[AD] New gfx api

[Angelo Mottola <a.mottola@xxxxxxxxxx>]

> Checked it out, and ported the changes I made to the graphics code to 
> it.
> The attached patch contains a good deal of the inversion to convert the
> old display API with a newer one (essentially, the one I posted 
> earlier).
> The underlying gfx_driver structure has not changed though, so this 
> does
> not add actual new functionality (well, transparent triple/double
> buffering) to the code, although it should be flexible enough to do 
> more
> than the current code does.

Checked out your patch, and looks cool, it can be applied for me; any 
needed fix can be applied later.
One thing: aren't displays supposed to be event sources?

I was thinking about this whole events stuff... Currently, you have to 
create at least one event queue and attach event sources to it. Then 
you regularly query this queue for events.
This requires some lines of code to initialize a program that properly 
gets input. Can't we have an internal global queue, on which all event 
sources are automatically attached when created? Then an user could 
just specify NULL as event queue and this global queue is used instead. 
What do you think?

--
Angelo Mottola
a.mottola@xxxxxxxxxx
http://www.ecplusplus.com