On Tue, 21 Jan 2003, Fabio Alemagna wrote:
[...]
> > You miss, that in X, you never talk to the graphics HW directly. X will
> > send you an event when you iconify, and will nicely gobble up all
> > your funny drawing requests if you don't act on it.
> 
> Not completely true: if backing store is active the rendering will still
> happen on a backbuffer.

Yes. But with X, you must go through X for all graphic operations you do;
so the X server can easily manage that feature. For example, X may delay
updating the backbuffer until it is needed.

[...]
> Yes it would. The world is full of examples where it works. Even in AROS
> it works, so I don't see why it shouldn't work in KGI.

Even if I fully agree with Andreas (and probably most of the computer
graphics community) on the technical viewpoint wrt to automatic
backbuffer-ring of applications; I do think too (like you) that such a
feature should be available. Why? Because there are many people that
expect it. Whether I think they are wrong in this expectation is not
relevant if it makes more people happy/interesting with/in GGI/KGI.

 However, due to the difficulty of doing this correctly at the kernel
level (because of all the things Andreas outlined: userspace/kernel
synchronization, graphic context preservation, VRAM protection and
"swapping"), I really think such a behavior should be provided by libggi.
At the libggi level, we are pretty near from the X level, the library
controls most of the accesses to the framebuffer or accel and certainly
can handle a signal itself; so it could provide for most applications a
good backbuffer. (It may imply to prohibit access to the DirectBuffer if
such backbuffering is done, but most the time, only simple apps want to
have a permanent backbuffer.)

Rodolphe


Reply via email to