On Feb 25, 2020, at 2:33 AM, Rob Jarratt
<robert.jarratt at ntlworld.com> wrote:
The fact that the bridge turns everything into a single Ethernet segment is
one of the reasons why I wrote my own user mode router. It interoperates
with the bridge and also SIMH DDCMP (although I haven't tried that for a
while and it looks like I haven't documented the fact either!).
Regards
Rob
I should mention that Rob's work was what inspired PyDECnet.
On Feb 25, 2020, at 5:22 AM, Keith Halewood
<Keith.Halewood at pitbulluk.org> wrote:
All the negative comments I?ve seen about Multinet...
I?ve never had a problem with it, connecting nodes across the Internet via TCP.
Yes, it often can look ok.
Part of my dislike of it comes from the fact that it is obvious it was created by people
who had never paid serious attention to the DECnet architecture specifications. And part
of it comes from the hacks and workarounds I needed to make in PyDECnet to get it to work
-- for some definition of "work". The need for those hacks is a direct
consequence of the ignorance of the Multinet designers.
I know from experience it can never be truly reliable in the sense of dealing well with
nodes restarting or flaky networks. The most I can possibly do is work around its
fundamental design errors to the point that links will, in general, come up eventually and
will stay up if the winds are fair.
paul