declared rectangle. This policy would be benefitial to Allegro's
widgets I think, if there was not this backward compatibility.
But I won't talk any more about the GUI things, I did it too much already ;)

-- 
Vincent Penquerc'h
Windows NT - New Trial