[Massiv-devel] Re: [massiv-dev] Fwd: implementing object system with CORBA

[Derek <pc...@gm...>]

Hi, Marek:

I see it. Thank you for your comments.

Massiv core features seems promising. What's its status now? Is it
still under development by others than David Li?

Regards
Derek

On 8/22/05, Marek Vondrak <ma...@us...> wrote:

> -------------------------------------------------------------------------=
---
> ----
>=20
> Hello.
>=20
> The primary reason is that it was our school assignment to write such a
> system. The other reasons are that we do not find CORBA suitable for the
> development of massively multiplayer games. If you know CORBA I believe y=
ou
> will agree with the following arguments. Basically, what a MMO developer
> wants, is not a RPC only system like CORBA or DCOM, but requires at least
> some support for object state serialization and replication, so that the
> simulation could be presented to clients. Support for consistent simulati=
on
> archivation is required on the server side. Not any of these requirements
> can be fulfilled with basic CORBA, because ORB cores do know nothing abou=
t
> CORBA object structures but their interfaces (except for so called CORBA
> value types, in other words ORBs can not serialize objects, can not acces=
s
> their attributes, etc. but can only call object methods through published
> interfaces). Also since MMO game components are usually written in a sing=
le
> language (mostly C++), it is highly encouraged to use that middleware tha=
t
> integrates with that particular language as much as possible so that
> programmers are not forced to write akward (unnatural) code. CORBA, on th=
e
> other side, is a multilanguage middleware and is "mapped" to multiple
> languages (OMG standard defines how CORBA objects described by IDL have t=
o
> be coded in that particular language). I reckon you would agree with me t=
hat
> CORBA mapping to C++ is _horrible_ and unusable.
> But enough. If you are still not convinced, try comparing the list of
> features supported by CORBA and Massiv side by side and try to answer to =
the
> following questions "Do I really need to use this CORBA feature in my MMO
> game?", "Is it useful to have Massiv core do this thing for me
> automatically?", "Can this Massiv feature be ever supported by ORB core o=
r
> do I have to implement it myself on the application level?". Maybe you
> should look at the sources of the Demo (the "managed" part, say
> src/demo/lib/server) to see how a MMO game can be written in Massiv and w=
hat
> (little) differences there would be if the game was written as a single
> stand-alone (not distributed) application.
> However to be faithful I have to say that most people failed building the
> demo application if anything or were discouraged from using Massiv due to
> its complexity. As it is with all middlewares I have seen, one has to get
> acquainted with the guts of the middleware in order to write even the
> tiniest "hello world" application on top of it. Usually, the initializati=
on
> part is the most complex one and requires the knowledge of the whole syst=
em.
> However I still believe that Massiv might be useful for those people how =
are
> fluent in modern C++, are qualified enough to write a simple network-base=
d
> game without the use of a network middleware and can understand how the
> principles used in their network game are abstracted in Massiv.
>=20
> -- Markoid
>=20
>=20
>=20
>