[ax25-layer2] Re: ax25-layer2 Digest, Vol 11, Issue 1

[pete at ae5pl.net]

I disagree.  First we must define what we are doing in such a manner
that it doesn't matter what platform people want to develop to.  In
doing so, we must also remember that there are some already developing
new TNCs and there are some who have developed some types of soundcard
interfaces as well as generic TNC interfaces.  Whatever we do must be
usable by them as well which means that we must first define what we
want to do and then define algorithms (not code) that anyone can design
and code from.

The first thing on the list is to nail down and document _what_ we want
the implementation to do (specifications).  Second, we define _how_ we
implement what we want to do (algorithms).  Then we can code to those
algorithms and specification.

To that end, I have obtained the domain name ax25.net and will have a
web site up by the end of the week.  At that web site, we can maintain
as many pages as necessary to accurately document what we want to
implement and how to implement.  This SIG is where we need to discuss
the issues but we also need to have a concise repository that someone
can go to and say "Oh, I can code to that" or "Maybe this needs to be
changed".  Additions, changes, and deletions will be discussed here and
I will keep the web site up-to-date.

Also, I have already coded the NSR in Java and some others have done it
using various platforms.  While Java is great for multipurpose computer
platforms such as PCs because it is OS independent, it doesn't scale
well down to proms, etc.  But at least we know things like this can be
coded and tested without a huge amount of effort.

As for networks to test on, we have a number of them and we can also
establish test networks with specific configurations.  This last part is
actually the first testing that should occur (alpha testing) before
taking it to a live network (beta testing).

Will things happen if we follow an orderly course of action?  Much
faster to the desired end of a usable implementation.  If some of you
wish to start coding without a specification and algorithms to code to,
feel free to.  However, be advised that much of that coding could be in
jeopardy depending on what is created as specifications on this list.
Also understand, the specifications and algorithms aren't cast in
concrete once "finished".  However, they stand far less chance of being
a moving target than if you code now and then try to create the
specifications.

73,

Pete Loveall AE5PL 

> -----Original Message-----
> From: James Ewen
> Sent: Monday, September 26, 2005 8:25 PM
> To: Discussion of Link Layer use of AX.25
> Subject: Re: [ax25-layer2] Re: ax25-layer2 Digest, Vol 11, Issue 1
> 
> Bill Beech wrote:
> 
> > I agree we must get some code working first, then play with it, and 
> > finally document it, then the question becomes on what 
> platform do we 
> > build the reference implimentation?
> 
> The best platform is one that allows the most experimentation 
> for the least cost. The ability to change programming very 
> easily is of paramount importance.