On Tue, 28 Nov 2000 13:00:12 -0500, Kevin Lawton wrote:
>I have not heard a single good argument as to why we should lock
>our selves down to SDL.
I think the only issue is performance. The guest OS driver can either
speak one of two protocols: SDL, or something else. If it talks SDL,
there are three conversions: guest OS interface -> SDL -> Host OS
interface. If you add an additional layer (or look at SDL as being the
"host OS interface"), you actually have 4 conversions: guest OS
interface -> PlexDL -> SDL -> Host OS interface. In the first example,
you write one thing: a guest OS driver. The guest OS driver depends
on something that's already written: the SDL interface. With the
second option, you have to write two things: a guest OS driver, and a
PlexDL to host OS interface. This increases the amount of work.
To make things easier, you could start out by just writing a PlexDL to
SDL interface. This gives you the speed (of getting it done, not
performance) of using SDL, but without tying yourself to it. This
gives you the ability to use something other than SDL, of course,
including going directly to the host OS interface where SDL is not
available, for for better performance. Of course, it would probably be
just as easy to write an SDL environment for other host operating
systems as it would be to write a PlexDL environment, especially if
you're using a subset of the SDL toolkit...
One more thing: I don't see how requiring SDL is going to increase
resource requirements. SDL isn't really doing anything that PlexDL
won't have to do. It seems to me to be 6 of one, half-dozen of the
other... Unless you think you're going to write something more
efficient than SDL for each host platform on TOP of developing
Plex86...
Just my $0.01 (not enough intelligence for $0.02)...
Tim Massey
