[aprssig] proper formatting of IS->RF APRS messages

[Andrew Pavlin]

Greetings.
It appears that I screwed up my APRS program YAAC over a year ago, and no one noticed until last month!
I happened to review the specifications at http://www.aprs-is.net/IGateDetails.aspx and saw something that seemed rather odd.-----------------------------

 IGates must use the 3rd-party format on RF of 
IGATECALL>APRS,GATEPATH:}FROMCALL>TOCALL,TCPIP,IGATECALL*:original packet data where GATEPATH is the path that the gated packet is to follow on RF. This format will allow IGates to prevent gating the packet back to APRS-IS. The format of the 3rd party path (TCPIP,IGATECALL*) is mandatory; APRS-IS paths MUST be removed before gating to RF. ------------------------------
It seemed weird to me, but I changed my code so the RF-transmitted packet used my I-gate station's RF callsign-SSID and tocall on the RF AX.25 packet header, and the original station's callsign and tocall only appeared in the 3rd-party header.
Now this is causing duplicate processing of APRS text messages, because an I-gate using my software is one of (but not the only) source to the complaining user's station, and the end-recipient is receiving the message over RF with both the original station's callsign and the Tx I-gate's callsign (in separate RF packets).

I looked at some other open-source APRS software's Tx I-gate implementations, and they preserve the original sender's source and destination fields in the RF AX.25 frame, but replace the digipeater list with "TCPIP" and the Tx I-gate's RF callsign-SSID (i.e., from where it was relayed back to RF). This makes a whole lot more sense to me. But is it correct?

So what _is_ the correct behavior? Related to this, what is the correct contents of the GATEPATH? I'm assuming I should put RFONLY and NOGATE in it (just so stupid receiving I-gates might not send the packet back in again). Do I need to put "TCPIP,myrfcallsign-ssid" in the RF digipeat path as well? If so, I assume I should put them in first and  mark them as already-digipeated, so any additional digipeat aliases (to support multi-hop Tx I-gating) would be next. Should RFONLY and NOGATE always be last in the RF digipeat list (assuming we don't have viscous digipeaters)?
Awaiting advice from the I-gate masters....
Andrew, KA2DDOauthor of YAAC
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.tapr.org/pipermail/aprssig_lists.tapr.org/attachments/20220101/b0997b85/attachment.html>