Text 35183, 87 rader
Skriven 2009-08-19 19:26:30 av Grant Taylor (108634.fidonews)
     Kommentar till en text av Nick Andre (108633.fidonews)
Ärende: Re: FidoNews 26:32 [02/05]: Rebuttals To Previous Articles
==================================================================
  Re: Re: FidoNews 26:32 [02/05]: Rebuttals To Previous Articles
  By: Nick Andre to Grant Taylor on Wed Aug 19 2009 09:30 pm

 > I've read this thread and decided to chime in with my own two cents...

Welcome.

 > I understand what you're trying to get at with the whole IPv6 thing but
 > unfortunately, there is just not enough incentive to write such a "helper"
 > to "shim" the legacy FTN/BBS stuff to play nice on IPv6.

You make a very good point.  Will there be enough demand to get over the
initial momentum for someone to actually write a shim.

 > And if there was, it takes a considerable amount of effort to diagnose,
 > troubleshoot and properly test to make sure that such a shimmy allows the
 > FTN/BBS stuff to work 100% perfectly as it would in its PSTN existance.

Thus far I have talking about a shim to interface between IPv4 and IPv6.  So I
don't know that it is appropriate to fall all the way back down to the PSTN
level that modems operate at.  Though doing so would be far more efficient and
problematic as you point out.

 > I think a good solution, if it REALLY came down to "IPv6 or die", would be
 > to run the legacy stuff in a Boches or VMWARE emulated session as we now
 > live in the age of multi-core CPU's and hardware that is light years ahead
 > of what we ever dreamed of back in the "glory days" of AT's with their
 > ultra-fast 16Mhz CPU's.

I think the "IPv6 or die" will be very difficult to enforce (other than the
lack of available IPv4 addresses).

I don't see how any sort of virtual machine will make a difference.  You still
have to have some form of connectivity between systems.  If said connectivity
is IPv4 based we are back at Michiel's original posting.

 > Thats far easier than writing shimmies. When Ray Gwinn wrote that Vmodem
 > product that Michiel mentioned, it took him a considerable amount of time
 > testing the shit out of it to make sure that it worked perfectly. And it
 > was written in a time where authors of FTN mailers and BBS software were
 > still around. None of them are around anymore... except a small few.

I agree that writing what was effectively a modem emulator / gateway would be
very difficult.

Ultimately we are dealing with character data that was transmitted across a
serial line with out any special packet headers or flags.  Everything was
originally character based.  Thus it should (I say should because I don't have
first hand experience) be trivial to write something that receives said
character stream via IPv4 and sends it via IPv6 and vice versa.

 > Take Synchronet as an example. Its open-sourced but you'd have to be
 > extremely knowledgable not with just C but with understanding exactly whats
 > happening in the low-level communications stuff, with each of its server
 > modules like its mail, file transfers and all that.

Yes, to implement IPv6 support directly in Synchronet.

 > FTN mailers really wouldn't be that tough - as most of them are
 > fundimentally build around PSTN. What you need to get your head around is
 > that you CANNOT change how they work. Forget about this stupid "mailer over
 > Telnet" shit because shimmying that is going to be more trouble than its
 > worth. Trust me, you will end up screwing around for months, if not a few
 > years, trying to re-tool and fine-tune a legacy PSTN mailer to integrate
 > with the Internet. Its gratifying work, don't get me wrong... but... its
 > *work*.

I think BinkD is a perfect example of this.  It is an FTN mailer on one side
and an IPv4 application on the other side.  I'm hoping that someone either
extends BinkD or uses it's FTN portion and adds IPv6 support.

Using something like BinkD means that SysOps will just have to revert to an
external mailer rather than using the one that might be integrated in to their
BBS.

 > The EMSI, Zmodem and Wazoo protocols are fairly straightforward but the
 > implementation of them is (was) a total nightmare to do. With the exception
 > of Binkleyterm and another program, the source code for all FTN mailers is
 > no longer around.

Does Husky come in to this at all?  (I have not looked close enough to know.)



Grant. . . .
--- SBBSecho 2.12-Win32
 * Origin: Vertrauen - vert.synchro.net (1:103/705)