On Saturday 12 June 2004 04:48, Chris Metzler wrote:
> This is very very cool. I have one recommendation -- put a copy of
> this message on the wiki? Updated in the future as there's more to
> say about it, that sort of thing? I've been searching through
> mailing list archives the last couple of days, looking for descriptions
> of the possible elements (and possible values of elements) for different
> .xml files. Putting this stuff in the wiki would be good.
Yes, I was thinking about that as well. Just haven't had the time to do so
> Do you have any kind of feel for what the overhead of this is? How
> many concurrent routes can the traffic manager handle at once without
> significantly impacting the simulator? How many AI aircraft can
> that component handle?
The overhead for the traffic manager itself is pretty low: It checks the
position of only one aircraft, on every update, so the more traffic you add,
the longer it takes before each position is updated. I've added close to 6500
aircraft to my test schedule, and then it took about 90 seconds, to complete
the update cycle. This is also why I'm using a slightly conservative value of
500 nm distance to user as the cut-off distance to hand over traffic to the
AI Model subsystem and start a more detailed simulation. I don't know how
large the overhead of the AIModel subsystem will become once we start
simulating high traffic loads in the vincity of the user.
> Is this a general thing? For instance, let's say two different planes
> fly from EHAM to KSFO. Will they both have to use that same EHAM-KSFO.xml
> as their flight plan? I looked in the PH-KC? files and didn't see an
> entry that specified the flight plan file, which makes me think the flight
> plan would be generic to any planes flying that route. True? I guess
> the auto-flightplan-generation will change this, but I'm just curious
> how it works.
Yes, for now these are generic routes. The traffic manager composes the name
of the flightplan based on departure and arrrival airport, and tries to open
the file with the matching name. I haven't really worked out the details of
the future automatic Flightplan generation scheme though...
Flightgear-devel mailing list