[drats_users] Using D-Rats on a reflector

[Nate Duehr]

Dan Smith wrote:
> Hi James,
> 
>> I was listening to Dan, KK7DS, on the REF002A reflector at around 11:30
>> PM EDT.
> 
> Ah, very good.  That was literally my first time on a reflector.  It's
> pretty neat, but it seems a little fragile, even from a voice perspective.

I had a brief e-mail discussion with Robin about one of the things I saw 
this week... if the header packet is lost, no audio comes through.

How this gets a little weird, is that the header can get lost 
"anywhere"... so what I heard during a Net was this:

- "W1ABC go ahead with your comments."
- [Long silence]
- "Thanks for your comments.  Any other stations with comments or 
questions?"

This means the header for that middle transmission was lost somewhere 
between the Reflector and W0CDS' gateway.  We never heard the middle 
transmission, but "everyone else" probably did.

So yeah... "fragile" is kinda a good word for it.  When header loss 
happens during a round-table, it gets REALLY confusing.

> Let me ask you a question.  Even though I think I knew better, I wasn't
> always leaving much of a gap between key-down and starting to talk.  Was
> I being cut off in the beginning of a voice transmission some of the time?

Here's what I think happens sometimes.

On the sending end... the person unkeys and their Gateway sends back 
"RPT?" to their rig and an ID.

On your end, receiving... if you key up and start talking before THEIR 
end has finished that ID transmission, I think they never hear you.

> If using a reflector results in, say, a 2-second period where packets
> might get dropped, then I could compensate for this in D-RATS.

It's really a timer from "last unkey"... delay after every unkey of the 
receiver enough time to let everything "settle", I think is the only way 
you could deal with it.

Also Multicast on a Reflector is probably a mess... if one station isn't 
hearing the sending station's packets and tries to send a NAK while the 
others are in the middle of copying data, it won't "interrupt", but the 
timing then gets really strange for the re-tries.  Multicast would get 
really problematic if there was "header loss" to only one connected gateway.

> That's a good question, and I was wondering the same last night.  I
> would guess that the way the repeater is linked to the reflector, that
> you'd still go to everyone and suffer the same issues.

You do NOT want to mix "push" or "Icom" routing with Dplus Reflectors. 
Try it out with voice sometime.  You get both dplus sending a stream to 
the far end and Icom's software *at the same time* and the results out 
the far-end are a garbled mess.

If you "Icom/push" callsign route INTO a Reflector-connected Gateway, 
dplus does the "right thing" and sends that transmission onto the 
Reflector, but replies to "CQCQCQ" on the Reflector won't get BACK to 
that person who just called.  A bit confusing for users, thus it's 
important to voice announce WHAT you're doing when you use the Icom 
callsign routing.  That way someone on the Gateway being called can 
one-touch back and say "hey, we're on the Reflector".

>> What I'm seeing is everyone is trying to use D-Rats with CQCQCQ like a
>> broadcast when I would think that you want to send the message or file
>> to a specific station just like APRS.

Using callsign routing should work fine with D-RATS but NOT in 
combination with a Reflector link.  Also I'm not sure if the rig sends 
the "RPT?" and "UR?" indications back to D-RATS, indicating "no you 
didn't route" or "yes you did route"... if that *is* available out the 
serial port, that would be pretty helpful for Dan... he could watch for 
that indication, in a special "callsign routed" mode.  But it'd have to 
be a user-controlled thing, since Reflectors and Dplus Gateway to 
Gateway links do NOT provide these indications.  (You always get "RPT?" 
back.)

> I've added a knob to let you tweak the number and size of packets that
> get sent per round of a file transfer.  Let me add something to scale
> the size of the "warm up" header and perhaps a delay setting.  I'll post
> a beta and then people can play with it and report back.

Cool! I need to get fired up to play here, but off to lunch at the 
moment, and I've been working (real work... like the paid kind) all 
morning.  Yuck.

Nate WY0X