[aprssig] are write-only APRS-IS clients valid?

[Lee Bengston]

Yes, that's a very clear and helpful description.

Thanks,
Lee - K5DAT


On Tue, Dec 3, 2013 at 11:58 PM, Pete Loveall AE5PL Lists <
hamlists at ametx.com> wrote:

> Actually, a better way to think of it is that a server is configured such
> that all clients are considered bidirectional IGates.  Lynn is correct in
> stating that the server does not differentiate between IGates and other
> clients.  As he stated and has been stated throughout this thread, the
> server maintains a separate "recently heard" list for each validated client
> on a limited feed and passes message packets addressed to any station on
> that "recently heard" list to the client (the client is always considered
> to be on that list).  The server will also pass the next valid posit from
> the message originator to the client so it can be sent out "as a courtesy".
>  Finally, if any packets are seen originating from a station on the
> "recently heard" list that were directly injected into APRS-IS (TCPIP* is
> the only "digipeater" in the path), those packets are sent to the client.
>  If a client is not an IGate, its "recently heard" list is empty but the
> above processing is still performed (of cours
>  e netting no packets beyond messages and associated posits addressed to
> the client).
>
> It is important that the server makes no decisions for the client
> regarding what packets are going to be sent to the client.  It is 100% per
> the above algorithm so if the client is an IGate, it gets sufficient
> information to perform its IGate function.  As also stated earlier, the
> "filters" (better term "subscriptions") that a client can set are additive
> to the above algorithm but do not override it.  Finally, full feed ports do
> not implement the above algorithm since all packets the server sees are
> passed to the client anyway.  This algorithm only exists for limited feed
> ports.
>
> The server's job is to pass packets to all clients as quickly and
> efficiently as possible while eliminating duplicates and preventing loops
> and mangled packets.  When a client connects to a limited feed port, the
> server restricts the packets passed to the client to the above algorithm
> (the client is always considered "recently heard") and any filters that
> might also be defined, if any.  This assumes, of course, that the client
> has provided a valid passcode at login.
>
> It is the responsibility of the client software to function as an IGate
> per the spec if it is an IGate and it is the responsibility of the server
> to make sure that all packets as defined in the above algorithm are always
> passed to the client.
>
> Don't put too much "intelligence" into how a server operates.  They are
> designed to move volumes of packets to all clients without regard to what a
> particular client is.  That design includes ensuring that limited feed
> ports pass the minimum packets necessary to fully support messaging to the
> client and for any clients located behind the directly connected client
> (i.e., an IGate).  That design is port and connection related, not client
> related (again, the server does not know what function a client might
> perform).
>
> Hope this helps.
>
> 73,
>
> Pete Loveall AE5PL
> pete at ae5pl dot net
>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.tapr.org/pipermail/aprssig_lists.tapr.org/attachments/20131204/55bfc573/attachment.html>