[HECnet] ANF10 network --> DN200 (RJE, Decnet) --> should be Ddcmp --Re3

[Paul Koning]

Correction...

Character mode sync in SIMH seems utterly trivial; just treat it as an async stream except that the device would supply sync bytes when the software isn't feeding it data for a while.  Perhaps just a couple, just enough to act as a boundary; no need to flood the connection with data.

HDLC mode is different; HDLC is inescapably a packet mode service since the flags are packet delimiters.  A small complication is that it can carry any bit count, not just whole bytes.  So the network encoding of a simulated HDLC device needs to handle that.  A UDP encoding could just be a packet containing the data bytes, with one byte appended to specify how many bits in the last data byte are meaningful.  (So a typical whole-byte packet might be "abcdefgh\010".)  A TCP encoding  could be similar with a packet length at the start.  If it's desirable not to buffer whole frames, the stream could be broken into fragments with the last one (where the flag appears) marked as last.  That would be necessary if arbitrary length frames are wanted, though a sensible protocol doesn't use frames longer than a thousand bytes or so since the 16-bit CRC isn't strong enough for really long frames.

	paul

> On Nov 18, 2021, at 7:12 PM, Paul Koning <paulkoning at comcast.net> wrote:
> 
> It seems like a very simple matter to simulate it as a basic character device, separate from the KMC pairing.  The only real question is how you represent a sync data stream in a simullated device.  Is it just the byte stream over a TCP connection?  That would be easy for synchronous character mode.  If  you also want to simulate HDLC sync devices, that's a different matter entirely; there isn't an obvious way to do that.
> 
> In other words, if someone wants this, just design how it communicates and do the code, it doesn't sound hard.
> 
> 	paul
> 
>> On Nov 18, 2021, at 6:35 PM, R. Voorhorst <R.Voorhorst at swabhawat.com> wrote:
>> 
>> No, the Dup is only simulated as Kdp/Dup combo, hence the Ddcmp characteristics of the pair.
>> 
>> 
>> Reindert
>> 
>> -----Original Message-----
>> From: owner-hecnet at Update.UU.SE [mailto:owner-hecnet at Update.UU.SE] On Behalf Of Johnny Billquist
>> Sent: Friday, 19 November, 2021 00:31
>> To: hecnet at Update.UU.SE
>> Subject: Re: [HECnet] ANF10 network --> DN200 (RJE, Decnet) --> should be Ddcmp --Re2
>> 
>> Fair enough.
>> 
>> But DUP11 is a sync character based interface. So that one too requires that you do DDCMP in software. But I didn't think simh simulated a DUP11?
>> 
>>  Johnny