[drats_users] Possible Who Is Online feature request

[Dan Smith]

> Think it would be useful to have Ping replies be "ACK'ed"?
>
> State machine/flow would be something like this: 
>
> - Station A sends CQCQCQ ping.
> - Stations B & C do as they do today, pick a random number of seconds
> from 0-10 and reply.
> - Station A "ACK"'s each reply as it comes in.
> - Stations B & C track whether their ping reply was ACK'ed.  If it
> wasn't, they pick another random number of seconds and try again up to a
> specific (LOW!) number of retries.

I think that this would quickly turn into a storm, which would be
*really* unappreciated by voice users.  If you have ten stations on
frequency, then you'd have something like 20-30 individual
transmissions (assuming that a decent number of the original 10
replies step on each other).  If you're unlucky, even a retry number
of 3 would make that pretty heavy.  Imagine if someone did that on a
reflector?  It'd be bad enough if they did a regular ping with what we
have now on a reflector.

> This way, if ping replies are hammered by voice traffic, another
> station picking the same "random" reply interval, etc... the station
> requesting the original Ping to CQCQCQ always is assured of getting
> a reply from each station on-frequency.

Another thing to remember is that I can't reliably tell when the
transmitter is keyed, and I can't flush the radio's buffer.

So, imagine that someone sends a CQ ping, and then someone starts a
2-minute-long voice transmission.  Stations B and C in the above
example send their reply, say, three times all of which is queued.
Then, when the person speaking shuts up, everyone keys immediately and
we get a huge burst of garble and nothing makes it through.

> Same algorithm for individual station pings, of course... but less
> "important" for those in some ways.

In response to a recent suggestion, I'm working on a connectivity test
function, which is similar to ping, but tries with increasing packet
sizes (in both directions) and marches upwards only if the remote
station is successfully replying to the pings.  This sort of situation
doesn't bother me because it's not something that could trigger a huge
storm if you show up on frequency not knowing that there are 20
stations online.

Something we could try to do in order to help make the CQ ping a
little more flexible is a dynamic reply timer.  Let the person asking
for the CQ ping specify the interval in which he wants to wait.  If
you start with a 10-second interval and you notice that a lot of
people step on each other (or you get fewer replies than expected),
try again with a 30-second interval.  Limit it to 60 seconds so
someone can't ping you and specify that you reply 48 hours later, of
course.

What do you think of that?

-- 
Dan Smith
dsmith#danplanet.com, s/#/@/
www.danplanet.com
KK7DS