Text 60, 202 rader
Skriven 2006-04-11 19:11:00 av Michiel van der Vlist (2:280/5555)
     Kommentar till text 50 av Robert Couture (1:229/2000)
Ärende: FD & -Unpublished-
==========================
Hello Robert.

11 Apr 06 08:59, you wrote to me:


 RC>>> Sorry, I cannot conform any issues with the nodelist.

 RC> Conform?  Sheesh!  I meant confirm of course ... :)

;-)


 MV|>> There are no "issues with the nodelist". Unless of course one
 MV|>> wants to create One out of thin air. The facts are:

 MV|>> 1) Irex does not need a nodelist.

 RC> Yes.

??? Is that a conirmation or a denial?


 MV|>> 2) Irex can use the nodelist to help create an entry in the node
 MV|>> manager. If it finds connect information in the nodelist when
 MV|>> creating a new entry in the nodemager, it will ask if you wish to
 MV|>> use that information instaed of having to enter it manually.

 RC> Yes I remember that.

 MV|>>    In that it does better than most IP mailers that do not use
 MV|>> the nodelist at all, but not as good as most POTS mailers that
 MV|>> fully use the nodelist.

 RC> Well yes.  It certainy does not do as well as POTS mailers.

Which unfortunately it shares with most other IP mailers. I attribute this to
the Fidonet community failing to reach a standard and inventing a new way to
list Ip nodes every six month. The developpers just gave up trying to keep up
and made Ip mailers that do not need the nodelist. :-(

 RC> However, it does require a compiled nodelist,

On;y *if* one wishes it to use the nodelist at all....

 RC> but does not actually do it, itself.  You need to locate a third party
 RC> util for it to work with the nodelist.  Not a big deal.

An advantage actually. It can use the index files of a number of third party
nodelist compilers. Including V7. So why bother with reinventing the wheel?

 MV|>> I suppose one could call that an issue, but if one does, then
 MV|>> almost all IP malers have issues with the nodelist.

 RC> But of course. :)

I should add that I forgot one more thing that Irex can do with the nodelist:
it cand send Crashmail to nodes not in the node manager using the information
from the nodelist directly. Provided the info is listed in a format it
recognises.

 MV|>> On my personal wish list is that it optionally automatically
 MV|>> updates the connect information in the node manager when the
 MV|>> information in the nodelist changes. F.e. You decide to change
 MV|>> dns provider and your connect information in the nodelist chanGes
 MV|>> from hub2000.darktech.org to hub2000.no-ip.org. Presently I would
 MV|>> have to manually update the information in the node manager.

 RC> To me.  This is an issue for Irex.  It uses the nodelist, but does not
 RC> actually *use* it.

See above: same as many other IP mailers.

 RC> That does seem kind of stupid.  Why bother with using the nodelist at
 RC> all if it doesn't actually take advantage of the dynamic nature of the
 RC> nodelist.

Because the deveoppers got fed up with the childish way the FidoNet community
Kept coming up with new ways to list IP nodes every six month...

 MV|>> I think one should make a difference between items on the wish
 MV|>> list and bugs.

 MV|>> The above is not a bug.

 RC> Perhaps not a bug, but certainly not a good implementation.

If you don't like it, why not vote with the feet instead of keep whining?

The reality of today is that we as a FidoNet community are too small a group to
make demands of the software developers. We just have to make do with the
snippets thrown at us. that's life...


 RC>>> The only thing I do remember about it was that it required a
 RC>>> compiled nodelist before it would use it.

 MV|>> And that you remember wrong. Irex can use the nodelist, but does
 MV|>> not *need* it.

 RC> No.

Yes.

 RC> My experience was that it would do nothing if the nodelist was
 RC> not compiled.  But, I may be wrong.

You are wrong. Irex does not need a nodelist. Period. Of course if you do not
need a nodelist you will have to manually enter all the required data in the
nodelist manager for all the nodes you wish to directly connect to.

Also when you *do* specify that a nodelist should be used, it probably will
complain if it can only find a raw nodelist and no index files.


 RC>>>   The other bit was about the hardcoded fidonet.???  I just can't
 RC>>> remember .net or .org.

 MV|>> .net. It uses fidonet.net as a last resort if it can not find any
 MV|>> other connect info. It is indeed hard coded. Not a big deal as
 MV|>> far as I aM concerned as I don't use that method anyway. I think
 MV|>> it is one of those things that looked like a good idea at the
 MV|>> time but that showed some serious drawbacks on closer inspection.

 RC> Yes.  It was not well thought out.

The fidonet.net idea or the Irex's implementation? I think both.

 RC> Having it configurable would have been better in case the fidonet.net
 RC> domain was changed for whatever reason.

Merely having it configurable would not do. it does not take a fortune teller
to predict what will happen if fidonet.net goes belly up. There will be *more
than one* replacement! That means that some will be reachable via
Newfidonet.net, some via fidonet.xxx and others via fidoclassic.org. :-(

It will have to be configured for more than one damain and it will have to try
them until it finds one that results in a connect. It will have to remember
that one for each link, so that it does not have to try again at the next
connect. Until the connect fails and then it will have to retry.

This is getting messy. Forget about the fidonet.net method, it was not such a
good idea after all. Connect info should be in the nodelist. Period.

 RC>>> My personal opinion is that we need to move to a *standard* for
 RC>>> all Internet nodes.  My personal choice is BinkP as it is the
 RC>>> most "mailer-like."

 MV|>> I am all with you there. This multitude of connection protocols
 MV|>> is a liability rather than an asset.

 RC> Here we go agreeing again.  What is this hobby coming to?  ;)

Down the drain. ;-)

 MV|>> But... of course the best way to promote Binkp as the universal
 MV|>> standard for IP is to stop supporting all the other IP
 MV|>> protocols....

 RC> I stopped supporting other protocols the moment I stopped using Irex.

,10,runekeep.darktech.org,NewsGroup<->FidoNet_Gatewy,Robert_Couture,
000-24-150-164-1150-164-144,300,MO,CM,IBN,IFT,IMI,IUC,
ITX:fidonet@runekeep.arktech.org

Then how about the above? I see four other protocols besides binkp.

 RC> It seems stupid to use so many different protocols.

No one has ever been able to explain the added value.

 RC> It would have been unheard of during the POTS days of FidoNet.
 RC> Imagine having to support 15 different negotiation protocols back
 RC> then.  Preposterous. :)

You are forgetting two things For POTS: 1) there is a common denominator:
FTS-1. 2) There is a fall back mechanism. If the other guy does not support
EMSI, there is FTS-6 to fall back on. If that is not supported either, there is
the fall back to FTS-1. And the fall back is automatic So one can always make a
connect, even without having to know the capabilities of the other side in
advance.

Plus of course that nearly everoyne supports EMSI as that was obviously
superior over FTS-1.

With IP the situation is different. Because of the multitude of protocols and
the lack of a common denominator, there is a fair chance that one can not
connect because one does not share a common protocol.

It is like the situation with power plugs that I run into in South Africa in
The seventees. FIVE different power plugs in use, all incompatible. Now how
does having FIVE different power plugs help to improve connectivity. When you
come in a hotel there is an 80% chance that your raizer plug won't fit in the
receptacle unless you carry four adapters!

At least *they* came to their senses and when I returned in the ninetees for a
bussines trip, they had depricated all but one.


Michiel

--- GoldED+/W32-MSVC 1.1.5-b20060315
 * Origin: http://www.vlist.org (2:280/5555)