<html>
<head>
<meta http-equiv="content-type" content="text/html; charset=UTF-8">
</head>
<body>
<p>The story behind the latest posting from Rob Vella KK9ROB,
regarding the "ASL Registration Code" post. Posted here to be a
matter of record. <br>
</p>
<p>--------------------------<br>
</p>
<p>The concept for the new registration was originally the idea of
Jason Kendall VE3YCA, however Jason was not able to spare cycles
to work on the new code as his job and life keeps him gainfully
employed. Around September 2020, when Rob joined in a discussion
on development with several of us, registration came up as well as
the desire to replace the Asterisk 13 based registration servers,
since they were still running a hack that was done to chan_iax by
me (with Tom Hayward KD7LXL doing a minor initial mod that led to
the start of the complete hack).<br>
</p>
<p>On 11 September 2020, Rob sent me a DM asking about ASL
registration and said "I've done a fair bit of socket programming
for a work project. You think it would be a worthwhile venture to
try to code an IAX server? I was thinking about it earlier.
Sniffed some packets, got a familiar with the registration flow.
It's only a few packets that are sent and looks pretty
straightforward."</p>
<p>"It should be fairly straightforward to do so. There's only 5-7
total packets to do a registration"</p>
<p> I told him he could and that there was an RFC for the protocol.</p>
<p>Then on 13 September from Rob:<br>
"won't be able to make the morning meeting but wanted to let you
know that I was successful in getting an IAX server setup. have
the full registration flow working (without verifying password,
just the packets) through UDP. going to work more on it tomorrow
to get is authenticating against real data."<br>
<br>
<img src="cid:part1.889A6003.9BBE621F@arrl.net" alt="" width="814"
height="198"></p>
<p>"just need to do some refactoring. assuming that it's going to
hit an HTTP endpoint for registration, as Jason expressed, so i
setup something really basic to just authenticate or reject. i'll
be implementing that packet tomorrow."<br>
<br>
"so far it's super fast. less than 15ms to do a full registration.
excited to see where it goes."</p>
<p>(of course there is more, like on 15 September the DM about "i
implemented client functionality, so as a validation measure, i
could literally forward the reaffic back out to a real reg server"
or the reply by me asking if this was in pyton "no. its in NodeJS.
send the hate. lol"<br>
</p>
<p><br>
</p>
<p> Jason and Rob did quite a bit of testing on systems that were
setup just for this purpose. Rob even requested and was provided
the packet captures from the live registration servers and said it
helped a lot.<br>
</p>
<p>Rob's front end was talking to the back end API that Jason wrote
in Python. In fact, both pieces of code still exist in the now
pttlink registration servers and were running on the ASL servers.
Unfortunately, there was a bug in the IAXjs code that had to be
addressed, and Jason was able to do a work around to fix it.<br>
</p>
<p><br>
</p>
<p>And for the record: Jason actually published an updated version
of the Python code about 3-4 months ago here:
<a class="moz-txt-link-freetext" href="https://github.com/apprpt-central">https://github.com/apprpt-central</a><br>
<br>
<a class="moz-txt-link-freetext" href="https://github.com/Apprpt-Central/pyIAX-Register">https://github.com/Apprpt-Central/pyIAX-Register</a></p>
<p><br>
</p>
<p><br>
</p>
</body>
</html>