Subj : binkley multinode, was something new
To : Sean Dennis
From : mark lewis
Date : Thu Oct 15 2009 07:18 pm
ml> i do EMSI with zmodem transfers over here all the time... the problem
ml> with using the older FTN transfer protocols over telnet is the
ml> timing... the older protocols (xmodem, ymodem, zmodem and all their
SD> It's also the protocol driver.
right... the protocol driver (the actual protocol executable) is where the
coder would have to tighten or loosen the timing windows so the protocols would
work over slower connection tunnels...
SD> I use P for OS/2, a real 32-bit driver that's better than anything
SD> else I've ever used, save for PD-ZModem for DOS. Absolutely
SD> painless to use and really fast to boot; it was written by someone
SD> who knew OS/2 really well. PD-ZModem was written by someone who
SD> knew how to optimize their DOS stuff to the T. Used PDZ with
SD> ProBoard under DOS (OS/2-DOS) and it was amazingly fast, almost as
SD> fast as P!
SD> But I digress. :)
i know the feeling... when joho had to rewrite FD and did it with an eye
specifically aimed at using the internet for the connection, he had to tweak
several of the protocols for looser timing windows to allow those ACKs and such
plenty of time to return thru the congested pipeline... since he had the 3rd
party sources for the protocols (more expensive library licensing) he was
using, this was a pretty trivial matter... the real problem comes with software
that uses 3rd party libraries that are only available in compiled unit or obj
format (free or low cost licensing)... without the sources or some way to
adjust the timings via parameters in the routine's call vectors, these are
frought with peril... a very congested pipeline and even tcp/ip packets
arriving out of order can easily cause a protocol to stop and backup or even
miss ACKs and NAKs...
