|
|
|
26-Sep-2003 14:34 |
|
|
This is Part Two in the history of FidoNet. It turned out
that the original FIDOHIST.DOC (now called FIDOHIST.DC1, or just
"Part One") was useful, and many people read it. Unfortunately,
by the time everyone read it, it became totally obsolete. Oh
well. Here is Part Two.
FIDOHIST.DOC covered the early history of FidoNet, why it
was done, how it was done, and the reasons for the organization
and obscure rituals surrounding node numbers. If you haven't read
it yet, I suggest you do now, because I'll probably refer to
things that won't make any sense otherwise.
The original FidoNet was organized very simply; each
FidoNet system (each node) had a number that served like a phone
number, uniquely identifying it. The NODELIST, generated by the
folks in St. Louis that had all FidoNet nodes in it, contains
information on all known FidoNet systems. Every system in FidoNet
had a current copy of the NODELIST, which served as the directory
of systems.
(In the interests of brevity I'm leaving out huge amounts
of information; I hope you have read FIODHIST.DOC by now ...)
FidoNet has been growing steadily since it started by
accident in May 84 or so. The node list continued to get out of
hand; the original FIDOHIST.DOC was written to try and help
smooth things out. It is impossible to overemphasize the amount
of work involved in keeping the node list accurate. Basically,
the guys in St. Louis were keeping track of hundreds of FidoNet
systems in Boston, Los Angeles, London, Stockholm and Sweden, and
publishing the results weekly. There has never been such a
comprehensive and accurate list of bulletin board systems
generated.
We talked for many months as to how we could possibly
find a solution to the many problems; it was at the point where
if a solution was not found in a few months (by Aug. 85 or so)
that FidoNet would collapse due to the sheer weight of it's node
list.
The newsletter, FidoNews, was, and still is, an integral
part of the process of FidoNet. FidoNews is the only thing that
unites all FidoNet sysops consistently; please keep up to date on
it, and stock it for your users if you have the disk space. And
contribute if you can!
There were many constraints on the kind of things we
could do; we had no money, so it had to be done for zero cost.
Centralization was out, so obviously localization was in; just
how to do it was a total unknown. We thought of going back to
having people in different areas handle new node requests in
their area, but that always generated confusion as to who a
person should go to, how to avoid having someone requesting a
node number from different people simultaneously, etc etc.
The old method of routing was very different than the
current method, and much more complex; instead of Fido
automatically routing to hosts, each sysop had to specify (via
the ROUTE.BBS file) how all routing was done in the system. The
was done originally by hand, later by John Warren's (102/31)
NODELIST program.
Then of course there was the problem that no matter what
we did, it would not be done overnight. (ha ha.) It would take
many weeks at the least, possibly months, so that whatever we did
had to be compatible with the old method as well.
We went through probably hundreds of ideas in the next
few months, some possibly useful, some insane. Eventually the
insanity boiled down to a pretty workable system. We chatted by
FidoNet and by voice telephone. Eventually, we settled on the two
part number scheme, like the phone company does with area codes
and exchanges. It accommodated backwards compatibility (you can
keep your present node number) and the new "area code" (net
number) could be added into an existing field that had been set
to zero. (This is why everyone was originally part of net #1).
When a fortunate set of circumstances was to bring Ezra
Shapiro and me to St. Louis to speak to the McDonnell Douglas
Recreational Computer Club on XXXX 11th, we planned ahead for a
national FidoNet sysops meeting that weekend. Ken and Sally
Kaplan were kind enough to tolerate having all of us in their
living room.
The people who showed up were (need that list) The
meeting lasted ten continuous hours; it was the most productive
meeting I (and most others) had attended. When we were done, we
had basically the whole thing layed out in every detail.
We stuck with the area code business (now known as net
and region numbers) and worked out how to break things up into
regions and nets. It was just one of those rare but fortunate
events; during the morning things went "normally", but in the
afternoon solutions fell into place one by one, so that by late
afternoon we had the entire picture laid out in black and white.
Two or three months of brainstorming just flowed smoothly into
place in one afternoon ...
What we had done was exactly what we have now, though we
changed the name of "Admin" to "Region", and added the
"alternate" node and net numbers. (We still seem to be stuck with
that terrible and inaccurate word, "manager". Any ideas?) I
previously had a buggy test hack running using area codes, and
the week after the meeting it was made to conform to what we had
talked about that Saturday.
When version 10C was done, it accomplished more or less
everything we wanted, but it sure did take a long time. 10C was
probably the single largest change ever made to Fido/FidoNet, and
the most thoroughly tested version. At 10M, there are STILL bugs
left from that early version, in spite of the testing.
Once the testing got serious, and it looked like we had a
shippable version, St. Louis froze the node list, and started
slicing it into pieces, to give to the soon-to-be net and region
managers. (That word again.) This caused a tremendous amount of
trouble for would-be sysops; not only was it difficult enough to
figure out how on earth to get a node number, once they did they
were told node numbers weren't being given out just yet.
Explaining why was even harder, since FIDOHIST.DC2 (ahem) wasn't
written yet. (I have to agree, this thing is a little bit late)
It was a typical case of those who already knew were informed
constantly of updates, but thee in the dark had a hard time.
Things were published fairly regularly (am I remembering
"conveniently" or "accurately" on this part?)
Eventually, 10C was released, and seemed to work fairly
well, ignoring all the small scale disasters due to bugs, etc. We
couldn't just swap over to the new area code business until very
close to 100% of all Fidos were using the new version. This was
(for me) an excruciating period, basically a "hurry up and wait"
situation. There had not been a node list release for a month or
two, and for all practical purposes it looked like FidoNet had
halted ...
Finally, on June 12th, we all swapped over to the new
system; that afternoon, sysops were to set their net number (it
had been "1" for backwards compatibility), copy in the new node
list issued just for this occasion, and go. I assumed the result
was going to be perpetual chaos, bringing about the collapse of
FidoNet. Almost the exact opposite was true; things went very
smoothly (yes, there were problems, but when you consider that
FidoNet consists of microcomputers owned by almost 300 people who
had never even talked to each other ...)
Within a month or so,just about every Fido had swapped
over to the area code, or net/node architecture. With a few
exceptions, things went very smoothly. No one was more surprised
than pessimistic I. At this time, August, I don't think there is
a single system still using the old node number method.
This is all well and fine as far as the software goes,
but it made a mess for new sysops. For us sysops who have been
around for a while, there was no great problem, as we saw the
changes happen one by one. However, new sysops frequently came
out of the blue; armed with a diskette full of code, they
attempted to set up a FidoNet node.
Actually, I don't understand how anyone does it. The
information needed is not recorded in any place that a non sysop
could find. On top of that, most of it is now totally wrong! If
you follow the original instructions, it said "call Fido #1 ..."
if you found a real antique, or "call Fido #51 ..." if it is more
current. Of course now it tells you to find your region manager.
"Region manager???" Well, a list of region managers was published
in FidoNews, but unless you read FidoNews, how does anyone ever
find out? I'll probably never know.
ANYWAYS ... the original reason for all the changes was
to DECENTRALIZE FidoNet. It just wasn't possible for Ken Kaplan
to keep accurate, up to date information on every Fido in the US
and Europe. The decentralization has been more or less a total
success. The number of problem introduced were negligible
compared to the problems solved, and even most new problems are
by this time solved.
It is interesting to note that with the hundreds of
systems there are today, the national FidoNet hour is less
crowded than it was when there were only 50 nodes.
Please, keep in mind that no one has done anything like
this before, we are all winging it, and learning (hopefully) as
we go. Please be patient with problems, none of us is paid to do
this, and it is more and more work as time goes on. Somehow it
seems to all get done ...
HOW TO GET A NODE NUMBER AND ALL THAT
20 August 1985
This is by necessity a very general idea of how it's
done, and you were warned earlier that this may be obsolete this
very minute; with that, here's the "current" process for starting
up a new FidoNet node.
You can of course skip all or part of this if you've done
this before; if you haven't, well, be prepared for a lot of
searching and asking questions.
Of coursve, you need to have your Fido BBS system running
first. It's probably best that you play with it for a while, and
get some experience with how it all works, and whether you have
the patience to run a BBS. It can get exasperating, and you will
never find time to use the computer ever again.
Obtain the most recent copy of the nodelist possible; this
may take some searching. If you get totally lost, you can always
contact Fido 125/1 or Fido 100/51; though these are very busy
systems, they both usually have the very latest of anything, and
can direct you to the right place.
The big problem here is to find out if you are in a net
or not, and if not, then who your region manager is. If you are
in a large city (Los Angeles, Cincinnati, etc) then there is
probably a net in your area. Look through the node list (use the
N)odebook command in Fido, or a text editor) for the right area
code or city.
If there is no net in your area, then you are part of a
region. This is a little harder, because regions are large, and
sometimes cover many states. Look at all the regions in the node
list, you should find a region that fits you.
Once you find this, you have to contact the net or region
manager to get your node number. Exactly how this is done depends
on who the manager is, and how sticky they are fir details. A
near universal requirement is that you send your request via
FidoNet, not by manually; this isn't done to make you life
difficult, but to ensure that your system is really working
right. IF you manage to get a FidoNet message to the manager, its
usually safe to assume that you're system is working OK. If you
get a reply in return, then you know both directions work.
It is usually each sysops' responsibility to go get the
latest nodelist and newsletters; they are not distributed to all
systems because of the expense. (Though, I'm trying to get them
distributed to more places than they are now, it's sometimes very
difficult to get a copy of the nodelist!)
Again, read the fidonet newsletter regularly; it is about
the only way to stay in contact with the rest of the net.
Programs, problems, services, bugs and interesting announcements
can always be found there. FidoNews articles don't come out of
thin air; send in anything you think might be of interest. They
don't have to be lifetime masterpieces, or even well written.
Please remember the entire network is made of the sysops;
there is no central location from which good things come, the net
consists entirely of the sysops and their contributions. If you
don't do it, chances are no one else will!
Tom Jennings
20 Aug 85
Ken Kaplan Fido 100/51314/432-4129
Tom Jennings Fido 125/1415/864-1418
Ben Baker Fido 100/10314/234-1462
|
|
|
26-Sep-2003 14:34 |
|
|