Hello Bj”rn.

25 Mar 06 02:23, you wrote to me:

Sorry for the late response. There were a couple of other things - mostle
not Fido related - o tghe priority list.

 MvdV>> A new sysop (node- or point) could easily make this mistake.

 BF>    You conveniently left out all the other factors I listed, that also
 BF> must be met to make this elusive 000-call happen.

I concentrated on the main issue: configuring a system as to make this
possible. As for accidentally setting the crash attribute; it happens. We
all have seen numerous examples, even from those that should be a long way
from the starting point of the learning curve.


 MvdV>> IMNSHO this 000- thing was a bad idea from the start.

 BF>    I think it was brilliant in all it's logic. IP-numbers are phone
 BF> numbers on the internet and what country code would be more logical
 BF> for internet than 000?

It may have looked liked a good idea at the time, but somtimes it happens
that good ideas are not so good on closer inspection. IP-numbers may in
some way be the equivalenet of telephone numbers, but the *are not*. Whwn
you atteempt to dial an IP number on an ordinary telephonem the results are
unpredictable. As for 000, the logical assigment for that is the access
code for out of planet dialling.

 MvdV>> The main reason I think this was a bad idea is that POTS
 MvdV>> mailers were never designed to have to deal with numbers that
 MvdV>> should never be dialled.

 BF>    Fidonet was never designed to use internet in the first place. It
 BF> sure as hell is fortunate, that some of us never let that prevent us
 BF> from experimenting with different ways to overcome that shortcoming.

Indeed. When experimenting to find new and better ways one should however
also be prepaired to drop methods that on investigation turn out to be not
such a good idEa after all instead of foolhardedly insisting on it.

 MvdV>> One simply can not expect all legacy  mailers to handle
 MvdV>> the situation correctly.

 BF>    Why not?

Because developers and programmers arent clearvoyant. When they developed
those mailers, they expected *dialable* phone numbers in the phone number
field, except for the string -Unpublished-. They never expected someone to
put non dialable numbers in there and so they did not take measures to see
tha it is handled properly.

 BF>  Some, at least FrontDoor, can, and has been doing so from
 BF> the early days of the internet boom.

The world of FidoNet is bigger than the world of FronTDoor.  Because some
cars can double as a boat, we can not expect all cars to do that.

 MvdV>> I see no gain in using this method, only loss.

 BF>    Yeah, well, it's like with all those fancy nodelist flags that
 BF> almost nobody can see any gain in: as long as one, single node can
 BF> benefit from it, that's at least *one* good reason to allow it.

You have a point: up to a point. But I think we should make a distinCtion
between things that only take up disk space for most and things that
actually cause problems for others and/or things that are detrimental to
the smooth operation of the network.

And of course you conveniently leFt out all *other* arguments against this
method of listing:

1) There are mailers out there that have no documented way of stopPing them
from attempting to dial *something* when encountering anything other than
-Unpublished-. They can translate the number in to something else, but they
can not be stopped from attempting to dial at all. Forcing people to change
their trusted and knwons software is infair and detrimental to the smotth
operation of the network.


2) There is only room in the phone number field for ONE entry. That means
that dual capable systems: POTS and IP, have to be listed with two or more
node numbers. That robs owners of intelligent mailers of letting the
software find out their best method of connection by itself. No mailer can
determine if a second listing is a true aka of one and the same system, a
second system of the same sysop or a completely different system of another
sysop that happens to Have the same name. Even humans make mistakes there.
Unneeded multiple listings with different associated connection methods are
detrimental to the smooth operation of the network.



Michiel

--- GoldED+/W32 1.1.5-0613