--- Artur Kornacki <[EMAIL PROTECTED]> wrote: > Level 1 protocolIn level 1 protocol there is field > "16b Lengte bericht". I > assume this is length of whole packet. > > But if this data are being send with UDP protocol, > this information is > avaible from UDP header ( and it's easily accessible > ) - so there is no > reason to include this value in level 1 packet. > I also found that out. But some ppl like TCP more and to help those out, packet length was required. (It has been downsampled to 10bits though)
> > > With UDP packet size there is also involved > something like MTU (maximum > ... > minimal supported MTU (but i don't remember value > right now). > Minimal required MTU to be supported by a TCP/IP implementation is 576bytes (-8 for UDP-header) so our set maximum of 512 'pure' data should allow for all thinkable extensions(header data). > > > One more thing - compression. Becouse of possibility > of packet lost you need > to compress each packet independly (size 50-200 > bytes). And I have real > doubts that classical compression (like gzip, bzip2) > will work without > increasing packet size. > As has been said in previous threads on the subject, its optional, so it can be disabled and defaults to NONE. (ZLIB and LZO are being implemented atm) > > > Before you continue to work on creating this > protocol, I think you should > play Warbirs 2.7x. You can do it for free on Russian > server > http://home.udmnet.ru/aike/. Play for 2 weeks. After > that you will have real > understanding of what needs to be done. > > In short - these should be your minimal goals: > > 1) 200 players > > 2) Each player connected with max. 21 kb (modem > connection). > > 3) Ping 600 ms and more - and game is playable > > 4) Plane warping, after huge packet lost > My goals were not that big: 1) 30 players (within FOV) 2) min connection 4k (56k6 modem), no maximum just yet. 3) ping around 250ms (which is fair) 4) plane warping is a 'fact of life' no 100% fixes exist or can exist without lowering ping-limit. Plz note that Flight Gear is a 'development/experimental' enviorment, we can't just do whats best for 1 situation, we must cope with ppl who have different purposes for multiplayer capabilities. We personally want 'bots' and normal multiplayer. Some other ppl like 'dog-fighting', but as long FG doesn't support it I won't implement it either. Some other ppl want to use MP for even more specialized purposes or handling 'customized' subprotocols. To support that, we can't use 'server side master control' such as all other MP games have (unless you know a peer-to-peer MP game ?) We only send specialized info that we know, everything we can't parsed must be handled by the client(s). That requires a totally different approach then a single purpose optimized flight simulator. > > > With best regards, Artur Kornacki > Nobody has ever written a engine from scratch without a lot of prior network programming experience that supports 100+ players (ID software started with 8! and they only support around 32 atm) The best MOPG I know and ran on a single server was Subspace. All other massive online multiplayer games run on server clusters. Do you have a cluster at home that I can borrow ? :) This will be our first MPE from scratch. So don't expect a commercial quality engine that needs all the clients updating everytime the server says 'burb'. My 2 cents. __________________________________________________ Do You Yahoo!? Yahoo! Autos - Get free new car price quotes http://autos.yahoo.com _______________________________________________ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel
