(not very) Massive Multiplayer Games

I've just been reading a book on MMPORGs (massive multiplayer online
roleplaying games) and it's piqued my interest. Despite never having
played one for more than a few minutes and being generally clueless on
networking and transactional issues, I now feel compelled to try writing
a toy one. Massive multiplayer Pong maybe :)

Since it's a fun project, I'll be doing it in Smalltalk, and Dolphin is
my favourite Smalltalk development environment, so here I am...

Is there any Dolphin code out there which already acts as a multiuser
server - not necessarily for gaming, but anything where heaps of people
connect over the same port and expect decent performance. It would be
even better if it used UDP, as that seems to be the favoured protocol
for these games.

For that matter, if there's any Squeak of VW code doing the same thing,
I'm sure it would be worth a read too. And if anyone has any favourite
websites discussing the design of these beasts (I read gamasutra.com,
but am otherwise pretty out of touch), I'd be delighted to get any
pointers to good sites.



                       thanks for any suggestions,



                                     Steve

Hi Steve,

> Is there any Dolphin code out there which already acts as a multiuser
> server - not necessarily for gaming, but anything where heaps of people
> connect over the same port and expect decent performance.

Not sure if it is the best approach, but a tuplespace would be one
approach.

John Small has a Tuplespace implementation as part of his
GenericServer. I can't find his website at the moment ... but I have an
old version ... email me if you want it and can't find it.

> For that matter, if there's any Squeak of VW code doing the same thing

Sounds like a good excuse to play with: http://www.opencroquet.org/

Have fun!
Steve

Would rST be of any help ?
If it is, then there is a port of rST for dolphin in
http://www.smalltalking.net/Goodies/Dolphin/index.htm

regards
martin

"Steve Taylor" <[hidden email]> escribió en el mensaje
news:VRh3g.15143$[hidden email]...

(mailed and posted)

Steve Alan Waring wrote:

> Not sure if it is the best approach, but a tuplespace would be one
> approach.

Which I'd never heard of before... Googling around, it looks like an
interesting thing to play with. I've only glanced at a few pages so far,
and I'm not clear on whether they're suited to client-server systems as
to cooperating servers - I'm contemplating a system with a central
server and realtively dumb and untrusted clients implemented in Flash.

> John Small has a Tuplespace implementation as part of his
> GenericServer. I can't find his website at the moment ... but I have an
> old version ... email me if you want it and can't find it.

Yes please. I had a look around the web a while ago and found a
dolphin/tuplespace which led me to a 404 page. This time I can't even
find that much. My Google-Fu is weak today.

> Sounds like a good excuse to play with: http://www.opencroquet.org/

I was looking at Croquet a while ago and initally thought it was just a
facny 3d interface - then I did a bit of a double take. There seems to
be quite a lot going on under the hood. Again, it seems to be looking
more at cooperating servers rather than a firmly client/server based
model. but maybe that's just my casual initial reading.

> Have fun!

It looks like an interesting can of worms!

> Steve


                    Steve

Martin Rubi wrote:
> Would rST be of any help ?
> If it is, then there is a port of rST for dolphin in
> http://www.smalltalking.net/Goodies/Dolphin/index.htm

I'd planned to do the server in Smalltalk and the clients in Flash, so
it wouldn't be immediately applicable - but it's likely to be a good
source of comms code, and worthwhile in itself if I ever want to think
about worlds running on different servers communicating and cooperating.

Thanks for the pointer!

> regards
> martin


                   Steve

> I was looking at Croquet a while ago and initally thought it was just a
> facny 3d interface - then I did a bit of a double take. There seems to
> be quite a lot going on under the hood. Again, it seems to be looking
> more at cooperating servers rather than a firmly client/server based
> model. but maybe that's just my casual initial reading.

That's not necessarily a bad thing.  A central server tends to bog down
under load, where Reed's versioned data has a chance to avoid that problem.

My main questions with Croquet are:

  (1) is Tweak's shotgun wedding of events and methods a good idea or a
mistaken attempt to make things easy on novices?  At first glance, it
looks like a needless comlexity disguised as a simplification.

  (2) What happens when one of us edits a class (add/remove instance
variables) both of us use in collaboration?  Given that (AFAIK) the
cooperating clients tend to compute locally, it might be ok.  But how
does the state get shared?  If it is via a serializer, the iv layouts
should be a problem.  If the system seemlessly shares code, there is an
obvious security concern.  If is also possible that they expect code to
be the same on both ends, with the users changing only data???

If nothing else, I look forward to seeing what they have built.  Based
on the little I saw of the first public release of Croquet, it will be a
strong statement about what Squeak can do.

Any ideas?

Have a good one,

Bill


--
Wilhelm K. Schwab, Ph.D.
[hidden email]

Hi Steve

email sent!

Steve