[haiku-appserver] Re: update code.

  • From: Adi Oanca <adioanca@xxxxxxxxxxxxxx>
  • To: haiku-appserver@xxxxxxxxxxxxx
  • Date: Thu, 31 Mar 2005 14:47:07 +0300

OKay, thanks Rudolf!

Guess we should stick to the original plan. :-)


Rudolf wrote:
> Adi,
>>>Please remember that you cannot do this: the current driver arch 
>>>NOT support you blitting outside the screen's area (framebuffer or 
>>>whatever). So you cannot blit in 'a part of free video memory'!
>>      A virtual screen like the one Allegro uses.
>>      Why can't we use that?
> In theory you could, but I'd consider that a tweak(!).
> The is a problem: You would need to make sure you find a way to 
> distinguish between 'user virtual screens', and 'offlimits 
> virtualscreens.
> Picture this:
> ->1. my monitor is set to 640x480 resolution.
> ->2. I set a virtual screen to 1024x768 resolution. 
> (Note that THIS is the intended use of virtualscreens. If I move the 
> mouse beyond the part visible on my monitor, the driver automatically 
> takes care of moving my 'viewport' to keep the mouse visible. The 
> driver also automatically blocks the move to go outside this 
> virtualscreen. Actual use of this is for instance on laptops (I use it 
> in the Neomagic driver). I also use this a lot when testing stuff. BTW, 
> I don't understand why the ScreenPrefs app does not allow you to set 
> such a workspace!. I created my own app accessing BScreen to set these 
> for me.)
> OK, now for completing what I want you to picture:
> ->3. Now on top of 1. and 2. you create an even larger virtual space 
> for your 'backbuffering' schemes (or whatever you'd like to call it). 
> You now need to prevent the moving mouse to scroll/pan to this part of 
> the virtualscreen. THIS is the problem. There's no way to do that 
> without altering the way an accelerant works, and creating an  
> interface for that to the app_server someway.
> ->Or, we would need to establish now that we want virtualscreens to be 
> used ONLY for your backbuffering: hence the 'intended' use of 
> virtualscreens, is now a HACK. In other words: we redefine what's 
> normal and what not.
> We _could_ do that I guess: this requires no chance in the driver. IT 
> would even work as normal on R5 and such (doing old 'intended' 
> virtualscreens), and on Haiku (doing new 'intended' virtualscreens).
> You'd better modify BSCreen however: user access to create 
> virtualscreens should be blocked now. 
> My thoughts:
> I'd say doing this is offlimits for R1.
> Rudolf.

Other related posts: