Ramon van Handel wrote:
> The problem with using X is that it is not portable to other
> platforms...
> We already thought of the solution: use SDL in stead of X.
> Base your primitives on SDL, which is itself portable.
> Why switch back to X now?
Drew Northup wrote:
> I agree......., X really isn't that portable in a non-network environment
> (like that inside of an IA-32 AT PC). Also, all of the X handlers out there
> (for things like WinNT) cost big $$ these days (unless you don't mind your
> system dying every 2 hours.........), and I thought that was something that
> we were trying to avoid.
Well, first I think people got me wrong. My point is that our virtual
graphics card should spit out a _generic_ set of primitives, for which
there are GUI _specific_ methods. One set of specific methods for
SDL, one for X, one for Win32, etc.
Architecting our video handling to expect, assume, and rely on SDL
is a broken idea. SDL looks quite cool, and I think using it might
make a good primary use target. Not all users will have it installed,
so we need to have X specific methods too. Since our host port is
only Linux/X now, the most sensible thing to do is at least have
a set of X methods. SDL methods could be designed in parallel.
One of the design methodology principles we should follow with plex86
(the interstitial commentary in _Open Source Development with CVS_ is
quite enlightening on this kind of stuff) is that we should make
regular progress and periodically work toward usable software releases.
Otherwise we keep jamming new stuff in plex86, without ever making
useful software for the end user.
Anyways, why don't I rephrase this and say, let's write X and SDL
methods in parallel. But we at least ought to have X. I'm not
keen on having plex86 _dependent_ on X, but I think it's an excellent
idea to have SDL so we can make the video processing kick-ass.
That said, some further thoughts from my neural net. It is
not necessarily the case that all graphics primitives protocol
has to be piped back to the host in the future. If you have
video memory mapped in via a direct video interface, we could
direct some processing to the local monitor or even guest
space. This way we can have our primitives cake and eat
it too. :^) And get working on the interface using host
space for now...
-Kevin
--
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
Kevin Lawton [EMAIL PROTECTED]
MandrakeSoft, Inc. Plex86 developer
http://www.linux-mandrake.com/ http://www.plex86.org/
