[HECnet] DECnet/Python

[Peter Lothberg]

The MCB is a phase-3 node. I guess T20 used the same stuff as T10
for the MCB, code is here: 
http://pdp-10.trailing-edge.com/decnet_mcb_cusps_703a/index.html

It's RSX "based" so maybe Johnny could make it Phase-4? 

On tops10 the problem with the MCB and DDCMP etc was solved with the 
"DDP" device that made a line (sync or async) on a ANF10 DN87* 
frontend a DDCPM device.

That way a Tops10 monitor that was phase-4 could speak sync/async 
to another phase-4 node.

I sent the tapes and disk packs with MRC's 4.x and 5.x tops20 
for 2020 to LCM.

The 4.x DECnet over DDCMP DUP/KMC for 2020 has reduced "segment size' 
from 576 to 390 or something like that due to limited buffer space.


-p 


----- Original Message -----
> From: "tommytimesharing" <tommytimesharing at gmail.com>
> To: "hecnet" <hecnet at Update.UU.SE>
> Sent: Wednesday, October 27, 2021 8:29:35 PM
> Subject: Re: [HECnet] DECnet/Python

> I had been wanting to put some DDCMP code back into later versions of
> Tops-20.  I don't believe it exists in 5, 6 or 7.  I'm not sure if it
> made it into MRC's version of 5 for the KS and those sources have yet to
> be recovered.
> 
> A KL class machine communicates with DECnet via the CI, the NI or a DTE,
> which is a connection to a DN20.  A DN20 is a PDP-11 running DECnet (I
> think it's MCB), but I believe you are limited to Phase III.  The last
> time I was in that particular neck of the monitor woods, I believe I saw
> that the Phase IV router code would not use a DN20 as a router.  I'm
> pretty sure it would hand packets to to any node that the DN20 knew about.
> 
> It didn't look like it would be a big deal to remove that restriction to
> the DN20, but my recollection was that they never spoke full Phase IV.
> So I would need something simulating a 'smart' DN20.  Maybe a brilliant
> DN20.
> 
> Or I could make the KMC/DUP look like a MASSBUS device, adapt the 4.2
> UNIBUS based drivers to use that hardware paradigm and then simulate the
> hardware and then graft the 4.2 DDCMP code to be another transport
> option for the router.
> 
> For some very strange reason, this is my idea of fun.