Re: TITHmailer
By: Deuce to deon on Mon Nov 17 2025 02:49 pm

Howdy,

Yeah, a lot of where FTN ended up is just inertia. There's various proposals from the 80s and 90s rotting away suggesting similar things, but they never caught on in FidoNet, so most of them are half-baked at best.

I agree. Language is also an issue there, where content is written in english that is too generic, open for multiple interpretations of the actual intent.

I too thought about written my version of those specs - but glad to see that you've started that bandwagon :)

It's been an interesting journey even for just the small bit I've done so far, realizing that the point I was starting at wasn't quite right... pretty sure the first thing I'll be creating now will be the bundler, which will be the thing that implements the Binkley-style outbound, then the unbundler, which will implement the inbound, THEN the mailer.

Nice. I was going to work on the mail bundling first (since I noticed your specs starting to define that).

The assumption of default direct has some interesting implications for echomail that I'm still working through, it still keeps collapsing into a simple star topology, which means the SEEN-BY for pretty much every post will either be only two nodes, or a copy of the whole nodelist. It feels like just a usable PATH will do the job, and a robust MSGID will allow loops to be detected and mitigated/fixed.

It might be worth considering multiple star topologies? IE: Distributed, otherwise the network falls apart when the 1 hub at the center of the star topology goes AWOL (which happens too often in othernets)...

But agree, a MSGID should be the definitive determination of duplicate content, and it might be a dynamic value being the "sender's ID" and a "timestamp". There might also need to be a "context" element to to avoid the situation that two users (in two different msg areas), post a message at the exact same "timestamp".

If a message has a digital signature added to it (as a kludge so its not displayed for example), then the origin line could be completly dynamic and cosmetic.

Anyway just spit balling ideas without really thinking them through...


...δεσ∩

---
■ Synchronet ■ AnsiTEX bringing back videotex but with ANSI

Re: TITHmailer
By: deon to Deuce on Tue Nov 18 2025 09:26 am

It might be worth considering multiple star topologies? IE: Distributed, otherwise the network falls apart when the 1 hub at the center of the star topology goes AWOL (which happens too often in othernets)...

So it's really up to the individual sysops... when you can connect directly to any system on the network at no cost, you can link your area from any BBS. This is completely supported today on the existing infrastructure, and the default echomail routing is still somewhat different to the default netmail routing.

The main point is that a tree structure to avoid loops or to minimize costs is obsolete.

But agree, a MSGID should be the definitive determination of duplicate content, and it might be a dynamic value being the "sender's ID" and a "timestamp". There might also need to be a "context" element to to avoid the situation that two users (in two different msg areas), post a message at the exact same "timestamp".

Yeah, even the weakest definitions for MSGID require it to carry both the origin system address and an identifier that is unique on that origin system for a sliding window of at least three years... the "official" standard one in FTS-0009 is pretty terrible (origin followed by a 32-bit hex number that must be magically pulled out of magicland) but even that will work if implemented per the spec (with a separate ticket service that hands out sequential numbers, there's whole other standard proposals for giving out numbers because of this).

Most of the actualy MSGIDs you see on FidoNet are better, so if you just treat the whole thing as a string it'll work... though you still need to build hash trees to be able to deal with them in reasonable timeframes.

If a message has a digital signature added to it (as a kludge so its not displayed for example), then the origin line could be completly dynamic and cosmetic.

TITH won't actually put kludges in the message text, so all the administrative "stuff" that TITH adds will be "somewhere else"... though it will get smeared all over the message text per tradition when it goes via a non-TITH interface. I'm considering initially just having a TITH-only net with everything routed through one system so FidoNet as a whole has a single node to yell at about any bits they don't like.

Anyway just spit balling ideas without really thinking them through...

Yeah, that's how it begins...
---
■ Synchronet ■ The future of BBSing

Deuce wrote to deon <=-

TITH won't actually put kludges in the message text, so all the administrative "stuff" that TITH adds will be "somewhere else"...
though it will get smeared all over the message text per tradition when
it goes via a non-TITH interface. I'm considering initially just
having a TITH-only net with everything routed through one system so FidoNet as a whole has a single node to yell at about any bits they
don't like.

Anyway just spit balling ideas without really thinking them through...

Yeah, that's how it begins...

So, it sounds like this is more involved than maybe originally
thought...

So maybe the naming convention (This Isn't That Hard) should be
re-thunk, and changed to "This Isn't That Simple".

TITSMailer has a nice ring to it, and rolls right off the tongue. ;-)



... If it has tits or tires sooner or later it's going to give you trouble!
--- MultiMail/Linux v0.52
■ Synchronet ■ Palantir BBS * palantirbbs.ddns.net * Pensacola, FL

Re: Re: TITHmailer
By: Gamgee to Deuce on Tue Nov 18 2025 08:17 am

Anyway just spit balling ideas without really thinking them through...

Yeah, that's how it begins...

So, it sounds like this is more involved than maybe originally
thought...

So maybe the naming convention (This Isn't That Hard) should be
re-thunk, and changed to "This Isn't That Simple".

TITSMailer has a nice ring to it, and rolls right off the tongue. ;-)

Heh, it'll still be FidoNet, so it certainly needs to remain complex. It's certainly a better name (TITHmailer is terrible). Alas, the TITH project already has "tith" in to many places for me to be arsed to change it.

That said, for the mailer, I can certainly go with "This Is That Simple" since the mailer will just take up to one bundle from here and put it there, and take up to two bundles from there and put them here. Of course, the server needs to parse the bundle on the fly, so it gets a bit complex again.
---
■ Synchronet ■ The future of BBSing