[Xastir-dev] a readable version of the latest official APRS spec?

[Tom Russo]

On Tue, May 24, 2005 at 09:38:24PM -0400, we recorded a bogon-computron collision of the <n2ygk at weca.org> flavor, containing:
> Sorry for the off-topic post but I felt this group might be more 
> appropriate to ask this question of:  I'm trying to figure out how to 
> make aprsdigi (aprsdigi.sourceforge.net) do whatever the fix14439.html 
> document of Bob's says.  I must say, frankly, as usual with Bob's 
> documentation, it is very long and tends to ramble in a manner that 
> makes it difficult to read.  In fact, right now, on my slow inkjet 
> printer, it's even difficult to print:-)  I found reading the official 
> APRS Protocol Specification at www.tapr.org to be much more useful when 
> developing aprsdigi.  However, there seems to be no update to that spec 
> despite the fact that Bob's web pages claim the working group voted on a 
> 1.1 spec a year ago.  Did the TAPR coordination of this die?

I find these documents similarly tedious to slog through.  It does not appear
that there is any official protocol document other than the spec you mention.
There are plenty of errata, and they're all there on Bob's web site.  

> I do not have the time to read the mailing lists lately and am just 
> trying to keep my code current.  Who knows, one day I may even ask Bob 
> to include it in the list of digipeaters.  Aprsdigi is in use at a small 
> number of sites (many of which I set up:-) in the greater NYC area and 
> there have been a few others who've donwloaded it and I'd like to make 
> sure it is correct.  For example, what the heck does WIDE2-1 mean in 
> fix14439?  I aways though the first digit was the original value of the 
> digipeat count and the second is of course the SSID which gets counted 
> down to zero.  So WIDE4-4, WIDE2-2, etc. would make sense as an initial 
> value.  Now I see this WIDE2-1.  My brain hurts.

In a nutshell, as I understand it:

 0) relay and wide are no longer to be used for aliases by any digi
 1) all wide digis should be widen-N, and possibly SSn-N (statewide, to 
    allow long paths within a state, but not polluting neighboring states) 
 2) in special circumstances, fill-in digis that used to be "relay" should
    instead alias as "wide1-1", but only when necessary to get a mobile in that
    area to a real widen-N
 3a) in areas where complete coverage is had by widen-N's, mobiles would use 
     WIDE2-2
 3b) in areas where fill-ins are needed, mobiles would use WIDE1-1,WIDE2-1
    to effect a 2-hop path, with the first being a fill-in digi.
 3c) BUT: since in areas where there is complete coverage by WIDEn-N's, 
   WIDE1-1,WIDE2-1 is the same as WIDE2-2
 3) SO: mobiles should use WIDE1-1,WIDE2-1 regardless

 4) In parts of the country where it is necessary to go more than two hops,   
    WIDE3-3 might be OK
 5) Paths of length WIDE4-4 or longer should be discouraged by making wide
    digis alias them --- a wide digi so configured would turn WIDE4-4 into
    it's own call and mark it digipeated, making an obnoxious path into a
    single hop
 6) in extremely congested areas, even 2 hops might be aliased away

This replaces the old recommendation that all home stations alias as RELAY, that
even WIDEn-N digis alias as RELAY *and* WIDE, and thatmobiles use RELAY,WIDE or 
WIDE,WIDE2-2.  The new recommendation enables real WIDEn-N to desupport both 
RELAY and WIDE aliases.  Desupporting those aliases in high-wide digis 
improves dupe elimination.

I think that is a somewhat accurate summary, but after months of seeing the
recommendations change over and over again, I can't be entirely sure.  YMMV,
IANAL, may cause drowsiness, use caution while driving, discontinue use if rash
develops.

You ask what the heck WIDE2-1 is.  Well, WIDE2-2 would normally be turned into 
WIDE2-1 by the first WIDEn-N digi it hits.  That WIDE2-1 would be turned into 
WIDE2 by the next digi it hits (that has not already digipeated the packet).  
By 3a) above, one would want all 2-hop mobile paths to be WIDE2-2, but 
sometimes we have 3b) instead.  By using WIDE1-1,WIDE2-1, we can have a one-hop followed by another one-hop path, and take advantage of good dupe checking
if both are real WIDEn-N's, and fill-in digis if the first is just a home
station helping out.  But the WIDE2 remains after all the digipeats are done
to indicate to the receiving station how long the path was supposed to be.

Ex:
In an area where all digis are high, widen-N's:
     KM5VY-9>APX151,WIDE1-1,WIDE2-1....
   hits a wide digi:
     KM5VY-9>APX151,WIDE1*,WIDE2-1....
   then hits another wide digi:
     KM5VY-9>APX151,WIDE1,WIDE2*...
   and goes no further.  This seems to me to be indistinguishable from the 
   last hop of the three-hop path WIDE1-1,WIDE2-2, so you can't know fore
   sure it was only 2 hops, but there you have it.

In an area where the first wide digi is unreachable except through what used
to be a relay under the old paradigm, say, K5DIG:
    KM5VY-9>APX151,WIDE1-1,WIDE2-1...
  becomes
    KM5VY-9>APX151,K5DIG*,WIDE2-1...
  then it hits the wide digi:
    KM5VY-9>APX151,K5DIG,WIDE2*...
  
The good thing is that under no circumstance would there be a cascade of 
dupes the way the old RELAY,WIDE2-2 would when the initial packet is heard by 
multiple WIDEn-N's all answering to "RELAY" as well.

-- 
Tom Russo    KM5VY     SAR502  DM64ux         http://www.swcp.com/~russo/
Tijeras, NM  QRPL#1592 K2#398  SOC#236 AHTB#1 
 "The only thing you can do easily is be wrong, and that's hardly
  worth the effort." -- Norton Juster