> The first important decision now is, do we > > multi-threading or multiplexing-IO ? For > multiplexing, > > PLIB can be used. For multi-threading, we got a > > problem. > > Another issue is how to get the code to compile on > > other platforms, because threading and sockets > aren't > > 100% portable, to say the least. > > PLIB's socket library is reasonably portable. It > runs on every platform that FlightGear runs on. > There are many "portable" open source threading > libraries available, Boost.Thread, ZThread and > CommonC++ to mention just a few. > > Bernie
Another issue here is performance/stability. Threading gives best performance(and most problems) with multiple processors. With a single processor the overhead might not measure up against performance. I don't have such an archictecture so bugs might not surface at my platform. Our first prototype will use multiplexing I/O, if it can handle 10+ users at once then we might stick with it. If it can't handle at least 10 users we will probably switch to threading. This should not effect clients since we will be using Flight Gears own scheduler for that. There might be a parameter to expect X packets per second spreaded of the second to minimize network overhead (like 10 for the 56k modem users, network users can turn this value to 100 or over). I don't know if we will be using ratios (56k modems have 33k6 up and +/- 45k down(on bad lines). I personally got 256k up and 512k down (in kilobit/sec) so there might be a ratio involved. __________________________________________________ Do You Yahoo!? Yahoo! - Official partner of 2002 FIFA World Cup http://fifaworldcup.yahoo.com _______________________________________________ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel
