Re: [AD] OS X full screen resize (was Re: Allegro 5 TODOs (from wiki))

[Evert Glebbeek <eglebbk@xxxxxxxxxx>]

On 13-Dec-08, at 4:36 AM, Milan Mimica wrote:
> Uh? Doesn't ex_threads2 draw to the display from 6 different threads  
> and none of
> them created the display, and it still works?

That's what I thought it did too, but no. That's not what it does. All  
threads draw to their own memory bitmaps, which are drawn to the  
screen in the main loop.

> al_set_current_display() will make the context current to the thread  
> calling the
> function, using pretty much the same routines on all three platforms:
> glXMakeCurrent/wglMakeCurrent/makeCurrentContext

Hmm.
Ok, this is weird. I'm not quite sure what I was doing differently  
before, but trying it again now it does actually work. :-/
Odd. I guess that comment is outdated then and my claim that it didn't  
work due to some messup somewhere.

> That shouldn't matter really.

No, but it would if the context wasn't set as current in the thread  
that was transfering the bitmaps - which I guess is more or less what  
I was trying to say.
That also suggests that my fix to get destroy_display() to not crash  
on fullscreen displays under OS X isn't entirely correct and may only  
work if the display that is destroyed is set as "current" in the  
calling thread. Without a second display that's hard to test, but I  
can make the destroyed display current while the cleanup operations  
are performed and then switch it back to the previous state. In that  
case, the loop transferring the bitmaps can go back into  
destoryDisplay as well.

I'll have a look at it this afternoon.

Evert