[Xpilot-hacks] XPilot 5.0.0 (XPilot++) Development

Kristian Söderblom ksoderbl at niksula.hut.fi
Sun Jan 18 13:09:17 AST 2004


Hi.
	I describe in this mail some of my views. However, as you may
know, I don't speak for the whole ng team, some other members may have
other views of this situation.

On Fri, 16 Jan 2004, Jarrod Miller wrote:

> Well, here's the thing...As of right now, the client hasn't visually
> been changed much at all,

As you may know, the ng client has had some visual changes. One is the
addition of the custom server textures and full color mode. Then there is
the addition of a lot of client color options and other functionality,
including better support for high fps. Additionally, there is the work on
an SDL/OpenGL client, which is still a work in progress, but is maturing
quite fast and works very well in real playing (assuming you have good
opengl support).

> although Dik removed the MFC parts I believe. There's been some work to
> incorporate the INI tools
> dik wrote up for xp++, but other than that I don't think he's changed
> much. I know for a fact that I haven't
> changed any of it at ALL, although it is up to the latest 4.5beta

One thing I've done is that in the next version of ng, extraversion X3,
the Xresource stuff is removed. I've rewritten the client option code
in ANSI C. This code also gives some advantages over the old code,
because each part of the code can declare own options, all the options
don't have to be in one array in the code. Also the options are
initialized in a type safe way. Also configure.c can be written so that no
changes are needed to it when new options are added/old ones removed.
This code also easily supports a client set command for client options.
There is a few bug fixes and missing features todo in the new option
code, but I think it's ok.

Another idea in this code is that the Unix xpilotrc file  format could be
used also on windows. We've tested the SDL windows client and seen this
code works on windows. Now I'm asking, would it not be a good idea that
the configuration file would be movable between the different systems?
What is the benefit of having this INI file format on windows, except
maybe that windows people are used to it?

> I am of course very aware of whats you guys have been doing, and I think
> the feeling pretty much is that we
> DON'T want to lose your work. Just let me be totally clear on that fact.
> There's a LOT the "ng" project brings
> to XPilot, and I want to see it all brought in. Which,  of course leads
> us to the difficulty of how to do so.

I've thought a bit about merging ng and xp++. One problem is that the ng
polygon protocol is not complete yet. Another is of course that there's
been changes in both xp++ and ng, compared to e.g. 4.5.4. I won't make
this mail longer by going trough the technical details of how I would
like this merge to be done, assuming it was to be done. I could describe
this in another mail if anyone is interested. However one precondition
for the merge is that both teams agree this should be done. Unless this
is possible, both projects could stay separate. In this case the ng team
would probably just continue working on ng, possibly supplying patches to
xp++ to move over fixes/features.

> What I would like to see happen is this. XPilot++ goes in as 5.0.0. Our
> server is fully backward compatible, see,
> so it's not going to break old clients at that point. And once it is in
> fact in and released as 5.0.0, we start integrating
> all the new code from "ng", and when thats done, we roll with 6.0.0.

This sounds ok to me. My work on ng has always been based on the
assumption that the ng code should be merged into the official version.
Since the code would be moving from ng to xp++, Bucko would have to agree
this is a good idea. I believe Uoti's view was that moving to C++ would
not be beneficial. I don't know if this is a problem, since we can
continue to develop the polygon protocol using a C code base, moving it
over to xp++ (and C++) when we think it is good enough.

> And this "ng" merge should be very high priority. I will throw all
> I can into writing a mapeditor for ng format maps.

Juha Lindström (Baron) has worked a lot on a Java map xp2 editor, maybe
you could take a look at that.

--
kps



More information about the Xpilot-hacks mailing list