At 19:07 on 11/28/2000 +0100, Ramon van Handel <[EMAIL PROTECTED]> wrote:
> Well, the basic idea is that SDL will have to have abstracted
> the interface already in order to be so portable, so...
> let these guys do the hard work for us! we're not graphics
> experts, we do VMs. Let them do the graphics abstraction sort
> of stuff, and we use it. SDL itself is already an abstraction
> of the graphics interface.
I don't think there is anything wrong with using the SDL api's as
inspiration for the plex86 guest <=> host graphics protocol, but all the
same there is no reason to use SDL's API exactly. We're going to have
special needs, without a doubt. For example we're going to need to
provide API calls which map well to DirectX's notion of hardware
accelleration. Not knowing anything about directx, i'm going to guess
that these won't be exactly like what SDL would provide.
By all means, let's create a protocol which is similar to SDL if its API seems
appropriate- As you said, they've done the hard work of designing a decent
abstraction, and it will make the host-side GUI bindings easy for us :)
But let's not limit ourselves by using "SDL" as the protocol between the
guest and host. I don't think it's practical or advisable to tie
ourselves so fundamentally to a separate library which has drastically
different design goals than plex86.
--Josh