Letters to the Editor

• Lisa Gronk explains mesh connections
• Alec Grynspan disputes facts
• John Souvestre notes increased publicity
• Dallas Hinton debunks Email Surcharges

Echomail, Zone 1 Backbone
by Lisa Gronke, 1:105/40

In last week's FidoNews, you wrote....

> The current issue rocking the tranquility of echomail distribution
> seems to be an ongoing argument between John Souvestre and Alec
> Grynspan. John has, in the recent past, disconnected from all NAB
> Star systems except George Peace in an effort to prevent the
> generation of duplicates. Alec has recently expanded his echomail
> connections to span both the Z1B distribution and the NAB
> distribution, effectively establishing a mesh which can lead to
> duplicates.

There seems to be some confusion about the meaning of the word "mesh."

I view "mesh" as a synonym for "fully connected polygon," a topology that uses standard FidoNet SEEN-BY technology and does not lose messages or create dupes. It is true that a "fully connected polygon" which is broken (i.e. not fully connected) will generate dupes. The solution to that is to either repair the broken connection or change the topology to hub-and-spoke (incidentally, a "broken triangle" IS hub-and-spoke).

Alec, on the other hand, is trying to implement redundancy (i.e. receiving more than one copy of each and every echomail message) with software that was specifically designed to avoid redundancy.

NNTP, for example, is designed for redundancy.

server: I've got news article 12345678@psg.com
client: gimme
server: here it comes
server: I've got news article 23456789@agora.rdrop.com
client: don't send, already have it
etc.

Using NNTP, a client can connect to as many news servers as he wishes (until the gimme ratio becomes zero).

Alec wants to receive two or more copies of each echomail message, sending the first one he receives to his downlinks and throwing the second, third etc. copy away. One problem he has is that "duplicate" messages from different uplinks are not really identical (they have different SEEN-BY lines).

(----)

I figured that any multipath connection is a mesh of sorts, whether seen-by's control the dupes or not. However, you're the term-meister... I'll find another word in the future. How about "mess?"

Isn't what Alec is doing somewhat the same as the sequence you describe with the NNTP server, except that he's using CRC calculations for the message identifier (or MSGID supplied), and he's discarding the mail he doesn't want rather than telling the server not to send it:

server: Here's message 83923
client: got it for the first time, I'll pass it on.
server: Here's message 69364
client: got it for the third time... in the trash!

You raise a good point that Alec will receive different SEEN-BY lines depending on which path gives him a particular message first. But isn't this somewhat analogous to the ZHUB practice (past if not current) of stripping all SEEN-BY's except for their immediate downlinks (Tiny SEEN-BY's)? How was that controlled?

To be honest, I don't see why duplicates aren't a bigger problem than they appear to be. With everyone able to tunnel through the internet and get echomail anywhere they want, there can't really be a pure implementation of a fully connected polygon... someone's going to open it up somewhere. And most tossers don't store CRC codes long enough to detect duplicates which are received more than a few days apart.

Doug Myers

Echomail from Zone 1 Backbone
Alec Grynspan, 1:341/11

>> generation of duplicates. Alec has recently expanded his echomail
>> connections to span both the Z1B distribution and the NAB
>> distribution, effectively establishing a mesh which can lead to
>> duplicates.

Doug, please don't play reporter when you don't have the facts.

1. I've been connecting to both backbones since day 1. The only problem until recently was that I used a single outbound.

No dupes in over a year.

2. I am not intending to establish a mesh. I am ensuring the most complete feed - for reasons that go beyond FIDO.

Consider what's been happening with your Email feed. The mail moves thru here fast enough - and goes nowhere because of bottlenecks in ERN. To sites more directly connected, the speed is in minutes.

Also - John did not disconnect from the nodes to ensure no dupes. That came after he had already disconnected from 104/1.

(----)

Alec, I appreciate your efforts to provide facts where you feel they are missing, but I don't need your demeaning advice about my play time. We're all playing here - that's the nature of a hobby. I don't plan on criticizing you for playing mail mover :)

Doug Myers

Netmail from John Souvestre, 1:396-1

> 2) There are more problems with echomail distribution
> than there were when the structure was more monolithic.

Wrong. They are just more public now since multiple distribution systems are involved.

Netmail from Dallas Hinton, 1:153/715

Your article about email surcharges is, I'm glad to say, a hoax. It's been reported and debunked several times over the past 2 or 3 years.

There is a web site (I've forgotten the address) which tracks such hoaxes -- you might want to check it out....

Cheers... Dallas

I kinda figured it was a hoax. It sounded like one of those urban legends which finds a life of it's own, not impeded by reality. I wasn't really trying to alert anyone... it just made a good lead in to the rest of the editorial's caustic remarks :)

Doug Myers