[vworld-tech] Middleware Reviews/Comparisons

Martin Piper MartinP at argonaut.com
Fri Jun 11 04:11:40 PDT 2004

For a games point of view I'm quite happy to see a departure from the
methods used several years ago. Back then the most common methods used were
client-server with the game code responsible for serialising data. Looking
back I can see why because a lot of games developers were still using C and
some had just stopped using assembly. Come to think of it, there are some
developers that still use C over C++. While not impossible with C to
abstract a nice data driven distributed object system, Sony SCE-RT DME(1)
springs to mind, the job becomes easier with C++ classes.

It could be said that the introduction of consoles with working C++
compilers helped is some small way to allow the games industry to mature and
use object orientated methods. Speaking from persona experience while
developing PS1 games the conscious decision was made to not use C++ because
at the time the compiler was judged to be not quite ready. Game code has
also increased in complexity driven by the need to deliver more interactive
worlds. At Argonaut the average age of the company has increased over the
last ten years, this must also have some kind of impact on the design of the
code produced. Anyway, I digress and I'll go wandering down memory lane if
left unchecked.

However back to the present where nearly every project, in this office here
at least, uses C++ or other similar object concepts.
It may come as no surprise to some that distributed object systems for games
probably started off in the military simulation field(2). These types of
simulation probably laid the groundwork for the MMO games we see these days.

So, what are the advantages of using a high level distributed object system?
For me the biggest advantage is that such systems are, by in large, able to
be plugged straight in to existing single player code and produce a quick
working multi-player demo. This can usually be accomplished with a minimal
amount of existing code by inserting extra C++ classes in to the existing
game class hierarchy. Imagine trying to do the same if the project is
largely C based and the game entities have a looser concept of data that
needs propagation. Creating quick a proof of concept is useful when
demonstrating to producers/managers.

I think I've gone on for long enough for now, I've got work to do. :)

(2) http://www.simulation.com/training/Military_Sim/military_sim.html

-----Original Message-----
From: Jim Purbrick [mailto:Jpurbrick at climaxgroup.com]
Sent: Thursday, 10 June, 2004 11:19 AM
To: 'vworld-tech'
Subject: [vworld-tech] Middleware Reviews/Comparisons

<takes deep breath before raising head above parapet again/>

I'd really like to hear more too. In fact I'd like to hear more about all
the current middleware options and how they differ in terms of philosophy
and technology. I went through most of the brochures a couple of years ago,
but I didn't come across ReplicaNet and presumably there are others that
were either under my radar at the time or have either appeared, changed
substantially or vanished in the mean time. With Martin and Adam on the list
we should be able to get some good insight in to the current state of the

> -----Original Message-----
> From: Bruce Mitchener [mailto:bruce at cubik.org]
> Sent: 04 June 2004 16:16
> To: vworld-tech
> Subject: Re: [vworld-tech] Combining networking modes?
> Martin Piper wrote:
> > I'll refrain from going in to too much depth however, since 
> it might be seen
> > as being a bit off topic. :) (I can talk for hours on this topic.)
> I certainly don't mind hearing more.
>   - Bruce
vworld-tech mailing list
vworld-tech at cubik.org

More information about the vworld-tech mailing list