On Wed, 2012-05-16 at 01:31 +0100, Peter Morgan wrote:
> I think maybe we need to also invite the fgms team into this discussion
> The only way really to get a position sat on webserver atmo
> is to poll a fgms telnet server.. upon its admin port..
> eg telnet
> telnet mpserver01.flightgear.org 5001
> return lines..
> which are then "parsed" into various formats..
> so we need to cover this scenario across anyones installation.. to
> make it usable..
> another thought..
We have discussed this off list, but to repeat
some of my ideas, which of course can be 'corrected'
1. I do not think it is good to bash one or each public
FG server on their telnet 'admin' port every few
seconds, to get current online pilot information...
The present fgms code drives a new process (originally
a fork(), but now a pthread in later code if configured)
into existence to reply to each 'telnet' query...
quite a load...
2. I agree and LIKE the idea of a NEW world 'map' of
all FG pilots online... that would be GREAT, in addition
to the existing 'pigeons code' ;=))
As suggested off list, my idea to achieve this is to
host a NEW fgms server... and get all other current
FG mp servers to relay packets to it... And that
means applying to Curt to get a DNS record -
Now this new server would be -
(a) serving the FG community, and
(b) you can either 'bash' its telnet port, it is
your server to treat as you like ;=)) or
(c) as provided in the fgms code, connect another
server to it to provide data in any form required.
The fgms code already has a 'crossfeed' config
parameter, so can feed every packet received to
that connected 'crossfeed'... with the header
set to RELAY_MAGIC... even before 'Blacklisted'
or 'Invalid' packet rejection...
That 'crossfeed' data can then be accessed ANY way you
desire, and produce pilot information in ANY form
needed... on a private or public ports... without
impacting the FG community mp servers...
As I have stated I would assist to develop this
'crossfeed', which in essence is just another fgms
server, given some desired specifications...
And that could perhaps include providing say a
pilot's historical 'track', and perhaps even some
'predicted' track... to show on the 'map'...
But the FIRST step is housing and setting up a
new mp server, and getting all others to relay
But as usual, this is just my 10 cents ;=)) Maybe
everyone is happy with telnet 5001 server
'bashing' as the source of the 'map' data...