[aprssig] Re: More thoughts on Connected Packet - telpac_node

[Wes Johnston]

I used to play with NOS and (so far as I know) got the first http bbs on the air
in SC back in 1995 using win3.11 and an ethernet to kiss encapsulation driver. 
The host was JNOS run by a neighboring ham, kd4htu.  It was a real pain in the
backside to make it all work.

In those days we had fixed IP addresses and didn't use DHCP.  One of the
complaints I saw in this thread yesterday was about how difficult it is to know
the local stations and how difficult it is to setup NOS or a linux box.  DHCP
solves the problem of knowing the local callsigns/addresses.

What I got to thinking about on the way to work this morning was that linux now
uses sound card packet and certainly speaks AX25.  I know you can run samba and
any other network protocol over ax25 on a TNC or sound card modem.  So it
appears we have the ability to do TCP/IP on air with linux.  This would include
DHCP - which solves the "how do I connect to the local network?" problem.

In the windows world, we have the wonderful AGWPE and it has network drivers as
well.  Using AGWPE, you can pop email, surf the web.  It will support DHCP too.

With windows IIS and linux SME 6.0 server, it is really quick to setup local
email with or without an external internet connection.  In either case, win or
linux, because both use standard protocols, it won't matter if the clients are
win and the server is linux or vise versa.

That said, what I do remember about tcp/ip is that it's really chatty on air. 
In a direct link to the server it's OK at 1200 (make no mistake it was
excruciatingly slow even in 1995), but it's impossible (practically speaking)
once you start having to route on air via net/rom.  TCP/IP uses what I'll call
"disposable" packets.  Net/rom delivers EVERY packet.  So when the two mix, my
TCP/IP station would assume a packet had been lost and send the same packet
again.  My local Net/ROM node would accept this packet and add it to the
backlogged queue and keep trying.  Net/ROM would deliver each and every packet
eventually.  Problem is that by the time it delivered the 2nd copy of a given
packet to the far end of the link, the far end didn't care.  The plethoria
(sp?) of packets would cause the network to effectively collapse on itself. 
Lesson learned was to route TCP/IP the whole way.

Where am I going with all of this?  I have not played with winlink, but will. 
If it turns out that winlink does something to reduce the "chattiness" of
TCP/IP on air, it's a winner.  I think it does.  But if it ends up being just
as chatty, we may as well run a more standard set of TCP/IP protocols on the
air.

I'm actually looking forward to playing with winlink.  The only problem I have
is that it may not run in linux... does it work with WINE?

Just my $.02 worth.
Wes


--



Quoting Bill Vodall <wa7nwp at jnos.org>:

>
> > > No reason they can't be -- specially if running Linux.   telpac_node is
> > > trivial to install and run.  It's sitting ready in the background and
> > > nobody
> > > knows it's there until it's needed.
> > >
> > >  http://www.jnos.org/blog/2004_11_01_jnos_archive.html#110140782217158315
> > >
> > > Bill
> >
> > Bill, tell me about the Linux Telpac!  I have not heard a thing about it.
> >
> > Jim
>
> Not much more to say beyond the notes on the blog.
>
> It's been mentioned a couple time on the wl2kemcomm list.  I downloaded
> the source, compiled it and installed on a couple systems.  Both worked fine.
>
> (I won't mention the other system where I'm still fighting with the
> compile...)
>
> Bill
>
>
> _______________________________________________
> aprssig mailing list
> aprssig at lists.tapr.org
> https://lists.tapr.org/cgi-bin/mailman/listinfo/aprssig
>